GuamYNandFadeyiMO(2020).Optimisationofshelteredwalkwaysperformancetomitigate wind-drivenraininSingapore.BuiltEnvironmentAppliedResearchSharing#03,ISSUU DigitalPublishingPlatform. ©BEARSreservestherighttothisappliedresearcharticle
Yi-Ting Natalyn Guam and Moshood Olawale Fadeyi,*
Sustainable Infrastructure Engineering (Building Services) Programme, Singapore Institute of Technology, 10, Dover Drive, Singapore 138683, Singapore
*Corresponding author’s email: fadeyi.moshood@singaporetech.edu.sg
Sheltered walkways are common urban features in Singapore, as pedestrians utilise them as pathways to public transportation locations in Singapore. The walkways are essential in reducing users' thermal discomfort and wetness in a tropical environment like Singapore, with high solar intensity and frequent and heavy rains The need for protection from winddriven rain is of particular concern in the design of sheltered walkways in Singapore. This part of the pedestrian’s comfort is often neglected. Sheltered walkways are built with vertical and 30° angled rainfall in mind. The considerations of the influence of wind are often neglected in the design of sheltered walkways, which results in a larger area of walkways receiving a high amount of rain penetration. To better evaluate the design of sheltered walkways, it is crucial to understand the pedestrian’s perspective, the correlation between wind and rain, and its effects on sheltered walkway design. Findings from the use of 3-Dimensional Computational Fluid Dynamics (CFD) simulations of wind flow and winddriven rain for sheltered walkway designs are presented in this study. Additionally, a proposed microcontroller prototype, Arduino UNO, is built to validate the feasibility of the design solution. This study demonstrates the importance of considering wind-driven rain during the sheltered walkway design process.
Keywords: Shelterwalkways,Wind-drivenrain,LandTransportAuthority,Computationfluid dynamics, Prototype design, Walk2Ride
AWalk2RideprogrammeispromotedbytheLandTransportAuthority(LTA),whichprovides anetworkofshelteredwalkwaystoallowpedestrianstoenjoyconvenientconnectionstopublic transportation locations, no matter rain or shine. Given the tropical climate of Singapore, weatherprotectionisoneofthecrucialfactorsfortheconsiderationsinthedesignofsheltered walkways.However,theexistingshelteredwalkwaysdesignedbyLTA,arebuiltwithvertical and 30° angled rainfall in mind. The considerations of the influence of wind are neglected in the design of a sheltered walkway, which results in a larger area of the walkway receiving a high amountofrainpenetration,andcause discomforttopedestrians.
With the advancement in technologies, the use of CFD simulations to study the interaction between both wind and rain has been widely adopted in many environmentally sustainable design(ESD)firms.Theaerodynamicseffectsonbuildingshavealsobeenwidelystudiedusing full-scaleexperiments(Pabiouetal.,2015),windtunnelexperiments(DuthinhandSimiu,2011), and numerical models (Foroushani et al., 2014). However, there is minimal knowledge on the airflow behaviour of sheltered walkways in tropical climates. Subramanian et al. (2019) conducted subjective and objective comfort measurements to improve the performance of sheltered walkways in Singapore. Yet, the focus of the study is on the influence of solar heat gain on sheltered walkways. As the influence of rain is also an important aspect of pedestrian comfort in tropical environment like Singapore, 3D CFD studies on sheltered walkways are required for a detailed analysis of the wind flow and wind-driven rain patterns. As LTA has standardisedthedesignofshelteredwalkwaysacrossSingapore,similarwind-flowpatternsand wind-drivenraindistributionscanbepredictedbyperformingCFDsimulations.
Thisstudyaimstooptimisethe performance ofshelteredwalkwaysin Singapore,byproviding a solution to mitigate the effects of wind-driven rain on pedestrians utilising the sheltered walkways. As an initial step, a literature review and survey are conducted to understand better thebackgroundofshelteredwalkwaysandthepedestrian’sperspective.A3Dsteady-stateCFD Simulationisappliedtoastandaloneshelteredwalkwaytoinvestigatetheeffectsofwind-driven rain under various wind and rain conditions. The flow field is modelled using the ReynoldsAveraged Navier-Stokes (RANS) equation together with the �� �� turbulence model; where�� represents the turbulent kinetic viscosity and�� represents the turbulent dissipation (Chiu et al.,
2017). The choice of the turbulence model is based on the existing CFD Methodology for Singapore’s Building and Construction Authority Green Mark, which is further explained in Section3.2.Eulerianparticletrackingisadoptedforwind-drivenrainCFDsimulationtoobtain raindrop trajectories on sheltered walkways. The performance of the different mitigation methods is evaluated based on a decision matrix, as explained in Section3.3. Last, a prototype of a sheltered walkway is designed to conduct experiments and of showcasing the mitigation methods.
Thisstudyisalsointendedtosupportfutureshelteredwalkwayresearchanddesign.According totheministryoftransportSingapore,morethan60newmassrapidtransit(MRT)stationsare beingbuilt.Hence,100kmofthenewshelteredwalkwaywillbeconstructedby2029,toserve the extensive public transportation network. Consequently, research on sheltered walkways is expected to increase significantly due to the increasing numbers of new sheltered walkways planned. It is also crucial to recognise that rain penetration on pedestrians provides a severe level of discomfort; which is generally not considered in pedestrian comfort studies for shelteredwalkways.
This studyisconductedinthecontextof Singapore.Singaporeislocatedatthesoutherntipof Malay Peninsula, at1.29°North and 103.85°East;andisclassified undera tropical rainforest climate withno distinctive seasons. Itsclimate has a uniformtemperature,high humidity,and abundance of rainfall of about 2165.9mm. At the same time, the wind speed experiences significant variances due to the monsoon seasons over a year. The climatic data used in the studywereobtainedfrommeteoblue.
(a):FrontViewofShelteredWalkway
(b):SideViewofShelteredWalkway
Figure1:ModelledShelteredWalkwayConfiguration
The sheltered walkway configuration used in this study is based on the LTA and Urban Redevelopment Agency (URA) Walking and Cycling Design Guide for Covered Linkway (Figure 1a and 1b). A typical sheltered walkway has a dimension of 2.4m or 3m width and standing at 2.4m height (Figure 1a). Besides, it has a sloped roof with an inclination of 3° (Figure1b).Theheightandwidthratioisdesignedtoprotectpedestriansfromlightrain,while thedownwardslopedroof isdesignedtodirectrainfallintothepavementdrainagesystem.
As part of the research to assess the existing sheltered walkwaydesign to the general public, a surveywasconducted.Thegeneralpublicofallagegroups,livinginSingaporewasinvolvedin thesurveystudy.Respondentswereaskedtoanswertheeffectivenessoftheshelteredwalkway, duringaheavydownpour,basedontheirdailyroutine.
Apreliminarysurveywascarriedoutwithfive randomparticipantstoassessthe practicability and ease of understanding for each question. Based on the feedbacks retrieved from the five respondents, the questionnaires were adjusted and distributed to the general public. The questionnaire includes a prologue, which describes the purpose, background, and problem statement of this study. The prologue ensures that respondents understand the context of the survey before attempting it. During theconduct of the survey, a total of eight questions were asked to the general public. The flow chart of the survey questionnaire is given in Figure 2, whilethesurveyquestionsandtheiransweroptionsareaslistedinTable 1.
