investigation of elms on alcator c-mod€¦ · antenna limiter is plotted below for each frame. ......
TRANSCRIPT
Sequence of tree 10 msec exposures take 1 msec apart
The fluctuation/striation lifetime is much shorter than the 1 msec frame time. The intensity variation along the edge antenna limiter is plotted below for each frame.
Striations/fluctuations as seen in a 2 ms exposure- there is a slight narrowing of thecharacteristic polloidal size
Investigation of ELMs on Alcator C-ModJ.L Terry, A.E. Hubbard, I. Cziegler, J.A. Snipes, J.W. Hughes, M. Greenwald, B. LaBombard, Y.Lin, P. Phillips*,P.B. Snyder+, L. Sugiyama, S.M. Wolfe, S. Wukitch
Plasma Science and Fusion Center, MIT, Cambridge, MA, *Univ. of Texas at Austin - Fusion Research Center, +General Atomics
Motivation and Introductionl Understanding ELMs is of crucial importance both because of they limit the pedestal pressure gradient and because of danger they pose to the first wall and divertor l Large, discrete ELMs are typically NOT present in C-Mod discharges - EDA H-mode with the "Quasi-Coherent Mode" as the pedestal relaxation mechanism is typical
l Recently, a new region of operational space has been accessed where discrete, relatively large ELMs are common
l We present here results of an investigation of those ELMs: - "Global" characteristics: plasma parameters, ELM energetics, edge stability - ELM dynamics: precursor oscillation, pedestal/SOL perturbation, filament ejection, filament propagation, relative timing of ELM perturbations
1050
6280
14 @
0.9
s
typical
time (s)0 0.2 0.4 0.6 0.8 1.0 1.2 1.4
HITER98(y,2) factor
Teo (keV)
GPC
(R=0
.67)
- 10
6050
3022
MHD
- 10
6050
3022
2
1
0
neo (1020/m3)TS
(R=0
.68)
- 10
6050
3022
6
4
2
0
1.5
1.0
0.5
0- these discharges exhibit other interesting transport properties, see M. Greenwald in C-Mod Oral Session Tues. afternoon
- Teo > 4 keV
ne typically low with neo~2x1020 m-3
non-typical shape for C-Moddlower > 0.75, dupper~0.15
- Confinement can be good in these ELMing H-modes (HITER98(y,2) >1) ~
Prf (MW)
6
4
20
Da
Core Plasma Characteristics
Edge Operational Space for Pedestal Relaxation in "Non-typical" Shape
- ELMs occur at relatively low n* (0.2<n*<1)
- discrete ELMs, variable size, probablyType I
- steady EDA, QC mode, no ELMs
- small ELMs with EDA, at high power & pressure
for q95~3.5, dlower>0.7 Zeff taken to be 1
1000
800
600
400
200
0
T epe
d (eV)
neped (1020 m-3)
1 20 3
ELM Pedestal Energeticsl Energy loss per ELM typically 10-25% of Wped
-100 0 100 200t-tcrash (ms)
0.42
0.46
0.50 Te (keV)
1060
5030
22:E
CE
ch 6
t c=0
.843
49s
12 ELMavg
ped
l Pedestal is not destroyed by ELM
DW
ELM
/Wpe
d
C-Mod (w strong shaping)
l 10% loss in Tpedestal
l nepedestal loss is less certain,
but in range 0 < Dne < 15%
0.01 0.1 1 100.00
0.05
0.10
0.15
0.20
0.25
n* ITER
n* (neoclassical)
}JET
JT-60UDIII-DASDEX-U
}l Consistent with multi-machine observations for Type I ELMs
from A. Loarte, et al.,JNM 313-316 (2003)
1050
6280
14 @
0.9
s
nozzle
Edge TS
Bdot
Bdot
ELM Dynamics Investigated usingMagnetic-pickup-coils, Da Gas-Puff-Imaging, and ECE
views for GPI
Toroidal view of Outboard Edge
limiter
Top Viewgas puff
Bdot coils
l GPI localizes emission toroidally- primarily sensitive to changes in ne, with some sensitivity to changes in Te
Bdot coilsvery closeto plasma
Fine Time-scale ELM Evolution
end with ~10% of Wped lost
H-mode pedestal close to stability boundary
peeling/ballooningmode growth
"filaments"ejected
dBq/dt
-100 0 10088
89
90
91
92
Maj
or R
adiu
s [c
m]
Shot
105
0628
018;
to=9
1608
2 m
s
0-50 50t-tfilament start (ms)
radial filament ejection
t-tfilament start (ms)
steady precursor growth
fil. ejectionsteady precursor growth fil. ejection
ELM Precursor
GPI Da Emissionfrom ped.
