11MSEC-0015 Experimental Analysis of Flow Past NASCAR COT Rear Wing Versus Spoiler Traveling Forward and Backwards using PIV Measurements and Flow Visualization

Author, co-author list (Do NOT enter this information. It will be pulled from participant tab in MyTechZone) Affiliation (Do NOT enter this information. It will be pulled from participant tab in MyTechZone)

Copyright © 2011 SAE International

ABSTRACT

There have been claims that the rear wing on the NASCAR Car of Tomorrow (COT) race car causes lift in the condition where the car spins during a crash and is traveling backwards down the track at a high rate of speed. When enough lift is generated, the race car can lose control and even become airborne. To address this concern, a new rear spoiler was designed by NASCAR to replace the wing and prevent this dangerous condition. This paper looks at the flow characteristics of both the rear wing and the new spoiler using particle image velocimetry (PIV) to provide both quantitative and qualitative analysis as well as hydrogen bubble flow visualization. These experiments are done in a continuous flow water tunnel having a cross section of 1.0 m2 using a simplified 12% scale model COT body with either a wing or spoiler attached. Flow structures are identified and compared for both the wing and spoiler under Reynolds number conditions between 1x105 and 3x105. We also review the same conditions when the car is traveling backwards as it might during a crash. This paper highlights the differences and similarities between the two devices, providing insights into the advantages and disadvantages of the new design.

INTRODUCTION

The NASCAR Car of Tomorrow (COT) is a racing vehicle designed to operate at high speeds (>300 kph) in close proximity to other racecars. It was introduced in 2007 as a completely new chassis and body design from the previous generations of cars in the racing series. Aerodynamic research played a crucial role in the design and evolution of the vehicle for both performance and safety. The wing (Figure #) was designed to provide downforce (negative lift) on the rear axle while producing a “smoother” or less turbulent wake than would the same car fitted with a spoiler. The intent of this was to allow trailing vehicles to approach in the car’s wake without losing control due to the stochastic airflow [1]. (“Stochastic” reads strange. How about “highly turbulent”?)

(I think it might also be appropriate to mention the highly effective ‘Splitter’ that provides downforce at the front of the car).

While the winged car met these design criteria, a safety issue arose in situations where the car was turned around backwards during high-speed travel—not an uncommon situation in this race series. When this happened, it was found that the car would generate rear lift. This lift led to loss of handling control and could even cause the vehicle to rise off of the road surface completely and fly into the air. This problem existed for previously designed racecars in the series and a solution was found and already in use. The solution consisted of devices commonly referred to as “roof flaps.” The “roof flaps”, shown in Figure #, are passive aerodynamic panels located at the rear of the roof that are hinged at the leading edge. When the car is traveling forward and travelling smoothly through somewhat ambient air, the roof flaps remain flat and do not disturb the flow. When the car is turned around backwards, the wake is a high speed flow of air being dragged along and results in a low-pressure bubble at the rear of the roof that pulls the flaps up and into the air stream. When the panels are extended, they provide a large amount of drag and spoil the lift over the car and prevent the car from lifting [2]. These roof flaps were carried over to the COT design; however, they did not function consistently on the winged vehicle.

Page 1 of 11

Figure # - Parametric model of COT with rear wing attached. http://www.creativecrash.com/rhino3d/marketplace/3d- models/vehicle/cars/c/nascar-cot-stock-cars-joe-gibbs-racing-pack

Figure # - Illustration of roof flap function. Nelson, G. et al [2]

This roof flap image shows the air going right through the spoiler and might not be what we want.

There have been many suggestions made as to why the winged COT had this safety issue; however, there is no research on the topic in the open literature. There have been two popularly made suggestions as to why this safety issue exists with the winged car, and it is unclear which, if either, is responsible. The first suggestion is that the wing is designed to create downforce when traveling forward but when facing backwards, creates lift instead. The second suggestion is that the dominant effect is the ineffectiveness of the roof flaps in deploying. NASCAR has chosen to replace the wing with a deck-mounted spoiler (Figure #) to alleviate the problem. The effectiveness of the spoiler as well as the underlying physical cause of the wing’s safety issue have yet to be fully understood.

Page 2 of 11

Figure # - Parametric model of COT with deck mounted spoiler. http://www.creativecrash.com/rhino3d/marketplace/3d- models/vehicle/cars/c/cot-nascar-2010-collection

NASCAR as well as independent race teams have done extensive experimental and computational testing on these vehicles, but have generally not made their results publically available. Additionally, research on these vehicles has been primarily done in wind tunnels and using CFD. While both of these methods provide useful lift and drag measurements, they do not provide the same insight as experimental flow visualization in a water tunnel. Flow visualization on these vehicles has been done using smoke sheets in the airflow stream as well as through the use of complex wake imaging systems [3]. The use of smoke visualization is limited, as it does not generally provide detailed images of coherent structures in a turbulent flow. Wake imaging systems employing pressure probes are usefulgood tools for validatingto help validate results, but still do not provide detailed flow structure visualization. CFD analysis can provide detailed flow structure images; however, accurate predictions in regions of flow separation (as in the wake) can be difficult to make and the validity of these results should always be verified with experimental results [4].

