INSTITUTE OF PHYSICS PUBLISHING PLASMA PHYSICS AND CONTROLLED FUSION
Plasma Phys. Control. Fusion 46 (2004) 551–572 PII: S0741-3335(04)70218-3
ELM characteristics in MAST
A Kirk
1
, G F Counsell
1
,HRWilson
1
, J-W Ahn
1
,RAkers
1
,ERArends
2
,
J Dowling
1
, R Martin
1
, H Meyer
1
, M Hole
3
, M Price
1
, P B Snyder
4
,
D Taylor
1
,MJWalsh
5
,YYang
6
and the MAST team
1
1
EURATOM/UKAEA Fusion Association, Culham Science Centre, Abingdon,
Oxon OX14 3DB, UK
2
FOM Institute for Plasma Physics ‘Rijnhuizen’, PO Box 1207, 3430 BE Nieuwegein,
The Netherlands
3
Applied and Plasma Physics, School of Physics, University of Sydney, NSW 2006, Australia
4
General Atomics, PO Box 85608, San Diego, CA 92186-5608, USA
5
Walsh Scientific Ltd, Culham Science Centre, Abingdon, Oxon OX14 3DB, UK
6
Institute of Plasma Physics, Hefei 230031, People’s Republic of China
Received 10 October 2003
Published 23 February 2004
Online at
stacks.iop.org/PPCF/46/551 (DOI: 10.1088/0741-3335/46/3/009)
Abstract
Edge localized mode (ELM) characteristics in a large spherical tokamak (ST)
with significant auxiliary heating are explored. High confinement is achieved in
mega ampere spherical tokamak (MAST) at low ELM frequencies even though
the ELMs exhibit many type III characteristics. These ELMs are associated
with a reduction in the pedestal density but no significant change in the pedestal
temperature or temperature profile, indicating that energy is convected from the
pedestal region into the scrape-off layer. Power to the targets during an ELM
arrives predominantly at the low field outboard side. ELM effluxes are observed
up to 20 cm from the plasma edge at the outboard mid-plane and are associated
with the radial motion of a feature at an average velocity of 0.75 km s
−1
. The
target balance observed in MAST is potentially rather favourable for the ST
since H-mode access is facilitated in a regime where ELM losses flow mostly
to the large wetted area, outboard targets and, in addition, the target heat loads
are reduced by an even distribution of power between the upper and lower
targets.
1. Introduction
An edge localized mode (ELM) is a violent repetitive fluctuation that appears near the edge of
a plasma in some improved confinement (H-mode) regimes in magnetically confined plasmas.
ELMs give rise to rapid transient increases in particle and energy fluxes to the divertor targets.
Divertor power handling in a spherical tokamak (ST) may be of even greater significance than
in conventional devices due to the reduced wetted area of the inboard strike-points due to the
small major radius. A good understanding of the effects of ELMs on target power loading is
0741-3335/04/030551+22$30.00 © 2004 IOP Publishing Ltd Printed in the UK 551
552 A Kirk et al
central to evaluating the potential of the ST as a future burning plasma device and can be used
to refine estimations for ITER.
Plasmas in the mega ampere spherical tokamak (MAST) typically have a major radius,
R ∼ 0.85 m and a minor radius a ∼ 0.65 m. MAST has operated with plasma current
up to I
p
∼ 1.35 MA and is equipped with two neutral beam lines, which have so far
provided deuterium injection into plasmas with total powers in excess of P
NBI
∼ 2.9MW
(and will be capable of up to 5 MW). All the discharges described in this paper are for
deuterium injection into deuterium plasmas. The toroidal field on axis is usually in the range
0.35 T <B
φ
< 0.55 T, and due to the low aspect ratio, the toroidal field varies between 1.7 T
at the inboard mid-plane and 0.25 T at the outboard. This strong variation across the plasma is
a distinct feature of the ST.
MAST is well equipped with diagnostics for the study of ELMs, including a 300-point
Thomson scattering system [1, 2], a fast reciprocating probe system equipped with a radial
array of triple probes [3] and arrays of high spatial and temporal resolution Langmuir probes
covering all four targets (3 mm spacing inboard, 10 mm spacing outboard) [4]. Figure 1 shows
a poloidal cross section of the MAST vessel with a typical equilibrium superimposed. The
locations of the main diagnostics used in this paper are indicated.
Preliminary results on ELMs were presented in [5] based on analysis of individual shots.
