I want to model the transient separated flow around a blunt body, e.g. a simple building model mounted on the ground using SST-DES model. In the meshing step, I estimate the minimun grid space y+ is about 10, and the grid growing factor is about 1.1~1.2. The parameters of the SST-DES model almost keep the defaults. During the iteration, the monitoring variables almost keep constant after several hundred iterations. After nearly 1000 time steps, I check the results of the blending function F2 in the post,and find that the LES region is very small. Only a small region after the building is calculated using LES, therefore, it could not obtain the transient flow structure around the building like the example given by CFX. I think the reason might be the grids in the detached regions are too coarse to active the LES.

1. Besides the method of refining the grids, could the parameter C_des also be decreased from the default value 0.61 to a smaller one since CFX adopt the multiplier of C_des and Delta(grid space) as the length scale? For the computation resource for me is limited, and the current mesh cells reach nearly 2 million.

2. Time step also affect the result. How to set the time step appropriately to aviod such problem?

Thank you!

http://portal.acm.org/ft_gateway.cfm...TOKEN=35797720

Thank you,Joe! I have downlaoded the paper. I want to ask is there any detailed paper or document about the CFX DES simulation?