Edge particle transport and heat flux in
stationary high-confinement
ELM-free regimes
Alcator
A. E. Hubbard1,
C-Mod
With thanks for contributions from
M. Fenstermacher2, A. Garofalo3, S.P. Gerhardt4, R. Maingi4,
D. G. Whyte1, K. Burrell3, A. Diallo4, T.E. Evans3, B.A. Grierson4,
J. W. Hughes1, C.J. Lasnier2, B. LaBombard1, G. McKee5, R. Nazikian4,
D. Orlov3, T.H. Osborne3, T. Petrie3,J. Rice1, W. M. Solomon4, P. Snyder3,
J. L. Terry1 J. Walk1, S. M. Wolfe1, A. Wingen3
1MIT Plasma Science and Fusion Center
2Lawrence Livermore National Laboratory,
3General Atomics, 4Princeton Plasma Physics Laboratory
5Univ. Wisconsin
21st International Conference on Plasma Surface Interactions
Kanazawa, Japan, May 28, 2014
Edge particle transport and heat flux in stationary
high-confinement ELM-free regimes
• Introduction: Why are these regimes needed?
• Candidate high-confinement regimes without ELMs.
– H-mode with Resonant Magnetic Perturbations.
– QH-mode.
– I-mode.
– Enhanced Pedestal H-mode.
– Not covered: ELM mitigation by pellets (Talk by L. Baylor),
vertical jogs, SMBI, high * regimes such as EDA, Type V…
• Pedestal physics and core performance
(Mainly drawing on US results, as part of 2013 Joint Research
which focused on comparing physics mechanisms).
• Heat and particle flux to boundary.
– Initial results and remaining issues.
2
A. Hubbard, PSI 2014
Stationary high-confinement regimes
without large ELMs are NEEDED for fusion!
1. L-mode energy confinement would lead to fusion reactors which are
much too large to be economic. (too bad, since L-mode regime would
make power handling easier).
2. Standard ELMy H-modes give sufficient energy confinement. BUT,
work by PSI community has clearly shown that transient heat load
due to ELMs will damage ITER PFCs. DEMO Limits will be even lower.
Evolution of tungsten samples
during 0.5 ms simulated ELM pulses
A. Loarte,
Nucl.
Fusion 54
(2014)
033007
Acceptable operating space
A. Zhitlukhin, et al., J. Nucl. Mater. 363–365 (2007) 301
A. Hubbard, PSI 2014
3
Stationary high-confinement regimes
without large ELMs are NEEDED for fusion!
1. L-mode energy confinement would lead to fusion reactors which are
much too large to be economic. (too bad, since L-mode regime would
make power handling easier).
2. Standard ELMy H-modes give sufficient energy confinement, BUT,
work by PSI community has clearly shown that transient heat load
due to ELMs will damage ITER PFCs. DEMO Limits will be even lower.
3. Standard “ELM-free” H-mode is not stationary.
Particles and impurities accumulate, leading to H-L transition.
Need some means to ‘flush’ particles, continuously.
(Or don’t confine them well in the first place)
Developing methods to mitigate ELMs, or new stationary
regimes which avoid them, is not optional.
Luckily several attractive possibilities now exist.
Focus here on ELM-suppressed and ELM-free regimes.
A. Hubbard, PSI 2014
4
Resonant Magnetic Perturbations
• Resonant Magnetic
Perturbations use 3D fields
generated by external “RMP”
coils to modify pedestal
transport and stability.
• RMP experiments have been
conducted on several
tokamaks with a range of coil
sets and conditions.
T. Evans, PSI 2012
– DIII-D, AUG, KSTAR, MAST,
NSTX, JET
• Strong mitigation, or complete
suppression, of ELMs has been
observed in many cases.
• Currently planned on ITER.
