1. Enrich the tip nodes with the Heaviside function normally but enforce the enriched dofs as zero in the boundary condition.
This is easy to achieve, also leads to lower leak-off and a drastic pressure drop at the tip. Also, there will be unbalanced internal force vectors at the tip nodes due to the application of Dirichlet boundary condition.
Adopted in the current repository.
On 02/11/2020,
02/11/20
Go back to the previous tip enrichment scheme, where udofs and pdofs of the tip edge nodes are enriched. But only udofs are enforced to be zero. After a second thought, this may be more reasonable because the gradient of pore pressure in the tip element is still discontinuous. When we do not have enriched pdofs, the pore pressure at the two nodes tend to have the same high pressure as the internal crack, which will require a high positive leakoff.
2. Do not enrich the tip ndoes at all.
This was also achieved in one version of the package. But this version leads to a smooth pressure field near the tip and causing a "higher" leak-off, which was not wanted during the verification process in spring 2020.
I need to double-check the theory and the implementation of leak-off. Perhaps, the second approach is more logical as neither the displacement nor pressure dofs need to be additionally specified in boundary condtions. The high leak-off may be attributed to other reasons.
1. Enrich the tip nodes with the Heaviside function normally but enforce the enriched dofs as zero in the boundary condition.
This is easy to achieve, also leads to lower leak-off and a drastic pressure drop at the tip. Also, there will be unbalanced internal force vectors at the tip nodes due to the application of Dirichlet boundary condition.
Adopted in the current repository.
On 02/11/2020,
2. Do not enrich the tip ndoes at all.
This was also achieved in one version of the package. But this version leads to a smooth pressure field near the tip and causing a "higher" leak-off, which was not wanted during the verification process in spring 2020.
I need to double-check the theory and the implementation of leak-off. Perhaps, the second approach is more logical as neither the displacement nor pressure dofs need to be additionally specified in boundary condtions. The high leak-off may be attributed to other reasons.