ms 7037, first revision, feb.2001
Overview of TCV Results
H. Weisen, S. Alberti, C. Angioni, K. Appert, J. Bakos1, R. Behn, P. Blanchard, P. Bosshard,
R. Chavan, S. Coda, I. Condrea, A. Degeling, B. P. Duval, D. Fasel, J.-Y. Favez, A. Favre,
I. Furno, P. Gomez, T.P. Goodman, M. A. Henderson, F. Hofmann, R.R. Kayruthdinov2,
P. Lavanchy, J. B. Lister, X. Llobet, A. Loarte3, V.E. Lukash4, P. Gorgerat, J.-P. Hogge,
P.-F. Isoz, B. Joye, J.-C. Magnin, A. Manini, B. Marlétaz, P. Marmillod, Y. Martin,
An. Martynov, J.-M. Mayor, E. Minardi5, J. Mlynar, J.-M. Moret, P. Nikkola, P. J. Paris,
A. Perez, Y. Peysson6, Z.A. Pietrzyk, V. Piffl7, R.A. Pitts, A. Pochelon, H. Reimerdes,
J.H. Rommers, O. Sauter, E. Scavino, A. Sushkov2, G. Tonetti, M.Q. Tran and A. Zabolotsky
Centre de Recherches en Physique des Plasmas
Association EURATOM-Confédération Suisse
École Polytechnique Fédérale de Lausanne
CH-1015 Lausanne, Switzerland
1
KFKI, Budapest, HU; 2RRC, Moscow, RUS; 3EFDA-CSU, Garching, D;
Troisk, RUS; 5Istituto di Fisica del Plasma “P.Caldirola”, Milano, I;
6CEA, Cadarache, F; 7IPP, Prague, CZ
4TRINITI,
E-mail address of main author: Henri.Weisen@epfl.ch
Abstract. The TCV tokamak (R=0.88m, a<0.25m, BT<1.54T) is equipped with six 0.5MW gyrotron
sources operating at 82.7 GHz for second harmonic X-mode ECH. By distributing the ECCD current
sources over the discharge cross section, fully driven stationary plasmas with Ip=210kA, ne0=2.1019m3
, Te0≈4keV, were obtained for the full discharge duration of 2s. Highly peaked electron temperature
profiles with Te0 up to 12keV were obtained in central counter current drive scenarios with off-axis
ECH. Absorption measurements using a 118 GHz gyrotron have demonstrated the importance of
suprathermal electrons for third harmonic absorption. A coupled heat-particle transport phenomenon
known as “density pumpout”, which leads to the expulsion of particles from the plasma core, has been
linked to the presence of m=1 modes, suggesting that it is due to the existence of locally trapped
particles associated with the loss of axisymmetry. Highly elongated discharges have been developed
with Ohmic heating (κ<2.8) and off-axis ECH. The latter exhibit considerably improved vertical
stability due to current profile broadening. A “gateway” for Elmy H-modes has been discovered, which
allows stationary Ohmic ELMy H-mode operation in over wide range of elongation, triangularity and
density. Divertor detachment experiments suggest the existence of recombination pathways other than
three-body or radiative processes.
1
1. Recent Advances
The TCV (Tokamak à Configuration Variable, R=0.88m, a<0.25m, BT<1.54T) with a vessel
elongation of 3, was designed to be a highly versatile facility destined for the investigation of
the effects of plasma shaping on confinement and stability [1]. Following the installation of a
flexible electron cyclotron heating (ECH) and current drive (ECCD) system, now totalling
2.8MW of power available to the plasma at the second cyclotron harmonic (82.7GHz), TCV
has delivered significant achievements to the fusion community. One of the most important of
these is probably the first demonstration of steady-state fully non-inductive ECCD operation
for 2s at plasma currents of up to 210 kA [2,3]. These experiments also established the
necessity of tailoring the driven current profile by suitably distributing the six available sources
over the plasma cross section in order to avoid disruptive MHD instabilities and have provided
a first validation of theoretical predictions for ECCD efficiencies in steady-state conditions.
In this paper the abbreviation ‘ECH’ shall refer specifically to electron cyclotron heating
without generation of driven current, as obtained when the RF waves are launched at right
angles to the magnetic field, while ‘ECCD’ shall refer specifically to current drive. ECH and
ECCD have also proven to be powerful tools for current profile modification for the purpose of
establishing and controlling improved core confinement (ICC) modes and for improving
vertical stability at high elongation. Reversed or weak central shear discharges at moderate
elongation (κ~1.7), produced by central counter-current drive (CNTR-ECCD) in combination
with pure ECH deposited off-axis for improving MHD stability, have lead to stable electron
confinement enhancements HRLW=3.5 over Rebut-Lallia-Watkins scaling, limited in duration
by the length of the RF pulse. With Ohmic heating alone, the most elongated plasmas can only
be vertically stabilized at high plasma current and the highest elongation achieved so far is 2.8
with an edge safety factor q95=2.5. The addition of a moderate level of off-axis ECH power
2
(1MW) has recently allowed the minimum stable plasma current for κa=2.4 to be reduced
three-fold to 300kA, corresponding to q95=8.2. These experiments open up a wide and
promising operational domain for future investigation of improved confinement modes at high
elongation.
Investigations of the scaling of sawtooth inversion radii and profile shapes have revealed that
these depend solely on the parameter <j>/q0j0 (defined in section 6) in Ohmic plasmas and
both on this parameter and the deposition profile with ECH or ECCD. The results show that,
due to the significant effect of the central elongation, current profile width and inversion radii
do not become excessive at high elongation. A study of sawtooth behaviour as a function of
elongation revealed a marked decrease of sawtooth periods and crash amplitudes for κa>2,
both with ECH and in purely Ohmic plasmas, attributed to a reduced internal kink stability
margin.
