[mk1]

Hello everyone and thanks for taking the time to help me out.

I'm new to OpenFoam and I'm currently interested in applying
different kinds of viscosity models in a Taylor-Couette flow (see the Mesh.pdf in the attachments). The mesh itself seems ok (checkMesh ok) and an incompressible Newtonian liquid flow using icoFoam yields sensible results in this geometry.

However, implementing a new solver (see viscousLambdaIcoFoam.C) using
the nonNewtonianIcoFoam as a starting point has proven difficult. In essence, I want the viscosity to be time-dependent. I approached the problem by designating "Lambda" as an uniformly valued field in the
case file (see Lambda, createFields.H. "fixedWalls" denotes the outer cylinder (stator) and "movingWall" the inner cylinder (rotor)) and assigning an ODE to this in the solver to induce evolution in time. An effective viscosity based on the updated values of Lambda is then calculated directly in the solver and inserted into the UEqn
in viscousLambdaIcoFoam.C. The solver compiles without problems.

However, in doing so, two problems seem to emerge. First off, regardless of what solver I specify in the fvSolution for Lambda, the solver always
displays "diagonal" as the solver used for Lambda (see viscousLambdaIcoFoam.log). Secondly, I do not know if the ODE for Lambda is actually solved correctly (or at all), since the solutions seem bizarre to say the least (see Lambda_t_0_001.pdf. x=0 in this Fig. is the rotor, x=0.0011 the stator and time=0.001 s) as the "stairs" seen in the results certainly should not exist. Any ideas on what's going on?

Any help is much appreciated!