CFD - Computational Fluid Dynamics, Motorsport, Formula 1

[AeroGT3]

Getting one or two results on an individual study does not circumvent the maths behind it.

It's more than one or two results, and none of them "circumvent" the math. That is totally contradictory to what you just said. If CFD IS math, then how can a CFD case circumvent the math that created it You can calculate all day on your calculator, but if Fluent calculates no real error with the coarser mesh, who is right? Does your calculator over ride what an actual Fluent simulation says?

Oh, and yes. I've seen massive differences in simulation quality going from wall function to low-Re modelling.

And even from going from a disc of 1.25 to 1.1.

A disc. Great. What changes have you seen on an F1 car? I posted my result. That was an F1 front wing. And I have the Cp on the surface and I have the Cl and Cd.

Since you love math, let me talk about that. I did a richardson extrapolation on all three of those meshes, and the middle mesh was basically converged with Cp on 2 random surface points AND on Cl and Cd overall. Let me pull an excerpt from my report:

Perhaps even more interesting is their level of accuracy with respect to the absolute converged values provided by the Richardson extrapolation. The results for the coarse, baseline, and fine grids are each within 1.79, 0.28, and 0.113 percent of the continuum (zero cell size/infinite cell).

Let me explain that. The 30M cell mesh I am advocating was shown in a grid convergence study to be 0.28% off of the 4th order extrapolated value for an "infinite" cell mesh. Your glorious 120M cell fine mesh was 0.113% off the continuum mesh. Is that really worth it? Is that necessary? Is that even within the error bar of the wind tunnel, or the level of accuracy afforded by k-e or k-w? What would be the points of gaining 0.1% accuracy with your mesh when the SOLVER or turbulence models are only 2% accurate?

Mate, you are TALKING about results on a disc and I am SHOWING results on F1 geometries. Which is more relevant to F1 solutions?

Using a case which couldn't be tested in the wind tunnel, did not use symmetry, and was a FORENSIC TROUBLESHOOTING application of CFD and NOT a DESIGN PROBLEM is not valid. Forensic trouble shooting is nowhere near the majority of teams' CFD cases. 100M cells is not that common.

Even with that most rudimentary method of comparison (the pretty picture that tells very little) there are differences in your pressure zones.

You produce some graphs of pressure along various Y locations (plotting pressure against spanwise location) and you'll see proper differences.

That's why I did the extrapolation and checked the convergence of Cp, Cl, and Cd

Can you name the papers?

If not, just give me the conference name and location.

AIAA, Reno, January 2008 I believe.

SO 30 million is ok as long as you only model the most rudimentary aspects of the car and do it in conditions which are of little actual use?

You can model yaw, etc. with 50-60M cells . . . still no need for "well over 100M" on design cases. That simple is a waste.

[AeroGT3]

Hey Robert, haven't talked to you in a while. Still in SLO?

Yes, but only for 3 weeks. Then it's off the to EU!

[connollyg]

Slim, maybe you missed my post earlier, are you looking at porting CFdesign onto GP-GPUs?

kilcoo, do you see the use of GP-GPU's as mainstream or a transient fix until the current bottlenecks in AMD's (Clock-speed) and Intel's (FSB) offerings get fixed?

Do you think that going down the FPGA route would be better (assuming that somebody bites the bullet and ports their code)?

G

[connollyg]

More cells => more nodes for the same turnaround => more parallel licenses => more $$. There is that possible incentive we need to beware of.

Well that is point i was trying to make earlier, if 30M cells is all that is required, why are all the teams rushing to buy bigger and bigger machines, in terms of $$$ todays 2K core machine is probably the same cost as the 600 Core machine that teams bought 2-3 years ago and that is including the licences.

G

Easy, speed. Bigger machines turn around simulations faster.

