See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/230271809

Fundamentals of Wind‐Tunnel Design

Chapter · December 2010

DOI: 10.1002/9780470686652.eae532

CITATIONS READS
75 87,690

3 authors:

Louis Cattafesta Christopher J. Bahr

Illinois Tech NASA
333 PUBLICATIONS   8,664 CITATIONS    60 PUBLICATIONS   1,014 CITATIONS   

SEE PROFILE SEE PROFILE

Jose Mathew
LM Wind Power Group
9 PUBLICATIONS   514 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Three-dimensional control of cavity flow oscillations View project

Three Dimensional, Turbulent Wall Jet View project

All content following this page was uploaded by Louis Cattafesta on 16 November 2017.

The user has requested enhancement of the downloaded file.

Fundamentals of Wind-Tunnel Design
Louis Cattafesta, Chris Bahr, and Jose Mathew
Department of Mechanical and Aerospace Engineering, University of Florida, Gainesville, FL, USA

From a fundamental viewpoint, one can look at the dimen-

1 Introduction 1 sionless form of the governing equations of fluid motion,
2 Facility Characteristics 2 which contain various dimensionless parameters: Strouhal
number St, Reynolds number Re, Euler number Eu, and
3 Wind-Tunnel Design 4
Froude number Fr. When the energy equation is considered,
4 Facility Characterization 8 additional dimensionless parameters appear, such as the Eck-
5 Conclusions 9 ert number Ec, Mach number M, and Prandtl number Pr. Ad-
Related Chapters 9 ditional dimensionless parameters also appear due to bound-
Nomenclature 9 ary conditions, etc.
Subscripts 9 Wind tunnels can also be classified based on their opera-
tional flow regime and corresponding relevant dimensionless
References 10
parameters. It is common to characterize an incompressible
gaseous flow, such as air, primarily by its Reynolds num-
ber. Compressible gaseous flow is primarily characterized by
1 INTRODUCTION its Mach and Reynolds numbers. Liquid flows are character-
ized primarily by the Reynolds and Froude numbers. These
Wind-tunnel testing of full/model-scale components is a parameters are critical, as one of the key, though often unre-
widely employed technique that guides detailed design de- alizable, goals of wind-tunnel testing is dynamic similarity,
cisions in thermal-fluid systems and allows for fundamental where all relevant dimensionless parameters match between
research of fluid phenomena. Many instances occur where model and full scale.
theoretical and/or computational techniques are inadequate, From a practical viewpoint, wind tunnels are often clas-
either due to the complexity of the problem or the lack of sified according to size. For example, low-speed (M < 0.3)
suitable computational resources. Wind-tunnel testing often wind tunnels vary from small tunnels (i.e., test sections with
serves as the most cost-effective approach for these reasons, dimensions less than approximately 1 m × 1 m) to large tun-
as well as the expenses involved in many forms of full-scale nels capable of testing full-scale automobiles and trucks and
testing (which can be related to wind-tunnel tests through large models of aircraft components. High-speed transonic
appropriate parameter matching). and supersonic wind tunnels are also fairly common, but due
Wind tunnels are extensively used in research institutes, to power requirements, these tunnels are typically relatively
universities, industry, and governmental agencies for a wide small in size compared to their low-speed counterparts.
variety of applications and can be classified in many ways. Wind tunnels are also classified according to application.
Some examples include tunnels with moving ground planes
for automotive testing, icing tunnels for studying the effects
Encyclopedia of Aerospace Engineering. of ice formation on aircraft wings, climate tunnels for sim-
Edited by Richard Blockley and Wei Shyy


c 2010 John Wiley & Sons, Ltd. ISBN: 978-0-470-68665-2 ulating various environmental conditions, smoke tunnels for
2 Experimental Techniques for Fluid Dynamics and Thermal Science

