Residential SHS Exposure

Transcription

Using Computer Simulation to Explore Multi-Compartment
Effects and Mitigation Strategies for Residential Exposure to
Secondhand Tobacco Smoke
by
Neil Edward Klepeis
B.A. Chemistry (Colgate University) 1989
M.S. Chemistry (Stanford University) 1992
A dissertation submitted in partial satisfaction of the
requirements for the degree of
Doctor of Philosophy
in
Environmental Health Sciences
in the
GRADUATE DIVISION
of the
UNIVERSITY OF CALIFORNIA, BERKELEY
Committee in charge:
Professor William W. Nazaroff, Chair
Professor Robert C. Spear
Professor Michael E. Tarter
Spring 2004
The dissertation of Neil Edward Klepeis is approved:
Chair
Date
Date
Date
University of California, Berkeley
Spring 2004
Using Computer Simulation to Explore Multi-Compartment Effects and
Mitigation Strategies for Residential Exposure to Secondhand Tobacco Smoke
c
Copyright 2004
by Neil Edward Klepeis
All Rights Reserved.
1
Abstract
by
Neil Edward Klepeis
Doctor of Philosophy in Environmental Health Sciences
In this dissertation, I quantitatively explore parameters influencing residential
exposure to secondhand tobacco smoke (SHS). I use a dynamic, multizone exposure simulation model to generate individual and population inhalation exposure
metrics for carbon monoxide, particles, and nicotine, studying how SHS exposure
changes in response to several key variates: (1) different numbers of well-mixed
zones; (2) variation in scripted and empirically observed smoker and nonsmoker
location patterns; (3) household door and window positions; (4) variation in air
flow patterns; and (5) pollutant-specific dynamics. I simulate cases involving unrestricted occupant behavior and when conscious strategies are used to mitigate
exposure. The model employs estimates of physical and environmental parameter
inputs that are representative of conditions in a typical US residence. Two house
types are considered, one dominated by a single, well-mixed zone, and one consisting of four distinct main rooms and a central hallway. The results of simulation
experiments show that the multi-compartment character of a house substantially
influences 24-h average residential SHS exposure concentrations for nonsmokers.
Depending on occupant location patterns, exposures occurring in a house with
multiple main rooms can be substantially larger or smaller than exposures occurring in a house that is well represented by a single, well-mixed room. The loading
of reversibly sorbed nicotine onto household surfaces can result in a doubling of
2
exposure concentrations. The operation of the air handling system generally decreases exposures, because of increased infiltration from duct leaks. The operation
of the air handling system or building cross-flow may slightly increase exposures
for some individuals due to an increased rate of SHS transport from active smokers
to nonsmokers in different rooms. Short of a total ban, the most effective particle
exposure mitigation strategies involve isolation of the active smoker or a ban on
smoking when the nonsmoker is at home. The potentially more practical strategy
of closing doors when house occupants follow unmodified location and smoking
patterns is not very effective. Opening windows or operating particle filtration
devices are, by themselves, moderately effective.
Professor William W. Nazaroff
Dissertation Committee Chair
1
Summary
by
Neil Edward Klepeis
Doctor of Philosophy in Environmental Health Sciences
The purpose of this dissertation is to contribute a more comprehensive and
thoroughly quantitative look into the circumstances surrounding residential exposure to secondhand tobacco smoke (SHS) than has been offered previously. I aim
to provide information on the magnitude of the effects that various environmental
and behavioral factors can have on exposure. With the advent of relatively complete and advanced knowledge of SHS pollutant dynamics and emissions characteristics, US building characteristics, and, especially, representative time-activity
and time-location profiles of the US population, an exploratory investigation into
residential SHS exposure, which synthesizes a broad array of available and relevant data, is timely and promises to highlight important relationships and reveal
fresh insights. The findings may be useful to health practictioners, to public health
researchers, and to those seeking to advance the science of exposure.
The dissertation consists of four main parts plus a set of technical appendices.
Part I contains chapters giving a broad overview of the dissertation and background material on the current state of knowledge for the health and exposure
aspects of SHS and the science of exposure assessment in general. Part II develops a sophisticated event-based (mechanistic) simulation model of residential SHS
exposure by first summarizing results from a selection of previously published
and unpublished studies on emissions, housing characteristics, and human activ-
2
ity patterns, which will be used as specific model inputs, and then outlining the
structure of the exposure model. Part III applies the simulation model developed
in Part II as part of three distinct approaches or tiers, each building on the last.
Tier I consists of a preliminary analysis of multi-compartment effects based on
scripted time-location patterns. For this tier, I examine intermediate simulation
output variables, including air flow rates and profiles of room concentrations, in
more detail than for the other tiers. In Tier II, I make use of observed human
room-to-room residential location patterns to estimate exposure frequency distributions. Tier III is a systematic analysis of mitigation strategies and their effects
on the frequency distribution of SHS exposures for a selected cohort of individuals
who spend the majority of their time at home during a 24-h exposure period and
in whose homes more than 10 cigarettes are smoked during the same time period.
Part IV of the dissertation contains an evaluation of the simulation model and an
overall summary and conclusions.
I review background material in Part I, Chapter 2, summarizing many efforts
relevant to SHS and health, including exposure surveys, indoor air quality monitoring and modeling, exposure prevalence, health effects, and health interventions.
I also present a basic definition of exposure and methods of exposure measurement
and discuss current efforts in activity-based exposure simulation. In general, exposure is defined as the confluence of an agent and a target in both time and space.
The adverse consequences of being exposed to another’s smoke are firmly established, and include the cause or exacerbation of asthma, as well as increased risk
of heart disease and lung cancer. With a fairly high smoker prevalence of 20−30%
in the US, exposure to SHS still presents a problem, although a number of intervention trials suggest that further education and attempts to mitigate exposure
can reduce in-home exposures and perhaps lead to a higher incidence of quitting.
For many studies, there is a lack of detailed information on mechanisms of exposures associated with current intervention and education efforts. The results of the
current modeling study will help to inform these efforts. Activity-based human exposure models and multizone models of indoor air spaces have been established
3
since the 1980’s. A primary issue of concern with regard to the accuracy of indoor
air models has been the rapidity of mixing within single zones and the estimation
of interzonal flow parameters. Although small-scale studies provide evidence that
the multizonal character of homes can have an impact on exposure, the degree to
which the restricted movement of pollutants and human beings among different
rooms of a house can affect exposure is currently unclear. An aim of the current
study is to address this question.
In Part II, Chapter 3, I establish the state-of-knowledge regarding particle, carbon monoxide, and nicotine emissions from tobacco products, since these species
are the ones I use to represent the many and diverse components of SHS. In addition, I introduce a general method of inferring size-specific mass emission factors
and deposition rates for indoor sources that makes use of an indoor aerosol dynamics model, measured particle concentration time series data, and an optimization routine. I apply the method using data from original chamber experiments.
In Part II, Chapter 4, I describe and analyze data on room-to-room human
movement patterns from an important modeling resource, the National Human
Activity Pattern Survey (NHAPS), which was performed for the USEPA starting
in 1992. The NHAPS database contains minute-by-minute timelines for thousands
of Americans as they traveled between the individual rooms of their homes, as
well as to work, school, and on various modes of transit. These data on room-toroom movement are used in this dissertation to provide the basis for simulating
realistic variation in residential exposure to SHS.
In Part II, Chapter 5, I present estimates for ventilation and flow-related parameters and dimensions for detached homes in the US and elsewhere. A typical American detached residence consists of a one or two story house with 4 to 6
rooms and a volume of 300 m3 . House leakage rates, which are induced by wind
and indoor-outdoor temperature differences, have a central tendency towards 0.5
h−1 . A variety of studies provide information on inter-room air flows, which are
on the order of 100 m3 h−1 when separated by an open doorway, and air flows
through open windows, which are commonly between 100 and 200 m3 h−1 . The
4
design of a mechanical air handling system of a house may include the deliberate
introduction of outdoor air into the house, i.e., it may provide forced-fan ventilation. In this case it is called a heating, ventilation, and air conditioning (HVAC)
system. However, most homes in the US have HAC systems, which do not include
a ventilation component. Therefore, the houses I simulate as part of the current
work are equipped with HAC, rather than HVAC, systems. In practice, HVAC
and HAC systems have unwanted leaks in their ductwork, which can lead to the
increased infiltration of outdoor air into the house when system fans are activated.
Part II, Chapter 6 introduces an original SHS exposure simulation model, which
takes time-activity profiles for one or more smoker-nonsmoker pairs as its fundamental input and deterministically calculates exposure metrics over a single 24-h
period. The simulation model, which is the centerpiece of my dissertation research,
constitutes a module of a more general exposure analysis modeling package for
conducting research in human exposure, which is described in Appendix D. The
model is not intended to produce stochastic simulations such as might be used in
a risk assessment investigation, but rather to generate distributions of exposure
over a range of carefully controlled environmental and behavioral factors. After
the selection of housing characteristics, the position and timing of smoking events
are uniquely assigned based on: (1) a fixed number of cigarettes smoked during
the day; and (2) the location profile of the smoker. Together, the defined residential
environment and source activity are used to calculate a minute-by-minute pollutant concentration time series for each room of the house. The nonsmoker exposure profile is determined by matching the nonsmoker location and room concentrations in time. In addition to 24-h exposure concentrations, the model reports
the nonsmoker’s 24-h individual intake fraction (the cumulative mass of cigarette
emissions inhaled by the nonsmoker divided by the total mass of cigarette emissions into the house over a 24-h period) and the equivalent ETS cigarette intake
(mass inhaled divided by the mass emissions of a single cigarette) for each simulated smoker-nonsmoker pair. Underlying the model is the numerical solution of
a series of coupled linear differential equations, where each equation accounts for
5
pollutant levels in a single air or surface compartment.
As part of the Tier I initial simulation analysis of 24-h residential SHS exposure
to particles and nicotine (Part III, Chapter 7), I define scripted nonsmoker location
patterns corresponding to the extremes of “follower” and “avoider” behaviors in
which the nonsmoker’s movement is either perfectly correlated or perfectly anticorrelated with the smoker. I also define a third intermediate nonsmoker mobility
pattern in which the nonsmoker spends a portion of the day in the company of the
smoker. On top of these scripted activities, I explore the effect of door and window
positions, the preloading of walls with nicotine, symmetric as well as asymmetric
flow in house air movement patterns, and the operation of a central mechanical air
handling system. The air handling system provides heating and air conditioning
(HAC), but does not introduce outdoor air by design. Its operation is assumed to
increase the house infiltration due to leaks in the supply ductwork. To simplify
the results, simulations were limited to two 287 m3 houses, one dominated by a
single, large zone and one consisting of four main rooms, plus a bathroom and
a central hallway. All physical and environmental model input parameters are
assigned one or two point estimates, which are determined in Part II to be common
or reasonably representative values for a typical US residence.
Based on the Tier I simulation experiments in Chapter 7, I find that the multicompartment character of a house heavily influences the exposures of nonsmokers.
Particle exposures occurring in the 4-room house can be twice as large as exposures
in the house with a single, large room, in which prolonged time spent away from
the smoker is not possible. When going from “follower” to “avoider” behavior,
the 24-h average particle exposure concentration can decrease by a factor of 3. The
variation in 24-h exposure concentrations for nicotine with respect to nonsmoker
activity pattern is even more dramatic. The sorption of nicotine onto household
surfaces results in an approximate halving of exposure concentrations. The operation of the HAC system decreases exposures because of increased infiltration,
and asymmetric flow patterns through the house slightly increase exposures for
nonsmokers spending time downwind from the active smoker.
6
The Tier II and Tier III analyses, contained within the second and third chapters
of Part III (Chapters 8 and 9), build on the scripted simulation analysis in Chapter 7 by introducing variation in human location patterns inside a residence and,
thereby, generating complete frequency distributions of 24-h exposure metrics. To
isolate the effects of human mobility, all other model inputs are kept at the same
point values as in the Tier I analysis. The Tier II analysis (Chapter 8) considers
circumstances that are not associated with conscious efforts to mitigate exposure,
such as intermittent or continuous HAC operation and long-term nicotine loading
of household surfaces. The broad findings for these simulation experiments are
similar to those for the scripted scenarios in Chapter 7.
For the Tier III analysis in Chapter 9, I focus on specific changes in human behavior, including isolation of the smoker from the nonsmoker, changes in door and
window positions, and the operation of particle filtration devices, for the purpose
of reducing, or mitigating, SHS particle exposure. Each strategy is compared to the
frequency distribution of exposure for a base scenario, for which no conscious mitigation strategies are enforced. The median 24-h average base SHS particle exposure concentration was 32 µ g m−3 . The two most effective SHS particle mitigation
strategies involve: (1) the isolation of the smoker in one room of the house where
the door is closed and the window is open; and (2) a temporal ban of smoking in
the house during times that the nonsmoker is at home. These strategies result in a
median difference from the base condition of about 30 µ g m−3 . Closing the door to
the room containing the isolated smoker is also, by itself, an effective strategy with
a median difference of about 20 µ g m−3 from the base condition. These strategies
represent extreme scenarios, which may or may not be practical. However, potentially more practical strategies involving the closing doors during the “natural”
location patterns of either the smoker or the nonsmoker are not very effective at
reducing exposure. After smoker isolation and temporal bans, the next most effective mitigation strategies involve the continuous use of particle filtration devices
in smoking rooms or the opening of one or more windows by the house occupants
during smoking episodes. These two strategies result in exposure reductions that
7
are close to the case of smoker isolation with the door closed.
In Chapter 10 of Part IV, I evaluate the simulation model with respect to observed SHS-associated concentrations. The model predictions of time-averaged
SHS room concentrations and personal exposure concentrations are in generally
good agreement with average SHS-associated concentrations measured during a
number of field studies for comparable settings, averaging times, and number of
cigarette sources. However, the model does not take into account potentially large
transient peaks that are observed when concentrations are monitored close to a
cigarette source. Therefore, the model predictions in this dissertation are most
relevant for nonsmokers and smokers that do not spend time in close proximity.
In addition, understanding of the behavior of sorbing chemical species, such as
nicotine, needs to be refined in part so that predictions of indirect SHS exposures
resulting from nicotine desorption can be more accurately characterized. In the future, model predictions should be systematically compared to empirical distributions of personal exposure determined from either intensive studies using scripted
activities or large-scale surveys of population exposure, such as the USEPA’s Particle Total Exposure Assessment Methodology (PTEAM). In addition to evaluating
model performance, these comparisons can serve to calibrate the exposure simulation model and interpret particular features of observed distributions.
In Chapter 11 of Part IV of this dissertation, I provide an overall summary and
conclusions for the work presented in this dissertation. I outline the features and
advantages of the modeling approach I have used, suggest possible improvements
to the modeling framework, and discuss the work in the context of public health research, education, and future research in human exposure assessment. The knowledge and understanding imparted by this dissertation can benefit SHS researchers
in epidemiology, risk assessment, health intervention, public outreach, and help
in the establishment of guidelines for SHS-related indoor air quality. Future exposure research should involve the intensive study of source-receptor proximity
effects, the monitoring of pollutant dynamics in real homes, the collection of longitudinal activity pattern data for multiple household members, and the recognition
8
and characterization of the complex human relationships occurring within household ecologies.
Appendices A−D contain supplementary technical information on topics related to activity patterns, exposure modeling, and indoor air quality modeling.
Appendix A presents raw NHAPS data on residential location patterns in the US,
broken down by various demographic groups. Appendix B presents the derivation of mathematical forms for single and multizone indoor air systems, which are
used in the current work to simulate room concentrations of SHS particles, nicotine, and carbon monoxide. Appendix C presents an interactive program for estimating two-compartment model parameters, implemented in the Perl programming language with graphical extensions. Finally, Appendix D describes a software package for simulating human exposure and accomplishing various tasks
in exposure-related data analysis and research. The package was developed and
used as part of the current work and contains a generic framework for managing
exposure calculations. It is implemented in the R statistical programming environment.
i
In memory of my father,
James Emrich Klepeis, Jr.
10.January.1937−27.October.1998
ii
Preface
Many lives have been touched by the use of combustible tobacco products. Both
firsthand and secondhand exposure to tobacco smoke emissions have been unequivocably associated with a wide variety of serious or life-threatening diseases.
Secondhand tobacco smoke has been the most ubiquitous form of indoor pollution in the developed countries and continues to be so in many parts of the world.
While there is a US trend towards smoking bans in indoor workplaces, social
venues, and public buildings, approximately a quarter of Americans continue to
be smokers and considerable secondhand exposure to tobacco smoke still occurs
in households and automobiles, which are locations where children are at risk, as
well as outdoor settings.
My own life and those of my family have been touched by tobacco. My father passed away in 1998 from lung cancer after spending 40 years of his life as
a heavy smoker. My older brother, who may have borne the brunt of household
exposure to tobacco smoke, has suffered from asthma and allergies ever since he
was a small child. Lung cancer and the induction or exacerbation of asthma are
both established health effects that result from tobacco smoke exposure.
As a youth, I used to wrangle with my father over his smoking in the house, the
basement, or the family car, which resulted in fairly effective bans on smoking. I
was also generally obsessed with a potentially unhealthful environment and food
supply. I would sometimes come home from school to complain that my mother’s
dinner was full of nitrites or some other presumably highly toxic constituent. During my morning paper route, I was annoyed at the smelly tailpipe emissions from
the huge and inefficient cars of that era. Once I reached college, a freshman course
iii
in ecology captured my imagination and idealistic spirit more than any other,
for which I wrote an impassioned diatribe against the use of untested food additives, focusing on monosodium glutamate, which is a common flavor enhancer
that causes a variey of abnormal neurologic effects.
Later on in college, I became entranced with the traditional “hard” disciplines
of physics, chemistry, and mathematics. Quantum mechanics in particular was
of special interest to me. I ended up staying in chemistry, probably because it allowed me to dabble in a whole array of aligned areas, most notably computer programming. In my later college years, I became fascinated with simulating physical
phenomena on computers, such as the properties of waves, or using computers to
process complex data, such as in the interpretation of electron diffraction patterns
to determine molecular structure. Ultimately, I ended up in a chemistry graduate
program at Stanford University where I produced a master’s thesis in computational quantum chemistry, studying the energy surface for a fairly obscure little
molecular anion.
The approximately six years I spent pursuing physical science as an undergraduate and graduate student were extremely enlightening and exciting. But upon
leaving the fairly staid east coast for California, I felt an upsurge of dissatisfaction and frustration. Maybe it was being alone in a new place, or culture shock,
or exposure to a host of new social possibilities, or the thought that I might be
stuck studying tiny inconsequential chemicals for the rest of my life! Managing
to finish master’s work at Stanford, I reevalulated my priorities and found that,
while I loved equations and computers, chemicals, by themselves, were actually
very boring things. I was also interested in people, art, literature, music, social
context, connections, and a certain vague thing referred to with the overused term
“the environment.”
I waded through a variety of small-time environmental jobs and volunteering experiences in transportation and woodsmoke activism and education, before
eventually finding a kindred spirit who would lead me back to science, but this
time with a very compelling context, a convergence of disciplines, and more stim-
iv
ulating personalities. This is the point at which tobacco smoke reentered the picture.
When I first met Wayne Ott, he was immersed in a tobacco smoke field study
of bars and restaurants. He was also a Ph.D. environmental engineer, a visiting
scholar in the Department of Statistics at Stanford Univeristy, a USEPA employee,
the author of books on air pollution and statistics, a fierce advocate of science and
Macintosh computers, a jazz afficionado, a connoisseur of good food and conversation, a holder of strong political views, and, as fate would have it, a major player
in the field of exposure science.
Wayne provided the means for me to immediately enter a field of intellectual
pursuit that possessed the desired balance of technical content, societal importance, and nexus for a multitude of converging fields, including building engineering, environmental monitoring, public health, statistics, mathematics, and computers. He also provided a flexible, open, and trusting atmosphere, exuding the thrill
of scientific discovery, and fostering my very rapid assimilation into indoor air and
exposure research. In the first year or so after meeting Wayne, I worked with him
on a state-sponsored Stanford study along with his colleagues Paul Switzer and
John Robinson, helping to conduct studies of tobacco smoke emissions in smoking lounges and homes, and writing my first human exposure simulation model.
I have since had other jobs in exposure science and, as suggested by Wayne, I applied to a Ph.D. program at the University of California at Berkeley, matriculating
in the fall of 1997.
My Ph.D. dissertation reflects a penchant for digging into the technical details
of a problem, especially in terms of traditional mathematics, physical science, and
computing fields, but also my dual inclinations of working toward a meaningful
social and political purpose and synthesizing diverse areas of academic study. Reflecting these interests, I have sought to make my doctoral work technically sophisticated from a physical science perspective, while making my results accessible to
a broad range of scientists in different fields, and to the public at large. My hope
is that this work, and my subsequent efforts, will help to educate and persuade
v
individuals in our society to prevent disease in human beings, as well as damage
to other natural systems, by understanding their causes and taking appropriate
protective steps.
Tobacco smoke may appear to be a particularly easy target for intervention,
because it is not essential to short or long-term survival of our species. The steps
needed to reduce or eliminate exposure in a residential setting may seem straightforward. However, as I have discovered in the course of my doctoral work and also
by growing up in a household with a smoker, many social, economic, and political
factors, both within a household and in the greater society, hold the reins. Effective reduction of exposure involves a mixture of technical and socio-behavioral
measures. This lesson has spurred me to think about new ways to understand how
and why people behave the way they do, forming complex interactions between
themselves and their environment, which, of course, is the realm of geography, sociology, and psychology. In the future, I plan on exploring how these fields can be
fused with current methods in exposure science to better identify and understand
ways of reducing or eliminating human exposure to toxic substances.
Wish me luck :-) . . .
NEK, 28-January-2004
vi
Acknowledgments
As my Ph.D. dissertation research advisor, Professor William Nazaroff has provided important insight and guidance to me in developing research approaches,
posing research questions, and building an appropriate framework for simulating
human exposure. I greatly appreciate, and have greatly benefited from, his sharp
intellect, thoroughness, thoughtfulness, adaptibility, careful organization, frankness, attentiveness, generosity, broad knowledge of indoor air quality, conscientious devotion to my professional development, and, especially, his understanding and appreciation of differences in creative thought processes. His adherence
to high standards of achievement have brought me much closer to my potential in
science, through both his example and his implicit expectation of excellence. He
has not only contributed, in a mentor-student capacity, to my success in this particular endeavor, but will remain a lasting model throughout my career in scientific
research.
From 1999 to 2003, Thomas McKone oversaw much of my research on exposure
modeling and aerosol model optimization in his capacity as the Lawrence Berkeley National Laboratory (LBNL) principal investigator for a US Environmental
Protection Agency (USEPA) University Partnership Agreement (UPA). The UPA
agreement was established at LBNL via Interagency Agreement DW-988-38190-010 with the US Department of Energy (DOE) under Contract Grant No. DE-AC0376SF00098. Halûk Özkaynak of the USEPA National Exposure Research Laboratory (NERL) served as the UPA project director during the tenure of my research
activities. Other members of the partnership were Rutgers University, led by Paul
Lioy and Panos Georgopoulos of the Environmental and Occupation Health Sci-
vii
ences Institute (EOHSI), and Stanford University, led by Professor Paul Switzer
and Wayne Ott of the Department of Statistics.
Michael Apte, Lara Gundel, and Richard Sextro of Lawrence Berkeley National
Laboratory (LBNL) invited me to collaborate with them on an experimental investigation into cigar and cigarette emissions characterization in the summer and fall
of 1998. This study was supported in part by the California Tobacco-Related Disease Research Program (TRDRP) under grant 6RT-0307 to Lawrence Berkeley National Laboratory, and by the Assistant Secretary of Conservation and Renewable
Energy, Office of Building Technologies, Building Systems and Materials, Division
of the US Department of Energy (DOE) under Contract DE-AC03-76SF0098. Additional support for this work was provided by the TRDRP under grant 6RT-0118
to Stanford University, Department of Statistics. I would like to express appreciation for the technical assistance of Doug Sullivan in setting up the chamber experiments and of Scott Baker, a visiting Stanford student, who helped with sample
filter preparation and experimental procedures.
Before beginning the pursuit of my Ph.D. degree at the University of California
at Berkeley in the fall of 1997, I worked in Las Vegas, NV supporting the USEPA’s
exposure research efforts and the analysis of human activity pattern data. I had
the opportunity to work with Andy Tsang, a statistician studying at the University
of Nevada at Las Vegas (UNLV), and Joseph Behar and William Nelson, both of
USEPA-NERL. I thank Andy for introducing me to approaches in analyzing large
survey data sets, as well as enlightening me during many statistically-oriented
discussions. I thank Joe for imparting wisdom and perspective, and both Joe and
William for recognizing the importantance of my and Andy’s work and working
hard to make it possible.
I would like to express special thanks to Wayne Ott, a 30-year veteran of the
USEPA, and currently a visiting scholar and consulting professor at Stanford University. Wayne led me into the field of public health and exposure science, providing friendship, and invaluable mentorship and guidance during the early stages of
my career in exposure science research. I am indebted to him because of his con-
viii
fidence in me, his frank and honest advice, our thought-provoking discussions,
and his willingness to involve me in every aspect of his research from the very beginning. Much of the groundwork for this dissertation, including both controlled
and field monitoring of airborne pollutants, human exposure modeling, single and
multizone indoor air quality modeling, and, perhaps most importantly, human activity pattern analysis, were laid during my work with Wayne, which began in
1993 and continues to this day. In this respect, Wayne Ott has made a very large
general contribution to this dissertation.
A number of other colleagues have provided encouragement, friendship, collaboration, and/or guidance either early in my career in exposure science, during
my coursework at Berkeley, or on the road to completing my Ph.D. These folks
include Mary Rozenberg, Lance Wallace, Valerie Zartarian, Andrea Ferro, Robert
Spear, Michael Tarter, Kirk Smith, Rob Harley, Dave Mage, Niren Nagda, James
Repace, John Robinson, Kathy Vork, Julian Marshall, Melissa Gonzales, Mike Wilson, Agnes Bodnar-Lobscheid, Laura Gunn, Sarah Jump, Lily Panyacosit, Liza
Ryan, Katherine Zandonella, Geniene Gefke, Rajan Mutialu, Justin Girard, Jennifer Hearst, David Pennise, Chris Kirkham, Chris Erdmann, Jennifer Mann, Sue
Chiang, Stephen Hern, Tom Phillips, Tracy Thatcher, William Engelmann, Sally
Liu, Brett Singer, Bill Riley, Alvin Lai, Christine Little, Doug Sullivan, Ed Furtaw,
Edmund Seto, Lee Langan, Mark Nicas, Max Zarate, Mike Sohn, Mike Green, Tom
Cooper, Dino Capeto, Edward Costello, Peggy Jenkins, Song Liang, Lin Wei Tian,
Tom McCurdy, Leon Alevantis, Joe Eisenberg, John Roberts, Jed Waldman, and
Chin Long Chiang.
The primary software tool I have used throughout my Ph.D. research has been
the freely-available R system for data analysis and graphics,1 augmented by the
GNU Scientific Library (GSL). I used R to implement the simulation model for
secondhand smoke exposures, which was developed and applied as part of my
Ph.D. research, to analyze and summarize both simulated and observed data, and
to create nearly all of the presentation graphics included in this dissertation. I
1 Visit
the R home page at http://www.r-project.org
ix
thank all of the original and continuing contributors to the GSL and R projects for
making their flexible and extremely high quality products available to me, and everyone, at no cost. The R programming environment, which is quite similar to the
popular S-plus system,2 was developed by an inspired and inspiring community
of first-rate educators, scientists, and statisticians who have generously donated
their labor, in terms of software design, programming, and support, for the greater
public good. It appears that R is rapidly becoming the premier tool for modeling,
exploration, analysis, and visualization of data, likely due in no small measure to
the philosophy of openness possessed by its creators.
In addition to R, I also made liberal use of other freely-available software tools,
including the venerable vector drawing program Xfig. I used the free GNU/Linux
operating system as my platform for conducting simulation trials as well as for
general computing. This dissertation was typeset using the free LATEX document
preparation system.
I would like to thank my long-time partner, Natalie Broomhall, for understanding all about this crazy Ph.D. thing and making my life interesting and complete
during much of these past years during which we’ve lived on our isolated Watsonville farm, with me spending most days (and nights!) hunched up over my
dusty computer monitor. As soon as I get those signatures, I’ll make it up to you,
I promise.
Oh, and Obie and Annie (our two labrador-aussie crosses), you guys were a
great help too.
Finally, I want to thank my three brothers, John, Keith, and Peter, for not picking on me too much for being the last Ph.D. in the family, my mother, who always
had great faith in me and made sure that I got into that algebra class in the 8th
grade, placing me on the road to educational fulfillment, and my father, whose
idealism and intense interest in science, and in the entire world around him, still
lives on in all of his sons.
2 See
http://www.insightful.com/products/splus/
x
Layout of This Document
This dissertation consists of 11 chapters and 4 appendices grouped into five parts:
I. Introduction and Background, II. Model Development, III. Model Application,
IV. Model Evaluation, and V. Appendices. On the page immediately after each
part’s title page, I have included a one or two-sentence summary of the contents of
each component chapter. The chapter title, sometimes condensed, is printed along
the top heading of each page. At the end of most chapters in parts II and III, I have
included a section entitled “Summary and Conclusions”, which gives an overview
of the chapter, highlighting its main points or findings. At the end of each chapter,
I include a listing of all references cited in that chapter. At the end of this document
is an index containing page references to key words and ideas occurring in the text.
xi
Contents
Abbreviations
xvi
List of Tables
xviii
List of Figures
xxiii
I Introduction
1 Research Overview
1.1 Broad Dissertation Outline . . . . . . . . . . . . . . . . . .
1.2 Modeling Approach . . . . . . . . . . . . . . . . . . . . .
1.3 Design of Simulation Experiments . . . . . . . . . . . . .
1.3.1 Tier I. Scripted Occupant Movement . . . . . . . .
1.3.2 Tier II. Realistic Variation in Occupant Movement
1.3.3 Tier III. Exposure Mitigation Trials . . . . . . . . .
1.3.4 Summary of Specific Analysis Factors . . . . . . .
1
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
3
4
7
13
14
15
16
17
2 Background
2.1 Defining and Measuring Exposure . . . . . . . . . . . . . . . . . .
2.1.1 Concept and Mathematical Formulation of Exposure . . .
2.1.2 Practical Measures of SHS Exposure . . . . . . . . . . . . .
2.2 Health Risks of SHS . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.3 Field Studies of SHS Exposure and Multiple Compartment Effects
2.4 Children’s Residential SHS Exposure . . . . . . . . . . . . . . . . .
2.4.1 Exposure Prevalence . . . . . . . . . . . . . . . . . . . . . .
2.4.2 Household Restriction Effectiveness . . . . . . . . . . . . .
2.4.3 Intervention Strategies . . . . . . . . . . . . . . . . . . . . .
2.4.4 The Need for Better Exposure Measures . . . . . . . . . . .
2.5 Models of IAQ and Exposure . . . . . . . . . . . . . . . . . . . . .
2.5.1 IAQ Model Validity . . . . . . . . . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
23
24
24
29
30
36
50
50
55
61
66
67
68
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
xii
CONTENTS
2.6
2.7
2.5.2 Exposure Simulation . . . . . . . . . . . . . . . . . . . . . . . . 72
Summary and Conclusions . . . . . . . . . . . . . . . . . . . . . . . . 76
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
II Model Development
93
3 Emissions Characterization
3.1 Human Smoking Patterns . . . . . . . . . . . . . . . . . . . . . .
3.2 Cigar and Cigarette Experiments . . . . . . . . . . . . . . . . . .
3.3 Estimating Particle Emissions with an Aerosol Dynamics Model
3.4 The Size Distribution of Particle Emissions . . . . . . . . . . . .
3.5 Size-Integrated Particle Emissions . . . . . . . . . . . . . . . . .
3.6 Particle Deposition . . . . . . . . . . . . . . . . . . . . . . . . . .
3.7 Emissions and Dynamic Behavior of Gaseous Species . . . . . .
3.8 Summary and Conclusions . . . . . . . . . . . . . . . . . . . . .
3.9 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
95
98
100
104
111
116
120
126
129
134
4 Human Activity Patterns
4.1 Time Spent in Broad Locations Over a 24-h Period
4.2 Time Spent at Home in Different Rooms . . . . . .
4.2.1 Time Spent by Age . . . . . . . . . . . . . .
4.2.2 Time Spent by Gender . . . . . . . . . . . .
4.2.3 Time Spent by Day of Week . . . . . . . . .
4.2.4 Time Spent by House Size . . . . . . . . . .
4.3 Summary and Conclusions . . . . . . . . . . . . .
4.4 References . . . . . . . . . . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
140
143
147
153
153
158
158
163
164
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
5 Housing Characteristics
5.1 Mixing Within a Single Zone . . . . . . . . . . . . . . . . . . . . . . .
5.2 Zone Volumes and Surface Areas . . . . . . . . . . . . . . . . . . . .
5.3 Air Exchange with the Outdoors . . . . . . . . . . . . . . . . . . . .
5.4 HVAC Systems: Recirculation, Outdoor Air Delivery, and Duct
Leakage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.5 Estimates of Interzonal Air Flow Rates . . . . . . . . . . . . . . . . .
5.6 Illustrative Simulation of Tracer Gas Concentrations in a House . .
5.7 Summary and Conclusions . . . . . . . . . . . . . . . . . . . . . . .
5.8 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
165
. 166
. 168
. 172
.
.
.
.
.
178
182
186
193
196
6 Model Structure
200
6.1 Model Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 201
6.2 Treatment of Chemical Species . . . . . . . . . . . . . . . . . . . . . . 203
xiii
CONTENTS
6.3
6.4
6.5
6.6
6.7
Treatment of Residences . . . . . . . . . . . . . . . . . . . . . . . . .
6.3.1 Specification of Volume, Rooms, and Layout . . . . . . . . .
6.3.2 Specification of Air Flow Conditions . . . . . . . . . . . . . .
Treatment of Residential Activity Patterns . . . . . . . . . . . . . . .
6.4.1 Mapping Sampled Occupant Locations to Simulated Rooms
6.4.2 Specification of Mitigation Scenarios . . . . . . . . . . . . . .
6.4.3 Simulation of Smoking Patterns . . . . . . . . . . . . . . . .
Combining House and Occupant Information . . . . . . . . . . . . .
6.5.1 Synchronization of Simulated Events . . . . . . . . . . . . .
6.5.2 Calculation of Room Concentrations and Exposure . . . . .
Summary of Input and Output Simulation Variables . . . . . . . . .
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
III Model Application
228
7 Tier I. Scripted Occupant Movement
7.1 Model Input for Scripted Scenarios . . . . . . . . . . . . . . . . . . .
7.2 Intermediate Output: Occupant Interaction, Air Flows, and Room
Concentrations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
7.3 Simulated Exposures by House Type, Flow Scenario, and Nonsmoker Activity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
7.4 Summary and Conclusions . . . . . . . . . . . . . . . . . . . . . . .
7.5 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
8 Tier II. Realistic Variation in Occupant Location
8.1 Seven Simulation Trials of Unrestricted Exposure . . .
8.2 Base Exposure Distributions . . . . . . . . . . . . . . .
8.2.1 Particles . . . . . . . . . . . . . . . . . . . . . .
8.2.2 Carbon Monoxide . . . . . . . . . . . . . . . .
8.2.3 Nicotine . . . . . . . . . . . . . . . . . . . . . .
8.3 The Effect of Air Flow Patterns . . . . . . . . . . . . .
8.3.1 HAC Operation . . . . . . . . . . . . . . . . . .
8.3.2 Asymmetric Leakage Flow . . . . . . . . . . .
8.4 Comparison of All Unrestricted Scenarios . . . . . . .
8.5 Sensitivity to Environmental and Physical Parameters
8.6 Summary and Conclusions . . . . . . . . . . . . . . .
8.7 References . . . . . . . . . . . . . . . . . . . . . . . . .
205
205
208
213
214
216
217
218
218
220
221
227
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
230
. 231
. 241
. 258
. 262
. 267
.
.
.
.
.
.
.
.
.
.
.
.
268
269
274
274
275
278
284
284
284
294
295
298
300
9 Tier III. Mitigation Strategies
301
9.1 Fixed Simulation Inputs: Cohort and Physical-Environmental Characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 302
xiv
CONTENTS
9.2
9.3
9.4
9.5
9.6
9.7
9.8
9.9
9.10
Programmed Mitigation: Twenty-Five Scenarios
Temporal Smoking Bans . . . . . . . . . . . . . .
Single-Door or Single-Window Strategies . . . .
Door and Single-Window Combined Strategies .
Multi-Window Strategies . . . . . . . . . . . . . .
Smoker Avoidance and Isolation . . . . . . . . .
Portable Filtration Devices . . . . . . . . . . . . .
Summary and Conclusions . . . . . . . . . . . .
References . . . . . . . . . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
IV Conclusions
304
311
311
314
316
322
323
326
327
328
10 Model Evaluation
330
10.1 SHS Concentrations in Rooms . . . . . . . . . . . . . . . . . . . . . . . 331
10.2 SHS Personal Exposure . . . . . . . . . . . . . . . . . . . . . . . . . . . 334
10.3 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 337
11 Overall Summary and Conclusions
11.1 A New Exploratory Modeling Tool . . . . . . . . .
11.2 Findings: Sensitivity of Exposure to Key Variables
11.3 Potential Enhancements to the Simulation Model .
11.4 Improving Public Health Research and Education
11.4.1 Epidemiology . . . . . . . . . . . . . . . . .
11.4.2 Public Health Interventions . . . . . . . . .
11.4.3 Educational Materials . . . . . . . . . . . .
11.4.4 Guidelines for Residential Air Quality . . .
11.4.5 Health Risk Assessment . . . . . . . . . . .
11.5 Future Exposure Research . . . . . . . . . . . . . .
11.5.1 Proximity Effects . . . . . . . . . . . . . . .
11.5.2 Residential Pollutant Monitoring . . . . . .
11.5.3 Residential Activity Patterns . . . . . . . .
11.5.4 Modeling Social Ecologies . . . . . . . . . .
11.6 References . . . . . . . . . . . . . . . . . . . . . . .
V
Appendices
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
339
340
343
345
346
347
347
349
349
350
351
352
352
353
354
356
357
A Raw Activity Pattern Data
359
A.1 Interview and Data Format . . . . . . . . . . . . . . . . . . . . . . . . 359
A.2 Plots of 24-h Time-Location Profiles . . . . . . . . . . . . . . . . . . . 360
A.3 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 371
xv
CONTENTS
B Compartment Model Equations
B.1 Single-Zone Model . . . . . . . . . . . . . . . .
B.2 Generic First-Order Compartmental Systems .
B.3 Multi-Compartment Indoor Air Quality Model
B.4 References . . . . . . . . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
372
373
374
376
380
C Interactive Two-Compartment Computer Program
382
C.1 User Interface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 383
C.2 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 385
D Software Package for Human Exposure Research
387
D.1 The ESM Function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 388
D.2 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 392
Index
393
xvi
Abbreviations
Listed below are some common abbreviations used in this document to refer to
health, professional, or government agencies, chemical species, aspects of building
engineering, statistical parameters, and physical or time units. Items are arranged
alphabetically.
ASHRAE
CARB
CFD
CO
cm
d
ETS
ft
g
GM
GSD
h
HAC
HVAC
IAQ
K
L
m
mg
µg
µm
min
ml
mo
NCI
NO2
American Society of Heating, Refrigerating, and Airconditioning Engineers
California Air Resources Board
computational fluid dynamics
carbon monoxide gas
centimeters, 10−2 m
days
environmental tobacco smoke, a.k.a., SHS
feet
grams
geometric mean
geometric standard deviation
hours
heating and air conditioning
heating, ventilation, and air conditioning
indoor air quality
degrees Kelvin
liters, 10−3 m3
meters
milligrams, 10−3 g
micrograms, 10−6 g
micrometers, 10−6 m
minutes
milliliters, 10−3 L
months
National Cancer Institute
nitrogen dioxide gas
xvii
perfluorocarbon tracer gas
particulate matter with aerodynamic diameters smaller
than 2.5 µ m
RSP
respirable suspended particles, generally equivalent to
PM2.5
s
seconds
SF6
sulphur hexafluoride tracer gas
SHS
secondhand smoke, a.k.a., ETS
TPM
total particle mass
US
United States of America
USDHHS, DHHS United States Department of Health and Human Services
USEPA, EPA
United States Environmental Protection Agency
WHO
World Health Organization
y
years
PFT
PM2.5
xviii
List of Tables
1.1
1.2
1.3
1.4
2.1
2.2
2.3
2.4
Assumptions for the Physical Behavior of Pollutants . . . . . . . . .
Emission Factors, Standard Concentrations, and Figures of Merit for
Carbon Monoxide and Particulate Matter . . . . . . . . . . . . . . .
Summary of Different Combinations of Scenario Factors Considered
in the Analysis of Residential SHS Exposure Using Scripted Occupant Activity Patterns . . . . . . . . . . . . . . . . . . . . . . . . . .
Summary of 31 Different Combinations of Scenario and Pollutant
Factors Considered in the Analysis of Frequency Distributions for
Unrestricted (Tier II) and Restricted (Tier III) Residential SHS Exposure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. 11
. 12
. 20
. 21
Reported Emission Factors for Gas-Phase Components of SHS . . . .
Reported Particle-Phase Components of SHS . . . . . . . . . . . . . .
Different Measures of Residential SHS Exposure . . . . . . . . . . . .
Health Effects Causally Associated with Exposure to Environmental
Tobacco Smoke with Annual Morbidity and Mortality Estimates for
the US . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.5 Environmental Tobacco Smoke Particle Concentrations Measured
During Field Surveys . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.6 Summary of Two-Room SF6 Tracer Experiments in a Large Ranch
Style House . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.7 Average SHS Nicotine and Respirable Suspended Particle Concentrations Measured in Multiple Rooms of Two Residences . . . . . . .
2.8 Surveys on the Prevalence of Household Smoking Restrictions and
Children’s Exposure to SHS . . . . . . . . . . . . . . . . . . . . . . . .
2.9 Surveys of Household Smoking Restrictions and Corresponding Reduction of Children’s Exposure to Secondhand Smoke . . . . . . . . .
2.10 Three Recent Controlled Trial Intervention Studies for the Reduction of Children’s Exposure to SHS Based on Household Smoking
Restrictions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.11 Studies Evaluating Models of Residential Multizone Transport of Indoor Air Pollutants, Single-Zone Mixing, and Source-Proximity Effects
25
26
31
34
38
45
47
52
58
63
69
xix
LIST OF TABLES
2.12 Examples of Some Existing Regulatory and Exploratory Inhalation
Exposure Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
3.1
3.2
3.3
3.4
3.5
3.6
3.7
4.1
4.2
4.3
4.4
4.5
4.6
4.7
California and US Federal Concentration Guidelines for Carbon
Monoxide and Particulate Matter . . . . . . . . . . . . . . . . . . . .
The Estimated Size Distributions of SHS Particle Emissions . . . . .
Reported Size-Specific Tobacco Particle Emissions for Cigarettes and
Cigars . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Summary of Cigar and Cigarette Experiments and Filter-Based SHS
Particle Emissions . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Reported Environmental Tobacco Smoke Particle Mass Emissions
from Cigarettes and Cigars . . . . . . . . . . . . . . . . . . . . . . . .
Reported Environmental Tobacco Smoke Nicotine Emissions from
Cigarettes and Cigars . . . . . . . . . . . . . . . . . . . . . . . . . . .
Reported Environmental Tobacco Smoke Carbon Monoxide Emissions from Cigarettes and Cigars . . . . . . . . . . . . . . . . . . . .
. 97
. 112
. 117
. 119
. 121
. 130
. 131
Overall Weighted Statistics for Time Spent by NHAPS Respondents
and Time Spent in the Presence of a Smoker in Six Different Grouped
Locations Over a 24-h Period Starting at 12:00 AM on the Diary Day
Weighted Statistics for Mean Percentage of Overall Time Spent and
Time Spent with a Smoker by NHAPS Respondents in Six Different
Grouped Locations Over a 24-h Period Starting at 12:00 AM on the
Diary Day . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Overall Statistics for Time Spent by NHAPS Respondents Living in
Detached Homes in Different Rooms of Their Residence Over a 24-h
Period Starting at 12:00 AM on the Diary Day . . . . . . . . . . . . . .
Overall Statistics for Time Spent by NHAPS Respondents in the
Presence of a Smoker Living in Detached Homes in Different Rooms
of Their Residence Over a 24-h Period Starting at 12:00 AM on the
Diary Day . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Statistics for Time Spent by NHAPS Respondents Living in Detached Homes During Continuous Individual Episodes in Different
Rooms of Their Residence Over a 24-h Period Starting at 12:00 AM
on the Diary Day . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Statistics by Age for Time Spent by NHAPS Respondents Living in
Detached Homes in Different Rooms of Their Residence Over a 24-h
Period Starting at 12:00 AM on the Diary Day . . . . . . . . . . . . . .
Statistics by Gender for Time Spent by NHAPS Respondents Living
in Detached Homes in Different Rooms of Their Residence Over a
24-h Period Starting at 12:00 AM on the Diary Day . . . . . . . . . . .
145
146
148
149
152
154
159
xx
LIST OF TABLES
4.8
4.9
5.1
5.2
5.3
5.4
6.1
6.2
6.3
6.4
6.5
6.6
7.1
7.2
7.3
7.4
7.5
7.6
Statistics by Day of Week for Time Spent by NHAPS Respondents
Living in Detached Homes in Different Rooms of Their Residence
Over a 24-h Period Starting at 12:00 AM on the Diary Day . . . . . . 161
Sample Size by Number of Rooms and Floors for NHAPS Respondents Living in Detached Homes . . . . . . . . . . . . . . . . . . . . . 163
Frequency Tabulation for Floor Area and Estimated Volume of OneUnit Residential Buildings by Number of Rooms and Number of
Stories: Unweighted Results from the 2001 American Housing Survey, n=29,356 Telephone Respondents . . . . . . . . . . . . . . . . .
Air Flow Rates Measured During Six Two-Compartment SHS Particle Experiments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Summary of Fourteen Two-Room SF6 Tracer-Gas Experiments on
the Effect of Door Position on Air Movement with Estimated Flow
Rates Between the Source Room and Test Room and the Estimated
Overall Two-Room Air Exchange Rate . . . . . . . . . . . . . . . . .
Steady-State Concentrations for Tracer Gas Simulations of a Continuous Source . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Simulated Separate and Multi-Use Room Types as a Function of the
Number of Main Rooms in a House . . . . . . . . . . . . . . . . . .
Room Categories for NHAPS 24-h Diaries . . . . . . . . . . . . . . .
Model Response Variates . . . . . . . . . . . . . . . . . . . . . . . . .
Model Input Parameters – Explicit Key Variates . . . . . . . . . . .
List of "On" Conditions for the 23 Environmental Scenario Binary
Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Derived Quantities – Implicit Key Variates . . . . . . . . . . . . . .
. 170
. 183
. 189
. 193
.
.
.
.
207
214
222
223
. 225
. 227
Fixed Model Input Parameter Values: Physical and Environmental
Quantities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Levels Considered for Five Model Input Scenario Variables: House
Type, Nonsmoker Activity, Flow-Related Conditions, Flow Symmetry, and Initial Nicotine Surface Concentrations . . . . . . . . . . . . .
Room Characteristics for Each Type of Simulated House . . . . . . .
Percentage of the Day Nonsmoker Spends in Rooms with the Smoker,
At Home During Smoking Episodes, and in Rooms During Smoking
Episodes by Activity and House Type . . . . . . . . . . . . . . . . . .
Simulated 24-h Mean Whole-House Air-Exchange Rate by Flow
Symmetry, Flow Scenario, and House Type . . . . . . . . . . . . . . .
Simulated 24-h Mean Flow Out of Smoking Rooms Into Other
Rooms During Smoking Episodes by Flow Symmetry, Flow Scenario, and Nonsmoker Activity for House #2 . . . . . . . . . . . . . .
234
236
238
242
243
249
LIST OF TABLES
Simulated 24-h Mean Flow Into Nonsmoker Rooms from Other
Rooms During Smoking Episodes by Flow Symmetry, Flow Scenario, and Nonsmoker Activity for House #2 . . . . . . . . . . . . .
7.8 24-h Simulated Mean Particle Exposure Concentration by Flow
Symmetry, House Type, Flow Scenario, and Nonsmoker Activity .
7.9 Correction Factors for Simulated 24-h Mean Particle Exposure Concentration Predicted by a Simple Single-Zone Model Across Flow
7.10 Simulated 24-h Individual SHS Particle Intake Fraction by Flow
7.11 Simulated 24-h Equivalent ETS Cigarette Particle Intake by Flow
7.12 Simulated 24-h Mean Nicotine Personal Exposure Concentration by
Flow Symmetry, Initial Surface Concentrations, Flow Scenario, and
Nonsmoker Activity for House #2 . . . . . . . . . . . . . . . . . . .
xxi
7.7
8.1
8.2
8.3
8.4
8.5
8.6
Descriptions of Seven Unrestricted Residential SHS Inhalation Exposure Scenarios . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Particles. 24-h Average Nonsmoker SHS Particle Inhalation Exposure Concentration, Individual Intake Fraction, and Equivalent ETS
Cigarettes by Number of Cigarettes Smoked Indoors and Fraction
of Time Spent by the Nonsmoker at Home . . . . . . . . . . . . . . .
CO. 24-h Average Nonsmoker SHS Carbon Monoxide Inhalation Exposure Concentration, Individual Intake Fraction, and Equivalent
ETS Cigarettes Statistics by Number of Cigarettes Smoked Indoors
and Fraction of Time Spent by the Nonsmoker at Home . . . . . . .
Nicotine Fresh. 24-h Average Nonsmoker SHS Nicotine Inhalation
Exposure Concentration, Individual Intake Fraction, and Equivalent
ETS Cigarettes Statistics for Fresh Surfaces by Number of Cigarettes
Smoked Indoors and Fraction of Time Spent by the Nonsmoker at
Home . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Nicotine Loaded. 24-h Average Nonsmoker SHS Nicotine Inhalation
Exposure Concentration, Individual Intake Fraction, and Equivalent ETS Cigarettes Statistics for Surfaces Preloaded with Nicotine
by No. of Cigarettes Smoked Indoors and Fraction of Time Spent by
the Nonsmoker at Home . . . . . . . . . . . . . . . . . . . . . . . . .
HAC 10%. 24-h Average Nonsmoker SHS Particle Inhalation Exposure Concentration, Individual Intake Fraction, and Equivalent
ETS Cigarettes Statistics for Intermittent Awake-Time HAC Operation by Number of Cigarettes Smoked Indoors and Fraction of Time
Spent by the Nonsmoker at Home . . . . . . . . . . . . . . . . . . .
. 250
. 260
. 261
. 263
. 264
. 265
. 272
. 277
. 280
. 286
. 287
. 289
LIST OF TABLES
HAC 100%. 24-h Average Nonsmoker SHS Particle Inhalation Exposure Concentration, Individual Intake Fraction, and Equivalent ETS
Cigarettes Statistics for Continuous Awake-Time HAC Operation by
Number of Cigarettes Smoked Indoors and Fraction of Time Spent
by the Nonsmoker at Home . . . . . . . . . . . . . . . . . . . . . . .
8.8 Asymmetric Flow. 24-h Average Nonsmoker SHS Particle Inhalation
Exposure Concentration, Individual Intake Fraction, and Equivalent
ETS Cigar-ettes Statistics for Asymmetric Flow Conditions by No.
of Cigarettes Smoked Indoors and Fraction of Time Spent by the
Nonsmoker at Home . . . . . . . . . . . . . . . . . . . . . . . . . . .
8.9 Calculated Geometric Means and Geometric Standard Deviations
for Distributions of 24-h Mean Nonsmoker SHS Particle, Carbon
Monoxide, or Nicotine Exposure Concentration, Individual Intake
Fraction, and Equivalent ETS Cigarette Intake Across Each Scenario
in Households Where More Than 10 Cigarettes Were Smoked and
Nonsmokers Spent More than 23 of their Time . . . . . . . . . . . . .
8.10 Sensitivity of the Geometric Mean of Nonsmoker SHS Particle Exposure Metrics to Eight Physical and Environmental Parameters . .
xxii
8.7
9.1
9.2
. 290
. 293
. 296
. 299
Descriptions of Each Residential SHS Inhalation Exposure Mitigation Strategy Arranged by Group . . . . . . . . . . . . . . . . . . . . . 307
Statistics from the Simulated Distribution of 24-h Nonsmoker SHS
Particle Inhalation Exposure Concentration for each Exposure Mitigation Strategy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 312
10.1 Comparison of Simulated and Observed SHS Respirable Suspended
Particle and Nicotine Concentrations Measured in Rooms of Residences or in a Furnished Chamber . . . . . . . . . . . . . . . . . . . . 333
A.1 Example 24-h Recall Diary Containing Beginning & Ending Times,
Activity, Location, Presence of a Smoker, and Time Spent for 22 Microenvironments Visited on the Diary Day . . . . . . . . . . . . . . . . 361
A.2 The Original NHAPS 24-h Recall Diary Locations . . . . . . . . . . . 362
A.3 The Original NHAPS 24-h Recall Diary Activities . . . . . . . . . . . 363
B.1 Response Variables and Observable Physical Input Parameters of a
Multi-Compartment Indoor Air Quality Model for Airborne Particulate Matter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 378
D.1 Component Subpackages for a Generic Human Exposure Research
Software Package . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 389
xxiii
List of Figures
1.1
1.2
Graphical depiction of the scientific process . . . . . . . . . . . . . . . 8
Graphical depictions of conceptual and mechanistic models for residential exposure to SHS . . . . . . . . . . . . . . . . . . . . . . . . . . 10
2.1
Schematic of a ranch style house in which two-zone SF6 tracer gas
concentrations were measured for different door configurations . . . 44
Time series plots of SF6 tracer gas concentrations measured in a
ranch house for different door configurations . . . . . . . . . . . . . . 46
Interacting factors relating to reduction and/or elimination of residential SHS exposure . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
2.2
2.3
3.1
3.2
3.3
3.4
3.5
3.6
3.7
3.8
3.9
Schematic of a chamber used for cigar and cigarette emissions characterization experiments . . . . . . . . . . . . . . . . . . . . . . . . .
Response surface for optimizing emission and deposition rate parameters of an aerosol dynamics model . . . . . . . . . . . . . . . .
Optimal fit of an aerosol dynamics model to the particle mass concentration time series measured during a Cigarillo chamber experiment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Optimal fit of an aerosol dynamics model to the particle mass concentration time series measured during a regular cigar chamber experiment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Estimated size distribution of particle mass emissions determined
from eight experiments . . . . . . . . . . . . . . . . . . . . . . . . . .
Fit of a lognormal distribution model to the size distribution of particle mass emissions determined for a regular cigar chamber experiment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Particle deposition rate for different air speeds and room furnishing
levels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
SHS particle deposition rate measured in a chamber for different fan
speeds . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
SHS particle deposition rate estimated using an aerosol dynamics
model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. 101
. 108
. 109
. 110
. 113
. 115
. 124
. 125
. 127
LIST OF FIGURES
xxiv
3.10 Evolution of the mass size distribution of SHS particles after a
cigarette was smoked in a chamber . . . . . . . . . . . . . . . . . . . . 132
4.1
4.2
4.3
4.4
4.5
Legend for plots of the hourly fraction of time spent in different residential locations . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Hourly fraction of time spent by NHAPS respondents in different
residential locations, overall and exposed to SHS . . . . . . . . . . .
Hourly fraction of time spent by NHAPS respondents of different
ages in different locations in and around detached houses . . . . .
Hourly fraction of time spent by male and female NHAPS respondents in different locations in and around detached houses . . . . .
Hourly fraction of time spent by NHAPS respondents on weekends
and weekdays in and around detached houses . . . . . . . . . . . .
The rate of SF6 tracer gas mixing measured in a chamber for quiescent and sunlight-driven cases . . . . . . . . . . . . . . . . . . . . .
5.2 The surface-to-volume ratio for bare rooms with wall dimensions
between 1 and 10 m . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.3 Increase in house air flow rate due to the opening of windows . . .
5.4 Schematic of HAC-related flow rates in a detached house with four
main rooms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.5 Schematic of a house where multi-room measurements of CO tracer
gas were made after a cigar was smoked in the kitchen . . . . . . .
5.6 Time series plots of CO concentrations measured in a house after a
cigar was smoked in the kitchen . . . . . . . . . . . . . . . . . . . .
5.7 Schematic of the first floor of a townhouse where two-room SF6
tracer gas measurements were made for different door positions . .
5.8 Time series plots of SF6 concentrations measured in two rooms of a
townhouse for different door positions . . . . . . . . . . . . . . . . .
5.9 House schematics with interzonal air flows corresponding to four
simulated tracer gas scenarios . . . . . . . . . . . . . . . . . . . . . .
5.10 Simulated tracer gas concentrations in a 4-room house for the case
of a 10-min cigarette source in the kitchen . . . . . . . . . . . . . . .
5.11 Simulated tracer gas concentrations in a 4-room house for the case
of a 10-min cigarette source in the bedroom . . . . . . . . . . . . . .
. 150
. 151
. 157
. 160
. 162
5.1
6.1
6.2
6.3
. 169
. 173
. 177
. 180
. 184
. 185
. 187
. 188
. 191
. 194
. 195
Logical flow of a simulation model for predicting residential SHS
exposure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 204
Graph depicting potential interzonal flows for a 3-room house . . . . 208
Graph depicting potential interzonal flows for a 6-room two-level
house. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 209
LIST OF FIGURES
6.4
6.5
6.6
xxv
Simulated flow rates between zones of a house for four illustrative
scenarios. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 210
Time-location plot for 139 NHAPS respondents . . . . . . . . . . . . . 215
Example time series plots for simulated smoker and nonsmoker location, smoking activity, house configuration, room particle concentration, and occupant exposure concentration . . . . . . . . . . . . . . 219
7.1
Time series plots for scripted smoker location, individual cigarette
events, extended smoking episodes, and HAC operation . . . . . .
7.2 Floorplans and interzonal flows for two simulated housing types .
7.3 Scripted smoker and nonsmoker time-location patterns used to simulate residential SHS exposure . . . . . . . . . . . . . . . . . . . . .
7.4 The simulated 24-h average air flow rates between zones of House #1
for six possible scenarios and initially symmetric boundary flow
patterns. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
for six possible scenarios and initially symmetric boundary flow
patterns . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
for six possible scenarios and initially asymmetric boundary flow
patterns. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
7.7 Simulated concentration time series plots for base conditions in
House #1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
7.8 Simulated concentration time series plots for scripted effects of
closed doors and avoidance behavior . . . . . . . . . . . . . . . . . .
7.9 Simulated concentration time series plots for scripted effects of window cross flow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
7.10 Simulated concentration time series plots for the scripted effects of
HAC duty cycle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
7.11 Simulated concentration time series plots for the scripted effect of
initial surface loading of nicotine . . . . . . . . . . . . . . . . . . . .
8.1
8.2
8.3
8.4
Floorplan of a 4-room house used to simulate frequency distributions of unrestricted residential SHS exposure . . . . . . . . . . . . .
Frequency distributions of the number of cigarettes smoked in the
house, the fraction of the day spend by nonsmoker at home, and the
fraction of the day smoker and nonsmoker spend in the same room
for simulations of unrestricted SHS exposure . . . . . . . . . . . . .
Log-probability plot of simulated 24-h average SHS particle exposure concentration for the base unrestricted case . . . . . . . . . . .
Log-probability plot of simulated 24-h average SHS carbon monoxide exposure concentration for the base unrestricted case . . . . . .
. 235
. 239
. 240
. 246
. 247
. 248
. 251
. 253
. 255
. 256
. 257
. 270
. 273
. 276
. 279
LIST OF FIGURES
8.5
8.6
8.7
8.8
8.9
Time series plots of the simulated 24-h average SHS air nicotine concentrations for a 4-room house over 5,000 sequential days . . . . . .
Time series plots of the simulated 24-h average surface nicotine concentrations for a 4-room house over 5,000 sequential days . . . . . .
Log-probability plots of simulated 24-h average SHS nicotine exposure concentration under fresh and loaded wall conditions . . . . .
Log-probability plot of simulated 24-h SHS particle exposure for the
base case and two cases with HAC activity . . . . . . . . . . . . . .
Log-probability plot of simulated 24-h SHS particle exposure for the
base case and a case with asymmetric flow conditions . . . . . . . .
Frequency distributions of the number of cigarettes smoked in the
house, the fraction of time spent by nonsmoker at home, and the
fraction of time smoker and nonsmoker spend in the same room
for simulation experiments involving different SHS exposure mitigation strategies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
9.2 Median 24-h average simulated interzonal flow rates for SHS exposure mitigation strategies . . . . . . . . . . . . . . . . . . . . . . . . .
9.3 Log-probability plot of simulated 24-h average SHS particle exposure for the base case and time ban mitigation strategies . . . . . . .
9.4 Log-probability plot of simulated 24-h average SHS particle exposure for the base case and single-door or single-window mitigation
strategies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
9.5 Log-probability plot of simulated 24-h average SHS particle exposure for the base case and combined single-door and single-window
mitigation strategies . . . . . . . . . . . . . . . . . . . . . . . . . . .
9.6 Log-probability plot of simulated 24-h average SHS particle exposure for the base case and combined two-door and single-window
mitigation strategies . . . . . . . . . . . . . . . . . . . . . . . . . . .
9.7 Log-probability plot of simulated 24-h average SHS particle exposure for the base case and multi-window mitigation strategies for
symmetric flow conditions . . . . . . . . . . . . . . . . . . . . . . . .
9.8 Log-probability plot of simulated 24-h SHS particle exposure for
the base case and multi-window, cross-flow mitigation strategies for
asymmetric flow conditions . . . . . . . . . . . . . . . . . . . . . . .
9.9 Log-probability plot of simulated 24-h average SHS particle exposure for the base case and smoker avoidance or isolation mitigation
strategies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
9.10 Log-probability plot of simulated 24-h average SHS particle exposure for the base case and the case of continuous particle filtration
in each smoking room . . . . . . . . . . . . . . . . . . . . . . . . . .
xxvi
. 282
. 283
. 285
. 288
. 292
9.1
. 303
. 310
. 313
. 315
. 317
. 318
. 320
. 321
. 324
. 325
LIST OF FIGURES
xxvii
10.1 SHS respirable suspended particle concentrations measured in the
living room of a smoking house over 2.25 days . . . . . . . . . . . . . 335
10.2 Time-location profiles for participants of the USEPA particle TEAM
study in Riverside, CA . . . . . . . . . . . . . . . . . . . . . . . . . . . 338
A.1 Legend for raw time-location plots . . . . . . . . . . . . . . . . . . .
A.2 Time series event plot for time spent in different rooms of detached
homes by NHAPS respondents . . . . . . . . . . . . . . . . . . . . .
homes by NHAPS respondents of different genders . . . . . . . . .
homes by NHAPS respondents for weekends and weekdays . . . .
homes by NHAPS respondents of different ages . . . . . . . . . . .
C.1 Screen shot of the graphical user interface windows for an interactive two-compartment computer model . . . . . . . . . . . . . . . .
C.2 Main graphical user interface window for a two-compartment interactive computer model . . . . . . . . . . . . . . . . . . . . . . . . . .
C.3 Parameter graphical user interface dialog box for a two compartment interactive computer model . . . . . . . . . . . . . . . . . . . .
C.4 Data and plot graphical user interface dialog boxes for a two compartment interactive computer model . . . . . . . . . . . . . . . . .
. 365
. 366
. 368
. 369
. 370
. 383
. 385
. 386
. 386
D.1 Graph of data flow between elements of an exposure simulation
model for residential SHS exposure . . . . . . . . . . . . . . . . . . . . 390
1
Part I
Introduction
2
The following two chapters provide a general introduction to the dissertation by first giving an overview of the research and then providing
a background discussion.
Chapter 1 (page 3) provides an overview of the modeling approach,
stating its specific objectives and how they fit into the current state of
knowledge regarding residential exposure to secondhand smoke.
Chapter 2 (page 23) provides a background discussion on exposure concepts, secondhand tobacco smoke health effects, indoor air field studies
of SHS and multi-compartment effects, the exposure of household occupants, emphasizing children, to secondhand smoke, and exposure
modeling.
3
Chapter 1
Research Overview
This dissertation is concerned with the use of simulated experiments to: (1) quantitatively explore the effect that multizonal transport of pollutants and household
occupants can have on residential secondhand tobacco smoke (SHS) exposure; and
(2) increase our understanding of the effectiveness of specific SHS exposure reduction measures. My underlying assumption in this research is that SHS has been
established as a hazard, both in terms of the raw toxic potential of its constituents,
and in terms of concentrations that can occur in typical residences (see Chapter
2). Rather than studying specific health effects or the individual or population risk
associated with exposure to SHS, my focus is entirely on the quantification of exposure across a variety of residential scenarios. Nevertheless, my results are relevant
to a variety of health and risk-related efforts, and especially to inverventions for
reducing children’s exposure (Chapters 2 and 11).
The sciences of indoor air and exposure are becoming well-established. A number of studies have been conducted that are relevant to the current work, including
sophisticated inhalation exposure modeling studies, extensive personal SHS exposure monitoring studies, controlled small-scale studies of the effect of a closed or
open doorway between two rooms, and studies of the concentrations of airborne
pollutants that can occur in different rooms of a multizone household (Chapter 2).
However, there is a dearth of scientific information on precisely how exposure
can vary within the ecology of a typical dwelling, particularly with respect to the
use of windows and interior doors. It is unclear exactly how the movement and
CHAPTER 1. RESEARCH OVERVIEW
4
behavior of human beings in a multi-compartment context affects their exposures.
The studies of previous investigators do not reflect the complexities of a population
of real-life homes. Because occupants move about a home with two or more rooms,
any of which might contain a smoker, and open and close doors and windows or
operate ventilation and filtration systems, exposures cannot be determined based
simply upon concentrations in designated smoking and nonsmoking rooms over
fixed time periods. The accurate estimation of the distribution of exposure for a
population requires an accounting of different types of homes and the variation
of in-home behaviors of smokers and nonsmokers. Such an accounting is a core
feature of the current research.
From a population perspective, SHS exposure occurring in apartments, mobile
homes, or attached homes might be statistically different than that in detached,
single-family homes, because of their generally smaller volumes and differences in
natural ventilation patterns. However, the mixing and flow-related housing characteristics of all residences, which are of key importance in understanding multizone pollutant concentrations and exposures, are expected to be generally similar.
For convenience, and to simplify my analysis of residential SHS exposure, I consider typical floor plans and house volumes for detached, full-size homes, which
represent the most common type of residence in the US.
In the remainder of this chapter, I first give a broad overview of my research,
including each component of this dissertation. I then describe my general modeling approach in more detail, including how it fits into the larger exposure science
context. Finally, I describe the design of specific simulation experiments, which
fall into three distinct tiers of analysis.
1.1 Broad Dissertation Outline
This dissertation aims to address the following broad research question: Given
physical and environmental conditions in a typical detached US residence, how
does the movement of pollutants and persons amongst distinct zones of the house
affect the exposure of household occupants to secondhand smoke (SHS)? A closely
5
related question is also of interest: How much can household mitigation strategies that make use of filtration devices, or modified location patterns or door and
window-related behavior patterns reduce occupants’ SHS exposure?
The approach I use to answer these broad questions employs a model that simulates multizone indoor air pollutant concentrations and exposure profiles for a
household containing a pair of inhabitants, one smoking and the other nonsmoking, explicitly taking into account room-to-room movement profiles of house occupants and the dynamics of interzonal flows. In developing the simulation model
(Part II, Chapters 3−6), I synthesize a wealth of currently available knowledge and
data on pollutant dynamics (Chapter 3), human activity patterns (Chapter 4), and
building characteristics (Chapter 5) into a flexible computer-based framework for
exploring residential SHS exposures for both individuals and populations (Chapter 6 and Appendix D).
My model-based analysis, the results of which comprise the central part of
the dissertation (Part III, Chapters 7, 8, and 9), makes use of trial simulation experiments in which most physical and environmental variables corresponding to
factors such as house size and layout, outdoor air infiltration, interzonal air flow
magnitudes, cigarette emission characteristics, and pollutant dynamics are generally kept fixed at one or two values. The individual effects of these parameters on
indoor air pollutant concentrations are generally well understood.
However, the combined effects of physical and environmental parameters on
residential SHS exposure within a complex household ecology are not understood
as well. Occupant locations, activities, and preferences are expected to introduce
variation in pollutant concentrations and exposures even when the physical dimensions of the system are constrained. Therefore, by selecting a small physical
domain, I can focus on the elucidation of the broad effects of multizonal location
patterns of household occupants and the changing positions of doors and windows.
In addition to the direct effects of human activity on SHS exposure, in Chapters 7 and 8 I also explicitly study the effects of three secondary factors, which are
6
related to weather patterns and household smoking history and are relatively understudied in the context of residential SHS exposure. These are the operation of
a central air handling system, i.e., a heating and air conditioning system (HAC) or
a heating, ventilation, and air conditioning system (HVAC), the existence of symmetrical versus asymmetrical air flow patterns, which are driven by temperature
and/or wind, and high levels of pollutant surface concentration for sorbing and
desorbing compounds, which result from chronic smoking activity.
For Tier I of the analysis (Chapter 7), I execute simulation experiments for three
different occupant patterns, one for the case where the smoker and nonsmoker
pair spend all of their time in the same room, one for the case where they are
never in the same room, and an intermediate case. In Tiers II and III, I examine the
frequency distribution of exposures produced by introducing realistic variability
in occupant location, first for common unregulated household scenarios in which
no conscious efforts are made to affect ones’s exposures (Chapter 8), and second for
deliberate attempts to mitigate exposure by operating filtration devices, regulating
the opening of windows and the closing of interior doors, and/or constraining the
movement of household occupants (Chapter 9).
An underlying concern with any modeling approach to explore residential SHS
exposure is the accuracy of the model predictions. An important followup to the
application of my simulaton model to explore residential SHS exposure is a test of
the model predictions against observed data. Appropriate test data would include
measured concentration profiles in multiple rooms of a house for different smoking patterns, and systematically varied window, door, and HAC configurations, as
well as continuous personal concentrations measured for people spending time in
a house where smoking and location patterns are recorded. While a complete validation exercise is beyond the scope of this dissertation, I use comparisons to other
investigator’s measurements of SHS-related nicotine and particle concentrations
in residential settings to perform a preliminary evaluation of the overall model
performance (Chapter 10).
7
1.2 Modeling Approach
The results of this research contribute to the advancement of the science of exposure primarily through the consolidation of established scientific knowledge in
the form of a simulation model (Chapters 3−6) and the generation of testable hypotheses (predictions) based on established theory (Chapters 7−9). In addition, I
conduct a preliminary evaluation of my theoretical predictions through comparison with experimental observation (Chapter 10). These three elements, theory
development, prediction, and observation, are central to the traditional method of
science as illustrated in Figure 1.1.
The theoretical underpinnings of my research into residential SHS exposure,
formalized as a computerized simulation model, are dependent upon an understanding of two fundamental elements. The first is that of the physical processes
tobacco smoke gases and particles undergo after they are emitted at specific points
and times in a house, and the second is the physical movement and behavior of
smoking and nonsmoking individuals as they travel between rooms of the house.
These two elements, the physical motion of pollutants and the physical motion of
people, interact to result in a large range of possible exposure profiles for occupants of multi-room dwellings.
The interplay of pollutant and human dynamical behavior can be understood
on a simple level in terms of a conceptual model involving the permeation of pollutants throughout a house, the persistence of those pollutants within each room,
and the proximity of the nonsmoking receptor to the active smoker. In this model,
which is illustrated in the top panel of Figure 1.2, source and receptor behavior
patterns result in a defined proximity profile. The permeation of pollutant from
the source to receptor location and persistence of pollutant in the receptor location
lead to a particular exposure profile. A more sophisticated mechanistic model, illustrated in the bottom panel of Figure 1.2, explicitly considers the location profiles
of source and receptor persons, and how their activities contribute to a unique concentration profile in each room. The receptor and concentration profiles are then
combined to determine the exposure profile. This second model forms the basis
8
Theory
Formulate
Revise
Calibrate
Hypothesis
Observation
Figure 1.1: This figure depicts the canonical method of scientific inquiry, which
consists of a repeating cycle of theory development, prediction (hypothesis generation), and observation. The main focus of the current research is to adapt and
simplify established exposure theory (i.e., models) and make predictions for a limited, and narrowly defined set of secondhand smoke exposure scenarios occurring
in residences for the particular purpose of exploring the expected effect of multiple compartments and various mitigation strategies on exposure (see Chapters
8 and 9). Futher simplication of theory is warranted if the effect on exposure for
particular scenarios is minimal, i.e., exposure is relatively insensitive to changes
in mitigation-related human behavior or to movement of people and pollutants
amongst distinct zones. The sensitivity of exposure to various physical and environmental parameter ranges is also explored (see Chapter 8). In addition, the
current work includes preliminary comparisons of predicted and observed exposure concentrations, reevalation of the exposure model, and suggestions for future
model revision (see Chapter 10).
9
for the simulation model used in the current research.
The design of the simulation model, which is presented in detail in Chapter
6, involves first tracking the real-time changes in occupant location and smoking
activity occurring in the simulated house over a 24-h period. Based on these occupant profiles and the time-varying configuration of window, door, HAC/HVAC,
and portable filtration devices, the time series of pollutant concentrations in each
room of the house are calculated using an indoor air quality model, which dynamically accounts for the introduction and removal of pollutant mass in each room.
Finally, exposure is assigned by matching the location of the receptor (nonsmoking
occupant) at a particular minute with the concentrations occurring in each room at
the same time. Once the minute-by-minute exposure time series is established for
each simulated individual, a variety of absolute and relative SHS exposure metrics
may be calculated, including 24-h average, peak, and integrated exposure concentrations, as well as intake fraction (taken over the total mass of all in-house
cigarette emissions) and ETS-cigarette equivalent intake (taken over the mass emissions of a single cigarette).
The physical processes that airborne pollutants are expected to undergo in a
multizonal indoor residential setting are fairly well established and have been successfully characterized using mathematical mass-balance models (see Chapter 2).
Several key assumptions are typically made when modeling indoor air quality and
exposures. Pollutants are assumed to mix very rapidly within single zones of the
house due to air convection before they are removed by ventilation or dispersed
to other rooms. Inter-room transport, which is driven by air exchange across doorways and HAC/HVAC-related recirculation, is assumed to be much slower than
in-room mixing. The assumption of virtually instantaneous mixing of pollutants in
individual rooms ignores any source-proximity effects in which a receptor might
experience extremely high, though short-lived, pollutant concentrations. The three
removal processes that particle and gaseous emissions undergo are surface interactions, including deposition and sorption, filtration, and outdoor transport (ventilation). The last of these is determined by the residence’s leakage characteristics,
10
Conceptual Model
Source Proximity
Pollutant Persistence
Exposure
Pollutant Permeation
Mechanistic Model
Source Location
Source Activity
Room Concentrations
Receptor Location
Receptor Activity
Exposure
Figure 1.2: A simple conceptual model for the exposure of house occupants to secondhand smoke takes into account the proximity of a person to the smoker, the
degree of permeation of smoke throughout the house, and the degree to which
smoke persists in the house, whether in close proximity to the smoker or in further
reaches of the house where smoke has permeated. A more detailed mechanistic
model, which forms the basis for the quantitative occupant exposure model used
in the current work, considers the movement and activities of both source and receptor house occupants, which directly affect pollutant room concentrations. Occupants may open and close doors and windows or activate and deactivate ventilation and filtration systems while in particular rooms. The coincidence of the
receptor and concentrations in particular rooms results in exposure. A more complex mechanistic model, different than that shown or considered explicitly in the
current work, might take into account the interaction between source and receptor
locations and activities, e.g., for spouses, siblings, or parents and their children.
11
Table 1.1: Assumptions for the Physical Behavior of Pollutants
1. Mixing of emitted pollutants occurs so rapidly that concentrations can be
considered to be instantaneously distributed within each zone (room).
2. Pollutants are transported between rooms by direct air flow between
doorways and possibly through air recirculation via the HAC/HVAC
system.
3. There is no source-proximity effect for pollutant concentrations within a
single zone, though there may be a proximity effect across multiple zones,
if a particular zone has stronger air connections to the source room than
other rooms.
4. Pollutants are removed from rooms by ventilation, occurring though
building leakage, open windows, or an HVAC system, or by surface
interactions, such as particle deposition or the sorption of semi-volatile
species.
5. There is no pollutant loss during inter-room transport, except perhaps by
filtration in the HVAC or HAC system. For example, no particles are lost
due to impaction or interception occurring within interior door cracks, i.e.,
a closed door has zero particle removal efficiency for particles travelling
through cracks along the door moulding, jamb, or floor.
door and window positions, and perhaps the operation of an HVAC system. Semivolative compounds, such as nicotine, may also desorb from surfaces. The model
assumptions are summarized in Table 1.1.
To represent the complex mixture of gaseous and particulate species in SHS, I
consider emissions of carbon monoxide (CO), which is an inert gas, nicotine, which
is a highly sorbing semi-volatile gas, and respirable particulate matter (RSP), i.e.,
PM2.5 . Only those emissions that are attributable to indoor-generated tobacco
smoke are considered. Outdoor concentrations of these species are considered to
be zero. The treatment of particulate matter is for total mass concentrations, since
including information on the size distribution of particles will likely not increase
the accuracy with which my research goals can be addressed. Since both PM2.5 and
CO are USEPA criteria air pollutants pollutants, they have associated health-based
National Ambient Air Quality Standards (NAAQS), which correspond approxi-
12
Table 1.2: Emission Factors, Standard Concentrations, and Figures of Merit (FOM)
for CO and PM2.5
SHS Emissions
NAAQSa
Time
FOMb
Pollutant
[mg cig−1 ]
[µ g m−3 ]
Period
[m3 cig−1 ]
CO
40−80
10,000
8-h
4−8
PM2.5
8−12
65
24-h
120−180
a The
USEPA National Ambient Air Quality Standards (NAAQS).
FOM is calculated as the ratio of the SHS emission factor to the NAAQS concentrations and
indicates the amount dilution air per cigarette that is needed for SHS to be in an acceptable range
assuming no contributions from other sources.
b The
mately to upper bounds on acceptable average concentrations over specific time
periods. These standard concentrations are summarized in Table 1.2, along with
their SHS emission factors (see Chapter 3), and a figure of merit (FOM), which indicates the minimum amount of dilution air per cigarette that is needed for SHS to
be in an acceptable range. Based on the results shown, CO is expected to be a less
toxic constituent of SHS than PM2.5 .
Large data sets on the movement of human beings among different rooms
of their home have recently become available with the completion of several
computer-assisted telephone surveys of human activity patterns occurring (see
Chapter 4). As exposure to SHS is dependent upon a number of very brief smoking events, lasting a few minutes each, and removal mechanism having associated
time scales of only a few hours, these activity pattern data, which span a single
day’s worth of activities, are well-suited to be used as a basis for exploring the
dynamics and determinants of SHS exposure.
The usefulness of the current approach lies in consolidating theory to bear on
