Dear professors,
Hello. I recently set up a two-dimensional box model with two parts. In the initial setup, the ASPECT’s solver did not converge, which caused a runtime error. So I adjusted the value of the CFL number.
Now the program works, but the problem is that the time step interval is too short. In fact, I’m not very proficient with aspects, so I’m wondering how to appropriately increase the interval of each time step while keeping the software running steadily.
If anyone can help me, I will be very grateful.
callisto.prm (2.8 KB)
Best wishes,
Lu
@ddarkerlu Yes, your time step is on the order of 3 days now – it’s going to take a lot of time steps to simulate geologic time scales this way 
In essence, the time step is computed based on the formula
Delta t = CFL * velocity / mesh size
You set CFL=0.1. Does it have to be this small to achieve convergence?
Of course, the other two factors on the right hand side are also relevant. I assume that you have fairly large velocities and small mesh sizes? Are the velocities you get realistic?
Best
W.
@ddarkerlu - I’m not familiar with the dynamics of callisto, but I think a time step of 3 days is not unexpected for velocities on the order 4.7e3 meters per year. As @bangerth noted, I think the key question is what in your model setup is producing such high velocities. This could be due to the driving forces, rheology, or some combination thereof.
Thank you very much for your reply. In my previous setup, the CFL number is indeed only convergent for very small values.Your reply also provides me with a very clear idea that my boundary velocity model may not be set very reasonably. At the same time, does the grid setting need to be more refined?
Thank you very much for your critical questions to me. In fact, I am just beginning to get involved with Callisto. I will look again to see if the velocities setting is reasonable, which is a factor that I did not care about when setting the model
