| International Atomic Energy Agency Nuclear Fusion
Nucl. Fusion 55 (2015) 104001 (14pp) doi:10.1088/0029-5515/55/10/104001
Overview of the JET results
F. Romanelli, on behalf of JET Contributors
a
JET-EFDA, Culham Science Centre, OX14 3DB, Abingdon, UK
E-mail: francesco.romanelli@jet.efda.org
Received
25 November 2014, revised 12 January 2015
Accepted for publication 2 February 2015
Published 27 March 2015
Abstract
Since the installation of an ITER-like wall, the JET programme has focused on the consolidation of ITER design choices and the
preparation for ITER operation, with a specific emphasis given to the bulk tungsten melt experiment, which has been crucial for
the final decision on the material choice for the day-one tungsten divertor in ITER. Integrated scenarios have been progressed
with the re-establishment of long-pulse, high-confinement H-modes by optimizing the magnetic configuration and the use
of ICRH to avoid tungsten impurity accumulation. Stationary discharges with detached divertor conditions and small edge
localized modes have been demonstrated by nitrogen seeding. The differences in confinement and pedestal behaviour before
and after the ITER-like wall installation have been better characterized towards the development of high fusion yield scenarios in
DT. Post-mortem analyses of the plasma-facing components have confirmed the previously reported low fuel retention obtained
by gas balance and shown that the pattern of deposition within the divertor has changed significantly with respect to the JET
carbon wall campaigns due to the absence of thermally activated chemical erosion of beryllium in contrast to carbon. Transport
to remote areas is almost absent and two orders of magnitude less material is found in the divertor.
Keywords: JET, tokamaks, magnetic confinement
(Some figures may appear in colour only in the online journal)
1. Introduction
The European fusion programme is moving into the phase
of implementation of its Roadmap [1] that foresees the use
of JET in Horizon 2020 as the main risk-mitigation element
for the preparation of ITER operation. In 2004 the JET
programme in support of ITER was launched consisting of
three main elements [2]: (i) the installation of an ITER-like
wall (ILW) to reproduce the same material mix for the plasma-
facing components (PFCs) of the nuclear phase of ITER;
(ii) the development of plasma regimes of operation in the
same configuration as ITER; and, (iii) a deuterium–tritium
experiment to test the integrated regimes with the fuel mix of
ITER. During the last couple of years, the ILW characterization
has been almost completed. Significant progress has been
made on scenario development and the preparation of the DT
campaign has started.
In 2013 and 2014, JET carried out experimental
campaigns in deuterium at plasma currents I
p
up to 4 MA,
magnetic field B
t
up to 3.7 T and auxiliary power up to P
NBI
∼
27 MW, P
ICRH
∼ 7MW, P
LHCD
∼ 3 MW. The upgraded
components of the neutral beam system (the newly configured
sources, the actively cooled ducts and the high voltage power
supplies (HVPS)) have all separately achieved the design
targets. However, problems with the old HVPS equipment
and a major water leak in one injection box at the end of
2013 have prevented the achievement of the maximum neutral
beam power needed for a full scenario optimization with the
a
See the appendix.
ILW at plasma currents beyond 2.5 MA. Nevertheless, after
three years of operation with ILW (referred to in the following
as the JET-ILW configuration) the JET heating systems have
improved on the performance achieved with the carbon wall
(the JET-C configuration) (figure 1) The JET experience with
the ILW shows the need of a careful preparation (as now
integrated in the ITER research plan with the choice of the
W divertor from the start of ITER operation) and a proper
real-time protection system for the PFCs in order to achieve
continuous improvement in the plasma performance.
During this period, the ILW characterization has focused
on the consolidation of the basis for the ITER decision on the
first divertor [3, 4] with a specific emphasis given to the bulk
tungsten melt experiment to understand the effect of shallow
melting due to transient events. For this purpose, a protruding
divertor lamella has been installed in the most internal stack of
the horizontal tile. A series of seven reproducible discharges
produced shallow melting of tungsten and demonstrated the
possibility of operating JET in these conditions without the
occurrence of disruptions. The molten layer dynamics has
been successfully modelled using the MEMOS code to validate
the model and to allow a meaningful extrapolation to ITER.
The results have been crucial for the final decision in favour of
the tungsten divertor from the start of ITER operation.
In the first JET campaigns with the ILW [5–7] fuel
retention and material migration studies had priority [8–11].
