Disruption Mitigation with

High-Pressure Gas Jets

D. Whyte1, R. Granetz2, V. Izzo2, T. Biewer2, M. Reinke2, J. Terry2,

A. Bader2, M. Bakhtiari1, T. Jernigan3, G. Wurden4

Workshop on Active MHD Control 2005

Madison, Nov. 2, 2005

1 University of Wisconsin2 MIT Plasma Science and Fusion Center3 Oak Ridge National Laboratory4 Los Alamos National Laboratory

Disruption mitigation with

high-pressure noble gas jet

High-pressure noble gas jets can mitigate 3 problems arising from

disruptions, without contaminating subsequent discharges.

1) Divertor thermal loading: sudden heat load ablates/melts divertor material

Solution: Deliver large quantities of impurity into core plasma to dissipate

high fraction of plasma energy by relatively benign, isotropic radiation

2) Halo currents: large mechanical J B forces on vessel/first wall components

Solution: Rapid thermal quench, resulting in a plasma that remains centered

in vessel during current quench, substantially reducing vessel halo currents

3) Runaway electrons: Relativistic MeV electrons from avalanche amplification

during current quench in large-scale tokamaks

Solution: Suppression by large density of bound electrons in plasma volume.

These issues are particularly severe for ITER

Outline

• New experimental results from Alcator C-Mod

• Modeling: Gas delivery and radiation

• ITER:

• Using disruptions for Tritium recovery

• Beryllium wall melting and MHD stability

Why do high-pressure noble gas jet

experiments on C-Mod?

Gas jet mitigation has been studied in DIII-D (D. Whyte et al, PRL 89, 55001)

It was postulated that the impurities penetrated as a neutral gas jet, since

Pjet (20-30 kPa) > Pplasma (8 kPa vol. avg, 30 kPa on axis)

Alcator C-Mod has:

• O(10x) higher pressure

• O(10x) higher Wth density

• O(10x) higher Wmag density

• Metallic wall

• Faster disruption timescale

– Challenging test of ability to convert

plasma energy to radiation on a fast

enough timescale (~1 GW)

Cooling front propagation into DIII-D

Specific goals of initial C-Mod experiments

• Study penetration of gas jet/impurities:

–Fast camera, Te and X-ray profiles

–NIMROD modeling, KPRAD modeling

• Disruption mitigation:

–Halo currents (current quench time, vertical displacement)

–Thermal deposition to divertor (IR camera, radiated fraction)

• Engineering/operational issues:

–Optimization of gas jet system (quantity & speed at LCFS)

–Reliability, reproducibility, post-disruption recovery

Not addressed: gas jet mitigation of actual disrupting plasma

–Gas delivery speed; realtime disruption sensing

C-Mod gas jet system optimized

based on DIII-D experience

To

vacuum

vessel

Plenum (70 bar) filled with

He, Ne, Ar, or Kr

Fast valve (ORNL)

Tokamak valve at port flange

Valve outside TF/ neutrons +

Gas delivery through ~m tube

desirable for application of

gas jets on ITER

C-Mod gas jet system optimized

based on DIII-D experience

Outlet nozzle is

extremely close to

plasma edge (2-3 cm)

to maximize gas

injected into plasma.

Nozzle is pointed at

plasma center

Injects 0.5-1.0 1023

atoms in a few ms

(plasma inventory is1.5-3.0 1020 D+, e–)

Example gas jet shot (Helium)

Gas jet valve fires at t=0.8 s

Gas jet valve fires at t=0.8 s

Example gas jet shot (Helium)

Thermal quench occurs a

few milliseconds later

n > 2x1021 m-3

Example gas jet shot (Helium)

Gas jet valve fires at t=0.8 s

Followed by current spike

and current quench

– Loss of vertical stability

– Halo currents

Thermal quench occurs a

few milliseconds later

n > 2x1021 m-3

Effect of gas jet on current quench

Typical (no gas jet) Argon gas jet

Fastercurrentquench

Less verticaldisplacement

Less halocurrent indivertor

Halo current reduction improves with Z

Fraction of energy radiated increases with Z

IR imaging of divertor surfaces

Inb

oard

Outboarddivertor

wall

Floor

Thermal deposition is not toroidally uniform,

but rather concentrated at leading edges

Temperature image from IR camera

Gas jet reduces energy deposition on

divertor surface

Temperature

differences

evident

during

cooldown

High-speed camera images indicate

shallow penetration of gas jet as neutrals

200 _s frame

Analysis of images of neutral gas

jet shows predominantly toroidal

flow, not deep radial penetration

divertor

High-Z vs. Helium penetration characteristics.

