1
DESIGNANDCONSTRUCTIONOFALOWSPEEDWINDTUNNEL
By
JonathanD.Jaramillo
Athesissubmittedinpartialfulfillmentoftherequirementsforthedegreeof
BachelorofScience
HoughtonCollege
May2017
SignatureofAuthor…………………………………………….………………………………….DepartmentofPhysics
May13,2017……………………………………………………………………………………………………………...
Dr.KurtAikensAssistantProfessorofPhysics
ResearchSupervisor……………………………………………………………………………………………………………...
Dr.BrandonHoffmanAssociateProfessorofPhysics
2
DESIGNANDCONSTRUCTIONOFALOW
SPEEDWINDTUNNEL
By
JonathanJaramillo
SubmittedtotheDepartmentofPhysicsonMay13th,2017inpartialfulfillmentofthe
requirementforthedegreeofBachelorofScience
Abstract
Despiteadvancesincomputationalaerodynamics,windtunnelsareandwillcontinuetobe
acornerstone in thedesignprocess forawiderangeofvehicles.Thismostlystems from
thedifficultiesofaccuratelyandefficientlypredictingturbulentflowfieldscomputationally.
Toexpandin-houseaerodynamicscapabilities,ageneral-purposelow-speedwindtunnelis
beingdesignedandbuiltintheHoughtonCollegePhysicsDepartment.Thiswindtunnelis
designed to reach test section speeds of up to 44.7m/s (100mph). To aid in the initial
design, semi-empirical formulas are used to estimate aerodynamic efficiencies and the
required fan-blower power as a function of various design choices. Tunnel geometry is
selected to optimize test section air flowquality, test section size, anddiffuser angle (to
avoidboundarylayerseparation),whiletheoveralltunnelsizeisconstrainedtofit inthe
allottedlaboratoryspace.Theproposedclosed-circuitwindtunnelisverticallyorientedto
reducefootprint,andis4.72m(15.5ft)longby1.67m(5.5ft)by0.762m(2.5ft)wide.The
overall design is presented and current design and construction progress is highlighted.
Additionally,futureresearchstudiesthatcouldutilizethewindtunnelarediscussed.
ThesisSupervisor:Dr.KurtAikensTitle:AssistantProfessorofPhysics
3
TABLEOFCONTENTS
Chapter1BackgroundandMotivation...................................................................................51.1. ApproachestoFluidMechanics.................................................................................51.2. HistoryofExperimentalAerodynamics..................................................................91.3. Objective.........................................................................................................................11
Chapter2Theory........................................................................................................................132.1. FluidMechanics............................................................................................................132.1.1. TheoreticalDevelopment.................................................................................................132.1.2. Quasi-One-DimensionalFlow.........................................................................................142.1.3. Bernoulli’sPrinciple...........................................................................................................17
2.2. WindTunnels................................................................................................................182.2.1. ImportantParameters.......................................................................................................182.2.2. BasicDesignDecisions......................................................................................................212.2.3. SectionLossandPowerConsiderations....................................................................242.2.4. FansandDriveSystems....................................................................................................312.2.5. ConstructionRequirements............................................................................................31
2.3. MeasurementTechniques.........................................................................................322.3.1. PressureMeasurement.....................................................................................................322.3.2. VelocityMeasurement.......................................................................................................34
Chapter3DesignConsiderationsandMethodology.......................................................373.1. BasicDesignDecisions...............................................................................................373.2. Constraints,Methodology,andDesign.................................................................383.3. CurrentProgress..........................................................................................................40
Chapter4ConclusionsandFutureWork............................................................................464.1. Conclusions....................................................................................................................464.2. FutureWork..................................................................................................................46
4
TABLEOFFIGURES
Figure1.AnexampleofagridmeshfortheNACA0012airfoil...................................................6Figure2.AdiagramoftherelativeaccuracyandcostofvariousCFDmethods....................8Figure3.AsketchofBenjaminRobin’swhirlingarm....................................................................10Figure4.AsketchoftheWrightbrothers’windtunne.................................................................11Figure5.Quasi-one-dimensionalflowthroughaduct...................................................................16Figure6.AnaerialviewoftheNationalFull-scaleAerodynamicsComplex........................22Figure7.TheNACA6-inchopenthroatwindtunnel......................................................................23Figure8.AschematicoftheUCDavisAeronauticalWindTunnelFacility...........................24Figure9.Adepictionoftheconicalangleforacone-shapeddiffuser.....................................27Figure10.Anexampleofcamberedairfoilturningvanesinacorner....................................28Figure11.Simplifieddiagramofau-tubemanometer..................................................................33Figure12.Asimplifieddiagramofapitot-statictube....................................................................35Figure13.Schematicoftheoverallwindtunneldesign................................................................38Figure14.Asketchoftheinsidedimensionsofthewindtunnel..............................................41Figure15.Thesteelframetosupportthewindtunnelwhencomplete................................42Figure16.AphotoofthefanandaCADdrawingofthefan........................................................43Figure17.Custom-madesheetmetalcomponentandsheetmetalturningvanes............43Figure18.Apreliminarydiagramoftheproposedwindtunnel...............................................47Figure19.ResearchersatNASAGlennproduce3-dscansoficeaccretiononairfoils....48Figure20.Ademonstrationofaplasmaactuatoractivelycontrollingtheairflow..........49
5
Chapter1
BACKGROUNDANDMOTIVATION
1.1. ApproachestoFluidMechanics
Since the timeof theWrightbrothers, theneed for reliableaerodynamicpredictionshas
only increased. Demand from both military and private aeronautics sectors has driven
rapid increases in flight technology. Similarly, the automotive, nautical, civil engineering,
andmanyotherindustrieshaveincreasedthedemandforacomprehensiveunderstanding
of fluid flowaroundobjectssuchascars,shipsandbuildings.Therearethreetechniques
frequentlyusedforsolvingaerodynamicdesignproblems–theoretical,computational,and
experimental–eachhavingdistinctadvantagesanddisadvantages[1][2].
Theoreticalmethodsrefer toanalyticalapproaches to solving thegoverningequationsof
fluiddynamics.Thedistinctadvantagetoanalyticalmethods isthattheyresult inclosed-
formsolutions.Theformulasobtainedfromananalyticalapproachcangiveagreatdealof
insight as to how the system will behave under different conditions. Unfortunately,
analytical solutions are often impossible to obtain, except in caseswhere the governing
equationsare linearorcanbeapproximatedassuch.Onewaytoobtain lineargoverning
equations is to consider only simple geometries, such as flow through an infinitely long
pipe. That said, many problems of interest involve complex geometries such as cars or
planes.However,theoreticalapproachesareusefulforgainingintuitionforhowgivenflow
fieldswillbehaveandunderstandingbasic flowphenomena.Theoreticalapproachesalso
continue to play a vital role in validating computational methodologies. Because
computational solutions are approximations, it is helpful to test the accuracy of
computationalalgorithmsonproblemsforwhichthereareanalyticalsolutions.
Thesecondapproachtotheaeronauticdesignprocessiscomputational,oftenreferredto
as computational fluid dynamics (CFD). These methods use computer algorithms to
approximatethesolutionstothegoverningequationsoffluidmechanics.Foragivenflow
problemthegeometrymustbemodeledusingCAD.Initialandboundaryconditionsmust
6
thenbedefinedandtheflowdomainmustbediscretizedintoagrid.Inmanyapproaches,
the solution is assumed tobe constant in each cell and thegoverningequations are also
discretizedtomodelhoweachcell'ssolutionchanges.Anexampleofagridforanairfoilis
showninFigure1.Onceanalgorithmischosen,thesimulationcanbeperformed.
Figure1.Anexampleofagridmesh for theNACA0012airfoil.Gridpointdensity increasesnear the leading and trailing edge of the airfoil becausethe solution near these points varies greaterwith space and time. FiguretakenfromRef.[3].
ThefieldofCFDhasseentremendousgrowthinthepastfiftyyearsduetothecontinuous
increase in computational power and memory of computers. The rate of computer
evolution during the 1960’s and 70’s led some researchers to predict that computer
simulationswouldeventuallybeabletomoreaccuratelyanalyzefreeflightscenariosthan
windtunnelsandwouldevenreplacethemcompletely[4].Unlikeanalyticalmethods,CFD
has the capacity tomodel complex geometries and does not need approximations to be
made to thegoverningequationsof fluidmechanics toobtain solutions.Thismeans that
fewphysical assumptions need to bemade to find a solution. CFD can also obtain time-
resolved flow solutions for any problem. Depending on the methodologies chosen, it is
feasibletoperformviscoussimulationsonafullaircraftinlessthanadayusingapersonal
computer[5].Furthermore,thereareproblemsforwhichexperimentaldataisdifficultor
impossibletoobtain.Forexample,takingmeasurementsinsideofajetengineisnotoften
practical. For this reason,CFD is an indispensableengineering tool that candramatically
reducetheamountofexperimentaltestingneededinanaerodynamicdesignprocess.
7
DespitethemanyadvantagesofCFD,therearesomedisadvantagesthatinhibitCFDfrom
completelyreplacingwindtunneltesting.Likeanyfieldthatreliesonnumericalmethods,
CFD is subject to truncation errors in floating point calculations. Round-off error is
introduced in CFD because computers hold a finite number of digits in arithmetic
operations. This error is proportional to the number of grid points. CFD also introduces
discretization error. Discretization error is defined as the difference between the partial
differential equation solution and the discretized approximation to the differential
equation solution when round-off error is excluded [2]. Additionally, it might seem as
though solutions to increasingly complicated problems can be found more easily with
advancesincomputationalpower.However,thisisnotnecessarilythecase.Thisisdueto
difficultiesinsoftwareparallelization.ItisalsoimportanttopointoutthatwhileCFDcan
predictaerodynamicperformancewithincreasingaccuracy,engineersareofteninterested
in other features such as reliability, operability, and maintainability. For example, CFD
alonecannotpredict ifanenginecanrestartatcruisingspeed,orhowmany flighthours
canbeachievedbeforeserviceisneeded.
Another difficulty associated with CFD is that of turbulence modeling. One approach to
modelingfluidflowiscalledDirectNumericalSimulation(DNS)simulations.Thismethod
does not make approximations to the governing equations but often requires orders of
magnitudemoremeshpointsthanacomputationallycheaperalternativesuchasReynolds-
AveragedNavier-Stokes (RANS) [4] [6].Becauseof its computational cost,DNS is almost
exclusivelyusedfortheoreticalanalyses.WhileDNScanbeusedtostudymoreproblemsas
computational power increases, its application to most practical engineering problems
remainsinthedistantfuture.Alternatively,RANSisoneofthemostcommonCFDmethods
usedinengineering.ItreliesonstatisticallyaveragingtheNavier-Stokesequationsintime.
