Skip to Main content Skip to Navigation

Model-based control-oriented scenario construction in tokamaks

Abstract : This thesis is concerned with developing and applying numerical tools in order to optimize the operation of the poloidal magnetic field (PF) system in tokamaks. The latter consists of a set of coils and power supplies which have the purpose of controlling the plasma shape and position, as well as driving the plasma current. The global context of our work is introduced in Chapter 1. Chapter 2 describes our approach, which consists in applying optimal control methods to the Free-Boundary plasma Equilibrium (FBE) problem, which is composed of a force balance equation in the plasma coupled to Maxwell’s equations in the whole tokamak. The numerical tool employed here is the FEEQS.M code, which can be used either (in the “direct” mode) as a solver of the FBE problem or (in the “inverse” mode) to minimize a certain function under the constraint that the FBE equations be satisfied. Each of these 2 modes (“direct” and “inverse”) subdivides into a “static” mode (which solves only for a given instant) and an “evolution” mode (which solves over a time window). The code is written in Matlab and based on the Finite Elements Method. The non-linear nature of the FBE problem is dealt with by means of Newton iterations, and Sequential Quadratic Programming (SQP) is used for the inverse modes. We stress that the “inverse evolution” mode is a unique feature of FEEQS.M, as far as we know. After describing the FBE problems and the numerical methods and some tests of the FEEQS.M code, we present 2 applications. The first one, described in Chapter 3, concerns the identification of the operating space in terms of plasma equilibrium in the ITER tokamak. This space is limited by the capabilities of the PF system, such as the maximum possible currents, field or forces in the PF coils. We have implemented penalization terms in the “objective” function (i.e. the function to be minimized) of the “inverse static” mode of FEEQS.M in order to take some of these limits into account. This allows calculating in a fast, rigorous and automatic way the operating space, taking these limits into account. This represents a substantial progress compared to “traditional” methods involving much heavier human intervention. The second application, presented in Chapter 4, regards the development of a fast transition from limiter to divertor plasma configuration at the beginning of a pulse in the WEST tokamak, with the motivation of reducing the plasma contamination by tungsten impurities. Here, FEEQS.M is used in “inverse evolution” mode. Data from a WEST experimental pulse is used to set up the simulation. The FEEQS.M calculation then provides optimized waveforms for the PF coils currents and power supplies voltages to perform a fast limiter to divertor transition. These waveforms are first tested on the WEST magnetic control simulator (which embeds FEEQS.M in “direct evolution” mode coupled to a feedback control system identical to the one in the real machine) and then on the real machine. This allowed speeding up the transition from ~ 1 s to 200 ms.
Complete list of metadatas

Cited literature [106 references]  Display  Hide  Download
Contributor : Abes Star :  Contact
Submitted on : Tuesday, June 16, 2020 - 10:29:14 AM
Last modification on : Wednesday, October 14, 2020 - 4:24:25 AM


Version validated by the jury (STAR)


  • HAL Id : tel-02869594, version 1



Xiao Song. Model-based control-oriented scenario construction in tokamaks. Automatic Control Engineering. COMUE Université Côte d'Azur (2015 - 2019), 2019. English. ⟨NNT : 2019AZUR4089⟩. ⟨tel-02869594⟩



Record views


Files downloads