�������
���
H-mode pedestal and threshold studies over an expanded operating space on
Alcator C-ModPresented by Amanda Hubbard
With Contributions from J. Hughes, I. Bespamyatnov*, T. Biewer#, E. Edlund,
M. Greenwald, B. LaBombard, L. Lin, Y. Lin, R. McDermott, M. Porkolab, J. Rice, W. Rowan*, J. Snipes, J. Terry,
S. Wolfe, S. Wukitchand the Alcator C-Mod Group
MIT Plasma Science and Fusion Center*Univ. Texas Fusion Research Center
#Oak Ridge National LaboratoryResearch supported by U.S. Dept. of Energy
APS-DPP Meeting, Philadelphia31 October 2006
�������
���OUTLINE
• Introduction – Overview of H-mode regimes and
prior pedestal results
• Influence of magnetic field B on– L-H threshold– Pedestal scalings– H-mode operating space, regimes
• Role of field and current direction– Review of results on configuration
dependence– Flows and L-H thresholds with
reversed field, current – Edge profile evolution in L-mode– Pedestals with reversed field
• Summary
Motivation:H-mode profiles are ‘stiff’; Stored energy W proportional to pe,ped, across all the operating space discussed.
0.0 0.5 1.0 1.5 2.0pe,95 (1023eV m-3)
0
50
100
150
200
250
Sto
red
Ene
rgy
(kJ)
Reversed BT
BT=7-8 T
�������
���Several H-mode regimes obtained on C-Mod
ELM-free H-mode• Low particle
transport, transient H-modes.
(a)
(b)
(c)
(d)
(e)
(f)
(g)
(h)
(i)
(j)
(k)
(l)
(m)
(n)
(o)
Enhanced D-Alpha (EDA)
• Quasicoherent mode leads to steady ne.
• Favored by higher q, ν*.
Small (“Type II”?) ELMs• Small ELMs on top of high Dα.• Occurs at higher pressure than
EDA.• Also gives steady ne.
Discrete ELMs(not shown)
• Recent observation, will be focus of talk by J. Terry, thissession.
�������
���
High-resolution diagnostics measure pedestal electron profiles and ionization rates
• Key diagnostic is Edge TS– Top view, 1.5 mm δR,
16 ms δt.– ECE used for faster Te.
• SOL profiles from scanning probes.
• Dα profiles from camera enable derivation of neutral and ionization rate profiles, key to fueling and nepedestal.– Find LD ≤ Lne <λion, λCX
• Diffusivity Deff derived from source, ne profiles. – Find Deff “well” in
pedestal, decreasing with higher Ip.
Edge Thomson + scanning probes
Inverted Dαbrightness from tangential camera
nD, Sioncomputed from aboveIonization
rate
Neutraldensity
D-alphaemissivity
Electrondensity
Electrontemperature
-10 0R - RLCFS (mm)
-20 10 20
0 1020
m-3
123
keV
0.0
0.2
0.4
kW
m-3
02
4
m-3
1016 1017 1018
0.0
0.2
0.4
1024
m-3s
-1
�������
���Key prior pedestal scalings (mainly at B=5.4 T)
Hughes, PoP 2006,IAEA 2006
• Strong correlation of nped with Ip, weaker dependence on neutral source.– Gas puffing has little effect in strong barriers, ie high Ip H-modes. – SOL largely opaque to neutrals in these cases.
• Narrow ne, Te, pe pedestal widths (∆ ~2-5 mm). – Little systematic variation with Ip, ne etc.
Base casePuffed with ~15 torr-L
n e (10
20m
-3)
-10 0R - RLCFS (mm)
-15 -5 -10 0R - RLCFS (mm)
-5 -10 0R - RLCFS (mm)
-5 -10 0R - RLCFS (mm)
-5
�������
���
• Strong correlation of nped with Ip, weaker dependence on neutral source.– Gas puffing has little effect in strong barriers, ie high Ip H-modes. – SOL largely opaque to neutrals in these cases.
• Narrow ne, Te, pe pedestal widths (∆ ~2-5 mm). – Little systematic variation with Ip, ne etc.
