Fast equilibrium reconstruction for tokamak discharge control based on GPU

Yue, Xie; Xiao, Bingjia; Luo, Z.P.; Guo, Youming

doi:10.1088/0741-3335/55/8/085016

Cited by 32 publications

(30 citation statements)

References 18 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The almost rectangular geometry of the TCV vacuum vessel makes the choice of a rectangular integration domain straightforward: to satisfy the conditions (9), the boundary is placed between the vacuum vessel and the tile aperture (see Fig. 1).…”

Section: Grid and Poisson Solvermentioning

confidence: 99%

“…Ignoring, for the present, disruptive and vertical displacement events, the required cycle time is imposed by the characteristic time constant of the vacuum vessel image currents and the corresponding response time of the poloidal field coil power supplies, typically 1ms. Despite a continuous increase in the computation speed of processors, only a few of the available real time equilibrium reconstruction codes can approach this figure, by distributing the computation load across -3 -several processors requiring advanced programming techniques [8] or dedicated hardware [9]. For the real time implementation of LIUQE on TCV, careful attention to the efficiency of competing numerical techniques in all steps of the algorithm together with an efficient, yet user friendly, code generation using the SIMULINK programming environment yielded a cycle time of less that 200µs on a single INTEL processor core.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Tokamak equilibrium reconstruction code LIUQE and its real time implementation

Moret

Duval

et al. 2015

Fusion Engineering and Design

113

119

View full text Add to dashboard Cite

Highlights.• Algorithm vertical stabilisation using a linear parametrisation of the current density • Experimentally derived model of the vacuum vessel to account for vessel currents• Real-time contouring algorithm for flux surface averaged 1.5D transport equations• Full real time implementation coded in SIMULINK runs in less than 200µs• Applications: shape control, safety factor profile control, coupling with RAPTOR Abstract. Equilibrium reconstruction consists in identifying, from experimental measurements, a distribution of the plasma current density that satisfies the pressure balance constraint. The LIUQE code adopts a computationally efficient method to solve this problem, based on an iterative solution of the Poisson equation coupled with a linear parametrisation of the plasma current density. This algorithm is unstable against vertical gross motion of the plasma column for elongated shapes and its application to highly shaped plasmas on TCV requires a particular treatment of this instability. TCV's continuous vacuum vessel has a low resistance designed to enhance passive stabilisation of the vertical position. The eddy currents in the vacuum vessel have a sizeable influence on the equilibrium reconstruction and must be taken into account. A real time version of LIUQE has been implemented on TCV's distributed digital control system with a cycle time shorter than 200µs for a full spatial grid of 28 by 65, using all 133 experimental measurements and including the flux surface average of quantities necessary for the real time solution of 1.5D transport equations. This performance was achieved through a thoughtful choice of numerical methods and code optimisation techniques at every step of the algorithm, and was coded in MATLAB and SIMULINK for the off-line and real time version respectively.

show abstract

Section: Grid and Poisson Solvermentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Tokamak equilibrium reconstruction code LIUQE and its real time implementation

Moret

Duval

et al. 2015

Fusion Engineering and Design

113

119

View full text Add to dashboard Cite

show abstract

“…A parallelized solver using multiple CPU's and employing discrete sine transforms instead of cyclic reduction has also been developed [11,12]. A highly parallelized magnetic equilibrium reconstruction code for EAST on a GPU has also been demonstrated [13]. A number of possibilities to add internal constraints to the magnetic equilibrium and the method for constraining the central safety factor, q(0), to a particular value has been reported [14].…”

Section: Introductionmentioning

confidence: 99%

Improvements for real-time magnetic equilibrium reconstruction on ASDEX Upgrade

Giannone

Fischer

McCarthy

et al. 2015

Fusion Engineering and Design

View full text Add to dashboard Cite

Real-time magnetic equilibria are needed for NTM stabilization and disruption avoidance experiments on ASDEX Upgrade. Five improvements to real-time magnetic equilibrium reconstruction on ASDEX upgrade have been investigated. The aim is to include as many features of the offline magnetic equilibrium reconstruction code in the real-time equilibrium reconstruction code. Firstly, spline current density basis functions with regularization are used in the offline equilibrium reconstruction code, CLISTE [1]. It is now possible to have the same number of spline basis functions in the real-time code. Secondly, in the presence of edge localized modes, (ELM's), it is found to be necessary to include the low pass filter effect of the vacuum vessel on the fast position control coil currents to correctly compensate the magnetic probes for current oscillations in these coils. Thirdly, the introduction of ferromagnetic tiles in ASDEX Upgrade means that a real-time algorithm for including the perturbations of the magnetic equilibrium generated by these tiles is required. A methodology based on tile surface currents is described. Fourthly, during current ramps it was seen that the difference between fitted and measured magnetic measurements in the equilibrium reconstruction were larger than in the constant current phase. External loop voltage measurements and magnetic probe pairs inside and outside the vessel wall were used to measure the vacuum vessel wall resistivity. This is the first step towards including vacuum vessel currents during the plasma current ramp in the real-time equilibrium reconstruction. Fifthly, the introduction of a constraint of the safety factor on the magnetic axis is found to be a helpful method to improve the prediction of the location of rational surfaces for NTM stabilization and disruption avoidance experiments. Soft X-ray tomography is used to assess the quality of the real-time magnetic equilibrium reconstruction using this internal constraint.

show abstract

“…We have tried to use the GPU-accelerated version of CUBLAS to accelerate, but the acceleration effect cannot reach the expected demand, because CUBLAS is mainly designed for large-scale computation andǴ I p p is only a middle-scale task. For this reason, a special parallel algorithm is designed based on the CUDA architecture [18].…”

Section: Parallelization Of Flux Calculation On Boundary Gridmentioning

confidence: 99%

Acceleration optimization of real-time equilibrium reconstruction for HL-2A tokamak discharge control

Xia

Ling

et al. 2018

Plasma Sci. Technol.

View full text Add to dashboard Cite

Real-time equilibrium reconstruction is crucially important for plasma shape control in the process of tokamak plasma discharge. However, as the reconstruction algorithm is computationally intensive, it is very difficult to improve its accuracy and reduce the computation time, and some optimizations need to be done. This article describes the three most important aspects of this optimization: (1) compiler optimization; (2) some optimization for middle-scale matrix multiplication on the graphic processing unit and an algorithm which can solve the block tri-diagonal linear system efficiently in parallel; (3) a new algorithm to locate the X&O point on the central processing unit. A static test proves the correctness and a dynamic test proves the feasibility of using the new code for real-time reconstruction with 129×129 grids; it can complete one iteration around 575 μs for each equilibrium reconstruction. The plasma displacements from real-time equilibrium reconstruction are compared with the experimental measurements, and the calculated results are consistent with the measured ones, which can be used as a reference for the real-time control of HL-2A discharge.

show abstract

Fast equilibrium reconstruction for tokamak discharge control based on GPU

Cited by 32 publications

References 18 publications

Tokamak equilibrium reconstruction code LIUQE and its real time implementation

Tokamak equilibrium reconstruction code LIUQE and its real time implementation

Improvements for real-time magnetic equilibrium reconstruction on ASDEX Upgrade

Acceleration optimization of real-time equilibrium reconstruction for HL-2A tokamak discharge control

Contact Info

Product

Resources

About