Computational Fluid Dynamics for Urban Design:The ...papers.cumincad.org/data/works/att/ijac20109103.pdf · gain as well as pollution accumulation can be minimized and/or to generate

33

Computational FluidDynamics for UrbanDesign:The Prospects forGreater Integration Daniel Hii Jun Chung and Malone-Lee Lai Choo

issue 01, volume 09international journal of architectural computing

03_Jun 08/06/11 9:23 am Page 33

34

Computational Fluid Dynamics for Urban Design:The Prospects for Greater Integration Daniel Hii Jun Chung and Malone-Lee Lai Choo

Abstract

Computational Fluid Dynamics (CFD) has always been used in the fieldof architecture, urban design and urban planning to understand thepatterns of wind flow through the built environment. Its analysis isimportant to evaluate whether the natural ventilation through a site isadequate to mitigate heat and pollutant to achieve better humancomfort in dense urban environments. However, given the complexoperational requirements, the response to wind flow is not always doneearly enough to support planning and design.This paper seeks toillustrate how CFD analysis can aid planning and design of urban areasand investigates the workflow requirements, in the hope of making theCFD simulations more accessible to the practices and contribute todesign decisions. It also looks at the present technologicaladvancements and future prospects to assess the scenarios whereemerging technologies can make CFD simulation more readily availablewith affordable and even mobile hardware installations.

03_Jun 08/06/11 9:23 am Page 34

1. INTRODUCTION

Today, sustainability has become a major concern in urban development asenergy consumption, excessive waste and pollution are threatening theresource equilibrium of cities. Issues such as urban heat islands,overcrowding, traffic congestion, uncomfortable pedestrian spaces andinadequate open spaces have resulted in deteriorating living and workingconditions which affect human health and wellbeing.

One of the ways to mitigate high energy consumption is to plan anddesign urban spaces with due regard to existing ventilation patterns andheat concentration at the local level.This can be done by appropriateanalyses at the urban design stage, using computational fluid dynamics(CFD), a technique pioneered by Harvard Lomax [1].The technique can beused to analyze prevailing wind direction and speed as well as heat radiationpatterns. It is particularly useful in high density and high rise contexts wherethe micro built environment had extensively altered the natural ventilationof the space. High rise and compact building structures naturally trap heatand pollution on building surfaces, vehicles, human and artificial landscapesaround the city. Even where releasing the heat is possible, it would take along time before neutral air temperature can be achieved.As a result, humancomfort is compromised and outdoor activities reduced.

The world’s population is increasing and it is expected that compact andhigh density living in cities will be inevitable. However, as noted, crowdedenvironments cause the accumulation of heat, air pollution and waste. Insuch environments, high energy consumption is expected for cooling andlighting in order to make spaces more conducive and comfortable for workand living. In addition, accumulation of heat would aggravate the Urban HeatIsland (UHI) effect [2], which results in high usage of air conditioning forindoor spaces.An indirect impact of this on urban planning and design in thetropics is that people would prefer staying in air-conditioned indoorenvironments, leaving planned outdoor spaces under-utilized. Planners anddesigners have to respond to this by having better understanding ofventilation behaviour and requirements at the city scale, and having themethods and tools to deal with it.

Traditionally, land use planning addresses urban functions and land useintensity requirements through two-dimensional zoning approaches. In urbanareas where the built form and cityscapes are important, such land uses aresimulated in urban designs and then translated to planning guidelines thatestablish the relationship of form, mass, space and function, and guide designersto achieve better quality living spaces. However, issues such as air stagnationand flow in the immediate urban context are usually not analyzed at theplanning stage, and very often, this is left to architects at the building designstage to assess the need for better ventilation through design interventions.Toachieve better spaces at the urban level, we propose that CFD analysis shouldbe done at the urban design stage.The process is illustrated in Figure 1.

35Computational Fluid Dynamics for Urban Design:The Prospects for Greater Integration

03_Jun 08/06/11 9:23 am Page 35

However, we need to explore how this can be effectively done given currenttechnologies and practical operational requirements.

1.1. Objective

The aim of this paper is to illustrate the use of CFD in urban design and toinvestigate how CFD can be made more accessible to planning and designpractice and research.At the current stage of development, CFD is stilllargely beyond the reach for many professional practices. Even at theuniversity level, it is mostly offered on a shared platform between theacademic, research and computing services support community.Theproblem with this situation is that there is usually a contention for thesimulation facility, with queuing time taking away valuable research time.

