pdf - dmg - University of Washington
Transcription
pdf - dmg - University of Washington
An Architecture of Relationships Built on the Use of Parametric Modeling and Evaluative Analysis in Design Scott Crawford A thesis submitted in partial fulfillment of the requirements for the degree of Master of Architecture University of Washington 2009 Program Authorized to Offer Degree: Department of Architecture University of Washington Graduate School This is to certify that I have examined this copy of a master’s thesis by Scott Crawford and have found that it is complete and that any and all revisions required by the final examining committee have been made. Committee Members: Brian McLaren Mehlika Inanici Rick Mohler Date: In presenting this thesis in partial fulfillment of the requirements for a master’s degree at the University of Washington, I agree that the Library shall make its copies freely available for inspection. I further agree that extensive copying of this thesis is allowable only for scholarly purposes, consistent with “fair use” as prescribed in the U.S. Copyright Law. Any other reproduction for any purposes or by any means shall not be allowed without my written permission. Signature: Date: Table of Contents List of Figures Introduction Chapter 1 Design of the Process and Object 1 3 1.1 Integrated Design 4 1.2 Systems Thinking 10 1.1.1 Integration of Building Systems 1.1.2 Integration of Design Process 1.1.3 Summary 1.2.1 1.2.2 1.2.3 1.2.4 Systems Modeling Simulation Summary Chapter 2 Case Studies 14 2.1 Jean-Marie Tjibaou Cultural Center 15 2.2 Beijing National Aquatic Center 20 2.3 Abu Dhabi Performing Arts Center 25 2.1.1 2.1.2 2.1.3 2.1.4 2.1.5 2.2.1 2.2.2 2.2.3 2.2.4 2.2.5 Project Description Integration of Building Systems Evaluating the Design through Simulation Analysis of the Design Process Summary Project Description Integrating Building Systems Generating the Design through Simulation Analysis of the Design Process Summary 2.3.1 Project Description 2.3.2 Generating the Design through Simulation 2.3.3 Summary i iv 2.4 Swiss Re Tower 2.4.1 2.4.2 2.4.3 2.4.4 2.4.5 Project Description Integration of Building Systems Generating & Evaluating the Design through Simulation Analysis of the Design Process Summary Chapter 3 Simulation Software 33 3.1 Digital Morphology 34 3.2 Digital Performance Analysis 41 3.3 Iterative Design Process 45 3.1.1 3.1.2 3.1.3 3.1.4 3.1.5 The Element of Geometry Creation of Systems through Geometry Form Generation Experience with Geometry Summary 3.2.1 Analysis of Psychometric Parameters 3.2.2 Experience with Analysis 3.3.1 Adaptation of an Iterative Design Process 3.3.2 Limitations of Iterative Design Process Chapter 4 Past Projects 48 4.1 Wallingford Library - Winter 06 49 4.2 Museum of Steel - Winter 07 65 4.1.1 4.1.2 4.1.3 4.1.4 4.1.5 4.1.6 4.1.7 ii 28 4.2.1 4.2.2 4.2.3 4.2.4 4.2.5 Project Description Rebuilding the Base Geometry Analysis of Sun Penetration Analysis of Lighting Levels Analysis of Thermal Performance Analysis of Acoustic Performance Summary of Analytic Studies Project Description Initial Development of Geometry Rebuilding the Base Geometry Developing the Space Frame Component Modeling the Systems of Geometry 4.2.6 Summary of Morphological Studies Chapter 5 Design Project 5.1 The Concept 81 5.2 Role of the Definition 84 5.3 Iterative Design 90 5.4 Relational Kit of Parts 93 5.1.1 Building Concept 5.1.2 Site Response 5.2.1 Definition as Diagram 5.2.2 Flexible, Relational Geometry 5.3.1 Influence of Daylighting 5.3.2 Balance of Influences 5.4.1 Establishment of Rules 5.4.2 Possibilities through Digital Fabrication Conclusions Bibliography Appendix A Site Analysis Appendix B Lighting Analysis Pocket Material: DVD of presentation/modeling files iii 79 96 99 102 105 Table of Figures Fig. 1.1 Puzzle as a metaphor for design 4 Fig. 1.2 Physical layering of building systems 5 Fig. 1.3 Visual integration of systems that reinforce spiralling geometry 6 Fig. 1.4 Performance integration of roof and mechanical systems in Kansai Airport by Renzo Piano 6 Fig. 1.5 Mutual reshaping of objective and subjective 7 Fig. 1.6 Nesting of architectural systems 10 Fig. 1.7 Breakdown of the components that make up the simplest form of a system. 11 Fig. 1.8 Interaction between desired and current state of a model 12 Fig. 2.1 Jean-Marie Tjibaou Cultural Center 14 Fig. 2.2 Beijing National Aquatic Center 14 Fig. 2.3 Abu Dhabi Performing Arts Center 14 Fig. 2.4 Swiss Re Tower 14 Fig. 2.5 Site Plan of the built design 15 Fig. 2.6 Local dwellings with post and beam structure 15 Fig. 2.7 Wood ribs connected by steel members 15 Fig. 2.8 Iroko wood structure and slats 16 Fig. 2.9 Diagram of the building response under conditions with strong winds 16 Fig. 2.10 Wind tunnel analysis 17 Fig. 2.11 Structural simulation analysis 17 Fig. 2.12 Transverse section of competition design 18 Fig. 2.13 Transverse section of built design iv 18 Fig. 2.14 Rendering of Beijing National Aquatic Center (left) and Beijing National Stadium (right) 20 Fig. 2.15 Plan of Beijing National Aquatic Center 20 Fig. 2.16 Ventilation diagram illustrating the ‘greenhouse’ walls 21 Fig. 2.17 ETFE skin panels of the Water Cube 21 Fig. 2.18 Process of extracting structure from block of Weaire-Phelan foam 22 Fig. 2.19 Interior of the Water Cube 22 Fig. 2.20 3D model printed from the parametric model was used to illustrate structural concept to clients. 23 Fig. 2.21 Rendering of the building viewed from the SE 25 Fig. 2.22 Rendering of the building viewed along water 25 Fig. 2.23 Basic set of geometries from growth simulation 26 Fig. 2.24 Iterations from interaction of growth simulation and programmatic diagrams 27 Fig. 2.25 CFD study illustrating stresses induced by wind loads (red indicates highest stress) 28 Fig. 2.26 Diagram depicting the building form concept that was arrived at after CFD analysis 28 Fig. 2.27 6th floor depicting the triangular voids located between the offices ‘fingers’ 29 Fig. 2.28 Spiraling thermal chimney 29 Fig. 2.29 Aerodynamic building form causes the air to travel is an smoother path compared to a typical tower 30 Fig. 2.30 Diagrid glazing system and building structure that has not yet received aluminum covering 30 Fig. 2.31 Physical model used to mock up preliminary ideas about the structure and mullion order 31 Fig. 3.1 Plan, section, elevation explored in serial sequence in the traditional design process 34 Fig. 3.2 Design tasks such as morphology, documentation, analysis, and representation are performed in parallel 34 Fig. 3.3 Polygonal surface created in Sketchup 35 Fig. 3.4 NURBS surface created in Rhino is smoother than the surface created in Sketchup 35 v Fig. 3.5 Point cloud created in Grasshopper can be used to define multiple sets of relationships between points 35 Fig. 3.6 Example of a geometric system 36 Fig. 3.8 San Carlo alle Quattro Fontane - Borromini 39 Fig. 3.7 Camp Nou stadium by Foster+Partners 39 Fig. 3.9 Diagrams from Sun, Wind, & Light depicting rules of thumb for solar and wind orientation 41 Fig. 3.10 Examples of graphical illustrations produced by Ecotect. Lighting analysis (top) Shadow analysis (bottom) 43 Fig. 4.1 Second Floor 49 Fig. 4.2 Ground Floor 49 Fig. 4.3 Section Perspective 49 Fig. 4.4 Lighting model of surface geometry 50 Fig. 4.5 Thermal/Acoustic/Wind model of zone geometry 50 Fig. 4.6 Shadow studies and Sunpath Diagrams for the 21st of June, September/March and December 52 Fig. 4.7 First Floor - March 21st at 12pm 53 Fig. 4.8 Second Floor - March 21st at 12pm 53 Fig. 4.9 No Skylights - March 21st at 12pm 53 Fig. 4.10 Ecotect Lighting Study for June 21st at 8am, 12pm, 4pm 54 Fig. 4.11 Ecotect Lighting Study for March 21st at 8am, 12pm, 4pm 54 Fig. 4.12 Ecotect Lighting Study for December 21st at 9am, 12pm, 3pm 54 Fig. 4.13 Wall section layer properties in Ecotect 55 Fig. 4.14 Zone thermal properties in Ecotect 55 Fig. 4.15 Internal (hourly) temperature on hottest day of year 56 Fig. 4.16 Internal temperature on March 21st 56 vi Fig. 4.17 Internal temperature on coldest day of year 56 Fig. 4.18 Hourly Internal Temperature (zone colors blue-central library space, green-storage, orange-meeting) 57 Fig. 4.19 Monthly Heating/Cooling Loads (zone colors blue-central library space, green-storage, orange-meeting) 57 Fig. 4.20 Passive Gains Breakdown 58 Fig. 4.21 Ventilation gains 58 Fig. 4.22 Gains Breakdown 59 Fig. 4.23 Direct Solar Gains 59 Fig. 4.24 Building Fabric Gains 59 Fig. 4.25 Interzonal Gains 59 Fig. 4.26 Central Library Space Passive Gains Breakdown 60 Fig. 4.27 Meeting Room Passive Gains Breakdown 60 Fig. 4.28 Initial attempt: Linked acoustic ray analysis of children’s area 61 Fig. 4.29 Second attempt: Linked acoustic ray analysis with addition of adult stacks to second level 61 Fig. 4.30 Reverberation time for wood ceiling 62 Fig. 4.31 Reverberation time for fabric ceiling 62 Fig. 4.32 Reverberation time for acoustic ceiling tiles 62 Fig. 4.33 Site plan 65 Fig. 4.34 Initial sketches of building form 66 Fig. 4.35 Early attempts at rationalizing roof form 66 Fig. 4.36 One of the final iterations of the base geometry 67 Fig. 4.37 Exploded axonometric of the final building form 67 Fig. 4.38 Decomposition of points into X, Y, Z coordinates, Z coordinates replaced with values from a sine function 68 vii Fig. 4.39 Creation of rail and section curve 68 Fig. 4.40 Division of surface into a grid of points 68 Fig. 4.41 Steps for rebuilding the Level 1 Base geometry 69 Fig. 4.42 Paracloud surface created from points imported from Rhino 70 Fig. 4.43 Paracloud surface with added depth creating ribs 70 Fig. 4.44 Paracloud cell matrix for defining components 70 Fig. 4.45 Steps for developing space frame upon base geometry 71 Fig. 4.46 Development of Triangular grid 72 Fig. 4.47 Space frame component defined by two different sized intervals. 73 Fig. 4.48 Two versions of the space frame generated with the change of only the rail curve 73 Fig. 4.49 Wall space frame 74 Fig. 4.50 Ground plane and wall derived from perimeter curves of roof geometry. 74 Fig. 4.51 Ground plane reconfigured with a small number of changes to the cull pattern 75 Fig. 4.52 Roof panel cull patterning studies 76 Fig. 5.1 Competition program for elementary school 79 Fig. 5.2 Cascade Playground and Land Use of Surrounding Blocks 80 Fig. 5.3 Current site conditions 81 Fig. 5.4 Site Cleared for Design Project 81 Fig. 5.5 Site Butterfly Shadow Diagrams 8am-4pm Zoning Height 85’ 82 Fig. 5.6 Sun angle diagram (points on curves represent mid-day) 82 Fig. 5.7 March Butterfly Shadow Diagrams over course of design 83 Fig. 5.8 Undeveloped plan 84 viii Fig. 5.9 Week10 Grasshopper Project definition 84 Fig. 5.10 Week10 Grasshopper Project definition 85 Fig. 5.11 Week10 Grasshopper definition of gym roof structure and geometry it is dependent on. 86 Fig. 5.12 Parametric relationship of room area to length along curve and width of extrusion 87 Fig. 5.13 Examples of a bezier curves 87 Fig. 5.14 Adaptation of site to building form 88 Fig. 5.15 Week10 Building Program Layout 89 Fig. 5.16 Week10 Site Program Layout 89 Fig. 5.17 Evolution of Roof structure 90 Fig. 5.18 Analysis comparison of Useful Daylight Index 100-2000 lux 91 Fig. 5.19 Analysis comparison of Useful Daylight Index 100-2000 lux of classrooms and gym 92 Fig. 5.20 Sample of three steps for generating the floor structure 93 Fig. 5.21 Exploded axonometric of classroom 94 Fig. 5.22 Layers of classroom structure 95 ix Acknowledgements I would like to first acknowledge the mentorship, critiques, and friendship provided by Professor Brian McLaren throughout this thesis project. From the beginning of this process, Professor McLaren helped to guide me through the development of my research without explicitly telling me what to do. That flexibility helped me to grow greatly as a student of architecture. I would also like to acknowledge the other members of my thesis committee, Professors Mehlika Inanici and Rick Mohler. Without Professor Inanici’s class on simulation and her feedback throughout the thesis much of my work would not have been possible. Professor Mohler has always pushed me to rationalize and understand the details of my designs and for that I appreciate his involvement with this thesis. I would also like to acknowledge Professor Rob Corser’s feedback and insights into parametric modeling during the final four weeks of the thesis project. I would also like to acknowledge the Fall 2008 Architecture Thesis Cookie Club for the delicious cookies, twice a week distractions, and mostly importantly their friendship. Lastly and most importantly, I would like to acknowledge my wife, Christina, and daughter, Mikayla, whom were always patient with me during these three years of school and especially during this final quarter when I spent more time at school than at home. Without them to come home to I would not have been able to complete this program. Dedication This thesis is dedicated to my father, Robert Dexter Crawford, who introduced me to many ideas before I was capable of fully comprehending them, but without those conversations I wouldn’t be who I am today. Introduction 1 Design can be thought of as a process of exploring the establishment of relationships between objects, forces, culture, context and other influences. Geometric and proportional relationships are used to influence form. Rules of thumb have been established which make suggestions for the relationship of a building to its site. As time has past the number of these relationships have increased both in number and scope, making the balancing of them more difficult and all the more necessary. Parametric modeling and evaluative analysis are two emerging design tools that have the potential to further extend and unify this idea of an architecture of relationships. Evaluative analysis offers a platform for designers to move beyond simple rules of thumb and explore the performance of their designs, while parametric modeling sets up a new way of conceiving and manipulating geometric relationships. Both of these types of software have capabilities beyond typical design tools because they exploit the computers ability to do a large number of calculations in a short period of time, making feasible tasks that would be unreasonable for an individual to attempt on their own. This thesis is concerned with what roles these tools can play in design, how their early integration changes the design process, and ultimately their impact on architecture. Before exploring the integration of these tools into a design process an attempt will be made to establish what is meant by the term integration, and how that applies to design. From that point a series of case studies are explored which show how this takes place in an architectural setting. The new capabilities brought by evaluative analysis and parametric modeling create the necessity for the acquisition of new skills. This experience 2 was achieved by revisiting two past projects through the scope of either parametric modeling or evaluative analysis. The final piece of this thesis addresses the initial questions surrounding the impact of the integration of these tools on design by attempting to use parametric modeling and evaluative analysis software from the beginning of the design of a new project. Chapter 1 Design of the Process and Object 3 “The impulse to learn in children goes deeper than desires to respond and adapt more effectively to environmental change. The impulse to learn at its heart, is an impulse to be generative, to expand our capability.” -Peter Senge, The Fifth Discipline As buildings and design increase in complexity due to addition of new elements it becomes necessary to develop a framework for integrating these new elements with the old, rather than simply coordinating them for the purpose of removing conflicts. Integration has the potential to not only strengthen the relationships between elements but also to entirely transform the elements of a building or design process by offering insight into issues that may go unnoticed when examined in isolation. Systems thinking offers a way of analyzing and breaking down a system into smaller parts, thereby making it easier to study the interaction between elements. Systems thinking applied to the design process analyzes the structure of a design process and opens an understanding of how to integrate new tools into the process. In the design of a building, systems thinking works to investigate how systems interact with the goal of creating crossover functions between systems that improve performance. Because neither the design process or building is influenced by a single factor there is the need for finding a way of balancing the various influences. Through the investigation of these theories a design process model will be assembled with the intent of integrating new elements into a design process or building. 4 1.1 Integrated Design concept objective data trimmed by concept a) Integrate - combining with another to create a whole1 Integration is one of the main jobs of an architect who is responsible for synthesizing a wide variety of information coming from different fields such as engineering, art, or psychology. The process of integration can be likened to the piecing together of a puzzle except the initial pieces have not yet been cut to fit together. An architect uses their experience and education to cut pieces into what is perceived to be an appropriate shape. Leftover information is either discarded or later found to be necessary, leading to a reformatting of the pieces either on a micro-scale that effects a small number of pieces or a macro-scale that requires the majority of pieces to be reconfigured. Building systems are an area in which integration is a necessity. The structural, mechanical, and other building systems must coexist to some degree and even cooperate or strengthen each other. In the book Integrated Buildings, Leonard R. Bachman classifies building integration into three categories, physical, visual, and performance integration.2 Integration of the design process is also necessary and often overlooked because of the difficulty in defining its parameters. In general a design process can be thought to consist of both objective and subjective information, a set of tools that provide discrete information, and a collection of experiences both individual and collective.3 Not only is integration necessary within the design process and building systems but also between these areas. fit by designer Fig. 1.1 Puzzle as a metaphor for design 1 Oxford American Dictionary 2 Leonard R. Bachman, Integrated Buildings, Wiley & Sons, NJ, 2003, p. 3 3 Leonard R. Bachman, Integrated Buildings, Wiley & Sons, NJ, 2003, p. 6 5 When the various pieces of the design process become integrated they have the ability to develop a stronger focus or ‘big idea’ for organizing and integrating building systems. Without this integration the design process and building becomes stratified into a series of steps or layers that only add on to what came before rather than reshaping and strengthening what was previously conceived. With integrated design the focus is on studying the building through a holistic approach that tries to work with all of the building systems starting in the initial phases of the design. The intention is to understand how the systems interact with each other so that they can be integrated which can take place in different ways. 1.1.1 Integration of Building Systems Fig. 1.2 Physical layering of building systems Physical Integration Buildings systems are typically broken down into separate layers such as structural, mechanical, or lighting. By separating the layers there is removal of the possibility of interference between them but at the cost of the need for additional space. Careful planning during the design process can mesh these systems together, lessening the overall space they require. This planning is typically difficult because of the complexity involved with visualizing how these systems interact. Computer 3D simulation has improved the ability of building systems layout by illustrating where possible interferences may take place. Physical integration also deals with systems on the scale of details where materials with different qualities need to be joined together or kept separate. Here again 3D simulation allows the designer to work at a variety of scales in one digital model. Before a building is ever built the assembly and accessibility of connections can be examined virtually. 