Now the focus of JET experiments has shifted towards
integrated scenario development [12] to achieve low PFC heat
loads [13–15] and avoid tungsten accumulation [16, 17] with
0029-5515/15/104001+14$33.00 1 © 2015 EURATOM Printed in the UK
Nucl. Fusion 55 (2015) 104001 F. Romanelli
Figure 1. Picture of the interior of the JET vessel with an all carbon wall (JET-C) on the left and with the ITER-like wall (JET-ILW) with
beryllium main chamber and tungsten divertor on the right.
the ultimate goal of achieving high performance plasmas [ 18,
19] in view of the DT experiment. Magnetic geometry,
strike point location and divertor pumping were established
as key aspects for achieving good H-mode confinement,
leading to the re-establishment of long-pulse (∼9 s) high-
confinement H-modes at 2.5 MA. This had to be combined
with the control of tungsten accumulation by central heating
using ion cyclotron resonant heating (ICRH) and sufficient gas
fuelling. Significant effort was devoted to the use of impurity
seeding [20] to produce core-divertor compatible scenarios
which are essential for ITER, as well as high radiative scenarios
which are required for DEMO. Stationary discharges have
been produced by nitrogen seeding with fully detached divertor
legs and small edge localized modes (ELMs). The use of N-
seeding clearly increases the pedestal pressure but the global
confinement is still 15% below the IPB98(y,2) scaling.
Comparisons between baseline and hybrid plasmas have
blurred the distinction between the two scenarios, which now
appear on JET-ILW as a single operating domain. Dedicated
power scans to elucidate the dependence on plasma beta of the
energy confinement time have confirmed that on JET the power
degradation is normally much weaker than the prediction of the
IPB98(y,2) scaling.
Post-mortem analyses of the PFCs retrieved from the
first ILW campaigns [21–24] have confirmed the previously
reported low fuel retention obtained by gas balance, with the
measured deuterium inventory below 0.25% of the injected
amount of deuterium (a factor 16 below that for the carbon
wall). These studies show that the reduced material erosion
and migration lead to reduced trapping of fuel in deposited
beryllium layers which incorporate less fuel in comparison
with carbon layers. In addition, the pattern of deposition
within the divertor has changed significantly with the ILW in
comparison with the JET carbon wall [25] due to the lower level
of re-erosion. Transport to remote areas is almost absent, with
the only significant beryllium deposits (∼15 µm) found on the
apron of the inner divertor. Overall, one order of magnitude
lower deposition rate is found in the divertor compared to
the carbon wall, with deposits in the floor of less than 2 µm
compared with more than 200 µm thick deposits found after
the last JET carbon-wall campaign due to long range migration
via chemical erosion. These results are well reproduced by the
WallDYN code. The WallDYN extrapolation to ITER shows
a reduction in the retention by more than a factor ten and
implies that at least 3000 full power DT discharges could be
produced on ITER before reaching the T-inventory limit. This
is comparable with the time estimated to obtain a substantial
erosion of the divertor.
The investigation of effective runaway electron (RE) beam
mitigation methods has continued. Spontaneous generation
of REs is not normally observed in JET-ILW. Runaways can
be generated through massive Ar injection and accelerated
up to 20 MeV with the production of runaway currents up to
1.2 MA. On JET, runaways can be suppressed by D
2
injection
provided injection takes place before the thermal quench.
Thereafter, runaway suppression using high-Z noble gases has
been attempted but was found so far ineffective.
In section 2 we give an overview of all major results
obtained with the ILW and the implications for ITER operation.
In section 3 the development of ITER-relevant scenarios
in the new all-metal environment is presented, along with
confinement and edge pedestal physics results. Conclusions
and perspectives are presented in section 4.
2. Operation of JET with the ILW
2.1. Bulk tungsten melt experiment
A dedicated experiment has been carried out in JET to address
the uncertainties associated with predicting the impact of
transient tungsten melting in ITER due to ELMs [26–29]. The
horizontal tile of the JET divertor is made of solid tungsten
arranged in four stacks of lamellae in order to minimize the
electromagnetic loads during disruptions (figure 2). JET’s
large size makes it possible to produce repetitive ELMs
with sufficient energy (∼300 kJ per ELM) to melt tungsten.
