Stability of a Taylor-relaxed cylindrical plasma separated from
the wall by a vacuum layer
Citation for published version (APA):
Schuurman, W., & Weenink, M. P. H. (1988). Stability of a Taylor-relaxed cylindrical plasma separated from the
wall by a vacuum layer. (EUT report. E, Fac. of Electrical Engineering; Vol. 88-E-207). Eindhoven University of
Technology.
Document status and date:
Published: 01/01/1988
Document Version:
Publisher’s PDF, also known as Version of Record (includes final page, issue and volume numbers)
Please check the document version of this publication:
• A submitted manuscript is the version of the article upon submission and before peer-review. There can be
important differences between the submitted version and the official published version of record. People
interested in the research are advised to contact the author for the final version of the publication, or visit the
DOI to the publisher's website.
• The final author version and the galley proof are versions of the publication after peer review.
• The final published version features the final layout of the paper including the volume, issue and page
numbers.
Link to publication
General rights
Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights. • Users may download and print one copy of any publication from the public portal for the purpose of private study or research. • You may not further distribute the material or use it for any profit-making activity or commercial gain
• You may freely distribute the URL identifying the publication in the public portal.
If the publication is distributed under the terms of Article 25fa of the Dutch Copyright Act, indicated by the “Taverne” license above, please follow below link for the End User Agreement:
www.tue.nl/taverne
Take down policy
If you believe that this document breaches copyright please contact us at:
openaccess@tue.nl
providing details and we will investigate your claim.
Stability of a Taylor-Relaxed
Cylindrical Plasma Separated
from the
Wall by a Vacuum Layer
byW. Schuurman and M.P.H. Weenink
EUT Report 88-E-207 ISBN 90-6144-207-9 November 1988
ISSN 0167- 9708
EINDHOVEN UNIVERSITY OF TECHNOLOGY
Faculty of Electrical Engineering Eindhoven The Netherlands
STABILITY OF A TAYLOR-RELAXED CYLINDRICAL PLASMA SEPARATED FROM THE WALL BY A VACUUM LAYER
by
w.
Schuurmanand
M.P.H.
Ween ink
EUT Report 88-E-207 ISBN 90-6144-207-9
Eindhoven
November 1988
association agreement of Euratom and the "13tichting Voor Fundamenteel
Onderzoek der Materie" (FOM) with financial support from the
"Nederlandse Organisatie voor Wetenschappelijk Onderzoek" (NWO)
and Euratom.
This report Was previously published as "Rijnhuizen Report 88-177".
CIP-GEGEVENS KONINKLIJKE B1BLIOTHEEK, DEN HAAG
Schuurman,
w.
Stability of a Taylor-relaxed cylindrical plasma separated from the wall by a vacuum layer / by W. Schuurrnan and M.P.H.
Wee~ink.
-
Eindhoven: Eindhoven University of Technology,Faculty of Electrical Engineering. - Fig. - (EUT report,
ISSN 0167-9708; 88-E-207)
Met
lit.
opg., reg.IS8N 90-6144-207-9
S1SO 535 UDC 537.5 NUG1 812
ABSTRACT
The minimum energy principle of Taylor is extended to the case where a vacuum layer is situated between the weakly resistive plasma and the per-fectly conducting wall. The second variation is investigated for periodic
perturbations (m and/or
k~O)as well as for symmetric perturbations (m-k-O).
A necessary and sufficient criterion for stability is derived by means of
the Rayleigh ratio method and turns out to be identical with the result of
the resistive mode analysis. Stability diagramsA
vs k are presented. A new instability region at m-I is found at negative values of k, already present when .\ is small. The symmetric mode becomes more unstable as the vacuum layer grows thicker.Schuurman, W. and M.P.H. Weenink
STABILITY OF A TAYLOR-RELAXED CYLINDRICAL PLASMA SEPARATED FROM THE WALL
BY A VACUUM LAYER.
Faculty of Electrical Engineering, Eindhoven University of Technology,
The Netherlands, 1988.
EUT Report 88-E-207
Address of the authors:
Prof.dr. M.P.H. Weenink,Faculty of Electrical Engineering,
Eindhoven University of Technology,P.O. Box 513,
5600 MB
EINDHOVEN,
Contents 1. 2. 3. 4. 5. Abstract Introduction
The relaxation model
Derivation of the first and second variations Stability of the symmetric equilibrium
A) with respect to periodic perturbations
(m and/or k ~ 0)
B) with respect to symmetric perturbations
(m
=
k=
0)Discussion and numerical results Acknowledgement References Figures Appendix A Appendix B Appendix C i i i 1 3 5 10 10 15 20 23 24 26 37 38 40
1 . I NT[{ODUCT lUN
Taylor' 5 variational principle (1974) minimizes in the cylindrical
approximation the magnetic energy of a toroidal plasma column enclosed by a
perfectly conducting metal wall. The minimization is subj ected to the
constraints of constant
longitudi~al
flux and magnetic helicity K - JA-B dT(Lagrange multiplier -
A/2).
These constraints are assumed to be valid on a
timescale intermediate between the ideal MHD timescale and the diffusion timescale. The minimizing magnetic field satisfies the Euler equation
v
XB
AB ,(1)
and its Bessel solution has among others the merit of correctly demon-strating the reversal of the longitudinal field component, as observed in reversed field pinch experiments.
A minor shortcoming of this theory is its inability to describe plasma
equilibria with a finite pressure, and equilibria with a current densitythat falls off rapidly towards the metal wall. To the latter problem a
number of solutions have been proposed (ORTOLANI,
HASEGAWA,
SCHOENBERG
(1984». In some of these solutions, the role of a vacuum layer between the
plasma and the wall is acknowledged without, however, incorporating this
layer into the variational principle. KONDOH (1985) simulated the layer of
enhanced resistivity near the wall by generalizing the helicity invariant
to
ff(~)A.B
dr, where
~
is the poloida1 magnetic flux and
f(~)
a function
gradually decreasing from 1 at the centre of the plasma to 0 at the wall.
On the contrary, in our work f(~) steps down from 1 inside the plasma to 0
in a vacuum layer surrounding the plasma. This choice enables us to treat a
free-boundary problem, in which the position of this boundary is varied
along with the other quantities in the variational principle. The totalmagnetic energy (plasma+vacuum) is minimized subject to the plasma helicity
and other constraints, necessary to let the occurring boundary terms in the principle vanish.The first variation of the energy, put equal to zero, yields the
equilibrium states. This part of the analysis has recently been published
elsewhere (CENTEN et al., 1986) and is therefore dealt with rather briefly.
