Processing math: 100%

Simulation Prediction of Heat Transport with Machine Learning in Tokamak Plasmas

  • Machine learning opens up new possibilities for research of plasma confinement. Specifically, models constructed using machine learning algorithms may effectively simplify the simulation process. Previous first-principles simulations could provide physics-based transport information, but not fast enough for real-time applications or plasma control. To address this issue, this study proposes SExFC, a surrogate model of the Gyro-Landau Extended Fluid Code (ExFC). As an extended version of our previous model ExFC-NN, SExFC can capture more features of transport driven by the ion temperature gradient mode and trapped electron mode, using an extended database initially generated with ExFC simulations. In addition to predicting the dominant instability, radially averaged fluxes and radial profiles of fluxes, the well-trained SExFC may also be suitable for physics-based rapid predictions that can be considered in real-time plasma control systems in the future.
  • Article Text

  • Artificial intelligence is currently regarded as a Priority Research Opportunity (PRO) in fusion control,[1,2] which has been effectively implemented in plasma-shape parameter reconstruction, surrogate model construction, and perturbation detection.[35] As high-performance computing resources become more powerful, machine learning approaches are regarded as essential tools for modeling and big data exploitation, including analysis and operation of fusion plasmas.[69] The emergence of these methods provides opportunities and applications for the exploration of complex phenomena. For instance, in micro-turbulence simulation models, the application of neural networks (NNs) for the construction of fast surrogate models has been proposed[10,11] in recent work.

    Much effort has recently been devoted to explaining the underlying physics of turbulent transport in tokamak plasmas. Direct numerical simulations, which have also become powerful tools in turbulence research, have revealed a great deal about it. Within the discharge time scales, time-evolved tokamak simulations have also been performed typically using ‘integrated modelling’.[12] An essential part of integrated models is the prediction of turbulent fluxes, especially in the tokamak core, where plasma micro-instabilities usually dominate the transport[13,14] process. For normal calculation of tokamak discharges, nevertheless, it requires huge computational cost to compute fluxes, especially using nonlinear models. Benefiting from the development of machine learning, these approaches have been applied in plasma research in various fields, such as the regression of energy confinement scaling nonlinearly,[15] database construction,[16] rapid determination of equilibria parameters,[17] reconstruction of temperature profile,[18] analysis of charge exchange spectrum on the JET device,[19] classification of disruption,[20] and development of surrogate models.[21,22]

    Motivated by the need for further prediction of transport characteristics caused by micro-turbulence with real-time character, in this work, the properties of electron heat transport, mainly in L-mode plasma conditions, are firstly introduced, using the Extended Fluid Code (ExFC). In particular, the electron heat pinch is found, which can be compared with theoretical and simulation research.[23] In order to shorten the execution cost of ExFC, ExFC-NN[24] is extended. Here, a surrogate model based on machine learning has been raised, which is named as SExFC. The time cost of building a machine learning-based surrogate model mainly lies in the training process, while the evaluation process is pretty rapid, which in our case only costs around several microseconds. SExFC can predict the type of dominant turbulence [including pure ion temperature gradient (ITG), pure trapped electron mode (TEM), as well as mixed ITG and TEM], radially averaged fluxes [〈Qer, 〈Qir and 〈Γr], and radial profiles of perturbations [Ti(r), Te(r), and n(r)], corresponding radial profiles of fluxes [Qi(r), Qe(r), and Γ(r)], and so on.

    ExFC is a recently developed global-type transport gyro-Landau fluid code containing the 3-dimensional finite difference approach. ExFC aims to investigate the physics of flux (external source terms)/gradient-driven turbulence in tokamak plasmas with multi-scale and multi-mode properties. The functions are simulated as follows:[2528]
    dtne=ωdte(n0φTe0nen0Te)+Dn2ne,

    (1)
    dtTe=Te0ωdte[(ς1)(φτenne)(2ς1)Te](ς1)8meTe0/miπ||Te+DTe2Te,

    (2)
    dtΩ=Ti0a(rlnn0+rlnTi0)θ2φ+fcarlnn0θφυ+ωdi[(1+fc/τ)φ+Ti+ftτinne]+ftωdte(φTeτenne)+DU2Ω,

    (3)
    dtυ=Tiftτinne(1+fc/τ)φ+Dυ2υ,

    (4)
    dtTi=Ti0(ς1)υ+Ti0ωdi{(ς1)[(1+fc/τ)φ+ftτinne]+(2ς1)Ti}(ς1)8Ti0/π||Ti+DTi2Ti,

