Will a CFD only gamble pay off?

What with virgin and usf1 choosing to design there cars soley from the use of CFD and not wind tunnels i wondered are there any cases of this approach actually working? i know that lola's last f1 car was designed without a wind tunnel and was a complete dog of a car, although this was before CFD was in use, also if virgin or usf1 produce decent cars would other F1 teams choose this approach to help cut down on costs/staff?

It's a big big gamble I think. Way to big.

What with virgin and usf1 choosing to design there cars soley from the use of CFD and not wind tunnels i wondered are there any cases of this approach actually working? i know that lola's last f1 car was designed without a wind tunnel and was a complete dog of a car, although this was before CFD was in use, also if virgin or usf1 produce decent cars would other F1 teams choose this approach to help cut down on costs/staff?

BMW were heavily into CFD, and it didn't work for them.

It's certainly a bold move. And a great question to ask. Of course time will tell. Branson has made a lot of noise about the fact that his team will not do things like every other team.
They're either ahead of the game or taking a silly gamble. If I was being cynical I would say they are simply looking to cut corners on development costs.
Still, if it works, watch every other team sell their wind tunnels.

Wirth's LMP cars were almost exclusively done with CFD weren't they? And very competitive. I suspect this is the way things are going in future, it's just that no one has found "the trick" for getting it consistently right. I wonder if Wirth has with his LMP success... and if so, we could be in for quite a shock.

Wirth's LMP cars were almost exclusively done with CFD weren't they? And very competitive. I suspect this is the way things are going in future, it's just that no one has found "the trick" for getting it consistently right. I wonder if Wirth has with his LMP success... and if so, we could be in for quite a shock.

Years since I used CFD but in those days CFD was great for evaluating possible changes to an existing design not for producing hard numbers from blank sheet of paper stuff.

They are focusing only on CFD because it's cheaper than build windtunel for new team not because it's better than windtunel.


Even if it will turn out good move, there is no need of dispose already existing windtunels

BMW were heavily into CFD, and it didn't work for them.

Very true, although i think that was more down to the technical staff not interpreting the rule changes aswell as some of the other teams. BMW's progress heavily using CFD when the rules were more understood in 06, 07 and 08 was pretty impressive, overall they probally out developed all the teams between 06 and 08 other than mclaren and maybe ferrari, which shows how effective it can be, although they also had a wind tunnel unlike virgin and usf1.

If you have no money you get more done with CFD than tunnel.

As I recall, McLaren reportedly made more development gains this season using CFD than the Wind Tunnel.

CFD is getting to the point at which it's going to be very precise and effective. I think Virgin is starting what will be a trend.

As I recall, McLaren reportedly made more development gains this season using CFD than the Wind Tunnel.

CFD is getting to the point at which it's going to be very precise and effective. I think Virgin is starting what will be a trend.

I wonder though was it crap cfd results that got them into the mess in the first place

What with virgin and usf1 choosing to design there cars soley from the use of CFD and not wind tunnels i wondered are there any cases of this approach actually working? i know that lola's last f1 car was designed without a wind tunnel and was a complete dog of a car, although this was before CFD was in use, also if virgin or usf1 produce decent cars would other F1 teams choose this approach to help cut down on costs/staff?

Just to clarify, USF1 have said that they will verify and test their car in the Windshear windtunnel in Charlotte. Virgin have said that their car will not see a windtunnel at all.

Nick Wirth is a CFD zealot, and Wirth Research is a CFD company, so it's not surprising that he is trumpeting the techniques. However, he has had real success with the Acura cars, and many year of experience. (There are plenty of people who say that the Simtek cars were fundamentally good designs that were hampered by other factors.)
Obviously the falling cost of processing power relative to building prototypes is driving CFD takeup, but there are also other factors. People will say that a computer simulation is not real life, but then neither is a 60% scale wind-tunnel model. Both need a great deal of mathematical conversion and interpretation of results. And of course, you are limited to a maximum wind-speed for tunnel testing now too.
One last thing, I don't know how much what has been said is purely advertising, but according to their website, Wirth have their own wind-tunnel (don't know how big or what spec.), they are choosing not to use it.

I wonder though was it crap cfd results that got them into the mess in the first place

No, because they use a wind tunnel too. What you design in CFD then you test it in the tunnel.