1 10 1000
2
4
6
8
10
12
ν*
αMHD
Key prior pedestal scalings (mainly at B=5.4 T)
• pped, ∇pe scale with Ip2.– Soft limit with higher αMHD at lower
ν*.– Suggests critical gradient
behaviour in pedestal as in SOL.
EDA H-modes
Hughes, PoP 2006,IAEA 2006
Base casePuffed with ~15 torr-L
n e (10
20m
-3)
-10 0R - RLCFS (mm)
-15 -5 -10 0R - RLCFS (mm)
-5 -10 0R - RLCFS (mm)
-5 -10 0R - RLCFS (mm)
-5
�������
���
2005-6 campaigns expanded H-mode parameter space
• C-Mod uses exclusively RF heating, primarily 6 MW ICRF.
• BT range for near-central heating constrained by fRF.
• 2.6 < BT < 8 T enabled by variable fRF (50-80 MHz), and D(He3) as well as D(H) heating.
• 0.4 < Ip <1.7 MA, 2.6 < q95 < 9.5– Can now better separate Ip,
q dependences.
• Also a 2006 mini-campaign with reversed Ip, BT at 5.4 T, 0.8 MA.
ITER BT5.3 T
q 95 ~
10
q 95 ~
7
q 95 ~
5
q 95 ~ 3.5
q95 ~ 2.5
Effective range of priorscaling studies
Recent extensions to parameter space
Pedestal data obtainedNormal B dir. Reversed B
0.0Ip (MA)
0.5 1.5 2.01.0
BT (T)
0
2
4
6
8
10
�������
���As expected, L-H thresholds increase with BT
• Total power thresholds for L-H transition are 2.7-5 MW for 8 T, vs typically 1-2 MW for 5.4 T.
• Edge Te is also substantially higher at transition, ~ 300-450 eV vs100-200 eV.– Higher edge Te, lower ν*
at L-H transition likely affects the n-T trajectory of the following high field H-modes.
0 2 4 6 8 100
2
4
6
Pth
resh
(M
W)
L-H Power threshold
5.4 T, 0.8 MA
7.9 T, 0.9-1.3 MA
0 2 4 6 8 10BT (T)
0
200
400
600
Te,
95 (
eV)
Te,95 at L-H
�������
���
0.0 1.0 2.0 3.0
n (10 m )20 -3
e,ped
0
2
4
6
8
10∆ (mm)
n0.46 MA
0.77 MA
0.98 MA
Extended scalings confirm pedestal widths are insensitive to BT, Ip
• Bulk of pressure width data ~2-5 mm over 2.6-8 T, 0.4-1.7 MA.– Rules out ρpol, ρtor scalings of pedestal width.
0 2 4 6 8 10Toroidal field B (T)
0
2
4
6
8
10P
ress
. Ped
esta
l Wid
th (
mm
)
0.0 0.5 1.0 1.5 2.0Plasma Current Ip (MA)
0
2
4
6
8
10
Pre
ss. P
edes
tal W
idth
(m
m)
• Exception is at lowest Ip(and nped) where ∆ increases.– Correlated with deeper
neutral penetration depth LDand may represent a fueling effect as on lower n devices.
∆p vs BT ∆p vs Ip
BT =5.4 T
�������
���
High B H-modes have same pedestal pressure as lower field cases, but higher Tped
• At given Ip, pressure pedestal profiles are independent of B, ie. αMHD~const. (even though most discharges do not have evident MHD).
• Stored energy, τE are also the same, since profiles remain “stiff”.
• Balance between T, n shifts.– High Te at L-H may contribute.
High Bnped lower
Tped higher
pe(r)~same
0
1
2
3
4
ne (
10
20 m
-3)
5.4T, 1.2 MA7.9T, 1.2 MA
0.0
0.2
0.4
0.6
T e (eV
)
-10 -5 0 5R - R
LCFS (mm)
0
5
10
15
20p
e (k
Pa)
1050624022
1060426015
0 1 2 3 4Density Pedestal (1020m-3)
0
200
400
600
Te
pede
stal
(eV
)
BT ~ 5.4 TBT ~ 8 T
p=constIp=1.2 MA
�������
���Higher B discharges therefore have lower ν*ped
• 7.9 T H-modes tend to be shorter, with ne rising.
