CFD - Computational Fluid Dynamics, Motorsport, Formula 1

[miqi23]

Hey,

I'm new to the forum as well. I'm an Aerospace engineering junior at Cal Poly San luis obispo, and I'm the Aero lead for our Formula SAE team. This is the first year Aero is being done (I literally pulled a team together in the past 2-3 weeks) and I'm trying to find the best way to make use of CFD to cut down testing time. I've been learning Floworks, and it seems relatively user friendly, but I'm curious as to you guys' thoughts on how accurate and applicable it is/would be to wings and underbody elements of a formula style car at low speed and low reynolds numbers (Less than 1 million.)

I'm especially concerned with its ability to model the ground effects on the front wing and the underbody. How can the stationary ground be modeled with respect to the moving undertray? Any suggestions? I've heard it can be done in Fluent, but that it can be complex and quite difficult. While we have access to Fluent and Gambit, and the computing resources as well, we don't yet have the experience to use either. Also, I've run the 2 element model on page two of this thread, and I keep getting stall on the trailing edge of the first element, which generally agrees with what theory would suggest. Does anyone care to comment? I adjusted the minimum gap size and wall thickness to correspond with the features of the geometry, but had to modify the gap to be larger as gap thickness for the smaller spacing caused insane run times. I tried more advanced manual meshing/refinement, but the calculations were going to take longer than I would have the computer for. My results are shown below. Note the quality is poor as the original bmp was changed to a jpeg. Thanks in advance for any comments, inputs, or advice you may have.
[IMG:1071:726]http://i23.photobucket.com/albums/b397/ ... oodCFD.jpg[/img]

In FSAE or Formula Student or whatever they call it in different countries, I believe their is no importance of aerodynamics since the speeds are very low, however its worth while investing some time to understand the basics of it by applying it to the FSAE car.

keep it simple. Use basic principles like keeping the flow attached as long as possible and analysing the wing section at differernt heights above the ground to sort of simulate vibration etc.

Use standard wing sections like they do in F1 for the main element and flap. Try select one of the NACA sections since wind tunnel data is available and you can easily select the one that gives the best lift to drag ratios and suits well the circuit you are going to race on.

I think its part of your majot project as well, so going through design practices like this would gurantee you good grades and the same time you may produce a better car.

Hope this helps and good luck.

P.S. Since your run times are long, try analysing your wing in 2D rather than 3D.

[AeroGT3]

In FSAE or Formula Student or whatever they call it in different countries, I believe their is no importance of aerodynamics since the speeds are very low, however its worth while investing some time to understand the basics of it by applying it to the FSAE car.

keep it simple. Use basic principles like keeping the flow attached as long as possible and analysing the wing section at differernt heights above the ground to sort of simulate vibration etc.

Use standard wing sections like they do in F1 for the main element and flap. Try select one of the NACA sections since wind tunnel data is available and you can easily select the one that gives the best lift to drag ratios and suits well the circuit you are going to race on.

I think its part of your majot project as well, so going through design practices like this would gurantee you good grades and the same time you may produce a better car.

Hope this helps and good luck.

P.S. Since your run times are long, try analysing your wing in 2D rather than 3D.

Aerodynamics actually are of great benefit, even in FSAE. The cars that run aero dominate the autocross and skidpad competitions, and with cars as low as 300 lbs (like ours), 100 lbs of downforce at 40 MPH makes a huge difference (+30%!!) in terms of traction. Other teams have posted numbers and actual data, and have proven their cars to do considerably better than 100 lbs of downforce. With our car being extremely powerful with respect to the competition, grip is the only thing missing in making a first place car.

Standarf airfoils would be a great place to start, but unfortunaly we have two important requirements for which there aren't many solutions. Unlike F1, our Reynold's number is low - under 600,000. Which means that NACA and other published data don't really fit us well and won't perform like they do on F1 cars and aircraft. Secondly, we need high lift. Many of the low Re airfoils available are for gliders and model airplanes, which produce relatively low Cl's. We need Cl's of 2.5-3+, which means we need multielement wings, but also high lift airfoils. Standard airfoils are limited in this range. Another factor is that NACA airfoils are generally symmetric or slightly cambered. We need supercritical airfoils to deal with the huge pressure gradiants we'll see on such high lift wings. As you know, adverse pressure gradients are a cause of stall, which we very much want to avoid. I am writing vortex panel code that will simulate our wings much like xfoil, etc. But my program will model ground effects, multi element airfoils, and most importantly, be able to perform computerized optimization.

Unfortunately, panel methods aren't accurate at high angles of attack. So, while we can use my code to weed out bad designs at low angles of attack, we need CFD to verify stall angles, etc.

I am very interested in 2-D cfd, as I'd actually like to compare inifnite wings (airfoils) to historical data, i.e. test a NACA in Floworks and see how accurate it is, and how well I can use the software. I have searched the help files, but come up with very litte. Could anyone chime in with advice on how to do 2-D CFD in Floworks? Many thanks for your help,

[hardhgear]

Hi guys
I think floworks is not suitable package for this type of advandec analysis
which required software like fluent.
at least in floworks u don't have any control over the grid it's almost structured and not fine one.
If i was some correct me.
Thanks

[miqi23]

Dear All,

I was wondering if it would be possible to make this thread lively once again. I have read all the messages on this thread and looks like it used to be very lively once with a lot of contribution being made by every one.

