See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/234899988

Stabilization of sausage and kink instability modes of a plasma pinch by radial

oscillations

Article  in  Physics of Plasmas · March 1995

DOI: 10.1063/1.871463

CITATIONS READS

8 842

3 authors, including:

Michael A. Liberman
Nordic Institute for Theoretical Physics
413 PUBLICATIONS   4,497 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Bose-Einstein condensation of excitons in nano- and microstructures under the influence of the electromagnetic fields View project

Effect of particle clustering on radiative transfer in turbulent flows and mechanism of dust explosions View project

All content following this page was uploaded by Michael A. Liberman on 05 June 2015.

The user has requested enhancement of the downloaded file.

 Stabilization of sausage and kink instability modes of a plasma pinch by radial
oscillations
A. B. Bud’ko, Yu. P. Kravchenko, and M. A. Liberman

Citation: Physics of Plasmas 2, 792 (1995); doi: 10.1063/1.871463

View online: http://dx.doi.org/10.1063/1.871463
View Table of Contents: http://scitation.aip.org/content/aip/journal/pop/2/3?ver=pdfcov
Published by the AIP Publishing

Articles you may be interested in

Coupling of sausage, kink, and magneto-Rayleigh-Taylor instabilities in a cylindrical liner
Phys. Plasmas 22, 032706 (2015); 10.1063/1.4915520

The effect of cooling on sausage and kink oscillation modes in expanding coronal loops
AIP Conf. Proc. 1564, 194 (2013); 10.1063/1.4832817

Simulation of high-energy particle production through sausage and kink instabilities in pinched plasma
discharges
Phys. Plasmas 13, 082106 (2006); 10.1063/1.2245577

Kink instabilities in the belt pinch

Phys. Fluids 17, 447 (1974); 10.1063/1.1694735

Stabilization of Sausage Instability in Semiconductor Plasma by Axial Magnetic Field

Appl. Phys. Lett. 19, 439 (1971); 10.1063/1.1653762

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
130.238.7.40 On: Fri, 05 Jun 2015 11:11:23
 Stabilization of sausage and kink instability modes of a plasma pinch
by radial oscillations
A. B. Bud'ko, Yu. P. Kravchenko, and M. A. Liberman
Department of Physics, Uppsala University, Box 530, S-751 21, Uppsala, Sweden and P. Kapitsa Institute
for Physical Problems, 117334, Moscow, Russia
(Received 30 August 1994; accepted 27 October 1994)
The growth of the global sausage (m =0) and kink (m = 1) perturbations of a Z-pinch subject to
radial oscillations is considered. It is demonstrated that the oscillations result in significant reduction
of the growth rate of both kink and sausage instability modes with wavelengths long compared to
the pinch radius. The analysis of stability is carried out in two ways. The first method is based on
the averaging magnetohydrodynamic equations over the period of radial oscillations. The second
one consists in the analysis of the growth of Fourier-components of perturbations. Numerical
simulation demonstrates that even moderate radial oscillations cause reduction of the growth rate of
long-wavelength sausage instabilities and complete stabilization of long kinks. This can be
understood as a result of the effective gravitational field produced in the pinch by the oscillations.
The effect in question can explain the anomalous stability of pinches with respect to the kink
perturbations observed in experiments. © 1995 American Institute of Physics.

I. INTRODUCTION includes the drift-kinetic effects and the influence of resonant

particles, demonstrated significant decrease of the growth
The comprehensive analytical and numerical analysis rates of these perturbations. However, neither these models
developed in the recent decades and originating from the nor the analysis including the effect of viscous damping26
earlier studies of the magnetic confinement fusion predicts could explain the observed stability of long-wavelength kink
very poor hydrodynamic stability properties of a stationary modes.
Z-pinch. I - 3 Ideal magnetohydrodynamics gives the value of Thus, the conclusion has been made that the stabilization
the typical growth time of the most dangerous sausage of the long-wavelength kink modes in Z-pinches can be theo-
(m = 0) and kink (m = 0) modes of the order of the Alfven retically understood with the aid of the perturbation analysis
transit time. These predictions decreased the interest to of nonlinear regime which has been carried out only numeri-
Z-pinches as a possible fusion reactor for years. cally under a very simplified assumption of skin current. 27
Nevertheless, a number of experiments performed dur- Another possibility involves the idea of dynamic stabili-
ing the last decade have demonstrated sufficiently improved zation and was discussed for the first time by Osovets. 28
stability of Z-pinch configuration. The lifetimes of the Using simple estimates he demonstrated that the kink modes
pinches produced in experiments with fiber-initiated dense can be effectively stabilized by an external high-frequency
Z-pinches,4-6 compressionaf· 8 and gas-embedded magnetic field. A mechanical analog of this effect is a pen-
Z-pinches,9-11 with imploding gas-puff Z-pinches I2 - J7 and dulum whose point of support executes high-frequency os-
the straight Extrap configuration l8 •19 were of the order of cillations with a small amplitUde which makes the vertically
I0 1-10 2 TA (TA is the Alfven transit time) which indicates upward position stable.29 The dynamic stabilization of a
that Z-pinch systems are capable of greater stability and can Z-pinch by a high-frequency multipole magnetic field was
provide much higher density of energy and time of plasma confirmed experimentally.30.31
confinement than was expected from earlier investigations. Although in the recent experiments with stabilized
The striking stability with respect to the sausage modes Z-pinch configuration there was no high-frequency external
observed in the experiments with compressional and gas- field, we use the idea of dynamic stabilization in a different
embedded Z-pinches can be explained, in principle, by the way to explain the suppression of the long-wavelength kink
ideal magnetohydrodynamic (MHO) theory20 with the as- modes. We assume that the pinch is not in a stationary but in
sumption that after the peak current plasma is rapidly cooled a dynamic equilibrium, namely, that it is subject to radial
and that the unperturbed profile of pressure slowly decreases oscillations.
with radius satisfying the Kadomtsev condition, I as well as Usually Z-pinch is considered as a plasma column hav-
by finite plasma conductivity effects 21 concerning fiber- ing a constant radius. However one should remember that the
initiated pinches. pinch is created by heating and ionization of an initially cold
The kink mode cannot be stabilized by the appropriate column of gas or solid and this process is inevitably con-
choice of unperturbed profiles within the scope of the ideal nected with transitional phenomena. In other words, the
MHO model of linear stability. The stability of short- pinch should be treated as a substantially dynamical system.
wavelength (compared to the pinch radius) kink modes can The duration of the pulse of current which is used for the
be explained if we take into account the finite ion Larmor production of Z-pinch is not infinitely large in comparison
radius. Such investigations carried out with the aid of the with the characteristic time scale of the pinch plasma dynam-
Hall MHO mode1 22•23 and the Vlasov equation,24.25 which ics. Therefore oscillations of the plasma created by the dis-

