Task 3
Analysis of constraints and search universe in parameter identification of thermelastic-viscoplastic and elastoplastic constitutive models.
It was demonstrated that the success of the material characterization through numerical constitutive models and their parameters depends of the physical constraints introduced in the identification (optimization) process. Previous works [23,PT5] show that the non-introduction of correct process constraints lead to set of parameters with no physical meaning, although they lead to low objective functions (in the optimization process). Additionally, full-filed measurements limitations will also bring constraints that must be taken into account in an accurate identification process.
In this task, the first part of the work is concerning the analysis of the selected constitutive models. The thermoelastoviscoplastic models generally are
(i) unified with internal variables or
(ii) the product of several terms that add the influence of the different phenomena (creep, temperature, etc.).
In the first case, the formulation include several progressive internal variables that characterize the history of deformation, hardening and other phenomena. The second case of models include all models built based on elastoplastic models that were enhanced in order to include creep, temperature and other effects. Nevertheless, for both cases, a selected number of models will be implemented in FEA by the use of a user routine and analysed. These user routine will also be part of the database created in the following task.
Conventional model calibration procedures are shown to introduce non-physical artefacts into constitutive models. These manifests as, for example,
(i)
non-zero hydrostatic stress or through-thickness strain generated under in plane shear stress, that violate the definition of the shear loading condition;
(II)
non-zero deviatoric stress in pure tensile strain, violating the definition of pure traction;
(III)
spurious stress developed during a theoretical bulge test;
(IV)
equivalent results for multiple set of parameters due to multiple scale parameters;
(V)
etc.
To overcome this problems, very careful and selected constraints must be introduced in the identification/calibration procedure for each constitutive model. The search universe for each parameter of each constitutive model must also be very carefully analysed in order to define the parameter’s universe.
Additionally, in the overall material characterization process, imaging requirements cannot be absent. Indeed, DIC is a low pass spatial filter (i.e. strain and temperature gradients will be smoothed out by DIC and thermography, respectively) and its spatial resolution is a key issue, as well at its noise performance. The solution here will be using synthetic image deformation [27].
This task is of maximum importance and it will have researchers in fully exclusivity for it.