BIM-basedParametricBuildingEnergyPerformanceMulti-Objective Optimization

Mohammad Rahmani Asl1, Michael Bergin2, Adam Menter3, Wei Yan41PhD Candidate, Department of Architecture, Texas A&M University 2ResearchScientist, Autodesk Inc. 3Sustainability Education Program Manager, AutodeskInc. 4Associate Professor, Department of Architecture, Texas A&M University1https://sites.google.com/site/bimsimgroup/people/students/mohammad-rahmani-asl2http://www.autodeskresearch.com/people/michaelbergin 3http://www.adammenter.com/4http://faculty.arch.tamu.edu/wyan/1,4{mrah|wyan}@tamu.edu 2,3{michael.bergin|adam.menter}@autodesk.com

Building energy performance assessments are complex multi-criteria problems.Appropriate tools that can help designers explore design alternatives and assessthe energy performance for choosing the most appropriate alternative are in highdemand. In this paper, we present a newly developed integrated parametricBuilding Information Modeling (BIM)-based system to interact with cloud-basedwhole building energy performance simulation and daylighting tools to optimizebuilding energy performance using a Multi-Objective Optimization (MOO)algorithm. This system enables designers to explore design alternatives using avisual programming interface, while assessing the energy performance of thedesign models to search for the most appropriate design. A case study ofminimizing the energy use while maximizing the appropriate daylighting level ofa residential building is provided to showcase the utility of the system and itsworkflow.

Keywords: Building Energy Performance Analysis, Building Information Model(BIM), Parametric Modelling, Parametric Energy Simulation, Multi-objectiveOptimization

INTRODUCTIONDue to the considerable impact of buildings on theenvironment, it is essential for designers to recognizethe importance of improving or optimizing buildingenergy performance in the early design stage. En-ergy performance-based design is a highly complexand labor-intensive process. Designers deal with acomplex Multi-Objective Optimization (MOO) prob-

lem to minimize capital and operating costs whilemaintaining occupants comfort (Wang et al., 2005;Wright et al., 2002). This complexity comes from thelarge number of interrelated parameters involved insustainable building design such as building geome-try, space layout, materials, sites, weather data, userbehaviors, etc. There is a lack of easy-to-use and effi-cient tools to help architects explore design alterna-

Contribution 224 (Preprint) - figure and table placement subject to change- eCAADe 32 | 1

tives and understand their impacts on building en-ergy performance. Consequently, design practition-ers either decide not to consider energyperformanceof their designs and instead follow general rules-of-thumbs, which may result in inefficient building de-signs, or seek help from building energy experts tosimulate building design alternatives. Since trans-ferring an architectural design model to an energymodel is a time consuming and error-prone process,thedesigners andenergyexperts have to select a lim-ited number of design alternatives for energy analy-sis, which result in unoptimized design solutions.

Current building energy modeling tools do notsupport comprehensive parametric relations amongbuilding objects for simulation in tools such as En-ergyPlus. For instance, if a wall is transformed inan energy model, none of the related objects in-cluding windows, shading devices, rooms, roofs, andfloors will be updated automatically. In other words,parametric intents that are embedded in parametricBuilding InformationModeling (BIM) are not embed-ded in the energy models. As a result, a manual up-date of the model data is needed before running thesimulations but this is complex, tedious, and error-prone.

In order to fulfill the requirements of low en-ergy building design there is a need for an innova-tive designmethodology and integrated design pro-cess. The integration of parametric modeling andBIM is the new trend of buildingmodeling, which cangreatly benefit sustainable buildingdesign. Paramet-ricmodeling enables the creative exploration of a de-sign space by varying parameters and their relation-ships (Azhar and Brown, 2009). BIM is a model-basedprocess that provides methods and tools for creat-ing and managing building projects faster and moreeconomically (Eastman et al., 2011). BIMmay containmost of the data needed for building energy perfor-mance analysis and if used appropriately can save asignificant amount of time and effort in preparing in-put data for building energy simulation while reduc-ing errors (Kumar, 2008).

