Virgin Racing VR-01 Cosworth

[SZ]

I agree SZ, however, could this have something to do with them trusting CFD all the way?
Wonder how much, if any, windtunnel calibration they have bothered with?

Performance is one thing, but getting a car together - and the money involved to do so - big challenge, big, big challenge. Two new teams failing on this miserably.

In time, the performance may come. They may yet surprise us all. Anyone got any rear shots?

[xpensive]

I was thinking more in terms of speed to complete the design, would you save time by trusting the CFD like Virgin?

[SZ]

Whether they had tunnel testing or not wouldn't have added to their development time, it'd significantly add to the pace of aero performance development as you can bang through a number of mods in the tunnel far quicker than you can in CFD. So however quick the car is (and whatever NW says) they'd have had a quicker car had they a tunnel too. If it's quick, all it'll mean is that his team can use CFD better than other teams. Not that CFD replaces the wind tunnel. Sure, having a tunnel program takes more budget, they'd have needed tunnel time in lieu of a suitable facility on site (very few that can be used for F1 development free now - most of those companies only work with a single client in each race category anyway), a whole model making team, etc. All this adds to budget though, not time.

The heads of department deciding which part goes on the car in the aero development program essentially have more options to chose from at the end of the day when your aero program is broader. You still only need enough parts to build a complete car (not all the options) ready to launch it. Whether they're fast or not doesn't matter too much, so long as you've a decent upgrade pre race 1, you can go testing meaningfully.

But does CFD, or any other CAE tool help? I'd argue that aero aside, the key roadblocks to getting a car design past the FIA so that you can assemble it are all done with CAE tools anyway. Budget might dictate how they build a few components - where their FEA runs out and they start destructively testing, better funded teams may have gone for a few extra ply of CF where a few more tests might have revealed a more on-the-limit solution... but that's strictly a function of budget. The key work takes the same tools and the same amount of time in any team.

Aero doesn't delay launch, that's whether you let CFD, tunnel, astrology/numerology guide you...

Lotus started a bit later so no surprises they're not first out of the blocks.

As for the other two, well, you really do need money and planning to go racing. And you can't get money if you don't have planning to sell to potential sponsors.

[timbo]

As for the other two, well, you really do need money and planning to go racing. And you can't get money if you don't have planning to sell to potential sponsors.

I wonder whether "selling the Manor plan" thing was aided by a certain mr.Donelli.

[Richard]

Regarding fuel tanks...

There is no rule on fuel tank size, but why let the facts get in the way of conjecture on a technical forum?

Its very easy to find the rules, you just google "F1 FIA rules" and you get taken straight to them!

http://www.fia.com/sport/Regulations/f1regs.html

[Sisso KavanhaK]

Yes! They exists! Razzia and Digrassi Gogogo!!!!
[youtube]http://www.youtube.com/watch?v=jJO4OFPNTuU[/youtube]

[fizzer]

as you can bang through a number of mods in the tunnel far quicker than you can in CFD.

Do you have a reference for that? Maybe in the early days of CFD that was the case, but now computing power is so relatively cheap, I believe you can run many more variations through CFD compared to constructing and testing physical models.

My understanding is that CFD is faster, but could have some inaccuracies that need to be verified in real world testing (track or tunnel).

[Pandamasque]

[youtube]http://www.youtube.com/watch?v=VJH8CSmcUJA[/youtube]

[SZ]

Do you have a reference for that? Maybe in the early days of CFD that was the case, but now computing power is so relatively cheap, I believe you can run many more variations through CFD compared to constructing and testing physical models.

My understanding is that CFD is faster, but could have some inaccuracies that need to be verified in real world testing (track or tunnel).

Experience, and I've posted a typical test cycle before.

It takes days to run a half car with reasonable accuracy on most supercomputing clusters at one fixed position.

It takes well under an hour to do the same in a wind tunnel. Teams that can afford a test program tend to build the models concurrently whilst testing goes on.

There's stuff you can do in CFD - aspects of flow interrogation - that's not possible in most wind tunnels (certainly not any F1 facilities). This renders CFD complimentary, in that flow interrogation is much easier. The really, really advanced stuff very few F1 teams have working. There's little evidence that the VR-01 has any such methods applied to it.

[PlatinumZealot]

Days? Are you absolutely sure?

Whatever it is, Wirth found it easier to just use CFD alone. And here is the Virgin 1, running on the track!

I guess Wirth is a master at interpreting and adjusting the CFD results. A simple weapon in the hands of a master can be devastating!

[timbo]

Whatever it is, Wirth found it easier to just use CFD alone. And here is the Virgin 1, running on the track!

I guess Wirth is a master at interpreting and adjusting the CFD results. A simple weapon in the hands of a master can be devastating!

And the car looks, well, simplistic.