the issue of multi-compartment effects, stemming from both pollutant and human
movement from room-to-room of a house, and in exercising a simulation model to
study the theoretical effects of varying conditions on exposures. To simplify the
study of effects, ill-understood key parameters (e.g., occupant location, window
and door position, HAC/HVAC operation, and air flow symmetry) are isolated by
fixing at plausible values other parameters for which the effects are better under-
13
sood (e.g., house size, layout, deposition, interzonal air flow rates, air exchange,
sorption, emission rate, and emission duration). Thus, the analysis is conditioned
on a physical and environmental configuration that one might commonly find in
homes so that a systematic range-finding study can be performed across realistic
values of the key variables of interest. The aim is to examine how established indoor air and exposure theory responds to a particular, tightly controlled domain
of model inputs, rather than to attempt to predict exposures or the sensitivity of
exposures for an actual population.
1.3 Design of Simulation Experiments
The primary goals of the study are to discover, through controlled simulation experiments, how the movement of pollutants and persons amongst distinct zones
of a house can be expected to affect exposure to secondhand smoke (SHS), and
how much effect might result from conscious household mitigation strategies that
make use of door and window-related behavior patterns to reduce exposure. Results from pursuing these goals are presented for three three tiers of analysis in
Chapters 7, 8, and 9 of this dissertation. In the interest of clarity, and to retain focus
on a small number of key variables, all three tiers use the same fixed set of physical
and environmental input parameter values, which are selected to represent typical
conditions in a US household.
The first tier of analysis sets out to demonstrate broadly how the multizone
structure of a house can contribute to wide variation in SHS exposure. The second
tier incorporates wide variation in location patterns to investigate frequency distributions of SHS exposure for a selected cohort. The final tier studies changes in
the exposure frequency distribution of a cohort as a result of attempts to mitigate
exposure. For simulations of cohort exposure, I use a stopping rule to judge when
enough people have been sampled and the distribution has stabilized. The metric is the ratio of the half-length of the 90% confidence interval to the distribution
mean. When this metric is stable to within 10%, I conclude that a sufficient sample has been drawn to represent realistic variation in human movement patterns.
14
Typically, between 500 and 1,000 sampled individuals are sufficient to achieve this
criterion.
1.3.1 Tier I. Scripted Occupant Movement
The results of the first tier of simulation analysis are presented in Chapter 7 of
this dissertation. As part of this analysis, I investigate the degree to which the
general multi-compartment character of a residence can influence the exposure
of its occupants to particles and nicotine in SHS, by devising a series of controlled
simulation experiments, which serve to flex the model between extremes in inputs.
Since proximity between nonsmoking and smoking occupants is expected to be of
central importance, I use three scripted receptor movement patterns to cover the
gamut of proximity, ranging from lock-step coincidence to complete avoidance
of the smoker. These scripted patterns are analyzed in combination with factor
levels for different scenarios, house layout, and symmetry of air flow across house
boundaries. The two different simulated houses are identical in size. One consists
of a single large multi-use room and the other is a house with four main rooms
and a connecting hallway. Both houses have HAC systems, which do not include
a forced ventilation component, but whose operation may increase infiltration due
to duct leakage.
The scenarios consist of three different door and window practices for smoking
and nonsmoking inhabitants. These operations are used to introduce impediments
to the permeation of pollutant from the smoking room or to promote enhanced
ventilation with outdoor air. Either continuous or intermittent HAC operation
may also occur, which in addition to increasing infiltration rate by a small amount
also increases the transport of pollutants among rooms of the house. As part of
the Tier I analysis, I take a first look at the effect of surface nicotine residue on
exposures and estimate the error made under the common simplifying assumption
that houses can be represented by a single well-mixed zone.
The fixed physical and environmental model input parameters used in each
simulation experiment, which include values for cigarette source strength, smok-
15
ing duration, house volumes and surfaces areas, house air exchange rate, door and
window air flow rates, and particle deposition rate, are selected based on common
values reported in the scientific literature. The range of possible values is discussed
in Chapter 3 of this dissertation, where I present the results of various studies on
the magnitude and properties of cigarette emissions, and in Chapter 5, where I
present data on the size and air flow characteristics of residences.
1.3.2 Tier II. Realistic Variation in Occupant Movement
The second tier of analysis, presented in Chapter 8, builds on that of Tier I, using
the same set of fixed physical and environmental input parameter values. However, for this analysis I introduce realistic frequency distributions of occupant location patterns, using a cohort sampled from the results of a recent telephone-based
survey, to estimate a base frequency distribution of exposure to SHS particles, nicotine, and carbon monoxide. To understand the features of this base distribution,
and its sensitivity to different conditions, I perturb the base distribution by conducting simulation trials with the sampled cohort across a variety of “natural”,
i.e., unregulated or unrestricted, scenarios. These scenarios, which do not involve
conscious attempts to regulate (mitigate) exposure and may be expected to occur
in a typical residence, include the continuous and intermittent operation of a residential HAC system, symmetric and asymmetric flows across house boundaries,
and the case when household surfaces are initially suffused with nicotine, such as
might occur after years of chronic cigarette smoking.
In addition to an exploration of the impact that various typical non-intervening
scenarios might have on the base distribution of exposure for a cohort of persons
with a wide range of time-location patterns, I also explore the local sensitivity of
frequency distributions of SHS exposure to small perturbations in values of the
physical and environmental parameters. Although the original selected values
place simulated exposures in the middle of those that might be expected for a
typical household in the US, and interpretation of the effects of scenarios might
be obfuscated by including variation in these input parameters in the simulation
16
experiments, it is important to understand generally how much the distribution of
exposures might change if different fixed values are chosen.
1.3.3 Tier III. Exposure Mitigation Trials
The third tier of analysis, which is presented in Chapter 9, uses a portion of the
sampled cohort of persons used for Tier II and the same set of fixed physical and
environmental parameter values. Since I am interested in the more intense exposures to SHS, which would result in the most benefit from successful mitigation
strategies, the mitigation simulation trials are limited to that segment of the cohort for which more than 10 cigarettes are smoked at home during the day and for
which the nonsmoker spends more than
2
3
of their time at home. For this cohort,
I perturb the base distribution of exposures with different exposure mitigation
strategies, i.e., deliberate attempts to regulate the behavior of household members
for the purpose of reducing occupant exposure to SHS. I evaluate the practicality and effectiveness of modifying occupant location patterns, specific door and
window-related behaviors, and the continuous operation of portable filtration devices in rooms of the home where smoking is allowed.
I explore changes in the frequency distribution of SHS exposure relative to two
bounding cases. In the first, base case, no migitation strategies are active. In the
second case, the smoker is not allowed to smoke in the house when others are also
at home. For intermediate scenarios, the nonsmoker occupant, the smoker occupant, or both occupants are directed to change their location in the house and/or
to close doors or open windows in rooms they occupy in response to smoking
episodes. Smoking episodes are defined by contiguous periods of time during
which a smoker is in a single room of the house and smokes for a portion of the
time.
For most simulation trials, the occupants do not change the time they spend in
rooms of the house, following their “natural” pattern. However, for several trials,
the occupant location patterns are modified so that nonsmoking occupant avoid
rooms where the smoker is active and the smoker is isolated in a single desig-
17
nated smoking room. Door- and window-related behavior is superimposed onto
the original or modified time-location patterns. When the smoker and nonsmoker
happen to occupy the same room during a smoking episode, the doors are left
open. If any window-related mitigation strategies are in effect, the windows are
also left open for time the ocupants spend in the same room during a smoking
period.
The rationale for different types of mitigation strategies is as follows: Avoiding
or isolating a smoker diminishes exposures by removing the direct proximity of the
nonsmoker to SHS emissions. The closing of interior doorways in the house during
smoking epidodes impedes the flow of air, and therefore SHS pollutants, from
the smoking room to adjacent rooms. The opening of a window during smoking
episodes enhances the removal of SHS pollutants through ventilation, exhausting
polluted air to the outdoors before it can move to other rooms. The opening of two
windows, one by each of the occupants, can induce a cross-breeze in the house and
lead to enhanced removal of pollutant from the house, although it is also possible
that pollutants will be carried more quickly in the direction of an occupant. The
mitigation strategy with the most benefit relative to the base condition is expected
to occur when the nonsmoker avoids the smoker, especially if the smoker closes
the door to their room during smoking and/or simultaneously opens a window
to increase ventilation. Combinations involving just the closure of a door or just
the opening of window by a smoker are expected to be of the next greatest benefit,
whereas similar action by the nonsmoker while they are in a different room than
the smoker is expected to be of less consequence.
1.3.4 Summary of Specific Analysis Factors
For the Tier I analysis, I examine SHS exposures across a total of 144 combinations
of eight different scenario, pollutant, and activity-related factors. These factors and
their corresonding levels are:
House Type (2 levels). Either a house with two rooms, one single large living area
and a bathroom, or a house of equal volume that has four main rooms.
18
Flow Pattern (2 levels). The air flows across house boundaries are either symmetric, such as might occur for temperature-driven flows, or asymmetric, such
as might occur for wind-driven flows.
Nonsmoker Activity (3 levels). Location patterns for nonsmokers who “follow”
smokers from room-to-room, who “nap” during the day away from the
smoker, or who “avoid” the smoker entirely.
Door-Related Behavior (4 levels). Either the smoker, the nonsmoker, or both occupants close doors of rooms they occupy during smoking episodes, or the
house remains in the base state where doors are always left open except during sleeping hours or time spent in the bathroom.
Window-Related Behavior (4 levels). Either the smoker, the nonsmoker, or both
occupants open windows of rooms they occupy during smoking episodes, or
the house remains in the base state where windows are always closed.
HAC Activity (3 levels). Either the HAC system is active intermittently for 10% of
the time when at least one occupant is awake, or it is active continuously for
100% of the time an occupant is awake, or it remains inactive for the entire
day (the base state).
Pollutant Type (2 levels). Either particles or nicotine.
Initial Surface Concentration (2 levels). Either surfaces are initially devoid of
nicotine concentrations or they are loaded at levels that might occur from
chronic smoking activity.
The base condition for each particular house, pollutant, and nonsmoker activity
corresponds to factor levels for which flow is symmetric, doors are open during
awake and non-bathroom times, windows are closed, and the HAC is off. The
fundamental set of six combinations across the three scenario factors for window,
door, and HAC configuration, are presented in Table 1.3. These combinations are
19
crossed with complete combinations of analyses for house, flow patterns, and nonsmoker activity factors for simulations involving particle emissions. For nicotine
emissions, they are crossed with surface concentration, flow pattern, and nonsmoker activity. The nicotine analyses are performed for the 4-room house only.
A summary of all the combination of factors considered in the second and third
tiers of analyses is given in Table 1.4. A total of six different analyses are performed
for Tier II, and 25 different analyses are performed for Tier III. The factors shown in
the table are the same as those described above for Tier I, except for the following.
Only the 4-room house is considered, there is an additional pollutant factor level
for carbon monoxide, nonsmokers may avoid rooms with active smoking, smokers may be forced to only smoke in the living room, smokers may be forced to
curb their smoking when other residents are at home, and there may be a portable
filtration device continuously active in smoking rooms. Also, activity patterns for
smoker and nonsmoker pairs are randomly sampled over a broad range of values
based on observed population data rather than taking on a small number of fixed
values. As with Tier I, the base condition requires that doors are open during waking hours, windows are always closed, and the HAC is always off. In addition, for
the Tier II and III base conditions, smokers and nonsmoker follow their “natural”
location patterns, smokers are allowed to smoke at any time of day in any room
of the house, except the hallway and bathroom, and portable particle filtration is
never used. Other factor levels represent perturbations of the base condition.
20
Table 1.3: Summary of Different Combinations of Scenario Factors Considered in
the Analysis of Residential SHS Exposure Using Scripted Occupant Activity Patterns (Tier I, Chapter 7)
Analysis
Door
Window
HAC
No.
Closinga
Openingb
Activityc
1
−
−
−
2
Smoker
−
−
3
Smoker
Smoker
−
4
−
−
10%
5
−
−
100%
6
−
Both
−
For simulations of SHS particle exposures, the factor combinations in these table were further
crossed with factor levels corresponding to “follower”, “avoider”, and “napper” nonsmoker behavior patterns, two types of houses, one dominated by a large, well-mixed space (House #1) and
one with four distinct main rooms (House #2), and both symmetric and asymmetric flow patterns.
For SHS nicotine exposures, the same factors levels were used, except for the omission of the first
type of house. The base case for analysis occurs under symmetric flow conditions when all interior
doors are open during waking hours not spent in the bathroom, all windows are closed at all times,
and the HAC system is inactive for all times. Adherence to this base condition for individual factors is indicated by dashes (−) in the door, window, and HAC columns.
a Either doors remain open during waking hours not spent in a bathroom (−), or the smoker
(Smoker) closes the door of rooms they occupy during smoking episodes.
b Either windows are always closed (−), or the smoker (Smoker) or both the nonsmoker and smoker
(Both) open(s) the window of rooms they occupy during smoking episodes.
c The HAC system is either always off (−), or it is active during either 10% or 100% of waking hours.
[Continued.]
Tier III
Analysis
Tier
Tier II
Analysis
No.
1−3
4
5
6
7−8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
c Door
Closing
−
−
−
−
−
S
NS
−
−
S
S
NS
NS
Both
Both
Both
−
S
NS
Both
b Flow
Pattern
Sym
Asym
Sym
Sym
Sym
Sym
Sym
Sym
Sym
Sym
Sym
Sym
Sym
Sym
Sym
Sym
Sym
Sym
Sym
Sym
Opening
−
−
−
−
−
−
−
S
NS
S
NS
S
NS
−
S
NS
Both
Both
Both
Both
d Window
Activity
−
−
10%
100%
−
−
−
−
−
−
−
−
−
−
−
−
−
−
−
−
e HAC
Modification
−
−
−
−
−
−
−
−
−
−
−
−
−
−
−
−
−
−
−
−
f Location
Activity
−
−
−
−
−
−
−
−
−
−
−
−
−
−
−
−
−
−
−
−
g Filtration
Table 1.4: Summary of 31 Different Combinations of Scenario and Pollutant Factors Considered in the Analysis of
Frequency Distributions for Unrestricted (Tier II, Chapter 8) and Restricted (Tier III, Chapter 9) Residential SHS
Exposurea
21
c Door
Closing
−
S
NS
Both
−
S
S
−
b Flow
Pattern
Asym
Asym
Asym
Asym
Sym
Sym
Sym
Sym
Opening
Both
Both
Both
Both
−
−
S
−
d Window
Activity
−
−
−
−
−
−
−
−
e HVAC
Modification
−
−
−
−
Avoid
Iso
Iso
−
f Location
Activity
−
−
−
−
−
−
−
S
g Filtration
base case for analysis occurs under symmetric flow conditions when occupants follow their “natural”, unmodified location patterns,
the smoker can smoke at home at all times not spent in the hallway or bathroom, all interior doors are open during waking hours not
spent in the bathroom, all windows are closed at all times, the HAC system is inactive for all times, and no portable filtration devices are
ever active. Adherence to this base condition for individual factors is indicated by dashes (−) in each column. For analysis #8 the smoker
cannot smoke in the house when others are home, but otherwise the analysis is for base conditions. Analyses #1−3 are for particulate
matter, carbon monoxide, and nicotine in SHS, respectively. All other listed analyses were performed for particles only.
b Possible scenarios for flow patterns are symmetric (Sym), for which flows across all building boundaries are initially balanced in incoming
and outgoing directions, or asymmetric (Asym), for which flows across building boundaries are initially allowed to have directionality.
c Either doors remain open during waking hours not spent in a bathroom (−), or the smoker (S), the nonsmoker (NS), or both (Both)
close(s) the door of rooms they occupy during smoking episodes.
d Either windows are always closed (−), or the smoker (S), the nonsmoker (NS), or both (Both) open(s) the window of rooms they occupy
during smoking episodes.
e The HAC system is either always off (−), or it is active during either 10% or 100% of waking hours.
f Location modification involves avoidance of the smoker by the nonsmoker during smoking episodes in unmodified locations (Avoid) or
isolating the active smoker in the living room by themselves (Iso).
g Portable filtration devices are either never active (−), or they were active continuously in rooms where smoking is allowed to occur (S).
a The
Table 1.4. Continued.
Analysis
Analysis
Tier
No.
Tier III
24
(Cont.)
25
26
27
28
39
30
31
22
23
Chapter 2
Background
Secondhand tobacco smoke, or secondhand smoke (SHS) for short, is the smoke
produced when someone other than oneself produces chemical emissions in the
process of smoking a cigarette, cigar, or pipe. Several alternate names for this type
of pollutant have emerged, including environmental tobacco smoke (ETS), passive smoke, and involuntary smoke [Chapman, 2003]. In scientific circles, SHS
is defined as the stream of particles and gas that has emanated from the burning
tip of a cigarette or other tobacco product (sidestream smoke), or has been drawn
through the product and exhaled by the smoker (mainstream smoke), and then has
undergone varying amounts of dispersion, dilution, and transformation in the surrounding environment. Environments typically include residences, workplaces,
automobiles, or the outdoors.
SHS consists of thousands of organic and inorganic chemical species in both
gaseous and particle phases [Jenkins et al., 2000]. Tables 2.1 and 2.2 list a number
of recent laboratory studies on the composition of SHS along with a summary of
their methods and findings. Daisey et al. [1998], Singer et al. [2003], and others
present per-cigarette emission factors for a variety of volatile organic compounds
in SHS. Of the compounds they analyzed, nicotine, acetaldehyde, formaldehyde,
toluene, ammonia, acetonitrile, and isoprene were among those with the largest
SHS emissions. In terms of inorganic gas SHS emissions, the carbon oxides (carbon
dioxide and carbon monoxide) contribute heavily [Löfroth et al., 1989; Martin et al.,
1997; Jenkins et al., 2000] as do nitrogen oxides [Löfroth et al., 1989].
CHAPTER 2. BACKGROUND
24
Because it has been associated with a variety of health problems at typical residential levels, SHS is of much interest as an environmental contaminant. The
adverse consequences of smoking inside homes have been established and quantified. It is the responsibility of those in health-related fields to educate the public,
providing incentives and strategies for reducing or eliminating SHS exposure, especially for children. To properly inform public health professionals, it is important to have a standardized formulation of exposure and to have reliable quantitative information on the magnitude of SHS exposures and the factors having the
largest influence on exposure.
In the first two sections of this chapter, I present a general quantitative definition of inhalation exposure, evaluate various practical SHS exposure measures,
and summarize the general health effects that have been associated with SHS at
levels typically found in homes. Next, I summarize work by exposure and indoor air quality scientists, which bears generally or specifically on SHS exposure.
These experimental studies, conducted both in the field and in controlled settings,
provide baseline residential SHS particle exposures and establish the importance
of multiple compartments in driving variation in exposures. I then take an indepth look at children’s residential exposure to SHS, including the prevalence
and effectiveness of household restrictions on smoking, and intervention strategies aimed at changing household smoking behavior. Finally, I present a series
of published multi-compartment indoor air quality modeling efforts, examining
the success with which models have been applied to residential pollutant concentrations, and review a sample of inhalation exposure simulation models currently
under development.
2.1 Defining and Measuring Exposure
2.1.1 Concept and Mathematical Formulation of Exposure
Some of the earliest formal codification of exposure concepts appearing in the
scientific literature is due to Ott [1982, 1985, 1990]. Focusing on exposure to air
SS
SS
MS+SS
SS
SS
SS
Löfroth et al. [1989]
Eatough et al. [1989]
Martin et al. [1997]
Daisey et al. [1998]
Singer et al. [2002]
Singer et al. [2003]
C, M, TD,
GC-MS
C, M, TD,
GC-MS
C, M, TD,
GC-MS
C, H, TD,
GC, IR
C, M, DEN,
IC, GC-MS
C, M, F, IR,
GC
Methodb
Of 29 gas-phase organic compounds, those with steady-cycle exposure-relevant
emission factors close to 1,000 µ g cig−1 or greater were acetaldehyde, formaldehyde,
isoprene, toluene, acetonitrile, and nicotine, and those in the range 300−1000 µ g
cig−1 were acrolein, 1,3-butadiene, 2-butanone, benzene, pyridine, pyrrole,
3,4-picoline, and 3-ethenyl-pyridine.
Of 26 gas-phase organic compounds measured for different ventilation rates in a
stainless steel or furnished chamber, only acetonitrile, toluene, isoprene, and nicotine
had exposure-relevant emission factors that were consistently above 800 µ g cig−1 .
Isoprene consistently had the largest emission factor, with a range of 2000−5000 µ g
cig−1 , and nicotine emissions ranged from about 400 µ g cig−1 in the fully furnished
chamber to about 3,700 µ g cig−1 in the bare stainless steel chamber.
Of 22 measured gas-phase organic compounds, those having emission factors near or
exceeding 1,000 µ g cig−1 were acetaldehyde, formaldhyde, and nicotine, and those in
the range of 300−1000 µ g cig−1 were benzene, pyridine, pyrrole, toluene, and
3-ethenyl-pyridine.
Formaldehyde, acetaldehyde, acetone, ammonia, nicotine, isoprene, and acetonitrile
had average yields over 1,000 µ g cig−1 . 3-Ethenyl-pyridine, toluene, and
1,3-butadiene had average yields exceeding 300 µ g cig−1 . The yield for carbon
monoxide was 58 mg cig−1 and yields for NO and NOx exceeded 1,000 µ g cig−1 .
HNO2 was found to be the major gas-phase inorganic acid and NH3 , nicotine,
pyridine, 3-ethenyl-pyridine, and myosmine were the principal gas-phase bases.
Sizeable yields were measured for carbon monoxide (67 mg cig−1 ) and yields
exceeded 1,000 µ g cig−1 for nitrogen oxides, nicotine, formaldehyde, acetaldehyde,
propene, ethane, ethene, and isoprene. Yields for acrolein, benzene, 1,3-butadiene,
and propane exceeded 300 µ g cig−1 .
Chemical Composition
a SS=sidestream smoke; MS=mainstream smoke. b C=chamber experiments; M=machine smoked; H=human smoked; F=filter sampling; FID=flame ionization detector;
DEN=denuder; TD=thermal desorption; IC=ion chromatography; GC-MS=gas chromatography/mass spectrometry; IR=infrared absorption
Sourcea
Study
Table 2.1: Reported Emission Factors for Gas-Phase Components of SHS
25
MS+SS
Rogge
et al.
[1994]
H, C,
MOUDI, F,
FID, IC
Kleeman
et al.
[1999]
Organics: SHS particles are predominantly organic compounds in every particle size range.
Inorganics: elemental carbon and the following trace elements and other species were detected:
−2
+
Na, K, V, Mn, Br, Sb, La , Ce, Cl− , NO−
3 , SO4 , NH4 ; the size distribution of these species, as for
the total particle size distribution, had a single mode between 0.3 and 0.4 µ m.
Inorganics: major elements associated with smoking were K, Cl, and Ca.
Organics: the following classes of species were detected (with compounds having emission rates
greater than 100 µ g per cigarette listed in parentheses): n-alkanes (hentriacontane,
tritriacontane); iso and anteisoalkanes; isoprenoid alkanes; n-alkanoic acids (hexadecanoic acid);
n-alkenoic acids; dicarboxylic acids; other aliphatic and cyclic acids; n-alkanols; phenols
(1,4-benzenediol); phytosterols (stigmasterol, β-sitosterol); N-containing compounds (nicotine,
3-hydroxypyridine, myosmine); polycyclic aromatic hydrocarbons.
Inorganics: in 77 homes with smoking, smoking contributed the following mass percentages: S
(11%); Cl (72%); K (70%); V (16%); Zn (14%); Br (44%); Cd (75%); estimated emission rates (µ g per
cigarette): S (65); Cl (69); K (160); V (0.37); Zn (1.2); Br (3.0); Cd (0.32).
Organics: 59.5% organic carbon by mass. Inorganics: species present above 0.01% by mass: S
+
−2
0.14%; Cl 0.23%; K 0.41%; elemental carbon 0.49%; Cl− 0.28%; NO−
3 0.071%; SO4 0.059%; NH4
0.04%.
Organics: main classes: n-alkanes, branched alkanes, bases, sterols, fatty acids, sterenes; µ mol/g
(std. dev.): nicotine 467 (144); myosmine 35 (21); nicotyrine 14 (11); cotinine 20 (11); cholesterol
1.41 (0.33); stigmasterol 2.9 (1.6); campersterol 1.53 (0.58); β-sitosterol 2.2 (1.8);
24-methylcholesta-3,5-diene 2.1 (2.0); 24-ethylcholesta-3,5,22-triene 1.60 (0.81); solanesol 22.2
−
−2
+
(3.3). Inorganics: species detected: Cl− , NO−
2 , NO3 , SO4 , NH4 ; species detected above 50
+
−
µ mol/g: Cl− , NO3 , NH4 ; elements detected: K, Ca, Ti, Ba, V, Cr, Mn, Fe, Co, Ni, Cu, Zn, Pb, As,
Se, Br; elements detected above 50 µ mol/g: Ca, K.
Chemical Composition
a The listed studies all involved cigarettes (i.e., no cigars). b C = chamber; M = machine-smoked; F = filter-based sampling; D = Denuder; FID = flame ionization detector,
NPD = nitrogen-phosphorus detector; GC-MS = gas chromatograph – mass spectrometer; H = human smoker; HD = emissions captured in a hood; FID = flame-ionization
detector; MOUDI = micro-orifice uniform-deposit impactor; Fd = field sampling; XRF = X-ray fluorescence; IC = ion chromatography. c This effort was part of a New York
State study carried out in Onondaga and Suffolk counties [Sheldon et al., 1989]. d This effort was part of the USEPA’s Particle Total Exposure Assessment Methodology
(PTEAM) study of 178 homes in Riverside, CA.
MS+SS
H, Fd, XRF
Özkaynak MS+SS
et al.
[1996]d
H, HD, F,
GC-MS
H, Fd, XRF
MS+SS
Koutrakis
et al.
[1992]c
C, M, F, D,
GC-FID,
NPD,
GC-MS, IC
H, HD, F,
XRF,
GC-MS
SS
Benner
et al.
[1989]
Methodb
Hildemann MS+SS
et al.
[1991]
Source
Study
Table 2.2: Reported Particle-Phase Components of SHSa
26
27
pollution, Ott [1982] provides the following definition of exposure: “exposure of
person i to pollutant concentration c is viewed as two events occurring jointly:
person i is present at a particular location, and concentration c is present at the
same location.” More recently, Zartarian et al. [1997] presented a unified theoretical framework where exposure is defined generally as contact between an agent
and a target. Their definition applies equally well to inhalation, dermal, and ingestion types of exposure. Zartarian et al. define an “instantaneous point exposure”
as the joint occurrence of point i of a target being positioned at location (xi ,yi ,zi )
at time t, and an agent of concentration Ci present at the same location (xi ,yi ,zi )
at time t. To distinguish it from exposure, dose is generally defined by Zartarian
et al. as the amount of agent that enters a target after crossing a contact boundary.
Exposure may occur without dose, but not vice versa.
The basic mathematical formulation of air exposure was established in original
contributions by Fugas [1975], Duan [1982], and Ott [1982, 1984], and came to be
called the indirect exposure assessment approach in contrast to direct approaches, in
which exposure is measured using personal monitoring devices (see Section 2.3 on
page 36). They introduced the basic concept of calculating exposure as the sum of
the product of time spent by a person in different locations and the time-averaged
concentrations in those locations. In this formulation, locations are termed microenvironments and assumed to have homogeneous pollutant concentrations. This
formulation is written mathematically as follows:
m
Ei =
∑ Ti j Ci j
(2.1)
j=1
where Ei is the integrated exposure for person i, Ti j is the time spent in microenvironment j by person i, Ci j is the time-average concentration person i experiences in
microenvironment j, and m is the number of different microenvironments. The calculation amounts to a weighted sum of concentrations. Each discrete time segment
with its associated discrete concentration need not be sequential in time, i.e. there
may be discontinuities in time and space, although Equation 2.1 is usually applied
to contiguous time segments adding up to some convenient duration, such as a
28
single day. An overall average exposure concentration is calculated by dividing Ei
by the total time spent in all microenvironments.
The basis for the temporally and spatially discrete Equation 2.1, in which Ci j are
supplied as mean concentrations or concentrations that are constant during each
corresponding time period Ti j , can be considered to arise from a fully continuous
formulation:
Z t2
Ei =
t1
(2.2)
Ci (t, x, y, z) dt
where Ci (t, x, y, z) is the concentration occurring at a particular point occupied
by receptor i at time t and spatial coordinate [ x, y, z]. If discrete microenvironments are considered rather than fully continuous space, then the following semicontinuous formulation applies:
m
Ei =
∑
Z
j=1
t j+1
tj
Ci j (t)dt
(2.3)
where Ci j (t) is the concentration experienced by the receptor in the discrete microenvironment j at a particular point in time t. In Equation 2.2 the exposure
trajectory of the receptor is followed explicitly with no discontinuities, whereas
in Equation 2.3 there are no time discontinuities within any given microenvironment, but microenvironments need not correspond to contiguous time periods. If
we apply Equation 2.3 to a series of three subsequent microenvironments, the sum
of integrals can be written as a sum of {mean-concentration×elapsed-time} products, i.e., the form of Equation 2.1, as follows:
3
Ei =
=
∑
Z
j=1
Z t2
t1
t j+1
tj
Ci j (t)dt
Ci1 (t)dt +
Z t3
t2
Ci2 (t)dt +
Z t4
t3
Ci3 (t)dt
= Ci1 Ti1 + Ci2 Ti2 + Ci3 Ti3
3
=
∑
j=1
3
Ci j Ti j =
∑ Ci j Ti j
j=1
where Ci j = Ci j is the average concentration in each microenvironment.
(2.4)
29
A simplified, population version of Equation 2.1 can be derived in terms of the
total time spent by all receptors in each microenvironment, if the same microenvironment concentrations are used for every person:
n
Ẽ =
m
m
n
m
j=1
i =1
j=1
∑ ∑ C j Ti j = ∑ C j ∑ Ti j = ∑ C j T̃j
i =1 j=1
(2.5)
where n is the number of people in the population, C j is the average exposure
concentration in microenvironment j assigned to every person i, Ẽ is the sum of
the exposure over all members of the population, and T̃ j is the total time spent by
all persons in microenvironment j.
2.1.2 Practical Measures of SHS Exposure
Studies of health effects due to residential SHS, or the effectiveness of household
smoking restrictions or interventions, depend on accurate measures of SHS exposure. These studies typically make use of one or more of four general exposure
measures [Hovell et al., 2000c]: questionnaire-based self-report of exposure; environmental nicotine concentrations; environmental measurements of respirable
suspended particles (RSP) concentrations; and biological levels of the nicotine
metabolite cotinine.
Eight specific SHS exposure measures are summarized in Table 2.3 along with
their classification as a direct or indirect measure, or a dose measure, which, like
indirect exposure measures, may indicate that exposure has occurred but not have
a clear quantitative relationship to true exposure. Indirect questionnaire-based
measures may consist of information on the existence of smoking bans in a home,
the subjective time spent exposed to nearby smoking, or the number of cigarettes
smoked in different locations or time periods. Fixed-site concentration measurements of airborne SHS consituents, such as RSP or nicotine, are considered indirect exposure meaures, whereas the most direct measure of exposure is provided
by real-time or integrated personal concentrations [Jaakkola and Jaakkola, 1997].
While direct measures of personal nicotine or particles intrinsically contain information on variation in the proximity of a subject to smokers and variation in
30
environmental conditions, fixed-site monitoring of residential SHS concentrations
requires supporting information to provide a reasonably accurate estimate of exposure.
Although convenient integrated and real-time personal particle monitors are
available [Brauer et al., 1999], RSP is not specific to tobacco smoke and the use
of these monitors, and the associated data analysis, can be awkward, unfamiliar
or complicated. Perhaps for these reasons, measurement of airborne nicotine and
urinary or serum cotinine in body fluids are the most widely used measures of
SHS exposure in public health studies. Nicotine is a major constituent of SHS and
found uniquely in tobacco emissions. It can be efficiently collected and analyzed in
both particulate and vapor phases using a two-filter assembly where one filter has
been treated with sodium bisulfate [Hammond et al., 1987]. Cotinine is the major
metabolite of nicotine and occurs in readily measurable quantities in the plasma,
urine, or saliva of persons who have been exposed to SHS [Benowitz, 1996, 1999].
The use of nicotine and cotinine are complicated by the peculiar indoor dynamics of nicotine relative to many other SHS species. Nicotine sorbs rapidly to indoor
surfaces, and may desorb from surfaces at a later time, contributing to SHS-related
exposures that occur many hours after smoking has stopped. Similarly, accumulated nicotine levels in house dust may interfere with SHS (air) levels [Hovell et al.,
2000c]. Children’s exposure to house dust in smoking homes, or their ingestion or
close contact with nicotine-contaminated objects due to earlier smoking, may lead
to measureable levels of cotinine in body fluids. Another serious concern in using cotinine to estimate SHS exposure is that the metabolism of cotinine varies for
different members of an exposed population.
2.2 Health Risks of SHS
In recent years, several governmental and international health or environmental
groups have undertaken reviews of the evidence for a causal link between SHS exposure and adverse health outcomes [USEPA, 1992; OEHHA, 1997; UKDH, 1998;
WHO, 2000]. In this section, I review the specific findings of these and other re-
Indirect Air Sample
Direct
Indirect Air Sample
Direct
Dose
4
5
6
7
8
Personal real-time concentration of SHS marker (e.g., PM2.5 )
Biological Sample One-time or repeated concentration of SHS biomarker (e.g.,
cotinine)
Air Sample
Fixed-location real-time concentration of SHS marker (e.g.,
PM2.5 )
Personal integrated concentration of SHS marker (e.g., nicotine
or PM2.5 )
Fixed-location integrated concentration of SHS marker (e.g.,
nicotine or PM2.5 )
Duration of exposure episodes
Number of cigarettes smoked at home with possible time,
location, or receptor-presence specificity
Binary exposure classification: smoking allowed or disallowed
at home with possible time, location, or receptor-presence
specificity
Description
ng L−1
µ g m− 3
µ g m− 3
µ g m− 3 h
µ g m− 3 h
h
−
−
Units
Only the “Direct” type of exposure measures can be considered an accurate quantitative assessment of exposure according to the definition given in Section 2.1.1. The other measures are qualitative descriptions of potential exposure (#1), estimates of exposure co-variates
(#’s 2 and 3), fixed-site concentration measurements, which may not reflect actual exposures experienced by household occupants, or measurements of dose, which indicates the occurrence of exposure but is not likely to have a clear quantitative relationship to exposure (#8).
a
Indirect Questionnaire
3
Air Sample
2
Instrument
Typea
1
No.
Table 2.3: Different Measures of Residential SHS Exposure
31
32
ports. Each study looked at current epidemiological and toxicological research
and reached very similar conclusions about how SHS impacts health under typical
exposure conditions, specifically that SHS poses a substantial health risk. Based
largely on US government reports [USEPA, 1992; OEHHA, 1997], the US Department of Health and Human Services classifies SHS as known to be a human carcinogen in their 10th Report on Carcinogens [USDHHS, 2002].
USEPA [1992] established a link between exposure to SHS by nonsmokers and
lung cancer and, by weight-of-evidence analysis, found that SHS belongs in a category of compounds that the USEPA classifies as Group A (known human) carcinogens. SHS was estimated to result in 3,000 lung cancer deaths among nonsmokers
per year in the US. SHS was also found to be causally related to symptoms of respiratory irritation, middle ear disease, lower respiratory infections (up to 300,000 annual cases in children), and increased severity and frequency of episodes of asthma
(up to 1,000,000 children affected). For this report, the hazard identification of SHS,
the weight of evidence analysis for non-cancer effects, and estimates of the public
health impact of SHS were based on epidemological studies. A surrogate for exposure to SHS for non-cancer effects was usually parental or spousal smoking status.
The analysis for the weight-of-evidence conclusion for lung cancer was based on
animal bioassays and genotoxicity studies, biological measurements of human uptake of tobacco smoke components, and epidemiologic data on active and passive
smoking. Thirty epidemiologic studies on SHS lung-cancer effects for normally occurring SHS levels were reviewed, which used spousal smoking as a surrogate for
SHS exposure.
OEHHA [1997] undertook a comprehensive review of SHS health effects starting in 1992 after the USEPA [1992] report was published. The OEHHA report underwent substantial review before the final draft was released and was viewed as
“the most current and definitive statement of the science applicable to [SHS].”1 The
major findings were that SHS exposure can be causally linked to nearly a dozen
fatal and non-fatal general maladies, ranging in severity from low birth weight,
1 As
stated by James Pitts, Chair of the Scientific Review Panel, in a letter dated July 18, 1997.
33
respiratory distress, and ear infections to Sudden Infant Death Syndrome (SIDS),
lung cancer, and heart disease mortality. Based on number of cases, heart disease
was seen as the primary fatal endpoint. Table 2.4 contains a summary of health
effects that were reported to be causally associated with SHS, including morbidity
and mortality estimates for the US. These results show that exposure to SHS is a
major public health problem that can have a large impact on the health of those
who are exposed, especially children.
Other reports present similar conclusions. In the United Kingdom Department
of Health’s “Report of the Scientific Committee on Tobacco and Health” [UKDH,
1998] various other governmental and scientific reports are summarized. The major conclusion of the report is that, “Exposure to environmental tobacco smoke is
a cause of lung cancer and, in those with long-term exposure, the increased risk is
on the order of 20−30%.” Other conclusions concerning non-cancer effects mirror
those in the OEHHA [1997] and USEPA [1992] reports. The second edition of the
WHO Air Quality Guidelines report [WHO, 2000] summarizes the health risks for
SHS as follows:2
ETS has been shown to increase the risks of health effects in nonsmokers exposed at typical environmental levels. The pattern of health
effects from ETS exposure produced in adult nonsmokers is consistent
with the effects known to be associated with active cigarette smoking.
Chronic exposures to ETS increase lung cancer mortality. In addition,
the combined evidence from epidemiology and studies of mechanisms
leads to the conclusion that ETS increases the risk of morbidity and
mortality from cardiovascular disease in nonsmokers, especially those
with chronic exposure. ETS also irritates the eyes and repiratory tract.
In infants and young children, ETS increases the risk of pneumonia,
bronchitis, and fluid in the middle ear. In asthmatic children, ETS increases the severity and frequency of asthma attacks. Furthermore, ETS
reduces birth weight in the offspring of nonsmoking mothers ...
Populations at special risk for the adverse health effects of ETS are
young children and infants, asthmatics, and adults with other risk factors for cardiovascular disease ...
2 Here
and in other quotations, SHS is referred to by its synonym, ETS.
34
Table 2.4: Health Effects Causally Associated with Exposure to Environmental Tobacco Smoke with Annual Morbidity and Mortality Estimates for the US
Class of Effect
Type of Effect
Developmental Low birthweight; small for
gestational age
Respiratory
Carcinogenic
Cardiovascular
Estimated Morbidity &
Mortality [year−1 ]
9,700−18,600 cases
Sudden Infant Dealth
Syndrome (SIDS)
1,900−2,700 deaths
Acute lower respiratory
infections in children (e.g.,
bronchitis, pneumonia)
150,000−300,000 cases,
7,500−15,000
hospitalizations, 136−212
deaths
Asthma induction
8,000−26,000 new cases
Asthma exacerbation
400,000−1,000,000 children
Chronic respiratory
symptoms in children
−
Middle ear infections in
children
0.7−1.6 million physician
office visits
Eye and nasal irritation in
adults
−
Lung cancer
3000 deaths
Nasal sinus cancer
−
Heart disease mortality
35,000−62,000 deaths
Acute and chronic coronary
heart disease morbidity
−
Source: OEHHA [1997].
35
A clear quantitative dose-response, or exposure-response, relationship for the
health effects associated with SHS exposure generally has not been established.
Such a relationship would be useful in estimating health risk and establishing lowrisk levels of exposure. In the WHO [2000] report, the following general guidelines
are given for establishing levels of SHS exposure that can be expected to have a
health impact:
ETS has been found to be carcinogenic in humans and to produce
a substantial amount of morbidity and mortality from other serious
health effects at levels of 1−10 µ g m−3 nicotine (taken as an indicator
of ETS). Acute and chronic respiratory health effects on children have
been demonstrated in homes with smokers (nicotine 1−10 µ g m−3 ) and
even in homes with occasional smoking (0.1−1 µ g m−3 ). There is no evidence for a safe exposure level. The unit risk of cancer associated with
lifetime ETS exposure in a home where one person smokes is approximately 10−3 .
The development of a clear dose-response for SHS itself is hindered by epidemiological studies that typically use exposure surrogates in the form of questionnaires, measured air nicotine concentrations, or cotinine biomarker concentrations, none of which provide precise estimates of personal SHS exposure (see
above). Nevertheless, a number of studies using surrogate measures of exposure
do find a clear trend of increasing health risk with increasing exposure. For example, USEPA [1992] describes eight studies for a lung cancer endpoint in nonsmoking women in which there is a significant upward trend in relative risk for
increasing intensity of spousal smoking in terms of cigarettes smoked per day.
Because SHS, which consists largely of sidestream smoke, is chemically similar to mainstream smoke, the dose-response of SHS and lung cancer can potentially be understood in terms of active smoking, for which there is an apparent
nonthreshold relationship, using the somewhat controversial notion of cigarette
equivalents. Uncertainties in this approach arise because the proportional yield of
chemical species in SHS is different than that for undiluted mainstream tobacco
smoke. A related approach is to use established risk factors associated with particular chemical components of SHS together with estimates of exposure concen-
36
trations to calculate long-term health risks. Nazaroff and Singer [2004] use generic
reference concentrations [OEHHA, 2002] and risk factors [USEPA, 2003] for a set
of hazardous air pollutants present in SHS to estimate hazard indices and cancer risks. Based on exposure estimates for a well-mixed house, their results show
that the volatile and semi-volatile SHS species acetaldehyde, acrolein, acrylonitrile,
benzene, 1,3-butadiene, and formaldehyde have either a hazard index greater than
1, indicating a significant risk of adverse non-cancer effect in the exposed individual, a cancer risk in excess of 1 per million, or both.
2.3 Field Studies of SHS Exposure and Multiple Compartment Effects
As stated above, SHS is a complex mixture of gaseous and particulate species.
A large proportion of SHS is composed of toxic or irritating gaseous chemicals,
such as nicotine, carbon monoxide, and nitrogen oxides. However, particles contribute a larger proportion of SHS by mass than any individual gaseous species,
except for the carbon oxides. SHS particles are predominantly organic compounds,
which include a variety of toxic polycyclic aromatic hydrocarbons. In general, the
measurement of particle mass or number concentration has fairly well-established
and easily-performed methods and the dynamics of particles are reasonable well
understood. Unlike chemically reactive species or species that may desorb from
surfaces, such as nicotine, the behavior of particles after they are emitted follow
relatively simple physical processes. While SHS particles may undergo a small degree of coagulation, evaporation, or condensation, their dynamics are expected to
be dominated by removal through ventilation or active filtration, and irreversible
deposition onto surfaces. Perhaps most importantly, particles have the characteristic of being able to penetrate and deposit deep in a person’s lung, so that they
pose a substantial risk for long-term ailments such as lung cancer.
For this array of reasons, much research into SHS exposure has been devoted to
measuring and modeling particle concentrations associated with SHS. A primary
limitation of using particles as the measured SHS component of interest, such as
37
for a marker for other SHS species, is that they are not specific to SHS. Table 2.5
summarizes the results of 23 large-scale field studies in which SHS-associated particle concentrations were measured in residential environments or using personal
monitors, i.e., sampling devices attached to a person as they travel from place-toplace.
For 24-h, 1-week, or 2-week PM2.5 samples3 taken as part of the PTEAM, Harvard Six City, and New York State studies [Özkaynak et al., 1996; Neas et al., 1994;
Spengler et al., 1985, 1987; Leaderer et al., 1990], smoking contributed an average
of approximately 30 µ g m−3 to overall average particle concentrations in monitored houses. Quackenboss et al. [1989] report mean PM2.5 particle concentrations
measured in 98 homes as a function of the number of cigarettes smoked per day.
Smoking one pack of cigarettes per day contributed 12 µ g m−3 , on average, above
levels with no smoking and smoking more than a pack contributed an average of
45 µ g m−3 . Similarly, Koistinen et al. [2001] report mean PM2.5 measured in smoking homes to be about 13 µ g m−3 higher than in non-smoking homes. Personal
monitoring results reported as part of CIAR tobacco-industry-sponsored studies
span a large range with reported mean or median RSP or PM3.5 concentration
increases for participants living in smoking homes in the range of 0 to about 30
µ g m−3 [Phillips et al., 1996, 1997a,b, 1998a,b,c,d,e,f,g,h, 1999; Phillips and Bentley, 2001]. In contrast, a study funded by R.J. Reynolds (RJR) tobacco company
reported personal PM3.5 concentation for participants living in smoking homes to
be about 60 µ g m−3 larger than for those living in nonsmoking homes [Heavner
et al., 1996]. Some of the RJR and CIAR results are difficult to interpret in terms of
strict residential and non-residential contributions.
Before embarking on an exposure modeling investigation that presumes a significant multi-compartment effect on long-term SHS concentrations, i.e., significant differences between air pollutant concentrations in different rooms of a house
over extended time periods, and, therefore, potential SHS exposures, it is important to establish empirical evidence for this effect. Such evidence for the effect
3 PM
2.5
consists of particles with aerodynamic diameters less than 2.5 µ m.
101 households; personal
and household
monitoring;
Kingston-Harriman, TN
300 households with
children; Watertown, MA;
St. Louis, MO;
Kingston-Harriman, TN
98 households; Tuscon,
AZ
359 stratified households
with valid data;
Onondaga and Suffolk
Counties, NY
585 office environments
1,273 households with
children aged 7 - 11;
Caucasian
178 random nonsmokers
aged 10 - 70; personal
monitoring; households;
Riverside, CA
Harvard Six City d
Harvard Six City d
-
New York State e
-
Harvard Six Cityd
PTEAM c
Spengler et al. [1985]
Spengler et al. [1987]
Quackenboss et al. [1989]
Leaderer et al. [1990]
Turner et al. [1992]
Neas et al. [1994]
Özkaynak et al. [1996]
Continued.
Subjects/Locations
Surveyed
Study Name
Investigators
PM2.5 ; 12-h samples;
day/night
PM2.5 ; 2-wk samples
PM3.5 ; 10 samples per
hour
PM2.5 ; 1 wk samples
PM2.5
PM2.5 ; 1-wk samples
PM3.5 ; 24-h samples
Methods
27-32 SHS contribution to
PM10 /PM2.5 ; day or night
annual means: 48.5 SH;
17.3 NSH
means: 46 SW; 20 NSW
geometric means: 29-61
SH; 14-22 NSH
means: 27 SH <= 1 pack
d−1 ; 61 SH > 1 pack d−1 ;
15 NSH
means: 30 greater in SH
than in NSH
means: 74 SH; 28 NSH
Concentrations (µ g m−3 )
31 homes with smokers;
61 samples day + night
580 consistently SH; 470
consistently NSH
331 smoking offices; 254
nonsmoking offices
238 SH; 121 NSH
45 NSH; 26 SH <= 1 pack
d−1 ; 17 SH > 1 pack d−1
NA
28 SH; 73 NSH
Sample Characteristics
Results b
Table 2.5: Environmental Tobacco Smoke Particle Concentrations Measured During Field Surveys a
38
190 working and
nonworking nonsmokers;
personal monitoring;
Stockholm, Sweden
154 office workers and
housewives; personal
monitoring; Barcelona,
Spain
monitoring; Turin, Italy
monitoring; Paris, France
CIAR g
CIAR g
CIAR g
CIAR g
Phillips et al. [1996]
Phillips et al. [1997a]
Phillips et al. [1997b]
Phillips et al. [1998a]
Continued.
104 nonsmoking married
female subjects over 25;
personal monitoring;
households; workplaces;
New Jersey and
Pennsylvania
RJR h
1564 subjects; personal
monitoring; households;
workplaces; 16 US cities
Heavner et al. [1996]
f ,g
Sixteen cities/CIAR
Jenkins et al. [1996]
Subjects/Locations
Surveyed
Study Name
Investigators
Table 2.5. [Continued]
RSP; 24-h samples
RSP; 24-h samples
RSP; 24-h samples
RSP; 24-h samples
PM3.5 ; ∼ 14-h samples at
home; ∼ 7-h samples at
work
PM3.5 ; 8-h sample at work;
16-h sample at home
Methods
medians: 62 SH; 36 NSH;
80 SH/SW; 64 SH/NSW;
43 NSH/SW; 35
NSH/NSW
59 NSH/SW; 55
NSH/NSW
85 SH/ any workplace; 40
NSH/any workplace; 94
SW/any home; 52 NSW
/any home
medians: 39 SH; 18 NSH
means: 89 SH; 28 NSH
(without regard to work)
means: 44 SH; 20-21 NSH;
49 SW; 18 NSW
51 SH; 44 NSH; 45
SH/SW; 13 SH/NSW; 59
NSH/SW; 10 NSH/NSW
36 SH; 47 NSH; 21
SH/SW; 9 SH/NSW; 51
NSH/SW; 24 NSH/NSW
43 SH; 42 NSH; 25
SH/SW; 3 SH/NSW; 36
NSH/SW; 5 NSH/NSW
9 SH; 31 NSH
29 SH; 58 NSH
SH: 306; NSH: 2078; SW:
331; NSW: 867
Results b
39
238 random nonsmoking
office workers and
monitoring, Prague, Czech
Republic
241 random office workers
and housewives; personal
monitoring; Kuala Lumpur,
Malaysia
319 nonsmokers; personal
monitoring; Sydney,
Australia
office workers and
monitoring; Hong Kong
office workers and
monitoring; Lisbon, Portugal
office workers and
monitoring; Beijing, China
office workers and
monitoring; Bremen,
Germany
CIAR g
CIAR g
CIAR g
CIAR g
CIAR g
CIAR g
CIAR g
Phillips et al. [1998b]
Phillips et al. [1998c]
Phillips et al. [1998d]
Phillips et al. [1998e]
Phillips et al. [1998f]
Phillips et al. [1998g]
Phillips et al. [1998h]
Continued.
Subjects/Locations Surveyed
Study Name
Investigators
RSP; 24-h samples
RSP; 24-h samples
RSP; 24-h samples
RSP; 24-h samples
RSP; 24-h samples
RSP; 24-h samples
RSP; 24-h samples
Methods
medians: 36 SH; 25 NSH; 39
NSH/SW; 23 NSH/NSW
100 NSH/SW; 95 NSH/NSW
NSH/SW; 34 NSH/NSW
NSH/SW; 43 NSH/NSW
SW; 16 NSW
NSH/SW; 43 NSH/NSW
NSH/SW; 30 NSH/NSW
21 SH; 60 NSH; 18 SH/SW; 6
SH/NSW; 49 NSH/SW; 36
NSH/NSW
56 SH; 46 NSH; 46 SH/SW;
31 SH/NSW; 47 NSH/SW;
27 NSH/NSW
24 SH; 56 NSH; 28 SH/SW; 7
SH/NSW; 61 NSH/SW; 21
NSH/NSW
35 SH; 35 NSH; 21 SH/SW;
43 NSH/NSW
30 SH; 48 NSH; 20 SW; 60
NSW
42 SH; 51 NSH; 30 SH/SW;
44 NSH/NSW
54 SH; 39 NSH; 64 SH/SW;
20 NSH/NSW
Results b
40
196 nonsmoking office
workers and housewives;
personal monitoring; Basel,
Switzerland
office workers; personal
monitoring; Bremen,
Germany
201 random adults aged
25-55; household,
workplace, and personal
monitoring; Helsinki,
Finland
CIAR g
CIAR g
EXPOLIS i
Phillips et al. [1999]
Phillips and Bentley [2001]
Koistinen et al. [2001]
PM2.5 ; 48-h samples
RSP; 24-h samples
RSP; 24-h samples
Methods
averages: 21 SH; 8.2 NSH;
30 SW; 9.5 NSW; 31
personal active smoker; 17
personal SHS-exposed NS;
9.9 unexposed NS
smoking locations: 48-53
(winter); 22-30 (summer);
nonsmoking locations:
22-28 (winter); 17-20
(summer)
39 SH/SW; 24 SH/NSW; 27
NSH/SW; 26 NSH/NSW
57 SH; 135 NSH; 46 SW; 105
NSW; 48 active smokers; 9
SHS-exposed NS; 137
non-SHS exposed NS
winter: 49 SW & SH; 53
NSW & NSH; summer: 52
SW & SH; 50 NSW & NSH
26 SH; 60 NSH; 25 SH/SW;
28 NSH/NSW
Results b
a The listed studies are limited to those that are large (n > 100) and/or probability-based with city-wide or larger scope, and where SHS-related and non-SHS-related particle
levels were reported.
b
NSH = nonsmoking home; SH = smoking home; NSW = nonsmoking workplace; SW = smoking workplace; NS = nonsmoker; NA = not available.
c PTEAM = USEPA’s Particle Total Exposure Assessment Methodology.
d The Harvard Six City study is described with preliminary results in Spengler et al. [1981]. The six cities were: Portage, WI; Topeka, KS; Kingston-Harriman, TN;
Watertown, MA; St. Louis, MO; Steubenville, OH.
e
The main reference for the New York State study is Sheldon et al. [1989].
f The sixteen cities surveyed were: Knoxville, TN; Portland, ME; San Antonio, TX; Fresno, CA; Boise, ID; Seattle, WA; Baltimore, MD; Columbus, OH; Daytona Beach, FL;
Buffalo, NY; St. Louis, MO; Grand Rapids, MI; Camden/Philadelphia, NJ,PA; Indianapolis, IN; Phoenix, AZ; New Orleans, LA.
g These studies were sponsored by the now defunct Center for Indoor Air Research (CIAR), which was affiliated with the tobacco industry. The Phillips et al. studies
typically recruited subjects that stayed mostly at home (single monitor subjects; SH or NSH) and/or those that were both at home and at work (dual monitor subjects; SH or
NSH with SW or NSW). Sample sizes are total study subjects. The statistics presented represent both time in and out of work for dual monitor subjects. In contrast, for the
Phillips et al. (1998g) study the listed 24-h results reflect pumps that were shut off when subjects were not at home or at work (although 72-h and continuously-sampling
24-h samples were also collected).
h Conducted by R.J. Reynolds Tobacco Company, Research and Development.
i The main reference for EXPOLIS is Jantunen et al. [1998].
Subjects/Locations
Surveyed
Study Name
Investigators
41
42
of multi-compartment pollutant and personal behavior on residential exposure to
SHS is best investigated by performing intensive multi-room pollutant monitoring studies of homes including personal exposure monitoring, preferably coupled
with activity diaries for house occupants. While sufficiently detailed information
is lacking in the large-scale SHS field studies described in Table 2.5, several smaller
experimental studies, some specific to SHS and some not, provide direct evidence
of the multi-compartment character of homes.
Wayne R. Ott performed ten SF6 tracer experiments in a large 750 m3 ranchstyle house to mimic scenarios that might occur in a household where nonsmokers attempt to avoid smokers [Ott, 2002]. For example, the smoker might be sequestered into a separate room or the nonsmokers might try to isolate themselves
from the smoker behind closed doors or simply on the other side of the house. To
limit his study to the effect of interior doors on room concentrations, during all experiments exterior windows and doors were kept closed and the central forced-air
system was inactive. For each experiment, SF6 was released at a constant rate of
200 ml min−1 for 30 min in a “source room,” which, for each experiment, was the
103 m3 kitchen. Measurements of SF6 were taken both in the source room and a
“test room” (an adjacent den, the living room, or three different bedrooms even
further away). For 4 of the experiments, the test room’s door was closed and for
one of these the source room’s door was also closed. For the other 6 experiments,
both the source and test room doors were left open. A detailed schematic showing
the layout of the house, monitor positions, and room volumes is given in Figure 2.1
and Figure 2.2 shows the concentration time series that was measured during each
experiment. Table 2.6 contains a summary of all ten experiments, including calculated absolute means and maxima in each room taken over a 20−24-h period, the
time it took to reach a maximum concentration in the test room, and relative measures consisting of ratios of test room means and maxima to those for the source
room.
For experiments involving the den, the living room, and the closer bedrooms
as test rooms where doors were open (experiments 1, 2, 8, 9), the test room mean
43
concentration was 48−75% of the mean concentration in the source room (kitchen).
The only other experiment where the test mean was approximately 50% or greater
was experiment 7 where the den, which shares a wall with the kitchen but whose
entrance is down the hall, was the test room. Its door was closed during the experiment. For the open-doors experiment involving the farthest bedroom as a test
room (experiment 5), the mean was only 18% of the source room mean, reflecting
the relatively slow rate of air flow between these rooms. When the bedroom doors
were closed (experiments 3 and 6), their means were only 2−8% of the kitchen
mean. Closing both the den and the kitchen door also had a dramatic effect (experiment 10), reducing the den mean to 8% of the kitchen mean. These results
show that the separation of rooms by a hallway and the closing of interior doors
can significantly reduce the rate of pollutant transport between zones of a house.
In another series of real-world experiments, Löfroth [1993] measured the roomto-room concentrations of nicotine, respirable suspended particles (RSP), and isoprene during replicate experiments in a 140 m3 3-bedroom apartment and a 300 m3
3-story, 3-bedroom townhouse. During the experiments, cigarettes were smoked
every 20−30 min over an air sampling time period of 200 min. Interior doors were
left open during the experiments. Cigarettes were also smoked for 90 min prior
to the sampling period at the same rate. The ventilation rates for each location
were approximately 0.5 h−1 during each experiment, and the outdoor RSP levels
were under 50 µ g m−3 . The results of the four total experiments (see Table 2.7)
showed that average particle concentrations in the living rooms, where smoking
took place, were 300−500 µ g m−3 taken over the 200 min sampling period, isoprene concentrations were 50−160 µ g m−3 , and nicotine air concentrations were
31−87 µ g m−3 . The particle concentrations in all three rooms of the smaller apartment were approximately uniform (a maximum difference of about 20%), whereas
levels in rooms on other floors of the townhouse were 30−40% lower than in the
living room. In contrast to RSP, the nicotine concentrations in the apartment were
about 14 − 32 of the concentration in the living room, and they were about
1
3
of living
room concentration for a room on the same floor in the townhouse and 80−90%
Source Monitor
SF 6 Source
Kitchen, 103 m³
Test Monitor
Dining Room, 58 m³
Main Hallway, 34 m³
Den, 69 m³
Front
Door
Front
Hall,
23 m³
Test Monitor
Living
Room, 112 m³
Closet, 24 m³
Linen
Bathroom, 20 m³
Test Monitor
Bedroom #3,
56 m³
Bathroom, 12 m³
Test Monitor
Bedroom #2,
57 m³
Test Monitor
2nd Hallway, 20 m³
Figure 2.1: Schematic, including room volumes, for the layout of a large ranch-style house in which SF6 concentrations were measured during 10 two-zone experiments. The SF6 source was located in the kitchen with a
“source-room” monitor also in the kitchen. The “test-room” monitor was moved between the den, the living room,
and the three bedrooms. See the text, Table 2.6, and Figure 2.2 for the results of these experiments.
Total Volume = 714 m³
(omitting the garage)
Ceiling Height = 8.3 ft
Garage
Laundry, 40 m³
Bedroom #1, 86 m³
44
45
Table 2.6: Summary of Two-Room SF6 Tracer Experiments in a Large Ranch Style
House
Source Room
Mean
No.
Test Room
a Max
[ppm] [ppm]
b Room c Door
Vol
d Mins
Mean
[m3 ]
Max
[ppm] [ppm]
Max
e Mean e Max
%
%
1
2.5
46
Den
O
69.1
73
1.8
8.4
75
18
2
2.2
73
Bed 3
O
55.8
145
1.6
6.5
73
8.9
3
1.6
68
Bed 1
C
55.8
95
0.11
1.1
6.9
1.5
4
2.0
35
Bed 1
O
86.2
88
0.029
0.11
1.4
0.31
5
3.1
38
Bed 1
O
86.2
243
0.57
2.9
18
7.7
6
1.9
72
Bed 3
C
55.8
86
0.15
0.38
8.0
0.53
7
2.6
79
Den
C
69.1
141
1.4
6.6
54
8.3
8
2.5
92
Liv
O
111.7
78
1.3
9.4
52
10
9
2.0
70
Bed 2
O
57.0
137
0.93
4.4
48
6.3
10
4.7
75
Den
C
69.1
131
0.39
2.7
8.4
3.6
These SF6 tracer experiments were carried out in 1997 by Wayne Ott in Atherton, CA from April
21 to April 30. The time interval between SF6 measurements was 1-2 min, and SF6 concentrations
are presented here in parts per million (ppm). For each experiment, the SF6 source was active for
30 minutes during each experiment in the “source room”. The SF6 flow rate was approximately
200 ml min−1 throughout each experiment. The designated source room for each experiment was
the residence’s kitchen with a volume of 103.3 m3 . The total number of minutes over which SF6
concentrations were measured beginning when the SF6 source became active ranged from approximately 1,200 to 1,440 minutes.
a The reported maximum in the source room may reflect transient peaks due to imperfect mixing.
b The test rooms contained no SF sources, varied in volume, and were located between 9 and 24 m
6
from the source room.
c O = door to test room left open; C = door to test room left closed
d The minutes to the maximum SF concentration in the test room beginning when the SF source
6
6
became active. If the test room concentration time series was low and broad, each maximum and
the corresponding minutes to maximum might reflect instrument noise rather than the actual peak.
e Mean % and Max % are the ratios of Test Mean and Test Max to Source Mean and Source Max
multiplied by 100. Since the reported maxima for the source and test rooms may contain biases due
to imperfect mixing or instrument noise, the Mean % may be a better indicator of relative concentrations for some experiments.
See Figure 2.1 for a schematic of the house layout and Figure 2.2 for time-series plots of experimental SF6 concentrations in each room.
46
0
200 400 600
1
2
Kit: Open
Den: Open
Kit: Open
Bed 3: Open
0
200 400 600
3
Kit: Open
Bed 1: Open
4
5
Kit: Open
Kit: Open
Bed 1: Closed Bed 1: Open
80
60
SF6 Concentration (ppm)
40
20
0
6
Kit: Open
Bed 3: Closed
7
8
Kit: Open
Den: Closed
Kit: Open
Liv: Open
9
10
Kit: Open
Bed 2: Open
Kit: Closed
Den: Closed
80
60
40
20
0
0
200 400 600
0
200 400 600
0
200 400 600
Elapsed Minutes
Kitchen
Den
Bedroom #3
Bedroom #1
Living Room
Bedroom #2
Figure 2.2: Plots showing the time series of SF6 concentrations measured in a large, ranchstyle house during 10 two-zone experiments. Door positions, i.e., either fully “open” or
“closed”, for each of the two rooms where concentrations were measured are given at the
top of each plot. Line colors for the plotted data series designate the "source" and "test"
rooms where measurements took place for each of the experiments. The kitchen was the
source room for all experiments, and the test room varied between the den, living room,
and each of three bedrooms. These plots reflect raw, unpublished data obtained from
Wayne Ott. See Figure 2.1 for a schematic of the house layout and Table 2.6 for more
information on the experiments, including room sizes and room concentration statistics.
47
Table 2.7: Average SHS Nicotine and Respirable Suspended Particle (RSP) Concentrations [µ g m−3 ] Measured in Multiple Rooms of Two Residences
Pollutant/
Living
Residence
Experimenta
Room
Kitchen
Bedroom
Study
Attic
Apartment
Nicotine A
31
−
20
8
−
(140 m3 )
Nicotine B
35
−
24
10
−
RSP A
350
−
350
290
−
RSP B
410
−
420
330
−
Townhouse
Nicotine A
63
20
11
8
9
(300 m3 )
Nicotine B
87
32
9
10
12
RSP A
380
−
270
−
240
RSP B
480
−
290
−
280
Source: Löfroth [1993].
a Two experiments, designated as A and B, were conducted in each residence. Smoking took place
in the living room at each residence every 20−30 min before and during a 200-min sampling period.
lower for rooms on other floors.
Löfroth concluded that there is efficient dispersion of particles between rooms
of homes, which means that occupants cannot escape exposure to SHS. Closed
doors are expected to decrease the movement of SHS, but he concludes that it
would be impractical to maintain a closed door for the several hours that would
be required to clear the generated SHS at the given ventilation rates, although he
does not provide any elaboration on exactly why such a strategy would be untenable. He does not consider timing of cigarettes with closing of doors, opening
windows, or some combined strategy, or any data on time-activities of house occupants. He notes that, since nicotine is known to disappear more rapidly than
other SHS components, exposure to nicotine is more dependent than exposure to
particles on the distance between the smoker and passive smoker.
In the 1970’s and 80’s indoor air quality researchers conducted numerious studies concerned with household gas stove or heater emissions, in which either carbon
monoxide (CO) or nitrogen dioxide (NO2 ), two prominent products of incomplete
48
combustion, were measured in multiple rooms of multiple houses. These studies
include those by Sheldon et al. [1989], Palmes et al. [1977], Wade et al. [1975], Sterling and Sterling [1979], and Palmes et al. [1979]. Average concentrations measured
in these studies, typically over time periods of a day or week, broadly reflect the
location, activity, and emissions of gas appliances, the movement of air between
rooms, and the residence time of different pollutants, although finely resolved activities in homes were not recorded. Because of its reactivity, NO2 gas is expected
to have less persistence than the non-reactive CO gas, which may contribute to
larger differences in NO2 concentrations between compartments. For this reason,
the results for NO2 reported here are not directly applicable to concentrations of
some SHS constituents in homes, but they do inform the broad issue of inter-room
pollutant variability.
Sheldon et al. [1989] report on a large study of indoor air quality for which
indoor and outdoor concentrations were measured of the combustion products
NO2 , CO, SO2 , and respirable suspended particles (RSP) across a probability sample of 400 homes located in New York State’s Onondaga and Suffolk Counties. For
houses in Onondaga County that had a gas stove, 7-d average NO2 levels in the
living room were 79% of those in the kitchen, where gas ovens and stoves were
located. In Suffolk County living room levels were 71% of those in the kitchen.
For a subset of 13 homes in the Sheldon et al. study, real-time concentrations of
CO and NO2 were measured in multiple rooms along with daily activity patterns
for ovens, stoves, and smokers. For 5 of these homes, the 7-d average concentrations in the living room was 37−77% of that in the kitchen. I analyzed the daily
averages for 4 of the homes selected for real-time measurements, reading data from
a series of plots in the authors’ final report, and found much variation in the relative concentations in the kitchen versus the living room or den. For one house
in particular, the kitchen levels were consistently 2−3 times higher than those in
the living room. For other homes, daily averages in the kitchen could range from
approximately equal to those in the living room to more than twice as high. This
observed variation is presumably due to day-to-day changes in flow patterns or
49
applicance activity in the monitored houses.
Other studies of interzonal concentrations from combustion applicances were
less extensive than the one by Sheldon et al. and not probability-based, being
limited to a small number of homes. Palmes et al. [1977] measured NO2 concentrations in kitchen and non-kitchen areas of 10 homes with gas stoves, finding that
1-week average concentrations in the non-kitchen averaged about 50% of those in
the kitchens. As a follow-up, Palmes et al. [1979] studied NO2 levels in 12 Florida
homes with either gas stoves or space heaters, finding that 1-week average concentrations in sourceless rooms could be as low as 15−27% of rooms with space
heaters and as low as 35−48% of rooms with stoves, i.e., the kitchen. In a study of
four houses with gas stoves, in which NO, NO2 , and CO were continuously measured in three different rooms, Wade et al. [1975] found that daily living room and
bedroom NO2 concentrations were consistently as low as 50−60% of concentrations in kitchens.
In addition to the multi-home field studies described above, Miller and Nazaroff
[2001], Ott et al. [2003], and De Gids and Phaff [1988] report results from experimental studies in single houses or multi-room test facilities, in which the effect of
interior door positions was explicitly investigated in the context of SHS exposure.4
Miller and Nazaroff [2001] report that after a 10-min cigarette was smoked, the 1-h
average particle concentration in a nonsmoking room was 85% of that in the smoking room when the adjoining door was left open. When the door was fully closed,
the nonsmoking room concentration was only 3% of that in the smoking room.
Ott et al. [2003] also report a dramatic effect of door position for an experiment
in which the door connecting a room where a cigar was smoked and an adjacent
living area was open 3 inches and the door to a bedroom (one room removed) was
fully closed. The 1-h average CO concentration in the adjacent living area, calculated after smoking stopped, was 60% of the average concentration in the smoking
room and the concentration in the far bedroom was only about 1−2% of the concentrations in either of the other two rooms. De Gids and Phaff [1988] present data
4 Data
from the first two studies have been used to make quantitative estimates of air flow rates
between rooms for use in modeling indoor air pollutant concentrations (see Chapter 5).
50
from what appears to be a continuous-release CO-tracer gas experiment in a house,
during which the position of the door between the downstairs living area and the
upstairs bedrooms was kept closed for 3 h and then opened for 2 additional hours.
When the door was closed, the concentrations in the upstairs hallway and one of
the bedrooms were negligible, although they increased to levels comparable to the
living room when the door was opened.
2.4 Children’s Residential SHS Exposure
For the research I present in this dissertation, I focus on factors that modify residential exposure to SHS. Because of the large amount of time spent at home with
family members and visitors who may smoke, one’s residence is the location where
the most current SHS exposure occurs. My central hypothesis is that the movement of individuals about their homes, coupled with the location of the smokers
and the configuration of doors, windows, and ventilation systems, likely leads to
substantial variation in exposure, presenting possibilities for exposure reduction
that occupy the middle ground between unrestricted smoking and a complete ban
on smoking in the home. Although I don’t limit my analysis of SHS exposure
to children in particular, young children are the demographic group of most concern with respect to SHS exposure occurring in homes, because they are typically
unknowing victims and are particularly susceptible to a large number of adverse
health effects. They also are apt to be exposed at high levels if given to the care of a
smoker, and may spend more time at home than members of smoking households
who work or attend school.
2.4.1 Exposure Prevalence
How many children in the US are exposed to SHS at home and how many parents
who smoke, or who have visitors who smoke, are concerned enough about exposure to restrict household smoking? In this section, I present the results of studies
that have gauged the extent of children’s exposure in terms of the percentage of
51
children in the US that have some SHS exposure at home and the percentage of
homes with smokers that have enacted some sort of household smoking restrictions for the protection of children and other household members. The studies are
summarized in Table 2.8.
The 1988-1991 National Health and Nutritional Examination Survey (NHANES
III) was a nationally representative cross-sectional survey conducted in the US. It
included questionnaire information on 16,818 persons aged 2 mo and older and
measurements of serum cotinine for 10,642 persons over 4 y old. Based on this
survey, Pirkle et al. [1996] report that 43% of children between the ages of 2 mo
and 11 y lived in a home with at least one smoker and 37% of adult non-tobacco
users lived in a home with at least one smoker. However, serum cotinine levels
suggested that SHS exposure was even more widespread. Eighty-eight percent
of surveyed persons who were non-tobacco users had detectable levels of serum
cotinine, and increased levels were associated with the number of smokers in a
household. In 1999-2000 a new NHANES was conducted and participants had
median serum cotinine levels that were 70% lower than median levels observed
during the 1988-1991 NHANES, suggesting a dramatic reduction in exposure to
SHS for the general US population [USDHHS, 2003].
Several recent human activity pattern surveys in the US and Canada contain
data on the time spent exposed to SHS in residences. A general limitation of these
surveys is that exposure was recorded from each respondent’s subjective evaluation of the presence of one or more smokers in each location they visited. This
measure of exposure may have resulted in biased estimates of exposure prevalence. Smoker presence may not have been reported in particular rooms or locations where residual SHS was present. Smoker presence may also have been reported for extended periods where smoker presence, and therefore SHS exposure
was minimal.
The 1992-94 US National Human Activity Pattern Survey (NHAPS) interviewed 9,386 randomly-selected repondents in the contiguous 48 states, who provided a 24-h recall diary containing the locations they visited along with the con-
Method
Telephone Interviews
Personal Interviews
Personal Interviews
Personal Interviews
Questionnaire
Reference
McMillen et al. [2003]
Pizacani et al. [2003]
Kegler and Malcoe
[2002]
Schuster et al. [2002]
Klepeis et al. [2001]
MMWR [2001]
Borland et al. [1999]
Norman et al. [1999]
MMWR [1997]
Pirkle et al. [1996]
Leech et al. [1996]
Jenkins et al. [1992];
Wiley et al. [1991b]
Wiley et al. [1991a]
1,200
1,762
10,642
−
7,000
2,500
−
9,386
45,335 original
respondents
380
> 6,000
1,503 (2000);
3,002 (2001)
Sample
California; children
under 12
California; adults
and youth > 12
4 Canadian Regions
US
US
California
Victoria, Australia
20 US states
US
US
Oklahoma; rural,
low-income
Oregon
US
Population
25% of children are exposed at home
26% exposed at home
30−34% of all ages exposed in all
locations
43% of US children 2 mo to 11 y
exposed at home
70% to 96% of homes with a smoker
allow smoking; 22% of all children
are exposed in their homes
24% of all homes and 57% of homes
with smokers allow smoking
67% of smokers allow smoking
around children (1997)
21% to 39% of all homes allow
smoking
26% of all ages exposed at home
35% of children live in homes where
smoking is allowed
51−57% of all caregivers allow
smoking in homes
50% of children in smoking
households are exposed
38% of smokers allow smoking in
front of children (2001)
Results
Table 2.8: Surveys on the Prevalence of Household Smoking Restrictions and Children’s Exposure to SHS
52
53
current presence of active smoking. Klepeis et al. [2001] report that 44% of all
responents said they were exposed to SHS with 26% being exposed in the home,
which is the highest percentage of all studied locations (residence, office-factory,
restaurant-bar, vehicle, outdoors, and other indoor locations). This percentage of
persons who report being exposed at home is lower than the percentage of children living in a home with a smoker as reported by Pirkle et al. [1996]. This discrepancy might be explained by the existence of smoking restrictions in the homes
of NHAPS respondents that contain smokers, which would reduce the reported
amount of time spent in the presence of a smoker.
For the 1994-95 Canadian Human Activity Pattern Survey (CHAPS), 2,381 respondents were interviewed by telephone in Toronto, Vancouver, Edmonton, and
Saint John [Leech et al., 1996], giving, as with NHAPS, a 24-h recall diary that included a code for the presence of a smoker in different locations. According to
Leech et al. [1999], 30-34% of adults, youth, and children that were part of CHAPS,
reported being exposed to SHS. Children reported being exposed mostly at home
between the hours of 4:00 PM and midnight, with exposure occurring largely in
living rooms and bedrooms.
Two earlier human activity pattern studies were conducted on 1,762 randomlyselected California adults and youth during 1987-88 [Jenkins et al., 1992; Wiley
et al., 1991b] and 1,200 children under age 11 during 1988-89 [Wiley et al., 1991a].
For these studies, 62% of adults and youth and 38% of children said they were
exposed to SHS in any location, and 26% of adults and youth and 25% of children
were exposed at home. The reduction in overall reported exposure between the
late 1980’s and mid-1990’s is apparently due to that occurring outside the home,
since the percentage of those reporting home exposure remained approximately
the same.
Schuster et al. [2002] analyzed data from the 1994 National Health Interview
Survey (NHIS) (n=45,435) and Year 2000 Objectives supplement, which are representative of the US civilian population. They report that 35% of US children live
in homes where residents or visitors smoke more than 1 d per week. For 16% of
54
nonsmokers with children, other residents or visitors are allowed to smoke in the
home, and 6% of homes where no residents smoke allow smoking by visitors. As
with the human activity pattern surveys, these results are based on actual smoking
in homes instead of whether or not smokers simply live in the home, which is the
case for the results of Pirkle et al. [1996].
McMillen et al. [2003] analyzed nationally representative cross-sectional data
from telephone surveys on smoking attitudes and behavior conducted in the US
during the summers of 2000 and 2001. The overall proportions of households reporting total smoking bans were 69% for 2000 and 74% for 2001, an appreciable
increase, and 95% of respondents believe inhaling SHS harms infants and children. However, in 2001 only 30% of smokers reported a household ban and 38%
allowed smoking in front of children, whereas 86% of nonsmokers had a household ban and 95% never allowed smoking in front of children. The lower rate of
household bans among smokers occurs even though 90% of smokers recognize
the adverse effects of SHS for children and infants and 50% of smokers think it is
unacceptable for parents to smoke in front of children.
Kegler and Malcoe [2002] conducted personal interviews with 380 rural, lowincome parents and guardians of children aged 1−6 residing in Oklahoma. For
White interviewee’s, 43% of all caregivers had complete smoking bans in homes
and 40% in cars, whereas 49% of Native American homes and 35% of cars had
smoking bans. Overall, only 22% of smokers reported a complete ban. The low
rates of smoking bans relative to the general population suggested to the authors
that low-income, rural people may need focused intervention efforts.
Borland et al. [1999] report on SHS-related data collected during face-to-face
surveys of about 2,500 randomly selected adults each year between 1989 and 1997
in Victoria, Australia. The authors note that in 1992, there was a concerted media
campaign urging smokers who couldn’t quit to at least not expose their children
to cigarette smoke, suggesting they should smoke outside and ban smoking in
their homes. This campaign is credited by the authors with reducing the amount
of smoking around children, and they suggest that more intensive interventions
55
may not be necessary in the face of the apparent effectiveness of campaigns that
explicitly communicate the importance of protecting others from exposure in the
home. Between the earlier and later surveys, the prevalence of not smoking in the
presence of children rose from 14% to 33%.
Pizacani et al. [2003] used data from a population-based cross-sectional telephone survey of over 6,000 Oregon residents conducted in 1997 containing questions on tobacco use. Of households with at least one smoker, 38% had full smoking bans and 33% had partial bans (smoking restricted by time and/or place), but
nearly 50% of households with children and a smoker did not have a full ban in
place, with 38% having a partial ban.
Norman et al. [1999] report on the results from a representative sample of almost 7,000 Californian adults aged over 18 who were interviewed by telephone
during 1996 and 1997. They found that, overall, 76% of all adults have complete
home smoking bans, with 43% of adult smokers having a complete smoking ban.
There was a clear decrease in the prevalence of complete smoking bans for respondents as the proportion of their friends who smoke increased. Except for cases
where more than half of their friends smoked, smokers with children at home were
only slightly more likely to have smoking bans than smokers without children.
2.4.2 Household Restriction Effectiveness
Restrictive approaches to reducing household exposures to SHS include limiting,
or eliminating, the number of cigarettes that are smoked in the house, i.e., a partial
or total smoking ban, or changing the location (i.e., room) or timing of smoking
activity to avoid the presence of nonsmoking family members. It is not clear exactly how effective approaches that stop short of a total ban on household smoking
are. In addition to restrictions on location and smoking intensity, exposure reduction efforts may include separate or simultaneous directives to block the physical
movement of smoke around the house, e.g., by closing doors, or enhancing the
removal of smoke by opening windows or operating filtration devices. These possibilities, which are a focus of the original research presented in this dissertation,
56
have not received much attention in the SHS health or exposure literature. In this
section, I first review the results of studies that provide evidence for reductions in
smoking behavior as a result of the establishment of household smoking restrictions. Next I review studies, typically cross-sectional surveys, that provide some
evidence, however limited, for the effectiveness of particular types of household
smoking restrictions in actually reducing SHS exposure.
There is evidence that partial and total bans on household smoking behavior can result in significantly enhanced quitting behavior. In the context of interventions (see below), Gehrman and Hovell [2003] report an interdependence
between different intervention targets. Efforts towards cessation could lead to reduction strategies, and exposure reduction could result in cessation. Wakefield
et al. [2000b] analyzed data from a 1996 cross-sectional survey of over 17,000 US
students aged 14 to 17 y that contained questions on smoking restrictions at home,
school, and in public places. They found that home smoking restrictions reduced
the likelihood of smoking uptake by students with total bans being more effective.
The effect was larger than for bans on smoking in public places or at school. The results apply for cases where parents both were and were not smokers. Gilpin et al.
[1999] report an association between recent quit attempts and intention to quit
with social pressure from within the household, which appears to be expressed
in terms of house smoking restrictions. They also found that light smoking increased in homes with more smoking restrictions. Their results are based on a
population survey of over 8,900 people in California. Based on a large, nationally representative sample of current and recent former adult smokers over age
18 conducted during 1992-93 in the US, Farkas et al. [1999] found that partial and
total bans on smoking in homes were associated with higher rates of smoking cessation attempts, successful cessation, and light smoking. Total bans were more
effective than partial bans, in which smoking was restricted to certain places in the
household or certain times. Household restrictions were found to be more strongly
associated with quitting behaviors than workplace restrictions.
Farkas et al. [1999] argue that household bans may be especially effective in
57
reducing smoking behavior, relative to bans in other locations, because the motivation of the smoker’s spouse or children may be for the smoker to quit completely rather than simple compliance with a ban. The ban may be accompanied
by strong social pressure to quit. In addition, they hypothesize that requirements
to change location may disrupt cues that elicit the smoking response, such as finishing a meal, or force the smoker to choose between pleasurable activities, such
as smoking and watching television, or delay smoking for essential activities, such
as getting dressed before going out. In the workplace, smoking can occur fairly
regularly, such as immediately before or after work or during breaks, and there are
likely to be fewer social impediments to smoking.
While the most desired household restriction in terms of reducing occupant
exposure to SHS is a total smoking ban, where no one is allowed to smoke tobacco products indoors, either due to their quitting or because they are forced to
smoke elsewhere, the intermediate solution of a partial ban on smoking in terms
of location within the house or times when smoking is allowed also has the potential for a protective effect on exposure. From many cross-sectional observational
studies of children’s exposure, based on either measures of nicotine, the nicotine
metabolite cotinine, or self-reported exposures, there is evidence that partial bans
on household smoking, e.g., designated smoking areas or times, lead to substantial
reductions in the exposure of young household occupants to SHS. Table 2.9 summarizes the results of seven recent studies on children’s SHS exposure at home,
which are also discussed in more detail below.
Biener et al. [1997] analyze data from a 1993 telephone survey of over 1,600 adolescents and found that households with designated smoking areas and total bans
were significantly associated with reductions of approximately 10 and 30 h per
week, respectively, in mean hours of self-reported exposure. Since exposure was
self-reported, actual exposure may have been biased relative to direct measures of
exposure.
A cross-sectional telephone study presented by Pizacani et al. [2003] found that
bans were strongly associated with awareness of the harm of SHS. Household
urinary cotinine-to-creatintine ratio
self-report
Bakoula et al. [1997]
Biener et al. [1997]
1,600
> 2,000
1,602
314
urinary cotinine
urinay cotinine-to-creatinine ratio
Blackburn et al. [2003]
242
Henschen et al. [1997]
urinary cotinine; nicotine
Berman et al. [2003]
6,000
249
self-report
Pizacani et al. [2003]
Restrictions associated with 10−30 h
of exposure reduction per week.
Exposure reduced by 40% using
room restrictions or enhanced
ventilation.
Exclusive maternal smoking is
associated with higher exposure than
exclusive paternal smoking.
Room restrictions resulted in
reduced exposure, but not as much
as for a complete ban.
Banning smoking led to significant
exposure reduction, but less strict
measures had no effect.
Complete bans and room or time
restrictions resulting in reduced
exposure.
Restrictions were associated with
low levels of exposure.
Sample Results
Wakefield et al. [2000a] randomized trial
Measure
Reference
Table 2.9: Surveys of Household Smoking Restrictions and Corresponding Reduction of Children’s Exposure to
Secondhand Smoke
58
59
smoking restrictions were found by the authors to be associated with low levels
of self-reported SHS exposure, although partial bans were less effective.
Wakefield et al. [2000a] conducted a cross-sectional randomized trial of urinary
cotinine-to-creatinine ratios for 249 asthmatic children aged 1 to 11 who had at
least one parent who smoked. They found that limiting smoking in the home to
rooms where the child rarely visited resulted in reduced SHS exposure compared
to unrestricted smoking. But this strategy, or one where exceptions were made to
a complete ban, resulted in measurably larger exposures than for a complete ban
on household smoking.
Blackburn et al. [2003] conducted a cross-sectional survey across 314 smoking
