Oscillation for the average vertical heat flux and rising velocity versus time

Hi all,
When I plotted the average vertical heat flux, viscosity and rising velocity from the depth-average.txt versus time to see whether the plume reached a steady state, I observed this kind of ‘sawtooth’ pattern.


We are not sure whether this is the ‘drunken sailor’ effect due to the free surface.
When we looked into the paraview output to see why the ‘sawtooth’ pattern. We found that the refinement of the grid changed to the opposite from one timestep to the next.


We think this might be the reason for the ‘sawtooth’ pattern but we don’t know the reason and how to solve it.
I would greatly appreciate it if you could help us with this issue. Thank you very much!

Cheers,
Ziqi

@maziqi96 - Thanks for posting and a few questions:

  1. What is the adaptive refinement criteria you are using?
  2. Have you tried using a constant grid resolution (or an initial adaptive one that then stays static through time)?
  3. Are there similar fluctuations in the surface topography (if using a free surface) or dynamic topography (if using a free-slip boundary)?

Cheers,
John

Hi John,
Thanks for your timely response.

What is the adaptive refinement criteria you are using?

Subsection Mesh refinement
set Refinement fraction = 0.7
set Coarsening fraction = 0.25
set Initial adaptive refinement = 4
set Initial global refinement = 1
set Time steps between mesh refinement = 5
set Strategy = viscosity, minimum refinement function, maximum refinement function

Have you tried using a constant grid resolution (or an initial adaptive one that stays static over time)?

Not yet. I will give it a try. But I don’t think we can do an initial adaptive one that then stays static through time because we want to see the plume-lithosphere interaction and diffusion and dislocation creep viscosity will change when the plume rises and hits the lithosphere.

Are there similar fluctuations in the surface topography (if using a free surface) or dynamic topography (if using a free-slip boundary)?
I just plotted out the maximum topography versus time and there seem to be similar fluctuations in the surface topography (we use a free surface in our models)


Hope this helps.

@maziqi96 - Thanks for the update.

The surface topography also has minor fluctuations along an otherwise smooth curve approaching steady state. The tests with temporally constant resolution will hopefully be informative.