search

GEOS-DEV / GEOS

GEOS Simulation Framework
GNU Lesser General Public License v2.1
210 stars 84 forks source link

Fixed issue #2853 with a flag #3108

Closed liyangrock closed 3 months ago

liyangrock commented 5 months ago

This PR addressed issue #2853 by adding a flag to determine if the lagging update should be applied to the fracture stencil weights.

This was done specifically for the hydrofracture solver, as it was observed that issue #2853 was currently causing performance degradation only for this particular solver. Therefore, the flag was exclusively introduced to the hydrofracture solver.

For further discussions related to issue #2853, please see #2922 .

Guotong-Ren commented 5 months ago

To understand, is the flow area in fracture from last time step? Do you have any thoughts on evaluation of the time step size impact on the solution?

liyangrock commented 5 months ago

To understand, is the flow area in fracture from last time step? Do you have any thoughts on evaluation of the time step size impact on the solution?

Indeed, the calculation of fracture stencil weights is performed using the hydraulicAperture from the previous time step. I have conducted a comparison between dt=0.25s and dt=5.0s for viscosity-dominated KGD cases. The resulting net pressure and aperture at the mouth are displayed below.

The difference between the numerical result (with dt=0.25s) and the analytical result can be attributed to the fact that the numerical scenario employs a low toughness value (i.e., 1e4), whereas the analytical solution does not incorporate any toughness.

kgd_verification

The input files are included here. kgdVis.zip

paveltomin commented 3 months ago

@rrsettgast how to merge this?

rrsettgast commented 3 months ago

@paveltomin . Can you create a branch on the main repo that contains these changes? that would be easiest for me...otherwise I have to just merge it and hope for the best.

rrsettgast commented 3 months ago

replaced with https://github.com/GEOS-DEV/GEOS/pull/3196