In addition, there is the operating system compatibility problem.As anexample, in the High Performance Computing Centre of the NationalUniversity of Singapore, the platform is running in both Linux and UNIXplatforms. Most of the application software that is currently being used inthe Computer-Aided Architectural Design (CAAD) field are running ineither Windows or Macintosh platform.This is a major incompatibility issuethat makes it difficult to integrate to the workflow.Apart from that, mostcommands are inputted line by line without Graphical User Interface (GUI).This makes it very tedious for designers who are used to operating on theGUI mode. In situations where the more traditional approach of using windtunnels is resorted to, major capital investment would still be required interms of space and physical facilities, which would be problematic if spaceand funding resources are constrained.

Hence, while it is recognized that CFD can provide architects withinsights into the effects of unusual architectural forms on the micro climateand give clues on potential building services requirements [3], it is still notcommonly done in design, let alone at the earlier planning and urban designstages. In practice, researchers or practitioners in the field would sub-contract the CFD simulation work to consultants or specialists. However,this usually takes up too much time and does not allow for interactiveresponses given that design is always evolving according to many factors.There have been some intentions to integrate it into the design studios butthe results have not been satisfactory [4], [5] because of the incompatibilityof the CAAD formats with CFD software, the time consuming process toanalyze different designs and lack of scientific knowledge to prepare the 3Dmodels for CFD simulation.

! Figure 1. Planning and DesignProcess

36 Daniel Hii Jun Chung and Malone-Lee Lai Choo

03_Jun 08/06/11 9:23 am Page 36

Fortunately, today, technology has reached a stage where local desktopsand workstations are starting to be able to do some degree of complicatedsimulation which, only a few years ago, remained in the realm ofsupercomputing.This opens up new opportunities to consider theintegration of CFD simulation at the earlier stages of the planning anddesign processes.

In the context of a research project that we are undertaking, weexplored the use of CFD at the urban design stage, and evaluate thepotential for improved work flows that better integrate CFD operations, asa prelude to the main research. In itself, our main research focuses on theevaluation of urban densities against a defined set of environmental factors,and one of the parameters is wind flow in the urban environment. Ourpremise is that different urban typologies will make a significant impact onwind flow through the city, as evidence from another study which showswind velocity ratios difference for different urban layout [6] can vary asmuch as 0.17. For our research in the context of a city with hot and humidclimate, we envisage that UHI intensity [7] will increase significantly as theurban density increases, and the city gets more compact with higher buildingheight to width ratios. Here, the study of ventilation using CFD simulationsis relevant, but of equal relevance is how such simulations can be mademore readily accessible for mainstream applications.

2. RELATED WORKS

Various studies have been done to clearly establish that wind flow and howit is harnessed in urban design has a significant impact on the thermalcomfort of urban dwellers.An experiment was conducted [8] in Hong Kongduring summer to find out the importance of wind in response to thermalcomfort. Subjects were requested to sit under a sun umbrella and behind awind break to measure their response towards changes in the ambientconditions. It was found out that by suppressing wind, neutral sensationdropped more drastically than removing the umbrella.An increase of 0.3-1m/s of wind speed helps to lower the temperature by 2 degree Celsius.Thisillustrates that wind is extremely crucial in maintaining people’s comfort inthe urban environment and hence, the relevance of CFD analysis.

CFD simulations have been used for both outdoor and indoorenvironments. In both cases, the aim is to understand airflow so that heatgain as well as pollution accumulation can be minimized and/or to generategood cross-ventilation for greater human comfort. In the indoorenvironment, CFD is used to analyze building designs for efficient ventilationsystem management [9] and ventilation systems for damage control [10]. Inthe outdoor environment, CFD has been used to evaluate the effects ofheat gain from materials, air-conditioning condensers and building geometryin the urban environment [11], [12].


03_Jun 08/06/11 9:23 am Page 37

Research has validated that CFD results are quite reliable whencompared with wind tunnel and site measurement results, although thereare some differences under strong and mild wind conditions [13].Tominagaet al [14] did the same validations and found CFD techniques to beparticularly useful for predicting wind environments in pedestrian areas.