The first of three gyrotrons, destined for operation at the third harmonic X-mode (118GHz),
has been brought into service for physics investigations of the role of target plasma conditions
on wave absorption. These experiments have demonstrated that absorption of X3 power is
vastly enhanced when a suprathermal electron population produced by X2-ECCD is present in
the plasma. Total absorption (within 10% error bars) was achieved for the 470kW of injected
X3 power with as little as 350kW of X2-ECCD for target plasma conditioning, as compared to
25% absorption with the same power of X2-ECH.
Although Ohmic H-modes are easy to obtain in TCV, even with the ion ∇B drift direction
away from the X-point, transitions to steady-state ELMy H-modes are only obtained in a
3
narrow region of the operational domain. Remarkably, after such a transition this attractive
confinement mode is very resilient to subsequent changes of elongation and density.
Detachment in open divertor geometries and pure deuterium is found to be easier than
expected from simulations, suggesting important contributions from recombination pathways
other than three-body and radiative processes.
TCV has also been used as a test-bench for the non-linear evolution code DINA [2] with the
purpose of validating it for ITER poloidal field coil feedback controller design.
2. Fully Non-Inductive Operation with ECCD
The six ECH launchers in TCV allow independent steering of the heating sources in both the
poloidal and toroidal directions; this high flexibility matches that of the TCV control system,
permitting the entire vast range of shapes that can be created to be heated in an accurately
localized manner. The six sources can be employed to tailor the current profile in a stationary
manner, as will be required in a prospective reactor for MHD stabilization and optimization of
performance. Crucial to the stationarity requirement is the ECH pulse length capability which
substantially exceeds the current redistribution time from Ohmic to non-inductively driven
profiles (typically < 0.5 s).
4
19
(d)
4
2
10
(kA)
−3
(10 m )
20
0
(e)
0
(f)
5
e0
n
e,av
(MW)
(c)
−2
OH
(V)
V
loop
0
EC
(b)
2
P
3
2
1
0
I
p
(a)
T (keV)
I (kA)
200
100
0
0
2
l
i
(g)
1
0
0.5
1
1.5
2
2.5
3
Time (s)
Fig. 1 Steady-state, fully non-inductive 210 kA discharge in a single-null diverted plasma with 2.8
MW of distributed ECCD: time histories of (a) plasma current, (b) EC power, (c) edge loop voltage,
(d) current in the Ohmic transformer primary, (e) line-averaged density, (f) peak electron temperature, (g) internal inductance
P=2.7 MW, Ip=210 kA
P=1.35 MW, I =160 kA
(a)
(c)
(b)
(d)
p
20
10
P
abs
(MW m−3)
30
J
EC
(MA m−2)
0
30
20
10
0
0
0.2
ρ
0.4
0.6
0
0.2
ρ
0.4
0.6
Fig. 2 Flux-surface-averaged (a)-(c) absorbed power and (b)-(d) driven current density as functions of a normalized radial coordinate proportional to the square root of the plasma volume, for
two different discharges. All profiles are calculated by TORAY. Line-averaged density 1x1019 m-3 (ab) and 1.2x1019 m-3 (c-d), central electron temperature 3.1 keV (a-b) and 3.7 keV (c-d).
5
Studies have been carried out over the past year in TCV at increasing power levels to
demonstrate stable, steady-state current profile control and to validate ECCD physics. After the
first demonstration of fully non-inductive operation in steady state with ECCD in a tokamak
was performed in TCV [3], work has progressed to a present record non-inductively driven
current of 210 kA [4], shown in Fig. 1. The current is sustained non-inductively for 2 s, while
plasma conditions relax over a time scale of less than 0.5 s. The current in the Ohmic
transformer primary is kept constant by feedback throughout the ECCD pulse; thus, in the
stationary phase during which the currents in all the shaping coils are also constant, no flux is
supplied to the plasma. The bootstrap current fraction is calculated to be 8% in this discharge.
The ECCD efficiency is defined as η=Iec/Pec, where Iec is the driven current and Pec the power.
A common figure of merit is the product neR0η, where ne is the line averaged density, which
for the discharge in Fig. 1 corresponds to 0.0073 [1020A/W/m2]. Higher figures of merit, up to
0.016 in the same units, have been obtained by concentrating all sources in the centre, but such
discharges become MHD unstable and disrupt. It was experimentally found that at each power
level there is a minimum stable driven current profile width, below which such MHD instabilities develop and lead to disruption. As these instabilities are driven by current or pressure
gradients, which increase with power for constant profile shape, the deposition width must be
increased with increasing power. This results in an effective degradation of the maximum
global ECCD efficiency with power. By way of illustration of this effect, Fig. 2 shows the
power and current deposition profiles calculated by the linear ray tracing code TORAY [5] (with
the Cohen package [6] for current drive estimation) for the case of Fig. 1 and for a case with
similar plasma conditions but only three sources (1.35 MW) and a steady-state non-inductive
current of 160 kA. Each case corresponds to the narrowest stable deposition profile identified
over a series of discharges at each power level.
