RF Supplementary Heating for Toroidal Reactors

RFSupplementaryHeating for T. Consoli, Grenoble Toroidal Reactors (Centre d’Etudes Nucléaires) In a toroidal configuration of the Tokamak or Stellarator type, two fundamental objectives have to be reached, the confinement, and the heating of a dense and hot plasma during a time interval τE, so as to satisfy the condition: nr ≥ 1014 requi red by the Lawson criterion to reach the D,T thermonuclear region. In Tokamaks and, in part, in Stellerators, the induced toroidal current not only provides for plasma confinement and equilibrium, but also supplies energy through ohmic heating directly to the electrons and indirectly to the ions by electron-ion collisions. It is universally admitted that this energy source is effective in raising the electron and ion temperatures to a few keV. Nevertheless, this collisional energy transfer is insufficient to heat the plasma of a future toroidal reactor to ignition temperatures. Conse quently, considerable attention has to be given to the question of supple mentary heating schemes which can raise the temperature reached by ohmic heating to the ignition tempe rature. As a result, for some years now, supplementary heating has begun to occupy the centre of the fusion pro gramme. In big Tokamaks like JET, T.F.T.R., T20and JT 60 where significant thermo nuclear energy production is expec ted, the additional power needed lies between 20 and 100 MW depending on the value of nτE which in its turn depends on the loss mechanism that is considered to be the most impor tant : Bohm, pseudo-classical, trapped electron or trapped ion instabilities. It is however clear that in a first phase, the injected power will be smaller, of the order of 10 MW and with a long pulse (~10s). Notice that this is already a big jump when com pared with the powers we are injec ting now which are of the order of a few hundred kW during a time 10 to 100 ms. Presently, the two main supplemen tary heating methods are neutral beam injection and radio frequency (RF) heating processes, based respectively on particle-particle collisions and wa ve-particle interactions. Of these, radio frequency wave-plasma interactions are very attractive, covering as they do a large variety of frequency depen dent mechanisms for the absorption by a plasma of the energy carried by a wave. As appears in Fig. 1, there exist various domains of interest, subdivi ded according to decreasing wave lengths into three classes : low, high, and very high frequencies. From the coupling view-point, the low frequency group utilizes loops, which can be put outside the plasma vessel (Toroidal Drift Magnetic Pum ping) or inside, (as for Transit Time Magnetic Pumping, Alfen Waves Re sonant Heating and Ion Cyclotron Resonant Heating). In the second and third groups (the high and very high frequencies) the waves are launched by guides. We must remark that for high toroidal magnetic field systems B ≥ 5T and for large Tokamaks, 2 r ≥3m, I.C.R. waves can also be launched by ridge or loaded guides. The possibility of using external coils in the case of very low frequency heating as in T.D.M.P., or of launching Classification of Wave-Plasma Interactions Fig. 1. Absorption mechanisms : I - Ohmic Heating, II - Toroidal Drift Magnetic Pumping, III - Transit Time Magnetic Pumping on the ions, IV - Alfven Waves Resonant Heating, V - Transit Time Magnetic Pumping on the electrons, VI - Hybrid Ion Cyclotron Resonant Heating, VII - Magnetosonic Waves, VIII - Lower Hybrid Resonant Heating, IX - Electron Cyclotron Resonant Heating. 7 the waves by guides directly connec ted to the toroidal chamber, as in the case of high frequency heating schemes (Lower Hybrid Resonant Heating) constitutes an undeniable advantage. We intend to report mainly on three very promising methods : T.T.M.P., I.C.R.H. and L.H.R.H. which have been tested experimentally on toroidal de vices. It seems, however, appropriate to mention first the potentialities of two other methods (T.D.M.P. and Elec tron Cyclotron Resonant Heating) which are, from a purely speculative viewpoint, very interesting for the fu ture thermonuclear reactor. T.D.M.P. and E.C.R.H. Heating Schemes These two methods, which are very attractive because of their launching systems, use respectively very low and very high frequencies and are at the state of proposals only. Indeed the first heating scheme has not been tested experimentally yet. T.D.M.P. non-collisional RF heating was proposed recently by E. Canobbio (VIIIth I.A.E.A. Conf. Berchtesgaden, October 1976). It requires frequencies so low, (of the order of kHz) that they can be supplied by a rotating ma chine, connected directly to coils (Fig. 2) external to the plasma vessel, which the waves penetrate easily. Suitably phased and modulated, they interact with the vertical component of the toroidal drift velocity. The coupling coils may be the same coils used for the generation of the vertical magnetic field necessary for the plasma equilibrium. The electro magnetic energy carried by the exci ted waves is absorbed by Landau-like damping, if ω0 > v, where v is the col lision frequency for the energy ex change of the resonant ions. Fig. 2. The T.D.M.P. showing the position of the excitation coils. 8 The space averaged power, absor bed by the ions of the plasma is pro portional to (nTi)2 and to the square of the relative plasma displacement. This means that heating is more effec tive in dense and hot plasmas. The optimum frequency heating in the case of the JET parameters (Ti = 3 keV, Ro = 3m, B = 3 Tesla, b, the vertical extension of the plasma = 1.2 m, b/a = 1.6) is around 1 khz. Cannobio also showed that by ex ploiting the existence within the plas ma of the M.H.D. resonant surfaces, T.T.M.P. can be produced at similar low frequencies, but with much higher heating efficiency. At the other end of the frequency spectrum E.C.R.H. has also very at tractive features. This heating scheme requires the realization of a powerful millimetric source (the gyroklystron) now in development. Although preli minary experiments with 28 GHz 200 kWCWwill be undertaken in 1978 at Oak Ridge, a full scale realistic test at a frequency near to 1011Hz (B (toroi dal) ~ 4 T, ne ~ 1014 cm-3) will be possible only after 1980. The two physical mechanisms of wave energy absorption involved are well known. One is the collisionless damping of microwaves propagating at a frequency near the electron cy clotron frequency, and the other one is the linear conversion of electro magnetic wave energy into longitudi nal waves near the upper hybrid re sonance. In an inhomogeneous plasma, the penetration of the wave to the reso nant region through the external eva nescent layer, requires the launching of the wave from the higher magnetic field side of the torus. The wave which has reached the resonance region, looses its energy by resonance or by Landau damping. In the resonance region, the electrons gain energy from the RF field through a stochastic pro cess. The m (...truncated)