Benjamin WelleStanford University
Grant SoremekunPhoenix Integration
Geometry, Structural, Thermal, and Cost Trade-Off Studies using Process Integration and Design Optimization
Improving Multi-Disciplinary
Building Design
Improving Multi-Disciplinary
Building Design
An academic research center within the Civil and EnvironmentalEngineering department at Stanford University: Research focus is on the Virtual Design and Construction (VDC) of
Architecture – Engineering – Construction (AEC) projects in collaboration with our industry partners
Introduction to theCenter for Integrated Facility Engineering
(CIFE)
Conceptual Phase – Model Based Design
OverviewThe time required for model-based structural and energy performance analysis feedback means few (if any) alternatives are evaluated before a decision is made.
ObjectiveDevelop/utilize a platform to integrate CAD and analysis tools for design exploration and optimization that:
Can interface with commonly used design tools in AEC industry Can support the following:
Software automation Software integration Data visualization Simplification of running of trade studies
Provides a robust, flexible and extensible environment
IntuitionProviding designers with this platform will allow them to systematically explore larger design space more efficiently and better understand those design spaces, resulting in higher performance and cost-effective design solutions.
Multidisciplinary Optimization Process
Energy and Daylighting Optimization Process
geometry definition parameters
Energy Analysis
Tool: EnergyPlus
Actor: Mechanical Engineer
Run-time: 0
Energy Analysis Results:
Energy Consumption: MJ/m2/yearSolar Heat Gains: MJ/m2/yearLighting Intensity: MJ/m2/yearLighting Multiplier: 0-1Cooling Intensity: MJ/m2/yearHeating Intensity: MJ/m2/yearElectricity Costs: $/yearGas Costs: $/yearTotal Costs: $/year
Adjust building geometry to minimize annual energy cost while meeting energy and daylighting constraints
Load Batch File
Tool: RunEPlus
Actor: Mechanical Engineer
Run-time: 0
Architectural Geometry from DP
Wall surface coordinates: X1, Y1, Z1, X2, Y2, Z2, X3, Y3, Z3, X4, Y4, Z4
Roof surface coordinates: X1, Y1, Z1, X2, Y2, Z2, X3, Y3, Z3, X4, Y4, Z4
Floor surface coordinates: X1, Y1, Z1, X2, Y2, Z2, X3, Y3, Z3, X4, Y4, Z4
Window surface coordinates: X1, Y1, Z1, X2, Y2, Z2, X3, Y3, Z3, X4, Y4, Z4
Create Variables Input Macro File
Tool: J-Script
Actor: Mechanical Engineer
Run-time: 0
Execute Main Input Macro File
Tool: EPMarcro
Actor: Mechanical Engineer
Run-time: 0
Execute Variables Input Macro File
Tool: EPMarcro
Actor: Mechanical Engineer
Run-time: 0
Create EnergyPlus Input File
Tool: EPMarcro
Actor: Mechanical Engineer
Run-time: 0
Proof of Concept Case Study: Classroom
Design Variables Building orientation (0) Building length (L) Window to wall ratio (W) Structural steel sections
Constraints Fixed floor area Structural safety Daylighting performance
Objectives Minimize first cost for structural steel Minimize lifecycle operating costs for
energy
L
O
steel frame
column
beam
girder
Window to Wall Ratio
Orientation
DESIGNPROCESSOPTIMIZATIONINTEGRATION TRADES SIMULATION VISUALIZATION
www.phoenix-int.com© Copyright 2007 Phoenix Integration, Inc. All Rights Reserved
DESIGNPROCESSOPTIMIZATIONINTEGRATION TRADES SIMULATION VISUALIZATION
www.phoenix-int.com© Copyright 2008 Phoenix Integration, Inc. All Rights Reserved
• 13 year history
• Provide process integration and design optimization (PIDO) software and services to customers in aerospace, defense, civil, oil and gas, financial
• Evolved out of a research program at Virginia Tech
• Office locations Philadelphia, PA (Corporate) Blacksburg, VA (R&D) California (Sales) North East (Sales)
• World-wide sales in North America, Europe, and Asia
Phoenix Integration
DESIGNPROCESSOPTIMIZATIONINTEGRATION TRADES SIMULATION VISUALIZATION
www.phoenix-int.com© Copyright 2007 Phoenix Integration, Inc. All Rights Reserved
DESIGNPROCESSOPTIMIZATIONINTEGRATION TRADES SIMULATION VISUALIZATION
www.phoenix-int.com© Copyright 2008 Phoenix Integration, Inc. All Rights Reserved
Phoenix Value Proposition
1. Improve your decision making capability Automate runs of existing
tools to quickly gather information
Apply intelligent algorithms to identify the best solutions
2. Manage design data Knowledge Capture, Search
and Reuse Collaboration and
Synchronization Data Pedigree/Traceability
DESIGNPROCESSOPTIMIZATIONINTEGRATION TRADES SIMULATION VISUALIZATION
www.phoenix-int.com© Copyright 2007 Phoenix Integration, Inc. All Rights Reserved
DESIGNPROCESSOPTIMIZATIONINTEGRATION TRADES SIMULATION VISUALIZATION
www.phoenix-int.com© Copyright 2008 Phoenix Integration, Inc. All Rights Reserved
AoA: Analysis of AlternativesCAIV: Cost As an Independent VariableSoS: Systems of SystemsDFSS: Design for Six SigmaMDO: Multi-Disciplinary Optimization
ModelCenter
Parameter Sweeps, DOE, Monte Carlo, Optimization, Add your own…
Run Matrix
Multi-Disciplinary Trade Studies
Process Results
Energy Model
Impact of Design Variables on Energy Performance
Design of Experiments (DoE) allow for the visualization of the design space and an understanding of variable sensitivity and performance trends.
