Wind tunnel testing

Aeronautical / Aerospace Engineering (B.E./B.Tech) · Tools and careers · free study notes for exams, placements and interviews

In short

A wind tunnel is a ducted facility in which air is driven past a stationary scale model so that the forces, moments, pressures and flow patterns on a real aircraft, vehicle, building or blade can be measured in controlled conditions. It works because of dynamic similarity: if the right non-dimensional numbers (Reynolds, Mach) match, the non-dimensional coefficients measured on a 1/10 model apply to the full-size body. Wind tunnel testing is the experimental half of aerodynamics — it validates CFD, certifies loads, and supplies the aerodynamic database for flight simulators. In India it is examined in B.E. aerodynamics and aerodynamics-lab papers, in GATE Aerospace (flow similarity, measurement, tunnel types), and practised at CSIR-NAL Bangalore (NTAF, 1.2 m trisonic tunnel), DRDO's ADE/DRDL, VSSC and LPSC (ISRO), HAL, IITs/IISc and Tata/Mahindra automotive aero centres.

Explained

What a wind tunnel is and why it is used

Principle of relative motion: aerodynamic force depends only on the relative velocity between body and air, so a fixed model in moving air is equivalent to a moving body in still air.
Purpose: measure lift, drag, side force and the three moments; map surface pressure; find stall angle, control effectiveness, stability derivatives, hinge moments, flutter boundaries, store-separation loads, inlet/intake distortion.
Why not only fly or only compute: flight tests are costly and risky; CFD still needs validation for separated, transonic and turbulent flows. Tunnel data is the reference ("truth") dataset.
Model: geometrically scaled, stiff, smooth, instrumented; mounted on a sting (rear), strut, or three-point support, or on a floor balance for ground-vehicle work.

Classification of wind tunnels

By return path:

Open-circuit (Eiffel/NPL type): air drawn from the room, passes once, exhausts out. Cheap, no heat build-up, easy flow visualisation with smoke; but noisy, power-hungry and sensitive to room draughts.
Closed-circuit (Prandtl/Göttingen type): air recirculates in a closed loop with corner vanes. Better flow quality, 3-5 times lower power for the same test speed, quieter; costlier and needs a cooler because the air heats up.

By test-section pressure/condition: open-jet (free boundary), closed test section, slotted/perforated wall (transonic).

By speed regime (M = test-section Mach number):

Type	Mach range	Drive / features
Low-speed subsonic	M < 0.3 (incompressible)	Fan/axial blower, continuous
High subsonic	0.3 < M < 0.8	Fan + cooler, compressibility matters
Transonic	0.8-1.2	Slotted or perforated walls to absorb reflected shocks
Supersonic	1.2-5	Convergent-divergent nozzle, needs pressure ratio; intermittent blowdown or induced-draught
Hypersonic	M > 5	Heated/stagnation storage, vacuum tank, shock tunnel, gun tunnel; very short run times

By run mode: continuous (motor-fan), blowdown (high-pressure reservoir → test section → atmosphere; run time seconds), induced/indraft (atmosphere → test section → vacuum tank), shock tunnel (milliseconds, for hypersonic/real-gas work).

Special tunnels: icing tunnel (water spray), smoke tunnel, water tunnel/towing tank (low speed, good visualisation), spin tunnel (vertical, free-spinning model), climatic/automotive tunnel with rolling road, environmental/boundary-layer tunnel for buildings (long fetch to grow an atmospheric boundary layer), free-flight and store-separation tunnels, propulsion/altitude test facility.

