CEscoffier DMUG2013w and some applications

Transcription

CEscoffier DMUG2013w and some applications
1
CALPUFF MODEL OVERVIEW AND
SOME APPLICATIONS
Dr. Christelle Escoffier
9 December 2013
DMUG meeting, St Martin in the fields, London
Content
 CALPUFF model characteristics
 CALPUFF validation studies for near-field and longrange applications
 CALPUFF applications
 Cumulative Impact
 Odour nuisance evaluation
REGULATORY STATUS
 CALPUFF is used in over 100 countries and it is accepted for use
in near-field/far-field applications by many international
environmental agencies in Australia, Canada, Chile, China,
Iceland, Italy, New Zealand, Saudi Arabia, South Africa and others
 CALPUFF is one of the two main USEPA Guideline Models
 CALPUFF is the preferred model recommended for BART assessments
(EPA,2005)
 CALPUFF is used by United Nations/International Atomic Energy
Agency (IAEA, 2010) in its SIMPACTS software package (a
simplified approach for estimating impacts of electricity generation)
– used worldwide in developing countries
 CALPUFF is recommended by
 U.S. Federal Land Manager’s Air Quality Related Values Workgroup (FLAG,
2010)
 U.S. Interagency Workgroup on Air Quality Modelling – Phase 2 (IWAQM,
1998)
OVERVIEW
 Non-steady-state Lagrangian Puff Model
 Design for fenceline (~meters) to long-range transport
(hundreds of kilometres)
 3D-variable meteorological data (to represent complex
meteorology events)
 Sea breeze, complex terrain, stagnation
 2D-spatial variability of terrain and landuse and their
associated parameters
 Temporal change of landuse can be modelled (snow,
flooding, etc)
OVERVIEW
 Averaging time from 1-hour to one+ year
 Sub-hourly time steps allowed in version 6 of models
• Point, area, line, and volume sources
• Buoyant or non-buoyant sources
• Special modules for fires, flares and overwater platforms
 Open source model – can be used for research,
development, regulatory applications, etc…
 Available at http://www.src.com
 New GUI (windows 7 and 64bit compatible) developed
by Exponent (available soon)
CALPUFF IS A NON-STEADY-STATE
LAGRANGIAN PUFF MODEL
 Allows variable curve trajectory
 Follow terrain in valley / variability in wind speed/direction
 Meteorological conditions vary and are not assumed
steady-state
 Spatial variability due to winds and turbulence fields
 Spatial variability of geophysical surface conditions
 Retains information of previous hour
 Stagnation, Recirculation, Fumigation
 Allows calm and low wind speeds
 Includes causality effects (plume does not extend to
infinity, size of plume depends on wind speed)
EXAMPLE:
Tall stack point source - coastal area
Tall stack = 75 m
No building downwash
Display:
Wind vectors
Mixing height (m)
Concentrations (μg/m3)
Land/Sea breeze is shown on
a 2-day period
EXAMPLE:
Tall stack point source - coastal area
Tall stack = 75 m
No building downwash
Display:
Wind vectors
Mixing height (m)
Concentration (μg/m3)
It shows:
Curve trajectory
Remember previous hours
Spatial variability of wind
& landuse
Spatial variability of mixing
height
Causality effects
EXAMPLE OF NON-STEADY-STATE MODELS
 Lagrangian Puff models
 CALPUFF
 SCIPUFF
 Particle Dispersion models
