Formula 1 (F1) racing is made for TV: the larger-than-life driver and owner personalities, the colorful cars, and the excitement of the high-speed race itself draw millions of fans. Of course, not seen are the long hours of development and testing undertaken to give one car a slight edge over another.
At speeds greater than 200 mph, the role played by aerodynamics is critical. Ensuring sufficient downforce to improve cornering while simultaneously reducing drag is being approached from two directions. The traditional method uses wind tunnel testing to evaluate all aspects of a car�s aerodynamics. Even the driver�s helmet shape must be considered.
By modern standards, wind tunnel testing is only part of the story. Today, virtually all F1 teams use at least one wind tunnel, but they either own or have access to a supercomputer. The computer runs detailed fluid dynamics analyses so designs are as theoretically correct as possible before undertaking wind tunnel testing. Computational fluid dynamics modeling often is the first route to improved performance, but physical testing still is necessary to prove a new design�s effectiveness and that changes have not caused unexpected problems.
Wind Tunnel Operation
Conceptually, wind tunnel operation is straightforward. A large fan either blows or sucks air through a test area where forces on the DUT can be measured. For aircraft engine testing, open-circuit tunnels have an advantage. Combustion products and heat from the engine are exhausted from the end of the tunnel, and only fresh air is drawn in the inlet end. This configuration also benefits cold-weather engine testing in locations such as Canada.
In general, however, closed-circuit tunnels that recirculate the same air can be better controlled. For example, the air can be exceptionally clean because once the tunnel is full of clean air, large volumes don�t need to be replaced. The temperature of the air can be closely controlled. In addition, the tunnel environment is totally isolated from the effects of wind gusts so the pressure profile across the test area is more uniform.
Many different types of wind tunnels have been developed during the 100 years since Eiffel conducted his early experiments. As wind-speed requirements have increased, so too has the power needed to drive the giant fans. For example, the National Research Council of Canada operates a closed-circuit tunnel driven by a 9,000-hp motor to achieve 180 ft/s velocity through the 30′ x 30′ x 75′ test section.
Ames Air Force Base houses the largest low-speed wind tunnels. The speed range in the 40′ x 80′ test section is continuously variable from 0 to 300 knots while the 80′ x 120′ test section is continuously variable from 0 to 100 knots. The air in the tunnel is driven by six 40′ dia, 15-bladed, variable-pitch fans powered by 40-pole, 6,600-V, three-phase synchronous motors each rated at 12 MW (18,000 hp) continuous with 2-hour 25% overload capability (22,500 hp).
These very large tunnels are used for full-scale testing of helicopters or small aircraft. They also can be used for testing scale models of very big aircraft. A large tunnel and a correspondingly large model ensure very close correlation between observed behavior in the wind tunnel and what can be expected in actual flight.
The effects of parameter scaling are encompassed in the dimensionless Reynold�s number (Re):
The test objective is to achieve the same Re for a scale model test that the full-size DUT will see in service. From the formula, it�s easy to see that if the characteristic length is reduced, as occurs when a scale model is tested, the flow velocity must be increased or the fluid viscosity decreased to compensate. As test speeds increase and achieving large test areas becomes impractical, smaller models must be used. Rather than increase wind speed beyond a certain point, some tunnels operate at pressures much higher than atmospheric and also may use gases more dense than air.
These extremes aren�t required for F1 wind tunnels. One of the newest, a Honda facility opened last year at the company�s racing team operations center in Brackley, U.K., accommodates a full-size car, achieves 180-mph test speeds, and is driven by a 3,000-hp motor.
A somewhat larger tunnel was built in 2004 for Sauber in Switzerland and now is run by BMW. This tunnel requires a 4,000-hp motor to drive its axial fan compressor, but its 15-m2 cross-section can accommodate full-size production cars and small vans as well as F1 cars. In addition, its long test section allows two model cars to be placed in line to simulate the disturbances caused by overtaking.
A tunnel�s axial fan increases air pressure. High velocity is achieved by forcing the high-pressure air through a restriction or nozzle. When the air leaves the restricted area, it expands and accelerates.
The design of the nozzle is critical to achieving uniform flow throughout the test area. For example, in the NASA Langley 14′ x 22′ tunnel, the flow in the test section is uniform with a velocity fluctuation of 0.1% or less. Subsonic airflow satisfies conservation of mass as well as Bernoulli�s equation so the ratio of test area to nozzle area can be calculated given the pressure developed by the fan and the desired test velocity.