WhileRANSmethodscanbeusedforcasesofmildseparation,theyoftenarenotascapable
in cases ofmassively separated flows [5]. Another approach to CFD is called large-eddy
simulation (LES). LES is like RANS in that it makes assumptions that result in
approximationstotheNavier-Stokesequations.However,insteadofaveragingtheNavier-
Stokes equations, it filters the Navier-Stokes equations, removing high wave-number
components[7].ThecomputationalcostandaccuracyofthismethodfallsbetweenRANS
8
andDNS.Figure2showstherelationshipbetweencomputationalcostandthenumberof
assumptions associated with each model. It should be stated that despite these
shortcomingsandthosediscussedpreviously,CFDhasproventobean invaluabletool in
the aerodynamic design process, dramatically reducing the number of design cycle
iterationsneededinthedesignprocess.
MoreAssumptions FewerAssumptions
LeastExpensive MostExpensive
RANS LES DNS
Figure 2. A diagram of the relative accuracy and cost of various CFDmethods.
The third approach to obtaining aerodynamic predictions is experimental. Historically,
windtunneltestinghasbeenusedasthebasisforthedesignprocessofallcontemporary
flyingvehicles,startingat thetimeof theWrightbrothers.Regardlessof the fact that the
numberofmilitaryaircraftprogramshasshownadownwardtrendsince the1950’s, the
numberofwindtunneltesthoursperprogramhasbeenonasteadyrise[8].Thiscanbe
attributedtothefactthatasaircraftbecomemoreadvanced,theamountoftestingneeded
to perfect a given design increases. Historical trends show that the amount of
experimentallyacquireddataforanaircraftprogramhavebeenandwillcontinuetogrow.
Windtunnels,however,arenotwithouttheirlimitations.Forexample,thecosttobuildand
maintainawindtunnelvariesdependingon itssizeand features,but it israrelyacheap
endeavor.Thisisbecausewindtunnelscantakeyearsandcountlessman-hourstodesign
andbuild.Makingaccuratemodelswithproperlycalibratedequipmentisadifficultprocess
andonlyincreasestheoperatingcostsofawindtunnel.Themainchallengeassociatedwith
windtunneltestingisproperlyreplicatingflowparameters.Forlowspeeds,this involves
building amodel of the object of interest and replicating the desired Reynolds number.
AchievinghighReynoldsnumbersintestscanoftenprovetobedifficultwithoutexpensive
infrastructure,andinsomecasesthedesiredReynoldsnumbercannotbeexperimentally
achieved at all. Furthermore, care must be taken to avoid measurement inaccuracies
9
related to the proximity of the tunnelwalls to the test object [9]. The fact remains that
despite growths in computational methodologies, CFD has not, and will not completely
replace experimental methodologies for the foreseeable future in solving aerodynamic
designproblems[1].
1.2. HistoryofExperimentalAerodynamics
Themodernconceptofawindtunnelhasonlybeenaround for150years.Prior towind
tunnelsthemostsystematicapproachtoexperimentalfluiddynamicsinvolvedattachinga
model to the end of a whirling arm. Benjamin Robins (1707 – 1751) is attributed with
employingthefirstwhirlingarmmechanismandheusedittomakelargeadvancementsin
our earlyunderstandingof fluidmechanics [10]. Figure3 is a sketchofRobins’whirling
armmechanism.ItwasnotuntilSirGeorgeCayley(1773-1857),whoalsousedawhirling
arm mechanism, began his work that experimental aerodynamics would contribute the
vital testdataneededtodesignwhat isbelievedtobethe firstrecordedheavier-than-air
vehicle inhistory [10].Cayley’s first successfulunmannedgliderhadawingspanof200
squarefeet.By1852hemadeatriplanethatintegratedmanyconceptsofmodernplanes:
fixedwings,fuselageandtail.Cayleywasthefirsttoconceptuallyseparatethemechanisms
behindliftandthrust.PriortoCayley,mostdesignsforflyingmachinesincorporatedone
mechanism forboth lift andpropulsion:ornithopters that flappedmechanicalwings [10,
11]. With this flying mechanism paradigm shift came the need for a better way to
experimentally measure forces such as lift and drag. Frank H. Wenham (1824-1908) is
frequently credited with designing and building the first wind tunnel [10]. As a council
member of the Aeronautical Society of Great Britain, he received funding to design an
apparatusthatpulledairthrougha3.66m(12ft)tubepastamodelusingasteam-engine-
poweredfanblower.
While the years after the creation of the first wind tunnel (1871) were full of ground
breakingexperimentsperformedbyWenhamandhis colleagues, thequestion remained:
didexperimentalresultsobtainedwithascalemodelbearanyresemblance to thoseofa
full-sizedaircraft[10]?Itwasnotuntil1885thatOsborneReynoldsshowedthattheflow
patternsoverascalemodelwouldbesameforafull-scalemodelifacertainparameteris
10
thesame inbothexperiments[12].Thisparameter,whichwillbediscussed later, isnow
known as the Reynolds number. The significance of this discovery is that wind tunnels
could be used to generate meaningful experimental results for full-scale models by
completingtestingonscalemodelsiftheReynoldsnumberisproperlyreplicated.Another
fifteen yearswould pass beforeWilbur andOrvilleWright, frustratedwith the accepted
aerodynamicdesigntablesof theirday,woulduseresults fromtheirownwindtunnel to
buildtheworld’sfirstpowered,manned,heavier-than-airflyingvehicle[11,10].Asketchof
theWrightbrothers’windtunnelisshowninFigure4.
Figure3.A sketchofBenjaminRobin’swhirling arm.A string iswrappedarounda cylinderandamass is tied to theotherend (bottomright).Thestring ispassedoverapulleyandthecylinderandattachedarmrotateasthemassfalls.Themodelofinterestisattachedtotheendofthearm.ImagetakenfromRef.[13].
Sincetheearlydevelopmentofthemodernwindtunnel,avarietyofdifferentwindtunnels
havebeendevelopedformanyspecialpurposes.TheseincludetheNACAvariabledensity
tunnel,whichwascapableofperformingexperimentsathighReynoldsnumbers[14]and
thewindtunnelat theNASAAmesResearchCenter,witha40by80-foottestsectionfor
full-scale testing [15]. Further examples of specialized wind tunnel applications include
verticaland/orshortlandingandtakeoffvehicle(V/STOL)testingandacoustictesting.For
moreexamplesofmorerecentspecializedtunnelsandapplications,seeBarlow[1].
11
Figure4.A sketchof theWrightbrothers’wind tunnelused todesign thefirst heavier-than-air powered flight vehicle. The fanwas belt driven andpowered by a small gas engine the Wright brothers used for poweringmachineryintheirbikeshop.Theairwaspushedthroughthetunnelbythefan and, as a result, the Wright brothers struggled to find a way tostraightentheswirlingmotionoftheairasitpassedthemodel.ImagetakenfromRef.[10].
1.3. Objective
As mentioned before, analytical and computational approaches to aerodynamic design
problemsarenot always sufficient.Asa result,Houghtonhas startedadesignproject to
buildanewgeneral-purpose low-speedwind tunnel toexpand its in-houseaerodynamic
testingcapabilities.ThiswindtunnelwillgiveHoughtontheabilitytoperformawiderange
of experiments. Specific examples include testsonplasmaactuators for flowcontrol and
testsonaircraftmodelswithicebuildup.Plansforfutureresearchwillbediscussedlater.
Furthermore, a general-purposewind tunnelwill giveHoughton the ability toperforma
varietyoflaboratoryactivitiesaspartofafluiddynamicscourse.
Thisthesisdiscussestheoveralldesignofthewindtunnelandthedesignprocessused.The
flow velocity throughout the tunnel was estimated using a quasi-one-dimensional
12
assumption.Giventhevelocity,semi-empiricalequationswereusedtopredict thetunnel
efficiency based on the air properties (e.g., test section speed and viscosity) and sizing
choices.AnOctavescriptwaswrittentoautomatethecalculations foravarietyofdesign
choices and estimate fan power requirements for each. Once construction of the wind
tunneliscomplete,thiswillprovideanotheropportunityforresearch:testingtheutilized
designprocess.Aretheestimatedsectionlossesandoverallpowerestimatescomparable
tothosemeasuredinthewindtunnel?Theseresultscouldbevaluabletothoseseekingto
designandbuildsimilarwindtunnelsinthefuture.
Theremainderofthisworkisasfollows.Chapter2describesthetheoryoffluidmechanics
andconceptsbehindwindtunneldesign.ThisincludestheNavier-Stokesequations,quasi-
one-dimensional flow, semi-empirical component efficiency equations, andmeasurement
techniques.Chapter3describesthespecificdesignprocessusedalongwithexplanationsof
designchoicesmade.Furthermore,Chapter3coversthedesignandconstructionprogress
made thus far. Lastly, Chapter 4 covers futurework and possible experiments inmore
depth.
13
Chapter2
THEORY
2.1. FluidMechanics
2.1.1. TheoreticalDevelopment
There are three fundamental principles which govern fluid mechanics. These are
conservationofmass,Newton’ssecond lawofmotion,andconservationofenergy.These
threeprinciplesresultinequationswhichrelatevariousquantities(e.g.pressure,density,
velocity)astheyvaryintimeandspace. Onlyabriefoverviewoftheseequationswillbe
presented.Foramorein-depthdevelopmentseeAnderson[11]orCurrie[16].
First,considertheequationwhichfollowsfromtheprincipleofconservationofmass.Itcan
bewrittenas
!"∗
!$∗+! "∗&'
∗
!('∗ = 0 (1)
where"is the fluid density,+is the vector fluid velocity,$is the time,-is the vector of
spatial coordinates, anda superscriptasterisk indicatesdimensionalquantities.Equation
(1)isoftenreferredtoasthecontinuityequation[2].