The accent lies on the full investigation of the stability of the
equi-librium states by means of the minimization of the second variation derivedIn Section 2 the model is described, after which the first and second variation are derived in Section 3. The accessorial problem (minimization
of the second variation of the magnetic energy) is solved for periodic (m
and/or
k~O)and symmetric modes
(m~k-O)in Section 4. Finally, in Section 5,
2.
THE RELAXATION MODEL
We consider a toroidal plasma surrounded
by
a vacuum layer and aperfectly conducting metal wall with circular cross-section (radius b). The plasma boundary is allowed to move during the relaxation. In the variational principle, both the vector potential and the position of the plasma edge are varied and acquire stationary values at minimum total magnetic energy of the configuration. Constraints needed in this model are the conservation of the toroidal fluxes in the plasma and in the vacuum
region,
and of the magnetic helicity of the plasma. The plasma-vacuum
interface is assumed to be an ideal conductor, so that the peloidal vacuum
flux
is also unchanged during the
relaxation. The
latter invariant,
however, can be omitted if a surface current at the plasma boundary is absent.The mathematical formulation becomes as follows. We minimize the total magnetic energy
W B B2
J
*
dr
V
0 P+J
V v B2 Vdr
21' owith respect to variations in V , A and A . V
P P v P
(2)
and V are the volumes of
v
and A
and A
are the
p v
the plasma and the vacuum region, respectively,
corresponding magnetic vector potentials. The conserved quantities are:
I' K -
J
A Bdr ,
0
V P P magnetic plasma helicity
P
toroidal plasma flux
1/>pt
J
B·
dS ,
S P
pp
1/>vt
J
B • dS ,S
v
toroidal vacuum flux
vp
1/>vp
J
B• dS .
S
v
poloidal vacuum flux
vt
where S , Spp
vp
and Svt are the poloidal cross-section of the plasma,(3)
(4)
(5)
(6)
the
poloidal cross-section of the vacuum, and the toroidal cross-section of thevacuum (e.g. in the median plane), respectively.
The toroidal geometry calls for an additional constraint, the flux
through the hole of the torus (REIMAN, 1981) ,
B • dS
f
Cmt
A • de ,
v
(7)
where Sh is the cross-section of the hole of the torus and C is a contour
mt
on the metal wall encircling the hole.
Furthermore, there are a number of boundary condi tians. First I we
have the continuity of the magnetic pressure across the plasma boundary r :
p
1
2"
o B2 p (r ) p B2 v (r ) pThe vector potential can be taken continuous across the boundary r : p
A
(r ) -
A(r ) .
p p
v
p
(8)
(9)
Finally, since the plasma boundary and the metal surface are flux
surfaces, we have n (r ) p p n (r ) • B (r ) - 0 . m rn v m (10) (11)3.
DERIVATION OF THE FIRST AND
SECONlVARIATICiNS--Of the four invariants defined in Eqs (3) - (6) needed in the
mini-mization of WB, only the helicity K is taken account of by means of a
Lagrange multiplier (-
A/2).
The remaining three enter into the conditions
on the free and rigid boundaries. We therefore minimize the generalized energy functional
f
8 2
8 2
fA'
F
5-E.. dT +
f
21"V dT -
A BdT
(12)
21"
21"
V
P P V 0 V 0 0p
v
p
*
Note
that the
integralsin Eq. (12)
have to be read asf~
dT
etc.21"
0
The vector variables A, B, r are decomposed into equilibrium parts A ( )'
P P v
B
p(v)
(index p indicating the plasma, index v the vacuum), aer , and theperturbations
5A ( )' 5B ( )'
5a.The latter vectors are Fourier-analysed in
p v
p v
i(m8+kz)
Band z, and modes (m,k) proportional to e are considered separately. There is some freedom in the definition of the displacement oa of the plasma-vacuum boundary.
In
Appendix A, we prove that we may conveniently choose oa in the radial direction:5 (8
a ,z)
-5
aei(m8+kz)
er (13)
In order to evaluate the first and second variation of F, we
deter-mine the energy difference LlF -
F[A+5A]-F[A] between the perturbed state
and the equilibrium state in toroidal geometry:
I" o
t.F
-2rrR
2rr
r
J
z-O
8-0
a+oa(8,z)
J
r-O
A2 (A +5A )·(B +5B )}rdrd8dZ +
p
p
p
p
2rrR
2rr
+
r
f
bf
(14)
Z-O
8-0 r-a+5a(8,z)
2rrR
2rr
r
I
z-o
8-0
aIH
r-O
B2_
~
A .B }rdrd8dz
-p 2 P P2rrRt 2rr
f f
z-O 8-0
bf
1 82 rdrd8dz .
2
vr-a
I,
The terms are re-arranged as follows:
a+oa(8,z)
+
f
a
b
{ B -oB
p p- A (A -oB +B -oA )}rdr +
2 P P P P(oB )
2 PA oA -oB }rdr +
2 P P+
f
a+oa(8,z)
(B oB )rdr
-v vf
t
B~
rdr +
f
t
(OB
V)2rdr d8dz .
a+Oa(8,z)
b
I
a a(15)
The first integral
I,
in Eq. (15) is partially integrated:
21rR
21r
r
f
z-O
8-0
a+oa(O, z)f
o
5A -(VxB -AB )rdr +
p
p
p
a+5a(O,z)+
f
o
a+5a(8,z)
f
5A -(VxB -AB )rdr+(a+5a(O,Z)){B
(a+5a(O,z))p
p
p
pz
o
where the result that OA
p8
and 5Apz can be considered real has been
Functions of a+oa in the integrand are developed in Taylor series
around r=a. Derivatives like B' (a), A' (a) are eliminated with the aid of
pz
pz
VxB - AB and B -
VxA,and Egs (C5,6,lO,11) are applied. I, then becomes
2"R
2"
a+6a(9,z)
J
tJ
I, -
J
6A o(VXBp-AB )rdrd9dz+2,,2R (6a) 2{-2l (B 2 ._B2)
-P
P
tpo
pz
a°
(17)In the second integral 12 between square brackets in Eg. (15), B (r)
pand A (r)
p
are developed in Taylor series in r-a in the interval
(a,a+oa).