6
The fundamental principles of ECCD have been known for twenty years [7], but a thorough
experimental validation is still missing. One of the principal aims of the ongoing experiments
on TCV is to provide that validation, particularly through measurements of the efficiency η. In
linear theory the efficiency has the theoretical dependence η = T e η T ' ⁄ [ Rn e ( Z eff + 5 ) ] ,
where Te is the electron temperature, R is the major radius, ne is the density, Zeff is the effective
ion charge and η T ' is a function of the trapped particle fraction and of the parallel wave
number [8]. The average efficiencies measured on TCV in a series of fully non-inductive
discharges with varying deposition profile widths have shown a dependence on the minor
radius of the magnitude predicted by theory, giving a clear indication of the existence of
trapped particle effects [4]. Furthermore, the absolute value of η T ' agrees with predictions by
the linear TORAY code to within 30%, over a range of profiles resulting in a variation in η T ' of
a factor of 3.5. Nonlinear effects thus do not appear to be significant in the present conditions.
This observation is in conflict with theoretical predictions [9] indicating that efficiency should
be enhanced in TCV conditions owing to quasilinear modification of the electron distribution
function. This discrepancy may be due in part to the radial diffusion of fast electrons, an effect
not considered in the theoretical analysis.
3. Quasi-Stationary Improved Core Confinement by Shear Optimization
An Improved Core Electron Confinement (ICEC) regime has been obtained by current profile
tailoring with ECH and ECCD in quasi-steady state conditions [10]. The intense central
counter current drive (CNTR-ECCD) produced in the ICEC regime and related equilibrium
modelling suggest the presence of a strong reverse shear profile in the plasma core. In most
tokamak experiments reversed shear profiles are obtained only transiently by heating during
7
the ramp-up phase of the plasma current. Taking advantage of the very localized power
deposition, which is typical of ECH and ECCD, and making use of the flexibility of the beam
launching system on TCV, an optimized current density profile could be established and
maintained for the duration of the heating pulse, corresponding to 200 energy confinement
times and several current redistribution times. For this purpose a combination of 4-5 X2
heating beams was used to deliver a total power in the range 1.8-2.3 MW to the plasma. The
optimized scenario begins with off-axis heating just outside the q=1 surface using two or three
X2 beams followed 300 ms later by central power deposition with a strong counter current
drive (CNTR-ECCD) component using two beams. The discharge evolution is shown in Fig. 3
for two examples with central CNTR-ECCD and for central ECH. Typical plasma parameters
were δ95≈0.2, κ95≈1.6, q95≈6, Ip≈200kA and <ne>≈1.5×1019m-3.
Central CNTR-ECCD leads to significantly higher peak electron temperatures (8-12keV) than
central ECH (~4keV) and better global plasma performance. The delay between the beginning
of the off-axis ECH and the on-axis CNTR-ECCD, which is of the order of the current redistribution time, is necessary to establish a suitably broad and MHD stable target current profile for
the central CNTR-ECCD pulse. If the delay is reduced to zero, or if the off-axis ECH is
altogether omitted [11], high temperatures (~10keV) and confinement can also be obtained,
although transiently, frequently terminating with violent, sawtooth-like MHD collapses.
Similar high performance phases, in the presence central CNTR-ECCD, have been reported
from ASDEX-UPGRADE [12]. Unlike discharges with central CNTR-ECCD only, the ICEC
plasmas shown in Figs.3 & 4 are MHD quiescent, like #18639 and #18518, or else only exhibit
benign levels of sawtooth activity, like #18635, with crash amplitudes of ~3keV and periods of
~20ms, which far exceed the global energy confinement time of ~5ms.
8
shots 18604, 18635, 18639
a)
Teo [keV]
10
ρh = 0.35
ρh = 0.3
ρh = 0.32
counter ECCD
on axis
ECH off-axis
10
8
6
4
+counter-ECCD on-axis, 2.25MW
8
6
4
ECH on axis
2
a)
12
Te [keV]
12
shot 18518
14
1.35MW
ECH off-axis
2
OH only
7
b)
6
5
3
q
HRLW
b)
4
2
4
3
2
1
0
0
counter ECCD or ECH on axis
1
ECH off axis
0.5
1
time [s]
0
0
1.5
0.2
0.4
ρ
0.6
0.8
1
Fig. 3 a) Evolution of central electron tempera-
Fig. 4 a) Electron temperature profiles as simu-
ture from Thomson scattering and b) enhance-
lated using RLW transport model for 3 phases
ment factor over RLW energy confinement time
of an ICC shot, together with measurements.
for different ECH/CNTR-ECCD scenarios com-
green: OH phase, red: 1.35 MW off-axis ECH,
bining off-axis ECH (0.9MW) at different loca-
blue: combination off-axis ECH and 0.9MW
tions ρh with delayed central ECH or CNTR-
central CNTR-ECCD.
ECCD. green: with 0.9MW central ECH
b) Corresponding calculated safety factor pro-
blue and red: with 0.9 MW central CNTR-
files with uncertainties (shaded) related to
ECCD
location of central CNTR-ECCD sources.
The achieved electron energy confinement times exceed the one predicted by the Rebut-LalliaWatkins (RLW) global electron confinement scaling law [13] by a factor of ~3.5, as seen in
Fig. 3b. The enhancement over the ITER-89P global confinement law [14] (which includes
ions) during the CNTR-ECCD phase is a factor of 1.5. It should however be noted that, while
RLW scaling agrees with the measured confinement during the Ohmic phase, ITER-89P over-
9
estimates the Ohmic confinement by a factor of two. Owing to the low densities necessary in
TCV for application of efficient ECCD, electrons and ions are decoupled, resulting in core ion
temperatures of only a few hundred eV, as measured using a neutral particle analyser. This,
together with the main ion dilution by carbon impurities, leads to the ion contribution to the
stored energy being negligible (<10%) and is the likely reason for falling short of ITER-89P
predictions in ordinary L-modes at low density.
