Hi all,
I am trying to use free surface to a VEP deformation model but see the large velocity (~20cm/yr) at the bottom at initial time step, even though free surface is only added to top of the model.
I didn’t set any of velocity at the boundary. Other boundaries are all free-slip.
For temperature setting, top and bottom are constant, and the rest is insulating.
I am new to free surface model but I think free slip models don’t show this amount of velocity at initial time step.
Is it natural to have this effect at initial time step when free surface is added to a model?
If not, could I get an advice how I can remove this effect?
Here are the images of the model at initial step:
-
Temperature
-
Velocity with Free Surface at Top Boundary
-
Velocity only with Free Slip
Cheers,
Hyunseong
Hi @hyunseong96
Thanks for posting the message to the forum and welcome!
I didn’t set any of velocity at the boundary. Other boundaries are all free-slip.
Can you try setting the bottom boundary to free-slip or no-slip in combination with a free surface top boundary? I believe the default is to set a boundary to only have traction constraints if it is not specified to have velocity boundary conditions.
That would certainly explain the anomalous velocities.
Cheers,
John
Hi @jbnaliboff
Thanks for the reply!
Sorry I think I made a bit of confusion.
Originally, top boundary is free surface and rest of them are free-slip. I meant to say I didn’t use any prescribed velocity boundary.
so I tried no-slip boundary at the bottom. Initial velocity at the bottom of domain does disappear but convection doesn’t develop.
Is there other way I can minimize unexpected velocity?
Thanks again for your help!
One update is that I changed geometry of this model from spherical shell to a box with similar aspect ration of geometry.
There is no such error. Could it be related to a geometry?
Cheers,
Hyunseong
Originally, top boundary is free surface and rest of them are free-slip. I meant to say I didn’t use any prescribed velocity boundary.
Ah, sorry for the confusion.
One update is that I changed geometry of this model from spherical shell to a box with similar aspect ration of geometry.
Oh, that’s interesting. I recall @maaikeweerdesteijn observed strain rate anomalies in spherical models of GIA-related viscoelastic benchmarks. Is the mesh uniform or is there by chance mesh refinement near or at where the anomalies are located?
Cheers,
john
Hi Hyunseong,
Some more things to think about: When you use a free surface, the mesh is not just deformed at the surface, but all nodes are moved up or down a little bit, and based on the mesh, the solution can change slightly. However, since this happens in time step 0 for you, I am assuming at that point in time you do not actually have a deformed mesh? In other words: Do you start with an initial topography? In addition, I would also test if these velocities become smaller with a higher mesh refinement, so if the problem is actually related to the mesh. Finally, how does your viscosity look like? How low does it get in the layer where you see these high velocities?
-Juliane
Hi John,
There was refinement at the top where free surface is applied, but not at the bottom.
Now that I tried refinement at the bottom, it did reduce the velocity!
I guess I will have to make finer mesh for the initial time step and early few steps then change into coarser mesh.
Thanks for your help!
-
without bottom refinement:
-
with bottom refinement:
Cheers,
Hyunseong
Hi Juliane,
I didn’t set any initial topography. On topography output at 0 yr, there is value around 1e-10 but I guess these are negligible.
But like you mentioned mesh refinement helped, as shown from the figures uploaded to reply to John!
Also, I see that viscosity is around 1e15 Pa*s where the velocity occurs. However, it is a numerical model for icy surface of Europa so may differ from models of rock mantle.
Thanks for your help!
Cheers,
Hyunseong
