Divertor Physics in Stellarators

Transcription

Divertor Physics in Stellarators
531st Heraeus Seminar, April 2013
Max-PlanckInstitut für
Plasmaphysik
Physics of stellarator divertors
Thomas Sunn Pedersen, for the W7-X team
Director of Stellarator Edge and Divertor Physics
with special contributions from
H. Hölbe, J. Boscary, T. Eich, Y. Feng, M. Hirsch, R. König
T. Sunn Pedersen
531st Heraeus Seminar
1
Outline
• 
Intro: 2D (tokamak) divertors and 3D (stellarator) divertors
• 
• 
LHD helical divertor versus Wendelstein-line island divertor
Scrape-off layer widths in tokamaks and stellarators
• 
Experience from W7-AS divertor operation
• 
Status report on the W7-X divertor(s)
• 
• 
Scraper element prototype
Summary
531st Heraeus Seminar
2
Limiter vs 2D divertor vs 3D divertor: schematic
Limiter plasma
531st Heraeus Seminar
Single null
divertor
Double null
divertor
Island
divertor
3
The tokamak divertor vs a stellarator island divertor
2D vs 3D - experiments
531st Heraeus Seminar
W7-AS
4
Stellarator island divertor principle
From J. Kißlinger, W7-X divertor design review
531st Heraeus Seminar
5
The W7-X edge topology
•  “Standard configuration”: Edge iota=1=5/5=m/n
• 
Island chain consists of five independent island bundles
•  “High iota”: edge iota=5/4.
• 
Island chain is one long bundle
•  “Low iota”: edge iota=5/6
• 
Island chain is one long bundle
•  In all three cases the magnetic shear is low and the island
are large and can be diverted
From J. Kisslinger, W7-X divertor design review
531st Heraeus Seminar
6
The W7-X divertor 3D layout
Ten identical, discrete divertor units, each aligned to the local
magnetic field (4-7 degree inclination angle in W7-X)
Simple field line diffusion model (Kisslinger) gives maximal loads of
10 MW/m2
From J. Kisslinger, W7-X divertor design review
531st Heraeus Seminar
7
LHD edge topology and helical divertor
divertor plate
ergodic layer
LCFS
divertor leg
•  Large shear makes it difficult to make large islands
•  But it allows overlap of islands with different low-order resonances è
stochastic layer is formed easily
•  Due to the existence of this stochastic layer, the quantity L|| varies strongly
from field line to field line
•  Field lines are diverged “naturally” by helical coils via an “edge surface
layer” to a helically continuous divertor relatively far away from the plasma
N. Ohyabu et al. Nuclear Fusion 34 p. 387 (1994)
531st Heraeus Seminar
8
Background: tokamak edge and divertor physics
• 
Large amount of data available from machines with a spread in size,
shape, and B-field strength
• 
Potentially problematic projections to ITER and a tokamak reactor:
•  Wetted area (next slides)
• 
• 
• 
ELM heat loads on divertors (not mentioned further in this talk)
Impurity accumulation in ELM-free H-mode
• 
Identification of various attractive regimes that might solve these
problems:
• 
Detachment, radiating mantle (solve wetted area problem)
• 
EDA H-mode, Improved L-mode, RMP’s for ELM suppression
(solve ELM problem without creating impurity accumulation )
Uncertainty in how well these attractive regimes are ITER relevant
531st Heraeus Seminar
9
Background: tokamak edge and divertor physics
• 
Large amount of data available from machines with a spread in size,
shape, and B-field strength
• 
Potentially problematic projections to ITER and a tokamak reactor:
•  Wetted area (next slides)
• 
• 
• 
ELM heat loads on divertors (not mentioned further in this talk)
Impurity accumulation in ELM-free H-mode
• 
Identification of various attractive regimes that might solve these
problems:
• 
Detachment, radiating mantle (solve wetted area problem)
• 
EDA H-mode, Improved L-mode, RMP’s for ELM suppression
(solve ELM problem without creating impurity accumulation )
Uncertainty in how well these attractive regimes are ITER relevant
531st Heraeus Seminar
10
Empirical extrapolations of λq to ITER (T. Eich et al. PRL 107 2011)
−0.8±0.1
0.1±0.1
0±0.1
λq (mm) = (0.7 ± 0.2)⋅ Btor
⋅ q1.05±0.2
⋅ PSOL
⋅ Rgeo
95
λq[mm] (exp.)
