[camoesas]

Hi,

I am simulating a simple test case of a flat plate. The case consists of one freeslip inflow length followed by a heated no slip wall. As Solver I am using some kind of pisoFOAM with a DES turbulence model with a SA for the RANS Region. The mesh consists of 1.2 Mio cells.
As I am using the DES I cannot thread the mesh as two-dimensional with the "empty" boundary condition. These walls are modelled as symmetryPlane.
The wall threathment is a low Re one.

The simulation is converging, but extremely slow. Examining the solver log I have seen that the max Courant Number is relatively high: ~37 and the Number of Iterations for some variables are extremely high:

• nuTilda: >500
• Ux > 200 (main flow direction)
• Uy > 250 (flow direction in the needless third dimension)
• Uz > 350 (flow direction normal to the surface)
• p = 1000

My first idea was to lower the time step by one order of magnitude. The Result is, that the max Courant Number is 3.7, as expected. And also the solution for the above mentioned variables is faster.



But the Number of Iterations for Uy (~850) and Uz (~200) is still extremely high and the solution is slow.





Are there any obvious reasons for that or some hints to fasten the solution?


Thanks!