Thread: Wind tunnel use in 2008

It's hardly a big restriction. Only one wind tunnel and you can only use it 5 days a week?!

it doesn't sound like much, but i believe some teams have had access to 2 tunnels running 24hrs a day, 7 days a week

Thats 168hrs a week per tunnel, 336 potential hrs per week,

now they'll be down to 15 runs in an 8hr day, so either 40hrs a week, unless they can still run 24hrs a day, in which case 120hrs a week,

either way thats a big potential reduction, given that they are starting from a blank sheet on the aero this year it could slow the speed of development quite a lot.

if this is in place the teams who have had the time to spend on 2009 aero during 2008 may be in the box seats if they've got it right, as others may take longer to catch up?

Originally Posted by Robinho

either way thats a big potential reduction, given that they are starting from a blank sheet on the aero this year it could slow the speed of development quite a lot.

But the article was written in December 2007 and reporting that these restrictions were to be put in place for the 2008 season.

If you're right that teams have had access to 2 tunnels running 24hrs a day, 7 days a week this year then I assume that the radical cost-cutting measures proposed were never introduced

That's where CFD comes in. You can do a lot of pre-testing (so to speak) on computers and then stick the models that you think will work in the tunnel. I'm sure I read that the FIA want to restrict the amount of CFD that can be done, but it will be extremely difficult to police that.

Originally Posted by Andrewmcm

That's where CFD comes in. You can do a lot of pre-testing (so to speak) on computers and then stick the models that you think will work in the tunnel. I'm sure I read that the FIA want to restrict the amount of CFD that can be done, but it will be extremely difficult to police that.

The argument had traditionally been that CFD costs even more than wind tunnel testing so restricting wind tunnel use will actually push costs up.

I'm not sure that's an up to date argument with modern hardware.

Originally Posted by Mark

The argument had traditionally been that CFD costs even more than wind tunnel testing so restricting wind tunnel use will actually push costs up.

I'm not sure that's an up to date argument with modern hardware.

It's not only a question of hardware development, it's also about the development of the codes used. You may considerably reduce computation times with the use of newer and better optimized codes.
But I agree, computation power is getting cheaper by the day.

Originally Posted by Mark

The argument had traditionally been that CFD costs even more than wind tunnel testing so restricting wind tunnel use will actually push costs up.

I'm not sure that's an up to date argument with modern hardware.

It's a bit of a catch 22.

ioan is sort of along the right line with part of his post. Continued development of SW gives us more advanced code.

However, it ultimately does come down to the capability of the HW. The faster I/O calculations, the more you crunch a second.

HW always comes down in price if you remain at a static level but the high end stuff will always command a premium because of the edge it offers.

Originally Posted by Knock-on

It's a bit of a catch 22.

ioan is sort of along the right line with part of his post. Continued development of SW gives us more advanced code.

However, it ultimately does come down to the capability of the HW. The faster I/O calculations, the more you crunch a second.

HW always comes down in price if you remain at a static level but the high end stuff will always command a premium because of the edge it offers.

ioan is certainly correct in that in such a complex system the software, either a licence or custom made is going to be by far the major cost in the equation, that and paying the experts you need to run the system. The actual hardware they are likely to use is commodity boxes, which in the scheme of things are very cheap.

Originally Posted by Mark

ioan is certainly correct in that in such a complex system the software, either a licence or custom made is going to be by far the major cost in the equation, that and paying the experts you need to run the system. The actual hardware they are likely to use is commodity boxes, which in the scheme of things are very cheap.

Not necessarily.

What would you need for a basic system.

3 Kva per rack with Fibre backbone and other HW
Initial 36 TB of storage such as Netapps.
Suitable processing capacity

I don't know how much the SW is but you looking at a couple of hundred grand there per year for a basic setup.

Then when something like optic or Kubit processing comes out, they will all jump again.

Originally Posted by Mark

ioan is certainly correct in that in such a complex system the software, either a licence or custom made is going to be by far the major cost in the equation, that and paying the experts you need to run the system. The actual hardware they are likely to use is commodity boxes, which in the scheme of things are very cheap.