In the absence of direct measurements of the current density profile, the evolution of the qprofile and its relation to the reduction in transport coefficients is inferred from time dependent
simulations [15] using the PRETOR code [16] and the RLW local transport model. The RLW
transport parameters depend strongly on the magnitude of the magnetic shear, but are assumed
not to depend on its sign. An example is presented in Fig. 4a, with simulated temperature
profiles for the Ohmic, off-axis ECH and high performance ICEC phases, together with the
experimental data from Thomson scattering, showing good agreement. Within the framework
of the local RLW heat diffusivity model, used in these calculations, the negative shear zone
near the center is essential for the confinement enhancement [15]. The corresponding qprofiles in Fig. 4b show reversed shear during the central counter ECCD phase. The central
CNTR-ECCD current is calculated to be 125kA. The range of plausible q-profiles has been
computed by PRETOR used as a fixed-boundary equilibrium solver (without transport model)
constrained by the experimental density and temperature profiles, the effective ion charge, the
edge loop voltage, and reflects the high sensitivity of the calculation to the location of the
CNTR-ECCD source taken from TORAY calculations. This sensitivity to the ECCD source
location has motivated an experimental scan of the position of the CNTR-ECCD component
which has shown that an outward displacement of only 10% of the minor radius caused a 40%
10
reduction in the central plasma temperature, which is also borne out by the PRETOR
simulations [17].
4. Full Absorption of ECH Power at the Third EC Harmonic in X-Mode
Plasmas in the TCV Tokamak have, for the first time, been heated using the first of three
0.5MW gyrotrons to be deployed at a frequency of 118 GHz [18], corresponding to the third
EC harmonic. One of the motivations for X3 ECH in TCV is the possibility of heating at
densities which are inaccessible with the X2 ECH system. The experiments reported here were
aimed at establishing the importance of the plasma conditions, mainly electron temperature
and suprathermal tail electron distributions, for the absorption of X3 ECH power [19]. For this
purpose the plasmas were preheated with different power levels of X2 ECH and ECCD. The
X3 wave was launched from the low field side via one of the upper lateral launching mirrors
normally used for X2 heating [20].
11
Fig. 5 Launching geometry and typical time traces for the X2 and X3 ECH. Top to bottom: RF power,
poloidal beta, loop voltage, soft X-ray signal, hard X-ray signal, peak electron temperature.
PX2=PX3=0.47MW, Ip=200kA.
The target plasmas used in these experiments have the parameters R=0.88m, a=0.25m, κ=1.31,
B0 = 1.42T, ne(0)= 2.5×1019m-3. The launching geometry and time traces of relevant
parameters are shown in Fig. 5 for a typical discharge. The top trace shows the timing X2
ECCD and X3 ECH. In all experiments the X2 power was kept constant from 0.3s to 1.3s
whereas the X3 power was applied from 0.5s to 1.2s and included a phase with 100%
modulation at 237Hz (0.8s to 1s).
The total stored energy variation was measured during the modulated part of the X3 RF pulse
using a diamagnetic loop. The modulation frequency of fm=237Hz was chosen such that
1⁄2πf<τe~5ms, where τe is the electron energy confinement time. While the X3 power
12
(0.47MW) and launching geometry, aimed at the plasma centre, were kept constant, different
X2 conditions were investigated, including variations of the toroidal launch angle φ, the power
deposition radius, the total X2 power and the plasma current. A toroidal injection angle scan of
the X2 launch, with PX2≈PX3=0.47MW, has revealed a clear asymmetry. X3 absorption is
highest on target plasmas with X2 injected with φ=+130, corresponding to CO-ECCD. Fig. 6
shows the X3 absorbed power fraction, versus X2 preheat power for three X2 launching angles
corresponding to CO-ECCD (φ=130), ECH (φ=00) and CNTR-ECCD (φ=-130). For COECCD target plasmas, within the experimental error bars, nearly 100% single pass absorption
is obtained. The interpretation of the measured absorption as being due to single pass
absorption is supported by a polarization scan of the X3 RF beams, from X-mode to O-mode,
as well as by a poloidal scan of the launch angle, from central to off-axis deposition. These
unfavourable conditions lead to a strong reduction of the measured X3 power absorption,
which is not consistent with a picture attributing absorption to a multi-pass effect involving
many internal reflections in the vacuum vessel.
Calculations of the theoretical absorption with the TORAY ray tracing code [5], which makes
the assumption of an isotropic, Maxwellian velocity distribution, are in fair agreement with the
experimental results corresponding to ECH preheating. This observation is also in agreement
with X3 absorption measurements by Pachtman et al. [21]. However, the measured X3
absorption exceeds that predicted by TORAY by a factor of up to 3 for the CNTR- and COECCD cases. The only explanation of the discrepancy is that a large fraction of the X3 power is
absorbed by energetic tail electrons created by X2 ECCD. The presence of these is confirmed
by the measurement of photon spectra using an energy resolving hard X-ray camera and a high
field side ECE radiometer [22]. The ECE radiometer detects suprathermal radiation levels
exceeding the thermal level by up to a factor of 5, while effective X-ray photon temperatures in
13
the range 12-30 keV, depending on the ECCD injection angle, are measured in the presence of
ECCD and X3 ECH. The insert in Fig. 6 shows that hard X-ray (>10keV) emission is highest
with X2 CO-ECCD and lowest with X2 ECH in all phases of the experiment.