The questions in Table 1a were designed to get an overview of the respondent’s general feedback with regards to their daily routine. Respondents are questioned on their daily utilisation of sheltered walkways and the efficiency of the existing sheltered walkway design to determine the practical use of sheltered walkways in Singapore. This is carried out to understandbetterrespondent’spresentimpressionoftheexistingshelteredwalkwaydesign.It is important to note the respondent’s frequent means of transportation as well, as Sheltered Walkways are not only positioned near bus stops and MRT stations. Those travelling via private orpersonaltransportation,likewise, usetheseshelteredwalkwaysatotherlocations.
ThequestionsinTable1bweredesignedtogetanoverviewofthefeedbackfromrespondents whofrequently travelaround Singapore bypublictransportation. Itiscrucialtoconsidertheir feedbacksas shelteredwalkways are provided primarilyfor pedestrians travelingto busstops orMRTstations. Sincetheprimaryintentionofthe studyis toevaluatethe design ofexisting sheltered walkways to mitigate wind-driven rain, respondents were given a scenario-based question to comprehend their decision making in times of heavy downpour. Besides the scenario-based questions, respondents were questioned with an open-ended response on their decisioninutilisingtheshelteredwalkwaysduringtimesofheavyrain.ThequestionsinTable
1c were designed to get a comprehensive understanding of the respondent’s viewpoint of the efficiency of the existing sheltered walkways during times of heavy downpour. Respondents are posed with other scenario-based questions, and the questions are provided with on-site photographs.Thephotographsareprovidedtoensurethatrespondentsunderstandthesituation andanswerthequestionsaccordingly.
ThequestionsinTable1dweredesignedtogetanunderstandingoftherespondent’sviewpoint ofthenewshelteredwalkwaysdesign.Therefore,itiscrucialtoquestiontheirthoughtsonthe advantages of the existing sheltered walkways design. By recognising the advantages of existingshelteredwalkwaysdesign,certainfeaturesoftheexistingshelteredwalkwayscanbe incorporated into the new design. Since sheltered walkways are ultimately designed for the pedestrians living in Singapore, the design should appeal to these pedestrians. Thus, respondents were questioned on the type of designs they would appreciate, for the new shelteredwalkways.Theansweroptionsprovidedaretheexistingpassivedesignforrainshade intheresidentialareasandotherfacilities.Inaddition,respondentsweregiventheopportunity to provide interesting designs for the new sheltered walkways, via an open-ended answer option.
1a:SectionOne–GeneralFeedback
Question
1 Howoftendoyouutilizethe shelteredwalkways?
AnswerOptions
MorethanTwiceaday(Everyday)
Onceaday(Everyday) ThreeDaysaweek
OneDayaweek
2 Doyouthinkshelteredwalkways aredesignedefficiently? Yes No
3
Whatisyourfrequentmeansof transportationonaweeklybasis?
a. PublicTransportation(Bus,MRT,LRT) b. PrivateTransportation(Grab,Gojek,Taxi) c. PersonalTransportation(Car,Van,Motorcycle)
4
1b:SectionTwo–PublicTransportation Question
Inthisscenario,whichwillyou prefer?
5 Providesomereasonsonthe willingnesstotakePublic Transportationinsuchaweather
AnswerOptions
i. StillWalktotakePublicTransportation ii. JustbookPrivateTransportation
Open Ended Paragraph Answer
1c:SectionThree–Efficiencyof ShelteredWalkways Question
AnswerOptions
6
Inthisscenario,wouldyouprefera moreefficientshelteredwalkway? i. Yes ii. No
1d:SectionFour-ShelteredWalkwayDesign Question
AnswerOptions
7
Whatistheadvantageofexisting shelteredwalkwaydesign?
8
Whatkindofdesignswouldyou lovetoseeforthenewsheltered walkways?
Open Ended Paragraph Answer
RetractableRollerShadeattheside Louvresattheside
ExtendedWidthoftheShelteredWalkways
Others: Open ended Answer
Thesurveywasconductedforonemonth,fromthe9thofSeptember2019tothe9thofOctober 2019.Thesurveyreachedoutto208participants.Thesurveyisdistributedviadifferentmodes of social media, suchasWhatsApp andTelegram. Respondentscanclick the provided link in thetextmessageandattemptthesurveythroughanonlineplatform,GoogleForms.Thereturn ratioofthe surveywas89.4%,whichisatotalof186respondents.
Figure3ashowstherespondent’saverageusageofshelteredwalkways,basedonaweek.Most of the respondents (60.8%) stated they use of sheltered walkways for more than twice a day,
while22%oftherespondentsutiliseshelteredwalkwaysonceaday,bothdaily.Aminorityof respondents(7%)statedtheyuseshelteredwalkwaysforaboutthreedaysaweek,while10.2% of the respondents utilise sheltered walkways one day a week. Supplementary to the survey results,oneoftherespondentsexplainedthatthelackofuseisasaresultofabsenceofsheltered walkwaysinthevicinity.
The impressionofthe efficiencyoftheexistingshelteredwalkwaydesign isdependentonthe pedestrians, who utilise these sheltered walkways. From Figure 3b, more than half of the respondents(65.6%)agreedthattheexistingshelteredwalkwaysareefficient,while 34.4%of the respondents disagreed. This impression of anefficient sheltered walkway could be forged asaresultofrespondentsutilisingshelteredwalkways, particularlyduringnon-rainyperiods.
Themostcommonmodeoftransportationisbypublictransportsuchasbuses,MRTs,andlight railtransit(LRT).Tobespecific,respondentswhotravelviapublictransportationaccountsfor 78.5%, while those that travel via private and personal transportation accounts for 2.7% and 18.8%, respectively (Figure 3c). Considering the cost of owning a personal vehicle and the surgeinprivatetransportationpricesinSingapore,itisreasonablethatmostrespondentstravel via thecheapestformoftransportation.
BasedonFigure3c,atotalof146respondentsdeclaredthattheytravelviapublictransportation. However,whenrespondentsarefacedwithaheavydownpourscenarioinshelteredwalkways, 17.8% of respondents are more willing to book private transportation to travel to their destination (Figure 3d). The remaining 82.2% of respondents are still inclined to commuting via public transportation, despite the physical discomfort from the rain. Based on the openendedanswers,themajorityrespondedthattheyare notwillingtopayextrafortransportation as private transportation prices usually are hiked during periods of heavy downpour in Singapore.
Although65.6%ofthetotalrespondents(Figure3b)agreethattheexistingshelteredwalkways are efficient, the numbers drastically fell after respondents are placed in scenarios-based questions.AsshowninFigure4e,themajorityoftherespondents(95.7%)wanttohaveamore efficient sheltered walkway, while 4.3% of the respondents disapprove. The impression of respondents could have changed based onthe heavy downpour scenario, provided in some of
thequestions.Beforethesurveyquestionnairecomestoaclosure,respondentswererequested toselectthetypesofdesigntheywouldappreciate,forthenewshelteredwalkway.Sincethere are about 246km of existing sheltered walkways in Singapore, it is impractical to design entirelynewshelteredwalkways. Therefore, thetypes of designs to mitigate wind-driven rain shouldbeeasilyinstalledonexistingshelteredwalkways.Asshelteredwalkwaysaredesigned fortheuse ofpedestrians, respondentsmustbe involvedinthe designprocess of the sheltered walkways,ensuringthattheybenefitfromthenewshelteredwalkways.
