How to set the normal direction of a boundary to free slip？

ASPECT
1

Hi all,

I used box with lithosphere boundary indicators build a 3D continental extension model. Now I want to prescribe the right lithosphere extends at 5mm/yr, and set the normal direction of the front and back lithosphere boundaries to free slip, so that material can flow in and out
according to the flow induced by the moving plate. I tried to prescribe the y component to be zero traction. But the model cannot converge. The model boundary conditions are set as follows:

####### boundary condition #######
# 0:left, 1:right, 2:front, 3:back, 4:bottom, 5:top, 6:left lithosphere, 7:right lithosphere, 8:front lithosphere, 9:back lithosphere
subsection Boundary velocity model
  set Tangential velocity boundary indicators = 0, 1, 2, 3, 4, 6
  set Prescribed velocity boundary indicators = 7 x:function
  subsection Function
    set Variable names      = x,y,z
    set Function constants  = v=0.005
    set Function expression =   v; \
								0; \
								0 
  end
end

subsection Boundary traction model
  set Prescribed traction boundary indicators = 8 y:zero traction, 9 y:zero traction
end

subsection Boundary composition model
  set Fixed composition boundary indicators = 8, 9
  set List of model names 		    = initial composition
end

I want to know if there are any issues with such a setup, and how I could improve my model. This is my complete prm file.
test_extension_3D_decoupling_1.prm (11.9 KB)

2

@egg_wang Before we go and debug your input file, can you describe in more detail what “But the model cannot converge” means? It is often useful to be specific about what goes wrong and how it goes wrong.

Best
W.

3

Hi @bangerth , I’m sorry I didn’t make my question clear. When I calculate using the settings above, the time step becomes very short, about only 7-8 years, and it terminates after running for more than a hundred years. These are the error output files.
log.txt (101.9 KB)
job-err.txt (6.4 KB)

Best,
Wang

4

I tried to set the whole 8 and 9 to open boundaries with zero traction. However, I am still facing the same problem. Does anyone know how I can implement this example?

5

@egg_wang - A very brief reply (traveling).

You should use a traction boundary with the lithostatic pressure applied (this is one of the options for the traction boundary model) instead of a zero traction boundary. The latter will cause material to just flow rapidly out of those boundaries due to the large pressure gradients.

At the bottom of this page there should be a list of suggested and related topics where similar questions have been discussed. My recollection is that they contain PRM examples for how to setup the lithostatic pressure traction BC.

John