0-100 50
1060
5030
22t=
0.93
4113
s;1
TABK
- frequencies: 200-400 kHz, slowing to 70-100 kHz
- toroidal mode number is around 10, constant during growth phase
- precursor oscillation begins growing - often with large growth rate
- osc. seen on outboard GPI emission with radial location in and at top of pedestal, radial structure seen
hi-pass-filtered dBq/dt
0.55
0.60
0.65
Teped (keV)
time at whichfilaments are first seen in outboard SOL
- at filament ejection, a higher freq. perturbation (0.5-1 MHz) is seen, the "hi-freq mag. osc." - use as t=0 from now on
dBq/dt
2
8
-100 -50 0 50
-100 -50 0 50
Da monitor
-100 -50 0 50
-100 -50 0 50
1050
6230
33t=
0.78
4215
s;v
iew
2
Dt from onset of "hi-freq. mag. osc." pert. (ms)
ELM Dynamics in Radial Dimension - GPI Da emission
- much more complex time history with multiple ejections apparent from the radial array of views- "primary" ejection occurs after ped. is perturbed
time
(ms)
sepx
limite
rsh
adow
r (cm) 1050628014: t=0 is t=0.918943 s
-1 0 1 2 3
070
0tim
e (m
s)
2 6
Damonitor
2 frames from (6ms/fr) movie
-50
0
100
200
1 cm
time ave.subtracted
1050628019 frame 229
Radial
Verti
cal
during filament ejection(12 ms after frame on left)
before filamentejection
primary
- "primary" propagates rapidly outwardwith radial velocities ~0.5 to >6 km/s
time
(ms)
sepx
r (cm)-1 0 1 2 3
VR=3.2 km/sradial velocity (km/s)
# of
ELM
prim
arie
s in
ve
loci
ty in
terv
al
0 1 2 3 4 5 6 70
10
20
30
40
time of "primary" ejectiononset of outboard perturbation
time
(ms)
2 6
Damonitor
1050628014: t=0 is t=0.918943 s
070
0
-50
0
100
200
- multiple "secondaries" occur after "primary"- typically slower than "primary"
time
(ms)
sepx
limite
rsh
adow
r (cm)-1 0 1 2 3
# of
ELM
s
# of secondaries0 1 2 3 4 5 6 7
0
10
20
30
l likely that not all secondaries are seen in limited view
time
(ms)
2 6
Damonitor
1050628014: t=0 is t=0.918943 s
070
0
-50
0
100
200
large case - aMHD=8.8 - n >15 unstable
dPdr
note: no modes on inboard side
ped
dPdr
pedmedium case - aMHD=3.9 - all n stable
ELITE: both cases unstable ton=6,12, 20 modes
mode structurefrom ELITE
aMHD=8.8 caseped
M3D:
prelim. conclusions: - ELMing pedestals are close to and sometimes exceed peeling/ballooning instability boundary, - for unstable case, modes are predominantly ballooning with weaker peeling component - including previous ELITE analyses of equilibria of typical C-Mod shape, these atypical shapes are less stable for similar pedestal profiles
Pedestal Stabilty Analysis using MHD codesELITE (ideal MHD) and M3D (extended MHD)
press. pert. (n=12)
*see L. Sugiyama(Thurs. morn.)
2 atypically-shaped EFIT equilibria analyzed (meas. ped. press. profiles used)
0 100
dBq/dt
hi-pass-filtereddBq/dtf=0,Z=0.1m
1050
6280
14;t0
=0.9
9784
5 s
rel.
Four
ier a
mp.
30
20
10
0
freq. (MHz)0 0.2 0.4 0.6 0.8 1.0 1.2
1050
6280
14; t=
0.997
85 s
Freq. spectrum of hi-pass-filtered dBq/dt sig.
The "Hi-Frequency Magnetic Oscillation"
l Onset of "hi-freq. mag. osc." nearly coincident with onset of filament ejection
l Onset of "hi-freq. mag. osc." is nearly simultaneous on coils separated toroidally and poloidally - with systematic differences of up to ~10 ms
time difference relativeto onset of "hi-freq. mag. osc."
-60 -40 -20 0Dt (ms)
0.0
0.2
0.4
0.6
0.8
1.0
norm
aliz
ed P
DF
(# o
f ELM
prim
arie
s) onset of filaments
10-10-30-50
Dt from onset of "hi-freq. mag. osc." pert. (ms)
ECE Observations during ELM
0.850.900.951.00
1060
5030
22: t
o=96
2252
ms
-1000
100200
Dt f
rom
ons
et o
f "h
i-fre
q. m
ag. o
sc."
