scieee Science in your language
[en] (orig)

Exponentially growing tearing modes in Rijnhuizen Tokamak Project plasmas

Abstract

The local measurement of the island width w, around the resonant surface, allowed a direct test of the extended Rutherford model [P. H. Rutherford, PPPL Report-2277 (1985)], describing the evolution of radiation-induced tearing modes prior to disruptions of tokamak plasmas. It is found that this model accounts very well for the observed exponential growth and supports radiation losses as being the main driving mechanism. The model implies that the effective perpendicular electron heat conductivity in the island is smaller than the global one. Comparison of the local measurements of w with the magnetic perturbed field showed that w1/2 was valid for widths up to 18% of the minor radius.

Read accessible full text

Exponentially growing tearing modes in Rijnhuizen Tokamak Project plasmas

Author: F. Salzedas,F. C. Schuller,A. A. M. Oomens,RTP Team
Year: 2002
DOI: 10.1103/PhysRevLett.88.075002
Source: https://repositorio-aberto.up.pt/bitstream/10216/23182/2/60795.pdf
VOLUME 88, NUMBER 7 PHYSICAL REVIEW LETTERS 18F
EBRUARY 2002
Exponen ially G owing Tea ing Modes in Rijnhuizen Tokamak P ojec Plasmas
F. Salzedas,* F. C. Schülle , A. A. M. Oomens, and he RTP Team
FOM-Ins i uu oo Plasma ysica “Rijnhuizen,” Associa ion Eu a om-FOM, T ila e al Eu egio Clus e ,
P.O. Box 1207, 3430 BE Nieuwegein, The Ne he lands
(Recei ed 14 Ma ch 2001; published 5 Feb ua y 2002)
The local measu emen o he island wid h w, a ound he esonan su ace, allowed a di ec es o
he ex ended Ru he o d model [P. H. Ru he o d, PPPL Repo -2277 (1985)], desc ibing he e olu ion
o adia ion-induced ea ing modes p io o dis up ions o okamak plasmas. I is ound ha his model
accoun s e y well o he obse ed exponen ial g ow h and suppo s adia ion losses as being he main
d i ing mechanism. The model implies ha he e ec i e pe pendicula elec on hea conduc i i y in he
island is smalle han he global one. Compa ison o he local measu emen s o wwi h he magne ic
pe u bed ield ˜
Bshowed ha w~˜
B1兾2was alid o wid hs up o 18% o he mino adius.
DOI: 10.1103/PhysRe Le .88.075002 PACS numbe s: 52.55.Tn, 52.35.Py
The iden i ica ion o he main d i ing mechanism o he
obse ed MHD (magne ohyd odynamics) ins abili ies in a
okamak plasma is o conside able in e es . Such knowl-
edge o he na u e o he ins abili y can be e y help ul
in he choice o a s abiliza ion p ocedu e needed o a oid
deg ada ion o con inemen o plasma dis up ion. Dis-
up ions a e he mos dange ous ins abili ies in okamak
plasmas, and he p ecu so o densi y limi dis up ions is
o en domina ed by an m兾n苷2兾1MHD ins abili y [1]
(whe e mand na e he poloidal and o oidal Fou ie mode
numbe s, espec i ely). In his Le e , we s udy he expo-
nen ial inc ease o an m兾n苷2兾1 ea ing mode obse ed
p io o a densi y limi dis up ion [see Fig. 1(a)] in RTP
(Rijnhuizen Tokamak P ojec , ci cula limi e , mino a-
dius a苷0.16 m and majo adius R0苷0.72 m). Simila
exponen ial g ow h is obse ed also in o he okamaks [2].
Compa ison wi h he ex ended Ru he o d [3] model sup-
po s adia i e ene gy losses as he main d i ing mecha-
nism o his ea ing mode. To a oid con usion wi h he
well-known neoclassical ea ing modes, we, hence o wa d
denomina e hese ins abili ies as adia i e induced ea ing
mode o RTM. The s udy o RTM is also e y impo an
because u u e la ge okamaks such as ITER a e expec ed
o wo k close o 100% edge adia ion o educe he mal
load on he di e o .
In he plasmas s udied he e, he plasma cu en IP苷
100 kA, he sa e y ac o qa艐4, he a io o he plasma
ene gy o he magne ic ield ene gy, b,was low, and a
he ime he RTM is obse ed collisionali y is high. So
neoclassical e ec s a e dis ega ded. The illing gas was
He and he densi y was amped up using Ne gas injec-
ion. Bo h noble gases a e no abso bed by he essel wall.