This paper expands on these preliminary results and presents a systematic study of ELM
Lower
outboard
target
Upper
outboard
target
Upper
inboard
target
Lower
inboard
target
Line of sight of
Thomson scattering system
Reciprocating probe
D Views
Outboard target
Inboard target
X-point
Scale (m)
Figure 1. Poloidal cross section of MAST with a typical equilibrium. Shown on the figure are the
locations of the targets, the mid-plane location of the reciprocating probes and the line of sight views
of the Thomson scattering system and Dα views of the X-point and upper inboard and outboard
targets.
ELM characteristics in MAST 553
characteristics as follows: section 2 describes the global characteristics of the ELMs observed
in MAST, including a discussion of their type. Section 3 describes the effect of ELMs on the
target and section 4 describes evidence for the radial extent of ELMs.
2. ELM characterization
Figure 2 shows the Dα emissions from the inboard and outboard targets as a function of time
for six discharges with the same shaping, plasma current and density for different values of
auxiliary heating power (P
NBI
). With increasing P
NBI
, the Dα emissions increase at each
ELM and the ELM frequency decreases. Figures 3(a) and (b) show the ELM frequency as
a function of line-averaged density and P
SOL
(defined as the sum of the Ohmic and auxiliary
power input to the core plasma less the rate of change of stored energy and radiated power),
0.0
2.
0
4.
0
0
1
2
3
4
5
0.180 0.190 0.200 0.210
0.0
2.
0
4.
0
0
1
2
3
4
5
0.0
2.
0
4.
0
0
1
2
3
4
5
0.0
2.
0
4.
0
0
1
2
3
4
5
0.0
2.
0
4.
0
0
1
2
3
4
5
0.0
2.
0
4.
0
0
1
2
3
4
5
0.18 0.19 0.20 0.21
0.
0
2.
0
4.
0
0
1
2
3
4
5
0.18
0 0.190 0.200 0.210
0.0
2.
0
4.
0
0
1
2
3
4
5
0.0
2.
0
4.
0
0
1
2
3
4
5
0.0
2.
0
4.
0
0
1
2
3
4
5
0.0
2.
0
4.
0
0
1
2
3
4
5
0.0
2.
0
4.
0
0
1
2
3
4
5
0.18 0.19 0.20 0.21
Figure 2. The Dα signal at the upper inboard and outboard targets showing the variation of ELM
frequency with auxiliary heating power at fixed plasma current and density.
Line-average density (10
18
m
-3
)
P
SOL
(kW)
ELM freq (Hz)
ELM freq (Hz)
(a) (b)
0 500 1000 1500 2000 250030 35 40 45 50 55
0
500
1000
1500
2000
2500
3000
0
500
1000
1500
2000
2500
3000
Figure 3. ELM frequency versus line-average density for P
SOL
in the range 1.0–1.5 MW and ELM
frequency versus P
SOL
for line-average density in the range (30–40) × 10
18
m
−3
.
554 A Kirk et al
0 750 1500 2250 3000
0.00
0.01
0.02
0.03
0.04
0.05
0 750 1500 2250 3000
0.00
0.01
0.02
0.03
0.04
0.05
W/W
n
e
/n
e
ELM freq (Hz) ELM freq (Hz)
(a) (b)
Figure 4. (a) The fraction of stored plasma energy released by an ELM (calculated using the
change in stored energy from EFIT). (b) The fractional change in the line-average density due to
an ELM.
respectively, using data from the entire database of analysed ELMy H-mode discharges. Each
point represents the weighted mean of the data in that interval, and the error bar represents the
standard deviation of the distribution. The decrease in ELM frequency with decreasing density
and increasing power is typical of type III ELMs in conventional aspect ratio devices [6–8].
The energy released during the ELM has been calculated from the change in total stored energy
(W ) from EFIT equilibrium reconstruction [9], using EFIT runs with a 200 µs time step.
The fraction of stored energy (W/W ) released from the plasma due to an ELM is shown in
figure 4(a). At low ELM frequencies, up to 3% of the stored energy is released by an ELM.
A drop in the line-averaged density is observed at each ELM, and if the change of the global
density profile is small, this drop can be used to estimate the fraction of particles released during
an ELM. Figure 4(b) shows a plot of the fractional change in line-averaged density as a function
of ELM frequency. At low ELM frequencies, up to 4% of the particles in the core are released
during an ELM. These results are similar to those observed in COMPASS-D [10]. The fraction
of stored energy released by an ELM is given by W /W = n/n + T /T , where the first
term describes the convective losses and the second term the conductive losses. Bearing in
mind the relative uncertainties in the methods used to calculate the respective quantities, the
fact that n/n is greater than or equal to W/W at low ELM frequencies suggests that ELM
conductive losses in MAST are small for beam powers up to 2.5 MW.
