Check this link out:

http://www.turbosquid.com/FullPrevie....cfm/ID/213499

It's 140 bucks, but it seems to be a pretty good model. Can your software convert from .3ds?

BTW, do you have a trial version of the CFD software?

 

Wow awesome! I'd really like to see this thing develope!

 

Here's another one:

http://www.3dcadbrowser.com/preview.aspx?ModelCode=6139

Most of these formats have the model rendered in polygons, for games/graphics. Should be possible to work with for CFD...

Craig

 

i dont want to sound like a dick but all this really seems to do is show some aerodynamic traits of a block of wood that kind of resembles an s2000.

you dont have anything that accounts for the air that goes through the radiator, as well as your rear bumper is smoothly attached to the bottom of the car. in reality it catches air lick a small parachute. you have no way of accounting for all of the various shapes that stick down from under the car.

i think at best this model shows us the flow over the top of the car, but that doesnt really mean anything. in reality with the soft top, its not that smooth, it has waves built in, and so forth. and what about the air that flows in through the radiator. it is basically forced out of the bottom of the engine bay,once it collides woth all of the stuff in the engine bay as well as the fire wall.

again, im not trying to sound like an asshole but i dont think that this is in any way an accurate depiction of the air flow.

maybe i'm wrong. i would love to see this thing developed further, but unless there is underhood detail and much more, it doesnt really help anything.

please prove me wrong

 

Dude, it was just a hack. Something he was playing around with. That doesn't mean you can't learn something from even a simple model, but you have to remember that it is just a simple model.

I'm sure the Honda engineers have models that account for all the secondary flows and the like. Even then, though, no model is perfect. You always have to remember what the limitations on the model are, and only ask questions that make sense within those limitations.

 

Well, it means, of course, that the flow goes over the top of the car, for one thing. And the comparison of top up/down flows is really quite interesting, and certainly valid to first order. To the extent that Slim's able to refine the modeled car's geometry, even this level of simulation will continue to produce interesting results, and I'd like to see them. (I didn't think of it before, but the business about top-down/windows-up would be even more interesting that refining the windshield shape.) Getting flow through the radiatior right is something that would be a neat trick -- it would depend on how the CFD package allows (or doesn't) flow obstacles that are porous.

Aside from those technical details that I'm guilty of having started the discussion of, the reasonableness of simulations like this depends on what scale you're looking at -- how small are the wiggles in the flow field that you're concerned about relative to the scale of the object (and, of course, to the mesh size, which I don't think we've seen). What he's shown seems just fine. On the other hand, if he puts on a spoiler, there really won't be a steady flow to show -- the turbulence (assuming it emerges from the model) will be intermittent and the only real way to depict it is with a movie or with statistics.

And, Mike, if this is just a hack, then these guys are really good! HPH

 

If the desired information is just the external vehicle aerodynamics, you can get an 85-95% answer without worrying about underhood flow, small variations in the top shape, back bumper, etc.... However, it is important to get the vehicle's outer mold line geometry correct to begin with. In this case, I think we're just looking at general flow trends around a generic S2K-shaped class of vehicle. To get my 85-95% answer, we'd need an accurate S2K geometry and a full N-S simulation.

I think slimjim was pretty clear about the approach up front, though, so nobody should be expecting more than what he detailed.

Craig

 

i see what you mean, and again i wasnt trying to shit in anybodies frruitloops, i just think that the air flowing through the radiator is a big deal. the air that enters the radiator is at probably the highest pressure, and i would imagine that it accounts for a great deal of drag. it scatters about in the engine compartment, and has a ton of stuff to collide with.

i would be interested in seeing an example with spoilers of different heights and angles, and then run them again with the oem hardtop, or mugen hardtop, and then run them again with the mooncraft hardtop. i think the moon craft hardtop would show a significant advantage. it would be similar to the top up/ top down comparison in my opinion. the angle of the mooncraft hardtop would let the air meet the spoiler a little less turbulent. at least thats my theory.

or at the least i think you could mount the spoiler a few inches lower and still get the same results as the oem top with the spoiler sitting a few inches higher.

 

and also, i just watched a video about the 3d modelling they do for ALMS corvette. they ca account for every nut and bolt that in anyway effects airflow.

but, that costs about .5 million bucks.

maybe i was expecting way to much

 

Randazzo?


There is a huge difference with the windows up. I'll get to that soon hopefully.

Fooled me

Read my earlier posts. Modeling the splattered bugs on the headlights would take eons

A hack at best. Like I said, this took 20 minutes. I am more interested in the basic shapes right now. And frankly, while it doesnt look as sweet, if the general shape is similar, the airflow will be similar. Plus I can turn these around a LOT faster than a more complicated model.

I'll post top-down, rear wing in a few minutes.

Just bear with me here. These are nothing more then cool things to look at that show...oh yeah, thats why the air does that. Food for thought only. This is in NO way a 100% accurate representation of the S2000. It's only as accurate as my model (which sucks) and my mesh sizes (which suck). But I'd rather throw these simpler analyses up at a rate of two per day than one per two weeks.