Improving Coil Optimization Using Structural Risk Surrogates

Early coil optimization often overlooks mechanical constraints, risking structural deflection under Lorentz forces that can degrade plasma-facing magnetic fields. We aim to address this gap by providing a metric for introducing said risk early into the optimization process, allowing us to filter coil sets with minimal computational expense. By coupling finite element simulations of coil deflection with field recalculations, normal field errors are computed on the plasma boundary, and the geometric and loading conditions under which structural deformation leads to noticeable field degradation are identified. Through this, reduced-order surrogate metrics/models are constructed that help predict "risk" from early-stage coil optimization features (e.g., curvature, spacing, field gradient), without requiring full FEA within a given structural context. This combined approach would enable early-stage screening of mechanically fragile designs and suggests a path to integrating structural robustness more directly into the optimizer. This closes a design loop between magnetic performance, structural feasibility, and plasma-facing field quality.