PFC/RR-87-8 DOE/ET-51013-225 UC20 G Evolution of Poloidal Variation of Impurity Density and Ambipolar Potential in Rotating Tokamak Plasma, Part I Sigmar, D.J.; Zanino, R.t; Hsu, C.T. Plasma Fusion Center Massachusetts Institute of Technology Cambridge, MA 02139 tDipartimento di Energetica, Politecnico di Torino, Italy September 1987 This work was supported by the U. S. Department of Energy Contract No. DE-AC0278ET51013. Reproduction, translation, publication, use and disposal, in whole or in part by or for the United States government is permitted. Evolution of Poloidal Variation of Impurity Density and Ambipolar Potential in Rotating Tokamak Plasma, Part I D. J. Sigmarl R. Zanino2 C. T. Hsu 1 Massachusetts Institute of 2 Technology, Cambridge, MA, USA Dipartimentodi Energetica,Politecnico di Torino, Italy Abstract When the (beam driven) toroidal plasma rotation (Ve1 ) reaches thermal velocity (vik,) levels such that Vthz < VZ ~ Vei~Vthi where i denotes the hydrogenic and z the impurity ions, the particle density n3 and the ambipolar potential ( can develop strong poloidal variations in the flux surface. The rapid time evolution of nj(r,9), V4(r, 0), and the plasma flow velocity components in the flux surface V11 , V 1i is calculated in Part I. In a forthcoming Part II these results will be evaluated numerically using a spectral code to describe the strong poloidal couplings. Results will be compared to data from the ASDEX tokamak. I. Introduction The connection between neutral beam induced plasma rotation and impurity transport has been the subject of numerous investigations [1,2,3], mainly focusing on the radial (across the flux surface) transport of impurities in experiments with neutral beam heating. In most of these works, the steady state beam-plasma equilibrium was analyzed describing the potential variations of all quantities A, in the simplest form A(r,0) = X + Ac cos 0 + 1 A" sin0 where A ''/~j was ordered - e = r/Ro. However, using soft X-ray imaging techniques Smeulders [4] obtained results from the ASDEX tokamak of the form 2 n, (r,6) = ' cos 6 + n' sin M8) + t=1 with 0(1), not 0(e) - Fourier amplitudes. This necessitates a theoretical approach different from that used by previous authors. This is the subject of the present paper which will be restricted to the Pfirsch-Schliiter regime of collisionality for all species, and will assume the impurity strength parameters a = nZ 2 /ni to be of 0(1). The transient time evolution of the particle densities, flow-velocities and ambipolar potential is treated. The nonlinear couplings introduced by the inertia and other terms in the momentum balance equations are fully retained and contribute essentially to the final steady state poloidal impurity distribution. A complete description results of the fast time evolutions of the poloidal rotation and the ambipolar electric field (first described in a simplified model in Ref. [5]). On the slow time scale of the toroidal evolution, the neutral beam momentum source Mj is explicitly kept, balanced by the Braginskii viscous force modelled by a phenomenological momentum drag term - mjnjvd, Vj where vd,j is taken from experimental observations. (A first principle explanation of this drag as a gyroviscous force was given previously in Ref. [6]. Its fully selfconsistent theory is still under investigation [7].) The slow (diffusion) time scale evolution of toroidal rotation (driven by the radial flow) is not treated in this report. In Section II, the underlying fluid equations are formulated. Section III contains a summary. Appendices A and B contain details on the parallel viscous force and inertia, Appendix C gives a simplified physical model for the time evolution of the radial ambipolar field, E,, and Appendix D justifies the neglect of the radial flow veloctiy on the fast time scale of interest here. 2 II. Fluid Equations We adopt the usual (#,9,<p) flux coordinate system of an axisymmetric tokamak. The particle conservation equation for the ion species j = i, z is i+ V - njj = 0 (1a) For the purpose of this paper, namely to calculate the evolution in the flux surfaces where the slower radial diffusion flow -is frozen in, the component V - VO is ordered out (cf. Appendix D) and (1a) becomes 9n - 1 a + -- J'90 JnfVj . VO = 0 (1b) where J- 1 = - VO is the inverse Jacobian of the equilibrium. The momentum conservation equation is ejnj -V + -' x B + R j + Mi - mnjvd-i V c (2) where Rj is the collisional friction, Mj the beam momentum source and IIj the Braginskii viscous stress tensor. The last term models an effective drag force needed to describe those experimental observations where the observed rotation damping is faster than the classical perpendicular viscosity. (Cf. Ref. [6, 8] for a discussion.) The system is closed invoking charge neutrality E ejnj(4,0) = 0 j=e,i,z and the Vp component of Ampere's law c - .c + - -E - Vi = VO - V x B= 0 cO 3 (3a) where the last step requires axisymmetry. Using c dV) Ampere's law becomes 5wIV&Ik2 4(-1) is discussed below. = 47rcj- VO (3b) From parallel electron momentum balance and assuming all temperatures are constant in the flux surface T = Tj() (4) O = TB -Vn.+ enB - V (5a) one obtains Combining Eqs (3) and (5) one can express B.V4 in terms of the ionic density variations in the flux surface: ejnjB -V4/T. = ZjnjB - Vln(n, + Zn,) (5b) For the parallel friction we simply use R1i = -mnjviz(Vjj - VII.), Ri1 = -Rj (6) and the only stress tensor element needed explicitly is given in detail in App. A, Eq. A.4. The required ordering V ~ vt1, is incorporated by noting in Eq. (2) that to order 61[6 = p/a, the Larmor radius expansion parameter,} -VO(-0 + '(0) A. V =(-'0 or x A/c =0 40) = 0(-)(0) (7a) (7b) Then, (0)= c 8(_ VII b -8 4 ) b x VO B ) with the rotation frequency w 9 (9) which can be assumed given at t = 0. Here, I = RBW = I(M), b B/B, Vp = e ,/R, and the magnetic field is B = VW x VO + IVW. -(0) The total lowest order flow velocity V V(0) = can then be written as V 1 -+ (kg- = (iiI+cw R2 b- WR2 (10a) whence 4 +w 0 = -(o R 2 VW. =V -- B, = wR 2 1 - R2 )2 /R 2 B 2 (10b) (BB) (10c) (lvl V+ + 'i WI) = 0 (11) I) ( Using (10), the continuity Eq. (1b) is rewritten as +n Using (9), also the parallel inertial term (V -VV) - b can now be worked out and one finds the result given in App. B, Eq. (B. 1). Combining these equations (including Eq. A. 4) there results for the parallel momentum balance Eq. (2) of ion species j mi VIj + m, - B/2(b-. = -3/ Z2b-= -V B [(Vi -wR2] + Tj$-Vn;+mingva'Vix +w) 2V)-vv+ + Z VnIj- b.VB ' B I I -- -B b.MVBI I flj ZjTeb VlIn(ni + Zn,.) - mjnjv3 3 , (VIj1 - Vj1jj) + M111 5 J (12) To obtain the evolution of w, we take the ix V- projection of Eq. (2), i.e. the radial component of the perpendicular momentum balance, and sum over species which yields the radial current -ic= V#b- x + min ++ i - II + minvdjdf}-Ia (13a) Here, the lowest order 11j of Eq. (10a) can be used for V. Before summing over species, in tokamak flux coordinates, this equation is j eC V -V0 = -mjni -m [i , jn (I IN 2 b- VB V1\+ + B w (T1i. Vnj + ejnjb - V) vo B-n2 IVik120- "' + 2 (Mi) 4[ \-.2 B2 2B / j-R - V (II 11 /B3 ) + (R (13b) ) On summing over species the b- V0 term and the friction term R( will annihilate. Equa- tion (13b) can also be obtained by subtracting the toroidal (R 2 VW) component of the momentum balance from the parallel component given in Eq. (12) as can be seen from the identity (used in Eq. (8) before) x -. = - (R A where A represents the momentum balance equation. This reveals the meaning of (MI) as the perpendicular component (in the - b x V0 direction) of the momentum source term, projected onto the toroidal (p) direction. We then utilize Ampere's law Eq. (3b) and take the flux average of Eq. (13b) to obtain (E(mn) + 42 at+ b-VR +2 +2Z~n~nR (mnV R 6 12 Z(mnVgb. (mnvd)w VB) +ii (mnb -VR) 2 V01 (b . VB) IHII) (13c) B2R2 Here R 2 B 2 where we dropped the species index for brevity and E = = 12 + IVfI 2 was used (and for a low 3-equilibrium IVik is a flux surface