1
X2/CNT-ECCD φ =-130
300
0.6
X2/ECH
200
0.4
Hard-x [a.u]
X3 Absorbed fraction (DML)
400
0.8
0.2
central chord
Eph > 10 keV Te(0) = 2keV
X2+X3
100
C0-ECCD
CNT-ECCD
X2
ECH
0.3
0.4
0.5
time[s]
0
0
500
1000
1500
X3 Absolute absorbed power (DML) [kW]
X2/CO-ECCD φ =130
0
2000
X2 Injected Power [kW]
Fig. 6 Measured X3 absorption using the DML versus X2 preheat power for three different X2 launching configurations:
- CO-ECCD (green, φ=13o),
- CNTR-ECCD (red, φ=-13o),
- and ECH (blue, φ=0o).
3rd harmonic ECH (φ=00) is kept constant at 0.47MW with central deposition.
Insert shows hard X-ray (>10keV) signals for X2 Co- (green) and CNTR-ECCD (red) as well as X2
ECH (blue) cases. X3 ECH is applied from 0.4 s when the large rise in hard X-ray emission is observed.
The energies of the suprathermals are too low for the Ohmic electric field to be likely to be
responsible for the toroidal asymmetry. This asymmetry is attributed to the geometry
dependence of the overlap of the birth region of suprathermals generated by X2 ECCD with
the region of X3 deposition [23].
14
5. Particle Transport with High Power ECH and ECCD
A coupled heat and particle transport phenomenon, leading to particle depletion from the
plasma core is observed in a variety of plasma conditions with centrally deposited ECH and
ECCD in TCV. This phenomenon, known as “density pumpout” causes inverted density
sawteeth in the core of sawtoothing discharges and leads to stationary hollow profiles in the
absence of sawteeth. The density pumpout has been linked to the presence of (n,m)=(1,1)
MHD modes and can be suppressed by stabilizing the mode by means of operation at high triangularity. The correlation of pumpout with loss of axisymmetry suggests that neoclassical
transport processes involving locally trapped particles other than those arising from the
toroidal field ripple and analogous to those in heliacs, may account for the phenomenon in
tokamaks as well.
neo[m-3/1019]
1.6
SX0[W/m2]
1.8
1.8
1.76
(a)
550
1100
900
(b)
(c)
0.89
450
0.895
time[s]
0.9
(d)
1.115
1.118
time[s]
1.121
Fig. 7 Sawteeth on central Abel-inverted density (top) and raw X-ray signal (bottom) for Ohmic
heating (left) and ECH (right) with δa=0.22 and PECH= 1.45MW.
Fig. 7 shows the difference in sawtooth behaviour at low and high heating power at low triangularity. For PECH>0.5 MW density sawteeth are inverted. The situation is different at high triangularity (δa>0.3) when both X-ray traces and central densities have “normal”, triangular
15
sawteeth. The essential difference appears to be that at low triangularity a (1,1) magnetic island
is present during the sawtooth ramp phase, whereas the plasma is MHD-quiescent during the
ramp phase at high triangularity.
1.6
Discharge No. 18549
Discharge No. 18811
1.4
ne[m310-19]
1.2
1.2
0.8
Fig. 8 Examples of elec-
0.8
0.4
0
-0.2
tron density (top) and
0.4
temperature
0
0
0.4
0.2
Z[m]
0.6
-0.2
0
0.2
Z[m]
0.4
0.6
(bottom)
profiles in a Co-ECCD
(left) and a Counter-
6
Te[keV]
5
4
4
3
3
2
2
1
1
-0.2
ECCD discharge (right).
6
5
0
0.4
0.2
Z[m]
0.6
-0.2
0
0.2
Z[m]
0.4
0.6
The convective heat flux associated with the “pumped-out” particles is a small fraction,
estimated to be less than 10%, of the loss power from the core. In sawtoothing plasmas
strongly hollow density profiles cannot develop because the sawtooth crashes regularly flatten
density and temperature profiles. However with ECCD many situations arise when sawteeth
are stabilized for long enough (typically 10 ms or more) for the hollowness to become
significant enough for the Thomson scattering system, as shown in Fig. 8.
One of the first explanations for “pumpout” in tokamaks by Hsu et al [24], based on the
production of an excess of banana electrons and a poloidal charge asymmetry, is not consistent
16
with our observations . We propose that a loss of axisymmetry provides the crucial physics for
this phenomenon by allowing the existence locally trapped particles, which are not confined in
the core region [25]. The presence of an (n,m)=(1,1) island causes the core to be helically
displaced. The vicinity of the displaced core acquires stellarator-like features. As a result
trapped particle orbits exist even at the magnetic axis, just as in a heliac configuration. This
region may act as a sink from where locally trapped particles are lost to beyond the non-axisymmetrical region (typically outside q=1). The coexistence of locally and toroidally trapped
particles within the q=1 surface can also be expected to give rise to competing transport
phenomena, including pumpout, since the neoclassical off-diagonal terms associated with
these two classes of particles have opposite signs [26]. In steady state the resulting profiles
should be characterized by ∇n e ⁄ n e = – d 12eff ⋅ ∇T e ⁄ T e , where d12eff is the effective nondiagonal term resulting from all relevant transport processes. Near the displaced axis the effect
of locally trapped particles dominates (d12~1 in the long mean-free path regime), giving rise to
an outward flux after a sawtooth crash such as to set up a hollow density profile, while further
away toroidally trapped particles (d12=-1⁄2), or an anomalous mechanism, are most important,
giving rise to an inward pinch .