This allowed a be e and ep oducible con ol o he elec-
on densi y han wi h hyd ogenic plasmas. Mo eo e , he
densi y a dis up ion could be kep below he cu o den-
si y o he elec on cyclo on emission (ECE) he e odyne
adiome e , used o measu e he ime e olu ion o he elec-
on empe a u e a 20 adial posi ions along a ho izon al
cho d. Elec on empe a u e and densi y we e measu ed a
h ee ime poin s wi h a high spa ial esolu ion Thomson
sca e ing (TS) sys em. Saw ee h a e always obse ed in
hese discha ges [Fig. 1(b)].
One o he di icul ies in he analysis o he beha io o
ea ing modes is o ela e he measu ed pe u bed poloidal
magne ic ield ˜
Bu共 c兲[Fig. 1(a)], whe e cis he adius
o he pickup coils, wi h he unknown pe u bed adial
magne ic ield ˜
B 共 s兲a he mode esonan su ace, s, ha
is he quan i y whose dynamics is p edic ed by heo y [4].
The usual p ocedu e consis s [5,6] o app oxima ing he
o oidal plasma by a cylind ical one and hen o use he
FIG. 1. (a) Ampli ude o ˜
Bu. (b) ECE Tein he cen e (chan-
nel 10) and in he icini y o he q苷2su ace (channels 6
and 5) [see Fig. 2(a)]. (c) D 兾acalcula ed wi h (4). The
solid ci cles indica e he island wid h es ima ed om TS [see
Figs. 2(c)–2(e)]. The g ay line is he i o Eq. (10). (d) D 兾a
plo ed agains ˜
Bu. The dashed (solid) g ay line ep esen s
Eq. (6) [Eq. (5)]. Symbol 䉭(¶) e e s o LFS (HFS).
075002-1 0031-9007兾02兾88(7)兾075002(4)$20.00 © 2002 The Ame ican Physical Socie y 075002-1
VOLUME 88, NUMBER 7 PHYSICAL REVIEW LETTERS 18F
EBRUARY 2002
cu en p o ile as ound om one o he me hods desc ibed,
o example, in [7], o sol e he equa ion
d2c
d 211
dc
d 2µm2
1m0
djz0
d
Bu0共12n
mq兲∂c苷0, (1)
using ˜
Bu共 c兲as a bounda y condi ion, o ind ˜
B 共 s兲, whe e
cis a scala po en ial ha is ela ed o ˜
Bby
˜
B苷=c 3 ˆz苷21
≠c
≠u ˆ 1≠c
≠ ˆ
u.(2)
The zcomponen o he equilib ium plasma cu en is jz0,
and Bu0is he poloidal componen o he equilib ium mag-
ne ic ield.
In he case ha he plasma cu en ou side he esonan
su ace can be neglec ed and he wall ac s as a pe ec
conduc o , Eq. (1) has an explici solu ion. Then, using
(2), he mcomponen o ˜
Bua cis ela ed o he m
componen o ˜
B a sby
˜
Bum共 c兲苷˜
B m共 s兲µ s
c∂共m11兲11共 c
w兲2m
12共 s
w兲2m,(3)
whe e wis he adius o he essel wall.
In he me hods which use (1) and (3) wi h measu emen s
o ex e nal pickup coils, he mode e olu ion is desc ibed
ia he ampli ude o he pe u bed magne ic ield. In wha
ollows, ano he me hod o measu e he spa ial ampli ude
o he pe u ba ions, di ec ly and locally a ound he eso-
nan su ace, will be used. Basically, i elies on he as-
sump ion ha close o and ou side o he sepa a ix o a
ea ing mode he elec on empe a u e is a lux unc ion
[8]. In his case, a change Dwo he island wid h will
cause a p opo ional displacemen D o he lux su aces
in he neighbo hood o he sepa a ix. The displacemen
D is gi en by
D 苷
DT
dT0
d
,(4)
whe e T0is he empe a u e p o ile, measu ed adially
along he Xpoin and DTis he di e ence in empe a-
u e be ween he wo poin s a he wo angles ha co e-
spond wi h he Xpoin and he Opoin [9]. I is also
implici ly assumed ha he plasma is incomp essible. Fig-
u e 1(c) shows D 兾a, de i ed bo h om he low ield side
(LFS), channel 6 o he adiome e , and on he high ield
side (HFS), channel 16. The posi ion o hese channels is
shown in Fig. 2(a) and a pic o ial illus a ion o D 共LFS兲
is shown in Fig. 2( ). I is expec ed ha he smalle al-
ues o D ha e lowe accu acy since he island sepa a ix
OXO
OX O
m/n=2/1 mode
chn.6
∆
∆T
chn.