Parts of a subsonic tunnel, in flow order

Effuser/inlet with honeycomb and screens: honeycomb straightens flow direction; wire-mesh screens break large eddies and reduce turbulence intensity.
Settling chamber: large area, low velocity; lets disturbances decay. Stagnation conditions (p₀, T₀) are measured here.
Contraction (nozzle): accelerates the flow; contraction ratio = settling-chamber area / test-section area, typically 6-10 for low-speed tunnels. It raises velocity and reduces turbulence intensity roughly in proportion to the contraction ratio, so a bigger contraction gives a cleaner, more uniform test flow.
Test section: where the model sits; needs uniform velocity (within ~0.5%), low turbulence (0.1-0.5%), low flow angularity (<0.5°), optical access for visualisation. Slightly diverging walls compensate boundary-layer growth so the axial pressure gradient is near zero.
Diffuser: decelerates flow to recover pressure; half-angle kept small (about 3-5°) and area ratio moderate to avoid separation. Recovered pressure is the main power saving.
Fan/drive section: axial fan with anti-swirl stators, driven by an electric motor; downstream of the test section so the fan's wake does not pollute the test flow.
Corners (closed circuit): four 90° turns with cascades of turning vanes; breather slot near the test section in open-return tunnels to keep static pressure near atmospheric.
Cooler/heat exchanger in closed circuits; anechoic lining in aeroacoustic tunnels.

In a supersonic tunnel the contraction is replaced by a convergent-divergent nozzle whose area ratio sets the Mach number; downstream a second throat/variable diffuser recovers pressure and reduces the required pressure ratio. Starting the tunnel needs a higher pressure ratio than running it (the normal shock must be swallowed through the test section).

Similarity, scaling and the Reynolds number problem

Non-dimensionalising the Navier-Stokes equations gives the governing parameters: Reynolds number (inertia/viscous), Mach number (compressibility), plus Strouhal (unsteady), Prandtl (heat transfer), specific-heat ratio γ, and geometric similarity.
Complete similarity requires geometric + kinematic + dynamic similarity. Then C_L, C_D, C_M, C_p of the model equal those of the aircraft.
Rule of regimes: below M ≈ 0.3 only Re need be matched; above it both Re and M must match; in hypersonic flow, wall temperature ratio and real-gas effects matter too.
The scale-effect problem: a 1/10 model at the same speed and conditions gives 1/10 the Reynolds number, so the boundary layer is relatively thicker, transition is late and separation/stall behaviour differs — drag and C_Lmax come out wrong.
Remedies: pressurised tunnels (raise ρ), cryogenic tunnels (lower T → higher ρ and lower μ, as in the European ETW), higher speed, larger models, and artificial transition — trip wires, grit or carborundum strips placed per Braslow's criterion to fix transition where it would occur in flight.
Mach-Reynolds trade-off: a blowdown tunnel can vary stagnation pressure to sweep Re at fixed M, which is how Re-effects are quantified.

Measurement and instrumentation

Velocity/pressure: Pitot-static tube (total minus static), static pressure taps on the wall and on the model, multi-tube manometers, electronic pressure scanners, five-hole probe for flow direction, Pitot rake for wake survey.
Forces and moments: internal strain-gauge balance inside the model on a sting (six components); external balance (platform/pyramidal) under the floor; wake rake / momentum method for profile drag; floor balance with rolling road for cars; load cells for engine thrust.
Flow field: hot-wire anemometer (turbulence, high frequency response), Particle Image Velocimetry (PIV) with seeded particles, laser sheet and camera pairs (whole-field, non-intrusive), Laser Doppler Velocimetry (point velocity).
Visualisation: smoke/tufts/china-clay and oil-flow for surface streamlines and separation; pressure-sensitive paint (PSP) for continuous C_p fields; Schlieren (density gradient — shows shocks as sharp lines), shadowgraph (second derivative of density), interferometry (density itself).
Thermal/other: infra-red thermography for transition, microphone arrays and beamforming for aeroacoustics, accelerometers and high-speed cameras for flutter and aeroelastic tests.
Calibration: tunnel is calibrated empty for velocity uniformity, turbulence level, flow angularity and axial pressure gradient; the balance is calibrated with dead weights and a calibration body; a standard model (e.g. AGARD-B for supersonic) cross-checks facilities.

