CFD Online Discussion Forums

CFD Online Discussion Forums (
-   Main CFD Forum (
-   -   Contradictory Results from different software (

Yogibear October 19, 2011 14:20

Contradictory Results from different software
If ive put this in the wrong place ill apologise wasent sure where it should go.

Over the past 3 years ive done a lot of work with Star CCM+ whilst at University doing reasearch into racing car wings.

Ive recently started trying to use Solidworks Flo-simlation to get results as well is producing the designs and to to a degree validate my previous reasearch. I have set the simulations up identically but am getting very different results , in the region of 50% difference between the results from Star CCM and Flo-Simulation.

Feedback from other students is that Flo-simulation isnt as good as Star CCM but im strugling to believe that the results can be that different. If anyone could offer some advice it would be appreciated.

giuli@ October 20, 2011 10:17

...same number of cells?? same wall resolution?? same wall function?? same turbulence model?? same numerical scheme/order of accuracy??

ChunkyPastaSauce October 20, 2011 18:41

You should have some experimental data to check your results. Without it youre unlikely to know if you're getting correct results (even if the two packages get the same results). Optimally the data should come from your specific item you are doing analysis on. If that cant be had then search papers where someone has done something similar, simulate that and check to see if the results a reasonable. If so, proceed with your geometry.

Kevin De Smet October 22, 2011 05:58

Yeah, it's always the easiest thing to say "flo isn't as good as star" that isn't an answer, it's a statement. While true it may be that star calculated additional physics that flo doesn't and that it is important for your analysis, you should first check to check all of giuli's remarks.

Yogibear October 24, 2011 09:28

The experiments ive undertaken are very simple , a NACA 0012 wing in a wind tunnel with the same velocity and angle of attack. Ive used this as ive allready done a lot of work in this area and have verified results from using star ccm. The results matched publiished results very closly and the generated CL and CD were almost perfect.

Ive run the same experiment in flow-works and the results are wrong and are quite obviously wrong from looking at the visualisations its quite clear to see.

Pressure Star CCM+
Pressure Solidoworks Flo-works
Velocity Star CCM+
Velocity Solidoworks Flo-works

Kevin De Smet October 24, 2011 12:22

RANS codes don't handle these kinds of problems well, Floworks is a RANS turbulence solver only. It has problems especially where the re-attachment zone is.

Yogibear October 24, 2011 18:32

OK, im guessing Flow-works isnt the one to use for this kind of analysis then.

Is their anything that can be done to improve it or is it simply the case of using different software.

giuli@ October 25, 2011 03:44

I think RANS modelling is good enough for 2D airfoils, and different solvers should give very similar results.
Did you consider the questions I was asking above? I think the answer of your problem is there..
Just a suggestion, if you want to compare results, use the same scale-range.

Try openfoam, it is for free and it makes you be aware of the conditions you put


Yogibear October 25, 2011 09:55

Thanks, I am wanting to try open foam but im waiting for a second PC to run it on as I dont want dual operating systems on my workstation.

When I had set up the run I had tried to match all of the conditions that I could between both software's but was unable to find corresponding options for everything in flow-works. Was wondering if their were some hidden advanced menu's in their or what you see is what you get.

Simply wante to increse my work throughput as I can do things in flow-works far faster than in ctar-ccm so if the results were going to tally was going to use flow-works.

Kevin De Smet October 25, 2011 13:53

Flo is good really, it's just very dependent on the problem. Unfortunately many times problems are not easy to identify unless you have previous knowledge into the kind of models you are simulating.

All times are GMT -4. The time now is 13:54.