6 Visual Integration Separate systems combine to form the image of the building through aesthetic factors such as color, size, shape and placement.4 When visual integration does not occur the image of the building may suffer leaving a building that is not experientially enjoyed and gains little value in the eye of society. Computer simulation and the availability of rendering software that takes advantage of realistic lighting and material properties allows architects to work out the visual integration of their building during the design process. Fig. 1.3 Visual integration of systems that reinforce spiralling geometry Performance Integration Components of a building must be able to work together or at least not prevent other systems from properly functioning. Instead of only avoiding interference it is possible to integrate two building systems into one thereby replacing repetitive pieces. An example would be using a triangular grid structure that removes the need for a separate lateral bracing system. This type of integration has cost reduction implications as well as the ability to lower the complexity of having to layer multiple systems onto each other. Simulation analysis software can be used to understand the relationships between building systems and may inform an architect as to how they might be integrated. 1.1.2 Integration of Design Process Fig. 1.4 Performance integration of roof and mechanical systems in Kansai Airport by Renzo Piano Integration of the Subjective and Objective Architecture is a unique field because it requires the balance of objective information, such as technology, performance and program data, with 4 Leonard R. Bachman, Integrated Buildings, Wiley & Sons, NJ, 2003, p. 4 7 objective objective bias equal reshaping subjective bias subjective Fig. 1.5 Mutual reshaping of objective and subjective more loosely constructed ideas like concepts, aesthetics and intuition that might be based on past experience. The architect takes on the role of both artist and scientist and is responsible for determining what balance between the two is most appropriate for any given design problem. The better an architect becomes at integrating the subjective and objective parts of design the less the building will resemble a collection of performance features placed alongside superficial aesthetic concepts. This balance does not take place with decisions about what should be kept or discarded, but instead comes from understanding how to use the various sources of information to reshape each other. A simple example would be using objective data regarding the ‘most efficient’ building orientation for solar performance and allowing this form to both influence and be influenced by concepts related to circulation and program organization. Neither one dictates the direction that the other should take but instead the interaction between the two sources is manipulated until an acceptable outcome is reached. Comprehensive Toolkit The integration of objective and subjective realms is partially dependent on a balanced set of design tools. The information provided by these tools can also be classified as either objective or subjective. If a toolkit is too heavily biased in one direction then the design will likely lack in the other. Computer simulation software has the ability to provide a subjective virtual experience of a building design through animations and renderings, while also offering the capability for form generation and manipulation. Other simulation software provides more objective, quantitative feedback that is associated with the performance of a design related to parameters such 8 as heating loads, lighting levels, or cost implications. In either case the information provided by the software must be interpreted by the architect in relation to the other data. If the other data does not exist because of the lack of tools to address those conditions then the design may struggle to bridge between these two realms. Another issue is how and when to integrate these tools into a design process. Generally speaking, the earlier a tool or information is incorporated into a design process the more significant a role it can play. Every tool cannot be integrated into the design process at the beginning, but instead a judgement must be made concerning what tools would be most effective during the initial design decisions. Lighting analysis would not be appropriate at the earliest stages if a building form has not yet been decided. An analysis of solar exposure or climate data may be an appropriate starting position but explorations into form generation might also be valid. In the end these decisions cannot be prescribed because they are unique to the qualities, comforts, and intuition of each designer. The importance of the comprehensive toolkit lies in being able to integrate the early ideas generated through one medium with the feedback from another, and continuing this regenerative process throughout design. The Role of Experience (Individual and Collective) Every designer has a unique background and mental framework that informs their design process. In addition to these personal experiences they have available to them the vast collection of experiences associated with the history of architecture and the greater collection of experiences of human history. Similar to the comprehensive toolkit these experiences can be classified as either subjective or objective. The majority of an individual’s 9 design process may be shaped by subjective forces related to design creativity or intuition, but their design will inevitably be influenced by objective knowledge concerning codes, specifications, space requirements and building program.5 On top of these individually held ideas there is a need to integrate one’s ideas with the surrounding intellectual world in which they were shaped. No idea is created on its own but instead evolves in relation to other ideas. Rather than trying to isolate one’s ideas from an historical understanding, an idea can be greatly improved by applying lessons previously learned, skipping the vast field of errors that have already been confronted. 1.1.3 Summary Integrated design is an approach that can be useful to an architect but there needs to be an to be an idea of what is being integrated. The subjects of integration can range from the systems of a building in terms of their physical, visual, or performance integration or it can be about the design process and the information, tools, and experience that are required within it. Integration must occur not only within the design process and building systems but also between the design process and the building systems. Decisions made during the design will later impact how the building systems perform. Likewise, the performance requirements of a building system, such as the impact of gravity on the structure, influences activities of the design process. In order for integration to occur these complex processes and objects must be broken down into simpler pieces. Systems thinking is a way in which to break down these pieces. 5 Leonard R. Bachman, Integrated Buildings, Wiley & Sons, NJ, 2003, p. 8 10 1.2 Systems Thinking Systems thinking originated in the 1970s in response to the study of ecosystems and other natural systems. During this time there was an effort to describe how these complex systems functioned leading to the need for a method of dissecting a system into smaller and smaller subsystems while still retaining an understanding of how to reassemble them back into a whole. In order to facilitate this understanding, systems thinking uses models to simplify real systems and then studies these models through simulation. zone space component assembly site Fig. 1.6 Nesting of architectural systems Similar to natural systems, buildings can be thought of as having their own “complex energy and material systems that have a lifespan.”1 The methods of systems thinking can be applied to the field of architecture as a way of understanding the range of systems that a building is part of from the scale of urban systems to the individual detail of material systems. This approach can also be applied to an analysis of the design process in order to improve the understanding of how the process is carried out and what subsystems might exist within it. In order for this to occur a system must be defined, then a model generated which is then tested through simulation. Each of these steps help to form an understanding of a problem and how the variables interact. With this knowledge it becomes possible to tie together aspects of the problem that may have previously seemed unrelated, therefore opening up opportunities for transforming a project beyond what may have been previously considered possible. 1 The Emergence and Design Group, Emergence in Architecture, Architectural Design, May 2004, p. 6 11 element attribute relationship Fig. 1.7 Breakdown of the components that make up the simplest form of a system. 1.2.1 Systems Systems are composed of: a) Elements - the parts that make up the system, b) Attributes - the characteristics of the elements c) Relationships - the associations between elements and attributes2 Based upon this definition a wide variety of systems can be found within a building such as the structural system, mechanical systems, or electrical systems. A system can be considered to be an element of a larger system as is the case with a structural system within a building system that then fits into an urban system. The design process can also be described as a system. A defining feature of a system is emergence, when the relationships created in a system are not found in the attributes of the system’s elements.3 Systems can be classified into a number of different typologies, but for the purpose of this thesis the types that are of most interest are: a) Morphological system - understanding of relationships between elements and attributes but little knowledge about the transfer of energy b) Cascading system - flow of energy from one element to another is understood but with little knowledge about the quantitative relationships between elements4 These two system types become important in the simulation of buildings. 2 Michael Pidwirny, Definitions of Systems and Models, http://www.physicalgeography.net/ fundamentals/4b.html 3 The Emergence and Design Group, Emergence in Architecture, Architectural Design, May 2004, p. 6 4 Michael Pidwirny, Definitions of Systems and Models, http://www.physicalgeography.net/ fundamentals/4b.html 12 Action 1.2.2 Modeling Gap Addition Current State Desired State A model is a ‘simplified representation of a system created to promote an intentional development of understanding of the real system.’5 In architecture, models can take the form of physical scaled models or virtual digital models, even two dimensional drawings, sketches and diagrams serve as models of the intended building. The accuracy with which they represent the final building varies but each one is used to promote an understanding of the real system. In modeling, a desired state creates an action that produces an addition that creates the model or adds to the current state, closing the gap between the model and reality over time. As design progresses a model evolves to describe a more defined system that comes closer to resembling what the final building will be like. In systems thinking this evolution occurs is in response to feedback from simulation. 1.2.3 Simulation Fig. 1.8 Interaction between desired and current state of a model Simulation is the ‘intentional manipulation of a model in order to perceive interactions not apparent in real time or space.’6 The model is then redefined in response to the feedback provided through simulation. Feedback occurs in two forms: a) Reinforcing - a change in one part of the system leads to a change in another, which in turn, amplifies the change in the first b) Balancing - a change in one part of the system causes a change in another part, which in turn, counteracts the change in the first 5 Gene Bellinger, The Model, http://www.systems-thinking.org/modsim/modsim.htm 6 Gene Bellinger, The Model, http://www.systems-thinking.org/modsim/modsim.htm 13 Simulation takes place in different ways depending on whether the system being modeled is a morphological or cascading system. In the case of the morphological system, the model of the system is analyzed in terms of formal and conceptual qualities. This analysis can be as simple as judging whether a design aesthetically matches one’s design goals. The model for a cascading system is analyzed in terms of flows and forces that move through a building, for instance the loads experienced by a structural system can be simulated to predict the building’s probable response. Here it can be seen that a system, such as the structure of a building, can be both a morphological and cascading system which is called a process response system.7 In the end the most important part of simulation is the feedback that it provides because this is what is used either to push the model towards the desired state or redefine the desired state. 1.2.4 Summary Systems thinking extends the idea of the building or a design task beyond its immediate goal or context and establishes the larger framework in which those elements exist. While certain conventions in architecture might establish hierarchies that a building fits within, systems thinking offers a way of exploring where, why and to what those relationships correspond. When the ideas behind systems thinking become conscious to a designer they possess the ability of intentionally exploring a problem through the use of a model and simulation. Systems thinking must be adapted to the specific dimensions of a design process or building in order to explore those relationships. 7 Michael Pidwirny, Definitions of Systems and Models, http://www.physicalgeography.net/ fundamentals/4b.html Chapter 2 Case Studies Fig. 2.1 Jean-Marie Tjibaou Cultural Center Fig. 2.2 Beijing National Aquatic Center Fig. 2.3 Abu Dhabi Performing Arts Center Fig. 2.4 Swiss Re Tower 14 Through analyzing the design process and work of others it becomes possible to see what characteristics shaped the design process and to understand how simulation tools were incorporated. The following four case studies share similarities in their design process such as building upon a personal or collective set of experience or the incorporation of passive heating and cooling strategies. However, distinctions can also be found in how they arrived at their final design whether mostly through an artistic concept related to past form or to a physical concept related to efficient performance. Each of the case studies depicts a different balance of the evaluative and generative approaches to the use of simulation. In the first case study, JeanMarie Tjibaou Culture Center, the Renzo Piano Building Workshop utilizes digital modeling and performance analysis for the purpose evaluating a design that had already gone through conceptual and schematic development. Peddle, Thorp & Walker in collaboration with Arup explored the use of the computer as a generative tool with a focus on performance analysis in their design for the Beijing National Aquatic Center. Generative tools were also used by Zaha Hadid architects in the design of the Abu Dhabi Performing Arts Center but for the purpose of exploring form in relation to concepts of growth. In the last case study, Foster + Partners combine the generative and evaluative approach into a cyclic process where morphology is analyzed based upon performance which then redirects the shaping of the form. 15 2.1 Jean-Marie Tjibaou Cultural Center Location - Nouméa, New Caledonia, 1991-98 Design Team - Renzo Piano Building Workshop, Arup, Agibat Engineers The design of this building is an example of an evaluative-dominated approach to simulation. The simulation tools were used for the purpose of validating a design that was nearly complete. Fig. 2.5 Site Plan of the built design Fig. 2.6 Local dwellings with post and beam structure 2.1.1 Project Description The design for the Jean-Marie Tjibaou Cultural Center began with a competition held in 1991. Renzo Piano Building Workshop proposed creating three functional “villages” arranged along a central spine. On the southern side of the spine are ten wooden “cases” reaching to a height of 92ft(28m) while on the northern side there are three clusters of flat modern “boxes” roofed in metal and glass. The vertical surfaces of the majority of the spaces are clad in wood and glass louvers, part of the passive cooling strategy. The wooden “cases” are the main focus of the design, drawing inspiration from the vernacular architecture of the local Kanak people who use natural materials such as palms and vines to construct their buildings.1 Piano’s “cases” abstract the local dwellings’ center post and spoke beam structure into a double-skinned shell structure made of iroko wood. The double shell is formed by a pair of laminated iroko ribs, one is curved and the other is straight, held apart by horizontal steel tubes with stainless steel cables diagonally spanning between members. The inner shell is Fig. 2.7 Wood ribs connected by steel members 1 Sara Hart, Double Indemnity, Architecture, Oct 98, p.152 16 responsible for carrying the load of the roof and creating enclosure. The outer shell ribs of each “case” are the framework for a series of horizontal iroko slats spaced closely together in the middle of the shell, but spaced more openly at the top and bottom to allow wind to pass through the structure. 2.1.2 Integration of Building Systems Fig. 2.8 Iroko wood structure and slats Fig. 2.9 Diagram of the building response under conditions with strong winds New Caledonia has a humid tropical climate with annual average lows of 64°F(18°C) and highs of 83°F(28°C) and humidity ranging from 60-90%. There is a monsoon season during part of the year with high force winds, heavy rain and the occasional hurricane. The wind typically comes from two directions, on and offshore. The intention from the beginning of the design was to use this wind to cool the building while also providing fresh air. The form of the double-skinned shells began as a reference to the local buildings, but later became the mechanism for controlling natural ventilation. The spacing of the horizontal slats on the outer shell were controlled to allow wind to pass through the upper and lower portions of the shells. Aluminum louvers on the inner shell coincide with these openings allowing for a variety of configurations to be achieved for the most efficient cooling of the building. The double-skinned shells are located on the southern side of the building enabling their vertical extension to shade the roof from direct sun, causing the air between the two layers of skins to heat up and rise out of the cavity. When the wind is stronger the louvers can be controlled to let wind create a negative pressure at the top of the shell, pulling air out of the building or the louvers can be opened to allow 17 air to pass directly through. 