Public Abstract

DE-SC0021385: High-performance Scenario Development in NSTX-U via Physics-based Modeling and AI+Model-based Control

Award Status: Active

Institution: Lehigh University, Bethlehem, PA
UEI: E13MDBKHLDB5
DUNS: 808264444

Most Recent Award Date: 05/18/2026
Number of Support Periods: 6
PM: King, Joshua

Current Budget Period: 03/01/2026 - 02/28/2027
Current Project Period: 03/01/2026 - 02/28/2031
PI: Schuster-Rosa, Eugenio

Supplement Budget Period: N/A

Public Abstract

Realizing the potential of the Spherical Tokamak (ST) concept hinges on the development of high-confinement, low-disruptivity, advanced scenarios. The Lehigh University (LU) Plasma Control Group (PCG) possesses unique expertise in understanding and controlling the complex dynamics of these advanced ST plasmas, crucial for a compact U.S. Fusion Pilot Plant (FPP). A core strength of the LU-PCG is model-based control, integrating physics-based understanding with data-driven techniques. This approach, underpinned by advancements in plasma physics and subsequent control-oriented modeling, is central to the LU-PCG’s work, enabling the incorporation of plasma physics into control design. The LU-PCG employs a unique suite of modeling/analysis tools: CGYRO, a gyrokinetic code for both predictive/interpretive turbulence simulations and reduced-model development for the National Spherical Torus Experiment-Upgrade (NSTX-U); MMM, a fluid-based multiscale transport model suitable for ST conditions (high β, low aspect ratio) and used for understanding and predicting plasma dynamics; and COTSIM, a modular 1.5D control-oriented simulator in Matlab/Simulink®, integrating neural-network (NN) surrogates (MMMnet, NUBEAMnet, TORICnet) and fixed/free-boundary solvers to enable fast discharge simulations, scenario optimization, control synthesis/testing, and reinforcement learning. These tools synergistically advance the physics understanding and control solutions needed for advanced scenarios in NSTX-U. During the FY2020-25 period, the LU-PCG achieved significant progress in: i- validating a reduced microtearing mode transport model against gyrokinetic simulations and improving agreement with electron temperature profiles in NSTX; ii- enhancing MMM with a new two-fluid-based electron temperature gradient (ETG) model and achieving predictions for ion/electron transport in NSTX discharges that align closely with CGYRO simulations and experimental observations; iii- uncovering complex ETG threshold behavior in ST and Kinetic Ballooning Mode (KBM) turbulence regulation via global gyrokinetic simulations, demonstrating the role of self-generated zonal flows; iv- identifying novel electron-scale tearing-parity modes driven by negative density gradients; explaining observed power flow in high-β ST regimes; v- developing MMMnet, a fast NN surrogate for MMM9.1, integrated into COTSIM to enable fast, accurate, self-consistent, equilibrium+transport simulations for NSTX-U; vi- advancing model-based control capabilities for simultaneous profile (e.g., q) and scalar regulation, leveraging real-time optimization techniques for both continuous-time and discrete-time actuator dynamics. The overall objective of this project, which is aligned with the NSTX-U Five-Year-Plan 2026-2030, is for the LU-PCG to impact the NSTX-U program through the development of model-based and AI-driven advanced-scenario control approaches, by advancing turbulence transport theory through MMM/CGYRO-based core/pedestal transport modeling at low collisionality and high β. The LU-PCG’s FY2026-2031 effort will focus on: i- utilizing MMM and CGYRO to characterize and validate confinement and transport behavior in low-collisionality+high-β plasmas, quantifying scaling with collisionality and the impact of shaping; ii- investigating the impact of NBI/RF-driven rotation and fast-ion populations on core and pedestal transport and turbulence suppression; iii- conducting linear and nonlinear, local and global, multiscale CGYRO simulations, complemented by synthetic diagnostics for direct comparison with NSTX-U turbulence measurements; iv- developing NN surrogate models to enable real-time transport and source prediction for integrated simulations in COTSIM; v- enhancing core-edge modeling in COTSIM by implementing a model for anomalous poloidal momentum transport and coupling with equilibrium solvers to simulate L-H/H-L transitions and pedestal evolution; vi- integrating equilibrium and profile control in COTSIM, considering plasma shape as both a constraint and a controlled variable; vii- optimizing the full discharge based on COTSIM for robust realization and sustainment of advanced scenarios; viii- integrating reduced models of the Scrape-Off Layer/wall/divertor regions into COTSIM to enable heat-flux management within advanced-scenario control strategies; ix- developing state observers based on COTSIM to enhance real-time plasma state estimation, prior to implementing the LU-PCG’s control solutions on the NSTX-U Plasma Control System (PCS). This multidisciplinary project provides unique opportunities for young scientists, offering training in fusion and plasma physics, control theory, machine learning, and computational methods. The LU-PCG’s collaboration with NSTX-U and the Princeton Plasma Physics Laboratory (PPPL) provides access to world-leading facilities and expertise.