The objective of this study is to qualitatively characterize the two aerodynamic devices (rear wing and spoiler) and evaluate the effectiveness of the new design (the spoiler) using an adaptive correlation particle image velocimetry (PIV) technique in a continuous-flow water tunnel with 10% scaled models. The wing and spoiler are compared under forward- and reverse-facing vehicle conditions. Because only the flow differences between the wing and the spoiler are under investigation, a simplified car body model with a smooth underbody and no ground plane is employed. Velocity field measurements are taken on the central plane of the model in the spanwise direction. The resulting velocity fields are time averaged and the data is analyzed and discussed.

While the Reynolds numbers are much lower for this study than those in the full-scale case, experiments done in other water tunnels have proven to be qualitatively accurate when vortex flows are more dominant than the viscous flow effects [5]. Data obtained from NASA’s water tunnel visualization studies on models as diverse as combat aircraft and gurney flaps have shown excellent correlation with tests in wind tunnel and real-world experiments at Reynolds numbers that are many orders of magnitude higher [5-7].

EXPERIMENTAL EQUIPMENT AND SETUP

This section describes the equipment and methodology used to perform the experiments in this paper. The PIV system consists of a double-head Nd-Yag Laser employing a submersed laser light periscope, timing controlled CCD camera, synchronizer, and computer for control and post processing. The test section of the continuous-flow recirculating water tunnel is1 m deep x 1 m wide x 3 m long and fully visible through tempered glass windows. The output energy of the Nd-Yag laser is #### mJ with a double wavelength of 512 nm and 1064 nm. The CCD camera has 8-bit and has #### x #### pixel resolution. The mean thickness (front to back depth), of the vertical laser light sheet is 2 mm which is approximately 0.4% of the characteristic length (% CL) of the model. The seeding particles are silver-coated hollow glass spheres with a mean diameter of #### mocrons. The model vehicles, shown in Figure #, are thermoformed clear polycarbonate shells mounted to stainless steel bases. These 10% scale models are 510 mm x 190 mm x 150 mm (length x width x height). The models are fixed inverted in the tunnel to allow future un-submerged load cell force measurements. They are attached on the underside of the cars by two thin stainless vertical steel plates, which are only 1.5 mm thick (<0.3% CL) in the spanwise direction and therefore do not have a significant effect on the underbody flow and trailing wake. Both models have the same geometric shapeproperties; however, one model has a wing affixed to the rear deck while the other has a spoiler. On the winged car model, the wing does not include endplates that are used in the actual COT (see Figure #) because only the center plane is being Page 3 of 11

analyzed and these should not have a significant effect. This exclusion of endplates allows a more complete profile view of the wing for the PIV camera. Additionally, the chord of the rear pivoting wing on the COT is adjustable from 0 to 16 degrees from horizontal. In this study, the model’s wing is fixed at the (race car typical) maximum 16-degree angle of attack.

Figure # - Schematic layout of experiment (THE ABOVE PICTURE A PLACEHOLDER)

The schematic experimental layout is shown in Figure #. Laser light travels down the periscope, which is located approximately 7 car lengths downstream of the model, and is split into a vertical and spanwise thin laser sheet. The laser sheet is centered on the model and illuminates the front, top and rear of the model as well as a small portion of the rear underbody. This large coverage area is possible through the use of a clear polycarbonate model that allows laser light to pass through and out the front of the car. For the two dimensional, time-averaged analyses, a single camera is used and 500 image sets or pairs are captured. The laser light is pulsed and image pairs are synchronously captured. To optimize velocity measurements, the time between images in each pair was set at 3000 µs (0.003s) and the image pairs are captured at 7.4 Hz. Each image pair yields one vector map through use of a Fast Fourier Transform (FFT) based cross-correlation algorithm [#]. With a measurement plane scale of #### pixels/mm, interrogation windows of 16 x 16 pixels, and #### % overlapping, an initial interrogation area of #### and #### is used. The spatial resolution of the velocity vectors in the measurement plane is ### vectors/mm corresponding to one vector every ### % CL.

RESULTS AND DISCUSSION

The model with the spoiler affixed will be referred to as SM (Spoiler Model) and the model with the wing will be referred to as WM (Wing Model). The configurations investigated include both models (wing and spoiler) under forward- and rear-facing conditions at three different mean-stream velocities of 0.25 m/s, 0.50 m/s and 0.65 m/s. These velocities correspond with Reynolds numbers equal to 1.3, 2.6 and 3.3 (x 105) respectively. It was found that the velocity did not affect the flow structures. Although this is a relatively small range, the result is consistent with the assertion that the predominantly vortical flow is qualitatively independent of Reynolds number. Data from only one velocity is shown in this paper, as showing all cases would be redundant.