Figure3a
Figure3b
Figure3c
Figure3d
Figure3e
Inrecentyears,thedevelopmentofcomputersimulationsenableslarge-scalesimulationstobe conducted to understand fluid flows, which allows a virtual analysis of the performance and effectiveness of the existing or intended design. Raman et al. (2018) explain the concept of CFDapplication,whichincludesacombinationofphysics,mathematics,computertechnology, and flow fields. CFD is capable of predicting the behaviour of fluids by solving the NavierStokesequationbysatisfyingthreeconservationlaws,whicharemass,momentum,andenergy.
To further understand how wind-driven rain affects the sheltered walkways, CFD simulations were conducted on a 3D modelled sheltered walkway. The type and methodology of the simulationstobeconductedisbasedonSingapore’sBuildingandConstructionAuthority(BCA) Green Mark Criteria, Technical Guide, and Requirements. Table 2 shows the atmospheric boundaryconditions, which provides an estimation of the suitable wind speedfor wind-driven rain simulation, based on National Environmental Agency’s (NEA) 32-year weather data at a reference height of 15m. Furthermore, BCA states the four different raindrop sizes to be analysed, 0.5mm, 1.0mm, 2.0mm, and 5.0mm. With that, wind-driven rain simulation was carriedoutinaccordancewithBCA’sstandardsforastrictidentificationprocessoftheseverity ofrainpenetrationinashelteredwalkway.
www.bca.gov.sg)
of simulation time. As it would be biased to simulate specific wind directions according to specific estates, the wind direction and speed will not be accurately simulated as per current environmental conditions. With that, to determine the best and worst-case wind-driven rain scenario, the lowest and highest wind speed from Table 2, is used as the simulation parameter (Table3a).Thewinddirectionisdecidedbasedonthemostimpactfulwind-drivenrainscenario, as seen in Figure 4a. The sheltered walkway is designed as a control model for all four simulationstoensuretheaccuracyofresults.
Shelteredwalkwaysaredesignedtoallowpedestrianstoenjoyconvenientconnectionstopublic transportation locations, no matter rain or shine. Hence, simulation must be performed on a model that includes both a sheltered walkway and a human to mimic the real-life experience. From this, simulation results will exhibit the severity of wind-driven rain, on both the human and sheltered walkway’s footpath. The simulation parameters for the human model simulation (Table3b)isdeterminedbytheworst-casewind-drivenrainscenario,whichis8.4m/s.
3a:ExistingShelteredWalkwayModel Wind Speed Wind Directions
2.9 m/s North South 8.4 m/s North South
3b:HumanwithShelteredWalkwayModel Wind Speed Wind Directions
8.4 m/s North South
The new sheltered walkway should be designed efficiently to mitigate wind-driven rain on pedestrians.Basedonthesurveyresults(Figure3),eachdesignsolutionissimulatedtoidentify themostdesirableshelteredwalkwaytomitigatewind-drivenrain.Thesimulationsperformed foralldesignsolutionsoughttohavethesamecontrolparameters,suchasthewindspeedand winddirection.Fromtheearliertwogroupsofsimulationmentioned,thewindspeedandwind direction adopted is 8.4m/s and South Wind, respectively. In addition, a fair evaluation is conducted byperformingthe simulation consistentlyon the same base model, which includes
bothshelteredwalkwaysandahuman.Figure4cshowstheillustrationoffourdesignsolutions implementedonanexistingshelteredwalkway.
Figure4a:BasicSimulationModel Figure4b:HumanSimulationModel
Figure4c:ShelteredWalkwayDesignSimulationModel
Figure4:SimulationModels
To assess the naturally ventilated building designs in Singapore, BCA launched a CFD VentilationSimulationMethodologyandRequirementin2008.AccordingtoPohetal.(2019), the CFD Ventilation Simulation Methodology takes reference from different international standards,toenhancethe limitedguidelinesonnaturalventilationinSingapore.Therefore,by integratingtheBCA Green Mark Scheme into theCFD simulationsof a sheltered walkway, a guided process is engaged to ensure the accuracy and quality of simulation results. An opensource CFD Solver, OpenFOAM, solve the performed numerical CFD simulation. The preprocessingofthenumericalCFDsimulationsisachievedbyusinggraphicalinterfacesoftware, BIMHVACTool,designedbyTianBuildingEngineering.
The geometry of the sheltered walkway itself has a dimension of 2.5m height and width. The length of the sheltered walkway is insignificant as the study is focused on the depth of rain penetrationonthewidthoftheshelteredwalkway’sfootpath.ToensurethattheCFDnumerical simulationisperformedaccurately,itrequiresacomputationaldomainthatsatisfiesthedirection blockageratioasrequiredbyBCACFDVentilationSimulationRequirements.Acomputational domainisdefinedasaregioninspaceinwhichnumericalequationsoffluidflowsaresolvedby CFD (Çengeland Cimbala 2006). Tominaga etal. (2008) pointed out that the frontal blockage ratioshouldbebelow3%toavoidanyartificialaccelerationofthefluidflowduringtheconduct of the CFD Simulation. The frontal, lateral, and vertical blockage ratio equations are provided by BCA, as shown below. The existing sheltered walkwaydesign has a computational domain dimension of 700m x 200m x 30m (L x B x H), while the human simulation model has a dimension of 250m x100mx 15m. Table 4 showsthe blockage ratioof the sheltered walkway simulations.
Measured from the direction of the wind flow
LateralBlockageRatio
Measured from the lateral direction of the wind flow
VerticalBlockageRatio
Measured from the tallest building in the domain
Existing Sheltered Walkway
2.86% 1.25% 8.33%
Sheltered Walkway with Human Geometry and Design 2.8% 2.5% 16.67%
With only the computational domain, it is impossible to solve the complicated Navier-Stokes equation. Hence, OpenFOAM is essential to assist in the discretisation process of the computationdomain. Discretisationapproximatesthe complexequations allowing them to be easilysolvedbycomputers(Sosnowski,2018).Therefore,thediscretisationprocessgenerates computational grids with a multiblockmesh generator (blockMesh) provided by OpenFOAM (Soner and Ozturan 2015). However, for complex geometries like sheltered walkways, it requires a new process known as the snappyHexMesh. It refines the existing mesh iteratively andintegratesthe meshto the surfaceofthegeometry. Toensurethequalityofthesimulation results, BCA has provided a computational grid size guideline (Table 5). The computational grid of the existing sheltered walkways and the sheltered walkway withhuman geometry can beobservedinFigure5.
Table5:RecommendedComputationalGridSize (source: www.bca.gov.com.sg)
This section intends to provide information about the specific boundaryconditions, turbulence models, and solver settings required to perform the numerical CFD simulations. To perform a wind-drivenraincase,3DCFDwindflowsimulationshouldbeperformedbeforehandtocapture the realistic wind flow patterns over the modelled sheltered walkway. Thereafter, a Eulerian multiphase model is performed to calculate the specific catch ratio distribution on the human bodyandfootpathsofshelteredwalkways.Thespecificcatchratioisdeterminedbythreefactors, the referenced wind speed, referenced rain intensity, and the horizontal distribution of 17 differentraindropletssizes(Blockenetal.,2005).
