COTSIM-based Feedforward Control Optimizer for Model-based Scenario Planning in NSTX-U

D. Dang, B. Leard, T. Rafiq, E. Schuster

Division of Plasma Physics (DPP) Annual Meeting of the American Physical Society (APS)

Pittsburgh, PA, USA (Remote), November 8-12, 2021

Abstract

By combining plasma-response predictive models and nonlinear optimization techniques, the scenario planning problem in tokamaks like NSTX-U can be written as a model-based, constrained, feedforward-control, optimization problem as originally proposed in [1]. In this optimization problem the time evolutions of actuators such as the powers of heating and current drive (H&CD) sources are determined by minimizing a cost function quantifying the distance between the actual and desired plasma state. The optimization problem, which admits any user-defined cost function, is solved subject to plasma-dynamics constraints, actuator constraints such as the maximum amount of H&CD power or the total plasma current ramp rate, and plasma-state constraints such as the minimum-value of the safety-factor profile or the maximum value of normalized beta. The plasma-dynamics constraints for NSTX-U are provided in this work by the Control Oriented Transport SIMulator (COTSIM). The resulting optimization problem is solved by employing the Sequential Quadratic Programming (SQP) technique. The technique is illustrated by several scenario planning problems in NSTX-U.

[1] Y. Ou, C. Xu, E. Schuster et al., Plasma Physics and Controlled Fusion, 50 (2008) 115001.

*Supported by the US DOE under DE-SC0021385.