792 Phys. Plasmas 2 (3), March 1995 1070-664X195/2(3)1792111 /$6.00 © 1995 American Institute of Physics
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
130.238.7.40 On: Fri, 05 Jun 2015 11:11:23
 charge should persist during the experiment although their cally symmetric, so all the functions depend only on rand t
effect on global thermodynamic characteristics may be be- and the only nonzero component of velocity is the radial
low the sensitivity of measurements. component U r' We introduce the self-similar form of unper-
It is generally believed that plasma flow very soon turbed profiles
(-10\ TA) becomes self-similar. The evolution of the plasma
radius depends upon the profile of the current pulse, and a ur(r,t) = Roa(t)~, (5)
constant radius corresponds to a certain profile shape, i.e., to p(r,t) = poa(t) -2N( g), (6)
a unique point in the appropriate functional space. Of course,
in real experiments current never behaves exactly in such a p( r,t) =poa(t) -2 y p( g), (7)
way. Therefore one should expect initiating of self-similar
oscillations, whose frequency is about T A' For example, for B<p(r,t) = B <poa(t) -\ H <p( g), (8)
pinches described in Ref. 5 it is - 10 GHz and is comparable B z(r,t) =B zoa(t) -2 Hz< g), (9)
to the inverse characteristic time of discharge.
The present analysis demonstrates that such oscillations where ~= rl R(t) is the self-similar radius, R(t) is the time-
of a Z-pinch suppress the growth of the global kink modes dependent characteristic radius of the pinch, aCt)
which can be understood as an effect of the intrinsic effective = R(t)/ R o , Ro = R(t = 0); B <pO, B zO, Po, and Po are normal-
gravitational field. The evolution of the perturbations is not ization constants, and N (/;), p (g), H <p(~)' and H z(~) are
exponential in this case and can be treated only by approxi- time-independent dimensionless functions describing self-
mate methods. We develop two approaches to the problem. similar profiles of flow.
One of them consists in averaging the equations, describing The self-similar profiles satisfy the following conditions
the development of instabilities, over the period of oscilla-
tions assuming that the growth time of instabilities is longer (10)
than this period. The second method uses no averaging and is
based on the Fourier analysis of initial linearized equations demonstrating that one of the functions can be chosen arbi-
of MHD. It allows us to reduce the initial system of partial trarily. The corresponding equation of motion is
differential equations to a system of ordinary differential
equations and to perform stability analysis of the reduced 2.. a \-2'" b a -3+ a -\ =0,
TAa-jJa 1- (11)
system numerically.
The plan of the paper is as follows. In Sec. II, we derive
where the parameters {3 = /LoPo 1B;0
and b = B;o/ B;o are
characteristics of kinetic pressure and axial magnetic field,
basic equations describing the development of instabilities in
respectively. The requirement for a= 1 to be the position of
an oscillating pinch within the frame of a self-similar model.
equilibrium in the absence of oscillations leads to the equal-
In Sec. III, we present the method of averaging over a period
ity b + (3= 1. The time unit is the Alfven transit time
of oscillations and discuss obtained results. The formulation
of the method of expansion into series and comparison of
TA = ~ /LoPoRo I B<po. The integral of (ll) yields the instanta-
neous oscillation velocity
results obtained by both techniques are given in Sec. IV.
Conclusions are presented in Sec. V.
r: a2=~(a2-2Y_
A y_ 1 max
a 2- 2y ) +b(a- 2 -
max
a- 2) -2In~
a max '
II. LINEARIZED EQUATIONS FOR PERTURBATIONS
(12)
The dynamic eqUilibrium of a screw pinch subject to
radial oscillations is described by the self-similar solutions, where a max is the maximal radius of the oscillating pinch
obtained by Kulikovskii 32,33 and Felber. 34 Here we outline normalized to the equilibrium value. The minimal radius
the main steps of the procedure. amin is the second root of the equation a
= 0 and can be
We start from the basic system of ideal MHD equations easily obtained numerically as well as the period T of oscil-
(in SI units) lations

ap
at + V·(pu)=O, (1) T=
Io
T
dt=
famax

amiD
da
--
a(a)'
(13)

aB where the positive root of Eq. (12) must be taken. In the limit
- -Vx(uxB)=O (2)
at ' of small amplitude of oscillations the value of T is
TA7rji(y-l+b(2-y»-1/2, in the nonlinear regime val-
ap
at +(u·V)p+ yp(V·u)=O, (3) ues of amin and a max versus T are plotted in Fig. 1 for
y=5/3.
Small perturbations of the eqUilibrium of the dynamic
p(au +(u'V)u) +Vp-
at
~(VXB)XB=O.
/Lo
(4) pinch are described by the linearized equations of the ideal
MHD
Here p is the mass density, p, u, and y are pressure, mass
velocity, and the ratio of specific heats, and B is the magnetic
field. In case of a Z-pinch the unperturbed flow is cylindri-
a;' +V'(p'u)=-V'(pu'), (14)

