Discussion Forum

Hi All,

I'm having trouble maintaining the the incompressibility constraint (div u =0) on a Stokes flow model I set up. I've implemented my model using the weak form; I'm solving in the true overdamped limit (du/dt = 0). The flow in my problem is driven by a body force which comes from some other fields (their time derivatives are not zero, so this problem is not quasi-steady). When the driving force is large (I would say, on the order of 10x the viscous dissipation), incompressibility becomes very poorly maintained ~O(0.1) (unacceptable!).

My question is: How can I structure my solver to hold the continuity/pressure equation to a higher tolerance. I've essentially used the default settings up until now (with consistent initialization turned off). It seems a segregated solver that iterates more on the continuity equation is what I want, but my attempts to do so seem to just add computational time without increasing quality. Has anyone run across this issue?

Thank you!

More details:
Stokes and Continuity are implemented in their own weak form PDE modules so that I can assign a lower order element to the continuity equation (seems necessary, and is recommended in some FEM texts as well as by people in this forum). I've had pretty good luck getting reasonable results in cavity flow and other benchmarking problems.