• Most to date are ELM-free, not EDA.– These cases had lower ν*
and/or lower αMHD than typical EDA.
• Some discharges at high α, q95 had weak quasi-coherent modes, but not steady ne.– These have low ν* but
higher αMHD , continuing prior trends and extending EDA operation space.
ν*ped computed taking Zeff=1, lower bound
0 2 4 6 8 10ν*ped
0
5
10
15
20
α MH
D
BT 4.5-6 T, EDA
BT~8T, weak QC
BT~8T, ELM-free
�������
���
0.0 0.5 1.0 1.5 2.0Plasma Current Ip (MA)
0
1
2
3
4
Ped
esta
l Den
sity
(10
20m
-3)
BT 2.6-6.3 T
BT 7.5-8 T
nped is still linked to current. But, at high B, scales better with q95.
0 2 4 6 8 10q95
0
1
2
3
4
Ped
esta
l Den
sity
(10
20m
-3)
BT 5-6.3 TBT 7.5-8 T
�������
���
0 2 4 6 8 10q95
0
1
2
3
4
Ped
esta
l Den
sity
(10
20m
-3)
BT 5-6.3 TBT 7.5-8 T BT 2.6-4 T
0.0 0.5 1.0 1.5 2.0Plasma Current Ip (MA)
0
1
2
3
4
Ped
esta
l Den
sity
(10
20m
-3)
BT 2.6-6.3 T
BT 7.5-8 T
nped is still linked to current. But, at high B, scales better with q95.
The key result in all B ranges is that for high n and strong transport barriers, nped is not an easily controlled independent variable.– Largely determined by target plasma n, I, B.– Target n is constrained when available power is near Pthresh.– Low gas fueling efficiency, related to opaque SOL. – Transport sets ne gradient.– Potential issues for ITER H-mode scenarios.
Low B npeddata do not fit q95scaling; clearly more complex.
�������
���
Influence of Magnetic Field Direction and Configuration
�������
���SOL flows appear linked to the changes in
L-H threshold with configuration
• Well-known that L-H power threshold is ~2x HIGHER with Bx∇B drift away from active x-point. Tedge is also higher.
• Prior C-Mod experiments found strong parallel flows in inner SOL, reversing with USN vs LSN configuration. These affect core toroidal rotation Vtor(0).
• In unfavorable (USN) case, starts with more counter-current rotation, apparently further from L-H threshold conditions.– In all cases, flows and Vtor
increment co-current with increasing power and pressure. (no ext torque)
• L-H transition occurs at similar rotation values in each case, but requires more power in USN than LSN.
– Likely linked to differences in Er, shear.
-40
0
40
1 2 3 4
Core Rotation
SOL, near separatrix
1 2 3 4
0
20
Total Input Power (MW)
1 4
Toro
idal
Vel
ocity
(km
s-1
)
Lower NullUpper Null
L H(at transition)
Results consistent with SOL flows causing the differences in Pthresh with configuration (not necessarily the transition itself).
LaBombard, Nucl. Fus. 2004, PoP 2005
�������
���
Reversing B and Ip removes ambiguities in comparing different magnetic configurations
C-Mod has only one (lower) “divertor” structure. This means:
• Upper tile configuration is more open than lower, not designed for high heat flux.
• LSN and USN shapes were not exactly symmetric.
Do these effects contribute to the observed differences in SOL, flows/rotation, profiles, threshold?
To find out, reversed I and B to compare in SAME configuration:
“Reverse B” has ion Bx∇B drift upward.
“Normal B” has drift downward.
C
G
D
�������
���
Key results confirmed by field reversal:Inner SOL flows are unaffected by I, B direction
Details in LaBombardtalk JO1/4Tues pm.
Parallel Flow in High-Field Side SOL2 mm outside separatrix
Para
llel F
low
Mac
h N
umbe
r
Bx B Bx B
0.5 1.0 1.5 2.0 2.5 3.0-1.0
-0.5
0.0
0.5
1.0
ne (1020 m-3)
Ip Ip
• Flow direction depends only on X-point location, NOT Bx∇B.Consistent with transport-driven flux. Similar Mach No. in forward, reversed B.