Figure 5(a) shows the variation of normalized confinement time, H
H
(≡τ
E
/τ
IPB98(y,2)
E
,
where τ
E
is the energy confinement time and τ
IPB98(y,2)
E
is the ITER energy confinement time
scaling [11]) with ELM frequency (f
ELM
) for a variety of MAST ELMy H-mode discharges.
H
H
rises rapidly from L-mode confinement levels (H
H
∼ 0.5) at f
ELM
1 kHz to H
H
∼ 1.5
at f
ELM
50 Hz. High confinement is thus achieved in MAST at a low ELM frequency even
though no clear transition to type I ELMs has been observed, e.g. an increase in ELM frequency
with increased beam power. Figure 5(b) shows a plot of H
H
versus the number of ELMs per
energy confinement time, which follows a Fishpool-like scaling [12]. The distribution has
been fitted to the Fishpool parametrization:
H = H
max
− af τ
E
(1 − e
−b/f τ
E
),
where f is the ELM frequency, τ
E
is the energy confinement time and H
max
, a and b are
parameters to be determined for the fit. The fitted curve shown in figure 5(b) gives a good
ELM characteristics in MAST 555
10 100 1000
f
ELM
(Hz)
0.0
0.5
1.0
1.5
2.0
0 20 40 60 80 100 120 140
0.0
0.5
1.0
1.5
2.0
H
H
(IPB98(y,2))
H
H
(IPB98(y,2))
(a) (b)
E
f
ELM
Figure 5. Variation of τ
E
(normalized to IPB98(y,2) scaling) with (a) ELM frequency and (b) the
product of ELM frequency and confinement time. The line in (b) shows the results of a fit to
Fishpool scaling.
description of the data and yields H
max
= 1.25 ± 0.1, a = 0.018 ± 0.02 and b = 44 ± 3. The
parameter 1/b ≈ W/W ≈ 2.2% is consistent with the measured values given above.
One alternative way in which ELMs can be characterized is to consider the discharges
in a plot of pedestal density versus pedestal temperature [13], where different regions can be
associated with different ELM types. For example, type I ELMs appear to be associated with
a line of approximately constant pedestal pressure [13], often in the vicinity of the ballooning
stability boundary. It is therefore interesting to see how MAST fits this picture. Thus, the
electron temperature, T
p
, and density, n
p
, at the top of the pedestal for a range of ELMy
H-mode regimes have been determined from Thomson scattering profiles obtained during
inter-ELM periods at a location 2 cm inboard of the radius of peak Dα emission from the
outboard mid-plane, a
Dα
. The radius a
Dα
= 2 cm is found empirically to provide a robust
estimate for the location of the top of the density pedestal in MAST H-mode regimes. This
constant offset is due to the fact that for pedestal densities greater than 3 × 10
19
m
−3
there is
only a small change in density pedestal width with changing plasma conditions [14]. Figure 6
shows the outboard density and temperature profiles for shot 6252, which have been fitted
using a modified hyperbolic tangent [15]. For the density profile, the location of the pedestal
determined from this fit is in agreement, within the errors, with the radius a
Dα
= 2 cm. With
the current Thomson scattering system, electron temperatures below 30 eV at the outboard side
are subject to a large systematic error. In order to produce a modified hyperbolic tangent fit to
the temperature profile, temperature values measured at the inboard side have been mapped to
the outboard side (open circles in figure 6). The value of the pedestal temperature from the fit
is in rough agreement with the value at the radius a
Dα
= 2 cm. In order to obtain the pedestal
temperature at the outboard side for the bulk of discharges, the value at the location of the
density pedestal has been used. Note that unless otherwise stated all calculations presented in
this paper assume that T
i
is equal to T
e
.
ELMy H-modes have been observed with n
p
between 1.5 × 10
19
and 5.5 × 10
19
m
−3
and with T
p
between around 80 and 200 eV. Figure 7 shows a plot of T
p
against n
p
where
both are normalized to B
φ
/q
95
, where B
φ
is the toroidal field on axis and q
95
is the safety
factor at the 95% flux surface. Often the density pedestal is expressed as a function of the
Greenwald density (n
GW
). The MAST discharges shown in figure 7 have values of n
p
/n
GW
in
the range 0.15–0.90. In fact the normalized quantity n
p
q
95
/B
φ
can also be expressed in terms
of the Greenwald density as n
p
(q
95
/B
φ
) = (f (S)/R)(n
p
/n
GW
), where R is the major radius
and f(S) is a function of the shaping of the plasma. On MAST, f (S)/R has been found to
be ∼8.8 for all the discharges in figure 7 and hence the Greenwald density limit occurs for
n
p
q
95
/B
φ
∼ 8.8.