(ms)
norm
. cha
nge
in s
igna
l
-100 -50 0 50 100Dt from onset of "hi-freq. mag. osc." pert. (ms)
steady precursor growth
filament ejection
0
50
100
-50
-100 -2 0 2 4
R-Rsep (cm)
Edge ECEemission
1060503022: to=962252 ms
l Onsets of Te pert. and "hi-freq. mag. osc." are nearly coincidentl Non-thermal emission present during filament ejection since effective Trad > Te
fil
ped
ped
dBq/dt
ECE eff.Trad
Effe
ctiv
e T r
ad (k
eV)
far SOL
Dt from onset of "hi-freq. mag. osc." (ms)
near SOLmid-ped.
steady precursorgrow
thfil. ejectionTe (mid-ped.)
1060
5030
22 to
=0.9
7684
6 s
-100 -50 0 50 1000.0
0.5
1.0
1.5
2.0
2.5
3.0
Observations at Inboard Edge
l No mode observed on inboard edge
l No filaments observed in inboard SOLl However, inboard edge is perturbed l onset of pert. occurs at or during precursor growth phase, before outboard filament ejection
l perturbation propagates into the inboard SOL and appears to be a relaxation
1060
7190
16: to
=861
856
ms
-100 0 100 200
chan
ge in
GPI
Da
em
issi
on d
ue to
ELM
8
04
03
0
1.0
0
1.0
0
1.5
0
r=-0.47cm
r=-0.15cm
r=0.15cm
r=0.50cm
r=0.85cm
r=1.19cm
Dt from onset of "hi-freq. mag. osc." pert. (ms)
steady precursor growth
filament ejectioninner wall
Time-Sequence of ELM Pertubations
l Onset of (outboard) filament ejection: ~coincident
Dt (ms)
norm
aliz
ed P
DF
(# o
f ELM
prim
arie
s)
l t=0 taken to be onset of "hi-freq. magnetic oscillation"l Onset of rapid growth of magnetic precursor amplitude: ~40 ms before l Onset of density (GPI) perturbation on outboard edge: ~10-20 ms before l Onset of density (GPI) perturbation on inboard edge: ~10 ms before l Onset of (outboard) Te pertubation: ~coincident
-60 -40 -20 0 200.0
0.2
0.4
0.6
0.8
1.0
filaments
inboard pert.outboard pert.precursor growth
Te pert.
Speculations about ELM Filament Ejection
* acknowledgement to A. Boozer - Columbia Univ. and V. Naulin - JET for helpful discussions; see also W. Fundamenski, et al. submitted to PPCF (2007)
l ECE sometimes shows "filaments" of non-thermal emission with effective Trad >1 keV, a possible result of MR causing parallel electron acceleration l non-thermal emission follows the onset of "hi-freq. magnetic oscillation", with the oscillation showing a broadly peaked 0.5-1 MHz spect.
Evidence for ejectionas electrostatic ExB convection
Possible Evidence for Magnetic Reconnection (MR)
l Similar to "blob" radial convection (likely elect. stat. ExB)l Vrad/Cs~1%l Observe poloidal variation in outboard filament ejectionl ELITE & M3D simulations show unstable modes with poloidal variation within flux surface
- MR near X-point causes Alfven wavepacket to bounce back and forth between high-shear X-pt regions, i.e.
along field line of length=2pRq95
bounce freq=VAlfven/2pRq95 =0.5 MHz for C-Mod
ELM Evolution Paradigm - Summary
end with ~10% of Wped lost
H-mode pedestal close to stability boundary
peeling/ballooningmode growth
"filaments"ejected
l Complex structure of filaments observed in single ELM eventl "Primary" filaments propagate radially thru outboard SOL with Vr of 0.5-7 km/s; radial size ~1 cm; poloidal size >4.5 cml Multiple "secondary" filaments follow l "Primary" ejection coincident with 0.5-1MHz hi-freq magnetic oscillation & onset of Teped decrease l Non-thermal ECE emission observed in outboard SOL coincident with "primary"
l Precursor oscillation, n~10, cntr-Ip rotation, localized in outboard ped., grows rapidly before filament ejectionl Rapid mode growth precedes filament ejection by ~40 msl No mode evident at inboard SOL - yet inboard SOL responds to perturbation before ejectionl Modeling with ELITE and M3D yield unstable modes for n >15 (ELITE), for n=6,12, 20 (M3D) Implication: non-typical shape unfavorable for ped stability
Summary and Conclusionsl Discrete, relatively large ELMs occur reproducibly at low n* in an non-typical C-Mod shape (dlower>0.7,dupper<0.2)l DWELM/Wped=10-25% - consistent with multi-machine scaling - but pedestal is not destroyedl Both inboard and outboard pedestals are perturbed before "primary" filament ejection l Timing of "primary" ejection and the 0.5-1 MHz "hi-freq. mag.
oscillation", as well as the non-thermal Trad observations, suggest
that magnetic reconnection may be occurring in ejection process
l ELITE MHD stability code shows closeness of pedestal profiles to peeling/ballooning instability boundary - for higher gradient ped. profiles, modes with n>15 are unstable and predominantly ballooning with weaker peeling component