Hi, Everyone,

I am working on the computation of turbulent mixing layers (using K-E and RSM model), my computation domain is 1.0m by 0.12m, inlet air velocities are 22m/s and 41m/s respectively, how many grids is suitable for this case? thanks.

(1). You can read the Boundary Layer Theory book by Schlichting. Near the end of the book, there is a chapter on the jet and wake mixing. (2). Using mixing length theory, there are analytical solutions presented in the book. (3). Based on that solution, you can plot the velocity profiles from the origin along the downstream direction. (4). The mixing region spread out at certain rate or angle. Outside this mixing layer, the flow will be uniform. (5). So, your mesh must be a fan shaped grid, to take care of the spreading of the mixing layer. (6). So, the number of mesh points in the y-direction must be enough to give you the smooth velocity profile, say at least 20 points inside the mixing layer. Outside the layer, you still needs mesh points to cover the free stream region. (7). If you are planning to use two-equation model, then you can plot also k-profile, to see if you need to add more points to handle the k-profile. (8). So, basically, you work backward from the known distribution profiles. You don't have to use the analytical solution profile to guide you, the experimental data profiles are also very useful. (9). Once you have the expanding mesh created, you can run several cases with increasing mesh density. In this way, you can finally obtain the mesh independent solution. (10). So, 100x50 would be a good starting mesh in the x- and y-directions. The x-mesh also must be expanding in the s-direction. The total x-direction mesh points naturally depends on the size of the domain you are going to cover.