Towards a data platform for multimodal 4D mechanics of material microstructures

This paper presents advances in the data management strategy applied to 4D multimodal mechanics of material sample microstructures. Guidelines to build a data platform allowing for complex workflows, involving several high-throughput experimental and numerical techniques, and complying with FAIR data management principles, are discussed. Next, their implementation within the open-source Python package Pymicro is presented, offering a high-level interface to build complex datasets through multimodal methodologies. Its capability to enable and automate complex workflows by building a digital twin of a commercially pure titanium sample under tension are then demonstrated. The digital twin contains microstructural and mechanical data for thousands of grains gathered on the same sample through synchrotron DCT and in-situ SEM experiments, as well as full-field numerical simulation. Finally, a local and statistical comparison between simulation and measurements of plastic slip and crystal rotation in hundreds of grains is shown, as an example of the contribution of this platform to multimodal data convergence and its importance for the development of a new generation of material models. • A data management proposition for multimodal microstructure mechanics. • An open-source Python package implementing a data platform interface. • A 4D multimodal digital twin dataset for a polycrystalline Titanium sample with thousands of grains. • A detailed comparison of plastic activity predicted by crystal plasticity in a 3D microstructure and observed experimentally.