Talladega Teardown: Roof Flaps and Catch Fences

“You have to understand that, like, for years, we have had wrecks like this every time we come to Talladega ever since the (restrictor) plate got here, and for years, it was celebrated. The media celebrated it, the network celebrated it, calling it the Big One, just trying to attract attention.” –Dale Earnhardt, Jr.

My least favorite part of every Talladega pre-race show used to be watching Elliott Sadler tumbling through the infield. And you know, regardless of which network is carrying the race, they’re going to show a montage of crashes. Most will also show Bobby Allison’s 1987 crash – the one that instigated restrictor plates. Did anyone else have a really sick sense of deja vu watching the 99 car of Carl Edwards go airborne and tumble along the fence?

In both crashes, the cars because airborne and, instead of hitting the wall, the cars ended up in the catchfence. In 1987, Bobby Allison’s car pulled about 100 ft of chain-link fence down and sprayed debris into the crowd. A number of fans were injured, including a woman who lost an eye.

Most of the publicity about track safety has focused on SAFER barriers; however, there are researchers improving catchfences as well and the results were evident Sunday. The 99 rolled along the fence, but if you compare the two incidents, today’s fence did a much better job. Fences have difficult tasks: They have to allow spectators to see while being strong enough to contain a flying racecar. A stock car going 190 mph into a rigid barrier would produce a peak force of about 9700 lbs. The force the catchfence experienced would be less because fences have some give, but that’s still pretty impressive for steel and wire, especially if you look at how the 1987 fence did.

Fans and pundits are calling for NASCAR to ‘fix the car’ so that it stays on the ground. If you watch closely, you’ll notice that the primary safety device that guards against cars going airborne did work. Edwards’ roof flaps deployed and, in fact, you can see the car start to lower when the flaps come all the way up. That’s when the 39 got into him and was just far enough under to send the 99 into the air.

Matt Kenseth took a nasty roll during the Nationwide series race at Talladega on Saturday. In the numerous replays of the accident, I kept looking for the roof flaps rising as the car spun sideways. As far as I can tell, the roof flaps didn’t deploy on his Nationwide car. (NOTE: As Nick points out in the comments, this wasn’t an aerodynamic issue – it was more of a mechanical issue, like a SUV rollover. Thanks to Nick for pointing this out.)

Roof flaps are the two inserts on the top of the car. One runs along a line from left to right and one is angled at 45 degrees to the first, as shown in the figure below (which comes from the original patent #5374098).

Roof flaps (the invention of which I detail in my book The Physics of NASCAR) are designed to keep cars on the ground. Faster-moving air exerts less pressure and slower-moving air exerts more pressure. A wing develops lift because the air flowing under the wing moves slower than the air going over the wing. That creates more pressure underneath the wing than over the wing, which generates a net force upward. You want that for an airplane, but you don’t want it for a race car.

A NASCAR race car is pretty stable when airflow comes from the nose to the tail. The problems start when the car turns sideways because a sideways racecar looks a little too much like a wing. Air flows over the roof of a sideways racecar very quickly. It stays attached to the car’s surface for a long time, and that creates a low pressure region on the top of the car. A little air (or another car) gets under the car and all of a sudden, the car is an airplane. This only happens when the car rotates enough, so you need a solution that only becomes active when the car is really yawed.

You want the air to detach from the car’s roof, which increases the pressure on the top of the car and decreases the lift. That’s where the roof flaps come in. As shown below, the roof flaps are flaps of metal that are normally flush with the roof. If the pressure on the roof gets low enough, the pressure differential between the underside of the flap and the top of the lap causes the roof flap to pop up. A tether keeps the flaps perpendicular to the roof surface until the car gets going the right direction.

There’s an additional bit of help as well from the cowl area – the part of the car where the windshield meets the hood which also has flaps, as shown below.

The central flap is the air intake for the engine and the two piece called out on either side are the cowl flaps. Cowl flaps work the same way as the roof flaps, opening when the pressure on top becomes much lower that the pressure underneath. The opening of the cowl flaps allows air to escape from under the car and that also decreases lift. (The patent number on that one is #5544931.)

