J.-M. Moret

Observer-based real-time control for the poloidal beta of the plasma using diamagnetic measurements in tokamak fusion reactors

The constant technological advances and progresses in tokamak research constantly show up new control challenges. In this context, the control of the poloidal beta has arisen as a relevant issue. In this paper a real-time observer based on the real-time analysis of diamagnetic measurements that has been developed for the Tokamak à Configuration Variable (TCV) is presented. The algorithm proposed combine measurements of the diamagnetic loops, flux loops and magnetic probes. Afterwards, some simulations are carried out with the purpose of testing the observer. Finally, once the observer has been developed and validated through simulations, it is implemented on the TCV reactor and applied to the real-time control of poloidal beta of the plasma. The results of the experiments show the feasibility of the observer for real-time tokamak plasma control purposes.

An overview of results from the TCV tokamak

The Tokamak Configuration Variable (TCV) tokamak (R = 0.88 m, a < 0.25 m, B < 1.54 T) programme is based on flexible plasma shaping and heating for studies of confinement, transport, control and power exhaust. Recent advances in fully sustained off-axis electron cyclotron current drive (ECCD) scenarios have allowed the creation of plasmas with high bootstrap fraction, steady-state reversed central shear and an electron internal transport barrier. High elongation plasmas, kappa = 2.5, are produced at low normalized current using far off-axis electron cyclotron heating and ECCD to broaden the current profile. Third harmonic heating is used to heat the plasma centre where the second harmonic is in cut-off. Both second and third harmonic heating are used to heat H-mode plasmas, at the edge and centre, respectively. The ELM frequency is decreased by the additional power. In separate experiments, the ELM frequency can be affected by locking to an external perturbation current in the internal coils of TCV. Spatially resolved current profiles are measured at the inner and outer divertor targets by Langmuir probe arrays during ELMs. The strong, reasonably balanced currents are thought to be thermoelectric in origin.

Optimization of experimental snowflake configurations on TCV

The design of a snowflake (SF) equilibrium requires a strong effort on the poloidal field (PF) currents in terms of MAturns and mechanical loads. This has limited the maximum plasma current in SF configurations on TokamakConfiguration Variable (TCV) to values well below the intrinsic magnetohydrodynamic limits. In this paper the definition of optimized SF configurations in TCV and their experimental tests are illustrated. The PF current optimization procedure proposed in Albanese et al (2014 Plasma Phys. Control. Fusion 56 035008) is adapted and applied to a SF scenario in TCV where the PF currents were close to their operational limits with the aim of reducing the total MAturns in view of higher values of the plasma current. This procedure optimizes the PF currents while fulfilling the machine technological constraints for a given bound on the tolerable plasma shape changes. The method exploits the linearized relation between the plasma-wall gaps and the PF currents. In the investigated TCV scenario the optimization procedure allowed a 20% increase of the plasma current while keeping the plasma shape alignment with respect to the nominal shape within a tolerance of 1 cm. The predicted optimization potential was confirmed in a TCV experiment.

TCV Advanced Plasma Control System Software Architecture and Preliminary Results

A new advanced plasma control system for the TCV Tokamak has been implemented based on an in-house developed high performance DSPs-based VME system, specially designed for real-time plasma control and event detection on fusion experiments. This paper describes the software architecture of the digital system as well as the integration of the physicists control algorithms in the DSPs software. The integration conditions on the TCV control system and the first results of the real-time control operation are also presented. Finally, a critical analysis of the system performance and the experimental operation of TCV APCS is discussed.

A 3D multi-mode geometry-independent RMP optimization method and its application to TCV

Resonant magnetic perturbation (RMP) and error field correction (EFC) produced by toroidally and poloidally distributed coil systems can be optimized if each coil is powered with an independent power supply. A 3D multi-mode geometry-independent Lagrange method has been developed and appears to be an efficient way to minimize the parasitic spatial modes of the magnetic perturbation and the coil current requirements while imposing the amplitude and phase of a number of target modes. A figure of merit measuring the quality of a perturbation spectrum with respect to RMP independently of the considered coil system or plasma equilibrium is proposed. To ease the application of the Lagrange method, a spectral characterization of the system, based on a generalized discrete Fourier transform applied in current space, is performed to determine how spectral degeneracy and side-band creation limit the set of simultaneously controllable target modes. This characterization is also useful to quantify the efficiency of the coil system in each toroidal mode number and to know whether optimization is possible for a given number of target modes.The efficiency of the method is demonstrated in the special case of a multi-purpose saddle coil system proposed as part of a future upgrade of Tokamak a Configuration Variable (TCV). This system consists of three rows of eight internal coils, each coil having independent power supplies, and provides simultaneously EFC, RMP and fast vertical position control.

New developments for real time plasma control system of TCV Tokamak based on FPGA

The vertical instability problem of highly-elongated plasmas has been investigated. A reduced-order model for control purpose has been derived from the full-order plasma RZIP model. Based on reduced model, an optimal control has been designed which gives better performance than a legacy PID control in the presence of measurement noise. The control algorithm is programmed in user-friendly Matlab-Simulink environment. The VHDL code is generated automatically with HDL coder. The control algorithm has been implemented into a COTS FPGA-based system. Hardware in the loop simulation has been performed to test the algorithm in the real world.

Electron Bernstein wave heating of over-dense H-mode plasmas in the TCV tokamak via O-X-B double mode conversion

This paper reports on the first demonstration of electron Bernstein wave heating (EBWH) by double mode conversion from ordinary (O-) to Bernstein (B-) via the extraordinary (X-) mode in an over-dense tokamak plasma, using low field side launch, achieved in the TCV tokamak H-mode, making use of its naturally generated steep density gradient. This technique offers the possibility of overcoming the upper density limit of conventional EC microwave heating. The sensitive dependence of the O-X mode conversion on the microwave launching direction has been verified experimentally. Localized power deposition, consistent with theoretical predictions, has been observed at densities well above the conventional cut-off. Central heating has been achieved, at powers up to two megawatts. This demonstrates the potential of EBW in tokamak H-modes, the intended mode of operation for a reactor such as ITER.

Hierarchical model predictive control in fusion reactors

Nuclear fusion still has to pass several milestones on its way to successful energy production, one of them being successful plasma stabilization. The present article presents an innovative control implementation related to plasma control for the generation of electricity with magnetically confined plasma. The implementation has been carried out over a specific device of tokamak type, called Tokamak à Configuration Variable (TCV). This novel hybrid control design allows for real time implementation of an optimal model predictive control over a large scale complex system with small time constant. At each closed loop iteration, the full model is first controlled by a straightforward controller, then the output values are used for model reduction so that finally the discretized control system is optimized only for the variables of interest. In the case of the TCV, this novel hybrid model predictive control enhances the power availability on the actuators and extends the pulse duration. Thus, this control strategy will help to achieve fusion energy stated goal of “ignition”, where nuclear fusion generates as much energy as it is needed to start the reaction.