[marluc]

Hallo,

I have just written a UDF for the AKN model (Abe-Kondoh-Nagano).
I tested it for a channel flow, a backward facing step and a ribbed duct.
Upon reaching a convergent solution with a k-e model with enhanced wall treatment I switch off the calculation of the turbulence transport equations and, after initialization of the UDS, I solve for their transport equations, in order to compute the user-defined turbulent viscosity to use in the momentum equations.

My problem is when I first start with the UDS because I often get AMG solver divergence at the first iteration. So I first set the URF to very low values and solve for each UDS separately and without the momentum equations activated. Then I start increasing the URF and later I activate the flow calculation as well.

I am wondering if someone has the same problem when dealing with UDS for a user defined turbulence model.
Furthermore, when using the AKN model implemented in Fluent I can keep the default URF for momentum, k and e and the calculation converges. Does someone have an explanation for this? (actually Fluent and I should solve the same equations!)

Thank you in advance for your answers.
L.