turbulence model in internal combustion engine

George

Hello, I have written a source code simulating 3D flow in an internal combustion engine (curvilinear, unsteady, compressible, turbulent). Until now I have incorporated the standard k-epsilon turbulence model and the RNG k-epsilon. Except from LES and RSM, which other turbulence model is the better for in-cylinder simulation? What about non-linear k-epsilon models?

When I use the RNG model, the code converges much more slowly than with the standard k-e model (50% more iterations). Is this expected?

Thank you in advance