ELM filament structure in the National Spherical Torus Experiment

R. J. MaquedaNova Photonics Inc., New Jersey

R. MaingiOak Ridge National Laboratory, Tennessee

R. E. Bell, B. P. LeBlanc, J. E. Menard, D. P. Stotler, S. J. Zweben

Princeton Plasma Physics Laboratory, New Jersey

K. TritzJohns Hopkins University, Maryland

S. A. SabbaghColumbia University, New York

and the NSTX Research Team 50th APS-DPP MeetingNov. 17-21, 2008Dallas – Texas

TI2.03

NSTXNSTX

2

Outline

• Short review of ELM filaments: experiment and modelling

• The NSTX experiment and ELM characteristics

• ELM filaments in NSTX: primary and secondary filaments*

- high speed movies

- gas puff imaging diagnostic

• Characteristics of secondary filaments

• Summary and discussion

* J. L. Terry et al., J. Nucl. Mater. 363-365 (2007) 994.

3

ELM structure and dynamics key for ITER

D. N. Hill, JNM 241-243 (1997) 182.

JET• Short-timescale loss of energy and particles

from edge plasma (H-mode transport barrier).

• Power density to plasma facing components

limits lifetime of material walls.

• ITER has adopted 0.5 MJ/m2 for the maximum

allowed energy load in 250 s.

• Yet, structure and dynamics of energy and

particle losses during ELMs in current

experiments not completelly understood.

• ELM filament structure and dynamics (this

talk) contributes to this knowledge base.

ELM: Edge Localized Mode

4

ELM evolves into filamentary structure in scrape-off layer

A. Kirk et al., PPCF 49 (2007) 1259.

MAST

• Filaments aligned with local magnetic field.

• Filaments propagate radially and toroidally.

Visible imaging

DIII-D

J. Boedo et al., JNM 337-339 (2005) 771.Reciprocating probe

ELM

5

Models show formation of filaments

• Peeling-ballooning invoked as driving mode for ELM.

DIII-DP. B. Snyder et al., PoP 12 (2005) 056115.

ELITE (linear)

BOUT (nonlinear)

pure n=20 seed “broadband n” seed-> nonlinear mode

coupling

20.1 s

36.4 s

36.8 s

17.6 s

34.2 s

37.4 s

To

roid

al a

ng

le (

1/5

toru

s)

Radius RadiusIntermediate mode number

6

Primary vs. secondary filaments

• Basic difference: origin.

• Primary filaments: high density (temperature) originating from main

instability driving the ELM.

• Secondary filaments: lower density (temperature) due to

confinement changes, secondary consequence of ELM event.

7

Some open questions regarding ELMs

• Experimental support for peeling-balloning mode comes from:

- (Some) Experimental data consistent with stability diagram.

- “Observation” of filaments during ELMs.

…are all the filaments observed during the ELM due to the driving

mode?

• Energy content in observed primary filaments only <25% of losses.

No shortage of possible mechanisms.

…what other mechanism for the energy losses can be of

importance?

8

National Spherical Torus Experiment

Center stackCarbon tiles

Typical NSTX parametersR ~ 0.85 m a ~ 0.67 m R/a ~ 1.3 ~ 2-2.3Baxis = 4.5 kG

Ip = 0.8-1.0 MA

PNBI < 7 MW

Te(0) ~ 1 keV

ne(0) ~ 6 x 1019 m-3

9

Type I and Type III ELMs in NSTX

0.2 1.40.6 1.0R (m)

1.5

1.0

0.5

0.0

-0.5

-1.0

-1.5

Z (

m)

0.30 0.23 0.240.31 0.32 0.25time (s) time (s)

0.8

0.4

0.0

60

100

140

0.2

0.6

1.0

0

200

400

0

400

800

0

400

800

0.8

0.4

0.0

60

100

140

0.2

0.6

1.0

0

200

400

0

150

300

0

150

300

Ip (MA)

WMHD (kJ)

D lower div. (a.u.)

D midplane (a.u.)

“Bolometer” SOL (a.u.)

“Bolometer” Ped (a.u.)

USXR hdown #13

USXR hdown #11 USXR hdown #13

USXR hdown #14

Type I (4.8 MW NBI) Type III (2.0 MW NBI)

124664 (0.30 s)124667 (0.24 s)

• ELM “types” assigned based on scaling with PNBI and <ne>

10

Type I and Type III ELMS in NSTX (cont.)

freq. (Hz)

W Wtot

W Wped

e*

Type I Type III

~120 ~460

~6% ~2%

~18% ~5%

~1.3 ~1.3

~0.8ne

ng

~0.5

ne (1019 m-3)

Te (keV)

R - Rsep (m)

MPTS

MPTS

• Region of interest extends from ~3 cm inside separatrix to far SOL

Notes: 1) Parameter space broader than indicated for both ELM types.2) Collisionality generally higher for Type III ELMs.

