Testing Aerodynamics Through Software

[ConnClark]

Just as an FYI for everyone doing CFD. The more correct the data you feed your simulation the more accurate results you will get.


I have been playing with the model of my car to get it right. I just did a check on the underbody airflow. At the bottom of the picture you can see where I have made the front wheel well more like my car. It has less flow separation (and less drag) than the front wheel at the top of the picture. Also the stream lines diverge less from the lower front wheel (also indicating less drag). Small changes can have a big impact on what your being shown and thus the conclusions you make from your model.

[seifrob]

any chance to test virtual AirTabs or other vortex generators?

[ConnClark]

Originally Posted by seifrob

any chance to test virtual AirTabs or other vortex generators?

Once I get my model and simulation to show me similar stuff to what I observe in reality I'll test virtual vortex generators

[KamperBob]

I'm digging your progress and enthusiasm. FYI, at a previous job coworkers did FEM for a living. They lived by the motto trust but verify. Results of any simulation not validated experimentally were taken with a grain of salt. Rock on!

[housedir]

That's really cool!!

[MileMonger]

I was wondering if Shortie, ConnClark or anyone else had any updates of their use with the software. Also brought this thread back up cause Falcon appears to have extended their preview till March 15, 2013. Gonna check it out later today!

You can find it by searching "autodesk falcon." (Can't post links yet.)

[ConnClark]

I'm still plugging away at it. I don't have enough memory to brute force things to get good results so I am tweaking my mesh at key places. My results are getting more believable.

Also here is a validation model with known wind tunnel results.


discussion
TUMWAER: Automotive

more discussion about the models
http://www.aer.mw.tum.de/en/research...otive/drivaer/

even more discussion with an email to request the models in IGES or STEP format.
http://www.aer.mw.tum.de/en/research...vaer/geometry/

a paper with CD and other info
http://delphi.com/pdf/techpapers/2012-01-0168.pdf

[Smurf]

I got it to work.. After you SketchUp your design, go to File, Export, 3D Model.. Save it as a .3DS file. Open Falcon, click the car, import your design.

However, according to Falcon, an S10 like mine (with ridiculous body kit, courtesy of the Warehouse model) lowered with simple cap and angled nose can achieve a 0.37.. I tried it without the nose first, and it said 0.50 CD. Intriguing to play with though, thanks for the find.

[Cd]

I tried the software this morning with a model of a first generation Insight.
I managed to get the numbers pretty correct at a simulated .28, which had me really enthused, but the Cd varies so much each time you run the software you get a different number - actually it keeps going up and down as the software crunches the data.

What am i doing wrong ?

[ConnClark]

Originally Posted by Cd

I managed to get the numbers pretty correct at a simulated .28, which had me really enthused, but the Cd varies so much each time you run the software you get a different number - actually it keeps going up and down as the software crunches the data.

What am i doing wrong ?

Your doing nothing wrong. In simulations you have vortices that form and shed. these give rise to the fluctuations. What you have to do is wait until the oscillations get relatively steady and average the value you are looking for through a few cycles.