flow visualization, propulsion tunnels for the evaluation of centrifugal fans or blowers either push or pull air through the
aircraft engines, spin tunnels to study the spin recovery of test section. Fans/blowers can be either shaft- or belt-driven,
an aircraft, and stability tunnels to study flight dynamics depending on acceptable costs and desired performance
(Barlow, Rae and Pope, 1999). characteristics.
As mentioned above, the key dimensionless parameter Compressor-driven facilities can provide large pressure ra-
of interest in low-speed wind-tunnel tests is usually the tios for relatively little cost and are often preferred for high-
Reynolds number, which is the ratio of inertial to viscous speed facilities that require high stagnation pressures. The
forces. For models exhibiting dynamic similarity, the forces trade-off is the fixed amount of air available for a test. Since
and moments on full-scale models can be obtained by scal- typical compressors cannot supply the continuous mass flow
ing the force and moment data. However, achieving Reynolds necessary, these tunnels often limit the duration of an ex-
number similarity is a difficult task, even in incompressible periment to a few minutes or less, depending on the initial
flows. Since tunnel power scales with ρV3 ATS , restrictions pressure, storage tank volume, and mass flow rate. Fan-based
on power consumption and available space often limit ATS . systems can operate continuously, but the cost scales dra-
Blockage constraints, in turn, limit model size. Assuming matically with volume flow rate and power requirements.
an air medium near standard conditions, the only way to Fans tend to work best with low-speed facilities. An exam-
match Re is to increase the tunnel velocity. In some cases, ple compressor-driven facility is NASA Langley Research
either a limit on maximum tunnel speed or the introduction Center’s 20-inch Supersonic Wind Tunnel (NASA Langley
of compressibility effects precludes Re matching. Thus, one Research Center, 2006). NASA Langley’s 14
× 22
Subsonic
is forced to accept the largest Re that can be achieved in the Tunnel is a larger, fan-driven facility (NASA Aeronautics Test
test section. Depending on the experiment of interest, this Program, 2007b).
limit may or may not harm the validity of the results. This
scenario is commonplace in small-scale aeroacoustic wind
tunnels (Mueller et al., 1992; Mathew, 2005). 2.2 Operating fluid
A quality aerodynamic wind-tunnel facility should pro-
vide a reasonable Re range, flow uniformity, and low tur- When dynamic similitude is difficult to achieve, several op-
bulence intensities. In addition, aeroacoustic (Duell et al., tions are available. First, the working fluid of the tunnel can
2002) and laminar-to-turbulent transition tunnels (Choi and be changed from that of the full-scale application. However,
Simpson, 1987) should also provide low background noise when the working fluid is not the ambient fluid or matched
and vibration levels. to ambient conditions, for example, pressurized air, or water,
a closed-circuit facility must be used and properly sealed to
avoid leakage.
2 FACILITY CHARACTERISTICS Another option is to use a pressurized gas as the working
fluid. For a given temperature, this leads to a density increase
The following sections discuss several primary wind-tunnel that facilitates Reynolds number matching. Allowances must
characteristics that define a facility. Essential references for be made for the increased dynamic loading, proportional to
2
these characteristics and the ensuing design process can be dynamic pressure 0.5ρ∞ V∞ , when density is dramatically
found in the following: Barlow, Rae and Pope (1999), Pope increased. NASA Ames’ 11 × 11
Transonic Wind Tunnel is

and Goin (1965), and Mehta and Bradshaw (1979). an example of a pressurized air tunnel (NASA Aeronautics
Test Program, 2007a).
Alternatively, flow properties can be altered by cooling the
2.1 Drive system gas medium. This can be done using a cryogenic system to
increase fluid density and reduce the viscosity. Such facilities
A defining characteristic is the tunnel drive system, which require significant thermal insulation, as well as large-scale
determines how the working fluid is moved through the test refrigeration systems. NASA Langley’s National Transonic
section. Different drive systems have distinct optimum oper- Facility is an example of such a wind tunnel (NASA Aero-
ational modes, whose selection is dependent on the medium nautics Test Program, 2007c).
and the operational regime. For experiments involving two-phase flow, cavitation phe-
For an air tunnel, two primary drive systems are a compres- nomena, or general maritime interest, a water tunnel can be
sor and fan. In the former, pressurized air is supplied from used. Water tunnels must again be sealed against leakage
a compressor (usually from storage tanks) through a con- and generally use pumps for operation. Maintenance costs
trolled valve or regulator to the tunnel. In the latter, axial or may be significantly higher, depending on the ability to iso-
 Fundamentals of Wind-Tunnel Design 3

late contaminants from the flow reservoir. As with the pre- and exhaust regions must be free of any obstructions. More
vious alternative-fluid-facility-types mentioned, water will seeding material is required for flow visualization as there is
dramatically increase model loading for a given flow speed. no simple mechanism to recycle the seeding particles. De-
The Garfield Thomas Water Tunnel Facility at Penn State pending on tunnel design and location, operation is depen-
University is an example of a large-scale water tunnel (The dent on local weather conditions. An alternative to controlling
Pennsylvania State University, 2007). Free-surface wind tun- the flow conditions (temperature and relative humidity) is to
nels are used to study air–water interfaces, and consist of both house the wind tunnel in a large closed room with an HVAC
an air tunnel and a water tunnel, with a joined test section be- system that has a larger capacity than the load imposed by
tween the two (Rasmi, 2002). Tow-tanks are used for a vari- the wind tunnel.
ety of maritime applications and consist of large water tanks Closed-circuit facilities, while having additional construc-
through which models can, as the name implies, be towed tion expense, have lower operating costs than open-circuit
(Naval Warfare Center Carderock Division, 2009). facilities under similar operating conditions. While purging
seed material requires more effort, less material is required
for a given experiment. On the other hand, the seed quickly
2.3 Duct circuit creates a homogeneous “fog” in the tunnel, allowing for uni-
form but global (as opposed to localized) seeding. The tunnel
Wind tunnels can be designed with two primary types of duct in most cases operates independently of local weather, and
circuits: open or closed (Figure 1). An open-circuit facility flow conditioning is simpler as the input flow conditions are
ingests air from ambient at one end of the tunnel circuit, regulated to a greater extent. However, due to the recycling
passes it through the tunnel, and exhausts to ambient. Closed- of air through the drive system, which is often located “in
circuit tunnels retain a fixed mass of air, not accounting for line” with the flow passage to provide cooling, thermo-
leakage, and circulate flow in a loop through the various wind- dynamic properties can change with air heating unless a flow
tunnel components. cooling control system is employed.
Open-circuit tunnels have generally lower fabrication
costs and occupy less space. Since they constantly cycle in
fresh fluid, localized seeding for flow visualization in the 2.4 Operational flow regime
test section is straightforward. Purging injected seed parti-
cles is simple. One disadvantage is that, for a given speed, an A final major distinguishing characteristic of wind-tunnel fa-
open-circuit tunnel usually requires more energy for opera- cilities is the regime of operation. This is determined by the
tion (Barlow, Rae and Pope, 1999). Also, significantly more desired experimental niche of the facility and plays a strong
flow conditioning may be required to reach flow quality com- role in the design.
parable to an equivalent closed-circuit tunnel. The tunnel inlet Supersonic facilities (M > 1) have high energy and devel-
opment costs. The facility must be designed to sustain large
internal loads from high-pressure differences across the walls
and (for hypersonic flow) potentially high temperature, as
well as forces generated by both stationary and moving shock
waves. The test section must be designed with an appropriate
converging–diverging nozzle for the supersonic Mach num-
ber of interest to minimize shock wave generation and sized
such that reflected shocks from a test model do not interfere
with any experiments (see Complex Internal Compressible
Flows). Instrumentation must be selected to withstand a large
variety of conditions. Often tunnel size is reduced for larger
Mach numbers to reduce energy costs.
Transonic facilities (M ∼ 1) simulate cruise conditions for
many civil and military aircraft. The flow around the model
under test may reach local supersonic conditions, but the over-
all flow field in the test section may still exhibit subsonic,
elliptic behavior. Slotted walls are usually used to accommo-
Figure 1. Schematic of (a) open-; (b) closed-circuit wind date shock waves with steep angles. Model scales are often
tunnels. significantly increased compared to supersonic facilities due
4 Experimental Techniques for Fluid Dynamics and Thermal Science