�������
���
Key results confirmed by field reversal:Inner SOL flows are unaffected by I, B direction
• But, since Ip is also reversed, flows are counter-Ip when Bx∇B is away from the X-point (‘unfavorable’), co-Ip in favorable cases.
Details in LaBombardtalk JO1/4Tues pm.
Parallel Flow in High-Field Side SOL2 mm outside separatrix
Para
llel F
low
Mac
h N
umbe
r
Bx B Bx B
0.5 1.0 1.5 2.0 2.5 3.0-1.0
-0.5
0.0
0.5
1.0
ne (1020 m-3)
Ip Ip
• Flow direction depends only on X-point location, NOT Bx∇B.Consistent with transport-driven flux. Similar Mach No. in forward, reversed B.
Unfavorablefor H-mode
�������
���
0 1 2 3-60-40-20
0204060
Ohm
ic V
tor (
km/s
)
Ohmic Vtor
0 1 2 30
1
2
3
4
Pth
resh
(M
W)
L-H P thresholdReverse BForward B
0 1 2 3n---e (1020 m-3)
0200
400
600
8001000
Te,
95 (
eV) Te,95 at L-H
Key results confirmed by field reversal:L-H Thresholds higher in Reversed B LSN
• Ohmic core rotation is more counter-Ip in reversed field LSN.– Co-Ip increment when power,
pressure increase.
• LSN power thresholds are much higher (2.7-3.7 MW) - “unfavorable”– Usual variability with wall
conditions.
• Threshold temperatures and gradients are also much higher (>400 eV), particularly near low nelimit.– Limit varies between
campaigns.
Co-IpForward B
Co-IpReverse B
LSN, 5.4 T
�������
���
0.0
89P
0
20 Vtor (0) (km/s)Vpol,edge x2
0.0
0.4
0.0
1.5
3.0
ne,95
0.8
Te , r/a=0.97 (keV) Ti , edge chord
0.65 0.70 0.75 0.80 0.85time (s)
0
2Dα
τE,L
0.65 0.70 0.75 0.80 0.85
40
HHITER89P
W
MHD(kJ)
ne (1020 m-3)
L H
ICRF on, P=3.4 MW
1
2
01
2
3
40
50
100
150
Edge Te(r) with unfavorable drift shows interesting evolution before L-H transition
• Edge Te profiles evolve on a slow time scale, 3-4 τE.
• Often a “break-in-slope” in Te(t), ∇T ~40 ms before L-H.– Two-phase H-mode transition?
• Steep Te gradients develop, beforechanges in ∇ne & Dα (the classic “L-H”) transition.
• Vtor(0) steadily reduces. – Smaller change in edge Vpol.
• Stored energy W, H-factor also increase gradually, H89P to 1.6 in L-mode.
• This L-mode evolution is NOT seen in favorable drift direction, even with high L-H thresholds (eg, 8 T).
• It is similar to behavior seen in AUG ‘Improved L-mode’ with unfavorable drifts. (Ryter, PPCF 1998).
Te break in slope
Co-Ip
�������
���
0.0
0.5
1.0
1.5
2.0
ne (
10
20m
-3)
Ohmic
Pre L-H
H-mode
0.0
0.2
0.4
0.6
Te (
ke
V)
-10 -5 0 5R - RLCFS (mm)
-300
-200
-100
0
Gra
d p
e/n
e (
ke
V/m
)
1060601029
“Pedestal” in Te develops prior to L-H transition
• Preliminary measurements from ambient B+4 spectroscopy near topof pedestal indicate that total Erdoes not change substantially until the L-H transition.– However, do not resolve the
region of steepest ∇Te.
• Te, pe gradients develop before L-H over a narrower region (~2 mm) than in later H-mode. – ∇pe/ne up to 200 keV/m!
∇pe/ne
ne
Te
CX details in Bespamyatnov, QP1/56,McDermottQP1/57Wed pm
-10 -5 0 5R - RLCFS (mm)
Er (k
V/m
)
-100
100
-200
200
0
Ohmic
Pre L-H
H-mode
Early L-mode
meas chords
Er
�������
���
Steady decrease in edge χeff is accompanied by changes in turbulent fluctuations
• Gradual decrease in magnetic fluctuations at outboard side, strongest in ~50-100 kHz band, accompanies 60% drop in edge χeff from power balance. Also a broadening, fluctuation increase at f>150 kHz; – Net decrease in integrated (5-250 kHz) during evolution is ~46% – Upshift but little change in net ne fluctuations by PCI (top view).