The second governing equation of fluid mechanics comes from the principle of
conservationofmomentum,whichisanapplicationofNewton’ssecondlawofmotiontoan
elementof the fluid.The two typesof forces that actona fluidelementarebody forces,
suchasgravitationalandelectromagneticforces,andsurfaceforces,suchaspressureand
viscous forces. In many aerodynamic problems body forces are insignificant. Therefore,
theyareneglected.Theresultingequationsaregivenby
!("∗&/∗)
!$∗+! "∗&/
∗&1∗
!(1∗ =
!!('
∗ −3∗4'/ + 5∗!&'
∗
!(/∗ +
!&/∗
!('∗ − 4'/
23!&1
∗
!(1∗ (2)
14
where3isthepressureand5istheviscosity.Equation(2)isavectorequationindicatedby
thefreeindex,8.
The lastequationof fluidmechanicscomes fromconservationofenergy.Conservationof
energystatesthatthechangeintotalenergy(internalpluskinetic)isequaltothesumof
thetotalworkdoneonthesystemandanyenergytransferredbythermalconduction.This
equationisgivenby
!!$∗
"∗9∗ +12"∗&/
∗&/∗ +
!!(1
∗ "∗9∗ +12"∗&/
∗&/∗ &1
∗
=!!('
∗ &/∗ −3∗4'/ + 5∗
!&'∗
!(/∗ +
!&/∗
!('∗ − 4'/
23!&1
∗
!(1∗ + 4'/;∗
!<∗
!(/∗
(3)
where;isthethermalconductivity,9istheinternalenergyand<isthetemperature.Itis
importanttonotethatonlyfiveequationshavebeengiven,buttherearesevenunknowns:
density, pressure, temperature, internal energy, and threevelocity components.To solve
theseequationstwomorealgebraicexpressionsforpressureandinternalenergyshouldbe
used.Theyaregivenby
3∗ = "∗=∗<∗ (4)
and
9∗ = >?∗<∗ (5)
whereEquation(4)istheidealgaslawandEquation(5)isanexpressionfortheinternal
energy in termsof temperatureandspecificheatatconstantvolume,>? . In thiscase,=is
thespecificgasconstant,whichhasavalueof287Jkg-1K-1forair.Thesetwoequationsare
often substituted into the energy equation to give five equations with five unknowns.
Together,thesefiveequationsareknownastheNavier-Stokesequations.
2.1.2. Quasi-One-DimensionalFlow
A thorough treatment of quasi-one-dimensional flow is given by Anderson [11] and a
similarapproachtohisisusedhere.Whenconsideringtheairflowthroughawindtunnel,
15
oneapproachistousetheNavier-Stokesequationstomodeltheflowinthreedimensions.
However, no analytical solution exists and CFD would take time and computational
resources.Amoreeconomicapproachexists. Inwindtunnels, thecross-sectionalarea,@,
variesonlyas a functionof the lengthalong the loopof the tunnel,(,meaning@ = @(().
Forfluidmechanicsproblemsinvolvingducts,suchasestimatingwindtunnelefficiency,it
maybereasonable toassumethat the flow fieldpropertiesareuniformacrossanycross
section,andhenceonlyvaryasafunctionof(.Thisassumptionisvalidincaseswherethe
flowisassumedtobesteady,i.e.,notime-dependence.Furthermore,itisassumedthatthe
boundarylayersalongthesurfaceareverythin,andthatchangesinareawithrespectto(
aresmall.Ifchangesinareawithrespectto(aresmall,thentheAandBcomponentsofthe
velocity are small, allowing the fluid motion to be treated as though it is only in the(
direction.A flow that approximatelymeets these requirements is referred to as a quasi-
one-dimensional flow. Although quasi-one-dimensional flow is only an approximation of
thetime-dependentthree-dimensionalflowthatisreallyoccurringinthewindtunnel,the
resultsaresufficientlyaccurateforuseinestimatingtheefficiencyofthewindtunnel.
Themainresultneededforthepresentworkistheincompressiblequasi-one-dimensional
flow version of the continuity equation which comes from Equation (1). Using the
divergence theorem and assuming a time-independent flow, the integral form of the
continuityequationis
whereCis the the surface that bounds a given volume of fluid. This equation can be
appliedtoaductliketheoneshowninFigure5.Thefluidintheductisboundedbycross-
sectional area@Don the left,@Eon the right, and thewallsof theduct itself. In this case,
Equation(6)canbewrittenas
"& ⋅ GHI
= 0 (6)
"& ⋅ GHJK
+ "& ⋅ GHJL
+ "& ⋅ GHMNOO
= 0. (7)
16
The last term describes the fluid flow that passes through the wall, which is zero.
BecauseGHpoints out of the volume, and x-velocity,u, and density are assumed to only
varywith(,theremainingtermscanbewrittenas
and
Therefore,fromEquation(7),
Equation (10) is the quasi-one-dimensional continuity equation. It is a statement of
conservation of mass and that mass flow through a duct is constant. In the case of an
incompressibleflowwhere"isconstant,
ForairflowatMachnumberslessthan0.3,thedensityvariesbylessthan5%.Therefore,
theincompressibleassumptionismadeinthedesignofthislow-speedwindtunnel.
Figure 5. Quasi-one-dimensional flow through a duct. The dashed linesboundthe fluidvolumeof interest. In thecaseof incompressible flow, theareaandvelocityarerelatedbyEquation(11).
"& ⋅ GH = −"D@D&DJK
(8)
"& ⋅ GH = "E@E&E.JL
(9)
"D@D&D = "E@E&E. (10)
@D&D = @E&E. (11)
@D
@E
&D &E
(
17
2.1.3. Bernoulli’sPrinciple
Bernoulli’sprincipleisastatementofconservationofenergy.Itmathematicallyexpresses
the idea that the sum of all forms of energy along a streamlinemust be the same at all
points.Thatis,
whereQis the acceleration due to gravity,Ais the elevation of the fluid,3is the static
pressure,&isthefluidvelocity,and"isthefluiddensity.Thefirstterm,DE"&E,isreferredto
as thedynamicpressure, andexpresses thekinetic energyof the fluid.The second term,
"QA,describesthegravitationalpotentialenergyofthefluid.Thelastterm,3, isthestatic
pressureofthesystemandcanbethoughtofasameasureofinternalenergyofthesystem.
Thisexpressionofconservationofenergyisonlyvalidunderacertainsetofassumptions.
Thefirstassumptionisthatthefluidisincompressible–thedensityisconstant.Secondly,
theflowissteadyanddoesnotchangewithtime.Thelastassumptionisthatviscousforces
arenegligible.
Whenconsideringwindtunneldesign, thechange ingravitationalpotentialenergyof the
fluidissmallcomparedtothatofthedynamicandstaticpressures.Asaresult,thistermis
oftenneglected.Thestagnationpressure,3R,isdefinedasthesumofthestaticanddynamic
pressures
The stagnation pressure of a system is constant under the same set of assumptions as
Bernoulli’sprinciple.Energy,however,isnotconservedinanon-idealsystem.Forexample,
viscous forces are not negligible throughout the wind tunnel. Therefore, inefficiencies
throughout the tunnel can be expressed as a drop in stagnation pressure. This idea is
fundamentaltothewindtunneldesignprocessandisdiscussedingreaterdetailinSection
2.2.3.
12"&E + "QA + 3 = constant (12)
3R =12"&E + 3. (13)
18
2.2. WindTunnels
In this section, the Navier-Stokes equations are presented in non-dimensional form to
highlighttheimportantparametersforwindtunnelapplications.Section2.2.2coverssome
basicdesignfeaturesofwindtunnelsandgivessomeexamplesofoperationalwindtunnels
with these features. Next, Section 2.2.3 discusses the process and equations by which
powerestimatesaremadeforwindtunneldesign.Lastly,Sections2.2.4and2.2.5discuss
optionsforfanselectionandotherconsiderationsforconstructionofawindtunnel.
2.2.1. ImportantParameters
Wind tunnel testing is most often completed on scale models in an effort to better
understandflowfieldsaroundafull-scaleobject.Howdoesoneknowifflowfieldsaround
ascalemodelaresimilartoflowfieldsaroundafull-scaleobject?Anderson[2]definestwo
flowstobedynamicallysimilarif:
1) Thestreamlinepatternsaregeometricallysimilar.
2) The distributions ofY∗
YZ∗ ,\∗
\Z∗ ,]∗
]Z∗, etc., where subscript r denotes a reference
quantity,arethesamewhenplottedagainstcoordinatesnon-dimensionalizedby
acommonreferencelength.
3) Theresultingnon-dimensionalizedforces(e.g.lift,drag)arethesame.
In general, flow solutions are a function of all of the parameters included in the
dimensional Navier-Stokes equations. However, the Buckingham Pi theorem [17] states
thatthenumberoffreeparametersintheNavier-Stokesequationscanbereducedtothree.
Todothis,theNavier-Stokesequationsmustbenon-dimensionalized.Thisisachievedby
usingthesubstitutions
3 =3∗
3^∗&^∗E (14)
& =&∗
&^∗
19
$ =$∗&^∗
_^∗
" ="∗
"^∗
( =(∗
_^∗
where,asbefore,theasteriskrepresentsadimensionalquantity,thesubscript`denotesa
referencevalue,and_islength.Itiscommontochoosereferencevaluesthathavephysical
significance.Examplesincludethechordofthewingofanaircraftfor_^∗ ,andthefreestream
velocityfor&^∗ .
Using the variable substitutions given in Equations (14), the non-dimensional Navier-
Stokesequationsare
!"!$+! "&'!('
= 0, (15)
! "&/!$
+! "&/&1!(1
=!!('
−4'/3 +5=9
!&'!(/
+!&/!('
− 4'/23!&1!(1
,
(16)
!!$
3a − 1
+12"&/&/ +
!!(1
3a − 1
+12"&/&/ &1
=!!('
&/ −34'/ +5=9
!&'!(/
+!&/!('
− 4'/23!&1!(1
+5
a − 1 b`cE=94'/
!<!(/
(17)
20
whereais the specific heat at constant pressure dividedby the specific heat at constant
volume.Toclosetheseequations,theidealgaslaw(Equation(4))isnon-dimensionalized
usingthesamereferenceparameters,producing
3 ="<acE. (18)
There are three dimensionless coefficients that appear in these equations. The first
dimensionlessparameteriscalledtheMachnumber.Thisisgivenby
c =&^∗
d^∗ (19)
whered^ isthereferencespeedofsound.NextisthePrandtlnumber.Itdescribestheratio
oftheviscousdiffusionratetothethermaldiffusionrateandisgivenby
b` =>\∗5∗
;∗. (20)
In this expression,>\∗ is the specific heat of air at constant pressure and;is the thermal
conductivity of air. The last dimensionless parameter is the Reynolds number which
describestheratioofinertialforcestoviscousforcesinthefluid.Thisparameterisgiven
by
=9 ="^∗&^∗_^∗
5^∗. (21)
The Reynolds number is the similarity parameter of most interest for low speed wind
tunnels(i.e.M<0.3)[11].
