Mesh Dependence in MenetreyWillam
Hello everyone,
I am trying to model the cracking behaviour of unreinforced concrete or screed. I am using the MenetreyWillamModel and according to the underlying crackband model, I need to define a tension softening behavior via the yield function Omega_t (curve of tensile relative stress over plastic strain kappa) without losing the objectivity of the model. I can either use exponential softening or linear softening. In the case of exponential softening the parameters fracture energy G_ft and residual relative stress Omega_tr have to be defined. In linear softening, the parameters are again the residual relative stress Omega_t and the plastic strain limit kappa_tr.
Now my question concerns the mesh dependency: In the MenetreyWillam documentation in the Ansys Help there is a parameter "effective element length" to "overcome meshdepenedent softening behavior". If I use exponential softening, the effective element length influences the model obviously through the yield function (equation 447 in the documentation). But in linear softening I have to define the yield function directly through the parameter of the plastic strain limit kappa_tr.
So the effective element length doesn't influence the material model in case of linear softening, does it? So the linear softening behavior is fully meshdependent, isn't it?
Or is there another way, in which the effective element length influences the material behavior in the case of linear softening?
Thank you in adavance for your help!
Kind regards
Katharina
Best Answer

jjdoyle PittsburghMember
Hi Katharina:
In general, softening plasticity models are always mesh dependent unless the formulation somehow takes into account the mesh size. The MenetreyWillam exponential softening option does account for the element size as you have pointed out. The objective here is for the plastic deformation, during softening, to dissipate the same amount of energy, independent of the mesh size.
The linear softening option doesn't do anything to address mesh dependency, so it will always be mesh dependent. The magnitude of this dependency might not be significant enough to affect the results, but the effect should be investigated.
It is generally recommended that you calibrate the model and results to an experiment that closely matches the real service conditions.
Regards,
John
Answers
Hi Katharina:
In general, softening plasticity models are always mesh dependent unless the formulation somehow takes into account the mesh size. The MenetreyWillam exponential softening option does account for the element size as you have pointed out. The objective here is for the plastic deformation, during softening, to dissipate the same amount of energy, independent of the mesh size.
The linear softening option doesn't do anything to address mesh dependency, so it will always be mesh dependent. The magnitude of this dependency might not be significant enough to affect the results, but the effect should be investigated.
It is generally recommended that you calibrate the model and results to an experiment that closely matches the real service conditions.
Regards,
John
Hi John,
thank you for your reply, it was very helpful! The magnitude of the dependency is higher for unreinforced concrete, I believe. And if experimantal data is neccessary for calibrating the model, I cannot really use it for predicting largescale behaviour...?
I have another question concerning the MetereyWillam Model, maybe you (or someone else?) can help with that as well: In the documentation I found the constriction "excluding plane stress" concerning the use of element type PLANE182/183 and "excluding layered structural solid" for SOLID185/186. Do you know, why plane stress and layered solids are excluded for the MenetreyWillam model? The documentation doesn't give any explanation as far as I can tell.
Thank you in advance!
Katharina