Virgin Racing VR-01 Cosworth

[SZ]

noname

No offence taken, I didn't read your comments as such.

F1 teams presently do a lot of simulations and analysis of new concepts in CFD. The time required to do this means it's usually quicker to, once a concept shows promise in CFD, throw a few variations on the theme into the tunnel and tweak it for performance at a wide variety of ride positions. CFD can't do this. One of the crucial bits in Wirth's approach has to be managing the simulation of a concept at various key ride positions chosen as a discretization of key track scenarios, such that it's as effective as intended no matter what the car's doing on track. Typical CFD is done at one, or a few ride positions. Any more is generally viewed as time consuming when the same thing in a tunnel is over very quickly; possibly one of Wirth's key innovations is a new approach to managing as much with CAE tools only. Maybe they're wind tunnel testing somewhere and not telling anyone. Who knows.

horse

Fluent isn't bad and they're not the only team to use it (many very fast teams do too). It's not the fastest or most flexible code, but know how to use it and it does the job. Very few teams use anything revolutionary.

Not as if CFD designs the car? In some cases I might disagree here... it's not an impossibility but the class of solvers required aren't commercial. They're in-house and fcking complicated to develop. Judging by the cars, not all teams have them (far from it).

[horse]

Not as if CFD designs the car? In some cases I might disagree here... it's not an impossibility but the class of solvers required aren't commercial. They're in-house and fcking complicated to develop. Judging by the cars, not all teams have them (far from it).

I'd like to see that software in action! I know F1 regulations are strict, but there must be so many variables that such an algorithm would have to vary with mechanical caveats for feeding the radiators, etc, that would be an amazing piece of work which could do it all. Still, I think if something was considered in isolation (like an endplate, perhaps) this would seem like a more tractable problem.

[SZ]

Do a little research... you'll find out what I'm talking about quickly enough (CFD solvers that can deform a shape based on a cost function). Plenty of examples online.

When you get a grasp of that you'll realise that applying it to a whole car is not a trivial exercise.

[PlatinumZealot]

You can make mistakes in the wind tunnel too, that is the thing. We always hear teams expressing huge relief that the real life test results of the car match up with the windtunnel. These are experienced people like Mclaren, Brawn and Ferrari.

Wirth more than likely did his homework, maybe he has done some past cross referencing between the wind-tunnel or even real life tests and CFD and maybe has found what needs to be adjusted from the CFD. He could have even developed his own software plug-ins that will do this, to bring it closer to not even the wind-tunnel, but closer to real life. We know that CFD is no windtunnel, but the windtunnel is still not a real track, So we cannot say that CFD without windtunnel is far worse when we have teams releasing windtunnel dogs every year.. I think it's all about the user of the tool and not the tool itself.

It's like art in a way. Along with the technical stuff, I like drawing and painting. I am learning to paint on the computer now. It is a different feeling than the paintbrush, the control of pressure and flow is slightly different but you can get very similar effects once you master painting on the computer. The computer will even let you surpass the brush because you can do so much more. You can go back and erase (good luck doing that with real paint!), you can change the colour by pushing a slider, control layers and add effects etc. The feedback and adjustment times are much faster and most Importantly a mistake is less costly! It may not be the same, it maybe better or worse but it is similar! And in the hands of a master it will be (well we hope) just as effective. That is the key point with the Virgin!

[horse]

It's like art in a way.

I'm sorry, I can't agree here, it's not art, it's science. You can have creativity and the tools can make that creativity easier, but in the end you are looking for a tool that can simulate reality as closely as possible. So if you don't test it against something real, frankly, it's not useful. Fluent is well validated for a number of real cases, and thus you can have some confidence in it's outcomes, but F1 is a unique formula, so you can only assume you are getting an inference to the answer rather than the true solution. McLaren, BMW, Honda are all good examples that even when CFD is used in conjunction with wind tunnels the results are not necessarily reliable. So take half of the scientific process away and there is twice the risk.

Do a little research... you'll find out what I'm talking about quickly enough (CFD solvers that can deform a shape based on a cost function). Plenty of examples online.

I could find some research for simple shapes, nothing as complicated as an F1 car. Can you give an example?

[PlatinumZealot]

I know, but Engineering is not a science, There are so many fudged up things in engineering it is not even funny.. but I didn't mean to talk about engineering, I was talking about use of tools; which one gives a desired result and at what price.

I was comparing the results of CFD and/or windtunnel to a painting of a real life object using a brush or a pen tablet.