MAST
NSTX
C-Mod
AUG
DIII-D
JET
Bpol,MP [T]
λq[mm] (exp.)
−1.19
λq (mm) = (0.63± 0.08) × Bpol,MP
C-Mod
AUG
DIII-D
JET
λq[mm] (regr.)
•  Extrapolations to ITER (Bpol=1.18 T) give rather robustly λq,ITER ≤ 1mm
•  Not a “carbon PFC” effect – seen in all-metal machines as well
Why does Bp determine λq in a tokamak?
•  Heuristic model (R. Goldston, Nuclear Fusion 52 013009 (2012))
•  Magnetic drifts cause cross field transport
•  Plasma flows along the B-field at v=cs/2 towards the divertor
•  My quick and dirty version of Goldston’s model:
v∇B+Rc = vD = 2T / eBR "
$$
Tmi L||
2T 2L|| mi
L||
⇒
λ
=
v
τ
=
=
4
=
4
ρ
#
T
q
∇B+Rc ||
i
eBR
eB
R
R
τ || = 2L|| /
T
$
mi
$%
•  Note the linear proportionality between L|| and λq
•  But where does Bp come in then?
Why does Bp determine λq in a tokamak?
L||
λq = 4 ρ i
R
B
L|| ≈ a t
Bp
"
$
$
miT a Bt
miT 4a
=
# ⇒ λq = 4
eB RBp
eBp R
$
$
%
Why does Bp determine λq in a tokamak?
L||
λq = 4 ρ i
R
B
L|| ≈ a t
Bp
"
$
$
miT a Bt
miT 4a
=
# ⇒ λq = 4
eB RBp
eBp R
$
$
%
•  Main scaling is with Bp, as seen in the data.
•  T at the separatrix does not differ substantially between the tokamaks or at
different heating powers in the study (“Tsep is clamped to 50 eV due to power
balance”)
•  Even though ITER may have Tsep≈200 eV, that just brings λq up from 1 to 2
mm relative to Tsep≈50 eV
Status: stellarator edge and divertor physics
• 
Divertor data only available from a few high-performance machines,
first and foremost:
• 
• 
• 
• 
Some optimism regarding wetted area
Theoretical predictions about impurity accumulation are troubling:
• 
• 
Ambipolarity constraint leads to inward pointing E-field (ion root)
•  Electric field causes inward transport of high-Z impurities
•  Need for avoiding edge impurity sources
Identification of various attractive regimes that might solve these
problems:
• 
• 
W7-AS (until 2002)
LHD (currently active area of research)
W7-AS detachment
•  W7-AS High Density H-mode (HDH-mode)
Uncertainty in how HDH will project to W7-X (and to a stellarator
reactor)
531st Heraeus Seminar
15
Status: stellarator edge and divertor physics
• 
Divertor data only available from a few high-performance machines,
first and foremost:
• 
• 
• 
• 
Some optimism regarding wetted area
Theoretical predictions about impurity accumulation are troubling:
• 
• 
Ambipolarity constraint leads to inward pointing E-field (ion root)
•  Electric field causes inward transport of high-Z impurities
•  Need for avoiding edge impurity sources
Identification of various attractive regimes that might solve these
problems:
• 
• 
W7-AS (until 2002)
LHD (currently active area of research)
W7-AS detachment
•  W7-AS High Density H-mode (HDH-mode)
Uncertainty in how HDH will project to W7-X (and to a stellarator
reactor)
531st Heraeus Seminar
16
The W7-X divertor 3D layout
Ten identical, discrete divertor units, each aligned to the local
magnetic field (4-7 degree inclination angle in W7-X)
Simple field line diffusion model (Kisslinger) gives maximal loads of
10 MW/m2
From J. Kisslinger, W7-X divertor design review
531st Heraeus Seminar
17
Long connection length may save us
• 
An important difference to
the tokamak divertor is that
a very long connection
length is possible
• 
The long connection length
comes from the fact that the
local rotational transform
inside the island is very low
• 
Illustrated here for
CNT with visualized
field lines using an
electron beam,
originally near 100 eV,
in neutral gas
531st Heraeus Seminar
18
λq is not determined by Bp in a stellarator
•  For the picture in CNT (at least 3*7=21 transits) we get L||=21*2πR=120R (R=0.3
m in CNT)
λq = v∇B+Rc τ || = 4 ρi
L||
> 480 ρi
R
•  And we have seemingly broken the correlation between Bp and λq
•  In W7-X, we would achieve
λq = v∇B+Rc τ || = 4 ρi
L||
≥ 100 ρi ~ 4 cm
R
Status: stellarator edge and divertor physics
• 
Divertor data only available from a few high-performance machines,
first and foremost:
• 
• 
• 
• 
Some optimism regarding wetted area
Theoretical predictions about impurity accumulation are troubling:
• 
• 
W7-AS (until 2000)
LHD (currently active area of research)
Ambipolarity constraint leads to inward pointing E-field (ion root)
•  Electric field causes inward transport of high-Z impurities
•  Need for avoiding edge impurity sources
Identification of various attractive regimes that might solve these
problems:
• 
W7-AS detachment (also observed in LHD)
• 
• 
W7-AS High Density H-mode (HDH-mode)
Uncertainty in how HDH will project to W7-X (and to a stellarator
reactor)
531st Heraeus Seminar
20
Can detachment be achieved in a stellarator? In W7-X?
