A question about computational parameters
If a simulation time is taking too many tile steps until it reaches the Max_NumTimeSteps parameter, but it has covered only a short of time simulated, what can be done in PARAM to optimize this.
I have increased the relative convergence criterion and it did help, but still it is taking sole time. What else can be done?
1 reply
-
You can play with solver parameters or convergence criteria, but I suspect your problem is "real." Real, as in your system is evolving in such a way that complex physics are occurring that force the simulator to use very small time steps. Some common examples would include: 1) localized regions of hydrate formation at high saturation (low effective k), 2) very sharp dissociation fronts near production points, 3) large flows of undersaturated water through the hydrate zone (resulting in slow dissolution of hydrate methane directly into aqueous phase).
Typically the way to deal with small time steps is to visualize and analyze system behavior, focusing first on what's happening in the elements identified in the output file as leading to convergence troubles. From there you can determine whether the problem is "challenging behavior" or something else.