I would greatly appreciate some advice on improving the convergence of my fluid dynamics Heat transfer model of a pipe which I have developed in Comsol multiphysics. To be a bit more specific the problem I have is a transient one and I am using the 'time discrete solver' as my model uses values from previous time steps. Lastly, I can of course speed up the simulation time by increasing the relative tolerance, but this is not what I am looking for as I don't want to reduce the accuracy.
Do you know what algorithm the 'time discrete solver' is currently using?
Do you know what type of matrix is being produced by your simulation?
If your matrix has special properties, you could use a more specialized solver and speed up convergence.
What's your IC? Did you solve stationary and use it as IC? A physical meaningful IC is very important.
Thank you both for your comments.
1/To be honest I am not sure, but I would be interested in knowing more about this.
2/ Yes I first run a steady state solver and use this as the initial condition.
I found that by reducing my time step I achieve a better convergence, albeit a slower simulation. Also for future reference to anyone else with a convergence issue in COMSOL, I would recommend this webpage:
https://community.cmc.ca/docs/DOC-1453
- Badges
- Science topic