2.1.3 Evaluating the Design through Simulation After Piano’s competition entry was chosen the client agreed to pursue the use of natural ventilation as long as it could be proven that comfort conditions could be provided at least 90% of the year.2 The burden was then on Piano to prove that their ideas could work. Fig. 2.10 Wind tunnel analysis The Renzo Piano Building Workshop is well known for their model shop that strives to produce physical models that not only visually represent their design intentions but also function as close to reality as possible. For the analysis of the Jean-Marie Cultural Center a wood model was built with functioning louvers for wind tunnel testing as well as a digital model to be used for Computational Fluid Dynamics analysis. The initial analysis of the models was disappointing, showing insignificant changes created by the opening of the louvers. A hole was then added to the roof of the building, creating the desired cross ventilation.3 Piano was not interested in placing a hole in the roof and instead they began to explore where other openings could be placed to facilitate air movement. The final decision was to create a series of internal patios with operable windows. The lessons learned from these model studies led to a computer program that controls the opening of the louvers to achieve the desired air flow of 3.28ft/s(1.5m/s). Another concern at the beginning of the design was whether the wood shells, because of their size and distinctive shape, would be able to structurally withstand the high winds and occasional hurricanes that hit Fig. 2.11 Structural simulation analysis 2 Mark Chown, Building Simulation, 8th IBPSA Conference, 2003, p. 22 3 Sara Hart, Double Indemnity, Architecture, Oct 98, p.152 18 the area. This was studied both through the use of a wind tunnel and CFD modeling software, showing that the structure would be able to withstand the wind forces. Ove Arup Engineers and Agibat Engineering also built and tested a full scale prototype of the laminated rib structure. The results of this analysis were used to rationalize the complex curved geometry into a structurally feasible assembly that in the end varied little from the initial form. 2.1.4 Analysis of the Design Process Fig. 2.12 Transverse section of competition design Fig. 2.13 Transverse section of built design Renzo Piano is known for his sensitivity to site and context, and being able to blend this with his understanding of technology. Over the years his projects have explored this approach to architecture and consequently he has built up a large body of experience that allows him to apply his design intuition in a manner that generally makes correct assumptions in the early stages of the design process. This design intuition, shaped by a collection of experiences, forms general rules of thumb such as the principles of the size of openings and the distance between those openings necessary for designing a space that takes advantage of natural ventilation. However, these rules of thumb do not guarantee the desired performance levels but instead are only a starting point that can be strengthened through the use of simulation tools as was evident in the Jean-Marie Cultural Center. Throughout the design of the Jean-Marie Cultural Center there was a continual rationalization of the building form as can be seen in the differences between the section that was presented at the competition stage of the design and the section of the built design. The curved roof disappeared from the main exhibit space, the mechanical systems were 19 removed, the structural ribs changed in profile and number. In the competition entry these systems started as place holders or in terms of modeling in systems thinking they represent the initial action or current state. As development of the design continued, issues arose about the building being too literal a translation of the local structures. This critique was used as feedback in addition to the other feedback about the structure, ventilation, fabrication. All of this feedback was used to bring the model of the design closer to the desired state of which the final building is the closest representation. 2.1.5 Summary In this project the simulation tools were used mainly for evaluative purposes applied to a design that had already been conceived to a fine level of detail. Even though these tools were not used at the beginning of the design process, the concept of natural ventilation and integration of the building systems were included at the beginning of the design increasing their potential influence on the final design. These concepts were likely based on general rules of thumb and experience within the firm. Had these ideas not been incorporated in the early stages of the design it might have been difficult to integrate the strategies of natural ventilation into the final design. If the simulation techniques had been incorporated earlier into the design process there may have been a considerable amount of time saved because of not having to redesign or the design may have developed to a higher level of performance. A building or design process that is conceived holistically from the outset has a greater ability of avoiding or foreseeing issues that may arise later. 20 2.2 Beijing National Aquatic Center Location - Beijing, China, 2004-08 Design Team - Peddle Thorpe & Walker, Arup, China State Construction Engineering Co., China State Construction International Design Co. The design of this building is an example of a generative-dominated approach to simulation. The simulation tools were used for the purpose of generating the building structure based on initially defined parameters. Fig. 2.14 Rendering of Beijing National Aquatic Center (left) and Beijing National Stadium (right) 2.2.1 Project Description The Beijing National Aquatic Center, nicknamed the Water Cube, is one of the many buildings constructed for the 2008 Beijing Summer Olympics. When completed it will be the largest Olympic Swimming Venue ever built. Beijing hoped to show that the Olympics could be environmentally sensitive as well as a force to bring China into the realm of environmentally responsible development. While the debate over the sustainability of development on this scale is outside the scope of this thesis many of the buildings, including the Water Cube, have employed design strategies to cut the amount of energy and resources involved in their functioning. At the outset of the project Arup outlined to the architects what they hoped to achieve based upon their past experience with aquatic centers.1 Since swimming pools need to be heated they came up with a technical concept of turning the walls of the aquatic center into a greenhouse-like space that would act as a buffer to the cold Beijing winters, build up usable heat from Fig. 2.15 Plan of Beijing National Aquatic Center 1 Tristram Carfrae, Box of Bubbles, Ingenia, Dec 2007. pp 46 21 the sun’s radiation, and let in large amount of daylight. PTW Architects approached the project with a symbolic concept of the balance between the square and circle in Eastern culture with the Water Cube being the square and the neighboring Beijing National Stadium, the Red Bird’s Nest, as the circle.2 2.2.2 Integrating Building Systems Fig. 2.16 Ventilation diagram illustrating the ‘greenhouse’ walls Arup’s technical concept of creating a set of walls and a roof that behave like a greenhouse merged several of the buildings systems into an holistic system that has more advantages as an assembly than as individual pieces. The 3D steel structure creates a wall plenum, 11.8ft (3.6m) deep and a roof plenum, 23.6ft (7.2m) deep, that is used to capture heat between its two layers of skin, preventing most of the sun’s radiation from penetrating the interior. This heat can then be used for heating the swimming pools and the rest of the spaces, lowering the demand on the heater.3 Cooling loads are reduced because hot air is allowed to stratify and occupy the upper portions of the large volume, holding the cooler air down around the spectators. This hot air can then be evacuated from the space through the roof. The walls and roof also allow large amounts of daylight to flood into the space, potentially creating a 55% energy savings related to lighting.4 Other savings included not having to fireproof the 22,000 steel members because the structure’s complex geometry was shown through computer simulation to be able to withstand a worst-case fire scenario.5 In all it Fig. 2.17 ETFE skin panels of the Water Cube 2 Tang Yuankai, Don’t Burst My Bubble, Beijing Review, Aug. 2007, p. 28 3 Tristram Carfrae, Box of Bubbles, Ingenia, Dec 2007. p. 50 4 ARUP, The Giant Greenhouse, http://www.arup.com/australasia/feature.cfm?pageid=3491 5 Tristram Carfrae, Box of Bubbles, Ingenia, Dec 2007. p. 49 22 was estimated that the aquatic center would use 30% less energy in its functioning compared to a building of similar size, type, and location. Fig. 2.18 Process of extracting structure from block of Weaire-Phelan foam Like the skin of the body the skin of the Water Cube is critical to its high performance. The skin is made of ETFE (ethylene tetrafluoroethylene), a high performance plastic that was developed in the 1970s for the aerospace industry.6 The material is 1/100 the weight of glass, transmits light better, is a better insulator, and is recyclable. In addition to these features the ETFE protects the steel from the natural elements, avoiding costly maintenance and corrosion over time. This material represents the integration of properties usually attributed to different material but here the integration of these properties allows for the useA downside is that the cushions are inflated to keep them rigid, requiring an air pumping system to constantly maintain the internal pressure. 2.2.3 Generating the Design through Simulation The concepts for the Water Cube were developed early on in the design process but it took time and research to generate the building’s structural form. Initial attempts at defining the form lead to clumsy schemes involving the stacking of cylinders that left many of the details unresolved.7 Further research into how the 3-dimensional wall and roof structure might be designed led to the discovery of work done by two Trinity College physics professors, Professor Denis Weaire and his assistant Dr. Robert Phelan. The two professors were examining soap bubbles as a model for 3D spatial optimization. In the 19th Century, Lord Kelvin did research into the structure of soap bubbles leading him to the creation of a 14-sided shape, Fig. 2.19 Interior of the Water Cube 6 Elizabeth Woyke, Material for an Architectural Revolution, Business Week, Apr. 24, 2007 7 Tristram Carfrae, Box of Bubbles, Ingenia, Dec 2007. pp 46 23 the tetrakaidecahedron. A century before that the Belgian scientist Plateau, studied the structure of soap bubbles and developed a series of “rules for the way they join together in three faces forming a line”8 Weaire and Phelan were able to push their description of the soap bubble structure further than their predecessors because of the use of advanced 3D modeling. Fig. 2.20 3D model printed from the parametric model was used to illustrate structural concept to clients. The design team then took these findings and created a parametric script that could virtually construct a volume of Weaire-Phelan foam in any size that they required. From here they trimmed the virtual block of foam down into a square plan that referenced the traditional Eastern quadrangle courtyard.9 Interior spaces were carved out of the foam, leaving behind the bubbles that would make up the building’s structure. Changes made through the parametric model would automatically recreate the 22,000 member structure in roughly 25 minutes. The parametric model was developed to automatically size the steel members, trimming as much weight as possible to allow the roof to span the long distances. Physical models were also able to be three dimensionally printed directly from the parametric model. 2.2.4 Analysis of the Design Process The expertise that Arup brought to this project led to a design that strategically targeted the performance demands of an aquatic facility. However, the end product of the design was not simply an optimization of structure, but instead became an artful expression of lessons learned from a natural system. Natural systems typically hide the secrets to the way in 8 Tristram Carfrae, Box of Bubbles, Ingenia, Dec 2007. p 47 9 Tang Yuankai, Don’t Burst My Bubble, Beijing Review, Aug. 2007, p. 28 24 which they function within layers of complexities, making them difficult to comprehend. The design team did not stumble upon the structural form of soap bubbles through the use of the computer, but built upon past research that was slowly developing a model of the system by discovering fragments of information. A deeper understanding was made possible only recently when the use of computer simulation allowed for a detailed investigation into the complex structure of soap bubbles. Steve Pennell, who is in charge of the structural drafting and CAD program at the Sydney office of Arup, was quoted saying that “three years ago, computer power would not have been able to cope with the aquatic centre because of its complexity”.10 2.2.5 Summary The generative simulation techniques used in the Beijing National Aquatic Center allowed for the creation and optimization of a structural form that prior had never been built. Two concepts came forth early on in the design, one for the treatment of the structure as a greenhouse and the other as the image of the building as water, that were later unified to form the image of the building. Within this integration there were many other levels of systems integration that contributed to the building efficient performance in multiple respects. As simulation techniques grow they will offer new possibilities and directions that architecture can take. Things that had once seemed too complex to solve may become feasible through a designer’s collaboration with the computer. In order for this collaboration to be successful, the designer or design team must have a reasonable understanding of what the computer is capable of achieving. 10 Arup, Arup wins design award for Beijing’s National Aquatics Centre, http://www.arup. com/australasia/newsitem.cfm?pageid=3488 25 2.3 Abu Dhabi Performing Arts Center Location - Abu Dhabi, United Arab Emirates, 2007Design Team - Zaha Hadid Architects The design of this building is an example of a generative-dominated approach to simulation. The simulation tools were used for the purpose of generating the building form based upon growth simulation scripting. 2.3.1 Project Description Fig. 2.21 Rendering of the building viewed from the SE Since the 1960s Abu Dhabi has amassed enough wealth to transform itself from an area once inhabited by Bedouin encampments into a modern capital of hotels and high rises. Plans are currently under way for a 670 acre cultural district on Saadiyat Island that will showcase the work of many well known Western architects such as Frank Gehry and Jean Nouvel, and the Iraq native Zaha Hadid.1 Hadid has been commissioned to design the Abu Dhabi Performing Arts Center to be located along the main axis of the cultural district, reaching out into the Persian Gulf. At 62m tall, the building will hold “five theatres- a music hall, concert hall, opera house, drama theatre and a flexible theatre.”2 In a press release for the project, Hadid described the building as “a sculptural form that emerges from a linear intersection of pedestrian paths within the cultural district, gradually developing into a growing organism that sprouts a network of Fig. 2.22 Rendering of the building viewed along water 1 Nicolai Ouroussoff, A Vision in the Desert, New York Times, Feb. 1 2007 2 Marcus Fairs, Zaha Hadid in Abu Dhabi Update, http://www.dezeen.com/2007/02/02/ zaha-hadid-in-abu-dhabi-update/ 26 successive branches.” 3 2.3.2 Generating the Design through Simulation Fig. 2.23 Basic set of geometries from growth simulation The form of the Abu Dhabi Performing Arts Centre’s is based upon algorithms that simulate the flow of energy through natural growth systems, similar to branching structures of trees and vines. For the Performing Arts Centre this energy comes from the urban traffic along the pedestrian corridor and the opposing flow of movement of the site out towards the sea. Initially this algorithm was used to generate a set of basic geometries which gave the designers an understanding of what was possible to achieve with this tool. Programmatic diagrams were then overlaid on top of these formal investigations in order to give context to this growth. This was done repeatedly in response to design discussions centered around the aesthetics as well as the functional viability of a particular iteration. As these iterations continued to develop, architectural systems of structure, circulation, and glazing, were integrated into the model. The form of these systems were based upon the overall form that was created in the growth simulation 2.3.3 Summary Unlike the other case studies, this project has not yet been built so there is less available information. However, from the information provided, the possibilities of this approach to design begin to surface. The use of 3 Marcus Fairs, Zaha Hadid in Abu Dhabi Update, http://www.dezeen.com/2007/02/02/ zaha-hadid-in-abu-dhabi-update/ 27 the growth algorithm allowed the architects to generate a building form in response to a particular energy or force while continuing to develop other subsystems of the design. This growth of the building form was not automatic, but contained as much design intention as any other design decision. The architects were responsible for deciding that the use of a growth simulation algorithm was an appropriate concept for the design of a Performing Arts Centre. The firm likely had little experience with this approach to design so the project serves as a design experiment in order to test the validity of this approach. There is also the conscious or subconscious interaction of the subjective and objective decisions that occur within the design process, but which are not found in natural processes. The architects needed to find a balance between the desired aesthetics and the feasibility of the project in order to develop an achievable design. For this project it seems as if the growth simulation was being used for formal purposes, but it can be imagined that the energy involved in this simulation could actually be tied to other areas of performance such as lighting or thermal conditions and have a design iteratively shaped in response. The choice of what force to respond to is ultimately left up to the design team and their intentions for the project. Fig. 2.24 Iterations from interaction of growth simulation and programmatic diagrams 28 2.4 Swiss Re Tower Location - London, UK, 2004-08 Design Team - Foster + Partners, ARUP The design of this building is an example of a evaluative-generative approach to simulation. Preliminary concepts suggested a direction that was then analyzed with simulation tools. This feedback reshaped the concept and the cycle repeated until a final design was achieved. Fig. 2.25 CFD study illustrating stresses induced by wind loads (red indicates highest stress) Fig. 2.26 Diagram depicting the building form concept that was arrived at after CFD analysis 2.4.1 Project Description Office space demands large amounts of fresh air, cooling, and light. At the same time, office towers require large amounts of structure to resist the increased force of wind experienced as they rise higher from the ground. For the Swiss Re Tower, Foster + Partners worked to address this set of demands with a design whose pieces take on multiple functions. Their work built upon ideas first developed with Buckminster Fuller in the 1970s for the Climatroffice; a free-form glass skin building with its own microclimate.1 The Climatroffice was never built because of the limited technology available to support the complex geometry. With the Swiss Re, the first few months of the design were spent exploring the most “efficient structure for the site” studying, with Computational Fluid Dynamics modeling, the impact that various forms would have on the loads generated by wind forces. By the end of the analysis the design team was looking at a tower that was circular in plan to make it more 1 Unknown, Swiss Re, Architect’s Journal, Apr. 15 2004, p. 65 29 aerodynamic and would use natural forces to provide ventilation. As the design progressed the form was constantly refined in response to simulation feedback, space layout and ideas related to the building fabrication. 2.4.2 Integration of Building Systems Fig. 2.27 6th floor depicting the triangular voids located between the offices ‘fingers’ A floor plate taken from the Swiss Re demonstrates the extent to which various building strategies were incorporated into the overall form. Each floor, circular in plan, has six radial fingers separated by triangular voids that allow air circulation between levels and bring in added natural light. Floors are rotated ten degrees from the floor below, creating a set of spiraling thermal chimneys that “tap in to the building’s pressure differential rather than just relying on the stack effect”.2 Fresh air is brought in through the chimneys, warmed by the sun and delivered into the offices in the winter. In the summer the chimney can be opened to the outside and the offices opened up to the chimneys causing warm air to be pulled out. From the exterior, the chimneys are evident with grey-tinted glass that has a highperformance coating to reduce solar gain. These strategies were predicted to lower energy usage by 50% compared to a traditional office building. Another benefit of the circular foot print was opening up the site for a public plaza. Normally these types of spaces are inhospitable because of the turbulence created by the wind hitting a flat facade. The circular plan causes wind to flow around the building, reducing uncomfortable air flow. The tower is tapered in section from the center towards the top and base, permitting more light to fall on the public plaza. This tapering of the tower alleviates the imposing experience of being next to a tower by making the Fig. 2.28 Spiraling thermal chimney 2 Austin Williams, Round Peg in a Square Hole, Architect’s Journal, Sept. 26 2002, p.30 30 middle of the tower appear to be the highest point. The building’s structure follows the same rotation of the floor plates, forming a diagrid structure that laterally braces itself and thereby frees the center of the building from needing a dense structural core. The diagrid is further broken down with window mullions that occur every five degrees. Curved glass panels would have been prohibitively expensive to fabricate so instead 5500 diamond-shaped, flat glass panels of different sizes for each floor were used to break down the building’s complex geometry. 