In this paper we investigate a systematic integra-

tion of BIM, parametric modeling, and building per-formance analysis to provide a new workflow thatmakes the parametric building energy performancestudy more accessible for innovative energy efficientbuildingdesign. Theworkflowuses aMOOalgorithmto explore the design space and provide a set of op-timal solutions to the designers.

BACKGROUNDThe conventional architectural design methodolo-gies focus on space and form. With the increasingimportance of building energy-efficiency, designershave to consider energy performance of their de-sign by exploring design alternatives that are morepromising to save energy in the conceptual designphase (Azhar et al., 2009). A considerable amountof literature has been published on building energysimulation tools. For instanceMaile et al. (2007) stud-ied the use of a selection of energy simulation en-gines and their user interfaces over different build-ing lifecycle phases. Also, Crawley et al. (2008) pro-vided a comparisonof the features and capabilities oftwenty major building energy simulation tools. Theliterature review of this paper is focused on build-ing energy simulation in conjunction with paramet-ric modeling, BIM, multi-objective optimization, andvisual programming, which are the techniques thatare used in the developed integrated system.

Parametric Modeling and Building EnergyPerformance AnalysisOneof themajorbenefitsofperformingenergy simu-lationduring thedesignprocess is to comparedesignalternatives using parameters and rules among ob-jects. Parametric modeling enables generative form-making and form-finding on the basis of aestheticand performancemetrics of buildings. Once the con-texts change in a later design stage, parametricmod-eling allows objects to automatically update (Aishand Woodbury, 2005; Stocking, 2009). Designers canintegrate parametric modeling into the process ofperformance analysis in different fields of buildingdesign, including, but not limited to, energy simula-

2 | eCAADe 32 - Contribution 224 (Preprint) - figure and table placement subject to change

tion (Paoletti et al., 2011; Pratt and Bosworth, 2011),structural analysis(Shea et al., 2005), and acousticsimulation (Wu and Clayton, 2013).

Parametric studies show a significant potentialcontribution to optimize the building energy per-formance (Naboni et al., 2013; Pratt and Bosworth,2011). Nonetheless, designers rarely use parametricbuilding energy performance analyses for the sakeofdue to the difficulty in preparing the energy mod-els as well as the long simulation run time. To solvethis issue, there are two common approaches: todevelop computational algorithms that reduce thenumber of runs (Coley and Schukat, 2002; Wetterand Wright, 2004), or to increase the computationalpower through cloud-based simulation (Garg et al.,2010; Zhang and Korolija, 2010; Zhang, 2009).

BIMandBuildingEnergyPerformanceAnal-ysisBIM is the process of generating and managing dig-ital representations of the building's physical andfunctional characteristics to facilitate the exchangeof information (Eastman et al., 2011). BIM representsthe building as an integrated database of coordi-nated information that can be used for the analysis ofthe multiple performance criteria including architec-tural, structural, energy, acoustical, lighting, etc. (Fis-cher, 2006). Performance-based design supportedby BIM is increasingly used in the building designdisciplines, allowing practitioners efficiently gener-ate andmodify building models (Fischer, 2006; Welleet al., 2011).

The existing studies that consider BIM as thecentral data model for building energy performanceanalysis aremainly focusedonautomaticpreparationof the building energymodel for various energy sim-ulation tools such as such as eQUEST (Maile et al.,2007), EnergyPlus (Maile et al., 2007; Bazjanac, 2008;Cormier et al., 2011), TRANSYS (Cormier et al., 2011),Ecotect and Green Building Studio (Azhar et al., 2009,2011), and Modelica-based tools (Yan et al. 2013).The common approaches in this type of research isto translate the BIM models to energy input files for

solving interoperability issues using Industry Foun-dation Classes (IFC) (Bazjanac, 2008; Morrissey et al.,2004) and to create an automatic link between BIMauthoring tools and building energy simulation en-gines (Yan et al. 2013).