A. Hubbard, PSI 2014
E. Daly, et al., Fusion Sci. Technol. 64 (2013) 168
5
Quiescent H-mode
• The Quiescent H-Mode (QH-mode)
provides particle control through a
continuous pedestal fluctuation called
the Edge Harmonic Oscillation (EHO),
which replaces ELMs.
W. Solomon, et al, APS-DPP 2013
– Has been seen on DIII-D, ASDEX-U, JET
and JT-60U.
• Needs strong rotation shear. Originally
achieved with counter-torque NBI, but
DIII-D now also has co-torque cases.
A. Hubbard, PSI 2014
G. McKee, et al,
APS-DPP 2013
6
I-mode regime has
energy barrier but no particle/density barrier
– On C-Mod, ASDEX-Upgrade, DIII-D.
• A broader ‘weakly coherent mode’ is
seen in the T pedestal region.
• ELM-free due to lower P.
L-mode
I-Mode
L-Mode
L-mode density
4
Core Te 4 ->8 keV
0
1.00 Te,ped (keV)
0.50
0.00
T pedestal
1.5
1.0 < P > (atm)
0.5
0.0
1.5 H
ITER98y2
0.5
I-Mode
I-mode
5
4 PICRF (MW)
3
2
1
0
2.0
1.5
1.0
20 -3
0.5 ne (10 m )
0.0
8 Te(0) (keV)
1120907028 1.1 MA LSN
• Achieves H-mode like confinement
without a substantial change in density
or impurity transport.
• Obtained with unfavorable BxB drifts.
1.0
0.5
0.0
2
High pressure
0.6
0.7
0.8
0.9
1.0
1.1
1.2
High confinement
Dα
ELM-free
1
0
0.5
L-Mode
0.6
0.7
Alcator
C-Mod
0.8
0.9
time (s)
1.0
1.1
1.2
A. Hubbard, IAEA 2012
J. Walk, APS 2013
A. Hubbard, PSI 2014
7
Ti vs. Major Radius
EP H-mode
Dα [Arb.]
WTOT [kJ]
1.4
1.2
1.0
0.8
0.6
0.4
0.2
0.0
6.0
134991
4.5
Ip
NB Power/10
Dα
3.0
1.5
0.0
400
333
267
200
133
67
0
8
density [1013 cm-3]
• Transition to EP H-mode with
higher E occurs on NSTX after
normal L-H transition (often
following an ELM).
• Regions of steep Ti and toroidal
rotation shear; location can vary.
• Duration has been progressively
extended. As is typical for STs,
ne is not fully stationary.
IP, Pinj/10 [MA, MW]
Enhanced Pedestal H-mode - a Highconfinement regime in the Spherical Torus
6
Stored energy
H98=1.7
Density
EP Hmode
4
H-mode
2
0
0.0
0.2
0.4
0.6
0.8
1.0
time [s]
H-mode
R. Maingi et al,
PRL 2010
Relatively little exploration of this regime to
date due to shutdown for NSTX-U.
A. Hubbard, PSI 2014
8
Key questions for all ELM-free regimes
1. Can they be obtained in the relevant conditions for
burning plasmas? (eg *, q95, Te/Ti, n/nG, rotation…)
2. How does energy confinement compare with ELMy Hmode?
3. Is particle transport sufficient to flush impurities?
Most research and progress to date has been on these questions.
Results are generally encouraging, with variation between regimes.
4. Is regime compatible with long pulse heat flux handling?
– Relatively little attention to date.
– Results and open issues in last part of talk.
9
A. Hubbard, PSI 2014
All regimes under consideration have shown good
confinement and access to low collisionality
2.0
H98(y,2)
1.5
• Representative but
incomplete data sets
shown (from US JRT 2013).
I-Mode (DIII-D and C-Mod)
EPH-Mode (NSTX)
QH-mode (DIII-D)
RMP (DIII-D)
DIII-D RMP
1.0
DIII-D QH-Mode
NSTX EP H-Mode
I-Modes
DIII-D
C-Mod
0.5
FY-2013 FES Joint Research
Target Report
0.0
0.01
•
0.10
10.00
All regimes have demonstrated H98(y,2) >1 and compatibility with lowcollisionality pedestals * < 0.15.
