Dynamic behaviour characterization of materials is still the cornerstone of many research programs in mechanics. It leads to the development of non-linear constitutive models (e.g., elasto-viscoplasticity) involving several parameters that need to be identified for each considered material. Usually, these models are characterized using simply the homogeneous response of uniaxial specimens submitted to different loading rates. Since the identified models are applied to predict the behavior of complex structures undergoing various stress states, this standard approach eventually leads to unsatisfactory predictions. To achieve a satisfactory level of prediction for a complex full-scale structure to in service loadings, a building block approach can be implemented.

Two options can be considered for a building block approach. The first option is to perform systematically a full characterization of the material behaviour mechanisms (e.g., anisotropy, strain-rate effects, . . . ) onto material samples submitted to a variety of loading conditions by studying their homogeneous response. With this option, the material database is necessarily large enough to expect a correct prediction of the mechanical behavior of complex components. However, this type of systematic extensive characterization is expensive and might be unnecessary for material that exhibit a rather simple mechanical behaviour.

Another option for the building block approach - perhaps more frugal - is to perform only a minimal characterization using material samples submitted to uniaxial loading with homogeneous responses and to move on to small-scale component tests based on constitutive models errors. From the observed discrepancy, it is possible to define additional tests on increasingly complex and
large components to update the database iteratively until the constitutive model can be validated. The latter approach is even more relevant for characterization under dynamic loading conditions for which experimental devices are more restricted to uniaxial loading : in this case, the validation of material models based on heterogeneous states (e.g., in terms of stress , strain-rates, . . . ) is more appropriate. In order to study heterogeneous fields, or more generally to analyze tests/structures, it is interesting to use full-field measurement techniques. Indeed, owing to recent spectacular improvements in optics, it is now possible to extract full-field kinematic maps (displacement, acceleration, strain) that are accurate enough using high-speed cameras, even at high temporal resolution (several thousands of images). Finally, for this second approach chosen for this work, the main challenge is to design informative complex components tests (including the specimen geometry as well as the metrological toolchain).

Applicant profile : MSc in Mechanical/Materials Engineering, Computer Science, Numerical Analysis or equivalent

Appreciated skills : Non-linear/Computational Mechanics, good proficiency with Python or DIC framework