FIG. 2. (a) ECE Te共 , 兲iso he ms. The channels posi ion is indica ed a he igh . A ske ch o he m苷2mode sepa a ix is
illus a ed by he do ed lines a ound he g ay a eas. (b) TS Tep o iles a 1,
2, and 3. (c) Zoom in on (b). Fo cla i y, only one
ypical e o ba is shown. (d) and (e) a e he same as be o e o ne共 兲. The dashed line indica es he calcula ed adial posi ion o he
q苷2su ace. ( ) Illus a ion o (4). G ay (black) line, ECE Te共 兲passing h ough he O(X) poin . (g) j共 兲and q共 兲calcula ed
om Te共 1兲.
075002-2 075002-2
VOLUME 88, NUMBER 7 PHYSICAL REVIEW LETTERS 18F
EBRUARY 2002
is u he away om he ECE channels. On he o he ex-
eme, i he sepa a ix c osses he ECE channel hen he
de ini ion (4) is no applicable.
A cha ac e is ic o m苷2islands in okamaks is isible
in he ECE measu emen s, as shown in Fig. 2(a), namely
ha he de o ma ion o iso he ms is symme ic in ela-
ion o he esonan su ace on he HFS [see illus a ion
in Fig. 2(a)]. On he LFS i is mo e di icul o localize
he Xpoin , since he island g ows asymme ically in e-
spec o he esonan su ace, i.e., much mo e owa ds he
cen e o he plasma han o he edge. In his case i is ex-
pec ed ha on he LFS he s ongly asymme ic island leads
o D ⯝w, while on he HFS a symme ic island gi es
D ⯝w兾2. Figu e 1(c) con i ms his expec a ion wi h
D 共LFS兲⯝2D 共HFS兲. Mo eo e , he es ima ed island
wid h om he TS [10], measu ed along a e ical cho d,
comes e y close o he alues o D measu ed on he LFS
[see Figs. 1(c) and 2(c)–2(e)].
In Fig. 1(d) D is plo ed agains ˜
Bu共 c兲, o alues up o
18% o he mino adius, which a e measu ed jus 100 ms
be o e he onse o he dis up ion. Wi hou any assump ion
on he plasma equilib ium, he bes i s o he expe imen al
da a gi e he ollowing o he LFS and HFS:
D 共LFS兲苷共1.04 60.08兲p˜
Bu共 c兲,(5)
D 共HFS兲苷共0.56 60.05兲p˜
Bu共 c兲.(6)
We can compa e he expe imen al ela ion (5) wi h he
well-known exp ession o he island wid h ha ela es he
spa ial and magne ic mode ampli udes:
w苷4s sq
mBuq0p˜
B 共 s兲.(7)
Because o he adia i e con ac ion o he cu en p o-
ile, he cu en o . scan be neglec ed so we can use
(3) o ob ain ˜
B o m苷2. Using he alues om Table I
ha we e calcula ed om TS Te共 1兲[see Figs. 2(b) and
2(g)], he ollowing is ob ained:
w苷共1.14 60.25兲p˜
Bu共 c兲.(8)
The ag eemen be ween (8) and he expe imen ally de-
i ed (5) is e y good. This suppo s he calcula ion o w
in he cylind ical app oxima ion used un il now.
Expe imen al e idence indica es ha he equilib ium pa-
ame e s, Bu共 s兲, s, and q0共 s兲, do no change no ice-
ably in ime du ing he island exponen ial g ow h. Fo
s his can be seen unambiguously a he HFS in Fig. 2(a).
Rela i ely o Bu共 s兲and q0共 s兲, since he cu en di usion
ime is ypically 10 20 ms, i is e y unlikely ha hese
TABLE I. Equilib ium pa ame e s a 苷 scalcula ed om
Te共 1兲in Fig. 2(b), assuming jz~Te共 1兲3兾2.
sTejq
0Buh
(m) (eV) (MA m22)(m
21)(T) (1026Vm)
0.113 127 0.211 32.7 0.165 4.12
pa ame e s change signi ican ly du ing he 0.6 ms o he
mode exponen ial g ow h. So, assuming ha du ing his
pe iod he equilib ium pa ame e s do no show no iceable
changes, Fig. 1(d) p o ides a di ec demons a ion ha (7)
is s ill alid up o island sizes o he o de o 18% o he
mino adius. This ea u e should be emphasized because
he use in he li e a u e (e.g., [5,6]) o (7) o island sizes
o his magni ude is based only on he magne ic measu e-
men s pe o med ou side he plasma, which a e ex apo-
la ed inwa ds up o s.