Majority of the 3D CFD wind flow studies are performed with a steady-state simulation approach. The CFD code, OpenFOAM, is used tosolve the 3D RANS equation togetherwith the realizable �� �� turbulence model. Van Hooff et al. (2011) explained the use of the realisable�� �� turbulence model to be directly relatable to the accuracy of the CFD wind-
driven rain simulation when compared to a full-scale measurement. Therefore, the use of the realisable�� �� turbulence model is adopted for this study. The parameters of the realizable �� �� turbulence model can be easily modified through the boundary conditions of the computational domain, according to the simulation requirements. The computational domain is bounded by six faces, where five boundary conditions are applied to each face. The computationaldomainanditsindividualfacesareshown inFigure6,whileTable6showsthe computationaldomain’sboundarycondition,providedbyalocalcompanyinSingaporecalled TianBuildingEngineeringPte Ltd.
Figure6:ComputationalDomain
Table6:BoundaryConditionsofComputationalDomain
At the inlet of the computational domain, it imposes a neutral Atmospheric Boundary Layer (ABL) to establish an environment that is similar to the natural environment that pedestrians experience. As stated by Poh et al. (2019), ABL is represented in the CFD simulation by an inbound vertical wind profile that is based on a Logarithmic Law with a reference height of 15m. Furthermore, the logarithmic wind profile determines the vertical distribution of the
Equation2:
average wind speed at different heights. The two equations provided below shows the derivation of the vertical wind profile (Figure 7). Equation two is applied to derive the wind speedatthe specifiedheight. ��∗ = �� ln ℎ+�� �� ��(��)= ��∗ �� ln ��+�� ��
Equation1: Where, ��∗ : ABLfrictionvelocity �� : vonKarmanconstant(0.42) �� : Aerodynamic roughnesslength �� : Specifiedvelocityatreferenceheighth ��: SpecifiedHeight
Figure7:VerticalWindProfile–2.9m/s(Left),8.4m/s(Right)
Since windissignificantlyinfluencedinthepresence ofnearbywalls, suchasthe bottom and solid geometry boundaries, a near wall treatment is applied using the standard wall-function method. Thewall-function methodmodelsthe velocityprofilefromthefirst point tothe wall. Thus, the near wall boundary conditions are provided by bridging the inner regions between wallsand developedturbulenceregions(LaunderandSpalding,1974).
3.2.2.2
Following the completed 3D CFD wind flow simulation study, the evaluation of wind-driven rain on the sheltered walkway is performed by a numerical CFD simulation that is presented based on the Eulerian multiphase model. The Eulerian multiphase model is applied instead of the Lagrangian multiphase model, as rain droplets in the Eulerian approach are regarded as a
continuum likewind, insteadof individual raindroplets. Huangand Li (2010)justifiedthatthe Eulerian multiphase model reduces the complexity of evaluating wind-driven rain parameters. Furthermore,wind-drivenrainsimulationsbasedontheEulerianMultiphasemodelareverified to be reliable and accurate when compared with the available numerical and experimental data (BlockenandCarmeliet,2007).
Fora wind-driven rain simulation, onlythe Top andFrontisspecifiedasan inletwhile the rest arespecifiedasawall.Attheinletofthecomputationaldomain,thespecificationoftherainfall intensity,raindropletssizes,andvelocityarerequired.Thevolumeofwaterperhourcanexpress rainfallintensity(mm/hr)inaunitarea.Forthestudy,arainfallintensityof25mm/hrisapplied foreachwind-drivenrainsimulationstudy.Therainfallintensityisselectedbasedonthehighest rainfallintensity,asshowninthepre-processinggraphicalinterfacesoftware,BIMHVACTool.
Todeterminethe severityofwind-driven rain onshelteredwalkways,17differentraindroplets sizesareinjectedintothecomputationaldomain.Kubilay(2014)explainsthatraindropletswith a diameter of smaller than 0.3mm are of nearly perfect spheres and also known as a drizzle. Hence, rain droplet sizes smaller than 0.3mm are not considered in this simulation study. In addition,consideringthatanyraindropletslargerthan6.0mmisknowntobreakapartwhenthey fall,theupperlimitforthelargestraindropletdiameterissetat6.0mm.
As the studyfocuses on wind-driven rain, an oblique rainintensityvector is formedas a result of the occurrence of wind and rain. Rain droplets motions for wind-driven rain is governed by gravity and drag forces, where the horizontal component shows the wind intensity on rain dropletsbasedondrag.Incontrast,theverticalcomponentshowsthehorizontalrainfallintensity based on gravity. By adopting the Eulerianmultiphase method, each class of rain droplet sizes is regarded as a different phase since similar rain droplet sizes interact with the wind flow in a similar manner (Kubilay, 2014). Once the governing equations for the rain phases are solved, thespecificcatchratiodistributionforeachraindropletsizecanbecalculated.Thespecificcatch ratiolocatesthecatchmentofraintrajectoriesontheshelteredwalkway.Thus,thespecificcatch ratioiscapableofdeterminingtheseverityofwind-drivenrainonshelteredwalkways..
As the sheltered walkways designs are evaluatedusing the BCA CFD Ventilation Simulation MethodologyandRequirementsTechnicalGuide,itrequirestheanalysisofonlyfourdifferent rain dropletsizes and their respective terminal velocity. In this section, three different groups of simulation results are evaluated. The three groups are existing sheltered walkways, a sheltered walkway with human geometry, and lastly, the different sheltered walkway design solutions. Thethree groupsareevaluatedbasedonthespecific catchratioof thedifferentrain dropletssizeanalysed.
Figure 8showsthe wind-flowpatternsof allthreegroupsonahorizontalplaneatthe height of 1.2m above ground level. From Figure 8a, due to the two different wind speed simulated, it utilisesadifferentvelocitymagnitudescaleduringthecomparison.Basedontheobservedwind velocityvectors,thecolumnsoftheshelteredwalkwayalterthewindflowpatterns,resultingin anincreaseinwindvelocityintheareasbetweenthecolumns.Thewindflowpatternissimilar across all four simulations of the basic simulation model as they are simulated with the same geometry. However, human geometry may change the wind-flow patterns in the sheltered walkways.TotesttheeffectivenessoftheshelteredwalkwaysthatLTAdesigned,thesheltered walkwayisalsosimulatedwithahumangeometry(Figure8b).
The addition of human geometry in the simulation drastically changes the wind-flow pattern aroundthecolumns.Asthenorthwindcomesintocontactwiththecolumnsfirst,windspeedis deceleratedbeforereachinghumangeometry.Subsequently,windvelocityisfurtherdecelerated. However,forthesouthwinddirection,theareabetweentheshelteredwalkwayroofandhuman geometry is significantly reduced. Based on the principals of fluid mechanics, a smaller area results in a higher velocity. Hence, the wind velocity above human geometry increased. Furthermore, theedgeofthe sheltered walkway createda corneringeffect, whichincreasesthe wind velocity as well. After testing the effectiveness of the sheltered walkways, the developed design solutions should mitigate wind-driven rain. Although the solution should be able to mitigate wind-driven rain, the design solution must also provide thermal comfort to the pedestrians.