As a follow-up of CFD simulations, there are many strategies to plan anddesign responses to further improve ventilation and airflow.These includefaçade design and articulations, window vent size, shape and location,envelope insulation, balcony provision, natural ventilation, centralized heatingand cooling systems, building material selection and micro-environmentdesign [15]. In addition, at the urban scale, building height variation [16],good city planning incorporating air paths [17], smart building orientationwith regard to prevailing wind direction [18] will all help to improveventilation.

Other strategies had been proposed to help ameliorate the UHI effect,for example, in the case of Singapore [19], these include introducing high-rise towers at intervals, maintaining height-to-width ratio of 0.6 to 0.7 togive maximum velocity at the centre of the urban canyon. Elsewhere,Malkawi et al [20] had done similar works by getting the building formgeneration to respond to CFD analysis results.There are also otherresearchers who have tried to combine it with other environmentalparameters, for example, energy with CFD simulation [21], solar with wind[22] as well as various other physical parameters [23].

However, most of the CFD analyses are still done in two-dimensionalmode and lack visualization of how the air flow pattern looks in three-dimension. In this experimental research, we seek to do three-dimensional(3D) CFD simulations in an urban setting to give a more complete picturesince air flows are volumetric. Most CFD research is also very independentof other inter-related environmental issues which we intend to address at alater stage in our main research.

3.THE SITE

The site of interest for the experimental CFD simulation is a planning areasituated at south-eastern part of Singapore around the Kallang River, asshown in Figure 2. It is one of the four growth areas identified in theSingapore Master Plan 2008, which is a statutory plan that shows thedevelopment patterns envisaged for a projected population of over 5.5million in the future. It is an inner urban area with a site area of 64 hectares(640,000m2) of prime developable land.The site is divided into four partsfor which different themes have been identified for their respective futuredevelopments.The themes are “Life in the Open”,“Life’s Beach”,“Buzz onthe Green” and “A Unique Work-Live Place” [24].We note that all the fourthemes focus on outdoor living and recreation opportunities.The mainthrust of the plan was to enhance the local environment to make it


03_Jun 08/06/11 9:23 am Page 38

conducive to outdoor activities, with attractive open spaces and pedestrianlinkages. In this context, it is crucial to have good ventilation in and aroundthe site to support these themes. Given the country’s hot and humidtropical climate, it would be difficult to realise this planning vision ofencouraging people to use the open areas if the air flow is poor and theenvironment is uncomfortable.

The planning authority of Singapore, namely, the Urban RedevelopmentAuthority (URA) has developed an urban design plan and simulated blockmodel for the development area as shown in Figure 3.The proposed PlotRatio (Floor Space Ratio) for the site ranges from 1.5 to 5.6. By referring tothe block model, as well as taking the proposal building footprint andproposed Plot Ratio, we deduced the height of the buildings to range from 3to 36 stories.

4.THE CFD PROCESS, RESULT AND VALIDATION

Using the above plans and information, we proceeded with the CFDprocess as shown in Figure 4.The diagram shows the entire process frommodeling to virtual reality (VR) and animation visualization. Solid modelswhich are fully sealed are required for effective CFD simulation as all thewind vector paths will respond to every surface on the scene.There are

" Figure 2.The site location and theMaster Plan land use

" Figure 4.The CFD process

" Figure 3.The urban designintentions of the planning authorityand the 3D model representation


03_Jun 08/06/11 9:23 am Page 39

some Computer Aided Architectural Design (CAAD) software that do solidmodeling like AutoCAD and Revit Architecture. For the CFD simulation itself,there are a few well known software available commercially, namelyCOMSOL Multiphysics, CD-adapco STAR-CD/CCM+ as well as ANSYS Fluent andCFX. STAR-CD which is used frequently in Hong Kong, such as in a recentconsultancy study on Building Design that Supports Sustainable Urban livingSpace in Hong Kong commissioned by the government [25]. For ourexperimental project, we used COMSOL because it has an affordableeducational version and accepts major CAAD formats. Final CFD resultscan be analyzed in ANSYS CFD-Post and later CEI EnSight, which is thesoftware for post-processing CFD virtual reality visualization and animation[26].

