Manor and the winds of change...

[egomeister]

double6 said:

I can just imagine the conversation at the first race ....

Race engineer: can I have the offsets for the front wing gurneys
Aero guy: yeah here you go..
Race engineer: that’s great but what about the clr offset...doesn’t the front wing gurney effect the rear of the car ?
Aero guy: not in a front wing only cfd model it doesn’t
Race engineer: okay can i have the ride height map
Aero Guy : yeah here you go
Race engineer : but theres only 2 points on the map
Aero Guy : do you know how difficult it is to change the ride height in a CFD model

I think you are dramatising a bit too much there

In a lot of ways cfd would be easier for data like this. It's a damn sight easier to run a cfd case before a race than it is to revert a wind tunnel model to the spec it was weeks/months ago!

It's not that its difficult to make these changes in cfd, more the time it takes to solve each case (ie, each ride height etc) - a whole lot longer than a 15min tunnel run!

[Alfie Noakes]

If they only use CFD it'll give Cosworth and excuse.

[entropy]

egomeister said:

entropy said:

Nick Wirth has his own R&D company and designed this year's Acura LMP1 mainly with CFD.

But the aero on an LMP is nothing compared to that of f1. I just can't see cfd providing accurate enough results to develop a car to the level of detail seen in f1.

So which is better? Wind tunnel or CFD? Aren't they both just as bad since simulations and the real world can have great differences?

[double6]

egomeister : for a full car CFD analysis 'most' of the f1 teams take about 24hours to run the solver, that's excluding gridding and CAD time. so for a ride height change you are probably looking at 3-4 days minimum to do a single ride height point....seem a lot when you are just looking at say a 5mm gurney change wouldn't you agree ?

a 15 min wind tunnel run will give you about 20 points on the ride height map...

Don't get me wrong CFD plays a major part in design and development...I just can't see how they would get enough information from it quick enough to be useful. In my opinion CFD is fantastic for developing a concept and analysis of the detailed flow structure around a car, it can help you understand what's really going on, not just guessing from a few numbers from the wind tunnel.

However wind tunnel data fills in the gaps for the multiple ride heights, wing sweeps, steered wheel, roll and yawed cases that can be really tricky to do in CFD.

I don't think the guys at wirth's are daft...I suspect they will do a lot of their mapping work in straight line testing or even at the track...which is also invaluable to back up the wind tunnel and CFD numbers.

and yes I was dramatising a bit

[egomeister]

entropy said:

egomeister said:

entropy said:

Nick Wirth has his own R&D company and designed this year's Acura LMP1 mainly with CFD.

But the aero on an LMP is nothing compared to that of f1. I just can't see cfd providing accurate enough results to develop a car to the level of detail seen in f1.

So which is better? Wind tunnel or CFD? Aren't they both just as bad since simulations and the real world can have great differences?

For evaluating broad concepts, or more specific simulations involving heat as well as airflow: CFD
For general aerodynamic development, and small iterative changes: Wind tunnel

[egomeister]

double6 said:

egomeister : for a full car CFD analysis 'most' of the f1 teams take about 24hours to run the solver, that's excluding gridding and CAD time. so for a ride height change you are probably looking at 3-4 days minimum to do a single ride height point....seem a lot when you are just looking at say a 5mm gurney change wouldn't you agree ?

a 15 min wind tunnel run will give you about 20 points on the ride height map...

Don't get me wrong CFD plays a major part in design and development...I just can't see how they would get enough information from it quick enough to be useful. In my opinion CFD is fantastic for developing a concept and analysis of the detailed flow structure around a car, it can help you understand what's really going on, not just guessing from a few numbers from the wind tunnel.

However wind tunnel data fills in the gaps for the multiple ride heights, wing sweeps, steered wheel, roll and yawed cases that can be really tricky to do in CFD.

I don't think the guys at wirth's are daft...I suspect they will do a lot of their mapping work in straight line testing or even at the track...which is also invaluable to back up the wind tunnel and CFD numbers.

and yes I was dramatising a bit

Sounds about right to me!

I don't think changing something like a gurney would take so long though. It's about 5 minutes CAD time, only a small mesh rebuild, then the solve time - which is the killer, hence the big teams investing in supercomputers! It would take ages to the full data map you can generate in the tunnel using cfd methods, but in a lot of cases you don't need all that data to see if an avenue is worth following.