quantity.) We note that using B 2 /47rmnc 2 = c2 /c 2 < 1 the term (B2)/47rc2 is negligible. (13c) is an equation for the time evolution of w(-O) oc E, of the form 8w + CjW2 + C2W + C3 = 0 C1 (14a) ( mn. VRj BT (14b) C2 =mnV VB + (o - D IBP2R E (77o ((. VB)(2b. VVI +V D = III. B lmV2b. mn I.bVB IIB BT C3 (b- VB) 1(mn) + B) B (14c) 2 - +-( 2 (14d) M/R) vi~o BB (B 2 )/47rc2 (14e) Summary We have obtained a closed set of Eqs. (11) for A~ni(O, B), (12) for AVj1 ia, ( ) and (14) for A-w(O)(j = i, Z) which describes the evolution of nj and VjIj in the domain 0 < 0 < 27r, on a fixed surface 4, and the evolution of w oc E, as an initial value problem. The evolution of nj(,0, 0) is driven by the poloidal flow V = (V + to steady state such that 2" !w)BE and will relax becomes a flux function. The evolution of V and thus V, is naturally driven by the parallel component of the momentum source M11 , which 7 simultaneously couples to the poloidal variation of the density via terms involving b.- Vnj in Eq. (12). For nwit is interesting to note that in addition to the explicit drive by the usually small toroidal projection of M 1 , w is driven indirectly by M11 through VII as shown in the first term of Eq. (14d). That is, even for purely parallel injection, a large radial electric field can be built up when V reaches a level VI ~ Vthi Concerning the damping mechanisms of the system, a simplified physical discussion of the poloidal damping of V is given in Appendix C. Here we only remark that the parallel viscosity involving ?7o provides a damping mechanism for both w and VII on the time scale r,, while the momentum drag term provides the toroidal damping mechanism on the time scale 1rT. However, it is worth mentioning that in most tokamak experiments the time scales for 7, and 1rT are not as widely separated as is usually assumed theoretically, involving the small classical perpendicular viscosity [9] q7 -ri (experimental) ~ (50 - 100)(i71/a 2 ~ io(liri)- 2 . As mentioned after Eq. (2), ) where a is the plasma radius, cf. Ref. [8]. After the transients A {n,, Vj,w} have vanished, Eqs. (11), (12) and (14) yield a steady state system with nontrivial solutions for Vig and V,, = BpK j (Ob)/nj maintained by the finite momentum sources. Note that the quadratic form of Eq. (14a) implies two possible roots for w - E, thus permitting a bifurcation of the steady state plasma. 8 Appendix A on B - V -I From Braginskii [9] the 0(8*) - part of the viscous stress tensor is 21 1(bb- 3I S with IiiI = -7OWii , and VY+(VY)T - W1 = bb: )] 1(V- (Al) - 3V-V =2b-VVII -2b-Vb-Y1 where 7o corresponds to i in Eq. (C5) such that 77o = -1. In the Pfirsch-Schliter regime, 77o = .96p/v. Using the lowest order Eq. (10) for V gives either WI1 = 2 [ VV + VB)V B1( (A2a) wb-VB] -- or W, ~ 2 tb V - (A2b) 2(b.- VB)] depending on the decomposition V = V+b+V1 =V RVp , V =VI+Vp Using 2 (B. b+ (b-Vb)rI - VHII one obtains B -V - 1 = B 3 /2 B -V(I/B (A) 3 /2 ) and using W11 from (A.1) in the (V,, V,) representation one obtains Eq. (17) of Ref. [6] for - -(0) B -V -H . Using (A.2a) gives B-V - =- B -BV)[B ( 2b-VVI + 9 B IW -vB] B2 -V)] (A4) Appendix B on the inertia term Using Eq. (10) for the lowest order V one has (V VY)$= = + ) (-VV bR - b-Vb V-1 +')(g. 1( b I\b (vi /Vi -. R VS- R2V) 2 b-VR 2 V. (Bl) - 2R2 (V +w = W~)-!2g w2 1+W B b S +W 2 V. - For the flux averaged equation determining E, ~ w (see Eq. (9)) the expression G,=(nm2VW. (- . V) g- is needed. It is worth mentioning that if '9 5 0 in Eq. (1a), G,, does not vanish, even for axisymmetric systems. This is readily seen if one uses the relation a ,(nV)+ av V - (nVv) = n- +n(V -V)V which is obtained invoking On &+ V -(nV) = 0 Then, (R 2 V - n(V -V)V) = ((V -R2V) ) + -V' - (R 2 VW -nVV- V4) (B2) The last term was obtained from the general flux averaged divergence formula (V. - V'A- VO) = 1) and from, nVV: VR 2 VW = 0 since VV is a symmetric and VR 2 VW an antisymmetric tensor (but only if last term in (B.2) vanished since V.VO = 0 (see Eq. (10)). Thus, from (B.2), (R 