•  Detachment is favorable for
tokamaks and divertors
alike, since it reduces the
divertor heat loads by a
large factor (~10)
•  In W7-AS the plasma
detached from the divertor,
ie. glowing ‘clouds’ of
recombining cold plasma
appeared above the divertor
plates
•  Occurs at higher densities:
No Greenwald density limit
in stellarators – access to
high density and
detachment should be
possible in W7-X and a
stellarator reactor
531st Heraeus Seminar
21
Favorable operating mode in W7-AS: HDH
High-Density H-mode observed
near the end of the W7-AS
program
•  Plasma density high and stable.
•  Good energy confinement.
•  Impurity density low and steady due
to:
•  Reduced impurity confinement
•  Reduced impurity influx
•  Basic physics of HDH not fully
understood;
•  Will it be reached in W7-X?
•  Is it a low-temperature phenomenon?
•  (resistive-ballooning modes?)
•  Is it similar to the C-Mod EDA mode?
531st Heraeus Seminar
22
Outline
• 
Intro: 2D (tokamak) divertors and 3D (stellarator) divertors
• 
• 
LHD helical divertor versus Wendelstein-line island divertor
Scrape-off layer widths in tokamaks and stellarators
• 
Experience from W7-AS divertor operation
• 
Status report on the W7-X divertor(s)
• 
• 
Scraper element prototype
Summary
531st Heraeus Seminar
23
First divertor: Test Divertor Unit (TDU)
Baffle modules
TDU
Vertical modules
TDU
Horizontal modules
•  25m2 TDU: solid graphite tiles, no active
cooling
•  Will be installed during the first operation phase
of W7-X (OP1)
•  Same surface geometry as the steady state
water-cooled divertor, also known as the High
Heat Flux divertor (HHF)
Toroidal
divertor closure
•  No power flux limit per se – ablation may occur
once a limit of 50-100 MJ/m2 has been reached
•  Purpose: development of discharge scenarios
that do not exceed 10 MW/m2 heat flux at full
power
531st Heraeus Seminar
24
TDU Manufacturing status: Complete
•  TDU has been manufactured and delivered; installation has not yet started
531st Heraeus Seminar
25
High Heat Flux (steady state water-cooled) divertor
50 61.5
CFC
TM1H-TM4H
HHF-modules
8
Adjustable Frame
Manifold
TM5H-TM6H
Clamped LHF-modules
19
Ø9
CuCrZr
Unit
Pumping
Slid
TM7H-TM9H
HHF-modules
TM1V-TM4V
Target Elements
HHF-modules
High power Low power
Module
Total area
10 MW/m2
19 m2
1 MW/m2
6 m2
Target units (10)
Target modules
Target elements
100
890 376
20
250 end of 2013
Plasma facing mat.
CFC
Graphite
531st Heraeus Seminar
26
Testing of the HHF divertor elements
•  Prototypes of the HHF divertor
elements have been tested
extensively
•  Elements withstand 10000 cycles of
10 MW/m2 applied for 10 seconds
•  Front target (at pumping gap) can
“only” take 3 MW/m2
•  Front targets may receive stronger
loads than that during a transient
phase, due to bootstrap current
evolution
•  A so-called “scraper element” has
been proposed to prevent these
overloads
531st Heraeus Seminar
Front targets
Upside down
target showing
CuCrZr cooling
structure and
front target
27
Scraper elements
•  Scraper does eliminate overload on the target tips (front tiles)
•  The overload scenarios are only relevant for OP2 (starting 2019)
•  But we can simulate the OP2 overload scenarios already in OP1 by seeking
out special configurations (see H. Hölbe, poster session)
531st Heraeus Seminar
28
Conclusions
• 
• 
• 
• 
• 
Stellarator divertors share many similarities with tokamak
divertors
World database is much smaller – less confidence in
projecting to next step devices
There may be some special advantages to the island
divertor concept – in particular heat loads
W7-X will deliver important data on island divertor
operation and performance
•  TDU divertor is manufactured
•  HHF divertor is in series production, with excellent test
results
Please visit the W7-X posters also: (Bozhenkov,
Drewelow, Hölbe, Preynas, Rodatos, Warmer)
531st Heraeus Seminar, Bad Honnef, 2013
29
Extra slides
• 
• 
• 
• 
Why do ST’s fit the simple Goldston-like theory?
