Hello, i'm trying to incorporate a nonlinear stress-strain relation into a turbulent flow solver. the flow solver is the one according to the book by ferziger and peric, "computational fluid dynamics". it uses colocated grid arrangement and a finite volume approach. the additional terms compared to a linear boussinesq approx are treated explicitly in the momentum conservation. the problem is that i can get a solution, but there are oscillations in the velocity- and pressure fields. without considering certain terms in the stress-strain model (especially the second derivative terms), the solution is without oscillations. How can this be explained ad what countermeasures can be taken to eliminate the oscillations ?

Hi,

I have implemented something similar a while back. If I remember correctly, the trick was to implement to non-linear stress term as an explicit correction onto the Boussinesq approximation (the effective viscosity part of the momentum equation remains unchanged). If you try to implement the whole div(R) explicitly OR rip out the parts corresponding to component-wise effective viscosity, you get oscillations.

Hi, I've implemented a cubic (nonlinear) turbulence model and found that in order to avoid divergence I needed to apply a form of under-relaxation to the quadratic and cubic terms. Obtaining a stable solution was tricky at the best of times!!

richard