    (5)
    in which ne is the normalized electron density, Te is the electron temperature, Ω is the vorticity defined as Ω=[fcn0(φφ)]/Te02φ, υ|| is the parallel ion velocity and Ti is the ion temperature. Furthermore, electrostatic potential averaged around the magnetic surface is represented by the symbol 〈φ〉. It is mentioned that vorticity contains both electron and ion temperature which means TEM and toroidal ITG modes are coupled. As a result, both ITG and TEM may be included in the multi-mode system. The gyroBohm type is applied as the normalization. The time coordinate is normalized by a/cs0 (cs0 = (Ti0/mi)1/2, ion sound speed):
    (tcs0a,rρs0,ρs0,a)(t,r,,),

    (6)
    and the spatial coordinate is normalized by ρs0(ρs0=mics0/eiB0, ion sound gyroradius):
    aρs0(nen0,TeTe0,eφTe0,υcs0,TiTi0)(ne,Te,φ,υ,Ti).

    (7)
    The q profile is displayed as an example in Fig. 1.
    Fig. Fig. 1.  Schematic of the q profile.
    The radial profiles are formed:
    n0(r)=ncexp{(Δr/Ln)tanh[(rr0)/Δr]},

    (8)
    Ts0(r)=Tscexp{(Δrs/LTs)tanh[(rr0)/Δrs]}.

    (9)
    Here, the subscript label s stands for the species of electrons and ions, the label c corresponds to the plasma center at the magnetic axis. The ratios of temperature and density are displayed as
    τ=Te0/Ti0,τen=Te0/n0,τin=Ti0/n0.

    (10)
    The subscript 0 represents the equilibrium quantity; a is the minor radius, and R is the major radius. The inverse aspect ratio is represented as ε = a/R. Operator ft=22r/R/π is the fraction of trapped electrons, then the fraction of passing electrons is fc = 1–ft. Operators of trapped electron precession drift and ion magnetic drift are defined as ωdte=2ελtqr1ϕ and ωdi=2ε(cosθr1θ+sinθr), respectively. The precession frequency of trapped electrons is λt = 1/4+2s/3, in which the magnetic shear is defined as s = rdq/qdr. The other operators are expressed as follows:
    dtf=tf+[φ,f],

    (11)
    [φ,f]=r1(rφθfθφrf),

    (12)
    in which f represents any field variable. The adiabatic compression index is ς = 5/3. The diffusivity terms, Dn = DTe = DU = Dυ = DTi = 4.8 × 10−3, are included to damp smaller-scale fluctuations. Equations (1)–(5) are numerically solved with ExFC in toroidal coordinate (r,θ,ϕ) and the simulation domains are [Lr,Lθ,Lϕ] ≈ [a,2π,2π]. More details about the code and models are included in our previous work.[24] In the following, all the results are calculated with usages of the gradient-driven version ExFC.

    In our previous research, a credible NN-based model, ExFC-NN, has been developed.[29] A clear and detailed methodology of the ExFC-NN construction is described in our previous work (Ref. [29]). The electrostatic ITG and TEM turbulence system is taken into account in ExFC-NN. The well-trained ExFC-NN perfectly reproduces and predicts what the ExFC can do while taking less calculation time. Detailed in Table 1, the applications of previous ExFC-NN have been concluded with predictions of the type of dominant turbulence (including pure ITG, pure TEM as well as the mixed ITG and TEM), radially averaged fluxes (〈Qer, 〈Qir, and 〈Γr), and radial profiles of perturbations [Ti(r), Te(r), and n(r)] as well as corresponding radial profiles of fluxes (Qi(r), Qe (r), and Γ(r)). The average region is in [0.3, 0.8] and time is also averaged in the nonlinear steady stage. Furthermore, the ExFC-NN has been considered to realize experimental analysis of the HL-2A fusion device which may provide reference for the design of experimental parameters. In this work, more plasma parameters are contained, i.e., the database has been expanded. Additional plasma physics is taken into account as a result. The extended ExFC-NN is named as SExFC with the differences between ExFC-NN and SExFC detailed in Table 1. Most importantly, it is critical to select the appropriate plasma parameters for constructing the dataset. Figure 2 shows the process of constructing the database and model in detail. According to the flow chart, further points are added iteratively. Though in the nonlinear steady state, the simulation results still have slight fluctuations in the nonlinear evolution.