Don't get me wrong, if they buy the SW (this may or not be the case, I have no ideea) than it's certainly not more expensive than the computational power they need (and which they need to change or heavily upgrade every 2 years at least).
If they do develop they own SW (it's not at all out of question even if complicated) than they need to pay lots of money for those who do it, but certainly they will have a SW where they can do whatever they want, knowing every little bit of it and being able to develop it further and further.

What I was trying to say is that with today's technology being available to every one of them, it is rather the SW that will make the difference, and if they got the better mathematicians, engineers and programmers than the advantage they can get over the competition is way bigger than the one they get buying better HW. So IMO CFD software is more important than HW and allows for higher gains in a strictly limited and controlled environment.

You are also right about the HW being cheaper by the day as it is highly standardized and thus even if some of it needs to be adapted or personalized it still isn't as difficult and expensive as as 10 years ago.

In the end the people and the knowledge will be most expensive of all.

I hope I managed to give some sense to what I was trying to say.

Originally Posted by ioan

Don't get me wrong, if they buy the SW (this may or not be the case, I have no ideea) than it's certainly not more expensive than the computational power they need (and which they need to change or heavily upgrade every 2 years at least).
If they do develop they own SW (it's not at all out of question even if complicated) than they need to pay lots of money for those who do it, but certainly they will have a SW where they can do whatever they want, knowing every little bit of it and being able to develop it further and further.

What I was trying to say is that with today's technology being available to every one of them, it is rather the SW that will make the difference, and if they got the better mathematicians, engineers and programmers than the advantage they can get over the competition is way bigger than the one they get buying better HW. So IMO CFD software is more important than HW and allows for higher gains in a strictly limited and controlled environment.

You are also right about the HW being cheaper by the day as it is highly standardized and thus even if some of it needs to be adapted or personalized it still isn't as difficult and expensive as as 10 years ago.

In the end the people and the knowledge will be most expensive of all.

I hope I managed to give some sense to what I was trying to say.

Very clear and I agree.

Fluid dynam Sw has advanced greatly as has HW. You can spend millions but without the right people with relevant experience to manage it, your wasting your money. Just ask Honda and their wind tunnel

The actual software used by the F1 teams perhaps isn't as big an issue as you might think. Most of the teams use Commercial Codes that no doubt they've modified to some degree. Hardware is reasonably cheap these days and you can end up with a 1024 CPU cluster for a few hundred thousand quid at most.

The key to obtaining good results from CFD is to understand the limitations of the model used and to interpret the results correctly - the failure of CFD to predict the performance of the CDG is a good example of that. That's where the teams will spend the money - getting people who can analyse the results and come to the right conclusions.

Originally Posted by Andrewmcm

The actual software used by the F1 teams perhaps isn't as big an issue as you might think. Most of the teams use Commercial Codes that no doubt they've modified to some degree. Hardware is reasonably cheap these days and you can end up with a 1024 CPU cluster for a few hundred thousand quid at most.

The key to obtaining good results from CFD is to understand the limitations of the model used and to interpret the results correctly - the failure of CFD to predict the performance of the CDG is a good example of that. That's where the teams will spend the money - getting people who can analyse the results and come to the right conclusions.

IMO the first thing to get the right results and than analyze them correctly is to understand at a 100% rate how exactly the SW is working, to know what kind of mathematical model it is used, to know how to impose the right conditions and how exactly that info is processed.

If as you say they use commercial software than they need to have people who take their time to dismantle the SW and understand it perfectly, otherwise it would be only an ordinary tool available to everyone.

Given the implications I , personally, would have had a custom in-house developed SW if I was one of the team managers. A software that will do things exactly as it's needed to be done, and that I can always improve using the latest research results in the domain.

Originally Posted by ioan

IMO the first thing to get the right results and than analyze them correctly is to understand at a 100% rate how exactly the SW is working, to know what kind of mathematical model it is used, to know how to impose the right conditions and how exactly that info is processed.

If as you say they use commercial software than they need to have people who take their time to dismantle the SW and understand it perfectly, otherwise it would be only an ordinary tool available to everyone.

Given the implications I , personally, would have had a custom in-house developed SW if I was one of the team managers. A software that will do things exactly as it's needed to be done, and that I can always improve using the latest research results in the domain.

Unless I could possibly avoid it, I would NEVER go for a bespoke development.

Why would you want to reinvent the wheel?

