About

Research Interests (Lay Summary)

Michael's research interests are in the continual improvement and development of computational tools used in the simulation of metal forming processes. A component of the research involves understanding the material behaviour of steel as it is processed, being subjected to large deformations. A variety of models ranging from the micro-scale to the component level are being developed and tested for this purpose. In addition, methods of linking the scales between the micro and component levels are a research topic. Deep learning models are a potential tool that could be used in improving the efficiency of simulations. Data collection on simulation software can allow future simulations to 'learn' from the results of those run previously.

Technical Summary

The simulation of metal forming processes involves the computational modeling of large-strain plasticity. Accurate plasticity models are required in order to predict the final component's geometry and mechanical properties. An efficient spectral method has been applied to describe the macroscopic response of pearlitic steel, based on the results of representative volume element (RVE) micro-mechanical simulations.

Deep learning is applied to track and predict the evolution of the solution variable until it converges at any particular load step in a solid-mechanics simulation. It has been shown that such an evolution can be learned, with the deep learning model being able to give a reasonable estimate to the final converged solution, reducing the number of subsequent solver iterations, and CPU time required.