Then the r-integration is performed first. The 8-z-integrations follow,with proper account of the cos (me+kz) -dependences in the integrand. The
result is2"
a+6a(e, z)
J
J
{t
B~(r)-
I
A
p
(r)oB
p
(r)}rdrd9dz
o
°
a- 2,,2Rt{-41 B
2
p
(a) - -2
1
B
2
p
.(a) + -4
A
(A
B -A
B )
+
op9 p9
pz pz a
+
(18)
b
J
{5Av ·VXBv}rdr +
a+5a(8,z)
J
{V.
(5AvXB) }rdrjd8dZ
a+5a(8,z)
-(a+5a(8,z»)B
(a+5a(8,z»)5A 8 (a+5a(8,z»)cos(m8+kz) +
vz
v+ (a+5a(8,z»)B 8(a+5a(8,z»)5A
(a+5B(8,z»)cos(m8+kz)-jd8dZ +
v
vz
bJ
Z~O 8~0a+5a(8,z)
(5A .VxB )rdrd8dz .
v v(19)
As in the first integral I, in Eq. (16), functions of a+5a are
ex-panded in power series around r=a. Derivatives of Bv8' Bvz' AvO' and Avz
are eliminated with VxB -
0 and B - VxA, and Eqs (C7,8,12,13) are applied.
We obtain the result
(20)
The fifth integral Is in Eq. (15) is reduced in the manner of 12 and
the result can most easily be found by accepting Eq. (18), changing the
indices p to
vand putting A-O ,
(21)
We substitute the integrals in Eqs (17), (18), (20) and (21) into
Eq. (15), leaving the third and the sixth integral in Eq. (15) unchanged.
First and second-order terms are collected and the following results are obtained:
I. First variation for cylindrically symmetric equilibria with perturbation
II.
.
i(mO+kz)
proportional to e : !lof
o2"R
2"
r
f
a+6a(O. z)f
z~O 0-0r-O
2"R
2"
+
r
f
bf
z-O
B-O r~a+oa(B.z)Euler equations:
VxB
P
AB .
p'
Second bation
VxB
v- 0
variation for cylindrically
.
i(mB+kz)
proportional to e :2"R
2"
a r ,V
P r , V v(22)
(23)(24)
symmetric equilibria with
pertur-r
f
z~O e~of
{~(5Bp)2
-
t
5A
p
'5B
p
}rdrdBdZ +
r~O+
2"R
2"
r
f
z-O
B-O bf
t(5B
v)2 r drdBdZ
r-a
+ ,,2R {B2 (a) _ B2 (a)}(5a) 2 .
t vz pz(25)
4. STABILITY OF THE SYMMETRIC EQUILIBRIUM
a) Stability with respect to periodic perturbations (m and/or k#O).
First, we determine the perturbed quantities in the vacuum region. When we write out the basic equations for 6B
v
'1
x
5Bv
o .
'1 • 5B - 0 • v
in cylindrical coordinates with a dependence proportional to find for the radial part of the z-component the equation
5B"
+
1 5B'vz r vz
m
Z- (k
z
+
?
)
5B - 0 . vzThe total solution thus becomes
5B vr 5B vz A, I m (kr) + AzK m (kr) • (26)
(27)
i(m8+kz) e , we (28)(29a)
(29b)
(29c)
where the prime now denotes differentiation with respect to the argument kr. The perturbation of the vector potential, SA, follows from
5B - '1
x
5Av v
(30)
Since it contains an arbitrary gauge we can leave the value of 5A (r) vr
open. It drops out in what follows so that we do not actually use the gauge
freedom. The components of
Eq. (30)
lead toSA (r) vz i k 5B vr r k:
f
5BvO (s)ds b r r im+
r-f
5BvO(s)ds + ikf
SAvr(s)ds b b rf
SA (s)ds vr b (31a) (31b)Equation (3Ib) satisfies the boundary condition oA (b) vz we fulfil the boundary condition oAve(b) ~
0
by putting- O. In Eq. (3Ia) oB vr (b) - 0, which
yields one relation between the integration constants AI. A2:
A,l' (kb) m
+
A2K' (kb) m ~ 0 . (32)Next we consider the second variation 62F derived in Section 3,
Eq. (25). Since the perturbation 6A in the plasma is arbitrary, we must find the minimum value of 62F (accessorial variational problem). To this end we define the Rayleigh ratio R
A:
(33)
RA ~
---J(VX6A)2dT
+
J(VX6A)2 dT
p
V
where P and V are shorthand notations for V and V respectively.
p v
Here we have assumed a cylindrical geometry with energies measured
per unit length in the z-direction (dr - rdrdB). We denote the denominator in Eq. (33) (the norm), by c 2 , so that the stability condition becomes
(34)
At the minimum of 52F, RA will reach a maximum, which we call
~
(wetake A positive; negative values of A could be taken into account by using
absolute values of A, Q and R
A). According to Eq. (34), the stability
criterion can now be written as
Q
>
A • (35)We use the shorthand notation y for 6A and g - Vxy for 6B. The maximum value of RA is determined by putting 6R
A ~ 0:
J(VXy)oydT -
~{B~z(a)
-B~z(a)}(6a)2
pJ
(Vxy) 2dT+
P J(vxy)2 d r Vo ,
or
2
f
(Vxy) oydT - 2;
{B~z
(a) -
B~z
(a) }(oa) 2
p
{f(Vxy
)2d
TP
+
f
(Vxy) 2 }dT
V
The last term between square brackets can be reduced as follows
of ;
(OB
v
)2dT - f(vxoy)ovxoYdT
f(vxvxy)ooydT
+
f
{oYX(VXy)}odS
V
V V S p, S v
In the vacuum we have VxVxy - 0 and oy - 0 on the wall at r - b
Therefore, we obtainof
t
(oBV)2dT - - f{oyx(vxy)}odsa .
V
We substitute this into Eq. (36) and mUltiply with c 2 . The result is
J 2 (Vxy) oOydT
P
- 27r{B2 (a) - B2 (a)}oao 2a
+
J(OYXy)odS
_
A
vz pz a 2"
f
(VXVxy) oOydT
P+
Jvo{OYX(VXY)}dT
PEquation (37) leads to the two following conditions.
- 0 .