Integration of BIM andparametricmodeling pro-vides a more effective process for performance-based design. Welle et al. (2011) created a thermaloptimization tool, ThemalOpt, which used BIM for ex-tracting the necessary information for thermal simu-lation and optimization. Rahmani et al. (2013) devel-oped Revit2GBSOpt, a plug-in for a BIM platform (Au-todesk Revit®), which integrates parametric BIM andbuilding energy performance simulation. Due to thecomplexity of parametric design study, an easy andvisual approach for designers to set up building pa-rameters and the inclusion of advanced, open sourceMOO algorithms are needed to improve the existingstudies, as presented in this paper.

Building Energy Performance OptimizationOptimization studies are being used in building de-sign after long being computationally intractable, onmulti-scale systems in various topics including opti-mizing construction costs (Radford and Gero, 1987),construction elements (Sambou et al., 2009), build-ing shapes (Wang et al., 2006), building envelopes(Bouchlaghem, 2000; Radford and Gero, 1987), Heat-ing, Ventilation, andAir Conditioning (HVAC) systems(Zhang et al., 2006), etc.

There are two common approaches to MOOproblems: 1) simple aggregation 2) Pareto Optimal.In simple aggregation, a composite objective func-tion is defined by combining all of the individualobjective functions. The composite objective func-tion can be determined with various methods, likeuse ofweighting factors. Determining the compositeobjective function needs knowledge of the relation-ships among individual objectives and their weight-ing factors (Fonseca and Fleming, 1993; Konak etal., 2006). Nevertheless, in building design these re-lationships are unknown in many cases. The sec-ond approach is to seek a set of promising solutions,

Contribution 224 (Preprint) - figure and table placement subject to change- eCAADe 32 | 3

known as Pareto-optimal set (Fonseca and Fleming,1993), given multiple objectives. Pareto Optimalitysupports decision making by finding the equally op-timal solutions such that it is not possible to improvea single individual objective without causing at leastone other individual objective to become worse off(Hoes et al., 2011). A posteriori set of preferencesmaybe used to evaluate the optimal solutions and findthe unique solution later by the designers (Gossardet al., 2013; Konak et al., 2006).

Visual ProgrammingWhile computer programming is often needed fordesigners to implement their sophisticated designintent (e.g. through the use of for-loop and con-ditional statements) in parametric BIM, visual pro-gramming interfaces can replace the conventionalelaborate coding with a visual metaphor of con-necting small blocks of independent functionalitiesinto a whole system or procedure (Boeykens andNeuckermans, 2009). Visual programming allowsusers create computer programs by manipulatingprogram elements graphically rather than textually.Based on a survey of 50 visual programming lan-guages (Myers, 1990), it is clear that a more visualstyle of programming could be easier to understandfor non-programmers or novice programmers (archi-tects normally fit into these categories). Examples ofvisual programming tools for architectural design areGrasshopper for McNeel Rhinoceros® and Dynamofor Autodesk Revit®.

METHODOLOGYIn this study an integrated system is developed forenabling designers to optimizemultiple objectives inthe early design process. A prototype of the system iscreated in an open-source visual programming appli-cation - Dynamo, which can interact with a BIM tool(Autodesk Revit®) to extend its parametric capabili-ties. The prototype contains a set of new functionnodes that can be used to optimize building energyperformance.

We have developed multiple Dynamo nodes to

contain essential functions for creating parametricBIM models in Revit and run parametric simulationsin GBS. A MOO algorithm (Non-dominated SortingGenetic Algorithm-II or NSGA-II, Deb et al., 2002) iscreated in Dynamo as a package of nodes that canhelp designers optimize multiple conflicting objec-tives and approach to a set of optimal solutions. TheNSGA-II node package is built based on the opensource code [1]. The node "NSGA-II" in Dynamo in-cludes a package of nodes and plays the main looprole for population generation in MOO to get to theoptimal solution (figure 1). The node "Initial Solu-tion Set" generates the initial set of random variableswithin the provided range and with the size of popu-lation defined by user. The output of this node is a listof variables and objective. The objective values arenull and they are assigned by "Population Evaluate"node which gets objective values as input parame-ters.

Figure 1Implementation ofNSGA-II in Dynamoto optimizedaylighting andenergy use

This workflow enables the Dynamo code to ac-cept objective functions as nodes or packages ofnodes. For instance, in this study the "LEED Daylight-ing" node is created as a package of nodes to calcu-late the LEED daylight values based on LEED Refer-ence Guide for Green Building Design and Construc-tion (USGBC, 2009) as an objective function.

The node "gbXMLExport" in Dynamo generatesenergy model data in the Green Building eXtendedMarkup Language (gbXML, 2014) format, which con-tains the necessary information for energy simula-tion, using Revit's Application Programming Inter-face (API). The "GBSProject" node is designed to cre-ate a new project in GBS by extracting the project in-formation from a BIMmodel such as the project loca-tion and the building type using Revit API, GBS API,and the Representational State Transfer (REST) pro-

4 | eCAADe 32 - Contribution 224 (Preprint) - figure and table placement subject to change

tocol. "GBSRun" is designed to create multiple runsin the GBS project and upload the exported gbXMLfiles to GBS for whole building energy analysis. Whenthe simulations are done, GBSRun retrieves the en-ergy simulation results for further analysis, optimiza-tion, and visualization (figure 2).

Figure 2Parametric BIM andwhole buildingenergy simulationintegration inDynamo

The presented system enables designers to ex-plore design alternatives and at the same time assessthe building performance to search for the most ap-propriate design.

CASE STUDYThe objectives of the optimization routine for thiscase study is tomaximize the number of rooms of theresidential unit that satisfy the requirements of theLEED IEQ Credit 8.1 for Daylighting while minimizingthe expected energy use. The simulation and calcula-tion of the energy use requires building informationthat BIM can provide, for example geometry informa-tion, physical material information, and location dataembedded within the model. The workflows devel-oped in this project can identify parameters from el-ements within the BIM and explore a set of scenariosfor energy performance and daylighting adequacy.

Climate and ContextThe geographic location of the home is in the cityof Indianapolis, Indiana, USA. The climate is domi-nated by heating loads with 5892 Heating DegreeDays (HDD) on a yearly basis. Due to site constraints,the long-axis orientation of the structure is fixed at 15degrees west of true north (figure 3).

Figure 3Case study buildingsite and floor plans

Model and Free Parameters (Decision Vari-ables)The residential home has six rooms at level one andtwo rooms at the second level that are included aspart of the daylighting calculation and energy use forthe entire building. The light admitted to the build-ing can enter via two fixed curtain walls that are notincluded as free parameters in the design space op-timization. These two curtain systems light the mainliving space in the first floor and the balcony in sec-ond floor. The rooms separated from the main livingspace by interior partitions are lit naturally by fixedwindows with a visual transmission coefficient of 0.9.The width and height of the windows are identifiedwithin the Dynamo interface as free parameters. Thedomains of the width and height of the glazing areaare set independently from 0.5' to 7.0' with an incre-ment of 0.1'.

Optimization AlgorithmThe NSGA-II algorithm is implemented with the in-put of a population size of 100 for each generation,with the maximum evaluations set at 1000 for a totalof 10 generations. The mutation probability is set at0.01. The crossover probability is set at 0.9 and boththe mutation distribution index and crossover distri-bution index are set at 20.0. Figure 4 shows the gen-eral overview of the MOO system designed for thisstudy and figure 1 shows its implementation in Dy-namo to optimize daylighting and energy use of thebuilding. The Pareto Optimal set from the NSGA-II al-gorithm is shown in figure 5. This graph shows the

Contribution 224 (Preprint) - figure and table placement subject to change- eCAADe 32 | 5

Figure 4General overview ofthe designed MOOsystem

result for 1000 runs for this experiment which tookabout 3 hours overall. This graph indicates that theoptimization routine begins to converge on the op-timal solution for each variable from the third gen-eration onward. From the graph in figure 6 it canbe seen that windows of various Widths from 1' to7' meet the requirements for more than 80% of therooms correlating with about $150 in variation forthe yearly energy cost. In this instance, windows be-tween the sizes of 3' and 4' in Height are evaluated, asthis parameter is preferred for the reasonof style to fitwith immutable horizontal datum elements. For de-sign variations within the bottom 30% of energy costand the full satisfaction of the daylightingmetric, thesmallest glazing Width is specified at 2' 8".

Figure 6Interactive parallelcoordinates plot forthe constraint andanalysis of designparameters.

Visualizing the results in an interactive parallelcoordinates plot allows the various iterations to beevaluatedby thedesigner. In figure 7 the chart showsthe sample of design variations that meet 100% ofthe LEED Daylighting requirements. Of these thelowest energy use calculated is $4,265 and the small-est window size is specified as a 5' width and 3.5'height.

6 | eCAADe 32 - Contribution 224 (Preprint) - figure and table placement subject to change

Figure 5Scatterplotshowing the ParetoFrontier with modelthumbnailssuperimposed onthe plot to illustratethe associationbetween thecalculated optimalsolutions and thebuilding forms.

CONCLUSIONThe investigation shows that the use of a BIM modelto generate a multiplicity of parametric design vari-ations for simulated and procedural analysis is a vi-able workflow for designers seeking to understandtrade-offs between daylighting and energy use. Theavailability of a cloud-based energy analysis tool en-ables thequick evaluationof hundreds of design vari-ations and the connection to a visual, parametric pro-gramming environment allows the design space tobe quickly and accurately specified.

Designers with limited parametricmodeling andprogramming experience may use the nodes pro-duced to perform a broad variety of design spaceanalyses. It is possible to optimize each window'swidth and height individually though this methodexpands the design space considerably. It is also pos-sible to include the angle of the building orienta-tion and the overall building footprint in the set offree parameters to be modified. For a broader de-sign space the number of iterations required may besignificantly increased toobtain reliable optimization

results.In addition to local variables such as window di-

mensions and material variations this system is ca-pable of producing design options in global buildinggeometries such as the footprint, the formof the roof,and the interior layouts. These design options areconsidered often by architects and engineers in thedesign process. The information embedded withinthe BIM can quickly be leveraged to obtain quantifi-able sensitivity of the performative implications to abroad set of possible design decisions.

Through the continued development of similarprojects to enable fast BIM-based simulation and rep-resentation of solution spaces and trade-offs, design-ers may be able to understand dependencies of de-sign options on the decision variables at the earlydesign stage without substantial expertise in energymodeling and daylighting analysis. For parametricanalysis, large changes in global building geometrycan lead to alterations in structural requirements andmechanical systems as well. Incorporating a broadervariety of simulations in different domains into the

Contribution 224 (Preprint) - figure and table placement subject to change- eCAADe 32 | 7

Figure 7Illustration of abi-directionalassociationbetween parallelcoordinates and 3Dmodel views

system will lead to more comprehensive explorationof the solution space and provide better decisionsupport for the stakeholders of building construc-tion.

AcknowledgmentThis research is partially supported by the NationalScience Foundation under Grant No. 0967446. Wewould like to acknowledge and thank Autodesk Inc.'sBuilding Performance Analysis and Dynamo teams,and all who contributed to our research for all theirhelp and support. Also, we would like to thankAlexander Stoupine at Texas A&M University for hiscontribution in this project.