Corrections and sources of error

Solid blockage: the model displaces area, so local velocity rises; wake blockage: the wake's momentum deficit raises velocity further in a closed section. Both make the true dynamic pressure at the model higher than the value taken from the empty-tunnel calibration, so the measured q is too low and the coefficients come out too high unless corrected.
Lift interference / wall-induced upwash: walls constrain the trailing vortex system, changing effective incidence and induced drag — corrected by classical (Glauert/Maskell) formulae or by wall-pressure-signature methods.
Horizontal buoyancy: the small axial pressure gradient in the test section produces a spurious drag — corrected from the empty-tunnel calibration.
Support interference: sting/strut changes the base flow; measured by image/dummy-support tests and subtracted.
Other errors: tunnel turbulence raising measured drag or early transition, aeroelastic deformation of the model, temperature drift of strain gauges, blockage in supersonic tunnels (too large a model unstarts the tunnel), Reynolds mismatch, and instrumentation lag.
Keep models small: usual rule is model frontal area under about 5-10% of test-section area for low speed, and far less for transonic/supersonic work.

Indian facilities, practice and safety

CSIR-NAL, Bangalore: 1.5 m low-speed tunnel, the National Trisonic Aerodynamic Facility (NTAF) 1.2 m blowdown tunnel covering subsonic to supersonic, a 0.6 m trisonic tunnel, hypersonic and shock tunnels, and the 4 m × 2.75 m-class low-speed facility used for aircraft and civil-engineering models. Used for LCA Tejas, SARAS, RLV, launch vehicles.
ISRO (VSSC, Thiruvananthapuram): 1 m trisonic tunnel and hypersonic facilities for PSLV/GSLV/LVM3 and RLV-TD aerodynamic characterisation.
IISc Bangalore: Laboratory for Hypersonic and Shock Wave Research (free-piston shock tunnels, HST series). IIT Kanpur, IIT Bombay, IIT Madras, IIT Kharagpur, MIT-Manipal and many AICTE colleges run subsonic tunnels for teaching and research.
Automotive/civil: full-scale climatic and aero tunnels in the Indian auto industry (e.g. facilities at ARAI Pune and the Chennai/Chakan OEM centres) and boundary-layer tunnels for wind loads on buildings.
Standards/codes: aircraft work follows DGCA/CAR requirements derived from FAR/CS-25 and military airworthiness (CEMILAC) specifications; wind loads on structures follow IS 875 (Part 3) — wind loads on buildings and structures, whose coefficients and terrain categories come largely from boundary-layer tunnel data; model-test practice for structures broadly follows AWES/ASCE 49-type guidance. ISO 5167 covers differential-pressure flow measurement. Laser work needs eye-safety control (IEC 60825 laser classes); tunnel operation is covered by the occupational safety and health provisions of the Occupational Safety, Health and Working Conditions Code, 2020, in force from 21 November 2025, which replaced the Factories Act, 1948. Compressed-air reservoirs for blowdown tunnels fall under pressure-vessel rules (SMPV/static and mobile pressure vessel regulations) — exact applicability depends on volume and pressure.

Key points

Wind tunnel - a duct in which controlled airflow passes a stationary scale model to measure aerodynamic forces and flow.
Dynamic similarity - matching Re and M makes model coefficients equal to full-scale coefficients.
Reynolds number Re = ρVL/μ - ratio of inertial to viscous forces; governs boundary layers, transition and separation.
Mach number M = V/a - compressibility parameter; must be matched above M ≈ 0.3.
Open vs closed circuit - open is cheap and simple; closed gives better flow quality and needs 3-5 times less power.
Contraction ratio (6-10 typical) - raises velocity and cuts turbulence intensity in the test section.
Settling chamber with honeycomb and screens - straightens flow and reduces turbulence before the contraction.
Diffuser - recovers pressure at 3-5° half angle; the main reason a closed tunnel is power-efficient.
Convergent-divergent nozzle - area ratio A/A* sets the supersonic test Mach number.
Blowdown tunnel - reservoir-driven, seconds of run time, variable stagnation pressure to sweep Reynolds number.
Shock/gun tunnel - millisecond hypersonic runs with high enthalpy; used for re-entry and scramjet work.
Internal strain-gauge balance - six-component force/moment measurement inside a sting-mounted model.
Blockage and wall interference - closed-section walls raise velocity and change effective incidence; corrections are mandatory.
Support interference - sting/strut effect removed by dummy-support (image) tests.
Trip strip - artificial roughness that fixes transition when the model Reynolds number is too low.
Schlieren/shadowgraph - density-gradient optics that make shocks and expansion fans visible.
PIV and hot wire - whole-field and point measurements of velocity and turbulence.
Tunnel unstart - an oversized supersonic model prevents the shock from being swallowed, so supersonic flow is never established.

Formulas and examples

q = ½ ρ V² — dynamic pressure, Pa (ρ in kg/m³, V in m/s)
V = √(2(p₀ − p)/ρ) — Pitot-static velocity, incompressible; p₀ total, p static, Pa
Re = ρ V L / μ = V L / ν — L reference length (chord) in m; μ in Pa·s, ν in m²/s
M = V / a, a = √(γ R T) — γ = 1.4, R = 287 J/kg·K for air, T in K
C_L = L / (q S), C_D = D / (q S), C_M = M / (q S c) — S reference area (m²), c reference chord (m)
C_p = (p − p∞) / q — pressure coefficient, dimensionless; C_p = 1 at a stagnation point
A₁V₁ = A₂V₂ — incompressible continuity; contraction ratio A₁/A₂ = V₂/V₁
p₀/p = (1 + 0.2 M²)^3.5, T₀/T = 1 + 0.2 M² — isentropic air relations (γ = 1.4)
A/A* = (1/M)[(1 + 0.2M²)/1.2]³ — nozzle area ratio for Mach M (γ = 1.4)
Energy ratio ER = (kinetic energy flux in test section)/(drive power input) = ½ρV³A / P
Correction (Maskell-type, closed section): V_corrected = V(1 + ε), ε ≈ ¼ (model frontal area / test-section area) for solid blockage of a bluff body
Tunnel power P = (½ρV³A) × ΔH-loss factor; power ∝ V³, so doubling speed needs eight times the power

Example 1 - Dynamic pressure and force from a Pitot reading. A subsonic tunnel shows a Pitot-static difference of 1,200 Pa; air density 1.20 kg/m³.

V = √(2 × 1200/1.20) = √2000 = 44.7 m/s.
q = ½ × 1.20 × 44.7² = ½ × 1.20 × 2000 = 1,200 Pa (as expected, q = Δp).
A wing model of S = 0.30 m² reads L = 180 N on the balance → C_L = 180/(1200 × 0.30) = 0.50.
If D = 10.8 N → C_D = 10.8/360 = 0.030, so L/D = 16.7.

Example 2 - Reynolds number matching and the scale problem. An aircraft flies at V = 70 m/s with chord c = 2.0 m in air of ν = 1.5 × 10⁻⁵ m²/s. A 1/5 model (c = 0.4 m) is tested in a tunnel at the same ν.

Flight Re = 70 × 2.0/1.5 × 10⁻⁵ = 140/1.5e−5 = 9.33 × 10⁶.
To match Re, V_model = Re ν/c = 9.33 × 10⁶ × 1.5 × 10⁻⁵/0.4 = 140/0.4 = 350 m/s — about M = 1.03, which would introduce compressibility that the real flight does not have. Matching Re and M together is impossible at the same ν.
Pressurising instead: at the same 70 m/s, Re_model = 70 × 0.4/1.5e−5 = 1.87 × 10⁶, i.e. 1/5 of flight. To restore Re at the same V and T, raise density 5 times → stagnation pressure ≈ 5 atm (μ is nearly independent of pressure). Hence pressurised, cryogenic or large tunnels, or a trip strip as a cheap partial fix.

Example 3 - Supersonic nozzle area ratio and stagnation conditions. Design a M = 2.0 test section with throat area A* = 0.01 m².

