G.D.Conway, 30-June-2015 : EFTSOMP (Lisbon) 1
EFTSOMP 2015 (Lisbon)
Recent GAM studies in ASDEX Upgrade
P.Simon*#, G.D.Conway, A.Biancalani, T.Happel, P.Manz*$, D.Prisiazhniuk, U.Stroth*$, and the ASDEX Upgrade Team
*Max-Planck-Institut für Plasmaphysik, Garching, Germany#IGVP, Universität Stuttgart, Germany
$Technische Universität München, Garching, Germany
● GAM parameter dependences – freq. & amp. scaling
● GAM structure & propagation
● Magnetic signature
● Impact of non-axisymmetric (resonant) magnetic perturbations MP
● GAM envelope detection – turb. interaction
G.D.Conway, 30-June-2015 : EFTSOMP (Lisbon) 2
GAM measurements from Doppler reflectometry on AUG
● Complex spectra from I/Q signal and determine Doppler peak fD & AD (using weighted average CoG or Gaussian fit)
fD = ku/2p AD ~ (dn)2
● Repeat process on sliding window to obtainfD(t) and AD(t) time series
● Power spectrum of fD(t) to find peak at fGAM
● Calculate GAM strength0 2 4 6 t (ms)
-0.6
-0.2
0.2
f D (
MH
z)
1 10 100 f (kHz)0.01
0.10
1.00
Pow
er (
kHz-
1 )
-2 -1 0 1 2 f (MHz)10-5
Pow
er S
f (a.
u.)
fD = Σ(f*Sf)/Σ(Sf)
AD10-4
10-3
10-2
10-1
A[kHz ]=2 f 1
f 2 S f D 4/1.5
f1
AGAM=2 A[ kHz ]/k ┴f2
G.D.Conway, 30-June-2015 : EFTSOMP (Lisbon) 3
GAM Frequency Scaling: κb Dependence
● Freq. scale factor G = wGAM Ro/cs : G ~ 2 for “core” (inside ped. circular kb 1) [Windsor, PF 1968]
● Edge GAMs (rpol > 0.95) show strong dependence on boundary elongation kb
● Conway empirical scaling good overall prediction of edge GAMs (especially limiter config.)
● Divertor data deviate more than limiter role of X-point?
f scale=c s
2 Ro4 [ 1
1b− o ]
1.0 1.2 1.4 1.6 1.8 2.0Boundary elongation kb
0
1
2
3
G =
wG
AM
R0/
c s
LimiterDivertor
√2
GConway
LimiterDivertor
Core GAMs
[Conway, PPCF 2008]
0
5
10
15
20
25
f GA
M (
kHz)
0 10 20 30fscale (kHz)
kb<1.2
1.2< kb<1.4
1.4< kb<1.6
kb>1.6
New AUG data
5 15 25
G.D.Conway, 30-June-2015 : EFTSOMP (Lisbon) 4
GAM Frequency Scaling: Gao-scaling
● Analytic Gao scaling: Influence of k lessstrong than Conway scaling
● All experimental data lie above Gao
● Gao (linear) gives min. wGAM – non-linearity & X-point etc. may raise GAM frequency
LimiterDivertor
[Gao, PST 2011]
q=4,=0.3, '=0
1.0 1.2 1.4 1.6 1.8 2.0
LimiterDivertor
GConway
Missing AUG data
GGao
fscale (kHz)Boundary elongation kb
kb<1.2
1.2< kb<1.4
1.4< kb<1.6
kb>1.6
0 10 20 305 15 25
f scale∝ 2b21
0
1
2
3
G =
wG
AM
R0/
c s
0
5
10
15
20
25
f GA
M (
kHz)
√2
[Simon, IRW 2015]
G.D.Conway, 30-June-2015 : EFTSOMP (Lisbon) 5
GAM Amplitude: Damping dependence on k & q
● GAM amp. generally increases with q, but falls at high q
● Shape / kb dependence also present
● Stronger variation for divertor configuration
● NEMORB simulations in progress
GA
M a
mp.
(km
/s)
Damping coefficient × 10102 3 4 5 6
q_local
0.00
0.01
0.02
0.03
0.04
0.05
0.06LimiterDivertor
kb<1.2
1.2< kb<1.4
1.4< kb<1.6
kb>1.6
0.1 1.0 10.0
GA
M a
mp.
(km
/s)
0.00
0.01
0.02
0.03
0.04
0.05
0.06
● Collisionless damping – w/o krri (Finite Orbit Width) corrections [Gao, PoP 2008]:
Strong freq. dependence
Dominant at low q
Dominant at high q
q5 exp(-q2)
G.D.Conway, 30-June-2015 : EFTSOMP (Lisbon) 6
GAM Radial structure
● 2 forms of GAM radial structure:
– low kb: freq. continuum
– high kb: freq. eigenmode
● Stronger GAM at low elongation kb (lim.)
● Collisionality & other dependences under investigation
limiter discharge
Pow
er (
arb.