2 VW -n(V -V)V) 5 0 if On 10 = 0). The Appendix C on the evolution of E, for a simplified model In an early paper focusing on a physical explanation, Hirshman [5] constructed a simplified model problem to show the evolution of Er, V and V., leaving out inertia, variations of n3 in the flux surface and assuming a phenomenological toroidal damping time rT such that r, << rT. Here r is the poloidal rotation damping time and the mechanism for rT can be a perpendicular viscous force. From the toroidal component of (2), and summing over species and using Ampere's law (3b), Hirshman obtained rigorously P( mjnjR2V g+ - (E - V4i)/47rc =0 + 0(-ri) (Cl) Thus, P. is a constant of the evolution on the fast r, time scale. Using Eq. (10a) for Vj, Eq. (C.1) can be seen to link the time evolution of Vi1 (or V,) with E, in accordance with our more general result, Eq. (13c). It is instructive to develop this link in more detail. From Eqs. (C.1), (10a) and cE.Vk = -W(VO)2 one gets a &8 K 47rmnc 2 rn22 B2 IB , - wR 2 B j~ - 2 =0 (C2) where the last term can be neglected since 4imnc2 /B 2 = c2 /C2 Equation (C2) implies that mn Vijj (mnR2) = Next, from parallel momentum balance (nrmjBV1 y) = - -V. -I) = - The last step was taken from Ref. [5]: (B - V - I) = 3((b . VB) 2 )ji 1 (V/B,) 11 B p1(C4) (C3) pi = RqnmVthv./[(1 + V.)(1 + -- /2v.)] (C5) where for concentric circular flux surfaces ((b - VB) 2 )/(B 2 2q 2 R 2 Thus, in the Pfirsch-Schlnter regime (e3/2V. > 1)Aj = n&mfvo2 /v,. Thenfrom Eqs. (lob), (C3) BpB P V + ~E TiVI (C6) Eq. (C4) thus becomes & ~>jminjj mjnjVj ~- -( Y Vpj m-j~ with the poloidal damping frequency given by = 2 Vi Finally, owing to Eq. (C6) the radial electric field, E,. = RBpw/c will relax to a quasisteady state at the same rate r; 1 as the poloidal flow. 12 Appendix D on neglecting V()- V. We neglected this term in the continutity equation (1b). inertial term ~ 2 V0 2 Scaling V through the in Eq. (13b), and Ve = V, from Eq. (10c), one obtains readily V )/V ~ 8,, the Larmor radius parameters - in the poloidal field. The diffusion driven radial flow TV is yet smaller than V,( by the factor (ni7r-i) rj1 is the collision frequency. 13 in an impure plasma, where Appendix E on alternate form for w evolution The purpose of this appendix is to obtain an alternative form of Eq. (13c) which explicitly contains information of toroidal angular momentum conservation which is useful for numerical computation. The flux-averaged equation, Eq. (13c) resulted from forcing the constraint w = W(V) (defined in Eq. (9)) on the unaveraged equation (13b). However, one always has the choice to multiply by any -varying function before flux-averaging and thus to arrive at different forms of the resulting equation. In Eq. (13c), we chose to annihilate the term involving Tb. Vn by determining E(B2 -Eq. (13b) ). 2 VC -Eq. (2) By taking E(R ), one finds the toroidal momentum conservation equation [Mn (V + (d =(C B VB B2 W ( - R2 C) VO) + (El) Therefore, instead of annihilating Tb . Vn, we preserve the form E determining Z( A (m n Eq. (13b) ) which yields aw+C'w2 +C2w+C1=0 (E2) where + 2 12 +31 1 E-B Bn 2 Bm(iVi - B (- 2 355(1 14 )I 2 2 |VB -- 1 = nm(j..V IV ( C,= C3~ w) by vB) 2 2 )] B2)+mnvd B + . B4 .VB B2 2b.-VVII + b.VB 1 B )I + (RM± ) +T( B D' = (V1 2 4-irc 2 + B By straightforwardly calculating D'- Eq. (E2), E(IVVJ Z(V1 m Eq. (11) ), V 2) mW Eq. (11) ), y Eq. (12) ), one can reconstitute the toroidal momentum conservation equation (El). 15 References [1] K. H. Burrell, T. Ohkawa, S. K. Wong, Phys. Rev. Lett. 47, 511 (1981). [2] C. S. Chang, Phys. Fluids, 26, 2140 (1983). [3] W. M. Stacey, D. J. Sigmar, Phys. Fluids 27, 2076 (1984). [4] P. Smeulders, Nucl. Fusion 26 (1986)267. [5] S. P. Hirshman, Nucl. Fusion 18 (1978)917. [6] W. M. Stacey, D. J. Sigmar, Phys. Fluids 28 (1985)2800. [7] C. T. Hsu, D. J. Sigmar, Sherwood Fusion Theory Conf. 1987 (San Diego) Paper 2C-6. [8] W. M. Stacey, M. A. Malik, Nucl. Fusion 26 (1986) 293. [9] Braginskii, Revs. Plasma Physics Vol. I, p. 250. 16