Public Abstract

DE-SC0010661: Toward Disruption-free, Machine-safe, High-performance Operation in ITER and FPP via Integrated Advanced Control

Award Status: Active

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

Most Recent Award Date: 09/18/2023
Number of Support Periods: 10
PM: Lanctot, Matthew

Current Budget Period: 07/15/2023 - 07/14/2024
Current Project Period: 07/15/2023 - 07/14/2026
PI: Schuster-Rosa, Eugenio

Supplement Budget Period: N/A

Public Abstract

Toward Disruption-free, Machine-safe, High-performance Operation in ITER and FPP via Integrated Advanced Control Eugenio Schuster (Principal Investigator), Lehigh University Plasma Control Laboratory (https://www6.lehigh.edu/~eus204/lab/PCL_fusion.php) Future fusion reactors like ITER or a compact Fusion Pilot Plant (FPP), as proposed by the U.S., will need to robustly operate within safety, stability, and controllability boundaries to eliminate the risk of uncontrolled transients leading to disruptions and damage of the plasma facing components. Robust, integrated, simultaneous control of a large number of plasma properties in the core, scrape-of-layer (SOL) and divertor is a critical technology needed to satisfy this need and to guarantee safe and disruption-free tokamak operation in both steady-state and pulsed machines. A model-based approach emerges as the only possible way of tackling this challenging, nonlinear, high-dimensional, control-design problem, while enabling real-time safety/stability/controllability boundary reconstruction, plasma state and parameter estimation from a limited number of noisy diagnostics, control integration by actuator allocators and reference governors, and both adaptation and optimization of the control response. The Lehigh University (LU) Plasma Control Group (PCG) has pioneered many control-oriented modeling and model-based control-design techniques for advanced-scenario control, strongly impacting not only DIII-D but also major experiments around the world (techniques developed at DIII-D have been and are being extended to NSTX-U, EAST, KSTAR, and ITER). In spite of the significant progress achieved, a great deal of research work is still needed for these advanced-scenario control solutions to reach the level of completeness, integration, and reliability required for ITER, FPP, and future fusion reactors. New and ongoing research work on both control mathematics/physics and machine learning (ML) will be carried out to continue LU-PCG’s educational and research project on integrated, advanced control for disruption-free, machine-safe, high-performance, tokamak-reactor operation. The overall objective of this project is to continue and extend LU-PCG’s work on integration of physics and operation in DIII-D, with the ultimate goals of increasing the physics productivity of experimental time in DIII-D by providing improved and expanded control capability for ongoing and future operation, developing and testing at DIII-D extrapolative solutions to several challenging control problems expected in ITER and FPP, and contributing to resolving important physics issues related to burning plasmas and advanced scenarios. LU-PCG’s research work within this project builds upon recent physics and engineering contributions, including: i- Development of ML-based surrogate models for transport/actuation in LU-PCG’s COTSIM (Control-Oriented Transport SIMulator) for both fast and accurate predictions; ii- Implementation of critical components of COTSIM in the DIII-D Plasma Control System (PCS) to enable real-time profile estimation by leveraging ML surrogate modeling; iii- Development of integrated solutions for advanced-scenario control by tackling the nonlinearity and infinite dimensionality of the plasma, exploiting spatially moving actuators leading to enhanced controllability and effective local regulation of moving targets, and integrating kinetic/magnetic profile control; iv- Design of actuator-sharing strategies for integrated control of multiple plasma scalars/profiles with simultaneous Neoclassical Tearing Mode (NTM) stabilization by both static and dynamic real-time optimization; v- Development of control-oriented models integrating Core, SOL, and Divertor (CSD) dynamics to assess ITER’s and FPP’s operational spaces. The research plan evolves around the following central topics, which have been carefully selected to maximize the impact on the DIII-D and ITER programs: (1) Enhancement of LU-PCG’s COTSIM prediction capabilities for control design at DIII-D and extrapolation to ITER and FPP by ML surrogate modeling, equilibrium/transport solver coupling, and core/edge integration; (2) Further development of adaptive, fault-tolerant, feedforward+feedback control/estimation solutions for advanced scenarios by incorporating new actuation capabilities planned for DIII-D (an upgrade of the heating and current-drive power is planned for DIII-D to enable access to reactor-relevant regimes in order to identify a viable FPP design and to ensure success of the U.S. mission in ITER); (3) Integration of competing controllers in actuator-constrained reactor-level tokamak operation by actuator-allocation algorithms enabled by real-time optimization and ML; (4) Coupling of global/local profile and scalar controllers with reference-governor schemes to regulate the proximity to the NTM stability boundary; (5) Development of density (profile) controllers by simultaneous gas puffing and pellet injection to enhance (DIII-D) or enable (ITER, FPP) operation; (6) Development of divertor-safe core-kinetic (burn) control solutions for both ITER and FPP via reference-governor-based core-edge integration and dynamic actuator allocation. A central goal of the upcoming phase of this ongoing long-term project is to increase the number of LU-PCG's PhD students stationed at DIII-D, tightly align their work with DIII-D’s goals, and enhance their pairing with DIII-D’s scientists in order to train the workforce needed by the U.S. fusion program.