households of parent’s knowledge and use of SHS harm reduction strategies. Urinary cotinine-to-creatinine ratios in infants were used to study the effect of mitigation strategies. They found that banning smoking in the home achieved a small,
but significant reduction in urinary cotinine-to-creatinine ratios, and less strict exposure measures had no effect when compared to no exposure reduction measures
at all. Less strict exposure reduction measures were defined as restrictions placed
on smoking in the vicinity of the baby or active steps that were taken to ventilate
rooms during or after smoking has taken place, such as opening windows or using fans or ionizers. Multiple measures were undertaken by most of the parents.
The exact distance of smoking from the baby (for measures that included restriction) was not specified. The researchers conclude that “parents would benefit from
more information on what measures actually work.”
Berman et al. [2003] report on SHS exposure, based on surveyed behavior, urinary cotinine, and air nicotine samples for 242 children with asthma who live in
homes with at least one smoker. Forty-two percent of homes allowed smoking only
in certain areas or at certain times, whereas 47% had a complete ban on smoking.
Both of these groups had significantly lower cotinine levels and air nicotine levels than for the group with no restrictions on smoking, with a total ban having
median concentrations that were less than a third of concentrations with a partial ban. They also found that children’s urinary cotinine levels were significantly
60
higher in those homes where the only smoker was the mother, suggesting that maternal smoking is an important factor in children’s exposure, presumably due to
the close proximity between mother and child versus other possible smokers in a
household.
Bakoula et al. [1997] conducted a cross-sectional study of over 2000 children
under age 14 in Greece, where 73% of the children were exposed to SHS from
at least one smoker in the household. In this study, questionnaires on various
exposure-related factors were administered and exposure was estimated using urinary cotinine-to-creatinine ratios for each child. The authors found that SHS exposure can be reduced by about 40% by parents taking simple precautions. Here,
precautions were defined by parents reporting “that they never smoke in the presence of their child, or they smoke only in restricted areas and regularly open windows to freshen the indoor air.” Avoiding smoking in the presence of the child
had a more significant impact than regular ventilation, although data on precise
behaviors associated with different precautions, e.g., duration of window opening, number of windows, room layout, or position of doors, was not determined.
The authors conclude that while smoking cessation is the optimal solution to home
SHS exposure, “simple common-sense measures that can be applied easily” can
reduce exposure until the objective of quitting is achieved. The authors also note
that urinary cotinine concentrations measured on Mondays, which indicate exposure over the weekend, were about 30% higher than on other days, reflecting the
importance of SHS exposure in the home. Every 20 m2 of floor area contributed to
a 9% drop in average exposure and the presence of central heating was associated
with a 14% drop in exposure.
Henschen et al. [1997], based on a longitudinal study of urinary cotinine for
602 elementary school children living in three towns in Germany, reported that
exclusive maternal smoking was associated with higher cotinine levels than exclusive paternal smoking, presumably because children spend more time with their
mother and/or more of the mother’s smoking takes place at home. Also, the size
of dwellings was negatively associated with higher cotinine levels, suggesting SHS
61
is diluted more in larger homes. Families with low education level tend to have
smaller homes on average, so this social group may have an increased burden of
exposure.
The findings in all of the studies reviewed in this section support the expectation that, while partial bans on household smoking can be effective in reducing
SHS exposure, total bans result in the most protective effect. Partial bans may take
the form of designated smoking areas where nonsmokers do not visit regularly or
enhanced ventilation in rooms where smoking takes place. They are likely to be
effective because nonsmoking occupants are less proximate to active smokers in
the house, doors block the passage of smoke to nonsmoking areas of the house, or
nonsmoking occupants are not at home when smoking occurs so that any exposure
is a result of the persistence of smoke after smoking has ended.
2.4.3 Intervention Strategies
The public health community appears quite united behind the need for householdbased SHS exposure interventions for children. Ashley and Ferrence [1998] discuss
how SHS control in homes must be a public health priority, since the home is the
most important site for exposure and children are particularly susceptible. Given
the extent of this public health problem, Gehrman and Hovell [2003] conclude that
“effective interventions must be developed immediately.”
However, while educational interventions can be implemented in the context
of fairly routine professional health care, governmental regulation of the home is
more problematic. Ashley and Ferrence note that the issues involved are diverse
and complicated, with social, economic, legal, and political factors resulting in
a lower level of support than for the workplace and public locations.5 One factor
contributing to lower support for home control measures that they cite is that some
believe government and external agents should not interfere with behavior in private settings. This belief contrasts with laws and regulations protecting children
5A
prime example of governmental regulation of smoking behavior is California’s Assembly
Bill 13, which when fully enacted in 1998 effectively banned smoking in all publicly accessible
restaurants and bars.
62
from abuse or requiring infant restraint in vehicles. Gehrman and Hovell point out
that:
Protecting children from ETS exposure in home environments has
many social and political implications, in that it is difficult to monitor and regulate behavior in private residential settings. The sancity
of the home is a widely held tenet in US society. Therefore, protecting
children from home exposure is a complex and sensitive issue.
While in the previous section I was concerned with evidence for the effectiveness of various approaches for reducing SHS exposure, particularly for children,
here I address ways that households may be drawn into implementing exposure
reduction measures through interventions by health clinicians. Interventions involve counselling of family members, especially smokers who are caretakers of
children, to effect behavior modification. Although the success of educational interventions is judged in large part on whether or not the targets of the intervention
respond by changing their long-term habits, the ultimate effectiveness of the intervention also relies on the effectiveness of the exposure reduction measures themselves, as discussed in the preceding section. The best way to evaluate the success
of an intervention is by conducting a randomized controlled trial in which members of a case group are targeted with the full intervention and a group of controls
with otherwise similar characteristics are not.
Three recent interventions for reducing children’s SHS exposure, which have
been published in the medical literature, are summarized in Table 2.10. These
studies are fairly representive of the different kinds of intervention, covering a
range of education intensity from the simple distribution of booklets accompanied
by follow-up telephone calls [Wakefield et al., 2002], to several nurse-led sessions
[Wilson et al., 2001], and finally to multiple intensive counseling sessions [Hovell
et al., 2000a]. Each of the studies used a controlled trial design.
Wakefield et al. [2002] conducted a controlled trial involving written and verbal
feedback to parents about their 1−11 year old child’s cotinine-to-creatine levels, as
well as information booklets, and phone calls encouraging a ban on smoking at
home. They found that minimal interventions, such as those in their study, are un-
63
Table 2.10: Three Recent Controlled Trial Intervention Studies for the Reduction of
Children’s Exposure to SHS Based on Household Smoking Restrictions
Reference
Method
Results
Wakefield et al. [2002] information booklets;
telephone calls; feedback
on urinary
cotinine-to-creatinine ratio
no greater change in bans
on smoking in homes or
habits to reduce children’s
exposure in intervention
group relative to controls
Wilson et al. [2001]
three nurse-led sessions;
asthma and exposure
education; feedback on
urinary cotinine
significantly lower odds
of having more than one
acute medical visit for
asthma in the educated
groups than in controls
Hovell et al. [2000a]
seven formal counseling
sessions; mother reports
of cigarettes smoked in
same room as child;
urinary cotinine
mother’s reports of
exposure and cotinine
decreased significantly
more in the counseled
group than in controls
likely to bring about significant reduction in exposure of children, recommending
more structured advice and support over a longer duration.
Wilson et al. [2001] conducted a controlled trial to determine the effect of interventions on asthma-related health care utilization for children aged 3−12. Their
intervention methods were more intense than those of Wakefield et al., involving
three nurse-led sessions on behavior-changing strategies, education about asthma,
and feedback on urinary cotinine levels. They found that their intervention resulted in reduced asthma health-care utilization. However, the effect of the intervention on actually reducing exposure, as measured by cotinine, was relatively
inconclusive. The intervention appears to have involved only vague discussion of
possible reductions in children’s exposure, such as identifying how and where in
the home the child is being exposed.
Hovell et al. [2000a] confirmed, through the use of a randomized, double-blind
trial, the efficacy of intensively and systematically counselling mothers with chil-
64
dren under age 4 on reducing their children’s exposure to SHS over a total of seven
sessions. As compared to controls, who received cursory advice on reducing exposure, mothers in the counselled group reported a significantly greater reduction
in the number of cigarettes the mother smoked in the same room as the child and
urinary cotinine levels decreased in the children of counselled mothers, whereas it
increased for the control group, which might be due to increased smoking activity. In this study, quitting smoking was not a required exposure reduction method
and counselled mothers did not quit at a larger rate than for the control group.
While parents could have smoked in a different room in response to counselling,
they could still have been close enough for the child to inhale smoke. The lack of
detailed information on reduction strategies and direct objective measures of exposure for this study, besides a general effort to smoke in a different room, prevents
quantitative evaluation of the effectiveness of specific types of reduction strategies.
Hovell et al. [2000b] reviewed eight studies on reducing children’s residential exposure to SHS. The authors place interventions for exposure reduction in
a behavioral-ecological context that includes “cultural contingencies of reinforcement.” Behavioral components, such as smoking outside, smoking away from the
child, smoking fewer cigarettes, or quitting smoking, are first identified as critical
to SHS exposure. Next, levels of environmental control are recognized, including how the biological addictive drive and the tobacco industry serve to sustain
or increase current smoking rates, timing, and context, with social pressure from
family and friends pushing for change. In addition, outside forces may intervene,
introducing additional influence designed to promote changes in smoking behavior through sustained reinforcement. These forces may take the form of clinicians,
media, broader society, and the legal system.
Hovell et al. found that no legal policies have currently been enacted to restrict children’s SHS exposure in homes, and although divorce and adoption courts
limit custody to protect children from SHS, there are no available studies. While
one-time clinical services appeared to be ineffective, repeated session counselling
showed promise, although few controlled studies have been performed. Coun-
65
selling has been targeted at parents and includes information on the health effects
of exposure to SHS and means for reducing SHS exposure.
Gehrman and Hovell [2003] conducted a more recent and complete critical review, evaluating nineteen SHS household intervention studies for children and
youth conducted between 1987 and 2002. These studies consisted mostly of controlled trials with parent reports as the measure of exposure. Their findings were
similiar those of Hovell et al. [2000b] in that intensive home-based interventions involving extended and close contact could be effective in general, and those based
on an explicit behavior-modification theoretical framework were more successful
than physician-based interventions or those that simply consisted of advice on the
harmful effects of SHS without providing strategies and skills for reducing exposure. Gehrman and Hovell note that smoking-related behaviors are difficult to
modify and researchers in the behavior sciences generally accept that simple acquisition of knowledge is usually insufficient to bring about changes.
Gehrman and Hovell make several specific recommendations for future intervention studies. Interventions should include a stepped-care approach consisting
of initial advice from a physician, printed materials, cotinine feedback, and leading to repeated and longer contact via phone calls and home visits with trained
counselors. The initial components must include basic education regarding the
harmful effects of SHS exposure. Their recommended approach is to involve parents with the skill training, problem solving, and goal-setting associated with successive strategies for reducing SHS exposure, using self-reinforcement to achieve
a final goal, which may be to smoke exclusively outdoors. This approach provides
a structure to make parents intimately involved with controlling their children’s
exposure, helping them to stay motivated and focused. Finally, they suggest targeting interventions towards children themselves, instead of exclusively to their
parents or care-givers.
Green et al. [2003] also reviewed a number of studies on attitudes and behaviors towards SHS, children’s health, and the indoor environment. They discuss
research needs and ways to “encourage reductions in domestic SHS levels.” They
66
conclude, similarly to the other literature reviews presented above, that traditional
health promotion campaigns have limited success in encouraging risk reduction
meaures in homes. While many nonsmokers and smokers are concerned about
children and secondhand smoke, many people continue to allow children to be
exposed in the home.
2.4.4 The Need for Better Exposure Measures
There is a general need for the refinement of SHS measures in public health studies to improve evaluations of the efficacy of interventions and to attain a better
understanding of the impact of specific exposure reduction strategies. Gehrman
and Hovell [2003] find that a major difficulty in analyzing interventions studies
was the lack of a standardized definition of SHS exposure, which many times was
given as “cigarettes smoked around the child.” In studies of residential SHS exposure mitigation strategies, too little detail has been gathered on the precise method
of exposure reduction. Data on room size, ventilation systems, and door positions
would be helpful in determining the degree of exposure reduction.
The use of cotinine and nicotine in public health studies clouds the findings of
intervention studies, and the general studies of household restrictions described
in Section 2.4.2, because nicotine rapidly sorbs to household surfaces, reducing its
transport into different rooms of the house from a given smoking room, whereas
other SHS constituents, such as particles and less reactive gases, may permeate
a house more uniformly. On the other hand, in houses with chronic smoking,
there may also be a persistent background level of nicotine and other sorbing SHS
species.
In addition, the widespread use of the biomarker cotinine, a nicotine metabolite, in health studies is hampered by variation among individuals with respect to
nicotine uptake, distribution, metabolism, and excretion. Since they are not direct
measures of SHS exposure, cotinine biomarker data, by themselves, do not contain
time or location-specific information on exposure, smoking activity, or other conditions and activities surrounding exposure. Circumstances under which exposure
67
occurs may be critical to efforts directed at behavior modification or estimation of
risk. Matt et al. [1999] and Hovell et al. [2000c] conclude that biological measures
based on cotinine are best combined with environmental, observation, and/or selfreport techniques.
Hovell et al. also recommend more research and development of real-time, continuous indicators of exposure, i.e., direct measurements of exposure (see Section
2.1.2 on page 29), along with the location and activities surrounding that exposure.
Real-time measures could be used to provide immediate feedback, and thereby facilitate intervention efforts, whereas other measures are either intermittent or their
analysis is delayed. Hovell et al. state that
The failure to obtain real-time and continuous (or close approximation)
measures of ETS exposure limits our understanding of true exposure
levels, cumulative exposure levels, and variability over time (that is,
exposure profile). . . . Without more information about variability over
time and associated events that might account for such variability, efforts to control ETS exposure will be compromised.
2.5 Models of IAQ and Exposure
From the studies reviewed in this chapter, it is apparent that there exists a need
in tobacco-related health research for an enhanced understanding of what residential SHS exposure concentrations can occur across a variety of different situations.
Field exposure measurements, whether part of health studies or stand-alone exposure monitoring surveys, establish basic relationships, challenging or confirming
hypotheses. But any given survey can only address a limited domain of possible exposures and scenarios. Exposure models, on the other hand, are uniquely
capable of generalizing experimental findings to make predictions across a very
wide range of arbitrary conditions. They serve as a tool with which to make sense
of observations and to identify areas where further experimental investigation is
warranted.
As in the case for the model of residential SHS exposure that I develop in the
current research, the central component of quantitative inhalation exposure mod-
68
els is typically an indoor air quality (IAQ) model. Such models encapsulate the
mechanisms by which tobacco smoke emissions lead to residential SHS concentrations, generalizing knowledge on the physical and chemical behavior of pollutants
in buildings to arbitrary environmental conditions. They take information on indoor emission patterns and the physical characteristics of a building, such as air
flow rates, zone volumes, and loss rates, and produce information on the time
evolution of airborne pollutant concentrations, as well as time-averaged concentrations.
2.5.1 IAQ Model Validity
For this dissertation, I am interested in the accuracy of multizone IAQ models used
in the assessment of residential exposure to particles or gases in SHS. While the
multi-compartment character of residences has been established empirically (see
above), it does not follow immediately that theory can accurately predict multizone concentrations over arbitrary time scales. Several published studies, some of
which are listed in Table 2.11, have evaluated the performance of multizone models in residential locations or explored the issues of mixing and source-receptor
proximity, which are typically neglected in model formulations.
Although some recent efforts, such as those by Ribot et al. [2002], involve modeling the distribution of pollutants within each room of a house using computational fluid dynamics (CFD), the central assumption of many zonal indoor air
quality models is that of uniform mixing of pollutants in individual rooms. Under
this assumption, any emitted pollutant is instantaneously mixed throughout the
zone of release. The implications of this assumption are that concentrations in a
particular room are the same everywhere for all times. In reality, it takes a finite
amount of time for emissions to mix within a room so that the average exposure
one receives while immediately next to an active pollutant source may be larger
than the average exposure at a more distant location, such as on the other side of
the room.
The increase in exposure that occurs when one is in close proximity to an in-
Real-time particle and CO monitoring in
a tavern and house
Real-time tracer and particle monitoring
in a two-room test facility
Tracer gas
Tracer gas
Tracer gas
Cigarettes and tracer gas
Tracer gas and incense stick
Cigarettes
Baughman et al. [1994]
Drescher et al. [1995]
Furtaw et al. [1996]
Mage and Ott [1996]
Miller et al. [1997]; Miller and
Nazaroff [2001]
Klepeis [1999]
McBride et al. [1999]
Ott et al. [2003]
a house
Real-time CO and particle monitoring
a house, tavern, and smoking lounge
Real-time SF6 monitoring in a chamber
Real-time monitoring of CO at 9 points
in a chamber
Grab sampling of SF6 at 41 points in a
chamber
Good agreement between measurements and two-zone
model parameterized from same experiment; error
surface shows relative insensitivity to flow parameters
Proximity to active particle sources of 1 m resulting in
mean concentrations averaging 3 times higher than
those at a fixed distant location. Proximate CO
concentration were also much higher than distant ones
during source-on periods.
Mixing of air pollutant in medium to large rooms is
fairly rapid in real locations under typical conditions
on the order of 12−15 min before average
concentrations at separated points are within 10% of
the room mean
Good agreement between two-zone model and
measurements
Use of a uniformly mixed assumption to determined
average exposures is generally valid for an intermittent
source if the source-off well-mixed time period is large
compared to the source-on plus mixing time periods
Average concentration a distance of 0.4 m from the
source was double the theoretical well-mixed
concentration for typical flow rates
Mixing times range from 2 to 15 min for forced
convection
Mixing times range from 7 to 15 min under natural
convection in which heat was added from solar
radiation or an electrical heater
Multizone model does a good job of predicting indoor
pollutant concentrations
VOC and particle samples in a house
Moth cakes, kerosene heater,
dry cleaned clothes, aerosol
spray, applied wet products
Sparks et al. [1991]
Good agreement between measured and modeled CO
Conclusions/Results
Real-time CO monitoring in a house
Tracer gas
De Gids and Phaff [1988]
Method
Source
Study
Table 2.11: Studies Evaluating Models of Residential Multizone Transport of Indoor Air Pollutants, Single-Zone
Mixing, and Source-Proximity Effects
69
70
door pollutant source has been well established. Rodes et al. [1991] review studies on the influence that a personal activity cloud has on exposure to gases and
aerosols. They find that the ratio of personal exposure measurements to microenvironmental (fixed) exposure measurements is typically 3−10 for occupational settings and 1.2−3.3 for residential settings. The elevated personal measurements
are attributed to proximity between pollution sources and receptors. McBride
et al. [1999] and Furtaw et al. [1996] conducted separate controlled experiments on
source-receptor proximity using either tracer gases or particle releases, reporting
that residential proximity effects can lead to average personal (proximate) to microenvironmental (distant) ratios of 2−3 for source-receptor distances of 0.4−1 m.
Time spent by an exposure receptor in close proximity to a source has the potential to create negative deviations of modeled concentrations from actual ones. If
the receptor spends a long time in the room after the source stops and the time the
pollutant takes to mix is short in comparison to that time, or the pollutant removal
time, then the receptor’s average exposure may not be much different from the theoretical well-mixed case. The discrepancy tends to decrease with added distance
between the exposure receptor and the source, although the exact relationship between concentration and source-receptor distance depends on the room’s air flow
patterns. So, errors in model predictions are dependent on the distance from the
source, the duration of the source, the time spent in the source room, the pollutant
removal rate, and the time it takes for pollutant to become uniformly mixed in the
room (the mixing time). The last of these factors, the time-to-mixing, is a function
of physical conditions, such as bulk air flow rates and room heat flows, which may
or may not be influenced by human activity, whereas the others are dependent
largely or solely on human behavior.
While not instantaneous, the mixing time in a room for pollutants emitted from
short sources under typical residential conditions is rapid, occurring within a 15
min period for forced convection or natural convection with sufficient energy content of the air [Baughman et al., 1994; Drescher et al., 1995]. Based on monitored
locations in a home, tavern, and smoking lounges where cigarette or cigar sources
71
of short duration were used, Klepeis [1999] found that averaging time periods of
12−15 min were sufficent to achieve average concentrations at different points in
a room that were within 10% of the overall room average. For continuous sources,
averaging times of up to several hours or longer are required.
While mixing and proximity are critical issues in the accuracy of indoor air
models, they may or may not result in unacceptably large errors in predicted concentrations. For example, if receptor locations are reasonably distant from active
sources, mixing rates are rapid, on the order of 10 min or less, and the time scale of
interest for the dynamics of the air pollutant system are on the order of 10 min or
greater, then we would expect that measurements of average concentration would
agree well with the predictions of single and multizone models. As it turns out,
several investigators have indeed discovered that measurements of particles and
gases can be well described by models that assume well-mixed rooms.
Most recently, Ott et al. [2003] parameterized a two-zone indoor air model with
real-time measurements of CO emitted from a cigar and found that the model
provided an excellent fit to measurements during both the peak and decay periods. However, these results may be slightly misleading because the parameterized model is not applied to an independent experiment. The authors reported
that slight changes in flow parameter values did not change the quality of the fit
appreciably. The experimental data and estimated flow rates for this study are
presented in Chapter 5 of this dissertation.
Miller et al. [1997] also parameterize a two-zone model using controlled tracer
gas releases, finding good agreement between prediction and experiment. Flow
rates determined from their experiments are also presented in Chapter 5. Based on
a concurrent set of experiments, Miller and Nazaroff [2001] find that size-specific
predictions of an aerosol model can accurately predict continuously measured particle concentrations in a two-room test house environment when smoking occurs
in one of the rooms.
De Gids and Phaff [1988] found very good agreement between measured and
modeled concentrations of CO tracer gas in a dwelling. During their experiments,
72
doors were opened and closed and air was mechanically recirculated or fresh outdoor air was introduced.
Another application of an IAQ model to residential air pollutant concentrations
is that by Sparks et al. [1991], who conducted a variety of experiments in a test
house using continuous sources of particles and gases from household products
including moth cakes, kerosene heaters, and wet paint. They found that concentration measurements taken in multiple rooms over time scales of hours or days
were well predicted by a model. While their experiments are not directly relevant
to SHS exposures, which occur on time scales of minutes to hours, they contribute
towards a general confidence that multizone indoor air models can provide accurate results.
2.5.2 Exposure Simulation
Data from empirical studies can establish broad trends in exposure magnitudes
across a number of homes or intensively characterize the dynamic behavior of pollutants in controlled settings, and IAQ models can generalize these data to other
situations. But data on human behavior and human-environment interactions are
needed for the realistic characterization of exposures and the extrapolation of concentration data to complex social ecologies, such as multi-person households.
In general, exposure simulation models incorporate the precise timing and context of source and receptor activity, providing a cogent, coherent, and compact
framework for use in pinpointing factors critical to understanding and predicting
levels of exposure. In representing the theoretical foundation of exposure science,
they serve to (1) consolidate experimental findings and extrapolate them to new
situations, (2) identify uncertain areas for further study, and (3) provide a platform
for generating new hypotheses. Fundamentally, exposure models simply match
environmental pollutant concentrations occurring along axes of time and space,
which may be available in raw form or determined from component models, with
the presence of one or more persons. Complexity arises when physical and environmental parameters and detailed profiles of the behavior of sources and recep-
73
tors, as well as interactions between the two, are incorporated into the model.
For example, for the case of indoor inhalation exposures in a multi-room context, a multizone IAQ model, such as those discussed in the previous section of
this chapter, is typically a key component of the overall simulation model. An
indoor air model usually has many parameters relating to emissions, transport,
and removal of pollutants. Using activity and location patterns for the source(s)
as input, the IAQ model provides the core engine for generating room concentration profiles. These concentrations are then overlaid with receptor movement patterns to reveal patterns of exposure. In the case of human-generated emissions, as
with SHS, both sources and receptors may follow complicated trajectories between
rooms where they may alter the configurations of doors, windows, air handling
systems, or air cleaning devices. This approach to modeling exposure is the one
followed in this dissertation to simulate and explore residential SHS exposures.
Exposure models currently under development can be crudely divided into
two camps, according to their primary intended purpose. The first camp, which
I will refer to as “exploratory”, is comprised of models that are experimental and
limited in scope, focusing on a particular domain of exposure scenarios. Their purpose is primarily scientific, developing methods or approaches, establishing mechanisms of exposures, empirically testing model assumptions, or exploring model
predictions in a formal sensitivity and/or uncertainty analysis. For this camp,
the prediction of exposures for arbitrary populations is less a priority than understanding how exposure occurs in limited settings. The current work falls into this
first camp. The second camp of models has a much broader scope and is intended
to support regulatory mandates, such as in the estimation of population risk assessments. Those who use these models generally are interested in applying them
to large groups of people, and therefore may incorporate sophisticated sampling
techniques, e.g., Monte Carlo or Latin Hypercube sampling, and stratification of
model inputs and outputs according to geographic or demographic characteristics. They likely describe multiple sources of pollution and a range of different
settings where exposure can occur.
74
In Table 2.12, I list a number of existing inhalation (air) exposure models, categorizing them by their general status in either the “exploratory” or “regulatory”
camps. Seigneur et al. [2002] and Price et al. [2003] both present fairly in-depth descriptions of many of these models as well as others. The entries in the table reflect
either significant efforts by a regulatory agency or efforts that have an associated
article in the refereed scientific literature, or both. As a whole, they are reasonably
reflective of the current state of inhalation exposure modeling, including efforts by
government, acedemia, and industry. Most of the models listed models are or have
been made available in distributable (executable) form.
In recent years, there has been an emphasis on the second, regulatory type of
model with the US Environmental Protection Agency (USEPA) investing considerable resources in its NEM, HAPEM, SHEDS, and APEX series of models [McCurdy, 1995; Rosenbaum, 2002; Burke et al., 2001; Richmond et al., 2002]. Because
inhalation is likely the most important exposure route for many toxic chemicals
and is mechanistically one of the simplest routes of exposure, and since extensive
air quality regulations are already concerned with air quality, these regulatory inhalation exposure models are among the most well-developed. They tend to be
statistically based, sampling from empirical or parameterized distributions of observed air concentrations and aggregate times spent in broad location categories
(e.g., home, outdoors, or automobile).
There exists a massive database of ambient air quality data to support regulatory and other predictive population exposure models, as mandated under regulations such as the US Clean Air Act. There is also a growing data base of personal
inhalation exposure monitoring data from studies such as EXPOLIS, NHEXAS,
TEAM, PTEAM [Koistinen et al., 2001; Sexton et al., 1995; Pellizzari et al., 1995;
Wallace, 1987; Özkaynak et al., 1996], and others (Table 2.5), and a large database
of microenvironmental model inputs to support the scope of regulatory modeling
efforts. The American Chemistry Council has funded two recent in-depth reviews
of data sets and reports having relevance to exposure modeling [Koontz and Cox,
2002; Boyce and Garry, 2002]. The USEPA Exposure Factors Handbook and Expo-
−
Richmond et al. [2002]
Briggs et al. [2003]
−
Expl
Reg
Reg
Expl
Reg
Reg
Reg
Reg
Reg
Expl
Expl
Expl
Expl
Expl
Expl
Classa
Northampton
EXPOLIS
USEPA
NIST
USEPA
USEPA
CARB
Harvard
USEPA
Carnegie-Mellon
−
USEPA
LBNL
LLNL
EPA
Developed Byb
Full Name or Description
Residential radon exposure model
European population particle exposure model
Air Pollutants Exposure Model and Total Risk Integrated Methodology
Exposure Event Module; criteria and hazardous air pollutants
Multizone simulation of air flows, contaminant concentrations, and
personal exposure.
Hazardous Air Pollutant Exposure Model; mobile source air toxics
Stochastic Human Exposure Dose Simulation – Particulate Matter
California Population Indoor Exposure Model
Benzene Exposure and Absorbed Dose Simulation
(Probabilistic) National Exposure Model; criteria pollutants
Model for Analysis of Volatiles and Residential Indoor Air Quality /
Total Exposure Model
Multichamber Chemical Exposure Model
Descendant of EXPOSURE and INDOOR models; simulates multizone
indoor air concentrations, individual exposure, and risk
A “macromodel” for indoor exposure to combustion products
Residential inhalation exposure model for volatile compounds in tap
water
Simulation of Human Activity Patterns and Exposure
Regulatory models used for development or enforcement of government regulations or for related risk assessments. These models are typically applied to large
populations and require extensive data inputs that are representative of the population being modeled; Expl: Exploratory models used for intensive scientific study of
particular exposure scenarios. These models typically treat an individual or narrowly defined cohort of people and have facilities for a detailed treatment of residences or
some other specific microenvironment.
b EPA: US Environmental Protection Agency, Washington, D.C. USA; NIST: National Institute of Standards and Technology, Gaithersburg, MD USA; CARB: California Air
Resources Board, Sacramento, CA USA; EXPOLIS: European Exposure Assessment Project; LLNL: Lawrence Livermore National Laboratory, Livermore, CA USA; LBNL:
Lawrence Berkeley National Laboratory, Berkeley, CA USA; Northampton: Contributed by academic researchers in Northampton, UK.
Information and downloads for the APEX, TRIM, HAPEM, and HEM regulatory models for criteria pollutants and air toxics can be accessed from the EPA website at the
following URL: http://www.epa.gov/ttn/fera/
a Regl:
EXPOLIS
APEX/TRIM Expo
Dols and Walton [2002]
Kruize et al. [2003]; Hänninen et al. [2003]
HAPEM
CONTAM
Rosenbaum [2002]
SHEDS-PM
CPIEM
Burke et al. [2001]
BEADS
Koontz et al. [1998]; Koontz and Niang
[1998]; Rosenbaum et al. [2002]
NEM-pNEM
MAVRIQ/TEM
MCCEM
Macintosh et al. [1995]
McCurdy [1995]
Wilkes et al. [1992, 1996, 2002]
Koontz and Nagda [1991]
RISK
Traynor et al. [1989]
Sparks [1988, 1991]; Sparks et al. [1993]
−
SHAPE
Ott et al. [1988]; Ott [1984]
McKone [1987]
Acronym
Reference
Table 2.12: Examples of Some Existing Regulatory and Exploratory Inhalation Exposure Models
75
76
sure Factors Handbook for Children are fairly comprehensive resources of appropriate inputs for predictive models [USEPA, 1997, 2002].
The original research I undertake in this dissertation is exploratory in nature,
focusing exclusively on SHS exposure in a multizonal residential context. Rather
than predicting exposures for an actual population of people using complex or
highly variable data inputs, I seek to investigate exposure relationships for a relatively small range of parameter values. This work has, as its earliest apparent precedents, models by Sparks et al. [1993], Sparks [1991], Koontz and Nagda
[1991], and Wilkes et al. [1992], which track the behavior of household occupants
and follow pollutant concentrations between rooms, incorporating more detailed
physical mechanisms of emissions, pollutant dynamics, and exposure than the
more statistical approach of regulatory air exposure models.
2.6 Summary and Conclusions
Secondhand smoke (SHS), also known as environmental tobacco smoke (ETS), is
a mixture of many gaseous and particulate components that has been associated
with adverse health in children and adults. Residential SHS exposure is of key importance because of the large proportion of time that is spent by people at home,
particularly children who may require close care by smoking adults. Field studies
provide evidence that SHS can contribute as much as 30 µ g m−3 or more to indoor particle concentrations, and therefore significantly influence exposures, and
that the multi-compartment character of homes, and especially the open or closed
states of interior doors, can lead to significantly different pollutant concentrations
between rooms.
It appears that between 25% and 50% of children in the US are at risk of some
exposure to SHS in their home. The ecology in these smoking households, and
particularly the prospects for reducing or eliminating SHS exposure, are socially
complex. Different factors for reducing exposure may occur simultaneously to different degrees and have common or interacting driving forces. In this web of interaction (Figure 2.3), both the greater society and health care professionals provide
77
incentives to limit smoking behavior. The act or process of instituting household
smoking restrictions may create inconvenience or alienation for smokers, putting
pressure on them to reduce the number of cigarettes they smoke or to quit completely.
A major challenge for the public health community is the development of effective intervention strategies that encourage lasting behavior modification and
result in the verifiable reduction or elimination of household SHS exposure. Interventions, and studies of intervention efficacy, would benefit from more accurate
and precise measures and understanding of exposure, potentially resulting from
the use of real-time exposure monitoring equipment and more detailed diaries of
in-home behavior. Currently, there is a lack of understanding as to precisely how
well different exposure reduction strategies perform.
Predictive models can play an important role in improving our understanding
of residential SHS exposure. Indoor air quality models, which take a variety of
physical parameters as input and have been shown to predict observed concentrations with good accuracy, can be used to generalize measured indoor air pollutant
concentrations to arbitrary houses under arbitrary conditions. Exposure simulation models fuse observed or modeled air pollution concentrations with human
activity patterns, generating exposure profiles for individuals or populations.
In the current work, I make use of established, and proven, techniques in modeling indoor air quality and exposure to shed light on the effectiveness of specific
residential SHS exposure mitigation strategies. This new information is expected
to inform public health researchers and practitioners in their efforts to reduce SHS
exposure.
2.7 References
Ashley, M. J. and Ferrence, R. (1998). Reducing children’s exposure to environmental tobacco smoke in homes: Issues and strategies. Tobacco Control, 7(1): 61–65.
Bakoula, C. G., Kafritsa, Y. J., Kavadias, G. D., Haley, N. J., and Matsaniotis, N. S.
(1997). Factors modifying exposure to environmental tobacco smoke in children
(Athens, Greece). Cancer Causes & Control, 8(1): 73–76.
78
An Eco−Social Model for Reduction or Elimination of
Household Exposure to Secondhand Tobacco Smoke
Society
Smoking
Behavior
Industry −
Advertising
Uptake
of Smoking
The Media −
Entertainment
Biological
Addiction
Social
Pressure
Light
Smoking
Quitting
Smoking
Health
Intervention
Exposure
Reduction
Govt.−
Health Policy−
Outreach
Exposure
Elimination
Health
Care
Decreased
Exposure
Time/Space
Restrictions
Ventilation
Measures
Isolation
Measures
Filtration/
Removal
Measures
Knowledge
of Adverse
Health
Effects
Knowledge
of Exposure
Reduction
Strategies
Household
Restrictions
Figure 2.3: A web of inter-relating factors associated with eventual reduction or
elimination of household secondhand smoke exposure. Elements in society encourage smoking behavior, which can be moderated directly or indirectly by health
care leading to decreased exposure. Society and health care can also reduce exposure by influencing the adoption of household restrictions, which can in turn
serve to either directly reduce exposures or do so indirectly by coercing a reduction in smoking behavior. One focus of original research in this dissertation is the
effectiveness of specific household restrictive measures in reducing, but not eliminating, exposure to secondhand smoke through the blocking of smoke movement
across doorways of a house and the enhancement of smoke removal by the opening of windows or use of active filtration.
79
Baughman, A. V., Gadgil, A. J., and Nazaroff, W. W. (1994). Mixing of a point source
pollutant by natural convection flow within a room. Indoor Air, 4(2): 114–122.
Benner, C. L., Bayona, J. M., Caka, F. M., Tang, H., Lewis, L., Crawford, J., Lamb,
J. D., Lee, M. L., Hansen, L. D., and Eatough, D. J. (1989). Chemical composition
of environmental tobacco smoke. 2. Particulate-phase compounds. Environmental Science and Technology, 23(6): 688–699.
Benowitz, N. L. (1996). Cotinine as a biomarker of environmental tobacco smoke
exposure. Epidemiologic Reviews, 18(2): 188–204.
Benowitz, N. L. (1999). Biomarkers of environmental tobacco smoke exposure.
Environmental Health Perspectives, 107(Suppl 2): 349–355.
Berman, B. A., Wong, G. C., Bastani, R., Hoang, T., Jones, C., Goldstein, D. R.,
Bernert, J. T., Hammond, K. S., Tashkin, D., and Lewis, M. A. (2003). Household smoking behavior and ETS exposure among children with asthma in lowincome, minority households. Addictive Behaviors, 28(1): 111–128.
Biener, L., Cullen, D., Di, Z. X., and Hammond, S. K. (1997). Household smoking
restrictions and adolescents’ exposure to environmental tobacco smoke. Preventive Medicine, 26(3): 358–363.
Blackburn, C., Spencer, N., Bonas, S., Coe, C., Dolan, A., and Moy, R. (2003). Effect
of strategies to reduce exposure of infants to environmental tobacco smoke in
the home: Cross sectional survey. British Medical Journal, 327(7409): 257–260.
Borland, R., Mullins, R., Trotter, L., and White, V. (1999). Trends in environmental
tobacco smoke restrictions in the home in Victoria, Australia. Tobacco Control,
8(3): 266–271.
Boyce, C. P. and Garry, M. R. (2002). Review of information resources to support
human exposure assessment models. Human and Ecological Risk Assessment, 8(6):
1445 – 1487.
Brauer, M., Hirtle, R. D., Hall, A. C., and Yip, T. R. (1999). Monitoring personal
fine particle exposure with a particle counter. Journal of Exposure Analysis and
Environmental Epidemiology, 9(3): 228–236.
Briggs, D. J., Denman, A. R., Gulliver, J., Marley, R. F., Kennedy, C. A., Philips,
P. S., Field, K., and Crockett, R. M. (2003). Time activity modelling of domestic
exposures to radon. Journal of Environmental Management, 67(2): 107–120.
Burke, J. M., Zufall, M. J., and Özkaynak, H. (2001). A population exposure model
for particulate matter: Case study results for PM2.5 in Philadelphia, PA. Journal
of Exposure Analysis and Environmental Epidemiology, 11(6): 470–489.
80
Chapman, S. (2003). Other people’s smoke: What’s in a name? Tobacco Control,
12(2): 113–114.
Daisey, J. M., Mahanama, K. R. R., and Hodgson, A. T. (1998). Toxic volatile organic
compounds in simulated environmental tobacco smoke. Journal of Exposure Analysis and Environmental Epidemiology, 8(3): 313–334.
De Gids, W. F. and Phaff, J. C. (1988). Recirculation of air in dwellings: Differences
in concentrations between rooms in dwellings due to the ventilation system,
Paper 17. In Proceedings from the 9th AIVC Conference, Gent, Belgium, pages 301–
310.
Dols, W. and Walton, G. N. (2002). CONTAMW 2.0 User’s Manual: Multizone Airflow and Contaminant Transport Analysis Software. National Institute of Standards,
Building and Fire Research Laboratory.
Drescher, A. C., Lobascio, C., Gadgil, A. J., and Nazaroff, W. W. (1995). Mixing of
a point-source indoor pollutant by forced-convection. Indoor Air, 5(3): 204–214.
Duan, N. (1982). Microenvironmental types: A model for human exposure to air
pollution. Environment International, 8: 305–309.
Eatough, D. J., Benner, C. L., Bayona, J. M., Richards, G., Lamb, J. D., Lee, M. L.,
Lewis, E. A., and Hansen, L. D. (1989). Chemical composition of environmental
tobacco smoke. 1. Gas-phase acids and bases. Environmental Science and Technology, 23(6): 679–687.
Farkas, A. J., Gilpin, E. A., Distefan, J. M., and Pierce, J. P. (1999). The effects of
household and workplace smoking restrictions on quitting behaviours. Tobacco
Control, 8(3): 261–265.
Fugas, M. (1975). Assessment of Total Exposure to Air Pollution. In Proceedings of
the International Conference on Environmental Sensing and Assessment, Volume 2 of
IEEE #75-CH1004-1 ICESA, Las Vegas, NV. Paper No. 38-5.
Furtaw, E. J., Pandian, M. D., Nelson, D. R., and Behar, J. V. (1996). Modeling indoor
air concentrations near emission sources in imperfectly mixed rooms. Journal of
the Air and Waste Management Association, 46(9): 861–868.
Gehrman, C. A. and Hovell, M. F. (2003). Protecting children from environmental
tobacco smoke (ETS) exposure: A critical review. Nicotine & Tobacco Research,
5(3): 289–301.
Gilpin, E. A., White, M. M., Farkas, A. J., and Pierce, J. P. (1999). Home smoking restrictions: Which smokers have them and how they are associated with smoking
behavior? Nicotine & Tobacco Research, 1(2): 153–162.
81
Green, E., Courage, C., and Rushton, L. (2003). Reducing domestic exposure to
environmental tobacco smoke: A review of attitudes and behaviours. Journal of
the Royal Society for the Promotion of Health, 123(1): 46–51.
Hammond, S. K., Leaderer, B. P., Roche, A. C., and Schenker, M. (1987). Collection
and analysis of nicotine as a marker for environmental tobacco smoke. Atmospheric Environment, 21(2): 457–462.
Hänninen, O., Kruize, H., Lebret, E., and Jantunen, M. (2003). EXPOLIS simulation model: PM2.5 application and comparison with measurements in Helsinki.
Journal of Exposure Analysis and Environmental Epidemiology, 13(1): 74 – 85.
Heavner, D. L., Morgan, W. T., and Ogden, M. W. (1996). Determination of volatile
organic compounds and respirable suspended particulate matter in New Jersey
and Pennsylvania homes and workplaces. Environment International, 22: 159–
183.
Henschen, M., Frischer, T., Pracht, T., Spiekerkötter, E., Karmaus, W., Meinert, R.,
Lehnert, W., Wehrle, E., and Kuehr, J. (1997). The internal dose of passive smoking at home depends on the size of the dwelling. Environmental Research, 72:
65–71.
Hildemann, L. M., Markowski, G. R., and Cass, G. R. (1991). Chemical composition
of emissions from urban sources of fine organic aerosol. Environmental Science &
Technology, 25(4): 744–759.
Hovell, M. F., Zakarian, J. M., Matt, G. E., Hofstetter, C. R., Bernert, J. T., and Pirkle,
J. (2000a). Effect of counselling mothers on their children’s exposure to environmental tobacco smoke: Randomised controlled trial. British Medical Journal,
321(7257): 337–342.
Hovell, M. F., Zakarian, J. M., Wahlgren, D. R., and Matt, G. E. (2000b). Reducing
children’s exposure to environmental tobacco smoke: The empirical evidence
and directions for future research. Tobacco Control, 9(Suppl 2): II40–II47.
Hovell, M. F., Zakarian, J. M., Wahlgren, D. R., Matt, G. E., and Emmons, K. M.
(2000c). Reported measures of environmental tobacco smoke exposure: Trials
and tribulations. Tobacco Control, 9(Suppl 3): III22–III28.
Jaakkola, M. S. and Jaakkola, J. J. K. (1997). Assessment of exposure to environmental tobacco smoke. European Respiratory Journal, 10(10): 2384–2397.
Jantunen, M. J., Hänninen, O., Katsouyanni, K., Knoppel, H., Kuenzli, N., Lebret,
E., Maroni, M., Saarela, K., Sram, R., and Zmirou, D. (1998). Air pollution exposure in European cities: The "EXPOLIS" study. Journal of Exposure Analysis and
Environmental Epidemiology, 8(4): 495 – 518.
82
Jenkins, P., Phillips, T., Mulberg, E., and Hui, S. (1992). Activity patterns of Californians: Use of and proximity to indoor pollutant sources. Atmospheric Environment, 26A(12): 2141–2148.
Jenkins, R. A., Guerin, M. R., and Tomkins, B. A. (2000). The Chemistry of Environmental Tobacco Smoke: Composition and Measurement. Lewis Publishers, Boca
Raton, second edition.
Jenkins, R. A., Palausky, A., Counts, R. W., Bayne, C. K., Dindal, A. B., and Guerin,
M. R. (1996). Exposure to environmental tobacco smoke in sixteen cities in the
United States as determined by personal breathing zone air sampling. Journal of
Exposure Analysis and Environmental Epidemiology, 6(4): 473–502.
Kegler, M. C. and Malcoe, L. H. (2002). Smoking restrictions in the home and car
among rural Native American and White families with young children. Preventive Medicine, 35(4): 334–342.
Kleeman, M. J., Schauer, J. J., and Cass, G. R. (1999). Size and composition distribution of fine particulate matter emitted from wood burning, meat charbroiling,
and cigarettes. Environmental Science & Technology, 33(20): 3516–3523.
Klepeis, N. E. (1999). Validity of the uniform mixing assumption: Determining
human exposure to environmental tobacco smoke. Environmental Health Perspectives, 107(SUPP2): 357–363.
Klepeis, N. E., Nelson, W. C., Ott, W. R., Robinson, J. P., Tsang, A. M., Switzer, P.,
Behar, J. V., Hern, S. C., and Engelmann, W. H. (2001). The National Human
Activity Pattern Survey (NHAPS): A resource for assessing exposure to environmental pollutants. Journal of Exposure Analysis and Environmental Epidemiology,
11(3): 231–252.
Koistinen, K. J., Hänninen, O., Rotko, T., Edwards, R. D., Moschandreas, D., and
Jantunen, M. J. (2001). Behavioral and environmental determinants of personal
exposures to PM2.5 in EXPOLIS - Helsinki, Finland. Atmospheric Environment,
35(14): 2473–2481.
Koontz, M. D. and Cox, S. S. (2002). Literature review on microenvironmental
modeling. In Levin, H., editor, Indoor Air – Proceedings of the 9th International
Conference on Indoor Air Quality and Climate, pages 304–309, Monterey, CA.
Koontz, M. D., Evans, W. C., and Wilkes, C. R. (1998). Development of a Model for
Assessing Indoor Exposure to Air Pollutants. A933-157, Geomet Technologies, Inc.
Prepared for California Air Resources Board, Sacramento, CA.
83
Koontz, M. D. and Nagda, N. L. (1991). A multichamber model for assessing consumer inhalation exposure. Indoor Air, 4: 593–605.
Koontz, M. D. and Niang, L. L. (1998). California Population Indoor Exposure Model
(CPIEM), Version 1.4F. A933-157 User’s Guide, Geomet Technologies. Prepared
for California Air Resources Board, Sacramento, CA.
Koutrakis, P., Briggs, S. L. K., and Leaderer, B. P. (1992). Source apportionment
of indoor aerosols in Suffolk and Onondaga counties, New York. Environmental
Science & Technology, 26(3): 521–527.
Kruize, H., Hänninen, O., Breugelmans, O., Lebret, E., and Jantunen, M. (2003).
Description and demonstration of the EXPOLIS simulation model: Two examples of modeling population exposure to particulate matter. Journal of Exposure
Analysis and Environmental Epidemiology, 13(2): 87–99.
Leaderer, B., Koutrakis, P., Briggs, S., and Rizzuto, J. (1990). Impact of indoor
sources on residential aerosol concentrations. In Walkinshaw, D., editor, Indoor
Air ’90. Proceedings of the 5th International Conference on Indoor Air Quality and
Climate, Volume 2, pages 269–274, Toronto, Canada.
Leech, J. A., Wilby, K., and McMullen, E. (1999). Environmental tobacco smoke
exposure patterns: A sub-analysis of the Canadian time-activity pattern survey.
Canadian Journal of Public Health, 90(4): 244–249.
Leech, J. A., Wilby, K., McMullen, E., and Laporte, K. (1996). The Canadian human
activity pattern survey: A report of methods and population surveyed. Chronic
Diseases in Canada, 17(3-4): 118–123.
Löfroth, G. (1993). Environmental tobacco smoke: Multicomponent analysis and
room-to-room distribution in homes. Tobacco Control, 2: 222–225.
Löfroth, G., Burton, R. M., Forehand, L., Hammond, S. K., Seila, R. L., Zweidinger,
R. B., and Lewtas, J. (1989). Characterization of environmental tobacco smoke.
Environmental Science and Technology, 23(5): 610–614.
Macintosh, D. L., Xue, J. P., Özkaynak, H., Spengler, J. D., and Ryan, P. B. (1995). A
population based exposure model for benzene. Journal of Exposure Analysis and
Mage, D. T. and Ott, W. R. (1996). The correction for nonuniform mixing in indoor
environments. In Tichenor, B. A., editor, Characterizing Indoor Air Pollution and
Related Sink Effects, pages 263–278, West Conshohocken, PA. American Society
for Testing and Materials.
84
Martin, P., Heavner, D. L., Nelson, P. R., Maiolo, K. C., Risner, C. H., Simmons, P. S.,
Morgan, W. T., and Ogden, M. W. (1997). Environmental tobacco smoke (ETS):
A market cigarette study. Environment International, 23(1): 75–90.
Matt, G. E., Wahlgren, D. R., Hovell, M. L., Zakarian, J. M., Bernert, J. T., Meltzer,
S. B., Pirkle, J. L., and Caudill, S. (1999). Measuring environmental tobacco
smoke exposure in infants and young children through urinary cotinine and
memory-based parental reports: Empirical findings and discussion. Tobacco Control, 8: 282–289.
McBride, S. J., Ferro, A. R., Ott, W. R., Switzer, P., and Hildemann, L. M. (1999).
Investigations of the proximity effect for pollutants in the indoor environment.
Journal of Exposure Analysis and Environmental Epidemiology, 9(6): 602–621.
McCurdy, T. (1995). Estimating human exposure to selected motor vehicle pollutants using the NEM series of models: Lessons to be learned. Journal of Exposure
Analysis and Environmental Epidemiology, 5(4): 533–550.
McKone, T. E. (1987). Human exposure to volatile organic compounds in household tap water – the indoor inhalation pathway. Environmental Science and Technology, 21(12): 1194–1201.
McMillen, R. C., Winickoff, J. P., Klein, J. D., and Weitzman, M. (2003). US adult
attitudes and practices regarding smoking restrictions and child exposure to environmental tobacco smoke: changes in the social climate from 2000-2001. Pediatrics, 112(1 Pt 1): e55–e60.
Miller, S. L., Leiserson, K., and Nazaroff, W. W. (1997). Nonlinear least-squares
minimization applied to tracer gas decay for determining airflow rates in a twozone building. Indoor Air, 7(1): 64–75.
Miller, S. L. and Nazaroff, W. W. (2001). Environmental tobacco smoke particles in
multizone indoor environments. Atmospheric Environment, 35(12): 2053–2067.
MMWR (1997). State-specific prevalence of cigarette smoking among adults, and
children’s and adolescent’s exposure to environmental tobacco smoke – United
States, 1996. Morbidity and Mortality Weekly Report, 46(44): 1038–1043.
policies and attitudes about secondhand smoke – United States, 2000. Morbidity
and Mortality Weekly Report, 50(49): 1101–1106.
Nazaroff, W. W. and Singer, B. C. (2004). Inhalation of hazardous air pollutants
from environmental tobacco smoke in US residences. Journal of Exposure Analysis
and Environmental Epidemiology, 14: S71–S77.
85
Neas, L. M., Dockery, D. W., Ware, J. H., Spengler, J. D., Ferris, B. G., and Speizer,
F. E. (1994). Concentration of indoor particulate matter as a determinant of respiratory health in children. American Journal of Epidemiology, 139(11): 1088–1099.
Norman, G. J., Ribisl, K. M., Howard-Pitney, B., and Howard, K. A. (1999). Smoking bans in the home and car: Do those who really need them have them? Preventive Medicine, 29(6 Pt 1): 581–589.
OEHHA (1997). Health Effects of Exposure to Environmental Tobacco Smoke. California Environmental Protection Agency, Office of Environmental Health Hazard
Assessment (OEHHA), Sacramento, CA. Final Report.
OEHHA (2002). All Chronic Reference Exposure Levels. California Environmental
Protection Agency, Office of Environmental Health Hazard Assessment.
Ott, W. (1982). Concepts of human exposure to air pollution. Environment International, 7: 179–196.
Ott, W. (1984). Exposure estimates based on computer generated activity patterns.
Journal of Toxicology: Clinical Toxicology, 21: 97–128.
Ott, W., Thomas, J., Mage, D., and Wallace, L. (1988). Validation of the Simulation
of Human Activity and Pollutant Exposure (SHAPE) model using paired days
from the Denver, CO carbon monoxide field study. Atmospheric Environment,
22(10): 2101 – 2113.
Ott, W. R. (1985). Total human exposure. Environmental Science and Technology,
19(10): 880–886.
Ott, W. R. (1990). Total human exposure - basic concepts, EPA field studies, and
future research needs. Journal of the Air & Waste Management Association, 40(7):
966–975.
Ott, W. R. (2002). Unpublished tracer data collected in paired rooms of a large
ranch-style house. Personal communication.
Ott, W. R., Klepeis, N. E., and Switzer, P. (2003). Analytical solutions to compartmental indoor air quality models with application to environmental tobacco
smoke concentrations measured in a house. Journal of the Air and Waste Management Association, 53: 918–936.
Özkaynak, H., Xue, J., Spengler, J., Wallace, L., Pellizzarri, E., and Jenkins, P. (1996).
Personal exposure to airborne particle and metals – Results from the particle
TEAM study in Riverside, California. Journal of Exposure Analysis and Environmental Epidemiology, 6(1): 57–78.
86
Palmes, E. D., Tomczyk, C., and Dimattio, J. (1977). Average NO2 concentrations
in dwellings with gas or electric stoves. Atmospheric Environment, 11: 869–872.
Palmes, E. D., Tomczyk, C., and March, A. W. (1979). Relationship of indoor NO2
concentrations to use of unvented appliances. Journal of the Air Pollution Control
Association, 29(4): 392–393.
Pellizzari, E., Lioy, P., Quackenboss, J., Whitmore, R., Clayton, A., Freeman, N.,
Waldman, J., Thomas, K., Rodes, C., and Wilcosky, T. (1995). Population-based
exposure measurements in EPA Region 5 – A phase I field study in support of
the National Human Exposure Assessment Survey. Journal of Exposure Analysis
and Environmental Epidemiology, 5(3): 327–358.
Phillips, K. and Bentley, M. C. (2001). Seasonal assessment of environmental tobacco smoke and respirable suspended particle exposures for nonsmokers in
Bremen using personal monitoring. Environment International, 27(1): 69–85.
Phillips, K., Bentley, M. C., Howard, D. A., and Alvan, G. (1996). Assessment of
air quality in Stockholm by personal monitoring of nonsmokers for respirable
suspended particles and environmental tobacco smoke. Scandinavian Journal of
Work Environment & Health, 22(S1): 1–24.
Phillips, K., Bentley, M. C., Howard, D. A., and Alvan, G. (1998a). Assessment
of air quality in Paris by personal monitoring of nonsmokers for respirable suspended particles and environmental tobacco smoke. Environment International,
24(4): 405–425.
Phillips, K., Bentley, M. C., Howard, D. A., and Alvan, G. (1998b). Assessment
of environmental tobacco smoke and respirable suspended particle exposures
for nonsmokers in Prague using personal monitoring. International Archives of
Occupational and Environmental Health, 71(6): 379–390.
Phillips, K., Bentley, M. C., Howard, D. A., and Alvan, G. (1998c). Assessment of
environmental tobacco smoke and respirable suspended particles for nonsmokers in Kuala Lumpur using personal monitoring. Journal of Exposure Analysis and
Environmental Epidemiology, 8: 519–542.
Phillips, K., Bentley, M. C., Howard, D. A., Alvan, G., and Huici, A. (1997a). Assessment of air quality in Barcelona by personal monitoring of nonsmokers for
respirable suspended particles and environmental tobacco smoke. Environment
International, 23(2): 173–196.
Phillips, K., Howard, D. A., Bentley, M. C., and Alvan, G. (1997b). Assessment of
air quality in Turin by personal monitoring of nonsmokers for respirable sus-
87
pended particles and environmental tobacco smoke. Environment International,
23(6): 851–871.
Phillips, K., Howard, D. A., Bentley, M. C., and Alvan, G. (1998d). Assessment
by personal monitoring of respirable suspended particles and environmental
tobacco smoke exposure for nonsmokers in Sydney, Australia. Indoor + Built
Environment, 7: 188–203.
Phillips, K., Howard, D. A., Bentley, M. C., and Alvan, G. (1998e). Assessment
of environmental tobacco smoke and respirable suspended particle exposures
for nonsmokers in Hong Kong using personal monitoring. Environment International, 24(8): 851–870.
Phillips, K., Howard, D. A., Bentley, M. C., and Alvan, G. (1998f). Assessment of
environmental tobacco smoke and respirable suspended particle exposures for
nonsmokers in Lisbon by personal monitoring. Environment International, 24(3):
301–324.
Phillips, K., Howard, D. A., Bentley, M. C., and Alvan, G. (1998g). Environmental
tobacco smoke and respirable suspended particle exposures for nonsmokers in
Beijing. Indoor + Built Environment, 7: 254–269.
Phillips, K., Howard, D. A., Bentley, M. C., and Alvan, G. (1998h). Measured exposures by personal monitoring for respirable suspended particles and environmental tobacco smoke of housewives and office workers resident in Bremen,
Germany. International Archives of Occupational and Environmental Health, 71(3):
201–212.
Phillips, K., Howard, D. A., Bentley, M. C., and Alvan, G. (1999). Assessment of
environmental tobacco smoke and respirable suspended particle exposures for
nonsmokers in Basel by personal monitoring. Atmospheric Environment, 33(12):
1889–1904.
Pirkle, J. L., Flegal, K. M., Bernert, J. T., Brody, D. J., Etzel, R. A., and Maurer, K. R.
(1996). Exposure of the US population to environmental tobacco smoke: The
third National Health and Nutrition Examination Survey, 1988 to 1991. Journal
of the American Medical Association, 275(16): 1233–1240.
Pizacani, B. A., Martin, D. P., Stark, M. J., Koepsell, T. D., Thompson, B., and Diehr,
P. (2003). Household smoking bans: Which households have them and do they
work? Preventive Medicine, 36(1): 99–107.
Price, P. S., Koontz, M., Wilkes, C., Ryan, B., Macintosh, D., and Georgopoulos, P.
(2003). Construction of a Comprehensive Chemical Exposure Framework Using Person
88
Oriented Modeling. The Lifeline Group. Developed for the American Chemistry
Council.
Quackenboss, J. J., Lebowitz, M. D., and Crutchfield, C. D. (1989). Indoor-outdoor
relationships for particulate matter — exposure classifications and health effects.
Environment International, 15: 353–360.
Ribot, B., Chen, Q. Y., Huang, J. M., and Rivoalen, H. (2002). Numerical simulations
of indoor air quality in a French house: Study of the distribution of pollutants
in each room, modelling contaminant exposure. In Levin, H., editor, Indoor Air
– Proceedings of the 9th International Conference on Indoor Air Quality and Climate,
pages 524–529, Monterey, CA.
Richmond, H. M., Palma, T., Langstaff, J., McCurdy, T., Glen, G., and Smith, L.
(2002). Further refinements and testing of APEX(3.0): EPA’s population exposure model for criteria and air toxic inhalation exposures. Presented at the
ISEA/ISEE Joint Conference, Vancouver, BC, August 11-15.
Rodes, C. E., Kamens, R., and Wiener, R. W. (1991). The significance and characteristics of the personal activity cloud on exposure assessment measurements for
indoor contaminants. Indoor Air, 2: 123–145.
Rogge, W. F., Hildemann, L. M., Mazurek, M. A., and Cass, G. R. (1994). Sources of
fine organic aerosol. 6. Cigarette smoke in the urban atmosphere. Environmental
Science and Technology, 28(7): 1375–1388.
Rosenbaum, A. (2002). The HAPEM4 User’s Guide: Hazardous Air Pollutant Exposure
Model, Version 4. ICF Consulting. Prepared for the U.S. Environmental Protection
Agency, Office of Air Quality, Planning, and Standards.
Rosenbaum, A., Cohen, J., Kavoosi, F., Lum, S., and Jenkins, P. (2002). The California Population Indoor Exposure Model, version 2: A user-friendly assessment
tool for population exposure to air pollutants. Presented at the ISEA/ISEE Joint
Conference, Vancouver, BC, August 11-15.
Schuster, M. A., Franke, T., and Pham, C. B. (2002). Smoking patterns of household
members and visitors in homes with children in the United States. Archives of
Pediatrics & Adolescent Medicine, 156(11): 1094–1100.
Seigneur, C., Pun, B., Lohman, K., and Wu, S. Y. (2002). Air Toxics Modeling. Atmospheric and Environmental Research, Inc., San Ramon, CA. Prepared for the
Coordinating Research Council, Inc. and the U.S. Department of Energy, Office
of Heavy Vehicle Technologies, National Renewable Energy Laboratory.
89
Sexton, K., Kleffman, D. E., and Callahan, M. A. (1995). An introduction to the
National Human Exposure Assessment Survey (NHEXAS) and related Phase I
field studies. Journal of Exposure Analysis and Environmental Epidemiology, 5(3):
229–232.
Sheldon, L. S., Hartwell, T. D., Cox, B., Sickles, J. E., Pellizzari, E. D., Smith, M. L.,
Perritt, R. L., and Jones, S. M. (1989). An Investigation of Infiltration and Air Quality. Final report, New York State Energy Research and Development Authority
(ERDA), Albany, New York. New York State ERDA Contract No. 736-CON-BCS85.
Singer, B. C., Hodgson, A. T., Guevarra, K. S., Hawley, E. L., and Nazaroff, W. W.
(2002). Gas-phase organics in environmental tobacco smoke. 1. Effects of smoking rate, ventilation, and furnishing level on emission factors. Environmental
Singer, B. C., Hodgson, A. T., and Nazaroff, W. W. (2003). Gas-phase organics in environmental tobacco smoke: 2. Exposure-relevant emission factors and indirect
exposures from habitual smoking. Atmospheric Environment, 37: 5551–5561.
Sparks, L. E. (1988). Indoor Air Model, Version 1.0. EPA 600/8-88-097a, U.S. Environmental Protection Agency, Research Triangle Park, NC.
Sparks, L. E. (1991). EXPOSURE Version 2: A Computer Model for Analyzing the
Effects of Indoor Air Pollutant Sources on Individual Exposure. EPA-600/8-91-013,
United States Environmental Protection Agency, Research Triangle Park, NC.
Sparks, L. E., Tichenor, A. B., White, J. B., and Jackson, M. D. (1991). Comparison of
data from an IAQ test home with predictions of an IAQ computer model. Indoor
Air, 4: 577–592.
Sparks, L. E., Tichenor, B. A., and White, J. B. (1993). Modeling individual exposure
from indoor sources. In Nagda, N., editor, Modeling of Indoor Air Quality and
Exposure, ASTM STP-1205, pages 245–256, Philadelphia, PA. American Society
Spengler, J. D., Dockery, D. W., Turner, W. A., Wolfson, J. M., and Ferris, B. G.
(1981). Long-term measurements of respirable sulfates and particles inside and
outside homes. Atmospheric Environment, 15(1): 23–30.
Spengler, J. D., Treitman, R. D., Tosteson, T. D., Mage, D. T., and Soczek, M. L.
(1985). Personal exposures to respirable particulates and implications for airpollution epidemiology. Environmental Science and Technology, 19(8): 700–707.
90
Spengler, J. D., Ware, J., Speizer, F., Ferris, B., Dockery, D., Lebret, E., and
Brunekreef, B. (1987). Harvard’s indoor air quality respiratory health study. In
Seifert, B., editor, Indoor Air ’87. Proceedings of the 4th International Conference on
Indoor Air Quality and Climate, Volume 2, pages 218–223, Berlin.
Sterling, T. D. and Sterling, E. (1979). Carbon monoxide levels in kitchens and
homes with gas cookers. Journal of the Air Pollution Control Association, 29(3):
238–241.
Traynor, G. T., Aceti, J. C., and Apte, M. G. (1989). Macromodel for Assessing Residential Concentrations of Combustion-Generated Pollution: Model Development and
Preliminary Predictions for CO, NO2 , and Respirable Suspended Particles. LBL-25211,
Lawrence Berkeley National Laboratory, Berkeley, CA.
Turner, S., Cyr, L., and Gross, A. J. (1992). The measurement of environmental
tobacco smoke in 585 office environments. Environment International, 18: 19–28.
UKDH (1998). Report of the Scientific Committee on Tobacco and Health. Department of
Health (UK), Department of Health And Social Services, Northern Ireland, The
Scottish Office Department of Health, Welsh Office, London, England. Official
United Kingdom Document.
USDHHS (2002). Report on Carcinogens. U.S. Department of Health and Human
Services, Public Health Service, National Toxicology Program, Washington, D.C.
Tenth Edition.
USDHHS (2003). Second National Report on Human Exposure to Environmental Chemicals. NCEH Pub. No. 02-0716, Department of Health and Human Servies, Centers for Disease Control and Prevention, National Center for Environmental
Health, Division of Laboratory Sciences, Atlanta, GA.
USEPA (1992). Respiratory Health Effects of Passive Smoking: Lung Cancer and Other
Disorders. EPA/600/6-90/006F, U.S. Environmental Protection Agency, Office of
Research and Development, Washington, D.C.
USEPA (1997). Exposure Factors Handbook. EPA/600/P-95/002F, United States Environmental Protection Agency, Office of Research and Development, Washington, D.C.
USEPA (2002).
Child-Specific Exposure Factors Handbook (Interim Report).
EPA/600/P-00/002B, U.S. Environmental Protection Agency, Office of Research
and Development, National Center for Environmental Assessment, Washington
Office, Washington, DC.
91
USEPA (2003). Integrated Risk Information System. U.S. Environmental Protection
Agency.
Wade, W. A., Cote, W. A., and Yocom, J. E. (1975). A study of indoor air quality.
Journal of the Air Pollution Control Association, 25(9): 933–939.
Wakefield, M., Banham, D., Martin, J., Ruffin, R., Mccaul, K., and Badcock, N.
(2000a). Restrictions on smoking at home and urinary cotinine levels among
children with asthma. American Journal of Preventive Medicine, 19(3): 188–192.
Wakefield, M., Banham, D., McCaul, K., Martin, J., Ruffin, R., Badcock, N., and
Roberts, L. (2002). Effect of feedback regarding urinary cotinine and brief tailored advice on home smoking restrictions among low-income parents of children with asthma: A controlled trial. Preventive Medicine, 34(1): 58–65.
Wakefield, M. A., Chaloupka, F. J., Kaufman, N. J., Orleans, C. T., Barker, D. C., and
Ruel, E. E. (2000b). Effect of restrictions on smoking at home, at school, and in
public places on teenage smoking: Cross sectional study. British Medical Journal,
321(7257): 333–337.
Wallace, L. A. (1987). The Total Exposure Assessment Methodology (TEAM) Study:
Summary and Analysis: Volume I. U.S. EPA, Washington D. C.
WHO (2000). Air Quality Guidelines, Second Edition. World Health Organization
(WHO), Regional Office for Europe, Copenhagen, Denmark.
Wiley, J., Robinson, J. P., Cheng, Y., Piazza, T., Stork, L., and Pladsen, K. (1991a).
Study of Children’s Activity Patterns. California Air Resources Board, Contract No.
A733-149, Sacramento, CA.
Wiley, J., Robinson, J. P., Piazza, T., Garrett, K., Cirksena, K., Cheng, Y., and Martin, G. (1991b). Activity Patterns of California Residents. California Air Resources
Board, Contract No. A6-177-33, Sacramento, CA.
Wilkes, C. R., Blancato, J. N., Hern, S. C., Power, F. W., and Olin, S. S. (2002). Integrated probabilistic and deterministic modeling techniques in estimating exposures to water-borne contaminants: Part 1, Exposure modeling. In Levin,
H., editor, Indoor Air – Proceedings of the 9th International Conference on Indoor Air
Quality and Climate, pages 256–261, Monterey, CA.
Wilkes, C. R., Small, M. J., Andelman, J. B., Giardino, N. J., and Marshall, J. (1992).
Inhalation exposure model for volatile chemicals from indoor uses of water. Atmospheric Environment, 26A(12): 2227–2236.
92
Wilkes, C. R., Small, M. J., Davidson, C. I., and Andelman, J. B. (1996). Modeling
the effects of water usage and co-behavior on inhalation exposures to contaminants volatilized from household water. Journal of Exposure Analysis and Environmental Epidemiology, 6(4): 393–412.
Wilson, S. R., Yamada, E. G., Sudhakar, R., Roberto, L., Mannino, D., Mejia, C., and
Huss, N. (2001). A controlled trial of an environmental tobacco smoke reduction
intervention in low-income children with asthma. Chest, 120(5): 1709–1722.
Zartarian, V. G., Ott, W. R., and Duan, N. H. (1997). A quantitative definition of
exposure and related concepts. Journal of Exposure Analysis and Environmental
Epidemiology, 7(4): 411–437.
93
Part II
Model Development
94
The following four chapters describe inputs and structural aspects of a
simulation model for residential SHS exposure. The three key areas of
data input are tobacco smoke emissions and dynamic behavior, human
time-activity patterns, and housing characteristics.
Chapter 3 (page 95) contains a description of emission factors and surface interaction parameters for particulate and major gaseous components of SHS. A method for determining size-specific emissions and its
application to data from a set of original cigar and cigarette chamber
experiments are also presented.
Chapter 4 (page 140) contains a description of activity pattern data from
the USEPA’s National Human Activity Pattern Survey (NHAPS). Attention is focused on the movement of people between rooms of their
residence.
Chapter 5 (page 165) contains a description of characteristics for the US
housing stock, including air-exchange rates, interzonal air flow rates,
air handling systems, mixing rates, house volume, and numbers of
rooms and floors.
Chapter 6 (page 200) contains a discussion of the design and key features of an original simulation model for residential exposure to SHS.
95
Chapter 3
Emissions Characterization and
Dynamic Behavior of Key SHS
Constituents
The amount of material emitted from burning cigarettes and cigars is a key factor in determining maximum and time-averaged air pollutant concentrations in
houses where smoking occurs. Air pollutant concentrations are also affected by
surface reactivity of emissions, which influences how rapidly species are removed
from air and the reentrainment of sorbed species into air via surface desorption.
Therefore, in this chapter, I am interested in characterizing emission factors for
cigarette and cigar SHS sources and also deposition, sorption, and desorption rates
of key SHS species for typical residential environments.1
It is a difficult to find compounds that reasonably represent the extremely varied makeup of SHS. However, three species – particles, carbon monoxide (CO),
and nicotine – are well-studied air pollutants that provide a reasonable representation of the array of SHS components in terms of phase and dynamic behavior.
Particles and nicotine react with environment surfaces by irreversible deposition
and reversible sorption, respectively. In contrast, CO is a nonreactive species that
is only removed from indoor air by ventilation. Together, these species represent
1 Much
of the material in this chapter has been published in Klepeis, N. E., Apte, M. G., Gundel, L. A., Sextro, R. G., and Nazaroff, W. W. (2003) Determining size-specific emission factors for
environmental tobacco smoke particles. Aerosol Science and Technology, 37: 780-790.
CHAPTER 3. EMISSIONS CHARACTERIZATION
96
a significant portion of total SHS emissions, and using these compounds in applications of the simulation model I develop in this dissertation will facilitate the
comparison of my simulation results to the findings of a large number of published
chamber and field studies.
Nicotine is a heavily studied compound not only because it forms the basis for
the addiction of smokers, but, since it is specific to tobacco emissions, it and its
metabolites have been widely used as chemical markers for bulk SHS concentrations and exposures (see Sections 2.1.2 and 2.4.4). Carbon monoxide has also been
used as a marker for SHS, because it is emitted in fairly large quantities and is
relatively easy to measure, although interferences can result from CO emissions
of household appliances and automobiles. Many field studies, some of which are
presented in Chapter 2, have focused on environmental concentrations of particles,
especially those under 2.5 µ m in diameter, i.e., PM2.5 . As with CO, it is sometimes
difficult to attribute field measurements of particles to SHS, because of interference
with other sources, including cooking, cleaning, or outdoor air.
In addition to being major components of SHS, PM2.5 and CO are also USEPA
criteria pollutants for which there are extensive monitoring networks and established California and Federal ambient air quality standards, which are summarized in Table 3.1 for different averaging times. Carbon monoxide is a poisonous
gas that interferes with the binding of oxygen to hemoglobin. Both pollutants
present special health hazards, because they are able to penetrate deeply into the
human lung.
In estimating exposure to SHS particles, which constitute a major component
of SHS by mass, I am generally focused on “SHS emissions” of particles as opposed to fresh mainstream or side-stream tobacco smoke emissions. My aim is to
determine the characteristics of “effective” SHS emissions, defined as the mass of
particles that have come to be dispersed in a previously pollutant-free room just after a cigarette (or cigar) has been smoked. Sidestream tobacco smoke is defined as
the undiluted plume coming from the smoldering end of the cigarette, and mainstream smoke is the undiluted puff of smoke that is drawn through the cigar or
97
Table 3.1: California and US Federal Concentration Guidelines for Carbon Monoxide (CO) and PM2.5
Averaging
California Standard
Federal Standarda
Pollutant
Time
[µ g m−3 ]
[µ g m−3 ]
CO
1-h
23,000
40,000
8-h
10,000
10,000
24-h
65
65
Annual
12
15
PM2.5
a Source:
USEPA National Ambient Air Quality Standards (NAAQS).
cigarette and then exhaled by the smoker (either a human or a machine). The particles in real or simulated SHS are derived from particles in exhaled mainstream
and sidestream smoke, but they are different in that they have undergone mixing
and dilution (i.e., dispersion), and perhaps deposition and filtration over varying
time scales and in a particular indoor setting (e.g., a home, an automobile, or a
workplace). While undergoing dispersion, the median tobacco smoke particle size
can shrink as particle mass evaporates [Hinds, 1978] or it can grow as particles
coagulate. The end result can be SHS particle size distributions that are different
from the distributions of fresh mainstream or sidestream smoke.
In the remainder of this chapter, I first briefly discuss human smoking patterns, which influence bulk SHS emissions in a residence over the course of the
day. The next two sections describe original SHS particle chamber experiments
and the interpretation of the size-specific particle concentrations measured in these
experiments with an aerosol dynamics model. The next three sections summarize
estimates of size-specific and integrated particle emission factors and particle deposition rates as determined from both my original experiments and previously
published studies. Finally, I include two sections discussing estimates of CO emission factors and emissions and surface-related characteristics of volatile organics,
specifically for nicotine.
98
3.1 Human Smoking Patterns
Key model input parameters include the time it takes to smoke a single cigarette,
the number of cigarettes smoked at home, and the magnitude of per-cigarette emissions. Since cigarettes are consumed over a short period compared to the time
typically spent in any given room of a house (see Chapter 4), the cigarette combustion duration is relatively unimportant compared to the actual magnitude of
emissions. Cigarettes can almost be treated as instantaneous releases of air pollutants, with the total number of releases being the critical quantity, although the
duration of the cigarette burn may have a small effect on peak air pollutant concentrations. Cigarette duration is expected to be in the range of 7−11 min. Ott et al.
[2003] report unpublished data on the duration of cigarettes that were smoked in
a Las Vegas casino, finding an average cigarette duration of 9 minutes with a standard deviation of 2 minutes. Smoldering regular cigarettes that are not actively
aerated by a machine or human smoker typically last approximately 10 minutes
before the burning tip reaches the filter. Longer cigarettes are expected to have a
proportionately longer smolder duration.
The number of cigarettes that are smoked in the house is expected to carry more
influence than cigarette duration in determining average concentrations of SHS air
pollutants both in the smoking room and in rooms with an open air pathway to
and/or from the smoking room. In the 1999 National Household Survey on Drug
Abuse (NHSDA), 77% of daily smokers reported smoking 6 to 15 cigarettes a day
with 19% smoking a pack of 20 cigarettes or more per day. For teenagers and young
adults under age 25, approximately 90% of daily smokers reported smoking 6 to 15
cigarettes a day, whereas 14% of young adults aged 26−34 smoked a pack or more
and 23% of older adults smoked a pack or more per day.2 Nazaroff and Singer
[2004] use a US per capita consumption rate of 117 packs, obtained from data on
taxes paid [TI, 1997], and the US prevalence of smoking [MMWR, 2001] to estimate
2 The
US Department of Health and Human Services, Substances Abuse and Mental Health Services Administration, conduct the NHSDA every year, which is an annual cross-sectional study on
the prevalence and incidence of drug, alcohol, and tobacco use of Americans 12 years of age and
older. The 1999 survey includes data from nearly 70,000 persons [Kopstein, 2001].
99
an average rate of 20−28 cigarettes consumed per smoker per day.
A reasonable number, if perhaps towards the upper end of the distribution, for
daily consumption of cigarettes for adults that may have children residing in their
homes is 1 or 2 packs of cigarettes. A fraction of these cigarettes will be smoked in
the home. This fraction will be depend largely on whether or not a parent works
at home. A reasonable assumption is to spread the number of cigarettes smoked
evenly throughout a smoker’s waking period so that a portion of these cigarettes
will fall naturally during periods where the smoker is at home. This strategy is
used for the simulation model presented in the current work (see Chapter 6).
The mass of pollutant emitted in a given smoking episode is the product of the
number of cigarettes that are smoked and the average mass that is emitted from
each cigarette. Investigators typically report the total mass of particles or gaseous
species that are emitted after an entire is cigarette is smoked, or sometimes the
mass that is emitted per unit time or per unit mass of consumed tobacco. The
total mass emitted per cigarette depends on the style and rapidity of smoking,
i.e., the frequency and volume of puffs, in addition to the type of cigarette [NCI,
1996]. However, due to their uniformity in shape and size in comparison to cigars,
cigarettes are expected to have a more narrow range in both duration of smoking
and the mass of emissions per cigarette.
Since few investigators have reported emission rates, and few if any have reported precisely how emissions change in time, a convenient assumption is that
smokers typically finish an entire cigarette. Therefore, it seems reasonable to take
established emissions per cigarette and assume they are emitted evenly throughout a given consumption period, even though in reality the emission rate may vary
in time in a way that is peculiar to each cigarette and each smoker. In the current
work, I model emissions from the smoking of multiple cigarettes over the course
of a day by placing individual cigarettes in time and assigning a fixed emission
rate for the time that they are active by dividing reported per-cigarette emissions
by the time spent smoking the cigarette.
100
3.2 Cigar and Cigarette Experiments
I conducted nine cigar (premium, regular, and cigarillo) and four cigarette (regular
and lights) smoking experiments in an unventilated 20 m3 chamber (see Figure 3.1
for a schematic). For eight of these experiments, valid measurements of the particle
number concentration were obtained (counts per cm3 ) in the chamber air every
minute using an optical particle counter (LASAIR; Particle Measurement Systems,
Inc., Boulder, CO), which registered particle counts in 8 size bins ranging from 0.1
to over 2 µ m based on the scattering of 633 nm light emitted from a HeNe laser.
The LASAIR input air stream was diluted with filtered air at 5 to 6 times the sample
air flow rate. Semi-continuous particle number concentration was also measured
using a differential mobility particle sizer (DMPS; TSI, Inc., St. Paul, MN), which
scanned and sized particles in 34 size bins ranging from approximately 0.01 to over
1 µ m in diameter over approximately 10- or 30-min intervals. The DMPS sizes
particles according to their mobility when charged and placed in an electric field.
The particle size measurement devices were placed outside the chamber and their
input air was drawn through sampling tubes located near the top of the chamber
door.
During most of the experiments, electrochemical measurements of carbon monoxide (CO) were made every minute using a Langan CO Personal Measurer (Langan Products, San Francisco, CA), which was placed inside the chamber and connected to a Langan DataBear digital logger. The CO measurements were used to
obtain air exchange rate values for the chamber, which ranged from 0.03 to 0.1 h−1 ,
by fitting a line to the natural logarithm of the decaying CO concentrations.
The interior surfaces of the smoking chamber consisted entirely of stainless
steel. In addition, two 4-foot by 8-foot sheets of upright gypsum wallboard (a
total of approximately 12 m2 of exposed surface area) were placed vertically in the
center of the chamber. The inside volume of the chamber was approximately 20 m3
and the surface area was approximately 57 m2 – including the wallboard – giving
a surface-to-volume ratio of 2.9 m−1 .
Pre-weighed cigars and cigarettes were smoked using a smoking machine
101
= Elevated Fans
Inside Width = 2.4 m
Particle Sampling Tubes
1
0
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0Door
1
0
1
Smoking
Machine
CO
Monitor
Cigar
Stand
1111111111111
0000000000000
0000000000000
1111111111111
Inside Height = 2.2 m
Volume = 19.9 m³
S/V = 2.9 m−1
Wall−
board
Inside Length = 3.7 m
Figure 3.1: A schematic of the experimental chamber showing the chamber dimensions and the approximate placement of the sampling tubes, cigar and cigarette
sources, CO monitor, two sheets of wallboard, and six mixing fans. The surface-tovolume ratio of the chamber, including the two sheets of upright wallboard, was
2.9 m−1 . The particle sampling tubes were connected to particle sizing instrumentation based on either light-scattering (LASAIR) or electrostatic mobility (DMPS).
102
(ADL II, Arthur D. Little, Inc., Cambridge, MA) at a standard rate of two 35-cm3
puffs per minute. The cigars were 13-cm regular Swisher Sweets, aromatic and
mild blend plastic-tipped Tiparillos (cigarillos), and an 18-cm Macanudo premium.
The cigarettes were Camel lights and Marlboro regulars.
I ignited the tobacco products using a hand lighter, whereupon I exited the
chamber and securely closed the airtight door. Both sidestream and mainstream
tobacco smoke were freely emitted into the chamber where they were thoroughly
mixed by six small fans – two aimed up at the plume with four more cycling air
clockwise around the chamber. The cigarettes were held in place by a customdesigned automatic smoking carousel connected to the smoking machine, while
the cigars were attached to a nearby stand and connected to the smoking machine
using a copper fitting, Teflon tape, and plastic tubing. During smoking, smoke
plumes observed through the chamber window became rapidly dispersed over a
period of a few seconds.
A timer was used to disconnect power from the smoking machine after a preset
smoking time (10−15 min for cigars and 5−8 min for cigarettes). The smoldering
cigars and cigarettes were rapidly extinguished from outside of the closed chamber
by forcing nitrogen gas in reverse direction through the cigarette or cigar.
Once the sources were completely extinguished, I began collecting total particle mass (TPM) on Teflon-coated glass fiber filters, sampling at approximately
18 L min−1 over time periods ranging from 30 to 60 min. The LASAIR and DMPS,
being activated before the sources were ignited to measure background levels,
were left to record particle number concentrations in the sealed chamber for at
least 12 h and, in some cases, up to 24 h after smoking. The background levels
were negligible compared to the peak concentrations in each experiment for each
particle size range.
TPM concentration was determined gravimetrically by weighing the accumulated mass on each filter with a Cahn-25 precision electrobalance and dividing it
by the volume of chamber air sampled. The filters were frozen following each experiment, and then thawed and equilibrated to ambient relative humidity prior to
103
being reweighed. The TPM emitted by a cigar or cigarette during each experiment
was estimated from the filter TPM concentrations by taking into account the loss
of mass from deposition and ventilation that occurred during sample collection.
The total particle removal rate for each experiment was estimated by fitting a line
to the logarithm of the decaying total particle counts as measured by the LASAIR.