to the desire to achieve Reynolds number similarity. Thus, sociated noise, as well as to minimize losses. Blockage effects
energy costs are still significant. Ejector systems are used to due to the test section model and the corresponding stream-
get the test section into the transonic range, as conventional line curvature should be minimal to emulate freestream con-
inlet schemes for both subsonic and supersonic wind tunnels ditions. Vibrations generated by the drive system should be
do not perform well (Bradshaw and Mehta, 2003). damped to minimize its influence on the experiment. In the
Subsonic facilities are used to simulate flight by slower case of aeroacoustic facilities, acoustic treatment is essen-
aircraft or examine take-off or landing configurations for tial to minimize disturbance noise from contaminating the
larger aircraft. Ground vehicle facilities also fall in this cat- acoustic measurements.
egory. Some facilities operate at or close to full scale, while
smaller tunnels are necessarily restricted to small-scale test-
3.1.2 Design process
ing. As mentioned above, for sufficiently low operational
speeds (M = 1), the flow is essentially incompressible and The design of a wind tunnel is usually constrained by bud-
M similarity is not required. get, operational and maintenance costs (see Project Manage-
ment: Cost Analysis), and the overall size of the facility. A
flow chart for the design of a low-speed, low-noise facility
is shown in Figure 2. In low-speed facilities, Reynolds num-
3 WIND-TUNNEL DESIGN
ber considerations typically drive the test section size to its
maximum value, subject to cost and size constraints. Test
Wind-tunnel design is a process where research goals are first
section size and desired flow quality determine the size of
set, which then establishes the design criteria. Because the
the flow-conditioning section. Trade-offs are often required
design typically involves fabrication, cost, space, and other
between the desired inlet contraction ratio and contraction
conflicting constraints, the process is iterative.
length in order to approach the required test section flow con-
ditions while remaining within allowable dimensions. The
inlet size determines the size of the settling duct that houses
3.1 Design criteria flow conditioners such as the honeycomb and screens. The
honeycomb affects the flow uniformity, and the screens and
Wind-tunnel design should be tailored to meet the specific subsequent settling duct length dictate the turbulence inten-
research goals and is subject to budget and facility space lim- sity in the test section. The diffuser should be designed to
itations. Guidelines for the detailed design of wind-tunnel maximize pressure recovery from the losses incurred by the
components are provided in this section, including flow con- test section, contraction, and flow conditioners. Although a
ditioners, contraction, test section, diffuser, tunnel driver, longer diffuser provides greater pressure recovery, size lim-
and optional components. Interested readers should refer to itations often constrain the diffuser length and height. Some
dedicated texts on this topic (Barlow, Rae and Pope, 1999; tunnels may require multiple diffusers connected by turning
Bradshaw and Mehta, 2003). sections to fit within facility constraints, as assumed in Figure
2. Care must be taken to ensure no flow separation occurs in
the ducts, which may also require acoustic treatment to min-
3.1.1 Research goals imize flow noise. The drive system must provide the desired
A good wind tunnel should enable measurements of accurate flow rate and pressure drop requirements while maintaining
steady or unsteady data, depending on the desired applica- reasonable energy efficiency. In a closed-circuit tunnel, the
tion. The turbulence intensity in the test section should be flow ultimately is routed back to the flow conditioners in front
low enough to facilitate study of the physical phenomena of of the contraction, while in an open-circuit the flow exhausts
interest. For example, boundary-layer transition experiments to the ambient.
(Saric, Reed and White, 2003) often require extremely low
turbulence levels. Aeroacoustic experiments of trailing edge
noise require low turbulence levels to minimize the effects 3.2 Component design guidelines
caused by the impingement of freestream turbulence on the
leading edge of a model (Mueller et al., 1992). The test sec- This section describes the design of various tunnel com-
tion flow should be uniform and devoid of any unintended ponents, including the flow conditioners, contraction, test
unsteadiness, while secondary flow effects should be mini- section, diffuser, and wind-tunnel drive. Optional compo-
mized in the contraction. Flow separation should be avoided nents such as a turning section, vibration isolator, and jet
inside the tunnel circuit to prevent flow unsteadiness and as- collector for open-jet test sections are also discussed.
 Fundamentals of Wind-Tunnel Design 5

