I was wondering about some of the assumptions made when modeling the aerodynamics of the F1 cars. As near as i can tell in the CFD models used to model the cars the car is stationary and the air moves over the car. That is the same with wind tunnels, the air is moved over the car. I understand that it is reasonable to assume that the car will behave the same whether moving through the air or remaining stationary when the air moves over the car. However i would really doubt that that it is a perfect assumption. Also i think that it would be a much worse assumption when the air that the car is moving through is not uniform in flow and density. I think that the air around an F1 track during a race is probably not anywhere near uniform. Particularly right behind another car. So my question is since F1 is so competitive why arent the teams modeling the cars moving through the air instead of the other way around. Or do they already do that? I'm particularly interested in the modeling stuff since i am an engineering student and would really love to work for a formula one team someday. I think that more and more complex models are the way of the future.

The short answer is that the only real difference is the rotation of the tires, but yes, it is huge. The top of the tire is moving into the wind twice as fast as the car and that is why the tires make so much drag and wheel skirts are so effects in reducing drag on a sports car.

thanks for the reply. It is an interesting point. i would think tho that with slicks the rotational effects wouldn't be that big although i dont doubt that they are significant. My big point is that in say a wind tunnel the air is moving in laminar flow over the car and in real world the car is moving through mostly turbulent air. I find it hard to believe that the car will behave identically in both situations. I know that they have rolling chassis wind tunnels which would accurately model the effects of the rotating tires so i am more interested in the differences in the other parts of it. Thanks again for the response.

Audiophile426 - 19 November 2009 10:58 PM

My big point is that in say a wind tunnel the air is moving in laminar flow over the car and in real world the car is moving through mostly turbulent air. I find it hard to believe that the car will behave identically in both situations. I know that they have rolling chassis wind tunnels which would accurately model the effects of the rotating tires so i am more interested in the differences in the other parts of it. Thanks again for the response.

The car, of course, would NOT behave identically, however, modeling for what are purely random and unpredictable circumstances would, I would think, be all but impossible.

I am quite certain that the teams do perform aero testing in various degrees of yaw, pitch and roll at varying speeds so they will have some idea how the aero forces change and how those changes affect the chassis.

However, I can't seem to rid myself of the notion that a car hurtling at, say, 160 mph around a race course would be so much more affected by the aero forces imparted as a result of the speed of the car, as opposed to whatever small differences wind direction, speed and air density would make by comparison, that it would make those variables, even if not completely unpredictable and/or unforeseen, all but completely and utterly inconsequential.

PS............rather than trying to model for these random conditions in the tunnel, teams use load cells and pressure sensors out on real race tracks to do some of their aero testing, if for no other reason than to verify what they think they have learned in the lab. The real world is the best 'model' for itself.

Audiophile426 - 19 November 2009 10:58 PM

thanks for the reply. It is an interesting point. i would think tho that with slicks the rotational effects wouldn't be that big although i dont doubt that they are significant. My big point is that in say a wind tunnel the air is moving in laminar flow over the car and in real world the car is moving through mostly turbulent air. I find it hard to believe that the car will behave identically in both situations. I know that they have rolling chassis wind tunnels which would accurately model the effects of the rotating tires so i am more interested in the differences in the other parts of it. Thanks again for the response.

The laminar flow issue is tricky. In theory, the boundary layer is in laminar flow, and that is important to aircraft people working with fully streamlined shapes, but the overall vehicle is well into the turbulent regime (meaning that the wake following the car is fully turbulent). A race car at a meager 100 mph has a Reynolds number of about 40,000 to 50,000 depending on what number you want to pick for the effective diameter. Stable laminar flow ends at 1000 and anything above 2000 is considered fully turbulent. The Reynolds number is a dimensionless number. It is defined as the ratio of inertial force to viscous force acting on the surface. It is calculated a little differently for fluid flow in a confined space like a pipe versus flow around a solid object in a bulk fluid such as a race car, but the concept is the same.

The value and accuracy of wind tunnel and Computational Fluid Dynamics data are as "accurate" as you want to make them. Over the past 50 years or so, the results have gotten better and more accurate, but it is still possible to get "robbed" by modeling data.