On the day following each experiment, the unsmoked portion of each cigar or
cigarette was weighed to determine the mass of tobacco that had been consumed
by smoking.
The particle count data measured with the LASAIR and DMPS instruments
were analyzed to estimate the mass of SHS particles emitted in each size range. The
LASAIR- and DMPS-measured particle number concentrations were converted to
particle mass concentrations for each particle size range by assuming that SHS
particles have a density of 1.1 g/cm3 [Lipowicz, 1988], that the logarithm of the
particle mass concentration is uniformly distributed within each size range, and
that SHS particles are spherical.
The LASAIR was originally calibrated with latex spheres, which have a different refractive index (1.588 - 0i) than SHS particles (1.532 + 0i; McRae [1982]). The
LASAIR data were post-calibrated by comparing the calculated LASAIR response
for SHS particles to those for latex spheres [Bohren and Huffman, 1983; Garvey
and Pinnick, 1983].
Sufficiently high quality data from the DMPS were limited to three of the cigar
experiments and these data did not, in general, possess high time resolution or uniformity since each scan lasted from 10 to 30 min and there was sometimes a delay
between scans. Therefore, with regard to the parameter optimization procedure
(see below), the use of the DMPS data was limited to providing initial estimates of
the proportion of mass emitted for particles smaller than 0.1 µ m, a diameter range
not sampled by the LASAIR.
For the parameter optimization procedure, initial particle emissions were estimated independently for each size range using the following formula:
E = φVC +
V
C
To f f − Ton peak
(3.1)
104
where E is the mass emission rate for each bin [µ g min−1 ], φ is the total loss rate for
each bin [min−1 ], V is the chamber volume, C peak is the peak particle concentration
for each size range above a zero background [µ g/m3 ], and C is the average concentration between the time the source started, Ton [min], and the time the source
ended, To f f [min]. The average concentrations, C, was approximated as
C peak
2 .
The
loss rate φ for each particle diameter range was estimated by fitting a line to the
logarithm of the decaying concentrations.
3.3 Estimating Particle Emissions with an Aerosol Dynamics Model
The aerosol dynamics model of Nazaroff and Cass [1989] was adapted to calculate
SHS particle concentrations for each experiment, taking into account the effects of
ventilation, particle coagulation, deposition, and direct emissions from a cigar or
a cigarette. Coagulation, deposition, and emission change the size distribution of
airborne particles. In contrast, ventilation removes particles at an equal rate across
all sizes.
Although creation of new particle mass can occur through condensation of
semi-volatile organic compounds (SVOCs) from particles, I do not expect this process to play an important role in SHS aerosol dynamics. Kousaka et al. [1982]
report that humidity only affects the growth of smoke particles under supersaturated conditions, which do not apply in my case. The shrinkage of particles owing
to SVOC evaporation can also occur for SHS, such as when rapid mixing and dilution occur after smoking. Both Ingebrethsen and Sears [1989] and Hinds [1978]
provide evidence of this phenomenon.
For experiments in the current work, the chamber ventilation rate and source
duration are known, and the rate of particle coagulation is calculated based on concentrations occurring in a given time step. The two remaining, unknown model
parameters are the magnitude of the particle mass emissions and the rate of particle deposition onto surfaces. In this chapter, I reference mass emissions in terms of
three types of emission factors: the mass emission rate (mg emitted per minute);
105
the total particle mass (TPM) emissions (mg emitted per tobacco source); and the
mass-normalized emissions (mg emitted per gram of tobacco consumed).
My task is to find values of the unknown model parameters for each particle size range (or bin) that result in the best fit of model predictions to measured
concentrations. In addition to tuning model input values, the fits also provide an
indication of model accuracy. I used the following steps to obtain optimal values
of mass emissions and deposition loss-rate coefficient for each measured particle
diameter from each of the eight chamber experiments in which valid LASAIR measurements were recorded.
Step 1. From the LASAIR data, initial guesses of the mass emission rate and deposition loss-rate coefficient were made for each optimization from observed
peak concentrations and decay rates for each size bin, assuming independence among the bins (see Equation 3.1). The initial estimate for deposition
loss rate was calculated by subtracting the ventilation rate from the overall
particle loss rate. These initial values are expected to be in error since the loss
or gain of particle mass in each bin also depends on coagulation.
Step 2. The aerosol dynamics model and a local grid search routine were used to
locate the optimal values of mass emission rate and deposition rate for each
particle size range starting at 0.1 µ m, which is the lower limit for the LASAIR,
and ending at the bin with a 2.0 µ m lower limit. For the goodness-of-fit statistic between modeled and observed mass concentration time series, the mean
absolute deviation was used, which is less sensitive to outliers than the mean
squared deviation. A time period of 8 h was selected for each experiment,
since it would capture time-dependent dynamics over a relatively long time
scale but avoid appreciable shifts in background concentration that appeared
to occur over 12−24 h time periods. The sample size of each series used in the
optimization was approximately 400 for each experiment. The optimization
surface was generally smooth with a clear minimum, as illustrated for the
0.1−0.2 µ m particle diameter range in the top panel of Figure 3.2. The bottom panel of Figure 3.2 contains an illustration of the grid search method for
106
the same experiment and size range. Since emissions in each bin can influence concentrations in adjacent bins through coagulation, the optimization
was repeated, using the final values from one run as the starting values for
the next run, until the starting values remained unchanged for all size bins.
Step 3. Using the above steps and the DMPS data from three cigar experiments,
where the DMPS scan times were 10 min or less, size-specific mass emission
and deposition rates were obtained in twelve aggregated size ranges from
about 0.009 to 1.154 µ m. This range appeared to encompass most of the particle sizes present in SHS. Each of the three experiments showed that about
20% of the emitted particle mass was smaller than 0.1 µ m. The lower end of
observed SHS particle sizes was 0.02−0.03 µ m.
Step 4. The LASAIR-based optimization (step 2) was repeated using initial estimates from the results of step 3 for particles smaller than 0.1 µ m. The emissions and deposition rates for particles smaller than 0.1 µ m were optimized
by calculating the mean absolute deviation between observed and predicted
concentrations across all larger sizes. Initial guesses for the other size ranges
(> 0.1µ m) were obtained from the ending values in step 2. As in step 2, the
overall optimization process was repeated (from lowest to highest bin) until the parameter values remained unchanged. The surface for the “indirect”
optimization of emissions smaller than 0.1 µ m was irregular and much flatter
than for other particle sizes, indicating more uncertainty.
Step 5. After the optimization procedure was completed for each experiment, a
lognormal distribution was fit to the optimized mass emission rate to obtain
estimates for the mass median diameter (MMD) and geometric standard deviation (GSD) of the emissions size distribution. These fitted parameters are
the same whether for the distribution of TPM emissions (mg), mass emission
rate (mg/min), or mass-normalized emissions (mg/g-smoked).
Step 6. To estimate uncertainty in the lognormal size distribution parameters,
steps 2−5 were repeated for slightly perturbed initial values in each size
107
range, including particles smaller than 0.1 µ m. Owing to the large computational time required for a single optimization, it was impractical to conduct
a large number of optimization trials. A comparison of final optimization
results to initial values provides an approximate characterization of uncertainty.
The fits between LASAIR-observed and modeled time series data were generally good with minima for the mean absolute deviation in each bin ranging from
0.4 to 3 µ g/m3 . Figure 3.3 shows an example optimal fit of a modeled time series
to an observed time series for four particle diameter ranges. For bins between 0.1
and 1 µ m, the error across all experiments was between 2 and 12%. A linear regression of predicted time series values (dependent variable) against the observed
values (independent variable) for these bins yielded coefficients of determination,
i.e., r2 values, between 0.8 and 1 across all experiments, except for the 0.3−0.4 µ m
particle size range in two experiments where r2 values were 0.6 and 0.7.
Small systematic deviations in the fits are apparent at the beginning of a few
time series, where the observed particle loss appeared to be faster than later in the
time series. This “early decay” effect may be due to evaporative losses as suggested by Ingebrethsen and Sears [1989] and Hinds [1978]. Figure 3.4 shows an
example optimal fit for which particles in the 0.3−0.4 µ m particle diameter range
appear to undergo evaporation during the first 1−2 hours after smoking stopped
– a behavior that does not appear to be well-captured by the model. Incomplete
mixing is unlikely to explain the early decay, since not all size ranges exhibited this
behavior for a given experiment (e.g., the 0.2−0.3 µ m diameter range in Figure 3.4).
The model seemed to accurately account for the effect of particle coagulation,
which influenced concentrations in the smaller size bins for times as long as 4 h
after the source was extinguished. Second-order coagulation processes caused the
concentrations for the lower diameter ranges to actually increase after the source
was extinguished. This behavior is captured by the model and occurs as emissions
in smaller size ranges transfer particle mass into larger size ranges.
108
0.16
Deposition Loss−Rate Coefficient [1/h]
Minimum
0.14
0.12
0.10
46
48
50
54
52
0.12
0.10
0.08
0.06
25
30
35
40
45
50
55
Emission Rate [µg/min]
Figure 3.2: The top panel shows contours of an optimization surface for particles with diameters of 0.1−0.2 µ m (for the Cigarillo #2 experiment) over a range
of emission rates and deposition rates. The bottom panel depicts an optimization
pathway, illustrating how a local grid search method was used to find the minimum point on the surface and the optimal values of model input parameters, here
0.125 h−1 for deposition loss-rate coefficient and 50 µ g/min for emission rate. The
circle size is proportional to the mean absolute deviation between elements of the
observed and modeled time series, indicating the error of the model in fitting the
measurements at particular values of the parameters.
109
Particle Mass Concentration [µg/m³]
0
100
200
300
400
500
50
Particle Diameter:
0.2 − 0.3 µm
Particle Diameter:
0.1 − 0.2 µm
40
30
20
10
0
50
Particle Diameter:
0.4 − 1.0 µm
Particle Diameter:
0.3 − 0.4 µm
40
30
20
10
0
0
100
200
300
400
500
Elapsed Minutes, t
Figure 3.3: The optimal fit of the model (smooth curve) to the particle mass concentration time series observed during the Cigarillo #2 experiment (dots) for four
particle diameter ranges. Smoking began at time t = 0 and lasted approximately
15 min. From the time series shown, it appears that incomplete mixing was not an
issue as the model, when optimal parameters were used as input, provided good
fits to the observed data. Mixing also did not appear to be an issue for the other
experiments.
110
0
100 200 300 400 500
Particle Mass Concentration [µg/m³]
80
Particle Diameter:
0.3 − 0.4 µm
60
40
Particle Diameter:
0.2 − 0.3 µm
20
0
0
100 200 300 400 500
Elapsed Minutes, t
Figure 3.4: The optimal fit of the model (smooth curve) to the particle mass concentration time series observed during the Regular Cigar #3 experiment (dots) for
two particle diameter ranges. Smoking began at time t = 0 and lasted approximately 10 min. It appears that incomplete mixing cannot account for the observed
model-measurement discrepancy, since the 0.2−0.3 µ m time series does not display the rapid decrease in concentration during the first 60 min, which is apparent
in the 0.3−0.4 µ m time series. Since the model does not take evaporation into account, it is likely that the discrepancy arises from evaporative loss [Hinds, 1978;
Ingebrethsen and Sears, 1989].
111
3.4 The Size Distribution of Particle Emissions
Table 3.2 contains best estimates of the SHS particle mass emissions size distribution, based on the LASAIR data collected during the eight original experiments
described above. Figure 3.5 presents the lognormal fits to the optimization results
of each experiment. The MMD’s for the emissions are close to particle diameters of
0.2 µ m for all source types (x = 0.20 µ m; s = 0.017 µ m; COV = 8%), with the GSD’s
ranging from 1.9 to 3.1 (x = 2.3; s = 0.37; COV = 16%). SHS particle emissions appear mostly to have diameters between 0.02 and 2 µ m – with no clear difference in
the estimated mass size distributions between cigars and cigarettes.
The uncertainty in these estimates is highest for particles smaller than 0.1 µ m,
since they are based on an indirect fitting procedure. However, after repeating the
procedure for each experiment for different starting points, the fitted MMD and
GSD were in the ranges 0.16−0.23 µ m and 1.9−3.1, respectively. These represent
differences of 0−13% in MMD and 2−10% in GSD from the values in Table 3.2.
These differences are comparable to the coefficient of variation (COV) across all
experiments and source types for the final results stated above. In other words,
parameter uncertainty appears to be on the order of parameter variability. In contrast, the initial guess for the size distribution, which (using the LASAIR data and
assuming independence between bins) did not consider coagulation or particle
mass below 0.1 µ m, had MMD’s and GSD’s that ranged from 0.22−0.30 µ m and
1.4−2.1, respectively – differences of 14−39% and 10−34% when compared to the
final results. The MMD’s are higher and the GSD’s are smaller (i.e., the distribution is more narrow) than for our final estimates, because of the neglected particle
mass. These differences give an indication of the maximum error one would expect when using the raw LASAIR data to directly estimate SHS particle emissions
characteristics.
Since the current approach neglects evaporation, results may be influenced by
the evaporation of SVOC particle constituents. The true emissions may be larger
in magnitude and occur at larger particle sizes than we determined. However,
using the optimal values of emissions and deposition rate for input, the model
112
Table 3.2: The Estimated Size Distributions of SHS Particle Emissionsa
Integrated SHS Particle Emissionsb
Experimentc
MMD
GSD
[µ m]
TPM
Rate
Mass-Normalized
[mg]
[mg/min]
[mg/g-smoked]
Regular Cigar #1
0.18
2.5
8.8
0.71
5.2
Regular Cigar #2
0.21
2.2
6.7
0.46
4.6
Regular Cigar #3
0.20
2.4
6.3
0.61
3.3
Premium Cigar
0.18
1.9
4.7
0.35
3.7
Cigarillo #2
0.23
3.1
2.8
0.19
2.9
Cigarette #2
0.21
2.1
5.5
0.90
7.6
Cigarette #3
0.20
2.1
5.0
0.68
7.0
Cigarette #4
0.19
2.1
5.1
0.71
7.1
a These
estimates are based primarily on the LASAIR data with initial guesses for particle mass
smaller than 0.1 µ m determined from DMPS-based optimization results. The size metric is particle
diameter measured in µ m. MMD is the fitted mass median diameter of each size distribution and
GSD is the fitted geometric standard deviation.
b The “integrated” total particle mass (TPM), emission rate, and mass-normalized emissions were
obtained by integrating the estimated size distribution of total mass emissions and dividing by
unity, the smoking time, or the mass of tobacco consumed, respectively.
c Table 3.4 contains data on smoking time and tobacco mass consumed for each experiment and
the results of five additional experiments for which only filter-based, non-size-specific total mass
emissions were determined.
0
500
1000
1500
0
500
1000
1500
0.02
0.1
Cigarette #3
Cigarillo #2
Regular Cigar #3
Regular Cigar #1
2
Particle Diameter, Dp [µm]
1
Cigarette #4
Cigarette #2
Premium Cigar
Regular Cigar #2
0.1
1
2
0
500
1000
1500
0
500
1000
1500
Figure 3.5: The estimated size distribution of particle mass emission rate (in µ g/min) and the corresponding lognormal fits for eight experiments. The text in each panel gives the type of source used in each experiment. See
Table 3.2 for the fitted mass median diameter (MMD), geometric standard deviation (GSD), and integrated mass
emissions. The average MMD and GSD were 0.20 µ m and 2.3, respectively.
Mass Emission Rate,
dE/dlog(Dp) [µg/min]
0.02
113
114
provides a good fit to SHS particle concentrations for most size ranges and for
most times (with model errors generally near or below 10% for 0.1 - 1 µ m particles),
and is therefore judged to be an appropriate tool for predicting concentrations and
exposures. The agreement between the model and the observed concentrations
suggests that the evaporation of SVOCs from SHS particles may not be a very
large effect.
The optimized DMPS particle mass emissions distributions, which were used
to give initial estimates for particle sizes smaller than 0.1 µ m during the LASAIRbased optimization procedure, all had MMD’s of 0.20−0.22 µ m with geometric
standard deviations of 1.81−1.86. See Figure 3.6 for a sample fitted lognormal distribution for DMPS data. The GSD values, which are lower than our final LASAIRderived values, may be due to a small amount of neglected mass greater than 1 µ m
in diameter. Also, the largest size ranges measured by the LASAIR and DMPS have
more uncertainty associated with them than the middle size ranges.
Based on optimization, approximately 20% of the DMPS particle mass was
found to be emitted for sizes smaller than 0.1 µ m for each of the experiments.
To compare these results to a direct estimate from the DMPS data alone (i.e., without applying the model-based optimization procedure), we fit a lognormal distribution to the measured particle number size distribution from the DMPS data
that was collected just after the cigar or cigarette was extinguished. The resulting count median diameters (CMD’s) were 0.07−0.09 µ m and geometric standard
deviations were all near 1.8 with a corresponding mass median diameter (MMD)
of 0.20−0.25 µ m [Hinds, 1982]. These values for MMD and GSD, which neglect
particle transformation processes that may have occurred in the first few minutes
after the smoke was mixed, are reasonably close to the estimates obtained by using
the optimization procedure. These results provide a measure of self-consistency in
our approach, and they support the practice of using sufficiently time-resolved
concentration measurements by themselves to provide estimates of size-specific
SHS emissions.
Several previous investigators have studied tobacco smoke particle size distri-
115
Mass Emission Rate,
dE/dlog(Dp) [µg/min]
500
Regular Cigar #2
DMPS
400
MMD = 0.22 µm
GSD = 1.83
300
200
100
0
0.01
0.1
1
2
Figure 3.6: The fit of a lognormal distribution to the size distribution of mass emissions in µ g/min based on the DMPS data collected for the Regular Cigar #2 experiment. The fit shown is representative of the quality of the fits for the other two
experiments where DMPS emissions estimates were obtained. All DMPS-based
MMD’s were near 0.2 µ m and GSD’s were near 1.8. The proportion of mass smaller
than 0.1 µ m in these fits was used as the initial guess for model optimization with
the LASAIR data, which resulted in our best estimates of the size distribution of
particle mass emissions (see Table 3.2).
116
butions [Keith and Derrick, 1960; Chang et al., 1985; Ueno and Peters, 1986; Chung
and Dunn-Rankin, 1996], but these studies have mostly been focused on mainstream or sidestream smoke, rather than SHS. Those few investigators that have
studied SHS particles directly have tended to provide only a cursory examination
of the particle size distribution at a particular moment in time [Benner et al., 1989;
Kleeman et al., 1999].
The results for the size distribution of SHS particle emissions presented in this
section are in generally good agreement with the findings of other investigators
who, although they may not have examined SHS per se, have studied mainstream
or sidestream smoke, typically after it has been aged and/or diluted. Because concentrated tobacco smoke undergoes coagulation and, in addition, evaporation can
occur during dilution, the size distribution of the smoke is sensitive to experimental conditions. Therefore, investigations of fresh or diluted-and-aged mainstream
or sidestream smoke are not likely to give results identical to ours. In addition,
most of these investigations have used a non-model-based approach to estimate
the emissions size distribution.
As summarized in Table 3.3, MMD values reported in the literature are in the
approximate range of 0.3−0.7 µ m for mainstream smoke [Chang et al., 1985; Anderson et al., 1989; Chung and Dunn-Rankin, 1996], 0.2−0.5 µ m for sidestream
smoke [Ueno and Peters, 1986; Ingebrethsen and Sears, 1989; Chung and DunnRankin, 1996], and 0.2−0.5 µ m for SHS [Nazaroff et al., 1993b; Kleeman et al.,
1999], with reported GSD values in the range of 1.2−2.0. In spite of the variation in
these reported results, SHS particle emissions appear to have a fairly narrow and
identifiable distribution. Nearly all freshly dispersed SHS particle mass lies in the
diameter range of 0.02−2 µ m.
3.5 Size-Integrated Particle Emissions
Table 3.4 contains a summary of each original chamber experiment described
above and the filter-based results for size-integrated mass emission rates, massnormalized emissions, and total particle mass (TPM) emissions. Equivalent TPM
MS+SS
MS
SS
MS
MS
SS
MS
—
—
—
SS
MS
MS+SS
SS
MS
SS
MS
MS+SS
SS
MS+SS
Cigarettes
Present Work
Anderson et al. [1989]
Benner et al. [1989]
Chang et al. [1985]
”
Chung and Dunn-Rankin [1996]
”
Hinds [1978]
”
”
Ingebrethsen and Sears [1989]
Keith and Derrick [1960]
Kleeman et al. [1999]
Ishizu et al. [1978]
Okada and Matsunuma [1974]
”
Sextro et al. [1991] f
”
Ueno and Peters [1986]
Cigars
Present Work
5
1
13
44
—
—
2
6
4
—
10
1
—
—
—
6
3
—
3
—
3
1
1
1
2
2
1
1
1
—
1
1
1
5
5
—
—
1
2
7
Sampleb
N T
M, C, OPC, DM
M, C, OPC, DM
M, C, DM
M, C, DM
M, C, OPC
M, C, OPC
M, C, AC-NS
M, C, CI
M, C, AC
M, C, OPC, DM
M, C, CON
H, C, MOUDI
M, C, OPC
M, C, OPC
M, H, OPC
M, C, OPC, DM
M, C, OPC, DM
M, OPC, DM, CI
M, C, OPC, DM
M, C, DM
Methodc
20-50K
—
10 e
6−18 e
—
—
10
10−100
100−700
—
295
—
1000
1500
1500
—
—
6-18 e
20-50K
80K
Dilution
—
—
0.10
(0.0084)
0.11
0.23
0.22−26
0.27
0.15
—
—
—
0.1
0.23
—
0.1
0.17
0.10 – 0.12
—
—
0.10
0.20 (0.02)
0.26
0.26
0.25−0.30
0.5
0.7
0.52
0.39−0.52
0.37−0.38
0.20
—
0.3-0.4
—
—
—
0.22
0.48
0.16
0.20 (0.01)
0.38 (0.02)
2.4 (0.44)
—
1.23
1.19−1.27
1.6
2.0
1.37
1.38−1.49
1.31−1.37
—
—
—
1.5
1.5
1.4 – 1.6
—
—
1.4 – 1.7
2.1 (0.0)
2.0 (0.05)
Size distribution, mean (std. dev.)d
CMD
MMD
GSD
[µ m]
[µ m]
a Mainstream emissions (MS), sidestream emissions (SS), or both (MS+SS). b Total number of cigars or cigarette experiments (across all source types) (N); number of types of
cigars or cigarettes (T). c M = machine smoked; H = human smoked; C = chamber experiment; OPC = optical particle counter; DM = differential mobility analyzer; CON
= conifuge; CI = cascade impactor; AC = aerosol centrifuge; AC-NS = aerosol centrifuge in non-spectrometric mode; MOUDI = micro-orifice uniform-deposit impactor.
d Particle size distribution characteristics are as follows: CMD is the count median diameter, MMD is the mass median diameter, and GSD is the geometric standard
deviation. e Primary dilution ratio. f As reported by Nazaroff et al. [1993b].
Sourcea
Study
Table 3.3: Reported Size-Specific Tobacco Particle Emissions for Cigarettes and Cigars
117
118
emissions determined by integrating the particle mass size distributions (Table 3.2)
were generally lower than those determined using filters (Table 3.4) with the in-situ
real-time measure yielding total particle mass emissions that were 54−84% of the
filter-based emissions (absolute differences were 0.9−4.3 mg per cigarette or cigar).
The larger values for filters may be a consequence of the collection onto the filters,
by sorption or condensation, of SVOCs that are present in SHS. These vapor-phase
compounds are likely not detected by the real-time particle sizing instrumentation.
Since sidestream nicotine emission factors of 5−7 mg per cigarette have been reported [Daisey et al., 1998], which are relatively large for a single SVOC species, it
is plausible that the sorption of nicotine and other SVOCs onto filters could contribute to the observed discrepancies [Mader and Pankow, 2001].
The equivalent integrated emissions were consistently higher for cigarettes
(7−8 mg/g-smoked and 0.7−0.9 mg/min) than for cigars (3−5 mg/g-smoked and
0.2−0.7 mg/min). The total particle mass emitted by the cigarillos and premium
cigar and their emission rates were markedly lower than for the other types of
cigars and for cigarettes, although this finding may be an artifact of leakage around
the end-fittings during smoking (the cigarillos had plastic tips and the premium
cigar was rather bulky). The mass-normalized emissions (mg/g-smoked), which
may be more appropriate for direct comparisons, showed consistent results among
different types of cigars.
In alignment with the current work, Ueno and Peters [1986] and Chang et al.
[1985] report equivalent total mass emissions from real-time instruments (using
an electrical mobility analyzer and condensation nucleus counter) that are substantially smaller than determinations based on direct mass measurements (a cascade impactor in their case). Chang et al. [1985] found that when their primary
dilution ratio for mainstream smoke was increased from 6 to 18, the equivalent
TPM measured from their electrical mobility analyzer decreased dramatically (18
mg/cigarette down to 2.0 mg/cigarette) while the TPM measured with the cascade
impactor remained approximately the same (19−21 mg/cigarette). For sidestream
smoke, Ueno and Peters [1986] report cascade impactor TPM measurements of
119
Table 3.4: Summary of Cigar and Cigarette Experiments and Filter-Based SHS Particle Emissions
Smoking
Tobacco Mass
Duration
Consumed
TPM
Rate
Mass-Normalized
[min]
[g]
[mg]
[mg/min]
[mg/g-smoked]
Regular Cigar #1
12.5
1.71
10.8
0.86
6.3
Regular Cigar #2
14.8
1.46
8.1
0.55
5.6
Regular Cigar #3
10.3
1.92
9.5
0.92
4.9
Regular Cigar #4
11.3
1.41
11.8
1.04
8.4
Regular Cigar #5
14.0
1.50
7.3
0.52
4.8
Premium Cigar
13.4
1.26
5.6
0.42
4.5
Cigarillo #1
10.1
1.02
4.0
0.67
6.6
Cigarillo #2
14.8
0.96
4.0
0.27
4.1
Cigarillo #3
14.8
1.36
6.3
0.42
4.6
Cigarette #1
5.5
0.73
9.8
1.79
13.4
Cigarette #2
6.1
0.72
7.0
1.15
9.7
Cigarette #3
7.4
0.72
9.3
1.3
13.0
Cigarette #4
7.1
0.72
7.3
1.03
10.1
Experiment
a The
Filter-Based SHS Particle Emissionsa
total particle mass (TPM) emitted by a cigar or cigarette during each experiment was estimated from filter data by taking into account the loss of mass from deposition and ventilation that
occurred during sample collection. We estimated the effective total particle removal rate for each
experiment by fitting a line to the logarithm of the decaying total particle counts as measured by
the LASAIR. Emission rate and mass-normalized emissions were calculated by dividing the TPM
emissions by the smoking time or mass of tobacco consumed, respectively.
120
6.0−9.6 mg/cigarette across all primary dilution ratios (6−18) compared to equivalent TPM from the electrical mobility analyzer of 1.3−2.3 mg/cigarette. As far as
I know, these discrepancies have not been resolved and will require further investigation.
Gravimetrically determined values for SHS particle emission factors reported
in the literature are in the approximate range of 8−20 mg per cigarette smoked
[Hammond et al., 1987; Eatough et al., 1989; Löfroth et al., 1989; Hildemann et al.,
1991; Leaderer and Hammond, 1991; Koutrakis et al., 1992; Özkaynak et al., 1996;
Martin et al., 1997; Daisey et al., 1998] and approximately 6−50 mg/g-smoked
for cigars [CPRT Laboratories, 1990; Leaderer and Hammond, 1991; Nelson et al.,
1998, 1999; Klepeis et al., 1999b]. These ranges indicate substantial amounts of unexplained variability, probably stemming from different experimental conditions
and methodologies (e.g., sampling volume) in addition to the variety of tobacco
products that were used.
3.6 Particle Deposition
The particle deposition velocity is defined as the net flux density of particle mass
to a surface (mass per area per time) divided by the particle concentration in air
(mass per volume), giving it units of length per unit time [Nazaroff et al., 1993a].
The deposition velocity varies with particle diameter. If vd (s) is the deposition
velocity onto a surface at position s, then the area-weighted deposition velocity
over all room surfaces, vd , is defined as
1 R
S s
vd (s)ds. The deposition velocity can be
used to describe the first-order loss of particles to surfaces as follows:
S
dC
= −C vd
dt
V
(3.2)
where C is the room particle concentration [µ g m−3 ], V is the room volume [m3 ],
and S is the total room surface area [m2 ]. For simplication, Equation 3.2 is typically
written in terms of a deposition loss-rate coefficient defined as β =
S
V vd
for a spe-
cific room. Both Thatcher et al. [2002] and Lai [2002] present size-specific particle
deposition velocities and loss-rate coefficients from studies in residences.
MS+SS
SS
MS+SS
MS+SS
MS+SS
SS
MS+SS
MS+SS
MS+SS
SS
MS+SS
SS
Current Work
Daisey et al. [1998] f
Hammond et al. [1987]g
Hildemann et al. [1991]
Leaderer and Hammond [1991]h
Martin et al. [1997]i
Ott et al. [1992]
* Sextro et al. [1991] j
* Sextro et al. [1991] j
SS
SS
MS+SS
Cigarettes
* Current Work
* Ueno and Peters [1986]
Ueno and Peters [1986]
[Continued.]
Sourceb
Studya
1
1
3
2
6
4
4
10
1
50
50
50
1
6
2
Samplec
M, C, OPC,
DM
M, C, F, TPM
M, C, F, PM2.5
H, C, F, RSP
H, HD, F
H, C, F, RSP
M, C, F, TPM
H, C, F, RSP
H, C, PZ, RSP
H, C, MR, RSP
M, C, MR
M, C, OPC,
DM
M, C, OPC,
DM
M, OPC, DM
M, CI
Methodd
1.3 – 2.3
6.0 – 9.6
—
8.4 (1.4)
8.1 (2.0)
12.7 (2.1)
20 (4.2)
17 (2.1)
10
13.7 (4.1)
11.6 (3.6)
39.1 (1.7)
49
—
5.2 (0.3)
—
—
0.7
1.3 (0.33)
—
1.7 (0.28)
—
—
—
1.2
1.1
3.6
7.0
2.4
0.76 (0.12)
—
—
—
11.6 (1.9)
12.4 (1.3)
26 (4)
—
27 (3.4)
10 - 11
24.5
20.7
69.8
—
—
7.2 (0.3)
Particle emissions, mean (std. dev.)
Total
Rate
Normalized
(mg cig−1 )
(mg min−1 )
(mg g−1 )e
Table 3.5: Reported Environmental Tobacco Smoke Particle Mass Emissions from Cigarettes and Cigars
121
MS+SS
MS+SS
—
MS+SS
MS+SS
MS+SS
MS+SS
Cigars
* Current Work
Current Work
CPRT Laboratories [1990]o
Klepeis et al. [1999b]
Leaderer and Hammond [1991]
Nelson et al. [1998]
3
13
1
1
6
20
3
Samplec
1
1
2
178
178
M, C, OPC,
DM
M, C, F, TPM
—
H, PZ, RSP
H, C, F, RSP
H, C, F, RSP
H, C, F, RSP
Methodd
M, OPC, DM
M, CI
H, PZ, RSP
H, PM2.5
H, PM2.5
7.5 (2.8)
—
88
—
50 (24)
93 (41)
5.9 (2.3)
0.63 (0.26)
—
0.98
—
5.0 (2.4)
9.3 (4.1)
0.46 (0.21)
5.5 (1.4)
10.3 (2.4)
8.2
48 (9)
6.4 (4.1)
12 (5)
3.9 (0.9)
Particle emissions, mean (std. dev.)
Total
Rate
Normalized
2.0 – 18.5
—
—
19.1 – 20.7
—
—
—
1.43 (0.01)
—
12.7
—
—
13.8 (3.6)
—
—
a An asterisk (*) indicates that size-specific SHS particle emissions were measured in this study. b Mainstream (MS), sidestream (SS), or both kinds of emissions (MS+SS)
were studied. c The number of different experimental sites or individual types of tobacco sources used. d Key: H = human smokers; M = machine-smoked; C = chamber
experiments; HD = emissions collected in a hood; F = filter-based sampling; OPC = optical particle counter; DM = differential mobility analysis; MR = Miniram optical
scattering monitor; PZ = piezobalance; RSP = respirable suspended particulate matter; PM2.5 = particulate matter smaller than 2.5 µ m in diameter; TPM = total particulate
matter; CI = cascade impactor. e Particle mass emitted per unit mass of tobacco product combusted. f Results shown for six commercial cigarettes, 62.5% of top-selling
California cigarettes, ca. 1990. g 40 cigarettes of each type were smoked for each experiment. Emission factors were calculated from information presented in the paper.
h
The cigarette types smoked in this study represent 48% of the US market, ca. 1987. Results summarized for US commercial cigarettes only. Danish cigarettes and research
cigarette (Kentucky 1R 4F) are omitted from this table. Forty cigarettes of each type were smoked for each experiment. i This study presents weighted results for 50 top-selling
US cigarette brands, which comprise 65% of the US cigarette market ca. 1991. Each cigarette was tested twice with an 11 min smoking duration and about 0.56 grams of
tobacco consumed. j As reported in Nazaroff et al. [1993b]. k The range of results shown is due to varying dilution ratios with more dilution leading to a lower equivalent
yield for DM measurements. l This study estimated the average total particle emission rate for cigarettes from real-time measurements in airport smoking lounges over a
period of 2-3 h. m Emissions were estimated in the cited reference by fitting nonlinear regression model to average PM2.5 concentrations for 178 homes with valid data out
of 394 (excluding homes with pipe or cigar smoking and fireplaces) total in New York state. n Emissions were estimated in the cited reference by fitting nonlinear regression
model to average PM2.5 concentrations for 178 homes in Riverside, CA. o As cited in NCI [1998], pp 169 and 178.
Sourceb
MS
MS
MS+SS
MS+SS
MS+SS
Studya
* Chang et al. [1985]k
Chang et al. [1985]k
Klepeis et al. [1996]l
Koutrakis et al. [1992]m
Özkaynak et al. [1996]n
122
123
Thatcher et al. [2002] present results from eight studies conducted in full-sized
rooms or buildings where particles with diameters of 0.1−0.5 µ m had deposition
loss-rate coefficients ranging from approximately 0.01 to 1 h−1 . The wide range
of particle loss rates likely reflects the diversity of air speed and furnishing levels
present for each set of experimental conditions as well as variation in measurement
techniques. Thatcher et al. also present new data on the effect that air speed and
room furnishings have on indoor particle deposition rates (Figure 3.7). The minimum particle size they measured was 0.55 µ m, which had deposition loss rates
of 0.09−0.27 h−1 depending on the degree of furnishing and changing air speed.
The largest size corresponding to tobacco smoke particles, i.e., approximately 2
µ m, had deposition rates ranging from about 0.4 to 1 h−1 . Their general finding
was that airspeed and furnishings caused deposition rates to vary for a particular
particle size by a factor of approximately 2−3, whereas, for a given set of air speed
and furnishing conditions, the deposition rate varied by a factor of about 50 across
particle sizes (0.5−10 µ m). Thatcher et al. note that because air exchange rates in
dwellings are typically greater than 0.1 h−1 (see Chapter 5 on page 165), 0.1 h−1
may be the minimum particle deposition rate that deserves close attention.
Xu et al. [1994] measured the effect of air speed, but not furnishings, on sizespecific deposition loss rates, specifically for SHS particles. They found a general
increase of deposition rate as a function of air speed with rates of 0.006−0.3 h−1
for all measured particle sizes (about 0.06−2 µ m) and 0.006−0.1 h−1 for particles
in the 0.1−1 µ m range (Figure 3.8). In light of the results of Thatcher et al., if
furnishings were present, then the deposition rates might be expected to lie in the
range 0.01−0.9 h−1 .
Figure 3.9 presents estimates of deposition loss-rate coefficient across six particle diameter ranges and corresponding to the eight experiments in Table 3.2. Results for the lowest and highest diameter ranges were most uncertain because of
either an indirect optimization approach (0.02−0.1 µ m) or measurement scatter
from sparse particle counts (1−2 µ m). Also in Figure 3.9 are SHS deposition lossrate coefficients determined by Xu et al. [1994] in chamber experiments where four
124
Deposition Loss−Rate Coefficient [h−1]
Deposition Loss−Rate by Furnishing Level and Air Speed
0.5
1
Bare
10
Carpeted
Furnished
6
1
0.1
0.5
1
10
0.5
1
10
Particle Diameter [µm]
0 cm/s
5.4 cm/s
14.2 cm/s
19.1 cm/s
Deposition Velocity by Furnishing Level and Air Speed
0.5
Deposition Velocity [m h−1]
Bare
1
10
Carpeted
Furnished
6
1
0.1
0.03
0.5
1
10
0.5
1
10
Particle Diameter [µm]
0 cm/s
5.4 cm/s
14.2 cm/s
19.1 cm/s
Figure 3.7: Particle deposition loss-rate (top) and deposition velocity (bottom) determined by Thatcher et al. [2002] for a variety of air speeds and furnishing levels.
125
Particle loss rate coefficient [h −1 ]
0.4
0.1
Fan Speeds
0 rpm
430 rpm
2000 rpm
3070 rpm
0.01
0.1
1
Particle diameter, Dp [µm]
Figure 3.8: Particle deposition rates for SHS measured by Xu et al. [1994] in a lowventilation chamber for four different fan speeds.
126
small wall fans were operated over a range of different speeds. Their results for
the case of the maximum fan speed of 3070 rpm, which is similar to the speed of
the six wall fans used in my experiments, are shown.
Since this dissertation is concerned with simulating total SHS particle exposures, it is important to estimate a total deposition loss-rate across all SHS particle
sizes. Since deposition velocity is defined in terms of the mass flux to a surface, the
total deposition rate is calculated by integrating the deposition loss-rate coefficient
with respect to size, weighting the integrand by the SHS mass particle size distribution. The use of a total (size-integrated) deposition rate can be problematic,
because the mass particle size distribution may change over time. For example,
under low air-exchange conditions, the size distribution may become appeciably
narrowed before particles are completely cleared due to preferential removal of
ultrafine and coarse particles.
Based on the results discussed above, the particle deposition loss-rate rates for
sizes relevant to the bulk of SHS particles, and occurring in homes that are furnished and have typical residential air speeds, are likely to be in the range of 0.01
to around 0.5 h−1 . Using the estimated mass size distribution for SHS particle
emissions from the current work, which has a GM of approximately 0.2 µ m and a
GSD of approximately 2, in combination with the reported size-specific deposition
rates presented above, the total deposition loss-rate is likely to be approximately
0.1 h−1 or less, which is near the lower end of deposition rates that would strongly
compete with the ventilatory removal of SHS particles.
3.7 Emissions and Dynamic Behavior of Gaseous
Species
In Chapter 2, I presented information on the particulate and gaseous composition
of SHS. While particles comprise a substantial portion of total SHS emissions, they
contain many toxic species, and they have the ability to deposit deeply in the lung,
a variety of toxic gases constitute a large proportion of SHS emissions. These gases
fall roughly into two types: (1) those that interact fairly strongly with the surround-
127
Deposition Loss−Rate Coefficient [1/h]
10
Present Work
Xu et al. (1994) − fans at 3070 rpm
1
0.1
0.01
0.02
0.1
1
2
Figure 3.9: Size-specific particle deposition rates estimated in the current work.
Each box represents the range (top and bottom limits) and median (center line) of
the deposition rate across a given diameter range. Dashed lines indicate results
with higher associated uncertainty. The filled circles represent the results of Xu
et al. [1994] for experiments in which four small fans were operating at 3070 rpm,
air speed conditions similar to my experiments.
128
ing environment; and (2) those with relatively low reactivity, either with surfaces
or other airborne species.
In contrast to SHS particles, which undergo irreversible deposition, nicotine
and other volatile SHS species sorb rapidly to indoor surfaces [Löfroth, 1993;
Van Loy et al., 1998; Piadé et al., 1999]. Over time, they accumulate and may desorb from surfaces, reentering indoor air over time. Nicotine makes up the largest
proportion of volatile organic species in SHS, a fact that has contributed to its extensive use as an SHS tracer in indoor air and exposure studies (see Chapter 2).
Some of the measured emission factors associated with nicotine for cigarettes are
given in Table 3.6. A variety of investigators have reported “effective” emission
factors specifically for nicotine (see Table 3.6) that implicitly incorporate surface
sorption (1−3 mg cig−1 ). A few have estimated the true mass yield of nicotine
emissions from cigarettes (5−7 mg cig−1 ).
Singer et al. [2003] report exposure relevant emission factors (EREF’s) for nicotine
and other volatile organic species present in SHS. These EREF’s represent the mass
available per cigarette for daily inhalation exposure taking into account adsorption and reemission processes. Of the 29 compounds they studied, only nicotine
and 3-ethenylpyridine were markedly affected by surface sorption processes. The
eight compounds that were found to have EREF’s of 500 µ g cig−1 or larger at a
ventilation rate of 0.6 h−1 were formaldehyde (950), acetaldehyde (2360), acrolein
(610), isoprene (2950), acetonitrile (1080), toluene (990), 3-ethenylpyridine (530),
and nicotine (1660).
Van Loy et al. [2001] report on an empirical study of the interactions between
nicotine and common elements of residential environments. They find that a linear kinetic model provided a good fit to their data for experiments involving carpet
and wallboard, so that the dynamics of nicotine can be captured with linear sorption and desorption coefficients. Their estimated values for these coefficients were
5.3 m h−1 and 0.00012 h−1 , respectively. The sorption coefficient is analagous to the
particle deposition velocity described above in Equation 3.2. I incorporate these
parameters into the simulation model described in Chapter 6 to explore the effect
129
of nicotine sorption and desorption processes on residential exposure to SHS. The
IAQ model equations I use to describe the dynamics of SHS nicotine are presented
and discussed in Appendix B.
Carbon monoxide (CO) is a nonreactive SHS gas, which does not react with
surfaces or other SHS components. Its removal is driven solely by ventilation.
Therefore, CO-specific terms in indoor air quality models are limited to the source
term, so that varying emission rates for CO, or for a completely different nonreactive species, can be easily treated.
CO is one of the primary products of incomplete combustion processes and,
in general, its measurement has been standardized with convenient techniques.
As a result, many measurement studies of CO have been conducted. As evident
from the results of CO emissions characterization studies for cigars and cigarettes
listed in Table 3.7, CO contributes more to SHS on a mass basis than particles.
Cigarette emissions of CO are approximately 40−80 mg cig−1 and cigar emissions
are 160−1200 mg cig−1 .
Much of this chapter is devoted to a model-based method for estimating sizespecific particle mass emission factors for indoor sources, and to reporting the
results of applying this method to data from a set of original cigar and cigarette
chamber experiments. The model was fit successfully to the observed data. Figure 3.10 contains illustrative fits of a lognormal distribution to the predicted mass
size distribution of SHS particles for one of the experiments. The estimated particle loss-rates are shown in the figure inset. The curves for two different times
show how the particle size distribution evolves due to the processes of ventilation,
deposition, and coagulation.
Full knowledge of the size distribution of SHS particle emissions is essential in
understanding their dynamics and in predicting important health related behavior, such as lung deposition or removal by portable or structural filtration technology. However, the focus of this dissertation is on using a simulation model to
130
Table 3.6: Reported Environmental Tobacco Smoke Nicotine Emissions from
Cigarettes and Cigarsa
Emissions
Source
Sample
Method
[mg cig−1 ]
SS
6
M, C
5.4
MS+SS
4
H, C
1.2
Löfroth et al. [1989]b,c
SS
1
M, C
0.80, 3.3
Eatough et al. [1989]a
MS+SS
50
H, C
7.3
Leaderer and Hammond
MS+SS
10
H, C
1.2
MS+SS
50
H, C
1.6
SS
6
M, C
0.92
MS+SS
2
M, C
0.94
Singer et al. [2002]b
SS
−
M, C
0.4−3.8
Singer et al. [2003]b
SS
−
M, C
0.8−3.1
MS+SS
5
M, C
0.66
Study
Cigarettes
Daisey et al. [1994]a
Hammond et al. [1987]b
[1991]b
Martin et al. [1997]b
Daisey et al.
[1998]b
Klepeis et al.
[1999a]b
Cigars
Klepeis et al. [1999a]b
a See
notes for Table 3.5.
values reflect an attempt to measure the total nicotine SS yield.
b Reported values are effective emission factors, which implicitly incorporate the effect of nicotine
sorption onto surfaces.
c The first, lower reported value corresponds to a series of experiments in which more adsorbent
surfaces were present, including two persons, a television, crib, chair, and curtain, and the relative humidity was 50−60%. The humidity was only 30% in the second series of experiments and
furnishings were absent, which is reflected in the lower effective emission factor.
a Reported
131
Table 3.7: Reported Environmental Tobacco Smoke Carbon Monoxide Emissions
from Cigarettes and Cigarsa
Emissions
Source
Sample
Method
[mg cig−1 ]
Current Work
MS+SS
3
M, C
78 (4.7)
Rickert et al. [1984]
MS+SS
2
H
41−67
SS
1
M, C
67
Ott et al. [1992]
SS
1
M, C
65.8
Klepeis et al. [1996]
MS+SS
2
H
83
Martin et al. [1997]
MS+SS
50
H, C
55
MS+SS
6
H, C
55
Current Work
MS+SS
5
M, C
161 (28)
MS+SS
6
H, C
432
Klepeis et al. [1999b]
MS+SS
H
630−1200
MS+SS
H, C
873 (511)
Study
Cigarettes
Cigars
a See
notes for Table 3.5.
20
132
400
300
Dep. loss coeff. [h−1 ]
Particle mass conc., dM / dlog(Dp) [µg m −3]
Air Exchange Rate: 0.03 h−1
1
0.1
0.01
1
0.1
470 Minutes
After
Smoking
Dp [µm]
200
1 Minute
After
Smoking
100
0
0.1
1
Particle diameter, Dp [µm]
Figure 3.10: The mass size distributions of SHS particles shortly after a cigarette
was smoked in a 20 m3 room and after nearly eight hours had elapsed. The bellshaped curves are fitted lognormal distributions. The earlier, more broad distribution has MMD = 0.23 µ m, GSD = 2.2, and a total mass concentration of 300 µ g m−3 .
The later, more narrow distribution has MMD = 0.29 µ m, GSD = 1.4, and a total
mass concentration of 100 µ m m−3 . The size-resolved particle deposition rates,
determined for this particular experiment, are shown in the upper left inset. In
addition to deposition onto surfaces, particles were removed at a ventilation rate
of λ = 0.03 h−1 . The particle size distribution becomes narrower in time because
larger and smaller particles deposit more quickly than those near the mode of the
distribution (0.2 – 0.4 µ m). In addition to the effects of deposition, coagulation
causes the mode to shift toward larger sizes.
133
explore the effects of human location, and door and window positions, on SHS
exposure, rather than on SHS particle dose or the performance of particle filters.
Including size information in the simulation of SHS exposure will not significantly
enhance the study of these factors, and would, instead, potentially obscure the
main findings, adding complexity in the analysis of simulation results, and requiring considerably more overhead in computer resources. Therefore, to simplify the
simulation of residential SHS exposure in this dissertation, I limit my treatment to
size-integrated emission factors and dynamic characteristics. I use the results of
original research and reviews of prior work presented in this chapter to calibrate
my simulation model with size-integrated values for particle-related parameters.
While not incorporated directly in the simulation model, the size distribution of
SHS particles informs the selection of integrated parameter values. Based on the
results of chamber studies, I estimate the total loss-rate coefficient for irreversible
deposition of SHS particles to be approximately 0.1 h−1 . From a variety of studies,
I find the total SHS particle emissions of a cigarette to be close to 10 mg cig−1 .
I reviewed the literature for values of CO and nicotine SHS emissions and dynamic parameters so that exposure to these species can also be simulated. For
these gas-phase species, I find that SHS emissions aer approximately 50 and 5 mg
cig−1 , respectively. Carbon monoxide is a nonreactive gas, while nicotine strongly
reacts with surfaces. To allow for an accounting of the sorption and desorption of
vapor-phase nicotine to and from residential surfaces, I use the reported sorption
and desorption coefficients of 5.3 m h−1 and 0.00012 h−1 as parameters in a linear
model, which also depends on the surface-to-volume ratio of the modeled indoor
space.
Two final emissions-related parameters that are required to simulate residential SHS exposure are the per-cigarette duration, which is close to 10 min, and
the number cigarettes that a typical medium-level smoker consumes, which has a
rough mean of 1−2 packs (20−40 cigarettes) per day.
134
3.9 References
Anderson, P. J., Wilson, J. D., and Hiller, C. F. (1989). Particle size distribution
of mainstream tobacco and marijuana smoke. American Review of Respiratory
Disease, 140: 202–205.
Benner, C. L., Bayona, J. M., Caka, F. M., Tang, H., Lewis, L., Crawford, J., Lamb,
J. D., Lee, M. L., Hansen, L. D., and Eatough, D. J. (1989). Chemical composition
of environmental tobacco smoke. 2. Particulate-phase compounds. Environmental Science and Technology, 23(6): 688–699.
Bohren, C. and Huffman, D. R. (1983). Absorption and Scattering of Light by Small
Particles. John Wiley and Sons, New York.
Chang, P. T., Peters, L. K., and Ueno, Y. (1985). Particle size distribution of mainstream cigarette smoke undergoing dilution. Aerosol Science and Technology, 4:
191–207.
Chung, I. and Dunn-Rankin, D. (1996). In situ light scattering measurements of
mainstream and sidestream cigarette smoke. Aerosol Science and Technology, 24:
85–101.
CPRT Laboratories (1990). Development of Methods for the Characterization of Toxic
Constituents in Cigars, Pipe Tobacco, and Smokeless Tobacco. SSSC Contract H40781-C230/01-SS, Tobacco Branch, Health Canada. Cited on pp. 169 and 178 of NCI
(1998).
Daisey, J. M., Mahanama, K. R. R., and Hodgson, A. T. (1994). Toxic Volatile Organic Compounds in Environmental Tobacco Smoke: Emission Factors for Modeling
Exposures of California Populations. Final Report Contract A133-186, California
Air Resources Board, Sacramento. LBL No. 36379.
Daisey, J. M., Mahanama, K. R. R., and Hodgson, A. T. (1998). Toxic volatile organic
compounds in simulated environmental tobacco smoke. Journal of Exposure Analysis and Environmental Epidemiology, 8(3): 313–334.
Eatough, D. J., Benner, C. L., Bayona, J. M., Richards, G., Lamb, J. D., Lee, M. L.,
Lewis, E. A., and Hansen, L. D. (1989). Chemical composition of environmental
tobacco smoke. 1. Gas-phase acids and bases. Environmental Science and Technology, 23(6): 679–687.
Garvey, D. M. and Pinnick, R. G. (1983). Response characteristics of the Particle
Measurement Systems Active Scattering Aerosol Spectrometer Probe (ASASPX). Aerosol Science and Technoogy, 2: 477–488.
135
Hammond, S. K., Leaderer, B. P., Roche, A. C., and Schenker, M. (1987). Collection
and analysis of nicotine as a marker for environmental tobacco smoke. Atmospheric Environment, 21(2): 457–462.
Hildemann, L. M., Markowski, G. R., and Cass, G. R. (1991). Chemical composition
of emissions from urban sources of fine organic aerosol. Environmental Science &
Technology, 25(4): 744–759.
Hinds, W. C. (1978). Size characteristics of cigarette smoke. American Industrial
Hygiene Association Journal, 39: 48–54.
Hinds, W. C. (1982). Aerosol Technology: Properties, Behavior, and Measurement of
Airborne Particles. John Wiley and Sons, New York.
Ingebrethsen, B. J. and Sears, S. B. (1989). Particle evaporation of sidestream tobacco smoke in a stirred tank. Journal of Colloid and Interface Science, 131(2): 526–
536.
Ishizu, Y., Ohta, K., and Okada, T. (1978). Changes in the particle size and the
concentration of cigarette smoke through the column of a cigarette. Journal of
Aerosol Science, 9: 25–29.
Keith, C. and Derrick, J. (1960). Measurement of the particle size distribution and
concentration of cigarette smoke by the "conifuge". Journal of Colloid Science, 15:
340–356.
Kleeman, M. J., Schauer, J. J., and Cass, G. R. (1999). Size and composition distribution of fine particulate matter emitted from wood burning, meat charbroiling,
and cigarettes. Environmental Science & Technology, 33(20): 3516–3523.
Klepeis, N. E., Apte, M. G., Gundel, L. A., Nazaroff, W. W., and Sextro, R. G.
(1999a). Characterizing ETS emissions from cigars: Chamber measurements of
nicotine, particle mass, and particle size. In Indoor Air ’99. Proceedings of the 8th
International Conference on Indoor Air Quality and Climate, Volume 2, pages 903–
908, Edinburgh, Scotland.
Klepeis, N. E., Ott, W. R., and Repace, J. L. (1999b). The effect of cigar smoking on
indoor levels of carbon monoxide and particles. Journal of Exposure Analysis and
Klepeis, N. E., Ott, W. R., and Switzer, P. (1996). A multiple-smoker model for predicting indoor air quality in public lounges. Environmental Science & Technology,
30(9): 2813–2820.
136
Kopstein, A. (2001). Tobacco Use in America: Findings from the 1999 National Household Survey on Drug Abuse (Analytic Series A-15). DHHS Publication No. SMA
02-3622, Substance Abuse and Mental Health Services Administration, Office of
Applied Statistics, Rockville, MD.
Kousaka, Y., Okuyama, K., and Wang, C. (1982). Response of cigarette smoke
particles to change in humidity. Journal of Chemical Engineering of Japan, 15(1):
75–76.
Koutrakis, P., Briggs, S. L. K., and Leaderer, B. P. (1992). Source apportionment
of indoor aerosols in Suffolk and Onondaga counties, New York. Environmental
Science & Technology, 26(3): 521–527.
Lai, A. C. K. (2002). Particle deposition indoors: A review. Indoor Air, 12: 211–214.
Leaderer, B. P. and Hammond, S. K. (1991). Evaluation of vapor phase nicotine
and respirable suspended particle mass as markers for environmental tobacco
smoke. Environmental Science and Technology, 25: 770–777.
Lipowicz, P. J. (1988). Determination of cigarette smoke particle density from mass
and mobility measurements in a Millikan cell. Journal of Aerosol Science, 19(5):
587–589.
Löfroth, G., Burton, R. M., Forehand, L., Hammond, S. K., Seila, R. L., Zweidinger,
R. B., and Lewtas, J. (1989). Characterization of environmental tobacco smoke.
Environmental Science and Technology, 23(5): 610–614.
Mader, B. T. and Pankow, J. P. (2001). Gas/solid partitioning of semivolatile organic
compounds (SOCS) to air filters. 3. An analysis of gas adsorption artifacts in
measurements of atmospheric SOCS and organic carbon (OC) when using teflon
membrane filters and quartz fiber filters. Environmental Science and Technology,
35(17): 3422–3432.
Martin, P., Heavner, D. L., Nelson, P. R., Maiolo, K. C., Risner, C. H., Simmons, P. S.,
Morgan, W. T., and Ogden, M. W. (1997). Environmental tobacco smoke (ETS):
A market cigarette study. Environment International, 23(1): 75–90.
McRae, D. D. (1982). The refractive-index of individual cigarette-smoke droplets.
Journal of Colloid and Interface Science, 87(1): 117–123.
policies and attitudes about secondhand smoke – United States, 2000. Morbidity
and Mortality Weekly Report, 50(49): 1101–1106.
137
Nazaroff, W. W. and Cass, G. R. (1989). Mathematical modeling of indoor aerosol
dynamics. Environmental Science & Technology, 23(2): 157–166.
Nazaroff, W. W., Gadgil, A. J., and Weschler, C. J. (1993a). Critique of the use of
deposition velocity in modeling indoor air quality. In Nagda, N., editor, Modeling
of Indoor Air Quality and Exposure, ASTM STP-1205, pages 81–104, Philadelphia,
PA. American Society for Testing and Materials.
Nazaroff, W. W., Hung W. Y., Sasse, A. G. B. M., and Gadgil, A. J. (1993b). Predicting regional lung deposition of environmental tobacco smoke particles. Aerosol
Science and Technology, 19(3): 243 – 254.
NCI (1996). Monograph No. 7, The FTC Cigarette Test Method for Determining Tar,
Nicotine, and Carbon Monoxide Yields of U.S. Cigarettes: Report of the NCI Expert
Committee. NIH Publication Number 96-4028, National Institutes of Health, National Cancer Institute, Bethesda.
NCI (1998). Monograph No. 9, Cigars: Health Effects and Trends. NIH Publication Number 98-4302, National Institutes of Health, National Cancer Institute,
Bethesda.
Nelson, P. R., Kelly, S. P., and Conrad, F. W. (1998). Cigars as a source of environmental tobacco smoke. Journal of Aerosol Science, 29(Supplement 1): S1307–S1308.
Nelson, P. R., Kelly, S. P., Conrad, R. W., Richardson, J., and Risner, C. H. (1999).
Characterization of environmental tobacco smoke from cigars. In Indoor Air ’99.
Proceeding of the 8th International Conference on Indoor Air Quality and Climate, Volume 2, pages 925–929. Edinburgh, Scotland.
Okada, T. and Matsunuma, K. (1974). Determination of particle-size distribution
and concentration of cigarette smoke by a light-scattering method. Journal of
Colloid and Interface Science, 48(3): 461–469.
Ott, W. R., Langan, L., and Switzer, P. (1992). A time series model for cigarette
smoking activity patterns: Model validation for carbon monoxide and respirable
particles in a chamber and an automobile. Journal of Exposure Analysis and Environmental Epidemiology, 2(Suppl 2): 175–200.
138
Piadé, J. J., D’Andrés, S., and Sanders, E. B. (1999). Sorption phenomena of nicotine
and ethenylpyridine vapors on different materials in a test chamber. Environmental Science and Technology, 33: 2046–2052.
Rickert, W. S., Robinson, J. C., and Collishaw, N. (1984). Yields of tar, nicotine, and
carbon monoxide in sidestream smoke from 15 brands of Canadian cigarettes.
American Journal of Public Health, 74: 228–231.
Sextro, R. G., Gross, E., and Nazaroff, W. W. (1991). Determination of Emissions
Profiles for Indoor Particle Phase Environmental Tobacco Smoke. Presented at
the 1991 Annual Meeting of the American Association for Aerosol Research. Traverse City, MI.
Singer, B. C., Hodgson, A. T., and Nazaroff, W. W. (2003). Gas-phase organics in environmental tobacco smoke: 2. Exposure-relevant emission factors and indirect
exposures from habitual smoking. Atmospheric Environment, 37: 5551–5561.
Thatcher, T. L., Lai, A. C. K., Moreno-Jackson, R., Sextro, R. G., and Nazaroff, W. W.
(2002). Effects of room furnishings and air speed on particle deposition rates
indoors. Atmospheric Environment, 36(11): 1811–1819.
TI (1997). The Tax Burden on Tobacco: Historical Compilation, Volume 32. The Tobacco
Institute, Washington, DC.
Ueno, Y. and Peters, L. K. (1986). Size and generation rate of sidestream cigarette
smoke particles. Aerosol Science and Technology, 5: 469–476.
Van Loy, M. D., Nazaroff, W. W., and Daisey, J. M. (1998). Nicotine as a marker
for environmental tobacco smoke: Implications of sorption on indoor surface
materials. Journal of the Air and Waste Management Association, 48(10): 959–968.
Van Loy, M. D., Riley, W. J., Daisey, J. M., and Nazaroff, W. W. (2001). Dynamic behavior of semivolatile organic compounds in indoor air. 2. Nicotine and phenanthrene with carpet and wallboard. Environmental Science & Technology, 35(3):
560–567.
139
Xu, M. D., Nematollahi, M., Sextro, R. G., Gadgil, A. J., and Nazaroff, W. W. (1994).
Deposition of tobacco smoke particles in a low ventilation room. Aerosol Science
and Technology, 20(2): 194–206.
140
Chapter 4
Human Activity Patterns
This chapter is devoted to characterizing the residential time-location patterns for
occupants of individual US households. In general, two essential ingredients for
the occurrence of residential exposure to SHS are: (1) the existence of SHS pollutant concentrations at a particular point in time and space in the house; and (2) the
presence of a person at approximately the same point (see Section 2.1.2). Because
of the key importance of the timing and location of exposed individuals in different
areas of a house, human activity patterns play a central role in simulating residential exposures to SHS. Since SHS pollutants are assumed to be well mixed within
individual rooms (see Chapter 5), spatial resolution for occupant location is only
needed at the room level. In the current work, I use the measured room-specific
time-location patterns described in this chapter, along with pollutant-specific parameters and housing characteristics, to explore frequency distributions of human
exposure to residential SHS (see Chapters 8 and 9).
In addition to their own location in the house, a nonsmoker’s exposure to emissions from cigarettes and cigars in homes is also strongly modulated by the location of smoking activity and various activities performed by each occupant in different rooms. The relative location of smokers and the nonsmoker, both during
and shortly after smoking activity, determines the intensity of direct nonsmoker
exposure to SHS emissions. By altering air flow patterns in the house, specific
kinds of occupant behavior influence the dispersion, dilution, and persistence of
SHS throughout a home. Relevant activities include door and window positions
CHAPTER 4. HUMAN ACTIVITY PATTERNS
141
and the operation of a centralized air handling system. These activities influence
direct exposure to emissions as well as more indirect exposures that may occur
some time after smoking has ceased or in areas away from active smokers.
At one extreme of potential SHS exposure, either no one smokes in the home or
no one other than the smoker is present, in which case receptor SHS exposure
is exactly zero. Towards the other extreme, the smoker and nonsmoker are at
home continuously and occupying the same room. For intermediate situations,
the smoker and nonsmoker occupy different rooms of the house for different time
periods, which may be coincident, overlapping, or nonoverlapping.
The most detailed and representative human activity and location study conducted for the US population is the National Human Activity Pattern Survey
(NHAPS), which was sponsored by EPA and carried out in the early-to-mid 1990’s
[Klepeis et al., 2001, 1996; Tsang and Klepeis, 1996]. NHAPS was patterned after a
previous set of studies conducted on the California population [Jenkins et al., 1992;
Wiley et al., 1991a,b]. The NHAPS repondents comprise a representative crosssection of 24-h daily activity patterns in the contiguous US. The 9,386 NHAPS respondents, who were interviewed by telephone, gave a minute-by-minute diary
account of their previous day’s activities, including the places they visited and the
presence of a smoker in each location. Detailed information was provided on the
rooms that each respondent visited while in residences, whether their own or one
they were visiting.
Thus far, there has been no detailed analysis or use of the NHAPS data with
respect to time-location profiles in individual rooms of homes, although a number
of modeling efforts make use of broad location categories, such as home, work,
school, and automobile (see Section 2.5.2). Since NHAPS contains the precise sequence and duration of room-to-room human behavior for a large sample of people, it represents a rich resource for use in understanding exposures to a variety of
pollutants in the residential indoor environment for which a single 24-h period is
an appropriate time scale, e.g., for SHS exposure.
NHAPS is a limited resource, since the interaction of smoking individuals, non-
142
smoking individals, and the house environment constitute a complex ecological
web, which cannot be fully characterized by independent activity profiles from
unassociated individuals. Coupled location and activity information for multiple
residents of a single household, which could be used to understand how relationships among individuals in a home can influence exposures, is missing from the
NHAPS data. However, the degree of confluence between multiple residents and
the resulting effect on exposure can be addressed in a simulation model to some
degree by selecting two individuals from the database who have certain relative
characteristics, such as a particular age, gender, and total number of minutes spent
at home, and time spent together in the same room. This approach is discussed
further in Chapter 6.
In addition, NHAPS data do not contain activity information on flow-related
activities, such as window, door, air handling, and filtration practices. This information is also generally unavailable from other sources. Using the residential
exposure simulation model developed in the current research, I systemically evaluate the effect of different flow-related activities on SHS exposure. This behavior is either superimposed upon the unmodified time-location profiles reported
by the NHAPS respondents, or the time-location patterns may be modified to remove time spent by the smoker and nonsmoker in the same room or to restrict the
smoker to particular rooms.
In this chapter, I conduct a detailed analysis based entirely on on the 24-h
NHAPS room-specific human location patterns in terms of the overall time spent
in different rooms of one’s own residence, including time spent in the presence of
a smoker. For analyses conditioned on residential locations, I only consider those
respondents who were reported to live in detached houses. To place the total time
spent at home into context, the first section below contains a brief analysis of broad
locations a person may visit throughout their day. Examples of raw activity pattern
data are included in Appendix A.
The statistics I present consist of aggregate time spent over a single 24-h period, including the sample size of people who report visiting a particular location
143
(N), the mean time calculated over all respondents – regardless of whether or not
they visited that particular location – the sample size of people who visited the
location (Doer N), the percentage of people who visited the location (Doer %), and
the mean time spent by those who visited the location. In addition, I calculate the
mean percentage of time spent in a given location with respect to the whole set
of possible locations, where the mean percentage of time spent across all locations
sums to 100%. For time spent exposed to SHS or in residential locations, I calculate the mean percentage of time spent by averaging over the percentage for each
individual, rather than by dividing the mean time spent in each location by the
mean overall time spent in all locations.1 Note that post-stratification weights as
described by Klepeis et al. [2001] were used in calculating 24-h statistics. For timeof-day analysis, I calculate the fraction of individuals who were in each of a set
of house locations (rooms) across multiple, consecutive 1-h time periods. During
each 1-h period, I select the house location that was associated with the most time
for a given individual as being representive of the entire 60-min period, although
an individual might also have spent sizeable fractions of time in another location.
4.1 Time Spent in Broad Locations Over a 24-h Period
As demonstrated by the results of the NHAPS analysis presented in Tables 4.1 and
4.2, the home is undeniably the location where one spends the bulk of one’s life.2
All but a very small percentage of sampled Americans spent time in their own
home on the day just before they were interviewed, being at home for a mean time
of more than 16 hours, or
2
3
of the day.
NHAPS contains data on the reported time spent in different locations while
in the presence of an active smoker, which is an indirect measure of exposure,
and therefore only indicates potential exposure (see Section 2.1.2). The reported
1 The
two calculations are only different if the total time spent in all categories changes between
respondents, such as for time spent exposed to SHS in different locations or for the time spent in
residential locations. The total time spent in an exhaustive set of NHAPS diary categories is 24-h
for all respondents.
2 Note that NHAPS is biased against people not living in homes with telephones. It omits people
who are homeless, on vacation, or who may be institutionalized or in the military.
144
presence of exposure is used here as a relative measure of exposure in different
locations. The measure may be useful to estimate exposure prevalence in different
locations, but bias may be inherent in the use of this measure of SHS exposure
to estimate exposure duration.3 Respondents can still be exposed to SHS when
no smokers are present. In addition a respondent may report the presence of a
smoker during an extended time period when active smoking only occurred for a
very brief time.
Circa 1992-94, when NHAPS was conducted, approximately 26% of respondents reported being in the presence of a smoker in their own home for a mean
time of more than 6 h, which was, on average, nearly 43% of the total time spent in
the presence of a smoker over a single day. While smoking prevalence has shifted
and there is currently more awareness of the dangers of SHS exposure, a significant number of households still contain residents who are exposed to SHS (see
Section 2.4.1). The particular household smoking rules that may be applied in individual houses has likely changed, but the room-specific rates of smoker presence
reported below give an indication of those rooms where residents are most likely
to be exposed.
It is worth noting that while in recent years exposure to SHS in office and bar
or restaurant settings may be diminishing, exposure occurring in vehicles and outdoors is probably at least the same and has likely increased as more smokers may
be forced to shift their smoking behavior to alternate locations. In 1992-94, 83%
of Americans reported spending some time in a car, with nearly 15% reporting the
presence of a smoker in that location for a mean time of more than one hour, which
was, on average, 15% of the total time spent in any location where a smoker was
present.
3 Due
to the imprecision of smoker-present data in NHAPS, and their receptor orientation, they
cannot directly be used to simulate individual smoking events in a household. Therefore, the simulation model presented in Chapter 6 simulates smoker activity using independent data on the
duration of smoking events and the number of cigarettes smoked in a day, superimposing these
data on a particular individual’s NHAPS time-location profile.
145
Table 4.1: Overall Weighted Statistics for Time Spent by NHAPS Respondents and
Time Spent in the Presence of a Smoker in Six Different Grouped Locations Over a
24-h Period Starting at 12:00 AM on the Diary Daya
Mean
Location
N
Doer
Time
Doer
Doer
Mean
[min]
%
N
[min]
Overall Time Spent
In a Residence
9196
990
99.4
9153
996
Office-Factory
9196
78
20.0
1925
388
Bar-Restaurant
9196
27
23.7
2263
112
Other Indoor
9196
158
59.1
5372
267
In a Vehicle
9196
79
83.2
7596
95
Outdoors
9196
109
59.3
5339
184
Time Spent with a Smoker
All Locations
9196
163
43.8
3949
372
In a Residence
9196
78
25.6
2331
305
Office-Factory
9196
16
4.3
394
363
Bar-Restaurant
9196
14
10.0
951
143
Other Indoor
9196
19
7.6
725
247
In a Vehicle
9196
11
14.5
1340
79
Outdoors
9196
24
11.4
1038
213
a Means and percentages have been calculated using sample weights, whereas the sample sizes N
and Doer N are raw counts.
146
Table 4.2: Weighted Statistics for Mean Percentage of Overall Time Spent and Time
Spent with a Smoker by NHAPS Respondents in Six Different Grouped Locations
Over a 24-h Period Starting at 12:00 AM on the Diary Daya
Mean Time
Mean Time
with a Smoker
%
%
In a Residence
68.7
42.7
Office-Factory
5.4
7.2
Bar-Restaurant
1.8
14.6
11.0
12.1
In a Vehicle
5.5
8.7
Outdoors
7.6
14.7
100.0
100.0
Location
Other Indoor
Total
a The
overall mean percentage time spent was calculated by dividing the mean number of minutes
spent by NHAPS respondents in each location by the total time spent on the diary day, i.e., 24-h =
1440 min. The mean percentage time spent with a smoker was calculated by dividing the time spent
with a smoker for each NHAPS respondent in each location by the total time spent by the same
respondent with a smoker on the diary day, and then averaging this value over all respondents.
147
4.2 Time Spent at Home in Different Rooms
A standout feature of time spent at home over the course of a day is that almost
98% of Americans spend time in the bedroom for a mean time of more than 9
hours, which is 58% of the time spent, on average, in any location in or around the
house (see Table 4.3). Consistent with the amount of time spent in the bedroom
is the fact that over 6% of Americans were exposed to SHS in the bedroom for
over 3 hours, on average, which is more than 15% of the total time they spent, on
average, being exposed in the home (see Table 4.4). The residential location with
the highest average percentage of time spent in the presence of a smoker was the
living room with 46%, although only an average of 19% of the total time was spent
in the living room. Besides the bedroom and the living room, the only other room
with a proportion of the total time spent in the presence of a smoker over 10% was
the kitchen with 13%, although half of this proportion of time was spent overall
(7%). Taken together, the kitchen, living room, and bedroom account for over 85%
of the total time spent at home. These three rooms account for 74% of the time
spent in the presence of a smoker, with 9% spent when moving from room-to-room
and almost 8% spent in an area just outside of the house.
From Figure 4.2 and the accompanying legend for different rooms in Figure
4.1, it is apparent that the largest fraction of individuals are in the bedroom until
about 9 AM and after 11 PM, as might be expected. During the middle of the day,
and especially between 6 PM and 10 PM, more people are in the kitchen and living
room than in any other room of the house, although about 40−60% of people are
away from home between the hours of 9 AM and 6 PM. The presence of a smoker
in the morning occurs primarily in the bedroom, shifting to the living room and the
kitchen during midday. In the afternoon and evening, more than twice as much
time spent in a smoker’s presence occurs in the living room than for any other
single room, although by midnight the bedroom has progressed to a comparable
level.
The duration of time spent in location episodes, i.e., continuous portions of
time in a given room, is relevant to the time scales over which persons might be
5895
5895
5895
5895
5895
5895
5895
5895
5895
5895
5895
5895
5895
5895
Living, Family, Den
Dining Room
Bathroom
Bedroom
Study, Office
Garage
Basement
Utility, Laundry
Pool, Spa
Yard, Outdoors
Room to Room
In and Out of House
Other, Verified
Refused to Answer
15
86
390
2393
1693
59
314
218
162
254
5756
4181
1150
4796
4548
N
0.3
1.5
6.6
40.6
28.7
1.0
5.3
3.7
2.7
4.3
97.6
70.9
19.5
81.4
77.2
%
Doer
0.3
1.9
6.3
54.6
40.2
1.0
3.9
5.2
3.2
9.8
547.4
24.5
13.8
199.5
75.3
[min]
Mean Time
131.4
129.1
94.5
134.5
140.1
98.4
72.7
141.4
117.2
227.1
560.6
34.5
70.6
245.2
97.6
[min]
Mean Time
0.0
0.2
0.6
5.0
3.6
0.1
0.4
0.5
0.3
0.9
58.0
2.7
1.3
19.3
7.2
%
Mean Timea
Time % was calculated by averaging over the individual percentages of time spent in each residential location reported by each
NHAPS respondent, rather than from the calculated mean time spent in each location.
a Mean
5895
N
Kitchen
Location
Doer
Doer
Table 4.3: Overall Statistics for Time Spent by NHAPS Respondents Living in Detached Homes in Different Rooms
of Their Residence Over a 24-h Period Starting at 12:00 AM on the Diary Day
148
5895
5895
5895
5895
5895
5895
5895
5895
5895
5895
5895
5895
5895
5895
Living, Family, Den
Dining Room
Bathroom
Bedroom
Study, Office
Garage
Basement
Utility, Laundry
Pool, Spa
Yard, Outdoors
Room to Room
In and Out of House
Other, Verified
Refused to Answer
0
9
29
284
177
7
23
14
17
31
364
108
93
828
486
N
0.0
0.2
0.5
4.8
3.0
0.1
0.4
0.2
0.3
0.5
6.2
1.8
1.6
14.0
8.2
%
Doer
0.0
0.4
0.8
7.2
5.1
0.1
0.3
0.4
0.4
1.1
12.5
0.6
1.3
29.4
7.2
[min]
Mean Time
252.2
164.6
148.7
169.9
111.4
79.1
186.4
141.8
200.2
202.3
34.8
83.7
209.6
87.4
[min]
Mean Time
0.0
0.5
0.9
9.3
7.7
0.3
0.5
0.6
0.6
1.2
15.4
1.5
2.5
45.9
13.0
%
Mean Timea
respondent, rather than from the calculated mean time spent in each location.
a Mean Time % was calculated by averaging over the individual percentages of time spent in residential locations reported by each NHAPS
5895
N
Kitchen
Location
Doer
Doer
Table 4.4: Overall Statistics for Time Spent by NHAPS Respondents in the Presence of a Smoker Living in Detached
Homes in Different Rooms of Their Residence Over a 24-h Period Starting at 12:00 AM on the Diary Day
149
150
Residential Locations Visited
Kitchen
Living Room, Family Room, Den
Dining Room
Bathroom
Bedroom
Study, Office
Garage
Basement
Utility Room, Laundry Room
Pool/Spa (Outdoors)
Yard/Other Outside House
Moving From Room to Room
Moving In and Out of House
Other Verified
Refused to Answer
Figure 4.1: Legend for the plots of the hourly fraction of time that NHAPS respondents spent in different rooms of their house presented in Figures 4.2−4.5.
0.0
0.2
0.4
0.6
0.8
Mid
6A
Time of Day
Noon
6P
Residential Locations
Mid
0.00
0.02
0.04
0.06
0.08
0.10
0.12
Mid
6A
Time of Day
Noon
6P
Mid
Residential Locations with SHS Exposure
Fraction of Individuals
Figure 4.2: Stacked bar charts showing the overall fraction of NHAPS respondents living in detached houses who
spent time in various locations (left), and who spent time in the presence of a smoker in various locations (right),
in and around their home during each hour of the day.
1.0
151
152
Table 4.5: Statistics for Time Spent by NHAPS Respondents Living in Detached
Homes During Continuous Individual Episodes in Different Rooms of Their Residence Over a 24-h Period Starting at 12:00 AM on the Diary Day
Percentiles
Location
25th
50th
75th
Mean
[min]
[min]
[min]
[min]
40
34.4
Kitchen
15
30.0
Living, Family, Den
30
60
120
91
Dining Room
20
30
60
43
Bathroom
10
15
30
21
Bedroom
30
105
300
170
Study, Office
60
90
180
126
Garage
16
60
120
91
Basement
30
60
120
93
Utility, Laundry
15.0
40
90
59
Pool, Spa
35.5
78
135
100
Yard, Outdoors
30.0
60
120
85
Room to Room
30.0
60
118
84
In and Out of House
15.0
30
120
80
Other, Verified
20
45
90
70
Refused to Answer
48
75
180
170
in close proximity to a smoker in a given room of the house. The distribution of
time spent in individual episodes for each residential location is summarized in
Table 4.5. Excluding the bedroom, the mean time spent in rooms ranged from
about 20 min in the bathroom to 126 min in a study or office. The mean time per
episode spent in the bedroom was 170 min, but this likely reflects two different
periods of continuous time just before or just before midnight.
153
4.2.1 Time Spent by Age
Persons over the age of 65, many of whom are presumed to be retired, reported
spending the least percentage of time in the bedroom of any age group at 47% and
the largest percentage of time, and absolute length of time, in the living room and
kitchen at 26% and 10%, respectively. Persons of working age (between 18 and 65)
had the second to least percentage of time in the bedroom at 58% (see Table 4.6).
While children under age 5 spent a somewhat smaller percentage of time in the
bedroom than older children aged 5−18, they spent the greatest absolute length of
time in the bedroom, on average, at more than 12 h.
It is evident from Figure 4.3 that some of the time young children under age
5 spent in the bedroom can be accounted for by midday naps between approximately 1 and 5 PM, whereas most older children and adults aged 18−65 are out
of the house during this time period. Adults over the age of 65 spent comparable
amounts of time at home as young children do, but their time is predominantly
spent in the living room, kitchen, and moving about the house. The large amount
of time that older adults spent in the kitchen corresponds to three rather large and
distinct peaks of activity approximately centered around 8AM, 1PM, and 6PM, and
corresponding to 20% or more of respondents.
4.2.2 Time Spent by Gender
The differences in location patterns for males and females are slight (see Table
4.7). Both groups spent approximately 57−59% of their time at home in the bedroom, on average, and 18−20% of their time in the living room. However, females
spent more than 8% of their time, corresponding to about 1.9 hours, at home in the
kitchen, whereas males spent aboue 6%, or 1.3 h, on average. Females also spent
slightly more time moving about the home, whereas males spent slightly more
time outside in the yard.
The time-location profiles of males and females are quite similar (Figure 4.4),
except that males spent slightly less time at home than females in the middle of the
day, and females have a noticeably higher peak of activity in the kitchen around
154
Table 4.6: Statistics by Age for Time Spent by NHAPS Respondents Living in Detached Homes in Different Rooms of Their Residence Over a 24-h Period Starting
at 12:00 AM on the Diary Day
Doer
Doer
Doer
Mean Time
Mean Time
Mean Timea
N
N
%
[min]
[min]
%
Kitchen
321
224
69.8
54.8
78.6
4.5
Living, Family, Den
321
262
81.6
187.8
230.1
15.7
Dining Room
321
75
23.4
14.3
61.3
1.2
Bathroom
321
210
65.4
23.9
36.5
2.0
Bedroom
321
312
97.2
721.7
742.5
62.9
Study, Office
321
3
0.9
1.6
166.7
0.1
Garage
321
4
1.2
1.0
83.5
0.1
Basement
321
5
1.6
1.5
95.0
0.1
Utility, Laundry
321
3
0.9
0.7
75.0
0.1
Pool, Spa
321
6
1.9
2.1
110.0
0.2
Yard, Outdoors
321
111
34.6
47.8
138.2
3.8
Room to Room
321
157
48.9
102.3
209.1
8.1
In and Out of House
321
22
6.9
10.7
155.5
0.9
Other, Verified
321
2
0.6
0.9
145.0
0.1
Refused to Answer
321
1
0.3
0.1
40.0
0.0
Kitchen
499
345
69.1
42.8
61.9
4.3
Living, Family, Den
499
405
81.2
141.1
173.8
13.5
Dining Room
499
140
28.1
15.0
53.5
1.5
Bathroom
499
348
69.7
19.9
28.5
2.1
Bedroom
499
489
98.0
652.8
666.1
68.2
Study, Office
499
3
0.6
0.1
24.0
0.0
Garage
499
5
1.0
0.7
70.0
0.1
Basement
499
21
4.2
6.1
145.5
0.6
Utility, Laundry
499
3
0.6
0.6
105.7
0.0
Pool, Spa
499
9
1.8
2.3
127.8
0.2
Yard, Outdoors
499
200
40.1
58.7
146.4
5.2
Room to Room
499
172
34.5
36.5
105.9
3.4
In and Out of House
499
29
5.8
8.5
146.0
0.8
Other, Verified
499
5
1.0
0.4
39.2
0.0
Refused to Answer
499
0
0.0
0.0
Age
0 −5
5−12
Location
Continued.
0.0
155
Doer
Doer
Doer
Mean Time
Mean Time
Mean Timea
N
N
%
[min]
[min]
%
Kitchen
447
307
68.7
35.8
52.1
4.0
Living, Family, Den
447
355
79.4
163.0
205.2
17.1
Dining Room
447
81
18.1
9.8
53.8
1.0
Bathroom
447
334
74.7
20.4
27.3
2.3
Bedroom
447
439
98.2
619.1
630.4
69.3
Study, Office
447
10
2.2
2.8
126.0
0.2
Garage
447
12
2.7
2.2
80.8
0.2
Basement
447
21
4.7
4.3
91.3
0.5
Utility, Laundry
447
5
1.1
0.7
64.8
0.1
Pool, Spa
447
4
0.9
0.9
96.2
0.1
Yard, Outdoors
447
122
27.3
29.7
108.9
2.9
Room to Room
447
117
26.2
20.0
76.5
2.0
In and Out of House
447
15
3.4
2.1
62.1
0.2
Other, Verified
447
2
0.4
0.1
32.5
0.0
Refused to Answer
447
1
0.2
0.4
180.0
0.0
Kitchen
3719
2875
77.3
74.5
96.4
7.5
Living, Family, Den
3719
2969
79.8
184.4
230.9
18.9
Dining Room
3719
671
18.0
12.5
69.5
1.2
Bathroom
3719
2769
74.5
25.8
34.6
3.1
Bedroom
3719
3623
97.4
513.2
526.8
57.8
Study, Office
3719
193
5.2
12.6
243.5
1.2
Garage
3719
105
2.8
3.8
133.7
0.3
Basement
3719
138
3.7
5.6
150.9
0.6
Utility, Laundry
3719
254
6.8
5.0
72.8
0.5
Pool, Spa
3719
32
0.9
0.9
102.2
0.1
Yard, Outdoors
3719
946
25.4
35.7
140.5
3.3
Room to Room
3719
1483
39.9
50.9
127.7
5.0
In and Out of House
3719
258
6.9
4.9
70.5
0.5
Other, Verified
3719
56
1.5
2.0
135.0
0.2
Refused to Answer
3719
7
0.2
0.3
144.3
0.0
Age
12−18
18−65
Location
Continued.
156
Doer
Age
65+
a Mean
Doer
Doer
Mean Time
Mean Time
Mean Timea
N
N
%
[min]
[min]
%
Kitchen
909
797
87.7
123.0
140.3
10.1
Living, Family, Den
909
805
88.6
315.6
356.4
26.2
Dining Room
909
183
20.1
19.8
98.5
1.6
Bathroom
909
520
57.2
23.9
41.8
2.1
Bedroom
909
893
98.2
532.4
541.9
46.5
Study, Office
909
45
5.0
9.7
196.4
0.8
Garage
909
36
4.0
3.6
91.4
0.3
Basement
909
33
3.6
5.0
137.6
0.4
Utility, Laundry
909
49
5.4
3.8
70.7
0.3
Pool, Spa
909
8
0.9
0.4
42.0
0.0
Yard, Outdoors
909
314
34.5
50.9
147.4
4.0
Room to Room
909
464
51.0
79.6
156.0
6.6
In and Out of House
909
66
7.3
11.1
153.0
0.9
Other, Verified
909
21
2.3
3.3
142.4
0.3
Refused to Answer
909
6
0.7
0.8
123.5
0.1
Location
Time % was calculated by averaging over the individual percentages of time spent in residential locations reported by each NHAPS respondent, rather than from the calculated mean time
spent in each location.
157
Residential Locations by Age
Mid 6A Noon 6P Mid
0−5
5−12
12−18
1.0
0.8
0.6
0.4
0.2
0.0
18−65
65+
1.0
0.8
0.6
0.4
0.2
0.0
Mid 6A Noon 6P Mid
Time of Day
Figure 4.3: Stacked bar charts, grouped by age, showing the fraction of NHAPS
respondents living in detached houses who spent time in various locations in and
around their home during each hour of the day.
158
6PM. The time that males and females spent at home during midday is associated
with yard activities more than at any other time of day, whereas for females there
is more activity moving from room to room in the house during midday than for
any other time of day.
4.2.3 Time Spent by Day of Week
As with differences according to gender, differences between 24-h aggregate time
spent into different home locations by day of the week are small (see Table 4.8). As
expected, respondents spent more mean absolute time in their bedrooms (9.7 hours
versus 9 hours), living rooms (4.6 hours versus 3.9 hours), and kitchens (1.7 hours
versus 1.6 hours) on weekends than on weekdays, although the mean percentage
of time spent in these locations was comparable.
On weekends, there are slightly fewer people in their bedrooms in the early
morning hours, probably due to late-night activities and the peak of people spending time in the kitchen in the morning is shifted from about 8AM to about 9AM
relative to weekdays. The time normally spent in the middle of the day out of the
house on weekdays, e.g., at work or school, is made up largely by time spent in
the living room, the laundry room, or outdoors in the yard.
4.2.4 Time Spent by House Size
The most likely house layout for NHAPS repondents was a single-storied residence with five or six rooms (Table 4.9). Two-storied residences most commonly
had six to eight rooms. While large residences are likely to have multiple bedrooms and living areas, NHAPS respondents only reported spending time in a
single generic room of each type. Therefore, as might be expected in this situation,
the time-location patterns in different sized houses are essentially indistinguishable. Irregularities only arise between groups when samples sizes are very small,
which is the case for homes with a small number of rooms, especially when combined with two or more stories.
159
Table 4.7: Statistics by Gender for Time Spent by NHAPS Respondents Living in
Detached Homes in Different Rooms of Their Residence Over a 24-h Period Starting at 12:00 AM on the Diary Day
Gender
Female
Male
Doer Mean
Time
Meana
Time
Doer
Doer
Mean
Time
N
N
%
[min]
[min]
%
Kitchen
3129
2565
82.0
91.4
111.5
8.4
Living, Family, Den
3129
2519
80.5
199.5
247.8
18.3
Dining Room
Bathroom
3129
3129
631
2222
20.2
71.0
15.0
25.4
74.3
35.8
1.4
2.7
Bedroom
Study, Office
3129
3129
3073
97
98.2
3.1
562.8
6.4
573.1
207.2
57.4
0.5
Garage
Basement
3129
3129
59
113
1.9
3.6
1.3
4.6
68.0
128.4
0.1
0.4
Utility, Laundry
Pool, Spa
3129
3129
263
36
8.4
1.2
6.0
1.2
70.9
102.1
0.5
0.1
Yard, Outdoors
Room to Room
3129
3129
784
1589
25.1
50.8
29.3
73.9
116.9
145.6
2.6
6.7
In and Out of House
Other, Verified
3129
3129
246
51
7.9
1.6
7.1
1.9
90.0
114.0
0.6
0.2
Location
Refused to Answer
3129
8
0.3
0.3
122.5
0.0
Kitchen
Living, Family, Den
2766
2766
1983
2277
71.7
82.3
57.1
199.6
79.6
242.4
5.8
20.3
Dining Room
Bathroom
2766
2766
519
1959
18.8
70.8
12.4
23.4
66.0
33.1
1.2
2.7
Bedroom
Study, Office
2766
2766
2683
157
97.0
5.7
529.8
13.6
546.2
239.3
58.7
1.3
Garage
2766
103
3.7
5.4
145.4
0.5
Basement
Utility, Laundry
2766
2766
105
51
3.8
1.8
5.9
1.5
155.3
82.0
0.6
0.2
Pool, Spa
Yard, Outdoors
2766
2766
23
909
0.8
32.9
0.8
52.6
92.5
160.1
0.1
4.7
Room to Room
In and Out of House
2766
2766
804
144
29.1
5.2
32.7
5.3
112.5
102.2
3.2
0.5
Other, Verified
2766
35
1.3
1.9
151.1
0.2
Refused to Answer
2766
7
0.3
0.4
141.6
0.0
a Mean Time % was calculated by averaging over the individual percentages of time spent in residential locations reported by each NHAPS respondent, rather than from the calculated mean time
160
Residential Locations by Gender
Mid
6A
FEMALE
Noon
6P
Mid
MALE
1.0
0.8
0.6
0.4
0.2
0.0
Mid
6A
Noon
6P
Mid
Time of Day
Figure 4.4: Stacked bar charts, grouped by gender, showing the fraction of NHAPS
respondents living in detached houses who spent time in various locations in and
around their home during each hour of the day.
161
Table 4.8: Statistics by Day of Week for Time Spent by NHAPS Respondents Living
in Detached Homes in Different Rooms of Their Residence Over a 24-h Period
Starting at 12:00 AM on the Diary Day
Doer
Day
Weekend
Weekday
Mean
Time
Meana
Time
Doer
Doer
Mean
Time
Location
N
N
%
[min]
[min]
%
Kitchen
Living, Family, Den
1939
1939
1435
1568
74.0
80.9
76.7
222.4
103.7
275.1
7.0
20.6
Dining Room
Bathroom
1939
1939
395
1282
20.4
66.1
15.5
24.2
75.9
36.6
1.4
2.6
Bedroom
Study, Office
1939
1939
1884
59
97.2
3.0
566.8
7.4
583.4
243.6
56.3
0.7
Garage
Basement
1939
1939
67
73
3.5
3.8
3.7
5.5
105.9
146.3
0.3
0.5
Utility, Laundry
1939
90
4.6
3.8
82.4
0.3
Pool, Spa
Yard, Outdoors
1939
1939
21
628
1.1
32.4
1.1
52.4
97.9
161.9
0.1
4.5
Room to Room
In and Out of House
1939
1939
740
146
38.2
7.5
54.8
7.6
143.7
100.7
4.9
0.7
Other, Verified
Refused to Answer
1939
1939
28
5
1.4
0.3
2.5
0.4
171.2
164.0
0.2
0.0
Kitchen
3956
3113
78.7
74.6
94.8
7.3
Living, Family, Den
Dining Room
3956
3956
3228
755
81.6
19.1
188.3
12.9
230.7
67.8
18.6
1.3
Bathroom
Bedroom
3956
3956
2899
3872
73.3
97.9
24.6
537.8
33.6
549.5
2.8
58.9
Study, Office
Garage
3956
3956
195
95
4.9
2.4
10.9
3.0
222.0
125.1
1.0
0.3
Basement
Utility, Laundry
3956
3956
145
224
3.7
5.7
5.1
3.9
138.9
68.8
0.5
0.4
Pool, Spa
Yard, Outdoors
3956
3956
38
1065
1.0
26.9
0.9
34.3
98.6
127.2
0.1
3.1
Room to Room
In and Out of House
3956
3956
1653
244
41.8
6.2
54.5
5.6
130.3
90.9
5.1
0.5
Other, Verified
3956
58
1.5
1.6
108.8
0.1
Refused to Answer
3956
10
0.3
0.3
115.1
0.0
Time % was calculated by averaging over the individual percentages of time spent in residential locations reported by each NHAPS respondent, rather than from the calculated mean time
a Mean
162
Residential Locations by Day of Week
Mid
6A
WEEKEND
Noon
6P
Mid
WEEKDAY
1.0
0.8
0.6
0.4
0.2
0.0
Mid
6A
Noon
6P
Mid
Time of Day
Figure 4.5: Stacked bar charts, grouped by day of the week, showing the fraction of
NHAPS respondents living in detached houses who spent time in various locations
in and around their home during each hour of the day.
163
Table 4.9: Sample Size by Number of Rooms and Floors for NHAPS Respondents
Living in Detached Homes
Number of Rooms
No. Floors
1
2
3
4
5
6
7
8
9
10
1
10
27
156
351
861
894
473
231
98
50
2
0
7
51
89
293
479
415
399
221
144
3
0
0
9
19
53
121
136
147
89
72
The human activity data presented in this chapter represents a rich data resource
for use in simulating exposure to SHS in homes. The data contain event time series
for a large and representative sample of Americans, recording their whereabouts
in their homes for a single day. These data form a foundation of behavior patterns for persons occupying a home, providing the substrate upon which smoking