Solid models have to be meshed [27] and scaled [28] correctly for thesimulation, as shown in Figure 6.The local Building and ConstructionAuthority [29] suggests the mesh density should be in the range of 0.5 to1m near buildings and at ground level to 10m at far field.We use theincompressible Navier-Stokes steady-state analysis in 3D dimension, whichdeals with incompressible isothermal fluid flow. In the CFD simulation of thefluids, there are three main techniques to simulate turbulence, the DirectNumerical Simulation (DNS), which computes directly the instant Navier-Stokes equation for the turbulence, the Large Eddy Simulation (LES) and theReynolds-Averaged Navier-Stokes (RANS) equations.They rank from themost detailed and accurate technique to those which are of reasonable butrelatively lower accuracy, which also means ranging from the mostcomputational intensive to the least [30]. In this regard, the DNS techniqueis not feasible for urban scale simulation [31].

We took four key climatic data for analysis, namely the annual meantemperature, the annual mean wind speed and direction, the air density andits dynamic viscosity of the annual mean temperature. Figure 5 shows the

! Figure 5.The National EnvironmentAgency and Department of Geographyweather data sites


03_Jun 08/06/11 9:23 am Page 40

sources of weather data in Singapore which is collected by the NationalEnvironment Agency (NEA) and can also be taken from the Department ofGeography of National University of Singapore.The NEA provides theaverage temperature for the area, which is 26.925 degree Celsius, about 300Kelvin.The air density is 1.184 kg/m3 and dynamic viscosity is 1.983 x 10-5kg/ms at 300 Kelvin.There are two predominant wind directions from theNorth-East at 2.4 m/s and South-East at 1.9 m/s.The middle picture inFigure 6 shows the site model being contained in a solid cuboid with thecorrect surface defined to represent the wind direction.The entire scenehas to be meshed before CFD solving takes place.

The results of the CFD analysis are shown in Figure 7. From the results, wecan immediately observe that the current urban design which is highlightedwith the square dotted line is creating a situation where heat and air flowwill be trapped, regardless of whether it is the North-East or South-Eastwind.This preliminary simulation highlights that planners and urban designswould have to respond with recommendations to alter the initial plannedproposal if it is the intention to improve the ventilation patterns in this site.From here, the process can be interactive, and various alternative planningand design variations can be generated and further tested.

" Figure 6. Buildings imported intoCOMSOL, defining wind direction onthe surface of the cube and meshing


03_Jun 08/06/11 9:24 am Page 41

Next, we did a validation in the HPC supercomputing environment togenerate results using ANSYS Fluent as shown in Figure 9.The processvalidates the lack of air movement at the same zone.Although it can domuch higher resolution airflow analysis in a shorter time, much work isinvolved in constructing the 3D model in ANSYS Gambit (3D modeling tool)to export to Fluent as it does not support most 3D CAAD formats.However, the benefit is that more of the other smaller zones around

! Figure 7. COMSOL North-East(above) and South-East (below) windflow analysis results


03_Jun 08/06/11 9:24 am Page 42

buildings with stagnant airflow can be identified in the exercise because ofthe larger number of iterations that can be done in a shorter time. Figure 8shows that it has other capabilities like checking the mesh for leaks beforeCFD simulation.This feature is good as it will save time later by avoidingproblematic wind flow simulations.

Apart from graphical visualizations, CFD can generate reports of thesimulation observed. Figure 10 and 11 shows the report generated from thesimulation.The graph shows the velocity magnitude with their distance tothe outflow of the wind from the site itself, after going through thebuildings.The charts show the minimum, maximum and average velocity ofthe atmosphere in the simulated sites as well as the histogram showing therange of wind speed (m/s) in the site.The figures show that the KallangRiver site maintains on average a higher amount of velocity from the north-east wind flow.At the same time, the lowest wind flow is also recordedfrom the same wind direction. In terms of overall percentage, the northwestwind direction would appear to have caused more areas of the site to havewind speed lower than 1m/s. In other words, the south-east wind directionresponds better to the Kallang River site by resulting in fewer spaces withlow wind speed.