 KSP (Kinematic Simulation Particle Model)
 TAPM (The Air Pollution Model)
 Hysplit (HYbrid, Single-Particle Lagrangian Integrated
Trajectory)
 Eulerian grid models
 CMAQ
 CAMx
 CALGRID
CALPUFF DESIGNED FOR NEAR-FIELD
 Stack tip downwash
 Building downwash (PRIME and ISC)
 Dispersion options
 PG Dispersion, Dispersion parameters internally calculated
 Plume rise
 Boundary layers (different for overwater and overland)
 Subgrid scale
 For coastline effect
 For complex terrain
 Fogging and icing – cooling tower
 Visible plume length calculation
CALPUFF DESIGNED FOR LONG-RANGE
 Chemistry schemes options
 Wet deposition
 Aqueous phase oxidation of SO2 in cloud and rainwater
 Scavenging coefficient method
 Function of precipitation type (rain vs. snow)
 Dry deposition
 Resistance-based deposition model
 Plume tilt effects (gravitational settling)
 Recirculation, Stagnation, Flow reversal
 Boundary conditions
 Puff Splitting / Vertical Wind Shear
CHEMISTRY
 Secondary Pollutant Formations (NO3, SO4, SOA)
 Original Scheme
 MESOPUFF II scheme (NOX, SO2) and RIVAD/ARM3 (NO2, SO2)
 User-specified diurnal cycle of transformation rate (NOX, SO2)
 New Schemes
 ISORROPIA (as in CMAQ)
 Aqueous phase oxidation of SO2
 For SOA (CalTech SOA routine from CMAQ-Madrid)
 NO2/NOX ratio (empirical function in CALPOST from NOX/NO2
measurements to estimate NO2 from NOX)
 Time decay
 23 nuclear species
 Parent and daughter species
 Half life decay (half life values – 2.5 minutes to 2-8 days to > 1
year)
METEOROLOGICAL DATA FOR CALPUFF
 Single-met station (same as for AERMOD or ISC) – lit version
 Plume will go in a straight line (directed by met station wind), cannot
reproduce changes due to changes in valley orientation
 3D-meteorological data (from CALMET diagnostic model) –
recommended way to use the model
 Allows full potential of model (including complex terrain modeling)
 2D geophysical domain needed (terrain + landuse + parameters)
 Input to CALMET
 Observations (upper air sounding and weather meteorological stations)
 Prognostic models such as WRF, MM5, ETA, RUC, TAPM
 Sources for WRF & MM5 (if used as input from CMAQ (already developed) or can
be purchased from various data providers as single point meteorological stations)
 Options to run CALMET
 Observations only (OBS mode)
 Prognostic models outputs only (NOOBS mode)
 Observations and Prognostic models outputs (HYBRID mode)
New GUI Interface CALApps: Example CALMET
New GUI Interface CALApps: Example CALMET
New GUI Interface CALApps: Example CALMET
Hourly Meteorological Data:
CALMET output (3D)
Or ISCMET output (1D)
Or AERMET outputs (1D)
CALPUFF DISPERSION MODEL
Emissions:
Varying emission files
CALPUFF
OUTPUT:
Predicted Concentration Fields
Predicted Dry Flux Fields
Predicted Wet Flux Fields
CALPUFF output list File
Relative Humidity Data
Temperature Data
Density Data
Optional Files:
-Hourly Ozone
- Subgrid Scale Coastal Boundary
-Complex Terrain Receptor Data
-Complex Terrain Hill Data
-Boundary File for Mass Flux
Diagnostics
-User-Specified Velocities
-User-Specified Chemical Conversion
Rates
-Input Restart File
Postprocessors to extract
and view data:
CALPOST
CALUTIL
CALSUM, etc...