Well of course thats true, but its this 30M v 100M cells thing, if 30M is sufficent and you can do a suitable turnaround (say <24 hours) with a 600 core machine, why would the teams be spending huge amounts of money on 4K machines and i know you can say well its to run multiple jobs, but i know that at least one team wants a big machine to run a single job on and still give reasonable turnaround times!

G

[connollyg]

[quote="AeroGT3]
You can model yaw, etc. with 50-60M cells . . . still no need for "well over 100M" on design cases. That simple is a waste.[/quote]

As i said in an earlier post i understand that transients are quite difficult to model, and it seems that this is the area that is causing the teams most trouble to fix.

You guys have already squabbled about tyre-wake and yaw, don't remember cross-winds or tyre dynamics being mentioned (there was an article snippet in Autosport this year discussing McLaren's advanced tyre modelling and how its meant they have coped well with the change to Bridgestones).

G

[slimjim8201]

Easy, speed. Bigger machines turn around simulations faster.

Slim, maybe you missed my post earlier, are you looking at porting CFdesign onto GP-GPUs?

I am not aware of any push towards GP-GPUs.

[dnomdec]

Is the advanced tyre modeling used in CFD? or just ADAMS?

[tamil.turbos]

Hi, Can anyone get me the profile of the wings of a F1 car.

[miqi23]

Make one of your own mate

[tamil.turbos]

I should know the dimensions; in particular the length, breadth and other stuff, like angle to which the wing is twisted.

If you know some dimensions please do post it.

Thanks for your reply.

[Tifoso]

Hi all,

during this school year (my last one before I enter the University) I had a subject called D.A.P.O, which is the equivalent to C.A.D. It was the first time that most of us experienced with AutoCad even though I had already done "something" with Catia and AutoCad.

Anyway, in March more or less we had to start a 3D project with AutoCad 07, and as I am crazy motorsport fan I convinced a friend to do this project with me. After some months of hard work, problems, more or problems and... yes, some joy, we finished this project which is just a resemblance of a single-seater car. Some dimensions are taken from the FIA rulebook, others from our eyes and imagination.

Here are some vids that I'd like to share with this community, I hope that you like it:

http://img204.imageshack.us/img204/6668/cp1aq8.jpg


http://www.youtube.com/watch?v=w7qHRJFpbqY
http://youtube.com/watch?v=DUyhlS2D-po
http://youtube.com/watch?v=kRo2lr2TLFo
http://youtube.com/watch?v=W9w1eplkxh0

[PlatinumZealot]

Pretty good =D> .. Very complete.. I can't tell the last time i completed a car model. Maybe you can test it in Solidworks.

[TudorMiron]

AeroGT3 and other experts of the field. I'd like to ask what would be considered as generally good enough number of cells for simulating GT type car - straight line, yaw, flowt hrough and out of front diffuser (so internal aero to some extent) etc. Number of cells within BL, height of BL? I haven't touched CFD for about 3 years now. What's the currently preffered turb. models this days?
Most definately ANSYS CFD will be used.

Thank You
Ted

[tomislavp4]

Hej guys, I have a question Can someone explain to me step by step (d´oh!) how the he.. du you calculate the Cd using CFD, say in COSMOS FloWorks??? Thanks!

[AeroGT3]

AeroGT3 and other experts of the field. I'd like to ask what would be considered as generally good enough number of cells for simulating GT type car - straight line, yaw, flowt hrough and out of front diffuser (so internal aero to some extent) etc. Number of cells within BL, height of BL? I haven't touched CFD for about 3 years now. What's the currently preffered turb. models this days?
Most definately ANSYS CFD will be used.

Thank You
Ted

Probably 30-60M cells (yaw, no yaw) if you really wanted to get everything dead nuts accurate (drag in particular.) If you dont mind the numbers being a bit off and just want to tell differences between models, maybe 10M? Its hard to say without knowing the level of detail present in the CAD. Height of the BL depends on # of cells and the initial height, which is determined by the y+ your aiming for. The number of layers and growth rate are put in place to get you a smooth volume transition with your tet cells.