Previous version of dvalue was incorrect and thus didn't help with convergence issues. The strain derivative of the permeability has to result in a 4x4 matrix for 2D or a 6x6 matrix for 3D. This is now corrected. Therfore Propertydatatype had to be extended by MatrixXd. Also for the correct computation of the dkde part of K_pu, the vectors grad p and (rho_fr*b) had to be lifted to Kelvin space. A simple tests has been added (with biot=1) which converges 5 times faster with the dvalue than without. With larger time steps it wouldn't converge without it.
There is an still an open question regarding the element order: With Taylor-Hood elements, the addition to the jacobian improves convergence as intended, but this is not so with linear elements. If we apply in this case a scaling factor of ~0.7 we still get well improved convergence, but this may also depend on the model and maybe Biot's coefficient.
Feature description was added to the changelog
Tests covering your feature were added?
Any new feature or behavior change was documented?