    Fig. Fig. 2.  Schematic diagram for the construction of simulation results in order to obtain a reliable model.
      Table 1.  Comparisons between ExFC-NN and SExFC.
    ExFC-NNSExFC
    InputR/Ln,R/LTe,R/LTi,R/Ln,R/LTe,R/LTi,
    parametersTi0(r),Te0(r),n0(r),εTi0(r),Te0(r),n0(r),q(r), s,ε
    Output 1Dominant type of turbulenceDominant type of turbulence
    (ITG, TEM, ITG & TEM)(ITG, TEM, ITG & TEM)
    Output 2Radial averaged fluxesRadial averaged fluxes
    Qer, 〈Qir, 〈ΓrQer, 〈Qir, 〈Γr
    Output 3Radial profiles of perturbations and fluxesRadial profiles of perturbations and fluxes
    Ti(r),Te(r),n(r),Qi(r),Qe(r),Γ(r)Ti(r),Te(r),n(r),Qi(r),Qe(r),Γ(r)
    Output 4Experimental processExperimental process
    Output 5Real-time evolution of fluxes and perturbations
     | Show Table
    DownLoad: CSV
    As for the description in the introduction, enhancing the understanding of physical mechanisms of anomalous transport, through both ion and electron channels, is of central importance for achievement of burning plasmas. The transport of heat across flux surfaces in magnetic fusion experiments has been observed in previous research. Accordingly, a thorough comprehension of the mechanisms of turbulent transport and improvement of energy confinement is of great significance from both theoretical and experimental points of view. An anomalous pinch (a net inward flow of particle/energy transport channel) was observed in various devices, such as ASDEX-Upgrade,[30] Tore Supra,[31] DIII-D,[32] and JET,[33] and attracted particular attention. Particle transport has been researched in our previous work, therefore, in the present research, the phenomenon of electron heat transport is described with simulations. The contour plots of radial averaged electron heat transport 〈Qer for given R/Ln = 1 and R/Ln = 9 are separately illustrated in Fig. 3. Directly, the density gradient R/Ln may cause more electron heat transport both in an outward and inward direction. In some parameter regions, 〈Qer ∼ 0 is observed. As shown in Ref. [23], the inward heat flows in tokamaks were analyzed. For local quasilinear heat fluxes, the expression was separated as follows:[23,34]
    qj=njχjjTj+TjχjnnjVjnjTj,

    (13)
    in which j denotes species.
    Fig. Fig. 3.  Contour plots of electron heat transport 〈Qer in the plane of (R/LTi,R/LTe) with (a) R/Ln = 1 and (b) R/Ln = 9.

    However, it is not unique because of the coefficients which depend on the profile scale lengths nonlinearly. The diffusivities and electron velocity are illustrated as well. The strength and direction of heat fluxes are determined through the relative strength of plasma temperature, density profiles as well as other associated scale lengths. Namely, the balance between temperature and density gradient driven flows plays an essential role in the direction of the heat flux. Moreover, electron heat pinch is observed in Fig. 4. A similar conclusion can be found in previous research[23] in which the heat pinch was determined by the ∇n component. Such a finding indicates a mechanism for heat flux pinch. Also, the results may be valuable for exploring the improvement of energy confinement. However, the heat pinch may require specified conditions, thus further underlying investigation will be needed in future work.

    Fig. Fig. 4.  Time evolution of radial electron heat flux Qe with R/Ln = 9: (a) R/LTi = 1, (b) R/LTi = 9.

    On the other hand, for a credible machine learning model, further comparisons between SExFC and ExFC are required. Figure 5 shows the regression plot for each of the outputs of the NN ensemble. The regression plot is obtained by feeding the validation data set into the SExFC regression and the ExFC models. The distribution of points along the line with slope unity demonstrates that the NN-based model can predict results of the full ExFC calculation with accuracy. The training data of radial averaged electron heat flux (〈Qer) is considered with the coefficient of determination metric ℜ2 = 0.97. A further comparison of 〈Qer between SExFC predictions and ExFC simulations is displayed in Fig. 6. By giving random plasma parameters selected out of the databases, the prediction results of SExFC are approximative to the simulation results of ExFC. The primary characteristics of transport can be captured by the SExFC.