" Figure 8. Gambit meshesexamination to avoid leaks


03_Jun 08/06/11 9:24 am Page 43

! Figure 9. Path lines coloured byVelocity Magnitude (m/s) of theNorth-East (above) and South-East(below) wind flow analysis in ANSYSFluent.The color of the lines rangesfrom red (higher velocity) to blue(lower velocity)


03_Jun 08/06/11 9:24 am Page 44

" Figure 10. North-East wind flow ofKallang River report in Fluent

" Figure 11. South-East wind flow ofKallang River report in Fluent


03_Jun 08/06/11 9:24 am Page 45

5.VISUALIZING THE RESULTS

Normally, the CFD analysis process ends at the post-processing stage,showing 2D still perspective images of the wind flow. Further than this, weproceeded to import the results into a visualization software like EnSight.EnSight can generate VR and animation visualization of the generated resultsfor presentation purposes. Figure 12 shows the interface of the KallangRiver model exported for VR and animation visualisations preparation.Thesevisualizations will help decision makers to see clearly how the spacedesigned is affected by the wind flow. It is a good preview of the design sointerventions can be made to further refine it.

! Figure 12. Kallang River 3D modelimported into EnSight

! Figure 13. EnSight VR visualizationwith the NVIDIA 3D Vision kit


03_Jun 08/06/11 9:24 am Page 46

Figure 13 shows how the process be visualized in the desktop environment,thus opening up the opportunities to reach out to the mainstream audiencein the design field. Figure 14 shows the output of the Kallang River CFDresults for presentation in EnSight in various views.

6. PROSPECTS AND CONCLUSION

The experimental work also demonstrated the speed of the simulations.Each of the wind direction simulations in the 32-bit COMSOL took about anhour to be completed.This quick process of getting the wind flow is veryencouraging and points to the great potential of integrating the process asan integral part of the urban design process.A lot of CFD software (ANSYSand CD-adapco software included) now have 64-bit desktop versions, whichmeans they are capable of utilizing more memory to generate morecomplex scenes.With these, we can already understand the airflowbehaviour and improve it at the early stage.Although we can get moreaccurate results by running more iterations, it is already very useful for theconceptual stage.This is a great improvement from the current processwhereby ventilation issues are addressed only at the later architecturaldesign stage when the planning guidelines are already determined.

Overall, the entire process is getting more affordable, easier and quickerto generate useful results on local desktops and workstations environments.It is also possible to do parallel CFD simulations in a professional practiceor research work environment by combining as many computers as possibleto share the simulation load for faster result generations.

ANSYS has now a framework called Workbench with the modeling toolDesignModeler integrated to support more 3D CAAD formats running on

" Figure 14. EnSight visualization ofKallang River CFD results from Fluentin top view, front view and bird eyes’view


03_Jun 08/06/11 9:24 am Page 47

Windows. DesignModeler makes it easier to generate 3D models than inGambit. Figure 15 shows the Workbench interface where the modeling,meshing, setup and solution of fluid flow in Fluent are all integrated into theWorkbench platform.The benefit of this is the ability to change parametersas well as compare between two simulations straight away, making thedesign process more parametric.

Figure 16 shows the side by side comparison of simulation A (northeastKallang River) against simulation B (southwest Kallang River) graphically inCFD-Post from what is connected in Figure 15 after simulation is done inFluent.The difference between them is generated in the last image below.With the ability to compare, we can further investigate the significance if weincrease the density (which can be done, for example, by increasing thebuilding heights) by comparing a precinct of plot ratio of 4 against 8, or bydifferent orientations and by different typologies.

! Figure 15.ANSYS Workbenchinterface comparing simulationA(North-East) and B(South-East)

! Figure 16.ANSYS CFD-Postinterface comparing simulationA(North-East) and B(South-East)


03_Jun 08/06/11 9:24 am Page 48

The potential of using VR software for CFD results for a morecomprehensive overall visualization is likely to become a norm in the future,especially for design professionals.They can analyze these results in 3D,propose alternatives and make various decisions before the actualimplementation processes take place. Other potentials of integration orrunning as stand-alone are in using Ecotect Analysis’ airflow capabilitiesintegration with the National Institute of Standards and Technology’s (NIST)Fire Dynamics Simulator and Smokeview (FDS-SMV), EnergyPlus’ naturalventilation simulation and IES VE-Pro’s Airflow CFD. Next Limit Technologies,the maker of the well known RealFlow software, is also releasing a new CFDsoftware named XFlow which may be faster to implement by using theparticle-based, fully Lagrangian approach. Finally, there is an open sourceCFD software called OpenFOAM (Open Field Operation and Manipulation)which is a great candidate for integration with Linux based CAD softwarelike Autodesk Maya and Side Effects Software Houdini. It is a free CFDsoftware package produced by OpenCFD Ltd. with features to solve complexfluid flows involving chemical reactions, turbulence and heat transfer, soliddynamics and electromagnetics. It includes tools for meshing, a parallelisedmesher for complex CAD geometries, and for pre- and post-processing.

