Hi, All

I just read a paper about using Monte Carlo method for CFD analysis. The authors run 100 times for their sensitivity analysis. In my case, this method is not feasible, since each of my run takes a long time. I would like to know what's the best way for CFD sensitivity and uncertainty analysis, how to significantly decrease the total number of runs (especially when use commercial CFD software)?

Waiting for your comments!

Thanks!

Jim

I think you should reformulate your question a bit Jim. What do you exactly mean with using MC for CFD? what type of flow (single/multi/reactive), or using MC after the CFD analysis has been performed?

Hi

You may have a look a section 8 (nonintrusive spectral projection) of "A stochastic method for fluid flow: II. Random process" by O. P. Le Maitre, M. T. Reagan, H. N. Najm, R.G. Ghanem and O. M. Knio, J. Comput. Phys. 181(1), pp. 9-44, 2001 Direct link here

Regards

Hi, Ridwan

In the paper I read, the MC was used for sensitivity analysis. It's a one-phase CFD modelling using phoenics.

In my case, it is a two phase problem and I am using CFX.

What I want to know is how to conduct sensitivity and uncertainty analysis for a time consuming CFD modelling. Since usually sensitivity analysis is based on lots of runs. However, 100 time runs may take me hlaf to one year.

Thanks!

Thanks, lionel That's a nice paper!