DIII-D RESEARCH OPPORTUNITIES FORUM FOR THE 2013 EXPERIMENTAL CAMPAIGN Review | Direct submission with log-in | Request submission without log-in
Print this page

Print this page
Title	369: First-principles-driven Model-based Current-profile Control in H-mode Discharges
Name:	Eugenio Schuster schuster@lehigh.edu	Affiliation:	Lehigh University
Research Area:	Plasma Control	Presentation time:	Not requested
Co-Author(s):		ITPA Joint Experiment :	No
Description:	Establishing a suitable current profile has been demonstrated to be a key condition for the achievement of advanced tokamak scenarios with improved confinement and possible steady-state operation. The present approach at DIII-D focuses on creating the desired current profile during the plasma current ramp-up and early flattop phases with the aim of maintaining this target profile during the subsequent phases of the discharge. Previous experiments on DIII-D showed that the high dimensionality of the problem, and the strong coupling between magnetic and kinetic variables, call for the design of a model-based, multi-variable controller that takes into account the dynamic response of the full current profile to the different actuators. The objective of this experiment is twofold. First, a control-oriented first-principles-driven model for the current profile dynamics in H-mode discharges will be developed and validated in DIII-D. Second, based on the developed and validated control-oriented first-principles-driven model, controllers for the regulation of the current profile in H-mode discharges will be designed and tested in DIII-D. Unique characteristics of the control approach are (i) the use of first-principles-driven models for the control synthesis, (ii) the integration of both static and dynamic plasma response models into the design of the feedback controllers, and (iii) the possibility of capturing the nonlinear dynamics of the system during the control synthesis. This experiment is a natural extension of the successful experiments on first-principles-driven model-based current profile control in L-mode carried out in 2011. Extending control scheme to H-mode first requires model extension. In the to-be-developed H-mode current-profile response model: i- the electron and density profile model must include edge transport barrier; ii- non-inductive current drive and heating systems must be modeled individually (not together); iii- the effect of bootstrap current must be included. The controllers developed from first-principles models in L-mode discharges have used so far three actuators - plasma current, beam total power and line-averaged density. By adding EC H&CD as an actuator and grouping the beams in different categories we intend to improve controllability for simultaneous current profile and beta_N regulation in H-mode discharges. The developed control-oriented nonlinear model for current profile response in H-mode discharges will be used to design feedback controllers, which will be tested in scenarios relevant to the Steady State Scenarios and Inductive Scenarios thrusts. It is important to emphasize that with the development of the DIII-D/LU profile control algorithm carried out in 2011, the PCS (plasma control system) at DIII-D does have now the necessary infrastructure for implementing such advanced profile controllers.	ITER IO Urgent Research Task :	No
Experimental Approach/Plan:	Open-loop optimal control laws will be expressed as time trajectories for the actuators: total plasma current, average plasma density, non-inductive current drive (NBI) power and heating (EC) power. The closed-loop controller will regulate in real-time these actuators based on real-time measurements of the q profile. We will assess the ability of the combined open-loop and closed-loop controllers to drive the current profile from an initial condition different from (but close to) the nominal one to a specific target profile. One additional goal of the controllers is to avoid MHD activity in the form of NTMs. Therefore, beta_N will also be regulated in closed-loop. Different initial and target profiles will be considered. The first-principles-driven, model-based, current-profile control experiment in H-mode will require two 2-hour evening sessions and at least one half-day session. It is important to emphasize that the to-be-developed nonlinear control-oriented model could find applications beyond feedback control design. First, this model could be used for feedforward control design (scenario planning). Determining whether a particular current profile is achievable given the initial conditions and actuators constraints, and eventually finding the actuators trajectories that are necessary to achieve a particular achievable current profile are two very important problems arising in tokamak operation that could find solutions by exploiting the developed nonlinear control-oriented model. Second, this model could be used for state estimation and prediction in real time. Noise could be separated from the actual plasma state (current profile) by using the developed nonlinear control-oriented model, which would filter any component of the estimated current profile not predicted by the model. Moreover, the plasma state (current profile) could be estimated in real time from a limited set of diagnostic (not including MSE for instance) by exploiting the prediction by the developed nonlinear control-oriented model. Finally, this model could be used as a simulation testbed. Controllers designed based on more simplified models, including identified linear plasma response models arising in data-driven modeling, could be tested in closed-loop simulations based on the developed nonlinear control-oriented model (controllers developed as part of proposal #316 could greatly benefit from this simulation capability). This will provide the opportunity of systematically comparing first-principles-driven and data-driven approaches to profile control.