)
1 10 10010-8
Frequency (kHz)
10-7
10-6
10-5
0.80 0.85 0.90 0.95 1.00
5
10
15
20
25
30
Freq
uenc
y (k
Hz)
-9
-8
-7
-6
-5
kb = 1.675
10
15
20
25
30
0.80 0.85 0.90 0.95 1.00
kb = 1.12
Radius ρpol Radius ρpol
#29722 Time: 1.17-1.37s #29722 Time: 3.51-3.71s
AD
fD
1 10 100Frequency (kHz)
AD
fD
G.D.Conway, 30-June-2015 : EFTSOMP (Lisbon) 7
GAM Propagation
● 2 chn. radial corr. fD filtered around fGAM (5 – 25kHz)
● Corr. pattern inclined inward GAM radial prop. (outward prop. also seen: towards GAM peak)
● Inclination falls at high kb, eigenmode vs continuum?
● Local Sl (w,kr) spectra inward kr ~ 0.67 rad/cm-1
Del
ay D
t (m
s)
-1.0
-0.5
0.0
0.5
1.0
Δr ~ 2.3 cmlimiter discharge
#29722 @ 3.51-3.71s
-0.05
0.00
0.05
Dt ~ 15 sμvr ~1.6 km/s
kb = 1.12
0.00
-0.05
0.05
kb = 1.67
0 10 15 20 25 30Frequency (kHz)
-1.0
0
1.0
Wav
enum
ber
k r (
rad/
cm)
5
#29722 @ 1.17-1.37s
17 kHz : 0.67 rad/cmvr ~1.59 km/s
0.80 0.85 0.90 0.95 1.00Radius ρpol
0.80 0.85 0.90 0.95 1.00Radius ρpol
In
war
d
Out
war
d
G.D.Conway, 30-June-2015 : EFTSOMP (Lisbon) 8
GAM magnetic signature: Divertor discharge
● Theory indicates m = ±2 magnetic component [Wahlberg, PPCF 2009]
● Doppler: strong eigenmode GAM at 15 kHz
● No Br signal near outer mid-plane, but weak at top
● For GAM expect: Bpol > Br
● Mode analysis: m ~ 2 structureBr coils (LFS, midplane)Bpol coils (poloidal coverage)
0.1
1.0
Pow
er (
arb.
) 15 kHz
0 10 20 30 40 50Frequency (kHz)
0 10 20 30 40 50Frequency (kHz)
0.01
0.10
1.00
Pow
er (
arb.
)
0.01
0.10
Pow
er (
arb.
)
1
10
100
Pow
er (
arb.
)
#29725
15 kHz
15 kHz
G.D.Conway, 30-June-2015 : EFTSOMP (Lisbon) 9
GAM magnetic signature: Limiter discharge at low k
● Doppler: GAM freq. continuum: 15 – 20 kHz
● Magnetics: approx. m = 2 mode structure
● Tilt due to choice of reference probe
● Why different fGAM at top & bottom?
0.01
0.10
1.00
Pow
er (
arb.
)
0.001
0.010
0.100
Pow
er (
arb.
)
0 10 20 30 40 50Frequency (kHz)
0.01
0.10
Pow
er (
arb.
) #29722
1.0 1.5 2.0 2.5Radius R (m)
1.0
0.5
0.0
-0.5
-1.0
Hei
ght z
(m
)
0.5
0.0
-0.5
20 kHz
15 kHz
G.D.Conway, 30-June-2015 : EFTSOMP (Lisbon) 10
Impact of RMPs on GAM
Radius rpol
0
10
20
30
40
Fre
quen
cy (
kHz)
10-5
10-6
10-7
10-8
10 100 kHz
Pow
er (
arb.
)
Er
ne
14.6 kHz
15.9 kHz
rpol ~ 0.99
1.00 1.050.95
110 1001
GAM
2nd GAM
LFfeature
Peaksplit
#29464MP off
MP off MP on
Er spectrogram
GAM
● Without MP: Strong GAM (flow peak) inside separatrix
● With MP: Flow peak weakens & freq. increases (nb. no Te change)
● Radial max. moves closer to Er min.
● dne increases, dEr decreases
= 45°, D = 180° n = 2, sig. resonantBT = -2.5 T, Ip = 0.8 MA
q95 ~ 5.2, no = 1.5×1019 m-3
10-5
10-6
10-7
10-8
Hole
Axisymmetric GAMs
Flow: n = 0, m = 0
Pres: n = 0, m = ±1
Mag: n = 0, m = ±2 ...
- -- -
[Conway, PPCF 2015]
G.D.Conway, 30-June-2015 : EFTSOMP (Lisbon) 11
Impact of RMPs on GAM
neo (1019 m-3)
Te ped (keV)
12
14
16
18
20
Freq
uenc
y (k
Hz)
0
2
4
6
IB
Mirnov coil dbr (B31-14)
2 3 4 5 6Time (s)
-1
1
0
Cur
rent
(kA
)
1.5
0.5
1.0
0.0
● Enhanced edge magnetic “signature” above MP threshold (in both dbr & dbq)
● Non-MP GAM normally only dbq signature
0.90
MP off
MP on(0.90 kA)
AUG #29464q = 4 5 6
0.95 1.00 1.05Radius rpol
-5
0
5
10
Er (
kV/m
)
= 45°, D = 180° n = 2 MP sig. resonantBT = -2.5 T, Ip = 0.8 MA
q95 ~ 5.2, no = 1.5×1019 m-3
“Mode” extent
GAM freq.