[noname]

It takes days to run a half car with reasonable accuracy on most supercomputing clusters at one fixed position.

It takes well under an hour to do the same in a wind tunnel. Teams that can afford a test program tend to build the models concurrently whilst testing goes on.

There's stuff you can do in CFD - aspects of flow interrogation - that's not possible in most wind tunnels (certainly not any F1 facilities). This renders CFD complimentary, in that flow interrogation is much easier. The really, really advanced stuff very few F1 teams have working. There's little evidence that the VR-01 has any such methods applied to it.

test in the tunnel may take an hour but it takes much longer to build and set up the model, the tunnel and postprocess the data. it takes time and, what's more important, requires a lot of skills and experience to do it right. in the recent years we had more than enough examples it's very easy to screw it up.

Wirth decided to rely purely on numerical simulations in the design phase not because he's arrogant, lazy or so much budget constrained but because he has a lot of confidence in the tools he's using.

the only way to build such a confidence is to devote a lot of time to develop the tools and this could be done only by comparing results with the reality. looking at his, and his team, past experience I have no reason to doubt they knew what they are doing. they've been successful in the lower formulas, time will tell if that's enough to succeed in F1.

I do have to admit I am anxious to see what VR-01 is capable of.

regards

[SZ]

n smikle

There are some good threads on how CFD works and what it takes to run a car. You have the model, you have the mesh of a given complexity, you have the turbulence model. The nature of flows about an F1 car and the accuracy required means you're not using particularly simplistic turbulence models. You're using tens of millions of elements for a half car. This is not a simple problem. That gives you one car, at one ride position, with one set of options. Once you've got a solution you can be smart about how you run the next problem but it's not super quick, and between major evolutions of a problem you certainly don't get away with running any problem quickly. CFD is great for flow interrogation, you can visualise almost everything you can't in a tunnel, can can explore why something works or not to a far greater degree than you can

I don't doubt Wirth has unique capacities in CFD. I don't doubt you can do a lot with it. I don't doubt his team brings a new perspective to how to use it, and in having less people, that they're probably unusually effective with it. It doesn't replace a wind tunnel though, as it's a complimentary tool.

CFD results are relative to the accuracy of your computational model. How you validate this without tunnel testing, I'm not sure. It could be spot on. Might not. I doubt there hasn't been an experimental validation somewhere in the plan, it's more likely that if the car's vaguely competitive, it's in Wirth's best interests to keep mum about it. Despite my doubts, I really hope Virgin does well. Nice vibe about them, go underdogs and all that.

noname

Model making and experimental testing happen concurrently, there is no waiting to run a test whilst the model is build and set up. There's far less data to post process from a tunnel run. Regards the skills, experience and propensity to fck it up... the same applies to CFD. They're complex processes and the people running the show need to be experienced and astute, and being human, occasionally make mistakes too.

Didn't say Wirth was arrogant or lazy. He certainly is budget constrained or he'd have both tools. I hope, for the same of F1, that he's better than the established players and using what his team's got.

"Comparing results with reality" means comparing with experimental data obtained in a controlled environment - that'd be a wind tunnel, not the track. See previous comments.

The major revolutions in F1 aren't down to having a tunnel or a supercomputing cluster, it's down to having innovative staff in an environment that allows them to thrive and effectively contribute. A fast VR-01 would be less of a testimony to the merits of CFD than it would be to Wirth's ability to grow and manage a competitive team.

[noname]

A fast VR-01 would be less of a testimony to the merits of CFD than it would be to Wirth's ability to grow and manage a competitive team.

only for this one reason I do wish all the best to Wirth and his team. for me that would be proof they did something "CAE industry" and its customers was not able to do (or at least were very resistant to) for decades - perform simulations, not only analysis.

words about laziness and arrogance were mine. I am sorry if I made impression they came from you.

[horse]

I thoroughly agree with SZ on this one. I think the CFD thing is a bit of good marketing rather than a revolution in practise. On Wirth Research's own website they state that

The company only uses Fluent solvers for its analysis.

Fluent is a good piece of software, but not revolutionary in any way. What Wirth has is experienced engineers who can interpret the results and have some feeling of what is right and wrong, however as SZ said without a wind tunnel it's hard to validate those assertions.

Even with a wind tunnel and excellent computing facilities you can get it wrong. BMW had all this at Hinwil for last season, but produced a dog, McLaren the same. The point being there is a lot more to building a car than aerodynamics and that is what CFD does, aerodynamics.

If Wirth had come out with "we have a excellent team of experienced race car designers", I would be much more convinced than "we have a big computer and Fluent".

EDIT: In addition it's not as if CFD designs the car. You build a model, test it, change it try another model. It's the skill of the user as to designing a model that can improve on the last one, nothing to do with the simulation (unless you have an awesome genetic algorithm and sole use of ORNL's Jaguar, maybe).