Honeycomb removes swirl from the incoming flow and

minimizes the lateral variations in both mean and fluctuat-
ing velocity (Mehta and Bradshaw, 1979). The yaw angle
for the incoming flow should be less than 10◦ to avoid
stalling of the honeycomb cells. Honeycomb comes in differ-
ent shapes such as circular, square, and hexagonal cross sec-
tions. Among these, hexagonal is usually the cross-sectional
shape of choice, as it has the lowest pressure drop coeffi-
cient (Barlow, Rae and Pope, 1999). Honeycomb cells have
been shown to have the best performance with a length-to-
diameter ratio of between 7 and 10 (Mehta and Bradshaw,
1979). Mehta and Bradshaw (1979) also state that the cell
size should be smaller than the smallest lateral wavelength of
the velocity variation. The honeycomb section should have
sufficient structural rigidity to withstand applied forces dur-
ing operation without significant deformation. Special con-
sideration may be required if the incoming Mach number to
the honeycomb section is high enough that flow choking is
possible.
Figure 2. Flow chart for low-speed, low-noise, wind-tunnel design. Tensioned screens are placed in the settling duct for the
reduction of turbulence levels of the incoming flow. Screens
break up the large-scale turbulent eddies into a number
3.2.1 Flow conditioners
of small-scale eddies that subsequently decay. Schubauer,
In most tunnels, the flow-conditioning section contains a hon- Spangenberg and Klebanoff (1950) state that the Reynolds
eycomb, screens, and a settling duct. An example honeycomb number based on the screen wire diameter should be less
section is shown in Figure 3. The honeycomb aligns the flow than 60 to prevent additional turbulence generation due to
with the axis of the tunnel and breaks up larger-scale flow vortex shedding.
unsteadiness. The screens cascade large-scale turbulent fluc- The spacing between the screens should be of the order of
tuations into smaller scales. These decay in the settling duct, the length scale of the large energy-containing eddies (Mehta
which must be sufficiently long to allow for sufficient de- and Bradshaw, 1979). Placing multiple screens in the settling
cay while minimizing boundary-layer growth (see Turbulent duct with varying porosity, with the coarsest screen being
Boundary Layers). closest to the incoming flow and the finest screen being closest
to the test section, has been shown to provide lower values of
test section turbulence (Watmuff, 1998). A settling chamber
is necessary after the screens so the smaller-scale fluctuations
generated by the wires can decay before accelerating through
the contraction.

3.2.2 Contraction
The inlet contraction plays a critical role in determining
the flow quality in the test section. The contraction accel-
erates and aligns the flow into the test section. The size and
shape of the contraction dictate the final turbulence inten-
sity levels in the test section (Derbunivich et al., 1987). The
contraction stretches vortex filaments, which reduces axial
but intensifies lateral turbulent fluctuations (Tennekes and
Lumley, 1972). The length of the contraction should be suf-
ficiently small to minimize boundary-layer growth and cost
but long enough to prevent large adverse pressure gradients
Figure 3. Schematic of a hexagonal honeycomb section. along the wall, generated by streamline curvature, which can
6 Experimental Techniques for Fluid Dynamics and Thermal Science

should determine which test section type is used. For 2-D

studies, the chord should not exceed 13 ∼ 23 span (Barlow,
Rae and Pope, 1999). It is also essential to have a blockage
ratio of <10% based on the frontal area of the model (Barlow,
Rae and Pope, 1999). A common rule of thumb in test section
sizing is to have rectangular dimensions with a ratio of about
1.4–1 (Bradshaw and Mehta, 2003).

