Why does the model run very very slowly after changing "box" to "Box with lithosphere boundary indicators"?

ASPECT
1

Hello everyone:
I wanted to fix the temperature of the lithosphere boundary, so I changed the “Box” to “Box with lithosphere boundary indicators”. However, after running the prm file, the model ran very slowly. No other Settings have been changed. Why is that? I’m completely puzzled.Thank you all.

image
image580×359 30.1 KB

2

@sxj Can you compare how many degrees of freedom you get with the two different choices of geometry and mesh? My suspicion is that the coarse mesh for the “Box with lithosphere boundary indicators” has many more coarse cells than “Box”, and consequently you have many more degrees of freedom and a shorter time step.
Best
W.

3

I’ve posted two comparison pictures for you to take a look. There are indeed differences, but I don’t understand what caused them. How should I adjust them? Thank you

44cd2b79c6599085ad7dc73e9039dd4
44cd2b79c6599085ad7dc73e9039dd41714×824 215 KB

3e5956bd506f88bf7983834ef0112e5
3e5956bd506f88bf7983834ef0112e51712×820 177 KB

4

@sxj The difference is due to the fact that the two geometries use a different number of cells for their coarse meshes, which you then refine a number of times. You can look at the coarse mesh if you set the number of refinement steps to zero and output the solution.

In your case, you end up with about 5x as many cells, and so you should expect that it takes about 5x as long to solve each time step. Each of these cells is also smaller by perhaps a factor of 2x, and so you should expect the time step to be smaller by about 2x just based on cells sizes.

Of course, in your case, the time step size is vastly smaller. This must be because the velocity is vastly larger for the box with lithosphere boundary indicators. Why that is I don’t know, but you should be able to compare the velocities in the first time step between the two simulations and see what is going on.

Best
W.

5

@sxj - To fully answer this question we really need to see the full PRM file. To the specific question here, how have the boundary conditions changed when you added the lithosphere boundary indicators? Are any of the model sides open (i.e., traction boundary)?

6

Thank you for your reminder and answer. Later, I found that after changing from “box” to “Box with lithosphere boundary indicators”, I forgot to specify the boundary of the lithosphere. After specifying the boundary conditions, the running speed of the model became normal. Thank you again for your answer!