Suggestions for convergence

 Register Blogs Members List Search Today's Posts Mark Forums Read

 November 7, 2002, 14:29 Suggestions for convergence #1 David Guest   Posts: n/a I am simulating gas-liquid two-phase fow in a pipe. The liquid seems to be converging but gas phase doesn't. Could anybody please give me some suggestions? Thanks David

 November 10, 2002, 17:32 Re: Suggestions for convergence #2 Astrid Guest   Posts: n/a Davedid, I have read something on this subject in the CFX-4 manual. If I remember it correctly it said that if you have convergence problems you have to use a (much) smaller timescale for the gasphase than for the liquid phase. Then, after you get a pseudo-steady state, you should switch to larger timescales. Astrid

 November 11, 2002, 00:43 Re: Suggestions for convergence #3 David Guest   Posts: n/a Thanks, Astrid. Do you mean that I can switch from a smaller timescale to larger timescale during the solver is running or after this run finished? How to implement it? Regards David

 November 11, 2002, 17:02 Re: Suggestions for convergence #4 Astrid Guest   Posts: n/a If you have little experience you should use physical timestepping and increase the timescale after you get the pseudo steady state. If you are more confident of the case you're running, you could increase the timescale automatically during the run. Use 'Auto Timescale' and all the options that are discussed in detail in the help. Good luck, Astrid

 November 12, 2002, 01:50 To Astrid #5 David Guest   Posts: n/a My simulation can converge to a critea between 0.001 and 0.0001 after the 20th step and then almost keep a constant residual but still decreasing very very slowly. It seems need a huge number of timesteps to meet a 0.0001 residual. I tried almost all sort of timestep, the results are same. The velocity balance is small while for other variables such as mass fraction the balance is very big. Could you please tell me how to do it? Thanks David