Hi. I'm applying cfd on a rectangular channel with 90 degrees inlet and outlet. Flow is laminar.

The mesh in the channel is structured mapping while those in the inlet and outlet are coopered following quad paving at the source faces.

When using SIMPLEC with skewness correction (1 to 3), the solution diverges (under-relaxation factors for pressure = 0.3, momentum = 0.4). It converges only with no correction or by using SIMPLE. Why is this happening?

Nice question (unusual) and I can't explain about the numerics. 2 comments tho -

Is the flow really laminar? Should the flow actually get unsteady at that Reynolds number where the jets enter the main channel? If that's the case, then perhaps the flow is damped in the SIMPLE solution, and the additional pressure corrector steps of SIMPLEC are revealing it's true nature...unsteady.

The other thing I can think of, you have a really low pressure underrelaxation. For SIMPLEC Fluent manual recommmends it is higher. Why are such low factors needed? Perhaps you need to run transient...

Maybe it helps, check, but don't trust my advice....

The flow is indeed laminar, at least numerically the highest Re is 600.

On the issue of unsteadiness, I noticed that under some conditions (higher flowrates), there are flow reversal at the outlet. The occurrence of flow reversal appears to be periodic wrt the number of iterations. That is, although the solution in the main channel has converged, it never converges at the outlet. Would this suggest that the flow outside the main channel is unsteady?

On pressure under relaxation, you are right...