Such uniform flow values ignore boundary effects very close to the tunnel walls as well as the interaction with the walls of any DUT-generated disturbances. To ensure valid test data, one source suggests keeping the DUT frontal area to less than 6% of the tunnel�s test-section cross-sectional area.
Using a small model does avoid wall effects but also lowers Re. You could use a very large tunnel and a full-size DUT, but that�s an expensive solution. Instead, test-area side walls have been made more flexible in an aerodynamic sense.
Some tunnels have slotted walls so that flow disturbances caused by the DUT can be dissipated rather than totally reflected. This makes the tunnel appear to be wider than it actually is. An earlier solution to the side-wall problem used open test sections where there were no side walls, but flow uniformity suffered.
A recent modification to the 10′ x 5′ wind tunnel at Imperial College in London added an outward bulge to the test-section walls. By allowing additional space for DUT-diverted airflow, side-wall effects were minimized. They cannot be entirely eliminated because the wall shape would need to be different for each DUT tested. However, this modification has allowed more realistic treatment of far-field streamlines.
Aircraft simply move through what ideally is a uniform air field. In contrast, an F1 car also is very close to a stationary track surface. To simulate the interaction between the car and track, a moving road runs underneath the car in the wind tunnel and causes the tires to turn. The surface of the moving road has the same velocity as the air. A sophisticated suction system controls the boundary layer at the front of the belt and ensures that the belt lies flat under the car.
Instrumentation
Much effort is spent in the design and operation of wind tunnels to ensure an undisturbed flow through the test area. Laminar flow along a DUT occurs at low Re, where viscous forces are dominant, and is characterized by smooth, constant fluid motion. Turbulent flow, on the other hand, corresponds to high Re and is dominated by inertial forces, producing random eddies, vortices, and other flow fluctuations.
The transition between laminar and turbulent flow often is indicated by a critical Re, which depends on the exact flow configuration and must be determined experimentally. A great challenge to aircraft designers is achieving completely laminar flow. It hasn�t been done, but because so much more energy is required to propel a plane to the same speed when turbulent vs. laminar flow is involved, interest remains high among researchers. For the same reason, laminar flow also is the goal for F1 car designers.
A critical Re exists for each part of an F1 car because the exact flow configuration in one area is influenced by that in another. For example, one race team has included a pair of small wings aft of the front suspension area. Their purpose is to remove sufficient turbulence from the airflow at that point to improve the aerodynamics at the rear of the car.
A recent research study of the transition from laminar to turbulent flow over a highly swept wing used nonintrusive infrared thermography. “The basic principle�for transition detection is differences in the heat-transfer coefficients of laminar and turbulent flows. The laminar boundary layer allows very low heat exchanges between the model surface and the surrounding freestream flow. Laminar regions are characterized by a very low heat-transfer coefficient and behave as a good insulator�.On the contrary, the turbulent boundary layer features high mixing and therefore high heat exchange. As a direct consequence, a surface characterized by a turbulent boundary layer will reach the temperature of the surrounding incoming flow faster than a laminar boundary layer.”
Figure 1 shows the results of one experiment with a 50� swept-wing model at Mach number 2.4 and -0.6� pitch. Airflow is from left to right. The diagonal yellow leading edge of the wing is very thin and has become warmer than the cooler laminar flow area immediately behind it.
Figure 1. Laminar Flow on 50� Swept-Wing Model at Mach 2.4
Courtesy of Arizona State University
This figure shows the laminar area extending to about 25% of the wing chord and ending in turbulent wedges caused by very small unintentional surface imperfections. The transition region extends noticeably farther back along the wing than in another report image made under the same conditions but with 0� pitch.1
In addition to temperature, air velocity and pressure also must be measured. A very common flow velocity sensor is the hot-wire anemometer. In one form of the device, an electric current heats a platinum or tungsten wire to a constant temperature. As the airflow increases, additional current is required to make up for the heat lost through convection. The additional current can be accurately related to air speed.
Pressure measurements account for many of the data acquisition channels used in wind tunnel work. Dynamic pressure is the pressure created by the flowing air and measured facing directly into the flow. Static pressure represents the overall air pressure and is measured at right angles to the flow.
Figure 2 shows a Pitot-static or Prandtl tube type of combined sensor. The pressure ports are arranged so that the sensor measures the difference between dynamic and static pressures.
Figure 2. Pitot-Static Tube
Courtesy of NASA
Bernoulli�s equation states that the sum of static and dynamic pressure equals the total pressure at a location. However, because the force of the moving air causes dynamic pressure, the equation is written as
Pt = Ps + ? x v2/2
where: ? = air density
v = air velocity
Ps = static pressure
Pt = total pressure
Given the air density, velocity can be calculated from pressure measurements.
