Improving Design Processes: Automate, Communicate, Decide Forest Flager Reid Senescu John Haymaker April 2, 2009 Member Companies Reid Senescu and John Haymaker © 2009 Forest Flager, Reid Senescu, John Haymaker CIFE Research Areas Master Planning Plot Variable: response (Model.class_cost_2.total_cost) Parametric Modeling 730872 691764 652655 613547 574438 535330 496221 457113 418004 378896 $720,000 $710,000 $700,000 $690,000 $680,000 $670,000 $660,000 $650,000 $640,000 $630,000 $620,000 $610,000 $600,000 $590,000 $580,000 total_cost $570,000 $560,000 $550,000 $540,000 $530,000 $520,000 $510,000 Optimization $500,000 $490,000 $480,000 $470,000 $460,000 $450,000 $440,000 $430,000 $420,000 $410,000 $400,000 $390,000 $380,000 $370,000 20 20.5 21 21.5 22 22.5 23 23.5 24 24.5 25 25.5 26 26.5 27 27.5 28 28.5 29 29.5 30 30.5 31 31.5 32 32.5 33 33.5 34 34.5 35 35.5 36 36.5 37 37.5 38 38.5 39 39.5 40 40.5 41 41.5 42 42.5 43 43.5 44 44.545 45.5 46 46.5 47 47.5 48 48.5 49 49.5 50 50.5 51 51.552 52.553 53.554 2.8 3 54.5 5555.5 56 56.5 57 57.5 58 58.5 59 59.5 60 bldg_length 3.2 3.4 3.6 3.8 4 5.4 5.6 5.8 6 4.2 4.4 4.6 4.8 5 5.2 6.2 6.4 6.6 6.8 7 7.2 7.4 7.6 7.88 9.4 9.6 8.2 8.4 8.6 8.8 9 9.2 num_columns_along_length Process Communication Collective Decision Making Integrated Concurrent Engineering 4D Construction Planning Design-Fabrication-Integration Automated Field Instructions Building Performance Monitoring feasibility Reid Senescu and John Haymaker design construction operation Outline: How do we improve design processes? Three Methods, Three Acronyms…One Integrated Idea AUTOMATE PIDO COMMUNICATE PIP DECIDE MACDADI Reid Senescu and John Haymaker Current Practice New analysis tools, new demand for sustainability. BUT, few iterations, lack of design space exploration PIDO Improves design process efficiency and effectiveness. BUT, at large initial investment and narrow scope PIP Ensures these improved processes have a global impact BUT, no way to capitalize on new information How do we AUTOMATE to improve efficiency and effectiveness? PROCESS INTEGRATION DESIGN OPTIMIZATION Reid Senescu and John Haymaker PIDO Current Practice g g low number of design iterations significant information management From a survey of a Arup engineers*: Duration of Design Iterations Initial Subsequent (avg.) 7 weeks 5 weeks Average Number of Iterations Per Project 2.8 Management (58%) Manage Information 58% Spec Plan Design Processes 6% Execute Design Processes Ex 36% * Flager, F. and J. Haymaker, A Comparison of Multidisciplinary Design, Analysis and Optimization Processes in the Building Construction and Aerospace Industries, in 24th International Conference on Information Technology in Construction, I. Smith, Editor. 2007: Maribor, Slovenia. p. 625-630 Reid Senescu and John Haymaker Process Integration Design Optimization (PIDO) Ex: Hypersonic vehicle design at Boeing Take-off gross weight ratio (TOGW) g Excess Propellant Fraction Reid Senescu and John Haymaker Flager & Haymaker, 2007 Proof of Concept Case Study: Classroom g Design Variables Building orientation (0) Building length (L) Window to wall ratio (W) Structural steel sections W L O g g Constraints Fixed floor area Structural safety Daylighting performance Objectives Minimize first cost for structural steel Minimize lifecycle operating costs for energy Reid Senescu and John Haymaker © 2009 Forest Flager, Reid Senescu, John Haymaker steel frame beam girder column Multidisciplinary Optimization Model Size of Design Space: 55,000,000 options Reid Senescu and John Haymaker © 2009 Forest Flager, Reid Senescu, John Haymaker MDO Runs Completed: 5600 (0.01%) RunTime: 34 hours Impact of Building Topology & Geometry on Structure Cost Variable: response (Model.class_cost_2.total_cost) Steel Cost vs. Plot Building Length and Number of Columns 730872 691764 652655 613547 574438 535330 496221 457113 418004 378896 $710,000 $700,000 $690,000 $680,000 $670,000 $660,000 $650,000 $640,000 $630,000 L $620,000 $610,000 $600,000 $590,000 $580,000 $570,000 total_cost Total Capital Cost of Steel Structure $720,000 $560,000 $550,000 $540,000 $530,000 $520,000 $510,000 $500,000 $490,000 $480,000 $470,000 $460,000 $450,000 $440,000 $430,000 $420,000 $410,000 $400,000 $390,000 $380,000 $370,000 20 20.5 21 21.5 22 22.5 23 23.5 24 24.5 25 25.5 26 26.5 27 27.5 28 28.5 29 29.5 30 30.5 31 31.5 32 32.5 33 33.5 34 34.5 3535.5 36 36.5 37 37.5 38 38.5 39 39.5 40 40.5 41 41.542 42.5 43 43.5 44 44.545 45.5 46 46.547 47.5 48 48.549 49.5 5050.5 51 51.552 3 52.553 53.554 54.5 5555.5 5656.5 5757.5 5858.5 5959.5602.8 bldg_length Building Length (L) Reid Senescu and John Haymaker © 2009 Forest Flager, Reid Senescu, John Haymaker 3.2 3.4 3.6 3.8 4 5.4 5.65.8 6 4.24.4 4.6 4.8 5 5.2 9 9.29.4 9.6 7.88 8.2 8.4 8.6 8.8 6.6 6.8 7 7.2 7.4 7.6 6.2 6.4 num_columns_along_length Number of Columns Along Length Impact of Steel Member Sizes on Structure Cost Identifying Promising Areas in the Design Space L Values for section types / building length that yield best designs Each line represents a single design Beam Section Type Reid Senescu and John Haymaker © 2009 Forest Flager, Reid Senescu, John Haymaker Cost Max DC Ratio Building Length Girder Sections Column Sections Beam Sections Total Cost Each point represents a single design Impact of Design Variables on Energy Performance Total Lifecycle Operating Costs vs. Orientation and Length Total Lifecycle Operating Costs ($/ 30 years) Less Efficient Length (mm) Orientation (deg) Reid Senescu and John Haymaker © 2009 Forest Flager, Reid Senescu, John Haymaker Most Efficient Design TradeTrade-offs for MDO Total Lifecycle Operating Costs ($/ 30 years) Total Capital Cost of Steel Structure ($) Building Length vs. Capital and Operating Costs Building Length (m) Reid Senescu and John Haymaker © 2009 Forest Flager, Reid Senescu, John Haymaker Building Length (m) L Testing the Scalability of PIDO g Case Study: Stadium Roof Member Sizing 65,000 seat stadium Designed my Arup Sport (UK) Symmetrical steel roof structure Finite Element Model of Roof Reid Senescu and John Haymaker © 2009 Forest Flager, Reid Senescu, John Haymaker Case Study: Stadium Roof Member Sizing g Rectangular Hollow Section (RHS) The Design Problem: 1955 members 150 load combinations ~10-30 sections choices per member Size of design space: 201953 combinations! Circular Hollow Section (CHS) Universal Beam (UB) Section Typical Member Types in the Roof Structure Reid Senescu and John Haymaker © 2009 Forest Flager, Reid Senescu, John Haymaker Case Study: Stadium Roof Member Sizing g Results: Comparision of Process Metrics for PIDO and Conventional Methods Time Duration Iterations* Design Method Set Up Per Iteration* Completed Conventional 1 day 4 hours 20 (approx.) PIDO 15 days 6 min 8,000 * Iteration involves the generation and analysis of a design alternative using model-based methods Reid Senescu and John Haymaker © 2009 Forest Flager, Reid Senescu, John Haymaker Case Study: Stadium Roof Member Sizing g Results: Comparision of Total Steel Weight Estimated Savings: $10 M USD PIDO Design 1146 (-19%) Baseline Design 1414 0 200 400 600 800 1000 Total Steel Weight (metric tons) Reid Senescu and John Haymaker © 2009 Forest Flager, Reid Senescu, John Haymaker 1200 1400 