Hi, we are trying to simulate a building to study the flow around it. The model is a 2d mesh imported from gambit. We are having problems with it, as the solution fails to converge. The solver is implicit and coupled, and we are using the SA viscosity model.

Our main concern are the boundary conditions. We have chosen a mass flow inlet and pressure outlets, and the buiding itself is formed of walls.

What would you change to make it converge? We're quite new and still learning this program.

the geometry of the domain is more or less like this, with a segregated mesh around the building (the small lines at the bottom of the domain). The mesh is smaller around the building, to simulate the boundary flow.

Please help us!!!!! ____________ | | | | | | | | | | | | | | | __ | |__||______|

Ooops sorry, the drawing wasn't exactly intended to be like this, if you have doubts about the geometry tell us.

I've done this topic before. Maybe I can help you a bit. Don't worry, it's not too difficult if you're dealing with 2D problem for that.

(1) For the turbulence model, you're now using Spalary-Allmaras Model. I used k-e model. maybe you can try.

(2)Boundary Condition: -for the building, you're right. they're stationary wall. -for the ambient, I used "velocity inlet" and "pressure outlet" for either 2 sides. for another 2 sides, I used "moving walling". Be aware of the dimensions of the block which simulates the ambient. it should be large enough, especially for the side near the pressure outlet.

For the convergence problem, it is affected by a lot of factor. i seem to be used segregated solver.

(1) convergence criterion - i used 10e-6 as convergence criterion. (2) relaxation factor - try to reduce a bit. (3) in discretization, using SIMPLEC algorithm for pressure-velocity coupling can speed up convergence.

Ok,I'll try to play a little bit with the convergence factors and see if it helps. As for the boundary conditions, fluent doesn't allow me to use the velocity inlet because I am using compressible flow. Maybe this is the problem after all, and the results are not very affected by the compressibility...i'll try changing it.

Thanks for helping