The MP4-24 born to be slow, it was bulky and underdeveloped, and KERS contributed a lot to worsen the aero. Aeroefficiency-wise it reached the mediocre level by the end of the season, that with the powerful engine, the good mechanics, KERS and an exeptional driver was capable for some very good results on some kinds of tracks.

The Acura ARX-02 which Wirth designed... even that wasn't 100% CFD. He admitted that they did spend 15 days in a wind tunnel with a scale model.

So, if they really do go 100% CFD with this car it'll be extremely interesting to see the result of it.

Edited by Xaus, 15 January 2010 - 23:34.

And of course, you are limited to a maximum wind-speed for tunnel testing now too.

I'd forgotten that. It's a low speed too, 50 km/h or something. Though how they police that I cannot see.

No, because they use a wind tunnel too. What you design in CFD then you test it in the tunnel.

The MP4-24 born to be slow, it was bulky and underdeveloped, and KERS contributed a lot to worsen the aero. Aeroefficiency-wise it reached the mediocre level by the end of the season, that with the powerful engine, the good mechanics, KERS and an exeptional driver was capable for some very good results on some kinds of tracks.

I don't really see it that way. Whatever about overall aero efficiency, it's clear the car had a 'bug' in high speed corners where they were losing the downforce off the rear wing. That was something that was not picked up by their cfd or windtunnel testing.

I'd forgotten that. It's a low speed too, 50 km/h or something. Though how they police that I cannot see.

I have just been to look it up. It's a little more than that. 50m/s which is just under 112mph.
However, I have been told by someone I trust. (He used long word like "Bernoulli" and "Reynolds number", and drew arcane symbols in the beer on the pub table.) that with the scale it is equivalent to testing at about 80mph.

CFD, like wind a tunnel, is primarily still an evaluation tool. It doesnt interpret rules or come up with ideas. You could evaluate and improve upon a bad idea for months and improve the design by ten fold but if it was a **** design in the first place you havnt achieved much.

McLaren had no problems with tunnel testing or CFD. The car matched their predictions exactly. The only problem was that everyone else had designed a quicker car.

I have just been to look it up. It's a little more than that. 50m/s which is just under 112mph.
However, I have been told by someone I trust. (He used long word like "Bernoulli" and "Reynolds number", and drew arcane symbols in the beer on the pub table.) that with the scale it is equivalent to testing at about 80mph.

80 mph. I vaguely remember reading something about doing aero testing using water at a low velocity. So that suggests there will be some quite big differences between what they see with air at 80mph compared with racing speeds of 3x that. Presumably to do with the mass of the air and its energy.

So maybe doing without windtunnels isn't such a big loss.

80 mph. I vaguely remember reading something about doing aero testing using water at a low velocity.

Water is 13x slower for same Reynolds number (i.e. identical effective speed after it is normalised by fluid type and object length).

Of course there are other properties to normalise too, and unfortuately they are not compatible I think.;)

Edited by V8 Fireworks, 16 January 2010 - 00:31.

McLaren had no problems with tunnel testing or CFD. The car matched their predictions exactly. The only problem was that everyone else had designed a quicker car.

No it didn't. Remember all the weird paint splattered all over their car, running the 2008 rear wing in testing?? Took them ages to figure out what the air was actually doing over the car.

Water is 13x slower for same Reynolds number (i.e. identical effective speed after it is normalised by fluid type and object length).

Of course there are other properties to normalise too, and unfortuately they are not compatible I think.;)

Ah

Well one way and another it looks like the sooner the teams can get their cars on track the better. And then the ability to develop quickly using honest-to-goodness real life data will be the key to the season.

After setting off in the right direction to start with, of course...(thinking of Newey and his big sheets of paper on a board )

What with virgin and usf1 choosing to design there cars soley from the use of CFD and not wind tunnels i wondered are there any cases of this approach actually working? i know that lola's last f1 car was designed without a wind tunnel and was a complete dog of a car, although this was before CFD was in use, also if virgin or usf1 produce decent cars would other F1 teams choose this approach to help cut down on costs/staff?