2.4.3 Generating & Evaluating the Design through Simulation Fig. 2.29 Aerodynamic building form causes the air to travel is an smoother path compared to a typical tower Fig. 2.30 Diagrid glazing system and building structure that has not yet received aluminum covering Beginning design studies focused on exploring what form would best perform on the site in relation to functional concepts about daylight, natural ventilation, and wind loads. Concepts were evaluated using CFD modeling, which took into consideration surrounding buildings and the weather data for the site and then applied this information to a variety of forms. Generated feedback was in the form of false color graphics and animations that showed the varying level of stresses that the building would experience from wind loads. That information combined with feedback of the amount of turbulence created from the wind hitting the face of the building further shaped the form of the building. The building form was also influenced by aesthetic considerations, particularly because of the fact that there are few towers within London. Once the general form had been shaped by this analysis the design team moved into fine tuning the geometry. For this part of the project they created a parametric model that controlled the overall form of the building 31 as well as the structural and mullion geometry. The design team did not immediately happen upon the final design but instead there was a progressive development of the project whereby one aspect of the project would be informed by the previous analysis, a design decision would be made and further simulation carried out. Use of the parametric model made this process efficient because the model could automatically be regenerated in response to changes input into the parameters. Fig. 2.31 Physical model used to mock up preliminary ideas about the structure and mullion order Parametric modeling also made it possible to rationalize the complex curved geometry, something that had previously prevented Foster + Partners from being able to carry out their related Climatroffice. The parameters are stored in a format similar to a spreadsheet which allows the model to be automatically regenerated. The spreadsheet was also sent to the fabricator who could then use those numbers to generate a set of shop drawings that Foster + Partners would be able to check against their original model. Tubular steel members were able to be designed with zerotolerance because of this precise method of form generation. 2.4.4 Analysis of the Design Process Foster + Partners came to the design of the Swiss Re Tower with concepts that they had previously conceived in collaboration with Buckminster Fuller. Through the use of technology they were able to finally realize the potential of these concepts by incorporating simulation early on in the design process. Simulation did not drive the design process though. Concepts that were initially brought to the design shaped the approach of the simulation analysis which in turn reshaped those original concepts and occasionally generated new concepts. This cyclic process of moving 32 between the concept and simulation continued throughout the design. A similar type of resonance occurred between the concepts of worker comfort and optimization of building form according to site qualities. An increasing understanding of the building’s response to wind loads suggested ways that natural ventilation and daylighting could be integrated into the building form. As this logic was followed through, other systems in the building began to fall in line with these concepts such as the structural system or even the aesthetic of the tower’s elevation. In the end a design was reached where nearly all of the parts build off of the same framework, integrating these various concepts and systems together. 2.4.5 Summary Through the integration of the evaluative and generative approach to design Foster + Partners were able to design an office tower that integrated its many building systems together. Early on it might have been unclear how the building would be informed by the initial concept of creating the most “efficient structure for the site”, but as the design progressed each of the systems grew around previous decisions. This was made possible by generating an idea, evaluating it, and then regenerating in response to the evaluation feedback, similar to the modeling loop previously discussed. Foster + Partners have a history of technologically influenced design that continues to evolve over time. This experience paired with simulation as both an evaluative and generative tool, shortens the time frame for achieving a deeper understanding of a building and its systems. Validation no longer needs to wait until construction, but instead lessons can be learned during the design, increasing the amount of knowledge that one design can contribute to the designer’s overall experience. Chapter 3 Simulation Software 33 When the computer first entered the architectural field it was used to replace the mechanical drafting that was previously done by hand. The method in which these digital drawings were created had few differences from their predecessors except that updates were thought to be more easily made. During its first few decades the computer did not significantly change the architectural field, but the technology continued to evolve. Digital two dimensional drafting, led to three dimensional modeling and then to rendering, and more recently there has been the development of software with the capability of going beyond three dimensional representation to include form generation, performance analysis, control of fabrication, and integrated documentation. Software may address dimensions of form, performance, fabrication, or documentation, but each piece of software will likely have strengths in a particular dimension and be weaker in others. This thesis focuses on software that address morphology and evaluative analysis. While these two types of software perform a particular set of functions within the design process, in the end there is the need for integrating both sets of tools into an iterative design process. 34 Sketches/ Diagrams 3.1 Digital Morphology Plan Input Section Input Elevation Model Fig. 3.1 Plan, section, elevation explored in serial sequence in the traditional design process Building form is most commonly explored and conveyed through two dimensional drawings and models. A design might start with a plan which then informs the production of a section, then elevation or back to the plan. All of these are then used in conjunction to create a model. Separation of the two mediums in the design process creates a cyclic relationship where either the drawings or models suggest the form of the other whose development then informs the regeneration of the initial media. A digital design process might still begin with hand sketches but a digital model allows for concurrently performed tasks that explore three dimensional space on a continuous spectrum of scales within the same model rather than the two dimensional, serially performed tasks traditionally used. 3.1.1 The Element of Geometry Design Tasks Model Fig. 3.2 Design tasks such as morphology, documentation, analysis, and representation are performed in parallel If geometry is treated as a system then its basic element is the point. Two points are connected through a relationship that defines a line. The attributes of the points are their coordinates defined relative to an origin. Traditional drafting uses the point for building up geometry as dimensions are laid out, lines are struck between them leading to the creation of shapes that represent a model of a building. Physical modeling references these points, lines, and curves, but is composed of 3D surfaces and volumes. Digital modeling is a combination of the two approaches, using both points, lines and curves, and shapes, surfaces and volumes. Digital 3D modeling software begins with the basic element of geometry, the point, and from there different software allows for varying levels of interaction with the geometry. 35 Fig. 3.3 Polygonal surface created in Sketchup Fig. 3.4 NURBS surface created in Rhino is smoother than the surface created in Sketchup point cloud ribs tubes Fig. 3.5 Point cloud created in Grasshopper can be used to define multiple sets of relationships between points The most basic interaction between points can be illustrated with Google’s 3D modeling software Sketchup. All geometry created in Sketchup is based upon the line. Points are not created in isolation of other points but instead become the defining attributes of lines. When a line is intersected with another line the point relationships are reconfigured, splitting the original line at the intersection into two lines. In Sketchup a circle is actually an approximation of a smooth curve broken up into a user-determined number of segments. This approach to 3D modeling is known as polygonal modeling where all shapes are built from line segments. In contrast, other software like McNeel’s Rhinoceros uses a system of geometry definition known as NURBS (Non-Uniform Rational B-Splines) to create lines and curves based upon a more complex set of relationships between points. With NURBS, lines and curves are defined by an evaluation rule that relates the degree, control points, and knots of a line or curve. The focus here is not on what each of these variables does but instead is about the level of accuracy with which Rhino can model complex surfaces and curves. These digital models can then later be directly used for form analysis such as an examination of a surface’s curvature or be used to drive Computer Numerically Controlled (CNC) machines for fabrication purposes. Another approach to handling points can be illustrated with the Rhino plugin, Grasshopper developed by David Rutten for McNeel Software like Grasshopper uses a spreadsheet or database of points, known as a point cloud, to define a surface in a variety of ways. These points can be edited individually or rules can be applied to a selection of points or the entire set to define the relationship for how the points interact such as creating lines, 36 Systems Object curves, or surfaces. Grasshopper also uses the relationships between four adjacent points to define a cell inside of which additional geometry can be positioned. 3.1.2 Creation of Systems through Geometry Range of Scales Components/ Groups Surfaces Lines Elements Fig. 3.6 Example of a geometric system Points After drawing a set of lines and surfaces a series of systems begin to emerge within a model. As one learns to use the software and becomes comfortable with digital modeling the distinction of these systems becomes apparent in the organization of the model through the use of: a) Groups - a collection of elements where changes to a copy do not effect other instances. b) Components/Blocks - a collection of elements where changes to a copy effects all instances. Groups and components may begin as placeholders in the overall model but as the understanding of the project becomes more elaborate so do the systems and elements contained within the groups and components. Each program may label these classifications differently but the use of groups and components has the ability to organize the building into discrete systems that can be altered in isolation or in reference to the rest of the building. In physical modeling and drafting, interactions between systems are analyzed through a layering process of redrawing or modeling separate pieces upon each other. A disadvantage to this approach is that the degree of influence between systems is limited because of the narrow view that is provided by a two dimensional plan, section or elevation. Digital modeling on the other hand allows not only for viewing these systems three dimensionally but also for switching between views of the entire building or isolated systems. 37 A more thorough understanding of the geometry of the systems is possible through the use of sub-models of sub-systems that can later be reintegrated into the overall model. Visual simulation feedback can be quickly achieved by turning on previously hidden systems and viewing them against the surrounding context. With this approach there is the ability to work with the whole building at once or at the scale of smaller systems before seeing how particular changes effect the whole building. A cyclic process is created similar to the additive layering that occurs with physical drafting and modeling but the digital model has the potential to receive a larger range of feedback than its physical counterpart. 3.1.3 Form Generation All digital modeling controls geometry through the use of parameters which are defined as “a numerical or other measurable factor forming one of a set that defines a system or sets the conditions of its operations.”1 In the case of geometry, a parameter could be variable like the position of a point, the diameter of a circle, or the color of a surface. Different software allows for different levels of interaction with the parameters. As an example Sketchup allows parametric control at the level of a line, while Rhino allows parametric control at the level of a surface, and Grasshopper allows parametric control of the all geometry. With this increased control comes an increase in the number of decisions and steps that are necessary to model a form causing a difference in the level of complexity that different software is capable of producing. 1 Oxford American Dictionary 38 Software that allows for more direct control and manipulation of these parameters has recently been gaining popularity. This control was initially achieved through scripting which is expressing a set of instructions and relationships for form generation in the code language of the software. Scripting can be used for defining already conceived forms or unknown forms can be generated from the relationships defined between parameters such as fractal or growth patterns. Capabilities of tools available in the original software are written through scripting so when a designer writes scripts they are customizing the software and are creating tools that perform a desired task. As scripting started to gain attention, software was created to bypass the need to learn the multiple scripting languages and instead a graphical interface was used that allowed for the precise control of parameters. Grasshopper is an example of logical modeling software. Relationships between parameters are based on mathematical formula that can control translation, scaling, rotation, and many other transformations of the geometric relationships. One key to the use of this type of software is the understanding of geometry and how to manipulate geometric systems. Without this understanding one can become paralyzed when faced with the long list of operations that are possible. 3.1.4 Experience with Geometry Just as the limits of mechanical drafting and physical modeling determine what can be drawn or built so does an architect’s understanding of geometry and how it is manipulated. Borromini and other architects of the Baroque period were able to create and build amazingly complex forms 39 using only 2D mechanical drawing and physical modeling, tools that were available to their predecessors but whose understanding of geometry was not available. Developments in the understanding of geometric systems throughout history have brought along with them changes in other fields such as architecture. This constantly evolving understanding can be seen in contemporary architecture where the computer has opened up new possibilities for how geometry can be manipulated and defined, leading to buildings whose forms explore this understanding. Fig. 3.8 San Carlo alle Quattro Fontane - Borromini Fig. 3.7 Camp Nou stadium by Foster+Partners A contemporary example would be Foster + Partners who are exploiting the ability of the computer to aid in the generation and rationalization of the complex forms. Forms similar to those that Foster + Partners are exploring have been proposed in the past in lesser detail but in the end not built because of a limited understanding of how to describe the geometry. Without the expertise they have developed through their Specialist Modeling Group, who are responsible for developing software within the firm, many of their buildings would not be possible. 3.1.5 Summary Modeling software is a critical piece in the design process because forms that are created are in part generated by the definitions associated with the tools of that software. For instance, in Sketchup a surface can be extruded perpendicular to the plane of the surface using the Push/Pull tool but can not be extruded in any other direction. This limitation is written into the software for how that particular tool operates. The designer learns these constraints over time and adjusts their design decisions accordingly. Distinctions between modeling software leads to different possibilities for 40 what can be formally achieved. As architects begin to learn a particular software the tools that are available to them open up new opportunities that might have previously been unattainable, influencing not only what they are capable of achieving but how they think and approach design. Experimentation with software leads to an experimentation with form in an attempt to understand what a particular tool is capable of contributing to the design process. This experimentation might try to represent abstract concepts such as time or movement through the form of the building or the building might be shaped to portray the impact of invisible, nonphysical forces. Much of this initial experimentation through scripting, trial and error, or chance will yield few applicable results in the beginning. However, over time experience with the tools will focus one’s approach if the experimentation is directed by a set of design goals that have developed in response to the capabilities of the tools that are being used. Formal experiments can lead to new considerations of what form can do other than changing the building’s aesthetic appearance such as investigating how structural stiffening can be achieved or manipulating a building’s form and orientation to maximize solar response. In order for this to occur, an architect must be able to analyze the building’s performance under these conditions. 41 3.2 Digital Performance Analysis Digital morphology can be considered the subjective side of the design process though the formal characteristics of a design can be based upon objective information, but where then does this information come from? Architectural design strategy books like Brown and DeKay’s Sun, Wind & Light present rules of thumb that designers can reference as starting points for their design decisions, unfortunately, they do not predict anything about the building’s actual performance. The danger with rules of thumb is that they can be applied in the wrong way without realizing deficiencies until the project is constructed. Another limitation is that they filter out possible solutions before their validity was ever tested in the context under investigation. This is where the use of performance analysis software can help to strengthen rules of thumb, eliminate the adoption of irrelevant rules, or inspire other appropriate unthought of solutions. There still exists though the possibility of applying the tool of performance analysis in the wrong way, creating meaningless results. Fig. 3.9 Diagrams from Sun, Wind, & Light depicting rules of thumb for solar and wind orientation Physical models are often used to simulate the performance of a building. These models and simulations are generally time consuming and inaccurate because of problems with scaling the data. An example would be the use of a heliodon to simulate daylighting levels but problems occur because of the approximation of materials and quality of light besides which the heliodon takes up a large amount of physical space and can be cumbersome to work with. Digital performance analysis creates a digital model whose geometric parameters are assigned additional parameters related to the properties of materials. Like physical modeling, the digital model is only as good as the definition of the material properties. Here the advantage is that a material 42 can be studied and quantified, and these physical parameters can then be attached to the digital model. Digital performance analysis looks at buildings as cascading systems concerned with the flow of energy between systems and elements. These flows of energy can arise from forces like the sun, wind, sound, and gravity. In order to analyze the impact of these forces there needs to be an initial understanding of what elements or systems they effect, and then what parameters or attributes of the element or system can be controlled in response to these forces. 