NEAR-FIELD VALIDATION
Example: Kincaid Tracer
• Kincaid Power Plant
• Tall stack (187m) located in flat farmland in Illinois
• 200 SF6 samplers in arcs of:
 7 arcs < 10 km: 0.5, 1, 2, 3, 5, 7, 10 km
 5 arcs > 10 km: 15, 20, 30, 40 and 50 km
• 171 hours of tracer experiments
• Data Quality Index (QI)
 QI = 3 (best quality, well-defined maximum observed)
 QI = 2 (a maximum is identified, but true max may be different)
 QI = 1, 0 – not used
Kincaid SF6 (Development) Ordered by Rank (QI=3) Arc-Max
3
1000
1:1 Line
2x Line (Overprediction)
2.8
2.6
0.5 Line (Underprediction)
Modeled Chi/Q (ns/m3)
2.4
2.2
AERMOD
2
100
CALPUFF v5.8 CTDM CALturb PsvPsw
CALPUFF v5.8 CTDM PG
1.8
CALPUFF w/ turb. disp
AERMOD v07026 NoSVW PsvPsw
ISCSCT3 ISCMET
1.6
CALPUFF w/ PG disp
1.4
1.2
101
1
10
ISCST3
1.2
1.4
1.6
1.8
2
100
2.2
Observed Chi/Q (ns/m3)
2.4
2.6
2.8
3
1000
Kincaid SF6 (Development) Ordered by Rank (QI=2,3) Arc-Max
3
1000
2.8
CALPUFF v5.8 CTDM CALturb PsvPsw
2.6
CALPUFF v5.8 AERMET CALturb PsvPsw
Modeled Chi/Q (ns/m3)
2.4
2.2
AERMOD v07026 PsvPsw
2
100
1.8
1.6
1.4
1.2
101
1
10
ISCSCT3 ISCMET
1.2
1.4
1.6
1.8
2
100
2.2
2.4
3
Observed Chi/Q (ns/m )
2.6
2.8
3
1000
LONG-RANGE VALIDATION
Example: European Tracer Experiment (ETEX)
• Designed for emergency
response model evaluation
• PMCH tracer release in Oct and
Nov 1994 from north western
France
• 12-hour release starting on Oct
23, 1994 at 16:00 UTC
• 3-hour average samples at
various times over 168
samplers in 17 countries
~300 km
Source: Scire et al. 2013
300 km x 300 km
box
• Most samplers over 300 km
away with tracer measured to
over 2000 km from release site
ETEX BASE – CALPUFF 36KM MMIF (run with some
errors)
ETEX 4 – CALPUFF 12-KM
CALMET
- Run on the left was performed by a third-party with some errors in settings and a lower
meteorological resolution. Meteorology is an important parameter for long-range
applications using lagrangian models
- By making setting corrections and properly using wind shear with the 12-km MM5 fields,
the model performance is dramatically better and comparable to other models results such
as CMAQ, HYSPLIT, FLEXPART (not shown)
Source: Scire et al. 2013
A Few Examples of CALPUFF Applications
 Regulatory applications (coal-fired, gas-fired, biomass
power plants, aluminium smelters, refineries, etc...)
 Road traffic modelling (special slug options for nearsource impact)
 City-wide modelling / Airport modelling
 Cumulative impacts assessment
 Forecasting system development
(MM5/CALMET/CALPUFF)
 Spray dispersion modelling (coupled with AGDISP)
 Odour nuisance evaluation
APPLICATION: Cumulative Impact Assessment in the USA
Sea Breeze Case – July 7, 1988 – 1:00pm LST
71.5W
72.0W
Site A is considered
70.0W the main source
and the location of
43.0Nthe meteorological
station used in
AERMOD
70.5W
71.0W
4760
Sea
4740
Breeze
Land
UTM North (km)
Breeze
C
B
4720
Sites B and C are the
locations of the
background sources
to be added to the
42.5N
cumulative impact
Assessment
1
2
4700
A
Wind vectors display
from CALMET model
3
4680
4660
4640
260
280
300
320
340
UTM East (km)
360
380
400
Within the PSD program,
42.0N if Source A’s impact is
above SILs, a cumulative
impact assessment is
required
420
APPLICATION: Cumulative Impact Assessment in the USA:
Sea Breeze Case – July 7, 1988 – 1:00pm LST
71.5W
72.0W
70.5W
71.0W
70.0W
UTM North (km)
43.0N
CALPUFF plumes
from 3 sources
4760
Project facility in
land breeze area (A)
4740
Two background
sources in sea
breeze area (B & C)
C
B
4720
1
42.5N
2
4700
AERMOD model
using
meteorological
observations (A)
close to project
facility
A
3
4680
4660
42.0N
4640
260
280
CALPUFF plume
direction
300
320
340
UTM East (km)
360
380
400
AERMOD plume
direction
420
To be applied for all
sources (A, B & C)
in the cumulative
application
APPLICATION:
Odour Nuisance Evaluation in NSW, Australia
Most odours are a complex mixture of many
odorous compounds
Odour is a Nuisance:
Odour detection is frequently below
the health standard
Dynamic Dilution Olfactometry
is the preferred method of measuring
odour and is endorsed worldwide
Compound
Workplace
standard
(ug/m3)
Odour
Threshold
(ug/m3)
Odour strength is quantified in terms of odour
units, a measure of odour concentrations
Ethyl
mercapton
43
0.075 – 2.5
Hydrogen
sulfide
470
0.7 – 7
Methyl
mercapton
33
0.04 - 4
APPLICATION:
Odour Nuisance Evaluation, NWS Australia
Why modelling odour with CALPUFF?
