Dear Suresh,
I must admit that the tangent stiffness for the isotropic damage model is probably not implemented yet.
I had a look at
IsotropicDamageMaterial :: give3dMaterialStiffnessMatrix
and at similar methods for plane stress, plane strain etc., and in all cases the material stiffness matrix can
be the elastic one or the secant one. In our simulations of localized failure we typically use the secant
stiffness, which gives a lower convergence rate but is more robust. With the tangent stiffness, convergence
is usually lost when softening occurs (of course this is our experience with other models, such as smeared
crack, because for the isotropic damage model the tangent stiffness is not available in oofem).
The implementation of the full tangent stiffness is available for a special version of the nonlocal model,
because the transition to localization is more gradual for the nonlocal model and the tangent stiffness
works fine (but is very expensive). You can have a look at the method
IDNLMaterial :: giveRemoteNonlocalStiffnessContribution
in idmnl1.C. The tangent stiffness is implemented here only for two expressions for equivalent strain,
namely standard Rankine eq. strain and energy-based eq. strain. One could use this for the local model, too,
by keeping the nonlocal characteristic length so small that there would be no nonlocal interaction among
Gauss points. Of course, running the code like this would not be very efficient compared to the pure local
model, but one could use the tangent stiffness. Otherwise the tangent stiffness should be implemented
for the local model. I am planning to do it, but it will take some time.
Sorry about the confusion, I was a bit surprised myself that the tangent stiffness is not implemented.
Best regards,
Milan Jirasek