Toward Disruption-free, Machine-safe, High-performance Operation in ITER and FPP via Integrated Advanced Control

Eugenio Schuster (Principal Investigator), Lehigh University

Plasma Control Laboratory (https://www6.lehigh.edu/~eus204/lab/PCL_fusion.php)

Future fusion reactors like ITER or a compact Fusion Pilot Plant (FPP), as proposed by the U.S., will need to robustly operate within safety, stability, and controllability boundaries to eliminate the risk of uncontrolled transients leading to disruptions and damage of the plasma facing components. Robust, integrated, simultaneous control of a large number of plasma properties in the core, scrape-of-layer (SOL) and divertor is a critical technology needed to satisfy this need and to guarantee safe and disruption-free tokamak operation in both steady-state and pulsed machines. A model-based approach emerges as the only possible way of tackling this challenging, nonlinear, high-dimensional, control-design problem, while enabling real-time safety/stability/controllability boundary reconstruction, plasma state and parameter estimation from a limited number of noisy diagnostics, control integration by actuator allocators and reference governors, and both adaptation and optimization of the control response.

The Lehigh University (LU) Plasma Control Group (PCG) has pioneered many control-oriented modeling and model-based control-design techniques for advanced-scenario control, strongly impacting not only DIII-D but also major experiments around the world (techniques developed at DIII-D have been and are being extended to NSTX-U, EAST, KSTAR, and ITER). In spite of the significant progress achieved, a great deal of research work is still needed for these advanced-scenario control solutions to reach the level of completeness, integration, and reliability required for ITER, FPP, and future fusion reactors. New and ongoing research work on both control mathematics/physics and machine learning (ML) will be carried out to continue LU-PCG’s educational and research project on integrated, advanced control for disruption-free, machine-safe, high-performance, tokamak-reactor operation. The overall objective of this project is to continue and extend LU-PCG’s work on integration of physics and operation in DIII-D, with the ultimate goals of increasing the physics productivity of experimental time in DIII-D by providing improved and expanded control capability for ongoing and future operation, developing and testing at DIII-D extrapolative solutions to several challenging control problems expected in ITER and FPP, and contributing to resolving important physics issues related to burning plasmas and advanced scenarios.

LU-PCG’s research work within this project builds upon recent physics and engineering contributions, including: i- Development of ML-based surrogate models for transport/actuation in LU-PCG’s COTSIM (Control-Oriented Transport SIMulator) for both fast and accurate predictions; ii- Implementation of critical components of COTSIM in the DIII-D Plasma Control System (PCS) to enable real-time profile estimation by leveraging ML surrogate modeling; iii- Development of integrated solutions for advanced-scenario control by tackling the nonlinearity and infinite dimensionality of the plasma, exploiting spatially moving actuators leading to enhanced controllability and effective local regulation of moving targets, and integrating kinetic/magnetic profile control; iv- Design of actuator-sharing strategies for integrated control of multiple plasma scalars/profiles with simultaneous Neoclassical Tearing Mode (NTM) stabilization by both static and dynamic real-time optimization; v- Development of control-oriented models integrating Core, SOL, and Divertor (CSD) dynamics to assess ITER’s and FPP’s operational spaces.

The research plan evolves around the following central topics, which have been carefully selected to maximize the impact on the DIII-D and ITER programs: (1) Enhancement of LU-PCG’s COTSIM prediction capabilities for control design at DIII-D and extrapolation to ITER and FPP by ML surrogate modeling, equilibrium/transport solver coupling, and core/edge integration; (2) Further development of adaptive, fault-tolerant, feedforward+feedback control/estimation solutions for advanced scenarios by incorporating new actuation capabilities planned for DIII-D (an upgrade of the heating and current-drive power is planned for DIII-D to enable access to reactor-relevant regimes in order to identify a viable FPP design and to ensure success of the U.S. mission in ITER); (3) Integration of competing controllers in actuator-constrained reactor-level tokamak operation by actuator-allocation algorithms enabled by real-time optimization and ML; (4) Coupling of global/local profile and scalar controllers with reference-governor schemes to regulate the proximity to the NTM stability boundary; (5) Development of density (profile) controllers by simultaneous gas puffing and pellet injection to enhance (DIII-D) or enable (ITER, FPP) operation; (6) Development of divertor-safe core-kinetic (burn) control solutions for both ITER and FPP via reference-governor-based core-edge integration and dynamic actuator allocation. A central goal of the upcoming phase of this ongoing long-term project is to increase the number of LU-PCG's PhD students stationed at DIII-D, tightly align their work with DIII-D’s goals, and enhance their pairing with DIII-D’s scientists in order to train the workforce needed by the U.S. fusion program.