Be o e he exponen ial g ow h s a s, i is obse ed ha ,
on a e age, he ampli ude o ˜
Bug ows algeb aically o
a pe iod o 艐20 ms. Du ing his pe iod [o which only
he las 4ms a e shown in Fig. 1(a)], he g ow h o he
mode is no mono onic bu shows an i egula modula ion
wi h he ampli ude now and hen dec easing p ac ically o
ze o. This modula ion is o en co ela ed wi h a saw oo h
c ash. This complex beha io will no be discussed he e.
Ins ead we will analyze only he exponen ial phase using
he ex ended Ru he o d model [3],
dw
d
苷C1
h
m0
D02C2
˜
PT
xisland
⬜e 具Te典w,(9)
whe e his he plasma esis i i y, C1苷1.22, and C2苷
0.9hjzq兾共Buq0兲wi h all he quan i ies aken a he eso-
nan su ace. ˜
PTis he o al powe densi y pe pa icle in
he island and he mode is des abilized i ˜
PTis nega i e.
xisland
⬜e and 具Te典a e he e ec i e pe pendicula elec on
he mal conduc i i y and he a e age elec on empe a u e
in he island, espec i ely.
F om 165.8 o 166.2 ms, he e olu ion o he ampli ude
o he mode magne ic ield could be well i ed wi h a
quad a ic inc ease wi h ime implying a linea inc ease in
island wid h. Du ing his pe iod, losses by adia ion inside
he island a e s ill no impo an , and so he second e m
o (9) is negligible. The alue o D0can hen be es ima ed
om he linea inc ease o he island wid h. Using he
alue o h om Table I, we ob ain D0苷2.
Be ween 166.2 and 166.8 ms, he inc ease in he mode
ampli ude is e y well desc ibed by an exponen ial i . This
can be explained by a sudden onse o he second e m o
(9) wi h ˜
PT,0. An ab up nega i e alue o ˜
PTis pe -
ec ly possible due o he na u e o adia i e powe losses.
This is an e ec o he in e se p opo ionali y be ween a-
dia ed powe densi y and he elec on empe a u e. I im-
pu i ies ha e been accumula ing inside he island, once he
empe a u e on he Opoin is low enough such ha adia-
i e ecombina ion o hese impu i ies can occu , a small
dec ease o empe a u e will lead o an inc ease in adia-
ion loss. Ene gy in he island is hen dissipa ed as e ,
se ing ab up ly ˜
PT,0. The hea lowing in o he island
can be enough o keep ˜
PTcons an , du ing he exponen ial
g ow h, as bo h ECE and TS indica e.
The high spa ial esolu ion empe a u e and densi y p o-
iles inside he island a e nei he la no mono onic, bu
i egula [Fig. 2(c)]. On bo h 1and 2p o iles, despi e he
075002-3 075002-3
VOLUME 88, NUMBER 7 PHYSICAL REVIEW LETTERS 18F
EBRUARY 2002
maxima and minima, in he island he a e age alue o Te
does no change. This indica es ha he hea low inside
he island is no pu ely di usi e. The ECE [Fig. 2(a)],
which has lowe spa ial esolu ion han TS and so does
no de ec he i egula i y ound by he TS p o iles, also
shows he island g owing wi hou signi ican changes in
具Te典. Seconda y maxima in Teand newe e also obse ed
in TEXTOR [11,12].