6. Dependence of Inversion Radii and Peaking Factors on Plasma Shape
When considering highly elongated tokamak designs, a frequently expressed concern is that as
a result of the high current carrying capacity of elongated plasmas, sawtooth in inversion radii
and consequently crash amplitudes may become excessive. In Ohmic plasmas we observe that
the normalised sawtooth inversion radius ρinv and the profile inverse peaking factors
(normalised widths) <pe>/pe0, <Te>/Te0 and <ne>/ne0 for electron pressure, temperature and
17
density, depend on the current profile peaking via the parameter <j>/(q0j0), where <j> is the
cross-sectionally averaged current density, irrespective of plasma shape and electron density
[27]. This parameter can be evaluated without knowledge of the still somewhat controversial
value of the axial safety factor since q0j0=B0(κ0+1⁄κ0)⁄(µ0R0), where κ0 is the axial elongation.
The core elongation is generally in good agreement with the elongation of emissivity contours
from X-ray tomography. In order to reduce the large scatter of the Thomson scattering
measurements at random times in the sawtooth cycle we define “clipped” profile widths as
<Te>/Te1 where Te1 is the electron temperature at the sawtooth inversion radius and Te=Te1 for
ρ<ρinv and Te=Te for ρ≥ρinv. These widths are sensitive to the profile in the confinement zone
(q>1) and are shown in Fig. 9 for a wide variety of Ohmic and ECH L-mode plasmas [28]. The
Ohmic data show a remarkably narrow distribution as a function of <j>⁄(q0j0) and are in good
agreement with an Ohmic relaxation model based on the assumption that the magnetic entropy
is stationary in time [29]. The parameter <j>⁄(q0j0) performs better than safety factor based
scalings such as 1⁄q95, for which the dispersion is larger, and is related to the popular database
variables q95, κ95 and δ95 [28]. With ECH heating <j>⁄(q0j0) remains the primary scaling
parameter, but now the profiles are modified by the ECH heat deposition profiles, the widths of
which are not matched in the experiments to <j>⁄(q0j0).
One of the benefits of high elongation in a large future fusion experiment is the ability to raise
the plasma current for a given value of the safety factor. There is a concern that the high plasma
currents sustainable at high elongation may lead to excessively large inversion radii and
sawtooth amplitudes, susceptible of producing seed islands that may trigger neoclassical
tearing modes (NTM’s) which degrade confinement. From the observed scaling for the
inversion radius, ρinv≈<j>/(q0j0), we can express the total current as
18
2
πa B 0
I p ≅ ρ inv ⋅ ---------------- ⋅ κ a ( κ 0 + 1 ⁄ κ 0 ) .
µ0 R0
This relation is similar to that obtained when scaling Ip at fixed edge safety factor, where
I p ∝ κ a ( κ a + 1 ⁄ κ a ) . The above results show that if sawtooth inversion radii per se are a
concern, the design plasma current and elongation can be scaled at fixed inversion radius
instead of fixed safety factor. This results in a small penalty in plasma current (compared to
fixed safety factor scaling), given by the ratio ( κ 0 + 1 ⁄ κ 0 ) ⁄ ( κ a + 1 ⁄ κ a ) . For typical plasma
parameters this ratio is still as high as 0.9 for κa=2 [28]. Moreover, sawtooth amplitudes are
seen to decrease strongly for κa>2 both in Ohmic and in ECH heated discharges [30],[31].
These observations suggest that operation at high elongation may be of interest for the
avoidance of NTM’s.
0.7
OH
ECH
0.7
0.6
<T̄e>/Te1
<T̄e>/Te1
0.6
0.5
0.5
0.4
0.4
1.0 <κa<1.25
1.25<κa<1.5
0.3
0.2
0.1
0.2
0.3
0.4
0.5
<j>/(q0j0)
1.0 <κa<1.25
1.25<κa<1.5
1.5 <κa<1.75
1.75<κa<2.0
2.0< κa<2.25
0.3
1.5 <κa<1.75
1.75<κa<2.0
2.0< κa<2.25
2.25<κa<2.6
0.2
0.3
0.6
0.35
0.4 0.45 0.5
<j>/(q0j0)
0.55
0.6
Fig. 9 Electron temperature inverse peaking factors in Ohmic (left) and ECH (right) plasmas.
19
7. Development of Highly Elongated Discharges
One of the main aims of the TCV tokamak is the creation and study of highly elongated
plasmas. Part of the motivation for this comes from the fact that the maximum plasma current
increases with elongation and according to Troyon scaling [32], the beta limit is proportional to
the normalized current, IN=I/(aBT). This favorable trend has been verified experimentally up to
elongations of κ=2.3 [33], but it is not known whether it continues to be valid at higher
elongation. The creation of elongated plasmas with κ>2.5 in a tokamak with conventional
aspect ratio is an extremely difficult task [34]. In Ohmic plasmas, axisymmetrical stability
imposes a lower limit to the normalized plasma current, which is necessary to produce a
sufficiently broad current profile, and simultaneously, the non-axisymmetrical modes impose
an upper limit to the current. The stable operating window between these two limits decreases
as the elongation increases.
Axisymmetrical stability can be improved in several different ways. First, the passive stability
can be improved by adapting the plasma shape as closely as possible to the shape of the
vacuum vessel. Second, the vertical position control system can be optimized such that
operation at very low stability margins becomes possible. Third, the current profile can be
widened either by operating at the maximum possible plasma current, by using a fast ramp-up
scenario or by applying off-axis ECH/ECCD. Non-axisymmetrical stability, on the other hand,
can be improved by operating at low beta or at low current, since at high elongation, the current
limit increases as beta decreases [35]. Clearly, the requirements for axisymmetrical and nonaxisymmetrical stability are partially contradictory. As a result, beyond a certain elongation,
the stable operating window shrinks to zero. Using the above methods in Ohmic plasmas, a
maximum elongation κ=2.8 with Ip⁄aBT=3.6MAm-1T-1 has been achieved in TCV (Fig. 10).
