erosion and confinement of tungsten in asdex upgrade · ☞ see poster exw/p7-25 (h.höhnle)...
TRANSCRIPT
23rd IAEA-FEC-Conference , 14.10.2010, Daejeon, Rep. Korea
ASDEX Upgrade
R. Dux, T.Pütterich, A. Janzer, and ASDEX Upgrade Team
Max-Planck-Institutfür Plasmaphysik
Erosion and Confinement of Tungsten
in ASDEX Upgrade
Main Advantage of Tungsten:Low Erosion Yield for Physical Sputtering
In the relevant temperature range W sputtering is mainly caused by light impurities
since impurities have • a higher mass • and a higher energy gain in the sheath (Z>1)
simple estimate of ion energy E = 2 k T + 3 Z k TB Bi e
10-5
10-4
10-3
10-2
10-1
100
W s
putte
r yi
eld
Yeff for 1% C4+Yeff4+
C4+
D+
1 10 100T (eV)
0
1
2
3
YD
/(f C
YC
)
W erosion yieldsfor sputtering by D and C
High tungsten influx expectedfor large temperatures → ELMs
23rd IAEA-FEC, 14.10.2010, Daejeon Ralph Dux
Main Disadvantage of Tungsten:Large radiative power at high temperatures
23rd IAEA-FEC, 14.10.2010, Daejeon Ralph Dux
-3
10 100T [keV]
1022
1023
n T
τ E [k
eVm
s]
c = 3.0E-5
W
c = 1
.2E-4
W
Ignition Curves with He and He+W
τ = 5τHe E
DT
+He
+W+W
• Ignition condition requests low central W concentration
• W-concentration of 3E-5 leads to an increase of the minimum triple product for self sustained burn by 20% .
• Operation window quickly closes for higher W-concentrations.
Elements Defining the Central Tungsten Density
23rd IAEA-FEC, 14.10.2010, Daejeon Ralph Dux
Central Tungsten Concentration depends on
• rate of eroded tungsten atoms at plasma facing components (PFC) - ionised very close to the surface in the scrape-off layer (SOL)
• ion transport - back to the surface
- across SOL into confined area
- in the edge transport barrier (ETB)
- inside of ETB especially in the very center of the plasma (control of neo-classical tungsten accumulation)
SOL
ETB
ASDEX Upgrade with W Plasma Facing Components
☞ see Poster EXW/P7-25 (H.Höhnle)
Measurement of Tungsten Source and Confined Tungsten Density
Source of neutral W from allmajor erosion areas• WI line at 400.9nm - heat shield: 9 lines-of-sight - outer limiters: 17 lines-of-sight - outer divertor: 12 lines-of-sight
W density in confined plasma• W quasi-continuum at 5nm emitted by ion stages existing around T≈1.5keV typ. r/a ≈0.8
1.0 1.2 1.4 2.0 2.2R[m]
-1.0
-0.5
0.0
0.5
1.0
z[m
]
# 22978 t=2.3s
23rd IAEA-FEC, 14.10.2010, Daejeon Ralph Dux
ELM resolved measurementsof total W erosion on: • inner column (∆t=1.2ms) • outboard limiters (∆t=0.5ms)
During ELMs drastic increase of W source at ouboard limiters by more than a factor of 10
mean temporal shape• rise time ≈ 0.35 ms• FWHM ≈ 0.85ms
Effect of Type-I ELMs on the W Erosion at the Limiters
2.5 2.6 2.7t (s)0
2
4
6
ΦW
1019
s-1
Total limiter source (10 LOS with ∆t=0.5ms)#25059 f_ELM = 20Hz,
ELM contrib.=72%
23rd IAEA-FEC, 14.10.2010, Daejeon Ralph Dux
-2 -1 0 1 2 3 4 5∆tELM (ms)
0
1
2
3
4
ΦW
10
19s-1
outer limiterΦW~ t/τc Exp[-t/τc]
τc=0.35ms
#25095t=1.9-2.9s
FWHM
W erosion at main chamber PFCs releases 10-20% of the W atoms eroded in the divertor.