REFERENCESAish, R and Woodbury, R 2005, 'Multi-level interac-

tion in parametric design', Smart Graphics, 3638,p. 151–162

Azhar, S and Brown, J 2009, 'BIM for Sustainabil-ity Analyses', International Journal of ConstructionEducation and Research, 5, pp. 276-292

Azhar, S, Brown, J and Farooqui, R 2009 'BIM-basedsustainability analysis: An evaluation of buildingperformance analysis software', 45th ASC AnnualConference, pp. 1-4

Bazjanac, V 2008 'IFC BIM-based methodology forsemi-automated building energy performancesimulation', Improving the Management of Con-struction Projects through IT Adoption, CIB W78,Universidad de Talca, pp. 292-299

8 | eCAADe 32 - Contribution 224 (Preprint) - figure and table placement subject to change

Boeykens, S and Neuckermans, H 2009 'Visual Pro-gramming in Architecture: Should Architects BeTrained As Programmers?', CAAD-Futures

Bouchlaghem, N 2000, 'Optimizing the design ofbuilding envelopes for thermal performance',Au-tomation in Construction, 10, pp. 101-112

Coley, DA and Schukat, S 2002, 'Low-energy de-sign: combining computer-based optimizationand human judgement', Building and Environ-ment, 37, pp. 1241-1247

Cormier, A, Robert, S, Roger, P, Stephan, L and Wurtz,E 2011 'Towards a BIM-based service orientedplatform: application to building energy perfor-mance simulation', Building Simulation 2011

Council, USGB 2009, LEED Reference Guide For GreenBuildingDesignAndConstruction, USGreenBuild-ing Council (USGBC)

Crawley, DB, Hand, JW, Kummert, M and Griffith, BT2008, 'Contrasting the capabilities of building en-ergy performance simulation programs', Buildingand Environment, 43, pp. 661-673

Deb, K, Pratap, A, Agarwal, S andMeyarivan, T 2002, 'Afast and elitist multiobjective genetic algorithm:NSGA-II', IEEE Trans. Evol. Comput., 6, p. 182–197

Eastman, C, Teicholz, P, Sacks, R and Liston, K 2011,BIM Handbook: A Guide to Building InformationModeling for Owners, Managers, Designers, Engi-neers and Contractors, Wiley

Fischer,M2006 'Formalizing construction knowledgefor concurrent performance-based design', 13thinternational conference on Intelligent Computingin Engineering and Architecture, pp. 186-205

Fonseca, CM and Fleming, PJ 1993, 'Genetic Algo-rithms forMultiobjective Optimization: Formula-tion, Discussion and Generalization', ICGA, 93, pp.416-423

Garg, V, Chandrasen, K, Tetali, S and Mathur, J 2010'EnergyPlus Simulation Speedup Using Data Par-allelization Concept', AMSE 2010 4th InternationalConference on Energy Suistanibility, pp. 1-6

Gossard, D, Lartigue, B and Thellier, F 2013, 'Multi-objective optimization of a building envelope forthermal performance using genetic algorithms

and artificial neural network', Energy and Build-ings, 67, pp. 253-260

Hoes, P, Trcka, M, Hensen, J and Bonnema, B 2011'Optimizing building designs using a robustnessindicator with respect to user behavior', BuildingSimulation Proceedings of the 12th Conference ofthe International Building Performance SimulationAssociation, pp. 1710-1717

Konak, A, Coit, DW and Smith, AE 2006, 'Multi-objective optimization using genetic algorithms:A tutorial', Reliability Engineering & System Safety,91, p. 992–1007

Kumar, S 2008, Interoperability between building in-formation models (BIM) and energy analysis pro-grams, Master's Thesis, University of SouthernCalifornia

Maile, T, Fischer, M and Bazjanac, V 2007, 'Buildingenergy performance simulation tools-a life-cycleand interoperable perspective', Center for Inte-grated Facility Engineering (CIFE) Working Paper,107, pp. 1-49

Morrissey, E, O’Donnell, J, Keane, M and Bazjanac,V 2004 'Specification and implementation of IFCbased performance metrics to support buildinglife cycle assessment of hybrid energy systems',Building Sustainability and Performance ThroughSimulation, Boulder, Colorado, USA

Myers, BA 1990, 'Taxonomies of visual programmingand program visualization', Journal of Visual Lan-guages & Computing, 1, pp. 97-123