Here are a couple of well-known examples where this happened: One was the "Porpoising Effect" in 1908s "Ground Effects" cars tht didn't take the springs into account, and the Mariner, a 12-Meter yacht in the America's Cup. The America's Cup "Meter Rule" measures several different points on the hull, including the waterline, and combines it all to come out with a single number, which in the America's Cup boats was twelve meters.

In 1974, designer Britton Chance thought he could get around the waterline length restriction by making the underbody (under water portion of the hull) end in a flat plate, like a speed boat. The idea was, this plate would cause a cavity like the one behind a speedboat, effectively lengthening the waterline length (which determines maximum speed by the length of the wave from front to rear), thus cheating the rule. Put simply, it didn't work. The Mariner never got the speed she needed for the "cavitation effect" and would up slow and sluggish to handle, and sporting a "rooster tail" behind her.

Now back to Formula One.

In the beginnnig, wind tunnel tests were carried out using small models of 1/8th to 1/6th scale or so, with the tyres fixed and resting on a solid surface. In the mid-70s, the Lotus team started testing on a "moving road plane," which is essentially a conveyor belt under the chassis, with the wheels rotating. This simulated the passage of the road under the car.

The results were more accurate than the fixed-wheels/no-moving-road wind tunnels. So what, you say, thte car is still standing still. True, but as Albert Einstein said, all motion is relative to the observer. In other words, air doesn't care if it's being whooshed past a model or the model is moving through the air, the same displacement happens either way.

The next advancement came in the size of the models. From the 1/8th scale, they went to 1/4 scale, 1/2 scale and possibly 2/3 scale. Teams sometimes even test full-scale cars in the wind tunnel, if they have one large enough.

The reason for this is the "scale factor." As you scale things up, the aerodyamic forces on the model increase exponentially. That includes the "unexpected" forces that can trip up a hypothetically good design when it is built full-scale. The closer you get to full size, the more apparent these unexpected forces become and the more accurate the tests are.

In the end, the aim of wind tunnel testing is not so much accurate numbers as a general idea of how the car will perform, and the percent of improvement a change makes over the baseline model.

Now about aerodymanic drag.

First, the explaniation of a word: "Drag Coefficient." This is the comparison of a car's form drag with that of a flat plate of the same frontal area held perpendicular to the wind flow. In truth, a flat plate's drag coefficient is closer to 1.28 or so, thanks to wake turbulence, but the idea is still the same. "The Broad Side Of A Barn" would have a drag coefficient, written symbolically as Cx or Cd of about 1.0. In case you're curious, a circular parachute has a drag coefficient of about 1.5.

I'm not privy to the drag coefficient numbers of a Formula One car, but I'm sure they're enormous. Indy cars are said to have drag coefficients of somewhere between 0.75 on a high speed oval (small wings trimmed flat) and 1.1 or so for high-downforce road racing configuration. That's an awful lot of drag, requiring an awful lot of horsepower to overcome at high speeds.

The drag coefficient of a "production based" race car (Think NASCAR or the Deutche Tourenrad Meisterschaft) is probably half that or less, though the total drag is somewhat more than half. This is because the bodywork covers the wheels, suspension and other high-drag parts of the car, making the airflow considerably smoother, in spite of having twice the frontal area or more.

The formula car's problem is that it has a lot of things sticking out in the air -- wheels, air ducts, suspension members, wings, air scoop, the driver's noggin, etc., all of which cause drag. Worse still, each piece is more or less separate, rather than integrated into the whole, as it would be on a "production-based" car or a sports-racer like an ALMS car.

Speaking of which... back in 1985 Road & Track decided to test the top speeds of various cars at Ohio's Highway Safety Research Centre. The list included a couple of race cars. One was Al Holbert's Porsche 962 and the other was Tim Richmond's Cheverolet Monte Carlo. You might expect the Porsche to have trounced the stocker because it was lower and "more aerodynamic," but that's not the case. Even with the LeMans under tray, the best the 962 could do was about 220 mph or so. The stocker turned close to 240 mph.

Why the difference? As I said, the stocker had no wings, no large radiator air scoops (which were in the Porsche's doors), no undertray, smaller tyres and a number of other things. It was simply a smoother shape. And when the crew taped up the air intakes, the car went 5 - 10 mph faster.