I have done a little CFD myself for my final year project; I couldn't have built 10 flame-tubes in that amount of time . I (nor the school) didn't want to pay the money for all the stainless steel i needed.. CFD was my only choice at the time.. In the end the flame tube didn't work out as I wanted anyway because I was inexperienced at CFD and had limited knowledge of turbine combustion chambers at the time. The flame tube didn't sustain the flame properly. BUT...who said I would have gotten it right If I made 10 or even 20 flame tubes? If I was a master at gas-turbine combustion I bet I would have gotten it right in both the CFD and tests.. but why do so many tests when being a master I know what shape tube is going to give good results anyway? So I would just run the best models in CFD, pick the best one and when the final product is made.. I know it will be accurate enough. I can even test that single final product in a real life situation instead of testing many.

Don't get me wrong I am not defending Wirth but I think the method is sound If he is a true master.

[ESPImperium]

Anyone know how many laps Timo Glock did on the first shakedown day???

I know the second shakedown that Di Grasi did they did 36 laps.

[mith]

Hi did 42 laps. Source is in Polish, so I won't post it here unless you ask.

[alleyoop]

42

http://www.pitpass.com/fes_php/pitpass_ ... t_id=39898

[ESPImperium]

Thanks.

That means the first virtual car in F1 history has done 99.918km on a actual track, even tho its only a school track, the same school track that Brawn GP started their first and only car on for the first time, the Silverstone Stowe Circuit.

[noname]

You can have creativity and the tools can make that creativity easier, but in the end you are looking for a tool that can simulate reality as closely as possible. So if you don't test it against something real, frankly, it's not useful.

you are absolutely right but if you have enough confidence in your tools you can delay the time when you have to validate the design against reality.

as an example - you can rely on even hand made assessment of the stress level in the bended beam or design riveted joint using data from the tables. however you do have to perform a huge amount of tests to build these tables.

the point is Wirth was able to create the tool sound enough, and learn how to use it, so he can avoid wind tunnel tests. I have no doubts this process required a lot of wind tunnel tests, but now they can live without it and start testing with the complete car.

in aviation aero guys are simulating flow between nacelles and the core or flow trough the turbine (fluid dynamics is not my strongest point so I will not draw any comparison with F1) and are validating results against real model having complete engine under the wing. I know they went along the long and rough way, with long hours spent on testing with real models, but if they've succeed I think it could be done in motorsport also.

[SZ]

Many of you are confusing what it takes to test a car accurately in CFD with what it takes to run a development program in CFD.

For some test cases, particularly bluff bodies, whilst it's easy to design a model in CFD getting accurate results is no laughing matter. With good computational power, meshing ability and the right turbulence model, it's then possible.

Once you've got that, to test and retest nearly endless tweaks to it takes a lot of time in CFD and a lot less in a wind tunnel.

That you can model a F1 car in CFD with relatively good accuracy is known, all the teams do it.

That you can run a development program in CFD at a competitive speed to your opposition is unknown, and for the moment not heard of. This is where Wirth would have to have something completely different to his opposition's approaches, as in this is where his team would have to bring something completely new to F1. I have no doubt that his team can be more effective with what they've got than the bigger teams - smaller teams are inherently easier to manage with less room for politics. Whether it's enough to stay competitive, who knows. What you need to develop for the current rules is certainly helped by having the established teams run their cars for a year first - much easier to pick what's a good and not so good design element when you've a season of cars to copy.

Suggesting teams release 'wind tunnel dogs' is a bit out of line. This hasn't been done since the late 90's. Teams have developed cars in CAE and in a wind tunnel for over a decade now, and putting out something that's not competitive isn't the tunnel's fault - it's a managerial fckup. It's a controlled test environment like any other engineering test tool. If you know how to use its limitations, you go far... it's not as if McLaren won't ever trust their wind tunnel again after last year's epic fail. They'll go through their processes to see where they went wrong, and amend as much to make sure that doesn't happen again. This is normal.

horse

The example you found for simple shapes are relevant. The 3D codes capable of doing something as complex as an F1 car, well, they're not commercial codes. They're developed in-house. Same theory applies.

[PlatinumZealot]

That you can run a development program in CFD at a competitive speed to your opposition is unknown, and for the moment not heard of.

The virgin car is out on the track and more parts are coming in next week! We will see how they keep the development up during the season. I think we will be surprised.


Here is some good news:

"We have acquired some good information to feed straight back into design for the development of new parts, which we may even be testing next week. All in all this is a very pleasing conclusion to a memorable week for Virgin Racing."

http://www.autosport.com/news/report.php/id/81316

[Saribro]

Nice.
I think I found my favourite team for 2010.

[SZ]

If they're in the pack at the Jerez tests, I'll be very happy.