Integrated computational risk assessment for fusion materials

The quantification of uncertainty in material behaviour and risk of failure is urgently needed to accelerate the design and qualification of fusion reactor components. In particular, plasma-facing materials present in first wall components and breeder blanket modules are subject to extreme temperatures, stresses, and radiation damage during their operating conditions, and are required to maintain structural integrity over a lifetime of at least 5-years.

While recent advances in physics-based material modelling and their deployment on high performance computer systems have made the in-silico qualification of fusion materials feasible, challenges still remain over the predictive reliability of such an approach. On the other hand, though high-fidelity experimental testing is now routine in the fusion materials community, the vast number of tests required to determine reliable statistical estimates of the occurrence of rare failure events makes it prohibitively expensive to perform.

In this project, we propose an integrated approach to risk assessment, relying upon a few targeted high-fidelity experiments and supported by higher throughput, lower fidelity simulations. Focusing on CuCrZr alloys, mechanical testing with high resolution digital image correlation will be performed to obtain full-field strain maps as a function of the local microstructure and applied load.

A Markov-chain Monte Carlo algorithm will then be used to train a lower fidelity crystal plasticity model with uncertainty bounds to the test data. The crystal plasticity model will then be used to perform high throughput simulations to estimate the material strength distribution as a function of microstructure, and statistical analysis of the tail end of this distribution will be performed to obtain material failure risk estimates.

This PhD project will be based at the University of Manchester and will engage with leading academics new to fusion to establish the framework for an in-silico approach to materials risk assessment. Candidates interested in numerical methods, programming, computational modelling and simulation and its application to fundamental materials science are encouraged to apply.

Although a background in engineering and or materials science is beneficial, it is not essential. Interested physicists and applied mathematicians willing to learn materials science and metallurgy would also make excellent candidates.