Mind you, it didn't generate the Porsche's enormous downforce, which would have made it faster in "real world racing," but R&T;weren't testing that. They just wanted to know, "how fast will it go."

Thanks for all the info guys. It is a very interesting topic and could be discussed indefinitely im sure. I agree that many things would be very similar between the car moving through the air and the air moving over the car. Especially in cases where rolling chassis wind tunnels are used. However i wonder what kind of models the teams use in their simulations. Im sure the differences are small but i gotta imagine that they are significant when you consider the fact that the whole field was often separated by only a second.

Also i dont doubt that F1 cars have a large coefficient of drag. Tho id bet that the Force India cars have much lower drag than most. It would be very interesting to see that data.

Wilmywood had an interesting point about the load cells on the cars. That seems like a great idea but wont help teams like USF1 during its development. They sound like they are heavily relying on their computer simulations to develop their car. Also with the limitations on testing the simulations will become much more important. Which brings me to another point. A few people mentioned that modeling is essentially arbitrary and implied that it was generally used for comparison type analysis. Which is probably true in a lot of cases and im sure it is valuable. However for a new team that is relying on computer simulations for most of the development i would think that comparison type analysis would be much less valuable. Therefore i would think that very accurate simulations would be absolutely essential. It will be very interesting to see how USF1 does next year.

Greywolf, i appreciate the lengthy response and the many examples. I would think that computer modeling is growing and improving incredibly fast so im sure that the modeling used now has very little in common with that used in the past. Its a great point tho that simulations can cause problems if trusted too much. I think the performance of USF1 next year will give us a bit of an idea of how good or "accurate" their simulations are.

Thanks again guys for all the info.

GreyWolf74 - 20 November 2009 09:51 AM

The next advancement came in the size of the models. From the 1/8th scale, they went to 1/4 scale, 1/2 scale and possibly 2/3 scale. Teams sometimes even test full-scale cars in the wind tunnel, if they have one large enough.

The reason for this is the "scale factor." As you scale things up, the aerodyamic forces on the model increase exponentially. That includes the "unexpected" forces that can trip up a hypothetically good design when it is built full-scale. The closer you get to full size, the more apparent these unexpected forces become and the more accurate the tests are.

Also a big factor in scale model testing is this: regardless of the scale of the model, the size of the molecules making up the atmosphere remains full scale, as does air density unless artificially altered. Makes a difference.

Audiophile426 - 19 November 2009 08:48 PM

I was wondering about some of the assumptions made when modeling the aerodynamics of the F1 cars. As near as i can tell in the CFD models used to model the cars the car is stationary and the air moves over the car. That is the same with wind tunnels, the air is moved over the car. I understand that it is reasonable to assume that the car will behave the same whether moving through the air or remaining stationary when the air moves over the car. However i would really doubt that that it is a perfect assumption. Also i think that it would be a much worse assumption when the air that the car is moving through is not uniform in flow and density. I think that the air around an F1 track during a race is probably not anywhere near uniform. Particularly right behind another car. So my question is since F1 is so competitive why arent the teams modeling the cars moving through the air instead of the other way around. Or do they already do that? I'm particularly interested in the modeling stuff since i am an engineering student and would really love to work for a formula one team someday. I think that more and more complex models are the way of the future.

CFD programs have improved a lot in the last several years. Some now include rolling ground/wheel models, yaw and steering movements, suspension movements and allow to put a car in front of, to the side in the modeling.
Many of CFD properties and "features" are also starting to show up in the data driven Simulation programs, where actual track data is run through the program and analyzed.
Also included is 7 poster simulations.

Thus shortening the valuable track time needed to full fill the "real life" end of the engineering.

Thats pretty interesting. Do you know if the teams write their simulator programs in-house or do they hire that out? and if they do purchase their simulators from third party companies do some of the teams use the same simulators? also what is a 7 poster simulation?

willmywood also has a good point about the scale.

1 - The difference between a fluid moving around a body and a body moving through a fluid.