3.2.1 Analysis of Psychometric Parameters Psychometric parameters address the relationship between natural and artificial forces that effect human senses such as lighting, acoustics, heating and cooling. Research in the field of psychometrics has established favorable performance levels for these parameters in terms of human comfort. These levels act as the baseline against which the results of performance simulation of a design is compared. The forces that are commonly tested are: a) solar - sun path, solar access, solar radiation b) thermal - heating/cooling loads, occupancy loads, wall construction c) acoustical - form of space, materials d) lighting - shadow studies, lighting levels, glare There is a certain amount of overlap between these different forces and the parameters that they effect. For example, solar analysis will also play a part in understanding the thermal forces as well as lighting which creates the 43 need for being able to concurrently analyze each of these forces. One of the better programs for performing this type of analysis is Ecotect, developed by Square One. For each of these parameters Ecotect is capable of producing graphical illustrations of analysis results, easing understanding of the results. Performance analysis of different parameters requires different considerations to be made when creating the digital model. Performance analysis models are usually a simplification of a formal model, removing elements that do not effect the parameters being tested. Instances where analysis is dependent mainly on geometry such as solar, shadow studies, and early lighting analysis, a model can be imported directly from another piece of software. The geometry will need to be assigned material parameters in the case of lighting analysis, but the relationships between the geometry do not need to change. When analyzing thermal or acoustic forces, which are both effected by volumetric parameters, there is the need to create models that define enclosed spaces referred to as zones. These models can be rather abstract representations of the building using only surfaces to define spaces. Each surface has parameters that define the materials and layers that might make up a wall, ceiling, or floor. Fig. 3.10 Examples of graphical illustrations produced by Ecotect. Lighting analysis (top) Shadow analysis (bottom) 3.2.2 Experience with Analysis In order for analysis to be accurate and effective the software user must have background knowledge about the factors that they intend to analyze. Without a prior understanding of how environmental factors effect a building’s performance the simulation has no basis for investigation. Starting with general rules of thumb allows for the creation of a set of hypotheses that can then be tested. These simulations produce 44 feedback that progresses the design, narrowing the field of options that are acceptable. In some instances an architect may be unsure about the feasibility of an idea but if there is knowledge of what should be tested for and how to perform that analysis then there is the possibility for discovering previously unthought of solutions. An architect’s understanding will continue to be strengthened over time leading to reformulated rules of thumb and criteria for what type of simulation is most appropriate for the given context. Performance analysis can have the largest impact when in is integrated into the earliest stages of the design process. In the essay, The Digital Design Ecosystem, Paul Seletsky describes this approach as pre-rational design where one is “using advanced computation to impart tacit and explicit experience into the earliest stages of conceptual explorations.”1 Design decisions can be tested to uncover their performance implications, preventing unexpected issues that may arise later in the design process due to delayed analysis. If performance analysis is delayed until the later stages of design most design decisions have already been developed to a point where redesign would be both time and cost prohibitive. In order for these tools to be integrated into the initial stages of design they need to closely interact with the tools involved in digital morphology, providing feedback for possible future directions. 1 Paul Seletsky, The Digital Design Ecosystem: Towards a Pre-Rational Architecture, AECbytes Viewpoint # 37, Apr. 8, 2008 45 3.3 Iterative Design Process Coupling of digital morphology and performance analysis leads to a process that expresses principles of emergence, the sum of the parts containing properties not found in any one part. Parameters for form generation can tend to be purely aesthetic considerations, but relationships can be created between subjective and objective parameters of design. In this case the parameters become the elements of a system and the relationship between them are defined by formula. These formula can take on the objective characteristics of a mathematical model, equating variables in relation to each other or a subjective approach of intuitively balancing between morphology and performance analysis. In either case the relationship that is defined sets up a hierarchy of importance with regards to the parameters. In an iterative design process the computer does not drive generation but instead facilitates the understanding of the complexity of relationships between parameters. A designer subjectively balances the influencing forces of a design, though one will likely dominate over the others. Project type, context, and concepts generated by the designer suggest possible interactions between these elements and determine what is of most significance to the final design. A museum will undergo different considerations because of its role as a cultural symbol and a place for viewing art whereas an office design would be driven more by performance and employee comfort, leading to a difference in the type of simulation that is appropriate. 46 3.3.1 Adaptation of an Iterative Design Process Project differences also have implications for when the analysis of the various parameters are integrated into the design process. A school because of its large number of rooms and scale of the spaces would likely not need in-depth structural analysis at the beginning of the design process. However, because of the tasks that take place within a school, early analysis of psychometric issues of lighting, thermal, and acoustic values would likely be most beneficial. On the other hand, the design of an enclosed stadium with a long span roof and the loads associated with the crowd of spectators would benefit from early explorations into the structural performance of the design. This reordering of tasks and tools within the design process is something that improves with time and experience. Flexibility of the design process allows this evolution and the feedback associated with it to constantly reshape the design process. Work that occurs early in an architect’s career may be completely different with regards to the process that is used at a later point in time. The development of the architect occurs through the constant reformulation of the design process in response to particular projects, the technology of that time and the accumulated experiences over one’s career. There is not necessarily an ideal process that one is working towards because the constraints and elements of the process are in constant flux. The goal instead is establishing a means of analyzing and adjusting the design process. 3.3.2 Limitations of Iterative Design Process 47 Currently communication between different software can be time consuming or not possible making it difficult to integrate these various tools that they each offer. As the use of the software continues to grow so will the support and development industry, establishing common formats for more efficient model exchange. There will likely still exist a difference in the way in which a model is built in response to the analysis of either morphology or performance. This can either be overcome with individual software taking on more tasks or breaking the modeling process into a set of models where each is used for a different type of analysis. The approach of breaking the a building into separate models is the current practice. Similar to the serially performed tasks of moving from plan to section to elevation, the process of studying the form of the building through one model requires then using that model to inform the development of a model to be used for performance analysis. This can lead to errors in translation of the model and large amounts of time that are spent rebuilding that model. More comprehensive software is beginning to be developed where performance analysis is incorporated directly into the morphological software through the use of plug-ins. While this may allow for the potential use of the same model there is still the fact that the model must be conceived of and constructed in a way that allows analysis to be performed upon it. These considerations have slight implications when thinking about creating a model for lighting analysis but have further reaching implications when a morphological model is to be used for thermal analysis. Given the current tools it becomes the task of the designer to balance these pros and cons when first planning how to go about constructing the building model. Chapter 4 Past Projects 48 Before attempting to integrate the parametric modeling and evaluative analysis tools into a design process there must first be a baseline understanding of what these tools are capable of. In order to develop this experience, each tool was applied, in isolation, to a previously completed studio design project. By removing the act of design and replacing it with the act of reconstruction more attention can be paid to the way in which the tool functions. Also by examining the tools in isolation of each other, the strengths and weaknesses of each can first be identified before attempting to balance their integration with each other as well as other pre-existing design tools. Both projects are taken from previous studios project in the University of Washington’s Master of Architecture program. 49 Study Room 4.1 Wallingford Library - Winter 06 Reading Area Stacks Study Room Fig. 4.1 Second Floor Staff Offices Computer Area Fig. 4.2 Ground Floor Fig. 4.3 Section Perspective Meeting Room Children’s Stacks This was the second studio project during the first year of the three year Masters of Architecture program and was the first project where I used digital modeling from the beginning of the design process. Throughout the process of designing and modeling this project there was a focus on the control and quality of light within the central library space, which was explored mainly through the use of shadow simulation within SketchUp. While these studies helped to direct the path of the design, in hindsight they may have failed to acknowledge many of the other important qualities related to natural light such as contrast, glare, solar radiation, and appropriate lighting levels for specific tasks and ignored other experiential qualities of space such as thermal, and acoustical. This project is being revisited for the purpose of applying evaluative simulation tools, specifically Ecotect, to understand the implications of past decisions that were based largely upon intuition. Another purpose is to explore how the information provided by these evaluative tools can be used in shaping the future designs. 4.1.1 Project Description The program for this project was to design a community library in the center of the Wallingford neighborhood of Seattle. In the final design, the service areas which were the work spaces, meeting room, bathrooms, computer area and offices were sheltered by the reading spaces and stacks which were located adjacent to the roof. From early on in the design the idea was to create an internally focused environment within the library, 50 filtering in views of the outside while bringing in light from above. This concept came to be represented by the idea of trees being bent over a raised platform, diffusing light into the interior of the central library space. A series of scattered skylights were used in the attempt to create this effect. The skylight placement was based on formal composition and a loose idea of concentrating the direct light in circulation areas that would receive less visually critical activity so that the light would then diffuse into the reading areas. 4.1.2 Rebuilding the Base Geometry Fig. 4.4 Lighting model of surface geometry Fig. 4.5 Thermal/Acoustic/Wind model of zone geometry Before the analysis of the previous design could begin it was necessary to reorganize and in some instances rebuild the original SketchUp model. Ecotect, an evaluative simulation software, performs two different methods of analysis, with each having a different set of geometric requirements. For analyses related to lighting the model needs to represent the geometry and its spectral surface characteristics, which are assigned in Ecotect. The more visually detailed this model and the more representative the surface characteristics, the more accurate the lighting and shadow analysis will be. Thermal and acoustic analysis requires a model that is divided into separate zones, representing different areas of enclosed space such as bathrooms, offices, or the meeting room. The central library space can be represented as one large space because it is open between the different areas. These models require a lot less geometry but the geometry must be much more precise and create a zone that is “water tight”, meaning each zone must be closed along all edges. Failure to achieve a “water tight” model will most likely lead to inaccurate results. 51 4.1.3 Analysis of Sun Penetration Early sketches in the design looked at how the form of the apertures could be shaped to possibly bounce light into the space but this idea later receded to the background because of the difficulty in understanding if these solutions would actually work. Instead, a simpler approach of using SketchUp to simulate how sunlight would travel through the space was used. This exercise was useful through out the design process and I now frequently find myself performing this task during design to quickly get an idea of the range of movement of the sun but it has its limit. While the time of day and day of the year could both be controlled within SketchUp, it was only possible to visualize a single point in time, making it difficult to get an holistic idea of the sun’s path. The use of Ecotect allows for the visualization of shadows over the course of an entire day, highlighting the sun’s path and areas that will experience the most direct sunlight. This can be seen in the plan diagrams on the following page, which were created for the summer and winter solstices, and spring and fall equinoxes (both produce almost identical results). The differing sun altitudes, high in the summer and low in the winter, and corresponding rise/set points of the sun, north of E-W in summer and south of E-W in winter, are depicted in the sunpath diagram. These images show that several of the areas experience large amount of direct sunlight such as in the stacks which would likely cause damage to the books, and in the reading area, possibly creating visual discomfort due to glare and large amounts of contrast. Study Room June 21st Reading Area Stacks Study Room September/March 21st December 21st Fig. 4.6 Shadow studies and Sunpath Diagrams for the 21st of June, September/March and December 52 53 4.1.4 Analysis of Lighting Levels Fig. 4.7 First Floor - March 21st at 12pm Fig. 4.8 Second Floor - March 21st at 12pm Fig. 4.9 No Skylights - March 21st at 12pm To get a better idea of what the natural light values are, measured in lux, a lighting analysis of the central library space on the second floor was done using the geometry from Ecotect and then processing the analysis through Radiance because of its more sophisticated lighting analysis algorithms. The lighting analysis starts in Ecotect after material properties have been assigned, followed by the creation of an analysis grid of the area to be tested. For this study the analysis grid was set to the footprint of the building but a smaller grid could be set to do detailed studies of individual rooms. Once the grid is established the geometry and grid are exported to Radiance where the lighting level calculations take place, and when complete are imported back into Ecotect and displayed in the images to the left and on the next page. Suggested lighting levels within a library are 540-810 lux for a reading area and in a computer lab suggested levels range from 160-320 lux. The results from the lighting analysis show that the levels within the reading areas are 600-1200 lux during December but can exceed 3000 lux in June, well above the suggested levels of 540-810 lux. When the option of having no skylights was analyzed it showed the natural light provided by vertical apertures would not be sufficient to avoid the use of supplemental electrical lighting. The large illumination difference between June and December suggests that even though the skylights may need to be slightly smaller they must also be capable of adjusting to the seasons by either a static or dynamic means. These results in addition to the results of the shadow analysis show that the skylights allow large amounts of direct sunlight into the central library space, making it necessary to prevent overheating due to solar radiation. 54 Fig. 4.10 Ecotect Lighting Study for June 21st at 8am, 12pm, 4pm Fig. 4.11 Ecotect Lighting Study for March 21st at 8am, 12pm, 4pm Fig. 4.12 Ecotect Lighting Study for December 21st at 9am, 12pm, 3pm 55 4.1.5 Analysis of Thermal Performance Fig. 4.13 Wall section layer properties in Ecotect Fig. 4.14 Zone thermal properties in Ecotect For the thermal analysis it was necessary to use the zoned model and to obtain weather data for the city of Seattle, WA which Ecotect then uses to calculate the effect of the weather on the building’s internal conditions. Whereas the lighting model was mainly concerned with a material’s surface characteristics, the thermal model requires the surfaces of the model to be given wall section layers which define the thermal and acoustic responses. In addition to the material properties, the properties of the zones must also be defined which include the type of HVAC system, the comfort band (range of accepted temperatures), the occupancy and schedule of use, internal gains from lighting and electronic equipment, and the infiltration rate of air between a zone and the outside. For the initial calculations the HVAC systems were set to NONE in order to get an idea of how the building would respond to the environment. Occupancy estimates were made for the main library spaces that would receive the most use such as the meeting room, offices, and reading areas. The rest of the properties were left at their default levels to initially control the fewest number of variables as possible. Once these settings were input then an interzonal adjacency was calculated, determining the influence that each zone has on another due to behaviors like thermal bridging or solar shading. The first analysis looked at the internal hourly temperatures to get a sense of how the building was responding to the environment. The following page contains images depicting the hottest day of the year, the coldest day and March 21st as a middle ground. In these graphs, the thick band of multi-colored lines represents the all of the zones’ interior temperature fluctuations with each zone represented by a different colored solid line. The dashed and dotted 56 Fig. 4.15 Internal (hourly) temperature on hottest day of year Fig. 4.16 Internal temperature on March 21st Fig. 4.17 Internal temperature on coldest day of year lines represent the environmental conditions such as outside temperature, solar radiation, wind speed. The red and blue bands indicate the upper and lower extents of the comfort band which falls within the black. From looking at these graphs and the data for the rest of the year it became apparent that the interior of the building hovered around the average of the daily minimum and maximum temperatures. By adjusting the building materials these curves can be subtly shifted because of each material’s different U-value, which is a measure of how well heat passes through a material. During design, this type of analysis could be used to study the thermal benefits of different materials as well as investigating the impact of alternate orientations or aperture locations to limit the impacts of direct solar gain. After studying the building without any mechanical systems, HVAC systems were assigned to the different zones and the heating loads for the entire year were calculated. The loads are displayed in a vertical bar graph where a bar for each month is divided up into segments that represent the percentage of energy use for each zone. It is apparent from the results shown on the following page that heating is the dominant load throughout the year with the central library space requiring the most energy. Within a design process, frequently repeated use of this type of analysis would show how the performance of the design was changing and hopefully progressing. Similar to the internal hourly temperature analysis, the heating/cooling load analysis has the potential of showing how the choice of certain materials could have energy implications. A situation could be imagined where a more expensive, high performance material is tested against a less expensive, conventional material in order to see whether the cost difference would be offset by the energy savings. 