CFD does offer huge flexibility though - you can do changes that would require a significant amount of work to the tunnel model, just as simply as changing a wing angle or a bargeboard.

[hairyben]

I thought cfd was good for exploring different directions, brainstorming perhaps, in an environment that's low-cost, instantly reversible, highly changeable and the windtunnel lets you verify whether or not the computer was talking st?

[Project 644]

Marcia said:

Project 644 said:

I wonder if the Duke of Rothsay will step in and settle the arguement about which aero is best?

He's dissapeared of the radar,not posted anywhere for months.

I believe he/it/she was arguing in "that" aero thread with a guy who's login is Navier_Stokes. He/it/she might have had a quick wiki on the name. Either way I haven't seen him/it/her since. I did like the way he/it/she found out what flying experience the real DoR had and put that in as his/its/her experience. A nice touch lost on most I thought.

[10 Pence Short]

Project 644 said:

Marcia said:

Project 644 said:

I wonder if the Duke of Rothsay will step in and settle the arguement about which aero is best?

He's dissapeared of the radar,not posted anywhere for months.

I believe he/it/she was arguing in "that" aero thread with a guy who's login is Navier_Stokes. He/it/she might have had a quick wiki on the name. Either way I haven't seen him/it/her since. I did like the way he/it/she found out what flying experience the real DoR had and put that in as his/its/her experience. A nice touch lost on most I thought.

I never really thought that deeply about him. I just thought he was a tw*t.

[skeggysteve]

Both CFD and windtunnels use a computer to give you the results.

What I can't understand why they rely on the computer.
Computers don't have the power to 'think' they only do what they are told to do - are programmed to do - so surely whoever programmed the computer software has decided what the results are going to be?

I understand that the software can/will be updated with real results but I just don't get the total reliance on CFD or wind tunnels

Any computer experts care to explain it to us?

...............................
If I remember correctly the first time a real driver took the 917 down the Mulsanne and said how unstable it was the Porsche engineers said the computer said it would be alrigth - the driver replied let the fking computer drive it then!

[groomi]

skeggysteve said:

If I remember correctly the first time a real driver took the 917 down the Mulsanne and said how unstable it was the Porsche engineers said the computer said it would be alrigth - the driver replied let the fking computer drive it then!

I imagine he meant drawing board...

[egomeister]

skeggysteve said:

Both CFD and windtunnels use a computer to give you the results.

What I can't understand why they rely on the computer.
Computers don't have the power to 'think' they only do what they are told to do - are programmed to do - so surely whoever programmed the computer software has decided what the results are going to be?

I understand that the software can/will be updated with real results but I just don't get the total reliance on CFD or wind tunnels

Any computer experts care to explain it to us?

...............................
If I remember correctly the first time a real driver took the 917 down the Mulsanne and said how unstable it was the Porsche engineers said the computer said it would be alrigth - the driver replied let the fking computer drive it then!

CFD is only an application of physics - a simulation. While admittedly the result you get is only as good as what you put in, CFD is a pretty mature simulation technology now and it is quite possible to build a competitive racecar solely using cfd simulations as the Acura demonstrates.

You say computers don't have the power to think, but it is getting that way now. You can perform cfd where you give the software a number of design parameters and a results target and it will work its way to picking the best solution for the criteria you have chosen.

A windtunnel only uses a computer for datalogging / processing. I don't see why you wouldn't rely on such things. You could take all the measurements mechanically and eliminate the computers, but I have no idea why you would do so.

As an aside, does anyone know how much grunt the Wirth cfd computer has?

[skeggysteve]

groomi said:

skeggysteve said:

If I remember correctly the first time a real driver took the 917 down the Mulsanne and said how unstable it was the Porsche engineers said the computer said it would be alrigth - the driver replied let the fking computer drive it then!

I imagine he meant drawing board...

Iain, I know what you mean but I'm sure I've read it somewhere but I can't find it - maybe just my old age memory

Edited by skeggysteve on Sunday 11th October 22:24

[Tripe Bypass]

skeggysteve said:

groomi said:

skeggysteve said:

If I remember correctly the first time a real driver took the 917 down the Mulsanne and said how unstable it was the Porsche engineers said the computer said it would be alrigth - the driver replied let the fking computer drive it then!