CFD is certainly the future. The thing that Wirth tends to forget is that all teams use CFD to design their cars, and use wind tunnels to back up the CFD results. Sauber has the most powerful computer in F1, combined with one of the best wind tunnels in F1. Take away the wind tunnel, and Sauber would still be more advanced in CFD technology, but with the wind tunnel, they have a huge advantage.

So, Wirth has a technology used by everyone else, but lacking a technology used by everyone else. I have yet to hear that Wirth has a better CFD computer than anyone else, so they are just lacking in resources and will fail as a result.

I am looking forward to seeing Branson as a trolley dolly.

As for USF1, they have promised to use a full scale wind tunnel. I doubt it will happen, since I doubt they will have a car to test, but nonetheless, the only team to refuse wind tunnels is Virgin.

I predict Virgin to produce a car which is slower than a GP2 car.

you cant get scale air aswell, well i guess you cant xD, maybe if you had some really fancy system... but the model will react differently to a full size car. the tunnel's are also limited speedwise, not just by regulations, but practicaly too. I suspect that CFD has reached a tipping point for atleast the newer teams, as rather than invest in windtunnel time, they would rather go cfd. the current f1 teams already have the needed stuff, apart from maybe torro rosso?

a good, if abit old interview with nick worth, tis from last year, not sure when tho, tail end of the alms season i guess

haha xD, nick says he has 4000+ pc's in that interview (more than sauber) with more on order

CFD seemed to work fine for the Space Shuttle.

Wirth's LMP cars were almost exclusively done with CFD weren't they? And very competitive. I suspect this is the way things are going in future, it's just that no one has found "the trick" for getting it consistently right. I wonder if Wirth has with his LMP success... and if so, we could be in for quite a shock.

His re-do of the Courage LMP2 was very successful, but his Acura LMP1 was less than successful this year. That may have been due to his idea using extremely wide front tires, which many felt was the wrong way to go.

Regardless, the big teams use CFD, but don't rely exclusively on it for a reason, so I think it's a bit of a gamble which may not result in what they anticipate.

i didnt follow the latter part of the season, but they seemed decent at the start, got pole at sebring... the wider front tyres was a attempt to steal something back, as the regs favour diesels a fair bit, acura dont sell any diesels i think, so they couldnt go down that route.

not sure if michlin devloped any proper front tyres for them either

i didnt follow the latter part of the season, but they seemed decent at the start, got pole at sebring... the wider front tyres was a attempt to steal something back, as the regs favour diesels a fair bit, acura dont sell any diesels i think, so they couldnt go down that route.

not sure if michlin devloped any proper front tyres for them either

They did get the pole at Sebring, but largely because the Audi and Peugeot teams sand-bagged. The diesels were running practice times up to 2 seconds faster than their qualifying times, and clearly were holding back (as was quickly revealed in the race).

They did need to get something back, but the issue was that it was very hard to make the car driveable with that much tire at the front, and it created significant balance issues. That wasn't something you wouldn't find out until testing, but at the same time, it was something seen as a possibility.

Like somebody else said, CFD is mainly a way to 'test' a car's aero. And just like a windtunnel, I imagine that its imperfect, and therefore, a variety of testing apparatus' would go a ways in ensuring the validity of the predictions made by either test.

We already see excuses that a team's windtunnel isn't calibrated properly or whatever and I can see the same thing happening with CFD.

Seems thats its a budget restriction more than anything and windtunnel use would be highly desired by these two teams, just to ensure that their CFD calculations can be confirmed by another sort of test.

CFD seemed to work fine for the Space Shuttle.

I suspect the heat transfer models are the important part there, not too many f1 car go fast enough to generate 2000 deg C at the nose, unless they are cheating.

Otherwise the shuttle is in space where there is no air, or plumeting from high atlitude like a brick isn't it?

Like somebody else said, CFD is mainly a way to 'test' a car's aero. And just like a windtunnel, I imagine that its imperfect, and therefore, a variety of testing apparatus' would go a ways in ensuring the validity of the predictions made by either test.

We already see excuses that a team's windtunnel isn't calibrated properly or whatever and I can see the same thing happening with CFD.

Seems thats its a budget restriction more than anything and windtunnel use would be highly desired by these two teams, just to ensure that their CFD calculations can be confirmed by another sort of test.

