# adjusttimestep led to extremely small timestep

 Register Blogs Members List Search Today's Posts Mark Forums Read

 May 28, 2013, 04:47 adjusttimestep led to extremely small timestep #1 Senior Member   Join Date: Jan 2013 Posts: 134 Rep Power: 13 dear all I'm simulating flow around an inclined cylinder at Re=3900 with pimpleFoam. I set my maxCo to be 1 and after some iterations, deltaT became very very small which made the simulation unable to progress. This warning even came out. Code: ```Courant Number mean: 0.00965307 max: 0.959928 deltaT = 3.9084e-11 --> FOAM Warning : From function Time::operator++() in file db/Time/Time.C at line 1029 Increased the timePrecision from 9 to 10 to distinguish between timeNames at time 0.0417741``` Before this I did several cases involving flow around non-inclined cylinder at Re=3900, it worked out fine; I also simulated flow around inclined cylinder at Re=100, it was also good. Does someone else encountered with this problem before?

 May 28, 2013, 05:40 #2 Senior Member   Olivier Join Date: Jun 2009 Location: France, grenoble Posts: 272 Rep Power: 17 hello, This mean that you have high non physical velocity somewhere. Check your boundary conditions (k & epsilon with bad initial guess ?) and your fvScheme (you may switch to Gauss upwind at the beginning). regards, olivier kkpal and Ethon like this.

 May 28, 2013, 07:50 #3 Senior Member   HECKMANN Frédéric Join Date: Jul 2010 Posts: 249 Rep Power: 16 This is common if you use a uniform initialization of your problem. The initial solution is too different from your final solution, therefor the simulation must go through some crazy steps that leads to your "divergence". You can try to run first with a fixed time step and once an approximate flow field is catch, you can turn to an adaptive time stepping. Another solution is to initialize your case with a steady state solution (from SimpleFoam for example). Attention, potentialFoam will not give you a good initialization as the boundary layer is not included. Therefor you might get the same "divergence" problem. kkpal and Ethon like this.