CFD for Internal Flows
 

Hi,

I've been running Jameson's Runge-Kutta time-marching scheme for a compressible internal flow simulation.

When plotting mass flow vs time at a certain location along the duct, I notice that the plot is sinusoidal type with decaying amplitude...but it decays very slowly and convergence could thus take a long time.

Appreciate advice on how this long settling time could be reduced.

Thanks in advance!

Re: CFD for Internal Flows
 

Your problem is how to accelerate convergence. Jamenson (JST) dissipation model is usually used together with multigrid, enthapy damping, local time stepping, and applies to transonic flow. Check if you have used these techniques.

Re: CFD for Internal Flows
 

I am also running Jame's Runge Kutta time marching. But for external flow, and I would like to do internal flow too. What is your boundary condition at the inlet and exit? Do you use Riemann invariant? Thanks for your info.

Re: CFD for Internal Flows
 

Hi,

Inlet: Total Pressure and Temperature,Flow Angle Exit: Static Pressure.

Yes, Riemann invariant used.

Regards

Re: CFD for Internal Flows
 

Thanks for your info. In Jameson's code, he solve density, den*u, den*v, den*w, den*E. So the boundary condition at inlet and exit need to specify the above 5 parameters, den,den*u,den*v,den*w,den*e. So at inlet, you use total pressure, temp and flow angle to calculate the above 5 parameters. But at the exit, only pressure.Do you use pressure to compute the above 5 parameters? Or extrapolate from the inside domain? If you use Riemann to extrapolate, pressure might not be the same as you specify. I know in physics we specify pressure at exit. But from computer coding, exit need to specify the above 5 parameters. Thank you again for your advice.