Hi folks, has any body in here experienced a situation whereby the wall temperature affects seriously the convergence of a simulation? I am simulating flow at high mach numbers. the free stream temperature is 262 k. i obtained converged solutions with a wall temp of 500k but cannot obtain a solution with 300k. The solution terminates with a FINMES error. will be grateful for some tips. regards, ben

Hi,

If you have natural convection effects activated the wall temperature defines the Rayliegh number of the simulation and therefore strength of the flow. Obviously in this case the wall temperature will effect convergence.

In your case you say that reducing the wall temperature lead to convergence problems. This sounds a little strange. What are you modelling?

regards, Glenn Horrocks

hi Glenn, thansk for your reaction to my post. I am dealing with hypersonic flow over a zero degree of attack plate. I seem to have found the reason for it. because of the high velocities involved, in my case mach 7.5 that is about 2376 m/s, the static temperature rises sharply in the boundary layer. for a free stream static temperature of 260k and a wall temperature of 300k, the static temperature reaches values of over a 1000k within 0.2mm distance from the wall. this leads to very high gradients. the effect is less with higher temperatures. since the cfx solver is implicit, besides a fine BL resolution, a good guess of the initial field is necessary. i am trying to get a profile around the midplate to use as initial value. seems to work well for laminar up to a convergence target of 1e-5 max residual. waiting to see if the transitional simulation will converge. thanks anyway. Glenn, have you worked with the SST two equation transition model before? i need to exchange experience with someone who has used this before. regards, ben

Hi,

I don't have much experience in modelling hypersonics so I don't think I can help you much. I have used the SST transition model, but again not in hypersonics.

So for some suggestions: Make sure the solver is doing the extra Pressure/mass solution each iteration, and carefully check the various convergence options - local timescale factor versus physical timestep.

Glenn Horrocks