    Fig. Fig. 5.  Regression histograms of electron heat flux 〈Qer comparing between SExFC model and ExFC simulations.
    Fig. Fig. 6.  Qer as a function of random parameters selected out of databases comparing between SExFC predictions and ExFC simulations. The dashed line represents the 〈Qer ∼ 0.

    Finally, in this work, a framework for a machine learning-based surrogate model is established and constructed, which is for gyro-fluid electrostatic toroidal plasma simulations with a global effect. The surrogate model, SExFC, is trained to predict the type of dominant turbulence, radially averaged fluxes, and radial profiles of perturbations in the nonlinear phase, including the electron temperature Te(r), ion temperature Ti(r) and density n(r), as well as the corresponding electron heat flux Qe(r), ion heat flux Qi(r) and particle flux Γ(r). Furthermore, the SExFC may also capture features of special inward transport. With the ability for real-time prediction, the possibility of taking physics-based turbulent transport information from massively parallel simulations into plasma control may come true.

    Acknowledgment: This work was supported by the National Natural Science Foundation of China (Grant Nos. 12205035, 11925501, 12275071, and U1967206).
  • [1]
    Humphreys D, Kupresanin A, Boyer M D, Canik J, Chang C S, Cyr E C, Granetz R, Hittinger J, Kolemen E, Lawrence E, Pascucci V, Patra A, Schissel D 2020 J. Fusion Energy 39 123 doi: 10.1007/s10894-020-00258-1

    CrossRef Google Scholar

    [2]
    Degrave J, Felici F, Buchli J, Neunert M, Tracey B, Carpanese F, Ewalds T, Hafner R, Abdolmaleki A, de Casas D L, Donner C, Fritz L, Galperti C, Huber A, Keeling J, Tsimpoukelli M, Kay J, Merle A, Moret J M, Noury S, Pesamosca F, Pfau D, Sauter O, Sommariva C, Coda S, Duval B, Fasoli A, Kohli P, Kavukcuoglu K, Hassabis D, Riedmiller M 2022 Nature 602 414 doi: 10.1038/s41586-021-04301-9

    CrossRef Google Scholar

    [3]
    Kates-Harbck J, Svyatkovskiy A, Tang W 2019 Nature 568 526 doi: 10.1038/s41586-019-1116-4

    CrossRef Google Scholar

    [4]
    Joung S, Kim J, Kwak S, Bak J G, Lee S G, Han H S, Kim H S, Lee G, Kwon D, Ghim Y C 2020 Nucl. Fusion 60 016034 doi: 10.1088/1741-4326/ab555f

    CrossRef Google Scholar

    [5]
    Abbate J, Conlin R, Kolemen E 2021 Nucl. Fusion 61 046027 doi: 10.1088/1741-4326/abe08d

    CrossRef Google Scholar

    [6]
    Zheng W, Wu Q Q, Zhang M, Chen Z Y, Shang Y X, Fan J N, Pan Y, Team J T 2020 Plasma Phys. Control. Fusion 62 045012 doi: 10.1088/1361-6587/ab6b02

    CrossRef Google Scholar

    [7]
    Böckenhoff D, Blatzheim M, Hölbe H, Niemann H, Pisano F, Labahn R, Pedersen T S, The W7-X Team 2018 Nucl. Fusion 58 056009 doi: 10.1088/1741-4326/aab22d

    CrossRef Google Scholar

    [8]
    Schmidhuber J 2015 Neural Netw. 61 85 doi: 10.1016/j.neunet.2014.09.003

    CrossRef Google Scholar

    [9]
    Rea C, Montes K J, Erickson K G, Granetz R S, Tinguely R A 2019 Nucl. Fusion 59 096016 doi: 10.1088/1741-4326/ab28bf

    CrossRef Google Scholar

    [10]
    Citrin J, Breton S, Felici F, Imbeaux F, Aniel T, Artaud J F, Baiocchi B, Bourdelle C, Camenen Y, Garcia J 2015 Nucl. Fusion 55 092001 doi: 10.1088/0029-5515/55/9/092001

    CrossRef Google Scholar

    [11]
    Meneghini O, Smith S P, Snyder P B, Staebler G M, Candy J, Belli E, Lao L, Kostuk M, Luce T, Luda T, Park J M, Poli F 2017 Nucl. Fusion 57 086034 doi: 10.1088/1741-4326/aa7776