Realizing the potential of the Spherical Tokamak (ST) concept hinges on the development of high-confinement, low-disruptivity, advanced scenarios. The Lehigh University (LU) Plasma Control Group (PCG) possesses unique expertise in understanding and controlling the complex dynamics of these advanced ST plasmas, crucial for a compact U.S. Fusion Pilot Plant (FPP). A core strength of the LU-PCG is model-based control, integrating physics-based understanding with data-driven techniques. This approach, underpinned by advancements in plasma physics and subsequent control-oriented modeling, is central to the LU-PCG’s work, enabling the incorporation of plasma physics into control design.

The LU-PCG employs a unique suite of modeling/analysis tools: CGYRO, a gyrokinetic code for both predictive/interpretive turbulence simulations and reduced-model development for the National Spherical Torus Experiment-Upgrade (NSTX-U); MMM, a fluid-based multiscale transport model suitable for ST conditions (high β, low aspect ratio) and used for understanding and predicting plasma dynamics; and COTSIM, a modular 1.5D control-oriented simulator in Matlab/Simulink®, integrating neural-network (NN) surrogates (MMMnet, NUBEAMnet, TORICnet) and fixed/free-boundary solvers to enable fast discharge simulations, scenario optimization, control synthesis/testing, and reinforcement learning. These tools synergistically advance the physics understanding and control solutions needed for advanced scenarios in NSTX-U.

During the FY2020-25 period, the LU-PCG achieved significant progress in: i- validating a reduced microtearing mode transport model against gyrokinetic simulations and improving agreement with electron temperature profiles in NSTX; ii- enhancing MMM with a new two-fluid-based electron temperature gradient (ETG) model and achieving predictions for ion/electron transport in NSTX discharges that align closely with CGYRO simulations and experimental observations; iii- uncovering complex ETG threshold behavior in ST and Kinetic Ballooning Mode (KBM) turbulence regulation via global gyrokinetic simulations, demonstrating the role of self-generated zonal flows; iv- identifying novel electron-scale tearing-parity modes driven by negative density gradients; explaining observed power flow in high-β ST regimes; v- developing MMMnet, a fast NN surrogate for MMM9.1, integrated into COTSIM to enable fast, accurate, self-consistent, equilibrium+transport simulations for NSTX-U; vi- advancing model-based control capabilities for simultaneous profile (e.g., q) and scalar regulation, leveraging real-time optimization techniques for both continuous-time and discrete-time actuator dynamics.

The overall objective of this project, which is aligned with the NSTX-U Five-Year-Plan 2026-2030, is for the LU-PCG to impact the NSTX-U program through the development of model-based and AI-driven advanced-scenario control approaches, by advancing turbulence transport theory through MMM/CGYRO-based core/pedestal transport modeling at low collisionality and high β. The LU-PCG’s FY2026-2031 effort will focus on: i- utilizing MMM and CGYRO to characterize and validate confinement and transport behavior in low-collisionality+high-β plasmas, quantifying scaling with collisionality and the impact of shaping; ii- investigating the impact of NBI/RF-driven rotation and fast-ion populations on core and pedestal transport and turbulence suppression; iii- conducting linear and nonlinear, local and global, multiscale CGYRO simulations, complemented by synthetic diagnostics for direct comparison with NSTX-U turbulence measurements; iv- developing NN surrogate models to enable real-time transport and source prediction for integrated simulations in COTSIM; v- enhancing core-edge modeling in COTSIM by implementing a model for anomalous poloidal momentum transport and coupling with equilibrium solvers to simulate L-H/H-L transitions and pedestal evolution; vi- integrating equilibrium and profile control in COTSIM, considering plasma shape as both a constraint and a controlled variable; vii- optimizing the full discharge based on COTSIM for robust realization and sustainment of advanced scenarios; viii- integrating reduced models of the Scrape-Off Layer/wall/divertor regions into COTSIM to enable heat-flux management within advanced-scenario control strategies; ix- developing state observers based on COTSIM to enhance real-time plasma state estimation, prior to implementing the LU-PCG’s control solutions on the NSTX-U Plasma Control System (PCS).

This multidisciplinary project provides unique opportunities for young scientists, offering training in fusion and plasma physics, control theory, machine learning, and computational methods. The LU-PCG’s collaboration with NSTX-U and the Princeton Plasma Physics Laboratory (PPPL) provides access to world-leading facilities and expertise.