A/A* = (1/2)[(1 + 0.2 × 4)/1.2]³ = 0.5 × (1.8/1.2)³ = 0.5 × 1.5³ = 0.5 × 3.375 = 1.6875 → A = 0.016875 m².
Pressure ratio p₀/p = (1 + 0.2 × 4)^3.5 = 1.8^3.5 ≈ 7.82; so with exhaust to 1 atm, the reservoir must supply at least ≈ 7.9 bar to run, and more to start (the starting ratio is roughly the running value times the normal-shock total-pressure loss, about 1/0.72 ≈ 1.4 times here, so ≈ 11 bar).
Temperature ratio T₀/T = 1.8 → with T₀ = 300 K, T = 167 K; a = √(1.4 × 287 × 167) ≈ 259 m/s, so V = 2 × 259 ≈ 518 m/s. Note the low static temperature — air driers are needed to prevent condensation.

Example 4 - Contraction ratio and power. A closed-circuit tunnel has a settling chamber 2.4 m × 1.8 m and a test section 0.8 m × 0.6 m.

Areas: 4.32 m² and 0.48 m² → contraction ratio = 9:1.
For a test speed of 50 m/s, settling-chamber speed = 50/9 = 5.6 m/s (low, so losses and turbulence there are small).
Kinetic-energy flux = ½ρV³A = 0.5 × 1.2 × 50³ × 0.48 = 0.5 × 1.2 × 125000 × 0.48 = 36 kW. With an energy ratio of 4 (typical closed circuit), shaft power ≈ 36/4 = 9 kW; an open-circuit tunnel with ER ≈ 1 would need ≈ 36 kW. Doubling to 100 m/s multiplies the KE flux by 8 → 288 kW.

Example 5 - Blockage correction. A car model of frontal area 0.05 m² is tested in a 1.0 m² closed test section at V = 40 m/s, measured drag 36 N.

Area ratio = 5%. Solid-blockage factor ε ≈ ¼ × 0.05 = 0.0125 → V_corr = 40 × 1.0125 = 40.5 m/s.
q_measured = 0.5 × 1.2 × 1600 = 960 Pa; q_corr = 0.5 × 1.2 × 40.5² = 984 Pa.
C_D measured = 36/(960 × 0.05) = 0.750; corrected C_D = 36/(984 × 0.05) = 0.732, i.e. about 2.4% lower. Ignoring blockage overstates drag.

Feature	Low-speed subsonic	Transonic	Supersonic blowdown	Hypersonic shock tunnel
Drive	Continuous fan	Fan + compressor	High-pressure reservoir	Shock/driver tube, vacuum tank
Run time	Hours	Hours	10-60 s	1-20 ms
Key hardware	Contraction, screens	Slotted/perforated walls	C-D nozzle, second throat	Diaphragm, heated driver gas
Main difficulty	Reynolds mismatch	Wall reflections, blockage	Starting pressure ratio, condensation	High enthalpy, very short data window
Typical measurement	Balance, pressure taps, PIV	Balance, PSP, Schlieren	Balance, Schlieren, pressure	Fast pressure/heat-flux gauges

Where it is used

Military and civil aircraft programmes: LCA Tejas, AMCA, HTT-40, SARAS and UAV development at HAL, ADA, DRDO-ADE with testing at NAL — force/moment databases, store separation, intake performance, flutter models.
Launch vehicles and re-entry: ISRO's PSLV, GSLV, LVM3, Gaganyaan crew module and RLV-TD aerodynamic and aerothermal characterisation in trisonic and hypersonic facilities at VSSC and IISc.
Automotive and rail: drag and cooling-flow optimisation with rolling-road tunnels at ARAI and OEM centres; crosswind stability for trains, buses and trucks.
Civil engineering and energy: boundary-layer tunnel testing of tall buildings, stadium roofs, bridges (vortex shedding, galloping) for IS 875 (Part 3) wind loads; wind-turbine blade and solar-panel array studies.
Job roles: experimental aerodynamicist, wind-tunnel test engineer, instrumentation and data-acquisition engineer, model design engineer (CAD + machining), aerodynamic analyst correlating tunnel data with CFD; recruiters include NAL, DRDO labs, ISRO centres, HAL, Tata Advanced Systems, L&T, Mahindra Aerospace, Airbus/Boeing India engineering centres.
Research and teaching: every AICTE aeronautical programme runs a subsonic tunnel lab (C_p distribution over an aerofoil, drag of bluff bodies, boundary-layer survey, smoke visualisation) — these labs are the commonest viva topic.