From Figure 8c, the thermal comfort for the Louvres is the highest, as compared to the rest of
thedesignsolutions.Theshelteredwalkwaydesignedwiththelouvresshowssimilarwind-flow patternswhencomparedtotheexistingshelteredwalkway.Therefore,thereisalikelihoodthat the sheltered walkways will be affected heavily by wind-driven rain, due to the similarities in wind-flow patterns. However, when Roller Rain Shade is implemented, the wind-flow pattern significantly changed due to the blockage of wind, from the rain shade. A cornering effect is noticed on the corners of the rain shade, which increases wind speed as more wind enters the sheltered walkway. This results in the thermal comfort of humans, not being compromised. Otherthanthethree-designsolutionmentionedabove,theothersolutionsthathavesimilarwindflow patterns are the angled rain shade with louvres. As both the rain shades are designed the same, it is evident that the wind-flow patterns would be similar. However, as both the slanted rainshadeisconstructedwithdifferentangles,thewindspeedexperiencedcandifferslightly.
Figure9showsthevarioussimulationresultsgraph,wherethedepthofrainpenetrationisplotted against the rain droplet size. The values provided are estimated based on the wind-driven rain simulationsperformed.Thedepthofrainpenetration(Figure9a)differsgreatlybasedonthetwo windspeedssimulated.FortheNorthWindSimulation,at2.9m/swindspeed,thedepthofrain penetration is noticed to decrease gradually from 0.5mm to 6.0mm. Hence, with higher wind speed, wind-driven rain intensityincreases, whichresultsin the rain droplets travellingfurther.
FortheSouthWindSimulation,basedonthedepthofrainpenetrationvalues,itcanbededuced that rain penetrates fully onthe footpath due to the lack of columns placedatthe exterioredge oftheshelteredwalkway.Theplacementofcolumnsinshelteredwalkwaysisveryimportantas it alters the wind-flow patterns to reduce the depth of rain penetration into the footpaths of shelteredwalkways.
The graphin Figure 9bshows the depth of rain penetration against the raindropletdiameter of thefivedifferentdesignsolutions.The valuesprovidedare estimatedbasedonthewind-driven rainsimulationsperformed.Basedonthedepthofrainpenetrationresults,theRollerRainShade proved to be the most effective solution out of the five design solutions. Considering that the Roller Rain Shade encloses the entire sheltered walkway, it is natural that wind-driven rain is eliminatedfromtheshelteredwalkways.Basedonthesimulationresults,thelouvresaredeemed tobeanineffectivesolutionasthedepthofrainpenetrationisthehighestamongstallfivedesign solutions.
From on-site observations, one side of the sheltered walkway is blocked by the development. Hence, a Roller Rain Shade fixed with louvres is simulated. From the results, it shows that the louvresmitigatewind-drivenrainbetterwheninstalledwithrollerrainshade.Simulationresults mightdifferifthesimulationisperformedwiththeprevailingwinddirectionfacingthelouvres. From the survey results, one of the design solutions is to extend the width of the sheltered walkways.Therefore,aslantedrainshadeissuggested.However,therearetwodifferentangles used to perform the simulation. This is done to analyse the level of rain penetration when the rainshadeisslantedatdifferentangles.Althoughthedepthofrainpenetrationissimilarforthe smallerraindropletsizes,the45°slantedrainshadepreventsfurtherrainpenetrationfromlarger dropletsizes.
Figure9a:DepthofRainPenetrationforExistingShelteredWalkwayDesign
Figure9b:DepthofRainPenetrationforShelteredWalkwayDesignSolutions
Figure9:DepthofRainPenetrationVSRainDropletDiametergraph
AccordingtoBCA,fourraindropletsdiameterisrequiredforanalysis.Hence,basedonFigure 10,acomparisonofthefourraindropletdiameterismadeforeachscenario.FromFigure10a, althougheachwindspeedanddirectiondemonstratedifferentseverityofwind-drivenrain,rain droplet of diameter 0.5mm shows a significant impact of wind driven rain on all scenarios. Considering that the horizontal rainfall intensity decreases as the rain droplet diameter decreases, with the same amount of wind-driven rain intensity applied; the rain droplet is capable of travelling afurther distance. Additionally, it can be observed that the South Wind with a wind speed of 8.4m/s is the worst-case scenario, based on the four rain droplet sizes compared. Hence, similar to the graph above, the worst-case scenario is determined to be the southwind,at8.4m/s.
Sheltered walkways are designed for safe and comfortable use for pedestrians. Hence, it is important to observe the severity of wind-driven rain on human geometry itself. Figure 10b showsthecomparisonbasedontheindividualscenariotodeterminetheseverityofwind-driven rain. For both the north and south wind scenarios, although the severity of wind-driven rain decreases as the rain droplet size increases, the full body is still affected by wind-driven rain. However, the north wind scenario is observed to be the worst-case scenario. This might be causedbythepositionofhumangeometry,whichissituatednearertotheedgeofthesheltered walkway. Hence, it directly impacts human geometry, which causes it to be fully wet, regardlessoftheraindropletsize.
Figure 10c shows the severity of wind-driven rain on the five design solutions. The louvres managetomitigatewind-drivenrainbyreducingtheseverityofitonthehumanbody.Yet,the rain droplets of diameter 0.3mm to 2.0mm affects more than half of the entire human body. However, when the louvres are simulated with the Roller Rain Shade, it is seen to be very effective in mitigating wind-driven rain. In addition, a close comparison betweenthe 30° and 45°slantedrainshadeisobserved.Bothslantedrainshadeiscapableofmitigatingwind-driven rain effectively as the human geometry on both simulations is observed to be dry. However, the rain droplet of diameter 0.5mm affects human geometry at the calf area, at a negligible extent. Although the calf area is affected, survey results proved that pedestrians do not mind unlesstheirfullbodyisdrenchedfromthecauseofwind-drivenrain.
Allinall,basedonthedepthofrainpenetrationintotheshelteredwalkwayandthewind-driven rain impact on human geometry, the design solutions that are the most effective are the Roller Rain Shade and the 45° slanted Rain Shade with Louvres. In addition, based on the simulation results for Louvres with Roller Rain Shade, it is evident that if the sheltered walkways are designed with both the 45° slanted Rain Shade and Roller Rain Shade, it would still mitigate wind-driven rain efficiently. Considering that not all sheltered walkways are obstructed at the NorthandSouthside, itisessentialtoalsodevelopa passivedesignsolution,insteadofonlyan active design solution, which is the automated roller rain shade. Hence, both the Roller Rain Shade and 45° slanted Rain Shade with Louvres can be installed at all types of sheltered walkwaysinSingapore.
NorthWind,2.9m/s
NorthWind,8.4m/s
SouthWind,2.9m/s
SouthWind,8.4m/s
Figure10a:Severityofwind-drivenrainforexistingshelteredwalkway
BlueArearepresentsa heavierrainpenetration(Wet)whileGreyareasrepresentsrelativelylittleornorainpenetration(Dry)
Figure10c:Severityofwind-drivenrainonshelteredwalkwaydesign solution
Figure10:Comparisonoftheseverityofwind-drivenrainbasedondepthofrainpenetration
The development of a prototype is crucial as it is directly related to the effectiveness of the design solution. Theeffectiveness of theprototype is deemedbyperforming anactualscaleddownexperimentinthenaturalenvironment,alongwiththecompletedcomputersimulations. The prototype is constructed based on the simulation results (Section 3.3), where the final design solutionsaredeterminedtobepreferable tomitigatewind-drivenrainonpedestrians.