More on Goldston’s theory extended to stellarators
Magnetic equilibrium diagnostics on W7-X
Video surveillance diagnostic on W7-X
531st Heraeus Seminar, Bad Honnef, 2013
30
Why do ST’s also fit the trend?
miT 4a
λq =
eBp R
•  Many tokamaks have 4a/R~1.3
•  But MAST has much larger 4a/R~ 4… shouldn’t it have a much wider λq ?
•  Maybe not: The L|| on the outboard side is comparably short since the pitch angle
on the outboard side is much steeper than in ‘regular’ tokamaks
Courtesy of NSTX group
An ELM from MAST
λq is not determined by Bp in a stellarator
•  For the picture in CNT (at least 3*7=21 transits) we get L||=21*2πR=120R (R=0.3
m in CNT)
λq = v∇B+Rc τ || = 4 ρi
L||
> 480 ρi
R
•  And we have seemingly broken the correlation between Bp and λq
•  In W7-X, we would achieve
λq = v∇B+Rc τ || = 4 ρi
L||
≥ 100 ρi ~ 4 cm
R
•  But the logic of Goldston’s calculation is broken: When you go around the torus
poloidally and toroidally, the magnetic drifts will tend to average out.
•  The λq calculation becomes much more complicated in this case. More work
is needed. But the situation should be better
λq caused by ExB drifts (blobs)?
•  What about ExB drift in the scrape-off layer?
•  Especially in L-mode, blobs are observed in the scrape-off layer and cause perhaps even dominate - cross-field transport
•  How can Goldston’s formula be so successful if so?
•  Because of quasineutrality, |eφ|/Te is order unity and that provides a
connection between ExB and magnetic drifts:
2T
eBR
ε T / e ε1 T
= 1
=
Bε 2 R ε 2 eBR
v∇B+Rc = vD =
E Δϕ eddy
vExB = =
B Breddy
•  Similar scaling, similar amplitude (?) would lead to similar λq – again, more work
is needed
Magnetic diagnostics on W7-X
•  Diamagnetic loops
(stored energy)
•  Rogowski coils
(toroidal plasma
current)
•  Segmented
Rogowski coils
(toroidal plasma
current distribution)
•  Mirnov coils (MHD
oscillations)
Fachbeirat meeting, May 2-4 2012
34
Foldable Diamagnetic Loop
Long-pulse operation requires elimination of thermal voltages
(would lead to large integration errors)
Foldable concept avoids segmentation of loop
Test loop shown does not have ECRH protection
Fachbeirat meeting, May 2-4 2012
35
Rogowski Coils
Rogowski coils →
measure the total
toroidal electric
current
Segmented
Rogowski coils →
measure the
poloidal distribution
of the toroidal
electric current
Location → outside
and inside the
plasma vessel
Fachbeirat meeting, May 2-4 2012
36
ECRH protection for Rogowski coils
Fachbeirat meeting, May 2-4 2012
37
Video Diagnostic In-vessel View
View from AEQ Port Opposite AEQ-­‐port 38
Fachbeirat meeting, May 2-4 2012
Toroidally Viewing Video Diagnostics
Objectives:
- plasma monitoring,
-  detect excessive heat load on
plasma facing-components
-  detect hot spots
-  determine plasma shape,
edge structure
-  not a spectroscopic system
à survey diagnostic
Should be straight-forward…?
39
Fachbeirat meeting, May 2-4 2012
ECRH, plasma radiation, plasma coatings….
•  Water cooled front plate with pinhole shields
plasma and ECRH radiation
•  H2 gas flow through pin hole prevents coatings of
optical components
•  Extra ECRH absorbed using ITO coating of mirror
40
Fachbeirat meeting, May 2-4 2012