20
With Ohmic heating alone vertically stable plasmas with such extreme elongations have only
been obtained at high values of normalized average current density, <j>*=µ0R0<j>⁄BT ~1.7.
With 1-2MW of ECH power deposited near or outside mid-radius, it is possible to create
highly elongated plasmas with much lower values of <j>*[31]. In the example of Fig. 11
plasma elongation was raised from κa≈1.75 to κa=2.4 corresponding to <j>*=0.46 and
<j>⁄q0j0=0.2, just by applying the ECH power at fixed quadrupole field. The resulting
broadening of the current profile leads to vastly improved vertical stability. The example
presented has a temperature profile width <Te>⁄Te0≈0.4, which is a factor ~1.5 times broader
than obtained in Ohmic plasmas at the same value of <j>/q0j0. These experiments open up a
wide operational domain for the investigation of confinement at high elongation.
Fig. 10 (left)
Record elongation in TCV (#19373).
κa=2.8, k0=2.15,
Ip=755kA, BT=0.8T,
δa=0.4, q95=2.5, <j>*=1.74
Fig. 11 (right)
ECH-assisted, highly elongated plasma
(#19533).
Magenta: LCFS of initial configuration.
Red: LCFS of final elongated plasma with
Ip=300kA, BT=1.43T, κa=2.4, q95=8.3,
<j>*=0.46. ECH ray trajectories from
TORAY are also shown, parts beyond X2
resonance are in black.
21
8. Ohmic ELMy H-mode accessibility
Ohmic H-modes are easily obtained in TCV, even with ∇B away from the X-point. In most
conditions these are ELM-free and terminate in a high density disruption. Transitions from Lmode to stationary ELMY H-modes for SN configurations with reversed ion ∇B are only
observed (at the nominal field BT=1.43T) in a narrow range of discharge parameters:
0.35MA≤Ip≤0.43MA, 4.5×1019m-3≤<ne>≤6×1019m-3, 1.6≤κa≤1.7, 0.5≤δa≤0.6. Moreover the
gap width between the inner wall and the plasma last closed flux surface must be between 1
and 3 cm. Outside this domain either L-modes or ELM-free H-modes are obtained. This would
be very restrictive, were it not for the high robustness of the ELMy H-mode, once it is
established. After the transition both the plasma elongation (together with the plasma current)
and the plasma density can be varied over a wide range as shown in Fig. 12 [36]. An ELMinsensitive magnetic position observer has been developed, which prevents an undesirable
vertical controller response during the ELM events [37].
1
L/H Phase
2.05
2
@ LH Trans.
0.9
ELMy
1.95
0.8
1.9
ne [1020 m−3]
Plasma elongation
L−mode
1.85
0.7
1.8
0.6
1.75
1.7
0.5
1.65
0.4
1.6
3.5
4
4.5
Plasma current [A]
5
0.2
5.5
5
x 10
0.4
0.6
0.8
1
Time [s]
1.2
1.4
1.6
Fig. 12 Operational domain for ELMy SN H-modes. a) Accessible elongations and currents b) Accessible density (time evolutions). All discharges transited through a “gate” in the operational domain indicated by the green symbols. Central insert: LCFS at transition and at highest elongation.
22
9. Detachment in Variable Divertor Geometry
Although the requirement of shape flexibility precludes the use of fixed baffle or optimized
divertor target structures, it does allow for the investigation of diverted equilibria not
achievable in more conventional tokamaks. One such configuration, shown in Fig. 13, has been
extensively used for studies of divertor detachment in Ohmic conditions and unfavorable ∇Β
drift direction with deuterium fueling only [38]. The equilibrium is simultaneously characterized by a very short inboard poloidal depth from X-point to strike point on a vertical target and
an extremely long poloidal depth to a horizontal target on the outboard side. Density ramp
discharges, invariably terminated by an X-point MARFE, leave the inboard target plasma
attached even at the highest densities, whilst clear partial detachment is observed at the
outboard target. Extensive modelling of this configuration using the B2-EIRENE coupled
package shows in fact that the outboard divertor achieves high recycling at very low densities,
with the rollover to detachment occurring near the outer strike point very soon after the density
ramp begins. The differences in detachment threshold at the two targets can be ascribed to a
large extent to divertor magnetic geometry.
a) #17823
b) #17824, fe = 2.8
c) #17823, fe = 6.4
d) #17822, fe = 9.3
Fig. 13 Distributions of Dα emissivity for varying outer target flux expansion at the same
degree of detachment.
23
Whilst there is little latitude for changing the inboard geometry, a series of experiments has
recently concentrated on studying the effect of outer target flux expansion, fe, on the
detachment behaviour. Fig. 13 (b,c,d) show inverted tangential CCD camera images of Dα
emissivity from the divertor volume for the same absolute degree of detachment (DOD) [39] as
fe increases. The DOD describes the extent to which the target ion flux obeys the scaling,
2
Γ ∝ n e predicted by the basic Two-Point Model [40] of the divertor for the high recycling
regime. By comparing Da emissivities for the same degree of particle flux detachment, any
differences due to magnetic geometry can be isolated. The distributions in Fig. 13 clearly show
the effect of plasma plugging by the expanded flux surfaces and can be qualitatively
reproduced by code simulations. Interestingly, however, the latter find the maximum in the
emissivity to be located at the strike point, in evident disagreement with experimental
observation [41].