No F1 team in their right mind would design a car solely on CFD, unless they are forced to, in the case of Virgin Racing. If the team had very deep pockets, they would surely sing a different tune. However if you HAVE to go all CFD, Nick Wirth is probably the best guy for it

I'd forgotten that. It's a low speed too, 50 km/h or something. Though how they police that I cannot see.

That's something I asked myself for months - how do they (the FIA) check if a team sticks to the limited amount of hours in wind tunnel, 60%-models, summer break...?

On topic: In their 2nd 'Bob Varsha Walk Around'-video one of the USF1 guys claims that for years planes are designed by CFD only "before they're ever flown" (@3min52sec). If that's true, Virgin's and USF1's approach can't be that wrong. But I guess that companies developing planes spend a little more than USF1 and Virgin do on their cars.

I suspect the heat transfer models are the important part there, not too many f1 car go fast enough to generate 2000 deg C at the nose, unless they are cheating.

Otherwise the shuttle is in space where there is no air, or plumeting from high atlitude like a brick isn't it?

Well they "flew" the shuttle first w/CFD. After the first test flight the pilots commented that it was extremely stable.

Well they "flew" the shuttle first w/CFD. After the first test flight the pilots commented that it was extremely stable.

Planes don't have complex surfaces like F1 cars. It's not that you can't do CFD for F1 cars but given the time pressure and the fact that some things are so much quicker to do in a windtunnel, it seems that you'd fall behind in the development race compared to other teams. And since you can't validate your CFD with windtunnel, if errors are missed in CFD you're screwed.

Planes don't have complex surfaces like F1 cars. It's not that you can't do CFD for F1 cars but given the time pressure and the fact that some things are so much quicker to do in a windtunnel, it seems that you'd fall behind in the development race compared to other teams. And since you can't validate your CFD with windtunnel, if errors are missed in CFD you're screwed.

USF1 said they will use Windshear tunnel for development, but I'm not sure about Virgin. It would seem logical that they would, but who knows.

USF1 said they will use Windshear tunnel for development, but I'm not sure about Virgin. It would seem logical that they would, but who knows.

Virgin claims they are relying solely on CFD (Nick Wirth) but Branson implied that this design route was the result of their small budget more than anything else

It's one of the the things I am looking forward to this year, let's hope we see a fair crack at it, and the car's performance isn't hobbled by engine or reliability issues.

Sure if you had bottomless pockets you do both, wind tunnels and CFD, 24x7, in multiple sites, in multiple time zones ... that's a no-brainer, but if you don't, and you have to deal with the harsh economic actualities, then betting it all on CFD is probably the smartest way to go.

Either way, what we can say for certain is that this year is the year the tide turned, Wirth has crossed the Rubicon, whatever problems or pitfalls do trip him up, a good healthy dose of Moore's Law will soon brute force a solution, and the next set of guys that do CFD only, will definitely make it stick. Everyone, say ta-ta to propellers, sticky-tape and model-shops, get your meshes and sliderules warmed up from here on out.

No, it's going to be a disaster. Look at Mclaren last year.

Computer models said the car was perfect, however real testing showed that it was the slowest car on the grid.

You need both CFD and wind tunnel testing, just like everybody is doing, because computer models are no replacement for real testing.

That doesn't make a lick of sense, McLaren has a huge windtunnel, they built the factory round it ... if flinging wind and smoke and toy cars is the only way to do it, then why did the McLaren not work ... you can't have it both ways, and you definitely can't then turn around and single-out for blame CFD on a car that was born, built and underdeveloped, in a tunnel.

Wind tunnel testing is not real testing, rolling roads aren't real roads, it's still just a very crude and limited approximation (and that's before the new speed limits, and scale-limits). The calibration required to map any kind of tunnel output to the real world is critical, just as critical as the mapping of computer model to the track. There's no shortcuts just cos you've got a propeller.

If anything your argument suggests that Wirth might be right, wind tunnels are eye-wateringly expensive and vicious time-vampires, they require a factory-sized army to feed parts in, and a squad of cleaners to empty the rubbish skips of exquisitely modelled, exotically constructed failed ideas.
Who knows, if McLaren had spent what little post 2008 development resource they had available on lighting up the processor racks and crunching the numbers, maybe they'd have spotted the front wing end-plates and moody rear-wing a whole lot sooner.