It makes no sense to ignore the wealth of developed code out there in favour of starting from scratch. All the modelling has been done previously, coded, tested and deployed. It will then have gone through user acceptance testing, bugs fixed and further versions developed.

I would dread to think how much money has be spent of flow dynamics by the oil companies, aviation companies, car manufacturers and the likes but I doubt that there's many models left that aren't covered.

All you need is the relevant Software, appropriate Hardware and people skilled in both using the application and interpreting / manipulating the results to produce meaningful feedback. It's nothing to do with the code or how the SW actually works.

Certainly, no team manager would decide to undertake a £million+ contract to develop a bespoke package from scratch.

No F1 team in their right mind would develop their own code. It takes a long time (i.e. years) to extensively validate a CFD code, and it's often the case that what appear to be good numerical schemes for certain situations don't work well together. Commercial code developers use test matrices that ensure consistent results are obtained when the software is modified so it removes the doubt from the end-user. Of course the results that the CFD spits out could well be absolute nonsense, but at least it'll be consistent nonsense.

I guarantee that not many people who work for motorsport teams know exactly how their codes work; their primary interest is in testing out geometries, looking at lift/drag plots and then sticking it on the wind tunnel model to ensure that the whole process works. They simply don't have time to worry about code development.

Originally Posted by Andrewmcm

No F1 team in their right mind would develop their own code. It takes a long time (i.e. years) to extensively validate a CFD code, and it's often the case that what appear to be good numerical schemes for certain situations don't work well together. Commercial code developers use test matrices that ensure consistent results are obtained when the software is modified so it removes the doubt from the end-user. Of course the results that the CFD spits out could well be absolute nonsense, but at least it'll be consistent nonsense.

I guarantee that not many people who work for motorsport teams know exactly how their codes work; their primary interest is in testing out geometries, looking at lift/drag plots and then sticking it on the wind tunnel model to ensure that the whole process works. They simply don't have time to worry about code development.

Put a lot more suscintly than a muddler like I ever could.

It would be like designing and building a Veyron from scratch for £8M rather than buying one off Bugatti.

(Yes, I know it would cost a lot more than £8M for a single car and that the figure is what each one cost for the excercise)

Originally Posted by Andrewmcm

No F1 team in their right mind would develop their own code. It takes a long time (i.e. years) to extensively validate a CFD code, and it's often the case that what appear to be good numerical schemes for certain situations don't work well together. Commercial code developers use test matrices that ensure consistent results are obtained when the software is modified so it removes the doubt from the end-user. Of course the results that the CFD spits out could well be absolute nonsense, but at least it'll be consistent nonsense.

I guarantee that not many people who work for motorsport teams know exactly how their codes work; their primary interest is in testing out geometries, looking at lift/drag plots and then sticking it on the wind tunnel model to ensure that the whole process works. They simply don't have time to worry about code development.

And what use does consistent nonsense have?
And not knowing how the code they use works means they will get in Honda like situations.
It might be cheap but it certainly isn't top drawer approach, IMO.
It might take time to extensively test and validate the code, but once it's done it's done and it's working as you wish.

I agree that I look to it from the POV of ultimate accuracy and performance, and maybe F1 teams aren't that much interested about this.

Originally Posted by Knock-on

Why would you want to reinvent the wheel?

It's not that simple a case.

Originally Posted by Knock-on

All the modelling has been done previously, coded, tested and deployed.

I doubt it, honestly. There are new theories and models being developed each and every day, that will only be implemented in an industrial code in the next years. What if you could have it a few months before the competition has it too?

Originally Posted by Knock-on

I would dread to think how much money has be spent of flow dynamics by the oil companies, aviation companies, car manufacturers and the likes but I doubt that there's many models left that aren't covered.

Everything is developed for specific uses, and if it isn't than it usually isn't as good as if it was developed accordingly.

Originally Posted by Knock-on

Certainly, no team manager would decide to undertake a £million+ contract to develop a bespoke package from scratch.

They do spend millions to gain an advantage of a tenth of a second over the opposition, why not spend it to get even more advantage. It might be priceless on the long term.

I don't know what they use, I just know what I would do if I had the budgets they have at their disposal.

Just out of interest, have you ever worked in the field of CFD ioan?