Virgin MVR-02 Cosworth

[marekk]

At the end of the day, it's not about weird holes in the rules. It's about having the right global design concept, and having a "feel" for how the flow bends and where it goes. That just comes with years of experience and some talent to start with.

Wind tunnels and CFD can only be used to check the validity of a given solution. They can't design the solution for you from scratch. That's the creative bit of the engineer's brain's job. And that's the beauty of it.

Unless you use brute computing force. One can just code boundary conditions based on rules and track data and iterate, changing shape of the car one step at a time and simulating a lap, until optimal solution is found.

There is only one problem - we are now at 10*10^15 Flops (IBM's Blue Gen), and we need about 10^15 computing clusters of this size to be able to finish our calculations the same year we started. And 10^30 bytes of disk space to keep the data.

AFAIK teams are allowed to use computing clusters of about 40*10^12 Flops (about 1000 times slower then blue gen)...

[PlatinumZealot]

You need a design philosopy first. Computer is not a very intelligent being, it's just a tool.

For intance iterating to change the shape...the engineer has to set the range of shapes that are to be changed to, before any iteration begins. The solution is just what is optimal of what the engineers sets, not what is optimal for all cases.

[raymondu999]

Yes. The engineer has to input the design into CAD first, and has to think of that design first. The CFD will probably just help in testing, not in designing.

[nacho]

I have thought so too, you draw up something and throw it into the CFD software and wait... Are there tools that modify the model within certain boundaries at the same time as performing analysis?

[raymondu999]

Basically something that evaluates all possibilities of design within a certain range? I suspect that, for example, with cascades, they could set it so that it tests all cascade widths between 10-80mm (I'm guessing at random numbers here) or something like that, but the basic design would have to be human

[beelsebob]

You need a design philosopy first. Computer is not a very intelligent being, it's just a tool.

For intance iterating to change the shape...the engineer has to set the range of shapes that are to be changed to, before any iteration begins. The solution is just what is optimal of what the engineers sets, not what is optimal for all cases.

Assuming you're responding ultimately to my waxing lyrical about how using genetic algorithms for this would be lovely: You're missing the point of genetic algorithms – the point of them is to remove the "engineer creates a design" stage and replace it with "computer creates thousands of random variants, and tests them all" – the computer then bases future random iterations on the designs that did well at that stage. This is why it needs extra computing power – it needs to run thousands of times more simulations than typically are needed, a) because the algorithm is coming up with thousands of variants and b) because it takes many more stages to work through to a working design.

In the past though, genetic algorithms have done some remarkable things. Circuits that previously had been human designed to use thousands of components reduced to a handful – one of which was disconnected from everything else and using an odd induction effect to do a complex job for example.

Anyway, back from fantasy land – yes, virgin need designers – probably better ones than they have.

[zgred]

[ESPImperium]

Looking at the team performances over the 2010 and 2011 season so far, the team is in need of a re think and re energiseation. The VR/01 was 7.1% behing P1 last year after event 2, and i take P1 as the time in Q3 for my sheets. This year its 6.2%, thats effectivly only a 0.9% performance gain. Hispania were 7.9% last year and this are 7.5%, but given their circumstances they have moved forward, and given Virgins resources, they have flatlined.

Im thinking that Timo Glock has the right thoughs, get the car into a wind tunnel and sort the CFD numbers out as there is something wrong. Ferrari are trying to get their numbers correlated with plenty of constant speed runs at 200kph in FP1 & FP2 at Sepang, but they will correct their CFD and tunnel numbers quickly with this data. I think that Virgin need to get their car to do plenty of constant speed tests and get their car into a tunnel to fix things, as something is wrong.

