thelfer
commented
5 months ago @octave-user This is seems a very interesting project !
The order of the strain tensor and deformation gradient is given here: https://thelfer.github.io/tfel/web/tensors.html
The finite strain case shall not be that challenging. The main difference is in the load function which requires to specify the stress measure and the consistent tangent operator.
auto o = FiniteStrainBehaviourOptions{};
o.stress_measure = FiniteStrainBehaviourOptions::PK2;
o.tangent_operator = FiniteStrainBehaviourOptions::DS_DEGL;
const auto d = load(o, library, behaviour, h);You can use the isStandardFiniteStrainBehaviour function to know if you shall use the "standard" load function or the one related to finite strain behaviour.
For example, this is the code that I wrote for the MFEM-MGIS project:
namespace mfem_mgis {
std::unique_ptr<Behaviour> load(const std::string& l,
const std::string& b,
const Hypothesis h) {
using namespace mgis::behaviour;
if (isStandardFiniteStrainBehaviour(l, b)) {
auto opts = FiniteStrainBehaviourOptions{};
opts.stress_measure = FiniteStrainBehaviourOptions::PK1;
opts.tangent_operator = FiniteStrainBehaviourOptions::DPK1_DF;
return std::make_unique<Behaviour>(mgis::behaviour::load(opts, l, b, h));
}
return std::make_unique<Behaviour>(mgis::behaviour::load(l, b, h));
} // end of load
} // end of namespace mfem_mgisI won't have the time to have a look at your code this week, but if you want, we could have a short meeting next week.
octave-user
Dear Mr. Thomas Helfer,
I would like to integrate your MFrontGenericInterfaceSupport library into the general purpose multibody dynamics solver MBDyn. So far, the small strain option is already working. However, the finite strain option seems to be more challenging. On one hand, I'm not 100% sure if I got the order of the tensor components right, and on the other hand I'm missing an option for the tangent operator. So, the derivative of the second Piola Kirchhoff Stress versus the deformation gradient "DS_DF" would be my favorite option.
You can find the current state of my implementation here: MFrontGenericInterfaceCSL and here: ComputeStressElasticDefGrad I'm also providing a link to my reference implementation of Mooney Rivlin which I'm using to validate the Signorini.mfront model. So, far I was able to run a simple uni-axial tensile test with this model, but the result was not as expected. Probably, I was using Cauchy stress instead of First Piola Kirchhoff stress, although the stress_measure was set to PK1. MooneyRivlinStressTensor
See also this link mbdyn-input-develop.pdf at page 313.
Any help would be appreciated.
Best regards, Reinhard