Because a streamline represents airflow at constant pressure, if a sufficient number of pressures can be measured throughout a test volume, then streamlines can be determined. Once an experienced designer knows how the pressure varies over the surface of a DUT, he can quickly suggest possible aerodynamic improvements.
Pressure acting on a DUT causes opposing forces, and these are measured by a force balance system in the model mounting hardware. Sometimes an airfoil is mounted on the test-area wall similar in concept to the half-ship models boat designers made in the past. It�s common for aircraft or engine models to be supported by a thin shaft or sting extending from the rear of the model. This arrangement avoids airflow disturbances that would interfere with those created by the DUT. Support structures also provide model-positioning capabilities.
F1 cars may be suspended by thin struts attached to each wheel hub so the wheels are just touching the rolling road. A single large vertical strut may support the car or car model, and the force balance system is attached to the strut. The objective is to ensure that sufficient lift or downforce and minimal drag are being generated.
The Model 8400 from Pressure Systems has become standard equipment in many wind tunnels. It is a pressure-measuring system based on piezoresistive sensors that provides temperature compensation and a fourth-order linearization for each sensor. A scanner digitizing interface has been developed that improves performance significantly and reduces the time required for periodic calibration.
The new capabilities depend on electronically stored calibration constants, which imply that only repeatable thermal effects can be corrected. Indeed, “pressure sensors initially are selected for low offset and span drift and good thermal closure [low hysteresis]. Additional temperature and pressure cycles are used to further select sensors with low pressure and temperature hysteresis and high accuracy over temperature and pressure points within the operating range.” Correction coefficients are computed and stored in the scanner�s nonvolatile memory.
Before this new process was adopted, a user could expect a static accuracy of �0.05% FS with a typical thermal offset shift of �0.015% FS/�C and a typical span shift of �0.005% FS/�C. With digital compensation, the combined thermal effects have been reduced to less than �0.002% FS/�C. In situ pneumatic calibration still may be used to adjust offset and span, but only two reference pressures are needed, and the operation requires two or three minutes every four hours instead of up to six minutes each hour with the original system.2
Large data acquisition systems made by Neff are typical of wind tunnel data acquisition equipment. However, more recently, some of this instrumentation has been replaced by PC-based modular designs.
For example, Mindready Systems developed a data acquisition system for the 31″ Mach 10 tunnel at NASA Langley to replace an earlier Neff system. The company used several National Instruments� (NI) technologies and products to create the new system, which is claimed to have improved flexibility and accuracy.
In one of the Langley wind tunnels, “Each of the four Neff front ends provides signal conditioning, filtering, and sample-and-hold ADC for 64 channels for a total capacity of 256 channels. Data can be sampled at an aggregate rate of 300,000 S/s, typically providing a rate of at least 1 kS/s for all available model instrumentation. The computer systems supporting the data acquisition system perform archiving and continuous data buffering necessary to provide high- quality dynamic data during tests.”3
Newer Neff systems feature separate signal conditioning, true differential amplification, and an ADC per channel as well as tight channel-to-channel synchronization, high resolution, and fast sampling rates. Applications featured on the Neff website highlight transient capture, for example, as required for airbag testing. The systems can be controlled by the company�s Mozayik Software running under Windows NT and Windows 2000, and sub-VIs are supplied to link LabVIEW to live or database data.
The 30′ Canadian National Research Council wind tunnel uses a separate electronically scanned pressure-measuring system. The data acquisition system is PXI-based with 64 channels of analog-to-digital conversion and 128 digital I/O lines. Remote NI FieldPoint modules provide 16 RTD inputs and eight ADC channels.
This system also includes a 12-axis motion-control system with programmable proportional integral differential controller. Another PXI-based data acquisition system adds 64 ADC channels with programmable signal conditioning and 96 digital I/O. Data processing is via LabVIEW and MATLAB.
HBM is another brand of data acquisition system that may be found in some wind tunnels. HBM develops and manufactures many types of sensors and transducers as well as instruments so it can supply complete application solutions. HBM and Neff, both founded more than 50 years ago, were among the few companies capable of instrumenting wind tunnels during the �60s and �70s, long before powerful PCs and fast data conversion ICs were available.
Pi Research develops instrumentation and control hardware and software specifically for automotive wind tunnel testing. The company offers a complete service including 24/7 support and capabilities such as wiring-harness design and construction for all test setups as well as for the cars themselves.