Background:	The Plasma Control Group at Lehigh University (LU) headed by Prof. Eugenio Schuster has been working on this problem for several years now. A preliminary first-principle control-oriented model of current profile evolution in response to auxiliary H&CD systems (NBI, EC) and electric field due to induction was developed for L-mode discharges [1]. Optimal open-loop control schemes were developed based on the simplified control-oriented model [2, 3]. These algorithms predict the open-loop (or feedforward) actuator waveforms that are necessary to drive the plasma from a specific poloidal flux initial profile to a predefined final profile during the current ramp-up. Data obtained from the 2008 1/2day experiment showed qualitative agreement between model prediction and experiment, and corroborated that the actuators constraints were correctly taken into account during the control synthesis. A reduced-order first-principles model was obtained from the original simplified control-oriented infinite-dimensional model and combined with Optimal Control and Robust Control theory to synthesize closed-loop controllers [4, 5]. Extensions of these controllers were tested in L-mode discharges in DIII-D in 2011 [6, 7, 8], which represents the first time ever model- based, first-principles-driven, full-magnetic-profile controllers were successfully implemented and tested in a fusion device. [1] Y. Ou, T.C. Luce, E. Schuster et al., Towards Model-based Current Profile Control at DIII-D, Fusion Engineering and Design 82 (2007) 11531160. [2] Y. Ou, C. Xu, E. Schuster et al., Design and Simulation of Extremum-Seeking Open-Loop Optimal Control of Current Profile in the DIII-D Tokamak, Plasma Physics and Controlled Fusion, 50 (2008) 115001. [3] C. Xu, Y. Ou, J. Dalessio, E. Schuster et al., Ramp-Up-Phase Current-Profile Control of Tokamak Plasmas via Nonlinear Programming, IEEE Trans. on Plasma Science, vol.38, no.2, pp.163-173, 2010. [4] Y. Ou, C. Xu and E. Schuster, Robust Control Design for the Poloidal Magnetic Flux Profile Evolution in the Presence of Model Uncertainties, IEEE Trans. on Plasma Science, vol.38, no.3, pp.375-382, 2010. [5] Y. Ou, C. Xu, E. Schuster et al., Optimal Tracking Control of Current Profile in Tokamaks, IEEE Transactions on Control Systems Technology 19 (2), 432-441 (2011). [6] J. Barton, M.D. Boyer, W. Shi, E. Schuster et al., Toroidal Current Profile Control During Low Confinement Mode Plasma Discharges in DIII-D via First-Principles-Driven Model-based Robust Control Synthesis, Nuclear Fusion 52 (2012) 123018 (24pp). [7] M.D. Boyer, J. Barton, E. Schuster et al., First-Principles-Driven Model-Based Current Profile Control for the DIII-D Tokamak via LQI Optimal Control, Plasma Physics and Controlled Fusion, under review. [8] M.D. Boyer, J. Barton, E. Schuster et al., Backstepping Control of the Toroidal Plasma Current Profile in the DIII-D Tokamak, IEEE Transactions on Control Systems Technology, under review.
Resource Requirements:	Machine time: Two 2-hour evening sessions + at least 1/2 day experiment. Actuators: All NB and EC H&CD systems at full power.
Diagnostic Requirements:	Core and tangential Thomson, CER, CO2, magnetics, MSE, ECH diagnostics, a reasonable set of fast ion instability diagnostics (UF interferometers, FIR scattering, ECE at 500 kHz, fast magnetics with fast delay set in the current ramp), FIDA. For closed-loop experiment real-time magnetic measurements and equilibrium reconstruction including the q-profile (EFIT and RTEFIT with MSE) and beta_N are essential, and real-time measurements of the ion and electron temperature profiles, as well as line-averaged plasma density or density profile are required.
Analysis Requirements:	Matlab. MDSPLUS.
Other Requirements:	--