Argon HeliumSXR(au)

tTQ

0

-2

High-Z vs. Helium penetration characteristics.

Mitigation effectiveness does not seem linked to

“strong” particle penetration as found in He case.

Argon Helium

~ Uniform Tedrop beforeCurrentquench

“Weak”particlepenetrationArgon@ r/a < 0.6

Cold frontpropagates

Correlated“Strong”He particlePenetration

Central impurity density (r/a<0.9)

nAr < 1019 m-3 nHe ~ 5x1020 m-3

NIMROD simulations (V. Izzo):

MHD plays a major role for high-Z cases

A 2/1 instability destroys outer flux surfaces, 1/1 mode flattens core temperature

Fast cooling of edge region triggers MHD modes:

Therefore only shallow (r/a>0.85) impurity penetration is

required to collapse core temperature on a fast timescale.

Effect of MHD mixing? Helium case

stands out again.

RMS amplitude of MHD

fluctuations from midplane

pickup coils shown relative to

the time of the final thermal

quench

Magnitude and history very

similar for Ne, Ar, Kr.

Helium gas jet leads to earlier

MHD rise…effect of forced

pressure gradient by particle

(ion) penetration?

He

Ar Kr

Ne

Summary: Operational results

• Helium, neon, argon, and krypton gas jets used successfully

• Very reproducible effects and timing (± 0.3 ms)

• Proved to be benign – no problem with following discharge

• No runaways generated (unlike with high-Z killer pellets)

Summary: Gas jet/impurity penetration

• Impurities do not penetrate far into plasma as neutral gas

• For higher Z gas jets (Ne, Ar, Kr) NIMROD modeling shows

triggered MHD playing dominant role

• Different picture for He gas jet with good particle penetration

and different MHD characteristics

Conclusion: Since deep gas jet neutral penetration is not required,

gas jet mitigation seems plausible in ITER and reactors.

Summary: Disruption mitigation

• Halo currents are reduced by as much at 50%

–Faster current quench --> less vertical displacement

– Improves with Z of gas (higher resistivity)

• More plasma energy (~100%) is converted to benign radiation

–Less heating of divertor surfaces

–Radiated energy fraction improves with Z of gas

Conclusion: Radiated power levels are high enough to affect

energy balance on the short C-Mod disruption timescale.

Higher Z impurities, which are better radiators, are more

effective.

Future work

• Extend to higher performance plasmas

–Higher Wth: 0.11 MJ --> 0.25+ MJ (ITER energy densities)

–Higher Ip: 1 MA --> 2+ MA

• Gas jet mitigation of disrupting plasmas

–Fire into programmed VDE’s

–Gas flow rate through system may matter: possible tradeoff in

Z of gas (higher speed vs better radiator; mixed gases)

–Realtime disruption sensing and gas jet firing (ultimate goal)

• Further analysis of energy accounting (particularly for He);

Address toroidal symmetry questions; NIMROD modeling using

KPRAD, NIMROD modeling of halo current reduction; etc.

Outline

• New experimental results from Alcator C-Mod

• Modeling: Gas delivery and radiation

• ITER:

• Using disruptions for Tritium recovery

• Beryllium wall melting and MHD stability

Frictional dissipation of the gas “shock” is important for gas

delivery to plasma on disruption timescale, and will set

requirements for disruption trigger time

Realistic Volumetric Impurity

Deposition intoRadiation Code

Friction-freeshock with uniformgas delivery

Gas delivery with friction

0-D KPRAD coupled with pipe code

We have have considered acurrent decay equation inKPRAD.

dIpdt

=2 R0L

Et

Input from pipe code

Assume uniformdeposition to plasmavolume

No other free parametersfor ionization/thermal balanceKPRAD calculation.

Coupled calculations of time-dependent gas

delivery and energy / radiation balance

compared to C-Mod data

0

1

2

3

4

5

6

7

8

0 5 10 15 20 25 30 35 40

Shot No.