‒
•
ν e,ped*
1.00
Unlike type-V ELMs (NSTX), EDA H-mode (C-Mod), HRS (JFT2M)
Wide ranges of q95 in each regime. RMP, QH-mode, and I-mode
extend to ITER-relevant q95 ~ 3.
A. Hubbard, PSI 2014
RMP ELM suppression is most effective in
certain windows of q95 and *
6.2 kA
5.2 kA
4.3 kA
T. E. Evans,
2014 USBPO seminar
• q95 ‘window’ is to be expected
given resonant conditions.
• Coil sets (and results) vary
between devices.
– For example, suppression on
AUG mainly for high *.
– DIII-D has high and low *
cases; physics may differ.
– Not yet a complete or consistent
understanding of suppression
conditions in all cases.
•
Increasing RMP coil currents (n=3 and n=1) expands the q95
suppression window by ~ 7x on DIII-D.
• ELMs can be suppressed even with some coils missing (D. Orlov, APS 2013)
• Additional experiments with multi-mode RMPs may lead to improved coil
designs with larger q95 suppression windows.
A. Hubbard, PSI 2014 => Goal is suppression for full ITER discharge.
11
In some cases (low *, strong shaping),
RMP causes density pumpout
T. E. Evans, 2014 USBPO seminar
•
•
•
RMP clearly provides strong particle transport (sometimes too strong?)
Less pumpout is seen in low , or higher *, plasmas.
Challenge is to optimize discharge, and increase density via fueling
while maintaining ELM suppression.
A. Hubbard, PSI 2014
12
Changes in pedestal height with RMP are
consistent with EPED model
P. B Snyder, PoP 19 056115 (2012)
R. Nazikian, APS 2013
D. Orlov, APS Invited 2013
• EPED model, based on combination of peeling-ballooning and
kinetic ballooning constraints, predicts limit in ELMing discharges.
– Good fit to measured pedestals in many tokamaks.
• Sets an upper bound to pressure gradient in ELM-suppressed
regimes, varying with nped. DIII-D RMP results are thus consistent.
• Can pedestals be returned to pre-RMP levels?
A. Hubbard, PSI 2014
13
In QH-mode, EHO keeps pedestal
below peeling boundary
• QH-mode operates at peeling
(current-driven) boundary.
• EHO is thought to be a
saturated peeling mode,
which keeps pedestal just
below limit without ELM crash.
– Pedestal needs to be at
peeling boundary, which
limited ne in the past.
• Stronger shaping has been shown to
increase limits, leading to progressively
higher density – and performance.
– H98 up to 1.8 has been achieved!
Osborne et al, J. Physics: Conf. Series
123, 012014 (2008).
• Calculations show ITER should always be
at the peeling boundary. (Burrell APS 2011)
A. Hubbard, PSI 2014
14
Good recent progress in raising QH mode
density
• Plasma density raised by gas
puffing into QH-mode.
• Higher absolute density
achieved with stronger shaping,
up to 0.8 nG
• Next steps will be to make QHmode discharges stationary,
while maintaining H98..
EPED again
predicts measured
pedestal pressure
trend with density.
A. Hubbard, PSI 2014
W. Solomon, et al,
APS-DPP 2013
15
I-mode regime has little power degradation, is
below pedestal stability limits
• Regime is most robust
at low *, low q95.
• H98 0.7-1.2.
• A key difference is
much less
degradation with
power.
– Suggests pedestal
not MHD-limited.