ELMs in NSTX: R. Maingi et al., Nucl. Fusion 45 (2005) 1066.

Inter-ELM profiles

11

Type I ELM Type III ELM

tELM = 250.729 mstELM = 305.607 ms

No interference filter~120000 frames/s

ELM filamentationPrimary filaments

• Type III ELMs usually have rotating structure (low n) before onset of primary filaments.

Wide angle view

• Primary filaments are “bright” filaments present during early onset of ELM event.

t - tELM (ms)t - tELM (ms)

Ed

ge

D (

a.u

.)

Ed

ge

D (

a.u

.)

12

Primary filaments show no periodic structure

0 50 100t - tELM (s)

0

2

4

6

8

Nu

mb

er o

f fi

lam

ents

Type I ELMs

1246

64

• Filaments aligned with magnetic field.

• Filaments evolve at different times with no periodic structure.

• Increasing number of filaments with time.

• Non-linear coupling of unstable modes appears to be of relevance for primary filament development.

-16

s

-8

s

0

s

+9

s

+1

7 s

+2

5 s

1246

64

tELM = 347.641 ms

Field lines atRsep = 9 cm

13

ELM filamentationSecondary filaments

tELM = 305.607 ms

• Secondary filaments seen over

an extended period of time

after the main ELM event.

Are the secondary filaments

just “more visible” due to an

increased neutral density due

to the ELM?

Wid

e an

gle

vie

wN

o i

nte

rfer

ence

fil

ter

Ed

ge

vie

w –

lo

w f

ield

si

de

, ~

mid

pla

ne

D f

ilte

r

24 c

m

Type I ELM

t - tELM (ms)

Ed

ge

D (

a.u

.)

separatrix

limiter shadow

14

ELM filamentationGas Puff Imaging

Wid

e an

gle

vie

wN

o i

nte

rfer

ence

fil

ter

Ed

ge

vie

w –

lo

w f

ield

si

de

, ~

mid

pla

ne

D f

ilte

r

24 c

m

Type I ELM

tELM = 384.790 ms

D2

puff

• Locally increase neutral

density above ELM level.

• Diagnostic puff injected in

poloidal plane within field of

view.

• Gas puff does not perturb

plasma nor filaments (blobs).

• Degas 2 simulation of D

emission (D. Stotler) indicates

overall structure is preserved,

despite non-linear

dependence on ne and Te.

separatrix

limiter shadow

t - tELM (ms)

Ed

ge

D (

a.u

.)

15

ELM filamentationGas Puff Imaging (cont.)

tELM = 262.593 ms

• Secondary ELM filaments

observed independent of increased

neutral density in scrape-off layer

due to ELM.

Wid

e an

gle

vie

wN

o i

nte

rfer

ence

fil

ter

Ed

ge

vie

w –

lo

w f

ield

si

de

, ~

mid

pla

ne

D f

ilte

r

24 c

m

t - tELM (ms)

Ed

ge

D (

a.u

.)

Type III ELM

D2

puff

separatrix

limiter shadow

16

0-50-100t - tELM (s)

0

0.2

0.4

0.6

I/I

-150-200

Rel. fluctuation level(RMS)

0-50-100t - tELM (s)

-150-200

k po

l (cm

-1)

8

6

4

2

Poloidal spectrum

1246

67

0 s +8 s

-17 s -8 s

-33 s -25 s

-50 s -41 s Type III ELMs

1246

67D

2 p

uff

, D

f

ilte

r

tELM = 267.320 ms

24 cm

24

cm

Primary filament formation similar to that seen in simulations

• Perturbation growth observed in RMS fluctuation level and poloidal FFT spectrum.

• FFT spectrum shows increase in the 2 cm-1 range, broadband.

• Non-linear coupling of many modes may be important.

17

t - tELM (ms)

Type III ELMsType I ELMs

Primary filaments

Secondary filaments

Ion diamagnetic drift(plasma rotation)

v r (

km/s

)v p

ol (

km/s

)Primary filaments move faster (radially)

than secondary filaments

• Primary ELM filaments are distinguished by their higher radial velocity, reaching ~8 km/s.