However, I am proposing that we include proper CFD analysis based on general Formula 1 aerodynamics and most importantly the analysis of the changes being made by the teams both in terms of Aerodynamics and may be heat transfer.

A good example would be an analysis of the Ferrari's front wing both fixed and while flexing at X m/sec. Or analysing the change flip up makes or etc etc.

I would need some volunteers to start this off nicely. I am aiming that we work as a proper team and learn at the same time from each other.

Some one having proper CAD skills would be great. Since I am busy most of the time, a little hand would be appreciated. Anyone interested can just reply on this thread.

For the time being I will make a list of possible projects that we can start with and post it on this thread later on.

Many Thanks.

[patankar]

miqi23,

I have some CFD experience in this area. I use Fluent/Gambit. I have a cluster at work I can use. I am interested in analyzing F1 technology, and I am willing to do some work on it! What geometries do we have so far? What were you thinking of working on.

Cheers

[AeroGT3]

So I've gotten pretty good with floworks, and have concluded that it just isn't up to the task of our analysis. We need much more accurate results, as we're going to mostly be using CFD in lieu of wind tunnel testing. So, we have opted to go with Floworks as my university has a 100-seat license. Our computers are slow though, so over the summer I will be building a linux workstation. I plan on using my existing 450 W power supply, 128 Meg AGP and whatever audio card I have. I have a few questions though . . .

First, which Linux distribution should I go with. I was thiking Gentoo or Red Hat Fedora 5.
Secondly, what kind of hardware? I plan on going with 4-6 megs of DDR RAM, and a 10k RPM harddrive, but what about processors? I was thinking AMD X2 Dual core over a single core with a higher frequency. Any recommendations?

I hope to be decent with Fluent by next year. Any suggestions on places to learn? I've started out with online tutorials from Cornell and recommendations from a CFD professor. BTW I have had all prerequisite courses in external aerodynamics and compressible flow. I could take grad level CFD courses if they were being offered right now.

[hardhgear]

From my little information
As more memory you will get the more number of cells in the grid and as mentioned before in this section
of the forumFluent needs 1 Mb of ram per 100 cell about the processor of course AMD X2 is one of the
more powerfull processors in the market now it takes the best processor at tomshardware for 2005
Also I found alot of company that sell pc for CAD/CAE application uses AMD Like http://www.xicomputer.com/
so the more powerfull processors the faster solution u will get
But try to find fluent 64-bit version

[AeroGT3]

Thanks for the info and the lead . . .

I am trying to get 32 bit fluent for my current windows computer, and I will try to get the 64 bit version for my new computer as well. Anyone have suggestions about what linux to run?

[miqi23]

Thanks a lot every one for the reply. I am little busy now days with some work, but I am trying to sort the work load asap. Will let you know when I am done!

Cheers

[hardhgear]

AeroGT3

Thanks for the info and the lead . . .

I am trying to get 32 bit fluent for my current windows computer, and I will try to get the 64 bit version for my new computer as well. Anyone have suggestions about what linux to run?

I think u can use Suse linux o redhat

[hardhgear]

AeroGT3

Thanks for the info and the lead . . .

I am trying to get 32 bit fluent for my current windows computer, and I will try to get the 64 bit version for my new computer as well. Anyone have suggestions about what linux to run?

I think u can use Suse linux o redhat

[RACKITUP]

Thanks for the info and the lead . . .

I am trying to get 32 bit fluent for my current windows computer, and I will try to get the 64 bit version for my new computer as well. Anyone have suggestions about what linux to run?

I use hummingbird to get gambit to work, and as Gridgen and Fluent 6.0 are windows based (my version at least) its a piece of piss to set up

[AeroGT3]

Does anyone here write their own CFD grid generators or solvers, for fun sorta?

I've been playing around writing my own. I've done 2-D Euler solvers and Thomas-Middlecoff Elliptic grid generators, and a hyperbolic BL algorithm that grows an elliptic grid outside the BL.

I want to try an unstructured grid generator. Anyone have experience writing these, or know how difficult it is?

[cfditya]

Dear All,

I was wondering if it would be possible to make this thread lively once again. I have read all the messages on this thread and looks like it used to be very lively once with a lot of contribution being made by every one.

However, I am proposing that we include proper CFD analysis based on general Formula 1 aerodynamics and most importantly the analysis of the changes being made by the teams both in terms of Aerodynamics and may be heat transfer.

A good example would be an analysis of the Ferrari's front wing both fixed and while flexing at X m/sec. Or analysing the change flip up makes or etc etc.

I would need some volunteers to start this off nicely. I am aiming that we work as a proper team and learn at the same time from each other.

Some one having proper CAD skills would be great. Since I am busy most of the time, a little hand would be appreciated. Anyone interested can just reply on this thread.

For the time being I will make a list of possible projects that we can start with and post it on this thread later on.

Many Thanks.

i think i can bring out some CAD models of some aerodynamic elements used in F1 cars........but since i dont have raw dimensions of elements wht do i do.......do i assume the dimensions and model them??????

[miqi23]

Cool, that would be good. Post some images first and see what others think about it. If it is worth while the thing, we can get it done shortly;)