1600 Goals of PIDO g Improve design knowledge early, maintain design freedom Reduced Time to Market 100% Improved Design Freedom Key Cumulative Design Freedom (Conv) Design Freedom (PIDO) Design Knowledge (Conv) Design Freedom (PIDO) 0% Improved Design Knowledge CD Reid Senescu and John Haymaker © 2009 Forest Flager, Reid Senescu, John Haymaker SD Design Timeline DD Current Practice New analysis tools, new demand for sustainability. BUT, few iterations, ineffective processes PIDO Improves design process efficiency and effectiveness. BUT, at large investment and narrow scope PIP Ensures these improved processes have a global impact BUT, no way to capitalize on new information How do we COMMUNICATE processes for global impact? PROCESS INTEGRATION PLATFORM Reid Senescu and John Haymaker PIP Scale beyond PIDO with design process communication g PIDO g Toolvis Tools shown below = 197 8 Campus Whole Building 85 Concept Design Dev. www.duncanwilson.com/toolvis Reid Senescu and John Haymaker © 2009 Forest Flager, Reid Senescu, John Haymaker Scale beyond PIDO with design process communication g PIDO g Structural Skills Search returned 189 software items This page has been accessed 13297 Reid Senescu and John Haymaker © 2009 Forest Flager, Reid Senescu, John Haymaker Case reveals struggle to SHARE processes Consider environmental impact for Steel vs. Concrete structure decision Successful Process Existed Actual Process Create Revit Structure Model Ask Forum, Colleagues, and Search Internet Export Model w/ LCA tags to IFC 2x3 Unable to perform model based LCA Embodied Energy Create Revit Structure Models ? ? Steel Concrete Reid Senescu and John Haymaker © 2009 Forest Flager, Reid Senescu, John Haymaker Analyze Model w/ LCA tags Embodied Energy g Steel Concrete Case reveals struggle to SHARE processes g Consider environmental impact for Steel vs. Concrete structure decision Sharing Today Actual Process Name Create Revit Structure Models Link Description Ask Forum, Colleagues, and Search Internet Arup contact Unable to perform model based LCA Embodied Energy Contact location Frequency of use ? ? Steel Concrete Reid Senescu and John Haymaker © 2009 Forest Flager, Reid Senescu, John Haymaker QA status Added value LCADesign 0 Sophisticated LCA tool that integrates with 3D models and produces environmental impact evaluation of material assemblies and buildings Henry Anning Brisbane None N/A High Case reveals struggle to SHARE processes g Consider environmental impact Steel vs. Concrete Structure Decision Successful Process Existed Actual Process Failed TakeCreate away from this case study: Ask Forum, Colleagues, and Search Internet Revit Structure – I could not find a design process Model to meet stakeholder goals – Even though this process had Export already Model been w/ LCAused tags in my firm ? ? Steel Concrete Reid Senescu and John Haymaker © 2009 Forest Flager, Reid Senescu, John Haymaker Analyze Model w/ LCA tags Embodied Energy Unable to perform model based LCA Embodied Energy to IFC 2x3 Steel Concrete Case reveals struggle to UNDERSTAND processes g Use model for Dayligting analyis to design Louver Configuration Export from Revit Architecture Model Time Per Design Task – Options Considered : 2-3 Louver Configurations – Total Design Time Per Option: 30 hours – Non-Value Added Time Per Option: 15 hours Use Radiance To Create Daylight Video Reid Senescu and John Haymaker © 2009 Forest Flager, Reid Senescu, John Haymaker Case reveals struggle to UNDERSTAND processes g Use model for Dayligting analyis to design Louver Configuration Export from Revit Architecture Model Time Per Design