57 Fig. 4.18 Hourly Internal Temperature (zone colors blue-central library space, green-storage, orange-meeting) Fig. 4.19 Monthly Heating/Cooling Loads (zone colors blue-central library space, green-storage, orange-meeting) 58 In order to better understand how to improve the design there needs to be an understanding of the thermal weaknesses and strengths of the building. The graph of the Passive Gains Breakdown shows the gains and losses that create the need for heating/cooling loads as a percent of the total energy usage. There are six variables that are being tested for in the Passive Gains Breakdown; Conduction, Sol-air, Direct Solar, Ventilation, Internal, and Interzonal. It is not the intention of this thesis to explain what each of these variables are but instead to show that major weaknesses of a design can begin to seen through these graphs. In the case of the library, conduction accounts for over 64% of the total energy loss, which again could be mitigated by choosing a material with a lower U-value. Fig. 4.20 Passive Gains Breakdown 10 12 14 16 18 20 22 hrs watts 26000 4 6 8 0 2 -26000 Jan Mar May Fig. 4.21 Ventilation gains July Sept Nov Each of these variables can then be examined in closer detail by producing a graph that averages the energy use per day for each month. These images offer a visual means of examining the complex relationships that occur between variables. For example on the following page there are graphs for direct solar, building fabric, and interzonal gains. In the building fabric and ventilation graphs the red areas represent near zero gains because the outdoor temperature is within the range of the comfort band due to the solar radiation that occurs during those months as can be seen in the direct solar graph. This then leads to the interior of the library not needing heat during the day so the interzonal gains are negative, the blue patch in the interzonal graph. Spaces with larger amounts of activity are giving off or losing heat to the other zones while during the evening hours and into the night these areas begin to absorb heat, the two yellow patches in the interzonal graph, because temperatures have cooled off outside. Analysis with the HVAC system set to none would make it possible to see how thermal lag of materials effected diurnal temperature swings. All of the 59 Fig. 4.22 Gains Breakdown watts watts 47000 1600 0 0 0 4 6 8 10 12 14 16 18 20 22 hrs watts 2800 -47000 2 -2800 Jan Mar May July Fig. 4.23 Direct Solar Gains Sept Nov Jan Mar May July Sept Fig. 4.24 Building Fabric Gains Nov -1600 Jan Mar May July Fig. 4.25 Interzonal Gains Sept Nov 60 analysis examples so far have focused on the interaction between all of the zones but it is also possible to visualize a single zone at a time to see whether it is gaining or losing energy. A comparison of the central library space and the meeting room illustrates the difference in loads created by internal activity in the blue band and the cyan band shows that the meeting room is losing heat to other zones while the library is not. Fig. 4.26 Central Library Space Passive Gains Breakdown Fig. 4.27 Meeting Room Passive Gains Breakdown For the use of thermal analysis tools to be effective they need to cross the entire range of scales from that of the site to the layers that make up a wall section. The larger of these scales represents systems of relationships whose attributes are dependent upon subsystems. Along with this heirarchy there must be the conceptual understanding of the physical principles that shape the exchange of energy, which can then be used to relate various scales in an effort to simulate an holistic model. The purpose is not to simply find out how much energy would be necessary to heat a building but instead to test forms, materials, or spatial arrangements in an attempt to enhance a relationship between the internal and external environments. A difficulty that arises in this approach to design are the implications that one area of performance such as designing for thermal or lighting performance then has on other variables such as acoustics. 61 Bounces 4.1.6 Analysis of Acoustic Performance Fig. 4.28 Initial attempt: Linked acoustic ray analysis of children’s area Bounces Fig. 4.29 Second attempt: Linked acoustic ray analysis with addition of adult stacks to second level Libraries are generally quiet places, allowing people to focus on a task without additional distractions. For this reason the children’s area of the library was chosen for the acoustic analysis in order to analyze how sound from the children’s area would effect the rest of the library. Acoustic analysis can happen in a variety of ways within Ecotect. Sound bounces off surfaces similar to the way that light diffuses through a space meaning that one of the ways for visualizing sound is to create a set of linked acoustic rays that display the bouncing rays of sound. In this analysis the detailed model of the surface geometry was used instead of the zone model because of the design to have visually accurate results. To start the analysis a speaker is placed within the model, in this case it was placed at about 3.5’ above the floor in the children’s area, and then the acoustic variables of azimuth angle (90°), angular increment (2°), and number of bounces (4) are entered. The purpose of this analysis was to get an idea of the behavior of the sound in the space rather than determining the relative decibel levels between spaces. The initial attempt of this analysis returned results that at first glance appeared unpromising. In the first figure, the sound originates from the dense yellow group of rays within the children’s area and then because of the surrounding walls and curved roof manages to reflect into the adult reading area on the second floor as early as the second bounce (cyan lines). The concave form of the roof focuses sound from the children’s area into the adult reading area, which would likely lead to acoustical conflicts. When this image was overlaid with the building section it was noticed that many of these second bounces were intersecting the stacks which would be full of books, a material that has good acoustic absorption characteristics. 62 Fig. 4.30 Reverberation time for wood ceiling Fig. 4.31 Reverberation time for fabric ceiling Fig. 4.32 Reverberation time for acoustic ceiling tiles The stacks were placed in the Ecotect model and the analysis was rerun creating a much different acoustic response. In the second attempt it can be seen that few second bounces make it into the reading area and that the majority of the third and fourth bounces remain near the ceiling. These results seem much more promising than the initial attempt and suggest the possibility of purposely controlling the parameters of the stacks, for example the placement or height, in order to control the sound originating from the children’s area. Another type of acoustic analysis is to test the reverberation time (RT) of a space. Reverberation is effected by the acoustic absorption of a material meaning that a high acoustic absorption coefficient allows less sound to bounce around the space. High RTs create an echo effect while low RTs can cause sound to be perceived as flat, or lacking tonal balance. For this analysis the zoned model was used because of the need for knowing the volume of the space. Ecotect has the ability to use three separate algorithms for calculating the reverberation time which are Sabine, Norris-Eyring, and Millington-Sette. For this analysis the Millington-Sette algorithm was used because of its ability to more accurately deal with a higher range of absorption coeffecients. Originally the ceiling material was intended to be wood slats but the RT analysis of this material showed did not fall within the recommended blue band. As alternatives, two other calculations were performed using fabric and then acoustic ceiling tiles. The results suggest that a fabric ceiling material would offer the best balance of reverberation times while the acoustic ceiling tiles were unable to hold the RT with the recommended band. 63 While these calculations can give a general idea of what the RT will be, they are not accurate enough to pinpoint the RT because they fail to account for the actual shape of the space. However, they do offer quick studies in the use of different materials that could be beneficial if coordinated with the early explorations in form. 4.1.7 Summary of Analytic Studies The omission of the stacks from the original model shows that the accuracy of simulation is dependent upon the accuracy of the model, something that was brought up in the previous section on systems thinking. The more encompassing the scope of the model and the issues it addresses, the more reliable the results will be. It is often thought that the earlier a task can be performed within the design process the greater the influence it can have on the end product, but with the evaluative analysis this does not necessarily hold true. Instead, a certain type of analysis can only enter into the design process when a sufficient amount of information is present within the model to accurately simulate the performance in relation to reality. Lighting and shadow studies are largely dependent upon the geometry of the building while thermal studies can be as greatly influenced by form as material considerations. Experience with the use of these tools would establish notions of when it is appropriate to begin a certain type of analysis but initially there is likely to be a trial and error approach in order to establish those experiences. As the analysis begins to examine multiple forces acting upon the building, there becomes the ability to integrate solutions rather than piling the solutions on top of each other. In this project the lighting analysis showed 64 that there were excessive amounts of daylight but also the need to control the light between seasons while the acoustic analysis showed that the wood ceilings would not give the desired reverberation times and instead a ceiling made out of a fabric material might perform better. Rather than address these two problems separately, it may be possible by looking at them together, to use the idea of a fabric ceiling to manage both light and acoustical forces, and then afterwards following up with an analysis of the thermal response. It would be difficult, if not impossible to achieve the desired response for each of these forces because of possible conflicts between solutions. Instead, varying weights need to be assigned to forces that are deemed more important. These priorities are the subjective decisions of the designer and heavily dependent upon the program of a project and directed by one’s intuition and experience. The software is not responsible for providing solutions but instead the results of the analysis must be interpreted by the designer and this understanding must then be shaped into a proposed solution. The software also doesn’t dictate how or what should be tested but again these decisions are left to the designer who must use their experience and intuition to make an informed decision. Luckily what the software excels at doing is providing instant feedback for the complex problems that it is asked to analyze. The power of the computer as a design tool is dependent on how this feedback is interpreted by the designer and then how it is applied. 65 Carrie Blast Furnace Hot Metal Bridge Fig. 4.33 Site plan 4.2 Museum of Steel - Winter 07 This project was an entry in the 2006-07 ACSA Museum of Steel Student competition as part of the ARCH 501 Tectonic studio. The three studio projects completed before this project all explored the use of the curve as a linear extrusion, and with this project the goal was to further that exploration through modeling more complex geometry. At the time of this project, the digital tools that I was using were limited to SketchUp, making it difficult to design and keep control of the complex geometry because of the way that geometry is defined by the intersection of lines to create surfaces. A re-examination of this project serves two purposes related to morphological simulation. The first is to gain experience with software that I did not have at the time of the original project, and secondly, to look at how the use of these tools might change the overall design approach of future projects. 4.2.1 Project Description The brief for the competition called for the design of a Museum of Steel on the site of the former Carrie Blast Furnace, located in Pittsburgh, PA. As part of the competition the structure needed to be steel and the brief called for the design of an iconic building. After site analysis and case studies of precedents in steel construction, the concept was to create a steel structure that appeared to spring from the landscape, opening itself up towards the two Steel Industry artifacts on site, the blast furnace and hot metal bridge. 66 4.2.2 Initial Development of Geometry Fig. 4.34 Initial sketches of building form Fig. 4.35 Early attempts at rationalizing roof form At first the form of the building was explored through sketches that later became the base drawings that the initial digital model was built upon. From the beginning there was an attempt to allow the geometry to appear as if it were free form but would actually be derived from controlled shapes. The first version of this approach was to take and cut a irregular section out of a torus, a circle revolved around a circle to create a donut. When this idea was then modeled and an attempt was made to extract a plan from this geometry it became apparent that there was a disconnect between the plan and form sketches. Neither one was informing the other which led to trying to rationalize the base geometry of the roof to allow for placing structure and walls. From this point a cycle developed whereby an attempt would be made at modeling the desired roof geometry and then explore how the perimeter wall would fit underneath the roof until that model was no longer functional. These obstacles usually forced a complete rebuild of the geometry but each time this occurred, a greater understanding of the elements and their relationships were brought to the next iteration. Eventually the curved roof geometry became divided by a triangular grid which dictated the angle of the enclosing walls. This triangular grid also served as the base for developing a space frame structure that had to be meticulously trimmed down to achieve the desired form. Though the base geometry turned out to be fairly simple, the complexity of constructing the geometry prohibited the ability to make large scale changes later in the design process. There was also no convenient way 67 Fig. 4.36 One of the final iterations of the base geometry of extracting the information from the model something that would be necessary for fabrication were this an actual project. One benefit that came from this approach was an introduction to understanding the individual parameters that were effecting the geometry. For instance, by the end of the design the number of segments in the profile circle was known and adjusted to control the number of panels that spanned across the roof. While the experience of constructing this model was invaluable it would be beneficial in future projects to be able to consciously approach the form modeling process , allowing for more control, and more freedom in morphological design studies. For that reason, time has been spent learning software like Rhino, which controls geometry with more accuracy and abilities as well as learning Paracloud which allows for the population of a surface with cellular components such as a space frame component. 4.2.3 Rebuilding the Base Geometry Fig. 4.37 Exploded axonometric of the final building form For the geometry rebuilding exercise with Rhino and Paracloud the original design was taken as the point to work towards with no changes being made. This started in Grasshopper which is a program that operates within Rhino. In Grasshopper geometry is controlled through the use of components that are connected together with wires similar to an electrical diagram. Each component has inputs and outputs, but they can control a wide range of parameters ranging from geometry to logic rules to transformations. The first step was the construction of a line between two points that represented the overall length of the base geometry. The x, y, z coordinates of these points were defined individually with sliders to allow for easy future adjustments. The next step was to divide the line into a number of points 68 Fig. 4.38 Decomposition of points into X, Y, Z coordinates, Z coordinates replaced with values from a sine function Fig. 4.39 Creation of rail and section curve where the higher the number of points the smoother the rail curve with be. These points are then decomposed (pComp) into their x,y,z coordinates and the Z coordinates are replaced with a values from a trigonometric sine (Sin) function. Because of the small value size returned from the sine function it is necessary to then multiply (Mult) these values before applying them to the Z coordinate. The range of the sine function as well as the multiplication factor are both controlled with sliders to allow adjustment. A curve is then interpolated through this new set of points to create the rail curve. In the original design a circle or arc was used for the profile, but here only a circle was used. To create a circle it is necessary to establish the vector that the circle would be perpendicular to, a center point, and the radius. The vector can be created by sampling the first two points (Item) from the rail curve and drawing a vector (VecPt2) between them. The first point of the rail curve is then used as the center point and the radius is established with the use of a slider. At this point the circle (section curve) and the sine curve (rail curve) are input into a sweep (Swp1) operation where the circle will be pulled along the sine curve forming a tubular surface. The last step in this sequence is to divide the surface in the U&V directions (NURBS surface) in order to apply the space frame component within Paracloud. It may seem like a lot of steps to create a simple piece of geometry but it allows the freedom to transform, replace, or mutate any of the original geometry at any step and have the end product regenerated. 