Tim

e (

ms)

Helium

Neon

Argon

Krypton

KPRAD

Time delay between Triggering valveAnd Beginning of Current Quench

Compared to KPRAD + Gas Flow Model

Current decay rate is reproducible &

set by gas species injected

• “Dip” in -dIp/dt at

t=0.5 - 1.5 ms is loss of

closed flux surfaces.

• Current quench rate clearly

controlled by resistivity of

impurity-dominated

plasma….

• Particle “penetration” is very

good in ~ zero Beta CQ

plasma as expected from gas

delivery.

Kr

Ar

Ne

He

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

0 5 10 15 20

Gas atomic number

L/

R d

ecay t

ime

for c

ore p

lasm

a (

ms)

KPRAD + Idealized shock

KPRAD + Frictional shock

Data

Coupled calculations of time-dependent gas delivery and

energy / radiation balance compared to C-Mod data

Physics in hand:

Gas delivery trigger needs,

Global energy balance,

CQ equilibrium Halo mitigation

/w runaway electron suppression

He Ne Ar

Current Quench L/R time Core cooling time

Physics missing:

Coupled roles of edge neutral

penetration, radiation, MHD, and

heat conduction

NIMROD + KPRAD

Outline

• New experimental results from Alcator C-Mod

• Modeling: Gas delivery and radiation

• ITER:

• Using disruptions for Tritium recovery

• Beryllium wall melting and MHD stability

Exploit rapid dissipation of plasma energy to wall

surface in order to desorb H/D/T without damaging

material surfaces.

-20

-10

0

10

20

30

40

50

-20 0 20 40 60 80 100

Stored energy at disruption (kJ)

Gas r

eco

vered

-

Gas i

np

ut

(To

rr-L

)

Plasma shots

No plasma

Energy density difference:Unmitigated disruption in C-Mod

compared to radiativeterminations in ITER.

Disruptions with closed pumpvalves for H/D particle balance

• C-Mod goal:

Control H/D ratio after vent and surface

cleaning for ICRH minority heating

– Surface analysis shows H/D ~ 5-10 in

near surface, due to H20 absorption.

– Large starting H+D inventory:

• H/(B+Mo) ~ 10-40% in first

micron(s)

• H inventory > 1000 Torr-L H2

• ITER goal:

Routine in-situ Tritium recovery

(PSI 2004)

• Expect threshold in local energy

density for removal: strong Tsurf

dependence on H diffusion in Mo/B.

– Estimate T th> 500-1000 C << Tmelt

H/D recovery by planned disruptions was

successful on C-Mod

Favorable scaling for ITER

0

2

4

6

8

10

12

14

16

18

20

H2

rem

oval

eff

icie

ncy

(To

rr-L

/ 1

5 m

in.

)

ECDC Fiducial

shotsq=2

terminations

VDE

terminations

Recovered ~30% H2 in single operation day.H/D reduced ~35%Wall H/D depleted

5-10x more efficient forH2 recovery thantypical techniques

Beryllium Melt Layer MHD

stability with disruptions

• Ideal rapid, quasi-uniform radiation flash desired for disruption mitigation can lead

to melting of the ~500 m2 Tm beryllium wall.

• Q=10 plasma termination produces a thin melt layer (~10’s microns) that remains

molten during thermal quench ~ 0.2 - 2 ms

• Because the layer melt simultaneously with the dissipation of plasma Wdia, the layer

always sees a radial inward, mobilizing JxB force resulting from induced eddy

currents by the expulsion of the diamagnetic toroidal flux.

• Surface tension acts as a stabilizing force, but for nominal tile size is insufficient to

balance JxB (~ 1mm surface “dimples” may work)

• As a result, 10’s of kg of Be will likely “splash” onto other main-wall surfaces

and divertor.

Disruption mitigation: sequence determined

by impurity “radiation instability”

• Impurities injected (gas, pellets) to dissipate

Wth by uniform radiation

• But Erad is predicted to arrive as a “burst”,

rather than uniformly in time

– Ionization instability when T < 200 eV.

– Not very dependent on mixing mechanism.

• Verified by fast bolometry on DIII-D.

Example: ITER Q=10, neon injection,

Prad spatial peaking factor =1.5 --> 50 µm thick Be melt layer over 300 m2

Energy &IonizationBalance

Be surfaceheating fromPrad

ImpurityInjectionRequired forRE control

Thermal quench

Sudden expulsion of toroidal diamagnetic flux leads to

inductive poloidal eddy currents in surrounding Be tiles that

produce radially inward JxB forces

• Relevant parameters for ITER Q=10

with Be first wall tiles.