• Further sharp decreases in all fluctuations, and in χeff , at L-H transition.
L H
Te break in slope
condeff
e eff
P2kn T
χ =∇
0.97<ψ<1.0
PCI
Magnetics
Time(s)
B
time (s)
0.70 0.75 0.80 0.85 0.900.0
0.2
0.4
0.6
edge χ
eff (
m2/s
)
0.44 m2/s
0.8
B, 50-100 kHz band.
edge χeff
0.17
0.03 m2/s
B (a.u.).
5
10
15
20
L H
Break-
in slope
�������
���
0 20 40 60 80 100 120 (keV/m)
0
1
2
3
4
5
Q/A
ne (
10
W
m)
-15
∇Te
0.44 m2/s
χL
χLH
χH
0.17
0.03 m2/s
m2/s
Early
L-Mode
L
H-Mode
H
time trajectory + = 2 ms
Pre-LH evolution is consistent with a “soft” transition
• Edge flux-gradient plot shows gradual increase in ∇T with near-constant Q, ne, after ‘break-in-slope’, – Appears to be a ‘soft’, second
order transition, as would result from –ve dependence of χ on T or ∇T.
– Contrasts with L-H transition, which is a rapid first order bifurcation.
– Consistent with the gradual decrease in turbulence.
• Regime may help identify which modes contribute to edge transport.
• Transport phenomenon has globally similar features to the ‘Intermediate Mode’ regime seen on DIII-D but no evidence of “bursty” fluctuations or fluxes. (Colchin, PRL 2002).
• More similar to ‘Improved L-mode’ on AUG with unfavorable drifts.
• Regime might be attractive as starting point for advanced scenarios: H~1.6, but low density.
GradientFl
ux
�������
���
H-mode pedestals in unfavorable configuration have lower nped, ν*
• In fully developed H-mode, pedestals in Reverse B LSN (unfavorable drift) tend to have lower n, higher T (up to 900 eV) than Forward B LSN with similar I, B, target ne. Pedestal widths, pressures are similar.– This leads to lower collisionality pedestals, 0.25 < ν*ped <2.5
• Dimensionless pedestal space (αMHD vs ν*ped) is quite similar to Forward B 8 T H-modes. Common feature in both cases is a high power and temperature (lower ν*) at the L-H threshold. Is the initial condition determining the final operating point?
Pedestal details in Hughes QP1/44 Wed pm
QC Mode details in Czieglertalk JO1/7 Tues pm,Lin QP1/63.
0 1 2 3Density Pedestal (m-3)
0
200
400
600
800
1000
1200
Te
pede
stal
(eV
)
For BT
Rev BT
p=const
BT=5.4 T, Ip=0.8-0.9 MA
0 2 4 6 8 10ν*ped
0
5
10
15
20
α MH
D
5.4 T, 0.8 MA7.9 T Rev BT, 5.4 T
�������
���
• C-Mod H-modes studies have been extended to high field (7.9 T) and to reversed (unfavorable) field and current direction.
• In both cases, power and edge Te H-mode thresholds are increased. – Pedestal widths, pressure limits, confinement ~ same.
• With strong barrier and opaque SOL, pedestal density is largely set by target I, B, n. Target n is constrained when Pthresh high. ITER may well be similar in these regards. Parameters at the L-H threshold affect the final H-mode.– Tped is higher and nped lower at high B and Reversed B. ⇒ Lower ν*.
• Reversed field results are consistent with SOL flows affecting threshold.• In unfavorable case, strong gradients in Te develop well before L-H transition in
particle confinement. Gradual decrease in χ is accompanied by changes in fluctuations.– Regime of interest for edge transport physics, and perhaps for future use in
advanced scenarios.
• 2007 experiments will exploit new cryopump to expand low ν* operation, and extend high field H-modes to higher Ip and PRF.
Summary: H-mode studies over expanded operating space on Alcator C-Mod