The difference in the equations is none… It is a frame of reference issue which isn’t accounted for in the equations… As far as the equations are concerned it is a velocity differential between the two and has no concern on which is not moving relative to the rest of the world. Your doubt needs to be removed because if you don’t accept that you can move a constant from one side of an equation to the other through simple algebra then you need to start accepting some other assumptions.

2 – Analysis should be done on the cars on non uniform airflow conditions.

The assumption of non-uniform flow and or inconsistent density having an impact on the effectiveness of an airfoil analysis is true. However because of other knowledge about fluid mechanics it is taken into account such as Froude number, Mach number and about 300 other dimensionless parameters used in computation fluid dynamics they don't worry about it because the software does it for them. They are not simply trying to figure out the exact amount of downforce or make a single airfoil work absolutely perfectly in a single state… There are loads of compromises all over the place. To best analyze this they check the flow of the system in each of what is called homeostatic conditions AKA unchanging conditions because the flow of air changes so quickly that to try to take advantage of the dynamics of an ever changing turbulent system is not nearly as important as making sure the airfoils work the way they want them to in relatively consistent conditions such as going around corners, braking and accelerating. The rate of change in the direction of the air flow is nowhere as close to how quickly the air reacts to those changes. As far as the density not being the same… that is what drafting is and you had better believe that they do those studies… In 2000 a car 5 car lengths behind the leader reduced the downforce by the lead car by up to 15% and went up from there the closer it got… Rumor was back in the day that Ferrari had different noses for who they thought was going to be ahead of them in the race because every car reacted differently and they had it all modeled out and worked up in the wind tunnels. It is not uniform but it is highly predictable and analyzable. The CFD models have no problem analyzing the system with two cars in it… They have no problem analyzing the airflow of every particle of air in an entire race on all the cars at all times… It is just a matter of how much money you want to throw at the super computers and how long you want to wait.

3 – I would suggest getting a student version of Solid works… It comes with flow works which will let you do everything you are dreaming about… 10 years ago when I started on it they were just starting the CFD stuff and analyzing the flow through a pipe took a few hours but now computers are much much MUCH faster and the software is comparatively easy to understand. I'm not positive but I think that is what USGP is using. You can get it through your student book store and compared to how much it will cost to get it in the real world it is a steal for the student edition. less than books this semester. As far as more and more complex models being the way of the future… You are about a decade behind. They are already modeling stuff at the microscopic scale and the programmers are getting better at making software that goes even further than we can even begin to think about. Software is the key but management of the people who know how to use it is the door.

4 – Tire rotation is HUGE!!! The air trapped in front of the tire causes by some estimates upwards of 30% to 40% of the drag on F1 car even though they only account for 15% of the frontal area of the vehicle. And that is probably on the low side by a pretty big margin now that the groves are no longer on the tires. The groves allowed air to travel through the tread of the tire at high speed instead of being forced around the outside of the contact patch.

5 – Laminar vs. turbulent airflow… Wilmywood has a lot of it right… lots of testing in different angles of attack pitch yaw and such give them a highly educated guess of what is going to happen under certain circumstances but it does not cover all… AND even Friday practice does not cover all as the McLaren diffuser stalling midseason shows. But it isn't just turbulent vs. laminar flow... At above .1 mach you can start to deal with fully developed viscous flow which is a whole other animal all together in the world of airfoils. The best model IS reality but the FIA won’t let them do that anymore (at least very much of it) and when you compare CFD to building 1/2 scale models to put into a multi-million dollar wind tunnel with pitot tubes and smoke machines for every evolution of design is compared to spending a few minutes and a few mouse clicks on a $10,000 computer with $30,000 of software with an engineer getting paid $80 an hour it makes a lot of sense to hire more CAD jockeys until you think you’ve got the final version and you want to show off for the camera crews. Wind tunnels are as close as you can get to the real proving grounds but CFD and FEA is the way of the future of development.

6 - as far as the molecules staying the same size and the models getting smaller so it having an effect on the experiment? Ua... no. I'm gonna pull this one from my Fluid Mechanics text book as a straight quote from Langhaar:

The motions of two systems are kinematically similar if homologous particles lie at homologous points a homologous times.

If you scale the prototype then you can scale other properties accordingly to get relevant and highly accurate data independent of the size of air molecules. It isn’t like they are throwing tennis balls and beach balls at the car