Balance of W Erosion during ELMs: Main Chamber versus Divertor
23rd IAEA-FEC, 14.10.2010, Daejeon Ralph Dux
0 1•1017 5•1017NW,div
0
2•1016
1•1017
NW
,HS
+N
W,li
m
20%
10%
W per ELMmain chamber vs divertor
W source balance between heat shield and outboard limitersdepends on distance to separatrix
1.0 2.0R[m]
-1.0
-0.5
0.5
1.0
z[m
]
#22978 t=3.5s
∆R
limiters
heat shieldat inner column
flux surface label ∆R- distance to separatrix measured on the outer equator
Expect the first limiting surface withlowest ∆R to be the predominant W source
23rd IAEA-FEC, 14.10.2010, Daejeon Ralph Dux
• between ELMs the balance is in acordance with a parallel loss onto the first limiting element (within uncertainties of the magnetic equilibrium and the erosion measurements)
• during ELMs the balance is shifted to the outboard side (filaments)
W Source Balance between Heat Shield and Outboard Limiters
23rd IAEA-FEC, 14.10.2010, Daejeon Ralph Dux
ratio of W influx limiter/heat shieldversus radial position of
the plasma column
-2 -1 0 1 2 3 4∆Rlim-∆RHS [cm]
0.1
1.0
10.0
ΦW
,lim
/ΦW
,HS
[1]
during ELMsbetween ELMsduring ELMsbetween ELMs
10-5
cW
0.1
1
10
ΦW
[1019
s ]-1
lim.HS
div.
0
5
cm ∆Rlim ∆RHS
0
10ms
t∆ ELM
0
2
4
1022
-1 s
ΦD,puff
1 2 3 4 5t [s]
0
5
P[M
W]
NBI
ECRH
#22978radial sweeps of outer plasma radiusat two puff levels
divertor source dominant
tungsten concentration modulates with the limiter source
mean tungsten concentrationincreases with the ELM period(ELM period is controled by puff level)
Comparison of different PFCs: Source Strength and Effect on W Concentration
23rd IAEA-FEC, 14.10.2010, Daejeon Ralph Dux
Modelling of W Edge Transport with 1D Impurity Transport Code STRAHL
STRAHL• solves coupled radial transport equations for all ion stages of several impurites (AUGD: C,O,W) • ne, Te, Ti are input parameters
• anomalous transport is set ad hoc (equal for all impurities)
• neo-classical transport from NEOART (including impurity-impurity collisions)
• in SOL parallel losses to divertor and limiter are approximated by volume loss rates (strong increase of loss rate when entering limiter shadow)
• W erosion rate at limiters calculated from impurity losses onto limiters times erosion yield
• account for promptly redeposited W 23rd IAEA-FEC, 14.10.2010, Daejeon Ralph Dux
τ SOL→div
core
divertor
τ lim
limite
r
SOL
Y ΓlimW
Γlim
rlim rbndrsep0
Peaking of impurity density in ETB due to collisional inward pinch
23rd IAEA-FEC, 14.10.2010, Daejeon Ralph Dux
CXRS measurements of impurity profile evolution in ETB during ELM cycle:
• He, C, Ne, and Ar • peaking of impurity density in ETB between ELMs• peaking increases with Z• flattening of gradient during ELM
• transport coefficients D and v in accordance with collisional transport• collisionality ν* of impurities in the Pfirsch-Schlüter regime
W has even higher ⟨Z⟩ in ETB:• higher ν* and collisional diffusion coefficient• stronger peaking
1ms before ELM
1ms after ELM
ELM
10.9 ρpol 10.9 ρpol
sepa
ratr
ix
all C
8
0
C-d
ensi
ty [1
0 m
]
17-3
C6+
-200
neoclass.