Naboni, E, Zhang, Y, Maccarini, A, Hirsch, E and Lezzi,D 2013 'Extending the use of parametric simula-tion in practice through a cloud based online ser-vice', BPSA - Conference of International BuildingPerformance Simulation Association, Bozen, Italy

Paoletti, G, Avesani, S, Exner, D and Lollini, R 2011'Designing low energy buildings: application ofa parametric tool and case studies', 27th Interna-tionalConferenceonPassiveandLowEnergyArchi-tecture, Louvain La Neuve, Belgium

Pratt, KB and Bosworth, DE 2011 'A method for thedesign and analysis of parametric building en-ergy models', International Building Performance

Contribution 224 (Preprint) - figure and table placement subject to change- eCAADe 32 | 9

SimulationAssociationConference–Building Simu-lation

Radford, ADandGero, JS 1987,Designbyoptimizationin architecture, building, and construction, JohnWiley & Sons, Inc.

Rahmani, m, Zarrinmehr, S and Yan, W 2013 'TowardsBIM-based Parametric Building Energy Perfor-mance Optimization', 33rd Annual Conference ofthe Association for Computer Aided Design in Ar-chitecture (ACADIA), Cambridge, Ontario, Canada,pp. 101-108

Sambou, V, Lartigue, B,Monchoux, F andAdj, M 2009,'Thermal optimization of multilayered walls us-ing genetic algorithms', Energy and Buildings, 41,pp. 1031-1036

Shea, K, Aish, R and Gourtovaia, M 2005, 'Towards in-tegrated performance-driven generative designtools', Automation in Construction, 14, pp. 253-264

Stocking, AW 2009 'Generative Design Is Changingthe Face of Architecture', Building Design

Wang, W, Rivard, H and Zmeureanu, R 2006, 'Floorshape optimization for green building design',Advanced Engineering Informatics, 20, pp. 363-378

Wang, W, Zmeureanu, R and Rivard, H 2005, 'Apply-ing multi-objective genetic algorithms in greenbuilding design optimization', Building and Envi-ronment, 40, p. 1512–1525

Welle, B, Haymaker, J and Rogers, Z 2011, 'Ther-malOpt: A methodology for automated BIM-based multidisciplinary thermal simulation foruse in optimization environments', Building Sim-ulation, 4, p. 293–313

Wetter, M and Wright, J 2004, 'A comparison ofdeterministic and probabilistic optimization al-gorithms for nonsmooth simulation-based op-timization', Building and Environment, 39, p.989–999

Wright, JA, Loosemore, HA and Farmani, R 2002, 'Op-timizationof building thermal designandcontrolby multi-criterion genetic algorithm', Energy andBuildings, 34, pp. 959-972

Wu, C and Clayton, M 2013 'BIM-based acoustic sim-ulation Framework', 30th CIB W78 InternationalConference, Beijing, China

Yan, W, Clayton, M, Haberl, J, Jeong, W, Kim, JB,Kota, S, Alcocer, JLB and Dixit, M 2013 'Interfac-ing BIM with Building Thermal and DaylightingModeling', 13th International Conference of the In-ternational Building Performance SimulationAsso-ciation (IBPSA), Chambery, France, p. 3521–3528

Zhang, Y 2009 '“Parallel” EnergyPlus and thedevelop-ment of a parametric analysis tool', IBPSA Confer-ence, p. 1382–1388

Zhang, Y and Korolija, I 2010 'Performing complexparametric simulations with jEPlus', Proceedingsof the 9th SET Conference, Shanghai, China

Zhang, Y, Lin, K, Zhang, Q and Di, H 2006, 'Ideal ther-mophysical properties for free-cooling (or heat-ing) buildings with constant thermal physicalproperty material', Energy and Buildings, 38, pp.1164-1170

[1] http://jmetalnet.sourceforge.net/index.html

10 | eCAADe 32 - Contribution 224 (Preprint) - figure and table placement subject to change