Pound for pound, bang for buck, a million quid of 2010 computer time will probably beat a million quid of hot-air ... if it was my money, I'd go in with my eyes open, I'd make all the right noises about state of the art, acknowledge it was a gamble, but a carefully calculated gamble, I'd bet on the software-stack, and I'd bet it all.

I don't really see it that way. Whatever about overall aero efficiency, it's clear the car had a 'bug' in high speed corners where they were losing the downforce off the rear wing. That was something that was not picked up by their cfd or windtunnel testing.

There was no bug in the second half of the season, and we have nothing proof there was a bug in the first half of the season. For high speed corners you need lot of downforce, and with efficiency (e.g. for Spa). The MP4-24 didn't have as much downforce as an RB5, BGP 001. In downforce levels, in high speed cornering, the McLaren was only mediocre at its best shape, falling behind cars like Toyota, Force India and even the updated BMW.

The MP4-24 was the bulkiest, fattest car of the field. Altough they cutfrom it as much as they could through the season, but even on the end it remained significantly bigger, than any other car. That's definitely not good to move forward in air. Mercedes and McLaren was quiet about it for obvious reason, but they know the choosing of the KERS path was a fail.

Edited by hunnylander, 16 January 2010 - 09:31.

There was no bug in the second half of the season, and we have nothing proof there was a bug in the first half of the season. For high speed corners you need lot of downforce, and with efficiency (e.g. for Spa). The MP4-24 didn't have as much downforce as an RB5, BGP 0001. In downforce levels, in high speed cornering, the McLaren was only mediocre at its best shape, falling behind cars like Toyota, Force India and even the updated BMW.

The MP4-24 was the bulkiest, fattest car of the field. Altough they cutfrom it as much as they could through the season, but even on the end it remained significantly bigger, than any other car. That's definitely not good to move forward in air. Mercedes and McLaren was quiet about it for obvious reason, but they know the choosing of the KERS path was a fail.

I can only ask you to go watch onboard footage from the first half of the season, and see for yourself how the unstable the rear end of the car was. They brought in an upgrade at germany that seemed to fix the bug and give stability to the car (and confidence to the drivers). I would agree in solving this they had lost ground to the rest in terms of overall efficiency

I can only ask you to go watch onboard footage from the first half of the season, and see for yourself how the unstable the rear end of the car was.

Lack of enough downforce.

They brought in an upgrade at germany that seemed to fix the bug and give stability to the car (and confidence to the drivers). I would agree in solving this they had lost ground to the rest in terms of overall efficiency

After Nürburgring, the MP4-24 had the best aero, the best aero-drag ratio? No! It was only good on tracks which have no too many such high speed corners which requires a lot of downforce. For medium and low speed corners, you don't need as much downforce in an efficient way. That why Hamilton could take two victories on low speed corner tracks.

Lack of enough downforce.


After Nürburgring, the MP4-24 had the best aero, the best aero-drag ratio? No! It was only good on tracks which have no too many such high speed corners which requires a lot of downforce. For medium and low speed corners, you don't need as much downforce in an efficient way. That why Hamilton could take two victories on low speed corner tracks.

Lack of downforce makes a car slow, not undriveable.

Where did i say it had the best aero??

Lack of downforce makes a car slow, not undriveable.

Where did i say it had the best aero??

Lack of downforce makes the car slow, hence 'undriveable' when the driver want to drive it as fast as the other cars go around him. It just a word, I saw Lewis and Heikki driving that 'undriveable' car even grabbing points with it. It's just a subjectively used word, an expression. They said many times, the car is slow, the car lacks downforce, the car lacks grip, the car is undriveable. All meant the same experience, just different verbal expressions.

Edited by hunnylander, 16 January 2010 - 11:47.

I have a vague memory that the first Stewart car was designed using only CFD back in 1996/97. Anyone else remember this?

/Viktor

Of course not. It's gonna be disaster.

Edited by Trust, 16 January 2010 - 10:09.

Who gives a **** ?!

It won't be a true racecar anyway, it will another racertainment irrelevant-mobile good for the sole purpose of being lapped or causing "Trulli trains".