Apart from just measuring wind against urban structures, designerswould need to take into account more realistic settings where the presenceof vehicles and trees canopy cannot be ignored [32].They would also needto take into account pollutants, especially from transport [33], momentumand heat concentrations [34].The building facade materials also need betaken into account for more realistic CFD simulation [17] instead of justthe default material for all solids. Finally, the relationship between outdoorand indoor CFD analysis which is lacking in current research will needfurther exploration.

In the end, detailed CFD simulation on the finalized design must be donein a more rigorous way using Realizable K-epsilon or Detached EddySimulation turbulence models to get more accurate answers for the windflow performance as well as covering a larger radius around the site to takeinto account the surrounding urban fabric’s impact towards wind flowperformance. More quantitative assessment of the wind flow performance(using wind velocity magnitude or wind velocity ratio) has to be done,especially at 2m pedestrian height. For the design process, finding therelationships between geometric variables with wind flow performanceindicators is useful as a quick estimation and prediction of the wind flowperformance before any simulation is done. Luc Adolphe [35] suggestedthree urban morphological parameters for wind flow, namely rugosity,porosity and sinuosity.The frontal area density [36], which is used toindicate the roughness of the urban fabric and to identify corridor paths, isa good geometric variable to understand the average façade areas that areblocking the wind from every direction. Explorations between therelationships between frontal area density [37] and ground coverage ratio as


03_Jun 08/06/11 9:24 am Page 49

well as wind velocity ratio [6] will be helpful for the use in the designprocess.

Harnessing the combined integration of Graphics Processing Unit (GPU)and CPU power for parallel computing, which is supported by CFD is nowa reality.The next step of our experimental research is to connect all thebridges from modeling to VR visualization of the CFD simulation results byusing the CPU and GPU combination processing. NVIDIA Tesla and ATIStream technology will give more mainstream users the capabilities toproduce CFD simulations.The NVIDIA (Compute Unified DeviceArchitecture) CUDA, which is a general purpose parallel computingarchitecture that leverages the parallel compute engine in NVIDIA GPUs tosolve many complex computational problems in a fraction of the timerequired on a CPU, is another potential language to embark in to get thefullest possible potential out of the GPUs.

Apart from these, we note Intel’s revelation of Larrabee in the firstquarter of 2010. Larrabee, which is a general-purpose computing on graphicsprocessing unit (GPGPU), will be the third competitor in the highperformance computing market. Indeed, this is an exciting time where CFDsimulations will no longer be in the realm of the inaccessiblesupercomputers. Finally, testing the simulations in Windows 7 will also beanother opportunity to be explored, with the potential for experimentationwith its multi-touch capabilities for VR visualization.

Today, 3D graphics is also moving into mobile devices with the availabilityof 3D enabled laptops with 120Hz monitors. ASUS was probably the earliestin the world to release the G51J-3D notebook in 2009. Since then, mostbrands are releasing their own versions along with the trend in 3D moviesand games. Figure 17 shows the world’s second 3D laptop by Toshiba with

! Figure 17.The Toshiba Satellite A665laptop with the NVIDIA 3D Vision kitmaking it convenient to visualize thesimulation in mobile


03_Jun 08/06/11 9:24 am Page 50

the 3D vision capabilities.Therefore, visualizing CFD results in the virtualreality environment in mobile is becoming a reality, thus making post-designdiscussions more convenient. In conclusion, the hardware is ready andgetting more affordable, and the convenience of setting up CFD simulationsis there. Hence, there should no longer be any excuse for not integratingCFD early into the planning and design process.

Acknowledgements

We would like to thank the lead specialist on High Performance Computing,Mr.Wang Junhong and chemical engineer Mr. Charles Hii Jun Khiong fortheir inputs on the CFD simulation.

