Speaker
Description
A major challenge for fusion reactors is the production of tritium fuel for sustained fusion reactions, as well as the extraction of heat to generate electricity. Both aspects are directly linked to the breeder blanket (BB) components in a reactor. An extensive literature indicates that various blanket designs have been proposed worldwide, including solid-pebble breeders, liquid lithium-lead breeders, and molten breeders, with different coolant options such as helium, water, or self-cooling mechanisms. However, all these designs remain at the conceptual stage and lack a generalized strategy for initial design methodology. The primary criterion for selecting a breeder blanket includes Tritium Breeding and power extraction. While ensuring the blanket's key function, it is essential to consider multiple factors, including electromagnetic loads, thermal cycling, material temperature constraints, neutronic parasitic capture, and structural integrity due to pressure loss. Addressing all these factors leads to an interconnected neutronic–hydraulic–thermal–structural problem, a complexity not effectively addressed in most existing designs. Although considerable studies have explored breeder blanket designs, limited attention has been given to a solution-agnostic blanket description. Furthermore, performing high-fidelity simulations of many blanket designs during the preliminary stage can result in high computational cost and complexity. To overcome these challenges, we aim to develop a systems engineering approach for breeder blankets, enabling the study of multiple concepts and the down-selection of design options at later stages. Current research focuses on developing the system analysis (SA) methodology based on energy balance to perform thermo-fluid analysis of the breeder blanket unit. The SA method addresses key challenges in fusion blanket modelling by providing a balanced approach between lumped models and high-fidelity simulations. It enables faster computation while maintaining reasonable accuracy and requires fewer assumptions than current lumped methods. Additionally, SA captures a higher level of physics than lumped models and supports multi-dimensional analysis as needed. Its computational efficiency allows rapid evaluation of different geometries, material properties, and operating conditions. Initial analysis and validation against CFD results demonstrate confidence in the SA approach in terms of both accuracy and computational speed compared with high-fidelity simulations. However, the robustness of the method is compromised due to the simplification of complex geometries and High-Fidelity flow physics. The next step of our research addresses this with the introduction of a correction function by employing Reduced Order Model (ROM) or Machine Learning (ML) techniques to augment the standard SA result accuracy and ability to handle complex geometries. This approach effectively maintains computational efficiency while delivering the high accuracy required for complex breeder unit designs.
| Speaker affiliation | Zienkiewicz Institute for Modelling, Data and AI, Faculty of Science and Engineering, Swansea University |
|---|