“(Drivers) have been saying this for years: racing like this is not a whole lot of fun.” –Dale Earnhardt, Jr.

In my opinion, calls for NASCAR to improve the roof flaps or other aerodynamic components so that they can keep the cars on the ground at Dega and Daytona regardless of circumstances are wrong-headed. NASCAR needs to decrease engine horsepower at the big tracks so that restrictor plates aren’t necessary. Slowing down the cars by making the restrictor plate hole smaller isn’t going to help. Most engine programs already have a totally separate plate engine program, so they’re putting a ton of effort into motors for just those four races already. Decrease the banking a bit, or make the banking different in different corners to introduce a ‘handling aspect’ that is present at Daytona and missing at Talladega. You can’t make the drivers ‘take it easy’, or suggest that changing the yellow line rule is going to solve the problem, because all it takes is a single freak coincidence. If Newman’s car hadn’t been right there, I suspect Carl’s car would have come down and hit more of the SAFER barrier instead of the catchfence. It looks from the most recent reports that the worst injury in the crowd is a broken jaw, but it could have been much, much worse.

What makes good racing? It’s the relative speed, not the absolute speed. If two cars are going nose-to-nose for the win, it’s just as exciting at 170 mph as it is at 190 mph. Does it really matter that much to you? It does to the drivers.

“I don’t know how I’d change this racing. I know it’s a spectacle for everybody and that’s great and all — but it’s not right to ask all these guys to come out and do this. What if the car goes up in the grandstands and kills 25 people? … I don’t know if I could live with myself if I ended up in the grandstands.” –Carl Edwards

NASCAR has a long, long history of being reactive. Here’s an instance in which NASCAR must be proactive and make changes before something really serious happens.

Here’s why I’m optimistic that might actually happen. I realized only recently that NASCAR hired Tom Gideon to be Director of Safety Initiatives for the R and D Center. Tom – someone whose integrity is respected throughout motorsports – was a prime moving force in motorsports safety at GM prior to his retirement. Tom will continue the tradition Steve Peterson started and, I’m sure, bring a number of his own initiatives to the job.

NOTE ADDED 4/28/09: A couple comments on some other reports: First, Reid Spencer reports that the roof flaps popped up in the wrong order; however, that should have very minimal – if any – impact on their function. Roof flaps are designed to disturb the flow of air over the roof of the car, as described above. The faster they deploy, the faster they can do their job. Even if one of the flaps ‘stuck’, it’s hard to believe that anything would have prevented the combination of the 09’s wake and the 39’s rapid approach from launching the 99.

“I would assume they just adapted their principles and the locations of the old style car to this new style car when it comes to the roof flaps and the cowl flaps and the things like that,” said Newman, who emphasized the importance of keeping the cars on the pavement.

Second, Ryan Newman has a Bachelor’s degree in vehicle dynamics. That doesn’t make him the authority on everything scientific. To Reid Spencer’s credit, he doesn’t let Newman have the last word. He cites Robin Pemberton, who notes that the roof flaps on the new car are larger to take into account the differing aerodynamics, and Bernie Marcus, Ford’s aerodynamicist, who points out that they did “at least three wind-tunnel tests” looking at differences in roof flap function on the new car. Unless Newman has some prior knowledge, it is arrogant and ignorant to suggest that NASCAR just transferred the roof flap design from the old car to the new car without testing it.

Patrick Canupp, Director of Aerodynamics at Joe Gibbs Racing, notes that he hasn’t seen the test data, but that NASCAR usually tests new car designs “at Lockheed where the yaw table can put the car at nearly any yaw angle.”, which means that they were in a position to look at roof flap function over a range of orientations of the car with respect to the airflow. Patrick also points out the most important element here – something that most media coverage glosses over. Engineering is not simple. You’re looking at complex machines and interactions between multiple complex machines. You can’t predict all the possible solutions and, even if you could, “that is still just a simulation of the actual event, whose detailed surface pressure history can be quite different,” Patrick notes.


