Divertor Heat Flux Control Design for High Heat Flux Tokamaks
David Eldon, General Atomics (Principal Investigator)
Jeremy Lore, Oak Ridge National Laboratory (Co-Investigator)
Himank Anand, General Atomics (Co-Investigator)
The principal goal of the project, “Divertor Heat Flux Control Design for High Heat Flux Tokamaks,”
is to develop an integrated toolset for time-dependent divertor heat flux simulation and controller design.
This toolset is intended to provide divertor heat flux control algorithms suitable for application to
tokamaks with high unmitigated divertor heat fluxes, a problem that is common to compact spherical
tokamaks (STs) and high-field standard aspect ratio tokamaks. The focal device for initial implementation
of this toolset will be the SPARC tokamak.
The proposal is a collaborative effort between Oak Ridge National Laboratory (ORNL) and General
Atomics (GA), leveraging the expertise in boundary simulation and controller design and implementation
from each institution. For a series of critical operational scenarios for SPARC identified by
Commonweath Fusion Systems (CFS), the team members at ORNL will perform high-fidelity timedependent
simulations of the tokamak boundary plasma using the SOLPS-ITER transport code. The team
members at GA will develop and implement synthetic diagnostics as scripts or plugins for the SOLPS
framework based on the proposed diagnostics measurement locations specified by the SPARC team. GA
and ORNL will use the SOLPS simulation results, including the synthetic diagnostic data for developing
reduced-order linear and non-linear models for the SPARC scrape-off layer plasma with real-time control
capability. Different modelling methodologies will be applied (dynamic mode decomposition, sparse
identification of nonlinear dynamics, occupation kernel, and first order plus dead time) to describe the
evolution of key operational parameters, such as the divertor heat flux, upstream density, and volumetric
quantities in response to external actuation, such as fuel ion and impurity gas puff magnitude, injected
power, and strike point position. Simple (proportional-integral-derivative [PID] controller) and complex
(model predictive controller [MPC]) control algorithms will be developed and their closed-loop
performance will be evaluated based on the lightweight reduced order model. In addition, a subset of
control variables depending on the type of the detachment regime will be identified and characterized
based on the assumptions for associated noise and delay for diagnostics provided by the SPARC team. In
preparation for SPARC experimental data, the reduced models, along with the developed PID and MPC
systems will be implemented and tested in closed-loop, with the help of the full high-fidelity timedependent
SOLPS model. The developed closed loop simulation environment will also be coupled with
the Integrated Plasma Simulator framework for understanding the effect on the core-edge integration with
the proposed divertor detachment controllers.
Given the extreme challenge of detachment control in SPARC, which is predicted to have extremely
narrow heat flux widths and high divertor fluxes, the proposed tool development of SPARC will prove
applicable and valuable to devices with common physics challenges, specifically compact designs such as
the ST.