Using CFD in Formula 1

[mikhak]

I just wanted to share some things about a seminar I attended a few days ago given by Dr. Luciano Mariella, head of the CFD group for the Ferrari F1 team. An interesting lecture I must say but I expected a bit more....but then I suppose they cant be giving away their secrets!!

Just some points from it:

As an example he used the simulation they ran for the FIA to investigate the effects of following in another car's wake. The computational box they used was 2 carlengths in front and 3 car lengths behind. They also did an actual experiment on the Monza track to investigate the wake. I think this was what led to the CDG wing idea. The CFD and track experiment really didn't tie up well quantitively because different speeds were used and the CFD used a constant ride height and constant gap between the cars whereas the track experiment had both varying. Seemed like a kind of half-assed effort on the CFD group to investigate this. I'm sure they had much more important stuff to be doing. But the results were loss in downforce obviously, cant remember the figure and a shift in aero balance to the front of the car. So the rear wing lost more downforce than the front.

For any of you interested in the CFD set-up well just a little bit of info. They ran one simulation of a car on it's own, no influence of wake or anything. They then saved the flow field behind this car at a certain distance and then applied these conditions at the inlet and ran another simulation.

In general, they simulate half the car where possible, eg. straight line performance. And then for yaw both sides need to be modelled obviously. He said that about 2/3 of the work focuses on cornering with 1/3 on straight line performance. These were just numbers he picked from his head so they're not exact or anything but just gives an idea.

An interesting point also was that the windtunnels are not used to answer the question 'why?'. Basically no design enters the windtunnel now without first being designed and “suboptimised” with CFD. The windtunnels guys have a certain number of aerodynamic parts they try and look at the forces and just answer the question of which one works better. The better ones are kept and the ones that dont provide a gain are discarded. If they want to know why the bad ones were bad then it's back to CFD to understand why. Makes sense I suppose as wind tunnels are now limited to no greater than 60% scale models and max speed of 50m/s.

Also about the windtunnel, he said they done some flow-visualisation when they first opened the windtunnel but after that practically no flow-vis.

He was quite annoyed at the double deck diffuser saga, even using the term "bullshit" i think to describe the whole thing. His team has obviously had to work like crazy to get this new diffuser for this weekend. Seems to have done a good job.

Sorry if this has bored you to tears.

[Gecko]

Thank you very much for the post, very insightful!

[tahadar]

great post, thanks for taking the time to sum up the experience! might I ask where the seminar was held?

[Jersey Tom]

The CFD and track experiment really didn't tie up well quantitively because different speeds were used and the CFD used a constant ride height and constant gap between the cars whereas the track experiment had both varying. Seemed like a kind of half-assed effort on the CFD group to investigate this.

Uhh.. there are some practical limitations in CFD. Making a "real time" mesh with two full cars, that varies distance and ride heights appropriately from suspension data? Good luck...

Interesting notes though.

[mikhak]

@tahadar The seminar was in KTH, Stockholm. I think they have a vehicle dynamics module and this Dr.Luciano Mariella gives a once off talk every year. I'm not taking any classes so I don't know for sure.

@Jersey Tom Yes I understand the limitations of simulating the exact track conditions. But the least they could have done was look at the effect of varying distances between the cars. This would've been relatively simple to do, just take the wake in a different position and apply to the inlet. He showed a nice graph with loss in downforce v distance from the car in front from the track experiment and then a single point on the same graph for the CFD data. So there was no trend to see what the CFD would predict the loss in downforce.

But then again this CFD data would be pretty useless unless they were going to take it further, there's no point in doing CFD just for the sake of it. The track test was relatively simple to do, showed what they wanted to know and is more like real conditions than the CFD could ever have been. I guess it's the only bit of CFD stuff he would show as it was a test for the FIA......all the rest is competitive information Ferrari will keep to themselves.

[ds2taieb]

Hi

Very interesting conclusions for CFD.

I'm a phD student in CFD. Now with the computation power available on different supercomputers and the improvements of the different methods (Large Eddy Simulation, moving mesh, and other), it can be possible to simulate the flow over the cars with all these phenomenon as overtaking and the height of ride. However it is possible with few academic code. I don't know how an automotive industrial (even a F1 team) can be interested with it.

@mikhad: do you have any support (slide, report) of this seminar? If yes, I'm very interesting in.

regards
David