Virgin also need a top quality designer and design team, Look at what Team Lotus have invested, Hispanis have Geff Willis a War Hardened veteran. I know that Pat Symmonds has been doing some consultancy work for Virgin, but what Virgin need to do is get a guy of comparible standard to take charge for their next car. I think that with the new Wirth Research CFD facility that will be opened sometime this month seemingly, they can gain that 2-3 seconds they need to get onto the Team Lotus coat tails. Also if i were Virgin id call the MVR/02 a day after Silverstone and do what Team Lotus did for the T127 and put 100% resources behind the next years car.

Im jusn not sure where to go with Virgin at the moment, Hispania are supposed to be the small team that struggles, but with Hispania getting massive direction from war hardened veterans, they are making strides and are looking more credible over Virgin as a F1 team and not one that should go back to the lower formulas before trying again.

[siwillems]

Without having any design flair and wind tunnel testing avaliable, were are the upgrades going to come from, presumably with the CFD approach the car is about as good as they can get it at the start of the season except perhaps for ironing out reliability problems and the most obviose of flaws in the design of the car.

I agree that brute force and organic algorithms can produce an efficient aero shape but the computing power required is far in excess of what resourses the teams are allowedto use; the only way this automated CFD approach would work was if the resourses could be used to find a better more efficient aero package by making slight refinments from what was a good starting place in the first instance.

Untill virgin start to use wind tunnel testing and use proper designers for their cars im affraid they will have a slow but reliable car.

[marcush.]

nick wirth has done outstanding LMP machinery on the same level with Porsche and speedwise not that far away from Audi or Peugeot.With CFD only aero.
I don´t think the CFD approach is holding them back.
the car looks clean if a little uninspired and simple to me but really they are learning the ropes of Formula 1 more than anyone else.I think they are prepared to go it the hard way and learn it all ground up whereas the others have brought in current or at least more recent F1 expertise to move forward.Pat Symonds will without a doubt open some eyes to reality there.
this year the car is running at least remember last year-HRT was catching them quickly in terms of miles covered .....this year the mechanical package is sound and clocking up the miles will build their knowledge base.
In this specilist world how can you catch up when your opposition does what you try to learn since dozenns of years plus with triple the workworce and x times your budget? It is an impossible feat really.you just don´t come in with a bang and destroy the old guard these days-not even with Willis or Gascoyne leading your technical department.

[PNSD]

IMO this car needed a strong 3-4 week wind tunnel program backed by PIV to first correlate baseline numbers. Then use only CFD.

We must remember the 'only CFD' approach is primarily based on money. They would use a wind tunnel if money allowed, except it doesn't, however surely a 3-4 week program could work?

[siwillems]

I think the problem really is one of attitudes, Nick Wirth believes quite passionatly that the CFD only approach is cost effective and that wind tunnels are an unneccersary expense. Who knows maybe the CFD only approach may work one day but at the moment their can be no real substitute for getting the actual race car in a wind tunnel and watching the flow over the actual working model warts and all so to speak.

[PlatinumZealot]

He uses the wind tunnel actually.. but he uses it like only once or twice.

[raymondu999]

really? Source? He often states on TV that he doesn't use tunnels at all, not even once to check the figures on CFD

[PlatinumZealot]

He said it a number of times.. this is one of them:

http://www.computing.co.uk/ctg/intervie ... in-f1-team

Computing: Can CFD modelling to design and test F1 cars completely replace physical wind tunnel testing?

Wirth: Even scale models within CFD by definition are not the real thing and you still have to take what you design to a wind tunnel or track.

But a huge part of the process is when people build components wrongly – we say this is the way we want it, and they do not do it, so the shape distorts under load.

All the teams do validation on that and that will continue; but the heavy lifting of design can be handled by CFD and we are now doing it all virtually.

Every eight-week cycle we report back to the FIA [F1’s governing body] saying how much time we spent testing the car in the wind tunnel, and how much with CFD and it is close to 100 per cent of that [the team’s allowance] in that period.

Read more: http://www.computing.co.uk/ctg/intervie ... z1J8eT4gFu
Computing - Insight for IT leaders Claim your free subscription today.