This sectionisintendedtohighlight thedevelopment ofthephysicalprototype. Asmentioned intheearlierpartofthestudy,solutionsshouldbedesignedtobeconvenientlyinstalledonthe existingshelteredwalkways. Asthedesignsolutionconsistsofbothpassive andactive design solutions, the two prototypes will be fixed on to the same main body to ensure consistency during the experiment. However, the automated roller rain shade also requires the use of an Arduino system, which is an open-source electronic prototyping platform. The Arduino prototypecontrolstheautomatedrollerrainshade,toallowtheplasticsheettorolldownwhen rain is detected. The Arduino prototype and its functions will be further explained in section 4.2.
Automaticrain-sensingrollingshadesareintroducedinthissectionofthestudy.Incomingrain is detected via the Arduino system, where the roller shade can be engaged when a heavy downpour is detected. The Arduino system consists of a raindrop sensor module that detects the intensity of the rainfall and acts as required. Rainfall intensity data is sent to the main Arduino UNO board, which engages the roller shade once requirements are satisfied. The 5V DCStepperMotorcontrolsthe rollerrainshade.
Thehardwarecomponentsusedinthisdesignareaslistedbelow.
ArduinoUNO
RaindropSensor
RaindropSensorModule
5VDCStepperMotor
ULN2003DriverModule
LEDIndicator
Buzzer
The system block diagram, flow chart, and overall circuitry of the Automated Roller Rain Shade are shownin Figure 12. The raindrop sensor detects the rain intensity while it’s sensor moduleacts asa resistor, whichchanges its resistance whenthe raindrop sensor circuit is wet ordry.TheArduinoUNOmicrocontrollerreceivesthesedatafromtheraindropsensormodule and transmits it to the ULN2003 Driver Module. The data retrieved by the ULN2003 driver modulecontrols the turning direction ofthe 5V DC stepper motor, whichis used to rotate the physical roller rain shades. When the 5V DC Stepper Motor is activated, the LED Indicator and Buzzer is activated at the same time as well. The LED Indicator and Buzzer ensures that allpedestriansnoticethattherollerrainshadeisactivatedandmoving.
Figure12a:SystemBlockDiagramofAutomaticRainsensingRollerShade
Figure12b:OverallArduinoCircuit
Figure12c:FlowChartoftheoverallArduinoSystem
Figure12:OverallArduinosystem
InthisArduinoprototype,theArduinoUNOmicrocontrolleristhemainbrainthatcontrolsthe ULN2003 Driver Module and the Raindrop Sensor Module. The main requirement for this system is the initialization of the raindrop sensor module and the ULN2003 Driver Module. Oncetherequirementsareappropriatelyset,thesystemcanoperatetogettheidealresults.
Theraindropsensordetectstherainintensityandsendsthedatacollectedtotheraindropsensor module. Thereafter, the retrieved data is transmitted to the Arduino UNO microcontroller. A setofcodesiswrittentotheArduinoUNOmicrocontrollertorecognisethedatareceivedfrom the raindrop sensormodule.ConsideringthattheRaindrop Sensor Module outputs the data in an analog voltage, the analog-to-digital converter in the Arduino UNO microcontroller reads thechangingvoltageandconvertsittonumbersforbettervisualisation.Thenumbersindicate the level of resistance of the raindrop sensor, ranging from 0 to 1024, where 0 represents the minimum limit, and 1024 represents the maximum limitthat the sensor can detect. To ensure that the raindrop sensor works efficiently and accurately, it is programmed to detect rain droplets on the raindrop sensor continuously. Therefore, the raindrop sensor automatically sendsdatatotheraindropsensormoduleevenwhenthereisnoraindropletdetected.
However,incaseswheretheraindropsensordetectsrainfall,theArduinoUNOmicrocontroller willsendtherequireddatatotheULN2003Drivermoduletoactivatethe5VDCStepperMotor. Anothersetofcodesiswrittentothe ArduinoUNO microcontroller,for the ULN2003Driver Module and 5V DC Stepper Motor, as they read a different library from the raindrop sensor module.ThesetofcodesallowstheArduinoUNOmicrocontrollertoinputasetofinstructions into the ULN2003 Driver Module, to activate the 5V DC Stepper Motor. The ULN2003 receivesstepsanddirectionalsignalsfromtheArduinoUNOmicrocontrollerandconvertsthe informationintoelectricalsignalstoactivatethe 5V DCStepper Motor.Theactivated5V DC Stepper Motor will rotate the rotorbased onthe numberof revolutions specified in the codes. Onerevolutionofthe rotorisequivalentto2048steps.SincetheRollerRainShadeisfixedto therotor,astherotorrotates,theRollerRainShadeisoperated.Additionally,oncethe5VDC StepperMotorisactivated,thebuzzerandLEDindicatorwillbeactivatedsimultaneously.
Based on the physical prototype developed, a physical experiment is conducted to determine the effectiveness of the active and passive design, in mitigating wind-driven rain. To prevent any interference of the experiment from the different design solutions, the experimentfor the automated roller rain shade prototype is conducted before the construction of the 45° Slanted RainShade with Louvres.Thereafter,the45°SlantedRainShadewithLouvres isconstructed onto the sheltered walkway prototype, where the experiment proceeds. The experiment is conductedinanaturalenvironmentwhereakestreldeviceisutilisedtomeasurethewindspeed, while a handheld fan is utilised to blow wind to the prototype. Rain is simulated by using a Handheld Pressure Sprayer to spray water onto the prototype. Similar wind speed reading is achievedtoprovethattheexperimentisconductedsimilarlyandaccuratelyforthecomparison ofresults.
Boththeautomatedrollerrainshadeand45°SlantedRainShadewithLouvresexperimentare conductedvia thesameprocessaswell.Thisistoensurethattheexperimentisdonesimilarly andaccuratelyforthecomparisonofresults.Theprocessisasdescribed:
1) Placethe prototypeatthesamelocationthatisblockedbynaturalwind
2) Standabout1mawayfromtheprototype
3) HoldtheHandheldPressureSprayeratchestlevel
4) Spraythe prototype forabout5seconds
Thefirstexperimentconductedwiththeautomated rollerrainshadeistotesttheefficiencyof the active design and, at the same time, the efficiency of the Arduino prototype. All of the hardwarecomponentsanddevicesareconnectedandfixedonthephysicalprototype.Thetotal number of revolutions required is calculated at 3.5 revolutions to ensure that the roller rain shade can cover the height of the sheltered walkway. This is done based on the height of the shelteredwalkwayandtherevolutions oftherotor. Astherain sensoris required tobe placed at a position that detects rain droplets easily, it is placed on top of the roof of the sheltered walkway.Thesecondexperimentisconductedwiththe45°SlantedRainShadewithLouvres, totesttheeffectivenessofthepassivedesign,inmitigatingwind-drivenrain.Figure13ashows the overall experimental setup for the experiment conducted with the automated roller rain shadeand45°SlantedRainShadewithLouvres.
TheexperimentalprocessforbothexperimentscanbeseeninFigure14.Astheraindropsensor detectsrain,the5VDCStepperMotorisengagedinextendingtherainshades.Thereafter, whenthesimulatedrainisstopped,therainsensorsensesnorain,the5VDCStepperMotoris activated,andtherainshadeisretractedtoitsoriginalposition.Ontheotherhand,the experimentconductedforthe45°SlantedRainShadewithLouvresissimple,asthereisno automationrequiredforit.
