Gyrokinetic modelling of the quasilinear particle flux for plasmas with neutral-beam fuelling. (9th January 2018)

• Record Type: Journal Article
• Authors: Narita, E; Honda, M; Nakata, M; Yoshida, M; Takenaga, H; Hayashi, N
• Is Part Of: Plasma physics and controlled fusion. Volume 60:Number 2(2018:Feb.)
• Publication Date: 2018-01-09
• Subjects: particle transport -- gyrokinetic calculations -- turbulent transport -- quasilinear model; Plasma (Ionized gases) -- Periodicals; Controlled fusion -- Periodicals; 530.44
• DOI: 10.1088/1361-6587/aaa02d
• Languages: English
• ISSNs: 0741-3335

Abstract

Abstract: A quasilinear particle flux is modelled based on gyrokinetic calculations. The particle flux is estimated by determining factors, namely, coefficients of off-diagonal terms and a particle diffusivity. In this paper, the methodology to estimate the factors is presented using a subset of JT-60U plasmas. First, the coefficients of off-diagonal terms are estimated by linear gyrokinetic calculations. Next, to obtain the particle diffusivity, a semi-empirical approach is taken. Most experimental analyses for particle transport have assumed that turbulent particle fluxes are zero in the core region. On the other hand, even in the stationary state, the plasmas in question have a finite turbulent particle flux due to neutral-beam fuelling. By combining estimates of the experimental turbulent particle flux and the coefficients of off-diagonal terms calculated earlier, the particle diffusivity is obtained. The particle diffusivity should reflect a saturation amplitude of instabilities. The particle diffusivity is investigated in terms of the effects of the linear instability and linear zonal flow response, and it is found that a formula including these effects roughly reproduces the particle diffusivity. The developed framework for prediction of the particle flux is flexible to add terms neglected in the current model. The methodology to estimate the quasilinear particle flux requires so low computational cost that a database consisting of the resultant coefficients of off-diagonal terms and particle diffusivity can be constructed to train a neural network. The development of the methodology is the first step towards a neural-network-based particle transport model for fast prediction of the particle flux.