• Within edge field-of-view, primary filaments seen only during early

ELM stages ( t - tELM < 50 s ).

• Secondary ELM filaments have similar characteristics to turbulent

blobs (vr ~ 1 km/s and vpol

predominantely in ion diamagnetic drift direction)*.

* S. J. Zweben et al., Nucl. Fus. 44 (2004) 134.

Velocities at R – Rsep = 2.5 cm

18

I /<

I >

inte

r-E

LM

Type III

Type I

SOL (R – Rsep = 5.5 cm)

Timescales for scrape-off layer activity different between ELM types

• Early scrape-off layer activity (D

emission) higher for Type III ELMs.

• Activity maintained for ~300 s for Type I

ELMs.

• Characteristic decay times longer for

Type I ELMs compared to Type III (300

s vs. 100 s).

• Emission composed of fine structured

filaments with poloidal auto-correlation

lengths of ~4 cm …similar to that seen in

turbulent blobs.

FWHM

Lp

ol (

cm)

t - tELM (ms)

19

Secondary ELM filamentation similar to L-mode turbulence and blobs

• FFT amplitudes increase without the appearance of modes, neither in time nor in poloidal spatial domains.

• Broad spectra similar to edge turbulence and “blobs”. [S. J. Zweben et al., Nucl. Fus. 44 (2004) 134].

505

FF

T a

mp

l. (r

el.) ELM

Inter-ELM

50.5

FF

T a

mp

l. (r

el.)

0.2

20

2

SOL (R – Rsep = 5.5 cm)

Type I ELM

10

1

Frequency (kHz)

kpol (cm-1)

20

Evaluation of density (and temperature) within filament

• Following work in J. R. Myra et al., Phys. Plasmas 13 (2006) #092509.

• The D image intensity (I) is given by:

I = no A L* F(ne,Te)

where no is the neutral deuterium density

A is the radiative decay rate

L* is the line of sight integration length (and calibration factor)

F(ne,Te) is the population ratio for the emitting energy levels, obtained from

Degas2 modeling (D. Stotler)

• The product (no A L*) is assumed constant in time, even as filaments move by.

• Assume filament convects plasma from its birth place, within the filament:

Te = Te(ne) (pedestal profile)

• Unique pair of ne and Te values obtained for each emitting filament.

21

ne

(1019

m-3)

R - Rsep (m)

MPTS

0 1 2 3 4 5

4

3

2

1

0

ne (1019 m-3)

v r (

km/s

)

Type I ELMs

Radial velocities present continuum between primary and secondary (blob) filaments

secondary

primary• The radial velocity of the filaments,

both primary and secondary (blobs),

have a positive scaling with their

density (and brightness).

• The radial velocities present a

continuum, with no break separating

primary from secondary filaments.

• Primary originate from top part of

pedestal, while secondary originate

from foot of pedestal.

Inter-ELM

22

0.5 1 5 10blob size â

0 2 4 6 8

v rex

p (

km/s

)

secondary

primary

4

3

2

1

0

10.0

1.00

0.10

0.01

vminth (km/s)

colli

sio

nal

ity

resistive balloning resistive

X-point

sheath-connected

2//14107.1

e

e

T

Ln

5/25/2//

5/15/4

018.0ˆe

b

TL

RBaa

22

2/310

min 104.1Ba

Tv

b

eth

Secondary filament velocity consistent with blob model

J. R. Myra et al., Phys. Plasmas 13 (2006) #092509

• Filaments marginally in sheath-connected regime.

• Radial velocity of secondary filaments consistent with predictions from

blob model vrexp ~ .

• Primary filaments (high Te) have lower velocities than predicted by model.

22

2/310

min 104.1Ba

Tv

b

eth

Sheath-connected regime

23

Summary

• Primary and secondary filaments during ELM event are

distinguished by their timing, velocity, Te and ne. ↔ Origin

• Secondary filaments have all the characteristics of turbulence born

blobs: velocities, time and poloidal spectra, birth location, etc.

• Formation of primary filaments qualitatively consistent with non-

linear evolution predicted by models. Coupling of unstable modes

an important aspect of evolution.

24

Discussion

• Study of modes driving the ELM made difficult by non-linear

evolution. “Filament counting” may be complicated by the

interaction between modes and presence of secondary filaments.

• Other experimental data (DIII-D, C-Mod, JET, ASDEX, TCV, etc)

also point to the presence of secondary filaments during ELM

event. These secondary filaments (L-mode edge phase)

represent an energy and particle loss mechanism that needs to be

considered as possible loss channel.