Load ramping to improve convergence

[RaJa]

Hey guys,

I'm currently struggling with a CFD problem that is a tough nut to crack. However, while combing through some COMSOL blogs, I came across something that is called viscosity ramping / load ramping for non-linear problems like turbulent flow. Now I'm wondering if ELMER is actually capable of that.

What I mean is:
Your start your CFD-problem with a higher viscosity of your medium as it actual has and so decrease the turbulences. The problem converges faster or even at all. Now you take this solution and use it as a starting point for the next iteration with a reduced vicosity. You repeat this cycle until you reached the real viscosity of your problem.
It seems that COMSOL can do this automatically. But is it possible to to something like that in ELMER?

Cheers,
Rainer

[kevinarden]

If you are running in the time domain than

Viscosity = Variable Time
Real
0.0 2.0
1.0 2.0
2.0 1.8
3.0 .9
4.0 .1
End

If you are doing steady state than you can change the Simulation Type to scanning which runs several cases
Simulation Type = Scanning
Timestep Intervals = 3 ! For 3 cases

time is still used but only as a variable to track the case number, then

Viscosity = Variable Time
Real
1 4.0
2 2.2
3 .5
End