  1. Most wind tunnels can only do a 6 degree yaw angle with a car, far more than when they’re actually running any Cup track.
    The issue with the 99 car is that it was punted by the 39 in which case roof flap effects are almost zero; this is similar to the fatal wreck of Indy cars at Lowe’s in 1999; one car wrecked, another car came along and punted the wheel of the first car into the stands (wheel and parts killed three fans).
    A second car hitting the parts of the first car in a wreck is common on most tracks. Springs coming out of a wrecked car are a real threat.
    The race tracks of today were designed in an entirely different era, not only were the cars slower, we were a great deal more casual about liability. The technology of race cars (speed and cornering) now is far beyond almost all tracks with the exception of the newer F1 layouts with huge run-offs and gravel traps. None of the ovals larger than 0.75 miles are safe for the speeds of modern cars (speaking as an engineer who does accident reconstructions).

  2. More interesting question is what has changed that makes it possible for two cars to out-run a train of a dozen; this was never possible with the Twisted Sister (car prior to the COT). Previously, the more cars in a line the faster they could go relative to one or two. That’s completely different, two are faster than ten,…what’s that about? We think we have a good idea but we’re running a number of calculations and doing CFD similations now to understand that one.

  3. Looking at some of the photos circulating on the ‘net, the fencing was breached in the 99 wreck on Sunday, only the restraining cables kept the car out of the crowd; a feature that was not part of the fence design in ’87.
    Look at the Le Mans race of ’55 where nearly a hundred people were killed when a fast car ran up over a slower one and was launched into the crowd. A wrecked Cup car turned around could allow its nose to act as a ramp for a second car with the same effect, the following car could be launched into the stands.
    In engineering one is responsible (legally) for considering the worst possible case which is “foreseeable,” and ever since ’87 a launched car, or parts of a car, is certainly a foreseeable case.

  4. How about a tire that is not as wide, that has a compound making it loose traction as it goes longer into the run?

    How about a small spoiler on the back like in the 70’s and raising the front spoiler to reduce down force making the driver lift in the corners.

    How about mandating more weight on the right side of the car?

  5. I am no physics expert but I suspect a 3400 lb car traveling 190mph in to a barrier will generate quite a bit more than 9,700 lbs of force.
    Also, if you examine the footage of the 99 car accident, you can see that the primary impact was not with the fence, it was the top of the roll cage into the top of the barrier wall. The fence did not take anything close to the full hit. More importantly, the car appeared to have been high enough at one point in it’s trajectory to have cleared that wall and struck the fence only. No one knows what would have happened if that was the case, but it would be relatively easy to find out. Simply use the video to arrive at the actual speed of the 99 car through the air when it struck the wall, and to measure the highest point in the trajectory, then build a ramp and drive a remotely controlled car in to the fence at that height and speed and see what happens. Lastly, several news reports said one of the spectators who was hit by debris, was struck by a “black, round, cylindrical object filled with bb’s or small round steel pellets” .What would be used in a Sprint Cup car that would fit that description?

  6. How about making catch fences just that – not PA speaker holders… Get the extra crap off the fence and there will be less to hit the fans.

  7. I hope Mr. Thompson will post the results of his calculations and simulations. In the meantime, my 7th grade engineering degree tells me that two are faster than twelve simply because the rear car of the two is now (the COT bumpers line up better)able to push the front car the entire way around the track. The twelve are still just bump drafting. You will probably argue that this phenomenon just began last weekend. I contend that it has been there since day one with the COT, the drivers have just now discovered it.

  8. Mr Gordon, the “black, round, cylindrical object filled with bb’s or small round steel pellets” sounds like one of the weights put in the hollow frame rails to manipulate both the weight balance of the car and the polar moment of inertia. Think of a figure skater as they do a spin, moving their hands and arms inward reduces the polar moment of inertia and they spin faster. Moving weights in the rails of car towards the center reduces the polar moment of interia making the car easier to rotate (turn), moving the weights out towards the ends of the rails reduces the tendency of car to rotate (makes it more difficult to turn). The rail weights are often solid lead blocks, about the size of a standard red brick, bolted into the rails, and I’ve seen them fly out of a car during a wreck in more than one instance. Figuring out how much weight and where are parts of the calculations done by a race engineer (vehicle dynamics).

