I am modeling flow going through an inlet port expanding into a manifold then branching into parallel channels before entering the exit manifold and contracting into a outlet port and is experiencing persistent convergence issue. With the default under-relaxation, all the residuals start decreasing monotically for about 25 iterations before diverging. I tried lowering the under-relaxation factor but it only delay the diverging. I have also tried using different element e.g. quad, tri, hybrid but nothing seems to help. With this diverging trend, it cannot even meet the default convergence criteria.

Any advice will be greatly appreciated.

Can you tell me a little more about your simulation? what solver are you using? what is the reynolds number or the velocity at the inlet port? is this macro or micro level?

Sure, I am using the segregated steady solver. Reynolds number is less than 1500 and velocity at the inlet port is 1m/s. It is in the mini- to micro- level.

Thanks

I should also mention that the flow makes a 90 deg bend from the inlet port to the manifold. Actually it is a bit like impinging on the manifold. Not sure is this giving rise to some unsteadiness...