4.2.4 Developing the Space Frame Component Fig. 4.40 Division of surface into a grid of points The division points that were created in the last step in Grasshopper are then imported, through a direct link, into Paracloud as a point cloud that is 69 Fig. 4.41 Steps for rebuilding the Level 1 Base geometry 1 Creation of a line 2 Division of the line into 23 points 3 Substitute Sine curve for Z values 4 Creation of curve thru points geometry created in Rhino 4.0 using Grasshopper 5 Creation of circle at first point in curve 6 Sweep of circle along curve 7 Division of surface in U & V direction 70 Fig. 4.42 Paracloud surface created from points imported from Rhino arranged in a spreadsheet and ordered based upon the number of division that were assigned in the U&V directions. For the purpose of the space frame the surface that was created needs depth added to it which is done through the cloud processing tool. Depth can be added to the surface either positively or negatively and in this case was projected normal perpendicular to the surface at each point. This then creates a rib structure with the specified depth that the space frame can be created within. Within each of the quadrangular cells of the rib structure a component can be placed. One way of doing this is by entering the points that define a set of lines through a cell matrix. These points are then recorded and can be populated onto the surface based upon different function. For the space frame the points create lines that become cylinders whose diameters are controlled by the Component column. Once these components are created and populated to the cloud it is possible to graphically alter their pattern through a spreadsheet to further shape the cloud. Fig. 4.43 Paracloud surface with added depth creating ribs Fig. 4.44 Paracloud cell matrix for defining components The initial geometry had to be imported from Rhino, thereby breaking the direct connection between the different systems of geometry because two pieces of software were now responsible for form. While each of these software has there advantages over the other, this break hinders the effectiveness of this approach to design by limiting the immediate visual feedback that might be possible if geometry was kept within one software. Since Paracloud was only used for creating the space frame and trimming the overall form it was decided to further explore how Grasshopper could be used to create a single model of the different geometric systems. Fig. 4.45 Steps for developing space frame upon base geometry 8 Geometry imported from Rhino 71 72 String of Points Profile X Profile Y Starting Point Cull Pattern: off-on-off-on Point Order Direction List2 Starting Point List1 Starting Point Fig. 4.46 Development of Triangular grid 4.2.5 Modeling the Systems of Geometry Reconstructing the space frame geometry within Grasshopper was at first difficult because of the lack of experience with the software and the unfamiliarity with its method for constructing geometry. Any space frame is dependent upon a grid which can either be rather or non-rational, but in either case there is a system of points that define the grid. In the original SketchUp model the space frame was created by intersecting 3 sets of planes, each with a different orientation, with the tube surface. The approach in Grasshopper is in one way simpler but in another more complicated, however, once this grid is set up within Grasshopper the points can then be referenced for creating the other systems of geometry. The first step in creating the space frame was to establish a grid on the surface of the tube similar to what had been done in the previous exploration with Grasshopper and Paracloud. When the division points are created on the surface there is a starting point and ending point though the surface, such as a sphere, might be continuous. The list of points is ordered based upon the U&V divisions, similar to dividing the surface into a series of rows (string) and columns (profile). A string is made up of a fixed number of points which are repeated as separate profiles along the surface. The list progresses through all the points in a string before moving on to the next profile. The list of points can be used with all of its original values or the list can be culled with an on/off pattern, in this case every other point is turned off. A line or multiple lines can be created on the surface by pairing the original set of division points with a copy of the list whose points have been shifted 73 to a value further along in the sequence. Lines of differing orientations can be defined through this process to establish the grid of the space frame, which increases in complexity as depth is added with an inset surface and that is then triangulated between these two surfaces. The number of points and cull patterns differed between the inner and outer surfaces in order to get the vertices of the outer triangles to align with the center of the triangles on the inner surface. Fig. 4.47 Space frame component defined by two different sized intervals. Fig. 4.48 Two versions of the space frame generated with the change of only the rail curve The first attempt at modeling the entire space frame on the tube surface demonstrated that the computer was struggling to efficiently handle this large amount of information, and it was also unclear how the tube surface would be trimmed to a final shape, similar to what had been done in Paracloud. It was decided to readdress the definition of the overall form and trim away a lot of the excess that was contained within the tube definition. The original form was based upon a sine wave with a circle swept along it, creating a lot of excess geometry. In the second attempt, the form was still defined by the two types of curves for the sweep operation, the rail and section curves, but this time the curves were used only to create the final building form. The triangular space frame grid that was previously developed for use with the tube could then be copied into this new definition. With this setup, at any time the rail or section curve could be adjusted in order to respond to site conditions, program, or other parameters, and the space frame would adjust as well. The difficulty with this form of design is that in order to maintain the flexibility of the Grasshopper definition, the geometry must be controlled through geometric rules rather than direct interaction. Increased familiarity with the software allows for better planning and organizational 74 Fig. 4.49 Wall space frame understanding of the model, making it easier for future changes at any level. As the definition grows in complexity, changes to the early values, such as the defining profile curve or the UV divisions, stop so that attention can be given to the next level of detail. The benefit here is that the designer does not have to be fully committed to all of the previous decisions in order to proceed, but instead a well defined form has the possibility of being radically altered at any level and at any time during the design process. Once the roof form was roughed out, its geometry was used as the starting reference for establishing the walls. By using a method similar to the selecting of points for the creation of the space frame, a set of relationships were defined that picked two strings of lines near an edge of the roof. The X, Y coordinates for the points in these strings were then copied and a value of 0 was substituted for the Z coordinates, making a copy of the strings of lines at ground level. Surfaces, lofted between these pairs of strings, were divided to align with the spacing of the roof grid, and in the same technique that was used to model the space frame for the roof, the midsection members of the space frame were applied to the wall. At the time when the wall was being defined, the radius of the pipe for the wall space frame was given a specific value but it was noticed that there is the possibility that this radius could be controlled by a function that sets the radius of the pipe based upon structural proportions of pipe length to cross sectional area Fig. 4.50 Ground plane and wall derived from perimeter curves of roof geometry. One of the difficulties that was experienced when first designing this project in SketchUp was coordinating the floor plan with the orientation of the structure along the edges. Again using the points that defined the roof, a series of point strings were pulled out and oriented onto the ground level. A curve was then created from these points and offset to make different 75 levels. It became possible to not only control how the floor interacted with the perimeter walls but also to follow the gesture of the roof with what was happening on the ground plane. The reliance on the grid of points and their order would sometimes make it difficult or nearly impossible to achieve the desired form. An available alternative would be to manually enter the geometry which would then be detached from the definition or to conceive of intricate ways of extracting bits and pieces of the desired geometry from the grid. The decision of whether to continue with the definition or abandon it in favor of manually editing the geometry is likely to be partially dependent upon one’s experience with the software. As familiarity with the software increases so does the understanding of how to organize and use geometric systems making it possible to more fully design the building with an all encompassing definition. Fig. 4.51 Ground plane reconfigured with a small number of changes to the cull pattern The last study that was performed was to create the roof panels and control their patterning. When originally performing this task in SketchUp each panel had to be selected individually, limiting the number of studies that could be done. To create the panels in Grasshopper, the same points that were used to create the lines on the roof surface were isolated into three different sets that would then define the vertices of the triangular panels. Once these panels were created, a cull pattern could be used to turn panels on or off. At first, understanding the numerical relationship’s effects on the pattern could be confusing, but by using the adjustable sliders within Grasshopper a repeating interval could usually be found. These values generally had some relationship to the original UV divisions that first divided the roof surface, and once a specific pattern was found its relationship to the overall geometry could be formulated to allow for consistent updating if values were changed. 76 Fig. 4.52 Roof panel cull patterning studies 77 The pattern of the roof panels were controlled only by aesthetic considerations, but this could be an area of design where Paracloud would come in useful because of its ability to control geometry and the population of a surface with components based upon data from a spreadsheet such as the data that is used in the Ecotect graphics for lighting analysis. An example would be to balance desired lighting levels with the placements of the skylights. The model would initially be studied with no skylights and the feedback of this analysis could then be used to arrange the placement of the skylights. This process could be repeated a number of times until the desired lighting conditions were established. 4.2.6 Summary of Morphological Studies It initially seemed like a large amount of time was invested in the creation of the definition for this project, but in the end this investment paid off with the ability to derive multiple building systems from the same base points and being able to make changes that updated the entire model. There is definitely limits to this approach to design though the limits seem to be related to one’s level of experience. Just as the analytical simulation required a large amount of background information and experience to create an accurate simulation so too does the morphological simulation. The process of designing using a tool like Grasshopper is at the same time drastically different and similar from how geometry is typically used in design. Grids are often used in the layout of structures or the arrangement of a curtain wall but they tend to be regular with systems manually laid out upon them. Grasshopper requires another skill in addition to the typically held visual understanding of geometry and that is an organizational understanding of geometry which has visual implications but requires 78 numerical interaction. If someone is unfamiliar with this logic it can be difficult to use the software or to even conceive of how it could possibly be applied. However, once a familiarity is developed, the approach to design becomes completely different and the building systems are integrally conceived through the reasoning that must take place when deriving different systems from a single base. With the combination of the two forms of information, the morphological and analytical, geometry can begin to develop a response to the environment where adaptations would be made during the design process in order to strengthen the relationship between these two areas. This is by no means an automatic process but instead requires experience and intimate knowledge of the software. The forms that are capable of being created with the generative software have few bounds and superficially complex geometry can be easily achieved. However, the power of this software lies in the intentional exploration of geometry and incremental building of complexity. By preventing a design’s geometry from becoming fixed in the early stages, a designer can more freely and quickly explore alternatives that might otherwise be unfeasible. This flexibility becomes essential when trying to apply analysis results that will likely change as the design progresses. This method works in contrast to the convention of moving from one phase of design to another and instead allows the conceptual phase of design, where the broad ideas that shape much of the design are formed, to be pulled through all of the other phases of design. A broad concept of organization can be used to direct the development of a geometric definition that can respond to changes at any point in the design process, thereby removing the pressure to solve the critical issues in the early stages of the design process. Chapter 5 Design Project Elementary School Program Number of Children = 400-450 Recommended Site Size = 5-10 acres Front Office Entry Lobby Administrative Offices 2 @ 120 s.f. each Nurses Office Teacher’s Lounge Teacher’s Restrooms 2 @ 20 s.f. each PTA / Volunteer Room Public Restrooms 2 @ 200 s.f. each Gathering Facilities Gym 50 x 84 Cafeteria Kitchen Music Art Media Science Library Classrooms 500 s.f. 240 s.f. 150 s.f. 500 s.f. 40 s.f. 200 s.f 400 s.f. 4200 s.f. 2000 s.f. 1000 s.f. 500 s.f. 500 s.f. 500 s.f. 500 s.f. 2000 s.f. 3/grade @ 500 s.f. each 10,500 s.f. Total Gross 23,730 s.f. Total Net Square Feet Plus 10% Allowance For mechanical areas, circulation, structure, etc. 26,000 s.f. Fig. 5.1 Competition program for elementary school 79 The case studies previously analyzed offer insight into a range of approaches that several architectural firms have taken when using parametric modeling and evaluative analysis tools within their design process. The revisited studio projects allowed for experimentation with the digital tools in isolation of the design process. The last research area of this thesis explores designing with the integrated use of parametric modeling and evaluative analysis tools from the beginning of a project. This investigation is concerned with how the tools are used within a design process, what restructuring is necessary because of this, and what implications these changes have on the practice of architecture. The parametric modeling and evaluative analysis tools will be used to explore a series of vignettes that implement these tools at a variety of scales, from the scale of the site, down to the joint. Because of this process of working at multiple scales, not all design decisions will be fully resolved, but rather when enough information is present, the project will move on to another scale of design. The program for the design project comes from the 2008-2009 competition entitled Life Cycle of a School, sponsored by the ACSA/AISC, which calls for the design of an elementary school that takes into consideration issues of prefabrication, disassembly, and reuse. The site location was left open by the competition, and after an abbreviated period of site analysis the South Lake Union region of Seattle was chosen for the school site, specifically the Cascade neighborhood. 80 Parking Lot Pon t H ar ris on St ius Multi Family Mi nor as om Terminal/Industrial Parking Fig. 5.2 Cascade Playground and Land Use of Surrounding Blocks N Office/Retail Th School/Church St Mixed Use Ave N Ave N 81 ot gL n i St k Par rrison a H Cascade People’s Center Av e St or Ave N N P-Patch N Fig. 5.3 Current site conditions om as in ius Th M 5.1 The Concept Pon t The Cascade Neighborhood playground and the adjacent parking lot to the north are being used as the site for the proposed elementary school. The functions of the playground will be added to the program of the school including the P-Patch garden and Cascade People’s Center, which serves as the neighborhood “living room”. In order to properly design a school on this site, Harrison St. will be removed between Minor Ave N and Pontius Ave N and the playground will be cleared of its current structures. While there are urban and sustainability issues present within these decisions, they are outside of the scope of this thesis because the focus is again on the design process. Before moving into the design vignettes it is important to explicitly address the site response and chosen building concept. While one of the predicted benefits of the use of parametric modeling is the ability to make large scale changes late in the design, there are limits to how far back changes can be made, and it is expected that this limit exists somewhere after the choice of a building concept. 5.1.1 Building Concept Fig. 5.4 Site Cleared for Design Project For this project the building concept is to treat the large spaces of the school, the library, cafeteria, and gym as light-wells that the classrooms and other smaller spaces receive light from, naturally lighting as many of the spaces as possible. This concept carries implications about the formal relationships between spaces and thereby constrains the future possibilities of the design, illustrating the importance the initial concept plays in design. 82 5.1.2 Site Response As a continuation of the site analysis, a series of butterfly shadow diagrams were created, which illustrate the pattern of shadows that fall on the site over the course of a day. The solstices and equinoxes were analyzed during school operation hours, 8am-4pm. All of the surrounding buildings were set to their maximum zoning height of 85’. The diagrams on the left show the shadow patterns while the lower diagram depicts the corresponding sun angles at noon on each of the days. This analysis became the starting point for the generation of the rest of the design. Areas of the site that typically fell under shadow were used for locating interior spaces, while sunnier locations were reserved for outdoor activities. June 21 March/Sept. 21 Harrison St Site December 21 N Thomas St Fig. 5.5 Site Butterfly Shadow Diagrams 8am-4pm Zoning Height 85’ Fig. 5.6 Sun angle diagram (points on curves represent mid-day) 83 This simple but effective analysis was used throughout the design, allowing a continuous investigation of how the building and site could be shaped in order to pull the site out of shadow. The lower diagrams show several iterations analyzed during the course of the design. Fig. 5.7 March Butterfly Shadow Diagrams over course of design 84 5.2 Role of the Definition Fig. 5.8 Undeveloped plan The use of parametric modeling was essential in being able to fluidly build iterations upon previous design attempts like those shown in the butterfly shadow diagram progression. From the beginning of the design process a definition was created in Grasshopper which evolved over the course of the design and in the end this definition represented every designed aspect of the project from the landscape to the joint. During the early stages of the design the definition had large sections that needed to be replaced. This was partially due to changes in the design, but mostly caused by learning how to efficiently model with this tool. 5.2.1 Definition as Diagram One way that the definition proved useful during design was as a diagram of the design process and overall design organization. The diagram on the left, shown larger and labeled on the following page, begins to give a sense of the interconnections between different pieces of the program such as connections between the building form and landscape. The piece of the definition within the dotted circle on the far left is the plan representation of the original concept; natural light being shared by the smaller spaces wrapping the larger spaces. The pieces farthest to the right of the definition are entirely dependent on the left side of the definition for the creation of their geometry. Their inputs may require surfaces, lines, points, vectors, etc, meaning that any changes to the plan ripples through the rest of the definition to the right. Fig. 5.9 Week10 Grasshopper Project definition 85 Classroom Gesture Plan Layout Gym Roof Cafeteria Library Site Fig. 5.10 Week10 Grasshopper Project definition 86 Layout Curve Gym Plan Roof Center Pt # of Structural Bays Roof Edge Curve Roof Ridge Section Profiles Cafeteria Plan Ceiling Curve Roof Ridge Library Plan Ceiling Curve Roof