Digitising Race-car for Aero analysis

[speed_12]

Hi,

I'm in the process of organising a dissertation project with a team I help with at race-weekends (I'm not going to give specific team/car info on here) to do some Aero analysis on their race-car.

The trouble is there may not be a digital model available to use and so I would have to make one myself.

I will be talking to my tutors at Uni and the team a lot to try and find the best way of modelling the car, but if anyone has any good tips then please fire away The CAD program I would be using will either be Solidworks or ProEngineer.

Cheers

[Rubinho]

When I wanted to do something like this I looked at PhotoModeler. This is Photogrammetry software which will build a 3D model of an object from photographs of it taken at different angles. It does this seemingly impossible feat by having intimate knowledge of the camera's optics and being calibrated against an optical scale. You assist the process by laying out the images and defining control points on the pictures. It outputs to various 3D image formats (IGES etc.)

I successfully used the demo version which is locked to using a set of demo pictures but I couldn't afford the full package ($895 + the cost digital camera with quality optics). I was going to do a similar (sounding) thing to you; create good quality, full car models of racecars and feed them into a CFD package for aero analysis.

[jhakewill]

There was an article about this a couple of months ago in Racecar Engineering - specifically about a service to do this for you I seem to recall.

Since this is something that happens for 3D games (although perhaps not at the level of accuracy you require), you might try asking on some 3D-modelling or game development forum!

James

[Locost47]

Looks like i've been beaten to it but i would also suggest scouring the net for 3D car models. Places like Turbosquid or 3dcafe offer extremely high resolution models of all sorts of road and race cars for really not a lot of money at all. A 500,000 polygon model of a McLaren SLR is about £28, for example, though many of the lower-resolution models are free and could still be a useful starting point if you've got no data at all.

Bear in mind that these models are often created by hand from sketches or measurements taken from die-cast models etc but if you shop around you should be able to find a decent representation of something resembling your car.

The models are usuall facetised for use in games or rendering packages but many are available in STL (stereolithography) format which will read into most CAD packages and CFD pre-processors (or you could cheat and use one of the 5 saves from an evaluation licence of Rhino to read in one of the fancy rendering package formats and write it out as STL).

You will have to do some clean-up before meshing the models though because they have been constructed for a different purpose. If you pick the right one then you'll just need to patch some holes and connect some edges to make watertight volumes and you're away.

Good luck!

[avsfan733]

some of the laser systems are very good. Do a little poking around see if anyone locally has one... product design firm maybe? your going to need a seriously dense points map. probably only want to do one side of the car though and then mirror it to reduce your area.

Are you looking to do engine compartment and underbody flow also or are thos going to be ignored and flat respectively?

[jhakewill]

The article in Racecar Engineering was in the December 2005 issue.
Cheers,
James

[Locost47]

Companies like 3D Scanners wll laser scan half a car (upper exterior surfaces only) and provide you with a decent surface mesh for about £5.5k. You could push the academic study side of things and try and get a discount or freebie (perhaps in exchange for them using the results for marketing purposes) but scanning whole vehicles isn't cheap. The results can be excellent though so if you can afford it, it's the best route.

[speed_12]

Blimey, I hadnt realised it would be quite as much as that!

The only problem I can see is that I will really have to model the underside of the car aswell as I am going to be analysing a front splitter (again, im not going to go into car details for confidentiality reasons) and so need to know what the air flow is doing above and below the car.

Actually im not sure why i didnt mention that to start off as it may have some bearing on what needs to be done. I'm thinking that just the front of the car could be modelled as the effects on the rear are limited....but then again there are effects on the rear so maybe a full car is best......blimey this is getting complicated!!

I will dig out the racecar article tonight and have a gander at it, but if anyone has anymore ideas go for it! Cheers to all those who have posted so far, keep it coming!!

[Locost47]

If you're mainly focussed on a splitter then you may not need the detailed geometry of the car, just the general shape, for an initial flow analysis at least.

What happens above and below the car is important, but not critical if the main thing you're interested in is maximising the downforce produced by the splitter or it's associated efficiency using CFD.

Details of the cooling system will be important though, as that will be a dominant influence on what the flow is doing in front of the bumper.

[speed_12]

Ah, that's brightened my day up a bit, I was kind of wondering where I was going to find £5.5k!

If its just a simplified model then It would be time consuming but not too difficult to knock up a CAD model.

I've been asked to come up with some analysis and ideas of improving the front splitter for higher front downforce.

From what I've read around teh subject so far, obviously the idea of the splitter is to tap into the high static pressure build up in front of the car which then gives low static pressure underneath the splitter (giving downforce).

So could I get away with just modelling the front 1/4 to 1/2 of the car, simplifying it (i.e. not model every single curve etc) and then running a simplified airflow over that model?

Also, what details of the cooling system would I need? I'm presuming its to do with the air temperature around the radiator and the effect that has on drawing air through the radiator/front of the car but please correct me if I'm talking rubbish!

[Locost47]

It's probably best to model the entire body of the car but the level of detail need not be huge. For a splitter the main effect of the rest of the body would be to determine where the stagnation point was on the front of the car. This is normally somewhere around the front bumper / number plate position on a saloon-type car but will migrate downwards when you start putting spoilers / splitters on.

It will also change in position depending on the amount of upwash generated at the rear of the car from diffusers or wings and that in itself is dependent on whether you're running a low or high downforce package so, IMHO, the exact location isn't so critical to pin down, meaning that the geometry of the rest of the car can be fairly basic but you do want it all there.

The cooling system will also dramatically effect the location of the stagnation point at the front and generally how the flow behaves in that area. It's nothing to do with the temperature, just how easy or difficult it is for air to escape the front of the car by running through the cooling system. Many race teams don't seem to manage the cooling airflow properly and it reduces downforce and, more critically, ultimate cooling capacity.

You'll need to model whether the radiator is sealed to the front bumper or not and whether it escapes out through the wheelarches, through vents in the bonnet or out of the bottom of the engine bay. Radiators are normally modelled in CFD as porous media, and you should find some guidance for setting them up in the users' guide of whatever software you're planning to use.