    CrossRef Google Scholar

    [12]
    Poli F M 2018 Phys. Plasmas 25 055602 doi: 10.1063/1.5021489

    CrossRef Google Scholar

    [13]
    Doyle E J, Houlberg W A, Kamada Y, Mukhovatov V, Osborne T H, Polevoi A, Bateman G, Connor J W, Cordey J G, Fujita T, Garbet X, Hahm T S, Horton L D, Hubbard A E, Imbeaux F, Jenko F, Kinsey J E, Kishimoto Y, Li J, Luce T C, Martin Y, Ossipenko M, Parail V, Peeters A, Rhodes T L, Rice J E, Roach C M, Rozhansky V, Ryter F, Saibene G, Sartori R, Sips A C C, Snipes J A, Sugihara M, Synakowski E J, Takenaga H, Takizuka T, Thomsen K, Wade M R, Wilson H R, ITPA Transport Physics Topical Group, ITPA Confinement Database and Modelling Topical Group, and ITPA Pedestal and Edge Topical Group 2007 Nucl. Fusion 47 S18 doi: 10.1088/0029-5515/47/6/S02

    CrossRef Google Scholar

    [14]
    Horton W, Hu B, Dong J Q, Zhu P 2003 New J. Phys. 5 14 doi: 10.1088/1367-2630/5/1/314

    CrossRef Google Scholar

    [15]
    Allen L, Bishop C M 1992 Plasma Phys. Control. Fusion 34 1291 doi: 10.1088/0741-3335/34/7/008

    CrossRef Google Scholar

    [16]
    Wakasa A, Murakami S, Itagaki M, Oikawa S I 2007 Jpn. J. Appl. Phys. 46 1157 doi: 10.1143/JJAP.46.1157

    CrossRef Google Scholar

    [17]
    Lister J B, Schnurrenberger H 1991 Nucl. Fusion 31 1291 doi: 10.1088/0029-5515/31/7/005

    CrossRef Google Scholar

    [18]
    Clayton D J, Tritz K, Stutman D, Bell R E, Diallo A, LeBlanc B P, Podestá M 2013 Plasma Phys. Control. Fusion 55 095015 doi: 10.1088/0741-3335/55/9/095015

    CrossRef Google Scholar

    [19]
    Svensson J, von Hellermann M, König R W T 1999 Plasma Phys. Control. Fusion 41 315 doi: 10.1088/0741-3335/41/2/016

    CrossRef Google Scholar

    [20]
    Vega J, Murari A, Dormido-Canto S, Moreno R, Pereira A, Acero A, Contributors J E 2014 Nucl. Fusion 54 123001 doi: 10.1088/0029-5515/54/12/123001

    CrossRef Google Scholar

    [21]
    Dong G, Wei X, Bao J, Brochard G, Lin Z, Tang W 2021 Nucl. Fusion 61 126061 doi: 10.1088/1741-4326/ac32f1

    CrossRef Google Scholar

    [22]
    Liu Y Q, Lao L, Li L, Turnbull A D 2020 Plasma Phys. Control. Fusion 62 045001 doi: 10.1088/1361-6587/ab6f56

    CrossRef Google Scholar

    [23]
    Nordman H, Strand P, Weiland J 2001 Plasma Phys. Control. Fusion 43 1765 doi: 10.1088/0741-3335/43/12/310

    CrossRef Google Scholar

    [24]
    Li H, Li J Q, Fu Y L, Wang Z X, Jiang M 2022 Nucl. Fusion 62 036014 doi: 10.1088/1741-4326/ac486b

    CrossRef Google Scholar

    [25]
    Beer M A 1995 Ph.D. Thesis Princeton University

    Google Scholar

    [26]
    Garbet X, Garzotti L, Mantica P, Nordman H, Valovic M, Weisen H, Angioni C 2003 Phys. Rev. Lett. 91 035001 doi: 10.1103/PhysRevLett.91.035001

    CrossRef Google Scholar

    [27]
    Garbet X, Dubuit N, Asp E, Sarazin Y, Bourdelle C, Ghendrih P, Hoang G T 2005 Phys. Plasmas 12 082511 doi: 10.1063/1.1951667

    CrossRef Google Scholar

    [28]
    Nordman H, Weiland J, Jarmén A 1990 Nucl. Fusion 30 983 doi: 10.1088/0029-5515/30/6/001

    CrossRef Google Scholar

    [29]
    Li H, Fu Y L, Li J Q, Wang Z X 2021 Plasma Sci. Technol. 23 115102 doi: 10.1088/2058-6272/ac15ec