Page 4 of 11

FORWARD-FACING DIRECTION

Figure # - Velocity vector maps of flow over COT models in forward-facing direction with free stream velocity of 0.25 m/s. Upper image is model with spoiler, lower image is model with wing. Flow direction to left. Red indicates fastest velocity and blue indicates slowest. (Vector fields are averaged over 500 images and ## seconds)

Figure # shows the velocity vector fields around both models with the flow going from front to the rear of the car (forward direction). It can be seen that the large-scale flow field is quite similar for both cars as would be expected. The flow upstream of the rear window is unaffected by either (rear downforce aero) device, therefore the only areas of interest in this case are the rear of the car and the wake. The scale of the wake in both cases is similar; however, the coherent structures in each are strikingly different. Figure # shows the streamlines around the models and Figure # shows a closer view of their wakes. The wake of the SM contains two large, counter- rotating vortices with a clockwise vortex in the lower region and a counterclockwise vortex in the upper region. The streamlines on the SM show heteroclinic (Sam: I’m not sure this is the right use of heteroclinic. The vortices you refer to are from highly averaged data and I don’t think there is a stable point in the actual flow) connections between the two vortices. The WM contains the same clockwise vortex in the lower region but a strong reverse-flow structure in the upper region of the wake. The focus of the lower vortex is in approximately the same location for both models. The difference between the flow structures in the two models is the expected result of fluid traveling underneath the wing. On the SM, all of the fluid traveling along the rear deck of the car is deflected upward, which creates a rolling vortex off of the tip of the spoiler (in reality and in real time, is this not a Vortex Street?). On the WM, much of the fluid traveling along the rear deck can continue in the free stream direction and carry more (free stream direction) momentum into the wake. This linear momentum sharply changes the trajectory of the fluid in the wake’s reverse-flow structure and rotating it back into the direction of the free stream flow.

Due to the large wake and relatively small wing or spoiler, it is assumed that inertial effects largely dominate the lift forces for this flow and that viscous forces are relatively small. It can be seen from Figure # that there is much more rotational motion coming off of the back of the spoiler than the wing. Viewed from the circulation theory of lift, this indicates that the spoiler is generating more downforce based on the shedding of vortices. This can also be inferred from the momentum theory of lift, which essentially examines the same fundamental phenomenon from a different perspective. There appears to be slightly more upward deflection of the fluid on the SM and conservation of momentum suggests that the larger deflection of fluid in the upward direction indicates a larger downward reactive force on the body of the car. Again, this suggests more downforce generation by the SM (this result is also found in comparisons of the “Aero-Maps” of the two configurations of the race cars at full scale wind tunnels).

The vortical structure in the upper region of the SM’s wake implies a much less “smooth” fluid field for a trailing car. In racing conditions, the upper portion of the wake would be contacting the top of a trailing car’s hood and flowing over its roof. Perhaps a better description is that the trailing car is driving through the wake of the car in front of it. The more linear flow in the upper part of the WM’s wake provides a less turbulent flow field on the hood of the trailing car and a more predictable aerodynamic force as a Page 5 of 11

function of time. The vortical structure of the SM’s wake creates an unstable flow field and makes it more difficult for the driver of the trailing car to maintain control of the vehicle.

Figure # - Streamlines around COT models in forward-facing direction with free stream velocity of 0.25 m/s. Upper image is model with spoiler, lower image is model with wing. Flow direction to left.

Page 6 of 11

Figure # - Vector map of wake behind spoiler model at free stream velocity of .25 m/s. Red indicates fastest velocity and blue indicates slowest. Flow to left.

Figure # - Vector map of wake behind winged model at free stream velocity of .25 m/s. Red indicates fastest velocity and blue indicates slowest. Flow to left.

REVERSE-FACING DIRECTION

Page 7 of 11

Figure # - Velocity vector maps of flow over COT models in reverse-facing direction with free stream velocity of 0.25 m/s. Upper image is model with spoiler, lower image is model with wing. Flow direction to left. Red indicates fastest velocity and blue indicates slowest.

Figures # and # show the velocity vector fields around the models when traveling in reverse. Again, the overall flow structures are very similar, but there are several very important differences between the two. As with the forward-facing direction, the area of interest is the region directly around the rear of the car. The flow over the SM is characterized by a single, large, counterclockwise, vortical wake behind the spoiler (ie. between the spoiler and the car’s roof) . This low pressure region or “bubble” sits on top of the rear deck and extends slightly over the rear portion of the roof before the free stream flow reattaches to the car’s body. The flow past the WM has three distinct vortices within its wake: a leading-edge, counterclockwise vortex on top of the wing, a trailing-edge, clockwise vortex from the bottom of the wing and counterclockwise vortex sitting over the rear deck and window. The difference in flow structure is again due to the fluid flow under the wing. The effect is perhaps even more pronounced in reverse because the wing acts almost like a funnel for the oncoming stream. The high-speed fluid flow under the wing forces reattachment of the mean stream to the roof.

