Hi all, I am trying to model a transient problem and I am having convergence issues. The system I am modeling is a reactor with pressure drop and the flow rate is very low (laminar). The reactor is filled with air and hydrogen is pumped into the recator at t=0. Hydrogen replaces air and fills up the volume. I didn't have any convergence issues when I run the steady state model with air. The problem is when I run the tranisent model, the continuity residuals didn't drop to 10-4, where as all the residuals dropped well to 10-6. How to define convergence. When the residuals are low it means we have minimised the error. Another signifance add to to the problem is as per my understanding for tranisent problems, set the number of iterations per time step size to be lower than 30. If you don't hit convergence then lower the time step size. But I lowered the time step size all the way down to 10-4 but still the continuity residauals didn't drop well. The problem being when the step size is small, and you take 50 iterations per time step , it is taking a whole lot of time in weeks to reach the total flow time. Even then I am not accepting the residual criteria but only trying to get a time step size where I have a stable solution and proceed to the next time step. I am also monitoring different flow characteristics with surface and point monitors. The boundary conditions I impose are: at the inlet Hydrogen mass fraction=1 at the outlet air mass fraction=1 laminar flow model. I do have porous zones for pressure drop in my system. Pressure Inlet and Pressure outlet boundray conditions. operating conditions: ambient. I am not solving for energy equations.

Any insight on transient flow problems and pressure correction and denisty difference issue in convergence will be greatly appreciated. I played a lot with under relaxation factors, and at present as I am operating in ambient conditions, my material properties are all based on mass based mixing laws. To reduce the complexity of the model and as the pressure drop is less than 10000 pascals I am calculating density based on mixing laws. In other words I am not into compressible flow. All your comments are most welcome, Thanks Sundar

Sundar: I m not exactly sure, but one way to check the convergence is to monitor the surface integrals over a surface, and see if that is constant. Thats what it says somewhere in the Fluent Manual too.. hope this helps! Atul