Deliberate shallow tungsten melting has been produced by
operating with the outer strike point on one protruding lamella,
intentionally modified and installed in one divertor module on
the innermost stack as shown in figure 2 [29]. A series of
seven identical 3 MA discharges with 23 MW of heating power
2
Nucl. Fusion 55 (2015) 104001 F. Romanelli
Figure 2. Schematic view of the lamella assembly in the modified
stack.
were produced. Within 1 s the base temperature of the lamella
was raised to a value well below the melting temperature, but
sufficiently high to facilitate the shallow melting by ELMs with
parallel power densities around 3 GW m
−2
during a further
0.5 s. The temperature on the top-side of the protruding and
normal lamellae is measured by the infrared cameras. The
top view does not allow resolved measurements on the side of
the special lamella and this has made the interpretation more
complex. The temperature measurements from the normal
lamellae do not suffer the same problems and are used to
deduce incident power fluxes as input for a 3D modelling
of the heat transport through the special lamella with the
exposed edge. The results are consistent with melting by
ELMs followed by re-solidification of tungsten in between
events. The melting produced an enhanced tungsten source
with occasional expulsion of small droplets (80–100 µm)
which did not significantly impact the main plasma and caused
no disruptions during the experiments or thereafter. Almost
1 mm depth (corresponding to a volume of ∼6mm
3
) was
moved/removed from the edge by several hundred ELMs
during these pulses. The temperature of the special lamella
is shown in figure 3 [30] and compared with that of a standard
lamella. The analysis confirms that transient melting during
the ELMs occurs and not bulk melting that continues across
the ELM cycle. From photographic evidence it is possible
to say that most of the molten tungsten moved along the
lamella edge although a precise mass balance will be possible
only after post-mortem analysis. The propagation of molten
material is consistent with a j × B driving force, where j is
the current density into the surface (mainly due to thermo-
electronic emission) and B is the local magnetic field [31].
Droplets on the lamella were seen to coalesce and grow, which
increases the risk for longer pulse duration above the melt
threshold. Nevertheless, the consequences of melting had no
significant impact on JET operation.
The inferred power load on the side of the JET special
lamella was substantially lower than expected. In order to
match the IR measurements, the tungsten evaporation rate
inferred from the W
I 400.88 nm line and the Planck radiation,
the side heat loads must be reduced by a factor 2.5 in these
H-mode discharges (a larger reduction factor of about five
is needed to match the measurements in L-mode). This
result is not yet understood: simulations including the gyro-
radius smoothing effects predict only a 20% reduction [32],
2000
Special lamella
Pulse Number 84778 Ref.
84779 Ref.
84778
84781 Ref.
84779
84782 Ref.
84781
84782
Standard (Ref.) lamella
1500
1000
500
0
2500
81410261
Temperature from IR (
o
C)
Time (s)
CPS14.984-14c
Figure 3. Temperature measurement for a few JET pulses used in
the bulk tungsten melt experiment. Temperatures are shown for both
the reference and the special lamella.
whereas the effect of vapour shielding may have contributed
to the power load mitigation in H-mode where the surface
temperature was near melting but could not explain the L-
mode results where the surface is much colder. Although not
understood, this finding has potentially positive implications
for ITER, which may be less sensitive than previously feared
to exposed edges created by chipping of mono-block edges or
components outside tolerance.
JET results are consistent with simulations of tungsten
melting and propagation using the MEMOS code [33], which
has been used to inform decisions on the choice of tungsten
as the material for the first divertor in ITER. The results
have also given confidence that flash melting of the tungsten
divertor elements by ELMs is relatively benign compared to
bulk melting as reported in other experiments [34].
2.2. Material erosion and migration, fuel recycling and
retention
The JET ILW experiment provided for the first time the oppor-
tunity to explore the coupling between tokamak plasma opera-
tion and plasma–surface interaction in the beryllium/tungsten
material environment of ITER, complementing earlier exper-
iments in other divertor tokamaks with metallic walls such as
ASDEX Upgrade [35–39] and Alcator C-Mod [40–42]. These
experiments are crucial in order to validate physics models and
modelling tools for ITER with regard to material erosion and
migration, fuel recycling and retention and impurity concen-
tration and radiation.
The observed strong inter-connection of plasma–surface
interaction with plasma-edge physics such as pedestal or
divertor properties revealed that the impact of the first wall
material on the plasma performance as well as the prominent
role of chemical erosion of carbon in the main chamber
and divertor was previously underestimated. The change in
material migration with the JET-ILW can be seen as one key
result as it impacts directly or indirectly on the majority of
plasma–surface interaction processes mentioned above and
contradicts partially the migration pattern predictions made
3
Nucl. Fusion 55 (2015) 104001 F. Romanelli
10
22
10
21
10
23
10
24
10
25
10
26
10
20
10
19
10
18
Long term retention rate (D/s)
Total wall flux Φ D/s
Pure-C
Be+W
ITER
H-mode
~Φ
0.5+/– 0.1
~Φ
0.7+/– 0.1
JET
Ohmic
WallDYN ITER-Case A
WallDYN JET-ILW
WallDYN ITER-Case D
WallDYN JET-C
JET-ILW Exp
JET-C Exp.