I imagine he meant drawing board...

Iain, I know what you mean but I'm sure I've read it somewhere but I can't find it - maybe just my old age memory

Edited by skeggysteve on Sunday 11th October 22:24

Frank Gardner said it.

Edited by Tripe Bypass on Sunday 11th October 22:54

[skeggysteve]

egomeister said:

a competitive racecar solely using cfd simulations as the Acura demonstrates.

You say computers don't have the power to think, but it is getting that way now. You can perform cfd where you give the software a number of design parameters and a results target and it will work its way to picking the best solution for the criteria you have chosen.

A windtunnel only uses a computer for datalogging / processing. I don't see why you wouldn't rely on such things. You could take all the measurements mechanically and eliminate the computers, but I have no idea why you would do so.

As an aside, does anyone know how much grunt the Wirth cfd computer has?

Stephen, thanks for replying.

Agree with you about the Acura - do you follow ALMS? Last nights race was a cracker!

Interesting stuff you say about computers getting close to thinking - is that down to the power/number crunching per milli/nano second thingy?

"for the criteria you have chosen"
Thats the bit I just can't get my head round - to put it in simple terms have computers advanced from the days of bullst in = bullst out?

Re: Windtunnels - I was very lucky to go on a Lola open day a couple of years ago. We got a demo of thier windtunnel in action. Very impressive bit of kit.
But all the data was being collected by computer, no bits of string or fag smoke to be seen.

I suppose my question is - just how much can you trust the computer?

I guess the answer to my question is: About the same as much you can trust a driver

[philis]

zac510 said:

philis said:

Wind tunnels are a waste of time for the spectator, yes they make the car go faster round corners, but it always results in the cars behind going slower with the whole "dirty air" conundrum.

So whats the point in racing?

The point of course is to get your car to the front, to win the race, to win the championship and all the fruits that go with it. They're not entirely just there to please you, despite what you seem to think!

Big Ross seems to think pleasing the spectators is fairly important too:
http://www.itv-f1.com/news_article.aspx?id=47125

[AndrewW-G]

egomeister said:

As an aside, does anyone know how much grunt the Wirth cfd computer has?

Cant comment on the Wirth Computer, but here's something on Albert2 BMW's CFD box from 2 and a bit years ago (applying Moores law would see it up by almost 75% today)

http://news.zdnet.co.uk/hardware/0,1000000091,3928...

[egomeister]

skeggysteve said:

Agree with you about the Acura - do you follow ALMS? Last nights race was a cracker!

Nope, didn't see this weeks race they do have a knack of producing some good racing over there though!

skeggysteve said:

Interesting stuff you say about computers getting close to thinking - is that down to the power/number crunching per milli/nano second thingy?

It's partly down to advances in software, and partly hardware - ie you can set the thing running and churn through a load of cases much quicker than you a few years ago. This makes an adaptive simulation much more viable.

skeggysteve said:

"for the criteria you have chosen"
Thats the bit I just can't get my head round - to put it in simple terms have computers advanced from the days of bullst in = bullst out?

Not really, the computer is only carrying out instructions from the engineer, so if there is a problem with what you tell it to do, there will be a problem with the solution. The trickiest area on cfd is probably turbulence modelling as this is often not a steady state condition, hence the software may only solve for snapshot of whats going on. This is why you need to correlate your results with actual testing whether it be wind tunnel or a full size test.

skeggysteve said:

Re: Windtunnels - I was very lucky to go on a Lola open day a couple of years ago. We got a demo of thier windtunnel in action. Very impressive bit of kit.
But all the data was being collected by computer, no bits of string or fag smoke to be seen.

I suppose my question is - just how much can you trust the computer?

Why wouldn't you trust the computer!? They've been around for a few years now you know! Are you scared of the computers in your road car? It wouldn't be feasible to collect the amount of data logged by the computer using manual methods, although you'd be surprised how often bits of string are still used to help visualise whats going on rather than just relying on the numbers!

[Finchy172]

Wirth runs his own CFD and research company.
This years Acura LMP1 car was solely built on CFD and has acheived very good results.

[Alfie Noakes]

de ja vu?