Description
Steady-State plume models
Non-Steady-State puff model
‘Memory’ of previous emissions
allows build-up of odours under
stagnant conditions which occur
typically through the night
Stagnation and retention
No ‘memory’ of previous hours
emissions
Treatment of calms
Unable to treat calms
Assume non-zero wind
Well-suited to treating calms. Puffs
are able to diffuse without being
advected.
Causality effects
Straight line trajectories, plume
travels to infinity even after 1-hour
Allows variable/curved trajectories
Spatial and temporal variability
Spatially constant met. Conditions
Spatially varying met. conditions
Recirculation / flow reversal
Not suitable
3D circulation through combination of
prognostic model data
Note: When high resolution meteorology is available , sub-hourly time steps could be modelled
with CALPUFF
APPLICATION:
Odour Nuisance Evaluation in NSW, Australia
Example: sewage treatment plant
-point sources, volume sources and
area sources are modelled
-Species modelled: Odour (in OU)
-2 OU and 7OU plotted
POPULATION DENSITY OF
AFFECTED COMMUNITY
(PERSONS PER KM2)
IMPACT ASSESSMENT
CRITERIA
(ODOUR UNITS)
NOSE RESPONSE TIME
AVG.
Urban areas (≥~ 2000) and/ or schools
and hospitals
2.0
~ 500
3.0
~ 125
4.0
~ 30
5.0
~ 10
6.0
Single residence (≤ ~ 2)
7.0
Source: Approved Methods for the Modelling and Assessment of Air Pollutants in NSW, DECCW 2005
Odour impact assessment (1-second averaging time (nose response time) )
Concentrations predictions are in 1-h average (if meteorological data are hourly)
A ‘peak to mean ratios’ method is applied to the 1-hr averaging period predictions of CALPUFF
Predictions are interpreted as one second averaging period values
APPLICATION:
Odour Nuisance Evaluation in NSW
Adjustment to 1-hr averaging period to sub-hourly period
1/5th Power Law
The 1/5th power law is frequently used to estimate < 1-hour odour concentrations.
C1 = C0 * ( t0 / t1 ) p
where :
C0 = the initial (1-hour average) concentration
C1 = the concentration at the desired avg. period
t0
= the initial (60-minute) averaging period
t1
= the desired averaging period (minutes)
p
= power law exponent (0.2)
OR
Peak-to-Mean ratio (PtM)
The PtM takes into account
the ratio of the peak smelled by
the nose over a short-period and
the average dispersion model
result over 1-hour
Tables with variable Peak-to-Mean ratios as a function of type of source, stability
and near-field/far-field are available in DECCW (2005).
CALPUFF has settings options to handle the variable peak-to-mean ratios
CONCLUSION
 CALPUFF is used in over 100 countries and it is accepted
for use in near-field/far-field applications by many
international environmental agencies
 CALPUFF is designed to model concentrations at a few
meters from sources and for long-range transport
 It is has been evaluated and validated for all ranges
 Inputs are 3D-variable meteorological data
 to represent complex meteorology events and complex terrain
(flow in valleys, slopes flow, sea breeze in coastal area, etc...)
 Because of its characteristics, it is a good tool for
applications involving (but not limited to)
 Sea breeze, complex terrain, stagnation, cumulative impact
analysis with background sources, odour modelling, etc...
THANK YOU FOR YOUR ATTENTION
Contact Information: Dr. C Escoffier
cescoffier@exponent.com