    CrossRef Google Scholar

    [30]
    Ryter F, Angioni C, Peeters A G, Leuterer F, Fahrbach H U, Suttrop P 2005 Phys. Rev. Lett. 95 085001 doi: 10.1103/PhysRevLett.95.085001

    CrossRef Google Scholar

    [31]
    Hoang G T, Bourdelle C, Garbet X, Giruzzi G, Aniel T, Ottaviani M, Horton W, Zhu P, Budny R V 2001 Phys. Rev. Lett. 87 125001 doi: 10.1103/PhysRevLett.87.125001

    CrossRef Google Scholar

    [32]
    Baker D R, et al.. 2000 Nucl. Fusion 40 1003 doi: 10.1088/0029-5515/40/5/301

    CrossRef Google Scholar

    [33]
    Garzotti L, Valovič M, Garbet X, Mantica P, Parail V, JET EFDA Contributors 2006 Nucl. Fusion 46 994 doi: 10.1088/0029-5515/46/12/002

    CrossRef Google Scholar

    [34]
    Fröjdh M, Strand P, Weiland J, Christiansen J 1996 Plasma Phys. Control. Fusion 38 325 doi: 10.1088/0741-3335/38/3/008

    CrossRef Google Scholar

  • Related Articles

    [1]ISOBE Mitsutaka, LIU Yi, YANG Jin-Wei, CHEN Wei, JI Xiao-Quan, YUAN Guo-Liang, ZHANG Yi-Po, ZOU Gui-Qing, LI Wei, LEI Guang-Jiu, DING Xuan-Tong, YAN Long-Wen, YANG Qing-Wei, DUAN Xu-Ru, OHDACHI Satoshi, MORITA Shigeru, TOI Kazuo. Beam Ion Transport from Neutron Emissivity Measurement in Deuterium Neutral Beam-Heated Deuterium Plasmas of the HL-2A Tokamak [J]. Chin. Phys. Lett., 2009, 26(10): 105201. doi: 10.1088/0256-307X/26/10/105201
    [2]GAO Hong, YU Qing-Quan. Heat Diffusion across a Strong Stochastic Magnetic Field in Tokamak Plasmas [J]. Chin. Phys. Lett., 2009, 26(6): 065201. doi: 10.1088/0256-307X/26/6/065201
    [3]ZHANG Xing, TAKAHASHI Koji, FUJII Motoo. Charge and Heat Transport in Polycrystalline Metallic Nanostructures [J]. Chin. Phys. Lett., 2008, 25(9): 3360-3363.
    [4]LI Heng-Chao, ZHANG Jia-Shu. Local Prediction of Chaotic Time Series Based on Support Vector Machine [J]. Chin. Phys. Lett., 2005, 22(11): 2776-2779.
    [5]FU Hong-Jun, DING Xuan-Tong, LIU Yi, XU De-Ming. Electron Heat Transport in an HL-1M Tokamak ECRH Plasma [J]. Chin. Phys. Lett., 2003, 20(11): 1973-1976.
    [6]WANG Ai-Ke, QIU Xiao-Ming. Drift Mode Growth Rate and Associated Ion Thermal Transport in Reversed Magnetic Shear Tokamak Plasma [J]. Chin. Phys. Lett., 2001, 18(5): 650-652.
    [7]QIU Xiao-ming, WANG Ai-ke. Pinches-Induced Transport Reduction in Reversed Shear/Optimized Shear Tokamak Plasmas [J]. Chin. Phys. Lett., 1999, 16(9): 662-664.
    [8]QIU Xiao-ming, WANG Ai-ke. Model for Heat Pinch in Reversed Magnetic Shear Tokamak Plasmas [J]. Chin. Phys. Lett., 1998, 15(10): 721-723.
    [9]HUANG Lin, QIN Xiaoming, DING Ning, LONG Yongxing. Heating Finite Beta Tokamak-Plasmas by Alfven Waves [J]. Chin. Phys. Lett., 1991, 8(5): 232-235.
    [10]LIU Min, WANG Long, LIN Ruiyan. SIMULATION OF THE RADIAL TRANSPORT IN CT-6B TOKAMAK DURING GAS PUFFING [J]. Chin. Phys. Lett., 1987, 4(3): 121-124.

Catalog

    Figures(6)  /  Tables(1)

    Article views (22) PDF downloads (56) Cited by()

    /

    DownLoad:  Full-Size Img  PowerPoint
    Return
    Return