behavior, air-flow-related behavior, and various mitigation strategies, can be superimposed. They provide a unique means to incorporate realistic variation in
human time-location patterns into a simulation, which can drive exploration of
factors causing variation in exposure. Since there is a natural diurnal cycle in human behavior, and smoking and SHS exposure events take place on times scales
that are less than a single day, the 24-h cross-section inherent in the NHAPS data
is suitable for the analysis of SHS exposure in homes. Although the data do not
contain information for multiple persons in a given household, thereby capturing
intrinsic interdependencies in behavior for the sampled population, the data can
be resampled to match individuals to one another that have varying amounts of
time spent together in different rooms of the house. In this way the effect on exposure of time-varying proximity to the smoker can be explored. The rooms where
the most time is spent, in order from greatest to smallest, are the bedroom, living
room, and kitchen, where the percentage of time spent in each of these rooms diminishes by a factor of three in turn. The most time spent exposed to SHS occurs in
164
the living room with approxiatly a third as much time spent exposed in the kitchen
and bedroom.
4.4 References
Jenkins, P., Phillips, T., Mulberg, E., and Hui, S. (1992). Activity patterns of Californians: Use of and proximity to indoor pollutant sources. Atmospheric Environment, 26A(12): 2141–2148.
11(3): 231–252.
Klepeis, N. E., Tsang, A. M., and Behar, J. V. (1996). Analysis of the National Human
Activity Pattern Survey (NHAPS) Responses from a Standpoint of Exposure Assessment. EPA/600/R-96/074, US EPA, Washington D. C.
Tsang, A. M. and Klepeis, N. E. (1996). Descriptive Statistics Tables from a Detailed Analysis of the National Human Activity Pattern Survey (NHAPS) Data.
EPA/600/R-96/148, US EPA, Washington D. C.
Wiley, J., Robinson, J. P., Cheng, Y., Piazza, T., Stork, L., and Pladsen, K. (1991a).
Study of Children’s Activity Patterns. California Air Resources Board, Contract No.
A733-149, Sacramento, CA.
Wiley, J., Robinson, J. P., Piazza, T., Garrett, K., Cirksena, K., Cheng, Y., and Martin, G. (1991b). Activity Patterns of California Residents. California Air Resources
Board, Contract No. A6-177-33, Sacramento, CA.
165
Chapter 5
Housing Characteristics
The purpose of this chapter is to establish typical or reasonable physical and enviromental parameters for houses located in the US. These parameters, consisting of
house dimensions (volumes and surface areas), single-zone mixing rates, outdoor
air-exchange rates, open-window air flow rates, and interzonal air flow rates, are
to be used either directly or indirectly in this dissertation as input into a multicompartment indoor air quality model. This model is the centerpiece of a larger
simulation model, which I use to calculate exposure to residential secondhand tobacco smoke (SHS) using both scripted and measured distributions of residential
human location patterns (see Chapters 7, 8, and 9).
One of the most important housing characteristics, and one which determines
the general validity of most indoor air quality (IAQ) models, is the rapidity with
which pollutants become mixed within a single zone. Generally, mixing within a
single zone is assumed to be so rapid that emitted pollutants can be considered instantaneously distributed throughout that zone. In contrast, mixing between zones
may be impeded by contricted passageways, or closed doorways, causing differences in interzonal concentrations. Various studies have established the importance of the degree of interzonal flow, especially with respect to door position, in
affecting inter-room concentration differences. Therefore, in addition to the issue
of single-zone mixing, the key characteristics of houses with respect to in-house exposures are expected to be the mere existence of multiple distinct rooms coupled
with relatively low inter-room air flow rates. Other housing characteristics, such
CHAPTER 5. HOUSING CHARACTERISTICS
166
as zone size, surface areas, and leakage rates, are also expected to affect exposures.
However, their effects are fairly well understood.
Since the approach I take in this dissertation is one of exploration rather than
exhaustive prediction, I seek central estimates for each quantitative parameter
rather than the comprehensive characterization of distributions. For analyses in
subsequent chapters, I select a residential environment and surrounding characteristics that are reasonably reflective of those in the US housing stock and use
them as a fixed platform upon which to conduct simulation trials. The purpose of
these trials is a better understanding of how changes in exposure occur in response
to variation in human location patterns and flow-influencing behavior within a
multizone context. Although the underlying focus is on detached, full-size homes,
which represent the most common type of US residence, the mixing and flowrelated housing characteristics presented in this chapter are generally applicable
to all types of housing.
5.1 Mixing Within a Single Zone
To apply an IAQ model, which is based on a mass-balance between zones that are
assumed to be instantaneously well-mixed, the rate of mixing in individual rooms
must occur rapidly over time scales relevant to human presence and human activities. The overall time scale of a single 24-h period, which I use in the present
work, seems appropriate to capture multiple smoking and exposure events over
the natural human diurnal cycle. From the human location patterns presented
in Chapter 4, it appears that the hourly fraction of individuals in different rooms
changes rapidly during the middle portions of the day. These data suggest that
occupant movement, smoking activity, door and window opening, and central air
handling – all activities that affect indoor SHS levels – can be expected to undergo
changes on the time scale of a few hours or less. Hence, for the well-mixed singlezone assumption to be reasonably valid and allow for the use of a simple mass
balance equation in realistically predicting time-varying room concentrations and
ultimately in assessing exposure time-profiles of house occupants to these concen-
167
tration profiles, rooms should become uniformly mixed in time periods well under
1 h.
Several studies published during the past decade report measured indoor
single-zone mixing rates. Klepeis [1999] reviewed the work of Baughman et al.
[1994], Drescher et al. [1995], and Mage and Ott [1996], and presents new data on
mixing rates measured in residential, commercial, and public settings.1 Generally, the mixing of point releases, such as for a single cigarette, was found to occur
rapidly enough so that different points in a room were in approximate agreement
within a period of 15 min or less in laboratory settings with typical natural convection or forced convection, or in realistic field settings, such as residences or taverns.
Selected data from Baughman et al. [1994], which are presented in Figure 5.1,
illustrate how, for rooms that have energy input equivalent to that from sunlight
shining through a window, complete mixing of a point pollutant release can occur
within about 5 min. The presence and movement of people within a room is likely
to inject energy on the order of (or greater than) that for impinging sunlight, so the
results reported by Klepeis for field settings appear consistent.
These findings support the use of a mass balance model under the well-mixedzone assumption to realistically describe dynamic and time-averaged air pollutant
concentrations resulting from short releases over time periods of 15 min or more
– and possibly over even shorter time periods. The mean occupancy time in major rooms of a residence, such as the living room, dining room, and kitchen, is on
the order of 70−245 min, with individual episodes for these rooms equal to 30−90
min, on average. An average of 9 h is spent in the bedroom (see Chapter 4). However, the results described so far may not be applicable to cases of extreme sourcereceptor proximity during active smoking. Errors in predicted exposure, i.e., differences from true exposure, can arise when individuals spend time in a room close
to an active, or recently extinguished, tobacco source. Both Furtaw et al. [1996] and
McBride et al. [1999] provide evidence that concentrations near an active source
can be substantially elevated, on average, by 2−3 times the expected room con1 Some
of the data used by Klepeis [1999] also appears in Ott et al. [2003].
168
centrations for the well-mixed room. These investigators find sharp spikes in concentration for locations in close proximity to sources, overlaying the gradual rise
in the room concentration. If these spikes are averaged over extended time periods
when the exposed person is not in close proximity to a source, or when there are
no active sources, then it is possible that the cumulative effect on exposure would
be small.
5.2 Zone Volumes and Surface Areas
Murray [1997] presents the results of an analyzing of two databases containing
house size information for residences throughout the US. One database is a representative sample of 7,041 individual households collected by the US Department of
Energy and the other is an amalgamation of data from over 4,000 homes collected
by the Brookhaven National Laboratory. The latter database is not necessarily representative of the US. House volumes were estimated from floor area data by assuming a ceiling height of 8 ft. For these data sets, house volumes were well fitted
by a lognormal model giving an overall geometric mean (GM) of 323 m3 and an
overall geometric standard deviation (GSD) of 1.8 for portions of homes that were
heated. No reliable information on room-specific volumes appears to be available.
The 2001 US American Housing Survey (AHS) collected information on floor
area from a representative sample of US households disaggregated by the number
of rooms and stories. A frequency tabulation of unweighted data for the segment
of the respondents (n=29,356) who lived in one-unit buildings is given in Table 5.1,
where house volume has been calculated from floor area estimates, following Murray’s use of an 8-ft ceiling height. The most common volume was in the range of
226−340 m3 for houses having five rooms and a single story.
The rate of surface deposition and sorption of air pollutants in a residence depends on the amount of available surface area in a given room. The typical quantity used in characterizing surface area, and as a parameter in air pollutant dynamics models, is the surface-to-volume ratio, which is a function of the amount of open
flat surface, e.g., floors, walls, and tables, as well as more finely textured surface
Sunlight, Min 5
Sunlight, Min 2
Sunlight, Min 10
Quiescent, Min 10
Sunlight, Min 15
Quiescent, Min 15
Figure 5.1: Panels showing snapshots of the extent of SF6 tracer gas mixing at four times and for two sets of conditions, as
measured in a 31-m3 room. The conditions are (1) “quiescent,” with windows covered and a minimum of air movement; and
(2) “sunlight,” in which sunlight entered through two 1.25 m2 windows, providing about 600 W of incident solar energy. The
area of each circle in a panel is proportional to the concentration of SF6 gas measured at a corresponding monitoring location.
The SF6 was emitted over a brief period from a point in the lower right portion of the room, to simulate the emissions from a
smoldering cigarette. Nine breathing-height sampling points (1.2 – 1.6 m high) were positioned in a rectangular pattern in the
middle of the room, and four additional sampling points were established 1 cm from the center of each wall. Under quiescent
conditions, the tracer remained incompletely mixed after 15 min, whereas the buoyancy induced mixing in the sunlight case
caused the tracer to become fully mixed within this period. (Note: Raw data from Baughman et al. [1994] were interpolated
with cubic splines to create the regular time series shown in this figure.)
Quiescent, Min 5
Quiescent, Min 2
169
No. Stories
One
Two
Three
> Three
One
Two
Three
> Three
One
Two
Three
> Three
One
Two
Three
> Three
No. Rooms
One
One
One
One
TOTAL
Two
Two
Two
Two
TOTAL
Three
Three
Three
Three
TOTAL
Four
Four
Four
Four
TOTAL
[Continued.]
75
24
9
2
110
103
19
10
8
140
45
6
1
3
55
< 113
14
3
0
2
19
< 500
843
299
66
20
1228
319
112
43
21
495
28
12
2
7
49
113-226
10
0
0
0
10
500 - 999
826
332
94
14
1266
91
42
2
7
142
11
3
2
0
16
226-340
2
0
0
0
2
1000-1499
10
10
3
1
24
0
1
1
1
3
223
157
53
6
439
a Floor
45
72
20
2
139
5
7
2
0
14
1
0
0
0
1
Area, ft2
1500-1999
2000-2499
b Volume, m3
340-453
453-566
0
0
1
0
0
0
0
0
1
0
16
21
7
1
45
1
3
1
1
6
0
0
1
0
1
566-679
0
0
0
0
0
2500-2999
32
21
10
3
66
3
3
1
1
8
0
1
0
0
1
> 679
1
0
0
0
1
> 2999
2060
926
259
48
3293
532
196
62
39
829
85
23
7
11
126
TOTAL
27
4
0
2
33
Table 5.1: Frequency Tabulation for Floor Area and Estimated Volume of One-Unit Residential Buildings by Number of Rooms and Number of Stories: Unweighted Results from the 2001 American Housing Survey, n=29,356
Telephone Respondents
170
One
Two
Three
> Three
One
Two
Three
> Three
One
Two
Three
> Three
Seven
Seven
Seven
Seven
TOTAL
Eight
Eight
Eight
Eight
TOTAL
> Eight
> Eight
> Eight
> Eight
TOTAL
0
2
3
3
8
1
4
4
0
9
8
3
6
0
17
12
11
18
1
42
< 113
29
13
13
1
56
< 500
12
15
11
2
40
13
14
11
1
39
46
34
28
1
109
233
123
73
10
439
113-226
641
264
71
9
985
500 - 999
40
52
36
2
130
42
83
57
7
189
353
261
178
11
803
1448
615
341
26
2430
226-340
2092
755
245
20
3112
1000-1499
72
97
95
16
280
170
213
164
22
569
758
569
344
35
1706
1450
791
495
39
2775
86
219
195
29
529
250
442
311
19
1022
459
525
367
43
1394
485
509
354
29
1377
Area, ft2
1500-1999
2000-2499
b Volume, m3
340-453
453-566
1044
264
543
287
229
144
24
10
1840
705
a Floor
69
242
226
34
571
128
301
221
21
671
152
243
209
23
627
131
195
155
14
495
566-679
69
118
43
1
231
2500-2999
105
523
662
85
1375
96
313
315
39
763
89
275
272
35
671
109
230
184
20
543
> 679
79
115
69
9
272
> 2999
area values estimated by telephone respondents, including main rooms only. b House volume calculated from floor area by assuming an 8-ft ceiling.
One
Two
Three
> Three
Six
Six
Six
Six
TOTAL
a Floor
No. Stories
One
Two
Three
> Three
No. Rooms
Five
Five
Five
Five
TOTAL
384
1150
1228
171
2933
700
1370
1083
109
3262
1865
1910
1404
148
5327
3868
2474
1620
139
8101
TOTAL
4218
2095
814
74
7201
171
172
in the way of carpeting or drapes. The composition of surfaces may also affect the
sorption of different chemical species.
To estimate the surface-to-volume ratio for a range of different bare rooms, containing no furniture, I generated all possible combinations of hypothetical rooms,
including closets and hallways, whose lengths and widths vary between 1 and 10
m and where the ceiling height was fixed at 8 ft, and calculated the room volume
and surface area for each combination. The distribution of surface-to-volume ratios as a function of room volume are presented in Figure 5.2 as a series of boxplots,
which display a clear trend of decreasing surface-to-volume ratio as room volume
increases. Small, bathroom-sized rooms have a surface-to-volume ratio of approximately 3−4 m−1 and house-sized spaces have ratios approaching 1 m−1 . These
ratios may be considered minimum values with larger values occurring for rooms
with different degrees of furnishings.
5.3 Air Exchange with the Outdoors
Natural leakage ventilation in residences occurs as air flows to and from the outdoors via small cracks and crevices in the building shell. The ventilation rate increases as exterior windows and doors are opened to appreciable widths, causing
air to enter and leave the building at elevated rates. As discussed in the literature [Awbi, 1991; Godish, 1989; ASHRAE, 1985; Wadden and Scheff, 1983], both
wind and indoor-outdoor temperature differences provide the driving force for
infiltration of air into residences. Awbi [1991] and ASHRAE [1985] provide equations for predicting infiltration flow rates. When there are indoor-outdoor temperature differences for a given home, a phenomenon arises, known as the stack effect,
whereby cold air flows in at the bottom of the structure and warmer air flows out at
the top during the winter, and in the reverse direction during the summer. Wind
drives flow by creating localized areas of positive and negative pressure on the
exterior of a house. Boutet [1987] provides illustrations of the natural flow of air
through residences for a variety of house layouts. Air entering particular rooms
moves about the house, and can flow along a fairly complex path in and amongst
173
4.5
Labels are Median Surface−to−Volume Ratio Values
4.0
3.5
3.5
3.1
2.5
3.0
2.2
2.1
1.9
2.0
1.6
1.4
]
,2
50
(1
,1
00
50
]
1.3
50
]
(1
(5
0,
10
0
]
(4
0,
50
]
(3
0,
40
]
(2
0,
30
]
20
0,
(1
(0
,1
0]
1.5
Surface−to−Volume Ratio [1/m]
Room Surface−to−Volume Ratio vs. Volume
Volume [m³]
Figure 5.2: Boxplots showing how the surface-to-volume ratio for bare rooms with
sides ranging from 1 to 10 m in length and a height of 8 ft varies as a function of
room volume.
174
doorways, walls, or furniture, before leaving the house via a boundary in the same
or a different room.
For purposes of modeling air pollutant dynamics in a house, it is convenient to
establish a base natural ventilation rate for houses in which all exterior doors and
windows are closed and no forced ventilation systems are operating. The effect
of windows, doors, and central air handling systems can then be considered as
perturbations to this base condition. Several published studies attempt to characterize home air-exchange rates for a fairly large population of dwellings, although
information about the number of open windows, doors, and the state of central air
operation is lacking. For example, Murray and Burmaster [1995] analyzed ventilation data for 2,844 US homes determined using a perfluorocarbon tracer (PFT)
technique. The database they used contained no information on door and window
status, or on the presence or operation of mechanical ventilation systems. They
found that the measured air-exchange rates across the US were fit well by a lognormal distribution with an overall GM of approximately 0.5 h−1 and an overall
GSD of 2.2.
A separate investigation of air-exchange rates, also using PFT, was conducted
in California by the Southern California Gas Company [Wilson et al., 1996] in a
total of over approximately 800 homes across two different surveys, one limited
to the Los Angeles area. Results from this study also show that air-exchange rates
follow a lognormal probability distribution with the statewide results in the winter
season having a GM equal to 0.4−0.6 h−1 and a GSD of 1.5−1.9. These winter
values are likely to be representative of times when exterior doors and windows
are closed.
As determined from continuous air exchange measurements in 16 naturally
ventilated homes by Kvisgaard and Coller [1990], where 63% of the total air change
was due to occupant behavior, windows and doors are important avenues by
which natural ventilation can be readily increased according to occupant preferences. A number of studies, summarized below, have reported quantitative information on the degree of extra ventilation offered by the opening of exterior
175
windows and doors in residences.
Van Dongen and Phaff [1990] undertook an extensive study of 31 Dutch apartments during the summertime, in which windows were fitted with sensors to measure the frequency and duration of window opening behavior and the width of the
openings over an 8-d period. They found that, on average, windows were opened
about 6 h per day with estimated air flow rates in excess of 100 m3 h−1 , when the
windows were opened to a width of 30 cm or more, and in excess of 200 m3 h−1 ,
when the balcony door was opened 30 cm or more.
Howard-Reed et al. [2002] analyzed hundreds of air exchange measurements
using SF6 tracer gas in two occupied residences, one a three-story attached townhouse in Virginia and one a two-story detached house in California, for which
one or more windows were opened different widths with areas ranging from 174
cm2 to over 18,000 cm2 . They found that window-opening can result in house airexchange rates that can be up to 3−10 times higher than closed-house air-exchange
rates, which averaged about 0.4 h−1 for the 510 m3 California house and 0.2−0.6
for the 400 m3 Virginia house across summer and winter months.
In an effort to estimate flow rates through open windows on a per-window
basis from the Howard-Reed et al. data, I normalized each measured increase in
air-exchange rate by the number of open windows and multiplied the result by the
house volume to obtain absolute magnitudes of increased flow rates. The results
from this analysis, broken down by the number of windows opened (one, two,
and overall) are shown in Figure 5.3 (top panels). The most frequent air flow rate
is between 100 and 200 m3 h−1 for experiments involving any number of open
windows. The distribution of air flow for single and multiple open windows is
similar. This result suggests that air coming in one or more windows will leave
via other windows or by enhanced leakage through building cracks if no other
windows are open, but opening multiple windows does not necessarily result in
larger flow rates through individual windows.
Alevantis and Girman [1989] conducted an extensive analysis of the effect on
ventilation rates of opening windows by different amounts and on different sides
176
of two 1300 ft2 (294 m3 ) detached houses and a two-story 878 ft2 (200 m3 ) condominium. They found that opening windows increased ventilation by factors of
between 1.2 and 11 times the original rate. As with the results of Howard-Reed
et al., the per-window absolute magnitude of flow increases tend to be between
100 and 200 m3 h−1 (Figure 5.3; bottom panels). The lower rate increases occurred
when windows were opened that were not in the direction of prevailing winds,
while opening windward windows resulted in the largest rate increases. Fully
opening a single windward window had approximately the same effect as fully
opening more than one leeward window, or opening multiple windward and leeward windows to a width of only 3 inches. The reported rate increases are calculated by controlling for wind speeds, i.e., the infiltration for closed-window and
open-window cases were compared only when wind speeds were similar. When
windows were closed, the authors found that wind could cause residential infiltration rates to increase by as much as 8.7 times the minimum rate.
Heiselberg et al. [1999, 2000] and Svidt et al. [2000] have studied natural air flow
through windows in a laboratory setting. They find that natural air flow through
a single window can be understood in terms of a thermal stack effect, which is
the main driving force in non-isothermal winter conditions, where air flows into a
room through the bottom half of a window and out of the room through the top
half of the window. In contrast, the main driving force across window boundaries
in isothermal summer conditions is wind turbulence. When there are multiple
window openings in a space, cross ventilation, which is typically wind-driven,
can occur when air flows in one window and out another on an opposite side.
The thermal stack ventilation effect can also contribute to flow between multiple
openings, especially when the openings are at different heights.
Panzhauser et al. [1993] present results of the simulation and measurement of
natural ventilation rates in residences. From experimental data on the volume of
air flowing through a swing-type window during the winter under purely stack
flow conditions (temperature difference of 10−20 K) and in the absence of wind
influence, they find that openings ranging from 10 cm to the maximum opening
177
Single Windows, n=64
Multiple Windows, n=30
200
400
All Windows, n=94
600
15
10
0
5
Frequency
20
25
0
0
200
400
600
0
200
400
600
Flow Per Open Window [m3 h−1]
Single Windows, n=11
Multiple Windows, n=13
All Windows, n=24
200 400 600 800
3
0
1
2
Frequency
4
5
6
0
0
200 400 600 800
0
200 400 600 800
Flow Per Open Window [m3 h−1]
Figure 5.3: Histograms of whole-house increases in flow rate due to the opening of one or
more windows and normalized by the number of open windows. The data were calculated
from air exchange and house volume data reported for two houses by Howard-Reed et al.
[2002] (top panels) and for two detached homes and a condominium by Alevantis and
Girman [1989] (bottom panels). The most common per-window flow rate increase is in the
range of 100−200 m3 h−1 for the first study in which windows were opened at various
positions to areas ranging from 174 cm2 to over 18,000 cm2 . For the second study, in which
windows were opened to widths ranging from 3 to 33 inches and on either the leeward
and/or windward sides of the house or sides that were not in the direction of prevailing
windows, per-window flow rates were most commonly under 100 m3 h−1 .
178
width of 14 cm corresponded to flow rates of 130−190 m3 h−1 . The flow rate
arising from wind-induced cross-ventilation through a 220 cm2 inlet and 750 cm2
outlet in a residence was measured to be between 110 and 140 m3 h−1 when the
wind speed was approximately 3 m s−1 . Using simulation, the authors show how
diagonal cross-ventilation rates can vary widely depending on wind speed and
wind direction, predicting flows ranging from 100 to 400 m3 h−1 at wind speeds of
1 m s−1 depending on the wind direction.
Using tracer gas injection, Roulet and Scartezzini [1987] measured air-exchange
rates in different rooms of an occupied 530 m3 , 10-room, three-level residence
over a period of 12 d for different door and window configurations, which were
recorded by the house inhabitants in diaries. Each of the rooms were connected to
a staircase through open doorways or loose-fitting doors. In the absence of wind,
a clear stack effect was evident in the house with fresh air entering through the
kitchen and living rooms on the first floor and exiting through bedrooms on the
upper level. When all windows were closed, the total mean air-exchange rate for
the house was 0.37 h−1 , increasing to 0.55 h−1 when one or two bedroom windows were opened, corresponding to an average increase in house air flow of 95
m3 h−1 . However, in some instances opening one or more windows could increase the house air-exchange rate by as much as 0.8 or 1 h−1 , corresponding to
an increase in house air flow over the closed-house mean by as much as 200−300
m3 h− 1 .
5.4 HVAC Systems: Recirculation, Outdoor Air Delivery, and Duct Leakage
The purpose of a residential heating, ventilation, and air conditioning (HVAC) system is to circulate heated or cooled air throughout the rooms of a house, drawing
air from one or more centrally-located return vents into a duct-work system, processing it, and pushing it out of individual room supply vents. For some systems, a
controlled quantity of fresh outdoor air may also be drawn into the system, which
is typically expressed as a percentage of the total supplied air flow rate. However,
179
to date, this practice has not been common for single family dwellings in the US.
Residential air handling systems that do not provide forced-air ventilation should
appropriately be called HAC systems, since they provide only heating and (often)
air conditioning. Filters may exist in HAC/HVAC return ducts and HVAC outdoor air supply ducts. Figure 5.4 contains a schematic of HAC flows for the case
of a four-room house.
According to Bearg [1993], total air supply rates from HVAC systems, calculated as the total volume of air supplied per unit time divided by the building
volume, range from 5 to 7 h−1 . Sparks et al. [1991] measured HVAC air flows using an anamometer in a 293 m3 three-bedroom test house, which was used as the
site of validation experiments for a multizone indoor air quality model. The air
flow through the single return, located in the central hallway, was 1760 m3 h−1 ,
which is equal to approximately six times the total house volume per hour and
was equal to the sum of air flows supplied to the bedrooms, bathrooms, and den.
About half of the total supplied air flow (830 m3 h−1 ) was supplied to the den,
which also comprised about half of the house volume.
For commercial buildings, the minimum amount of fresh outdoor air entering
the HVAC system through a make-up vent, and therefore the rooms containing
supply registers, is reported by Bearg [1993, p.43] to typically be 15% of the total
supply air flow rate. Wadden and Scheff [1983, p.144] cite 1980 ASHRAE guidelines for total delivery of fresh air into a 1-bedroom residence of about 300 m3 h−1 ,
mostly directed towards the kitchen. The current ASHRAE standard guidelines for
residential ventilation [ASHRAE, 2003] recommends similar ventilation rates. Formulae are given, which can be used to calculate the appropriate quantity of fresh
outdoor air to be delivered, either continuously or intermittently, by a whole-house
forced-fan residential HVAC system. For a house with a volume of approximately
400 m3 and 3 bedrooms, the recommended overall ventilation rate for an intermittent system with a 30% duty cycle is about 800 m3 h−1 . If the HVAC forces 2000 m3
h−1 through supply registers, then a fresh air supply rate of 800 m3 h−1 is equal
to 40% of the total air delivery rate. Because the existing US housing stock does
180
558 m3 h-1
100 m3
50 m3
Filter
50 m3
279 m3 h-1
7m3
40 m3 h-1
50 m3
1435 m3 h-1
30 m3
HAC System
279 m3 h-1
279 m3 h-1
Figure 5.4: Schematic of HAC-related flow rates for a hypothetical four-room
house, which contains a central hallway, where the HAC return is located, and a
master bathroom. The total HAC supply flow rate is set to the return rate of 1,435
m3 h−1 or five house volumes per hour. A particle filter is present on the return
register. Ideally, the same volume of supplied air flows through open doorways
back to the HAC return register, although in practice duct leaks or closed interior
doors may result in increased outdoor air infiltration or exfiltration through the
building shell caused by particular zones of the house becoming pressurized or
depressurized. Infiltration and natural ventilation flows are not shown.
181
not conform to current ASHRAE guidelines regarding ventilation, I do not include
HVAC-supplied outdoor air as part of the residential SHS exposure simulation experiments presented in Chapters 7−9.
When an HVAC system is activated during times when interior doors are open
and exterior doors and windows are closed, the total amount of positive air flow
delivered at each supply register is intended to be balanced by negative air flow at
the return register. However, in practice leaks in the ductwork can create net positive or negative pressurization of house zones with return leaks contributing to
positive pressurization and supply leaks contributing to negative pressurization.
These leaks are manifested as a larger total air-exchange rate for the house [Robison and Lambert, 1989; Cummings and Tooley, 1989; Lambert and Robison, 1989;
Modera, 1989].
Infiltration rates are typically measured with perfluorocarbon tracer gas (PFT)
or estimated from leakage area data obtained from blower door experiments. According to Modera [1989], who reviews a number of past studies, “...air infiltration
rates will typically double when distribution fans are turned on and...the average
annual air infiltration rate is increased by 30% to 70% due to the existence of the
distribution system.” The magnitude of the effect varies with the amount of leakage in a particular home’s ductwork. For example, while Parker [1989] observed
a 70% greater air-exchange rate (an absolute increase of 0.17 h−1 ) in houses with
forced air heating systems over houses with other types of heating and Cummings
and Tooley [1989] reported an average infiltration rate of 0.14 h−1 with air handlers
off and 1.42 h−1 when they were running, Robison and Lambert [1989] report only
a 10% increase in infiltration due to duct leakage. Lambert and Robison [1989],
using test data gathered from over 800 homes that had either forced-air heating
systems with ducts or were room-heater equipped, found that ducted “current
practice” homes were 26% leakier than unducted ones.
The position of interior doors may also have a large effect on residential infiltration rates, since, as Cummings and Tooley [1989] observe, when interior doors
are closed, the flow from forced-air supplies to the return is disrupted. This dis-
182
ruption causes an imbalance in the system and further increases in infiltration rate
beyond the effects introduced by duct leakage alone. They report that when interior doors of five different homes were closed while the air handler was running,
the infiltration rate increased from an average of 0.31 h−1 to 0.91 h−1 .
5.5 Estimates of Interzonal Air Flow Rates
Using tracer gas measurements in a test, Sparks et al. [1991] estimated inter-room
air flow rates for cases when the air handler was inactive as part of the validation
of their multi-zone indoor air quality model. Although their method for determining air flows is unclear, the authors assumed values of 100−120 m3 h−1 for air
flow to and from three bedrooms of a 293 m3 house and a central hallway, presumably through open doorways, which resulted in satisfactory agreement between
predicted and measured particle concentrations during several kerosene heater experiments.
Miller and Nazaroff [2001] conducted SHS particle and tracer gas measurement
and modeling studies for different ventilation scenarios in two-room test house,
where the air flows rate between rooms and to the outdoors were determined using a dual-tracer technique described by Miller et al. [1997] (see Table 5.2). The
differences in 1-h average PM2.5 concentrations between the two rooms (a designated smoking room and a designated nonsmoking room) were as much as 93
µ g m−3 when the adjoining door was closed, whereas 1-h average concentrations
were within 10 µ g m−3 when the door was left open. Closed-door air flow rates
between zones were very low at around 1 m3 h−1 . Open-door flows for the baseline case, in which there was no active ventilation or filtration, were near 60 m3
h−1 in both directions.
Ott et al. [2003] report three-room measurements in a one-story cottage where
one fully and one partially closed door exhibited substantial impact on inter-room
differences in measured carbon monoxide (CO) concentrations after a cigar was
smoked in one of the rooms. The layout of the house is shown in Figure 5.5 with
the CO concentration time series presented in the left panel of Figure 5.6. The
183
Table 5.2: Air Flow Rates Measured During Six Two-Compartment SHS Particle
Experimentsa
b Air
Scenario
F NS
FSN
Baseline
60
59
Segregationc
Exhaust Vent.
0.6
1.1
Flows [m3 h−1 ]
FSO
F NO
Air Exch.
FOS
FON
Filt
[h−1 ]
2.4
0.001
1.6
0.8
-
0.04
2.5
4.2
3.1
3.7
-
0.1
92
17
107
Enhanced Vent. 154
163
10
N Filtration
128
128
S Filtration
46
47
0.0
11
32
75
-
1.7
19
2
-
0.3
0.004
2.7
0.3
2.4
91
0.03
0.0
2.4
0.7
1.6
91
0.03
a The
data in this table were reported by Miller and Nazaroff [2001] and estimated through a technique described by Miller et al. [1997].
b "N" represents the nonsmoking room, "S" is the smoking room, "O" is the outdoors, and "Filt" is
for the recirculating fan. The designation FXY indicates flow from zone X to zone Y.
c The “Segregation” scenario was the only one for which the door between the smoking and nonsmoker room was closed.
right panel of Figure 5.6 shows the results of fitting a two-compartment model to
concentrations in the source room (kitchen) and an adjacent room (living room),
accomplished with the computer program described in Appendix C. The results
of the fitting procedure gave estimated interzonal air flow rates for the two rooms
of 103−130 m3 h−1 in either direction.
In the same split-level detached 510 m3 California townhouse where windowopening experiments were conducted [Howard-Reed et al., 2002], a series of 14
two-room SF6 tracer experiments were performed for different connecting door
positions [Ferro and Christiansen, 2001]. The door positions used were fully
closed, fully closed and sealed, fully open, and partially open at widths ranging
from 2.5 to 15 cm. Figure 5.7 shows a schematic, including room volumes and monitoring positions, for the residence’s first floor where the experiments were carried
out. The SF6 concentration time series for these experiments, which consisted of
measurements every 75 s, are shown in Figure 5.8. Once again applying the interactive computer program described in Appendix C to the measured time series,
I estimated the air flows for each room, including interzonal flows and flows be-
184
Front Door
Front Bedroom
Living Room
CO Monitor
Bathroom
CO Monitor
Connecting
Closet
Cigar Source
CO Monitor
Rear Bedroom
Kitchen
Porch
Back Door
Figure 5.5: Schematic of a small house where multi-room measurements of carbon
monoxide concentrations resulting from the smoking of a cigar in the kitchen were
made by Ott et al. [2003]. Figure 5.6 contains concentration data and the results of
a two-compartment model fit.
0
20
40
60
80
100
0
150
200
Living Room
Rooms
Elapsed Minutes
100
Bedroom
50
Kitchen
250
300
CO Concentration [mg/m³]
0
2
4
6
8
10
12
14
16
18
0
50
150
200
Elapsed Minutes
100
250
Living Room (Observed)
Kitchen (Observed)
Living Room (Model)
Kitchen (Model)
300
Figure 5.6: Plots of the time series of carbon monoxide (CO) levels measured in three rooms of a small residence
after a cigar was smoked in the kitchen for 15 minutes (left) and a two-compartment model fit to concentrations in
the kitchen and living room (right) as reported by Ott et al. [2003]. Only every 25th observation is shown in the plot
on the right. During this experiment, the door between the kitchen and living room was open three inches during
the experiment. See Figure 5.5 for the layout of the rooms in the house. The fitted air-exchange rate for the overall
two-compartment system was about 1 h−1 , corresponding to a volumetric flow rate of 70 m3 h−1 . The estimated
interzonal flows were 103 m3 h−1 from the kitchen to the living room and 129 m3 h−1 from the living room to the
kitchen.
CO Concentration [pphm]
120
185
186
tween each room and the outdoors. The estimated air flows, which are presented
in Table 5.3, show a clear effect of closing the door. Fully open door positions correspond to estimated source-room to receptor-room flows of 150−250 m3 h−1 and
fully closed door positions correspond to flows of 0.1−0.4 m3 h−1 . For intermediate door positions, the interzonal air flow rates ranged from 6 to 70 m3 h−1 .
5.6 Illustrative Simulation of Tracer Gas Concentrations in a House
In this section, I present the results of simulating tracer gas concentrations in a typical house with 4 main rooms (plus a bathroom and hallway) to illustrate generally how contaminant levels in different rooms in a residence might respond when
doors are left open or closed and when either short or prolonged releases occur
in different locations. This type of simulation for multizone indoor air pollutants
is central to the prediction of residential SHS exposures, as performed in Chapters
7−9. However, whereas particulate or semi-volatile constitents of SHS require consideration of surface interactions, in the treatment of an inert tracer gas presented
here, removal processes are limited to those involving air flow, i.e., air-exchange
with the outdoors.
The simulation is achieved through numerical solution of a system of coupled
linear differential equations, one equation corresponding to each of the six different compartments in the house. Appendix B presents the general form of these
differential equations, which are solved using a Runge-Kutta-type algorithm to
obtain a time-varying concentration profile for each compartment. To make the
simulated tracer gas concentrations somewhat representative of those that might
occur in a typical home in the United States, I use environmental and emissionsrelated parameter values discussed earlier in this chapter and in Chapter 3 to correspond to common household and smoking conditions. I use a total simulation
period of 12 h.
The total volume of the simulated house is 287 m3 , close to the most common
volumes determined from nationwide surveys, with the largest share of the vol-
187
Dining Area
Nook
Kitchen
Source
Living
Room
Op
Family
Room
en
460 m³
Bo
un
da
ry
Open Boundary
Laundry
Entrance
Source
Monitor
Stairs
Bath
Door
Garage
Receptor
Monitor
Bedroom
43 m³
Figure 5.7: Schematic of the first floor of a townhouse where two-room interzonal
air flow experiments were performed. The main living area, where the source was
active and the source monitor was placed, had a vaulted ceiling and was approximately 460 m3 in volume. The bedroom, where the receptor monitor was placed,
also had a vaulted ceiling and an approximate volume of 43 m3 . Each monitor
was 1 meter from the floor and the source monitor was located 5.7 meters from the
emitting source (approximately 200 cc min−1 of 99.8% pure SF6 ).
188
2 : FULLY OPEN
0
5
10
0 5 10 15 20
10
0
20
0 5 10
0 5 10
5 10 15 20
0
0 10 20 30
12 : FULLY OPEN
11 : OPEN 15 cm
0 5 10 15 20
10 : OPEN 15 cm
15 : CLOSED/SEALED
0 10 20 30 40
0 10 20 30 40
9 : OPEN 10 cm
20
8a : OPEN 5 cm
14 : CLOSED
7 : OPEN 2.5 cm
0
5
5
0
8 : OPEN 5 cm
0
5 10 15 20
5 : OPEN 10 cm
5 10 15 20
20
0 5 10
10
4 : OPEN 5 cm
S F 6 Concentration [ppm]
3 : OPEN 2.5 cm
15
1 : CLOSED/SEALED
0 200
600 1000
0 200
600 1000
Elapsed Minutes
Source Room (A)
Monitor
Receptor Room (B)
Figure 5.8: Time series plots of data from fourteen tracer-gas experiments conducted by Ferro and Christiansen [2001] in two rooms of a townhouse for a variety
of door positions. See Figure 5.7 for a schematic of the experimental setup and
Table 5.3 for the estimated interzonal air flow rates.
189
Table 5.3: Summary of Fourteen Two-Room SF6 Tracer-Gasa Experiments on the
Effect of Door Position on Air Movement with Estimated Flow Rates Between the
Source Room (A) and Test Room (B) and the Estimated Overall Two-Room AirExchange Rate
Opening Opening
Air Flows
Air
Door
Width
Area
A to B
B to A
Exchange
Exp
Position
[cm]
[m2 ]
[m3 h−1 ]
[m3 h−1 ]
[h−1 ]
1
closed/sealed
−
0.05
0.4
1.4
0.34
15
closed/sealed
−
0.05
0.1
2
0.39
14
closed
−
0.06
0.2
6
0.40
3
open
2.5 cm
0.22
6
12
0.26
7
open
2.5 cm
0.22
17
9
0.39
4
open
5 cm
0.37
56
29
0.27
8
open
5 cm
0.37
24
3
0.31
8a
open
5 cm
0.37
33
44
0.32
5
open
10 cm
0.67
58
41
0.32
9
open
10 cm
0.67
46
58
0.38
10
open
15 cm
0.99
57
69
0.44
11
open
15 cm
0.99
53
7
0.38
2
fully open
−
4.00
245
154
0.42
12
fully open
−
4.00
102
67
0.34
a The sampling interval for both monitoring instruments was approximately 75 s (1.25 min) and the
SF6 emission rate was approximately 200 cm3 min−1 or 1.3 g min−1 . The duration of the source
was 30 min for each experiment.
190
ume assigned to the kitchen-dining area (KIT-DIN; 100 m3 ) and equal volumes of
50 m3 assigned to the living (LIV), bedroom (BED), and auxiliary room (AUX) (extra bedroom or office space). The remaining volume was apportioned to a 30 m3
hallway and a 7 m3 bathroom, which is attached to the main bedroom (BED).
The source emission rate is equal to that which might be expected for carbon
monoxide emissions from a single burning cigarette, 50 mg cig−1 , which when
divided by a 10 min smoking period gives a rate of 5 mg min−1 . The source is
active in either KIT-DIN or BED. In the simulation, I consider both a short duration
of 10 min, which is approximately equal to the time need to smoke one cigarette,
and a continuous source lasting 12 h.
I chose a value of 0.5 h−1 for the leakage air-exchange rate, which is the value
of the GM from the Murray and Burmaster [1995] nation-wide study described
above, and values of 100 m3 h−1 and 1 m3 h−1 for open-door and closed-door interior air flows, respectively, which are fairly representative of values determined
from experiments in real and test houses [Ott et al., 2003; Miller and Nazaroff, 2001;
Ferro and Christiansen, 2001]. The total infiltration rate due to leakage is divided
amongst the rooms of the house in proportion to their volume. When a window
is opened in a particular room, the flow between that room and the outdoors is
increased by 150 m3 h−1 above the existing flow due to building leakage, which is
in the range of commonly observed increases in the intensive studies of HowardReed et al. [2002] and Alevantis and Girman [1989].
I simulate four different air flow scenarios corresponding to different configurations of door and window positions. The first scenario is a base case in which all
interior doors are open and all windows are closed. For the second scenario, the
base case is perturbed by closing all of the interior doors. The third and fourth scenarios are perturbations of the base case in which KIT-DIN and BED windows are
opened, respectively, corresponding to cases when the source is present in those
rooms. House schematics, including interzonal air flows for each of the four different scenarios, are given in Figure 5.9.
Results of the tracer gas simulations for 10-min cigarette sources are shown in
191
Doors Open,
Windows Closed
Doors Closed,
Windows Closed
25 m3 h-1
AUX,
50 m3
KIT-DIN,
100 m3
-1
100 m h
LIV,
50 m3
BED,
50 m3
Doors Open,
KIT-DIN Window Open
100 m3 h-1
25 m3 h-1
AUX,
50 m3
100 m3 h-1
100 m3 h-1
HALL, 30 m3
100 m3 h-1
3
-1
100 m h
BED,
50 m3
25 m3 h-1
15 m3 h-1
BATH, 7m3
LIV,
50 m3
25 m3 h-1
KIT-DIN,
100 m3
HALL, 30 m3
100 m3 h-1
25 m3 h-1
50 m3 h-1
AUX,
50 m3
100 m3 h-1
BED,
50 m3
Doors Open,
BED Window Open
25 m3 h-1
KIT-DIN,
100 m3
-1
1m h
25 m3 h-1
25 m3 h-1
200 m3 h-1
1 m3 h-1
3
100 m3 h-1
LIV,
50 m3
25 m3 h-1
100 m3 h-1
100 m3 h-1
15 m3 h-1
25 m3 h-1
1 m3 h-1
15 m3 h-1
100 m3 h-1
3
BATH, 7m3
HALL, 30 m3
15 m3 h-1
BATH, 7m3
LIV,
50 m3
1 m3 h-1
1 m3 h-1
HALL, 30 m3
100 m3 h-1
AUX,
50 m3
KIT-DIN,
100 m3
100 m3 h-1
100 m3 h-1
25 m3 h-1
50 m3 h-1
BATH, 7m 3
50 m3 h-1
BED,
50 m3
175 m3 h-1
Figure 5.9: House schematics with room volumes and interzonal air flow rates
corresponding to four different door and window configurations. These four flow
scenarios are used in simulations of tracer gas concentrations after either a 10-min
or 12-h release in either the kitchen-dining area (KIT-DIN) of the house or in the
bedroom (BED). Source positions are designated by filled circles.
192
Figures 5.10 and 5.11. The largest peak concentrations occur in the source rooms,
KIT-DIN or BED, when doors are closed. The BED concentration is higher than
that in KIT-DIN due to its smaller volume. The closing of doors also results in
practically zero concentrations in the other, non-source, rooms. When doors are
opened, peak concentrations occur in non-source rooms and the hallway, but at
a level much lower than for the source room, except for the bathroom, which is
connected directly to the bedroom. Opening a window in a source-room reduces
source-room peak levels somewhat below the case when interior doors are open
but all windows are closed, and only slightly decreases concentrations in nonsource rooms below the door-open-only case. It typically takes at least 5 h for
tracer emissions to be completely removed.
The steady-state tracer gas concentrations for simulations of continuous sources
in either BED or KIT-DIN, lasting the entire 12-h (720-min) period, are given in Table 5.4. When windows are closed, it takes more than 3 h of constant emissions
before steady-state conditions are reached. When a BED or KIT-DIN window is
open, steady-state concentrations are reached more rapidly, within about 1−2 h,
and the concentration is substantially lower than when windows are closed. Over
the course of a day in a household with a smoker, it is unlikely that steady-state
conditions could occur unless ventilation rates are high and cigarettes are chain
smoked for a large part of the day.
It is clear from these simulation results that the timing of a household member’s
presence in different rooms of a house can play a key role in determining their exposure to air pollutants emitted in one or more rooms over a 24-h period. The time
spent in source and non-source rooms and the positions of doors and windows are
important. By avoiding rooms where smoking occurs, entering rooms after smoke
has dissipated, or using doors as pollutant barriers and windows as sources of
increased ventilation, a person might reduce or even (in theory) nearly eliminate
their exposure. The effects of the multizonal character of a house on variation in
exposure, including the movement of pollutants and human beings amongst different rooms, is treated in Chapters 7 and 8. The effectiveness and practicality of
193
Table 5.4: Steady-State Concentrations for Tracer Gas Simulations of a Continuous
Source Emitting at 5 mg min−1
Source
Door/Window
Room
Config.
KIT−DIN
LIV
AUX
BED
HALL BATH
3,200
1,400
1,400
1,400
1,800
1,400
Doors Closed
5,900
12
12
11
310
2.4
Doors Open KIT−DIN
Window Open
1,550
700
700
680
880
660
Doors Open
1,400
1,700
1,700
4,000
2,100
3,800
Doors Closed
11
22
22
11,000
590
2,500
Doors Open BED Window
Open
2,100
1,600
1,600
2,900
2,000
2,800
KIT-DIN Doors Open
BED
Steady-State Concentrations [µ g m−3 ]
this type of behavior for use in exposure mitigation is the subject of Chapter 9.
A typical detached house in the US has four or five main rooms on a single story
with a total inhabited volume near 300 m3 , which implies rooms with a mean size
of 60−75 m3 . Rooms in a house with volumes of 40−100 m3 are expected to have
a base surface-to-volume ratio of at least 2 m−1 , with greater ratios arising from
the presence of furnishings. The mixing of short pollutant releases within single
zones in a residence occurs fairly rapidly, so that typically concentrations at different points in a room agree within a time period of 15 min or less, which is short
enough to support the common modeling assumption of instantaneous, or ideal,
mixing. Air-exchange rates for residences in the US due to leakage through small
building cracks have an expected value near 0.5 h−1 . The operation of HVAC systems can substantially increase the infiltration rates for a house due to duct leakage
0
100
200
300
400
0
200
400
BED
600
400
HALL
LIV
600
Elapsed Minutes
200
0
Doors Open
Doors Closed
Doors Open & KIT−DIN Window Open
0
200
400
BATH
AUX
600
0
100
200
300
400
Figure 5.10: KIT-DIN Cigarette Source. Plots of simulated tracer gas concentrations in a 6-compartment house for
three different flow scenarios when a single 10-min cigarette source was active in the kitchen-dining area. The 287
m3 house has 4 mains rooms (100 m3 kitchen-dining area (KIT-DIN) and a living room (LIV), bedroom (BED), and
auxiliary room (AUX) each 50 m3 in size) plus a 30 m3 hallway (HALL) and 7 m3 master bathroom (BED). The three
different flow scenarios correspond to cases when all interior doors are open (red curves), all doors are closed (blue
curves), and doors are open and the KIT-DIN window is open (green curves). See Figure 5.9 for air flow schematics
corresponding to each of the flow scenarios. The overall outdoor air-exchange rate of the house from leakage flow
is 0.5 h−1 with open-door inter-room air flow rates of 100 m3 h−1 , closed-door flow rates of 1 m3 h−1 , and open
windows contributing 150 m3 h−1 to per-room outdoor air-exchange rates above those for leakage. The emission
rate for the non-reactive tracer gas is 5 mg min−1 .
Tracer Concentration [µg m−3]
KIT−DIN
194
0
200
400
600
800
0
200
400
BED
600
400
HALL
LIV
600
Elapsed Minutes
200
Doors Open
Doors Closed
Doors Open & BED Window Open
0
0
200
400
BATH
AUX
600
0
200
400
600
800
Figure 5.11: BED Cigarette Source. Plots of simulated tracer gas concentrations in a 6-compartment house for three
different flow scenarios when a 10-min cigarette source was active in the bedroom (BED). The three different flow
scenarios correspond to cases when all interior doors are open (red curves), all doors are closed (blue curves), and
doors are open and the BED window is open (cyan curves). See the text and the Figure 5.10 caption for more
information on the simulations, and Figure 5.9 for air flow schematics corresponding to each of the scenarios.
Tracer Concentration [µg m−3]
KIT−DIN
195
196
and the closing of interior doors, which can cause uneven pressure distributions
within the house. When exterior windows are opened, the extra ventilation flow
is expected to be in the range of 100−200 m3 h−1 . Flow across fully open interior
doorways under normal outdoor leakage conditions (no open windows) appears
to be typically greater than 50 m3 h−1 in a single direction and maybe as much as
100 or 200 m3 h−1 . There is clear evidence that closed interior doors dramatically
impede the flow of air resulting in air flow rates as low as 1 m3 h−1 in either direction. However, when an HAC/HVAC system is operating, the flow rates across
closed doorways are likely to be much higher due to supplied air making its way
back to one or more return registers.
5.8 References
Alevantis, L. E. and Girman, J. R. (1989). Occupant-Controlled Residential
Ventilation. In The Human Equation: Health and Comfort, Proceedings of the
ASHRAE/SOEH Conference IAQ ’89, pages 184–191, San Diego. American Society for Heating, Refrigerating, and Air-Conditioning Engineers.
ASHRAE (1985). ASHRAE Handbook: 1985 Fundamentals. American Society of
Heating, Refrigeration, and Air-Conditioning Engineers, Inc., Atlanta.
ASHRAE (2003). ASHRAE Standard 62.2-2003. Ventilation and Acceptable Indoor Air
Quality in Low-Rise Residential Buildings. American Society of Heating, Refrigeration, and Air-Conditioning Engineers, Inc., Atlanta.
Awbi, H. B. (1991). Ventilation of Buildings. E & FN SPON, London.
Baughman, A. V., Gadgil, A. J., and Nazaroff, W. W. (1994). Mixing of a point source
pollutant by natural convection flow within a room. Indoor Air, 4(2): 114–122.
Bearg, D. W. (1993). Indoor Air Quality and HVAC Systems. Lewis Publishers, Boca
Raton.
Boutet, T. S. (1987). Controlling Air Movement: A Manual for Architects and Builders.
McGraw-Hill, New York.
Cummings, J. B. and Tooley, J. J. (1989). Infiltration and pressure differences induced by forced air systems in Florida residences. ASHRAE Transactions, 95(4):
551–560.
197
Drescher, A. C., Lobascio, C., Gadgil, A. J., and Nazaroff, W. W. (1995). Mixing of
a point-source indoor pollutant by forced-convection. Indoor Air, 5(3): 204–214.
Ferro, A. and Christiansen, C. (2001). Unpublished two-room tracer data from
door-opening experiments conducted in a townhouse. Personal communication.
Furtaw, E. J., Pandian, M. D., Nelson, D. R., and Behar, J. V. (1996). Modeling indoor
air concentrations near emission sources in imperfectly mixed rooms. Journal of
the Air and Waste Management Association, 46(9): 861–868.
Godish, T. (1989). Indoor Air Pollution Control. Lewis Publishers, Chelsea, MI.
Heiselberg, P., Dam, H., Sorensen, L. C., Nielsen, P. V., and Svidt, K. (1999). Characteristics of flow through windows. In HybVent: Hybrid Ventilation in New and
Retrofitted Office Buildings, First International One Day Forum on Natural and
Hybrid Ventilation, Sydney, Australia.
Heiselberg, P., Svidt, K., and Nielsen, P. V. (2000). Windows: Measurements of
air flow capacity. In Awbi, H. B., editor, Air Distribution in Rooms: Ventilation for
Health and Sustainable Environment, Volume II of Proceedings of the 7th International
Conference on Air Distribution in Rooms, pages 749–754, Reading, UK. Elsevier.
Howard-Reed, C., Wallace, L. A., and Ott, W. R. (2002). The effect of opening windows on air change rates in two homes. Journal of the Air and Waste Management
Assocation, 52(2): 147–159.
Klepeis, N. E. (1999). Validity of the uniform mixing assumption: Determining
human exposure to environmental tobacco smoke. Environmental Health Perspectives, 107(SUPP2): 357–363.
Kvisgaard, B. and Coller, P. F. (1990). The user’s influence on air change. In Sherman, M. H., editor, Air Change Rate and Airtightness in Buildings, pages 67–76,
Philadelphia, PA. American Society for Testing and Materials.
Lambert, L. A. and Robison, D. H. (1989). Effects of ducted forced-air heating
systems on residential air leakage and heating energy use. ASHRAE Transactions,
95(2): 534–541.
Mage, D. T. and Ott, W. R. (1996). The correction for nonuniform mixing in indoor
environments. In Tichenor, B. A., editor, Characterizing Indoor Air Pollution and
Related Sink Effects, pages 263–278, West Conshohocken, PA. American Society
McBride, S. J., Ferro, A. R., Ott, W. R., Switzer, P., and Hildemann, L. M. (1999).
Investigations of the proximity effect for pollutants in the indoor environment.
Journal of Exposure Analysis and Environmental Epidemiology, 9(6): 602–621.
198
Miller, S. L. and Nazaroff, W. W. (2001). Environmental tobacco smoke particles in
multizone indoor environments. Atmospheric Environment, 35(12): 2053–2067.
Modera, M. P. (1989). Residential duct system leakage: Magnitude, impacts, and
potential for reduction. ASHRAE Transactions, 95(5): 561–569.
Murray, D. M. (1997). Residential house and zone volumes in the United States:
Empirical and estimated parametric distributions. Risk Analysis, 17(4): 439–446.
Murray, D. M. and Burmaster, D. E. (1995). Residential air exchange rates in the
United States - Empirical and estimated parametric distributions by season and
climatic region. Risk Analysis, 15(4): 459–465.
Panzhauser, E., Mahdavi, A., and Fail, A. (1993). Simulation and evaluation of
natural ventilation in residential buildings. In Nagda, N. L., editor, Modeling of
Indoor Air Quality and Exposure, ASTM STP 1205, pages 182–196, Philadelphia,
PA. American Society for Testing and Materials.
Parker, D. S. (1989). Evidence of increased levels of space heat consumption and
air leakage associated with forced air heating systems in houses in the Pacific
Northwest. ASHRAE Transactions, 95(1): 527–533.
Robison, D. H. and Lambert, L. A. (1989). Field investigation of residential infiltration and heating duct leakage. ASHRAE Transactions, 95(3): 542–550.
Roulet, C. and Scartezzini, J. L. (1987). Measurement of air change rate in an inhabited building with a constant tracer gas concentration technique. ASHRAE
Transactions, 14(3): 1371–1380.
Sparks, L. E., Tichenor, A. B., White, J. B., and Jackson, M. D. (1991). Comparison of
data from an IAQ test home with predictions of an IAQ computer model. Indoor
Air, 4: 577–592.
Svidt, K., Heiselberg, P., and Nielsen, P. V. (2000). Characterization of the airflow
from a bottom hung window under natural ventilation. In Awbi, H. B., editor, Air Distribution in Rooms: Ventilation for Health and Sustainable Environment,
199
Volume II of Proceedings of the 7th International Conference on Air Distribution in
Rooms, pages 749–754, Reading, UK. Elsevier.
Van Dongen, J. E. F. and Phaff, J. C. (1990). Ventilation behavior in Dutch apartment
dwellings during summer. In Lunau, F. and Reynolds, G. L., editors, Indoor Air
Quality and Ventilation, Indoor Air and Ventilation Conference, Lisbon, Portugal.
Selper Ltd., London.
Wadden, R. A. and Scheff, P. A. (1983). Indoor Air Pollution: Characterization, Prediction, and Control. John Wiley & Sons, New York.
Wilson, A., Colome, S., Tian, Y., Becker, E., Baker, P., Behrens, D., Billick, I., and
Garrison, C. (1996). California residential air exchange rates and residence volumes. Journal of Exposure Analysis and Environmental Epidemiology, 6: 311–326.
200
Chapter 6
Model Structure
This chapter contains a description of a simulation modeling framework I have
devised to explore multizonal indoor exposure to airborne pollutants, specifically
for residential secondhand smoke (SHS). The form of the model follows that of
previous workers as discussed in Chapter 2. The model framework is flexible, allowing for specification of either single-valued or randomly-sampled input parameters. Stochastic inputs may take the form of empirical distributions or probability
models, such as lognormal or normal, and sampling of these inputs can follow a
stratified design. While the framework has been specifically designed to study residential SHS exposure, it can easily be expanded, through specification of a new
set of inputs, to study a variety of indoor pollutant sources and indoor settings.
In Chapters 3−5, I presented a critical review of empirical data on cigar and
cigarette emissions, characteristics of the US housing stock, and human timelocation profiles, illustrating the considerable data resources that are available to
support a modeling study of SHS exposure in US homes. In these chapters, I identify central tendencies or best estimates for each physical quantity, although the
data support the modeling of exposures across a wide range of housing types,
emissions patterns, and household occupant activity patterns. But rather than predicting the frequency distribution of exposure for a large population, such as the
US, the goal in this dissertation is to use a fairly small domain of possible inputs
to explore and quantify the effects of a few key variates on residential SHS exposures. In Chapters 7−9, I design and execute simulation trials to study the multi-
CHAPTER 6. MODEL STRUCTURE
201
compartment character of a typical house, and particularly how door and window
positions might accentuate multi-compartment effects and facilitate the mitigation
of SHS exposure. Even though I limit my analysis to one or two sets of emissions
and house characteristics, including a narrow or fixed set of air flows, emission
rates, and other physical and environmental parameters, wide variation in human
behavior patterns is expected to result in similarly large variation in exposures.
In this chapter, I first present the overall structure and flow of the simulation
model for residential SHS exposures, which allows for arbitrary variation in model
input. I then describe the specific features of the model, outlining how: (1) simulated residences are constructed, (2) house-related air flows are assigned, (3) the
movements of people are mapped to a simulated house, (4) smoking behavior is
simulated, (5) simulated events are synchronized, and (6) pollutant concentrations
and exposures are calculated. In the final section, I summarize the main response
(output) and explicit or implicit key (input) variables that are used in simulating
SHS exposures.
6.1 Model Design
Any exposure model must take into account the processes by which pollutant
emissions come into contact with the biological boundaries of a human being.
As discussed in Section 2.1.1, exposure models formalize the exposure process by
matching pollutant concentrations and people across time and space. In a discrete
formulation, integrated exposure is calculated as the sum of time-averaged concentrations corresponding to exposure episodes in different locations weighted by
the duration of each episode. When the timing of concentrations, personal movement or activity, and household conditions is arbitrary, and dynamics are fairly
rapid, i.e., on the order of minutes, such as for the residential system treated here,
then exposures are best characterized as a highly-resolved time series where different segments are aligned with particular rooms.
The model I use to explore residential SHS exposures tracks the individual
minute-by-minute movement of a smoker-nonsmoker pair as they travel among
202
rooms of a house over the course of a single day, which starts and ends at midnight, and, in the case of the smoker, smoking individual cigarettes in designated
areas. It also tracks key aspects of the house configuration over time, including the
opening and closing of windows and doors in particular rooms, the operation of a
central air system, i.e., an HAC or HVAC system, and per-room recirculating filtration or local ventilation. By precisely tracing these events across time and space,
the model is capable of resolving the impact of individual events on simulated
SHS exposure. For example, the exact timing of a smoked cigarette, the smokingroom door position, and the location of the nonsmoker, can lead to very different
exposures both during the smoking episode and when integrated over the course
of the entire day. Using highly resolved characterizations of household events, the
model can generate per-event and per-room exposure statistics (e.g., mean, maximum, minimum) in addition to 24-h integrated exposure, inhalation intake, and
intake fraction, which has been formally defined by Bennett et al. [2002].
The model can be executed for one or more households. To support a range
of possible model inputs for a simulated population, the model accepts lists of
values for each simulated household corresponding to most human or building
input parameters. These lists may contain a single value, which would be used
for all simulated households, or many thousands of values, which may either be
selected sequentially or at random. In some cases, if a list of input values is not
provided, the model will simulate the required value(s) automatically.
Human activity input parameter lists, passed in the form of composite timeactivity objects, are expected to be accompanied by a parallel list of associated
characteristics, such as age, gender, day of week, geographic region, season, or
housing type. These lists of characteristics are used to match individuals to a particular house, to match activities by day of week, or to produce matched pairs of
an adult smoker and a nonsmoker, e.g., a child, spouse, parent, or grandparent. In
addition, for population calculations, these characteristics may be used to simulate
a stratified sample.
Figure 6.1 depicts the logical flow of the model starting with selection of the
203
house and its occupants, proceeding through scenario selection and interzonal
air flow assignment, and ending with the calculation of room concentrations and
smoker and non-smoker exposure and intake. Each step in the simulation progression makes use of information produced in previous steps. Below, I present more
detail on each step of the simulation procedure. A description of the software used
to implement the model, including details on the central exposure simulation function, is given in Appendix D.
6.2 Treatment of Chemical Species
The simulation model can produce SHS concentrations and exposures for any
chemical species for which cigarette emissions rates and surface-interaction coefficients are available. The three major component species of SHS I treat in the
current work are carbon monoxide gas (CO), respirable suspended particles (RSP),
and nicotine (see Chapters 2 and 3). While CO is non-reactive and non-depositing,
and therefore its treatment only requires an emission rate, SHS particles will deposit on room surfaces, such as furniture, carpets, or drapes, and nicotine will sorb
to and desorb from the same surfaces. Currently, only a single type of surface in
each room of the house is differentiated in the simulations. The indoor air quality
(IAQ) component of the simulation model, which describes pollutant dynamics
and generates time series of pollutant concentrations in different rooms of house,
is described below. The explicit model equations are given in Appendix B.
The dynamics of airborne particles vary according to particle size. The size
distribution of SHS-particle mass is fairly well-characterized and lies primarily between 0.02 and 2 µ m with a mass median diameter of about 0.2 µ m (Section 3.4).
Although particle size is an issue in terms of deposition in the human lung, the
current work is limited to consideration of total mass exposure concentrations or
particle intake, which does not incorporate particle uptake in the lung but only
the total mass that is inhaled (and possibly exhaled). In the interest of simplicity, I
have chosen not to take into account differences in particle behavior according to
their size.
204
House
Occupants
No. Floors and Stories
Base Room Connections
Zone Volumes
Smoker Movement
Nonsmoker Movement
Scenario
Inter-Zonal
Flows
Room
Concentrations
Smoking Rooms & Times
Door Positions
Window Positions
Central Air Operation
Filtered Recirculation
Local Ventilation
Exposure &
Intake
Figure 6.1: The logical flow of the simulation model used to predict SHS-particle,
CO, or nicotine exposure occurring in the US population of single-unit houses.
First, a house of a particular size and layout is chosen and populated with smoker
and nonsmoker individuals who are closely tracked in time as they move about the
house. A particular scenario is assigned to each household reflecting a variety of
possible exposure control strategies or lack thereof. Based on the chosen scenario,
a timeline of air flows between zones of the house is established, which is used to
calculate SHS concentrations for each room and SHS exposure time series for each
occupant.
205
Based on reported deposition rates and the mass size distribution of SHS particles, a reasonable size-integrated value for SHS particle deposition appears to be
0.1 h−1 (see Section 3.6). Although particle deposition is an appreciable removal
mechanism, the value used in the simulation of SHS exposures is not expected to
be critical, since removal is likely to be dominated by ventilation, which can give
removal rates on the order of 0.5−1 h−1 or more for exchange with the outdoors
and as much as 5 h−1 for exchange between rooms. In addition, deposition, as it
relates to occupant exposure, is not under direct study.
6.3 Treatment of Residences
The critical features of a home, in terms of the disposition of pollutant emissions
and the subsequent exposure of its occupants, are the size, outdoor connectivity,
and inter-connectivity of its rooms, which determine how emitted pollutants are
dispersed within the house or removed from the house. The model is capable
of incorporating realistic home layouts, including numbers of rooms, numbers of
floors, and room volumes, and, ultimately, in simulating time-varying air flow patterns for a range of potential connection configurations, e.g., door and window positions and the intermittent operation of an HAC/HVAC system. The dimensions
and layout of the house are typically simulated during model execution or constructed separately and then provided as an input list. In the course of an exposure
simulation, connections between rooms, the outdoors, and the house HAC/HVAC
system are translated into actual air flow rates, which may vary in time as the door,
window, or HAC/HVAC configuration changes.
6.3.1 Specification of Volume, Rooms, and Layout
Unless a list of pre-defined house specifications are provided as a model input, the
model automatically simulates a house based on empirical values for house volume and the number of rooms and stories in the house (see Chapter 5). As the
number of rooms in each simulated house increases, the types of rooms are built
206
up from a small number of multi-use rooms to separate kitchen, dining, living,
sleeping, and auxiliary areas (the “main” living areas). Table 6.1 shows this progression starting with a “one room” house containing a combined kitchen, dining,
living, sleeping, and auxiliary area, and continuing up to a room with six completely separate main rooms. For multiple floors, the main rooms are placed on
upper levels in the order shown (i.e., the KIT-DIN dual-use room will always be
on the first level).
The house simulation does not distinguish among different types of auxiliary
rooms such as offices, second living rooms, or extra bedrooms. All houses have
a separate bathroom and a hallway connecting multiple rooms on each floor, although these spaces are not considered to be main living areas. The total volume
of the house is evenly divided among each floor with a predefined portion allocated to hallways. Separate rooms that serve the functions of kitchen, dining
room, living room, or master bedroom are weighted equally with auxiliary rooms
weighted
1
3
less. Supplemental rooms, including bathrooms, laundries, basements,
and garages, contribute an extra pre-defined volume to the house. Basements are
assumed to contribute one floor’s worth of extra volume.
The model applies a set of “connection rules” to determine how air can travel
among the rooms of the simulated house. Besides single-use dining rooms, all
main rooms are linked via a centralized hallway. There are no direct connections
between main rooms, except for single-use kitchens, dining rooms, and/or living
rooms, which are connected by a permanently open doorway. Adjacent floors are
linked through doorways between hallways on each floor, or directly to a single
multi-use room (or basement) on a given floor, if there is no hallway. A bathroom
on the same floor as the master (or only) bedroom is considered to be a master
bathroom inside of the bedroom, otherwise bathrooms are connected to the floor
hallway. Laundry rooms are connected to a single-use kitchen on the first floor,
if one exists. Otherwise, they are connected to the floor hallway. All rooms are
allowed to have a direct connection to the outdoors, through which air may flow
either via building cracks or an open door or window. Although it goes against
207
Table 6.1: Simulated Separate and Multi-Use Room Types as a Function of the
Number of Main Rooms in a House
No. of Main Rooms
Separate & Multi-Use Rooms
1
KIT-DIN-LIV-BED-AUX, BATH, HALL
2
KIT-DIN, LIV-BED-AUX, BATH, HALL
3
KIT-DIN, LIV-AUX, BED, BATH, HALL
4
KIT, DIN, LIV-AUX, BED, BATH, HALL
5
KIT, DIN, LIV, AUX, BED, BATH, HALL
6
KIT, DIN, LIV, AUX, AUX, BED, BATH, BATH, HALL
Abbreviations: KIT, kitchen; DIN, dining room; LIV, living room; BED, master (or only) bedroom;
AUX, auxiliary room such as office, second living area, or extra bedroom; BATH, bathroom; HALL,
hallway. The KIT, DIN, LIV, BED, and AUX are considered main living areas and included in the
room count. HALL and BATH are supplementary rooms. In addition to the rooms shown, a garage,
basement, or laundry room may be included in a given simulated house.
current ASHRAE recommendations, most forced-air systems for residences in the
US do not introduce fresh outdoor air, i.e., they comprise an HAC system with no
ventilation component. However, leaks in supply ducts may lead to the enhanced
exchange of air between the residence and the outdoors when an HAC system is
operating (see Section 5.4).
The above connection rules can be visualized with directed graphs containing
nodes for each type of room and connecting arrows (“edges”) representing pathways of non-zero air flow. For example, Figure 6.2 shows a simple house configuration with three main rooms, including a combined kitchen and dining room,
a combined living and auxiliary room, and a bedroom. A hallway connects the
main rooms and there is a bathroom located in the bedroom. Figure 6.3 shows the
more complicated case of a two-level house containing six main living rooms with
inter-connecting hallways and two bathrooms, one of which is contained within
the master bedroom on the second floor. For both configurations shown, each
room is connected to the outdoors and an HAC system, which supplies air to each
main room and upper-level hallway and receives return air from the lower-level
208
BED
LIV−AUX
HALL
KIT−DIN
BATH
Outdoors
HAC
Figure 6.2: A directed graph depicting interzonal flows for a simulated singlestory house with a single bedroom and attached bathroom, a combined living and
auxiliary room, a combined kitchen and dining room, and a central hallway. Each
room has a connection to the outdoors and either a supply or return register to the
HAC system. For the case shown, supply duct leakage causes loss of HAC air to
the outdoors.
hallway. Simulated supply leaks create a loss of HAC air to the outdoors.
6.3.2 Specification of Air Flow Conditions
In this section, I describe a general procedure for simulating residential air flow
patterns. Examples of simulated air flows for four scenarios are presented in Figure 6.4. I also present specific results of the procedure in Chapters 7 and 9 in association with the simulation of residential SHS exposures. Broadly, I model the flow
of air into and out of the rooms in each simulated house by establishing a base
state where all interior doors are open, all exterior windows are closed, and the
central air handling equipment is turned off. This base state is perturbed during
particular time intervals as conditions change, such as when one or more windows
209
BATH
HALL
BED
AUX
LIV
AUX
DIN
HALL
KIT
BATH
Outdoors
HAC
Figure 6.3: A directed graph showing interzonal flows for a simulated two-level,
six-room residential system. Each level has a bathroom and a centralized hallway.
Flows between rooms on each level are mediated through the hallway, except for
the kitchen, dining room, and living room, which are connected directly. Each
room has a connection to the outdoors and either a supply or return register to the
HAC system. For the case shown, supply duct leakage causes loss of HAC air to
the outdoors.
is opened or the HAC system is activated.
The characteristics of the base flow state depend on whether flows across building boundaries can be considered to be symmetric, i.e., balanced in either direction,
or asymmetric, i.e., unbalanced. Symmetric flows across the building shell might
occur when there are no strong directional driving forces, such as wind or temperature differences, or when there are temperature differences between the indoors
and outdoors that draw air in and push it out of rooms in equal measure, such as
for a vertical stack effect in a single-story structure. Asymmetric flows are likely
to occur when wind drives flow from one side of the house to another. Under
210
BED−3
LIV−AUX−2
HALL−4
BATH−5
KIT−DIN−1
HAC−7
Outdoors−6
1
2
3
4
5
6
7
1
2
3
4
5
6
7
Base State
1
2
3
4
5
0
0 100 0
0
0 100 0
0
0
100 100
100 100 100
0
0
0 100 0
30 30 30 10 3.5
0
0
0
0
0
6
30
30
30
10
3.5
7
0
0
0
0
0
0
0
2 Windows Open, Symmetric
1
2
3
4
5
6
0
0 100 0 180
0
0 100 0 180
0
0
100 100 30
100 100 100
0 10
0
0 100 0
3.5
180 180 30 10 3.5
0
0
0
0
0
0
7
0
0
0
0
0
0
1
2
3
4
5
6
7
1
2
3
4
5
6
7
HAC Active
1
2
3
4
5
0
0 432 0
0
0 432 0
0
0
471 100
100 100 100
0
0
0 139 0
63 63 63 10 7
299 299 299 0 35
6
7
30 0
30 0
30 0
10 1035
3.5 0
0
104
2 Windows Open, Asymmetric
1
2
3
4
5
6
7
0
0 452 0
0
0
0
0 100 0 335 0
0
0
100 135 35 0
100 435 169
0
0
0
0
0 100 0
35 0
352 0
0 52 0
0
0
0
0
0
0
0
Figure 6.4: Simulated air flow rates [m3 h−1 ] between zones of the simple house
(Figure 6.2) for four illustrative scenarios. The direction of flow is from zones listed
in rows to zones listed in columns. For asymmetric flow, inlet rooms are KIT-DIN
and HALL with all other rooms as outlets. For open-window scenarios, windows
are opened in KIT-DIN and LIV-AUX. The procedure for simulating residential air
flow patterns for each scenario is described in the text.
211
symmetric conditions, base state indoor-outdoor ventilation rates are assigned to
each room of a house in proportion to their volume. For asymmetric conditions,
the same total whole-house ventilation rate flows from the outdoors into a set of
“inlet” rooms, and to the outdoors via a separate set of “outlet” rooms. For symmetric base cases, the inter-room flows are balanced in either direction, but for
asymmetric cases, there is an air flow directionality through the house, i.e., a current, flowing between “inlet” and “outlet” rooms.
Representative values for base flows, which are used as default input parameters in the simulation model, are presented in Chapter 5. A representative value
for whole-house house leakage air-exchange rate is 0.5 h−1 , which is distributed
to rooms in proportion to their volume. The default symmetric flow across open
doorways is 100 m3 h−1 in both directions. Under asymmetric conditions, the path
of leakage flows through interior doorways for the base state is determined using
a heuristic for balancing flows, which is described below. To allow for backflow
across door boundaries, symmetric flows are superimposed over the asymmetric
flow path. In this way, asymmetric cases have comparable flows as for symmetric
cases, but with a relatively smaller flow directionality added to a base bidirectional
flow. This approach prevents bias from entering comparisons of SHS exposures for
asymmetric versus symmetric cases.
When the base state is perturbed by the closing of interior doors, default symmetric flows across doorways and symmeric contributions to asymmeric flows are
reduced from 100 to 1 m3 h−1 . For open-window perturbations, the overall house
outdoor air-exchange rate increases by a default amount equal to 150 m3 h−1 times
the number of open windows. For symmetric flow, 150 m3 h−1 is added to each
room’s leakage flow in either direction, thereby maintaining a room’s flow balance.
When flows are asymmetric and both inlet and outlet windows are open, the same
flow increase occurs simultaneously in each group of windows, divided equally
among windows in each group. If only inlet, or only outlet windows are open, the
flow increases for that group and the flow is balanced by adding infiltration or exfiltration to all rooms in the house. As with the asymmetric base state, closed-door
212
and/or window perturbations of the asymmetric base state are followed by application of the flow balancing heuristic described below to establish the direction of
excess air flow through the interior of the house.
For HAC perturbations of base states, a total of five house volumes per hour
of air flow, by default, are distributed to rooms with supply registers, in proportion to room volume. By default, 10% of the flow leaks from supply ducts to the
outdoors and is added back as infiltration into each room from the outdoors (in
response to depressurization). By default, no outdoor air is allowed to enter the
HAC system from the outdoors via designed fans. The activation of the HAC system induces recirculation of air so that it travels from supply vents in multiple
rooms to, usually, a single centrally-located return vent. When all interiors doors
are open, the supplied and returned flows are balanced. However, when some interior doors are closed, some of the supplied air may be diverted to the outdoors.
As with all asymmetric cases, the case of HAC perturbations, whether for the case
of open doors or closed doors, requires application of a flow balancing heuristic to
establish the directionality of flow through interior house boundaries.
A Heuristic for Balancing Flows
Because of the continuity requirement, flow into and out of the house, and into and
out of each room, must essentially be balanced. Unbalanced interior flows arise for
the base state under asymmetric flow conditions after per-room ventilation rates
have been assigned to “outlet” and “inlet” rooms. They also arise whenever a
base state, whether under asymmetric or symmetric conditions, is perturbed by
the HAC system potentially in combination with closed interior doors, or when
a base state under asymmetric conditions is perturbed with closed doors or open
windows.
For these cases of imbalanced flow, I have devised a heuristic to bring them
into balance. Generally, the procedure is as follows. Excess interior flow is fixed
for rooms with a single open or closed interior door with all excess flow going
through an open door (either incoming or outgoing), and excess flow divided be-
213
tween closed doors and the outdoors when there are no open doors. When doors
are closed, 10% of excess flow in a room (e.g., from HAC, infiltration, or an open
window) is assigned, by default, as exfiltration/infiltration for that room and the
rest (90%, by default) passes through the closed door, to the HAC return or to/from
an open outlet/inlet room. For rooms with multiple open doors, the excess flow is
evenly divided among them. As a last resort, excess flow can always be vented to
the outdoors.
The directionality of asymmetric or HAC-driven flows that are assigned at the
outset before the balancing procedure result in a definite flow pattern from inlets to
outlets or supplies to returns when the flows are balanced. When both are active,
the larger flows will dominate. The balancing procedure given here has only been
tested for house floorplans where air from most rooms is mediated by a central
hallway, but there may be a bathroom contained wholly within a bedroom, or a
dining room that is connected only to the kitchen and living room.
6.4 Treatment of Residential Activity Patterns
A key driving force in determining exposures is the interaction of time-activity
patterns for source and receptor individuals. However, precise data on smoking
activity in different rooms of a house are unavailable as are precise data on the
operation of central air systems or the positions of doors and windows. Therefore,
the simulation model generates hypothetical scenarios for activities by integrating
data from various sources, e.g., reported smoking rates and ventilation duty cycles, and considering specific exposure control strategies such as avoidance of a
smoker, isolation of a smoker, closing doors of smoking rooms, or opening a window whenever smoking occurs.
The location of the receptor and source individuals, and especially the relationship between their time-locations, plays a key role in determining exposure
outcomes and, fortunately, a substantial amount of data on the in-home locations
people visit is available as part of the USEPA’s National Human Activity Pattern
Survey (NHAPS) (see Chapter 4). Individual NHAPS respondents gave minute-
214
Table 6.2: Room Categories for NHAPS 24-h Diaries
Main Rooms
Bedroom; Kitchen; Dining Room; Living Room
Auxiliary Rooms
Office/Study
Optional Rooms
Utility Room/Laundry Room, Basement, Garage
Other
Bathroom, Moving from Room-to-Room (Hallway)
by-minute diaries for a single “diary day”, including the time they visited the indoor residential locations in their own home as listed in Table 6.2. These data provide a source of information about the duration and sequence of time spent in
various rooms of their home. Figure 6.5 illustrates the character of the NHAPS
time-activity data using plots of stacked location timelines, where each individual is represented by a thin horizontal strip with colors designating the different
rooms they were reported to visit.
I use the time-location NHAPS data for nonsmokers to provide simulation inputs for both smokers (over age 18) and nonsmokers. Smoking activity is superimposed on top of the in-home movement for the designated smoker. Smoker movement is likely to be biased if there are existing household smoking restrictions, so
it appears more accurate to use nonsmokers to represent unrestricted smokers. By
default, the NHAPS data are used to provide data on the “natural” locations of
persons in their homes, as reported in the unmodified respondent diaries. However, the model can also modify their location patterns as part of scenarios that
involve avoidance of the active smoker by the nonsmoker, or the isolation of the
active smoker in designated rooms. These and other options related to exposure
mitigation strategies are described below.
6.4.1 Mapping Sampled Occupant Locations to Simulated Rooms
While the time-activities of NHAPS respondents are fairly specific in terms of what
rooms are visited, there is a lack of specific information on the precise layout of
respondent’s homes including the exact number of bedrooms, bathrooms, living
215
Residential Time−Location Data
0.6
0.4
0.2
0.0
0.8
1.0
(NHAPS; n=139)
0
200
400
600
800 1000 1200 1400
Time of Day, min after midnight
Kitchen
Livrm, Familyrm, Den
Dining Room
Basement
Laundry room
Bathroom
Bedroom
Study, Office
Garage
Other Location
Figure 6.5: Plot showing the location time series for a sample of 139 NHAPS respondents. The event time series for the sample are represented by 139 vertically
stacked time strips where different colors correspond to times when an individual was reported to occupy a particular house location. White space corresponds
to time when an individual was reported to be in a location other than their own
home. The horizontal axis stretches across a single 24-h period, starting and ending
at midnight.
216
areas, or utility areas and what one or ones the respondent, in fact, visited on their
“diary day.” Therefore, in keeping with a desire to simulate realistic house sizes,
I have devised a standardized approach to mapping NHAPS time-activities to the
complement of rooms that are part of a given house layout. Lacking specific information from NHAPS, I superpose information based on the age of the respondent,
assigning adults to a master bedroom and bathroom and children to an auxiliary
space for sleeping and to a hallway bathroom located on the same floor as their
sleeping area. For houses with a limited number or rooms, occupants may be required to use the same bathroom, and an adult smoker and child nonsmoker may
be required to sleep in the same room (i.e., in LIV-BED-AUX or KIT-DIN-LIV-BEDAUX). Adult smoker and nonsmoker pairs with ages within 10 y of each other
always sleep in the same room, presumably as spouses or cohabitant couples.
6.4.2 Specification of Mitigation Scenarios
The model has a total of 23 binary input options for specific kinds of exposure mitigation strategies or related environmental and behavioral conditions. There are
options that allow or disallow smoking in each type of room and direct either the
nonsmoker or smoker to close doors or open windows during smoking episodes.
The smoker may be forced to not smoke inside the house while the nonsmoker is at
home. Other important options include the operation of portable filtration devices
or the use of local ventilation in either designated smoking or nonsmoking rooms,
and the operation of an HAC system.
Two important options result in modification of either the smoker’s or the nonsmoker’s time-location patterns. The first option forces the smoker to move to
one or more designated smoking rooms during smoking episodes. The number
of cigarettes the smoker consumes in the house remains unchanged. The second
option moves the nonsmoker to a room not occupied by the smoker during smoking episodes. These two options, in combination with scenario input variables
described below, can be used to set up a single, designated smoking room, in
which the smoker must always smoke and which the nonsmoker never visits dur-
217
ing smoking episodes. The rooms to which smokers and nonsmokers are moved
are randomly selected from available rooms. If no designated smoking or refuge
rooms are available, then the occupants are moved to an outdoor location.
When the nonsmoker and the smoker are present in the same room in the absence of any conscious strategies to reduce SHS exposure for the nonsmoker, the
door and window behavior of either occupant can take precedence. Without mitigation strategies for doors and windows, occupants only close their doors when
sleeping or in the bathroom, and they always leave windows closed. However,
when occupants close doors for the purpose of reducing the transport of SHS or
open windows to increase ventilation, the precedence of one or the other occupant
may actually result in increased SHS exposure for the nonsmoker. Therefore, when
any door-closing mitigation strategies are in effect and the smoker and nonsmoker
occupy the same room, the door is always left open during the smoking episode to
avoid build-up of SHS. Similarly, when any window-opening mitigation strategies
are in effect and the occupants are in the same room during smoking episodes, the
window is always left open to increase removal of SHS.
6.4.3 Simulation of Smoking Patterns
Smoking behavior is simulated by first picking the total number of cigarettes that
are smoked in a single day. This number is 1.5 packs, by default, where there are
20 cigarettes in a pack. I assume that these cigarettes are smoked in even intervals
throughout the smoker’s waking hours. By default, those cigarettes that are not
smoked in designated smoker areas during the unmodified smoker location profile are dropped. The dropped cigarettes constitute a reduced number of cigarette
