Divergence after many time steps
 

Hello, I am simulating a 3D turbulent in-cylinder flow (without combustion). I run my code and at 240 degrees after top dead center (after 240 time steps, during compression) the code diverges. I understand that in internal combustion engines the most crucial part is the flow through the valve, but at that moment the code performs quite well. The problem is afterwards, during compression, where there isn't any flow through the valve (intake valve closed at 180 degrees after top dead center). I tried to lower the underrelaxation factors (until 0.1 for all equations), but the result is the same. At around 240 degrees diverges. Any suggestions?

Thank you in advance! George