I am simulating flow through a CD nozzle when a shock is present in the divergent portion. With a coarse grid i got convergence in Steady run. Then i only refined the grid near the wall by reducing by half the grid spacing at the wall and by increasing grid spacing gradually with ratio 1.2. I am not getting convergence. For coarse grid the physical timestep was 1e-6 sec. Now I had to reduce the timestep to 1e-7 sec to avoid crashing. There is wall formation at the inlet even after 25000 timesteps. Please suggest some solution. I have not used turbulence for the coarse grid and so for the fine grid also.

Hi,

The wall at the inlet is due to a recirculation being present at the inlet. For a con/div nozzle this is almost certainly a numerical artifact caused by poor convergence.

What differencing scheme are you using?

Finer meshes allow you to resolve finer flow features and therefore you may be starting to pick up separations or vortex shedding which the coarser mesh smoothed out. Try using local timestep factor for a little while to improve convergence but always do the final run to convergence with a physical timescale. Using a coarser mesh result as an initial condition, interpolated onto the finer mesh can also improve things.

You should also find adding a turbulence model will help convergence, if in fact your flow is turbulent.

If nothing else works you will have to run the simulation as a full transient. This will result in a much longer run time but if you use the adaptive timestepping you can quickly find the timestep size it likes and with luck it should start settling down.

Hi all my self basappa studing mtech final year ,i am doing project on fuel injector assembly using catia v5/Meshing using hypermesh .can any body help me out for geeting the material on the same .Plz consider this as priority

Regards Basappa V