smoked if they were slated to occur in locations that are inside the house. In other
words, the smoker does not move to a designated smoking area in the house, or
to the outdoors, when their "time to smoke" happens to fall during a time when
they are in a nonsmoking house location. However, if the option for moving the
smoker to a designated smoking room is specified, then, as described above, the
smoker will consume all of their cigarettes that would normally occur in the house
218
as long as a designated smoking room is available. A scenario option also exists to
prevent the smoker from smoking in the house when the nonsmoker occupant is
at home. In this case, the smoker’s location profile in the house does not change,
but none of the cigarettes they would have smoked are allowed to be active.
6.5 Combining House and Occupant Information
A central and critical function of the exposure simulation model is to synthesize
time profiles of house and occupant-related information, generating room-specific
concentrations and 24-h exposure profiles. First, the time profiles for air flows and
pollutant emissions in each room are determined based on integrated and synchronized timelines for household events, including smoking activity, occupant
locations, door and window positions, HAC operation, and portable filtration activity. Next, the air flows and emissions are used in concert with other physical
data about the house, such as volumes, surface-to-volume ratios, and surfaceinteraction coefficients, to calculate room concentrations. Finally, the room concentrations and receptor time-location profile are merged, allowing for the calculation of exposure. Since the activity patterns for smoking and nonsmoking
occupants contain minute-by-minute information, it is possible to maintain this
level of precision throughout the simulation process. Figure 6.6 contains an integrated plot of time-profiles for selected household events, room concentrations,
and smoker/nonsmoker exposures calculated for an illustrative simulation.
6.5.1 Synchronization of Simulated Events
The time series for the different types of household events that are considered by
the model are combined into an integrated timeline to establish the time breaks
over which air flows, room concentrations, and exposures are to be determined. To
accomplish this task, I have devised a unified time-activity "master event object,"
which consists of a single vector of time breaks and a number of parallel vectors
containing codes for the different events that occur between each of the breaks. The
219
KIT−DIN
0 250
0 250
Concentration and Behavior Profiles
BED
0 250
0 250
LIV−AUX
0 250
HALL
BATH
Smkr Location
Cigarettes
Smkr Awake
0 250
Smkr Drs Open
Smkr Exp Conc
Nonsmkr Location
Nonsmkr Awake
0 250
Nonsmkr Drs Open
Nonsmkr Exp Conc
200
400
600
800
1000
1200
1400
Elapsed Minutes
Figure 6.6: Example simulated 24-h time-profiles for room particle concentrations [µ g m−3 ]
(top panels) and selected occupant-specific behavior patterns and exposure concentrations
[µ g m−3 ] (middle and bottom panels), occurring for the base flow state of the simple house
layout shown in Figure 6.3. Each profile starts and ends at midnight. Occupant-specific activity profiles are included for the cigarette, location, awake, and door behavior of a single
smoker and nonsmoker pair. The simulated exposure profile for each person is positioned
below each group of behavior profiles. Colors used to draw each room concentration are
matched with the color coding of the location profiles. White space in the activity profiles corresponds to an “absent from house”, “inactive”, “asleep”, and “closed” condition
for location, cigarette, awake, and door profiles, respectively. Color and gray segments
correspond to the opposite condition.
220
unified object contains the minimum number of time breaks required to retain the
information in each original event series. By merging each of the event time series,
I create a common time reference for subsequent calculations. Given a particular
time interval, the unified event object allows for immediate determination of the
occurrence or non-occurrence of specific events, such as whether or not a smoker
is active, what location the smoker is in, what location the receptor is in, whether
the door is closed in the smoking room, or whether the window is open in the
smoking room.
6.5.2 Calculation of Room Concentrations and Exposure
To calculate dynamic pollutant concentrations in each room of the house, I use a
generic multizone IAQ model that incorporates time-dependent information on air
flows and pollutant emissions. Appendix B describes the model formulation and
its solution. According to the flow simulating procedure described above, the master household event object is used to generate the interzonal flows for each distinct
household configuration that occurs during the 24-h simulation period. The flows
and room-specific pollutant emissions profiles are recoded into regular time intervals with minute resolution before they are input into the IAQ model. Such a high
resolution captures the highly dynamic nature of smoking behavior and human location patterns, and it also has the advantage that instantaneous concentrations are
approximately equal to time-averaged concentrations across each interval, with a
relatively small price to pay in terms of computational time.1 In addition, minutelong regular intervals allow for precise peak, per-event, and per-room exposure
statistics to be calculated.
The minute-by-minute SHS exposure concentration time series for the smoking
and nonsmoking occupants are determined by matching the minutes they spend
in each room of the house with the calculated concentrations in each room. In
this way, a measure of the confluence of a person and a series of concentrations is
1 The
computer time required to simulate a 24-h period for a single household, including occupants, household events, concentrations, and exposures, is typically 5−10 s.
221
determined, conforming to the accepted definition of exposure (see Section 2.1.1).
Based on each exposure concentration time series, integrated, 24-h mean, peak,
and per-room mean exposures are calculated. The exposure time series are also
used, in combination with at-rest inhalation rates tabulated by age and gender
[Layton, 1993] and the total mass of pollutant emissions, to calculate a pollutant
intake fraction for each person, based on time spent at home and in-home SHS
emissions.
6.6 Summary of Input and Output Simulation Variables
This section summarizes all of the input and output simulation variables that are
used to design and manage residential SHS exposure simulation trials. Each variable performs a role as a “key” or “response” variable. Table 6.3 contains the response, or output, variables, which depend directly on the set of inputs and constitute the primary variables of interest. Broad mitigation scenario specifications,
together with fine-tuned control over a IAQ model input parameters and personal
data, allow one to design a great variety of predictive and exploratory simulation
experiments. Table 6.4 lists all of the model input parameters, or key variables,
which explicitly condition the model response. Page references are included for
locations in this dissertation where appropriate input parameter values have been
discussed.
For simplicity, the 23 explicit key variables for mitigation scenarios take the
form of binary switches, with values corresponding to either an "on" or an "off"
condition. The "on" condition corresponds to the states listed in Table 6.5. The "off"
condition corresponds to the opposite state, e.g., having windows open instead of
leaving them closed, or allowing smoking only when others are not at home. Each
of the "on" states, when taken by themselves, would be expected to lead to more
potential SHS exposure for the nonsmoker with respect to the corresponding "off"
state.
The final set of simulation variables are those that arise implicitly in the course
222
Table 6.3: Model Response Variates
No.
Variate Description
Units
1
24-h exposure concentrations
µ g m− 3
2
Per-room mean exposure concentrations
µ g m− 3
3
Peak exposure concentrations
µ g m− 3
4
Intake fraction
−
5
Equivalent SHS cigarette intake
−
6
Ratio of paired smoker and nonsmoker (indirect) exposure
concentrations and intake fractions
−
7
Exceedance of state and federal air quality standards, binary
digit (1=exceeded; 0=not exceeded)
−
of the simulation (Table 6.6). Like the model outputs, they are dependent on the
chosen model inputs, but are used as indicators or summaries of simulated conditions, rather than simulation endpoints. In this way they are also a type of key
variable, because they may be used to disaggregate the model outputs. They arise
in the course of the simulation when aspects of the simulated house or occupant
activity are calculated for a particular scenario. These implicit key variables are not
specified as model inputs, because they depend on the interaction between house
occupants and their environment. For example, the correlation of smoker and nonsmoker behavior, i.e., the percentage of time they spend in the same room, is likely
to be an important indicator of nonsmoker exposure, but must be determined from
the randomly sampled activities of each occupant. Another example is the computation of total pollutant mass emissions, which is necessary for the calculation of
intake fraction and depends on the number of cigarettes smoked in the home.
Time-activity for occupants who smoke: time breaks, location
codes, and awake codes
Time-activity for occupants who don’t smoke: time breaks,
location codes, and awake codes
Number of cigarettes smoked in a day
Cigarette mass emission rate
Particle deposition loss-rate coefficient
Sorption and desorption coefficients
3
4
5
6
7
8
Housing data for each occupant, including number of floors,
rooms, and types of rooms
Continued.
Age, gender, geographic region, and education level for each
house occupant
2
9
A mitigation scenario specification given as a 23-digit binary
number
Variate Description
1
No.
Chapter 3, page 120
Chapter 3, page 126
h− 1
m h− 1 , h− 1
Chapter 5, page 168
Chapter 3, pages 116, 126
mg cig−1
−
Chapter 3, page 98
Chapter 4, page 140
Chapter 4, page 140
Chapter 4, page 140
Table 6.5, page 225
a Reference
−
−
−
−
−
Units
Table 6.4: Model Input Parameters – Explicit Key Variates
223
Surface-to-volume ratio for each room of the house
House volume as a function of rooms and floors
Closed whole-house air exchange rate
Interzonal flow rates for open doors
Interzonal flows amplification increments for closing doors and
opening windows
Central HAC system duty cycle and percentage time active
during waking periods
Inhalation rates by activity level, age, and gender
10
11
12
13
14
15
16
Chapter 5, page 182
Chapter 5, pages 174 and 182
m3 h− 1
m3 h− 1
Chapter 6, page 220
Chapter 5, page 172
h− 1
m3 h− 1
Chapter 5, page 168
m3
−
Chapter 5, page 168
m− 1
−
a Reference
Units
to a location in the current chapter or another chapter where appropriate input parameter values have been established or
relevant data bases have been discussed.
a Reference
Variate Description
No.
224
Smoking is allowed in the kitchen.
Smoking is allowed in the dining room.
Smoking is allowed in the main bedroom of the house.
Smoking is allowed in auxiliary rooms, such as an office or extra bedrooms where children may be
sleeping.
Smoking is allowed in the bathrooms of the house.
Smoking is allowed in the basement or garage of the house.
Smoking is allowed when others are at home.
The smoker does not move to a random designated smoking room when he/she smokes.
The nonsmoker does not move to a random room away from the active smoker.
The smoker leaves the door open in rooms where he/she smokes.
The smoker leaves the window closed in rooms where he/she smokes.
2
3
4
5
6
7
8
9
10
11
12
13 The nonsmoker leaves their room door open while smoking is happening in other rooms of the house.
Continued.
Smoking is allowed in the living room of the house.
"On" Condition
1
No.
Table 6.5: List of "On" Conditions for the 23 Environmental Scenario Binary Variables
225
"On" Condition
The nonsmoker leaves their room window closed while smoking is happening in other rooms of the
house.
There is no local ventilation in smoking rooms during smoking.
There is no local ventilation in nonsmoking rooms during smoking.
Changes in smoker/nonsmoker location, door and window position or location ventilation in response to
smoking activity occur only when a cigarette is active (versus for the entire time that a smoker spends in a
designated smoking room).
A recirculating filtration unit in smoking rooms is left off throughout the day.
A recirculating filtration unit in nonsmoking rooms is left off throughout the day.
The central air handling system is in operation throughout the day according to a particular duty cycle
(i.e., a repeating pattern of on- and off- time periods).
All of the main rooms in the house are combined into a single large zone.
Indoor doorway positions for the “base case” are randomly assigned instead of being all open. Doors will
still be closed as per the scenario variables above.
During non-smoking periods, smoker has precedence over nonsmoker with respect to door and window
positions in rooms that they occupy simultaneously, rather than vice versa.
No.
14
15
16
17
18
19
20
21
22
23
226
227
Table 6.6: Derived Quantities – Implicit Key Variates
No.
Variate Description
Units
1
Number of cigarettes smoked in the house
2
Total mass of SHS-related pollutant emitted into the house
mg
3
Mean whole-house air exchange rate
h− 1
4
Pollutant dispersion coefficient, mean absolute concentration
difference between rooms divided by mean room concentration
across all rooms
−
5
Correlation coefficient for smoker and nonsmoker time-activities,
percentage of time smoker and nonsmoker spend in the same
room over the 24-h period
−
6
Time nonsmoker spends in same room as smoker
h
7
Percentage of time occupants spend at home
−
−
6.7 References
Bennett, D. H., McKone, T. E., Evans, J. S., Nazaroff, W. W., Margni, M. D., Jolliet,
O., and Smith, K. R. (2002). Defining intake fraction. Environmental Science and
Technology, 36(9): 206A – 211A.
Layton, D. W. (1993). Metabolically consistent breathing rates for use in dose assessments. Health Physics, 64(1): 23–36.
228
Part III
Model Application
229
The following three chapters contain the results of applying a simulation model for residential secondhand smoke (SHS) exposure in three
separate tiers of analysis.
Chapter 7 (page 230) contains the Tier I analysis, which is a preliminary investigation of SHS exposure making use of a limited number of
scripted occupant location patterns.
Chapter 8 (page 268) contains the Tier II analysis, which examines frequency distributions of residential SHS exposure arising from realistic
variation in unrestricted smoker and nonsmoker behavior.
Chapter 9 (page 301) contains the Tier III analysis, which builds on the
Tier II analysis of residential SHS exposure by modifying the locations
and door and window-related behavior of smokers and nonsmokers
according to specific exposure mitigation strategies, or by introducing
portable filtration.
230
Chapter 7
Tier I. Analysis of
Multi-Compartment Effects Using
Scripted Occupant Movement
For this chapter, I execute initial simulation experiments using the simulation
model for multizonal residential SHS exposure described in Part II. In a multiple
compartment residential system containing active smokers, the proximity between
receptors and smokers, the magnitude of whole-house removal mechanisms (pollutant persistence), and the magnitude and direction of interzonal air flows (pollutant permeation) are all expected to control SHS concentrations and exposures.
But the ways in which these factors can intermingle is unclear. Therefore, in this
chapter I explore the magnitude of the effect on SHS exposure of a house’s multiple compartment character, including the relative locations of its occupants and
flow patterns induced by either design, occupants, or meteorology.
To represent both the particulate and volatile gaseous components of SHS,
which differ in their interaction with surfaces, I consider emissions of particles and
nicotine. Surface levels of nicotine are either initially zero or moderately loaded to
reflect chronic indoor smoking. Initial flows across house boundaries may be either balanced in each direction, i.e., symmetric, or they may be asymmetric, in
which case flows have a prevailing direction from one side of the house to another. Asymmetric interior flows may also occur for cases of either continuous or
intermittent central air handling operation.
CHAPTER 7. TIER I. SCRIPTED OCCUPANT MOVEMENT
231
Other physical model parameters whose general effects on indoor air quality
and exposure are understood, such as house volume, number of cigarettes smoked,
cigarette emission factors, deposition rates, sorption and desorption rates, flow
rates through doors, windows, and air handling systems, and receptor inhalation
rate, are held fixed for the analyses presented in this chapter. Fixing their values
clarifies and simplifies my analysis of other factors. I select single values of these
parameters that are reasonable or typical for US residences.
The primary focus of this chapter is the initial exploration of how much
occupant-related behaviors and the multiple compartment character of houses can
influence exposure. I consider two houses of identical size, one with a typical 4room layout and one dominated by a single large (and well-mixed) space. These
houses are intended for use in exploring the effect of a house’s multi-compartment
character on exposure and the commonly assumed condition in IAQ modeling
of instantaneous complete mixing throughout a house, rather than for studying
particular house layouts. I present results from a full dynamic treatment of exposure as well as 24-h average exposures from a simplified model, which is derived
in Appendix B. Occupant location and the positions of windows and doors are
expected to be critical determinants of residential SHS exposure. Therefore, I consider three different patterns of receptor location, in which the nonsmoker spends
progressively less time in proximity to the smoker, and three scenarios exploring
the effects of nonsmoking behavior with respect to interior door and exterior window positions. This initial investigation is a prelude to simulations of exposure
using more realistic variation in occupant location patterns, which are introduced
in Chapters 8 and 9.
7.1 Model Input for Scripted Scenarios
I use the same model input parameter values for both the full, dynamic treatment
of residential SHS exposures and the simplified treatment. In the simplified approach, I assume the home is represented by a single well-mixed zone, so that
exposure concentrations vary in time but not in space. Dependence on the com-
232
mon time that occupants spend at home is removed by assigning exposure based
on time-averaged pollutant concentrations. A correction factor, f , is used to compensate for pollutant dispersion amongst rooms and reduced time spent at home
during the averaging period. The simplified model of SHS exposure for airborne
particles can be formulated as follows (Appendix B):
y= f
ÑE
V ( A + D)
(7.1)
where y is the average airborne particle exposure concentration [µ g m−3 ] in the
house over the given averaging time T, which for convenience is set equal to 1 d,
Ñ is the number of cigarettes smoked in the residence over the one-day averaging
time [d−1 ], E is the particle mass emission factor [µ g cig−1 ], A is the air-exchange
rate [d−1 ], D is the deposition loss-rate coefficient [d−1 ], V is the total volume of the
residence [m3 ], and f is the correction, or adjustment, factor. The mass of pollutant
intake for the receptor can be estimated through multiplication of y by an individual’s breathing rate. The factor f will be larger the more time the receptor spends at
home and the more time spent in zones of average or higher-than-average concentration relative to zones of lower-than-average concentration. A value of 1 corresponds to the receptor spending time at home during and for substantial time after
cigarettes were smoked at home in zones with average air concentrations. The factor increases with an increase in time spent in the same room as the smoker, and
decreases if the receptor leaves the home during or shortly after smoking or moves
to a distant, relatively smoke-free location of the home. However, even with an appropriate correction factor, this simple single compartment model of time-average
exposure lacks the flexibility to accurately describe residential exposures in which
pollutant concentrations are significantly different in rooms of a house and occupants travel among multiple rooms.
The dynamic, multizone model incorporates the same parameters as in Equation 7.1, but it also incorporates additional physical and environmental input parameters. These include interzonal flow rates associated with doors, windows,
and an HAC or HVAC system, as well as particle deposition or surface sorption
233
and desorption (Table 7.1). I use fixed input parameter values, which represent
best estimates falling in the middle range of reported values (Chapters 3−6), for
all exposure calculations presented in this chapter. Note that while each parameter
value is fixed, pollutant emissions and house air flow characteristics change over
time due to the behavior of household occupants or the automatic operation of the
HAC system.
For simplicity, the behavior of the smoker is also held fixed for calculations
presented in this chapter. Figure 7.1 depicts the sequence of locations the smoker
visits in the household, as well as the timing of their smoking activity. The smoker
lights up 30 cigarettes during the day, approximately 19 of which are smoked in
the house.1 As discussed in Chapter 6, cigarettes are evenly spaced throughout
awake periods of the day with a portion of these cigarettes naturally falling within
periods spent at home.
Smoking episodes, shown as solid blocks of time, are defined as continuous
time periods for which a smoker is present in a room where smoking is allowed
and he or she smokes at least a portion of one cigarette. A new smoking episode
occurs whenever the smoker moves to a different room where they may initiate
smoking activity. Smoking is allowed in every room the smoker visits except the
bathroom.
Since, as discussed in Chapter 5, most house mechanical air handling systems
in the US do not provide ventilation, that is the case I explore here. Figure 7.1
shows the HAC system duty cycle for the case of intermittent operation where the
total on-time is fixed at 10% of the day with individual on-time episodes fixed at 10
min. The HAC only operates during times when at least one household member
is awake.
To address the primary focus of my investigation, which is the impact of household occupant behavior and multiple compartments on SHS exposure, I have iden1 The
procedure for simulating smoking activity, which is described in Chapter 6, involves superimposing a smoking pattern onto the smoker’s residential location pattern. In this process,
fractions of a cigarette may fall outside the limits of a time period spent at home, resulting in a
non-integer number of cigarettes being smoked in the house.
234
Table 7.1: Fixed Model Input Parameter Values: Physical and Environmental
Quantities
Smoker Activity
In main bedroom, living room,
kitchen and bathroom during the day;
Smokes in every room they visit
except the bathroom; At home 75% of
day; see text and Figure 7.1
Low-activity Inhalation Rate
0.0054 m3 min−1 , Chapter 6
Cigarettes Smoked Over Whole Day
30 d−1 , Chapter 3
Cigarettes Smoked at Home
19 d−1 between 6am and 10pm a
Cigarette Mass Emissions
10 mg cig−1 , Chapter 3
Cigarette Nicotine Emissions
5 mg cig−1 , Chapter 3
Cigarette Duration
7 min with 25−30 min breaks, Chapter
3
Particle Deposition Rate
0.1 h−1 , Chapter 3
Nicotine Sorption Rated
1.4 m h−1 , Chapter 3
Nicotine Desorption Rated
0.00042 h−1 , Chapter 3
Nicotine Initial Surface Conc.
50 mg m−2 , Chapter 8 b
Door Flow, Open Doorway
100 m3 h−1 , Chapter 5
Door Flow, Closed Doorway
Window Flow Addition, Open
Window
Base House Air-Exchange Rate
0.5 h−1 , Chapter 5
HAC, House Volumes of Flow
5 h−1 , Chapter 5
HAC, Intermittent Duty
10% c
HAC, Intermittent On-Time
10 min c
House Volume
287 m3 , Chapter 5
a Derived
quantity resulting from total cigarettes smoked and smoker location pattern.
b This approximate value for initital surface nicotine concentrations is determined from a simulation
experiment described in Chapter 8.
c HAC is active for 10% of the total time any occupant is awake for “on” periods of 10-min at a time.
For occupants that sleep 13 of the day, this corresponds to 1.6-h total duty, which is considered low.
d The surface-to-volume ratio for each room of the simulated houses, which are needed to simulate
nicotine sorption and desorption, are given in Table 7.3.
235
Midnight
6:00 am
Bedroom
Bathroom
Noon
Kitchen
Living Room
6:00 pm
Cigarettes
Smoking Episodes
Midnight
HAC
Figure 7.1: Concurrent 24-h time-activity patterns for smoker location, individual cigarettes smoked, and smoking episodes in individual rooms of House #2.
For House #1 bedroom, kitchen, and living room functions are served by a single
main room. Also pictured here is the HAC operation cycle. Smoking occurs in
every room the smoker visits during awake periods, except the bathroom. Over
the course of the day, a total of approximately 19 cigarettes are smoked inside the
house.
tified five study factors with 2−6 levels each. These factors, listed in Table 7.2,
correspond to model input scenarios for: (1) number of rooms and layout (House
Type); (2) nonsmoker location relative to the smoker (Nonsmoker Activity); (3)
door, window, and HAC configuration (Flow-Related Conditions); (4) flow patterns across zone boundaries (Flow Symmetry); and (5) the amount of sorbing
pollutant on household surfaces, which is reflective of the history of smoking behavior in the household (Initial Nicotine Surface Concentration).
The general effect of multiple compartments on exposures is explored by treating two homes having identical total volumes but different numbers of distinct
rooms. Home #1 consists of a single 280 m3 room, which satisfies most household
uses, plus a small (7 m3 ) bathroom (Table 7.3). House #2 consists of four main 50
or 100 m3 rooms satisfying distinct cooking/dining, living, working, and sleeping
uses, a 30 m3 hallway, and a 7 m3 bathroom. As evident from the schematics and
flow graphs shown in Figure 7.2, all of the rooms in each house have air connections to the outdoors through either leakage or a window, and to the HAC system
through either an air supply or return register. The air return is located in the main
236
Table 7.2: Levels Considered for Five Model Input Scenario Variables: House Type,
Nonsmoker Activity, Flow-Related Conditions, Flow Symmetry, and Initial Nicotine Surface Concentrations
House Type
1.
“1”
A single main room on a single floor
2.
“2”
Four main rooms on a single floor
Nonsmoker Activity
1.
“Follower”
Nonsmoker moves between rooms exactly with smoker
2.
“Napper”
Nonsmoker with smoker in living room during mid-day,
spending part of the day and the night in a separate bedroom
3.
“Avoider”
Nonsmoker always in different room from smoker, but may
move into rooms where smoking has occurred previously in
the day
Flow-Related Conditions
1.
“Base”
Internal doors all open during awake times, all windows
closed, and HAC off
2.
“Doors Closed”
Internal doors closed in rooms during smoking episodes
where smoker is present
3.
“SmkDrs-Closed/
SmkWins-Open”
Internal doors closed and windows open in rooms during
smoking episodes where smoker is present
4.
“HAC-10%”
Central air handling system is intermittently active for 10% of
waking hours; windows closed and internal doors open
5.
“HAC-100%”
Central air handling system is active for 100% of waking
hours; windows closed and internal doors open
6.
“Smk-NonSmk/
Wins-Open”
Windows open during smoking episodes in rooms where
smoker is present or nonsmoker is present; internal doors
open
Flow Symmetry
1.
“Sym”
Symmetric flow across window/door/wall boundaries, such
as might occur for local stack (temperature) effects or
non-directional turbulent flow. See text.
2.
“Asym”
Asymetric flow across door/window/wall boundaries, such
as might occur with wind-driven flow. See text.
Initial Nicotine Surface Concentration
1.
“0”
No nicotine is on room walls at the start of the simulation
2.
“50”
50 mg m−2 of nicotine are on the walls of each room at the
start of the simulation
Note: See Figure 7.3 for nonsmoker activity patterns. See Table 7.3 and Figure 7.2 for the layout of each house type.
237
room of House #1 and the hallway of House #2 with air supplies going to all other
rooms. The two rooms in House #1 are connected by a doorway. The air connection between rooms for House #2 is mediated by a hallway for all rooms except the
master bedroom and bathroom, which are directly connected by a doorway.
Except for time spent in the bathroom, House #1 occupants occupy the single
main room for the entire time they spend at home. In contrast, for House #2, which
contains four different main rooms, it is possible for the smoker and nonsmoker
to occupy separate rooms for much or all of the time they spend at home together.
I have established three different nonsmoker time-location patterns that provide
perspective on the influence of proximity to the smoker on exposure in House #2
(see Figure 7.3). The first nonsmoker profile, termed the “follower”, corresponds to
an infant, a small child, or perhaps a spouse, who spends all their time in the house
in the same room as the smoker. The second profile, the “napper”, corresponds
to a child or adult who spends some time in the same room as the smoker, but
takes a 2-h nap in the middle of the day in a separate bedroom and sleeps in a
separate bedroom at night. The third profile, the “avoider”, never spends any time
in the same room as the smoker. It is expected that larger amounts of time spent
in the same room as the smoker will result in higher exposures, although time
spent in rooms where smoking has just occurred may also contribute significantly
to cumulative exposure.
Occupants may open exterior windows and/or close interior doors in response
to smoking activity. Such activities result in a perturbation from the base conditions, where all exterior windows are closed and all interior doors are open, except
during time spent in the bathroom or in bedrooms during sleeping hours. Either
the nonsmoker, the smoker, or both may close a door or open a window in the
room in which they are residing during a particular smoking episode. This behavior occurs with respect to complete smoking episodes, so that door and window
positions are altered from the base condition for an entire smoking episode, rather
than being limited to the duration of individual cigarettes.
The treatment of interaction between smokers and nonsmokers in rooms they
LIV
BED
AUX
HALL
BATH
Living Room
Bedroom
Auxiliary Room
Hallway*
Bathroom
KIT-DIN
BATH
Bathroom
Kitchen-Dining*
MAIN
Abbrev.
Main*
a Rooms
7
30
50
50
50
100
7
3.5
2.2
1.9
1.9
1.9
1.6
3.5
1.2
[m−1 ]
[m3 ]
280
Ratio
Volume
287
287
[m3 ]
Volume
Total
+36
−1435
+251
+251
+251
+502
+32
+1260/−1435
[m3 h−1 ]
Flow
c HAC
3.5
15
25
25
25
50
3.5
140
[m3 h−1 ]
Flow
Leakage
d Base
These house characteristics are used for dynamic, multizone simulations of residential SHS exposure. See Figure 7.2 for house layouts
and a graphical depiction of direct connections between rooms (i.e., a closed or open doorway) for each house type. See Chapter 6 for a
detailed description of how flows are assigned.
a Rooms with an asterisk (*) are inlets rooms, i.e., rooms that experience a net inflow of air from the outdoors when asymmetric flow
patterns are specified, and all other rooms act as outlet rooms for which there is a net outflow to the outdoors.
b Minimum surface-area-to-volume ratios corresponding to rooms with listed volumes as calculated in Chapter 5.
c Air flow for supply (+) or return (−) to/from the house HAC system for each room of the house occurring when the HAC is active. The
supply rates shown have been reduced by 10% from the designed flow rate of 5 h−1 to account for supply duct leakage. See Chapters 5
and 6.
d Air flow between each room and the outdoors due to base leakage through building cracks and crevices when the HAC is inactive.
During HAC operation, supply duct leakage leads to additional infiltration.
2
1
No.
to-Volume
Room
b Surface-
Table 7.3: Room Characteristics for Each Type of Simulated House
238
239
HAC
Front
Door
Back
Door
Kitchen-DiningLiving-BedroomAuxiliary,
280 m3
Back
Door
HAC
HAC
Kitchen-Dining,
100 m3
Front
Door
Auxiliary,
50 m3
Hallway, 30 m3
HAC
HAC
Living Room,
50 m3
Bath,
7 m3
HAC
HAC
House #1
Bedroom,
50 m3
Bath,
7m3
HAC
HAC
House #2
KIT−DIN−LIV−BED−AUX
BATH
AUX
BED
HALL
LIV
KIT−DIN
BATH
Outdoor
HAC
Outdoor
HAC
Figure 7.2: Floorplans (top), showing room volumes and HAC supply and return flow
rates, and interzonal connection schematics (bottom) for the two house types considered
in dynamic multizone analyses in this chapter (also see Table 7.3). House #1 has a single
main room satisfying kitchen, dining, living, bedroom, and auxiliary uses. House #2 has
4 main rooms with a multi-use room for cooking and dining purposes. Both houses have
a single floor. Bathrooms and hallways are considered to be supplementary, and not main
rooms. Note that each room of the houses has a bidirectional connection to the outdoors,
e.g., via a window or wall. The main rooms in House #2 are connected by a doorway to
a hallway and not directly to other rooms, except for the master bathroom and the master
bedroom. When operating, the HAC system supplies a total of 5 house volumes of air
per hour divided among individual rooms on a volume basis, minus 10% due to supply
duct leakage, with zero supplied outdoor air. See Chapter 6 for a discussion of how house
air flows are assigned. The HAC return is located in the hallway for House #2 and in the
single main room for House #1. The HAC acts to recirculate all the supplied air between
the return and supply registers.
240
Smoker
"Follower" Non−Smoker
"Napper" Non−Smoker
"Avoider" Non−Smoker
Midnight
6:00 am
Bedroom
Kitchen
Noon
Living Room
Bathroom
6:00 pm
Midnight
Auxiliary Room
Awake Time
Figure 7.3: Scripted smoker and nonsmoker 24-h time-activity patterns used for
the occupants of House #2. The time-activities start and end at midnight. For
House #1, occupants spend all their time in a single, large main room or the bathroom. The locations House #2 occupants visit are designated by different colors
with a gray strip showing the time that each person was awake. Blank spaces
for locations designate periods when the person was outside of the residence. A
“follower” nonsmoker is one who consistently occupies the same locations as the
smoker. A “napper” nonsmoker sleeps in their own bedroom for much of the day,
and shares space with the smoker in the living room. The “avoider” nonsmoker
never occupies the same room as the smoker, using a separate bedroom, and opting to occupy the kitchen when the smoker is in the living room and vice-versa.
241
occupy simultaneously is a particularly important aspect of the simulation. In
keeping with the implicit goal of exposure mitigation associated with door closing
and window opening behavior, when the nonsmoker and smoker are in the same
room during smoking, flow scenarios are applied to result in the least expected
exposure for the nonsmoker. Thus, if either the smoker or nonsmoker is directed
to close doors during smoking episodes spent by themselves, the door of any room
they jointly occupy is left open instead to allow more rapid removal of SHS from
that room. If either the smoker or nonsmoker is directed to open a window during
smoking episodes, then the window of the room they share is always left open to
increase room ventilation.
7.2 Intermediate Output: Occupant Interaction, Air
Flows, and Room Concentrations
Dynamic, multizone simulation experiments were executed according to the fixed
model inputs and all combinations of the scenario levels defined in the preceding
section. Important characteristics of the simulated scenarios include time spent in
proximity to a smoker (Table 7.4), whole-house air-exchange rate (Table 7.5), and
air flow rates between rooms (Tables 7.6 and 7.7). Below, I review these generated
values for general validity and their consistency with expected occupant behavior
(Chapter 4) or housing characteristics (Chapter 5).
All occupants spend 75% of their day at home and 43% of the day at home
while smoking episodes are occurring (Table 7.4). The smoker and the “follower”
nonsmoker spend all of their time at home in the same room of the house. For
House #1, the “napper” and “avoider” nonsmokers also spend most of their time
in the same room as the smoker, since movement is restricted to a single multipurpose room and a small bathroom. In House #2, the “napper” spends considerably less time in the same room as the smoker (17% of the day), and the “avoider”
spends zero time in the same room as the smoker. While the “follower” is always
exposed to direct SHS emissions in a particular room of House #2, the “napper”
is only exposed indirectly through interzonal transport or when entering a room
242
Table 7.4: Percentage of the Daya Nonsmoker Spends in Rooms with the Smoker,
At Home During Smoking Episodes, and in Rooms During Smoking Episodes by
Activity and House Type
Nonsmoker
Activity
House
Type
In Room
with Smoker
Follower
1
2
Napper
Avoider
a All
75%
75%
At Home During
Smoking
Episodes
43%
43%
In Room During
Smoking
Episodes
43%
43%
1
2
72%
17%
43%
43%
42%
17%
1
2
71%
0%
43%
43%
41%
0%
household occupants spend 75% of their day at home.
where a smoking episode has previously occurred. The “napper” receives both
direct and indirect exposures.
The whole-house air-exchange rate (Table 7.5) indicates the degree of pollutant
removal through direct infiltation and exfiltration via the building shell. Under
symmetric flow, where flows are balanced across door and window boundaries,
the base house air infiltration rate of 0.5 h−1 is unchanged between base flow scenario conditions and for smoker closed-door conditions. Asymmetric flows resulted in air-exchange rates that were approximately equal to those for symmetric flows. HAC system operation increased the mean air-exchange rate slightly
in House #2 because of increased infiltration induced by supply duct leakage.
Opening a window in smoker rooms and/or nonsmoker rooms during smoking
episodes resulted in an increase of the air-exchange rate to 0.73−0.95 h−1 . This
variation is due mostly to different nonsmoker activity patterns in House #2 with
the largest increase occurring when windows are opened by both occupants for
“avoider” nonsmoker behavior. For this case, cross-flow conditions occur most
often when the “avoider” spends 100% of their time in a room separated from the
smoker.
While whole-house air-exchange indicates the overall removal of pollutants,
243
Table 7.5: Simulated 24-h Mean Whole-House Air-Exchange Rate by Flow Symmetry, Flow Scenario, and House Type
a Flow
House Flow
Sym
Type
Sym/
1
Asym
Scenario
2
0.50
SmkDrs−Closed
0.50
0.84−0.85
HAC−10%
0.50
HAC−100%
0.50
Smk−Nsmk−Wins−Open
0.85
Base
0.50
SmkDrs−Closed
0.50
SmkDrs−Closed/SmkWins−Open
0.73−0.85
HAC−10%
0.51
HAC−100%
0.58
a Air-exchange rates
Rate [h−1 ]
Base
Sym
b Air-Exchange
0.84−0.95
for House #1 are the same for asymmetric and symmetric cases. For House #2,
air-exchange rates for asymmetric flow conditions are approximately the same or slightly larger as
those for symmetric conditions, and are therefore not shown.
b Air-exchange rates were either the same across all types of nonsmoker activity or they had the
listed range across “follower”, “napper”, or “avoider” activity patterns.
244
room-specific outdoor air exfiltration and inter-room flows, which are modulated
by open windows, doors, and HAC operation, influence the pollutant concentrations in each room. The procedure for assigning residential flow rates is presented
in Section 6.3.2. Figures 7.4, 7.5, and 7.6 contain 24-h mean interzonal flow rates
corresponding to each flow scenario for symmetric House #1 flows, symmetric
House #2 flows, and asymmetric House #2 flows, respectively. All the flows presented in these figures correspond to “avoider” nonsmoker behavior, although different nonsmoker activity patterns do not change the flow patterns significantly.
Flow in and out of each room, the HAC, and the house as a whole are always
balanced within a tolerance of 1 m3 h−1 . When flows are initially asymmetric or
the HAC is active, flows have a prevailing direction through the house from inlet
rooms (KIT-DIN, HALL) to outlet rooms (LIV, BED, AUX, BATH) or from supply
to return registers, respectively. For symmetric flows or when the HAC is inactive,
flows are always balanced across room and building shell boundaries.
The average air flow out of rooms with a smoker or into rooms with a nonsmoker during smoking periods are two derived quanitites of particular interest.
Since House #1 is dominated by a single, well-mixed room, these parameters are
not as meaningful as for House #2, so I present results here for House #2 only
(Tables 7.6 and 7.7).
The mean air flow rate out of smoking rooms into other rooms, during smoking episodes under symmetrical flow, no HAC activity, and “follower” nonsmoker
behavior, are equal to the base state value of 100 m3 h−1 . As discussed above,
for scenarios involving the closing of the smoker’s door, the door remains open
for time that the smoker and nonsmoker spend in the same room. Therefore, the
closed-door smoking room outflows for “napper” and “avoider” are progressively
smaller as a function of the amount of time spent with the smoker. When the HAC
is active, outflows increase substantially in response to the HAC supply rate of
about 500 m3 h−1 into the KIT-DIN room, which is the location where much of the
smoking activity occurs. The KIT-DIN room is also an outlet room, from which
flows move towards the LIV, BED, AUX, and BATH rooms. Therefore, asymmetric
245
outflows are generally higher than the corresponding symmetric outflows. Asymmetric open-window flows are largest for “avoider” behavior, because two windows are opened in separate rooms during all smoking episodes.
The mean air flow rate into nonsmoker rooms from other rooms during smoking episodes does not include time spent in smoking rooms during smoking periods, so no flows are given for “follower” nonsmoker behavior, where the nonsmoker is continually in the same room as the smoker. The “avoider” and “napper” inflows do not change from one scenario to another under symmetric flow
conditions. Their magnitudes relative to the base flow of 100 m3 h−1 are reflective
of time spent in the bathroom with a closed door or napping in an auxiliary room
with a closed door. Under asymmetric flow conditions, the nonsmoker room inflow is typically larger than for symmetric conditions, because air travels from the
KIT-DIN room to the LIV and AUX rooms where “avoider” and “napper” nonsmokers spend time during smoking episodes.
To elucidate how airborne pollutant exposure concentrations are simulated, I
provide detailed room concentration profiles for seven different simulation scenarios. The first set of profiles are for particle concentrations in House #1, which is
dominated by one large compartment (Figure 7.7). The remaining six sets, which
are for airborne particle and nicotine concentrations in the six-zone House #2, illustrate the effect of closing doors and nonsmoker avoidance behavior (Figure 7.8),
the effect of cross-flow between two windows (Figure 7.9), the effect of intermittent and continuous HAC operation (Figure 7.10), and the effect of initial surface
nicotine concentrations (Figure 7.11).
For base conditions in House #1 (Figure 7.7), peak airborne particle concentrations reach a maximum of approximately 95 µ g m−3 in the main room. The 24-h
average concentrations are in the range of 41−43 µ g m−3 and the 24-h average personal exposure concentration for both occupants is 33 µ g m−3 . The limited capacity for movement of the occupants and the uniformity of pollutant concentrations
in each zone removes any dependence of exposure on occupant location.
For base conditions in House #2 (left side of Figure 7.8), when interior doors
246
KIT−DIN−LIV−BED−AUX−1
1
2
3
4
Smkr Dr Closed, Win Open
1
2
3
4
96
237
0
129
4
0
205
36
0
0
0
0
1
2
3
4
HAC 100%
2
3
96
140
4
6
22
102
4
1021
0
0
HAC 10%
2
3
96
140
4
4
3
14
4
144
0
0
BATH−2
1
Outdoors−3
1
HAC−4
Base State
2
3
96
140
4
4
0
0
1
2
3
4
96
140
0
1
2
3
4
Smkr Door Closed
1
2
3
96
140
96
4
140
4
0
0
0
4
0
0
0
4
0
0
0
121
240
897
1
1
2
3
4
99
154
126
1
2
3
4
Smk/Nsmkr Wins Open
1
2
3
96
237
129
7
205
39
0
0
0
4
0
0
0
Figure 7.4: House #1 Symmetric Flows. The simulated 24-h average air flow rates
[m3 h−1 ] between zones of House #1 for six possible scenarios and initially symmetric boundary flow patterns. These flows correspond to cases with “avoider”
nonsmoker behavior. The direction of flow is from zones listed in rows to zones
listed in columns.
247
AUX−3
BED−4
LIV−2
HALL−5
KIT−DIN−1
BATH−6
1
2
3
4
5
6
7
8
Smk Dr Closed and Win Open
1
2
3
4
5
6
7
0
0
0 77 0 84
0
0
0 84 0 50
0
0
0 71 0 25
0
0
0
67 96 30
77 84 71 67
0 15
0
0
0 96 0
4
84 50 25 30 15 4
0
0
0
0
0
0
0
8
0
0
0
0
0
0
0
1
2
3
4
5
6
7
8
HAC 100%
3
4
5
0
0 497
0
0
0 299
0
0
0 270
0
0
0
274
100 100 71 70
0
0
0 124 0
90 45 45 45 38
358 179 179 179 0
8
0
0
0
0
1021
0
0
1
2
3
4
5
6
7
8
HAC 10%
4
5
0 156
0
0 128
0
0
0 99
0
0
0
99
100 100 71 70
0
0
0 100 0
56 28 28 28 18
50 25 25 25 0
1
2
3
4
5
6
7
8
Smk/Nonsmk Wins Open
2
3
4
5
6
7
0
0
0 100 0 109
0
0
0 100 0 83
0
0
0 71 0 29
0
0
0
70 96 30
100 100 71 70
0 15
0
0
0 96 0
7
109 83 29 30 15 7
0
0
0
0
0
0
0
1
Outdoors−7
Base State
3
4
5
0
0 100
0
0
0 100
0
0
0 71
0
0
0
70
100 100 71 70
0
0
0 96 0
50 25 25 25 15
0
0
0
0
0
1
1
2
3
4
5
6
7
8
1
1
2
3
4
5
6
7
8
HAC−8
0
0
0
77
0
50
0
2
0
6
0
0
0
96
0
4
0
Smkr Door Closed
3
4
5
6
0
0 77 0
0
0 84 0
0
0 71 0
0
0
67 96
84 71 67
0
0
0 96 0
25 25 25 15 4
0
0
0
0
0
2
0
7
50
25
25
25
15
4
0
7
50
25
25
25
15
4
0
8
0
0
0
0
0
0
0
8
0
0
0
0
0
0
0
1
1
2
0
2
0
3
0
0
6
0
0
0
96
0
6
25
6
0
0
0
96
0
4
4
7
50
25
25
48
15
4
102
7
50
25
25
28
15
4
14
8
0
0
0
0
144
0
0
8
0
0
0
0
0
0
0
Figure 7.5: House #2 Symmetric Flows. The simulated 24-h average air flow rates
[m3 h−1 ] between zones of House #2 for six possible scenarios and initially symmetric boundary flow patterns. These flows correspond to cases with “avoider”
listed in columns.
248
1
2
3
4
5
6
7
8
Smkr Dr Closed Win Open
2
3
4
5
6
7
0
0
0 178 0 10
0
0
0 84 0 67
0
0
0 71 0 42
0
0
0
67 132 49
77 142 107 137
0 13
0
0
0 96 0
42
111 8
6 16 77 5
0
0
0
0
0
0
0
8
0
0
0
0
0
0
0
1
2
3
4
5
6
7
8
HAC 100%
3
4
5
0
0 569
0
0
0 273
0
0
0 245
0
0
0
228
100 110 81 90
0
0
0 98 0
111 20 21 21 89
358 179 179 179 0
8
0
0
0
0
1021
0
0
1
2
3
4
5
6
7
8
HAC 10%
4
5
0 228
0
0 124
0
0
0 96
0
0
0
93
100 132 103 129
0
0
0 96 0
77 3
4
8 74
50 25 25 25 0
1
2
3
4
5
6
7
8
Smk/Nonsmk Wins Open
2
3
4
5
6
7
0
0
0 292 0
0
0
0
0 100 0 152
0
0
0 72 0 40
0
0
0
70 135 41
100 250 108 142
0
8
0
0
0 96 0
41
192 2
3
9 73 2
0
0
0
0
0
0
0
1
AUX−3
BED−4
LIV−2
HALL−5
KIT−DIN−1
BATH−6
1
Outdoors−7
Base State
3
4
5
0
0 172
0
0
0 100
0
0
0 71
0
0
0
70
100 136 106 136
0
0
0 96 0
72 0
1
6 72
0
0
0
0
0
1
1
2
3
4
5
6
7
8
1
1
2
3
4
5
6
7
8
HAC−8
0
0
0
77
0
72
0
2
0
6
0
0
0
132
0
0
0
Smk Door Closed
3
4
5
6
0
0 147 0
0
0 84 0
0
0 71 0
0
0
67 132
119 106 132
0
0
0 96 0
1
1
7 72 0
0
0
0
0
0
2
0
7
0
36
36
36
7
36
0
7
2
36
36
36
7
36
0
8
0
0
0
0
0
0
0
8
0
0
0
0
0
0
0
1
1
2
0
2
0
3
0
0
6
0
0
0
106
0
3
25
6
0
0
0
128
0
0
4
7
0
36
36
53
2
36
102
7
0
36
36
38
7
36
14
8
0
0
0
0
144
0
0
8
0
0
0
0
0
0
0
Figure 7.6: House #2 Asymmetric Flows. The simulated 24-h average air flow rates
[m3 h−1 ] between zones of House #2 for six possible scenarios and initially asymmetric boundary flow patterns. These flows correspond to cases with “avoider”
listed in columns.
249
Table 7.6: Simulated 24-h Mean Flow Out of Smoking Rooms Into Other Rooms
During Smoking Episodes [m3 h−1 ] by Flow Symmetry, Flow Scenario, and Nonsmoker Activity for House #2
Flow
Flow
Sym
Scenario
Sym
Asym
Nonsmoker Activity
Follower
Napper
Avoider
Base
100
99
99
SmkDrs−Closed
100
42
4
100
42
4
HAC−10%
153
152
155
HAC−100%
517
516
516
100
99
99
Base
140
138
138
SmkDrs−Closed
140
78
40
163
137
98
HAC−10%
190
189
192
HAC−100%
539
538
537
163
252
298
250
Table 7.7: Simulated 24-h Mean Flow Into Nonsmoker Rooms from Other Rooms
During Smoking Episodes [m3 h−1 ] by Flow Symmetry, Flow Scenario, and Nonsmoker Activity for House #2
Flow
a Flow
Sym
Scenario
Sym
Asym
b Nonsmoker
Activity
Napper
Avoider
Base
44
95
SmkDrs−Closed
44
95
44
95
HAC−10%
44
95
HAC−100%
44
95
44
95
Base
78
117
SmkDrs−Closed
78
117
96
128
HAC−10%
73
114
HAC−100%
44
95
220
285
simulated HAC flow rate into main rooms of the house was approximately 250 m3 h−1 for
every room but KIT-DIN, for which it was 500 m3 h−1 .
b The “follower” nonsmoker is always present with the smoker so there are no corresponding flows
into nonsmoker rooms.
a The
251
40 80
BATH−2 − 41 µg m−3
Receptor − 33 µg m−3
0
40 80
0
0
KIT−DIN−LIV−BED−AUX−1 − 43 µg m−3
Source − 33 µg m−3
0
40 80
SHS Particle Concentration [µg m−3]
40 80
Room and Exposure Concentrations
200
400
600
800
1000
1200
1400
Elapsed Minutes
Figure 7.7: Simulated airborne SHS particle concentration and exposure time series
(µ g m−3 versus min) for rooms and occupants, respectively, in House #1 under
base conditions with “avoider” nonsmoker behavior. The nonsmoker (receptor)
and smoker (source) have near-identical movement patterns and exposure profiles,
reflecting their restriction to a single main room and bathroom and the uniformity
of pollutant concentration both within and between these rooms. The 24-h average
room and exposure concentrations are given in appropriate panels.
252
are open, except for occupied bathroom doors and bedroom doors during sleeping periods, and exterior windows (and doors) are closed, peak airborne particle
concentrations reach maximum levels in excess of 250 µ g m−3 during a 6-cigarette
(3-h) smoking episode in the living room. Single-cigarette episodes in the bedroom during morning and evening hours generate peaks of about 150 µ g m−3 . In
relatively large rooms where no smoking takes place, i.e, the hallway and auxiliary bedroom, levels can still reach 50−100 µ g m−3 during smoking episodes. The
24-h exposure concentration of the “follower” nonsmoker is 61 µ g m−3 , which is
greater than the 24-h average concentration in any of the rooms.
When doors are closed by the smoker in House #2 during smoking episodes,
concentrations are generally higher, relative to the base case, behind closed doors
in smoking rooms, and generally lower in others rooms (right side of Figure 7.8).
For this scenario, the 24-h mean concentration in the living room is 23 µ g m−3
higher than for the base case, and the smoking episode in the living room results in
peak particle concentration of approximately 500 µ g m−3 , which is twice the peaks
observed in the base case. The “follower” exposure increases to 81 µ g m−3 , which
is 20 µ g m−3 higher than for the base scenario. Avoidance behavior reduces the
“avoider” nonsmoker’s exposure dramatically with respect to the “follower” from
a 24-h average SHS particle level of 61 to 21 µ g m−3 . The “avoider” nonsmoker
still encounters fairly high peaks in their exposure concentration profile (100−300
µ g m−3 ) when they visit rooms soon after a smoking episode has occurred.
Opening one or more windows in House #2 increases its ventilation rate. Increased ventilation might have a local effect on a single room if air moves in and
out of a particular room’s window at equal rates, i.e., it is symmetric. However,
when the same amount of air moves in one window and out through a separate
room, perhaps through another open window, other rooms in the house will be
affected. The effect of window cross-flow versus localized window ventilation in
both the smoker’s and nonsmoker’s rooms is to remove pollutants more quickly
from rooms where smoking is occurring (Figure 7.9). Cross-flow involves the
induction of a prevailing current through the house, whereas for the symmetric
200
400
BATH−6 − 34 µg m−3
HALL−5 − 40 µg m−3
BED−4 − 35 µg m−3
AUX−3 − 31 µg m−3
LIV−2 − 55 µg m−3
KIT−DIN−1 − 49 µg m
−3
600
800
1000
Elapsed Minutes
1200
1400
200
400
BATH−6 − 25 µg m−3
HALL−5 − 26 µg m−3
BED−4 − 27 µg m−3
AUX−3 − 19 µg m−3
LIV−2 − 78 µg m−3
KIT−DIN−1 − 53 µg m−3
800
1000
Elapsed Minutes
600
1200
1400
Figure 7.8: The Effect of Closed Doors and Avoidance Behavior. Simulated airborne SHS particle concentration and
exposure time series (µ g m−3 versus min) for rooms and occupants, respectively, in House #2 under two symmetric flow scenarios and two nonsmoker (receptor) time-activity patterns. The left panel corresponds to base flow
conditions where interior doors are open (except when occupants are in the bathroom) and all windows are closed,
and to “follower” nonsmoker behavior. The right panel corresponds to flow conditions where doors are closed for
smoking rooms when a smoker (source) is present and to “avoider” nonsmoker behavior. The 24-h average room
and exposure concentrations are given in appropriate panels.
0 150
0 150
0 150
0 150
0 150
0 150
0 150
0 150
0 300
0 300
0 300
0 300
0 300
0 300
0 300
0 300
253
254
case flows are equal in either direction (to and from the smoking rooms). Therefore, symmetric flow can result in the slower overall removal of pollutants from
the house. While peak particle concentrations are similar between the two cases,
freshly generated tobacco smoke is carried more quickly to adjacent rooms for the
cross-flow case, increasing exposure for those with “avoider” behavior.
When the HAC system in House #2 is in operation, the rate of pollutant removal increases due to supply duct leakage, and the rates of interzonal mixing
are increased. Intermittent HAC activity during waking hours (left of Figure 7.10)
causes a reduction in 24-h mean concentrations of 2−6 µ g m−3 relative to the base
case, whereas continuous HAC activity (right of Figure 7.10) causes reductions of
14−28 µ g m−3 . Peak living room airborne particle concentrations are about 250
µ g m−3 for both base (no HAC) and 10% intermittent HAC conditions, but peak
concentrations during continuous HAC operation drop to about 140 µ g m−3 .
Whereas all of the previous example concentration profiles are for airborne particles in House #2, those shown in Figure 7.11 are for airborne vapor-phase nicotine. As evident from a comparision of the left sides of Figures 7.11 and 7.8, gaseous
nicotine is drawn into surface materials at a much more rapid rate than particles
are deposited onto room surfaces. When the walls of a home are initially free of
nicotine, there is little reemission of nicotine into air, so that airborne concentrations are fairly low in main rooms (0−5 µ g m−3 ) during times when cigarettes are
not being smoked. However, when 50 mg m−2 of reversibly sorbed nicotine is
initially present on the walls of each room of the home, not only are the peak concentrations in the living room greater than 80 µ g m−3 , but every room in the house
has a background level of nearly 20 µ g m−3 due to the reemission (desorption) of
nicotine vapor. The 24-h personal exposure of the “avoider” nonsmoker increases
from 1 to 11 µ g m−3 due to nicotine-loaded surfaces. In Chapter 10 of this dissertation, I compare simulated nicotine concentrations to various empirically observed
levels.
200
400
BATH−6 − 13 µg m−3
HALL−5 − 16 µg m−3
BED−4 − 14 µg m−3
AUX−3 − 12 µg m−3
LIV−2 − 21 µg m−3
−3
600
800
1000
Elapsed Minutes
1200
1400
80
200
400
BATH−6 − 14 µg m−3
HALL−5 − 13 µg m−3
BED−4 − 14 µg m−3
AUX−3 − 13 µg m−3
LIV−2 − 21 µg m−3
800
1000
Elapsed Minutes
600
1200
1400
Figure 7.9: The Effect of Window Cross-Flow. Simulated airborne SHS particle concentration and exposure time series
(µ g m−3 versus min) for rooms and occupants, respectively, in House #2 under symmetric and asymmetric flow
scenarios when two windows are open during smoking periods, one each by the smoker and “avoider” nonsmoker
occupants. The left panel corresponds to symmetric flow conditions where flow is balanced across each window,
and the right panel corresponds to asymmetric conditions where a cross breeze occurs between the windows. The
cross breeze flows from the kitchen and dining room areas of the house towards the living room, bedroom, and
bathroom areas. The 24-h average room and exposure concentrations are given in appropriate panels.
0
80
0
80
0
80
0
80
0
80
0
80
0
80
0
0 60
0 60
0 60
0 60
0 60
0 60
0 60
0 60
255
200
400
BATH−6 − 31 µg m−3
HALL−5 − 37 µg m−3
BED−4 − 33 µg m−3
AUX−3 − 29 µg m−3
LIV−2 − 49 µg m−3
−3
600
800
1000
Elapsed Minutes
1200
1400
200
400
BATH−6 − 20 µg m−3
HALL−5 − 22 µg m−3
BED−4 − 21 µg m−3
AUX−3 − 19 µg m−3
LIV−2 − 27 µg m−3
800
1000
Elapsed Minutes
600
1200
1400
Figure 7.10: The Effect of HAC Duty Cycle. Simulated airborne SHS particle concentration and exposure time series
(µ g m−3 versus min) for rooms and occupants, respectively, in House #2 where there are different HAC duty cycles,
but otherwise under base conditions and “avoider” nonsmoker behavior. The left panel corresponds to intermittent
HAC activity, lasting 10 min at a time, and equal to 10% of the total occupant awake times (when either nonsmoker
or smoker are awake), and the right panel corresponds to 100% HAC activity during awake times. The 24-h average
room and exposure concentrations are given in appropriate panels.
0 150
0 150
0 150
0 150
0 150
0 150
0 150
0 150
80
0
80
0
80
0
80
0
80
0
80
0
80
0
80
0
256
200
400
BATH−6 − 1 µg m−3
HALL−5 − 3 µg m−3
BED−4 − 2 µg m−3
AUX−3 − 1 µg m−3
LIV−2 − 8 µg m−3
−3
600
800
1000
Elapsed Minutes
1200
1400
40
200
400
BATH−6 − 14 µg m−3
HALL−5 − 16 µg m−3
BED−4 − 15 µg m−3
AUX−3 − 14 µg m−3
LIV−2 − 20 µg m−3
800
1000
Elapsed Minutes
600
1200
1400
Figure 7.11: The Effect of Initial Nicotine Surface Concentrations. Simulated SHS nicotine air concentration and exposure time series (µ g m−3 versus min) for rooms and house occupants, respectively, in House #2 where there
are different initial nicotine surface concentrations, but otherwise under base conditions with symmetrical flow
patterns. The nonsmoker occupant follows “avoider” behavior. The left panel corresponds to zero initial nicotine
surface concentrations (i.e., fresh smoking) and the right panel corresponds to initial surface concentrations of 50
mg m−2 of reversibly sorbed nicotine in each room. The 24-h average room and exposure concentrations are given
in appropriate panels.
0
40
0
40
0
40
0
40
0
40
0
40
0
40
0
0 40
0 40
0 40
0 40
0 40
0 40
0 40
0 40
257
258
7.3 Simulated Exposures by House Type, Flow Scenario, and Nonsmoker Activity
The symmetry of flows across internal and external building boundaries and nonsmoker activity patterns for House #1 have little effect on 24-h average nonsmoker
particle exposure concentrations. The base exposure concentrations for House #1
are 33 µ g m−3 across all nonsmoker behavior patterns (Table 7.8) for both symmetric and asymmetric flows. Opening or closing the House #1 bathroom door has
little or no effect on exposure, so door-related cases are omitted. Nonsmoker activity and flow symmetry also has little effect on exposure, because the occupants
spend nearly all of their time in a single main room. Turning on the HAC system
in House #1 intermittently for 10% of daily waking times reduces 24-h mean exposure concentrations from the base condition by only 2−3 µ g m−3 , while continuous
HAC operation during waking periods reduces mean exposure concentations by
12−13 µ g m−3 . Opening two windows decreases exposures by 13−16 µ g m−3 .
In general, the nonsmoker “follower” 24-h average SHS particle exposures are
substantially higher in House #2 than the corresponding nonsmoker exposures in
House #1 with a maximum value of 61 µ g m−3 for the base condition under symmetric flow conditions. In contrast to House #1, nonsmoker activity has a dramatic
effect on exposure in House #2. The House #2 “avoider” exposures are generally
much lower than House #1 exposures with the lowest exposures of 5 and 8 µ g m−3
occurring when the smoker closed doors and opened windows or both occupants
opened windows, respectively. The “avoider” exposures in House #2 are approximately 25−50% of “follower” exposures across all scenarios, with “avoider” exposures occupying an approximate mid-point between the other two. The difference
between exposures for different nonsmoker activities is smaller when the HAC is
active due to enhanced mixing of pollutants among rooms. However, the effects
of HAC operation on “avoider” exposure is not as large when the operation duty
cycle is 10%. As with House #1, flow symmetry in House #2 has a small effect on
average exposures.
Table 7.9 contains values of the correction factor f for the simplified single-zone
259
exposure model given in Equation 7.1. These values are obtained by dividing 24-h
average particle exposure concentrations predicted by the full dynamic model by
the simplified model prediction when using a value of f =1. The simplified model
with f =1 predicts a 24-h average exposure concentration of 46 µ g m−3 for all scenarios treated in this chapter. The calculated values of f provide a measure of the
error in the simplified model relative to the presumably more accurate predictions
of the multizone model. They can be used in Equation 7.1 to give 24-h average concentrations that match those of the more complicated multizone model. Values of
f range from 0.1 to 1.3, reflecting errors in the range of -90 to +30%. For House #2
the unadjusted single-box model tends to underestimate “follower” nonsmoker
exposure concentrations for base conditions, overestimate them when the HAC is
on continuously or windows are open, and strongly overestimate “avoider” exposure. The simplified model best matches the multizone model for “napper” behavior under base conditions or when the HAC is active intermittently for either
“follower” or “napper” behavior.
Since inhalation intake rate and cigarette emissions are held constant for each
simulation trial, the relative changes in intake fraction and equivalent ETS cigarette
particle intake across input scenario levels map exactly to the changes in 24-h average particle exposure (Tables 7.10 and 7.11). The individual particle intake fraction,
calculated as the ratio of 24-h nonsmoker inhaled intake of particle mass to the total mass of particles emitted by cigarettes in the home over the same 24-h period,
ranged from a low of 230 ppm, corresponding to “avoider” nonsmoker activity
in House #2 when both occupants opened windows during smoking episodes, to
over 2,600 ppm for “follower” behavior in House #2 under base conditions. Intake
fraction provides a measure of particle intake relative to the total mass of emitted
particles. I find here that SHS particle intake can be as much as about 0.3% of total
SHS particle emissions. This value is fairly small compared to the intake of the
smoker themselves; however, this individual intake fraction is much larger than
estimates of population intake fraction for pollutant releases to outdoor air from
motor vehicles [Marshall et al., 2003] or power plants [Levy et al., 2003], which are
260
Table 7.8: 24-h Simulated Mean Personal SHS Particle Exposure Concentration
[µ g m−3 ] by Flow Symmetry, House Type, Flow Scenario, and Nonsmoker Activity
Flow
House
Sym
Type
Sym /
1
Asym
Sym
Asym
a Exposure
2
2
Flow
Scenario
a Nonsmoker
Follower
Base
Activity
Napper
Avoider
. . . 33 . . .
HAC−10%
. . . 30−31 . . .
HAC−100%
. . . 20−21 . . .
. . . 17−20 . . .
Base
61
41
24
SmkDrs−Closed
61
37
21
24
13
8
HAC−10%
53
36
23
HAC−100%
30
23
16
24
13
5
Base
59
45
25
SmkDrs−Closed
59
43
23
26
19
11
HAC−10%
52
40
24
HAC−100%
31
25
17
26
19
9
concentrations are shown for each nonsmoker activity and flow symmetry in House #2.
Ranges are given for House #1 across all types of nonsmoker activity and flow symmetry.
261
Table 7.9: Correction Factorsa f for Simulated 24-h Mean Personal SHS Particle
Exposure Concentration Predicted by a Simple Single-Zone Model Across Flow
Symmetry, House Type, Flow Scenario, and Nonsmoker Activity
Flow
House Flow
Sym
Type
Sym/
1
Asym
Scenario
b Nonsmoker
Follower
Asym
a
2
2
Napper
Avoider
Base
. . . 0.7 . . .
HAC−10%
. . . 0.7 . . .
HAC−100%
. . . 0.4−0.5 . . .
Sym
Activity
. . . 0.4 . . .
Base
1.3
0.9
0.5
SmkDrs−Closed
1.3
0.8
0.5
0.5
0.3
0.2
HAC−10%
1.2
0.8
0.5
HAC−100%
0.6
0.5
0.3
0.5
0.3
0.1
Base
1.3
1.0
0.5
SmkDrs−Closed
1.3
0.9
0.5
0.6
0.4
0.2
HAC−10%
1.1
0.9
0.5
HAC−100%
0.7
0.5
0.4
0.6
0.4
0.2
The simple time-averaged single-zone exposure model discussed in Section 7.1 incorporates a correction factor f , which
accounts for time spent out of the house and unequal pollutant levels in different rooms of the house. Ranges of this factor
are calculated here for airborne particle exposure across each combination of flow symmetry, house type, flow scenario,
and nonsmoker activity pattern, by dividing the multi-zone results by the unadjusted single-zone result for base conditions.
The 24-h average exposure concentration predicted by the single-zone model using the model inputs listed in Table 7.1 and
Equation 7.1, using a value for f of 1, is 46 µ g m−3 .
b Corrections factors are shown for each nonsmoker activity and flow symmetry for House #2. Single values or ranges are
given for House #1 across all types of nonsmoker activity and flow symmetry.
262
under approximately 50 ppm and 1 ppm, respectively.
The equivalent ETS cigarette particle intake provides a measure of nonsmoker
particle intake relative to the emissions from a single cigarette. This quantity
ranged from 0.004 cig d−1 to about 0.05 cig d−1 for the scenarios mentioned. This
measure gives an indication of how many cigarettes worth of SHS yield a nonsmoker might inhale over 24 h. Comparisons to mainstream cigarette yields is
misleading, because the SHS yields of different species are not in the same proportion as for mainstream yields. However, keeping this in mind, and noting that
the ratio of SHS to mainstream smoke (MS) yields for particles can be close to 1
[Jenkins et al., 2000], the maximum intake of 5% of a single cigarette over a 24 h
period puts the nonsmoker’s exposure into perspective. Over months or years
of exposure, a person living with a smoker might inhale the equivalent of 100’s
of cigarettes worth of SHS particles. Nonsmoker inhalation intake of other SHS
species can be even higher if the ratio of SHS to MS yields exceeds 1.
The outstanding feature of 24-h mean nonsmoker nicotine exposure occurring
in House #2 is the sharp decrease in exposure across nonsmoker activity from “follower” to “napper” to “avoider” in the case of initially nicotine-free surfaces (Table 7.12). For all flow scenarios and symmetric or asymmetric flow, the “avoider”
nonsmoker receives 0.4−2 µ g m−3 while the “follower” receives 7−12 µ g m−3 .
However, when the walls are loaded with 50 mg m−2 of reversibly sorbed nicotine
the “avoider” exposure under the same conditions rises to 8−11 µ g m−3 and the
“follower” to 15−21 µ g m−3 , a higher and narrower range of exposure outcomes.
In this chapter, I conduct initial simulation experiments exploring the effects of
nonsmoker activity, flow symmetry, central air handling, and door and window
positions on residential exposure to the nicotine and airborne particles present in
SHS. I use scripted location patterns for house occupants in which the nonsmoker
follows the smoker’s movements exactly (“follower”), avoids the smoker completely (“avoider”), or spends a portion of the day with the smoker (“napper”). To
263
Table 7.10: Simulated 24-h Individual Particle Intake Fraction [ppm] by Flow Symmetry, House Type, Flow Scenario, and Nonsmoker Activity
Flow
House Flow
Sym
Type
Sym/
1
Asym
Sym
Asym
a Corrections
2
2
Scenario
a Nonsmoker
Follower
Activity
Napper
Avoider
Base
. . . 1400 . . .
HAC−10%
. . . 1300 . . .
HAC−100%
. . . 850−870 . . .
. . . 720−840 . . .
Base
2600
1700
1000
SmkDrs−Closed
2600
1600
880
980
550
320
HAC−10%
2200
1500
970
HAC−100%
1200
970
660
980
550
230
Base
2500
1900
1000
SmkDrs−Closed
2500
1800
950
1100
810
480
HAC−10%
2200
1700
1000
HAC−100%
1300
1000
700
1100
790
380
factors are shown for each nonsmoker activity and flow symmetry for House #2. Single values or ranges are given for House #1 across all types of nonsmoker activity and flow symmetry.
264
Table 7.11: Simulated 24-h Equivalent ETS Cigarette Particle Intake [cig d−1 ] by
Flow Symmetry, House Type, Flow Scenario, and Nonsmoker Activity
Flow
House Flow
Sym
Type
Sym/
1
Asym
Scenario
Asym
a Equivalent
2
2
Follower
Activity
Napper
Base
. . . 0.026 . . .
HAC−10%
. . . 0.024 . . .
HAC−100%
. . . 0.016 . . .
Sym
a Nonsmoker
Avoider
. . . 0.014−0.016 . . .
Base
0.048
0.032
0.019
SmkDrs−Closed
0.048
0.029
0.016
0.018
0.010
0.006
HAC−10%
0.042
0.028
0.018
HAC−100%
0.023
0.018
0.012
0.018
0.010
0.004
Base
0.046
0.035
0.020
SmkDrs−Closed
0.046
0.034
0.018
0.021
0.015
0.009
HAC−10%
0.041
0.031
0.019
HAC−100%
0.024
0.020
0.013
0.021
0.015
0.007
cigarette intakes are shown for each nonsmoker activity and flow symmetry for
House #2. Single values or ranges are given for House #1 across all types of nonsmoker activity
and flow symmetry.
265
Table 7.12: Simulated 24-h Mean SHS Nicotine Personal Exposure Concentration
[µ g m−3 ] by Flow Symmetry, Initial Surface Concentrations [mg m−2 ], Flow Scenario, and Nonsmoker Activity for House #2
Flow
Sym
Sym
Asym
Initial Flow
Surf. Scenario
0
Base
SmkDrs−Closed
HAC−10%
HAC−100%
Nonsmoker Activity
Follower
Napper
Avoider
12
12
7
11
7
7
7
6
3
6
5
4
1
1
0.4
2
2
1
50
Base
SmkDrs−Closed
HAC−10%
HAC−100%
21
21
15
20
16
15
16
16
12
15
13
11
11
10
10
11
10
8
0
Base
SmkDrs−Closed
HAC−10%
HAC−100%
12
12
8
11
8
8
7
7
4
6
5
5
2
1
1
2
2
2
50
Base
SmkDrs−Closed
HAC−10%
HAC−100%
21
21
16
20
16
16
17
17
13
16
14
13
11
11
10
11
11
9
266
examine the effects of a house’s multi-compartment character on exposure, exposures are simulated for two houses of identical volume, one dominated by a single,
large zone (House #1), and one with six distinct zones (House #2). Base conditions
in the house occurred for symmetric flows when door were open during waking
and non-bathroom times, windows were always closed, and the HAC system was
inactive.
As expected, nonsmoker exposures occurring among the multiple rooms of
House #2 exhibited much wider variation than exposures occurring in House #1.
Exposure concentrations in House #1 tended to be lower than those in House #2
for the “follower” and “napper” nonsmoker who spent some or all of their time
with the smoker. This result arises because pollutants are diluted into a single large
space. A simplified, single-zone exposure model with an adjustment factor equal
to 1 under- or overestimated the House #2 exposures predicted by the full multizone simulation model by 10−90%. The unadjusted model performed best when
the receptor only spent a portion of their time with the active smoker and under
base conditions.
Generally, each of the examined factors had a discernible impact on exposure,
although ignoring pollutant-specific effects, the effect of nonsmoker activity was
greatest. For nicotine, the effect of preloaded versus clear surfaces resulted in comparable variations in exposure as did “avoider” versus “follower” nonsmoker activity. The base 24-h particle exposure concentration in House #2 decreased from
61 to 24 µ g m−3 when the nonsmoker displayed “follower” versus “avoider” behavior. The corresponding nicotine exposure concentrations decreased from 21 to
11 µ g m−3 with surfaces loaded or from 12 to 1 µ g m−3 with fresh surfaces. The
opening of windows during smoking episodes resulted in the largest decrease in
exposures given the same nonsmoker activity. Doors generally had the smallest
effect. The effect of asymmetric flow was also relatively small, although exposures
could increase because of time spent “downwind” from the smoker, especially for
the “napper” and “avoider” nonsmokers. HAC operation reduces exposures because of increased outdoor air infiltration owing to supply duct leakage.
267
7.5 References
Jenkins, R. A., Guerin, M. R., and Tomkins, B. A. (2000). The Chemistry of Environmental Tobacco Smoke: Composition and Measurement. Lewis Publishers, Boca
Raton, second edition.
Levy, J. I., Wilson, A. M., Evans, J. S., and Spengler, J. D. (2003). Estimation of
primary and secondary particulate matter intake fractions for power plants in
Georgia. Environmental Science and Technology, 37(24): 5528–5536.
Marshall, J. D., Riley, W. J., McKone, T. E., and Nazaroff, W. W. (2003). Intake
fraction of primary pollutants: Motor vehicle emissions in the South Coast air
basin. Atmospheric Environment, 37: 3455–3468.
268
Chapter 8
Tier II. Frequency Distributions of
Unrestricted Exposure Based on
Realistic Variation in Occupant
Location Patterns
In this chapter, I conduct further residential SHS exposure simulation experiments, incorporating realistic variation in household occupant location patterns.
Here, I am interested in frequency distribution of exposure for “natural”, or unrestricted, situations that might be expected to occur for a typical US house under a
range of typical environmental conditions. By using a simulation cohort of 5,000
smoker/nonsmoker pairs, I am able to generate stable frequency distributions of
exposure for breakdowns of two key variates, the number of cigarettes smoked
indoors and the time spent at home by the nonsmoker. As in the previous chapter, most environmental parameters, including the house size and layout, are held
fixed, so that for a given physical configuration, variation in exposure is driven
by interacting patterns of smoker and nonsmoker location. I also carry out controlled trials on the cohort to explore the broad effects of HAC operation, air flow
symmetry, and surface concentrations of SHS pollutants. In addition to exploring
unrestricted distributions of residential SHS exposure, I calculate the distribution
of adjustment factors, f , for a simple single-zone exposure model, which can be
used to judge the appropriateness of such a model to represent population expo-
CHAPTER 8. TIER II. REALISTIC VARIATION IN OCCUPANT LOCATION
269
sures. The approach in this chapter is in contrast to that in Chapter 7, where I use a
small number of scripted human location patterns, and Chapter 8, where I systematically explore regulated human behavior related to specific exposure mitigation
strategies.
8.1 Seven Simulation Trials of Unrestricted Exposure
While the initial analyses in Chapter 7 were limited to a single inter-room location
pattern for the smoker, a fixed number of cigarettes smoked in the house, a small
number of different nonsmoker location patterns, and fixed times spent at home
by the smoker and nonsmoker, these constraints are relaxed in the current chapter. All of the simulations performed here are done for exposures occurring in a
4-room single-story house with an HAC system as illustrated by the floor plan in
Figure 8.1, which is identical to House #2 used for simulations in Chapter 7. However, the complex interplay of smoking activity in various rooms at various times,
nonsmoker presence in the home, and the proximity of the nonsmoker to the active
smoker is expected to result in broad variability in nonsmoker exposure.
I generate frequency distributions of particle, nicotine, and carbon monoxide
(CO) exposure for seven different scenarios (Table 8.1) by introducing inter-room
location data measured as part of a national human activity pattern survey (see
Chapter 4). For different scenarios, I consider changes in flow that might occur
when the HAC is in operation, or when there is either symmetric or asymmetric
flow across house boundaries. I consider only those behaviors that might naturally
occur in an unrestricted house where the smoker is free to smoke in any location
and at any time of his or her choosing. No regulations are imposed on smoker
behavior and no strategies are consciously applied to mitigate exposure.
I define base exposure distributions as those arising for the simulated cohort of
matched smokers and nonsmokers when windows are kept closed, interior doors
are kept open, except for bathroom or sleeping periods, and flows are symmetric. The SHS species were chosen to represent the particulate, gaseous, and semivolatile components of SHS (see Chapters 2 and 3). Particle and nicotine emis-
Back
Door
HAC
Kitchen-Dining,
100 m3
Front
Door
270
HAC
Auxiliary,
50 m3
Hallway, 30 m3
HAC
Living Room,
50 m3
HAC
Bath,
7m3
HAC
Bedroom,
50 m3
HAC
Figure 8.1: Floorplan of the 287 m3 single-story 4-room house (plus hallway and
bathroom) used in this chapter for simulations of SHS exposure frequency distributions. This house is identical to House #2 defined in Chapter 7. Each room of the
house has a bidirectional connection to the outdoors, e.g., via a window or wall,
and the main rooms are connected by a doorway to a hallway, except for the master
bathroom, which is only connected to the master bedroom. When operating, the
HAC system supplies a total of 1,292 m3 h−1 of air to different rooms in amounts
ranging from 502 m3 h−1 for the kitchen-dining area to 36 m3 h−1 for the bathroom. Air returns to the HAC through a hallway register. The HAC flows reflect
leakage from supply ducts equal to 10% of the designed flow rate of 5 h−1 . The
HAC enhances mixing of pollutants in the house by recirculating air between the
supply and return registers. The process of assigning house air flows, including
HAC flows and flows across indoor-outdoor and interior boundaries, is described
in Chapter 6.
271
sions, deposition, or sorption/desorption parameters are fixed at the same values
used for simulations in Chapter 7. Carbon monoxide emissions are set at 50 mg
cig−1 (Section 3.7), and loss to surfaces is set to zero, because CO is a nonreactive gas whose sole significant removal mechanism is by ventilation. Surfaces may
be initially free or loaded with reversibly sorbed nicotine resulting from chronic
smoking activity.
As described in Chapter 6, the variation in the timing and location of active
cigarettes follows naturally from the variation in selected smoker location time
series. Likewise, the absolute time spent at home and the common time spent
by nonsmoker and smoker in the same room result from the characteristics of the
selected location patterns. House occupants close doors for time they spend in the
bathroom or when they are sleeping. For time when the nonsmoker and smoker
occupy the same room, the door behavior of the nonsmoker takes precedence.
To represent adult smokers, I randomly selected a sample of 5,000 individuals
over the age of 18 from the NHAPS activity pattern study (Chapter 4). Each smoker
consumes 30 cig d−1 , although, depending on their location patterns, not all of
these are smoked in the house. Another 5,000 corresponding individuals were chosen of any age to represent nonsmokers. The two individuals were matched based
only on the day of the week and not on any housing-related variables, since the
reported activity pattern characteristics appear fairly uniform for different sized
homes. If the ages of the sampled smoker and nonsmoker were within 10 years,
and the nonsmoker was also an adult over age 18, then they were assumed to be
spouses that slept in the same bedroom. Otherwise, the nonsmoker was assigned
to sleep in the auxiliary room.
The distributions of three key characteristics for the sampled cohort of matched
occupants are pictured in Figure 8.2. These characteristics are the number of
cigarettes smoked in the house, the fraction of the 24-h day the nonsmoker spends
in the house, and the fraction of the 24-h day that the smoker and nonsmoker
spend in the same room. Most commonly, between 4 and 12 cigarettes are smoked
in the home over the course of the day. Most of the nonsmokers spend between
272
Table 8.1: Descriptions of Seven Unrestricted Residential SHS Inhalation Exposure
Scenariosa
No.
Abbreviation
Scenario Description
1
Particles−Base
Exposure to airborne particles under base
conditions
2
Particles−Asym
Exposure to airborne particles when leakage flow
is asymmetric, rather than being symmetric,
across indoor-outdoor building and room
boundaries
3
Particles−HAC 100% Exposure to airborne particles with HAC
operating continuously during times when at
least one occupant is awake
4
Particles−HAC 10%
Exposure to airborne particles with HAC
operating intermittently for 10% of the day for
10-min at a time during times when at least one
occupant is awake
5
CO−Base
Exposure to carbon monoxide under base
conditions
6
Nicotine Fresh
Exposure to airborne nicotine under base
conditions with initially clean room surfaces
7
Nicotine Loadedb
Exposure to airborne nicotine under base
conditions with nicotine-loaded room surfaces
a Except
where noted, base conditions apply, where the HAC system is inactive, all windows are
assumed to be closed, and all interior doors are open during non-bathroom waking hours for each
occupant.
b The initial nicotine surface concentrations in each room for this scenario are determined from a
residential simulation experiment of chronic smoking activity lasting 5,000 days (see Section 8.2.3).
The nicotine surface concentrations were 48, 75, 41, 48, 51, and 46 mg m−2 for the kitchen-dining
room, living room, auxiliary room, bedroom, hallway, and bath, respectively.
273
1000
300
200
300
0
5
10
15
20
25
30
Number of Cigs. [d−1]
0
0
0
100
100
500
200
Frequency
400
400
500
1500
500
600
Histograms for Key Exposure Variables
0.0
0.2
0.4
0.6
0.8
1.0
Time Fraction at Home
0.0
0.2
0.4
0.6
Smk/NSmk Correl.
Figure 8.2: Frequency distributions in the form of histograms for three key variables associated with the simulation cohort of 5,000 randomly selected smokernonsmoker pairs, which are expected to have a large effect on nonsmoker SHS
exposure: the number of cigarettes smoked in the house, the fraction of the day
the nonsmoking receptor person spends at home, and the fraction of the day that
the smoker and nonsmoker spend together in the same room.
50 and 80% of their day at home. However, more than two thirds of the smokers and nonsmoker matched pairs spend less than 10% of their time in the same
room of the house. Hence, the exposure frequency distributions generated in this
chapter for this cohort shed additional light on the variation in exposures among
nonsmokers that spend relatively little time in close proximity to the smoker.
Statistics for exposure metrics were calculated only for simulated households
with a non-zero number of cigarettes smoked indoors and non-zero time spent
at home by the nonsmoker. The original overall sample of 5,000 was reduced by
about 200 due to dropped smoker and nonsmoker pairs not meeting these crite-
274
ria. Zero exposures, which occurred when both the smoker and nonsmoker spent
some time at home, were retained in log-probability plots and in the calculation of
descriptive statistics. For simulated sample sizes in excess of 500, the mean and
median exposure concentrations were found to be stable, having a 90% confidence
band half-length equal to less than 10% of the mean or median. For the analyses
presented below, a sample of at least 500 was used for factor breakdowns by number of cigarettes smoked indoors and the fraction of time spent at home by the
nonsmoker.
8.2 Base Exposure Distributions
8.2.1 Particles
Figure 8.3 and Table 8.2 present the results of simulating the SHS particle exposure
distribution for the sampled cohort. Ten percent of the cohort had 24-h particle
exposure concentrations under 1 µ g m−3 with 10% over 39 µ g m−3 . Ninety-eight
percent of simulated 24-h mean particle nonsmoker exposure concentrations (n =
4,798) span five orders of magnitude from about 0.001 to nearly 100 µ g m−3 with
a median of 12 µ g m−3 . Similarly, 98% of simulated individual intake fraction and
equivalent ETS cigarettes, also shown in Figure 8.3, span 4−5 orders of magnitude,
from 10−6 to 6 x 10−2 and 10−6 to 4 x 10−3 cig d−1 , respectively. The most-highly
exposed upper half of the population have exposures that appear fairly well represented by a lognormal distribution. The overall arithmetic means for 24-h exposure
concentations, individual intake fraction, and equivalent ETS cigarettes are 17 µ g
m−3 , 1,100 ppm, and 1.2% d−1 , respectively.
Extremely low or zero exposure results when either a very small number of
cigarettes are smoked at home or a small fraction of time is spent by the nonsmoker at home, whereas the highest exposures result when more cigarettes are
smoked and more time is spent at home. A nonsmoker who spends under
2
3
of
his or her time at home experiences about half of the 24-h mean particle exposure
concentration encountered by the average nonsmoker who spends more than
2
3
of
275
his or her time at home. For a given amount of time spent at home, smoking more
than 10 cigarettes in the house results in an average tripling of mean SHS particle
exposure relative to smoking fewer than 10 cigarettes. Simulated nonsmokers who
live in a house where more than 10 cigarettes are smoked per day and who spend
more than
2
3
of their time at home have an average exposure concentration of 35
µ g m−3 , and they are clearly the subgroup with the highest potential exposure.
Statistics for the daily equivalent ETS cigarette particle intake follow the same
general pattern as the absolute particle exposure concentrations. The mean value
is about 0.5% d−1 when fewer than 10 cigarettes are smoked and the nonsmoker
spends less than 23 at home, and 2.5% d−1 when more than 10 cigarettes are smoked
and more than
2
3
of the day is spent at home. However, statistics for the individual
particle intake fraction have a different pattern. The average individual intake
fraction is 810−840 ppm when under
at home and 1400 ppm when over
2
3
2
3
of the time was spent by the nonsmoker
of the time was spent at home, regardless
of the number of cigarettes smoked in the house. This result is expected since,
for this metric, the total nonsmoker particle intake is normalized by total cigarette
emissions.
8.2.2 Carbon Monoxide
Although CO does not undergo surface deposition as particles do, the pattern of
CO exposure metrics is similar to that for particles (see Figure 8.4 and Table 8.3). In
homes where more than 10 cigarettes were smoked and more than
2
3
of the day was
spent by the nonsmoker at home the 24-h CO average SHS exposure concentration
is 200 µ g m−3 versus 35 µ g m−3 for particles. Ninety-eight percent of simulated
24-h CO exposure concentrations for the same subset of homes range from approximately 25 to 500 µ g m−3 . The magnitude of intake fraction and equivalent ETS
cigarettes is slightly larger for CO than for particles because particles have deposition as an additional loss mechanism, whereas loss of CO is limited to loss by
ventilation. While the total cigarette mass emissions for CO are greater than for
particles, it might be used as a marker for SHS exposure, with exposures that are
276
Particles
102
101
100
10−1
10−2
10−3
10−4
10−5
99
90
95
75
50
25
5
10
10−6
1
SHS Particle Exp. Conc. [µg m−3], iF, and ETS Cigarettes [d−1]
Cumulative Probability (%)
Exposure Conc.
Intake Fraction
ETS Cigs.
Figure 8.3: Particles. Log-probability plots of 24-h average SHS particle exposure
concentration, individual intake fraction, and equivalent ETS cigarette intake distributions for 4,798 nonsmoking individuals across the full range of number of
cigarettes smoked in the house and fraction of time spent at home. Forty-eight of
the simulated nonsmokers had zero particle exposure, reflected in the lowest cumulative probability shown (1%). See Table 8.2 for descriptive statistics calculated
from these distributions.
10−30 Cigs
Overall
0−10 Cigs
10−30 Cigs
Overall
0−10 Cigs
10−30 Cigs
No.
Cigs
Overall
0−10 Cigs
Fract.
at Home
Overall
< 2/3 at Home
> 2/3 at Home
< 2/3 at Home
> 2/3 at Home
Overall
< 2/3 at Home
> 2/3 at Home
< 2/3 at Home
> 2/3 at Home
Overall
< 2/3 at Home
> 2/3 at Home
< 2/3 at Home
> 2/3 at Home
n
nzero
4798 48
1430 38
984
1
1348
9
1036
0
4798 48
1430 38
984
1
1348
9
1036
0
4798 48
1430 38
984
1
1348
9
1036
0
Sample
Mean
17
6.5
11
18
35
1100
840
1400
810
1400
0.012
0.0050
0.0082
0.013
0.025
Std.
Dev.
17
7.5
9.4
14
21
880
920
980
670
750
0.013
0.0058
0.0071
0.011
0.016
Median
12
3.8
8.7
14
32
860
560
1100
650
1400
0.0083
0.0029
0.0063
0.011
0.023
Percentiles
10th
90th
1.0
39
0.11
17
1.2
25
3.0
36
12
63
140
2200
30
2100
270
2700
150
1700
550
2400
0.00074
0.029
0.00009
0.013
0.00075
0.019
0.0023
0.027
0.0078
0.046
are presented for households where more than zero time was spent at home by the nonsmoker and more than zero cigarettes
were smoked. Abbreviations: 24-h Avg = 24-h average exposure concentration (absolute exposure metric); iF = individual intake fraction
(relative exposure metric); ETS Cigs = equivalent ETS cigarettes (relative exposure metric); No. Cigs = number of cigarettes smoked in
the home by the smoker during the simulated 24-h period; Fract. at Home = fraction of the simulated 24-h period that the nonsmoking
receptor spent at home; n = total simulated sample size; nzero = sample size with zero exposure (included in statistics); Sample Mean =
arithmetic mean of exposure metric; Std. Dev. = standard deviation of exposure metric; 10th and 90th = 10th and 90th percentiles of the
distribution.
a Statistics
ETS Cigs
iF
Exposure
Metric
24-h Avg
Table 8.2: Particles. 24-h Average Nonsmoker SHS Particle Inhalation Exposure Concentration [µ g m−3 ], Individual Intake Fraction (iF) [ppm], and Equivalent ETS Cigarettes (ETS Cigs) [d−1 ] by Number of Cigarettes Smoked
Indoors and Fraction of Time Spent by the Nonsmoker at Homea
277
278
proportional to those for any nonreactive gaseous SHS constituent.
8.2.3 Nicotine
In a house with chronic cigarette smoking, nicotine and other sorbing chemical
species in the smoke will accumulate on household surfaces, such as walls, ceilings, floors, and furniture. This accumulation occurs because the rate at which
nicotine sorbs to surfaces (1.4 m h−1 times the air concentration and a typical
surface-to-volume ratio of 2−3; see Chapter 5) is much larger than the rate at
which it desorbs, i.e., is reemitted (0.00042 h−1 times the surface concentration).
When enough material has accumulated, the reemission rate can be sufficient to
contribute significantly to nicotine air concentrations, creating an elevated baseline exposure for any person occupying the house regardless of the daily timing of
smoking activity, number of cigarettes, or time spent by the receptor at home.
To explore the nicotine surface loading that might occur in a household with
daily smoking, I randomly selected 5,000 24-h individual daily smoker household
movement patterns. A different smoker entered the house on subsequent days so
that nicotine continually accumulated over a 5,000-day simulation period. I used
the same house as for the other simulations described in this chapter. The surfaceto-volume ratios of each room, which had a range of 1.6−3.5 m−1 were assigned
values based on calculations presented in Section 5.2.
Figure 8.5 shows plots of the simulated 24-h mean air nicotine concentrations in
each of the six rooms of the house (4 main rooms plus a hallway and a master bathroom). After 1,000−2,000 d, or about 3-5+ years, the air concentrations displayed
a stable minimum level in each room, between 10 and 20 µ g m−3 . The concentrations caused by smoking on particular days are added on top of the background
