Abstract Details
Abstracts
Author: Fu Lanke
Requested Type: Poster
Submitted: 2025-03-12 10:42:59
Co-authors: A. Kaptanoglu, E. Paul, A. Bhattacharjee
Contact Info:
PPPL
129 Faculty Rd.
Princeton, 08540
US
Abstract Text:
Balancing plasma performance and coil cost is a significant challenge when designing a stellarator power plant [1]. Most present stellarator designs are produced by two-stage optimization: the first for an optimized equilibrium and the second for a coil design reproducing its magnetic configuration. However, few proxies connect the coil and equilibrium stage [2]. Therefore, it is challenging to find a compromise between plasma and coils with this approach. In recent years, a new, single-stage approach has gained popularity [3, 4, 5]. This approach optimizes both plasma and coil degrees of freedom simultaneously. It improves the plasma-coil balance but substantially increases the dimensionality of the optimization problem. We present a flexible, differentiable coil proxy that directly predicts coil complexity during equilibrium optimization. Given a user-selected subset of supported coil engineering constraints, our proxy can directly estimate the values and gradients of any coil complexity measure as functions of the plasma boundary shape. Currently supported engineering constraints include curvature, Lorentz force, coil-coil spacing, and dipole array density. In addition, it produces a smooth surface current that can serve as the initial state for high-fidelity coil optimization. The value, gradient, and initial state evaluation require <1s on an Nvidia V100 GPU. Our proxy can offer the benefits of single-stage optimization without increasing the problem’s dimensionality. We present two numerical studies demonstrating the proxy’s effectiveness and validating the accuracy of gradient calculations.
References
1. https://doi.org/10.1109/FUSION.2009.5226449
2. https://doi.org/10.1088/1361-6587/ad1a3e
3. https://doi.org/10.1016/j.jcp.2022.111147
4. https://doi.org/10.1017/S0022377821000271
5. https://doi.org/10.1088/1361-6587/acd957
6. https://doi.org/10.1088/1741-4326/ada810
Characterization: 7.0
Comments: