Multiscale, Multi-fidelity and Multiphysics Bayesian Neural Network (BNN) Machine Learning (ML) Surrogate Models for Modelling Design Based Accidents

A floating nuclear power plant (NPP) is a site with one or more nuclear reactors located on a platform at sea.

It is an autonomous site that can provide electrical power and process heat to countries with a small land mass, but which are geographical close to the sea. These types of NPPs can also provide fresh clean drinking water to dry areas via desalination techniques.

They can be built using modern modular construction techniques at a factory or shipyard, eliminating the need to set up a nuclear licensed site for its construction and operation.

The location of these types of NPPs is also greatly simplified since it is not necessary to conduct viability studies on the land and land environment. However, the sea or coastal environment does make it necessary to take several factors into account.

These factors include the access for operational staff and equipment as well as the need to ensure that any radioactive material cannot leak into the sea.

Given the widespread development, and deployment, of large scale, medium scale and small modular PWRs, the aim of this PhD proposal is to focus on the analysis of these types of NPPs within this PhD proposal.

This enables Singapore to gain experience in understanding the operational and safety aspects of PWRs; and floating NPPs barges and platforms.

In addition, this proposal will enable Singapore to develop key skills and technology for modelling and simulating (M&S) the operational behaviour of PWRs as well as design basis accidents (DBA) in PWRs.

It will also build upon the current PhD studentship that is funded by National University of Singapore (NUS) Nuclear Research and Safety Initiative (SNRSI) which is focussed on developed a Bayesian neural network (BNN) based surrogate modelling and simulation (M&S) framework for investigating thermal fatigue issues associated with load following floating NPPs.