Phys.
This article is Plasmas, Vol.
copyrighted as2,indicated
No.3, March
in the 1995 Bud'ko, Kravchenko, and Liberman
article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. 793 to IP:
Downloaded
130.238.7.40 On: Fri, 05 Jun 2015 11:11:23
 same self-similar radius g and time. It is convenient to
2.0
choose T A as the unit of time. After substituting functions
(18)-(25) and unperturbed profiles (5)-(9) into Eqs. (14)-
(17) we can directly integrate Eqs. (14)-(16) over time and
1.5
express the perturbations of density, pressure, and magnetic
field via the displacement
----- 1>=0.5
(J. 1.0 ,, -1>=0.0 I a im
N'=-g ag(gNXr?-yNX<p-iKNXz, (26)
"-

0.5 a..
mm

(27)
4.5 5.0 5.5 6.0 6.5 7.0
T/'tA (28)

FIG. I. Interrelation between the period of oscillations T. expressed in the (29)

units of Alfven transit time 7 A, and minimal and maximal normalized radii
amin and a ma,,, respectively. for a pure Z-pinch (b=O) and a pinch with
applied axial field (b=0.5). The ratio of specific heats is y=5/3.
(30)

For convenience we have introduced here the normalized

aB' wave number K=kRo and function Hz(g)= Ji;Hz(g). Equa-
- -VX(uxB')=Vx(u'XB), (15) tion (17), which cannot be integrated directly, takes the form
at

a'
2
a Xr a dXr 1 [a f3 ap'
g - N' + -:::fi-
~ +(U·V)p'+yp'(V·u)=-(u'.V)p-yp(V·u'), (16) ---:;:z+2- - = - -
at a at N a a 'Y ag

au' ) i (m
-~ gHcp+KHz H;
_)
P ( Tt+(u' ·V)u+(u·V)u'

au ) I I
+~ a 2 I a - '}
I
ag(g HcpHcp) + a4 ag(HzH z ) ,
=-p' ( -+(u·V)u -Vp'+-[(VXB)XB'
at P-o
(31)
+ (VxB') xB], (17)
where primes denote the perturbed variables.
We seek for the solution of Eqs. (14)-(17) in the form
p'(r,t)=poa(t)-2N'(g,t), (18)
(32)
p'(r,r)=poa(t)- 2 'Yp'(g,t), (19)

B ;(r,t) = B <f!Oa(t) -1 H;( g,t), (20)

B~(r,t) =B cpoa(t) -1 H~(I;,t), (21)

B; (r,t) = B <f!Oa(t) -2 H; (g,t), (22) im H<f!H Z'} •

+ a2g (33)

aXr(g,t) Since this system explicitly depends on time, the evolution

u;(r,t)=Roa(t) at (23)
of perturbations is not exponential and can hardly be traced
analytically in general. Analytical treatment is possible in
I aXcp(g,t)
ucp(r,t)=Roa(t) at ' (24) two limiting cases. The first one corresponds to the case
where the growth time of perturbations is much less than the
, _ aXz(g,t) oscillation time scale. Such a situation has been considered
uz(r,t)-R o at . (25) in detail in Refs. 35 and 36. In the present paper we study the
opposite limit, namely, we assume that the growth time of
The dependence of these variables on coordinates 'P and z is long-wavelength sausage and kink perturbations is consider-
given by the factor exp(im'P+ ikz) and is omitted below. We ably more than the period of oscillations. Under this assump-
have expressed the perturbation of velocity via the displace- tion the analysis of stability can be performed with the aid of
ment vector X. Functions N, P, H, and X depend on the approximate methods described in the following sections.

794 Phys. Plasmas, Vol. 2, No.3, March 1995 Bud'ko, Kravchenko, and Liberman
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
130.238.7.40 On: Fri, 05 Jun 2015 11:11:23
 III. AVERAGING OVER THE PERIOD OF Now we assume that the perturbations grow exponen-
OSCILLATIONS tially with increment ()", i.e., their dependence on time is
A. Reduced equation for the radial displacement given by the factor exp«(}"t). Then the problem can be for-
mulated in the form
If the period T of oscillations is less than the growth
time of perturbations so that their ratio is a small parameter (40)
then the system (31)-(33) can be treated by means of the where F is a self-adjoint linear differential operator of the
method of averaging developed by Bogolyubov and summa- second order. Together with appropriate boundary condi-
rized in Ref. 37. We restrict out analysis with the zeroth- tions, which we discuss later, Eq. (40) constitutes an eigen-
order approximation and consider the system resulting from value problem with the real eigenvalues (}"2.
averaging Eqs. (31)-(33) over time: The assumed exponential form of the perturbations
makes it possible to solve Eqs. (38) and (39) with respect to
a2Xr a
-::-
N-..-=-i:.-N'-a-
--2- ap' --2 ( m -) ,
'Yj3-+a- i -H +KH H Xcp and Xz and rewrite (37) in the form
at k !:. a ag g cp Z r
2
d Xr dXr
G2~+Gl dg +GoXr=O. (41)

Functions G 2 , G 1, and Go are given by Eqs. (A17)-(AI9).

B. Boundary conditions
--=2. - --:::rim - ,
,
+a lKHzHcp-a THzHz' (35) The azimuthal and axial components of displacement
Xcp and Xz depend on Xr in a unique fashion [Eqs. (A15),
a2 X- - - -dH -
z (A16)] so boundary conditions should be imposed only on
N a/=-a 2- 2 'Y i j3KP'+a- 2 dg H;-iKHcpH~ Xr' The behavior of the radial component of perturbation
near the axis depends upon m and obeys the relations
--im
+a -2 T
H cp H'z' (36) Xr(O)=O, m=O,
(42)
where the operation of averaging is designated as dXr
d[(0)=0, m=1.