3.2.4 Diffuser
The diffuser decelerates the high-speed flow from the test
section, thereby achieving static pressure recovery and re-
Figure 4. Schematic of the contraction shape with matched poly-
ducing the load of the drive system. The flow field within
nomials.
the diffuser is influenced by the nature of the flow leaving
the test section. The orientation, size (blockage), and wake
lead to flow separation. While CFD may be used in modern development of the airfoil models are some of the factors
design schemes, Morel (1975) suggested a simple analytical that affect the diffuser entrance flow. The area of the diffuser
method of matched polynomials. A schematic of the con- should increase gradually along its axis, so as to prevent flow
traction shape polynomial is shown in Figure 4. The entrance separation. As with contraction sections, diffuser geometry
height of the contraction is Hi , and the exit height is He . The can be optimized.
total length of the contraction is L, and the two polynomials Barlow, Rae and Pope (1999) state that for a conical dif-
are matched (location, slope, and curvature) at a specified lo- fuser, the divergence half angle of the diffuser walls should
cation x = xm . At the exit of the contraction, any remaining be less than 3.5◦ for a “good” design. Mehta (1977) states that
“free” higher-order derivatives of the polynomial are set to the diffuser-included angle for a conical diffuser should be
zero to obtain a straight section. between 5◦ (for best flow steadiness) and 10◦ (for best pres-
In particular, Su (1992) recommended matching a cubic sure recovery). Klines’ flat diffuser curves (Runstadler, Dolan
polynomial at the contraction entrance with a higher-order and Dean, 1975) are generally used for the (non-CFD-based)
polynomial at the contraction exit. Contractions with mini- design of diffusers, as shown in Figure 5. The area ratio AR
mal flow nonuniformities can be designed by iteratively se- between the exit and entrance of the diffuser is plotted versus
lecting the entrance height, contraction ratio, match point, the ratio of diffuser length to the entrance height of the dif-
and length of the contraction. Techniques using 3-D poten- fuser. Three regions are shown on the plot. The design of the
tial flow simulations inside the contraction followed by the
application Stratford’s (1958) separation criteria for turbulent
4
boundary layers affected by adverse pressure gradients can Line
be applied to develop optimized contractions meeting design appreciable
stall
criteria for test section flow uniformity (Mathew, 2005). 3.5

3.2.3 Test section Stall Unsteady

3 flow
The test section can be of the closed-wall or the open-jet
AR

type. The test section design should allow for ease of ac-
2.5
cessibility and installation of wind-tunnel test models and
instrumentation. Aerodynamic performance of the models
can be better matched to full-scale performance in a closed 2 No appreciable
stall
test section; however, an open jet permits far-field acous-
tic measurements at the cost of potential test section jet
deflection, jet/collector interactions, and shear layer refrac- 1.0
0 5 10 15 20 25 30 35 40
tion (Amiet, 1978). In a closed-wall test section, acoustic
L /H i
measurements suffer from poor signal-to-noise ratios due to
[usually] turbulent boundary-layer pressure fluctuations and Figure 5. 2-D diffuser design curves. Adapted from Runstadler,
acoustic reverberation. The facility’s primary design purpose Dolan and Dean (1975)
c Creare, Inc.
 Fundamentals of Wind-Tunnel Design 7

diffuser is conducted by selecting a length for the diffuser, One of the main components of background noise comes
that is, within facility size constraints. Given L/Hi (where the from the fan or drive system. When designing a low-
height is dictated by the test section size), the corresponding disturbance facility, fan noise must be attenuated. The blade
value of AR is selected from the no-stall regions. Although passage frequency (BPF = R · Nblades ) and its harmonics
a greater pressure recovery can be achieved by operating in appear as discrete tones in the test section background noise
the “unsteady flow” regime, this can contribute to unwanted spectra at a fixed tunnel speed and may contaminate sensor
noise, as well as poor performance at off-design flow condi- measurements and affect flow physics. A spectrogram (con-
tions. If facility constraints limit the length of the diffuser or tour plots of sensor signal power vs. frequency) obtained via
a closed-circuit design is used, a turning section with guide short-time discrete Fourier transforms as the tunnel speed is
vanes can be used, and the diffuser can be broken into mul- increased is an effective tool to reveal blade passage con-
tiple sections. tamination (Duell et al., 2004). Such contamination can be
reduced via acoustic treatment of the tunnel circuit and other
acoustic paths (in open-circuit case) between the fan and test
3.2.5 Drive system selection section.
The drive system generates a volume flow rate and com-
pensates for the remaining pressure losses. The driver can
be a fan, blower, or a regulated compressed gas source. 3.2.6 Optional components
The relative advantages and disadvantages are discussed in
Section 2.1. For an open-jet test section, a collector is required at the
Fans are rated by the volume flow rate and the static pres- downstream end of the test section to capture the jet. The
sure drop they can overcome. Barlow, Rae and Pope (1999) resulting flow pattern is highly dependent on the room
provide a procedure for estimation of the losses incurred in characteristics and the test configuration/conditions. The
the tunnel circuit, aiding in fan selection. area of the collector is a design parameter that is best de-
Fan performance is characterized by fan load curves that termined via CFD. The negative aspects of the collector
are plots of fan efficiency and pressure loss as a function of include noise due to boundary-layer “scrubbing” and tur-
flow rate (Figure 6). Load curves are estimated for various fan bulent shear layer impingement on the collector. The col-
rotational speeds. The pressure loss calculation (Barlow, Rae lector is acoustically treated in an acoustic wind tunnel. Its
and Pope, 1999) leads to the wind-tunnel performance curve, effect on flow and acoustic measurements requires careful
which is an estimate of the static pressure loss for various assessment.
values of volume flow rates. The points where the pressure In closed-circuit tunnels where the ducts are joined to-
loss curves intersect the tunnel performance curve determine gether by 90◦ bends, flow separation will occur, leading to
the operating points of the wind tunnel. Fans provide optimal significant pressure drop, flow unsteadiness, and noise gen-
performance when the tunnel operating points fall near the eration. Turning vanes are installed in corner sections to mit-
maximum efficiency of the fan, as shown in Figure 6. igate these adverse effects. Collar (1937) and Salter (1946)
developed some early designs for turning vanes. Gelder et al.
(1986) developed turning vanes with a low loss factor. The
chord, thickness-to-chord ratio, shape, and the number of
turning vanes determine the efficiency, as defined by flow
losses, of the turning section. Modern CFD can now be used
to design advanced turning vanes that function well over the
entire speed range of the tunnel. In addition, the turning vanes
offer the possibility for cooling (or heating) the flow via heat
transfer between the tunnel flow and fluid passing through
the vane interior.
Mechanical vibrations generated by the drive system can
propagate into the test section through ductwork and the
ground. These vibrations must be minimized if aero-optic,
transition, or acoustic measurements will be conducted.
Vibration isolation devices consist of passive and active
mounts, as well as a flexible bellows section (Beranek and
Figure 6. Fan load curve. Vér, 1992; Mathew, 2005).
8 Experimental Techniques for Fluid Dynamics and Thermal Science