RainDroplets RainDroplets
Figure15showsthewind-drivenraineffectsonthefootpathoftheshelteredwalkwayafterthe experiment.BasedonFigure15,asmallamountoftinydropletsarenoticedonthefootpathof theshelteredwalkway.Sincetherainissimulatedbeforetherainshadeisactivated,tinydroplets fromtherainareexpected.However,forthe45°SlantedRainShadewithLouvres,itisseen thattherearesmalldropletsofwateratthebottomofthelouvres.Consideringthattherainshade isconstructedata45°angle,waterdropletsmighthavefloweddowntherainshadetothe footpathoftheshelteredwalkway.Furthermore,basedonthesimulationresults,theareabelow thelouvresareseentobewetaswell.Hence,boththeexperimentalresultsresonatewiththe simulationresults,asexplainedinsection3.3.3. AutomatedRollerRainShade(Left),45
Thissectionisintendedtodiscusstheoutcomesgatheredfromthesurvey,CFDsimulations,and lastly,theexperiments.Basedontheoverallstudy,wind-drivenrainissuccessfullymitigated fromthedifferentpassiveandactivedesignsolutionsconstructed.Thiscanbeseenfromthe simulationsandexperimentsconducted.Otherthantheexplanationsandoutcomesofthestudy, somelimitationsandrecommendationswillbeaddressedaswell.
Thesurveyresultsprovideanoverallevaluationofthesignificanceofshelteredwalkwaysin Singapore.Fromthesurveyresults,theeightquestionshaveprovidedsubstantialevidenceand rawdataonshelteredwalkways,inthisstudy.Basedonthestatisticalanalyticalresults,anindepthunderstandingofthepedestrian’sperspectiveisprovided.Furthermore,itprovedthat
there is a need to tackle the issues of wind-driven rain on sheltered walkways to ensure that pedestrians travel comfortably and safely. Hence, the survey results justify the problem statement,aspedestrianspreferamoreefficientshelteredwalkwayduringrainydays.Although five participants have practically assessed the survey questionnaire, the conducted survey questionnaire can be further improved. This is to ensure that all respondents have an in-depth understanding of each question, which lowers the probability of respondents answering incorrectly. Hence, a clearer explanation had to be provided to those respondents before they can continue attempting the questionnaire. To help respondents answer the questions accurately,answeroptionsprovidedshouldbeclearandprecise.
One limitation of the survey results would be the open-ended answers that respondents have providedforthefewquestionsthatrequirealongtextanswer.Sincerespondentscanfreelypen down their thoughts, their answers might not be relatable to the question. Thus, this affected the accuracy of selecting the main design tofocus on. With that, the decision-making process wasdifficult,andsimulationshadtobeconductedforallthreedifferentansweroptionstodeem the most efficient sheltered walkway design. One recommendation to prevent this limitation would be to remove the open-ended answer option, where respondents are fixed to only three answer options. Thus, this would provide an accurate representation of which sheltered walkwaydesign therespondentswouldappreciate.
The study uses CFD Simulations to assess the distribution of wind-driven rain on the sheltered walkway.Thesimulationstudyusesasteady-statesolver,whichthereforerepresentsanaverage value of the wind-flow. However, it provides a general guide of the influence of wind-flow on the shelteredwalkwayduringthe rainyperiod. Fromthe simulation studyperformed, itproved that the existing sheltered walkway design by LTA and URA are not up to standard as the sheltered walkways are strongly influenced by wind-driven rain. In addition, the study also showedthat the passive design of the sheltered walkway roof could significantly affectthe 3D wind-flowpattern,whichresultsinadifferentinfluenceinthesimulationresultsofwind-driven rain. Based on the design solutions as proposed in Section 5.3, the designs are evaluated by a decision matrix that consists of six components. Table 7 shows the overview of the wetted sheltered walkway area, while Table 8 shows the decision matrix and its components. The simulationresultsoftheshelteredwalkwaydesignshavebeenexplainedearlierinSection3.3.3.
Table8showsthecriteriaandweightageofthedecisionmatrix.Aseachcriterionhasadifferent level of importance, with relevance to the sheltered walkways, a weighing column is implementedtoconsidertheimportanceofspecificcriteria.Thecriteriaweightingusesasimple scale from 1 to 6, where 1 is the least important, and 6 is the most important. In addition, the scores of each sheltered walkway are on a scale of 1 to 5, where 1 is the unfavorable solution, and5isthebestsolution. Therefore,theshelteredwalkwaydesignwiththehighestscoreisthe mostfavorabledesignsolution.
FromTable8,themostfavorabledesignsolutionistheRollerRainShade,followedbythe45° SlantedRainShadewithlouvres.Themostimportantcriterionislistedasthe“impactofrainon pedestrians”. Since thefocus ofthe studyistheimpactofwind-drivenrain,the“impactofrain onpedestrians”shouldberankedfirst,andthe“depthofrainpenetration”isrankednext.Based onbothcriteria,theRollerRainShadescoredthehighestscorewhilethelouvresdesignscored thelowest,astheRollerRainShadedesigncanmitigatewind-drivenrainefficiently.Thelouvres, ontheotherhand,provedtobeineffectiveasthepedestriansarestillprimarilyaffectedbywinddrivenrain.However,whenthedesignsolutionsarecomparedtotheThermalComfortcriterion, the louvres design appears to have the highest score. The Thermal Comfort of each design is shown by the wind-flow patterns seen in Section 3.3.1. Due to the lack of obstructions, the louvresdesignbringsinmorewindtotheshelteredwalkway.However,whatissurprisingwould be the Roller Rain Shade design. Despite it being concealed, the wind is consistently entering theshelteredwalkway,ensuringthecomfortofthepedestrians.
Besides the comfort of the pedestrians, cost and ease of implementation are important in determining the feasibility of design solutions. As the Roller Rain Shade uses an active design solution,thecostofimplementationwouldbemoreexpensive,ascomparedtotheotherpassive design. Hence, the Roller Rain Shade scored the lowest as compared to the louvre design. In addition, the Roller Rain Shade with louvres scored the lowest in the maintenance and ease of implementation, as it requires two different design solutions. With that, various parties are neededtofixandmaintainthedesignsolutions,whichisatediousjob,ascomparedtotheother design solutions that require only one party. For both the 30º and45° Slanted Rain Shade with louvres, they are ranked equally as the materials and cost are the same, with the exceptionthat theyareconstructeddifferently.
Although the CFD simulation results proved the efficiency of the Roller Rain Shade and 45° SlantedRainShadewithlouvres,inmitigatingwind-drivenrain,theCFDsimulationsforwinddrivenrainwereconductedforonlyonereferencewindspeed(8.4m/s)andonewinddirection (South). Due to the different varying parameters, such as the different wind speed and wind direction in Singapore, to determine the effectiveness of a sheltered walkway design would have resulted inan excessive amount of computations and results. However, the limitation of having simulated one wind speed and wind direction would affect the estimated wind-driven rain results. As the aim of the study was to understand the effects of wind-driven rain on sheltered walkways in Singapore, further insights of the sheltered walkway design solutions should be obtained by comparing simulations with different wind speed and wind directions. Withthat,acompletesetofsimulationswithvaryingwinddirectionandspeedwouldprovide anestimateoftheimpactofwind-driven rainonpedestrians.