1.5
200
kJ
150
100
50
Computed using WMHD
H98,y2
WMHD
W=40 IpxPloss
1.0
I-modes
0.5
C-Mod
ITER
0.0 baseline
Alcator
C-Mod
2
3
A. Hubbard, PSI 2014
0
4
q95
5
5-6T, LSN
0
2
4
Ip(MA) x Ploss(MW)
6
6
16
I-mode regime has little power degradation, is
below pedestal stability limits
• Regime is most robust
at low *, low q95.
• H98 0.7-1.2.
• A key difference is
much less
degradation with
power.
– Suggests pedestal
not MHD-limited.
1.5
J. Walk, APS 2013, PoP 2014
Computed using WMHD
H98,y2
1.0
0.5
C-Mod
ITER
0.0 baseline
Alcator
C-Mod
2
3
A. Hubbard, PSI 2014
4
q95
5
• Wider p pedestal & L-mode density
pedestal leads to lower MHD.
• Explains lack of ELMs or energy
saturation.
6
• Pressure limit is not set by
stability, room to increase.
17
Upper bound to input power in I-mode is
set by I-H transitions
FY-2013 FES
Joint
Research
Target Report
• I-Mode thresholds, confinement and pedestals are being jointly
studied on C-Mod, DIII-D, AUG (ITPA). To be reported at IAEA 2014.
− Regime has been obtained with ICRH, ECH, co-and counter NBI.
• I-H threshold can be very high (up to 5 MW, 2x L-I) on C-Mod, but
so far is lower on other devices, leading to lower H98.
– Very recent C-Mod results indicate strong BT dependence of I-H
threshold and I-mode power window (best results are at 5-6 T).
A. Hubbard, PSI 2014
18
Changes in pedestal turbulence are associated with
I-Mode particle and energy transport
400
reflectometer
300
300
Te,edge
χ
eff
L-mode
1091203020
0.8
(m 2/s) 0.6
0.4
0.2
– Correlated with LCFS , consistent
with a key role in driving particle
transport.
Te/Te ~1% < ne/ne 10%.
• A low-f GAM (k=0 mode) is also
seen in v.
200
100
100
(keV/m) 80
60
40
20
0
• Weakly Coherent Mode appears at
higher frequency
Fluctuations
60-150 kHz (a.u.)
I-mode
200
Hubbard
PoP 2011
WCM
100
H
0.0
1.00 1.05 1.10 1.15 1.20 1.25
time(s)
A. Hubbard, PSI 2014
Alcator
C-Mod
Frequency (kHz)
Frequency (kHz)
• Edge eff correlates well
to the drop in mid-f
turbulence (~60-150 kHz)
as T pedestal forms
20
0
kθ=0
mode
-6 -4 -2 0 2
kθ (cm-1)
4
6
I Cziegler, PoP 2013
19
Enhanced Pedestal H-mode
features strong increase in ion energy
– Ion thermal transport drops to
~neoclassical.
– H98 up to 1.7 (~1.4 typical).
• Ti gradient scales with rotation
gradient, location varies.
Example of enhanced thermal and
particle gradients shifted inward
1.5 g)
[keV]
• ~75% of energy increase is in ion
channel.
133841
t=0.560000
•
Conditions for transition, extrapolation to
other devices?
Sufficient electron and impurity transport
to be stationary?
• Awaits experiments on NSTX-U,
other devices.
A. Hubbard, PSI 2014
0.0
1.25
10
1.30
1.35
1.40
i)
1.45
1.50
ne
10xnC
8
[1013 cm-3]
•
1.0
0.5
– Is rotation shear suppressing
residual turbulence?
• Regime little explored to date, many
questions remain, eg.
Ti
Te
6
4
2
0
-2
1.25
1.30
1.35 1.40
R [m]
1.45
1.50
S.P. Gerhardt, et al.,
submitted to Nuclear Fusion (2014)
20
ELM-free regimes can exhaust impurities as well
as or better than Type 1 ELMs
•
Confinement p obtained by puffing or
injecting impurities, measuring decay.
I-mode: Much smaller P than
H-mode, close to L-Mode.