CPS14.984-4c
Figure 4. Long-term retention rate predictions for ITER made by
WALLDYN [45].
for beryllium/tungsten PFCs on basis of migration physics
deduced in JET-C including physical erosion characteristics
of Be and W.
The primary source of Be sputtered in the Be main
chamber wall is a factor of five lower than C sputtered in
JET-C. The initial assumption that pure physical sputtering by
energetic charge exchange neutrals is solely determining the
main chamber source turned out to be invalid—low energetic
ions reaching the first wall also sputtered chemically the C
in JET-C. This chemical erosion at lowest energies (10 eV) is
absent due to energetic physical sputtering thresholds for Be—
even if chemical assisted physical sputtering has been found for
Be [43]. As consequence of the lower initial impurity source,
the inner divertor is only partially covered by Be and indeed
intact W surfaces are found. The local Be flux balance at the
strike-line area is in favour of Be reflection and re-erosion and
due to the low incoming flux not in deposition.
Furthermore, the virtual absence of chemical sputtering
of beryllium inhibits the cycle of multiple erosion/deposition
cycles within the divertor observed with carbon PFCs. Instead
beryllium remains deposited at the positions where it lands
after reflection or a physical sputtering process above the
energetic threshold. It should be noted that, due to the low
deposition rate and the limited operational time on JET, the
JET-ILW migration pattern could represent an intermediate
state with respect to long-pulse operation. This is supported by
the fact that two orders of magnitude less dust were recovered
compared with the carbon wall.
A special effort was made to accurately quantify the
reduction in long-term fuel retention with the JET-ILW,
already measured in gas balance experiments, through post-
mortem analysis. These results were used to validate the
WallDYN code [44] and to allow a meaningful extrapolation
to ITER. The code simulations reproduce both the reduction
in fuel retention and the observed migration pattern. The
remaining long-term retention is caused by implantation and
co-deposition with beryllium and residual impurities. Short-
term retention gained relative importance with respect to
the low level of long-term retention and impacts on the
recycling properties of both beryllium and tungsten and local
2.0
2.5
3.0
1.5
1.0
0.5
3.5
0.2 0.4 0.6 0.8 1.00
Toroidal field
(T)
Ar fraction in DMV
150+/
-
50kA
300kA
150kA
50kA
HXR traces only
No significant trace
CPS14.984-5c
Figure 5. Operational domain for RE production in JET-ILW.
plasma properties. Predictions for ITER using WallDYN
(figure 4)[45] indicate that more than 3000 full power DT
discharges are possible before reaching the fuel-inventory
limit, an amount comparable with that estimated for significant
divertor erosion. If confirmed, the need of frequent fuel
removal will be avoided.
2.3. Disruptions and the generation of REs
Successful disruption mitigation [46–48] with massive gas
injection of a mixture of argon and deuterium has been
obtained and is now mandatory for JET operation above 2 MA.
The radiation efficiency of massive gas injection using the
disruption mitigation valve (DMV) on JET (using a mixture
of 90%D
2
+10%Ar) is 90% at a low thermal energy fraction
W
th
/W
mag
of < 0.3 (where W
th
is the thermal energy and W
mag
is the magnetic stored energy), and reduces to 70% radiation
efficiency at W
th
/W
mag
∼ 0.5. In these discharges the thermal
stored energy was in the range 2.5–3 MJ. This behaviour is
independent of the amount of injected impurities as long as
the number of radiating atoms is above some limit. The
radiated energy shows a clear toroidal asymmetry caused by
a pre-existing locked mode. This asymmetry changes phase
when the mode is locked in a 90
◦
rotated toroidal position.
Massive gas injection into X-point and O-point of the locked-
mode island results in strong changes of the toroidal radiation
distribution. Future experiments with two DMVs will attempt
to minimize the radiation asymmetry. It should be noted
that typical values for the ratio W
th
/W
mag
in ITER will be
W
th
/W
mag
∼ 0.9. Hence, upgrades to the DMV systems are
being made in JET to increase the gas delivery capability, at
different locations, and with the capability to optimize the gas
mixtures. Furthermore, experiments on JET are being planned
at higher W
th
/W
mag
values.
