[CIG-SHORT] Time varying shear stress on a fault for time independent simulations.

[Brad Aagaard]

Josimar,

Look at the residual values during the solve. I think this is a solver 
tolerance issue. You will probably want to reduce the relative solver 
tolerance if the solution you are getting is not accurate enough.

My guess is that in the second time step you are simply picking up from 
where you left off on the solve in the first time step and getting 
smaller residuals. This is an effect of having an initial large residual 
in the elastic prestep with convergence due to a relative residual 
reduction and then starting with a smaller residual in the second time 
step and reducing it to a lower value.

Regards,
Brad


On 02/22/2018 09:22 AM, Josimar Alves da Silva wrote:
> Hi Brad,
> 
> I am looking to understand one of the outputs of  a quasi-static, 3D, 
> elastic simulation that I am working. My simulations contains two faults 
> (a thrust fault and a strike slip fault, see attached file for further 
> details) that intersect. I use FaultCohesiveKin object to specify zero 
> fault slip for all times during my simulations in both faults. The goal 
> is to look at the stresses distribution along the faults. Please refer 
> to the attached file for a complete description of boundary conditions 
> and geometry of the model.
> 
> At time step t=0, I observe that the left-lateral shear 
> stress oscillates wildly in both faults, but specially in the thrust 
> fault. The same occurs with less intensity on the up-dip shear and it 
> does not occur with the normal stress. I understand that t=0  time step 
> is the one where PyLith computes the quasi-static solution 
> (elastic_prestep=True).
> 
> In the next time steps (e.g. t=5 days) it seems that 
> the oscillations disappear, note that I have time independent boundary 
> conditions. I don't understand why this happens.
> 
> I suspect that something is wrong with my definition of the up_dip 
> parameter. Currently I have up_dir=[0,1,0] but I have tried also 
> up_dir=[1,0,0] without success. See attached for my complete .cfg file.
> 
> Would you know why the shear stress oscillates at t=0 days and why does 
> it get smooth at t=5 days ?  Would this be related to the up_dir 
> direction definition ?
> 
> thank you in advance
> Josimar
> 
> 
> 
> 
> 
> 
>