  9. We think J Gibson is on the right path; several papers published by SAE among others have plotted drag as function of drafting distance (space between the cars), usually expressed as a fraction of the wheel base, nominally 110 inches for a Cup car, so a fraction of 0.05 is 5.5 inches. The published data shows little reduction in drag with drafting distances under 0.1 for several different body styles, and almost no data for distances down in the 0.05 range; two cars actually in contact and pushing together has not been well investigated at all.
    The Talladega track pavement is unusually smooth both in terms of a lack of local bumps and long distance contour making it possible for two cars to stay in contact whereas the same play at almost all other tracks would have the cars banging into each other, rather than making a uniform controlled push. Essentially everyone now runs the spring suspension system all the way down to the bump stops on the shock, so the cars ride at a very uniform height. And the bumper alignment, as Gibson correctly points out, also makes it possible for one car to push another. With the Twisted Sister, pushing a car would have resulted in the rear wheels of the front car being lifted off the ground by the nose of the second car getting under the bumper of the first. Aerodynamics of the Twisted Sister favored a nose down, tail up attitude, which promoted the pushing car lifting the rear of the pushee (if there is such a word), this is not true the winged COT which favors a more level drive configuration.
    Also, two cars in contact have a combined 1400 hp and the drag of the connected vehicles is lower than a string of ten with even small spaces between them—at any rate that’s the idea,…we’re doing CFD runs to see if the pressure contours on the surface of the cars actually supports that idea. If not, we’ll dream up some other theory until we find one that does conform to the data.

  10. If NASCAR required changes to the front/rear bumpers of the COT where they would not be so easy to bump draft, maybe this kind of accident does not happen again.

    Put the bumper-to-bumper impact point even with the radiator. Maybe put a little “chrome horn” extension on the back bumper to make it a less than inviting target. Bump draft, if you wish, but risk losing an engine.

  11. I suggest making the front stretch tri-oval sections of Talledega and Daytona either a quad oval similiar to Charlotte, Atlanta, etc… or “D”-shaped like Michigan and California. All of these accidents where a car or truck almost ended up in the stands were in the tri-ovals. Reducing the angle of the tri-oval by creating a quad oval or continuos arc would reduce the angle the cars hit the wall or the fencing. It may not be all that difficult to create a quad oval in the infield in front of pit road and if flat, it may slow speeds down. Also, if the stands stay where they are, they will be well away from the track.

  12. To appreciate the real depth of the problem NASCAR has you should read the Jeff Gluck piece in Scene Daily, and all of the responses. This is not an engineering issue, it is one of responsibility, and liability vs selling tickets. NASCAR is in an awful bind, they have a monumental problem: a multi-billion dollar business and almost unlimited liability.

    There’s clearly a large contingent that simply doesn’t care if eight spectators were injured; apparently those fans don’t count as long as the drivers were not hurt.
    All I can say is I’m glad it’s not my name and PE license number on the design drawings for the Talladega fences. And when a car finally does go into the crowd at ‘Dega, it will in an instant end motorsports in this country, along with NASCAR, ISC and SMI. As a race car driver I hate to see that day coming; on the other hand I assure you I don’t own any ISC stock.
    By contrast look at WRC, people standing along side the road as cars race by at more than 100 mph,.. and sometimes a car goes into the crowd. Europe doesn’t have the liability laws we do and Europeans have the attitude that you’re responsible for your own actions: if you stand next to a road where cars are racing you might get hurt. If you have a problem with that, then stand in some other location.
    We don’t have that attitude, perhaps we should, but hurling a race car into the stands at Talladega isn’t the way for us to improve.

