Validation of Star CCM+ simulation?
 

Hi guys,
I'm doing a project on drag reduction of bluff bodies like trucks and buses, and running the simulations in star ccm+. There don't seem to be any big problems with the simulations because I get drag values around the typical values for trucks. But, my supervisors keep requesting me to validate the results.
I wanted to know if you guys think its really necessary to validate results obtained by star ccm, as I'm not writing my own solver code. And if anyone has validated their simulation results before, what might be a good place to start?

Thanks

AA.

I guess since it is your supervisor (especially if he does not really know/believe about simulation) it is good to do a validation provided time is not an issue :)
It depends on what kind of simulation that you would like to perform, this forum does provide you with some reference (Wiki -> Validation Cases) :)

Contact cd-adapco's support. Explain the type of simulations you are doing and I bet they can provide you with similar validation cases they have run. They keep a library of those. Not sure on how they convey these to inquiring parties

Thank guys for your suggestions. Yeah my supervisors are bit difficult to convince when it comes to simulations. We were supposed to do wind tunnel testing to validate the results but unfortunately the wind tunnel broke down.

The validation case given in wiki for ahmed body(which comes closest to my model) only provides the dimensions and other flow properties. I thought to validate..you would need to compare results (Cd etc.) or am i missing something here???:confused:

I guess i will contact CD-adapco and see what they say. Any other ideas and suggestions are welcome.

Once again thanks a lot.

Have you look at the reference given in the wiki?
That might help you I hope.
Yes, contacting CD-Adapco directly will be a good options.

thanks a lot for your suggestions :) but my supervisors are adamant that i do experimental validation :-/

You might consider doing a google searce for DOE and Heavy Vehicle Aerodynamics. A few years ago, some generic truck shapes were developed. Physical models went to NASA Ames and CAD data went to a bunch of CFD groups. I think it's all public domain.

I think your supervisor is right, you need to have experimental validation at least in a case that it is possible. For example you can use two turbulence models of k-omega and k-epsilon, how do you know which model is the best?
Experiment is time consuming but worth to do it.

Validating is little tricky issue. Imagine that you tried to validate it and compared the results produced by starccm+ or fluent or any other software and come to find out that they do not match the experimental data. Now at this point you conclude that the software is not good, but since the results are affected by many factors related to mesh, most of them are in your hands, you are never sure whether the software is wrong or the mesh you created was not good enough to give accurate enough results.

As arjun said, it may happen that you get two different results. In some cases the CFD prediction is correct but physically is different than experiment. Or the experiment is affected by some factors, that cause the result deviate from CFD results. Because of these issues you need to interpret and it is the reason, still human is needed :)