1= T1 IT0 f(t)dt. The boundary condition on the edge of the pinch column
corresponds to the interface between plasma and vacuum and
Terms in the left-hand sides of Eqs. (31) and (32) containing stems from the requirement for pressure and magnetic field
a vanish after the averaging. to be continuous on the displaced surface. In terms of the
The system (34)-(36) differs from that for a stationary physical radial displacement ~ r of the perturbed plasma this
pinch by presence of the additional term in (34) containing condition can be written as
cd a. It can be easily seen that it represents effective gravi-
tational field emerging in the plasma reference frame due to
the oscillations. As will be proved below (see Sec. III E.),
suppression of instabilities occurs exactly because of this
term.
Perturbations of pressure, density, and magnetic field are
given by Eqs. (26)-(30). Substituting them into (34)-(36)
we obtain where Re is the instantaneous pinch radius. 36 The displace-
a2 Xr aXr2 aXr . axcp
ment ~r can be expressed via self-similar functions as
N arr=Brr2 ag2 +B rr ! a[+BrroXr+1BrCPl ag
~r(t)=~r(O)+ J:U;dt=RoaXr It-R OJ:aXrdt
+ iBrcpoXcp+ iBrzl aa~z +iBrzOXz' (37)
=Roa(t)x,Ct). (43)
2
a Xcp. aXr. a
The integral in (43) containing vanishes due to the oscil-
N ---arr = zBcpr! a[ + lBcprOXr+ BcpcpoXcp + BcpzoXz' lating character of a. Therefore in self-similar variables the
(38) boundary condition on the edge takes the form

1 aX r ) m 2+a 2K2 Km(aK)

a2X~. aXr. (
N arr = IBzr! a[ + lBzrOXr+ BzcpoX",+ BzzoXz' (39) Xr ag g=l aK K~(aK)

where coefficients B are given by Eqs. (Al)-(AI4) in the

(44)
Appendix.

This article is copyrighted

Phys. as indicated
Plasmas, Vol. 2, No.3, in the article.
March 1995 Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions.
Bud'ko. Kravchenko. and Liberman Downloaded
795
to IP:
130.238.7.40 On: Fri, 05 Jun 2015 11:11:23
 (J

FIG. 2. The increment of sausage instability (m = 0) for the pure Z-pinch

(b=O) as a function of the normalized wave number K and maximal nor-
malized radius of the oscillating pinch a ma••

where the right-hand part for convenience is designated as

Q(a). After averaging we obtain (J

axr) g=1 =QXr(l)·

(a[ - (45)

If Eq. (41) has a singular point at one of the boundaries (it

happens for m = 0 at ~ = 1, when plasma density tends to
zero) then the last relationship is replaced by the requirement
(b)
for Xr to be regular at this boundary.
Equation (41) together with the corresponding boundary FIG. 3. The increment of sausage instability (m =0) for a pinch with axial
conditions (42) and (45) was solved numerically for the magnetic field versus the normalized wave number K and maximal normal-
cases of a stationary (a max = amin= 1) and oscillating pinch ized radius of the oscillating pinch a max for different values of the axial
for various amplitudes of oscillations. We used the following field: (a) b=(B,oIB",o)2=0.3 and (b) b=OA.
shape of self-similar profiles:
.N(~)=I-~2, P(~)=!(I_~2)2,
As one can see from the pictures the contraction of the
(46) spectral band occurs due to the suppression of short-
Hrp(~)=g,J!-te, Hz(g)=,J!g4_g2+!.
wavelength instabilities, while the growth rates of long-
Along with the pure Z-pinch (b=O) we have investigated wavelength perturbations still have nonzero, though substan-
pinches with applied axial magnetic field for values of b tially reduced values. A conclusion can be made that the
varying from 0.01 to 0.9. The obtained results are discussed long-wavelength sausage instabilities cannot be completely
below. stabilized by the oscillations but in the presence of axial
magnetic field their characteristic growth time can become
longer by several fold.
C. Sausage mode
The influence of oscillations on the long-wavelength
D. Kink mode
sausage instabilities exhibits itself in reduction of instability
increments. This effect is more pronounced for larger values For the kink mode the behavior of dispersion curves
of increment. For a pure Z-pinch (b = 0) slowly growing changes dramatically. Already small radial oscillations com-
long-wavelength instabilities remain almost unaffected by pletely suppress the growth of long-wavelength kink pertur-
the oscillations as shown in Fig. 2. The situation is changing bations. This is illustrated in Fig. 4 where we can see the
with emerging of axial magnetic field. As is known, suffi- dependence of instability increment versus normalized wave
ciently strong axial magnetic field makes a pinch absolutely number and amplitude of oscillations for the pure Z-pinch.
stable with respect to the sausage perturbations. I For the cho- The fiat site is the area where kinks do not grow. Cutoff
sen profiles (46) it happens when b=0.45. It turns out that if wave numbers as a function of the swing of oscillations are
the axial field is sufficiently close to this critical value then plotted in Fig. 5.
along with decreasing of the maximal increment of instabili- The situation becomes even more remarkable in the
ties the oscillations reduce the spectral width of the unstable presence of axial magnetic field. It is known that for a sta-
region as well. This effect is demonstrated on Fig. 3(a) for tionary pinch such a field suppresses the development of
b = 0.3 and Fig. 3(b) for b = 0.4. short-wavelength kink instabilities but fails to stabilize long