(36)
T. The Euler equation in the plasma:
Vxvxy - oVxy - 0 . (38)
II. Surface condition on the plasma boundary:
f
(6y xy -l
6y x(Vxy).1
6y x(Vxy )) .dS - 2" {B2 (a) _B2 (a)}6a6 2a - 0 •p p o p
p
0v
v
a
Avz
pz
(39)
where the indices p, v now denote values at the plasma side and at the
vacuum side of the plasma boundary respectively.
The solution of the Euler equation (38) with
yproportional to ei(mO+kz) is
well known:
6B r
where D is an integration constant.
(40a)
(40b)
(40c)
The Euler equation (38) can be combined with g -
Vxyto give after integration:
y
1 g+
Vx
o
x
( )
i(mO+kz)
X
r e
.
From Eqs (40) and (41) then follows:
SA r 6A z - v ~ z
(41)
(42a)
(42c)
In the complete solution appears a set of six constant parameters I viz.
four integration constants A}, A'2 I x(a) I and D, and the two undetermined
quantities
SA
(a)vz (see
Eq.
(31» and the amplitude of the radial pertur-bat ion 6a. As we will see there are six boundary conditions, homogeneous in the six constants. The resulting 6x6-determinant, put equal to zero, yieldsthe relation aU) we are looking for. Then aU) ~ A gives us the critical
value of A above which the plasma becomes unstable. The six boundary
conditions are (32), (CS,7,10,12) [combined with the solutions (31), (42) and (B2 ,4)
1
and the surface condition (39). The latter condition can be simplified by noting thatEq.
(C14) holds for all variations, not only for those that maximizeR
A
.
By taking the variation ofEq.
(C14),Sy(a)
(43)we conclude that
Sy
(a)x
y (a) - 0 inEq.
(39). Since S - 2na we obtain:p p a
B (a)g (a)
+
B .(a)g .(a)-pz pz p" p"
- B (a)g (a) - B .(a)g .(a) -
2~
{B2 (a) - B'(a)}sa
vz vz Vu Vu ""a zv zp
o .
(44)The 6x6-determinant
la
ij
l
mentioned above consists of the followingelements:
a" - I~(kb); a'2 - K~(kb); a'3
.'!'..
a'
a2S - 0; a26 - B (a); pz aS4 =- 0; ass ka m
o
J~{(a2_k2)l,a},
(a2 _k2)1, , a36 - -B p8 (a); B (a); vz_
~
{B2
(a) -82
(a)} .2Aa
vz
zp
(45)Fully written out, the determinant put equal to zero assumes the form
C>
{B2
(a) ZAavz
• {I' (kb)K'(ka) - I'(ka)K'(kb)}
m m m m
o .
(46)This relation between ex and A will be investigated numerically in Section 5, for the case in which the zero order surface currents vanish.
b) Stability with respect to symmetric perturbations (m-k-O).
Symmetric modes differ from helical modes in their behaviour with respect to helicity conservation. K must be conserved up to all orders of
the
perturbations.
In
the accessorial problem
(minimization of
S2F)condition that is both necessary and sufficient. The periodicity in
e
and zof the helical modes guarantees the vanishing of
oK
since it is a volumeintegral, linear in the variations 6A,
oS.
In the symmetric mode m=k-O.however, the constraint
oK
= 0 must be taken into consideration explicitly.For a moving plasma, K-conservation means
a+oa
J
(A +oA )o(B +oB )rdrp p p p
o
or,
with
an expansion inSa:
a a
J
A °B rdr p po
J
(A ° oBP
P
o
+ oA °B + oA °oB )rdr + aoa{(A +oA )o(B +OB)} +
p p p p p p p p r-a
+ 21 (oa)Z{rA oB }' -
a .
p p r-a
Up to first order we find:
a
J
(A °oB + B oM )rdr + aoaA (a)oB (a)p p p p p p
o .
a
(47)
(48)
Partial integration of the first of the two integrals in Eq. (48),
combined with the relation oA(a) ~ oaxBp(a), yields the invariant
c:
a
c
=f
o
B 00A
dr -a .
p
(49)
Before solving the accessorial problem, as in case a). we determine
the perturbed symmetric fields in the vacuum region:
oB vr
- a
SA vra
,
oBve cdr SAve Cz r2_b 2 2 roB
Cz SA-
-
c, In.E.
(50) vz vz bNext we turn to the accessorial problem, i.e. the minimization of
52F* -
f
H
p
5A ·5B
}dr
+
f.1
(5B
)2dr+
p p V 2 v
{ B2 (a) - B2 (a)}(5a)2 -
vz
pz
VAJ B ·5A dT •
p
p
p(51)
where energies are again taken per unit length in the z~direction (dT is
now rdrdO). and the constraint C has been incorporated with a Lagrange
multiplier -VA. The stability condition can now be put in the form
by the definition
p V
(5B )2dT .
v
and the introduction of a Rayleigh ratio R
A
:
f
(5A .Vx5A +
p
P
2vBP
·M )
P
~{B~z
(a) _ B2
pz
(a)}(5a) 2
P
RA
f
(Vx5A )2dT
+
f
(Vx5A )2dT
P
P
Vv
1
To determine the maximum of R
A, which we call
""
we put 5R
A
0:
1
5R -
A
~c
2" (B2 _B2 ) 5a52a ] _
). vz pz a1 c4
f
_ p
J{Y'VXY+2VB 'Y}dT -
p
~
A{B2 (a)-B
vz
2
pz
(a)}(5a)2]
f
_5 J (Vxy)2dT + 5 J (Vxy)2dT] - 0 .
p V(52)
(53)
(54)
We multiply Eq. (55) by c',
introduce 1 and perform a few partiala
integrations. Thus we arrive at
I.
The Euler equation in the plasma VxVxy
p
aVXy -
pavB .
p(56)
II. Surface condition on the plasma boundary,
identical to Eq.
(39) .The Euler equation constitutes two coupled equations for the magnetic field
perturbations oBe and oB
z
' both functions of r only. The solution reads:
avB
oBe
AJ, (ar)
-
a->.
0J,(>.r)
,
avB
oB
- AJ (ar)
-
a->.
0J (>.r)
(57a,b)
z
0 0From oB - vxoA follows
1
avB
DAe
- AJ,(ar)
0J, (>.r)
>.(a->.)
,
a
lAJ
avB
oA
z
a
(ar)
0J (>.r) + k .
(58a,b)
0>.(a->.)