Dimensionalanalysisofthegoverningequationsof fluidmechanicsshowsthattwoflows
willbesimilarif:
1) Thebodiesaregeometricallysimilar(i.e.,oneisascalemodeloftheother)
2) Thesimilarityparametersarethesameforbothflows
21
This is importantbecauseconductingexperimentsonscalemodels isoneof theprimary
activitiesofwind tunnels.Thismeans that full-scalebehavior canbepredictedbasedon
experimentalresultsifthenon-dimensionalflowparameterscanbeproperlyreplicatedin
thewindtunnel. However,itisdifficulttomatchboththeMachandReynoldsnumberin
mostcases,andmanytimesneithercanbematchedperfectly.Decisionsbasedonintuition
and knowledge of the system being testedmust bemade regardingwhich parameter to
morecloselyreplicateandwhetherreplicatingthemisnecessary[11].
2.2.2. BasicDesignDecisions
Thissectionwillhighlightsomeoftheimportantdecisionsinvolvedwithdesigningalow-
speedwindtunnelandgiveexamplesoftunnelsthatexhibitcommonfeatures.
It iscommonwhendesigningawindtunneltofirstconsiderthedesiredtestsectionsize.
The required size is determined by its intended application, laboratory size constraints,
andbudgetaryconstraints.Awiderangeofwind tunnelshavebeenbuiltbasedon these
constraints.Onthelargeend,theNationalFull-ScaleAerodynamicsComplexatNASAAmes
(seeFigure6)hastwotestsections.Thesmallertestsectionis12.19m(40ft)by24.38m
(80ft)andcanreachspeedsupto154.2m/s(345mph)andthelargertestsectionis24.38
m(80ft)by36.58m(120ft)andcanreachspeedsof51.41m/s(115mph)[18].Thelarger
testsectionisbigenoughtofitafull-scaleBoeing737init.Incontrast,itisnotuncommon
for educational wind tunnels to have test section diameters as small as 10 cm and be
limitedtospeedsoflessthan15m/s[1].Fordesignpurposes,agoodruleofthumbisthat
model airplanes and airfoils should take up at most 80% the width of the test section.
Furthermore,fortestarticleslikeairplanes,awidthtoheightratioof1.5iscommon.That
said,itisadvantageousformodelstobesmallcomparedtothesizeofthetestsection,as
this lowers the impact of the test section walls on the results [1]. For the same flow
velocity,largertestsectionsallowlargerobjectstobetestedandthereforealargerrangeof
Reynolds numbers can be achieved. The required power and cost of tunnel building
materials tend to vary with the test section hydraulic diameter squared [1], while the
power consumptionof the tunnel tends to increasewith the test section airspeed cubed
[19].
22
Figure6.Anaerialviewof theNationalFull-scaleAerodynamicsComplex.ImagetakenfromRef.[20].
Theseconddesignconsiderationtobemadeisifthetunnelistobeareturn(closedcircuit)
or non-return (open circuit) type. Closed circuit wind tunnels are designed to pull air
throughaclosedloop.Figure7isanexampleofaclosedcircuittunnelastheairpassesina
clockwisedirection.Conversely,opencircuittunnelspulltheairfromtheroomthroughthe
test sectionandexhaust itback into the room.Aschematicof theUCDavisAeronautical
WindTunnelFacility (AWT) is shown inFigure8.This is anexampleof a standardopen
returnwindtunnel. Mostsmallresearchtunnelswithtestsectionslessthan0.61m(2ft)in
diameter are open return [19], in part because of their lower construction costs. Open
circuitwind tunnels are often less energy efficient, but energy operating costs for small
windtunnelstendtoberelativelysmalltobeginwith[1].Therefore,largertunnelstendto
beclosedcircuitwitheitheroneortworeturnpaths.However,closedcircuittunnelsare
moreexpensivetoconstructandcanbemoredifficulttodesignthanopencircuittunnels.
Becauseopencircuittunnelsareopentotheatmosphereatthesettlingchamber,thestatic
pressureinthetestsection,wheretheairspeedisthehighest,islowerthanatmospheric
23
pressure.Thismeans that leaks in the test sectionare inevitablebecauseholesareoften
drilled into the test section to install models or probes. Open circuit tunnels are also
sensitive to obstructions in the room, such as furniture, other equipment or the walls.
Placinganopenreturntunneltooclosetoawallcanhaveadverseeffectsontestsection
flow.Therefore,openreturntunnelsrequirealargeroomastominimizethiseffect[19].
Figure 7. The NACA 6-inch tunnel is an example of an early open throatwind tunnel. Air is blown in a clockwise direction past the test section.ImagetakenfromRef.[21].
Another significantdesigndecision iswhether the test section itself isanopenor closed
type.Anopentestsection isa testsection that isnotenclosedonall sides.Conversely,a
closedtestsection isonewhich isclosedonallsides.Therearealsohybridandpartially
opentestsections.Figure7isanexampleofanearlywindtunnelwithanopentestsection.
Therearesomedisadvantagesforopentestsections.First,opentestsection(openthroat)
tunnelsdonotworkwithopencircuit tunnels thathaveablower in thediffuserbecause
theroomcanbecometooturbulent.Thismeansthateitherthetestsectionmustbeplaced
insideofaseparateair-tightroomorthetunnelmustbeclosedreturn.Openthroattunnels
24
can also suffer from pulsations, the sameway a pipe organ vibrates. If the object being
tested creates awake that is too big, the collector can interferewith the aerodynamics.
However,openthroattunnelsmakeiteasiertochangeandworkontestmodels,andmake
iteasiertoinsertmeasurementprobes.Ingeneral,closedthroatwindtunnelshavemore
advantages. However, several hybrid, partially open, convertible, or slotted wall wind
tunnelshavebeenmadeforuseintheautomotiveindustryandV/STOLdevelopment.
Figure 8. A schematic of the UCDavis AeronauticalWind Tunnel Facility.Thiswindtunnelisanopen-circuitdesign.Airentersthetunnelontheleft,is contracted through a nozzle, passes through the test section, expandsthrough a diffuser, passes the fan that powers the process, and exits thetunnelandreturnstotheroomontherightsideofthefigure.
2.2.3. SectionLossandPowerConsiderations
ToaidinthedesignoftheHoughtonCollegewindtunnel,ascriptwaswritteninOctave(an
open-sourceMATLAB alternative). This script has two primary purposes. The first is to
calculate the dimensions of various components of thewind tunnel given constraints on
selected components. The second purpose of this script is to estimate the required fan
specificationsneededtoachieveatestsectionspeedof44.7m/s(100mph).Thisisdone
by calculating the stagnation pressure drop across each component as a function of its
geometry,and theReynoldsnumberandMachnumber foreachcomponent.Thissection
will reviewthesemi-empiricalequationsused toestimate these losses. It is important to
notethattheseequationsarequasi-one-dimensional.Thecross-sectionalareaisallowedto
change throughout the wind tunnel and the air speed is only a function of the cross-
sectionalarea.
As mentioned in Section 2.1.3, inefficiencies can be thought of as a drop in stagnation
pressure throughout thewind tunnel.Wattendorf [22] considers losses in a return-type
windtunnelbyconsideringthe lossesofeachcomponent insuccession.Eckert,Mortand
25
Jope [23] take the same approach and compare the estimations given byWattendorf to
severalexistingwindtunnels.Asimilarapproachistakenhere.Thetunnelcanbebroken
into constant area sections, corners, diffusers, contracting sections, etc. In each of these
sections,apartfromthefan,thereareflowlossesduetofrictionbetweentheflowinggas
and solid boundaries. The successive drops in total pressure for each component are
balancedoutbythepressureriseacrossthefan.
The power loss, or drop in stagnation pressure, of each section is modeled as being
proportionaltothedynamicpressureattheentranceofthatsection.Forexample,thedrop
instagnationpressurefortheefgcomponentofthewindtunnelisgivenby
h3R' = i'12"&'
E = i'j' (22)
wherej' is the dynamic pressure at the entrance of theefgcomponent.i' is called the
pressure loss coefficient and can be estimated as a function of geometry of theefg
component and viscosity of the air. Then, the total efficiency of the wind tunnel is
calculated as the ratio of the flow’s kinetic energy in the test section to the energy
dissipatedinthetunnelduetofriction.Anexpressionforthisenergyratiois
kl =
12"&f
E
h3R''=
jfi' ' j'
(23)
wherethesubscripttreferstothequantityasmeasuredinthetestsection.Becauseboth
the numerator and denominator are effectively energy terms, both could also be
consideredasenergyperunittime.Thismeansthatthepowerdeliveredbythefanmustbe
at leastjf/kl . It is important tonotethat thisdefinition for therequired fanpowerdoes
not account for the electro-mechanical efficiency of the fan, but only the aerodynamic
efficiencyofthetunnelitself.
Giventhisdefinitionoftheenergyratio,itfollowsthatestimatingthelosscoefficient,i' ,for
eachcomponentofthetunneliskeytoestimatingtheaerodynamicefficiencyofthetunnel.
Wattendorf [22] provides semi-empirical formulas for the loss coefficient of a variety of
26
windtunnelcomponents.Thesearesummarizednext,startingwiththelosscoefficientfor
aconstantareasection.Thisisgivenby
iO = nopg (24)
wherenis the friction coefficient,pgis the hydraulic diameter, andois the length of the
component.Thehydraulicdiameter is a commonlyused termwhendealingwith flow in
rectangularducts.Itisdefinedas
pg =2dqd + q
(25)
whereaandbarethewidthandheightoftheduct.Thefrictioncoefficient,f,isgivenbythe
Prandtluniversallawoffriction[22]
1
n= 2 logDR =9 n − 0.8. (26)
This loss coefficient is used for the test section and any other sectionwhere the cross-
sectionalareaisconstant.
Whenconsideringthelosscoefficientforadiffuser,itiswrittenintermsoftheequivalent
conicalangle.Thisangle,asshownforaconeinFigure9, istheangleofexpansion.More
generally,itcanalsobeusedfordiffusersthatdonothaveacircularcrosssection.Inthis
case,theequivalentconicalangleistheangleofexpansionthatwouldoccuriftheentrance
andexitofthediffuserwerecircularandmaintainedthesamecross-sectionalareaoneach
endastheactualdiffuser.Mathematicallythiscanbeexpressedas
uv = tanwD12 @l − 1
o/pD (27)
where@listheratiooftheareaofthelargersidetotheareaofthesmallerside,pDisthe
hydraulicdiameterattheentrance,andoisthelengthofthediffuser.