796 Phys. Plasmas, Vol. 2, No.3, March 1995 Bud'ko, Kravchenko, and Liberman
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
130.238.7.40 On: Fri, 05 Jun 2015 11:11:23
 (a)
FIG. 4. The growth rate 0' of the kink instability for the pure Z-pinch as a
function of the normalized wave number K and the maximal amplitude of
oscillations a max • The flat site is the region where the kink mode is com-
pletely suppressed by the oscillations.
(J

kinks. I Radial oscillations, so to say, act from the opposite

side of the spectrum, thus resulting in complete stabilization
of the kink mode for all wave numbers. This effect is shown
in Fig. 6. It should be mentioned that in this case the ampli-
tude of oscillations necessary for the complete stabilization
.0
of kink perturbations for all wave numbers becomes quite
small as can be seen from Fig. 7, which shows the depen-
dence of the "amplitude of complete stabilization" on the (b)
parameter b.
FIG. 6. Spectra of growth rates of kink instability for a pinch with axial
magnetic field for different values of the field: (a) b=O.4 and (b) b=O.7.
E. The influence of the effective gravity
It is interesting to investigate the inner physical cause of
the suppression of instabilities. As was already mentioned its real value. The instability increments for an oscillating
above it is the effective gravity in the moving plasma refer- pinch obtained in this way almost completely coincide with
ence frame that results in the suppression. This may be easily those for the stationary one, and there is no sign of any
proved by the following method. stabilization. This is illustrated by Fig. 8 showing results
Let us put "by hand" the gravity to zero. It means that obtained by both methods. Dispersion curves for the station-
while computing coefficients B rrl' B rrO~'PO' and B rzO ary pinch and for the oscillating pinch with "zero gravity"
[Eqs. (A2), (A3), (A5), and (A7)] we take ii/a=O instead of are indistinguishable.

1.5
O.S

1.4
0.4

1.3
0.3
r;ja
~1 0.2
1.2

1.1
0.1

1.0
0.0 0.4 0.5 0.6 0.7 0.8 0.9
1.0 1.1 1.2 1.3 1.4 1.5 b
(Xmax

FIG. 7. The maximal amplitude of oscillations astab providing stability

FIG. S. The cutoff wave numbers of m= 1 mode versus the amplitude of against kinks for all wave numbers as a function of the parameter b char-
oscillations for the pure Z-pinch (b=O). acterizing axial magnetic field.

Phys.isPlasmas,
This article Vol.as
copyrighted 2, indicated
No.3, March
in the1995 Bud'ko, Kravchenko, and LibermanDownloaded
article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. 797 to IP:
130.238.7.40 On: Fri, 05 Jun 2015 11:11:23
 where Yi= y;(x,t), i=O, ... ,n-l. We use the approximate
0.50
method of solution of (47) based on the representation of the
nonsteady functions Yi at each instant of time by a finite
0.40 series of given linearly independent functions g k with time-
dependent coefficients A :
0.30
m-l
cr Yi(X,t) = 2: Aik(t)gk(X). (48)
0.20 k~O
- - -€Xmax=l.00

••••• €X =1.30 It is a well-known and thoroughly investigated method of

max
0.10
--"zero gravity" solution widely used in numerical computations. The choice
of the functions gk is not strictly limited as well as the
0.00 method of searching for the time-dependent coefficients. Af-
0.00 0.20 0.40 0.60 0.80 1.00 ter substituting the expansion (48) into Eq. (47) and fitting
the coefficients A we get a system of ordinary differential
equations of the order n X m
FIG. 8. Dispersion curves corresponding to the stationary pinch
(a ma,= 1.00, dashed line) and oscillating pinch. a max = 1.30, with "gravity" d
dtA=ZA. (49)
term (dotted line) and without it (solid line),

As it can be seen from (48) the growth rate of y is bounded

F. On the accuracy of the method from above by the growth rate of coefficients A. Therefore
We consider it necessary to discuss an important issue instead of stability analysis of the initial system (47) we may
concerning the approximate method of averaging used in this investigate the stability of (49), which gives us an upper
analysis, We expand initial equations into series in powers of estimation for the growth rate of y.
a small parameter representing the ratio of the oscillation If Z in (49) is a constant matrix then the analysis of
period to the characteristic growth time of instabilities. This stability is trivial. If Z is periodic with period T, which is the
parameter should be much less than unity, which can be ex- case for the problem in question, then we can use the known
pressed in terms of the increment as (T~0.2 (see Fig. 1). fact 3& that any solution of (49) can be represented in the form
Thus for regions with (T';3 0.2 the obtained results can some-
A(t) =X( t)A(O).
what deviate from real values of increment. But we are
mainly interested in the suppression of instabilities, in which
The matrizant XU) can be expressed as
case (T~O.2 and the procedure provides quite reliable results.
In particular, conclusions concerning complete stabilization XU) = F(t)exp(tK),
of kink instabilities «T= 0) are very well justified.
Due to the usage of two independent techniques we can where F(T+ t) = FU) is a periodic matrix and K is a con-
compare results obtained by the method of averaging with stant matrix, in accordance with the Floquet-Lyapunov theo-
those obtained by the analysis of series which is presented in rem. The exponent M = exp(TK) is the matrix of the mono-
the next section. It turns out that even for (T~ 0.5 and rather dromy operator of Eq. (49). To determine M we should
large amplitude of oscillations results obtained by both meth- m X n times integrate (49) from 0 to T with linearly indepen-
ods correlate within several percent despite the fact that the dent initial conditions A(O) and compose X(T) of columns
method of averaging is restricted to zeroth-order approxima- A (T). The most natural way is to choose A (0) as columns of
tion. an identity matrix. Yet due to the presence of boundary con-
ditions it may be impossible and initial conditions A(O) will
IV. THE METHOD OF EXPANSION INTO SERIES have other forms. In this case the matrix M is determined as
M=X(T)X-1(0). Logarithms of the eigenvalues rti of M
A. Basic formulation
divided by the period T serve as exact analogs of frequency
Another approximate method applicable for the analysis and growth rate eigenvalues. If we denote
of nonstationary partial differential equations is based on the
reduction of the equations to a system of ordinary differential 1 .
equations and on subsequent analysis of the obtained system. (T= -In ( max Irt;1 ) , (50)
T O~i~mn-l '
The suitability of the technique to an MHD stability problem
was demonstrated in Ref. 35. The main stages of the proce- then for any instant of time the norm of any solution of (49)
dure are the following. satisfies inequality
Let us consider a linear system of partial differential
equations IA (t)1 ~const· exp( (Tt) IA( 0) I, (51)

By; By) and the right-hand side of (51) is the exact estimation of the
at = Vij ax + WijYj, (47)
growth of A (t) on the time interval t.