4 FACILITY CHARACTERIZATION

A wind tunnel must be characterized to determine if the de-

sign criteria are satisfied. The most obvious test is to check
that the facility operates safely at desired flow speeds, but be-
yond this a series of careful experiments are recommended.
Some of the most common characterization experiments are
in the text below.

4.1 Flow uniformity Figure 7. Single-, dual-, and triple-wire CTA configurations.
Reproduced with permission from Dantec Dynamics (2009).
Clearly, uniform velocity in the test section is desired. The in-
viscid core of the test section flow should have as little devia- and Mehta, 2003), meaning the axial and lateral fluctuations
tion from a plug profile as possible. There are a large number can be drastically different. As such, a measurement scheme
of experimental techniques available to measure the mean should be used to resolve these two different velocity com-
velocity profile (Tavoularis, 2005), including traversing a ponents. In the simplest configuration, a measurement can be
pitot-static or pitot probe through the test section (see Pres- conducted with a single wire normal to the incoming flow,
sure and Velocity Measurements). This is a proven, inexpen- and then the wire can be rotated such that the test section ax-
sive, but time-consuming technique. A rake may be used as ial fluctuations would be rejected in a repeated experiment.
an alternative, so multiple spatial locations can be measured However, if two simultaneous velocity components are de-
simultaneously. Single or multiple hot wires configured for sired, a dual-wire probe configuration is usually used. For
constant-temperature anemometry (CTA), particle image ve- complete velocity steady and unsteady vector decomposi-
locimetry (PIV), and laser Doppler velocimetry (LDV) can tion, a triple wire assembly is required. These configurations
also be used. The flow uniformity is usually characterized as are shown in Figure 7. An array of hot wires can be used
either as a min–max or an rms deviation from the mean veloc- if spatial–temporal data or wave number spectra are desired.
ity in the inviscid core flow of the test section (Mathew, 2005). Turbulence intensity (TI) is typically computed as in equa-
tion (1) (expressed as a percent of the local mean velocity).
When computing TI, it is common to high-pass filter the data
4.2 Turbulence characterization at some cut-on frequency related to the mean flow speed and
test section size to separate true turbulence from large-scale
For many facilities, low turbulence levels in the test section facility unsteadiness, such as f > V∞ /(2LTS ) (Bradshaw and
are critical. While many full-scale applications may have sig- Mehta, 2003). Velocity spectra can be evaluated. An alterna-
nificant incoming turbulence levels, wind-tunnel tests often tive method uses LDV, although it requires flow seeding that
seek to isolate the effects of incoming freestream turbulence. must faithfully follow the turbulent flow (Tavoularis, 2005).
For example, in boundary-layer transition studies, signi- LDV can be used in single-, dual- (as shown in Figure 8),
ficant levels of incoming turbulence can alter the behavior of or triple-beam configurations to resolve one, two, or three
the boundary-layer transition location. In aeroacoustic stud- components of velocity. PIV provides 2-D or 3-D (if stereo)
ies, ingested turbulence can interact with model components data in a plane, enabling the computation of time-averaged
and generate additional “leading edge” noise, contaminating statistics. For more information on CTA, LDV, and PIV, see
measurements of other components of interest, such as airfoil Thermal Anemometry: An Introduction and Special Topics,
trailing edge noise (Brooks, Pope and Marcolini, 1989). Laser Doppler Velocimetry, and Particle Image Velocimetry.
As an unsteady velocity measurement, test section turbu-
u

lence levels are well suited for evaluation using constant- TI = (1)
V
temperature hot wire anemometry (Bruun, 1995). A single
hot wire can be traversed through the core of the test sec-
tion, and the bridge signal measured and analyzed at each 4.3 Acoustics and vibration
spatial location. Note that a single hot wire will resolve a
velocity magnitude normal to the wire line, with ambigu- If the facility is used for acoustic measurements, background
ous angle of incidence, while rejecting velocity fluctuations noise should be assessed. Ideally, an aeroacoustic flow facility
along the axis of the wire. It should also be noted that in most should have background noise levels at least 10 dB below
wind tunnels, test section turbulence is anisotropic (Bradshaw the acoustic source of interest in a test (Duell et al., 2002),
 Fundamentals of Wind-Tunnel Design 9

constraints and requirements. Completed facilities must be

carefully characterized to determine their characteristics.