level, appearing as random scatter above the constant background. In the living
room, where most smoking takes place, daily mean concentrations could exceed
40 µ g m−3 .
The emerging constant background concentrations are a result of accumulated
surface nicotine concentrations, which are plotted in Figure 8.6. The surface nico-
279
10
102
101
10−2
Intake Fraction
SHS Exp. Conc. [µg m−3]
Carbon Monoxide
3
10−3
10−4
10
10−2
99
95
90
75
50
25
10
5
10−3
1
Equiv. ETS Cigs. [d−1]
−1
Figure 8.4: CO. Log-probability plots of 24-h average SHS carbon monoxide exposure concentration, individual intake fraction, and equivalent ETS cigarette intake
distributions for 1,036 nonsmoking individuals in households where more than 10
cigarettes were smoked and the receptor spent more than 23 of their time at home.
See Table 8.3 for descriptive statistics calculated from the complete distributions.
10−30 Cigs
Overall
0−10 Cigs
10−30 Cigs
Overall
0−10 Cigs
10−30 Cigs
No.
Cigs
Overall
0−10 Cigs
Fract.
at Home
Overall
< 2/3 at Home
> 2/3 at Home
< 2/3 at Home
> 2/3 at Home
Overall
< 2/3 at Home
> 2/3 at Home
< 2/3 at Home
> 2/3 at Home
Overall
< 2/3 at Home
> 2/3 at Home
< 2/3 at Home
> 2/3 at Home
n
nzero
4798 48
1430 38
984
1
1348
9
1036
0
4798 48
1430 38
984
1
1348
9
1036
0
4798 48
1430 38
984
1
1348
9
1036
0
Sample
Mean
97
37
64
100
200
1200
960
1600
940
1700
0.014
0.0057
0.0094
0.015
0.029
Std.
Dev.
97
42
52
80
120
1000
1000
1100
750
830
0.015
0.0064
0.0079
0.012
0.018
Median
68
23
51
85
180
1000
650
1400
760
1600
0.010
0.0035
0.0074
0.013
0.026
Percentiles
10th
90th
6.7
220
0.91
93
7.8
140
20
200
73
360
190
2500
45
2300
350
3000
200
1900
660
2700
0.00097
0.033
0.00013
0.014
0.0010
0.021
0.0030
0.031
0.0096
0.053
were smoked. Abbreviations: 24-h Avg = 24-h average exposure (absolute exposure metric); iF = individual intake fraction (relative
exposure metric); ETS Cigs = equivalent ETS cigarettes (relative exposure metric); No. Cigs = number of cigarettes smoked in the home
by the smoker during the simulated 24-h period; Fract. at Home = fraction of the simulated 24-h period that the nonsmoking receptor
spent at home; n = simulated sample size; nzero = sample size with zero exposure (included in statistics); Sample Mean = arithmetic mean
of exposure metric; Std. Dev. = standard deviation of exposure metric; 10th and 90th = 10th and 90th percentiles of the distribution. Note:
At 25 ◦ C and 1 atm, 1 ppm of CO is equal to 1145 µ g m−3 .
a Statistics
ETS Cigs
iF
Exposure
Metric
24-h Avg
Table 8.3: CO. 24-h Average Nonsmoker SHS Carbon Monoxide Inhalation Exposure Concentration [µ g m−3 ],
Individual Intake Fraction (iF) [ppm], and Equivalent ETS Cigarettes (ETS Cigs) [d−1 ] Statistics by Number of
Cigarettes Smoked Indoors and Fraction of Time Spent by the Nonsmoker at Homea
280
281
tine concentrations reach a plateau after approximately 2,000 days (5+ years) of
habitual smoking. The surface concentration reached almost 80 mg m−2 for the
living room, and 40−50 mg m−2 for the other rooms. These levels represent surface concentrations that might be expected to occur in a home where unabated and
unrestricted smoking occurred for an extended period of several years or more.
However, the exposure model assumes that all nicotine sorption is fully reversible,
which, as suggested by the results of Piadé et al. [1999], may not be the case. Nevertheless, some reversibility in nicotine sorption is likely to occur in homes, so my
nicotine analyses likely bound true nicotine exposures for the given simulated conditions. I compare simulated and empirically observed air nicotine concentrations
in Chapter 10.
To examine the range of effects of nicotine-loaded walls on the distribution of
nicotine inhalation exposures, I conducted simulations using both intially fresh,
uncontaminated walls and initial surface concentrations approximately equal to
the simulated surface nicotine concentration at the end of the 5,000-day simulation period, i.e., 48, 75, 41, 48, 51, and 46 mg m−2 for the kitchen-dining room,
living room, auxiliary room, bedroom, hallway, and bath, respectively. Figure 8.7
contains plots of these distributions for households where more than 10 cigarettes
were smoked indoors over the day and where the nonsmoker spent more than
2
3
of
their time at home, i.e., households with high potential exposure. For this group,
the 24-h mean exposure concentration for loaded surfaces was 15 µ g m−3 , but it
was only 4.5 µ g m−3 for fresh surfaces (see Tables 8.4 and 8.5).
The distributions of exposure metrics for the cohort with high potential exposure and initially loaded surfaces are more nearly lognormal than for fresh surfaces and span a smaller range – within a single order of magnitude – with 24-h
mean exposure concentrations ranging from about 8 to 30 µ g m−3 , whereas the
fresh-surface distributions span closer to three orders of magnitude. The reduction
in the variance of exposures for the loaded-surface cases are caused by the minimum background level of air nicotine in each room, resulting from reemission of
nicotine from surfaces. Ten percent of fresh-surface nicotine exposures are under
282
0
KIT−DIN−1
1
2 3 4
LIV−2
5
AUX−3
Mean SHS Nicotine Air Concentration [µg m−3]
40
30
20
10
0
BED−4
HALL−5
BATH−6
40
30
20
10
0
0
1
2
3
4
5
0
1
2
3
4
5
Day / 1000
Figure 8.5: 24-h mean air nicotine concentrations [µ g m−3 ] simulated based on
random smoker activity in a 287 m3 house with 4 main rooms over 5,000 sequential
days. The nicotine sorption coefficient was set at 1.4 m h−1 and the desorption
coefficient was 0.00042 h−1 . The overall surface-to-volume ratio of the house was
1.9 m−1 and the air-exchange rate was 0.5 h−1 .
283
0
KIT−DIN−1
1
2 3 4
LIV−2
5
AUX−3
Mean SHS Nicotine Surface Concentration [mg m−2]
80
60
40
20
0
BED−4
HALL−5
BATH−6
80
60
40
20
0
0
1
2
3
4
5
0
1
2
3
4
5
Day / 1000
Figure 8.6: Surface SHS nicotine concentrations in mg m−2 simulated based on
random smoker activity in a 287 m3 house with 4 main rooms over 5,000 sequential
days. See the text and the caption to Figure 8.5 for simulation conditions.
284
0.73 µ g m−3 , whereas 90% of the loaded-surface nicotine exposures are above 9.8
µ g m− 3 .
8.3 The Effect of Air Flow Patterns
8.3.1 HAC Operation
When a home’s HAC system is activated, air is recirculated in the house so that
air pollutants in rooms where smoking occurs may be delivered somewhat more
rapidly to other rooms in the house. Leaks in the HAC system also lead to greater
dilution of air pollutants due to increased infiltration of outdoor air.
From the distributions of exposure metrics presented in Figure 8.8 for households with high potential exposure, i.e., where more than 10 cigarettes were
smoked indoors and the receptor spent more than
2
3
of the day at home, it is evi-
dent that the recirculation of pollutants by the intermittent operation of the HAC
system for 10% of waking hours caused the lower 25% of particle exposures to increase slightly. However, the increase in the home’s air leakage rate also resulted in
a slight decrease in particle exposures for the upper 75% of the distribution. Continuous operation of the HAC system during waking hours resulted in the upper
95% of the distribution having significantly lower exposures. The 24-h mean particle exposure concentrations for the group with high potential exposures dropped
from 35 to 32 µ g m−3 with 10% HAC operation and to 19 µ g m−3 with continuous
operation (see Tables 8.2, 8.6, and 8.7).
8.3.2 Asymmetric Leakage Flow
For the base symmetric air flow condition, which is in effect for six of the seven
scenarios treated in this chapter, the leakage flow going into the house from outdoors through any given room is balanced with flow exiting the same room to
the outdoors, and flow across each door boundary within the home is balanced
in either direction. This situation might apply to the case where only turbulent
flow is driving air movement, perhaps corresponding to the winter season when a
285
102
101
100
10−1
10−2
10−2
10−3
10−4
10−5
10−6
10−1
10−2
10−3
10−4
99
95
90
75
50
25
10
5
10−5
1
Intake Fraction
Nicotine
Fresh Surfaces
Loaded Surfaces
Figure 8.7: Nicotine. Log-probability plots of 24-h average SHS nicotine exposure
concentration, individual intake fraction, and equivalent ETS cigarette intake distributions under fresh and loaded wall conditions for 1,036 nonsmoking individuals in households where more than 10 cigarettes were smoked and the receptor
spent more than 32 of their time at home. See Tables 8.4 and 8.5 for descriptive
statistics calculated from the complete distributions, including overall and breakdowns by cigarettes smoked and nonsmoker time spent at home.
10−30 Cigs
Overall
0−10 Cigs
10−30 Cigs
Overall
0−10 Cigs
10−30 Cigs
No.
Cigs
Overall
0−10 Cigs
Fract.
at Home
Overall
< 2/3 at Home
> 2/3 at Home
< 2/3 at Home
> 2/3 at Home
Overall
< 2/3 at Home
> 2/3 at Home
< 2/3 at Home
> 2/3 at Home
Overall
< 2/3 at Home
> 2/3 at Home
< 2/3 at Home
> 2/3 at Home
n
nzero
4798 48
1430 38
984
1
1348
9
1036
0
4798 48
1430 38
984
1
1348
9
1036
0
4798 48
1430 38
984
1
1348
9
1036
0
Sample
Mean
2.1
0.82
1.5
2.1
4.5
270
210
350
200
370
0.0031
0.0012
0.0021
0.0032
0.0066
Std.
Dev.
2.6
1.2
1.6
2.2
3.4
300
300
350
220
270
0.0039
0.0019
0.0024
0.0034
0.0051
Median
1.1
0.22
0.83
1.3
3.8
160
64
210
120
330
0.0017
0.00034
0.0012
0.0020
0.0054
Percentiles
10th
90th
0.020
5.7
0.0016
2.6
0.027
4.0
0.086
5.4
0.73
9.0
5.3
690
0.76
640
11
870
8.1
510
65
740
0.00003
0.0084
0.0
0.0040
0.00004
0.0060
0.00013
0.0082
0.00099
0.014
of exposure metric; Std. Dev. = standard deviation of exposure metric; 10th and 90th = 10th and 90th percentiles of the distribution.
a Statistics
ETS Cigs
iF
Exposure
Metric
24-h Avg
Table 8.4: Nicotine Fresh. 24-h Average Nonsmoker SHS Nicotine Inhalation Exposure Concentration [µ g m−3 ],
Individual Intake Fraction (iF) [ppm], and Equivalent ETS Cigarettes (ETS Cigs) [d−1 ] Statistics for Fresh Surfaces
by Number of Cigarettes Smoked Indoors and Fraction of Time Spent by the Nonsmoker at Homea
286
10−30 Cigs
Overall
0−10 Cigs
10−30 Cigs
Overall
0−10 Cigs
10−30 Cigs
No.
Cigs
Overall
0−10 Cigs
Fract.
at Home
Overall
< 2/3 at Home
> 2/3 at Home
< 2/3 at Home
> 2/3 at Home
Overall
< 2/3 at Home
> 2/3 at Home
< 2/3 at Home
> 2/3 at Home
Overall
< 2/3 at Home
> 2/3 at Home
< 2/3 at Home
> 2/3 at Home
n
nzero
4798
0
1430
0
984
0
1348
0
1036
0
4798
0
1430
0
984
0
1348
0
1036
0
4798
0
1430
0
984
0
1348
0
1036
0
Sample
Mean
10
6.9
12
8.2
15
2600
3400
5100
790
1300
0.015
0.010
0.017
0.012
0.022
Std.
Dev.
4.5
2.2
2.6
3.2
4.4
12000
18000
13000
390
520
0.0068
0.0036
0.0047
0.0051
0.0074
Median
9.2
6.8
12
7.7
14
1400
1800
3000
730
1300
0.014
0.010
0.017
0.012
0.021
Percentiles
10th
90th
5.1
16
4.2
9.6
8.8
16
4.5
13
9.8
21
550
4300
880
5100
1700
8100
360
1300
720
2000
0.0073
0.024
0.0061
0.015
0.012
0.024
0.0066
0.020
0.013
0.031
a Statistics
ETS Cigs
iF
Exposure
Metric
24-h Avg
Table 8.5: Nicotine Loaded. 24-h Average Nonsmoker SHS Nicotine Inhalation Exposure Concentration [µ g m−3 ], Individual Intake Fraction (iF) [ppm], and Equivalent ETS Cigarettes (ETS Cigs) [d−1 ] Statistics for Surfaces Preloaded
with 50 mg m−2 of Nicotine by No. of Cigarettes Smoked Indoors and Fraction of Time Spent by the Nonsmoker at
Homea
287
288
10
101
100
10−2
10−3
10−4
10−1
10−2
99
95
90
75
50
25
10
5
10−3
1
Intake Fraction
Particles: HAC
2
Base State
HAC 10%
HAC 100%
Figure 8.8: HAC. Log-probability plots of 24-h SHS particle exposure concentration, individual intake fraction, and equivalent ETS cigarette intake distributions
for the base HAC-inactive scenario and two HAC-active scenarios and 1,036 nonsmoking individuals in households where more than 10 cigarettes were smoked
and the receptor spent more than 32 of their time at home. See Tables 8.2, 8.6, and
8.7 for descriptive statistics calculated from the complete distributions.
10−30 Cigs
Overall
0−10 Cigs
10−30 Cigs
Overall
0−10 Cigs
10−30 Cigs
No.
Cigs
Overall
0−10 Cigs
Fract.
at Home
Overall
< 2/3 at Home
> 2/3 at Home
< 2/3 at Home
> 2/3 at Home
Overall
< 2/3 at Home
> 2/3 at Home
< 2/3 at Home
> 2/3 at Home
Overall
< 2/3 at Home
> 2/3 at Home
< 2/3 at Home
> 2/3 at Home
n
nzero
4798 48
1430 38
984
1
1348
9
1036
0
4798 48
1430 38
984
1
1348
9
1036
0
4798 48
1430 38
984
1
1348
9
1036
0
Sample
Mean
15
6.2
10
16
32
990
790
1300
750
1300
0.011
0.0047
0.0076
0.012
0.023
Std.
Dev.
15
6.3
7.9
11
17
710
740
760
540
590
0.011
0.0049
0.0059
0.0089
0.013
Median
11
4.2
8.7
14
30
850
590
1100
620
1250
0.0082
0.0031
0.0062
0.010
0.021
Percentiles
10th
90th
1.4
35
0.26
15
1.5
22
4.1
31
13
54
210
1900
61
1800
420
2400
190
1500
610
2100
0.0010
0.026
0.00020
0.011
0.0014
0.017
0.0031
0.024
0.0089
0.040
a Statistics
ETS Cigs
iF
Exposure
Metric
24-h Avg
Table 8.6: HAC 10%. 24-h Average Nonsmoker SHS Particle Inhalation Exposure Concentration [µ g m−3 ], Individual Intake Fraction (iF) [ppm], and Equivalent ETS Cigarettes (ETS Cigs) [d−1 ] Statistics for 10% Intermittent
Awake-Time HAC Operation by Number of Cigarettes Smoked Indoors and Fraction of Time Spent by the Nonsmoker
at Homea
289
10−30 Cigs
Overall
0−10 Cigs
10−30 Cigs
Overall
0−10 Cigs
10−30 Cigs
No.
Cigs
Overall
0−10 Cigs
Fract.
at Home
Overall
< 2/3 at Home
> 2/3 at Home
< 2/3 at Home
> 2/3 at Home
Overall
< 2/3 at Home
> 2/3 at Home
< 2/3 at Home
> 2/3 at Home
Overall
< 2/3 at Home
> 2/3 at Home
< 2/3 at Home
> 2/3 at Home
n
nzero
4798 48
1430 38
984
1
1348
9
1036
0
4798 48
1430 38
984
1
1348
9
1036
0
4798 48
1430 38
984
1
1348
9
1036
0
Sample
Mean
9.2
3.9
6.5
9.6
19
600
510
780
440
770
0.0068
0.0030
0.0047
0.0073
0.013
Std.
Dev.
7.6
3.4
4.0
5.6
7.8
340
360
320
260
260
0.0057
0.0026
0.0030
0.0043
0.0061
Median
7.5
3.1
5.9
9.2
18
590
480
760
410
760
0.0055
0.0024
0.0041
0.0068
0.013
Percentiles
10th
90th
1.1
20
0.17
8.5
1.4
12
2.7
17
9.8
29
150
1000
37
1000
380
1200
130
800
430
1100
0.00080
0.015
0.00013
0.0066
0.00095
0.0090
0.0020
0.013
0.0063
0.022
a Statistics
ETS Cigs
iF
Exposure
Metric
24-h Avg
Table 8.7: HAC 100%. 24-h Average Nonsmoker SHS Particle Inhalation Exposure Concentration [µ g m−3 ], Individual Intake Fraction (iF) [ppm], and Equivalent ETS Cigarettes (ETS Cigs) [d−1 ] Statistics for Continuous Awake-Time
HAC Operation by Number of Cigarettes Smoked Indoors and Fraction of Time Spent by the Nonsmoker at Homea
290
291
home is heated and a local stack effect occurs in each room due to indoor-outdoor
temperature differences. Here, the flow between rooms with open doorways is 100
m3 h−1 and the balanced flows in and out of each room due to leakage are assigned
a portion of the 0.5 h−1 base outdoor exchange rate in proportion to room volume.
Leakage air flow rates for main rooms thus are in the range 25−50 m3 h−1 .
Under asymmetric flow conditions, air from the outdoors flows into a subset
of the home’s rooms (inlets) and exits back to the outdoors through the remaining
rooms (outlets), as if wind impinging on the house sets up areas of negative and
positive pressure on either side of the home, forcing air across the building shell.
In this case, air travels in a prevailing direction through the house from inlet rooms
towards outlet rooms, although bidirectional flow equal to the case of symmetric
flow is also present across each open doorway. See Section 6.3.2 for a discussion of
the simulation algorithm used to assign flows.
Figure 8.9 shows the differences between the distribution of exposure metrics
under symmetric and asymmetric flow conditions for individuals with the largest
potential exposure in homes. Although the medians for the two distributions are
32 µ g m−3 , the distributions diverge below the 25th percentile with the asymmetric case yielding slightly lower exposures than the symmetric case. The arithmetic
mean and standard deviation for the two distributions are similar across all subgroups (see Tables 8.2 and 8.8). The increased exposures for some individuals can
be explained by receptors who spend time in a downwind room during smoking
episodes in upwind rooms, where pollutants are removed more rapidly for some
occupants spending time in upwind rooms. However, both of these effects are
slight, because the directionality of flows is small compared to the base bidirectional flow across doorways. In Chapter 9, where I consider cross flow between
open windows, rather than just leakage flow, the effect of asymmetric flows is
larger.
292
10
101
100
10−2
10−3
10−4
10−2
10−3
99
95
90
75
50
25
10
5
10−4
1
Intake Fraction
Particles: Flow Symmetry
2
Symmetric
Asymmetric
Figure 8.9: Asymmetric Flow. Log-probability plots of 24-h SHS particle exposure
concentration, individual intake fraction, and equivalent ETS cigarette intake distributions for the base symmetric and asymmetric/directional flow conditions and
1,036 nonsmoking individuals in households where more than 10 cigarettes were
smoked and the receptor spent more than 23 of their time at home. See Tables 8.2
and 8.8 for descriptive statistics calculated from the complete distributions.
10−30 Cigs
Overall
0−10 Cigs
10−30 Cigs
Overall
0−10 Cigs
10−30 Cigs
No.
Cigs
Overall
0−10 Cigs
Fract.
at Home
Overall
< 2/3 at Home
> 2/3 at Home
< 2/3 at Home
> 2/3 at Home
Overall
< 2/3 at Home
> 2/3 at Home
< 2/3 at Home
> 2/3 at Home
Overall
< 2/3 at Home
> 2/3 at Home
< 2/3 at Home
> 2/3 at Home
n
nzero
4798 48
1430 38
984
1
1348
9
1036
0
4798 48
1430 38
984
1
1348
9
1036
0
4798 48
1430 38
984
1
1348
9
1036
0
Sample
Mean
17
7.0
11
18
35
1100
890
1400
830
1400
0.013
0.0053
0.0084
0.014
0.025
Std.
Dev.
16
6.9
8.5
12
18
741
760
780
570
620
0.012
0.0053
0.0064
0.0093
0.013
Median
13
4.9
9.8
15
32
960
720
1300
710
1400
0.0092
0.0037
0.0070
0.012
0.023
Percentiles
10th
90th
1.7
38
0.36
16
2.0
24
4.4
34
15
58
240
2100
75
1900
510
2500
210
1600
680
2300
0.0012
0.029
0.00026
0.013
0.0013
0.018
0.0034
0.027
0.010
0.044
a Statistics
ETS Cigs
iF
Exposure
Metric
24-h Avg
Table 8.8: Asymmetric Flow. 24-h Average Nonsmoker SHS Particle Inhalation Exposure Concentration [µ g m−3 ],
Individual Intake Fraction (iF) [ppm], and Equivalent ETS Cigarettes (ETS Cigs) [d−1 ] Statistics for Asymmetric Flow
Conditions by No. of Cigarettes Smoked Indoors and Fraction of Time Spent by the Nonsmoker at Homea
293
294
8.4 Comparison of All Unrestricted Scenarios
The most highly exposed individuals are of the most interest due to their likely
higher risk for adverse health effects from SHS exposure. As demonstrated in previous sections of this chapter, those nonsmokers who spend more than
2
3
of their
time at home and for whom more than 10 cigarettes were smoked at home during
the day tend to have higher absolute exposures to particles, CO, and nicotine for
all scenarios. Mean SHS exposure concentrations for these individuals were typically at least double those for individuals for whom fewer than 10 cigarettes were
smoked at home and/or who spent less than
2
3
of the day at home. In addition, the
exposure metrics for this subgroup of nonsmokers are approximately lognormally
distributed so that they can be conveniently compared in terms of their geometric
means (GM) and geometric standard deviations (GSD) as presented in Table 8.9.
One feature of the comparison among all the scenarios is that the GM of exposures does not change when leakage flows changes from symmetric to asymmetric,
although the GSD decreases slightly due to increasing exposures at the lower end
of the distribution. Also, continuous HAC operation can reduce the GM of absolute exposure by 40%, slightly reducing the GSD. Long-term loading of reversibly
sorbed nicotine on household surfaces can result in more than a quadrupling in
the GM of absolute nicotine exposure and an approximate halving of the GSD.
Reemission of surface nicotine raises exposures for all individuals in the population cohort, narrowing the distribution of possible exposures for those who reside
in households with chronic exposure.
Another enlightening comparison is that between the exposure metrics simulated for multiple household zones with variation in occupant locations and flow
conditions versus that predicted by a simplified single-zone exposure model. For
the simplified model, occupants are assumed to stay at home all day and fluctations in air-exchange rate from HAC operation or window configurations are neglected (see Equations B.5 and B.6). Reemission of chemical species by desorption
from household surfaces is also ignored. The unadjusted single-zone model, with a
value of 1 for the correction factor f , generally overestimates multizone exposures
295
for particles and CO exposure (Table 8.9). The GM of the single-zone correction
factor for all exposure metrics, i.e., the ratio of multi-zone to single-zone exposure
estimates, is 60−80% for base and asymmetric flow conditions. When the HAC
is running the GM is 40−70% across all metrics. However, for airborne nicotine
the correction factor GM is esimated to be 5−6 when walls are pre-loaded with
nicotine. If an effective emission factor of 1 mg cig−1 was used instead of the total
estimated SHS yield of 5 mg cig−1 , the estimated GM would be closer to 1. For
clean walls, the factor GM is only slightly above 1.0 for all exposure metrics, but
the estimated factor GSD is 2.8, indicating that the unadjusted simplified model is
expected to strongly underestimate the variation in nicotine exposure.
8.5 Sensitivity to Environmental and Physical
Parameters
The results presented in this chapter focus on elucidating the baseline frequency
distribution of residential SHS exposures for a cohort of individuals with typical
household movement patterns, which are considered fairly representative of behaviors for the US, as well as a variety of possible flow conditions. To simplify the
analysis and maintain focus on these primary variables, this cohort was examined
in terms of a fixed set of physical and environmental parameters, specifically a particular house size and layout, base outdoor air-exchange rate, surface deposition
rate, receptor inhalation rate, and doorway flow rate.
However, while these input parameter values were selected to be applicable to
typical residential conditions, it is desirable to understand how results may change
if somewhat different values are chosen. To this end, I introduced small positive
and negative perturbations in eight parameter values equal to 25% of a reference
(fixed) value, which was used to simulate exposures in previous sections of this
chapter. This reference case corresponds to the base simulation condition. A normalized sensitivity coefficient,
δ GM
δP ,
was independently calculated for each base
value and the two perturbations, where δ GM is the absolute change in the geometric mean of the exposure distribution divided by the base exposure value, and
296
Table 8.9: Calculated Geometric Means (GM) and Geometric Standard Deviations
(GSD) for Distributions of 24-h Mean Nonsmoker Particle, CO, or Nicotine Exposure Concentration [µ g m−3 ], Individual Intake Fraction (iF) [ppm], and Equivalent ETS Cigarette Intake (ETS Cigs) [d−1 ] Across Each Scenario and Limited to
Households in Which More Than 10 Cigarettes Were Smoked and Nonsmokers
Spent More than 32 of Their Time at Home (with GM and GSD of the Correction
Factor f for the Simplified Single-Zone Modela )
Metric
Scenariob
24-h Avg
iF
ETS Cigs
a The
GM
GSD
GM f
GSD f
Particles−Base
28
2.0
0.72
1.8
Particles−Asym
31
1.7
0.77
1.5
Particles−HAC 100%
17
1.5
0.42
1.4
Particles−HAC 10%
28
1.7
0.69
1.5
CO−Base
170
1.9
0.69
1.7
Nicotine Fresh
3.0
2.9
1.2
2.7
Nicotine Loaded
14
1.3
5.6
1.4
Particles−Base
1200
1.8
0.65
1.8
Particles−Asym
1300
1.6
0.70
1.6
720
1.5
0.38
1.5
Particles−HAC 10%
1200
1.6
0.63
1.6
CO−Base
1400
1.8
0.63
1.8
Nicotine Fresh
260
2.8
1.1
2.8
Nicotine Loaded
1200
1.5
5.1
1.5
Particles−Base
0.020
2.0
0.65
1.8
Particles−Asym
0.022
1.8
0.70
1.6
0.012
1.6
0.38
1.5
Particles−HAC 10%
0.020
1.8
0.63
1.6
CO−Base
0.024
2.0
0.63
1.8
Nicotine Fresh
0.0043
3.0
1.1
2.8
Nicotine Loaded
0.020
1.4
5.1
1.5
simplified time-averaged single-zone exposure model discussed in Section 7.1 incorporates a
correction factor f , which accounts for time spent out of the house and uneven pollutant concentrations (see Equations B.5 and B.6). The geometric mean of this factor is calculated here for the 24-h
airborne particle exposure for each scenario by dividing the exposure metric calculated for each
simulated individual by the unadjusted single-zone model result for base conditions. Parameter
values include a randomly assigned number of cigarettes smoked in each household and the fixed
cigarette emissions, inhalation rate, air exchange, and surface loss coefficients given in Table 7.1.
b See Table 8.1 for a description of each scenario.
297
δ P is the absolute change in the parameter value divided by the base parameter
value. These coefficients only provide a localized measure of the slope of the response surface in directions of particular parameters, and so they cannot be used
to predict changes in exposure across combined changes in a number of parameters or for large distances along any axis, but only for changes in exposure when a
single parameter is perturbed and the others are held fixed at their base value. A
value of unity for the normalized sensitivity coefficient signifies that the geometric
mean for the exposure distributions changes in equal proportion to the change in
parameter, either in the positive or negative direction indicated by the sign of the
coefficient. For small fractional changes, multiplying a given proportionate change
in a parameter by the coefficient gives the resulting proportionate change in mean
exposure.
So that exposures distributions were approximately lognormal, and the geometric mean accurately represents the median value, individuals were limited to
those having more than 10 cigarettes smoked at home and where the receptor spent
more than
2
3
of the day at home. Except for sensitivity coefficients for changes in
number of cigarettes smoked over the entire day, a sample size of 412 was used to
calculate the geometric mean for each perturbed parameter value. The sample size
was slightly lower or higher for fluctuations in the number of cigarettes smoked
per day. Sensitivity coefficients for perturbations in each of the eight physical and
environmental parameter values, as well as absolute perturbation increments and
corresponding parameter values, are given in Table 8.10 for each SHS particle exposure metric.
The coefficients for individual intake fraction and equivalent ETS cigarette intake are the same as for absolute exposure, except they vary in equal proportion
to perturbations in inhalation rate. Absolute exposure does not depend on inhalation rate at all. Unlike absolute exposure, intake fraction does not depend on the
magnitude of per-cigarette mass emissions and depends only slightly on the total
number of cigarettes a smoker smokes over the day. For either a positive or negative perturbations in cigarette emissions, absolute exposure increases or decreases,
298
respectively, in equal proportion.
For the given base parameter values, exposure is most sensitive to negative perturbations in house volume with a sensitivity coefficient,
δ GM
δP ,
of -1.22, i.e., there
is a 22% larger increase in exposure than the corresponding decrease in volume.
Exposure was also sensitive to a positive perturbation in house volume, which resulted in a 25% lower decrease in exposure ( δδGM
P = -0.75). Air-exchange rate has
a fairly large impact on exposure. A decrease in this parameter results in a 90%
proportionate increase in exposure and an increase results in a 61% proportionate
decrease. In contrast, exposure is fairly insensitive to changes in particle deposition rate, only increasing by about 15% and decreasing by about 14% for negative
and positive parameter perturbations, respectively. Exposure is insensitive to local
perturbations in inter-room air flow through open doorways and to the duration
of cigarettes. These sensitivity coefficients are less than 0.1 in absolute value for
both positive and negative perturbations.
This chapter explores the effect of variation in household occupant location on the
frequency distribution of nonsmoker exposure to SHS particle, carbon monoxide,
and nicotine air concentrations for unrestricted activites where no effort is made
to mitigate exposures. The typical operation of an HAC system is considered, as
well as the effect of asymmetric flows through the house and the effect of longterm nicotine loading on household surfaces. A fixed set of environmental and
physical simulation model input parameters corresponding to a typical 4-room
residence are used to simplify analysis, keeping it focused primarily on the effect of
occupant behavior. However, the sensitivity of exposures to small changes in eight
different physical parameters is explored to provide insight into how exposures
might change under different physical and environment conditions. The simulated
effects of the HAC, asymmetric flows, and nicotine surface concentrations are for
relatively narrow model inputs and may not represent conditions in all, or even
most houses.
0
0.25
0.00625
−0.25
0.005
0.00375
0.25
min−1
37.5
m3
0
30
−0.25
0.25
no. day−1
12.5
22.5
0
−0.25
10
min
0.25
12500
7.5
0
−0.25
10000
µ g cig−1
125
7500
0
0.25
100
−0.25
0.25
m3 h−1
0.625
75
0
−0.25
0.5
0.375
0.25
0.125
h−1
0
−0.25
0.1
0.075
0
0
0
29
0
29
0.12
33
29
0
−0.09
29
27
−0.01
0
29
29
0.02
30
0
0.25
36
−0.25
29
22
−0.02
29
29
0.02
−0.15
25
30
0
0.22
29
36
−0.03
0
28
0.04
29
−0.19
24
30
0
0.31
δG M
29
38
GM
0.00
−0.00
0.46
0.36
−0.05
−0.09
1.00
1.00
−0.07
−0.09
−0.61
−0.87
−0.14
−0.15
−0.75
−1.22
δGM
δP
24-h Mean Exposure, µ g m−3
0.00156
0.00125
0.00093
0.00126
0.00125
0.00128
0.00123
0.00125
0.00127
0.00125
0.00125
0.00125
0.00122
0.00125
0.00127
0.00106
0.00125
0.00152
0.00120
0.00125
0.00129
0.00101
0.00125
0.00163
GM
0.25
0
−0.25
0.01
0
0.03
−0.01
0
0.02
0
0
0
−0.02
0
0.02
−0.15
0
0.22
−0.03
0
0.04
−0.19
0
0.31
δG M
Intake Fraction
1.00
1.00
0.06
−0.12
−0.05
−0.08
0.00
−0.00
−0.07
−0.09
−0.61
−0.87
−0.14
−0.15
−0.75
−1.22
δGM
δP
0.0263
0.0210
0.0158
0.0234
0.0210
0.0191
0.0207
0.0210
0.0215
0.0210
0.0210
0.0210
0.0206
0.0210
0.0215
0.0178
0.0210
0.0256
0.0203
0.0210
0.0218
0.0171
0.0210
0.0274
GM
0.25
0
−0.25
0.12
0
−0.09
−0.01
0
0.02
−0
0
−0
−0.02
0
0.02
−0.15
0
0.22
−0.03
0
0.04
−0.19
0
0.31
δG M
Equiv ETS Cigs, d−1
1.00
1.00
0.46
0.36
−0.05
−0.09
−0.00
0.00
−0.07
−0.09
−0.61
−0.87
−0.14
−0.15
−0.75
−1.22
δGM
δP
P
For the sensitivity analysis, each parameter was varied around a central value in turn, keeping all other parameter values constant at their corresponding central values.
Vol=house volume (each room in equal proportion); Dep=particle deposition rate (same for each room); Aer=overall house air-exchange rate; Flow=flow rate across open
doorways; Mag=cigarette particle emissions magnitude; Dur=duration of a single cigarette; Cigs= number of cigarettes smoked in a single day (not all of which are in the
house); Inh=nonsmoker inhalation rate; GM = geometric mean of distribution for each exposure metric; δ P = normalized difference of parameter value from central value;
δ
δG M = normalized difference of exposure metric from central value; δGM = normalized sensitivity coefficient.
Inh
Cigs
Dur
Mag
Flow
Aer
Dep
0.25
h−1
358.75
−0.25
m3
0
215.25
Vol
δP
Units
287
Values
Name
Parameters
Table 8.10: Sensitivity of the Geometric Mean (GM) of Nonsmoker Inhalation Particle Exposure Metrics to Eight
Physical and Environmental Parameters
299
300
Higher geometric means (GM) of exposure occurred in homes where more than
10 cigarettes were smoked over the day and the nonsmoking receptor spent more
than
2
3
of the day at home, so comparative analysis was focused on this subset
of the simulated population. The operation of the HAC lowers the GM of exposure for this subgroup and increases exposure at the lower end of the distribution. The introduction of asymmetric flows did not change the GM very much,
but slightly decreased GSD, because the exposure of “downwind” receptors is increased. Loading of nicotine on surfaces significantly raises the GM of exposures,
and simultaneously lowers the GSD.
For a given change in house volume and per-cigarette mass emissions, there is
either an equal or approximately equal proportion of change in the GM of particle
exposure. There is close to an equal change in exposure for changes in air-exchange
rate. The sensitivity of absolute particle exposure to changes in other parameters
is appreciably lower.
8.7 References
301
Chapter 9
Tier III. The Effect of Mitigation
Strategies on Exposure Frequency
Distributions in Households with
High Potential Exposure
In the previous chapter (Chapter 8), I simulated baseline frequency distributions
of residential SHS exposure for a random selection of household occupant location
patterns, which are representative of persons living in the US. In the current chapter, I expand this analysis to consider conscious efforts to shield nonsmokers from
SHS exposure. To mitigate SHS exposure for nonsmokers, household occupants
close doors or open windows in the rooms they visit in response to smoking activity, either in the room they currently occupy or in a different room. They may also
change their location patterns to segregate the nonsmoker and the active smoker.
In addition, they may operate portable filtration devices in smoking rooms. For
25 simulation trials, I step a cohort of 1,037 matched smoker and nonsmoker pairs
through different mitigation scenarios, calculating the distribution of 24-h average SHS airborne particle exposure concentrations for nonsmokers. The cohort is
selected to represent cases where more than 10 cigarettes are smoked indoors at
home each day and the nonsmoker is at home for more than
2
3
of the day, and,
therefore, for which high exposures are more likely (Chapter 8). I evaluate the
effectiveness of each scenario in reducing population exposure by examining its
CHAPTER 9. TIER III. MITIGATION STRATEGIES
302
impact on the simulated frequency distribution.
9.1 Fixed Simulation Inputs: Cohort and PhysicalEnvironmental Characteristics
The potential for high SHS exposure in the sampled cohort is illustrated by histograms for the distribution of cigarettes smoked at home, the fraction of the day
the nonsmoker spends at home, and the fraction of the day the nonsmoker/smoker
pair spend in the same room, which are presented in Figure 9.1. Besides the age
of the smoker, the matching of occupants by day of week, the number of cigarettes
smoked, and the time spent at home by the nonsmoker, selected pairs in the cohort
have no other prescribed characteristics. A benefit of using a population with high
potential exposure to study the effects of mitigation strategies is that exposures are
approximately lognormally distributed so that their geometric means or medians
can be used to compare the central tendency across different simulation trials.
For each of the mitigation scenarios analyzed in this chapter, I limit my treatment of SHS species to airborne particles. As in the previous two chapters (Chapters 7 and 8), I also focus and simplify the analysis by assigning fixed values of
physical and environmental parameters corresponding to house size and layout,
base leakage-induced air-exchange rate, particle deposition rate, inter-room flow
rates, daily number of cigarettes smoked by the smoker, and cigarette duration
and emission characteristics, which are typical for US homes. The input parameter
values are given in Table 7.1 and include the characteristics for House #2, which is
described in Table 7.3 and pictured in Figure 8.1.
I found in Chapter 8 that the operation of a central heating and air conditioning
system (HAC), if run with a heavy duty cycle, could significantly decrease SHS exposures due to increased infiltration from supply duct leakage. However, a duty
cycle of 10% during waking hours did not result in much reduction in exposure.
In addition, both low and high duty cycles resulted in a mixing of SHS pollutants
in the house, which increased exposures for the lower portion of the distribution.
In this chapter, which is focused on strategies for mitigating exposures, I only con-
303
10
15
20
25
30
−1
Number of Cigs. [d ]
150
0
0
0
50
50
50
100
100
100
Frequency
200
150
150
250
200
Histograms for Key Exposure Variables
0.7
0.8
0.9
1.0
Time Fraction at Home
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
Smk/NSmk Correl.
Figure 9.1: Frequency distributions in the form of histograms for three key simulation variables for 1,037 simulated households where more than 10 cigarettes are
smoked over a 24-h period and the nonsmoker (receptor) occupant spend more
than 23 of their day. The variables are the number of cigarettes smoked in the house,
the fraction of the day the nonsmoking receptor person spends at home, and the
fraction of the day the smoker and nonsmoker spend in the same room.
304
sider cases in which the HAC is always inactive.
9.2 Programmed Mitigation: Twenty-Five Scenarios
Twenty-five exposure mitigation scenarios, which involve dynamic or continuous
changes in the house’s flow patterns, are listed in Table 9.1 under eight groupings.
The first scenario corresponds to base conditions where no mitigation strategies are
implemented. For this case, all interior doors are left open, except when occupants
are sleeping or using the bathroom. All windows are left closed during all times
of the day and no filtration devices are active. The second scenario corresponds
to a temporal ban on smoking, where smoking activity is prohibited during times
when the nonsmoker is at home, but otherwise base conditions apply. These two
scenarios are intended to provide initial reference cases that bound expected exposures occurring in a house without a complete ban on smoking. Exposures for
each of the other scenarios are expected to fall approximately between these two
bounding cases. For the 25th scenario, portable particle filtration equipment is continuously active in rooms where smoking occurs. The remaining scenarios fall into
six groups and they all involve dynamic changes in smoker or nonsmoker location
or door and window positions. These six groups are characterized as follows:
Door or Window. A smoker or nonsmoker closes a door or opens a window during smoking episodes (4 possible scenarios).
Door and Window. A smoker and/or nonsmoker closes a door and opens a window with only a single door ever closed during smoking episodes and a single window ever open (4 possible scenarios).
Doors. Both smoker and nonsmoker closed a door and one or the other optionally
opens a window during smoking episodes (3 possible scenarios).
Windows-Symmetric. Both smoker and nonsmoker open a window in their room
and one or both close a door during smoking episodes under symmetric flow
conditions (4 possible scenarios).
305
Windows-Asymmetric. Both smoker and nonsmoker open a window in their
room and one or both close a door during smoking episodes under asymmetric flow conditions (4 possible scenarios).
Avoid-Isolate. The nonsmoker avoids being in the same room as the smoker during smoking episodes or the smoker is isolated in the living room where they
may open the window and/or close the door (3 possible scenarios).
As discussed in Chapters 6 and 7, a smoking episode is defined as a continuous
time period during which a smoker occupies a particular room where smoking
activity is occurring, will occur, or has previously occurred.
For time that the nonsmoker and smoker spend together in the same room under base conditions, the door behavior of the nonsmoker takes precedence. However, when a door or window-related mitigation strategy is in effect, door and
window positions reflect an attempt to maximize reduction of the nonsmoker’s
SHS exposure. Hence, when door-closing strategies are active, the door to smoking rooms is always left open during smoking episodes when the nonsmoker and
smoker are in the same room. Similarly, when window-opening strategies are active, the window to smoking rooms is always left open when the nonsoker and
smoker are in the same room. Chapter 6 contains a general discussion of how
simulation scenarios are specified.
The conditions for symmetric and asymmetric flow scenarios are the same
as those used in Chapter 8 and described in Chapter 6. All scenarios involve
symmetric flow, except for those in the Windows-Asymmetric group. Figure 9.2
presents median 24-h interzonal flows calculated across the 24-h average flow
rates of all 1,037 simulated households corresponding to the first 21 mitigation
strategies listed in Table 9.1. For symmetric flow conditions, the living room and
kitchen-dining room doorway flows were 90−100 m3 h−1 , which is close to their
base, open-door values. Flows to and from the bedroom and auxiliary room are
smaller, because their doors are closed when occupants are asleep. For asymmetric
flow conditions and open-window scenarios, larger flows occur out of the kitchendining room, which acts as an inlet, and into the other main rooms, which act as
306
outlets.
As part of the discussion of simulation results for different mitigation scenarios
given below, separate log-probability plots are used to present the full distribution
of 24-h mean particle exposure concentration for the base case and each group of
scenarios given in Table 9.1. Along with the distributions of absolute exposure, I
also present frequency distributions of absolute differences, ∆, between individual
24-h average exposure concentrations for the base case and each mitigation scenario. The distribution of exposures corresponding to each matched pair of house
occupants are fairly stable with the 90th percentile confidence band half-length approaching 10% of the distribution mean.1
Descriptive statistics for each distribution, including means, standard deviations (Std. Dev.), medians, geometric means (GM), geometric standard deviations
(GSD), and 10th and 90th percentiles, are given in Table 9.2. Broadly, these results
show that mitigation strategies can cause decreases in the GM of 24-h average SHS
particle exposure concentration of 3 to 27 µ g m−3 , with the limits of this range
corresponding to scenarios when only the nonsmoker closes doors and when the
smoker is isolated in the living room with the door closed and window open, respectively. Changes in the distribution of exposure were sometimes marked by
either an increase or decrease in the GSD of as much as 20−50%.
For the base case, exposures of every member of the cohort ranged from 1 to
165 µ g m−3 . Individual exposures for all of the other scenarios were also positive,
except for the time ban scenario for which simulated exposures for 265 members
of the 1,037-member cohort (26%) are zero. Absolute changes in exposure for the
"Nsmk Door" (#4), "Nsmk Win" (#6), "Nsmk Door Nsmk Win" (#10), and "Avoid"
(#22) mitigation scenarios were zero for 70 (7%), 4 (0.4%), 4 (0.4%), and 148 (14%)
individuals, respectively. To represent the complete distribution in plots of distributions with zero values, I calculate probabilities using all data values. The plotted
distribution, minus the zero-valued exposures, is truncated at the percentile corresponding to the lowest non-zero value. Every value below this lowest percentile
1 The
ratio of confidence band half-length to the mean is a measure of the error between the
“true” population mean and the sample mean.
Smk Win
NSmk Win
5
6
[Continued].
NSmk Door NSmk Win A combination of scenarios 4 and 6
10
A combination of scenarios 4 and 5
NSmk Door Smk Win
9
Smk Door NSmk Win
The window is open in rooms where the nonsmoker is present
during smoking episodes.
The window is open in smoking rooms during smoking
episodes.
The door is closed in rooms where a nonsmoker is present and
the smoker is not present during smoking episodes.
The door is closed in smoking rooms during smoking episodes
when the nonsmoker is not present.
Same as #1, except smoker cannot smoke while nonsmoker is at
home.
Base state where all interior doors are open during waking
periods not spent in the bathroom and all windows are closed.
Scenario Descriptionb
8
Smk Door/Win
NSmk Door
4
Door and Win 7
Smk Door
3
Time Ban
2
Door or Win
Base
1
Bounding
Abbreviation
No.
Groupa
Table 9.1: Descriptions of Each Residential SHS Inhalation Exposure Mitigation Strategy Arranged by Group
307
A combination of scenarios 4, 5, and 6
A combination of scenarios 3, 4, 5, 6
16 NSmk Door Smk/NSmk Win
17 Smk/NSmk Door Smk/NSmk Win
[Continued].
The same as scenario 16, except with asymmetric
flow conditions
20 NSmk Door Smk/NSmk Win Wind
21 Smk/NSmk Door Smk/NSmk Win Wind The same as scenario 17, except with asymmetric
flow conditions
flow conditions
19 Smk Door Smk/NSmk Win Wind
flow conditions
15 Smk Door Smk/NSmk Win
13 Smk/NSmk Door NSmk Win
12 Smk/NSmk Door Smk Win
14 Smk/NSmk Win
11 Smk/NSmk Door
Wins-Asym 18 Smk/NSmk Win Wind
Wins-Sym
Doors
308
The same as scenario 23, except the smoker also opens the living
room window during smoking episodes in addition to closing the
living room door.
24 Isolate Smk Door/Win
A portable filtration device is operated continuously during
waking periods in each room where smoking occurs at a flow rate
of 80 m3 h−1 and 100% removal efficiency.
The smoker is restricted to being by themselves in the living room
during smoking episodes with the door closed.
23 Isolate Smk Door
25 Smk Filtration
Nonsmoker avoids rooms containing the smoker during smoking
episodes.
22 Avoid
Except where noted, flow patterns are symmetric. For each scenario, no HAC or HVAC system is ever active.
a Groups are generally defined as follows: Base: all windows are closed and interior doors are open during waking non-bathroom periods;
Door or Win: A single door is closed or a single window is opened during smoking periods; Door and Win: A single door is closed and a
single window is opened during smoking periods; Doors: Two doors are closed and possibly a single window is opened during smoking
periods; Wins-Symm: Two windows are opened and possibly one or two doors are closed during smoking periods where symmetric flow
conditions are assumed; Wins-Asym: Two windows are opened and possibly one or two doors are closed during smoking periods where
asymmetric flow conditions are assumed; Avoid-Isolate: Nonsmoker avoids rooms with smoking during smoking episodes or smoker is
isolated in a single room during smoking.
Filt: Portable filtration devices operate continuously during waking periods in rooms where smoking occurs.
b A smoking episode is defined as a continuous time period during which a smoker occupies a particular room where smoking activity is
occurring, will occur, or has previously occurred.
Filt
Avoid-Isolate
309
h− 1 ]
Mitigation
Strategy
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
1→5
100
100
94
99
100
100
94
94
99
99
90
90
90
100
94
99
90
202
191
197
188
Flow to HALL
2→5 3→5 4→5
100
65
64
100
65
64
86
65
61
98
63
64
100
65
71
100
65
71
86
65
67
86
65
67
98
63
70
98
63
70
82
63
61
82
63
67
82
63
67
100
65
71
86
65
67
98
63
70
82
63
67
100
65
69
86
65
65
98
63
68
82
63
65
5→1
100
100
94
99
100
100
94
94
99
99
90
91
91
100
94
99
91
100
94
99
90
Flow from HALL
5→2 5→3
100
65
100
65
86
65
98
63
104
65
104
65
92
65
92
65
101
63
101
63
82
63
87
63
87
63
104
65
92
65
101
63
87
63
183
99
167
99
178
97
162
97
5→4
64
64
61
64
64
64
61
61
64
64
61
61
61
64
61
64
61
128
125
128
125
1→7
50
50
50
50
63
55
63
55
63
55
50
63
55
68
68
68
68
2
4
3
6
2→7
25
25
25
25
63
46
63
46
63
46
25
63
46
67
67
67
67
90
90
90
90
AUX−3
LIV−2
6→7
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
38
38
38
38
7→1
50
50
50
50
62
54
62
54
62
54
50
62
54
67
67
67
67
104
104
104
104
BATH−6
HALL−5
HAC−8
BED−4
Flow from Outdoors
7→2 7→3 7→4 7→5
25
25
25
15
25
25
25
15
25
25
25
15
25
25
25
15
56
25
36
15
41
25
32
16
56
25
36
16
41
25
32
16
56
25
36
15
41
25
32
16
25
25
25
15
56
25
36
16
41
25
32
16
62
25
37
16
62
25
37
16
62
25
37
16
62
25
37
16
6
7
18
81
9
7
20
81
7
7
19
81
10
7
19
81
Outdoors−7
KIT−DIN−1
Flow to Outdoors
3→7 4→7 5→7
25
25
15
25
25
15
25
25
15
25
25
15
25
27
15
25
25
15
25
27
15
25
25
15
25
27
15
25
25
15
25
25
15
25
27
15
25
25
15
25
28
15
25
28
15
25
28
15
25
28
15
38
42
8
38
45
8
38
43
8
38
44
8
Figure 9.2: Tabulation of simulated median 24-h interzonal flow rates
for the first 21 mitigation strategies described in Table 9.1. Flows occur between
rooms of a 4-room house (see Figure 8.1) represented by lines between nodes of the
directed graph to the right. The medians are taken across 24-h average flow rates
for 1,037 simulated households where more than 10 cigarettes were smoked during
the day and the nonsmoker occupant spent more than 32 of their time at home.
Flows from main rooms to and from the central hallway and flows to and from
all rooms and the outdoors (node #7) are shown. All flows to and from the HAC
system (node #8) were zero for all scenarios. For asymmetric flow, inlet rooms
are KIT-DIN (node #1) and HALL (node #5). All other rooms are outlets. The
procedure for simulating residential air flow patterns is described in Chapter 6.
[m3
7→6
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
6
6
6
6
310
311
corresponds to zero exposure or zero exposure difference. In the calculation of
the GM and GSD in Table 9.2, only positive-valued exposures are used. The other
statistics in the table were calculated using both positive and zero-valued exposures.
9.3 Temporal Smoking Bans
A straightforward exposure mitigation strategy is to ban smoking activity in the
house during times when the nonsmoker is at home (a temporal ban). This strategy eliminates direct exposure during smoking episodes, either in proximity to
a smoker or in a separate room. Exposure only occurs for times when residual
SHS is present in the house. Figure 9.3 presents the distribution of 24-h average
particle exposure concentrations for the base case and the case of a temporal ban
on smoking. These two scenarios represent reference cases for residential SHS exposure. The base case corresponds to the nominal maximum exposure for each
cohort member. When the temporal ban is enforced, all individuals in the cohort
experienced a reduction in their SHS exposure and the exposure of 26% of the individuals is eliminated entirely. The median of the distribution of differences from
the base case for a temporal ban is close to 30 µ g m−3 . In the absence of a complete
ban on smoking, the temporal ban scenario is expected to result in the minimum or
close to the minimum exposure for each cohort member, although a temporal ban
in combination with other strategies would likely result in even lower exposures.
9.4 Single-Door or Single-Window Strategies
Some inhabitants of a particular residence may attempt to reduce nonsmokers SHS
exposure through the simple closing of a door or opening of a window. It is clear
from the preliminary simulations performed in Chapter 7 that closing doors in
a multi-room house can reduce exposure for nonsmokers who spend substantial
time in rooms separated from the smoker during smoking episodes and opening
windows can remove pollutant concentrations more quickly, thereby reducing ex-
Scenarioa
Base
Time Banb
Smk Door
NSmk Door
Smk Win
Nsmk Win
Smk Door/Win
Smk Door NSmk Win
NSmk Door Smk Win
NSmk Door NSmk Win
Smk/NSmk Door
Smk/NSmk Door Smk Win
Smk/NSmk Door NSmk Win
Smk/Nsmk Win
Smk Door Smk/NSmk Win
NSmk Door Smk/NSmk Win
Smk/NSmk Door Smk/NSmk Win
Smk/NSmk Win
Smk Door Smk/NSmk Win
NSmk Door Smk/NSmk Win
Smk/NSmk Door Smk/NSmk Win
Avoid
Isolate Smk Door
Isolate Smk Door/Win
Smk Filtration
Mean
35
3.7
31
32
13
15
9.9
13
12
14
31
9.9
13
11
9.0
10.1
8.9
12
11
12
11
24
12
2.7
14
Std.
Dev.
21
3.5
21
20
5.9
6.2
6.0
6.7
6.0
6.3
21
6.2
6.8
5.6
5.8
5.7
5.9
5.9
6.2
5.9
6.2
14
10
2.9
9.3
Median
32
3.1
28
29
12
15
9.3
13
11
13
28
9.4
13
9.9
8.1
9.3
8.0
11
10
11
10
21
9.3
1.9
12
GM
29
3.6
23
26
11
14
7.7
11
10
12
23
7.5
11
9.2
6.7
8.3
6.5
11
8.7
10
8.6
20
7.5
1.7
11
GSD
2.0
2.7
2.5
2.1
1.7
1.6
2.3
2.0
1.9
1.7
2.6
2.4
2.1
1.8
2.5
2.0
2.6
1.7
2.2
1.7
2.2
1.8
2.9
3.0
2.4
Percentiles
10th
90th
12
63
0.0
8.5
7.1
58
9.3
58
5.7
21
7.6
24
2.6
18
4.7
22
4.6
20
6.4
22
6.1
59
2.4
18
4.3
22
4.1
18
1.9
17
3.5
17
1.8
16
5.1
20
3.3
19
4.8
20
3.2
19
9.8
39
1.6
25
0.4
5.8
3.4
26
Table 9.1 for a description of each scenario and Figure 9.2 for the simulated median 24-h interzonal flows corresponding to each scenario. The sample size is 1,037 24-h
average particle exposure concentration values for each scenario. All values are positive, except for the "Time Ban" scenario, for which 265 values are zero. Values of zero
are included in the calculation of all "Time Ban" statistics, except for the GM and GSD.
a See
Filt
Avoid-Isolate
Wins-Asym
Wins-Sym
Doors
Door and Win
Door or Win
Group
Bounding
Table 9.2: Statistics from the Simulated Distribution of 24-h Average Nonsmoker SHS Particle Inhalation Exposure
Concentration [µ g m−3 ] for each Exposure Mitigation Strategy
312
313
102
101
100
10−1
102
∆ [µg m−3]
Base Case and Time Ban
101
95
90
75
50
25
10
5
100
Base
Time Ban
Figure 9.3: Log-probability plot of the frequency distribution for the 24-h particle
inhalation exposure concentration (top panel) for base case and time ban mitigation strategies for 1,037 (base case) or 772 (time ban) nonsmoking individuals with
non-zero exposures who occupied households where more than 10 cigarettes per
day were smoked and the nonsmoker spent more than 23 of their time at home.
Only positive exposures are included in the plotted data. The distribution of absolute change in individual exposure concentrations from the base case, ∆, is also
presented for the time ban mitigation strategy (bottom panel).
314
posure. However, here we examine how exposures change for a population with
a wide range of nonsmoker location patterns. Figure 9.4 presents the distribution
plots for the base case and for cases when either the smoker or nonsmoker close
their room door or either the smoker or nonsmoker open their room window during smoking episodes.
For exposure mitigation that depends on the smoker closing their door during
smoking episodes, the overall variation in 24-h particle exposure concentrations
increases with a drop in the lowest 50% of exposures. A similar but smaller effect
occurs when the nonsmoker closes their door. For both scenarios the room door
was left open whenever the smoker and nonsmoker occupied the same room during smoking. The median of the distribution of differences from base exposure for
each door scenario is 1−3 µ g m−3 .
When the smoker or nonsmoker opens their window during smoking episodes,
the reduction in 24-h mean particle exposure concentrations is much greater than
for door-closing scenarios. The median of differences in exposure from the base
condition was close to 20 µ g m−3 for both window-opening scenarios. The smoker
behavior is only a little more effective, because, for either scenario, a window is
opened whenever the nonsmoker and smoker are in the same room during smoking. All individuals experienced a reduction in exposure, although there was a
greater reduction in the upper portion of the distribution, which resulted in a decrease in the overall standard deviation and GSD of exposure. Those spending
more time in close proximity to the smoker, and therefore receiving the largest
base exposure, appear to receive the most benefit.
9.5 Door and Single-Window Combined Strategies
Instead of relying exclusively on a single closed door or a single closed window
to mitigate SHS exposure, occupants of a household may elect to have smokers
and nonsmokers close a single door and open a single window simultaneously. In
these scenarios a smoker or nonsmoker may act by themselves, affecting both the
door and window configuration in a single room, or they may close a door while
315
102
101
100
102
∆ [µg m−3]
1 Door Closed OR 1 Window Open
101
100
95
90
75
50
25
10
5
10−1
Base
Smk Door
NSmk Door
Smk Win
NSmk Win
Figure 9.4: Log-probability plot of the frequency distribution for the 24-h average SHS particle inhalation exposure concentration (top panel) for base case and
single-door or single-window exposure mitigation strategies for 1,037 nonsmoking individuals in households where more than 10 cigarettes per day were smoked
and the nonsmoker spent more than 32 of their time at home. The distribution of
absolute change in individual exposure concentrations from the base case, ∆, is
also presented for each mitigation strategy (bottom panel).
316
the other occupant opens a window. For time spent by both occupants in the same
room, both doors and windows are left open to maximize reductions in exposure.
As illustrated in Figure 9.5, the single door + single-window scenario having the
largest impact on the base frequency distribution of 24-h particle exposure concentrations is the one in which the smoker both closes the door and opens the window
during smoking episodes. However, all scenarios resulted in a lowering of exposure for all simulated individuals, with the median of differences in exposure from
the base case all clustered around 20 µ g m−3 . The scenarios diverged in the lower
halves of their difference distributions with scenarios involving a smoker’s door
or window causing the largest differences and the combined nonsmoker door and
window strategy having the smallest effect.
Another mitigation option, involving doors and a single window, is to insure
that both the smoker and nonsmoker close their respective doors during smoking
episodes, and one or the other also opens their window. The frequency distributions corresponding to these scenarios, are shown in Figure 9.6. The scenario corresponding to when both the smoker and nonsmoker close doors has little effect
on the upper half of the exposure concentrations distribution, because these exposures are for occupants who spend more time in the same room during smoking
periods. When they are together, doors remain open, which is the base condition.
The other scenarios result in substantial exposure reductions, which are similar to
those for the single-window scenarios described above. The median of differences
in exposure from the base case are again clustered around 20 µ g m−3 . While effective, none of the door and single-window combined strategies reduce exposures
as much as the temporal smoking ban, which resulted in a median for differences
from the base case that was near 30 µ g m−3 .
9.6 Multi-Window Strategies
The fifth and sixth scenario groups involve cases when two windows are opened
simultaneously, one by the smoker and one by the nonsmoker, and doors may
be closed by zero, one, or both occupants during smoking episodes. Under sym-
317
102
101
100
102
∆ [µg m−3]
1 Door Closed AND 1 Window Open
101
95
90
75
50
25
10
5
100
Base
Smk Door/Win
Smk Door/NSmk Win
NSmk Door/Smk Win
NSmk Door/Win
Figure 9.5: Log-probability plot of the frequency distribution for the 24-h average
SHS particle inhalation exposure concentration (top panel) for base case and combined single-door and single-window mitigation strategies for 1,037 nonsmoking
individuals in households where more than 10 cigarettes per day were smoked
and the nonsmoker spent more than 23 of their time at home. The distribution of
absolute change in individual exposure concentrations from the base case, ∆, is
also presented for each mitigation strategy (bottom panel).
318
102
101
100
102
∆ [µg m−3]
2 Doors Closed AND 0−1 Window Open
101
100
95
90
75
50
25
10
5
10−1
Base
Smk/NSmk Door
Smk/NSmk Door & Smk Win
Smk/NSmk Door & NSmk Win
Figure 9.6: Log-probability plot of the frequency distribution for the 24-h average
SHS particle inhalation exposure concentration (top panel) for base case and combined two-door and zero or one open window mitigation strategies for 1,037 nonsmoking individuals in households where more than 10 cigarettes per day were
smoked and the nonsmoker spent more than 23 of their time at home. The distribution of absolute change in individual exposure concentrations from the base case,
∆, is also presented for each mitigation strategy (bottom panel).
319
metric flow conditions, no directionality or cross-flow current of air exists in the
house. The two windows act independently to increase pollutant removal rates
for each separate room with no effect on other rooms. In contrast, under asymmetric flow conditions, cross-flow occurs through the house from inlet to outlet rooms,
modulated by doors closed by either occupant. Cross-flow was demonstrated in
Chapters 7 and 8 to result in somewhat increased exposures for those spending
more time “downwind” from smoking rooms.
Figure 9.7 contains frequency distributions of SHS particle exposure under
symmetric flow conditions and Figure 9.8 presents distributions under asymmetric cross-flow conditions. For each scenario, nearly all individuals experienced a
reduction in exposure. The opening of multiple windows increases removal rates
in the house so much that “downwind” effects, which acted to increase some exposures for cross-flow from leakage flow (see Chapter 8), appear to be overwhelmed.
However, the lower tail of exposures is higher for asymmetric flows than for corresponding scenarios under symmetric flow conditions.
As with single-window scenarios, the largest exposure reduction for all individuals occurs whenever the smoker closes their door and opens a window during
smoking episodes. When the smoker’s door was closed, there was a larger reduction in the lower half of exposures relative to the case of no closed doors or when
the nonsmoker closed doors. The position of the nonsmoker’s door had little impact on exposure, although it resulted in slightly lower exposures in the lower half
of the distribution for the case of symmetric flows. The median for differences in
exposure from the base concentration are just above 20 µ g m−3 , marginally better than the results for scenarios when a single window was opened by either the
smoker or nonsmoker. Overall, the multi-window mitigation strategies are the
most effective of any scenarios considered so far, except for the temporal ban on
smoking.
320
102
101
100
102
∆ [µg m−3]
2 Windows Open (Symmetric Flows)
101
95
90
75
50
25
10
5
100
Base
Smk/NSmk Win
Smk Door & Smk/NSmk Win
NSmk Door & Smk/NSmk Win
Smk/NSmk Door & Smk/NSmk Win
multi-window mitigation strategies under symmetric flow conditions for 1,037
nonsmoking individuals in households where more than 10 cigarettes per day
were smoked and the nonsmoker spent more than 23 of their time at home. The distribution of absolute change in individual exposure concentrations from the base
case, ∆, is also presented for each mitigation strategy (bottom panel).
321
102
101
100
102
∆ [µg m−3]
2 Windows Open (Asymmetric Flows)
101
95
90
75
50
25
10
5
100
Base
Smk/NSmk Win
Smk Door & Smk/NSmk Win
NSmk Door & Smk/NSmk Win
Smk/NSmk Door & Smk/NSmk Win
Figure 9.8: Log-probability plot of the frequency distribution for the 24-h particle
inhalation exposure concentration (top panel) for base case and multi-window mitigation strategies unde asymmetric flow conditions for 1,037 nonsmoking individuals in households where more than 10 cigarettes were smoked and the nonsmoker
spent more than 32 of their time at home. The distribution of absolute change in individual exposure concentrations from the base case, ∆, is also presented for each
mitigation strategy (bottom panel).
322
9.7 Smoker Avoidance and Isolation
From the mitigation strategies explored above, the opening of one or more windows during smoking episodes appears to present the most effective means of
reducing SHS exposure. Door position, by itself, has not presented an effective
means to reduce exposure, because the smoker and nonsmoker may spend significant amount of time in the same room or because the nonsmoker may enter
a room immediately after a smoking episode has occurred. Mitigation stategies
that involve modifying either the smoker or nonsmoker locations in response to
smoking activity are expected to be as or more effective, both by themselves and
in combination with the closing of doors and the opening of windows. These
“location modifying” scenarios are in contrast to all the other mitigation strategies in this chapter, which use the “natural” location patterns of the smoker and
nonsmoker, overlaying them with door or window-related behavior. In locationmodifying mitigation strategies, the smoker smokes the same number of cigarettes
in the house as before. Only the location of smoking is changed.
The first level of location-modifying mitigation I consider is to have the nonsmoker avoid rooms where the smoker is active during smoking episodes. In the
next level, the smoker is forced into isolation in the living room with the door
closed during smoking episodes, and, similar to the avoidance-only level, the nonsmoker avoids the living room during smoking episodes. For the final mitigation
level, the smoker both closes the door and opens the window during their solitary
smoking episodes in the living room.
The distribution of 24-h average SHS particle exposure concentrations for each
location-modifying mitigation strategy and the base case are presented in Figure 9.9. The effectiveness of the avoidance scenario is generally better than that
for the door-only scenarios presented above. The median of the distribution of
differences between the base case and the avoidance scenario is about 10 µ g m−3 .
However, scenarios where the smoker is also isolated in the living room with the
window open and the door closed, or just the door closed, are much more effective. The median of the distribution of differences between these cases and the
323
base case are approximately 30 and 20 µ g m−3 , respectively. The door-only isolation case is of similar effectiveness as single and multiple window scenarios with
“natural” location patterns. Isolation combined with a closed door and open window results in a median exposure reduction that is even better than the benchmark
case of a temporal ban on smoking. While the median of exposure concentrations
for the temporal ban is 3.1 µ g m−3 , the median for the last isolation case is only 1.9
µ g m−3 . However, unlike for the temporal ban, no exposures involving smoker
isolation are eliminated completely.
9.8 Portable Filtration Devices
An alternate approach to mitigating SHS exposure in a residence is to set one or
more portable particle filtration devices into operation throughout the day. This
strategy has the advantage of unattended operation, but the disadvantage of only
removing a portion of SHS pollutants, whereas door and window, i.e., air-flowbased, strategies act to reduce exposure to the full range of smoke constituents.
Figure 9.10 shows comparative plots of the frequency distribution of 24-h average SHS particle exposure concentrations for base conditions and when a portable
particle filtration device, which removes particle flowing through it with 100% efficiency and has an air flow rate of 80 m3 h−1 , is operating continuously in all rooms
where smoking is allowed to occur, i.e., the kitchen-dining area, living room, bedroom, and auxiliary room. The median for differences in exposure from the base
condition is about 20 µ g m−3 . This reduction in particle exposure with respect to
the base case is comparable to the best reduction in exposure, across the entire distribution, for mitigation strategies that increased ventilation by the opening of one
or more windows or when the smoker was isolated in the living room with the
door and window closed.
324
102
101
100
10−1
102
∆ [µg m−3]
Smoker Avoidance and Isolation
101
100
95
90
75
50
25
10
5
10−1
Base
Avoid
Isolate Smk Door
Isolate Smk Door/Win
smoker avoidance or isolation mitigation strategies for 1,037 nonsmoking individuals in households where more than 10 cigarettes per day were smoked and the
nonsmoker spent more than 23 of their time at home. The distribution of absolute
change in individual exposure concentrations from the base case, ∆, is also presented for each mitigation strategy (bottom panel).
325
102
101
100
102
∆ [µg m−3]
Portable Filtration
101
95
90
75
50
25
10
5
100
Base
Smoking Room Filtration
continuous 100% efficient particle filtration and an 80 m3 h−1 flow rate in each
smoking room for 1,037 nonsmoking individuals in households where more than
10 cigarettes per day were smoked and the nonsmoker spent more than 32 of their
time at home. The distribution of absolute change in individual exposure concentrations from the base case, ∆, is also presented for the filtration mitigation strategy
(bottom panel).
326
In this chapter, I build upon the SHS exposure simulations presented in the previous two chapters to explore the effectiveness of specific SHS exposure mitigation
strategies involving the closing of doors and opening of windows during smoking
episodes, as well as the banning of smoking during times when the nonsmoker is at
home, avoidance and isolation of the smoker, and the continuous use of portable
filtration devices in rooms where smoking is allowed. Observed room-to-room
movement patterns are used to generate frequency distributions of SHS particle
exposure for a fixed cohort of 1,037 matched smoker/nonsmoker pairs across 25
separate mitigation scenarios.
I found the most effective mitigation strategy to be isolating the smoker in a
room by themselves during smoking episodes where they close the door and open
the window. Imposing a temporal ban on smoking resulted in an exposure reduction that was nearly as large. Closing the door during solitary smoking episodes,
but not opening the window, reduced exposure by a smaller amount, comparable
to the reduction achieved when one or more windows were opened during smoking episodes that occurred anywhere in the house. Simply avoiding the smoker
when they were allowed to smoke in any room of the house was not as effective in
mitigating exposures, although it was more effective than door-only strategies for
which there was no modification of smoker location.
In the absence of a temporal smoking ban or strict smoker isolation, the opening of windows by either the nonsmoker, the smoker, or both occupants, or the
continous operation of an extremely efficient particle filtration device in smoking
rooms at a flow rate of 80 m3 h−1 or more, appear to be the most effective and practical strategies. However, particle filtration devices are not designed to remove all
non-particulate components of SHS, and therefore won’t protect house occupants
from their deleterious health effects.
The mitigation strategies explored in this chapter have different implications
with respect to energy utilization and occupant comfort. Doors are likely to have
the smallest energy requirement and smallest impact on comfort, whereas, de-
327
pending on climate and weather conditions, the opening of windows could have a
large impact on both energy use and occupant comfort. The use of air filtration devices would consume substantial amounts of electricity, and the noise associated
with their operation may cause occupant discomfort.
A major finding of this chapter is that, while doors are quite effective in blocking the passage of air pollution between two compartments, they are not necessarily effective in reducing exposure in a multizonal residential context. The simulated results in this chapter show that persons following typical, unmodified location patterns in their homes can experience some reduction in exposure by closing the door to rooms where smokers are active and isolated from other persons.
However, because smoking and nonsmoking household occupants tend to spend
time in the same room, the effectiveness of these door-related strategies is diminished. Also the doors-closed case does not speed removal of SHS pollutants from
indoor air and so delayed permeation into other rooms can still lead to inhalation
exposures. Those persons who already spend time removed from the smoker experience small reductions. For doors to be used as an effective means of exposure
reduction, smokers and nonsmoker must be in separate rooms for all or most of
the time during which smokers are active.
Löfroth [1993] concluded that it is impractical to use doors as impediments to
SHS emissions in the course of attempting to reduce occupant exposures. If sequestering a smoker in a single room behind a closed door is considered impractical,
then he appears to be correct in his original assessment. My simulations show that
the opening of windows can, by itself, also result in a substantial reduction of exposure and may present a more practical solution for particular seasons and/or
geographic areas.
9.10 References
328
Part IV
Conclusions
329
The following two chapters provide an evaluation of the simulation
model and model results, and an overall summary and conclusions.
Chapter 10 (page 330) contains a comparison of the simulation model
results in this dissertation with the results of indoor air and personal
exposure surveys.
Chapter 11 (page 339) contains a summary of the main findings in this
dissertation, a discussion of implications for public health studies and
education efforts, and suggestions for future work on SHS exposure
occurring in residences and exposure science efforts, in general.
330
Chapter 10
Model Evaluation
In this chapter, I conduct a partial evaluation of the simulation model, which has
been used in this dissertation to explore residential exposure to SHS. I look separately at the model predictions of indoor SHS air pollutant concentrations and
frequency distributions of personal SHS exposure, comparing both to observed
values. One of the largest concerns with regard to uncertainty in the model lies
with issues surrounding source proximity. Prior to becoming uniformly mixed
in a room, pollutants follow complex convective dynamics, which are not easily
characterized for arbitrary residential conditions. Another source of uncertainty
involves the surface sorption and desorption of semi-volatile species in SHS, including nicotine.
To test the overall performance of the exposure simulation model, its predictions can be compared to empirical distributions of exposure for real populations
of people. While predictions of room concentrations have been modeled accurately in the past, the direct verification of a multizone individual exposure model
has not received much attention. The movement and activity of persons between
and within zones, and the accompanying dynamics of air flow through windows,
doors, and HAC or HVAC systems, may strain the model assumptions and simplifications. Therefore, while not provided here, in the future a careful evaluation of the uncertainties in the model, and a comparison of predicted exposures
to observed ones, should be performed. This effort is critical to the verification of
theoretical conclusions.
CHAPTER 10. MODEL EVALUATION
331
10.1 SHS Concentrations in Rooms
As discussed in Chapter 2, a number of investigators have studied the performance
of multiple compartment indoor air quality (IAQ) models. Generally, these studies
found that models were in satisfactory agreement with observed concentrations.
However, few of these studies are optimal in terms of all of the following: (1)
specificity to tobacco smoke; (2) real-time measurements or those with moderately
high time resolution; (3) known occupant activity patterns; (4) a realistic residential
environment; (5) a substantial variety of house configurations; or (6) a variety of
SHS constituents. Therefore, it remains unclear precisely how well IAQ models
can predict residential SHS concentrations for different SHS species, and varying
occupant activities and environmental conditions.
These concerns aside, the SHS concentrations I simulate in the current work are
in generally good agreement with reported measurements of time-averaged SHS
particle and nicotine concentrations. A number of empirical studies in residencies
or a furnished chamber have resulted in measurements of nicotine and particulate matter concentrations associated with moderate or heavy smoking activity. In
these studies, average concentrations were reported over at least a 24 h period. The
simulated 24-h average PM2.5 concentrations in the current work are comparable
to these empirical results (Table 10.1). For example, the USEPA’s PTEAM study
determined an average PM2.5 concentration in smokers homes of 67 µ g m−3 . The
indoor concentrations used for this result cannot unambiguously be linked with
tobacco smoking, since particulate matter in this size range is sensitive to all combustion processes, including both tobacco smoking and cooking. Through the use
of step-wise regression, smoking was determined to contribute approximately 30
µ g m−3 of PM2.5 mass to indoor concentrations. This value is approximately in the
middle of the range of simulated concentrations I report in the current work.
Simulated 24-h average room concentrations of nicotine, which is a tobaccospecific marker, for both clean and loaded surfaces are also comparable to empirically determined values (Table 10.1). The range in simulated concentrations for
preloaded surfaces are closest to the ranges reported by Singer et al. [2002] and
332
Glasgow et al. [1998]. Concentrations measured in the work of Singer et al. [2002],
and most likely the other experimental studies, were influenced by reemission of
sorbed nicotine. Because the composition of surfaces in rooms and house ventilation rates affect nicotine sorption and desorption processes, the true accuracy of the
simulation is unclear. In addition, my simulation experiments assume that nicotine reversibly sorbs to surfaces so that sorbed nicotine is made available to reenter
the air through desorption at some later time. To the extent that some nicotine undergoes irreversible sorption, the model predictions will overestimate the true air
nicotine concentrations contributed by desorption from surfaces. Evidence from
Piadé et al. [1999] suggests that irreversible sorption can occur, but more research
undertaken in real residential settings is needed to quantify the effect.
Few data exist on the concentrations of SHS-related pollutants in real homes
along with well-characterized occupant activity patterns or environmental conditions. One step in the right direction is an unpublished data set consisting of
airborne particle concentrations measured in the living room of a 365 m3 residence [Ott, 2004] over a period of about 2.25 days. Two adult smokers lived in
the residence during the study period. The observed time series of continuous
1-min average particle concentrations is presented in Figure 10.1. The plotted concentrations are derived from laser particle counts of particles with diameters of
0.3−0.4 µ m, assuming a particle mass size distribution equal to that determined
in Chapter 3. The existence of several very high transient peaks suggests that the
smokers were in close proximity to the monitor for periods in the morning and in
the early evening. The exact time-location profile of each house occupant is unknown, but since the monitor was located in the living room, an assumption that
much of the smoking occurred in the same room as the monitor seems reasonable.
Most of the peaks in particle concentration were under 500 µ g m−3 with many
in the range of 100 µ g m−3 . These levels are in the range of those simulated in
Chapter 7 for a variety of nonsmoker and smoker activity patterns and ventilation
scenarios.
In any given household containing a child and his or her smoking caregiver, a
24-h
24-h
54-h
24-h
7-d
12-h
7-d
24-h
Simulated - Clean Surfacesb
Simulated - Loaded Surfacesb
Ott [2004]c
Singer et al. [2002]d
Glasgow et al. [1998]e
Özkaynak et al. [1993] f
Leaderer and Hammond [1991]g
Coultas et al. [1990]h
−
>70
>5
0−748 (148)
10−20
>25
19
19
[No.]
Range (Mean)
0.6−6.9
0.5−10
0.1−6.2 (1.2)
0.02−29.2 (5.4)
4.9−64
−
14−20
32−77
25−160
(67)
−
−
51
12−78
12−78
[µ g m−3 ]
[µ g m−3 ]
1−8
Range (Mean)
SHS RSP Conc.
Range (Mean)
SHS Nicotine Conc.
the simulation results from the current work, the listed studies are empirical studies of SHS-associated concentrations in real or simulated residences. The studies
have comparable averaging periods and involve monitors that were placed in residences where smoking occurred at an approximate rate of 10 cigarettes or more per day.
b
“Simulated” represents the range in room concentrations from simulation experiments presented in Chapter 7. Results are shown for cases when surfaces were initially free
of nicotine and when they were loaded due to chronic smoking in the house. The range in particle concentrations reflects a variety of nonsmoker and smoker activities and
ventilation rates. c The Ott [2004] data are the unpublished results of a study performed in a home with two smokers over a period of more than 2 days. See Figure 10.1 for a
plot of the measured data. d This study was conducted in a 50 m3 furnished chamber in contrast to the other empirical studies, which were performed in actual residences.
For most of the 24 experiments, 10 or 20 cigarettes were smoked over a typical time period of 3 h. The ventilation rate ranged from 0.31 to 2.0 h−1 . The concentrations
reported here are taken over a 24-h period across all furnishing and ventilation conditions. e For this study, passive nicotine was measured in frequently-used rooms of 39
homes with smoking occupants. Ninety-five percent of homes had one or two smokers. An average of 19−20 cigarettes per day were smoked in the homes. f This study
was the USEPA PTEAM survey of personal exposures and room concentrations in 178 homes where two 12-h samples were taken. The particle results presented here are
for combined 12-h samples from 31 nighttime and 27 daytime room samples. The nicotine results are for combined 12-h samples from 29 nighttime and 26 daytime room
samples. g For this study week-long samples of nicotine and RSP were made in the main living area of 96 homes. The results shown represent nicotine and RSP levels for
those 18 homes for which more than 70 cigarettes were reported to have been smoked over the 7-day period. In addition to cigarettes, homes may have contained other
particles sources, such as as stoves, heaters, and/or fireplaces. h For this study 24-h samples were made in 10 homes with smokers.
a Besides
Time
Averaging
Reference
Study/
Cig. Smoked
Table 10.1: Comparison of Simulated and Observed SHS Respirable Suspended Particle (RSP) and Nicotine Concentrations Measured in Rooms of Residences or in a Furnished Chambera
333
334
similar exposure pattern to that pictured in Figure 10.1 may occur, because source
and receptor are in close proximity for much of the day. The simulated results presented in the current work are, therefore, likely to be most accurate for situations
where receptor and source persons can be assumed to not spend much or any time
in extremely close proximity during periods of smoking activity.
10.2 SHS Personal Exposure
Some of the large-scale particle exposure surveys listed in Table 2.5 on page 38
report fixed-site indoor concentrations and/or average personal exposure concentrations for persons who spent some or most of their day in a residential location where smoking was allowed, as well as for persons who spent time in other
smoking and nonsmoking locations. The PTEAM study, which was an especially
in-depth survey of representative particle exposures for nonsmokers living in a
city in California, found that 12-h average daytime personal exposures for PM10
were 35 µ g m−3 (28%) larger than concurrent 12-h indoor average concentrations
for those living in homes with smokers [Özkaynak et al., 1993]. This result is attributed to a “personal cloud effect”. Personal exposures for those living in homes
with smokers were about 23 µ g m−3 larger, on average, than those without smokers.
In the current research, I find that simulated 24-h average personal exposure
concentrations for SHS particles range from 24 to 61 µ g m−3 for receptors in the
same room as a smoker in a 4-room house (Chapter 7). For those that avoid being
in the same room as a smoker, the exposure concentrations range from 5 to 25
µ g m−3 . These exposures are 1−3.5 times average concentrations in main living
areas with increases ranging from 0 to 69 µ g m−3 . For receptors that avoid being
in the same room with the smoker, personal exposures were generally lower than
room concentrations with differences as large as 57 µ g m−3 .
In light of the above, it appears as though my simulated personal SHS particle exposure concentrations are roughly consistent with the results of PTEAM.
It seems likely that at least a portion of the personal cloud, or the lack thereof,
1500
1000
500
0
Particle Mass Concentration [µg m−3]
6PM
Cigarette(s)
Midnight
6AM
6PM
Day: 7 AM − 11 PM
Time of Day
Noon
6AM
Night: 11 PM − 7 AM
Midnight
Noon
6PM
Figure 10.1: Plot of the total SHS respirable suspended particle (RSP) mass concentration time series measured as
consecutive 1-min averages over 2.25 days in the living room of a 365 m3 single-level, detached residence [Ott,
2004]. Cigarettes were typically smoked in the evening between 5 PM and 11 PM and in the morning between
7 AM and noon. The mean particle concentration over the 2+ day period was 51 µ g m−3 . Peaks during or just
after smoking activity were consistently close to 250 µ g m−3 and some transient peaks reached 2,000 µ g m−3 .
Background levels, which occured when no cigarettes were active, and after particles were cleared from the house,
were relatively close to 0 µ g m−3 , indicating the likely absence of any non-tobacco sources of particles in the
measured size range.
Noon
Multi−Day Particle Monitoring in a Smoking Household
335
336
for persons living in smoking homes, can be explained by multi-compartment
effects. However, a more precise evaluation of model predictions against the
PTEAM results would involve simulating the population of PTEAM participants
with matched house characteristics and personal activity patterns. Fortunately,
many auxiliary quantities were gathered as part of the PTEAM personal exposure
study, including house air-exchange rates and time-activity profiles for monitored
occupants.
Figure 10.2 contains a plot of the time-location profiles for all 178 PTEAM respondents sorted from bottom to top by the total amount of time each subject spent
at home. The most striking feature of these profiles is the overwhelming amount
of time spent at home over a time period of approximately 24-h. Even the 50% of
the sample that spent the least amount of time at home still spent the bulk of the
12-h period between 8 PM and 8 AM at home. These time-profiles can be used
to subdivide the 12-h average personal exposure concentrations measured as part
of PTEAM into groups that spent different amount of time at home. An improvement over the PTEAM time-activities would be to include time resolution of the
locations and activities of subjects in their homes, including the rooms that were
visited, the positions of doors and windows, as well as the use of combustible
tobacco products and other sources of air pollution.
In the future, a systematic comparison between the results of the simulation
model and a study like PTEAM, which is representative of a large population,
would be desirable. In general, a careful comparison of the distribution of simulated and observed exposures, taking into account specific housing characteristics
and occupant behavior patterns, potentially allows for three levels of analysis: (1)
an evaluation of the the general performance of the model; (2) calibration of the
model; and (3) use of the model to interpret features in the empirical exposure
distribution.
While PTEAM has collected some of the necessary variables to facilitate these
kinds of analysis, more studies are needed. A large study with a complete set of
variables, similar to or exceeding the level of the PTEAM effort, would be diffi-
337
cult and expensive. A more manageable approach might involve using carefully
scripted location and activity profiles for a small number of houses, where the
level of information detail could be expanded, including the use of real-time or
near real-time monitoring of room concentrations, personal exposures, personal
activities, house configuration, and environmental characteristics. By systematically varying each study factor, the simulation model could be thoroughly tested
across a variety of important scenarios.
10.3 References
Coultas, D. B., Samet, J. M., McCarthy, J. F., and Spengler, J. D. (1990). Variability
of measures of exposure to environmental tobacco smoke in the home. American
Review of Respiratory Disease, 142(3): 602–606.
Glasgow, R. E., Foster, L. S., Lee, M. E., Hammond, S. K., Lichtenstein, E., and
Andrews, J. A. (1998). Developing a brief measure of smoking in the home:
Description and preliminary evaluation. Addictive Behaviors, 23(4): 567–571.
Leaderer, B. P. and Hammond, S. K. (1991). Evaluation of vapor phase nicotine
and respirable suspended particle mass as markers for environmental tobacco
smoke. Environmental Science and Technology, 25: 770–777.
Ott, W. R. (2004). Unpublished particle concentration data measured in the living
room of a two-smoker household. Personal communication.
Özkaynak, H., Xue, J., Weker, R., Butler, D., and Spengler, J. (1993). The Particle
Team (PTEAM) Study: Analysis of the Data; Volume III. Contract Number 68-024544, U.S. Environmental Protection Agency, Research Triangle Park, NC.
338
0.6
0.4
0.2
0.0
0.8
1.0
PTEAM Time−Location Profiles; n = 178
8PM
MIDNT
4AM
8AM
NOON
4PM
8PM
Time of Day
INSIDE HOME
INSIDE OTHER
OUTSIDE HOME
OUTSIDE OTHER
TRAVEL ON ROADWAY
Figure 10.2: Vertically-stacked time-location profiles based on written diaries collected from the 178 participants of the USEPA’s PTEAM study [Özkaynak et al.,
1993, 1996]. The diaries for each person generally began when a crew arrived at
their home in the mid-evening to begin monitoring of airborne particle and nicotine exposures for the first ∼12-h (evening) monitoring period, and they ended
when monitoring was completed in the early evening of the following day after
the second ∼12-h (daytime) monitoring period. The five recorded location categories, identified by color, are an exhaustive set, consisting of time spent inside
the home, time spent outside the home, time spent traveling, and time spent in
some other indoor or outdoor location. The stacked profiles, each lasting a total of
approximately 24 h, are sorted from bottom to top according to the total amount
of time each person spent inside their home (shown in red). The results of the
PTEAM exposure monitoring survey, and others like it, can be compared to the
results of a simulation model to broadly evaluate the accuracy of the model.
339
Chapter 11
Overall Summary and Conclusions
Secondhand tobacco smoke (SHS) is a ubiquitous indoor air pollutant that has been
positively associated with a range of adverse health effects at typical residential
levels. With the advent of restrictions on smoking in workplaces in US and the
burgeoning US trend in banning smoking in public venues for eating and drinking,
the remaining locations for substantial SHS exposure are likely to be limited to
homes, automobiles, and outdoor public or private settings. Since people spend
the majority of their time in their homes, homes will likely remain the primary
location for continuing SHS exposure. Families occupying single-unit residences
will be a major target of public health interventions. Children, the demographic
group who have been shown to be especially sensitive to a number of SHS-related
illness and who spend as much or more time at home than any other group, are of
particular concern.
The current study on determinants of residential SHS exposure is important
and timely. By looking quantitatively at the causes of variation in exposure, from
the general multizonal nature of homes to the specific effects of occupant location,
door and window poisitions, and HAC/HVAC operation, this study informs the
field of exposure science by posing testable theoretical predictions. It also informs
the field of public health by advancing our understanding of how exposure might
be effectively controlled.