0A and k are constants that have to be determined from the boundary condi~
tions. As in the periodic case there are altogether 6 constants, A, V, k,
ell C2 and 8a. The corresponding 6 homogeneous equations determining these
constants are
(39),(C5,7,10,12) and
(49).They lead to a 6x6-determinant
with elements b
ij
:
b"
- aa{J (>.a)J (aa)+J,(>.a)J (aa)}; h"
0 0 0
aa { '
2}
-J,(>'a); b'5 - -J (>.a); b,. - -
~B (a)-B
(a) ;
o a a vz pz
b 2, - J,(aa); b 22 - J,(>.a); b 23 - h 2• - b 25 - 0; b 26 - J (>.a);
ob 3, - J (aa); b 32 - J (>.a); b 33 - 1; b 3• - b 35 - 0; b 36 - -J,(>'a);
o 0b61 - a-A A {Jo(Aa)JI(aa) 1);b46 -B I B (a); vz o
b
1
In -; hS5 - 0; hS6 = -aB
o (59)The determinant
Ibijl
put equal to zero yields the required functiona(A). According to Eq. (52) the marginal stability is determined by putting aU) ~ A.
5.
DISCUSSION AND NUMERICAL RESULTS
In the absence of zero-order surface currents, the marginal form of
Eq. (46), when
Qis replaced by A, is also found from a resistive mode
anal-ysis
(c,'-method, see e.g. GIBSON and WHITEMAN,
(1968». The differential
equations for the fields are the same in both cases. So, in order to prove the equivalence of the two theoretical approaches. we only have to show that the boundary conditions used are equivalent. Firstly. the continuity
of SB on the plasma surface. used in the resistive mode analysis. also
r
holds in the variational method. This follows from the continuity of B on
n
the surface, where
(60)
Since the zero-order field components are continuous, the continuity
of 6B
immediately follows. Secondly, the pressure balance at the perturbed
r
plasma surface has to be fulfilled:
B2 (a+6a) _ B2 (a+6a) .
p
v
(61)
Expansion in 6B and 6a yields the condition
[2B e(a) 6B
e
(a) + 2B
(a) 6B (a) + 6a(B2
e
' (a) + B2 '(a)}] 1 - [ ...
J •(62)
o oz Z 0 oz p vac
The first and second terms on both sides cancel due to Eq. (44),
(pressure balance at the unperturbed boundary). The third terms are also
found to cancel when the zero-order fields are substituted. Thus. we seethat the pressure balance indeed holds, which establishes the equivalence
we set out to prove. It is satisfactory that the agreement between the resistive mode analysis and Taylor's variational principle, already presentwithout a vacuum layer, still exists in the presence of such a layer.
It has been known for some time that a force-free plasma with a free
boundary is intrinsically unstable (e.g. KRUGER (1967) for ideal MHD plasma,
MILLER (1985) for a slightly resistive plasma). In the framework of our
variational analysis we can deduce the instability as follows. The
equilib-rium of a force-free plasma surrounded by a vacuum layer can be represented
by the equation
j - ).(r)
B ,
(63)where the function A(r) steps down from a constant value A in the plasma to zero in the vacuum. The second variation of the energy S2W can be expressed in the perturbation
roB :
r b
1.
I
--.,----;-r
.,,---,,-20 m2 + k2r2).'(krBe'mB)
.
B
k B
Z }(r6B )2] dr .
m
e
+
r z
r
(64)
The contribution to 52W due to the jump .6.)" at the plasma-vacuum interface
then goes like
a3 ~). 1
+
k" a 2 (6 Br) a
2
"-'-.-;;:2:::B-;(-7)"'--:(--;-)"'/-="2""(--;-) a . r
AD
a Bz a aB a
s
(65)
for a singular point r chosen to be near a, Since fj,),. is negative and
s
). . 2B
e
(a)B
z
(a)/aB 2 (a) is positive for the Bessel-function solution, .s2W
can be arbitrarily large and negative by having r
just inside the plasma
s
boundary (provided
~O),implying instability.
We start the numerical results with a presentation of a: VB ..\ plots
according to Eq. (46) for various values of
P -
b/a. We restrict our analysis
to a model without zero-order surface currents on the plasma-vacuum
inter-face. Equation (46) then reduces to
I' (kb)K' (ka)
m mI'(ka)K'(kb)
m
m
- H(a)V(),) , (66)I (ka)K' (kb)
m mI' (kb)K (ka)
m m where H(a) ~ (67)and
(68)The case {J=1 is outstanding here because it describes the stability of a plasma without a vacuum layer (Taylor). The degeneracy of this case is
apparent from Eq. (66) and leads to the straight horizontal lines H-O and
vertical lines V-O. In Fig. 5.1a we have plotted the lines H-O,
V~Ofor m-O
and
Ikl-
0.50, 1.25 and 2.00. Fig. 5.1b shows these lines for m-l and k--l.OO,
+0.50, 1.20 and 1.70. For
P~lthe degeneracy is removed, as can be seen by
comparing Figs 1 with Figs 2 (P-1.0l). For m-O Figs 3a, 4a, Sa show the a
vs >. curves for
P~1.10,1. 50 and 3.00 and 4 typical values of k. The
cor-responding curves for m-l (but p-10.00 instead of 3.00) are presented in
Figs 3b, 4b, 5b.
The transition stable-unstable can appear in two ways. First," a: can change sign tantamount to a transition from stability to instability
according to Eqs (34) and (35). Note that in this transition 02W jumps from
~
to
-~so that the ordering in small perturbations (oW etc) breaks down.
Since we do not have an exact calculation of the perturbation in the
generalized magnetic energy we are forced to accept the change in sign of a: as an indication of a transition stable-unstable. Secondly, a:(A) may cross
the line of marginal stability '"
=
>. (for convenience also indicated in
Figs 2 - 6) .
The familiar stability diagrams >. vs k are obtained from the a vs >.
curves by putting a: either equal to A or equal to O. The results are seen inFig. 6a (m-O) for p-l.02, 1.10, 1.15 and 1.20 and in Fig. 6b (m-l) for p-l.10,
1.50, 3.00, 5.00 and 10.00. Stable and unstable regions are designated by the
letters S and I respectively. Transition lines Q=A are drawn fully, the
a-O
transition is dotted. For comparison, the Taylor curve (~-1) is also shown
(dashed line). The instability of the configuration at m-l according to
Miller manifests itself by the new curve at negative values of k, causinginstability at small A.