Interview questions

Q: Why can a stationary model in moving air represent a moving aircraft? A: Aerodynamic forces depend only on relative velocity between body and fluid, so reversing which one moves leaves the flow field (and hence forces) unchanged.
Q: What must match between model and full scale for the data to be valid? A: Geometric similarity plus the governing non-dimensional numbers — Reynolds number always, Mach number as well above about M = 0.3, and Strouhal number for unsteady work.
Q: Why is it usually impossible to match Reynolds and Mach numbers together? A: A smaller model needs a higher velocity to keep Re, but that raises M; the fix is to change density or temperature (pressurised or cryogenic tunnels) rather than speed.
Q: What is the purpose of the contraction and what ratio is typical? A: It accelerates the flow to test speed and reduces turbulence intensity and non-uniformity; typically 6-10 for low-speed tunnels.
Q: Why is the fan placed downstream of the test section? A: So the swirl, wake and noise generated by the fan do not contaminate the test flow, and so the fan works on decelerated, lower-velocity air.
Q: Why does a closed-circuit tunnel need far less power than an open one? A: The diffuser and return circuit recover most of the kinetic energy so the fan only makes up circuit losses, raising the energy ratio from about 1 to 3-6; it does need a cooler.
Q: What is a trip strip and when is it used? A: A roughness element (wire, grit, tape) placed near the leading edge to force laminar-to-turbulent transition when the model Reynolds number is too low to transition naturally, making the boundary layer behaviour flight-like.
Q: Explain solid and wake blockage. A: The model's volume reduces the effective flow area and raises local velocity (solid blockage), and the wake's momentum deficit forces a further acceleration outside it (wake blockage); both make measured q too low, so coefficients need correction.
Q: How is a six-component internal balance different from an external balance? A: The internal strain-gauge balance sits inside the model on the sting and measures normal, axial and side forces plus three moments directly; an external platform balance sits outside the tunnel and measures reactions at the supports, with greater support interference.
Q: What does Schlieren show and how is it different from a shadowgraph? A: Schlieren responds to the first derivative of density (gradient) and gives sharp shock images; shadowgraph responds to the second derivative and is better for locating discontinuities but gives less quantitative information.
Q: What is tunnel unstart in a supersonic facility? A: If the model blockage is too large, the normal shock cannot be swallowed through the test section, so supersonic flow is never established and the data is invalid — model size and starting pressure ratio must be checked.
Q: Why does a supersonic tunnel need air drying? A: Expansion to the test Mach number drops static temperature far below ambient (167 K at M = 2 from 300 K stagnation), so moisture condenses and forms fog, changing density and releasing latent heat.
Q: Why is the test section slightly divergent in a low-speed tunnel? A: To offset boundary-layer growth on the walls so the effective area stays constant and the axial static-pressure gradient (and hence horizontal buoyancy drag) is nearly zero.

Common mistakes

Using dynamic pressure from the empty-tunnel calibration without applying blockage or buoyancy corrections, then quoting the inflated C_D as a result.
Assuming a 1/10 model at the same speed gives the same Reynolds number — it gives one-tenth, and the stall and drag behaviour change with it.
Applying incompressible Bernoulli (V = √(2Δp/ρ)) above about M = 0.3, where the compressible form of the Pitot equation is needed.
Forgetting that the C-D nozzle area ratio has two solutions (subsonic and supersonic) for the same A/A*, and that the supersonic branch requires the correct back pressure.
Mixing reference areas and lengths between model and aircraft when converting coefficients to forces, or using planform area in one and frontal area in the other.
Reporting sting-mounted data without subtracting support interference from dummy-support runs, especially base drag.
Forgetting unit consistency — mm of water on a manometer is about 9.81 Pa per mm, kgf on an old balance is 9.81 N, and gauge versus absolute pressure in blowdown calculations.
Treating the tunnel's turbulence level as irrelevant; a noisy tunnel transitions the boundary layer early and reports a different drag and stall angle than flight.