[CIG-SHORT] Bizarre tractions on dynamic fault with gravity

[Romain Jolivet]

Dear Brad et al., 

I am running a 2-D problem with an elastic rectangle (200km*50km) cut by a fault to simulate a thrust fault.
Gravity is on in this problem. For a picture, please look at the file geometry.png.

Because gravity is on, I have to prescribe initial stress conditions (isotropic stress tensor everywhere = rho*g*depth).
For the same reason, I prescribe a Neumann condition on the bottom boundary with sigma_n = rho*g*depth and tau = 0 Pa
 
The fault is a RateStateAgeing with slip strengthening properties. I push on the right side (dirichlet condition, 1 cm/year) and I want to see it sliding all the way, steady. The other side is fixed.
Because I do not want to wait for too many time steps, I prescribe an initial shear stress along the fault equal to the normal stress multiplied by the standard friction coefficient (normal stress which should be equal to rho*g*depth).
Figure initialTraction.png shows the prescribed traction on the fault (x-axis is depth in m, 0 m is the surface).

Still, when I run the problem, it turns out that the pre-step computes a normal traction on the fault that is not equal to rho*g*depth, at least not everywhere. It seems fine for all the nodes inside the material, but the two nodes on the edges are off the linear trend where they should be (see afterElasticPrestep.png). I might be doing something wrong, but I wonder how tractions are estimated on the edges of the fault?
This case leads to non-constant slip along the fault, especially for the first time step.

I could compute the pre-step and then feed the tractions as a spatialdb, but I feel this problem is too simple to require that and there should be a solution.

I joined a tar ball with the problem configure files, spatialdb, etc, if you want to try it (If you feel like numbers for stress are a bit awkward, it is because I want to do as if the whole thing was buried under 500 m of material. Removing that 500m of excess does not change anything in my case.).
Let me know what you think,
Romain

—————————————————————————————————————
—————————————————————————————————————	
!!!!!! ADDRESS CHANGED !!!!!

Romain Jolivet
Postdoctoral Fellow

University of Cambridge
Department of Earth Sciences
Bullard Labs
Madingley Rise 
Madingley Road
Cambridge CB3 0EZ
United Kingdom

email: rpj29 at cam.ac.uk
Phone: +44 1223 748 938
Mobile: +44 7596 703 148

France: +33 6 52 91 76 39
US: +1 (626) 560 6356
—————————————————————————————————————
—————————————————————————————————————
-------------- next part --------------
A non-text attachment was scrubbed...
Name: problem.tgz
Type: application/octet-stream
Size: 30720 bytes
Desc: not available
URL: <http://lists.geodynamics.org/pipermail/cig-short/attachments/20141111/e3da5c47/attachment-0001.obj>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: geometry.png
Type: image/png
Size: 30295 bytes
Desc: not available
URL: <http://lists.geodynamics.org/pipermail/cig-short/attachments/20141111/e3da5c47/attachment-0003.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: initialTraction.png
Type: image/png
Size: 30568 bytes
Desc: not available
URL: <http://lists.geodynamics.org/pipermail/cig-short/attachments/20141111/e3da5c47/attachment-0004.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: afterElasticPrestep.png
Type: image/png
Size: 35837 bytes
Desc: not available
URL: <http://lists.geodynamics.org/pipermail/cig-short/attachments/20141111/e3da5c47/attachment-0005.png>