Jdia

B

plasma

p

Inductivee.m.f.

plasma

Jtile x B

2 cm

B

2.5

cm

2.5 cm

1.5 V sToroidal Flux swing(DINA / Sugihara)

= (dtile )2 µo 4 ms >> melt

Low Be resistivity--> long skin time

= 4 10 8 m

Wall segments: ~ 106 Ltile,eddy < 10-8 H

Itile Nseg L> 100 A a

Jtile x Btor > 200 g

Prad

meltlayer

Stabilizing forces against jxB= p

Surface tension & Toroidal eddy current

• Surface tension resists change in shape of film.

• Approximate stabilizing body force by pinning film

near eddy current “corners”

• For d~50 microns, aS.T. ~ 30 g

• Stabilization can be improved by intentional

“dimpling” of original surface (like a golf ball) to

enhance capillary effect.

Eddycurrent

aJxBJtile x Btor

aS.T . dtile film

Surface tension

• The sudden inward radial movement of plasma at thermal

quench can induce toroidal eddy currents that produce radial

forces from IT x Bp

– Forces tend to be weaker since Bp ~ 10% BT

– Biggest stabilizing effect at inner midplane, de-stable at outer.

– More modeling required on poloidal dependence.

Bp

JT

Timing of Be melting and solidification critical to melt-

layer stability:

Scope Prad uniformity & Beta-Loss

1.E-06

1.E-05

1.E-04

1.E-03

1.E-02

0 0.0002 0.0004 0.0006 0.0008

Layer d

isp

lacem

en

t in

tau

_m

elt

(m

)

0.0E+00

5.0E+03

1.0E+04

1.5E+04

2.0E+04

2.5E+04

3.0E+04

0 0.0002 0.0004 0.0006 0.0008

Layer a

ccele

rati

on

(m

/s^

2)

0.0E+00

5.0E-04

1.0E-03

1.5E-03

2.0E-03

2.5E-03

0 0.0002 0.0004 0.0006 0.0008

t of final Beta collapse (s)

Melt

du

rati

on

(s)

Prad peaking

1.25 1.5

2.0 2.5

Neontermination

Melt duration increaseswith local Erad

Acceleration ismaximized whensurface “just melts”(max. surface current)

Radial displacement istypically larger than filmdepth in all conditionsUNSTABLE

Timing of Be melting and solidification critical to melt-

layer stability:

Scope Prad uniformity & Beta-Loss

Prad peaking

1.25 1.5

2.0 2.5

Neontermination

Melt thickness increasesAs Beta-drop becomesfaster

10’s of kg of molten Beare mobilized

Rdimple ~ mm stabilizesjxB movement bycapillary effect.

0.E+00

1.E-03

0 0.0002 0.0004 0.0006 0.0008

Dim

ple

siz

e

(m

)

0

10

20

30

40

50

60

0 0.0002 0.0004 0.0006 0.0008

Mo

lten

mass (

kg

)

0.E+00

5.E-05

1.E-04

0 0.0002 0.0004 0.0006 0.0008

Layer t

hic

kn

ess (

m)

ITER Summary

• Disruptions are an inevitable consequence of an experimental

tokamak.

• Controlling the timing and consequences of disruptions will be

an integral part of the operational availability and success of

ITER

– Controlled energy dissipation may help control Tritium inventory.

– Wall component viability.

– Reliable plasma breakdown and current ramp.

– Scientific aggressiveness in exploring burning plasmas.

“Shock-capturing” code developed for gas

flow down long pipes

Solves Euler’s equations for a compressible fluid.

Normalized Length along tube

Friction-free benchmark to exact solution

Friction factor dependent onReynold’s # & pipe roughness…but uncertain for transient flow

w =1

2fu u

DFriction force /wf = friction factor

Code

Data

Benchmarked “shock-capturing” code used for gas

flow down long pipes: ORNL data

f=0.07 provides good fitto measured exit pressure vs. time (inverted pressure signal)

Total gas flow rate~ 4x105 Torr-L/s ~ 1025/s

Data

Code

Gas valve pulse time (ms)S.K. Combs, et al .