Pretty CFD pictures...

[trebuchet03]

So I'm still learning how to use Cosmos.... Last night I did a run, but forgot to set my goals properly. So it calculated flow, but never calculated drag (basically 6 hours of wasted crunch time)

So I'm almost done, with the calculation goals set correctly...


Here's a picture of what came out... Mind you, this is not our final version - we have to make a person and vehicle fit in there properly... Not to mention, address certain issues with that shape.


This is at 50 ft/s - about 34mph - ~55kph.
Based on the solver working right now, there's about 1.4 poundf of drag (if I'm reading correctly). BUT, there's no ground effect (bonus points for catching that before I said it)... I gotta see about making a moving ground for measurements...

[Nerds laugh at me]

Wow -that is fantastic !

Sort of resembles a shoe in its present form ;-)

I remember seeing a show called Airwolf as a kid. The main star of the show had a 'secret aircraft blueprint' that was shown on a computer screen ( really just a basic wireframe ) Ever since seeing that show, I have wanted to do stuff with CFD design.

I have just always been poor in math ( and in the pocketbooks too ) ...so no CFD for me.
I can certainly admire it all though.

You guys are brilliant and I'm glad that I get to converse with you all.
Looking forward to more CFD posts like this one !

[Nerds laugh at me]

How was your mesh created ? Polygons or ...

Yet another question > What was your render time, and on how many machines ?

Oh what the heck. Another question > What would be the minimum system requirements for using a CFD program ?
I know that everyone says to use a network of supercomputers with CFD, but CFD was in use over fifteen years ago.
The Cray supercomputers that were used to render back then are now seemingly outclassed by a mere Playstation, so I would think surely you could render from home on a single computer ( Perhaps a quad core with lots of RAM )
I know it might take days or even over a week to run a calculation, but it can be done right ?

[trebuchet03]

Originally Posted by Nerds laugh at me

How was your mesh created ? Polygons or ...

Yet another question > What was your render time, and on how many machines ?

Cosmos does the mesh for you -- so you never have to look at it, unless you want to. The software crashed when I tried to view them (I've got a lot of crap running at the moment though) - but the peek I got looked like hexes or tets. I am assuming a "fluid cell" is Floworks speak for element - of which there are ~2.3million.

Processing... About 6.5 hours on 1 machine set to 1 step below max refinement (7 out of 8). It isn't as intensive as Fluent (another CFD suite available to me at school) - but to use Fluent, you need to manually create meshes in Gambit (which is one PITA and a half). The results you see are interpolated (estimated between actual calculation points) to give nice smooth curves - otherwise, you'd see blocky bits which aren't as visually appealing

I have just always been poor in math ( and in the pocketbooks too ) ...so no CFD for me.
I can certainly admire it all though.

I was actually amazed at how easy it was to work with this software... You don't have to create the mesh yourself and setting up a solve run is literally a wizard It's almost foolproof

[Nerds laugh at me]

Ohhhh crap !
Now I really want a good CFD program !

But is the program available to students ? ( Or must you use it in lab only ? )

[skewbe]

That does look like fun. Anything good in the public domain for 3d modeling and cfd?

[omgwtfbyobbq]

Openflower and gmsh seem to be the most popular.

[Snax]

Nice!

As for the design, it looks to me like if you could delay seperation of airflow around the mid-height of it, you could drastically reduce the downflow off of the top. Either that, or encourage seperation sooner off the top of it.

[trebuchet03]

Originally Posted by Snax

Nice!

As for the design, it looks to me like if you could delay seperation of airflow around the mid-height of it, you could drastically reduce the downflow off of the top. Either that, or encourage separation sooner off the top of it.

So here's the unexpected part... We got the cD data out of the CFD model... and apparently it's .14 BUT, there's no ground effect, no wheels etc.

But ya, it's likely we'll be adjusting some angle to keep laminar flow... But we need to get some Reynolds number information first, to make sure we're in the laminar regime in the first place