798 Phys. Plasmas, Vol. 2, No.3, March 1995 Bud'ko, Kravchenko, and Liberman
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
130.238.7.40 On: Fri, 05 Jun 2015 11:11:23
 B. The case of the oscillating pinch It has the canonical form (47). perturbations N' •...• V; being
We apply the method described above to investigate the the functions Yi. i = 0 •... ,7.
Computations in Ref. 35 were performed with Cheby-
stability of the system (14)-(17). As well as in the previous
shev polynomials; in the present analysis we use trigonomet-
method we seek for a self-similar solution. Perturbations of
ric series
density. pressure. and magnetic field are still being given by
Eqs. (18)-(22) but instead of introducing displacement X we I
write perturbations of velocity as Yi(g,t)=AiO(t)+ ~ Aij(t)sin(7Tjg)
j=!
u; (r,t) = Ro V;( g,t), (52)
21
u~(r,t) = Roa(t) V~( g,t), (53)
+ ~ Aij(t)COS(7T(j-l)g). (63)
u~ (r,t) = Ro V~ (g,t). (54) j=l+!

In this case Eqs. (14)-(16) cannot be explicitly integrated Fitting the coefficients A is performed by the method of col-
over time because their right-hand sides involve the time- location points, that is, we require Eq. (47) to be valid at a
dependent parameter a. As a result the evolution of pertur- given set of points g;, i =0, ... ,21, which are chosen on the
bations is now being governed by the following system of interval ]0,1[. It turns out that the results weakly depend on
equations: the particular choice of {O provided that the outer points are
sufficiently close to the bounds of the interval.
The number of terms in expansion (63) takes little effect
on final results. Values of (T obtained for 1=4 and 1= 8 differ
within 2 %. The results presented in the paper were obtained
(55)
with /=4.

C. Boundary conditions
-i'YKPV~ , (56) The transformation of boundary conditions to the finite-
dimensional approximation (48) requires special attention. A
condition imposed on a certain function Yq creates an addi-
tional relationship between functions Aqk and hereby reduces
the required number of collocation points by one. So for this
function we exclude from the consideration an arbitrary
point of collocation and require Eq. (47) to be valid for Yq
(58) only at m - 1 remaining points, replacing the lacking relation
by the imposed boundary condition. The removed point is
aH~ 1 _ av; 1 (dH z Hz) , im _ ,
-=--H - -- - + - V --H V usually the nearest to the bound in question if the condition
at a z ag a dg g , g z cp
is imposed on the value of Yq or any inner point if it involves
aYq/ax.
(59) Each boundary condition causes a linear relationship be-
tween elements of matrizant and makes it impossible to find
the reverse matrix X-! (0). In order to avoid it we had to
exclude certain rows and columns of the matrizant from con-
sideration. This operation cannot lead to loss of results be-
cause due to the linear relationship excluded elements cannot
grow faster than other summands.

I a 2 , 1 a - ,]
+ ag2 ag(g H~cp)+ ~ ag(HzH z) , (60)
D. Limitations of the method

av~ a,
1 [ im{3 , I d(gHcp) , As can be seen from Eq. (48) the method of expansion
Tt=--;;VCP+N-a2rgP+a2g dg H, into series is applicable only if functions Yi have "good be-
havior," i.e., they do not have singularities within or at the
iK _ , im _ ,] bounds of the interval of definition of Eq. (47) on the x-axis.
+ ~ HzHcp- a 4 g Hz Hz • (61) Otherwise one has to determine asymptotic behavior of y
near the singular points and allow for the singularities by
av; _ 1[ i{3K , 1 dH z , . , introducing normalizing multipliers in the expansion (48).
Tt- N - a 2r - 2 P + ~ dg H,-IKHcpHcp The self-similar profiles (46) result in singularity on the edge
of the pinch because the plasma density, which is involved

+~;gH~;]. (62)
reciprocally in Eqs. (60)-(62), tends to zero when g-.l. In
order to avoid the complicated procedure of normalization

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
Phys. Plasmas, Vol. 2, No.3, March 1995 Bud'ko, Kravchenko, and Liberman 799
130.238.7.40 On: Fri, 05 Jun 2015 11:11:23
 1.00
... - .....
---
1.00 ...
l-<:
0.80 _. - --'K'=10.0
,.
I

- - -'K'=30.0
.'

J.,
.~
0.60
I
I
I
I

I
I
0.40 , I
I I
0.10 --method of averaging I I
~ 0.20
I
•• - - • expansion into series /
/

0.00
0.1 10 0.0 0.2 0.4 0.6 0.8 1.0
Normalized pinch radius ~

FIG. 9. Spectra of increment of sausage instability in the absence of oscil- FIG. 10. Eigenfunctions of m == 0 instability mode of a stationary pinch
lations obtained by the method of averaging (solid curve) and the method of corresponding to different values of normalized wave number.
expansion into series (dashed curve).

rate those given by the method of averaging and, in accor-

we have performed computations for a pure Z-pinch with the dance with the previous discussion, serve as estimates from
following shape of profiles, which do not lead to singulari- above.
ties:
V. CONCLUSIONS
(64)

The growth rates of sausage instabilities were computed both In this article, we have presented a possible theoretical
by the method of averaging and the method described in this mechanism explaining anomalous stability of Z-pinches with
section. The results are discussed below. respect to long-wavelength kink perturbations observed in
experiments. It is shown that radial oscillations of a pinch
completely suppress the growth of the long kinks. For the
case of a pinch with axial magnetic field oscillations can
E. The results of numerical simulation stabilize the m == 1 mode for all wave numbers and signifi-
We start from the discussion of results obtained in the cantly reduce the rate of growth of sausage instabilities. For
absence of oscillations. Dispersion curves computed by both the m == 0 mode the oscillations tend to shrink the unstable
methods are presented in Fig. 9. For long-wavelength pertur-
bations (K"';' 1) results are almost identical and relative error
is
(T series - (T aver
--"-----"-< 1.4%. 0.30 1C=1.0
(T aver