References1. Lomax, H., Pulliam,T. H. and Zing, D.W., Fundamentals of Computational Fluid

Dynamics, Structural and Multidisciplinary Optimization, 24(6), 2002, 469-470.

2. Givoni, B., Climate Considerations in Building and Planning Design, John Wiley &Sons, Inc., 1997.

3. http://repository.tudelft.nl/assets/uuid:e06fb7be-e044-4a0d-9682-3aeadaf0a947/Koutamanis_2000_2.content.pdf [1-11-2009].

4. Tsou, J.-Y.,Applying Computational Fluid Dynamics to Architectural DesignDevelopment - Strategy and Implementation, CAADRIA ‘98 Proceedings of The ThirdConference on Computer Aided Architectural Design Research in Asia, Osaka, Japan,1998, 133-142.

5. Tsou, J.-Y., Lam, S. and Hall,T.W., Integrating Scientific Visualization with StudioEducation – Developing Design Options by Applying CFD, 21st Annual Conferenceof the Association for Computer-Aided Design in Architecture, Buffalo, New York, 2001,302-310.

6. Ng, E., Designing for Urban Ventilation, in Ng, E., ed., Designing High Density Citiesfor Social and Environmental Sustainability, Earthscan Publications Ltd, 2010.

7. Oke,T. R., Boundary Layer Climate, Methuen, 1987.

8. Cheng,V., Ng. E. and Givoni, B., Outdoor Thermal Comfort for Hong KongPeople:A Longitudinal Study, PLEA 2007: Proceedings of the 24th Passive and LowEnergy Architecture, Singapore, 2007.

9. Somarathne, S., Seymour, M. and Kolokotroni, M., Dynamic thermal CFDsimulation of a typical office by efficient transient solution methods, Building andEnvironment, 40(7), 2005, 887-896.

10. Steeman, H. J., Janssens,A., Carmeliet, J. and Paepe, M. D., Modelling indoor air andhygrothermal wall interaction in building simulation: Comparison between CFDand a well-mixed zonal model, Building and Environment, 44(3), 2009, 572-583.

11. Priyadarsini, R.,Wong, N. H. and Cheong, K.W. D., Microclimatic modeling of theurban thermal environment of Singapore to mitigate urban heat island, SolarEnergy, 82(8), 2008, 727-745.

12. Priyadarsini, R. and Wong, N. H., Building surfaces and their effect on the urbanthermal environment, Architectural Science Review, 48(4), 2005, 1-12.

13. Yoshie, R., Mochida,A.,Tominaga,Y., Kataoka, H., Harimoto, K., Nozu,T. andShirasawa,T., Cooperative project for CFD prediction of pedestrian windenvironment in the Architectural Institute of Japan, Journal of Wind Engineering andIndustrial Aerodynamics, 95(9-11), 2007, 1551-1578.


03_Jun 08/06/11 9:24 am Page 51

14. Tominaga,Y., Mochida,A.,Yoshie, R., Kataoka, H., Nozu,T.,Yoshikawa, M. andShirasawa,T.,AIJ guidelines for practical applications of CFD to pedestrian windenvironment around buildings, Journal of Wind Engineering and IndustrialAerodynamics, 96(10-11), 2008, 1749-1761.

15. Niu, J.-L., Some significant environmental issues in high-rise residential buildingdesign in urban areas, Energy and Buildings, 36(12), 2004, 1259-1263.

16. Takahashi, K.,Yoshida, H.,Tanaka,Y.,Aotake, N.,Wang, F.-L., Measurement ofthermal environment in Kyoto city and its prediction by CFD simulation, Energyand Buildings, 36(8), 2004, 771-779.

17. Ng, E.,An Investigation into Parameters Affecting an Optimum Ventilation Designof High Density Cities, The International Journal of Ventilation, 6(4), 2008, 349-358.

18. Ng, E., Policies and Technical Guidelines for Urban Planning of High DensityCities – Air Ventilation Assessment (AVA) of Hong Kong, Building and Environment,44(7), 2008, 1478-1488.

19. Priyadarsini, R. and Wong, N. H., Causes of urban heat island in Singapore, PLEA2009: Proceedings of the 26th International Conference on Passive and Low EnergyArchitecture, Quebec, Canada, 2009, 1-6.

