CFD - Computational Fluid Dynamics, Motorsport, Formula 1

[miqi23]

Sure, I can try and help..

[mahesh248f1]

YEPPPPP !!!

[Just_a_fan]

Renault built their own code, one of the guys did it went through QUB a few years before me.

A real, real smart man.


I think Renault and Boeing have a thing going now regarding computational methods.


[I had thought Renault and Airbus were involved in CFD in the past... ]

I think Lamborghini and Boeing share some computational systems and methodologies for advanced composites.

[shelly]

As far as I know

Reanault use starccm+ (commercial code - see cd adapco sponsor label on the crash structure). I think they have some inhouse software for optimisation on cfd.

[Th0R]

So basically it's more important to be able to write your own cfd code than to know some comercial software?

[shelly]

It depends on the role you want to fill. But profinecy with a commercial code is usually a prerequsite, as then you have to program extensions or customisations of that code

[Th0R]

Well, the comercial tools i know don't give you that much possibillity to change fundamental parts of the code. For example starCCM+ has its userfunctions but no way the exchange fundamental parts like the solvers. On the other hand programs like openFoam consume a lot of time to set up a simulation, but of course here you can do what you want with your code.

And if you think about it there are not many aspects which would justify developing your own code (because commercial tools have them covered pretty well)

[shelly]

Agree with you. Most of the work is done via commerial codes

[Jersey Tom]

Most important to just be a sharp engineer, know your stuff and be able to react to things.

[Th0R]

Most important to just be a sharp engineer, know your stuff and be able to react to things.

But that's more a personal skill than a technical one.... And you know you dont need to be an engineer to do cfd ...

[kilcoo316]

And you know you dont need to be an engineer to do cfd ...

Any clown can run the software.


However, it takes quite a skilled clown* to run the software properly, ensure they are running the software correctly and also interpret the results correctly.



*such clowns would typically need mathematics, physics or appropriate engineering training in the circus before being sufficiently skilled to undertake any task to any useful end.

[Th0R]

I was just aiming for the different backgrounds. I myself have more an applied mathematics / computer science background and would never dare to call me an engineer (but i'm planing to become one), but i can mostly tell you what's happening inside the cfd code. I know nothing about constructing things or materials or even producing things.

The thing is you always need a team to build such complex projects like a formula one car. Even Newey didn't design the RBRs by himself. And there are a lot of people in a team which aren't engineers.


And i love colorful pictures

[BalazsSzabo]

I am very interested in giving a try to CFD analyses. Although I didn't have any experience with it, I have been learning mathematics, can you give me some pieces of advice how to start it and which program to use? Thank you!

[KeithYoung]

I met THE top guy of CFD of Ferrari-F1 and I can tell you he isn't a monkey running code thrown at him. I asked him how he landed what in my opinion is just about the most cutting edge job that exists. He learned the ins and outs of aerodynamics and numerical methods. He didn't learn how to just use them, he completely understood them.

If you guys want to really know CFD, get Matlab or some equivalent and start writing simplified 2d solvers, even 1D. Start eliminating simplifying assumptions (If you've seen my website, you've seen how many simplifying newb like assumptions I make) used in your model. Work on increasing convergence, estimating boundary conditions, learn the nitty gritty numerical methods and in his words "Keep fighting"

I might have to just start a thread for that, as this may be detracting from the oringinal topic which was Russ and I and a few others just running code we didn't really understand.

By the way, if anyone knows how to get ahold of Russ please let me know. Thanks, and hope this post helps.

[Just_a_fan]

I used CFD in my Fire Engineering Masters degree. FDS by NIST if anyone is interested. Easy to use? Yes, prety much. Easy to get right? Oh wow, whole different question.

I spent several months and countless hours modelling some very simple cuboidal "room" fires and found that I was making assumptions that compounded errors and gave weird results. Easy to do and not obvious until you get the results some days later. One run took 47 days on a quad core PC with 8GB of RAM. And that was at a grid size of 50mm. So nowhere near fine enough for F1 aero work.

I spent a week in lectures with a Russian lecturer who could do the maths in his head whilst asleep! Anyone can chuck a pretty model at a basic CFD package and get some lovely looking results. But the chances of those results being meaningful? You have to ask? To use is one thing. To truly understand is another issue altogether.

Oh, and FDS uses LES rather than RANS (which a lot of CFD packages use) so can provide more useful turbulence modelling in the right hands - FDS is limited to cuboidal construction though - no curves. Even then the computational demands are huge.

The simple fact is - there is no simple CFD. Just CFD with varying levels of assumptions where accuracy and assumptions are usually inversely proportional...