The special case m-k-O is also treated without zero-order surface
currents. Equation (59) then reduces to a 3x3 determinant with elements
J,(aa); d '2 - J,(Aa); d
'3 ~J (Aa) ln
oP;
aa{J (aa)J (>.a) + J,(aa)J,(>'a»; d 22 - Aa{J
0 0 02 (Aa) + J;(>.a»;
J2(>.a)
o
Its numerical evaluation is shown in Fig. 7 for three values of
p:
fi -
1.00 (Taylor), 1.25 and 1.50. If we move from left to right, the first crossing of a particular Q: vs ..\ curve with the line cr ->.
determines thecritical value of A at which the instability sets in. This critical value if plotted vs {3 in Fig. 8. We observe that even this mode becomes more
unstable as the vacuum layer grows thicker.
Acknowledgement
The authors are greatly indebted to Dr. H.J.L. Hagebeuk and Mr. L.L.M.M. Verhoeven for their assistance with the numerical calculations. This work was partly performed under the Euratom-FOM association agreement with financial support from NWO and Euratom.
References
Centen, P. and M.P.H. Weenink, W. Schuurrnan
Minimum-energy principle for a free boundary, force-free plasma.
Plasma Phys. & Controlled Fusion, Vol. 28(1986), p. 347-355.
Gibson, R.D. and K.J. Whiteman
Tearing mode instability in the Bessel function model.
Plasma Phys., Vol. 10(1968), p. 1101-1104.
Hasegawa, A. and Y. Kodama, R. Gruber, S. Semenzato Study of reversed field pinch with surface current.
J. Phys. Soc. Japan, Vol. 53(1984), p. 1316-1325.
Houten, M.A. van and M.P.H. Weenink,
w.
SchuurmanOn the second variation of a minimum energy principle for a force free plasma with a free boundary.
In: Proc. 12th European Conf. on Controlled Fusion and Plasma
Physics, Budapest, 2-6 Sept. 1985. Contributed Papers, Part 2. Ed. by L. Pocs and A. Montva1.
Petit-Lancy-{Switzerland): European Physical Society, 1985.
Europhysics Conference Abstracts, Vol. 9F, Part 2. P. 382-385.
Kondah, Y.
An energy principle for axisymmetric toroidal plasmas.
J. Phys. Soc. Japan, Vol. 54(1985), p. 1813-1822.
Kruger, J.G.
Bijdrage tot de studie van stabi1iteitsprob1emen in de
kinetische plasmatheorie.
Verh. Kon. Vlaarnse Acad. Wet. Lett. & Schone Kunsten Belgie
Kl. Wet., Vol. 29 (1967), No. 97.
This papep by Kpugep is coveped fpom the gpeatest papt by the
foLLowing papeps by Kpugep and CaLLebaut.
Kruger, J.G. and D.K. Callebaut
Energy principle for gravitational and magnetodynamic stability with application to force free fields.
Mem. Soc. R. Sci. Liege, Seme serie, Vol. 15(1967), p. 175 et seq. Kruger, J.G. and D.K. Callebaut
On the least stable mode in cylindrical systems.
Z. Naturforschung, Teil A, Vol. 23a(1968) , p. 1357-1361.
Kruger, J.G. and D.K. Callebaut
Relations between adiabatic and incompressible (non-adiabatic) systems and their stability.
Z. Naturforschung, Teil A, Vol. 25a(1970) , p. 88-100. Kruger, J.G. and D.K. Callebaut
Variational principles for compressible and incompressible systems.
Z. Naturforschung, Tei1 A, Vol. 25a(1970) , p. 1097-1100.
Miller, G.
Error magnetic fields in a cylindrical plasma: Stability with zero pressure.
Ortolani, S. and R. Paccagnella, E. Zilli
Characteristics of magnetic field profiles in a reversed field pinch.
In: Contributed Papers Int. Conf. on Plasma Physics, Lausanne,
27 June-3 July 1984. 3rd Joint Conf. of the 6th Kiev Int. Conf. on
Plasma Theory and 6th Int. Congress on Waves and Instabilities in
Plasmas. Ed. by M.Q. Tran and M.L. Saw1ey.
Centre de Recherches en Physique des Plasmas, Ecole
Poly technique
Federale de Lausanne, 1984. P. 150.Reiman, A.
Taylor relaxation in a torus of arbitrary aspect ratio and cross section.
Phys. Fluids, Vol. 24(1981), p. 956-963.
Schoenberg, K.F. and R.W. Moses, Jr., R.L. Hagenson
Plasma resistivity in the presence of a reversed-field pinch dynamo.
Phys. Fluids, Vol. 27(1984), p. 1671-1676.
Taylor, J.B.
Relaxation of toroidal plasma and generation of reverse magnetic
fields.
•
aa
aa
•
H=O,
{J =
1.00 V=O m = 0 ka=±O.50,
,
,
•
,
Aa
•
aa
,
a.
•
H=O,
{J
= 1.00 V=Q m = 0 kS=±1.25, ,
•
,
Aa
,
aa
aa
H=O•
,
{J
= 1.00 V=O m = 0 ka=±2.00, ,
,
•
Aa
aa
H=O,
V=Q{J
= 1.00 m = 1 ka=-1.0Q,
,
,
,
Aa
•
•
H=O{J
= 1.00 V=O m = 1 ka=
0.50, ,
•
Aa
,
H=Q{J
= 1.00 v=o m =,.
ka= 1,20, ,
3•
,
Aa
•
•
H=O,~---~--~~---{J
= 1.00 V=O m=
1 ka= 1.70 , L -____ L -_ _ _ _ L-, __ -LL-____ L-__ ~ o 2 3 .Aa
Fig. la. a vs A plots for m=
0, Fig. lb. a vs A plots for m=
1,B
= 1.00 andIkaJ
= 0.50, 1.25,2.00
8
= 1.00 and ka=-1.00,0.50, 1.20,1.70a. a. a.
•
•
,
•
•
,
'V
,
,
,
•
•
,
•
/
"
•
•
•
,
:V
,
"
•
./
/
7
/
7
/
V
/
[7
v
/
v
/3:1.01 m=o ka=O.50•
•
"
A.
V
~
V
I~'"
1.01 m,..o ka= 1.00"
Aav
/
V
{) '" 1.01 m ; Q ka .. 1.50"
_.\ IIV
L
V
/~ = 1.01 m",o ka", 2.00"
AaFig. 2a. a vs
A
plots for m=
0,B
=
1.01 and ka=
0.50,1.00,1.50,2.00...