200
Density
B.A. Grierson, et
al., submitted to
Nuclear Fusion
D
150
ELMy
Fluorine
Alcator
C-Mod
τp
(ms)
(Injected Ca)
EHO
EDA H-mode
100
QH-mode: EHO gives slightly
smaller P than ELMy H-mode.
50
0
0.0
I-mode
L-mode
0.5
H98
1.0
1.5
RMP fields can have complex effects on
impurities, see presentations by Kobayashi and
Briesemeister this meeting.
A. Hubbard, PSI 2014
All low P operating scenarios have the
important advantages of low core
radiation, compatibility with seeding
and metal walls.
Key questions for all ELM-free regimes
1. Can they be obtained in the relevant conditions for burning
plasmas? (eg *, q95, Te/Ti, n/nG, rotation…)
2. How does energy confinement compare with ELMy H-mode?
3. Is particle transport sufficient to flush impurities?
• Much recent progress on these issues:
− Operating spaces have been greatly expanded.
− Low energy and high particle transport are favourable – though the
mechanism(s) for their separation are still not clear.
− At least RMP H-mode, QH-mode and I-mode appear promising
candidates from a pedestal and core perspective; no known ‘show
stoppers’ for burning plasma. EP H-mode requires more study.
• Key research issues do remain. For example:
− RMP ELM suppression: Suppression conditions? Reduce drop in pped.
− QH-mode: Can edge rotational shear be achieved with ITER tools?
− I-mode: Can I-H transitions be avoided at high enough power?
4. Is regime compatible with long pulse heat flux handling?
A. Hubbard, PSI 2014
22
Plasma-wall issues for ELM-free regimes
• ELM-free regimes share the major PWI advantage that heat load is
continuous, not pulsed. BUT, even the steady heat flux q// will,
as with ELMy H-mode, be a major challenge.
Potential issues for the new
regimes:
• How wide is the heat flux
footprint? Where does it peak?
ITER
Eich, et al., NF 53
(2013) 093031
• Higher particle transport
naturally favours lower density.
Can density be increased for
optimal power handling?
• Is regime compatible with
radiative divertor, detachment?
23
A. Hubbard, PSI 2014
RMPs: Strike point splitting seen for e* > 1
•
Divertor heat flux
profile splits into
distinct peaks
during e* ≥ 1
ELM mitigation
– Amplitude of
peaks evolves
with constant
I-coil current
T. Evans, et al., J. Phy. Conf. Ser.
7 (2005) 174
• Tsurf is greatly
reduced vs ELMy
H-mode.
A. Hubbard, PSI 2014
24
RMPs: Heat and particle fluxes differ for low e*
•
•
•
For e* < 0.35, get much less strike point splitting.
Heat flux q and particle flux (D, or CII) can be quite
different.
•
Striation of D agrees well
with magnetic connection
length footprints.
•
Heat flux often has single
peak, agrees with modeling
including ion drifts.
ELM suppression gives strong
reduction in peak heat flux.
A. Hubbard, PSI 2014
A. Wingen, Phys. Plasmas 21 (2014)
25
RMPs: Detached divertor has been
obtained, but not yet with ELM suppression
•
Combined neutral D2 and Ar
gas flow increase e* above
ELM suppression threshold
(e* ~ 0.35)
– ELMs return prior to divertor
detachment. More
experiments are needed.
A. Hubbard, PSI 2014
T. Petrie, et al., Nucl. Fusion 51 (2011) 073003
26
I-mode: Divertor heat flux largest on inner leg
• Because regime is obtained with reversed
drifts & flows, peak heat flux shifts from
outer to inner leg.
r
Bx∇B
r
Bx∇B
1101209014
1110309024
Alcator
C-Mod
J. Terry, PSI 2012
J. Nucl. Mat. 438 (2013)
• Can put heat on atypical surfaces which are
less well measured.