The shorter the wavelength the more the discordance; for

K == 30 it reaches 40%. For any K the estimation of increment 0.25
--method of averaging
(Tseries obtained by the method of expansion into series is cr
• - - - - expansion into series
higher than the value (T aver obtained by solution of the eigen- 0.20
value problem in accordance with the theory presented in ------- .. - ------- ... _----
Sec. IV A: (Tseries is an upper estimation while (Taver in the K=O.5
0.1S
absence of oscillations is the exact value. The tendency of
the error to increase with growth of K can be understood if
we consider the shape of eigenfunctions plotted in Fig. 10 for 0.10 L....L....L...LJ......I....L..J..-1....1.....L....L..L..JL.L.L...L.l..-I-..l.....L.L..J....L..L....L....L...J...JL.L.L...L.J......I....L..J

different values of K. For K== 1.0 the eigenfunction can be 1.00 1.05 1.1 0 1.15 1.20 1.25 I. 30 1.35
accurately represented by a finite sum of sines and cosines
while for K = 30 errors of such a representation would be
very high.
Figure 11 shows dependence of the growth rate of sau- FIG. 11. Increment of sausage instability of the oscillating pinch as a func-
tion of the maximal radius a max of oscillations for two different values of
sage perturbations on the maximal radius of oscillating pinch wave number: K == 0.5 and K = 1.0. Solid curves represent results obtained by
a max for K= 0.5 and K== 1.0. As we can see, results obtained the method of averaging while dashed curves correspond to the method of
by the method of expansion into series completely corrobo- expansion into series.

800 Phys. Plasmas, Vol. 2, No.3, March 1995 Bud'ko, Kravchenko, and Liberman
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
130.238.7.40 On: Fri, 05 Jun 2015 11:11:23
 region from the short-wavelength side of the spectrum while APPENDIX: COEFFICIENTS OF AVERAGED
for the kink mode both short- and long-wavelength perturba- EQUATIONS
tions become stable.
Upon substituting explicit relations (26)-(30) into the
The treatment of a plasma pinch as a column whose averaged system (34)-(36) with subsequent differentiation
radius is slowly and monotonically varying with time is only we finally get Eqs. (37)-(39) describing the displacement X
a crude approximation to the extremely complicated dynam- of the plasma. If we introduce auxiliary functions
ics of the system. At the initial stage of heating the non-
equilibrium state of the plasma inevitably results in excita- P Hep
W=-~ , J =~- ,
tion of oscillations which can be roughly treated as having
cylindrical symmetry. We do not discuss here the evolution then the coefficients B may be written as:
of such oscillations and consider a very simplified self-
similar model which can serve as a first approximation to the B rr2 = a- 2 'Y{3yP+ a- 2H 2ep + a- 4iI z2 ' (Al)
real picture. It is shown that such a correction of the theo-
retical model results in completely new effects, namely, in
total stabilization of long-wavelength kink instabilities and
reduction of the growth rate of m = 0 perturbations. Until
now no one theory based on the "classical" presentation of (A2)
plasma pinch as a cylinder with slowly varying radius could
explain this phenomenon of stabilization observed in experi-
ments.
The idea of dynamic stabilization of plasma is not a new
one but in previous experimental and theoretical works only
stabilization by external fields was considered. The principal
novelty of the present work resides in that we consider no
external stabilizing field and show that free oscillations of
plasma near its eqUilibrium configuration dramatically im-
prove the stability of the system with respect to global per-
(A3)
turbations. This may be treated as a result of effective gravi-
tational field created by the oscillations.
As mentioned above, for the case considered in the paper
the free oscillations of the plasma may be initiated during the
(A4)
initial stage of heating and no additional build-up is neces-
sary. It seems quite plausible that such a situation indeed
takes place in experiments with fiber-initiated pinches5
where no kink instability has been observed.
Results of the performed analysis show a new possible
method of suppression of global plasma instabilities. It con-
sists in deliberate initiation of oscillations by external fields
with resonant frequency. Due to the resonant character of the (A5)
oscillations the power necessary for their build-up and sus-
taining is expected to be small in comparison with previously (A6)
proposed schemes of dynamic stabilization.
Another important result obtained in the present work is
the development of the approximate method of stability
analysis of MHD equations based on expansion of solutions
into series of functions with time-dependent coefficients apd (A7)
subsequent analysis of obtained ordinary differential equa-
tions. It is shown that the method yields a very accurate
estimation of stability properties provided that corresponding
eigenfunctions are sufficiently close to their finite-
dimensional approximations by the series. This technique as
well as the method of averaging, whose suitability to the
problem is successfully demonstrated, may be of great inter-
est for the further development of ideas presented in the pa-
(A9)
per.