The design space can be explored from a wide range of perspectives, including general trends using surface plots, actual data points using glyphs, and sensitivity data using bar charts
Orientation (deg)
To
tal
Lif
ec
yc
le O
pe
rati
ng
Co
sts
($
/ 3
0 y
ea
rs)
Most Efficient
Less Efficient
Length (mm)
Total Lifecycle Operating Costs vs. Orientation and Length
Total Window Area
Total Operating Cost
Total Wall Area
Impact of Design Variables on Energy Performance (cont’d)
Total Lifecycle Operating Costs vs. Total Wall Area and Total Window Area
Optimization vs. DoE Results for Energy and Daylighting Performance
DoE- 1882 simulations Optimization-93 simulations
Optimum areas of design space
The correlation between the optimum designs using DOE and the optimizer was extremely high. Simulation time to achieve optimum designs was reduced by 95%.
To
tal L
ife-c
ycle
Co
sts
($
/ 3
0 y
ea
rs)
Total Life-cycle Operating Costs vs. Orientation and Length
Orientation (deg)
Length (mm)
Multi-Disciplinary Model
Size of Design Space: 55,000,000
MDO Run: 5600 (0.01%)
Time: 34 hours
Design Variables• Building orientation
• 0-180 deg, 10 deg inc • Building length
•4-14m, 1m inc• Window to wall ratio
•0.1 to 0.9, 0.1 inc• Structural steel sections
•Girders (65 types)•Columns (7 types)•Beams (65 Types
Structural Cost vs. Energy Cost with Pareto Front
Pareto Optimal Designs for Classroom MDOStructural First Cost vs. Energy Lifecycle Cost
Structural Cost ($)
Life
cycl
e E
nerg
y C
ost
($/
30
years
)
Pareto Optimal Designs for Classroom MDOBuilding Length vs. Energy Lifecycle Cost
Pareto Optimal Designs for Classroom MDOBuilding Length vs. Structural Cost
MDO Optimization of Structural vs. Energy Performance
Optimal Designs with Varying Objectives
DESIGNPROCESSOPTIMIZATIONINTEGRATION TRADES SIMULATION VISUALIZATION
www.phoenix-int.com© Copyright 2007 Phoenix Integration, Inc. All Rights Reserved
DESIGNPROCESSOPTIMIZATIONINTEGRATION TRADES SIMULATION VISUALIZATION
www.phoenix-int.com© Copyright 2008 Phoenix Integration, Inc. All Rights Reserved
Forest FlagerGrant Soremekun
Stadium Roof Structural Optimization
Studies
Stadium Roof Structural Optimization
Studies
x y
z
DESIGN LAYER
Scale: 1:727.8
DESIGNPROCESSOPTIMIZATIONINTEGRATION TRADES SIMULATION VISUALIZATION
www.phoenix-int.com© Copyright 2007 Phoenix Integration, Inc. All Rights Reserved
DESIGNPROCESSOPTIMIZATIONINTEGRATION TRADES SIMULATION VISUALIZATION
www.phoenix-int.com© Copyright 2008 Phoenix Integration, Inc. All Rights Reserved
ModelCenter
Web Browser
Compute ClusterMulti-processor Server
Spare Computers
Analysis LibraryAnalysis Execution
Trade Study ArchiveSoon: Bill of Analysis
Accelerating Design Studies
DESIGNPROCESSOPTIMIZATIONINTEGRATION TRADES SIMULATION VISUALIZATION
www.phoenix-int.com© Copyright 2007 Phoenix Integration, Inc. All Rights Reserved
DESIGNPROCESSOPTIMIZATIONINTEGRATION TRADES SIMULATION VISUALIZATION
www.phoenix-int.com© Copyright 2008 Phoenix Integration, Inc. All Rights Reserved
Preliminary CenterLink Results
• Load balance Energy Plus Trade Study90 Energy Plus Analyses
Single Machine – Run Time: 50 minutes
CenterLink– 4 Machines (Quad 4 processors)– Run Time: 7 minutes
DESIGNPROCESSOPTIMIZATIONINTEGRATION TRADES SIMULATION VISUALIZATION
www.phoenix-int.com© Copyright 2007 Phoenix Integration, Inc. All Rights Reserved
DESIGNPROCESSOPTIMIZATIONINTEGRATION TRADES SIMULATION VISUALIZATION
www.phoenix-int.com© Copyright 2008 Phoenix Integration, Inc. All Rights Reserved
Stanford Cluster
• 16 Blade Compute Cluster
• Dual Core / Quad 4 (128 nodes)
• Installed Jan / Feb - 09
Current and Future Work
General: Make software wrappers more robust / flexible More complex building types, Case Studies (ARUP, SOM, Gensler, Burro Happold, AKT) Topology changes Parallel computing to reduce trade study run times
Energy: Variable constructions, locations, HVAC equipment, internal loads, schedules, etc. Developing a scriptwrapper to handle any DP geometry (or from any other BIM tool) and convert it to EP
syntax (no macros)
Daylighting: Developing a Radiance wrapper with support from Zack Rogers Combine SPOT and DAYSIM engines to calculate dynamic daylighting metrics Automatic sensor grid generation, using construction data from EP Each room will become a separate Radiance run, and an include file will be generated for EP Developing methodology using translucent windows to reduce simulation time
CFD: Developing a Fluent wrapper with auto-meshing using Gambit for space temperature stratification, air
velocity distribution, and mean radiant temperature Construction properties and surface temperatures taken from EP Variable diffuser locations