– Very recently, I-mode was obtained near-DN
(Rsep -1.5 mm).
A. Hubbard, PSI 2014
I-mode: Heat flux width is comparable to H-mode
SOL (~Eich q) similar to or slightly larger in I-mode than in H-mode,
may reflect larger upstream pressure gradient scale lengths.
•
Possibly less Ip dependence, but measurements needed in a wider range.
• This means that, as for H-mode,
heat flux is a concern.
• Impurity seeding (Ne) has often
been used to reduce surface
temp.
mm
Alcator
C-Mod
– Works well due to low p.
– Near-term plans include
systematic seeding scans, assess
detachment.
J. Terry, PSI 2012
J. Nucl. Mat. 438 (2013)
A. Hubbard, PSI 2014
Heat flux in QH-mode
•
While it is not essential for the regime, many
QH-mode experiments are also in unfavorable
drift configuration, ie. More heat to ISP.
In early experiments with low ne, and counter
NBI, an anomalous heat flux was observed on
upper baffles, attributed to high energy ions
seen in SOL.
– Fast ions, both from pedestal and NB, are
likely affected by strong Er well, and EHO.
Upper divertor heat flux
4
3
OSP
ISP
(a)
Heat flux, MW/m2
•
1.5 s (ELMs)
3.0 s (QH)
anomalous peak
reflection
2
1
#110849
0
1.0
1.2
1.4
1.6
Major Radius (m)
1.8
2
1.5
Heat and particle fluxes are modulated by EHO.
• Recent success in extending QH-mode to
co-NBI, and to higher ne with gas puffs,
should be very beneficial for power handling.
0.5
– Detachment experiments are planned this year.
A. Hubbard, PSI 2014
C. Lasnier, PSI 2002
J. Nucl. Mat. 313 (2003)
Upper divertor heat flux
1.5 s (ELMs)
3.0 s (QH)
Langmuir Probes
1.0
Z (m)
•
(b)
r
Bx∇B
#110849
0.0
1.0
1.5
2.0
Major Radius (m)
2.5
29
For all ELM-free regimes, fueling is a promising
route to higher density
I-Mode (C-Mod)
QH-Mode (DIII-D)
I-Mode
4
0
2.5
PRF (MW)
1120907032
ne(1020m-3 )
2.0
1.5
Gas Fuelling ON
O
1.0
1.00 T
e,ped (keV)
Alcator
C-Mod
W. Solomon,
et al, APSDPP 2013
0.50
0.00
200
WMHD (kJ)
100
0
0.6
0.6
0.7
0.7
0.8
0.8
0.9
0.9
1.0
1.0
1.1
1.1
1.2
1.3
Frequency (kHz)
reflectometer (88 GHz, r/a~0.98)
400
300
n~e
200
100
0.6
0.7
0.8
A. Hubbard, PSI 2014
0.9
1.0
Time (s)
1.1
1.2
1.3
• What are limits to density?
• How does heat flux respond?
• Can we reach detachment
while maintaining good
properties?
30
Good progress in developing stationary
high-confinement ELM-free regimes
• Stationary high-confinement regimes, without large ELMs, are
essential for fusion. (Clear from prior studies by PSI community.)
• Several promising techniques and regimes now exist which appear
applicable to burning plasmas. These include:
o Suppression with RMP coils, planned for ITER. Engineering is
challenging especially for FNSF or DEMO.
o QH-mode and I-mode, which naturally provide high particle
transport without ELMs. Candidates for ITER scenarios.
o Enhanced Pedestal H-mode has very high energy confinement,
needs more assessment.
• RMP, QH–mode and I-mode all have pedestal fluctuations which
have larger effects on particle than heat fluxes. Why?
• These new regimes have had much less study of boundary heat
and particle fluxes than Type I ELMy H-mode.
o Increased effort by the PSI community is urgently needed to
integrate ELM-free regimes with boundary solutions!
A. Hubbard, PSI 2014