The company�s Mistral package of sensors, signal conditioning, and data acquisition is well focused on F1 test requirements. As a result, more general-purpose, high-bandwidth transient signal capture has not been emphasized, a 200-S/s sampling rate being the highest listed in recent product brochures.
In addition to dealing with low-level analog signals from strain gages and pressure transducers, Mistral also interfaces to a range of laser-displacement sensors. Laser sensors provide accurate, repeatable measurements and are nonintrusive so their use requires little if any change to the DUT.
Conclusion
For F1 testing, a maximum wind speed of 200 to 300 mph is adequate. Unless part of the car is resonating or the airflow produces a resonance in a cavity, not much changes aerodynamically on a racecar in a few milliseconds.
This is not true of supersonic aircraft. In particular, the intake unstart event can occur very quickly and has major consequences. A description of the means used in the SR-71 Blackbird engine nacelle to avoid an unstart event demonstrates the complexity of high-speed aerodynamics design:
“In order for a turbojet to run properly, air must arrive at the engine face at a relatively low, subsonic speed. Also, engines function less efficiently with a turbulent air supply, so the inlet system has to deliver smooth flow to the engine. The only way air decelerates from supersonic to subsonic is across a shock wave or a series of shock waves. At high Mach numbers, it is necessary to decelerate the air through a series of swept or oblique shocks to get it down to subsonic speed without dissipating too much energy. The SR-71 [shown in Figure 3] uses a large central spike in the inlet to generate and control the system of shocks needed to guide air into the engine.
Figure 3. NASA Dryden�s SR-71B Blackbird Showing
Engine Nacelle Spikes
Courtesy of NASA
“Because of the Mach number range over which the airplane flies, the spikes must move to keep the shocks in the right place. The nacelle also has bypass doors that allow some of the air taken into the inlet to bypass the engine and be dumped into the exhaust ejectors. Air is also bled through holes in the center portion of the spike body to control the boundary layer and prevent flow separation. Much development was required to get all of the components of the inlet system to work optimally.
“When the inlet works the way it should, the bow shock from the spike is swallowed by the inlet and impinges on the inside of the inlet just aft of the lip. If turbulence or a small change in engine settings disturbs the flow, the shock can pop out of the inlet and become a much larger bow shock extending into the airflow around the airplane. When this happens, it is called an inlet unstart: A dramatic and unpleasant event.
“When the inlet unstarts, the drag on the nacelle increases, and the thrust from the engine behind the inlet drops. The result is a violent yaw in the direction of the unstarted nacelle. An unstart at speed could bang the pilot�s head against the canopy. The inlet spike actuators and control system had to be developed to the point at which the inlet would only unstart under extreme conditions and would self-recover quickly.”4
Engine nacelle flow fields are complex and require detailed study. The same is true of wheel and suspension areas in F1 cars. Hundreds of pressure sensors may be embedded within a model to determine the air pressure and speed over the critical areas. This kind of work is one of the reasons that wind tunnel instrumentation must support so many channels of accurate, simultaneous data acquisition.
References
1. Zuccher, S., et al., “The Role of Infrared Thermography in the Study of Crossflow Instability at M=2.4,” Arizona State University, Mechanical and Aerospace Engineering, 2003.
2. Juanarean, D. and Klaser, H., “Using Digital Thermal Correction of Electronic Pressure Scanners to Improve Windtunnel Productivity,” Pressure Systems.
3. “Transonic Dynamics Tunnel: NASA Langley Research Center,” Wind Tunnel Enterprise.
4. Wainfan, B., “SR-71 Blackbird,” Flight Journal, February 2000.
|
FOR MORE INFORMATION |
| Canadian National Research Council | Open-Circuit Wind Tunnel | www.rsleads.com/706ee-198 |
| FLIR | SC3000 Infrared Camera | www.rsleads.com/706ee-199 |
| HBM | CANHEAD� Distributed Data Acquisition System | www.rsleads.com/706ee-200 |
| NASA Wind Tunnel Enterprise | Wind Tunnel Testing | www.rsleads.com/706ee-201 |
| National Instruments | Compact FieldPoint | www.rsleads.com/706ee-202 |
| Neff | System 495 | www.rsleads.com/706ee-203 |
| Pi Research | Mistral F1 Data Acquisition and Control System | www.rsleads.com/706ee-204 |
| Pressure Systems | System 8400 | www.rsleads.com/706ee-205 |
June 2007