Although after the installation of the ILW REs [49,
50] are rarely generated during spontaneous disruptions,
they can be generated, as in JET-C, using massive argon
injection. Argon dominates the disruption plasma content,
4
Nucl. Fusion 55 (2015) 104001 F. Romanelli
0.5
1.0
0
-
0.005
0.010 0.015 0.020 0.025
(MA)(m)
Z
pos
n
e,line
E
acc
~35 V/m
4
2
0
(10
4
counts.s
-1
)
HXR
0.8
1.0
1.2
0.6
(10
20
m
-2
)
-
20
0
-
40
-
60
(V
m
-
1
)
I
pla
Just after: TQ:
850 kA RE plateau
No RE
870 kA RE
DMV2 time
(no RE case)
DMV2 time
(RE case)
Just before TQ:
no RE
-
1
0
-
2
Time from DMVI openin
g
(
s
)
CPS14.984-10c
0.5
Figure 6. RE suppression by massive D
2
injection. The injection is
effective only if the injected gas reaches the plasma before the
thermal quench (red). A delay of about 2 ms results in the
production of a 850 kA runaway beam.
as shown in figure 5 [51], thus decreasing the effect of
the intrinsic impurities (carbon or beryllium) on the current
quench dynamics. The conditions in which RE appear are
similar between JET-C and JET-ILW: high toroidal field,
high accelerating electric field and low densities (leading to
lower critical electric field for RE generation) favour large RE
currents. They also show a strong dependence on plasma initial
shape and vertical stability.
Energies up to 20 MeV were measured for 200 to 380 kA
RE beams. RE beams up to 1.2 MA at I
p
= 2MA have
been produced by massive argon injection via DMV, leading
to substantial damage of tiles due to localized and toroidally
asymmetrical melting on the inner wall and upper beryllium
limiters [51]. It is to be noted that, within the uncertainty of
the time reference of the IR camera, the interaction between
the tile and the RE beam starts before the RE current decay,
possibly due to contact with the wall. After in-vessel visual
inspection, the affected tile showed clear signs of melting and
droplet ejection over an area of about 10 cm
2
.
Suppression of fully developed runway beams up to
1.2 MA was attempted using massive high-Z gas injections (up
to 45 bar l Kr or 24 bar l Xe). Although the fast camera showed
that some gas arrived at the plasma, no useful mitigation
was observed with no obvious change in hard ray emission
or current decay rate. However, early massive deuterium
injection was found to be efficient at totally suppressing the
runaway beam provided that the gas reached the plasma before
the mixing phase of plasma and incoming gas which occurs at
the thermal quench (figure 6).
3. H-mode physics in an all-metal environment
The qualification of the ELMy H-mode and of the hybrid
regime with the ILW has provided a number of results of direct
0.9
1.4
0.4
0
81216
0.9
1.4
0.4
0.9
1.4
0.4
0.9
1.4
0.4
0.9
1.4
0.4
0.9
1.4
0.4
P
NBI
(MW)
f
ELM
~ 15Hz
f
ELM
~ 40Hz
n/n
GW
= 0.8
Z
eff
= 1.4
n/n
GW
= 0.64
Z
eff
= 1.2
JET Pulse No: 82122
JET Pulse No: 82122
H
98y2
Time (s)
Conf. time normalized to ITER98 scaling
Central electron temperature (keV)
Line integrated density (1×10
19
m
-3
)
Gas dosing rate (1×10
22
/s)
Be II
CPS14.984-11c
20
Figure 7.
The optimization of the magnetic configuration led to the
re-establishment of long-pulse, high-confinement H-modes (red).
Shown in blue is the best discharge obtained in 2012 by controlled
gas puffing to produce high-frequency ELMs and avoid W
accumulation.
relevance for ITER. The baseline and hybrid scenarios have
been progressed towards ITER dimensionless parameters in
plasmas up to 4 MA, with Z
eff
values as low as 1.1 and the
domain of good confinement extended [52]. High performance
scenarios are being successfully developed within the ILW
constraints, namely the control of heat loads on the PFCs, the
minimization of core tungsten concentration and disruption
mitigation/avoidance.
3.1. Scenario development and optimization
H-mode and hybrid plasmas were obtained with energy
confinement enhancement factors (H
98
) in the range of 0.9–
1.2 [53] when compared with IPB98(y,2) scaling law [54].
It was previously reported that stationary H-modes could be
re-established with the ILW by avoiding W accumulation
through the production of frequent ELM regimes by gas
puffing [12]. This, however, had a detrimental effect on
confinement, possibly due to the pedestal cooling by gas
injection. During the last two years, stationary H-modes with
confinement encompassing the ITER value (H
98
= 1) have
been produced by placing the divertor strike points such as
to maximize pumping. In this way stationary, long-pulse
(∼9 s) H-modes have been produced at low value of Z
eff
(figure 7). Interestingly, confinement times in line with or
5