-100
0
He C Ne Ar
at ρ = 0.99pol
v/D [m ]-1
☞ see Poster EXC/P3-03 (B.Kurzan)
Radial Impurity Transport Parameters during ELM Cycle
0 2 4 6 8 10∆t (ms)
0
5
10
15
20
m2 /s
Diffusion Coefficient("anomalous")
ELM off:• low turbulent diffusion coefficient in ETB• neo-classical v and D dominate
ELM on:• high diffusion coefficient around ETB• no anomalous drift• neo-classical drift unimportant
Timing:• sudden switch-on of ELM transport and linear decay within 1ms
23rd IAEA-FEC, 14.10.2010, Daejeon Ralph Dux
0.01
0.1
1
10
D [m
2 /s]
WOC
an. ELM
an. no ELM
-8 -6 -4 -2 0 2 4Rlfs-Rlfs,sep (cm)
-300-200-100
0100
v/D
[m-1
]
COW
#22895
W profile evolution during an ELM cycle
23rd IAEA-FEC, 14.10.2010, Daejeon Ralph Dux
∆t=0ms
∆t=0.09ms
∆t=0.25ms
n W [1
015m
-3]
∆t=0.53ms
∆t=1.2ms
∆t=2.2ms
0
1
2
0
1
2
0
1
2
∆t=3.9ms
Rlfs-Rsep,lfs (cm)
∆t=6.9ms
-20 -20-20 -10 0 -10 0 -10 0 5
average
#22895(fELM=100Hz)
10
1
2
∆t [ms]
[10
19s-1
]
ΦΦ(1-fredep)
0 2 4 6 80
quasi-equilbirum = ELM averaged densities are constant in time
• flattening of W density gradient in ETB during ELM
• increase of SOL density due to W erosion
• re-build of strong gradient up to next ELM
Increase of Prompt Redeposition during ELM:immediate return to the surface during the first gyration
Increase of electron density and temperature in SOL causesdecrease of ionisation length
→ increase of prompt redeposition
⊗ Bλion
rL
23rd IAEA-FEC, 14.10.2010, Daejeon Ralph Dux
W Influx and Source
0 2 4 6 8 10∆t [ms]
0
1
2
[10
19s-1
]
⟨Φ⟩=3.3x1018s-1
ELM contrib.=70.%⟨1-fprompt⟩=36.%
ΦΦ(1-fprompt)
Study dependence of confinement time of W
23rd IAEA-FEC, 14.10.2010, Daejeon Ralph Dux
=0.6-7.5 ms
=25-200 Hz
=0.1-2 m s
independent scan of
• ELM frequency:
• parallel loss time to divertor:
• diffusion coefficient in SOL:
• turbulent diffusion coefficient in ETB:
-8 -6 -4 -2 0 2 4Rlfs-Rlfs,sep (cm)
0.01
0.1
1
10
D [m
2 /s]
WOC
=0.01-0.1 m s
confinement time of W:
τ increases about linearly with• parallel loss time from SOL to divertor• ELM repetition time
very weak dependence on diffusioncoefficient in SOL• drives radial flux from the source location to the inside and to the the outside• set DSOL to 1m^2/s
Confinement time of W depends on Elm frequency and parallel loss time
23rd IAEA-FEC, 14.10.2010, Daejeon Ralph Dux
10-4 10-3 10-2 10-1
τp,reg[s]
10-4
10-3
10-2
10-1
τ p[s
]
: temporal average over ELM cycle
=0.6-7.5 ms
=25-200 Hz
=0.1-2 m s
=0.01-0.1 m s
p
ms
Increasese of ELM frequency• invoked by increase of D puffing rate• decrease of W-confinement
Code results for SOL diffusioncoefficient = 1 m^2/s
Fit of τ (c ) by adjusting τ :• peaking across ETB ↓ for f ↑ • W loss per ELM ≈10%• τ ↑ for Φ ↑• prompt redeposition about 60%
Code results for three AUG H-mode plasmas with fELM=50,100,200Hz
23rd IAEA-FEC, 14.10.2010, Daejeon Ralph Dux
0
4
8
12 τp[ms]0
2
4ΦW,lim[1018s-1]
0
2
4cW[10-5 ]
0
22 -1
1
2 ΦD,puff[10 s ]