Theexperimentalresultsprovideanoverallevaluationoftheeffectivenessofthetwosheltered walkway design solutions. From the experiment conducted in Section 5, the two sheltered walkwaydesignsolutionshaveprovidedsubstantialevidenceontheeffectivenessinmitigating wind-driven rain, in thisstudy. Based on the experiment, both design solutions are conducted withthesamecontrolparameterstoensuretheaccuracyoftheexperimentalresults.Withthat, the experiments conducted can be easily compared, as they are completed under the same parameters. However, what was unexpectedabout the experimental resultsis thatit was quite similartowhatisseenfortheCFDsimulationresults. Thisprovedthatthephysicalprototype was constructed in accordance with the 3D model, which ensured the accuracy of the experimentaswell.
Although the experiments were performed similarly, there are several limitations that should be considered. To achieve an experiment that has added accuracy, the experiment should be conductedinacontrolenvironment.Acontrolledenvironmentcanincludeanacrylicboxthat is covering the physical prototype. With that, no additional natural wind will affect the experimental results, resulting ina greater extentof accuracy. In addition, theexperimentcan be performed with a count-down timer to time the length of spray. As the experiment was conducted based on the ability of the user to count, it might not be as accurate as the countdowntimer.
AlthoughsomestudiesonshelteredwalkwaysinSingaporewereperformedinthepast,3DCFD Simulation studies of wind-driven rain have not been performed. A full comfort study is conductedonshelteredwalkways,toassesstheseverityofwind-drivenrainonthefootpathsof sheltered walkways and the pedestrians. The purpose of this study is, therefore, to provide insightsintothewind-flowpatternsandwind-drivenraindistributionontheshelteredwalkways.
With that, government agencies such as LTA and URA can design sheltered walkways that provide comforttous,pedestrians.Sinceshelteredwalkwaysare designedforpedestrians,they mustplayapartindecidingif the existingshelteredwalkwayisefficientenough.Accordingto the results of the existing sheltered walkway design, 95.7% of the respondents wish to see a change in the current design. Based on the survey results, a series of new sheltered walkway designsaremodelled,toensurethatitefficientlymitigateswind-drivenrain.
Thenewshelteredwalkwaysdesigninthestudywereevaluatedbasedontheirabilitytomitigate wind-driven rain through a CFD simulation. However, the best-sheltered walkway design is determined by different design criteria, which are evaluated based on a decision matrix. From thehighwindspeedandwinddirectionproposedinthestudy,theapprovedshelteredwalkways designs are the Roller Rain Shade and the 45° Slanted Rain Shade with louvres. Both of the designs performed the best in mitigating wind-driven rain while ensuring that the cost and maintenance of the installation are affordable. In addition to that, the physical experiment is performedtotesttheabilitiesoftheactualproduct.Unexpectedly,the resultsoftheexperiment coincidewiththeCFDsimulationsresults.Therefore,itprovedthatboththeactiveandpassive sheltered walkway design could mitigate wind-driven rain efficiently by ensuring a safe and comfortablecommuteforpedestrians.
ThesupportoftheSingaporeInstituteofTechnologyincarryingoutthisappliedresearchstudy isgratefullyacknowledged.Ms.Yi-TingNatalynGuamdidtheworkandcontentsofthispaper as part of her BEng final year design project in the Sustainable Infrastructure Engineering (BuildingServices) programme. Dr. MoshoodOlawale Fadeyi guidedthedevelopmentof the prototype solutionandexperimentaldesigntotesttheeffectivenessofthe developedsolution. Dr. Fadeyialsocontributedtothedevelopmentofthisarticle.
Blocken, B., and Carmeliet, J. (2007). On the errors associated with the use of hourly data in wind-driven rain calculations on building facades. Atmospheric Environment, 41(11), 23352343.
Blocken, B., Carmeliet, J., and Poesen, J. (2005). Numerical simulation of the wind-driven rainfalldistributionoversmall-scaletopographyinspaceandtime. Journal of Hydrology,315(14),252-273.
Çengel, Y.A., and Cimbala, J.M. (2006). Introduction to Computational Fluid Dynamics. In FluidMechanics:FundamentalsandApplications,Firsted.,NewYork,McGraw-Hill.
Chiu,P.H.,Raghavan,V.S.,Poh,H.J.,Tan,E.,Gabriela,O.,Wong,N.H.,...andLeong-Kok, S. M. (2017).CFDmethodologydevelopmentforSingaporeGreenMarkBuildingapplication. Procediaengineering,180,1596-1602.
Duthinh,D.,andSimiu,E.(2011).TheUseofWindTunnelMeasurementsinBuildingDesign. WindTunnelsandExperimentalFluidDynamicsResearch,282-300.
Foroushani, S. S. M., Ge, H., and Naylor, D. (2014). Effects of roof overhangs on wind-driven rain wetting of a low-rise cubic building: A numerical study. Journal of wind engineering and industrialaerodynamics,125,38-51.
Huang, S. H., and Li, Q. S. (2010). Numerical simulations of wind-driven rain on building envelopes based on Eulerian multiphase model. Journal of Wind Engineering and Industrial Aerodynamics,98(12),843-857.
Kubilay,A.(2014).Numericalsimulationsandfieldexperimentsofwettingofbuildingfacades duetowind-drivenraininurbanareas(Doctoraldissertation,ETHZurich).
Launder, B. E., & Spalding, D. B. (1983). The numerical computation of turbulent flows. In Numericalpredictionofflow,heattransfer,turbulenceandcombustion(pp.96-116).Pergamon.
Pabiou, H., Salort, J., Ménézo, C., andChillà, F. (2015). Naturalcross-ventilationof buildings, anexperimentalstudy.EnergyProcedia,78,2911-2916.
Poh,H.J.,Chiu,P.H.,Ooi,C.C.,Raghavan,V.,Wan,S.,Xu,G....andMing,L.K.S.(2019). Development of GM2015 Computational Fluid Dynamics (CFD) Methodology for NaturallyventilatedNon-residentialBuildings(NRB)in Singapore. InIOPConferenceSeries: Earth and EnvironmentalScience(Vol.238,No.1,p.012079).IOPPublishing.
Raman, R.K., Dewang, Y., and Raghuwanshi, J. (2018). A review on applications of computationalfluiddynamics.InternationalJournalofLNCT,2(6),137-143.
Soner, S., and Ozturan, C. (2015). Generating multibillion element unstructured meshes on distributedmemoryparallelmachines.ScientificProgramming,2015. ArticleID437480.
Sosnowski, M. (2018). The influence of computational domain discretization on CFD results concerning aerodynamics of a vehicle. Journal of Applied Mathematics and Computational Mechanics,17(1),79-88.
Subramanian, R., Tunçer, B., and Binder, A. (2019). Thermal Comfort Based Performance Appraisal of Covered Walkways in Singapore. CAADRIA 2019, Victoria University of Wellington,Wellington,NewZealand.
Tominaga, Y., Mochida, A., Yoshie, R., Kataoka, H., Nozu, T., Yoshikawa, M., & Shirasawa, T. (2008). AIJ guidelines for practical applications of CFD to pedestrian wind environment around buildings. Journal of wind engineering and industrial aerodynamics, 96(10-11), 17491761.
Van Hooff, T., Blocken, B., & Van Harten, M. (2011). 3D CFD simulations of wind flow and wind-driven rain shelter in sports stadia: influence of stadium geometry. Building and Environment,46(1),22-37.