  13. Which device failed? The car produced a driver that ran away. The steel cables held. But the catchfence failed and debris came thru. Start working there. Like hockey and Lexan. But that was a quality NASCAR race.

  14. RAEckart states a common myth perception in science; there’s a huge difference between the fence didn’t fail this time, and the fences being “safe enough.” We’ve been plotting the path of the cars from photos available on the ‘net, and we’re of the opinion that if the 39 had hit the 99 just 2 feet farther forward along the side (closer to the center of gravity) and the cars had been 1 foot closer to the outside wall, then the dynamics of the event would have been sufficiently different for the 99 car to have made it into the grandstand. My engineering opinion is that this is an unacceptably narrow safety margin; it would also be a legal opinion in a liability trial.
    The essential variable in these calculations is what is the crush rate of a Cup car (force per inch of deformation per second of applied load), good data is available on production cars, relatively little on race cars. We have good numbers on the torsional stiffness of the Cup chassis which gives us a benchmark number for crush rates.

  15. Kirchhoff raises an interesting point about the role of the track geometry in a given event. However, recall that Michigan (MIS) also had a fatal accident. (my apologies here, memory is a little weak) I think in 1998 or 1999, a CART race and a tire from one car wrecking was punted by a second car and the tire plus suspension parts killed two or three people in the very top row of the grandstands.
    The critical issue is the momentum transfer during the impact event (think of the fraction of a second when a bat is in contact with a ball), it is the momentum transfer that determines the exit trajectories of the objects after impact. I don’t think track layout is the critical parameter.
    On track layout, look at the new F1 tracks: a wide track, then a grassy stretch, then 50 or 100 yards of what amounts to a sand trap (gravel), then a tire barrier, then a catch fence, then an open space and then the grandstands, and they’re elevated about 15 or 20 feet to boot. You couldn’t get an F1 car into the stands if you tried; the track is designed that way for that reason. Europeans may be willing to stand next to a road during an WRC race but even they’re not getting close to an F1 track on a curve. They may be willing to take some risk but they’re not crazy.
    They’ve changed, too. Look at vintage pictures of places like the Spa in Belgium, you’d see a nice old couple sitting at an outdoor table of a cafe, having a pint and F1 cars racing by in the street, not 10 feet away and not so much as a hay bale in sight. Times have changed.
    Being more safety aware is a good thing. Now that I have grandchildren, I’m much more in favor of safety measures, e.g., car seats, etc.

  16. I’d like to point out that the forces that caused Matt Kenseth’s car to flip were not aerodynamic, but mechanical. It wasn’t that the air got under his car and launched it- it was the fact that the tires gripped the asphalt and caused the car to turn sharply and roll over, just like an SUV rolling over on a sharp curve.

  17. Hi Nick – I’ve been watching the video and I think you’re right. There’s an interesting calculation for the students to do – would that have happened if the center of gravity were not 2″ higher? What is the torque difference?

  18. That’s a good question, Diandra.

    Thing is, Kenseth’s flip happened in the Busch (excuse me, Nationwide) Series… and they don’t use the new car, so I don’t think the height of the center of gravity in the Busch Series cars are any different than in years past…

  19. The safest thing NASCAR could do at Talladega is make a rule that Carl Edwards cannot ride around at the back of the pack all day, and then drive like a “pissed-off teenager” in the final laps to get to the front. Carl’s ‘strategy’ has caused late-race accidents in the last two races at Talladega… and I suspect it will happen at the next two if he keeps it up. This latest accident was just as avoidable as the last one. To make this about physics… two objects cannot occupy the same space at the same time… this is especially true for 200 mph stock-cars, regardless of their safety equipment. If you want to improve the safety at Talladega, start with the driver.

  20. Just found the blog and wanted to say hi. I am so delighted. Now I wonder how many of us physics professor NASCAR fans are there?

  21. Hi David! I think it’s a surprising number (at least it was suprising to me). In my seminar travels, I find that more physics professors are interested in open wheel racing – not sure why that might be, though. Welcome!