27
Figure 9. A depiction of the conical angle for a cone-shaped diffuser. Theequivalent conical angle is the angle of expansion thatwould occur if theentrance and exit of the diffuser were circular and maintained the samecross-sectionalareaoneachendastheactualdiffuser.
Oneofassumptionsmadeaboutdiffuserlossesisthatthetotallosscanbebrokenintothe
sumoffrictionlossesandexpansionlosses,asindicatedby
ix = iy + ivz (28)
where
iy = 1 −1
@lE
n8 sin uv
(29)
and
ivz = iv |@l − 1@l
. (30)
iv(u)isgivenbyEckertetal.[23]fordiffuserswithasquarecrosssection
iv(u) =
0.09623 − 0.004152ufor0 < u < 1.5°
0.1222 − 0.04590u + 0.02203uE + 0.003269uÖ
−0.0006145uÜ − 0.00002800uá + 0.00002337uâfor1.5° ≤ u ≤ 5°
−0.01322 + 0.05866ufor5° < u.
(31)
28
Recommendedpracticeisfordiffuserstomaintainanequivalentconicalangleof3°orless
withatotalarearatioof lessthan3[1]. Ingeneral,theriskofboundarylayerseparation
increaseswithalargerequivalentconicalangleandarearatio[24].Insomecases,diffusers
with a more aggressive expansion are used. So-called wide-angle diffusers are used to
increase the nozzle contraction ratio for awind tunnel, which can improve overall flow
qualityinthetestsection.Barlow[1]statesthatthesediffusersoftenhavearearatiosof2-
4 and22.5°equivalent conical angles.While they canbeused to increase test sectionair
flowquality,theyintroduceadditionallossesandrequirescreenstoavoidboundarylayer
separation.Awide-anglediffuserisnotusedinthepresentdesignsofurtherdetailsarenot
includedhere.SeeMehta[25]formoredetailsonwide-anglediffuserdesign.
Forclosedcircuitwindtunnels,cornersareequippedwithturningvanestoreducelosses
andstraightentheairflowafterthecorner.Figure10diagramsanexampleofacornerthat
usescamberedairfoil turningvanes.Thesevanescanvary ingeometryandchord-to-gap
ratio.Barlow[1]givesanestimationforthelosscoefficientofacorneras
iã = 0.10 +
4.55logDR =9ã E.áå (32)
wheretheReynoldsnumberisbasedontheturningvanechordlength.Equation(32)isa
conservative estimate and losses can often be reducedwith careful design or increased
withpoordesign.Consequently,theuseofCFDforpressurelossestimatescanbeusefulin
choosingtheairfoilshape,thechordlength,andthegap-to-chordratio.Equation(32)only
includeschordlength,limitingitspredictivepower.
Figure 10. An example of cambered airfoil turning vanes in a corner. Theflowdirectionisspecifiedbythearrow.DiagramtakenfromRef.[26].
29
The nozzle primarily serves two purposes. The first is to increase themean air velocity
beforeitentersthetestsection.Thisallowsfortheairspeedintherestofthetunneltobe
lower relative to the test section. This is beneficial because the loss in each component
tends to vary with the local flow speed squared. The second is that the total pressure
remains close to constant across the contraction and, therefore, variations in velocity
becomesmallerfractionsoftheaveragevelocityatagivencrosssection.Thismeansthat
lessscreens(discussedfurtherbelow)areneededtoreduceturbulence,thereforereducing
pressurelossesevenmore[27].Thelosscoefficientforanozzleisgivenby
içf = 0.32nN?éoçpfè
(33)
whereoçisthelengthofthenozzleandpfèisthehydraulicdiameterofthetestsection.In
this case,nN?écan be found using Equation (26) based on the average of the Reynolds
numbersattheentranceandexitofthenozzle.Whilethelosscoefficientwillalsodepend
onthecontractionshape,theeffectoftheexactshapeofthecontractionisrelativelysmall
becausethe losses fornozzlesaretypicallyontheorderof3%thetotalwindtunnel loss
[1]. To properly design a nozzle, it is common to use computer simulations to check for
boundarylayerseparationandoptimizetheflowuniformityatitsexit–thestartofthetest
section. See Mehta [27] for a more detailed approach to contraction geometry and
separationpredictions.
Screens and honeycombs are used to straighten the flow and increase flow uniformity
immediatlybeforethenozzle.Eckertetal. [23]givesrelationsforscreenlosscoefficients
which are based on the porosity. The porosity of awiremesh is dependent on thewire
diameterandweavedensity.IfGêisthewiredensityandëêistheweavedensity,thenthe
porosityisgivenby
íè = 1 −GMëê
E
(34)
30
andthescreensolidityis
ìè = 1 −íè. (35)
Typicalporosityvaluesforwindtunnelscreensarebetween0.5-0.8[1].Theexpressionfor
thelosscoefficientofascreenisgivenby
iê = iîïñóilçìè +ìèE
íèE. (36)
Additionally, Idel’chik [28] gives a loss coefficient curve for screens as a function of the
Reynoldsnumberbasedonthewirediameter,=9M .Barlow[1]providesabest-fitcurveto
thisdata
ilç = 0.785=9M241
+ 1.0wÜ
+ 1.01. (37)
Idel’chik [28]gives themesh factor, iîïñó, tobe1.0 fornewmetalwire,1.3 foraverage
circularmetalwireand2.1forsilkthread.
Eckertetal.[23]givesanexpressionforthelosscoefficientforahoneycombas
ig = ògogpg
+ 31íg
E
+1íg− 1
E
(38)
where
òg =0.375
hpg
R.Ü
=vôwR.Dfor=9ô ≤ 275
0.214hpg
R.Ü
for=9ô > 275
. (39)
In these expressions,pg is the hydraulic diameter of the honeycomb cell,íg is the
honeycomb porosity,=9ôis the Reynolds number based on the honeycomb material
roughness and incoming flow speed,ogis the honeycomb length in the flow direction,
andhisthematerialroughness.MehtaandBradshaw[29]suggestthatabout25,000total
31
cells inahoneycombare sufficient for straightening theair flow. It shouldbenoted that
whiletheliteraturecitedherediscussesthelosscoefficientsforhoneycombs,itisnotclear
howtheroughnesslength,h,isdefined.Asaresult,aconstantguessofig = 0.5wasused
forthepressurelosscoefficientofthehoneycombinthisstudy,assuggestedbyBarlow[1].
2.2.4. FansandDriveSystems
The thrust of a fan and the drag in the various tunnel components varies with the fan
angularvelocitysquared[1].Whilesmallerwindtunnelstendtoonlyhaveangularvelocity
control, it iscommonin largerwindtunnels tocontrol the fanbladepitch, too.However,
havingbothisnotessential.Forcontrollingthefanangularvelocity,mostnewfacilitiesuse
variable-frequency drives for AC motors but some use solid-state controllers for DC
motors, instead. In some cases, themotor isplacedoutside thewind tunnelwith a shaft
leadingtothepropeller,butthisdesignisnotessential.Itisusefultomountmotorsonto
anti-vibration mounts and connect them to the tunnel with flexible coupling to reduce
vibrations [29]. It should also be noted that many wind tunnels have fan straighteners
placed downwind of the fan to reduce swirl. BothMehta [29] andBarlow [1] givemore
detailsonthepowersectionofthewindtunnel.
2.2.5. ConstructionRequirements
Whenbuildingwindtunnels,Barlow[1]suggeststhatcomponentsshouldbedesignedto
withstand the maximum stagnation pressure with a safety factor of around 4. The
structuralstrengthofthewindtunnelisprimarilytoavoidvibrationandtomakesurethat
themotorwillstayinplacewerethefantolosehalfofitsblades.Vibrationcontributesto
noiseandcan,overtime,reducethestructuralintegrityofthewindtunnel.Forthisreason,
thesupportforvariouscomponentsshouldbestiffeneduntiltheirnaturalfrequencyiswell
abovethefrequencyofthefan[1].
Windtunnelsaremadefromavarietyofmaterialsincludingwood,plywood,thinorheavy
metal, concrete,orplastic [1].For small researchand instructional tunnels,plywood isa
goodchoicefortheconstructionmaterial.Itallowsforholesandchipstobeeasilypatched
andforthewholecomponenttobereplacedifneedbe.Plywoodcanbeboltedtogetherand
32
supportedbywoodorsteelsupports.Theinteriorofthetunnelshouldbesmoothedwith
cornersandseamstreatedwithcaulkingorarubbergasket.
2.3. MeasurementTechniques
The primary purpose of aerodynamic testing is to understand the flow field around an
object. The flow field refers to the velocity vector, temperature, pressure, viscosity, and
densityasfunctionsofpositionandtime.Thismeansthataccurateresultsnotonlydepend
on thedesignof the tunnel itself,butalso the techniquesused tomeasure the flow field.
One of the challenges associatedwithmeasuring the flow field around an object is that
insertinganinstrumentintotheflowfieldwillalmostinevitablydisturbit.Designingand
implementing minimally invasive measuring techniques can prove to be a challenge.
Furthermore, obtaining accurate and precise flow field measurements with high time
resolution requires careful design and planning. This section overviews some of the
techniquesusedtoobtainmeasurements.
2.3.1. PressureMeasurement
Measuringpressureisavitalcomponentofwindtunneltesting.Pressureisdefinedasthe
normalforceperunitarea.Inthecontextofaerodynamics,thisforceisexertedonanarea
element due to a time rate of change ofmomentumof the gasmolecules impacting that
surface[11].Pressureisafunctionofbothtimeandposition.
Onetechniqueformeasuringpressureisthemanometer.Themanometerisonetheoldest
devicesformeasuringdifferentialpressure,orthepressuredifferencebetweentwopoints,
andisalsooneoftheeasiesttobuild.Therearemanydifferenttypesofmanometersbuta
simpleonecanbemadebymeasuringthedifferencesinheightofaliquidsittinginaglass
u-shapedtubewithitstwoendsconnectedtothereferenceandtestedpoints.Thisiscalled
au-tubemanometer.SeeFigure11foranexample.Thedifferenceinpressureisrelatedto
theheightdifferenceintheliquidby[1]
h3 = 3D − 3E = hℎ sin í Q "y − "N (40)
33
wherehℎis theheightdifferenceof the liquid in thecolumns,íis theanglebetween the
horizontalandtheplanemadebythetwocolumns,Qistheaccelerationduetogravity,"y
isthedensityoftheliquidinthetubes,and"Nisthedensityofthefluidinthewindtunnel.