Previous studies of SHS exposure, whether model-based or experimental, have
not carefully examined the movement of pollutants and human beings amongst
CHAPTER 11. OVERALL SUMMARY AND CONCLUSIONS
340
different rooms of a house. In this dissertation, I have compiled a wide array of
data on cigarette emissions, housing characteristics, and human activity patterns
and devised an original simulation model, which consolidates and makes innovative use of these input data, especially room-to-room human location patterns, to
track individuals, pollutant concentations, and exposures on a minute-by-minute
basis. This model represents a formalization and extension of existing exposure
frameworks and is ideally suited for the structured design of simulation trials intended to elucidate mechanisms of residential SHS exposure. The results of an
ensemble of simulation trials comprise the main subject of the dissertation.
In this final chapter of the dissertation, I first summarize the characteristics and
significance of the SHS exposure simulation model, the development and application of which constitutes the centerpiece of my work. Next, I summarize the
results of applying the model, making suggestions for enhancements to the model
and for new analyses that might be performed with the updated version. The final
two sections of this chapter are devoted to putting my results and experience in developing and applying an exposure simulation model in the context of improved
public health initiatives and making recommendations for future modeling and
measurement-based exposure assessment studies.
11.1 A New Exploratory Modeling Tool
Two kinds of exploratory approaches might be used to study exposure mechanisms, one purely empirical and the other theoretical. Although the field of exposure science includes many mechanistic studies of pollutant dynamics and transport, the design and implementation of a series of experiments to explore the
mechanisms of exposure, incorporating measurements with high space-time resolution for both pollutants and people, is both technically difficult and expensive
and no such experiments for SHS exposure have been conducted. Also, there are
currently few active efforts that use a suite of well-designed simulated experiments
to explore exposure relationsips in detail, and none that attack the question of SHS
exposure. Therefore, because of the large burdens involved and a relatively low
341
return on effort for a purely experiment approach versus a model based one, and
because I would make a new contribution with either one, I chose to make use of
a custom computer simulation model.
In contrast to efforts that may involve the pure execution of theory to predict
cross-sectional results for a fixed population, the methodology I have used in this
dissertation is in keeping with the scientific tradition of conducting carefully controlled laboratory-based experiments to isolate the effects of a small number of key
variates on the outcome variate of interest. My outcome variate of interest is exposure to SHS in detached homes and the key variables studied here are the location
timelines for human beings in rooms of a house, the time-varying configuration of
doors and windows, and the operation of HAC/HVAC systems or portable filtration devices. To pinpoint the effect of key variables on the response variable, I held
a host of physical environmental variables fixed at values that might typically be
found for residences in the US. These conditioning variables include the house layout and dimensions, the base outdoor air-exchange rate of homes, inter-room air
flow rates, surface reactivity coefficients, and cigarette emissions characteristics.
In contrast to studies that consider purely statistical, or non-mechanistic, relationships between variates, this approach explicitly considers the physical mechanism
by which exposure occurs.
The rationale for this deterministic approach is that small-scale changes in individual behavior on time scales of minutes or hours are expected by themselves to
have large impacts on exposure, owing to changes in proximity of the nonsmoker
to the smoker, including time spent in separate rooms, outdoors, or away from
the house. The complex interplay among spatially and temporally varying factors – including cigarette smoking, interzonal flow, nonsmoker location, and flowinfluencing configuration choices – necessitates the use of a sophisticated, mechanistic simulation tool.
The simulation model incorporates a multi-compartment, mass-balance indoor
air quality model that accounts for pollutant emissions in any compartment at any
moment in time, transport of pollutant between compartments, pollutant removal
342
by means of outdoor air exchange, room-specific particle filtration, pollutant loss
onto surfaces by deposition or sorption, the possible reemission of pollutant from
contaminated surfaces, and HAC/HVAC operation, which may involve outdoor
air introduction (HVAC only), interior air recirculation, or particle filtration. For
all the simulations I perform in this dissertation, I assume an HAC system, which
does not incorporate forced-air ventilation. Duct leakage and pressurization from
closed interior doors, which may be associated with the home’s HAC system and
result in an increase in air infiltration and exfiltration rates, are also taken into
account. The indoor air quality model is currently equipped to treat a variety of
chemical species including airborne particulate matter, sorbing/desorbing semivolatile compounds such as nicotine, and nonreactive tracer gases such as carbon
monoxide or sulphur hexafluoride. The central assumption of the indoor air model
is one of instantaneous mixing of pollutants within each zone.
The indoor air model component is defined by a set of n coupled differential
equations, one corresponding to each air compartment (i.e., room), with an additional set of n linked equations corresponding to a surface compartment in each
room. The model differential equations are solved numerically to obtain the dynamic or time-averaged air and surface concentrations of pollutants in each room
for time resolutions as low as a minute. Theoretically, any number of room and
surface compartments can be treated. The model is also readily expanded to consider multiple pollutants.
Using my new simulation model, I executed a set of trials that were designed
to reveal the sensitivity of exposure to each key variable or combination of variables (see Chapters 7−9). Since the simulation model is based on well-established
exposure and indoor air theory, I have confidence that the results are an accurate
representation of nature given the set of controlled conditions. In particular, the
broad changes in distributions of exposure are those that one would expect to occur with real observations. The model results may also be considered as testable
scientific predictions. After comparison between real and predicted exposures, the
model structure can be improved and parameter values fine-tuned.
343
11.2 Findings: Sensitivity of Exposure to Key Variables
Broadly, my findings include a confirmation that the multi-compartment character
of homes can lead to substantial variation in SHS exposure. Throughout my analyses of both scripted exposures and frequency distributions of exposure in Chapters
7−9, it is apparent that different amounts of time spent in rooms either away from
a smoker or in the same room as a smoker can lead to large differences in exposure
for three major constituents of SHS: particles, nicotine, and carbon monoxide (CO).
The multi-compartment nature of a house can be enhanced and exploited for the
purpose of mitigating exposure through the use of open windows to supply local
ventilation, closed doors to impede air flow between rooms, or some combination
of the two.
With regard to exposure assessment, in general, my findings confirm the importance of characterizing SHS concentrations in the rooms that people visit, the
time spent in each room, and the activities peformed in each room. A careful spatial and temporal tracking of pollutants and people in the residential environment
permits a refined classification of exposure.
The geometric mean (GM) of simulated 24-h average SHS particle exposures
for a fixed population of households where more than 10 cig d−1 were smoked
and the nonsmoker spent more than
2
3
of the day at home ranged from 17 to 28
µ g m−3 across different flow scenarios. These results are similar to the observed
increase in average particle exposure for persons living with smokers versus those
who don’t. The GM of the individual particle intake fraction for the base exposure scenario and the same population was about 1200 ppm, which is much larger
than the estimated population intake fraction for motor vehicle and power-plant
emissions.
The dynamic behavior of specific SHS pollutants can strongly affect exposures.
Because nicotine sorbs very rapidly onto surfaces, the simulated relative differences in room-to-room concentrations during smoking is greater for nicotine than
for particles or CO. This effect has been observed in empirical house monitoring
344
studies. The result is that simulated avoidance behavior has a larger proportional
effect on exposure to nicotine than either particles or CO. When I simulated nicotine exposure for the case of substantial loading of nicotine on walls, the background exposure of all household occupants increased. Clean walls resulted in
as large a decrease in scripted exposures as did avoidance behavior by the nonsmoker. The 10th percentile of the 24-h average nicotine exposure concentrations
for clean and loaded surfaces was 0.73 and 9.8 µ g m−3 , respectively, and the 90th
percentile was 9 and 21 µ g m−3 , respectively. While the range of simulated 24-h
average nicotine rooms concentrations is similar to that observed in field studies,
the true behavior of nicotine in homes remains unclear. In particular, the sorption
of nicotine onto surfaces may not be fully reversible, as assumed by the simulation
model.
Human activity drives the largest variation in exposures for a given house and
pollutant type. For scripted location patterns, avoiding a smoker could result in a
3−5 times decrease in exposure from 24−61 µ g m−3 down to 5−25 µ g m−3 across
a variety of flow scenarios. The interquartile range in the simulated base frequency
distribution of 24-h particle exposure concentration, which was simulated for fixed
housing and pollutant characteristics, was over 30 µ g m−3 and the difference between the 90th and 10th percentile was over 38 µ g m−3 . This variation in exposure
is entirely due to variation in household occupant location patterns.
Additional variation in exposure resulted when mitigation strategies were enforced for a cohort of 1,037 simulated households for which more than 10 cig d−1
were smoked and the nonsmoker was at home for most of the day. The most effective SHS exposure mitigation strategy involved the isolation of a smoker in a single
room of the house during smoking episodes where the door was kept closed and
the window was opened during smoking. The next best strategy was the banning of smoking in the house for time periods when the nonsmoker was at home.
Avoidance of the smoker as they smoked throughout the house normally was less
effective, but it was more effective than closing doors during smoking episodes
for normal location patterns. Without the modification of smoker and nonsmoker
345
location patterns, the opening of windows and continuously active particle filtration devices in each smoking room were the most effective mitigation strategies.
In climates where it is feasible, the opening of windows during the normal routine of household occupants may be a more practical approach to mitigating SHS
exposure than avoiding the smoker or isolating them in their own room.
Apart from occupant activity, the operation of a house’s HAC system could
decrease simulated exposures a small amount, due to an increase in the house’s
infiltration rate caused by supply duct leaks. The directionality of leakage or openwindow flows across house boundaries did not change exposures much, but crossflows could slightly increase exposures for those who spent a significant amount
of time in “downwind” locations from the smoker.
11.3 Potential Enhancements to the Simulation Model
The simulation model I use in this dissertation could be revised to facilitate a more
in-depth exploration of several current key variates. These changes may involve
structural modification of the model or simply the expansion or refinement of input parameter values for the purpose of exploring a broader array of residential
exposure scenarios, environmental conditions, or pollutant properties. For example, the model could be applied to a broader array of housing types or a more
detailed set of flow patterns.
A second type of model enhancement would involve refining the mechanism
by which human behavior is characterized or modified for the purpose of mitigating exposure and considering more complex interactions between multiple individuals. Most of the mitigation strategies I consider are those where occupant
activity that changes the position of a door or window is superimposed over a
natural pattern of room-to-room movement for smokers and nonsmokers. For a
few simulation experiments, I examined the effect of a scripted “avoider” location pattern on exposure or considered mitigation strategies involving modification of location patterns, moving the nonsmoker to another room whenever smoking commences, and/or restricting smoking to a single room. However, further
346
exploration of the complex interaction between individuals in a household is warranted.
Most households contain more than two individuals with diverse roles, personalities, occupations, and smoker status. Therefore, an enhanced model might
describe SHS exposure occurring with families of three or more persons having a
range of different family roles, smoking behaviors, and characteristic activity patterns. Their location patterns could be matched according to typical household
relationships or demographic group. For example, a child and their smoking caregiver might be apt to spend all or most of their time in the same room. Using logical
family groupings, a systematic analysis of the effect of variation in the correlation
between particular classes of occupant activity patterns could be conducted.
A major structural enhancement to the model would involve adding another
layer of possible proximity between smokers and nonsmokers. Currently, the highest degree of proximity that makes any difference is when the nonsmoker occupies
the same room as the smoker. However, the distance between smoking and nonsmoking persons in a given room will likely result in different gradations of exposure. Unfortunately, there are limited data that could be used to support the
inclusion of a finely resolved proximity effect in an exposure simulation model. I
discuss the need for more experimental investigation of this effect below.
11.4 Improving Public Health Research and Education
The knowledge, understanding, and experience gained in formulating and applying a theoretical characterization of residential SHS exposure can be used to inform
health researchers, who require exposure assessment in the course of their studies,
and help them to design better self-report exposure questionnaires, time-diaries,
or physical exposure measures. While these researchers will likely not make use
of the mathematical and computer-based models I have used in this dissertation,
my broad findings on the need for careful tracking of exposures, concentrations,
and/or human behavior in time and space, and specific findings on the magnitude of effects associated with doors, windows, and other factors, may help to
347
guide and educate epidemiologists, risk assessors, public health clinicians, environmental regulators, and ordinary citizens.
11.4.1 Epidemiology
Environmental epidemiology studies that provide the link between SHS exposure
and acute or chronic adverse health effects can benefit from better measures of
SHS exposure. Many of the studies identifying a link between SHS and lung cancer depend on fairly crude characterizations of exposure, such as the number of
cigarettes a spouse smokes in the house per day. Despite fairly imprecise exposure
measures, an effect was established and in some cases a clear exposure-response
relationship was observed with the probability of an adverse outcome rising with
a larger number of cigarettes smoked. However, if questionnaires or diaries are
revised and expanded in light of sophisticated information on how exposure occurs in homes, such as that presented in this dissertation, or if they were used in
combination with better measurements of personal exposure, then a more well defined exposure-response relationship might be established for lung cancer, as well
as other SHS-related ailments, such as asthma, SIDS, or heart disease. Collection
of more and better data on exposure concentrations and/or the behavior of household occupants would contribute to a more complete characterization of exposure
for each family unit.
11.4.2 Public Health Interventions
Public health interventions aimed at reducing SHS exposure through partial bans
on smoking in the house or mitigation efforts, such as segregation of smoking and
nonsmoking household occupants, filtration, or enhanced ventilation can, like epidemiologic investigations, be informed by the work presented in this dissertation.
The model results for effects of doors, windows, and removal devices can be used
to help design logical and sound approaches for reducing the SHS concentrations
to which people may be exposed.
Some health researchers have suggested that the use of real-time, personal ex-
348
posure monitors coupled with time diaries with information on specific scenarios
in particular homes would help provide clearer evidence of the efficacy of different physical and social strategies for reducing residential SHS exposure (see Chapter 2). Time varying micro-level behaviors of smoking parents, such as the distance
of smokers from childrens’ rooms, the duration and timing of window-opening,
and the positions of doors, may have large impacts on the magnitude of exposure
for infants, children, and other house occupants. However, a complete experimental characterization of such a system is likely to be burdensome and would reduce
the number of households that could be sampled. One possible approach may be
to measure the real-time behavior of people and their impact on their environment
(doors, windows, and HAC/HVAC), and fuse this information with model predictions, or a set of guidelines, to estimate exposures. The model could be used to
better characterize variations in exposure than would be possible using traditional
self-report methods, such as questionnaires, or with biomarker approaches.
Putting the effectiveness of specific physical measures aside, the actual feat
of getting smokers to change their smoking patterns is the true crux of interventions. The intensive counselling of smokers, and their and their family’s close involvement with finding ways to reduce smoke exposure, has been shown to be
effective in getting intervention targets to modify their behavior. Barriers to behavior change may stem from smokers being unaware or skeptical of the health
risks of smoking, or they may consider it overly impractical or inconvenient to restrict their smoking. By leading smokers and nonsmokers into active discussions
and problem-solving sessions, they are more apt to assimilate and believe in the
importance and possibility of change. The knowledge imparted by the detailed
exploration of exposure presented in this dissertation will not only lend itself to
identifying physically effective exposure reduction techniques, but can be used as
material to drive dynamic and critical discussions between family members that
evaluate specific approaches, which may be especially practical or attractive to
particular households.
Although they are of importance in reducing health risks from SHS, partial
349
smoking restrictions and mitigation efforts may be considered an intermediate
step towards total relief from SHS exposure. Ultimately, the goal of health interventions, in terms of the highest protection from risks of adverse health, is to eliminate smoking entirely in homes where children or other nonsmoking residents
can be exposed, and even to reduce or stop smoking behavior itself. If partial restrictions can be instituted in a given household, inconvenience and constant or
increasing social pressure from other household members, coupled with outside
pressure from co-workers, friends, or the media, may convince smokers to either
never smoke at home, severely curtail the amount of their smoking, or to give up
their habit entirely.
11.4.3 Educational Materials
The exposure modeling results from Chapters 7−9 can be used to inform persons
requiring information on exposure to SHS, and for whom it is not possible to obtain
detailed exposure or concentration measurements. In the future it will be possible
to create statistical tables of exposure metrics (e.g., means and standard deviations)
as a function of smoking activity, door and window positions, room-specific filtration, HAC/HVAC use, and the relative time-profiles of smoking and nonsmoking
household occupants. These tabulated exposures will provide an accessible resource in the education of those interested in the protection of exposed persons, especially children, in the development of effective and practical exposure reduction
measures, and for tobacco-related researchers involved in epidemiological studies,
public health interventions, or risk assessments.
11.4.4 Guidelines for Residential Air Quality
Governmental agencies, such as the USEPA and CARB, work to establish standards for levels of ambient air pollution, which are designed to protect the health
of persons in the US, particularly those who live in cities suffering from motorvehicle induced smog. Carbon monoxide (CO) and particulate matter are two of
the USEPA’s criteria pollutants considered harmful to public health and the envi-
350
ronment, for which National Ambient Air Quality Standards (NAAQS) have been
established in accordance with the 1990 Clean Air Act. The 24-h NAAQS for PM2.5 ,
which is the size range encompassing most of the particulate matter in SHS, is currently 65 µ g m−3 and the annual NAAQS for PM2.5 is 15 µ g m−3 . The 1-h NAAQS
for CO is 10,000 µ g m−3 and the 8-h standard is 40,000 µ g m−3 .
While the CO standard would likely never be exceeded due to smoking activity
in homes (Chapter 8), some persons living with a smoker might have 24-h average
exposures that exceed the PM2.5 standard, and if they have chronic exposures for
most days of the year, the annual standard would be exceeded as well. However,
ambient air quality standards were not designed to be applicable to the range and
intensity of the toxic constituents in SHS, which have been shown to cause adverse
health outcomes for typical residential concentrations. While ASHRAE standards
for indoor ventilation exist, it would be beneficial for regulatory agencies, such as
the USEPA, to establish indoor air concentration guidelines, preferably for specific
sources like SHS, using estimates of risk based on established health and exposure
data. Such a guideline could be used by regulators, builders, and residents to take
a variety of steps towards improving health.
11.4.5 Health Risk Assessment
The time seems ripe for a national or state health or environment agency to enact
indoor air quality concentration standards or formal guidelines for SHS exposure.
To provide the exposure component for an SHS health risk assessment, data from
a modeling effort, such as the one presented in this dissertation, could be used
to estimate realistic frequency distributions of exposure for different SHS compounds. There has been some initial work in this area, which predicted the risk of
SHS-related cancer and non-cancer endpoints using reference concentrations and
risk factors for selected gaseous components of SHS along with estimates of average residential SHS exposure concentrations (based on exposure-relevant emission
factors and narrow characterizations of US housing stock and smoking behavior).
This work could be extended by using simulated frequency distributions of resi-
351
dential exposure to SHS particle and toxic gas species to more carefully examine
the probability of adverse health associated with specific behaviors and exposure
scenarios. One could then predict the fraction of the US population that would
be expected to suffer from ill health when no exposure reduction measures are in
place and compare it to the expected fraction when, for example, smokers are segregated behind a closed door during smoking episodes. The cutoff point where the
risk of adverse outcome is less than a prescribed probability could be established
as an indoor air concentration guideline.
11.5 Future Exposure Research
The study of SHS is warranted, in general, because of the magnitude of its associated risk, but also because plentiful data resources exist to support a comprehensive model-based investigation, from emissions to housing characteristics to
human activity patterns. My investigation has touched on most of the issues that
arise in any complete exposure assessment, e.g., appropriate space-time resolution,
interaction of source and receptor activity, pollutant dynamics, selection of parameter domain, and the design of real or simulated experimental trials. It provides
a reference, or perhaps a template or starting point, for future studies of smoking,
SHS, or other pollutants and pollutant sources.
Future exposure studies should seek to further verify and parameterize exposure models, including the physico-chemical models of pollutant dynamics upon
which they are likely to be based. As discussed earlier, these models provide
the means to extrapolate and generalize findings to arbitrary locations, situations,
and exposure scenarios. Aspects of simulation models for indoor air exposures
that need exploration and/or development include pollutant mixing and sourcereceptor proximity, general pollutant dynamics, flow rates, activity patterns, spatial and temporal resolution and extent, and the eco-social context of exposure.
Modern information technology, including microsensors and remote digital loggers, holds the promise of the more efficient collection of real-time exposure and
exposure-related data.
352
11.5.1 Proximity Effects
Close proximity to smoking individuals, and the elevated SHS exposure that may
result, have not been considered in this dissertation. In my modeling framework,
the maximum possible proximity to a smoker that can occur is when a receptor
occupies the same room as the smoker. All of the smoker’s emissions are assumed
to be immediately mixed throughout the room volume. This assumption is likely
to be fairly accurate, except for cases where the exposed person is very close to the
smoker. If one occupies the same room as a smoker and also is within approximately 1 m, then one’s localized exposure in time cannot be reliably estimated by
assuming a protective effect of rapid dilution in the room. Using a portable filtration device or opening a window in the room may reduce proximate exposure
to a degree, but the effect is likely to be different than for exposures at a farther
distance. On the other hand, it is possible that one’s average exposure over a sufficiently long period is not much different than the theoretical well-mixed case.
A careful investigation into the proximity effect for SHS exposure in the specific
case and emissions from an assortment of household products in the general case,
especially one that characterizes the distribution of exposure concentrations as a
function of distance and averaging time, is warranted.
11.5.2 Residential Pollutant Monitoring
More intensive air monitoring experiments in residential settings are necessary to
confirm and extend our understanding of how pollutants travel between rooms, to
and from household surfaces, and through HAC/HVAC systems. The rate of air
flow between zones of a house is an understudied area, especially as influenced
by interior door position, window positions, HAC/HVAC operation, and ambient
atmospheric conditions, e.g., indoor-outdoor temperature differences, and wind
intensity and direction. The interaction of semi-volatile compounds with surfaces,
which can lead to substantial indirect exposures, is another understudied area.
Controlled experiments should be designed to examine how recognized physical and chemical process affect exposure, and executed in a number of test houses
353
in which real-time measurements of particulate matter, tracer gases, and selected
reactive gases are made for different types of furnishings, house sizes, house layout and design, and weather. The houses should be representative of the range of
conditions that are expected to occur in the housing stock of a given population.
The degree of effort needed for each house restricts the number of homes that can
be studied. However, it is expected that experiments in only a relatively small
number of houses will be necessary to fully test and parameterize a mechanistic
indoor air model.
11.5.3 Residential Activity Patterns
Currently, there is a paucity of information available on human activity patterns
in a residential context. Two large-scale activity pattern studies conducted across
the US and in California have produced timelines of movement between specific
rooms of a house. The results of the nationwide survey have been used to develop the exposure simulation model in this dissertation (see Chapters 4 and 6).
However, no studies have appeared for a sizeable population that have collected
human activity pattern data simultaneously for all the members of a household or
across multiple days.
In this dissertation, I have not explicitly incorporated interaction amongst
household members into the simulation of SHS exposures. Random selection of
smoker-nonsmoker pairs resulted in considerable variation in the amount of time
the smoker and nonsmoker spent in the same room of the house. To fully understand how exposure to residential air pollutants, such as SHS, occurs, it is important to consider dependencies among members of a household, and possible
changes in activity patterns from day to day, perhaps in response to particular
exposure-relevant initiatives or changing source behaviors. The amount of time
that occupants spend together and in which rooms, and what activities are performed together or apart, coupled with the time-dependent nature of particular
pollutant generating patterns, e.g., smoking, and door, window, and HAC/HVAC
configurations, will all affect exposures. Careful consideration of relative move-
354
ments for occupants of different ages and relationships, e.g., child and caregiver,
would allow for a better understanding of how different demographic groups are
exposed.
Much of the activity pattern data collected to date consists of fairly crude location and activity categories. Future exposure studies should be focused on specific
types of exposure and measure as detailed information on exposure-related human
activities events as possible. The use of detailed micronvironment and behavior
categories results in a record of the micro-level behavior of human beings, which
can play a critical role in how exposure occurs and help to identify new and better
strategies for reducing or eliminating exposure. The collection of detailed information may be prohibitive for large studies, although the advent of sophisticated
electronic monitoring equipment may facilitate data gathering and management.
In the future, new human activity surveys should be performed for households on multiple days, perhaps in conjunction with residential pollutant monitoring studies, in which movement between rooms is precisely tracked for each
occupant, along with detailed information on the activities they perform in each
room. Particularly interesting would be an effort to study exactly how occupant
patterns change in response to public health interventions. Since human activities
are highly variable and involve complex interaction between persons, the number
of households worth of data will likely need to be considerably larger to fully encapsulate human activity dynamics for a given population than it would need to
be to understand and model arbitrary patterns of pollutant concentration.
11.5.4 Modeling Social Ecologies
Broadly, exposure science works to understand the mechanisms of exposure with
the hope of identifying effective means to reduce or eliminate it. As demonstrated
in this dissertation, the mechanism of exposure is intimately associated with human behavior dynamics. Doing small, simple, and inexpensive or zero-cost things
may go a long way in reducing exposure to SHS and other pollutants and safeguarding health. However, while strategies for reducing exposure may at times
355
appear to be straightforward from a technological or logistical perspective, there
may be sizeable hurdles to overcome in terms of house roles, personalities, habits,
and scheduling. The ecology of a household, consisting of multiple adults and
children of varying occupations, genders, ages, and personalities, is complex. For
public health intervention projects focused on households, as with urban planning or peace and nation-building efforts that have significantly larger scope, a
roadmap for improvement or recovery handed down from above is not enough,
and is likely to fail. The people who live in the households (or cities) must desire the change. They must understand the process, and be fully informed and
involved, as it progresses. The successful adoption of beneficial changes is three
pronged: (1) researchers must amass technical knowledge of potentially effective
solution paths; (2) researchers must gain an understanding of the interpersonal
relationships and dynamics of the population; and (3) researchers and members
of the target community must participate in the effective, and ongoing, discussion
and evaluation of technical knowledge and human factors.
For the case of residential SHS exposure, it is likely not enough that SHS has
known adverse health effects and that strategies such as isolating smokers, opening windows, or using filtration devices have been identified as effective means
of removing SHS-related indoor air pollutants. When this knowledge is fed into
the ecology of a smoking household by health care workers, the media, or other
elements of society, it may or may not have any lasting and beneficial effect. Nonlinear, feedback effects related to social pressure on smokers, interpersonal roles,
personal empowerment, and feelings of involvement in evaluating, identifying, or
implementing effective means of decreasing SHS exposure are likely to play important roles in the eventual reduction or elimination of exposure.
In the next logical phase in exposure science, physical models of pollutant dynamics should be fused with informed social models of human dynamics. New
developments in quantitative social science, including techniques of agent-based
modeling [Epstein and Axtell, 1996], show much promise as a means of understanding the complex, changing relationships in human ecologies. Some exposure
356
modelers are beginning to recognize that life-stage and life-role variables must be
included in studies of human activity to sufficiently explain and understand the
variation in human behavior that impacts exposure [Graham and McCurdy, 2004].
Exposure science would benefit by drawing liberally from fields such as cognitive
psychology, sociology, and geography, which are focused on human behavior and
the interactions amongst individuals and between individuals and the environment. Psychological, sociological, and economic forces act to provide context and
incentives for behavior inertia and modification. Incorporation of these factors
into theoretical descriptions of exposure will allow exposure assessors to study
how factors such as roles, empowerment, knowledge, perception, and beliefs contribute to a particular exposure landscape, and will facilitate the identification of
both physically and socially practical means for reducing or eliminating dangerous
exposures for a particular population.
11.6 References
Epstein, J. M. and Axtell, R. (1996). Growing Artificial Societies: Social Science From
The Bottom Up. Brookings Institution Press and MIT Press, Washington, D.C. and
Cambridge, MA.
Graham, S. E. and McCurdy, T. (2004). Developing meaningful cohorts for human
exposure models. Journal of Exposure Analysis and Environmental Epidemiology,
14(1): 23–43.
357
Part V
Appendices
358
The following four appendices contain supplementary information on
activity pattern data (Appendix A, page 359), the derivation of forms
for single and two-zone systems (Appendix B, page 372), an interactive program for estimating two-compartment model parameters (Appendix C, page 382), and a software package for simulating human exposure and accomplishing various tasks in exposure-related data analysis and research (Appendix D, page 387).
359
Appendix A
Raw Activity Pattern Data
This appendix provides a more detailed look at the 24-h diary component of
the 1992-94 National Human Activity Pattern Survey (NHAPS) database [Klepeis
et al., 1996, 2001; Tsang and Klepeis, 1996]. These diaries provide a minute-byminute account of the locations visited and activities engaged in by a representative sample of residents across the contiguous US. While Chapter 4 presents a
broad analysis of the resident-specific location data, which are used in this dissertation as the primary source of location information in simulated exposure to
residential secondhand tobacco smoke, here I describe the NHAPS data format for
the diaries and provide raw plots of time-location profiles.
A.1 Interview and Data Format
Each of the 9,386 persons interviewed as part of NHAPS reported the starting and
ending times of distinct microenvironments they visited on the day before they
were interviewed, starting and ending at midnight. They also reported whether
or not a smoker was present in each of these microenvironments. A microenvironment is defined as a unique combination of their location and the activity occurring in that location. Table A.1 contains an illustrative set of diary records for
a single individual with each record corresponding to a different microenvironment. This table contains codes for both the original activities and locations for
each microenvironment, as well as codes for reduced location and activity groups.
APPENDIX A. RAW ACTIVITY PATTERN DATA
360
Original codes for NHAPS locations and activities are given in Tables A.2 and A.3,
respectively. The reduced location codes are defined as follows: 10. indoors at
a residence; 20. outdoors at a residence; 30. in a vehicle; 40. near a vehicle; 50.
at some other outdoor location; 60. in an office or factory; 70. in a mall or other
store; 80. at school or in a public building; 90. at a bar or restaurant; and 100. at
some other indoor location. The reduced activities are: 00. an activity deemed to
be unrelated to exposure; 10. cooking or preparing food; 20. doing laundry, dishes
or cleaning the kitchen; 30. housekeeping; 40. bathing, showering, or using the
bathroom; 50. doing yardwork, gardening, or car or house maintenance; 60. doing
sports or exercise; and 70. eating or drinking.
A.2 Plots of 24-h Time-Location Profiles
To further explore the richness of available human activity pattern information
beyond the 24-h and hourly aggregate descriptions presented in Chapter 4, this
appendix presents raw data from the NHAPS study in the form of time-location
plots (see Figures A.1−A.5). These plots consist of vertically stacked strips, each
one corresponding to a single individual’s 24-h time-location profile. The strips are
divided into colored time segments where each color corresponds to a different
location at one’s own residence and white space indicates time spent outside or
away from home (see the legend in Figure A.1). In addition, the strips are stacked
bottom to top in order from the most time spent at home to the least time spent
at home, i.e., roughly 100% to 0% time spent in or around one’s home. Because
thousands or many hundreds of profiles are not discernible at the resolution of
most graphics devices, I reduced the number of profiles for each plot to a sample of
250 representative individuals. To create the sample, I chose every nth person from
the sorted list of individuals, where n was assigned so that the sampled individuals
spanned the entire range of total time spent at home. For example, for an original
population of 5,000 people, I would sample every 20th person. The original sample
sizes are given in the captions of each plot.
Figure A.2 shows a plot of time-location profiles that is representative of all
Time
0:00
1:45
2:00
11:00
11:05
11:15
11:25
11:30
11:37
13:37
13:44
13:54
13:57
15:30
15:33
16:30
17:00
19:00
19:10
19:25
19:35
21:00
No.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
24:00
19:35
21:00
19:10
19:25
17:00
19:00
15:33
16:30
13:57
15:30
13:54
13:37
13:44
11:30
11:37
11:15
11:25
11:00
11:05
1:45
2:00
Time
End
Studying
Travel related to shopping for food
Watching TV
Traveling to shopping
Shopping for food
Bathing or Showering
Watching TV
Traveling from bar
Watching TV
Traveling to bar
At bar
Preparing Meals or Snacks
Playing flag football
Traveling to home
Dressing or Personal Grooming
Traveling to play football
Preparing Meals or Snacks
Eating Meals or Snacks
Sleeping or Napping
Brushed teeth
At night club
Traveled home after night club
Summary
54
39
91
39
30
40
91
79
91
79
77
10
80
79
47
89
10
43
45
44
77
79
Activity
Detailed
00
00
00
00
00
40
00
00
00
00
00
10
60
00
00
00
10
70
00
40
00
00
Activity
Reduced
102
301
102
301
414
104
102
301
102
301
405
201
507
306
102
306
101
102
105
104
405
301
Location
Detailed
10
30
10
30
90
10
10
30
10
30
90
10
50
40
10
40
10
10
10
10
90
30
Location
Reduced
NO
NO
NO
NO
NO
NO
NO
NO
NO
NO
YES
NO
NO
NO
NO
NO
NO
NO
NO
NO
YES
NO
Present?
Smoker
180
10
85
10
15
30
120
3
57
3
93
10
120
7
5
7
10
10
540
5
105
15
[min]
Time Spent
a The respondent whose diary is shown in this table was an Hispanic male from Connecticut between the ages of 18 and 24 who was
interviewed on a weekend in the fall.
Start
Micro
Table A.1: Example 24-Hour Recall Diary Containing Beginning & Ending Times, Activity, Location, Presence of a
Smoker, and Time Spent for 22 Microenvironments Visited on the Diary Daya
361
362
Table A.2: The Original NHAPS 24-h Recall Diary Locations
Loc.
Code Description
OWN HOUSE
100
OTHER, HOME
101
HOME KITCHEN
102
HOME LIVING RM/FAMILY RM/DEN
103
HOME DINING ROOM
104
HOME BATHROOM
105
HOME BEDROOM
106
HOME STUDY/OFFICE
107
HOME GARAGE
108
HOME BASEMENT
110
HOME UTILITY RM/LAUNDRY RM
111
HOME POOL, SPA(OUTDOORS)
112
HOME YARD, OTHER OUTSIDE HOUSE
113
HOME MOVING FROM ROOM TO ROOM
114
HOME MOVING IN/OUT OF THE HOUSE
120
OTHER VERIFIED
199
REFUSED TO ANSWER
FRIEND’S/OTHERS’ HOUSE
200
OTHER, OTHER’S HOUSE
201
OTHER’S KITCHEN
202
OTHER’S LIVING RM/FAMILY RM/DEN
203
OTHER’S DINING ROOM
204
OTHER’S BATHROOM
205
OTHER’S BEDROOM
206
OTHER’S STUDY/OFFICE
207
OTHER’S GARAGE
208
OTHER’S BASEMENT
210
OTHER’S UTILITY RM/LAUNDRY RM
211
OTHER’S POOL, SPA(OUTDOORS)
212
OTHER’S YARD, OTHER OUTSIDE HOUSE
213
OTHER’S - MOVING FROM ROOM TO ROOM
214
OTHER’S - MOVING IN/OUT OF THE HOUSE
220
OTHER VERIFIED
299
REFUSED
TRAVELING
300
OTHER, TRAVEL
301
CAR
302
TRUCK (PICK-UP/VAN)
303
TRUCK (OTHER)
304
MOTORCYCLE/MOPED/SCOOTER
305
BUS
306
WALKING
307
BICYCLE/SKATEBOARD/ROLLER-SKATES
308
IN A STROLLER/CARRIED BY AN ADULT
310
TRAIN/SUBWAY/RAPID TRANSIT
Loc.
Code Description
311
AIRPLANE
312
BOAT
313
WAITING FOR BUS, TRAIN, RIDE (AT STOP)
314
WAITING FOR TRAVEL, INDOORS
320
OTHER VERIFIED
399
REFUSED
OTHER INDOOR
400
OTHER, INDOOR
401
OFFICE BLDG/BANK/POST OFFICE
402
PLANT/FACTORY/WAREHOUSE
403
GROCERY STORE/CONVENIENCE STORE
404
SHOPPING MALL/NON-GROCERY STORE
405
BAR/NIGHT CLUB/BOWLING ALLEY
406
AUTO REPAIR SHOP/GAS STATION
407
INDOOR GYM/SPORTS OR HEALTH CLUB
408
PUBLIC BLDG./LIB./MUSEUM/ THEATER
409
LAUNDROMAT
410
HOSPITAL/HEALTH CARE/DOCTOR
411
BEAUTY PARLOR/BARBER/HAIR
412
AT WORK/NO SPECIFIC MAIN LOCATION
413
SCHOOL
414
RESTAURANT
415
CHURCH
416
HOTEL/MOTEL
417
DRY CLEANER
418
OTHER REPAIR SHOP
419
INDOOR PARKING GARAGE
420
OTHER VERIFIED
499
REFUSED
OTHER OUTDOOR
500
OTHER OUTDOOR
501
SIDEWALK/STREET/NEIGHBORHOOD
502
PARKING LOT
503
SERVICE STATION/GAS STATION
504
CONSTRUCTION SITE
505
SCHOOL GROUNDS/PLAYGROUND
506
SPORTS STADIUM
507
PARK/GOLF COURSE
508
POOL, RIVER, LAKE
510
RESTAURANT/PICNIC (OUTDOORS)
511
FARM
520
OTHER VERIFIED
599
REFUSED
Table A.3: The Original NHAPS 24-h Recall Diary Activities
Act. Code Description
NON-FREE TIME
Paid Work
01
MAIN JOB
02
UNEMPLOYMENT
03
TRAVEL DURING WORK
05
SECOND JOB
08
BREAKS
09
TRAVEL TO/FROM WORK
Household Work
10
FOOD PREPARATION
11
FOOD CLEANUP
12
CLEANING HOUSE
13
OUTDOOR CLEANING
14
CLOTHES CARE
15
CAR REPAIR/MAINTENANCE
16
OTHER REPAIRS
17
PLANT CARE
18
ANIMAL CARE
19
OTHER HOUSEHOLD WORK
Child Care
20
BABY CARE
21
CHILD CARE
22
HELPING/TEACHING
23
TALKING/READING
24
INDOOR PLAYING
25
OUTDOOR PLAYING
26
MEDICAL CARE-CHILD
27
CHILD CARE
28
DRY CLEANING
29
TRAVEL, CHILDCARE
Obtaining Goods, Services
30
SHOPPING FOR FOOD
31
SHOPPING FOR CLOTHES HH ITEMS
32
PERSONAL CARE SERVICES
33
MEDICAL APPOINTMENTS
34
GOVT/FINANCIAL SERVICES
35
CAR REPAIR SERVICES
36
OTHER REPAIR SERVICES
37
OTHER SERVICES
38
ERRANDS
39
TRAVEL, GOODS AND SERVICES
Personal Needs and Care
40
WASHING, ETC
41
MEDICAL CARE
42
HELP AND CARE
43
EATING
44
PERSONAL HYGIENE
45
SLEEPING/NAPPING
47
DRESSING, ETC
48
NA ACTIVITIES
49
TRAVEL, PERSONAL CARE
Act. Code Description
FREE TIME
Educational
50
ATTENDING FULL TIME SCHOOL
51
OTHER CLASSES
54
HOMEWORK
55
USING LIBRARY
56
OTHER EDUCATION
59
OTHER TRAVEL, EDUCATION
Organizational
60
PROFESSIONAL UNION
61
SPECIAL INTEREST
62
POLITICAL/CIVIC
63
VOLUNTEER HELPING
64
RELIGIOUS GROUPS
65
RELIGIOUS PRICES
66
FRATERNAL
67
CHILD/YOUTH/FAMILY
68
OTHER ORGANIZATION
69
TRAVEL ORGANIZATIONAL
Entertainment/Social
70
SPORTS EVENT
71
ENTERTAINMENT
72
MOVIES/VIDEOS
73
THEATER
74
MUSEUMS
75
VISITING
76
PARTIES
77
BARS/LOUNGES
78
OTHER SOCIAL
79
TRAVEL, SOCIAL
Recreation
80
ACTIVE SPORTS
81
OUTDOOR RECREATION
82
EXERCISE
83
HOBBIES
84
DOMESTIC CRAFTS
85
ART
86
MUSIC/DRAMA/DANCE
87
GAMES
88
COMPUTER USE
89
TRAVEL, RECREATION
Communications
90
RADIO
91
TV
92
RECORDS/TAPES
93
READING BOOKS
94
MAGAZINES, ETC
95
READING NEWSPAPER
96
CONVERSATIONS
97
LETTERS, WRITING PAPERWORK
98
THINKING/RELAXING
99
TRAVEL RELATED PASSIVE LEISURE
363
364
NHAPS respondents who live in detached homes across the contiguous US (original sample size of 5,895). The first immediately obvious characteristic is that time
spent in the bedroom is dominant (shown in blue) with 7 AM as the approximate
central tendency for wake-up times regardless of total time spent at home. Scatter
around this central value appears to be fairly small – in the neighborhood of 1 h
or less for most respondents. The central tendency for bedtime is approximately
11 PM with a similarly small variation. This time is largely invariant with respect
to total time spent at home. The second most dominantly occupied room in the
house is the living room (shown in yellow). The time where the most people in the
US are in their living rooms is approximately 8:30 PM regardless of the total time
individuals spent at home. The variance for this time is somewhat larger than for
wake-up and bedtimes with an apparent range of plus or minus 2 h. The sorting
of the location profiles allows us to see immediately that only about 3% of the respondents did not spend any of the morning or the day at home with 1−2% not
spending any time inside their home at all. Also, about half of respondents spent a
great deal of time during midday outside of their home. The 50% of the population
spending the most time at home during the day had no discernible pattern in their
location, when arranged only by total time spent at home, with different amounts
of time spent moving about the house (magenta), in the kitchen (green), or in other
rooms.
Figure A.5 shows time-location profile plots for different age groups, revealing
overall patterns in the behavior of young children (under age 5), who are likely
to take a nap in the bedroom at midday, and older respondents (over age 65) who
spend more time in living areas of the home during the day than working-age
adults (ages 35−65). The first panel in Figure A.5 shows that many children under
5 are in their bedrooms between approximately 12 noon and 3 PM. Children between the ages of 5 and 18, shown in the next two panels, tend to spend less time
at home between the hours of 9 AM and 3 PM than do younger children. Children
under age 12 are typically in their bedroom by about 9 PM, whereas older children
retire slightly later and get up earlier. The 30% of children aged 12−18 who spent
365
Residential Locations Visited
Kitchen
Living Room, Family Room, Den
Dining Room
Bathroom
Bedroom
Study, Office
Garage
Basement
Utility Room, Laundry Room
Pool/Spa (Outdoors)
Yard/Other Outside House
Moving From Room to Room
Moving In and Out of House
Other Verified
Refused to Answer
Figure A.1: Legend for plots of the raw time-location profiles for NHAPS respondents presented in Figures A.2−A.4.
366
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.0
0.1
0.9
1.0
All Respondents in Detached Homes
Mid
3AM
6AM
9AM
Noon
3PM
6PM
9PM
Mid
Time of Day
Figure A.2: A time-location plot showing the location time series for a sample of
250 NHAPS respondents living in detached homes (original sample size: 5,895).
The event time series for the sample are represented by vertically stacked time
strips that have been sorted by the total time each person spent at home, where
different colors correspond to times when an individual was reported to occupy
a particular house location. White space corresponds to time when an individual
was reported to be inside a location other than their own home or outside their
home. The horizontal axis stretches across a single 24-h period, starting and ending
at midnight.
367
the most time at home get up about an hour or more later than other children of
the same age, though they appear to generally go to bed at the same time. Adults
aged 18−65 are distinguished from younger age groups in that many of these respondents are at home in the kitchen at about 6 PM, spend most of their time in
the living room between 7 and 10 PM, and go to bed at 10 or 11 PM. Older adults
over the age of 65 have a similar pattern to the 18−65 age group, except a larger
proportion spend time in the living room during evening hours as well as during
midday between 9 and 5 PM.
The patterns of time-location profiles for working-age (18−64) male and female respondents do not display as much overall difference as is evident between
respondents of different ages. However, as shown in Figure A.3, there is a distinctly larger number of female respondents than male respondents in the kitchen
between 5 PM and 7 PM. More female respondents are also recorded moving about
the house between the hours of 7 AM and noon. In contrast, more male respondents are out of the house between the hours of 9 AM and 5 PM. There seems to be
a trend in male respondents, who spend more total time at home, spending more
time in the bedroom, whereas this trend is not apparent for females.
Working-age respondents (18−64) who reported their time-location profiles on
weekends spent more time at home between the hours of 9 AM and 5 PM than
respondents giving weekday accounts (see Figure A.4). There is a somewhat larger
proportion of people spending almost no time at home at all on weekends. The
profiles also show that respondents tended to sleep 1−2 h later on weekends than
on weekdays, although the time they went to bed appears to be approximately
unchanged.
368
Females, Ages 18−64
Males, Ages 18−64
6AM
Noon
6PM
Mid
0.7
0.6
0.5
0.4
0.3
0.0
0.1
0.2
0.8
0.9
1.0
Mid
Mid
6AM
Noon
6PM
Mid
Time of Day
Figure A.3: A time-location plot showing the location time series for samples of 250
male and 250 female NHAPS respondents aged 18−64 living in detached homes
(original sample sizes: 1,960 females and 1,759 males). See the Figure A.2 caption
and the text for more information on plot construction.
369
Weekdays, Ages 18−64
Weekends, Ages 18−64
6AM
Noon
6PM
Mid
0.7
0.6
0.5
0.4
0.3
0.0
0.1
0.2
0.8
0.9
1.0
Mid
Mid
6AM
Noon
6PM
Mid
Time of Day
Figure A.4: A time-location plot showing the location time series for two samples
of 250 NHAPS respondents, interviewed either on weekends or weekdays, aged
18−64 and living in detached homes (original sample sizes: 2,513 weekday interviews and 1,206 weekend interviews). See the Figure A.2 caption and the text for
more information on plot construction.
370
Ages Under 5
Ages 5−12
6AM Noon
6PM
Ages 12−18
Mid
0.4
0.2
0.0
Mid
6AM Noon
6PM
Mid
Ages 18−65
6AM Noon
6PM
Mid
Ages 65+
Mid
6AM Noon
6PM
Mid
0.8
1.0
Mid
0.0
0.2
0.4
0.6
0.6
0.8
1.0
Mid
Mid
6AM Noon
6PM
Mid
Time of Day
Figure A.5: Time-location plots showing the location time series for samples of
250 NHAPS respondents of different ages living in detached homes (original sample sizes: 321 respondents aged 0−5 years; 499 respondents aged 5−12 years; 447
respondents aged 12−18 years; 3,719 respondents aged 18−65 years; and 909 respondents aged over 65 years). See the Figure A.2 caption and the text for more
information on plot construction.
371
A.3 References
11(3): 231–252.
Klepeis, N. E., Tsang, A. M., and Behar, J. V. (1996). Analysis of the National Human
Activity Pattern Survey (NHAPS) Responses from a Standpoint of Exposure Assessment. EPA/600/R-96/074, US EPA, Washington D. C.
Tsang, A. M. and Klepeis, N. E. (1996). Descriptive Statistics Tables from a Detailed Analysis of the National Human Activity Pattern Survey (NHAPS) Data.
EPA/600/R-96/148, US EPA, Washington D. C.
372
Appendix B
Model Equations for Single- and
Multi-Compartment Systems
The central tool for the research presented in this dissertation is a simulation model
used to predict inhalation exposures for pollutants in residential secondhand tobacco smoke (SHS). The simplest approach to simulate exposure would be to assume that a single well-mixed zone accurately describes SHS exposure concentrations in a home. However, as demonstrated in this dissertation, it is likely that a
careful accounting of the movement of people and pollutants amongst different,
distinct zones is important to understanding SHS exposure in residences. Such a
treatment allows for dynamic interzonal air flow rates and room-specific emission
rates. Therefore, my simulation model uses a sophisticated multi-compartment
indoor air quality model, which provides quantitative real-time estimates of air
and surface pollutant concentrations for different rooms, and incorporates timevarying values for all observable parameter inputs.
Below, I present dynamic and time-averaged solutions for the simple singlezone model, which is typically used as a first-order approximation for residential
pollutants. I then introduce the governing equations for a general multiple compartment system with specific application to residential indoor air quality, and I
describe a technique for numerically solving the system, obtaining both dynamic
and time-averaged solutions.
APPENDIX B. COMPARTMENT MODEL EQUATIONS
373
B.1 Single-Zone Model
The governing mass balance equation for pollutants entering and leaving a single
enclosed zone due to cigarette emissions, ventilation, and other first-order loss,
such as by particle surface deposition, is as follows:
dy(t)V
= −VAy(t) − VDy(t) + n(t)e
dt
(B.1)
where y(t) is the airborne concentration of the pollutant at time t [µ g m−3 ], V is
the volume of the zone [m3 ], A is the exchange rate with air from outside of the
zone [h−1 ], D is the rate of loss due to other first-order processes [h−1 ], typically
particle deposition, n(t) is the number of active cigarettes at time t, and e is the
emission rate for a single cigarette [µ g cig−1 h−1 ]. This equation assumes that air
from outside of the zone is pollutant free.
The time-averaged concentration in the zone is obtained by first integrating
Equation B.1 between times t1 and t2 , and dividing by the total time period T =
t2 − t1 :
δy
e
+ ( A + D) y = n
(B.2)
T
V
where y is the zonal time-averaged concentration for the period T, n is the mean
number of cigarettes being smoked at any time during this interval, and δ y is the
change in concentration between t1 and t2 . The volume, removal constants, and
cigarette emission rate are considered to be constant over the averaging time interval T. The first term in Equation B.2 can be neglected for long averaging times
and/or episodes where the concentration returns to the initial concentration. Also,
n can be expressed as
Nτ
T ,
where N is the number of cigarettes smoked over time
interval T, and τ is the duration of a single cigarette. Thus, Equation B.2 can be
solved for y as follows:
y = Ne
τ
1
T V ( A + D)
(B.3)
By setting E = eτ , where E is the mass emissions per cigarette [µ g cig−1 ], switching
units from hours to days, and considering a time period T of a single day, we obtain
374
a simplified result:
y=
where Ñ =
N
T
ÑE
V ( A + D)
(B.4)
is the number of cigarettes smoked in a day [cig d−1 ] and D and A
are in units of d−1 . When applying this equation to daily exposures occurring in a
house, a correction factor f may be used, which accounts for nonideal dispersion
within the zone, so that a modeled occupant can experience lower- or higher-thanaverage concentrations, and for time spent by the occupant outside of the home,
which may correlate in time with higher or lower SHS species concentrations. So,
the final equation for estimating the time-averaged exposure concentration to airborne cigarette emissions in a single-zone structure is:
y= f
ÑE
V ( A + D)
(B.5)
Both Nazaroff and Singer [2004] and Nazaroff and Klepeis [2004] have derived
closely related forms. The total pollutant mass intake rate is the product of the
average pollutant exposure concentration, y, multiplied by the inhalation rate Q
[m3 d−1 ], which is assumed to be constant over the course of the day. The daily
pollutant mass intake rate, I, can be written as:
I = fQ
ÑE
V ( A + D)
(B.6)
where I has units of µ g d−1 .
B.2 Generic First-Order Compartmental Systems
A generic ordinary differential equation (ODE) solver can solve a system of n firstorder ordinary differential equations of the form:
dyi (t)
= f i t, y1 (t), y2 (t), · · · , yn (t)
dt
(B.7)
where i = 1, 2, 3, . . . , n, and yi (t) are response, or dependent, variables that take
on specific values for each value of the independent variable t, which is usually
375
considered to be time. The solution consists of n time series, one for each response
variable yi (t). A typical problem contains of coupled response variables where
response variables influence each other in terms of simple linear combinations.
For this case, the system is one of n coupled linear equations with constant coefficients. Each equation is conceived as corresponding to a particular location,
compartment, or some other kind of distinct region or conceptual domain:
dy1 (t)
= k10 + k11 y1 (t) + k12 y2 (t) + · · · + k1n yn (t)
dt
dy2 (t)
= k20 + k21 y1 (t) + k22 y2 (t) + · · · + k2n yn (t)
dt
···
(B.8)
dyn (t)
= kn0 + kn1 y1 (t) + kn2 y2 (t) + · · · + knn yn (t)
dt
where yi , i = 1, 2, 3, . . . , n, are the response variables for each compartment, such
as mass or mass concentration, and ki j , i = 1, 2, 3, . . .,n; j = 0, 1, 2, 3, . . .,n, are
constant coefficients corresponding to rates of gain (positive) or loss (negative)
to/from the other compartments, which can be written in terms of observable parameters (see below). The ki j are strictly constant within any given time step, but
by solving the system individually for short consecutive time steps, and using the
final responses in one time step as the initial responses in the following time step,
an approximate solution can be obtained for parameters that vary in time.
One can solve the n-dimensional system in each time step for an arbitrary number of equations and for arbitrary coefficients ki j , across any parameter configuration, by making use of standardized Runge-Kutta-type routines, such as those
available in the GNU Scientific Library (GSL) [Galassi et al., 2003]. If, for example,
the observable parameters, and therefore the coefficients ki j , are supplied every
minute, then the GSL routine adaptively selects the best time increment to use
within each minute time step to evaluate the solution at the end of that step. If
the system responses for each compartment do not change rapidly over a minute,
then they can be used to accurately represent the state of the system during each
minute time interval. If this is not the case, then the calculation of time-averaged
response values may be more appropriate.
376
To obtain time-averaged concentrations for a particular time step, a system of
n linear algebraic equations is obtained by taking the integral of the above set of
differential equations with constant coefficients and dividing by the duration of
the time step:
A11 C1 + A12 C2 + · · · + A1n Cn = B1
A21 C1 + A22 C2 + · · · + A2n Cn = B2
···
(B.9)
An1 C1 + A2n C2 + · · · + Ann Cn = Bn
where Ci , i = 1, 2, 3, . . . , n, are the average concentrations in each compartment
i for the current time step. The coefficients A and constants B are obtained from
the constant ODE coefficients and instantaneous concentrations for the dynamic
solution to Equation B.8. The average concentrations Ci can then be determined
by solving the linear system using, for example, LU decomposition1 [Galassi et al.,
2003].
B.3 Multi-Compartment Indoor Air Quality Model
For a multi-compartment indoor air quality problem, the coefficients in Equation B.8 depend on the physical quantities listed in Table B.1, which include zone
volumes, zone surface-to-volume ratios, air flow rates, particle deposition rates,
chemical sorption and desorption rates, mass emission rates, particle filtration efficiencies, and building particle penetration efficiences. Nazaroff and Cass [1989]
introduce mathematical forms relating changes in aerosol concentrations to physical parameters. Neglecting different particle sizes and components, each compartment equation i, corresponding to a particular row in Equation B.8, has the
1 LU decomposition is a procedure for factoring an NxN matrix into a lower triangular matrix
(L) and an upper triangular matrix (U).
377
following form:
m,h6=i
dyi (t)
Ei
f hi yh m,h6=i f ih yi
pi f mi
=
+
y m − βi yi + ∑
− ∑
dt
Vi
Vi
Vi
Vi
h= 1
h= 1
m
f
f xi ∑h=1 (1 − ηhx ) f hx yh
η f
− ix yi − ii ii yi (B.10)
+
m
Vi
Vi
Vi
∑h=1 f hx
where n is the total number of compartments, index m = n + 1 corresponds to
the outdoors, and index x corresponds to the HAC/HVAC system. The terms
correspond in order to particle mass emissions into compartment i, infiltration of
outdoor particles into compartment i, removal of particles in compartment i by
surface deposition, cross flow from other compartments into compartment i, cross
flow from compartment i into other compartments, flow from the HAC/HVAC
system into compartment i, including loss to duct-based filtration, flow from compartment i into the HAC/HVAC system, and removal of particles through local
filtration.
Grouping terms by particle concentrations in compartment i, concentrations in
other compartments h = j where j 6= i, the outdoor concentration h = m = n + 1,
and stand-alone terms, Equation B.10 can be rewritten as follows:
(
)
m,h6=i
f ih
dyi (t)
f ix ηii f ii
f xi (1 − ηix ) f ix
= yi (t) −βi − ∑
−
−
+
dt
Vi
Vi
Vi
Vi
∑m
h= 1 f hx
h= 1
#
"
m, j6=i
f ji
f xi (1 − η jx ) f jx
+ ∑ y j (t)
+
V
Vi
∑m
i
h= 1 f hx
j=1
(
)
pi f mi
f xi (1 − ηmx ) f mx
+ ym
+
Vi
Vi
∑m
h= 1 f hx
+
(B.11)
Ei
Vi
Hence, the ki j coefficients in Equation B.8 can be written in terms of the physical
378
Table B.1: Response Variables and Observable Physical Input Parameters of a
Multi-Compartment Indoor Air Quality Model for Airborne Particulate Matter
Symbol
a Units
yi ( t )
µ g m− 3
Airborne particle concentration in room i at time t
ym
µ g m− 3
Outdoor airborne particle concentration
pi
−
f ii
m3 h− 1
Recirculating flow through a filtration device in room i
fi j
m3 h− 1
Air flow from room i to room j
f ji
m3 h− 1
Air flow from room j to room i
f mi
m3 h− 1
Air flow from the outdoors to room i via infiltration and
natural ventilation
f im
m3 h− 1
Air flow from room i to the outdoors via exfiltration and
natural ventilation
f xi
m3 h− 1
Air flow from HAC/HVAC to room i
f ix
m3 h− 1
Air flow from room i to HAC/HVAC
f mx
m3 h− 1
Air flow from outdoors to HAC/HVAC
f xm
m3 h− 1
Air flow from HAC/HVAC to the outdoors
βi
h− 1
Particle deposition loss-rate coefficient
Ei
µ g h− 1
Particle mass emission rate for room i
Vi
m3
Volume of room i
ηii
−
Removal efficiency for filtration device in room i
ηix
−
Removal efficiency for HAC/HVAC return filtration for
room i
ηmx
−
Removal efficiency for outdoor to HAC/HVAC filtration
a The
Name
Particle penetration efficiency from outdoors to indoors in
room i considering infiltration and natural ventilation
units given here form a consistent set. Efficiencies are dimensionless quantities between 0
and 1.
379
parameters for a multi-compartment indoor air quality system:
m,h6=i
f ih
f ix ηii f ii
f xi (1 − ηix ) f ix
kii = − βi − ∑
−
−
+
Vi
Vi
Vi
Vi
∑m
h= 1 f hx
h= 1
f ji
(1 − η jx ) f jx
f
ki j =
+ xi
Vi
Vi
∑m
h= 1 f hx
(
)
f
(1 − ηmx ) f mx
pi f mi
E
+ xi
ki0 = ym
+ i
m
Vi
Vi
Vi
∑h=1 f hx
(B.12)
where i = 1 . . . n and j = 1 . . . n, j 6= i. Using these assignments, and by varying
each physical parameter value between minute-long time steps, a system can be
solved where air flow rates, emissions, and other parameter values vary across
time for each compartment. I use this approach in the current research to study
the exposure of household residents to secondhand tobacco smoke when they visit
various rooms in a house as pollutant emissions, and door, window, and HVAC
configurations change arbitrarily over time.
To treat other chemical species besides particles, which sorb onto room surfaces
and potentially desorb back into the air, a minimum of 2n coupled compartment
equations are required, n equations for air compartments and n additional surface
compartments for each room. Because air concentrations are expressed in mass per
volume and surface concentrations are expressed in mass per area, it is convenient
to write the response variables in terms of the mass present in each compartment.
In this approach, the air compartment concentrations in Equations B.11 and B.12
are replaced by ỹi (t) = yi (t)Vi , where ỹi (t) is in mass units of µ g.
The simplest model for sorption/desorption processes is one that assumes linear rates in either direction. Such a model has been used to accurately predict
air concentrations of semi-volatile compounds [Van Loy et al., 1997, 2001]. In this
model, mass is added to the surface compartment and simultaneously subtracted
from air compartments through a linear sorption term, νi VSi ỹi (t). Thus, the k(i +n)i
coefficient is
νi VSii
i
and the portion of the kii for irreversible loss to surfaces, −βi , is
replaced with −νi VSi . Here, the parameter ν is the sorption coefficient in units of m
h− 1 ,
i
Si is the surface area in room i, and
Si
Vi
is the corresponding surface-to-volume
380
ratio for room i in units of m−1 .
The model predicts that pollutant mass sorbed into surface compartments is
reemitted into the air compartments at a rate of ξi zi (t), where ξi is the desorption
coefficient for room i in units of h−1 , and zi (t) is the mass in the surface compartment for room i at time t in units of µ g. The mass leaves the surface compartment
at the same rate that it enters the corresponding air compartment, so the ki (i +n)
coefficient is ξi and the k(i +n)(i +n) coefficient is −ξi .
Besides k(i +n)(i +n) and k(i +n)i , all coefficients for the surface compartment
equations are equal to zero. Thus, the second set of n equations for surface compartments in the 2n-compartment system have the following form:
dzi (t)
S
= νi i ỹi (t) − ξi zi (t)
dt
Vi
(B.13)
where i = 1, 2, 3, . . . , n.
B.4 References
Galassi, M., Davies, J., Theiler, J., Gough, B., Jungman, G., Booth, M., and Rossi, F.
(2003). GNU Scientific Library Reference Manual - Second Edition, Software Version
1.3. Network Theory, Ltd., Bristol, UK, http://www.network-theory.co.uk/.
Nazaroff, W. W. and Cass, G. R. (1989). Mathematical modeling of indoor aerosol
dynamics. Environmental Science & Technology, 23(2): 157–166.
Nazaroff, W. W. and Klepeis, N. E. (2004). Environmental tobacco smoke particles. In Morawska, L. and Salthammer, T., editors, Indoor Environment: Airborne
Particles and Settled Dust, pages 245–274, Weinheim, Germany. Wiley-VCH.
Van Loy, M. D., Lee, V. C., Gundel, L. A., Daisey, J. M., Sextro, R. G., and Nazaroff,
W. W. (1997). Dynamic behavior of semivolatile organic compounds in indoor
air. 1. Nicotine in a stainless steel chamber. Environmental Science and Technology,
31(9): 2554–2561.
381
Van Loy, M. D., Riley, W. J., Daisey, J. M., and Nazaroff, W. W. (2001). Dynamic behavior of semivolatile organic compounds in indoor air. 2. Nicotine and phenanthrene with carpet and wallboard. Environmental Science & Technology, 35(3):
560–567.
382
Appendix C
An Interactive Computer Program for
Estimating the Parameters of a
Two-Compartment Model
I have written a graphical computer program that interactively plots air pollutant
concentrations in two zones based on the configuration of a set of 15 slider controls. Each slider corresponds to a single model input parameter. This program is
intended to be used in the visualization of two-compartment air pollutant concentration profiles, exploration of the sensitivity of concentrations to physical parameters, and the estimation of air flow rates between building rooms. I have used the
program, along with tracer gas concentrations measured in a house, to determine
the inter-room air flow rates by varying input parameter values until an optimum
fit is achieved between theoretical and observed concentrations based on subjective visual evaluation or by least squares or least absolute difference. Figure C.1
is a screen shot of an active session from such an analysis, consisting of the main
control panel, a plot of modeled and observed concentration time series, and a
file of observed data inside the window of a text editor. The results of the fitting
procedure are given in Chapter 5 of this dissertation.
The computational engine of the program is a subroutine written in Fortran,
which calculates concentration profiles using an analytical solution to a twocompartment dynamic system. The solution is obtained using a software package
APPENDIX C. INTERACTIVE TWO-COMPARTMENT COMPUTER PROGRAM
383
Figure C.1: A screen shot showing three of the computer program’s windows,
including the main window with slider controls for each model input parameter,
columns of observed data inside of a text editor, and a plot of the observed data
superimposed with model predictions, which are based on the current parameter
configuration.
for symbolic mathematics.1 The basic model equations, their solution, and the estimation of model parameters have been described by Miller et al. [1997] and Ott
et al. [2003].
C.1 User Interface
The program’s graphical front-end is written in the Perl programming language
and makes use of the Perl/Tk user-interface library.2 The user moves 15 individual
sliders up and down to adjust the values of the corresponding model parameters.
After a slider movement, a plot of the modeled concentrations and specified empirical concentrations are immediately updated to provide instant feedback on the
1 MathematicaTM ,
2 See
version 3.0 by Wolfram Research; see http://www.wolfram.com.
http://www.perl.org or http://www.perl.com.
384
influence of the corresponding parameter on levels in each compartment. As evident from the detailed view of the program’s main window, shown in Figure C.2,
the parameters include room volumes [m3 ], room air-exchange rates [h−1 ], the percentage of total inflow that each room receives from the adjacent room, the source
emission rate for each room [µ g min−1 ], the duration of the source in each room
[min], the initial concentration in each room [µ g m−3 ], the outdoor concentration
[µ g m−3 ], the time resolution (interval between concentration values) [min], and
the total time period of the simulation [min]. The modeled time series is assumed
to begin at time t = 0.
In addition to purely interactive adjustment of parameter values and manual,
or “eyeballed”, estimation of parameters giving the best fit to observed data, the
software also has the facility to perform a simple grid search of the parameter space
using mean-squared or absolute deviations of predicted and observed values as
the response metric. A data file containing observed concentrations in each room
is specified by entering the filename into the textbox at the bottom of the window.
If no filename is specified, then only modeled (theoretical) concentrations will be
plotted.
The automatic optimation method allows the user to specify as many parameters as desired for optimization. As initial and outdoor concentrations, volumes,
time parameters, and emission rates are usually predetermined for a given experiment, the user typically can achieve a first guess for the parameters by manually
adjusting the four flow parameters and then fine tuning them with the automatic
grid search.
The dialog box shown in Figure C.3 contains options that can be set for each
parameter. In addition to graphical labels, the user can set a conversion factor, the
minimum and maximum parameter values that are allowed, and the small and
large increments to be used in the automatic fitting procedure.
Options for the observed concentrations in the two different rooms are set using
the lefthand dialog box shown in Figure C.4. Here, indices for the time and concentration column in the specified data file are given, as well as the starting place
385
Figure C.2: The main window where parameter values for the two compartment
model can be changed interactively using sliders.
in the file, the number of records that are skipped for use in plotting a sample of
all measurements, and offset and multiplicative factors for the time and concentration columns. In addition, the user can specify the range of time values that will
be used in the automatic fitting procedure described above.
On the righthand side of Figure C.4 is shown a dialog box with options for
plotting concentrations and a three-dimensional (3D) plot of the response versus
two arbitrary parameters. The X and Y ranges can be specified along with the size
of plotted points (observations), the width of lines (model), and the 3D viewpoint,
which is determined by X and Y rotations and elevation above the XY plane.
C.2 References
386
Figure C.3: The parameter window where labels and ranges for each model parameter can be specified.
Figure C.4: The data window where the observed data can be selected and manipulated (left) and the plot window where characteristics of plots can be specified
(right).
387
Appendix D
A Software Package for Conducting
Human Exposure Research
The simulation and analysis described in this dissertation was accomplished using
an original software package for human exposure research1 that I implemented in
a freely available statistical programming environment called “R” [The R Development Core Team, 2003; Venables et al., 2002; Ihaka and Gentleman, 1996].2 Some
external source code was written in C for computationally intensive tasks, such as
obtaining the numeric solution of a generic system of multiple coupled compartments (see Appendix B), and dynamically loaded when needed. An existing C
numerical library called the GNU Scientific Library (GSL) was used [Galassi et al.,
2003].
All package components, together with a short description of their contents,
are listed in Table D.1. The software package consists of five subpackages for
tasks related to generic exposure simulation, indoor air quality, human activity
pattern analysis, human inhalation, and miscellaneous utility routines. An additional three subpackages contain raw activity pattern data, air quality monitoring
data, and exposure monitoring survey data. Finally, a custom subpackage call
“ResSmoke” was written, which makes use of the other subpackages, to conduct
simulation experiments for secondhand smoke exposure occurring in multi-zone
1 See
http://exposurescience.org/her.html.
binaries can be downloaded from http://cran.r-project.org for WindowsTM , MacintoshTM ,
and LinuxTM platforms. General information on R is available from http://www.r-project.org.
2R
APPENDIX D. SOFTWARE PACKAGE FOR HUMAN EXPOSURE RESEARCH
388
residences. This final package implements the model design described in Chapter 6 and was used for all simulations performed in Chapters 7, 8, and 9.
D.1 The ESM Function
The central routine in the simulation subpackage is the exposure simulation model
(ESM) function, which controls the passing of data and function specifications between the elements of a particular simulation problem. This function takes three
component functions for input, one that prepares input for each simulated individual, one that calculates individual exposure, and one that prepares the final simulation output format for each individual from the raw simulation output data. Each
of these functions can call, in turn, any number of “helper” functions, which use
simulation data generated by previously called functions. Other inputs into the
ESM function determine whether or not individual characteristics will be stratified according to a prescribed probability scheme, the time period of the simulation, and the number of times the simulation will be repeated for the population,
i.e., cycled. Particular aspects of the simulation, such as the nature of data inputs
(e.g., single value, probability model, or empirical distribution) and the grouping
of different types of simulation input, are specified with customized option lists.
In this way, a standardized, but flexible framework for constructing any type
of exposure calculation is provided. Figure D.1 illustrates this framework in terms
of the relationships amongst exposure simulation functions and simulation input
types for the specific case of residential exposure to secondhand smoke. The following paragraphs summarize the approach to using this simulation environment
for the research application carried out in this dissertation.
For simulating secondhand smoke exposure in multi-room dwellings there are
four types of data input corresponding to source-person characteristics (smoker),
cigarette emissions characteristics, receptor-person characteristics (nonsmoker),
and the residential environment. The “helper” or sub-component functions that
operate on these inputs pass data along so that later functions can use the results
of earlier ones.
389
Table D.1: Component Subpackages for a Generic Human Exposure Research Software Packagea
Name
Description
heR.Misc
Miscellanous plotting and analysis functions;
log-probability plots; distribution fitting; descriptive
statistics; sample weighting; plot annotation; table
generation and manipulation
heR.Simulation
Generic framework for exposure simulation modeling;
illustrative example for multi-room-dwelling exposures
heR.Activities
Analysis and manipulation of activity pattern data;
time-activity plots
heR.IndoorAir
Indoor air quality modeling; multi-zone; trellis
concentration plots; model fitting; front-end for MIAQ
aerosol dynamics calculations
heR.Inhalation
Inhalation modeling and model parameters; lung
deposition efficiency; inhalation rate
heR.ActivityData
Activity pattern data and documentation from several
large US surveys; 1992-94 USEPA NHAPS; 1989-1992
California ARB surveys of children, adults, and youth
heR.MonitoringData
Air pollutant monitoring data from experiments in
homes and chambers; model fitting; indoor-outdoor
PAH; multi-room SF6 tracer experiements
heR.SurveyData
Exposure monitoring survey data from USEPA PTEAM
particle exposure assessment study in Riverside, CA
ResSmoke
Functions and datasets for simulating residential
exposure to tobacco smoke
a The
individual packages listed in this table constitute an integrated software package intended
to conduct scientific research, and for use in educating students, in the field of human exposure
to environmental contaminants, primarily for exposure occurring via the inhalation route. The
software package is implemented in the R statistical computing environment. See http://www.rproject.org.
Calculate Exposure Time Series in Each Room
Select Deposition and Desorption Characteristics
Assign Room−Specific Emissions Time Series
Simulate Airflows
Assign HAC/HVAC Duty Cycle
Assign Receptor Bedroom and Bathroom Based on Age
Assign Receptor−Related Door/Window Activity
Assign Source−Related Door/Window Activity
Select House Characteristics
Select Airflow Scenario
Select Receptor Person Activity − Matched Day of Week
Select Inhalation Rate
Select Source Person Activity − Over age 18
Select Smoking Characteristics
Assign Source Smoking Activity Pattern
Select Smoker Inhalation Rate
Step 1. Input Processing Function
Processed Inputs
Raw Outputs
Processed
Outputs
Raw Exposures
24−h Mean Receptor Exposures
24−h Mean Source Exposures
Intake Fraction
Equivalent ETS Cigs.
Final Population Outputs
Calculate Exposure Metrics
Generate Descriptive Statistics Results Tables
Generate Diagnostic Plots
Step 3. Output Processing Function
Assign Smoker Exposure Concentration Time Series
Assign Nonsmoker Exposure Concentration Time Series
Step 2. Exposure Function
Final Inputs
Outputs
Figure D.1: Schematic showing the flow of data between elements of an exposure simulation model for inhalation of secondhand smoke
in multi-room dwellings. There are three generic components of the model: a master control function with associated inputs and outputs
(shown in red) calls an input data processing function (in blue), followed by an exposure calculation function (in gold) and an output
data processing function (in magenta). Results generated by each component function are returned to the master function before the final
processed output for the simulated population is returned. The master function controls the base time period and time resolution for the
simulation, as well as the total number of individuals and number of adjoining time periods (i.e., simulation cycles).
House
Deposition
Air Exchange
Desorption
Flow Rates
Environment
Time−Activity
Inhalation Rate
Receptor
Number Cigs.
Mass Emissions
Cig. Duration
Emissions
Time−Activity
Inhalation Rate
Source
Raw Inputs
Define Time Period = 24 h
Define Time Interval − 1 min
Define Input Groups − Environment, Emissions, Source, Receptor
Specify Component Functions − Input, Exposure, Output
Cycle for Each Person and Repeated Time Period
Master Simulation Function
Relationship Amongst Data Inputs and Component Functions
for a Multi−Room Dwelling Inhalation Simulation Exposure Model
390
391
The source and emissions data inputs provide raw lists or empirical distributions for activity patterns, inhalation rates for different ages and genders, the
number of cigarettes smoked in a single day, cigarette total mass emissions, and
cigarette duration, which are sampled by the source subcomponent of the input
processing function for each subsequent individual in the simulation. The sampled source is required to be 18 years of age or older. The sampled number of
cigarettes they smoke during the day are evenly spread out during they times they
spend awake. The receptor data input is identical to that for the source, although
the input processing function selects a particular receptor activity datum to match
that for the source.
Once the source and receptor individual characteristics are assigned, the household environment they occupy is defined based on raw data inputs for house
size and layout, air exchange rates, deposition and desorption rates, and interzonal air flow rates. The environment is also characterized by a particular scenario
where receptors and sources may change the door and window configuration of
the house, heating, ventilation, and/or air conditioning (HAC/HVAC) duty cycle, room-specific filtration, rooms where smoking is allowed, and several other
parameters. The source and receptor activity patterns are augmented to include
information on their door and window activities and the cigarette emissions profile for each room of the house. A complete air flow profile amongst rooms, the
outdoors, and the HAC/HVAC system is also generated.
The final task of the input processing function is to calculate air pollutant concentration profiles in each room using emissions profiles, air flow profiles, as well
as deposition and desorption rates (as appropriate) selected from the raw environment data inputs. These profiles contain instantaneous concentrations for every
minute of the specified time period, typically equal to a single day. The minute-byminute concentrations are equal to minute average concentrations within a very
close approximation.
Once concentration profiles are available for each room, the exposure function
is called to create an exposure profile for receptor and source persons by matching
392
the time series of locations they visit with the corresponding concentrations. As
with the air pollutant concentration profiles, these minute-by-minute exposures
can be considered as minute exposure averages and used to calculate integrated
and 24-h average exposure concentrations.
The output processing function takes the full output list, which consists of the
raw output generated by each component function and their respective helper
functions, and calculates summary quantities of interest. These summary quantities include the 24-h average exposure concentration, individual intake fraction,
and equivalent ETS cigarette intake for each simulated household, and summary
statistics for the time spent by each individual at home, asleep, and in the company
of their matched occupant.
D.2 References
Galassi, M., Davies, J., Theiler, J., Gough, B., Jungman, G., Booth, M., and Rossi, F.
(2003). GNU Scientific Library Reference Manual - Second Edition, Software Version
1.3. Network Theory, Ltd., Bristol, UK, http://www.network-theory.co.uk/.
Ihaka, R. and Gentleman, R. (1996). R: A language for data analysis and graphics.
Journal of Computational and Graphical Statistics, 5: 299–314.
The R Development Core Team (2003). The R Reference Manual – Base Package Volume 1 & 2. Network Theory, Ltd., Bristol, UK.
Venables, W. N., Smith, D. M., and The R Development Core Team (2002). An
Introduction to R. Network Theory, Ltd., Bristol, UK.
393
Index
activity patterns
CAPS, 53
CHAPS, 53
for multiple days, 354
from PTEAM, 336
future studies, 353
more specific categories, 354
NHAPS, 51, 140
prevalence of SHS exposure, 51
time spent by age, 153
time spent by day of week, 158
time spent by gender, 153
time spent by house size, 158
time spent in broad locations, 143
time spent in different rooms, 147
air flows
across doorways, 182
for mitigation scenarios, 306
for scripted simulations, 242
house leakage, 172
HVAC flow rates, 178
simulation of, 208
though windows, 174
air standards
carbon monoxide, 12, 96, 350
for SHS, 350
particulate matter, 12, 96, 350
carbon monoxide
ambient air standards for, 12, 96,
350
emissions, 129
frequency distributions of exposure for, 275
chamber experiments
cigars and cigarettes, 100
for particle deposition, 120
for SHS particle deposition, 123
for SHS volatile organics, 128
children
prevalence of household SHS exposure, 50
proximity to a smoker, 332
residential exposure, 50
time spent at home, 153
environmental tobacco smoke, see secondhand smoke
ETS, see secondhand smoke
exposure
and social ecologies, 76, 354
definition of, 24
effect of asymmetric flow, 284
effect of avoidance and smoker
isolation, 322
effect of doors and windows, 311,
314, 316
effect of filtration devices, 323
effect of HAC, 284
effect of temporal smoking ban,
311
exploratory modeling, 340
field surveys, 36
frequency distributions of, 274
future research, 351
mathematical formulation of, 28
measures of, 29
mitigation strategies, 311
INDEX
394
models, 72
proximity effect, 70, 168, 332
scripted concentrations, 259
scripted ETS cigarette intake, 262
scripted intake fraction, 262
sensitivity to physical parameters,
295
single-zone correction factors, 232,
258, 259, 295
source proximity, 352
to carbon monoxide, 275
to nicotine, 262, 278
to particles, 274, 311
unrestricted scenarios, 294
indoor air
better residential monitoring, 352
effect of multiple compartments,
37
field surveys, 36
guidelines for quality, 349
mixing of pollutants, 9, 68, 166
models, 67
monitoring of, 71
proximity effect, 352
room-to-room SHS monitoring, 43
stove and heater emissions, 47
tracer gas monitoring, 42, 175
validity of models, 71
field studies
monitoring of a smoking household, 332
PTEAM, 37, 331, 334
residential SHS concentrations, 332
models
assumptions, 11
evaluation, 336
exposure, 72
general approach, 7
of IAQ, 67
parameter estimation, 104
single-zone correction factors, 259
tracer gas dynamics, 186
validity for IAQ, 68
health
education, 349
epidemiology, 347
future interventions, 347
improved studies of, 346
need for better exposure assessment, 66
NHANES, 51
risk assessment, 350
SHS health effects, 30
SHS interventions, 61
housing
air exchange rate, 172
base ventilation rate, 174
duct leakage, 178
HVAC systems, 178
interzonal air flow, 182
rate of pollutant mixing, 166
surface-to-volume ratio, 168
volume, 168
window air flow, 174
nicotine
emissions, 128
extended simulation experiment
for, 278
room concentration profiles, 254
scripted exposure, 262
surface dynamics, 128
surface loading, 278
particles
ambient air standards for, 12, 96,
350
deposition rates, 120
dynamic model, 104
equivalent ETS cigarette intake, 262
frequency distributions of exposure for, 274, 311
INDEX
intake fraction, 262
room concentration profiles, 252
scripted exposure, 259
size-specific SHS emissions, 111
total mass SHS emissions, 116
passive smoke, see secondhand smoke
scientific method, 7
secondhand smoke
composition, 23
gas phase, 24
particle phase, 24
dose-response for, 33
health effects of, 30
measures of exposure to, 29
SHS, see secondhand smoke
simulation
air flow balancing, 212
air flow conditions, 208
analysis factors, 17
current efforts, 73
design of experiments, 13
enhancements to the framework,
345
evaluation of, 330
example concentration profiles, 245
input and output variables, 221
intermediate output, 241
mitigation cohort, 302
mitigation scenarios, 216, 304
model structure, 200
selected input parameters, 231
smoking patterns, 217
species treated, 203
summary of findings, 343
synchronization of events, 218
tiers of analysis, 13
treatment of activity patterns, 213
treatment of residences, 205
smoking
effect of restrictions on exposure
reduction, 57
395
effect of restrictions on quitting,
56
interaction of eco-social factors, 76
patterns of, 98
prevalence of restrictions, 50

Residential SHS Exposure

Transcription

Similar documents

Sacred Heart School, Allentown

Newsletter 20

The 2010 Prairie Garden Available for Holiday Gift Giving (#50-09)

March 2013 Employee Newsletter

FOR SALE

Located on the shores of beautiful Lac Beauport, Villegia Manoir St

Knightly News - Stamford High School

Tobacco cessation - Interior Health Authority

The Hotel A simple façade of Higuera street hides the inside of this

- SHS Web of Conferences