...
...
f3 ".
1.01 m = 1...
ka=I.20..
,
'.'
..
,
..•
'.'
..
,
M,.,
A...
,
".
,.,
,
..
,
.
,
..
,.,
...
...
,.'
f3
= 1.01 m =1,
..
ka=I.70..
,
2.50 2.16 3.02 3.U,."
3.1" A. Fig. 2b.a
vsA
plots form 1,
S
= 1. 01 and-"
...
a. a.'"
'"
..•
..•
..•
..,
'., ~ =1.10 m =1 ka .... OO ka =0.50•
"
,
..
,.,
..,
...
'"
'.'A.
,i."
'"
a."'
.
.
,
••
...
"
••
••
..
,
f3
'"'.10 m.,
ka=O-so"
...
..
,
...
...
...
A.
A.
"
..,
a. a •...
..
.
...
...
.
,
...
,
..
I~'"
1.10'.'
moo ka", 1.50...
..
,
"
...
..,
..•
...
...
.
..
'"
. h "i •"
...
a. <to..
,
...
...
...
...
..
.
..
..
.
,
...
...
..
,
() '" 1. to...
m =1...
ka= 2.50..
.
"
'.'
...
.
.. ..
,
...
.
..
...
.
..
A.
A.Fig. 3a.
a
Vs
Aplots for
Fig. 3b.a
vs
A
plots for
m ~
0,
S
~1.10
and m1 ,
S
~1. 10
and'.'
a. a."
'.0 '"...
...
'.' LO 0.',.
•..
..
,
A.
a. Aa"
r---~---~~---~----'" a.,
..
, ,
,
..
...
...
ka=1.70,
..
,
..
,
..
"
2.$0 2.'. 3.01 3.21 ,.~ s.aoA.
A.,..
a. n. u...
,
..
,
..
..
,
,.
,
,
..
,
..
...
,
..
...
~ :1.50 >0. m.
,
,
..
ka 3: <1.00 '..
...
,
.•
,
.. ...
,.,
,
..
,
..
,
..
A.
A.
Fig. 4a. a vs
A
plots form = 0,
S
= 1.50 and ka = 0.50,1.00,1.50,2.00Fig. 4b. a vs
A
plots form
=
1,S =
1.50 andaa aa
,
.•
"
'.'
,
.•
,
..
.
..
M'.'
.
..
.
..
"\a "\a"
a.•
a •.
..
,
..
..
.
.
,
...
,
..
,
..
,
..
~ ;; 10.00 m.
,
,.,
k • • O.70•
,
..
"
,
..
,.,
,
.•
..,
...
.
..
..,
,,\. ,la"
,
..
aa aa,.,
...
...
•.
.
...
..
,
•.
,
...
,
..
,
..
,.
,
,
..
,
..
,
..
•
"
'"
,
..
,
..
..•
,.,
'.'
...
.
..
A.
,I."
...
aa aa'.'
...
'"
...
...
'-',.
,
'.'
,
..
','
,.
,
~ :: 10.00,
..
m.
,
,
..
ka =4.00 ,."
".
,
.•
,
..
.
.•
..
,
.
..
.,.
.
..
"\aA.
Fig. So. rt vs )., rIots for
:--'ic: •
5b. a vsA
plots for m ~ OfG
~ 3.00 and'"
~ 1 ,S
~ 111.or'
an(;Aa
Aa
_50 _ 4 5 _ 4 0 . 3 5 . 3 0 - 2 5 . 2 0 _1.5 .10_0.5 0 05 ' 0 15 20 25 30 35 40 4 5 5 0~i""'"
""'" _
I
7~~;
•. 0 ~",,,,,,,,,,,,,,", . . .= ... ,,,,,,,,,,,, ...
~•.
0 3. 3.5 3. ,5
3.02.'
~...
== ...
====== .... ========================j2 .•
2.0 2",05
m=O {3=1.02 LO•••
•••
,L--L __ L-~ __ L-~~~~~~~~ __ ~~ __ -'-~~ __ ~L~ _ _ ~~ _ _ ~-J, -5.0 -4.5 _4.0 -3.5 -3.0 _2.5 -2.0 _1.5 -'.0.0.5 o 0.5 1.0 1.5 2.0 2.5 3.0 3.15 4.0 ".5 5.0ka
Fig. 6al. A vs ka plot for m
o
andS
1.02_5.0 _ 4. S _4.0 _3.5 .3.0 -2·5 .2.0 _1.5 _1.0 -05
,
0.',.,
L, 24,.
3.' 3.'•••
..,
'.0,.,
...
6.0.-••
..
•
'.0I
' . 0..
,
•• 5 '.0 ...._---
--
•. 0 3 .• 3 .• 3. 05
3.0 5 2.6 2.0 2.0 LS L '5
m=O LO {3=1.10 LO,
..
0 . ' 0 0 -5.0 - ... 5 .4.0 -3.5 -3.0 -2.5 -2.0 -1.5 -1.0 -0.5 0 0.' LO U 2.0...
3.0...
'.0...
'.0ka
-"
• 4.5 -4.0 _3.5 _3.0 _2.5 .2.0 -1.5 _1.0 _0' 0 0.' '.0...
'.0...
3.0 3.' '.0•••
.~ '.0 ' . 0.-AS
•••
•••
'.0...
...
I
..
0..
.
'
....
·
..
'
....
'.0S
....
-
---
S
' . 0 3 .• 3 . ' 3.0 3 .0I
•••
'.0 . 0...
S
m=O...
'.0 f>=1.15 . . 0 0.' 0,•
0 0 _5.0 -" .5 -4.0 _3.5 -3.0 -2.5 -2.0 -1.5 _1.0 .0·5 0 0.5 '.0•••
'.0 2.5. 3.0 3.' '.0...
'.0ka
Fig. 6a3. A vs ka plot for m
o
and8
1.15.5.0 _ •. is - '.0 _3.5 .3.0 _2.5 .2.0 _1.5 -1.0 _0.5 0 0.' '.0
..
,
'.0...
3.0...
'.0'.'
'.0 '.0 ' . 0 "-.-Aa
...
•
,
'.0I
' . 0..
,
•••
'.0S
---
' . 0 3.' 3 .• 3.0I
3.0...
