Choice of the time step
Niels Montanari
How could be choose a time step to ensure the stability of the simulation? Is there any stability criteria, like the Courant-Friedrichs-Lewy condition in computational fluid dynamics?
I understand that the class ChRealtimeStepTimer has a member function SuggestSimulationStep that enforce real-time performance. Is there a similar function but for attempting to ensure stability?
Sincerely,
Niels
Alessandro Tasora
Hello Niels,
- there is no "automatic" tuning of the time step at the moment, you should provide a proper one at the beginning,
- you might compute a bound on dt using the CFL condition, as you said, if using finite elements only (enable chrono FEA unit, and MKL or MUMPS solvers - suggested) But adding constrained rigid bodies, and contacts maybe, make things more complicate.
- the default timestepper with ChSystemNSC (ie with non smooth contact formulation) is very stable and robust because it behaves like an implicit solver: in my simulations I rarely had to go below dt=0.001s; for robotics often 0.01s is good enough; but keep in mind that if you use FEA and you want also contacts, at the moment you can't use ChSystemNSC and you must use ChSystemSMC, ie penalty for contacts, which requires shorter time steps. Something that requires hand tuning of dt.
Regards
Alessandro Tasora
--
You received this message because you are subscribed to the Google Groups "ProjectChrono" group.
To unsubscribe from this group and stop receiving emails from it, send an email to projectchrono+unsubscribe@googlegroups.com.
To post to this group, send email to projec...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.
Niels Montanari
Got it! Thank you very much for your quick reply.
Sincerely,
Niels
To unsubscribe from this group and stop receiving emails from it, send an email to projectchron...@googlegroups.com.