Task – Options Considered : 2-3away Louver Configurations Take from this case study: – Design Time Per Option: – Total Process is often repeated 30 hours – Process is inefficient. – Non-Value Added Time Per Option: – No investment in improving 15 hours process Use Radiance To Create Daylight Video Reid Senescu and John Haymaker © 2009 Forest Flager, Reid Senescu, John Haymaker Improving design process by catalyzing emergent behavior g Ingredients required: – SHARE better practice with others – UNDERSTAND current practice to improve it SHARE + UNDERSTAND = COMMUNICATE g How can we improve design processes through improved communication? Reid Senescu and John Haymaker © 2009 Forest Flager, Reid Senescu, John Haymaker Students prototype Process Integration Platform Stanford Graduate School of Business Library g Team: Five student design consultants g Options: Two atriums g Goals: Be Beautiful, Provide Good Light, Minimize Net Energy Use…. Reid Senescu and John Haymaker © 2009 Forest Flager, Reid Senescu, John Haymaker Legend: Narrative convention (Haymaker, 2006) Process Integration Platform Tool Used Barrel color indicates the status of the source information. Green = up-to-date Red = not updated What does the information look like? Haymaker, J., (2006). Communicating, Integrating and Improving Multidisciplinary Design and Analysis Narratives, inJohn J. Gero (ed), Design Computing and Cognition, pp 635-65 Reid Senescu and Haymaker Designer browses project via PIP web tool Process Integration Platform Arup: 130676-Stanford GSB/Library Senescu R and Haymaker J (2008). Requirements for a Process Integration Platform, 7th Workshop on Social Intelligence December 2-5, 2008, San Juan, Puerto Rico. Reid Senescu andDesign, John Haymaker Designer adds daylight analysis process Process Integration Platform Arup: 130676-Stanford GSB/Library Reid Senescu and John Haymaker © 2009 Forest Flager, Reid Senescu, John Haymaker Designer searches for context appropriate process Process Integration Platform Arup: 130676-Stanford GSB/Library SEARCH RESULTS Reid Senescu and John Haymaker © 2009 Forest Flager, Reid Senescu, John Haymaker Community shares, understands, and automates process Process Integration Platform PROCESS INFO Arup: 130247-Green House/Day Light Analysis Project: Green House Projects With This Process Process Type: Day Light Software Required Developer: Robert Cole Process Used: 57 times Links Copied: 31 times Used Most By: Engin Ayaz Other Users of this Process More Usage Details SEARCH RESULTS COMMENTS jrogers: Should we invest in automating this process? Normanrock: Yes, I’ll send a meeting request to divide up the work. Reid Senescu and John Haymaker © 2009 Forest Flager, Reid Senescu, John Haymaker PROCESS USAGE Ave. Iteration Time 22 hrs # of Iterations 3.3 Community shares, understands, and automates process Process Integration Platform PROCESS INFO Arup: 130247-Green House/Day Light Analysis Project: Green House Projects With This Process Process Type: Day Light Software Required Developer: Robert Cole Process Used: 57 times Links Copied: 31 times Used Most By: Engin Ayaz Other Users of this Process More Usage Details SEARCH RESULTS COMMENTS jrogers: Should we invest in automating this process? Normanrock: Yes, I’ll send a meeting request to divide up the work. Reid Senescu and John Haymaker © 2009 Forest Flager, Reid Senescu, John Haymaker PROCESS USAGE Ave. Iteration Time 22 hrs # of Iterations 3.3 Designer copies another process module Process Integration Platform PROCESS INFO Arup: 31347-Atrium Apartments/Day Light Project: Atrium Apartments Projects With This Process