..,
'.0 • .0...
...
m=O...
S
13=1.20 • .0 0.' o .• 0 0 -!i.0 -" .5 -4.0 -3.5 -3.0 -2.5 -2.0 -1.5 -1.0 -0.5 0 0.' '.0..
,
'.0..
,
3.0 3.' •• 0..
,
'.0ka
4 r - - r - - r - - r - - r - - r - - r - - r - - r - - r - r r - - r - ' - - ' - - ' - - ' - - ' , r 7 , - - , - - , - - , 4 ~ I~,'
A8
"
S
3s
,.-.--
....... I
3...
...•....•...•••.•
2 2s
1
m: 1p=
1.10 ~5.0 _4.5 .4.0 -3.5 -3.0 .25 -2.0 _1.5 _1.0 _Q5 0 as 1.0 1.5 2.0 2S 3.0 3.5 4.0 4.5 5.00 ka Fig. 6bl. A vs k plot for m 1 andS
= 1.104r--,---.--~--~--r---r--,---.---r~ •• ,---r--,r--,---r---r--~-,.--,r--,---,4
/~,','
.-s
3 2I
I...
_
...
'
'.
'.
'.
....
Aa
s
I
___ ---,3... ...
2s
m=
1p=
1.50 7 0'.8
.5,0 -4.5 .4.0 -35 -3.0 -25 _2.0 .15 _1.0.05 0 os 1.0 1.5 2.0 2S 3.0 3.5 4.0 4.5 5 ka Fig. 6b2. A vs k plot for m 1 andS
= 1. 504r--r--r--r--r--r--r--r--r--r~r--r--r--r--r--r--V--y--r-,,-.4
,/ I
'\
(
"
1
~~/'
S
I
r:~:~...
I
"---=
3...
3-5
Aa
2 - - 25
I
1 - - 1 m=lf3.
3.OO ~5.0 -4.5 .4.0 -3.5 .3.0 -2S .2.0 _15 .1.0 _QS 0 OS 1.0 15 2.0 2.5 3.0 3.5 4.0 4.S 5.00ka
Fig.-6b3.
A
vs k plot for rn=
1 andS
=
3.002 - - 2
5
I
-
- 1 m= 1f3=
5.00 ~5~~~-4~.5~-4~~~.~~~.~~~O-.~U~_~2~~~_~~-_~,.0~_~~~O~~~~-,~~~~~~2~~~~~~3~~~~~-4~.O~~~~~~oka
4r-~~--~-r--r-~~--~-r~r-~~--~-r--r-~~--~-r-.4
.
,If"~
/
.,
I , , "
(
'\::--~/
I
31- Ir··.
-
3~---~\ ~
...
~
s
s
.1a
_ 2s
I
-
_ 1 ~5.0 -4.5 -4.0 -3.5 _3·0 -25 -2.0 -1.5 -1.0 .05 0 0.5 1.0 1.5 2.0 25 3.0 3.5 4.0 45 5.0° ka Fig. 6b5.A
vs
k plot for m 1 andS
10.00
o
•
•
• •
•
{3
=1.00 m=o ka=o•
A8
{3=1.25
m=o ka=o•
Aa
•
'0.
"
a e 10Aa
Fig. 7. Stability diagrams for
m~k~ 0 and S~ 1.00,1.25,1.50 .tl.a s
.
.•,
...
..
,
'.0
L._-''-_-' _ _ -'-_ _ ... _ _ ..L.._--I .. 0 '.0,
..
3.03.' {3
'.0Fig. 8. A vs S plot for m ~ k ~ 0 cr
Appendix A
If 6a(a) at the plasma boundary originally has three components in
the r-, 0- and z-directions, we perform a transformation to a system (A,oa)
indicated by bars, such that
6a
e
(a) -6a
z
(a) -
O.
According to Eq. (CI4) of
the Appendix C:
and
6An(a) -
-B(a)6a (a)
u
z
r
6A
(a)z
6K
(a) -0 ,
r -B(a)6a (a)
z
r
(AI)
6K
(a) -Bn(a)6a
(a) ,(A2)
Z u r
where the equilibrium relation
B(a) -
0has been used.
r
The transformed
Ii.
differs from the original
Aby the gradient of a
single-valued function ¢ which is chosen such that
6K
(a) -
O.
Therefore
r
6A (a)
r
ill.
8r
6A (a) -
6K
(a) - ik6¢
Z Z (A3)
Using Eqs (AI) and (A2) in Eq. (A3b,c) we find
im 6¢
+
B
(a)6a (a) - B (a)6a
(a)a
z
r
z
r
(A4)
These two equations can be solved for 6a
r
(a) and 6¢ and thus determine
the transformation
oa
~5a, oA
~ SA that was sought. The conclusion is thatin the main text we can take
6a
e
(a) -
6a
z
(a) -
0, 6A
r
(a) - 0, although in
the two volumes 6A
~O.
Appendix B
For further reference we give a listing of the equilibrium quantities
in the plasma and in the vacuum, and derive some useful relations between these quantities. In the case of a current 1* on the plasma surface, we haveB (a)
vz
B
pz (a)+
r*
0
B
.(a) -B
.(a)+
r*
Va pO' Z
(Bl)
We solve the Euler equations VxB
p
-
AB , VxB -
p v 0,using Eqs (Bl) and
(9),and the boundary conditions ApO(O)
- 0,A (b) - O. We obtain in the plasma
vz B o
J,(Ar) ,
B
- B
J(Ar) , A
pz 0 0 pzB
o (J (Ar) - J (Aa»
+
a (B J,(Aa)
+
r*)gn)?,
A 0 0 0 Z a
(B2)
and in the vacuumB
J (Aa)+l~B
-
~
(B J,(Aa)
+
r*)
vO
r
0 Z o 0 "aB
oAr
J,(h) ,A
- alB J,(Aa)
+r*)in b
vz 0
z
r
(B3)
From the continuity relation (8) it follows that
2B J
(Aa)r~+ 2B J,(Aa)r*
+
r~2+ r*2 - 0 ,
o
0 u 0 Z u Z (B4)so that r* has only one independent component. The number of invariants
matches the number of independent parameters: the
~
(Eqs (3)-(6»
determine the parameters B , A,
£
invariants K,
~Pt' ~vt'vp 0 a
*
i~
, and rO (or r;).
By
combining the Euler equationswith
r-differentiationsof the
A"