Figure 11. Simplified diagram of a u-tube manometer. The pressuredifferencebetweenp1andp2canbefoundfromtheheightdifferenceofthefluidinthetwotubesanddensityofbothfluids.
Another way to measure pressure is using a pressure transducer. The term pressure
transducerreferstodevicesthatprovideanelectricpotentialorcurrent inresponsetoa
pressureorchangeinpressure[1].Themostcommonpressuretransducersaremadefrom
asmalldiaphragmthatdeformswhenapressuredifferential is introducedacross it.The
most common and inexpensive ways of measuring the deformation of the diaphragm
includecircuits thatmeasureachange incapacitanceor inductancedue to thechange in
geometry.Thesetypesofpressuregaugescanbemadeinexpensivelyandrelativelysmall,
but they do require periodic calibration. This means that a manometer or some other
reliablepressuremeasuringtechniqueisstillrequiredforcalibratingpressuretransducers
inawindtunnel.
34
It is important tonote that there are a varietyof othermethods formeasuringpressure
that are not discussed in depth here. For example, some wind tunnels use pressure-
sensitivepaintwhichchangescolorwhenthepressureappliedtoitchanges.SeeGregoryet
al. [30] for a more comprehensive review of pressure-sensitive paints. Similarly,
piezoelectrictransducersproduceanelectricfieldinresponsetoanappliedpressure.See
Barlow[1]foramoreindepthlookatpressuremeasurementmethods.
2.3.2. VelocityMeasurement
Oneofthemostcommonwaystomeasurefluidvelocityisusingapitot-statictube.Pitot-
static tubesarean instrumentwhichyield thedifferencebetween the totalpressureand
the static pressure of a moving fluid. The density of the fluid can be found using the
equation of state based on ameasured temperature and static pressure. This allows the
velocitytobecalculated.Theprincipleofapitot-statictubeliesinBernoulli’sequation(see
Section2.1.3),whichisastatementofconservationofenergy.SolvingEquation(12)forthe
velocity,gravitationaleffectscanbeneglected,giving
& =2 3R − 3
". (41)
This instrument is made of two tubes, one inside the other, with one orifice facing the
airflowandthesecondperpendiculartotheairflow.Anexamplepitot-statictubeisgiven
inFigure12.Asdepicted,theorificeatAmeasuresthetotalpressure,3R,asitbringstheair
to a stop, and the orifice atB senses the static pressure,p, as the airmovespast it. The
pressuresfromthetwoorificesareconnectedtoamanometerorpressuretransducerand
the pressure difference is approximatelyDE"&E. From this the velocity can be calculated
usingEquation(41).
The tipofapitot-static tubecanbeeitherroundedorsquare.Whilesquare tipped tubes
aremore accurate at higher angles relative to the incoming flow [1], hemispherical tips
offer less flow field disturbances further downstream near the second set of holes.
However,tip-effecterrorsbecomenegligentifthesecondsetofholesisplacedmorethan5
or6tubediametersdownstreamfromthebase[31].
35
Figure12.Asimplifieddiagramofapitot-statictube.ThetotalpressureismeasuredatAastheairisbroughttoastopgoingaroundthebendontheleft. The static pressure is measured at B as the air moves past it. Thevelocity can be calculated based on the difference between these twopressures.
Anotherwaytomeasureflowspeedisthermalanemometry.Thismethodtakesadvantage
ofthethermalandelectricalpropertiesofathinwireorfilm.Thefasterairmovespastthis
wire or film, the faster thermal energywill be transferred from it. This property can be
usedtomeasurethespeedoftheair.
To accomplish this, an electric current is passed through the wire or film to raise its
temperatureabovetheadiabaticrecoverytemperatureofthegas[1].Becausethewireor
filmisnotinthermalequilibriumwiththefluid,energyistransferredtothefluid.Therate
of energy transfer to the fluid is therefore a function of the air velocity. An electronic
controlsystemmonitorsthetemperatureofthewire,whichisafunctionofitsresistance,
and adjusts the power in order to maintain a constant temperature in the wire. The
airspeed, therefore, is a function of the power provided by the control system. When
properly calibrated, accurate velocity measurements can be obtained with frequency
responses of up to 50 kHz [32]. This makes thermal anemometry ideal for measuring
velocitiesinturbulentflowfields.
Hot-wiresensorsuseathinmetalwire,withatypicaldiameterof0.5–55mandatypical
lengthof0.1-1mm.Thewireisoftencomposedofplatinumortungsten,or,lesscommonly,
platinum-rhodiumoriridium.Hot-filmsensorsaremadeofathinplatinumfilmdeposited
A
B
36
onto a quartz support with another layer of quartz on top to reduce electromechanical
effects.Hot-filmsaremorecommonlyusedinliquidswhilehot-wiresaremorecommonly
used ingases [32]. Furthermore, forhot-wires, singleprobes can containmultiplewires,
whichcanbeusedforresolvingflowdirectionaswellasvelocity.
Duetothedevelopmentofthedigitalcamera,particleimagevelocimetry(PIV)hasbecome
a viableway tomeasure the vector velocity of a flow field [33].This is accomplishedby
introducingtracingparticlestotheairandmeasuringtheirvelocityastheyfollowtheflow.
Thetracingparticlesneedtobesmallenoughtoproperlyfollowtheair.Theparticlesare
thenilluminatedandtwoimagesoftheparticlesarecaptured.Thevelocityoftheparticles,
and therefore the fluid, canbe calculated from the averagedisplacementof theparticles
within an interrogationwindow and the time between the two frames. This results in a
vectorfieldofthefluidflowattheresolutionofthechoseninterrogationwindowsize.PIV
systems are commonly used for two-dimensional flow field resolutions, but three-
dimensionalflowfieldscanberesolvedwiththeuseoftwocameras[34].LiuandKatz[35]
used a four-exposure PIV system tomeasure acceleration of a flow field. From this they
couldcalculatethepressuredistributionofthetestsection.PIVsoftwareisavailableboth
commercially and though open-source options. OpenPIV is one such open-source option
andthereareversionsforMatlab,PythonandC++[36].
37
Chapter3
DESIGNCONSIDERATIONSANDMETHODOLOGY
3.1. BasicDesignDecisions
Now that the theory and background information on wind tunnel design have been
presented,adiscussiononthedesignmethodologyusedforthisprojectwillbegiven.First,
general wind tunnel layout decisions will be discussed. Then the procedure used to
determine the dimensions of the wind tunnel is given followed by the loss coefficient
estimatescalculatedbasedonthisgeometry.Lastly,areviewofthecurrentprogressmade
onthedesign,construction,andpurchasingofneededmaterialsisgiven.
AsmentionedinSection2.2.2,mostwindtunnelswithatestsectionhydraulicdiameterof
lessthan0.61m(2ft)areoftheopenreturntype.Whiletheexactsizeofthetestsection
waslatercalculatedfromgeometricandpowerefficiencyconstraints,itwasapparentfrom
the beginning of the design process that the test section for theHoughton Collegewind
tunnel would be less than 0.61 m in diameter. However, due to the small size of the
laboratory,anopenreturntunnelwouldbeimpractical.Effectsfromtheendsofthetunnel
being too close to the walls of the roomwould sharply reduce efficiency, lowering test
section speeds. Another consideration is that open-return tunnels are louder than
similarly-sized closed-return tunnels. Sound levels are made even more important here
because of the small laboratory space and its proximity to classrooms and offices. As a
result,aclosedreturndesignwasselected.Figure13 isaschematic forageneralclosed-
returntunnel.
After deciding upon a closed type return, it was decided that including a wide-angle
diffuser (discussed in Section 2.2.3)was not beneficial to the overall design of thewind
tunnel. Wide-angle diffusers are placed upstream of the settling chamber. They are
designedtoslowtheairdownbeforepassingitthroughthehoneycombsandscreens.They
alsoallowforalargernozzlearearatiowhich,inturn,createsamoreuniformflowfieldin
thetestsection.
38
Theaforementionedcodewasusedtocalculatethesizeofvariouscomponentsofthewind
tunnel given area ratio constraints on the wide-angle area ratio, nozzle area ratio, first
diffuserarearatio,diffuserangleandatotallengthof4.72m(15.5ft).Aftertestingseveral
different combinations of these parameters, it became apparent that the desired test
sectionsizeandspeedswouldnotbepossibleifawide-anglediffuserwasincluded.
Figure13. Schematicof theoverallwind tunneldesign thathighlights themostimportantcomponents.
3.2. Constraints,Methodology,andDesign
Inthissection,specificconstraintsimposedonthewindtunneldesignarediscussed.Then
theoveralldesignmethodologyisdescribedasarethedetailsofthechosendesign.
Thefirstconstraintthatwasimposedonthedesignprocesswasatestsectionvelocityof
44.7m/s(100mph).Smallerspeedswouldnotbeidealforaccomplishingthetypesoftests
the physics department at Houghton College hopes to do with this tunnel. Specifically,
lowerspeedswould further limit therangeofReynoldsnumbers thatcouldbeachieved.
The second constraint placed on the design of the wind tunnel is the area ratio of the
nozzle.Itisidealtomaximizethisvaluebecauseitwillcausetheairintherestofthetunnel
tomoveslowerrelativetotheairinthetestsection.Becausethelossesofeachcomponent
varywiththelocalizedairspeedsquared,maximizingthisvaluewillincreasetheefficiency
ofthetunnel.Thismeansthatanozzleratioofascloseto6aspossiblewasdesired(based
onasuggestionfromMehta[29]).However,thespecificdesignofthecontractionrequires
39
computationalanalysistooptimizetheflowuniformityandpressureloss.SeeBrassardet
al[37],Fang[38],Fangetal.[39],Leifsson[40],Mikhail[41],andMehtaandBell[27]for
workonspecificcontractiondesigns.
Ultimately a nozzle area ratio of 5.45 and an area ratio of 3 for the first diffuser were
selected.Thiswasdoneusingcodethatiteratedthoughvariousnozzleratiosfrom5to6by
0.05andvariousdiffuserangles2°to3°by0.05°.Alengthbydiameterratioof2.5forthe
test section, 0.8 for the nozzle, and 0.5 for the setting chamberwere chosen based on a
suggestion given by Barlow [1]. The resulting test section size and required fan
specificationsneededtoachievethedesiredtestsectionspeedwereprintedtoatextfile.
