Hi Jorgen -
I've been using MCalibration to fit an Abaqus PRF model (both, 3 and 4 networks with power law flow) to some stress relaxation test data. A few the of tests included several segments:
1) Load to strain 0.1, hold for a few hours, unload
2) Load to strain 0.2, hold for a few hours, unload
3) Load to strain 0.3, hold for a few hours, unload
What I notice is that the model can predict the initial load-hold-unload cycle but in subsequent cycles the stress is increasingly overpredicted on loading. I realize the model is nonlinear so it is difficult to really make a lot of sense of the behavior from the parameters alone but I would still like to gain some intuition so I can try to start MCalibration off with a better initial guess.
Do you know of some simple way of relating a power law flow model to a relaxation time? The reason I ask is I was assuming that if the predicted viscous strain recovers too quickly on unloading that the stress would be overpredicted during the second and third loading segments (since, for a given total strain, only the elastic part is used to compute the stress). If I could relate a relaxation time to the power law model, then I could start each network off with different relaxation time "decades" similar to linear viscoelastic models.
Perhaps I am hoping for too much with such a complex constitutive law but I'd really appreciate your opinion. Thanks a lot for your time.