  22. Hi Ray – thanks for stopping by! I was going by the NASCAR rule book. The copy I saw says “Air Deflectors (Roof Flaps)”.

    So are Cowl Flaps not Cowl Flaps either?

  23. would it be possible to slow the cars down using a combination of rev limitation and gearing

  24. I would hazard a guess that in terms of fan safety, Nascar is much safer than major league baseball. It seems like a common occurance for a shattered bat or a fouled ball to sail into the seats. Are there injuries as bad as those at Talladega? Absolutely. Has there been a call for moving the seats back, aluminum bats, softer baseballs, screening….NO. What about hockey pucks sailing over the low screens at 100+ miles per hour.I think Nascar’s fan safety record would hold up rather well to other major sports. How about non-american “football”? How many people have been killed attending those “football” games.I think it is obvioius that Nascar has done an excellent job of protecting both fans and drivers. Can something go wrong, yes. You could get killed driving home from a race.

  25. They need to limit downforce, not horsepower. Make the cars unstable so the drivers have to slow down and drive them. Their skill would separate the packs.

  26. I hope that Mr. Thompson will publish the results of its calculations and simulations. Meanwhile, my 7th grade engineering degree tells me that two are faster than twelve simply because the car is rear of the two hour (COT bumper line up better) able to push the car in front all the way around the track . The twelve are still only bump drafting. You probably argue that this phenomenon has just started last weekend. I maintain that it was there for a day with the COT, the drivers have just discovered.

  27. By contrast look at WRC, people standing along side the road as cars race by at more than 100 mph,.. and sometimes a car goes into the crowd. Europe doesn’t have the liability laws we do and Europeans have the attitude that you’re responsible for your own actions: if you stand next to a road where cars are racing you might get hurt. If you have a problem with that, then stand in some other location.

  28. The car was not starting to lower when the roof flaps deployed; yes Newman’s hit threw the car high into the air, but the flaps were not going to plant it back to the ground. They have not been doing this in all the years they’ve been in use – not with Ward Burton’s flip at Talladega in 1995, not with Mike Skinner’s flip at Charlotte a few weeks after Talladega, and so forth.

    The blunt truth is the roof flap concept doesn’t work.

    The fencing at Talladega did what it was supposed to do – people screaming that it bent ignore that it is designed to bend and then throw the car back. NASCAR in just the last 13 seasons saw fencing shredded at Bristol in 1996 and Daytona in 2000, so to panic and think the sport is one pierced fence away from dying altogether is insanity and inanity.

  29. diandra, it is not possible to lower the banking or introduce a handling aspect to Daytona and Talladega because it degrades the competitiveness of the racing and adds NOTHING to safety. The restrictor plate tracks are far safer than non-plated tracks because the severity of crashes is lower (not only via the eye test but with measurement of Gs in the cars) and there is no reduction in the incidence of multicar wrecks on unplated tracks compared to the plate ovals.

    The reality is that only by restricting horsepower can the speeds be lowered and KEPT DOWN. Bolt-on aerodynamic devices like the roof rail the Nationwide Series presently runs on the plate ovals certainly help, but they’re not nearly enough. The cars can never be slowed by any package whose premise is that drivers can be made to lift – the fact is they will NEVER lift. Only through making it physically impossible to reach a certain speed at open throttle can the c ars be slowed down.

    Restrictor plates have proven over 21 years to be the ONLY solution not only to the two biggest tracks but to all the others, and the racing is so far superior to anything else that it isn’t even an issue. 56 lead changes among 26 drivers is always more importa nt than people screaming about a purely NOTIONAL risk.

  30. Just so you know, The devices on the top of the race car should be refered to as ROOF Spoilers – not Flaps. Spoilers decrease lift and create DRAG. Flaps increase lift.

    I know this is a knit-picky point, but calling them flaps is just as wrong as saying a driver starts on the Outside Pole. There is no such thing.

Leave a Reply

Your email address will not be published.


This site uses Akismet to reduce spam. Learn how your comment data is processed.