Priortothisrefiningprocess,afanhadbeenpurchasedandonlynozzleanddiffuserarea
ratiosthatallowedthefantofitwereconsidered.Moredetailsonthepurchasedfanwillbe
giveninChapter4.TheequationsgiveninSection2.2.3wereusedtocalculatethepressure
loss coefficients foreach section.Asmentionedbefore, these coefficientshavebeennon-
dimensionalizedbydividingthepressureloss ineachsectionbythedynamicpressureof
thetestsection.
Basedonthechosennozzlearearatio,allremaininggeometricdimensionsarecalculated
by the code and presented in Table 1. Figure 14 illustrates a sketch of the inside
dimensionsoftheproposedwindtunnel.Airwillmovecounterclockwisethroughtheloop.
The script also computes the loss coefficient, pressure loss, and air speed for each
component,giveninTable2.Itshouldbenotedthatthefirstdiffuseristhesectionwiththe
highest losses.Thismakessense,notonlybecause it is the longestcomponent,but ithas
the second highest air speed, behind the test section. Note that corners, although not
estimatedtomakealargecontributiontotheoveralllosses,havemuchlargerlossesifthey
arenotdesignedandbuiltproperly.Becausethenozzlemakesuponlytwopercentofthe
totallosses,theprimarygoalofitssubsequentdetaileddesignshouldbemaximizingflow
uniformity. Maximizing the efficiency of the nozzle is not a concernwhen its losses are
smalltobeginwith.
For the purpose of estimating required fan characteristics, the energy ratio (k^) is
calculatedusingEquation(23).Itsvalueis2.60whichmeansthatthewindtunnelrequires
40
1.73kWtobeaddedbythefantoreachtestsectionspeedsof44.7m/s(100mph).Thisis
equivalentto2.32hp.Giventhesizeofthetestsection,thefanwouldneedtodisplace3.84
m3s-1(8,130ft3min-1)ofairandhaveatotalpressureriseof452Pa(1.82in.ofwater).The
acquired fan (described in the next section) is capable of meeting these operational
requirements.
3.3. CurrentProgress
Manyoftherequiredmaterialsandcomponentsofthewindtunnelhavebeenpurchased.
See Table 3 for a list of all purchased items along with their cost, distributor and
manufacturer.Thisincludes19.1mm(¾in.)mediumdensityoverlay(MDO)plywoodand
13mm (½ in.) acrylic needed to build thewind tunnel. MDO plywood’s primary use is
buildingconcretemolds.Thismeansthatithasexcellencestructuralstabilityandoneside
thatissmoothed.Thesecharacteristicsareidealforbuildingourwindtunnelbecausethe
airpressure insidethetunnelwillbeaboveatmosphericandthesmoothnesswillreduce
friction.Furthermore,onesideof the tunnelwillbeconstructedoutofacrylicso thatair
flow throughout the tunnel can be visualized and important tunnel components can be
seen. Inaddition,asteel framehasbeenconstructedoutofUnistrut tosupport thewind
tunnelonceitiscomplete.(Figure15).Unistrutwasselectedasabuildingmaterialbecause
of its adjustability. The bolts which hold the joints together can be loosened and the
horizontalbarscanbemoved,allowingthetunneltobeplacedastomatchtheheightofthe
fan.
Forthefan,avaneaxialfanmanufacturedbyNewYorkBlower(Figure16)wasselectedto
meettherequirementscoveredinSection3.2.Thisfanis53.3cm(21in.)indiameterwith
ninebladespitchedat33.0°.Ithasa5.6kW(7.5hp)motorandcanspinatamaximumof
3500rpm.Becauseithasacircularcrosssection,twocustom-madetransitions(Figure17)
havebeen fabricatedbyNordfab tobeused to convert fromacircular cross section toa
square cross section. The fan was selected to match cross sectional area with the
components before and after it as closely as possible. The fan is driven by a variable
frequency drive (VFD) that has been purchased and installed in the laboratory. A C10
CustomizableVectorACDriveVFDmadebySaftronicswasselectedtodrivethefan.
41
Table1.Alistofthedimensionsofeachcomponentofthetunnel.Thisliststartsat the testsectionandworksdownstreamfromthere.Note that forthepurposeofestimatinglosses,themotorandtransitionsbetweensquareandcircularcrosssectionweremodeledasaconstantareasectionwithalosscoefficientcalculatedusingEquation(24).
Figure14.Asketchoftheinsidedimensionsofthewindtunnel.Theairwillpasscounterclockwisethroughtheloop.Notethatadetaileddesignforthenozzlehasnotbeencompleted.Thenozzleshownhereusesstraightwallstoconnectthesettlingchamberupstreamandthetestsectiondownstream.
Component UpstreamWidth(cm)
UpstreamHeight(cm)
DownstreamWidth(cm)
DownstreamHeight(cm)
Length(cm)
TestSection 35.9 23.9 35.9 23.9 73.3Diffuser1 35.9 23.9 49.5 49.5 192.4Corner1 49.5 49.5 49.5 49.5 49.5Edge1 49.5 49.5 49.5 49.5 45.6Corner2 49.5 49.5 49.5 49.5 49.5Fan 49.5 49.5 49.5 49.5 173.9Diffuser2 49.5 49.5 68.4 68.4 180.7Corner3 68.4 68.4 68.4 68.4 68.4Edge2 68.4 68.4 68.4 68.4 30.5Corner4 68.4 68.4 68.4 68.4 68.4SettlingChamber 68.4 68.4 68.4 68.4 34.2Nozzle 68.4 68.4 35.9 23.9 54.7
42
Figure15.ThesteelframeconstructedoutofUnistruttosupportthewindtunnelwhencomplete.TheacquiredvaneaxialfanfromNewYorkBlowerisalsovisible.
Table2.A tableof theestimated losscoefficient, stagnationpressure loss,percentage of total losses and velocity for each component of the windtunnel.Theliststartswiththetestsectionandprecedescounter-clockwiseasshowninFigure13.
Component LossCoefficient StagnationpressureLoss(Pa)
LossPercentage Velocity(m/s)
TestSection 3.07e-2 36.2 7.8 44.70Diffuser1 1.23e-1 144 31.3 44.70SafetyScreen 6.51e-2 76.6 16.6 15.69Corner1 2.13e-2 25.0 5.4 15.69Edge1 1.66e-3 1.95 0.4 15.69Corner2 2.13e-2 25.0 5.4 15.69Fan 5.82e-3 6.84 1.5 15.69Diffuser2 9.30e-3 10.9 2.4 15.69Corner3 6.22e-3 7.26 1.6 8.20Edge2 2.14e-4 0.252 <0.1 8.20Corner4 6.22e-3 7.32 1.6 8.20SettlingChamber 2.40e-4 0.283 <0.1 8.20HoneyComb 1.68e-2 19.8 4.3 8.20Screen 7.63e-2 89.8 19.5 8.20
Nozzle 7.80e-3 9.17 2.0 8.20
43
Figure16.Aphotoofthefan(left)andaCADdrawingofthefan(right).
Figure17.Custom-madesheetmetalcomponentfortransitioningbetweensquareandcircularcrosssections(left)andsheetmetal turningvanes forcorners.(right).
44
A variety of other products have been acquired for thewind tunnel also. Turning vanes
(Figure17)wereorderedfromAeroDynetobeinstalledinthecorners.Thesewillreduce
lossesastheairpassesaroundthecorners.AhoneycombsheetwasorderedfromPlascore,
Inc.aswellasscreensfromW.S.Tyler.AsafetyscreenwaspurchasedfromMetalsDepotto
beplacedafterthefirstdiffuser.Thisscreenwillhelptostopanythingthatcomesloosein
thetestsectionfromenteringthefan.
45
Table3.Alistofthematerialspurchasedsofaralongwithapriceandproductdetails.
TUNNELCOMPONENTS Price Shipping PurchasedFrom Manufacturer(ifdifferent) Item
$1168.31 $149.95 Professionalplastics ½”thickclearcastacrylicpaper-masked
sheet@+/-0.125”cuttolerance
$588.87 VFDs.com Saftronics C102007-1,ACVectorDriveVFD
$2,409.00 $172.00 AdvancedAir,Inc. NewYorkBlower Vaneaxial,21"Diameter,Arrangement
4M,33degreebladepitch.Comeswitha
7.5HPmotorwithamaximumspeedof
3600RPM.
$498.53 $199.80 LordandSons,Inc. AtkoreInternational,Inc. Unistrutandvariousconnections
$508.00 $100.00 AeroDyne H-E-PVane,200feet
$62.50 $13.46 Plascore,Inc. PC2-250B2.000-13,27"x27",2"thick,
1/4"cell,honeycomb.
$407.79 $- BelmontHardware 8Single-Sided,3/4"sheetsofMedium
DensityOverlay(MDO),2x4's
$117.92 $15.00 W.S.Tyler 4-27"x27"0.009"diameterwire,316SS
finemesh.Theseare16-countmesh
screenswith73.3%openarea.
$49.84 $35.13 MetalsDepot 20"x20"WeldedWireSteelMesh,1"x1"
cellswith0.118"diameterwire
$630.00 $203.63 ABFlow-tek NordfabDucting Customductsquaretocircular
transition:19.5"x19.5"to21.25"
diameter,length=18.25"
INSTRUMENTATION
$70.50 $16.96 Dwyer,Inc. 2-MARKIIM-700PAU-Tube
Manometers
$270.00 $28.68 UnitedSensor UnitedSensor PCC-18-KLandPCC-8-KLpitot-static
tubes
TotalProducts TotalShipping TOTAL $6,781.26 $934.61 $7,715.87
46
Chapter4
CONCLUSIONSANDFUTUREWORK
4.1. Conclusions
Thepurposeofthisprojectwastodesignageneral-purposelow-speedwindtunnel.Code
waswritteninOctavetocalculatethegeometryofeachcomponentofthetunnelgivenaset
of constraints and geometry choices. The area ratio of thenozzle and first diffuserwere
chosentomaximizetheratioofthetestsectionairspeedtotheairspeedintherestofthe
tunnel.Becausetheenergylossesineachcomponenttendtovarywiththelocalairspeed
squared,itisidealtomaximizethenozzlearearatiotoincreasetheefficiencyofthetunnel.