20. Malkawi,A. M., Srinivasan, R. S.,Yi,Y. K. and Choudhary, R., Decision support anddesign evolution: integrating genetic algorithms, CFD and visualization, Automationin Construction, 14(1), 2005, 33-44.

21. Zhai, Z.-Q. and Chen, Q.-Y., Solution characters of iterative coupling betweenenergy simulation and CFD programs, Energy and Buildings, 35(5), 2003, 493-505.

22. Capeluto, I. G.,Yezioro,A. and Shaviv, E., Climatic aspects in urban design—a casestudy, Building and Environment, 38(6), 2003, 827-835.

23. Murakami, S., Environmental design of outdoor climate based on CFD, FluidDynamics Research, 38(2-3), 2006, 108-126.

24. http://www.ura.gov.sg/MP2008/kallang_riverside.htm [1-11-2009].

25. http://www.bd.gov.hk/english/documents/index_CSBD.html [1-4-2010].

26. http://www.scientificcomputing.com/extreme-resolution-visualization.aspx?terms=visualization [1 -11-2009].

27. Asfour, O. S. and B. Gadi, M. B.,A comparison between CFD and Network modelfor predicting wind-driven ventilation in buildings, Building and Environment,42(12), 2007, 4079-4085.

28. Srebric, J.,Vukovic,V., He, G.-Q. and Yang X.-D., CFD boundary conditions forcontaminant dispersion, heat transfer and airflow simulations around humanoccupants in indoor environments, Building and Environment, X43(3), 2003, 294-303.

29. Building and Construction Authority, Building planning and massing, Green BuildingPlatinum Series,The Centre for Sustainable Buildings and Construction, Singapore,2010.

30. Letzel, M. O.,An, X. and Ng, E, Modelling the air flows in Urban Places in HongKong using Large-Eddy Simulation, PLEA 2007: Proceedings of the 24th Passive andLow Energy Architecture,Singapore, 2007.

31. Michelassi,V.,Wissink, J.G. and Rodi,W., Direct numerical simulation, large eddysimulation and unsteady Reynolds-averaged Navier—Stokes simulations ofperiodic unsteady flow in a low-pressure turbine cascade: a comparison,Proceedings of the Institution of Mechanical Engineers, Part A: Journal of Powerand Energy, 217(4), 2003, 403-411.

32. Mochida,A. and Y.F. Lun, I.Y. F., Prediction of wind environment and thermalcomfort at pedestrian level in urban area, Journal of Wind Engineering and IndustrialAerodynamics, 96(10-11), 2008, 1498-1527.


03_Jun 08/06/11 9:24 am Page 52

33. Tsou, J.-Y.,Yimin, Z. and Lam, S., Improving Air Quality of Public TransportInterchanges - Design Strategies to integrate CFD simulation in early designprocess Architectural Information Management, 19th eCAADe Conference,Helsinki, Finland, 2001, 54-59.

34. Baik, J. J., Kim, J. J. and Fernando, H. J. S.,A CFD Model for Simulating Urban Flowand Dispersion, Journal of Applied Meteorology, 2003(42), 1636-1648.

35. Adolphe, L.,A simplified model of urban morphology: application to an analysis ofthe environmental performance of cities, Environment and Planning B: Planning andDesign, 2001(28), 183-200.

36. Wong, M. S., Nichol, J.E.,To, P.H.,Wang, J.A simple method for designation ofurban ventilation corridors and its application to urban heat island analysis,Building and Environment, 45(8), 2010, 1880-1889.

37. Ng, E.,Yuan, C., Chen, L., Ren, C. and Fung, C.H. J., Improving the windenvironment in high-density cities by understanding urban morphology andsurface roughness:A study in Hong Kong, Landscape and Urban Planning, 101(1),2011, 59-74.


Daniel Hii Jun Chung and Malone-Lee Lai Choo

Centre for Sustainable Asian CitiesSchool of Design and EnvironmentNational University of Singapore SDE1 #04-244 Architecture DriveSingapore 117566

[email protected], [email protected]

03_Jun 08/06/11 9:24 am Page 53

03_Jun 08/06/11 9:24 am Page 54

Computational Fluid Dynamics for Urban Design:The ...papers.cumincad.org/data/works/att/ijac20109103.pdf · gain as well as pollution accumulation can be minimized and/or to generate

Documents