Newey doubts CFD-only designs

[horse]

Story on autosport: http://www.autosport.com/news/report.php/id/81372

"I think CFD is a very powerful tool, there is no doubt about it, and it is another way of simulating the real environment," Newey said at the Red Bull launch when asked for his opinion on the Virgin Racing car. "A wind tunnel is a simulation of the real world.

"CFD is an electronic simulation of a real environment, but it still has pitfalls - not least that every single run in CFD for a given attitude of the car, or ride height, or whatever it might be, is a discreet run. Whereas in the wind tunnel, what we call a normal run, will have 20 or more data points in it. In other words, that is equivalent to 20 runs in the CFD.

"That is a limitation of size really, so your CFD cluster has to be that much bigger to do that many runs. And there are some areas that CFD physically doesn't capture as well as a wind tunnel - like basic aerodynamic properties.

"So how well it turns out, we shall see. It is a different route, and my personal belief is that you still need to combine the two at the moment. But maybe their car will go very well and I will have to revise my opinion."

I thinks Newey shares the general consensus here that a CFD only approach is highly challenging, although he does not question the accuracy of the tools, just the amount of time required to test in CFD versus wind tunnels.

[Jersey Tom]

Well I'd have to agree with him. I don't think CFD-only designs will be particularly fantastic, but when you're strapped for cash.. your options are limited.

[Giblet]

If Moore's law keeps managing to hold true, then the time it takes to do 20 discreet runs will shrink year on year.

The wind tunnel will always be a wind tunnel, and it's capacity to do work will never increase.

CFD will overtake the usefulness of windtunnels in development someday, and I don't think it's as far off as many think.

[Jersey Tom]

There's a lot more to the limitations of CFD than just speed.

[Belatti]

What do you think about the USF1 claims in the RCE article that the company thats suposedly doing the CFD for them has a mesh size 10 times smaller than current systems capabilities... or something like that, I dont have the mag right now, my number of zooms is out

I think that CFD results are always stationary. The wake behind an object can change a lot if that given object is standing still or its oscillating at a frequency that can also be variable. I believe thats the current and bigger limitation in CFD, and to consider that you have to have the processing speed because a small and sensitive mesh is also required.

A ´perfect´ CFD would require not only the fluid moveing but also the analysed object vibration and movement at the same time.

[Giblet]

I could be very wrong, but I feel year on year, CFD will improve as a tool to use, and may catch people off guard in a few years with it's capabilities.

I would love to see an 'aero constructors' series where the chassis are all the same, GP2 or what have you, but the teams can all play with the body and wing shape, but using only CFD.

That might help push it forward even more.

A friend asked me once "If aeronautics is a science, why don't all the cars look exactly the same?".

Was a good question really, and my answer was that it is still as much black art as it is science.

[autogyro]

I have a feeling that Todt will soon put a stop to costly aerodynamic development.
The regs will set the aero to one basic design for the whole season in F1.
Then of course there could well be a series just for aero.
That might open it up to proper new ideas instead of variations on a well used theme.

[segedunum]

What do you think about the USF1 claims in the RCE article that the company thats suposedly doing the CFD for them has a mesh size 10 times smaller than current systems capabilities...

USF1 were supposed to have Windshear at their disposal........and it's all gone rather quiet. Now it's supposedly CFD only?

[tk421]

i think there's a quote from windsor in the usf1 thread saying that once their (CFD) car is complete, they'll go to windshear to have a couple of tests.
so i guess "cfd only" is true in terms of design, but they'll bolster that with a windtunnel test or two...not certain obviously!

[segedunum]

The problem is not computing power. The problem is calibrating CFD to reflect the real world to a high level of accuracy. If you don't have that accuracy, and more importantly, you cannot verify it, then you're going to be in a lot of trouble. I suspect Virgin might well have that problem. I don't agree with Nick Wirth at all. That CFD precision just isn't there in the tools available right now unless you have other real world tools to verify against. Teams have even had trouble verifying what's happening at their wind tunnels against what they actually see on-track, so how CFD-only helps I don't know.