The equivalent conical angle is constrained to reduce the chance of separation in the
diffusers,whichwouldgreatly increasethepressure loss.Giventhemaximumsizeof the
windtunnel,thecodeiteratesthroughnozzleanddiffuserarearatios,calculatingrequired
sizesforeachcomponentandensuringthereisenoughroomfortheselectedfan.
For each component, the code calculates the loss coefficient using a semi-empirical
equation.The losscoefficient isbasedstrictlyonthegeometryof thecomponentandthe
viscosityoftheair.Fromthis,thetotalefficiencyofthewindtunnelisestimated.Basedon
preliminaryanalysesusingthecode,afanwaspurchasedprovidingsufficientcapabilities
forthepresentapplication.Thewindtunnelandfanhavebeendesignedtoachieveatest
sectionspeedof44.7m/s(100mph)anddetailedsizinginformationhasbeengiven.The
proposedwindtunnelwillbe4.72m(15.5ft)longand1.67m(5.49ft)tall.Thetestsection
willbe0.239m(9.42in.)tall,0.359m(14.1in.)wideand0.733m(28.8in.)long.Todate,
manyofthenecessarycomponentsandmaterialsforthetunnelhavebeenpurchasedand
detaileddesignworkisinprogress.
4.2. FutureWork
In this section, immediate and more distant future work is discussed. In the immediate category,
a detailed design and drawing of the wind tunnel is under way, based on the basic design
outlined in this work. This will ensure that the plywood and acrylic will be cut to the correct
47
dimensions to ensure that each component has the correct interior dimensions of the air. Figure
18 is a preliminary version of this drawing, detailing the configuration of plywood and acrylic.
After this drawing is complete, construction of the tunnel can commence, with the exception of
the nozzle and corners. Currently, work is being done to simulate airflow through the nozzle to
select a contraction design which maximizes flow field uniformity in the test section. Similarly,
work is being done to simulate airflow through the corners of the wind tunnel. The goal of these
simulations is to determine the optimal spacing of the turning vanes to reduce losses and
maximize flow uniformity in the corners. After simulations on the corners and nozzle have
yielded optimal designs, they can be built.
Figure18.Apreliminarydiagramoftheproposedwindtunnel.
In the more distant future, and once the construction of the wind tunnel is complete,
instrumentation will need to be installed. This includes drilling holes to insert pitot-static tubes in
the settling chamber and test section. Also, a balance for supporting and measuring forces on a
48
model (e.g. lift, drag) needs to be installed in the test section. After the instruments have been
installed and calibrated, the tunnel can be used for experiments.A few options for experiments
are listed next.
As mentioned in section 1.3, there are many examples of future research that could be conducted.
One example is studying the effects of ice accretion on airplane wings. As ice builds up on
airplane wings it can adversely affect drag and lift and, consequently, the flight of the plane. Our
proposed wind tunnel is not capable of operating at freezing temperatures, but results from
NASA Glenn could be used to help us study this problem. The Icing Research Tunnel at NASA
Glenn can operate with air temperatures below freezing [42]. Water vapor is then injected into
the tunnel and ice begins to build up on the model they are testing. Researchers at NASA Glenn
scan the ice buildup to make a 3-d digital model. Figure19 is a picture of a model with ice
buildup and its corresponding 3-d model. These models can be replicated with a 3-d printer and
tested in the Houghton College wind tunnel.
Figure 19. Researchers atNASAGlenn can produce 3-d scans (left) of iceaccretiononairfoilsthathavebeenproducedintheIcingResearchTunnel(right).ImagestakefromRef.[42].
A second example of possible future research is plasma actuation. Plasma actuation is a
techniqueforactivelycontrollingairflow,e.g.,aroundanairfoil.Airfoilsatahighangleof
attackareatriskforseparationtooccur.Separationiswhentheboundarylayerbecomes
detachedfromthesurfaceofanobject.Foraircraft,separationonthewingsisreferredto
asstall.Whenstalloccurs,thereisarapidlossinliftandincreaseindrag.Thiscanbeseen
49
in the top image of Figure 20. Plasma actuators use a pair of electrodes with a high
potentialdifferencebetweenthem(3-12kV)atthefrontandmiddleofanairfoiltoionize
the air as it passes over the front edge of the airfoil. This attracts the ionized air to the
surface of the airfoil to reduce the chance of separation [43]. The bottom of Figure 20
demonstratesactivecontrolofairflowaroundanairfoil.
Figure20.Ademonstrationofaplasmaactuatoractivelycontrollingtheairflow around an airfoil. Smoke is being used to visualize the flowfield andclearly shows separation when the actuators are off (top). The flowsmoothlyfollowstheairfoilsurfacewhentheyareon(bottom).ImagetakefromPostandCorke[43].
Lastly,theHoughtonCollegelow-speedwindtunnelwillbeusedforteachingapplications
inthefuture.Specifically,itwillbeusedinthecontextofanundergraduatefluidmechanics
course. This canbeuseful in helping students understand fundamental fluidphenomena
and measurement methodologies. For example, Reynolds [44] suggests a lab for
undergraduates that focuses on plotting the lift of an airfoil as a function of air speed.
Undergraduatewind tunnel labsareuseful forbothphysicsandengineeringstudents,as
wellasnon-technicalstudentsinterestedinaviation.
50
R e f e r e n c e s
[1]J.Barlow,W.RaeandA.Pope,Low-SpeedWindTunnelTesting,3rded.(Wiley-Interscience,NewYork,1999).[2]D.Anderson,R.H.PletcherandJ.C.Tannehill,ComputationalFluidMechanicsandHeatTransfer,2nded.(Taylor&Francis,1997),pp.5-55.[3]P.CusdinandJ.D.Muller,ReportQUB-SAE-03-01d,Queen'sUniversity,2003(unpublished).[4]D.R.Chapman,AIAAJournal17(12),1293(1979).[5]E.M.Kraft,ITEA31(3),329(2010).[6]S.Chen,D.D.Holm,L.G.MargolinandR.Zhang,PhysicaD:NonlinearPhenomena133(1),66(1999).[7]C.Wagner,T.HuttlandP.Sagaut,Large-EddySimulationForAcoustics,1sted.(CambridgeUniversityPress,2007),pp.89-91. [8]M.R.Melanson,PaperNo.830presentedatthe46thAIAAAerospaceSciencesMeetingandExhibit,Reno,NV,2008(unpublished).[9]H.C.Garner,E.W.E.Rogers,W.E.A.Acum,andE.C.Maskell,NATOAdvisoryGroupforAerospaceResearchandDevelopmentReportNo.AD657092,1966(unpublished).[10]D.D.BaalsandW.R.Corliss,WindTunnelsofNASA,1sted. (ScientificandTechnicalInformation Branch, National Aeronautics and Space Administration, Washington, D.C.,1981).[11]J.D.Anderson,FundamentalsofAerodynamics,1sted.(McGraw-Hill,Boston,2001).[12]O.Reynolds,ProceedingsofTheRoyalSocietyofLondon35(224),84(1883).[13]B.D.Steele,TechnologyAndCulture35(2),348(1994).[14]E.N.JacobsandI.H.Abbot,NACAReportNo.416(1932). [15]J.V.Kirk,B.K.HodderandL.P.Hall,NASATND-4233(1867). [16]I.G.Currie,FundamentalMechanicsofFluids,3rded.(CRCPress,BocaRaton,2002). [17]E.Buckingham,PhysicalReview4(4),345(1914).[18]P.T.Zell,NASATM103920(1993).[19]P.BradshawandR.C.Pankhurst,ProgressinAerospaceSciences5,1(1964).[20] “AnaerialviewoftheNationalFull-scaleAerodynamicsComplex”[Online].NASA.https://www.nasa.gov/content/national-full-scale-aerodynamics-complex-nfac [21]“6-inchWindTunnel”[Online].NASA.https://crgis.ndc.nasa.gov/historic/NACA_Tunnel_One[22]F.L.Wattendorf,ProceedingsoftheFifthInternationalCongressforAppliedMechanics,(1938),pp.526-530.[23]W.T.Eckert,K.W.MortandJ.Jope,NASATND-8243(1976). [24]J.M.RobertsonandH.R.Fraser,JournalofBasicEngineering82(1),201(1960).[25]R.D.Mehta,ProgressinAerospaceSciences18,55(1977).[26]T.F.Gelder,R.D.Moore,J.M.SanzandE.R.McFarland,NASATM87146(1986).[27]J.H.BellandR.D.Mehta,NASACR17488(1988).[28]I.E.Idelʹchik,HandbookofHydraulicResistance,1sted.(TheIsraelProgramforScientificTranslation,TelAviv,1966).
51
[29]R.D.MehtaandP.Bradshaw,TheAeronauticalJournalofTheRoyalAeronauticalSociety83(826),443(1979).[30]J.W.Gregory,K.Asai,M.Kameda,T.LiuandJ.P.Sullivan,ProceedingsofTheInstitutionOfMechanicalEngineers,PartG:JournalOfAerospaceEngineering222(2),249(2008).[31]K.G.MerriamandE.R.Spaulding,NACATN546(1935). [32]G.Comte-Bellot,AnnualReviewOfFluidMechanics8(1),209(1976).[33]A.Melling,Meas.Sci.Technol.8(12),1406(1997). [34]M.S.Iliescu,G.D.Ciocan,andF.Avellan,ProceedingsoftheASME2002JointU.S.-EuropeanFluidsEngineeringDivisionConference,(2002),pp.311-316. [35]X.LiuandJ.Katz,ExperimentsinFluids41(2),227(2006).[36]Z.J.Taylor,R.Gurka,G.A.Kopp,A.Liberzon,IEEETransactionsonIntrumentationandMeasurement,59(12),3262(2010). See also http://www.openpiv.net [37]D.BrassardandM.Ferchichi,JournalofFluidsEngineering127(1),183-185(2005).[38]F.M.Fang,JournalofFluidsEngineering119(2),454(1997). [39]F.M.Fang,J.C.ChenandY.T.Hong,JournalofWindEngineeringandIndustrialAerodynamics89(3),247(2001).[40]L.LeifssonandS.Koziel,JournalofComputationalScience7,1(2015).[41]M.N.Mikhail,AIAAJournal17(5),471(1979).[42]S.Lee,A.Broeren,H.Addy,R.SillsandE.Pifer,NASATM217702(2012). [43]M.L.PostandT.C.Corke,AIAAJournal42(11),2177(2004).[44]R.S.Reynolds,JournalofAviation/AerospaceEducation&Research14(2),9(2005).