On the SM, it can be seen that the low-pressure wake extends over the back edge of the roof flap, which is marked in Figure #. The low pressure in the recirculating region of the wake is what causes the roof flaps to open. On the WM, the wake ends at the top of the rear window and the mean stream flow is attached over the roof flaps. It is clear that under these conditions, the low-pressure wake does not deploy the flaps. This explains why the roof flaps do not consistently function on the COT when a wing is affixed. Although this might be the primary reason that the winged car is more likely to lift off of the road, the claim that the wing generates lift in reverse can also be seen.

Similar observations about the downforce can be made as for the forward-facing case. Viewing the flow from circulation theory perspective, the clockwise, trailing-edge vortex shed by the underside of the wing indicates lift. The small, counterclockwise, leading- edge vortex off the top edge of the wing indicates that the flow is separated over the top of the wing and producing very little downforce. Additionally, the two devices can be compared from a momentum theory perspective. More upward deflection can be seen in the velocity field on the SM than on the WM, which indicates more downforce on the SM body.

Page 8 of 11

Figure # - Velocity vector map of flow past spoiler in reverse-facing direction at free stream velocity .25 m/s. Red indicates fastest velocity and blue represents slowest. Arrow indicates rear edge of roof flap.

Page 9 of 11

Figure # - Flow past wing in reverse-facing direction at free stream velocity 0.25 m/s. Red indicates fastest velocity and blue indicates slowest. Arrow indicates rear edge of roof flap.

SUMMARY

An experimental flow visualization study was carried out on simplified models of a NASCAR COT race car. Qualitative flow analysis showed that in the forward-facing condition, the COT fitted with a spoiler generates more downforce than the same car fitted with an inverted wing. This is due to the momentum of the vertical deflection of fluid caused by the spoiler. The wing allows fluid to flow underneath and does not cause as much of the fluid to change direction vertically. This same flow underneath the wing provides a less turbulent wake behind the car than does the largely vortical flow behind the spoiler. Under racing conditions, the wake behind the winged car creates smoother conditions for a trailing vehicle.

In the reverse-flow direction—as might be experienced during loss of control, or in a spin during a wreck—it is shown that the winged car is more likely to lift off of the road surface for two reasons. The first is the larger relative magnitude of the downforce created by the spoiler. Again, this is due to the upward deflection of fluid hitting the spoiler and the ability for fluid to travel underneath the wing. In fact, when the car is turned around backwards, the wing appears to generate positive lift. (What! Where did this come from?) The second reason the winged car is more likely to lift off is due to a reduction in the ineffectiveness of the roof flaps due to counter- rotating flows of the wing. On the spoiler model, the low-pressure wake behind the spoiler extends over the roof flaps and pulls them

Page 10 of 11

up. The flow underneath the wing, on the other hand, forces the free stream fluid to remain attached to the roof over the flaps. The attached free stream does not create low pressure over the flaps and they do not consistently deploy.

REFERENCES

1. Leslie-Pelecky, Diandra. “The Physics of NASCAR: How to Make Steel + Gas + Rubber = Speed”. London: Penguin Books Ltd, 2008. 2. Nelson, G., Roush, J., Eaker, G., Wallis, S., “The Development and Manufacture of a Roof Mounted Aero Flap System for Race Car Applications”, SAE Technical Paper 942522, 1994. 3. Brzustowicz, J., de La Rode, JM. “Experimental & Computational Simulations Utilized During the Aerodynamic Development of the Dodge Intrepid R/T Race Car”, SAE Technical Paper 2002-01-3334, 2002. 4. Hucho, W., “Aerodynamics of Road Vehicles”, SAE, 1998. 5. Cobleigh, B., Del Frate, J., “Water Tunnel Flow Visualization Study of a 4.4% Scale X-31 Forebody”, NASA Technical Memorandum 104276, 1994. 6. Johnson, S., Fisher, D., “Water Tunnel Study Results of a TF/A-18 and F/A-18 Canopy Flow Visualization”, NASA Technical Memorandum 101705, 1990. 7. Neuhart, D., Pendergraft, O., “A Water Tunnel Study of Gurney Flaps”, NASA Technical Memorandum 4071, 1988. 8.

CONTACT INFORMATION

Sam Hellman ([email protected])

Peter Tkacik ([email protected])

Mesbah Uddin ([email protected])

Scott Kelly ([email protected])

ACKNOWLEDGMENTS APPENDIX

Page 11 of 11