Ridge Fig. 5.11 Week10 Grasshopper definition of gym roof structure and geometry it is dependent on. 87 5.2.2 Flexible, Relational Geometry extrusion width Fig. 5.12 Parametric relationship of room area to length along curve and width of extrusion Fig. 5.13 Examples of a bezier curves The flexibility of the definition was used was to set up spatial, programmatic relationships that controlled the division of the plan into the smaller spaces. For this to occur a relationship was created between the square footage of each room and the width of the extrusion which would return the required length of the room along the curve. This length was not a rigid value but once established could be altered in order to allow for a given amount of error or to round the dimension a desired module. A key value to this relationship is the width of the extrusion. If it was necessary to have this width be variable along the length of the curve, then a different set of relationships would need to constructed, likely negating the relationship that had already been set up. The width of the extrusion was changed at several points during the course of design in response to lighting levels, suggested classroom proportions, and prefabrication/assembly delivery . If the building had the possibility to constantly change in response to feedback than it would be necessary to develop the landscape in a similar way. In order to achieve this flexible, relational geometry with the landscape, relationships were established between points along the perimeter of the site and points along the edge of the building through the use of Bezier curves. A Bezier curve is controlled by a vector at each endpoint that determines the direction and magnitude of curvature from the associated endpoint. If a Bezier is created with opposing vectors at the end points then a S-shaped curve will be created. Within the Grasshopper definition the length of this vector could be manipulated with a slider, allowing a fine sculpting of the landscape. Once an overall gesture for the landscape was established there was the ability to automatically update its 88 form in response to a new building form. The definition also allowed for simple changes to be made to the landscape at any time without losing all of the detail that had already been achieved. Fig. 5.14 Adaptation of site to building form 89 bathroom Cascade People’s Center kitchen classrooms labs offices entry offices play fields library classrooms play ground entry cafeteria orchard play courts gym lab entry Parking Garage P-patch Cascade People’s Center Fig. 5.15 Week10 Building Program Layout Green roof Fig. 5.16 Week10 Site Program Layout 90 5.3 Iterative Design The greatest benefit of the flexibility made possible with the use of parametric modeling was the ability to create iterations during the design process and morph these iterations in response to feedback from evaluative analysis. This process of iterative design is most apparent in the complex geometry of the space frame roof structure which enclosed the gym, cafeteria, library, and entry. A space frame structure was chosen because of the geometry of the plan layout and their ability to rationalize a complex, doubly curved surface by subdividing it with triangles. As the design developed the form of the roof was influenced by programmatic, structural, lighting, wind, and aesthetic concerns. 5.3.1 Influence of Daylighting The goal to naturally daylight the school led to a system of panels within the space frame that would bounce light into the interior, thereby avoiding direct sunlight which would cause glare and undesirable lighting conditions. In order to test the validity of this concept a series of lighting analyses were done at different stages of the roof ’s evolution. The roof structure was manipulated in several ways, over a range of scales in order to accomplish the uniform lighting conditions, including the altering of the overall plan geometry, differing patterns of paneling, and manipulating the orientation of individual space frame cells in an attempt to aim light into the space. The false color images shown on the following page represent the Useful Daylight Index (UDI) which is the percentage of the year that lighting levels fall within the range of 100-2000 lux. Fig. 5.17 Evolution of Roof structure 91 Once the desired lighting conditions were achieved within the gym then the classrooms were incorporated into the analysis and their design proceeded in a similar process of analysis to iteration to analysis. 5.3.2 Balance of Influences At the beginning of the design all of the spaces within the school were provided with excessive daylighting levels. While all of the spaces were not addressed during this research, the gym serves as an example of how these 0% 10 20 30 40 Fig. 5.18 Analysis comparison of Useful Daylight Index 100-2000 lux 50 60 70 80 90 100 92 levels can be brought into balance through the analysis of design iterations. The areas of the design that were altered in order to achieve the uniform levels within the gym would likely have repercussions on other areas of performance such as structural, thermal, or fabrication. In the design of an actual project the requirements of each of these forces would all need to be brought into a balance of sorts which in the end is determined, somewhat subjectively, by the design team and what they hold to be most important to the success of the design. The ability to constantly manipulate a model parametrically allows this balancing process to be explored from many different perspectives, illustrating the ways that these forces can influence each other. Week 8 Week10 Fig. 5.19 Analysis comparison of Useful Daylight Index 100-2000 lux of classrooms and gym 93 5.4 Relational Kit of Parts The final area of research of this thesis focuses on the idea of a kit of relational parts rather than the traditional approach of a kit of standardized parts. Within a parametric definition, the rules for creating a set of parts and the relationships shared with other parts are laid out. If for example, a set of rules are established for generating a structure from a base surface then these rules can then be applied to an unlimited number of surfaces. Because of the irregular form of the plan, each classroom ends up being a unique shape, making it an appropriate testing ground for this idea of a kit of relational parts. 5.4.1 Establishment of Rules The classroom geometry began as two surfaces, the floor and the roof/ walls, which were provided from the subdivision of the plan extrusion. An early concept for the classrooms was to be able to divide each classroom into bays that could be stacked on top of each other and shipped as prefabricated chunks to the site. This immediately set up a maximum dimension for how wide each bay could be, resulting in a division of the classrooms into three bays. From this point on this idea of three bays shaped the design of the classroom structure. Fig. 5.20 Sample of three steps for generating the floor structure The floor and roof structure are identical in terms of their system of components; a beam runs down the center of each bay and holds intersecting joists that cantilever out in both directions with their ends tied together with two secondary beams. The generation of this structure came from a series of the steps that further subdivided each bay along its length, 94 always taking care to create pieces that could be produced from flat sheet material. These divisions progressed down in scale until reaching the scale of the joint between materials. Each piece of the structure is notched to allow them to slide together for assembly. Within the parametric definition the depth, thickness, and angle of intersection are all quantities that are taken into consideration, making it possible to change any of these values and have the notched intersections update in response. The use of intersecting planes to create the floor structure made it possible to not only apply this definition to a plane but a curved surface as well. Once the rules for this structural system were established for the floor structure it was a simple matter of substitution to create the same structure for the curved roof surface. Many of the rules for creating these elements were conceived through hand sketches before attempting to define the geometry within the parametric model. The intent behind a sketch is to understand the relationships between pieces not to numerically define all of the dimensions. A parametric definition can be thought of as a recording of those relationships that were established in the original sketch, leaving the dimensions to be fixed at a later time. 5.4.2 Possibilities through Digital Fabrication Digital fabrication tools have made it possible to conceive of designing a structure where every piece can be unique. It could be argued that the shape of most buildings (flat surfaces) is largely influenced by traditional fabrication techniques rather than structural performance, lighting, solar exposure, etc. Digital fabrication tools have the ability to change the balanced between these different forces and in doing so open new Fig. 5.21 Exploded axonometric of classroom 95 possibilities within design. Structures could be designed in such a way that material would be removed where it was not needed and added to those areas needing additional strength, similar to the responsive growth process of bones and trees. While this approach offers the ability to optimize performance in response to a single force, the true power behind the software is the ability to manipulate the performance of multiple forces simultaneously. Fig. 5.22 Layers of classroom structure Conclusions 96 The goal at the beginning of this research was to integrate parametric modeling and evaluative analysis into the design process. The design project shows that not only was integration possible but also lead to possibilities that are typically not found in design. An example is the fact that an entirely new plan form was able to be generated at the end of the design without a loss of the detail that had already been modeled. This was not an effortless task, but required roughly four hours of work to adjust some of the relationships within the parametric model. All of the analysis that had been done up to that point was no longer valid since the building form and its orientation had changed, but this example shows the potential of making large scale changes late in the design process. There is a limit, however, to how far back these changes can be made, and that limit exists somewhere around the chosen concept. The concept behind the design project influenced the shape of the building and how spaces interacted. The relationships of the parametric model were then built upon this idea. Changes could be made to the interaction of those relationships but a new concept would likely not be able to be inserted at the scale of the overall form while retaining the details that were built upon the previous concept. These relational dependencies can be seen in the Grasshopper definition and how the wires tie different systems of the design together. By understanding what inputs are used to create a specific system it would be possible to extract a section of the definition, connect it to a new set of similar inputs and regenerate that system. This approach is similar to the detail libraries that many firms use for their construction documents which develop overtime in response to experience. However, the difference is that this parametric systems library can be defined to adjust to the context 97 in which they are placed. Much more information must be put into a parametric model to make this possible as well as a different approach to design. The immediate context as well as the universal context must be addressed for a parametric model to achieve this type of flexibility. The role of experience does not diminish with the use of parametric modeling and evaluative analysis, but instead is made more critical in the case of choosing a concept or constructing a parametric model. If experience is based upon the breadth of circumstances encountered, then parametric modeling and evaluative analysis tools multiple the pace at which experience can be gained. The power of parametric modeling to aid in the rapid production of iterations allows an architect to better understand the structure of the relationships they have established. More powerful though is the feedback that is generated in response to these iterations through the use of evaluative analysis. This feedback not only allows for improvement in the performance of a design but gives an architect a deeper understanding of these influencing forces. Rather than designing based upon rules of thumb, the architect is establishing their own rules of thumb that then inform their intuition and later response to design problems. While the flexibility of parametric modeling might make it possible to undertake evaluative analysis during later stages of the design process, it is still most powerful in the early stages of design. It is at this point that the concept or underlying framework of the parametric model is established and by incorporating an early response to external forces there will be more possibilities for addressing them in later stages of design. The design project focused on a limited number of influencing forces, mainly lighting, aesthetics, and structural organization. These forces have 98 an impact on other areas of design such as thermal performance, acoustics, and cost. As experienced is gained with this approach to design it is necessary to draw these other influencing forces into the design concept, parametric model and evaluative analysis. Through the exploring of iterations a hierarchy will be established that balances the interaction of these forces. It is the role of the architect to use their experience, intuition, and available tools to manipulate how that balance occurs. This approach of high performance design also requires that architects move beyond the role of organizer of information provided by consultants and instead begin to establish a deeper understanding of the underlying forces and parameters that each consultant or system is responding to. Imagine working with consultants to establish a set of rules or parameters for the design of an HVAC system. This HVAC definition might receive certain key inputs that describe the overall form, total square footage, mechanical chase sizing, and other variables necessary for the preliminary design of the system. The expertise of the consultant has not been diminished but instead the architect has made more of an effort to internalize the factors that influence the HVAC system, thereby creating a greater potential for integration and emergence between systems. The role and capabilities of the architect changes when parametric modeling and evaluative analysis are integrated into the design process. These new capabilities open up directions in design that were previously impractical, but also require a restructuring of the design process, including they way that architects interact with consultants. The goal of these changes is to allow for a more informed, flexible, and holistic exploration of design, further the idea of an architecture of relationships. Bibliography Aranda, Benjamin, Chris Lasch, and Sanford Kwinter. Tooling. New York: Princeton Architectural P, 2006. Bachman, Leonard R. Integrated Buildings : The Systems Basis of Architecture. San Francisco: Jossey-Bass, 2002. Blaser, Werner. Renzo Piano - Centre Kanak : Cultural Centre of Kanak People. Trans. B. Almberg and K. Steiner. New York: Birkhauser Verlag AG, 2001. Buchanan, Peter. Renzo Piano Building Workshop Vol. 2 : Complete Works. New York: Phaidon P, 1995. Collins, Peter, and Kenneth Frampton. Changing Ideals in Modern Architecture. New York: McGill-Queen’s UP, 1998. Ednie-Brown, Pia. “Falling into the surface: towards a materiality of affect.” Architectural design. Sept.-Oct. 1999: 8-11. Hansmeyer, Michael. “Algorithmic Architecture.” Algorithmic Architecture. Mar. 2008 <http://www.mh-portfolio.com/>. Hart, Sara. “Double Indemnity - Featuring a Complex Structural System and Automatic Louvers That Adjust to Shifting Winds, Renzo Piano’s Tjibaou Cultural Center Is More Than Meets the Eye.” Architecture : the AIA Journal. 87. 10 (1998): 152. Hensel, Michael & Menges, Achim & Weinstock, Michael. Emergence : Morphogenetic Design Strategies. New York: Academy P, 2004. “Home Page.” Active Matter Research Group. Apr. 2008 <http://www. activematter.org/wiki/pmwiki.php>. 99 100 Johnson, Steven. Emergence : The Connected Lives of Ants, Brains, Cities, and Software. New York: Scribner, 2001. Kalay, Yehuda E., and William J. Mitchell. Architecture’s New Media : Principles, Theories, and Methods of Computer-Aided Design. New York: MIT P, 2004. Kolarevic, Branko, and Ali Malkawi, eds. Performative Architecture : Beyond Instrumentality. New York: Routledge, 2005. Lerum, Vidar. High-Performance Building. New York: Wiley, 2007. Malkawi, Ali, and Godfried Augenbroe, eds. Advanced Building Simulation. New York: Spon P, 2004. Porter, Tom, and John Neale. Architectural Supermodels : Physical Design Simulation. New York: Architectural P, 2000. Radford, Anthony D., and John S. Gero. Design by Optimization in Architecture and Building. New York: John Wiley & Sons, Incorporated, 1988. Rahim, Ali. Contemporary Processes in Architecture. New York: John Wiley & Sons, Limited, 2000. Reiser, Jesse, and Nanako Umemoto. Atlas of Novel Tectonics. New York: Princeton Architectural P, 2006. Rotheroe, Kevin. “A Vision for Parametric Design.” Architecture Week. July 2002. Mar. 2008 <http://www.architectureweek.com/2002/0710/ tools_1-1.html>. Sasaki, Mutsuro, Toyo Ito, and Arata Isozaki. Morphogenesis of Flux Structure. London: Architectural Association, 2007. 101 Steele, James. Architecture and Computers : Action and Reaction in the Digital Design Revolution. Grand Rapids: Laurence King, 2001. Terzidis, Kostas. Algorithmic Architecture. New York: Architectural P, 2006. Willims, Austin, and Susan Dawson. “Round Peg in a Square Hole [Swiss Re, London].” Architects’ Journal. 216.11 (2002): 26-35. “WORKING DETAIL - Swiss Re’s Diagrid Perimeter Structure of Steel Columns and Nodes.” The Architects’ Journal. 216. 11 (2002): 34. Appendix A Site Analysis 102 Single Family Multi Family Other Housing Mixed Use Church School/Daycare Public Facilities Gov’t Services Office Retail/Service Entertainment Industrial Terminal/Warehouse Utility Parking Vacant Waterbody Fig A.1 Land Use Map of South Lake Union Region of Seattle (site marked with black dot) Open Space Cascade Neighborhood 103 2005 2008 Single Family Multi Family Other Housing Mixed Use School/Daycare Fig. A.2 Land Use Map of Residential development from 2005-08 in Cascade Neighborhood (site marked with white dot) 104 Parking Lot Pon t H ar ris on St ius Multi Family Mi nor as om Terminal/Industrial Parking Fig. A.3 Cascade Playground and Land Use of Surrounding Blocks N Office/Retail Th School/Church St Mixed Use Ave N Ave N Appendix B Lighting Analysis Wall1-Ceiling0 Wall1-Ceiling4 Wall3-Ceiling4 105 Wall0-Ceiling4 Wall1-Ceiling5 Useful Daylight Index 100-2000 lux Useful Daylight Index >2000 lux 0% 10 20 Fig. A.4 Daylighting analysis of classrooms 30 40 50 60 70 80 90 100 106 Iteration in Response to Feedback Initial attempt was to flood the gym with light through large openings. Structural bays were too large. Too much light in center of gym. UDI >2000 lux UDI0% 100-2000 lux 10 20 30 Fig. A.5 Daylighting analysis of gym - Iteration 1 40 50 60 70 80 90 100 107 Iteration in Response to Feedback Next attempt slightly altered orientation of the space frame units through their extrusion vector. Opening the possibility to aim the light. UDI 100-2000 lux 0% 10 UDI >2000 lux 20 30 Fig. A.6 Daylighting analysis of gym - Iteration 2 40 50 60 70 80 90 100 108 Iteration in Response to Feedback Third attempt decreased the structural bay and the extrusion vector is again manipulated to focus light in the center of the gym. UDI 100-2000 lux 0% 10 UDI >2000 lux 20 30 Fig. A.7 Daylighting analysis of gym - Iteration 3 40 50 60 70 80 90 100 109 Iteration in Response to Feedback Fourth attempt covered half of the openings thereby bringing the light to within the desired levels. UDI 100-2000 lux 0% 10 UDI >2000 lux 20 30 Fig. A.8 Daylighting analysis of gym - Iteration 4 40 50 60 70 80 90 100 110 Iteration in Response to Feedback As an extension of the fourth attempt the classrooms were added to the analysis to determine how well the gym was feeding them light. UDI 100-2000 lux 0% 10 UDI >2000 lux 20 30 Fig. A.9 Daylighting analysis of gym - Iteration 5 40 50 60 70 80 90 100