Moreover, if the absolute level of detachment is to be quantitatively matched by the code using
the three-body and radiative recombination processes commonly accepted to be responsible for
detachment in most tokamak divertors, then the rate coefficients for these processes must be
artificially increased by factors of 5 or more. This is a strong indication that other effects
dominate in the relatively low density plasma characterizing the TCV outer divertor. Such
pathways may include molecularly assisted recombination processes involving the deuterium
molecule (for example, inelastic charge exchange, D+ + D0 → D + (D2)+ followed by
immediate dissociative recombination, e + (D2)+ → D + D) or the recently proposed [42]
source of increased volume recombination involving proton charge exchange with hydrocarbon molecules. Such pathways are currently under further study.
24
Acknowledgement: This work was partly supported by the Swiss National Science
Foundation.
References:
[1] Hofmann F. et al., Plasma Phys. Control. Fusion 36 (1994) B277
[2] Karyutdinov R.R and Lukash V.E., J. Comp. Physics 109 (1993) 193
[3] Sauter O. et al, Phys. Rev. Letters 84 (2000) 3322
[4] Coda S. et al, Plasma Phys. Control. Fusion 42 (2000) B311
[5] Kritz A.H. et al, Proc. 3rd Varenna-Grenoble Int. Symposium on Heating in Toroidal
Plasmas, Grenoble, 1982 (CEC, Brussels), vol. II, p. 707
[6] Cohen R. H., Phys. Fluids 30 (1987) 2442
[7] Fisch N.J. and Boozer A.H., Phys. Rev. Letters 45 (1980) 720
[8] Alikaev V.V. and Parail V.V., Plasma Phys. Control. Fusion 33 (1991) 1639
[9] Harvey R.W., McCoy M.G. and Kerbel G.D., Phys. Rev. Lett. 62 (1989) 426
[10] Pietrzyk Z.A. et al., Phys. Rev. Lett. 86 (2001) 1530
[11] Pietrzyk Z.A. et al., Phys. Plasmas 7 (2000) 2909
[12] Wolf R.C. et al, Proc 18th IAEA Fusion Energy Conference (2000) IAEA-CN-77/EXP4/04
[13] Rebut P.H., Lallia P.P and Watkins M.L., Proc. 12th Int. Conf. Plasma Physics and
Controlled Nuclear Fusion Research, Nice 1988, edit. IAEA, Vienna, Vol 2., 191
[14] Uckan N.A. and ITER Physics Group in ‘ITER Physics Design Guidelines: 1989’, IAEA,
Vienna 1990, IAEA/ITER/DS/10
[15] Angioni C. et al., to be publ. in Theory of Fusion Plasmas, Ed. Compositori, Bologna 2001
[16] Boucher D. and Rebut P.H., in Proc. IAEA Tech. Conf. on Advances in Simulation and
Modelling in Thermonuclear Plasmas, Montréal, 1992
[17] Goodmann T.P. et al., Proc 18th IAEA Fusion Energy Conference (2000) IAEA-CN-77/
25
EXP4/09
[18] Alberti S. et al, Fusion Engineering Design 53 (2001) 387
[19] Alberti S. et al, Proc 18th IAEA Fusion Energy Conference (2000) IAEA-CN-77/PD/2
[20] Goodmann T.P. et al., In Proc. 19th Symp. Fusion Technology (Lisbon, 1996) Vol 1, Ed.
C. Varandas and F. Serra (Amsterdam: Elesevier), p 565.
[21] Pachtman A., Wolfe S.M., Hutchinson I.H., Nuclear Fusion, Vol.27, No.8, (1987)
[22] Alberti S. et al., submitted to Phys. Rev. Lett. (2001)
[23] Gomez P. et al., submitted to 26th EPS Conference on Controlled Fusion and Plasma
Physics, Funchal, Portugal, 18-22 June 2001.
[24] Hsu J.Y. at al., Phys. Rev. Lett. 53 (1984) 564
[25] Weisen H., Furno I., Goodman T. and TCV Team, Proc 18th IAEA Fusion Energy
Conference (2000) IAEA-CN-77/PDP/6, to be publ. in Nucl. Fusion (2001, ms 7126)
[26] Kovrizhnykh L.M., Nucl. Fusion 24 (1984) 851
[27] Weisen H. et al, Plasma Phys. Contr. Fusion 40 (1998) 1803
[28] Weisen H., Furno I. and TCV Team, to be publ. in Nucl. Fusion (2001, ms 7125)
[29] Minardi E. and Weisen H., Nuclear Fusion 41 (2001) 113
[30] Reimerdes H. et al, Plasma Phys. Contr. Fusion 42 (2000) 629
[31] Pochelon A. et al., Proc 18th IAEA Fusion Energy Conf. (2000) IAEA-CN-77/EXP3/10
[32] Troyon F. et al., Plasma Phys. Controlled Fusion 26, 209 (1984)
[33] Lazarus E.A. et al., Phys. Fluids B4, 3644 (1992)
[34] Hofmann F. et al., Nucl. Fusion 28, 399 (1998)
[35] Hofmann F. et al., Phys. Rev. Lett. 81, 2918 (1998)
[36] Martin Y. et al, Proc 18th IAEA Fusion Energy Conference (2000) IAEA-CN-77/EXP5/30
[37] Hofmann F. et al, 27th EPS Conf. Contr. Fusion and Plasma Phys., Budapest, 2000, P1.029
[38] Pitts R.A. et al., Journ. of Nucl. Mater. 290-293 (2001, March issue)
26
[39] Loarte A. et al., Nucl. Fusion 38 (1998) 331
[40] M. Keilhacker et al., Phys. Scr. T2/2 (1982) 443
[41] Pitts R.A. et al., Proc 18th IAEA Fusion Energy Conference (2000) IAEA-CN-77/EXP4/23
[42] R. K. Janev, T. Kato, J. G. Wang, Phys. Plasmas, 7 (2000) 4364
27