Multi-Objective Design Optimization for Fusion Energy Systems

This research presents Multi-Objective Optimization (MOO) of fusion concepts based on engineering, physics, and economic targets to develop efficient and cost-effective fusion energy systems. The method employs modelling parametric geometries, which are evaluated using comprehensive neutronics, thermal, and mechanical simulations, generating training datasets to develop surrogate models. These surrogate models are used to perform MOO, which uses an evolutionary optimization technique to handle complex, nonlinear, and multi-dimensional design spaces, allowing for exploration of diverse configurations and trade-offs between conflicting objectives. The MOO process incorporates equality and inequality constraints, and the ability to adjust weights assigned to engineering, physics, and economic targets which dynamically identifies solutions that align with prioritized goals. The results demonstrate that MOO can rapidly explore expansive parametric spaces and identify optimal designs with significantly reduced computational effort. Given that fusion systems are intricate, massively complex, and interconnected, traditional methods struggle to optimize each subsystem and the overall system cohesively. By addressing these challenges, this research provides fusion energy stakeholders—including designers, policymakers, and investors—with a systematic approach to achieving balanced, optimized designs that meet multiple competing criteria, thereby accelerating the path toward practical fusion energy deployment