RELATED CHAPTERS

The Role of Wind Tunnel Experiments in CFD Validation

Flow Visualization by Direct Injection Technique
Optical Flow Visualization
Shear Stress Measurements
Spectroscopic and Scattering Techniques
Temperature and Heat Transfer Measurements
Measurement of Aerodynamic Forces and Moments in Wind
Tunnels
Figure 8. Two-beam LDV configuration for two-component veloc-
ity measurements. Reproduced with permission from R. Holman
(2006)

c University of Florida. NOMENCLATURE
but this is rarely the case. Many measurement techniques A cross-sectional area (m2 )
are available for analyzing acoustic sources contaminated by a isentropic speed of sound (m s−1 )
background noise (Soderman and Allen, 2002), but all ben- BPF blade passage frequency (Hz)
efit from minimizing background noise levels as much as cp constant pressure specific heat (J kg−1 K−1 )
possible. Ec Eckert number V2 /cp T
Background noise levels are commonly measured using a Eu Euler number (p − p∞ )/ρV 2
single free-field microphone. This microphone can either be f frequency (Hz)
placed in the inviscid core of the test section (using a nose Fr Froude number V2 /gL
cone), flush mounted in the wall of a closed-wall test section, g gravitational acceleration (m s−2 )
or outside of the flow in an open-jet test section facility. The He duct exit height (m)
selection is typically based on where the majority of acous- Hi duct inlet height (m)
tic measurements will be made. If the overall sound pressure k thermal conductivity (W m−1 K−1 )
level (OASPL) is the quantity of interest, appropriate fre- L reference length scale (m)
quency bounds should be selected for the power integration. M Mach number V/a∞
For common applications where frequency shifts to full scale Nblades number of fan blades
are not a concern, A-weighting is often used (Duell et al., p pressure (Pa)
2002). Pr Prandtl number ␮cp /k
Vibration measurements can be conducted using single- or R fan rotation rate (Hz)
multi-axis accelerometers (Beranek and Vér, 1992). Absolute Re Reynolds number ρVL/µ
vibration levels can be evaluated. If any steps have been St Strouhal number fL/V
taken in the facility design and fabrication to mitigate vi- V reference velocity magnitude (m s−1 )
bration transmission through duct treatments, a transmission x streamwise location in contraction (m)
loss factor can be measured. This relates the vibration from T temperature difference (K)
a source, such as the tunnel drive system, to the vibration TI turbulence intensity
in the facility test section through correlation and coherence U local mean velocity magnitude (m s−1 )
analysis. u
local root mean square fluctuating velocity magni-
tude (m s−1 )
µ dynamic viscosity (kg m−1 s−1 )
5 CONCLUSIONS ρ density (kg m−3 )

Wind tunnels are an essential tool used by experimentalists to

complement analytical and computational methods. A wide SUBSCRIPTS
variety of wind-tunnel facilities exist, since tunnel design is
highly dependent on the application of interest. Tunnel design TS test section property
is an iterative process that must address various competing ∞ freestream conditions
 10 Experimental Techniques for Fluid Dynamics and Thermal Science

REFERENCES Aerospace Ground Testing Conference. AIAA Paper 92-3883.

