Correcting random divergence
 

Dear CFD Community,

In my experience of transient flows, I have had highly coupled nonlinear solutions seemingly diverge unexpectedly during simulation. When monitoring the residuals throughout the simulation, each equation converges to a consistently minimum value, but after hours of simulating, the residuals will blow up with no clear distinction as to which equation was the problem.

Does anybody have an opinion to this matter or a plausible method to correct it or at least their own problems in the past?

I am using an unstructured hybrid mesh with the explicit VOF model. It is here that I am expecting the problem to arise. Thoughts?

Hershey