F om he da a o Fig. 1(c), i is possible o es ima e
he a io o ˜
PT兾xisland
⬜e . Assuming ha ˜
PTis cons an
du ing he sho ime in e al in which he mode g ows
exponen ially, in eg a ion o (9) gi es
w共 兲苷
a1
a2
1µw02a1
a2∂e2a2 ,(10)
whe e w0苷w共166.2兲苷0.02a,a1苷c1
hD0
m0
苷8, and
a2苷
0.9hjzq
Buq0具Te典
˜
PT
x⬜e
.(11)
The i ing o (10) o he da a o Fig. 1(c) (dashed line)
gi es a2苷22.3 3103s21. F om (11) and Table I i is
ound
˜
PT兾xisland
⬜e 苷21.0 3106eV m22.(12)
Bo h ˜
PTand xisland
⬜e a e unknown, bu i is possible o
es ima e i s ange o alues. A he ime he TS Te共 1兲
p o ile is measu ed, he ene gy con inemen ime is
e苷3.4 ms which gi es a global e ec i e pe pendicu-
la elec on he mal conduc i i y xglobal
⬜e 苷a2兾共4 e兲苷
1.9 m2s21.I xisland
⬜e was he same as xglobal
⬜e , hen
˜
PT苷21.9 3105eV s21pe pa icle which means ha
du ing he 0.6 ms o exponen ial g ow h an a e age o
1140 eV pe pa icle would ha e o be dissipa ed in he
island. Such a alue is e y la ge since du ing he same
pe iod he empe a u e in channel 6 dec eases 215 eV,
while in channel 5 i inc eases only 65 eV [Fig. 1(b)].
The di e ence o 150 eV be ween hese wo channels is
almos 1 o de o magni ude smalle han he p e iously
ound alue 1140 eV. Assuming ha only 150 eV we e
dissipa ed in he island, hen ˜
PT艐22.5 3105eV s21
pe pa icle (which co esponds o 艐270 kW). This
alue is in good ag eemen wi h he adia ion losses in he
ou e plasma laye s a compa able empe a u e alues and
is e y close o he minimum elec on cyclo on powe o
90 kW ha was ound necessa y o s abilize he m苷2
RTM in simila discha ges [9,13]). So om (12) i ollows
ha xisland
⬜e 艐0.25 m2s21.
This alue o he e ec i e elec on hea conduc i i y
inside he island is almos 1 o de o magni ude smalle
han he global e ec i e elec on hea conduc i i y o he
plasma. This educ ion in di usi i y inside he island is in
ag eemen wi h p e ious indica ions [11,12,14]. Mo eo e ,
i is co obo a ed by he obse a ion in TS Te共 1, 2兲o he
seconda y maxima in empe a u e and densi y inside he
island, ela i ely o he sepa a ix. The Temaxima a e
mo e p onounced in TS Te共 3兲. Despi e his p o ile was
measu ed immedia ely a e he exponen ial g ow h ˜
Bu共 c兲
does no indica e any s ong change in he island s uc u e.
Only la e a 苷167.2 ms, du ing he dis up ion ˜
Bu共 c兲
shows a sha p spike, indica ing he p obable des uc ion o
he island.
One o he au ho s (F. Salzedas), was suppo ed by he
Po uguese Founda ion o Science and Technology FCT
unde P og ama PRAXIS XXI-G an No. BD/4531/94.
This wo k was pe o med unde he Eu a om-FOM Asso-
cia ion ag eemen , wi h inancial suppo om he Du ch
esea ch o ganiza ion NWO and Eu a om.
*P esen add ess: Cen o de Fusão Nuclea , Associa ion
Eu a om-IST, A . Ro isco Pais, 1049-001 Lisboa,
Po ugal.
Elec onic add ess: [email p o ec ed]
[1] F. C. Schülle , Plasma Phys. Con olled Fusion 37,A135
(1995).
[2] M. Schi enhelm e al., Nucl. Fusion 37,1255 (1997).
[3] P. H. Ru he o d, PPPL Repo -2277 (1985).
[4] P. H. Ru he o d, Phys. Fluids 16,1903 (1973).
[5] Z. Chang e al., Phys. Re . Le . 74,4663 (1995).
[6] B. Ca e as e al., Nucl. Fusion 19,1423 (1979).
[7] H. Sol wisch, Plasma Phys. Con olled Fusion 34,1669
(1992).
[8] R. Fi zpa ick, Phys. Plasmas 2,825 (1995).
[9] F. Salzedas, Ph.D. hesis, Uni e si ei U ech , 2000 (URL:
h p://www.lib a y.uu.nl/digia chie /dip/diss/1940669/in-
houd.h m).
[10] TS has a e y high spa ial esolu ion bu only one poin
in ime is gi en while ECE has a high empo al bu low
spa ial esolu ion. Bo h e ec s cause some e o in he
loca ion o he sepa a ix.
[11] P. C. de V ies e al., Plasma Phys. Con olled Fusion 39,
439 (1997).
[12] P. C. de V ies e al., Nucl. Fusion 37,1641 (1997).
[13] F. Salzedas e al. ( o be published).
[14] B. P. an Milligen e al., Nucl. Fusion 33,1119 (1993).
075002-4 075002-4