● dbr & dbq : Complex toroidal structure
● GAM interacts with MP field non axisymmetric (n 0) GAM
● GAM reduced in stochastic regions
“Mode”(n = 2, m > 7?)
- -- -
[Conway, PPCF 2015]
G.D.Conway, 30-June-2015 : EFTSOMP (Lisbon) 12
Impact of RMPs on GAM: Magnetic signature
0 10 20 30 40 50Frequency (kHz)
15 kHz
RMP off RMP on
0.01
0.1019 kHz
19 kHzno peak
AD
fD
15.3 kHz
1 10 100 kHz
19 kHz
#27652
Pow
er (
arb.
)P
ower
(ar
b.)
0.01
0.10
0.01
0.10
Pow
er (
arb.
)P
ower
(ar
b.)
0.01
0.10
0 10 20 30 40 50Frequency (kHz)
10-8
Pow
er (
arb.
)
10-6
10-5
10-4
1 10 100 kHz
Top
Mid
10-7
10-8
10-6
10-5
10-4
10-7
AD
fD
Pow
er (
arb.
)
38 kHz
G.D.Conway, 30-June-2015 : EFTSOMP (Lisbon) 13
GAM / turbulence interaction
Coh. g between Env(dn) (PCR) & flow (DR)
● Theory GAM modulates the HF density fluctuations
● Extract flow from envelope of high-pass filtered dne using Env(n) = {nn* + H(n)H*(n)} [Nagashima, PPCF 2007]
● Correlate Env(dne) {PCR} & fD {DR} ~ 0 cross-phase at tok. mid-plane (different tor. sectors)
● Env(A){DR} & fD {DR} ~ 0.0 Expect = p /2 at top?
0.0
0.4
0.8
f D:E
nv(A
) C
oh. g
2
60
0
p
f D:E
nv(A
)
16.5 kHz
AD
fD
Env(F(A))
0 20 40Frequency (kHz)
10-8
10-6
10-5
10-7Pow
er (
arb.
)
-p
p/2
-p/2
AUG #29722
[Prisiazhniuk, IRW 2015]
G.D.Conway, 30-June-2015 : EFTSOMP (Lisbon) 14
Conclusions
● GAM frequency > Gao formular (gives min. freq.)
– fGAM raised by non-linear effects and possibly higher shaping orders (X-point)
– Still to include Zeff in scaling
● GAM amplitude
– Scales roughly inversely with damping (drive effects under investigation)
– Different behaviour for divertor config.
– Numerical simulations progressing
● GAM structure & propagation
– Either radial continuum or eigenmode (k dependence – collisionality under investigation)
– Propagates mostly inward: kr ~ 0.7 rad/cm & vr ~ 1.6 km/s (radial acceleration under invest.)
– Roughly m = 2 magnetic structure (eigenmode vs continuum)
● External MPs – strong impact
– non-axisymmetric GAM structure?
– Stochastization weakens & ev. suppresses GAM despite turb. rise
● GAM – turbulence interaction evident
G.D.Conway, 30-June-2015 : EFTSOMP (Lisbon) 15
GAM Frequency Scaling: Core vs. Edge
● GAMs contribute to effective shearing rate & reduce turb. correlation length if fGAM < td-1
● Analysis of new limiter and divertor with varying κb line-up with previous results
● Core GAMs (limiter only) follow classic scaling (even with κb scan)
● Edge GAMs deviate from core scaling
Core: ρpol < 0.95 Edge: ρpol > 0.95 f scale=2cs /2 R0
0 5 10 15 20 25fscale (kHz)
0
5
10
15
20
25
f GA
M (
kHz)
kb<1.2
1.2< kb<1.4
1.4< kb<1.6
kb>1.6
LimiterDivertor
0
5
10
15
20
25
f GA
M (
kHz)
0 5 10 15 20 25fscale (kHz)
[Winsor et al., PF 1968]
kb<1.2
1.2< kb<1.4
1.4< kb<1.6
kb>1.6
G.D.Conway, 30-June-2015 : EFTSOMP (Lisbon) 16
GAM Amplitude: Dependence on k & q
LimiterDivertor
0.00
0.01
0.02
0.03
0.04
GA
M a
mp.
(km
/s)
3.4 3.6 3.8 4.0 4.2 4.4q95
-20
-15
-10
-5
0
Dam
ping
coe
ffici
ent ×
101
0kb<1.2
1.2< kb<1.4
1.4< kb<1.6
kb>1.6
Divertor
0.06
GA
M p
.t.p.
am
p. (
km/s
)
1.0 1.2 1.4 1.6 1.8
Limiter0.04
0.02
0.00
Boundary elongation kb