This article
Phys.is copyrighted as2,
Plasmas, Vol. indicated in the article.
No.3, March 1995 Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions.
Bud'ko, Kravchenko, and Liberman Downloaded to IP:
801
130.238.7.40 On: Fri, 05 Jun 2015 11:11:23
 (AlO) 4D. W. Scudder, Bull. Am. Phys. Soc. 30, 1408 (1985).
5J. Hammel and D. W. Scudder, in Proceedings of the XIV European Con-
(All) ference on Controlled Fusion and Plasma Physics. Madrid, 1987. edited
by F. Engelmann and J. L. Alvarez Rivaz (European Physical Society,
Petit-Laney, 1987), p. 450.
oJ. D. Sethian, A. E. Robson, K. A. Gerber, and A. W. DeSilva, Phys.
Rev. Lett. 59, 892 (1987).
7M. G. Haines, J. Phys. D 11. 1709 (1978).
8J. M. Bayley, P. Baldock, and A. E. Dangor, in Proceedings of the Sec-
(AI2) ond International Conference on Dense Z-pinches, Laguna Beach, CA.
1989, edited by N. R. Pereira. J. Davis, and N. Rostoker (American In-
stitute of Physics, New York, 1989), p. 481.
(A 13) 9 A. E. Dangor, M. B. Favre Dominguez. S. Lee, and E. Kahan, Phys.
Rev. A 27,2751 (1983).
Bzzo= - a 2- 2'Yf3K2yp- K2H; - a- 2m 2J 2. (AI4) 10J. E. Hammel, D. W. Scudder. and J. S. Schlachter, Nucl. lnstrum.
Methods 207, 161 (1983).
If perturbations grow exponentially then Eqs. (38) and (39)
"M. Skowronek, P. Pomeas. J. Larour, and A. E. Dangor, in Proceedings
can be solved with respect to the variables X if' and Xz : of the Second International Conference on Dense Z-pinches, Laguna
Beach, CA. 1989, edited by N. R. Pereira. J. Davis. and N. Rostoker
·C dXr ·C (AlS) (American Institute of Physics. New York, 1989), p. 491.
Xif'=1 11'1 dl; +1 ipOXn 12J. Shiloh, A. Fisher, and N. Rostoker, Phys. Rev. Lett. 40,515 (1978).
I3J. Shiloh, A. Fisher, and E. Bar-Avraham. App!. Phys. Lett. 35. 390
(1979).
·C dXr +·C (AI6)
Xz = 1 z I d I; 1 zOX" 14M. K. Matzen, R. J. Dukart, B. A. Hammel, D. L. Hanson,
W. W. Hsing, T. W. Hussey, E. J. Maguire, M. A. Palmer, and R. B.
where the coefficients C are Spielman, J. Phys. (Paris) 47. c6-135 (1986).
ISR. J. Dukart, T. W. Hussey, and N. F. Roderick, in Megagauss Technol-
ogy and Pulsed Power Applications. edited by C. M. Fowler,
R. S. Caird, and D. J. Erickson (Plenum, New York, 1987), p. 183.
16F. S. Felber, M. M. Malley, F. J. Wessel, M. K. Matzen. M. A.
Palmer, R. B. Spielman, M. A. Liberman,. and A. L. Yelikovich, Phys.
Fluids 31, 2053 (1988).
17F. S. Felber, F. J. Wessel. N. C. Wild, H. U. Rahman, A. Fisher,
C. M. Fowler, M. A. Liberman, and A. L. Velikovich, J. App!. Phys.
69, 3831 (I988).
18B. Lehnert, in Unconventional Approaches to Fusion, edited by
B. Brunelli and G. G. Leotta (Plenum, New York. 1982), p. 135.
19J. R. Drake, Plasma Phys. Controlled Fusion 26, 387 (1984).
_ I 2 20M. Coppins. I. D. CulverweIl, and M. G. Haines, Phys. Fluids 31, 2688
C zo - D[B ZlpoB iprO + (NO' - B ipipo)B zro ], (1988).
211. D. Culverwell and M. Coppins, Phys. Fluids B 2, 129 (1990).
D = (N0'2 - B iplpo)(N 0'2 - B zzo) - B ipZoB ZipO' 22G. O. Spies and M. Faghihi, Phys. Fluids 30. 1724 (1987).
23
0. Agren, Plasma Phys. Controlled Fusion 30, 249 (1988).
Coefficients G in Eq. (41) have the following form: 24 0. Agren, Plasma Phys. Controlled Fusion 31, 35 (1989).
2sH. O. Akerstedt, J. Plasma Phys. 41, 45 (1989).
G2=Brr2-BripICipl -BrzICzI, (AI7) 26 0. Eriksson, Phys. Ser. 35, 851 (1987).
27H. O. Eriksson and C. Wahlberg, J. Plasma Phys. 43, 397 (1990):
dCipl 28S. M. Osovets, SOy. Phys. JETP 12, 221 (1960).
GI=Brrl-Brip1 ~ -Brip1ClpO-BrlpOCIpI 29L. D. Landau and E. M. Lifshitz, Mechanics (Pergamon, New York,
1976), p. 95.
dC z1 30S. M. Osovets and U. I. Sinitsyn, SOy. Phys. JETP 21.715 (1965).
IIp. R. Forman, A. Haberstich, H. J. Karr, J. A. Phillips, and A. E.
-B r:;1 ~- BrzlCzO- BrzOCz1, (Al8)
Schofield, Phys. Fluids 14,684 (1971).
32A. G. Kulikovskii, Dok!. Akad. Nauk SSSR 114,984 (1957).
dClpo dC zo 33 A. O. Kulikovskii, Dok!. Akad. Nauk SSSR 120, 984 (J 958).
Go=BrrO-Brlp1 ~ -BrIpOCipO-Brzl ~ 34F. S. Felber, Phys. Fluids 25. 643 (1982).
35 A. B. Bud'ko, F. S. Felber, A. I. Kleev, M. A. Liberman, and
(AI9) A. L. Velikovich, Phys. Fluids B 1, 598 (1989).
- B rzoCzO- N0'2.
36 A. B. Bud'ko, M. A. Liberman, F. S. Felber. and A. L. Yelikovich.
Phys. Fluids B 2, 1159 (1990).
lB. B. Kadomtsev, in Reviews of Plasma Physics, edited by M. A. Leon- 37 A. I. Morozov and L. S. Solov'ev, in Reviews of Plasma Physics, edited
tovich (Consultants Bureau. New York, 1966), Yol. 2, p. 153. by M. A. Leontovich (Consultants Bureau, New York, 1966), Yol. 2,
2G. Bateman, MHD Instabilities (MIT Press, Cambridge, MA, 1978), p. p. 273.
102. 38y. A. Yakubovich and V. M. Starzhinskii. Linear Differential Equations
3J. P. Freidberg, Rev. Mod. Phys. 54,801 (1982). with Periodic Coefficients (Halsted. New York, 1975), Vol. I, p. 90.

802 Phys. Plasmas, Vol. 2, No.3, March 1995 Bud'ko,. Kravchenko, and Liberman
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
130.238.7.40 On: Fri, 05 Jun 2015 11:11:23
View publication stats