Process Type: Day Light Software Required Developer: Andy Key Process Used: 2 times Links Copied: 5 times Used Most By: jrogers Other Users of this Process More Usage Details SEARCH RESULTS COMMENTS jrogers: Does anyone else use this process? Reid Senescu and John Haymaker © 2009 Forest Flager, Reid Senescu, John Haymaker PROCESS USAGE Ave. Iteration Time 5 hrs # of Iterations 6 Designer pastes and combines process modules into GSB Process Integration Platform Arup: 130676-Stanford GSB/Library Reid Senescu and John Haymaker © 2009 Forest Flager, Reid Senescu, John Haymaker Designer uses arch. model and executes process Process Integration Platform Arup: 130676-Stanford GSB/Library Reid Senescu and John Haymaker © 2009 Forest Flager, Reid Senescu, John Haymaker Energy consultant knows daylight analysis is complete Process Integration Platform Arup: 130676-Stanford GSB/Library Reid Senescu and John Haymaker © 2009 Forest Flager, Reid Senescu, John Haymaker PIDO automated and PIP communicated….. Process Integration Platform Arup: 130676-Stanford GSB/Library How do we DECIDE on a design? Reid Senescu and John Haymaker © 2009 Forest Flager, Reid Senescu, John Haymaker Current Practice New analysis tools, new demand for sustainability. BUT, few iterations, ineffective processes PIDO Improves design process efficiency and effectiveness. BUT, at large initial investment and narrow scope PIP Ensures these improved processes have a global impact BUT, no way to capitalize on new information How do we DECIDE among these options? MULTI-ATTRIBUTE, COLLABORATIVE DESIGN, MULTIASSESSMENT, AND DECISION INTEGRATION Reid Senescu and John Haymaker MACDADI MACDADI Multi-Attribute, Collaborative Design, Assessment, and Decision Integration Reid Senescu and John Haymaker © 2009 Forest Flager, Reid Senescu, John Haymaker MACDADI: Team Model Eric Holder Yoko Matsu Sudhip Chaudri Matt Clough Ricardo Pitella Facility Manager Reid Senescu and John Haymaker © 2009 Forest Flager, Reid Senescu, John Haymaker MACDADI: Goal Model Reid Senescu and John Haymaker © 2009 Forest Flager, Reid Senescu, John Haymaker Metric -3 0 Operating Cost Per Year $20k $15k 3 $10k MACDADI: Preference Model Reid Senescu and John Haymaker MACDADI: Options Model Building Program Structural Systems Atria 4 Story Atrium 3 Story Atrium Mechanical Systems Reid Senescu and John Haymaker © 2009 Forest Flager, Reid Senescu, John Haymaker MACDADI: Analysis Model Reid Senescu and John Haymaker © 2009 Forest Flager, Reid Senescu, John Haymaker MACDADI: Value Model Value -32 -52 Benefit = Analysis x Preference Reid Senescu and John Haymaker © 2009 Forest Flager, Reid Senescu, John Haymaker Current Practice New analysis tools, new demand for sustainability. BUT, few iterations, ineffective processes PIDO Improves design process efficiency and effectiveness. BUT, at large initial investment and narrow scope PIP Ensures these improved processes have a global impact BUT, no way to capitalize on new information MACDADI Creates collborative rationale decision making. BUT….. We would like to hear what YOU think? Reid Senescu and John Haymaker Acknowledgements g Arup for research support – Especially: Cole Roberts, Key Anderson, Jake Wayne, Engin Ayaz, John Worley, Shruti Narayan, Martin Simpson, Chris Field, Stephen Burrows, and Jim Quiter g g Grant Soremekun at Phoenix for PIDO support Ben Welle for PIDO energy analysis results Reid Senescu and John Haymaker © 2009 Forest Flager, Reid Senescu, John Haymaker PIDO PIP MACDADI Integrated Concurrent Engineering session to Automate, Communicate, and Decide Reid Senescu and John Haymaker QUESTIONS?