Another problem is that too many use CFD as a toy. It's all to easy to play about with several designs and test them in CFD and 'shoot in the dark', as it were, rather than building the car with a central and defined philosophy in mind. I've certainly got the impression that's what Williams have been doing for a while.

[apexdc]

I have been following the development of Windshear's tunnel from the inception and am baffled why USF1 would make such a compete commitment to CFD for design without even the cross check of testing a large scale model in the tunnel. It might be a different story for a team that was physically located so far from any available tunnel, but to have one of the best in your backyard and not use it seems very risky.

Certainly the cost of the model is not insignificant, but the real cost in time and money of finding significant flaws at the virtual end of the design process in almost incalculable, especially to a new team.

Perhaps there is something about the true relationship between Windshear and USF1 that is affecting this decision???

[Mystery Steve]

Perhaps there is something about the true relationship between Windshear and USF1 that is affecting this decision???

The problem with USF1 is having (or not having) a car to test. I'm sure once they have a car built they will head over to Windshear and take some measurements. For a first design, I don't think they'll be any worse for the wear being CFD-exclusive during the design process. I'm not the foremost expert on CFD, but like with any engineering tool, if you truly understand how it works and what it's strengths and limitations are then you can at least make educated evaluations of different designs. This is true of not only aerodynamics, but also strength/structural mechanics, suspension design, etc.

I know I've said this before, but I'll say it again. Computer modeling simulations, laboratory/wind tunnel testing, and track testing all have drawbacks to them. However, they all offer advantages that make up for some of the others' deficiencies, and in that sense the different methods complement each other. The whole really is greater than the sum of the parts, and the teams that are most productive in utilizing the balance of resources will bring a competitive car to the track.

[horse]

Mike Gascoyne (reliably) weighs in on the side of the experimentalists...

Speaking about the CFD tactic, Gascoyne said: "I think it is an integral part, but it is not a complete part."

When asked about his reaction to Newey's comments, Gascoyne said: "Well, I think he is one of the best aerodynamicists in F1. I used to be an aerodynamicist, and my PhD was in CFD, and I think I would kind of agree with him.

"You look at BMW when Albert II was announced as one of the world's biggest supercomputers dedicated just to their CFD. If you look at Enstone, they built their environmentally-friendly CFD centre with a huge computing resource. I don't think these guys are idiots, and they also have wind tunnels.

"I know Bob Bell at Renault very well, he is a clever guy who gave me my first job in F1 and he is a trained aerodynamicist - and I think he thinks you need a wind tunnel. I think CFD is a very exciting technology and it is advancing, but is it an absolute? I don't think there are many people who think it is."

[RacingManiac]

I am pretty sure Virgin was going to use the tunnel at least once(or just once, I think it was in a press release), just as stated to calibrate the model. What they are foregoing is the repeated model building, and iterative running in the tunnel, and thats really the expensive part. I am sure if the economy of the team allows, they would not be doing "CFD-only". But such is the circumstances of the operation. Even when HPD was backing their Acura program they still only runs on peanuts compare to Audi/Peugeot....

I am not sure how big of a budget Lotus will have(they do have the backing of the oil-producing Malaysian gov.), while I am pretty sure Virgin and the still born Campos and USF1 are on the budget cap level...

[Terrible3]

In Wirths defense he has already produce two LMP cars that were both very successful. Sure the ARX-02a really had no competitors but Wirths ARX-01b is in my opinion is outstanding. The fact that they were able to take an existing chassis from courage and an engine from Honda and made it competitive against the RS spyder is impressive. The RS spyder has a bespoke chassis and engine allowing the gearbox and engine to be fully stressed members. The ARX-01b was also much faster than the lola chassis fitted with the same engine. The ARX-01b was completely developed via CFD.