0
4
8τSOL,div[ms]
0 100 200fELM [Hz]
0
10
20
nW,0.9/nW,sep
0 100 200fELM [Hz]
p W
SOL→div D
ELM
→ measurements of parallel losses needed
SOL→div
In ITER mucher lower collisional transport coefficients in ETBdue to higher temperatures and larger magnetic field strength
23rd IAEA-FEC, 14.10.2010, Daejeon Ralph Dux
0.0010.01
0.1
1
10
D [m
2 /s] W
ArHeO
an. ELM
an. no ELM
-30 -20 -10 0r-rsep [cm]
-60
-40
-20
0
v/D
[m-1
]
HeOArW
Assume W Limiters in ITER
Reference scenario for inductive operation (Q=10)
Collisional diffusion coefficient in ETB:• about 20 times below AUG values due to higher T and B
• W, Ar still in PS regime
• He: banana-plateau term contributes 60% at pedestal top an 10% near separatrix
apply transport model and assume turbulent diffusion in ETB to be a factor of 10 below collisional values
T =T = 4.8keVe,ped i,ped
e,pedn = 7.8 x10 m19 -3
TB = 5.3T, I =15MAp
Hec = 2%, c =0.9%, c =0.05%O Ar
In ITER low W peaking in ETB in relevant ELM frequency range
-30 -20 -10 0r-rsep (cm)
0
1
⟨nW
⟩ [10
16m
-3]
⟨Φ(1-fredep)⟩=4.4x1019s-1
5.0Hz10.0Hz16.7Hz25.0Hz33.3Hz
23rd IAEA-FEC, 14.10.2010, Daejeon Ralph Dux
NIP = 40MW, P =80MWα
heatP = 120MW, P =30MW, P =90MWrad,core sep
sep,ELMP = (1/3) P =30MWsep
ELM∆W ≤ 1MJ ⇒ f ≥ 30HzELM
For type-I ELMs:low maximum permissible ELM energy loss ∆W requires high ELM frequency
Effect on W peaking in pedestal:• time to build up strong gradient in ETB longer than typical ELM repetition time
• substantial peaking only for 5Hz if there is still sufficient ELM flushing
In ITER: global He confinement time not affected by low ETB transportat relevant range of ELM frequencies
23rd IAEA-FEC, 14.10.2010, Daejeon Ralph Dux
Exhaust of He across ETB
no divertor recycling• global confinement time limited by perp. transport accross ETB
s
with divertor recycling• realistic case but recycling model in STRAHL very crude
ELM
global confinement time of He doesnot strongly increase for f ≥10Hz.
τ SOL→div
core
divertor
limite
r
SOL
Γrec
Γlim
τ pump
τ div→SOL
0 10 20 30fELM [s-1]
05
101520253035
τ*H
e[s
]
02
4
6
8
10
1214
ρ=τ*
He/τ
E
perp. transport
+div. recycling
Conclusion
23rd IAEA-FEC, 14.10.2010, Daejeon Ralph Dux
Fast Measurements of W Erosion in Asdex Upgrade• For the divertor and the outboard limiters, the W erosion is mainly due to ELMs. • In the main chamber, the ELM erosion is mainly on the outboard limiters.
W Confinement → W Edge Transport during ELM Cycle• in-between ELMs: build-up of steep W density profile in ETB due to neo-classically dominated impurity transport• ELMs: impurity flushing from confined plasma • weak dependence of W confinement on diffusion coefficient in SOL • loss time to divertor and ELM frequency are the important parameters
ITER ETB with dominant collisional transport • combination of low collisional diffusion coefficient and high ELM frequencies leads to low impurity peaking and good He exhaust