Nashville, Tennessee, July 1992.
Amiet, R.K. (1978) Refraction of sound by a shear layer. J. Sound NASA Aeronautics Test Program (2007a) NASA’s Aero-
Vib., 58(4), 467–482. nautics Test Program 11-Foot Transonic Unitary Plan Facility,
http://www.aeronautics.nasa.gov/atp/facilities/documents/06am-
Barlow, J.B., Rae, W.H. Jr. and Pope, A. (1999) Low-Speed Wind
es11x11 dsi.pdf (accessed 18 June 2009).
Tunnel Testing, 3rd edn, John Wiley and Sons, New York.
NASA Aeronautics Test Program (2007b) NASA’s Aeronautics
Beranek, L.L. and Vér, I.L. (1992) Noise and Vibration Control
Test Program 14×22 Foot Subsonic Tunnel, http://www.
Engineering – Principles and Applications, John Wiley & Sons,
aeronautics.nasa.gov/atp/facilities/documents/03langley14x22
New York.
dsi.pdf (Accessed 18 June 2009).
Brooks, T.F., Pope, D.S. and Marcolini, M.A. (1989) Airfoil self
NASA Aeronautics Test Program (2007c) National Tran-
noise and prediction. NASA Reference Publication 1218, July
sonic Facility, http://www.aeronautics.nasa.gov/atp/facilities/
1989.
documents/07langleyntf dsi.pdf (accessed 18 June 2009).
Bruun, H.H. (1995) Hot-Wire Anemometry, Principles and Signal
NASA Langley Research Center (2006) The Langley 20-
Analysis, Oxford University Press, New York.
Inch Supersonic Wind Tunnel, http://wte.larc.nasa.gov/facilities
Bradshaw, P. and Mehta, R. (2003) Wind Tunnel De- updated/fluid dynamics/20inch supersonic2.htm (accessed 18
sign, http://www-htgl.stanford.edu/bradshaw/tunnel/ (accessed June 2009).
18 June 2009).
Naval Warfare Center Carderock Division (2009) Towing Basins,
Choi, K. and Simpson, R.L. (1987) Some mean velocity, turbulence http://www.dt.navy.mil/hyd/fac/tow-bas/index.html (accessed 21
and unsteadiness characteristics of the VPI & SU stability wind July 2009).
tunnel. Report VPI-Aero-161, December 1987.
The Pennsylvania State University (2007) Garfield Thomas Wa-
Collar, A.R. (1937) Some experiments with cascades of aerofoils. ter Tunnel Facilities, http://www.arl.psu.edu/facilities/gtwt.html
A.R.C., Reports and Memoranda. Technical Report 1768. (accessed 18 June 2009).
Dantec Dynamics (2009) http://www.dantecdynamics.com/Default. Pope, A. and Goin, K.L. (1965) High-Speed Wind Tunnel Testing,
aspx?ID=1057. John Wiley & Sons, New York.
Derbunivich, G.I., Zemskaya, A.S., Repik, E.U. and Sosedko, Y.P. Rasmi, S. (2002) Characterization of the University of Florida air–
(1987) Effect of flow contraction on the level of turbulence. Izv. water shear layer facility. M.S., MAE Department, University of
Akad. Nauk SSSR Mekh. Zhidk. Gaza, 2, 146–152. Florida.
Duell, E., Walter, J., Arnette, S. and Yen, J. (2002) Recent advances Runstadler, P.W., Dolan, F.X. and Dean, R.C. (1975) Diffuser Data
in large-scale aeroacoustic wind tunnels. 8th AIAA/CEAS Aeroa- Book, Creare, Inc. Hanover, New Hampshire.
coustics Conference and Exhibit. AIAA Paper 2002-2503. Breck-
Salter, C. (1946) Experiments on thin turning vanes. A.R.C., Reports
enridge, Colorado, June 2002.
and Memoranda, no. 2469.
Duell, E., Yen, J., Walter, J. and Arenette, S. (2004) Boundary
Schubauer, G.B., Spangenberg, W.G. and Klebanoff, P.S. (1950)
layer noise in aeroacoustic wind tunnels. 42nd AIAA Aerospace
Aerodynamic characteristics of damping screens. NACA TN
Sciences Conference. AIAA Paper 2004-1028. Reno, Nevada,
2001.
January 2004.
Saric, W.S., Reed, H.L. and White, E.B. (2003) Stability and tran-
Gelder, T.F., Moore, R.D., Sanz, J.M. and McFarland, E.R. (1986)
sition of three-dimensional boundary layers. Annu. Rev. Fluid
Wind tunnel turning vanes of modern design. 24th Aerospace
Mech., 35, 413–440.
Science Meeting. AIAA Paper 86-0044. Reno, Nevada, January
1986. Soderman, P.T. and Allen, C.S. (2002) Microphone measurements
in and out of airstream (Chapter 1), in Aeroacoustic Measure-
Holman, R., (2006). An experimental investigation of flows from
ments (ed. T.J. Mueller), Springer-Verlag, Berlin, Heidelberg &
zero-net mass-flux actuators. PhD. MAE Department, University
New York.
of Florida.
Stratford, B.S. (1958) The prediction of separation of the turbulent
Mathew, J. (2005) Design, fabrication and characterization of an
boundary layer. J. Fluid Mech., 5, 1–16.
anechoic wind tunnel facility. PhD. MAE Department, University
of Florida. Su, Y. (1992) Flow analysis & design of three-dimensional wind
tunnel contractions. AIAA J., 29(11), 1912–1919.
Mehta, R.D. (1977) The aerodynamic design of blower tunnels with
wide angle diffusers. Progr. Aerosp. Sci., 18, 59–120. Tavoularis, S. (2005) Measurements in Fluid Mechanics, Cam-
bridge University Press, New York.
Mehta, R.D. and Bradshaw, P. (1979) Design rules for small low
speed wind tunnels. Aeronaut. J., 83, 443–449. Tennekes, H. and Lumley, J.L. (1972) A First Course in Turbulence,
MIT Press, Cambridge, Massachusetts.
Morel T. (1975) Comprehensive design of axisymmetric wind tun-
nel contractions. J. Fluids Eng., 97(2), 225–233. Watmuff, J.H. (1998) Detrimental effects of almost immeasur-
ably small freestream nonuniformities generated by wind-tunnel
Mueller, T.J., Scharpf, D.F., Batill, S.M., Sullivan, C.J. and Sub-
screen. AIAA J., 36(3), 379–386.
ramanian, S. (1992) The design of a subsonic low-noise low
turbulence wind tunnel for acoustic measurements. AIAA 17th

View publication stats