Next-generation Advanced Modular Reactors are expected to aid in the worldwide transition to net-zero emissions. However, these reactors are designed to operate at very high temperatures, which makes the structural integrity of the reactor critical to its design. The structural integrity of previous nuclear reactor designs has been calculated using deterministic methods, but this approach can lead to overly-conservative (or in some cases, optimistic) assessments. Probabilistic structural integrity analysis methods offer the potential to increase understanding of uncertainty and variability in the reactor design, which can reduce design conservatism or increase robustness. Classical Monte-Carlo analysis has been widely used for probabilistic structural integrity analysis, but this approach can be computationally expensive, making it unsuitable for use when iterating and refining reactor designs. This paper assesses alternative probabilistic structural integrity analysis methods, from simple, pseudo-stochastic look-up curve methods to a more complex surrogate Monte-Carlo analysis method using Gaussian process regression. Of the pseudo-stochastic methods, the log-normalized damage margin (LNDM) method is recommended, as it is the most robust, even when used with complex creep laws. The surrogate Monte-Carlo analysis (using a Gaussian process regressor as a surrogate) was found to generate results that were just as accurate and precise as classical Monte-Carlo analysis, but with several orders of magnitude lower computational cost.

This content is only available via PDF.
You do not currently have access to this content.