llnl-pres-641496 resistive-mhd simulations of coaxial helicity … · 2013. 8. 16. · this work...

22
This work was performed under the auspices of the U.S. Department of Energy by Lawrence Livermore National Laboratory under Contract DE-AC52-07NA27344 LLNL-PRES-641496 Resistive-MHD simulations of Coaxial Helicity Injection (CHI) in NSTX Bick Hooper, LLNL NIMROD Team Meeting Boulder, CO August 13-15, 2013

Upload: others

Post on 02-Sep-2021

1 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: LLNL-PRES-641496 Resistive-MHD simulations of Coaxial Helicity … · 2013. 8. 16. · This work was performed under the auspices of the U.S. Department of Energy by Lawrence Livermore

This work was performed under the auspices of the U.S. Department of Energy by Lawrence Livermore National Laboratory under Contract DE-AC52-07NA27344"

LLNL-PRES-641496"

Resistive-MHD simulations of Coaxial Helicity Injection (CHI) in

NSTX !

Bick Hooper, LLNL

NIMROD Team Meeting"Boulder, CO"

August 13-15, 2013"

Page 2: LLNL-PRES-641496 Resistive-MHD simulations of Coaxial Helicity … · 2013. 8. 16. · This work was performed under the auspices of the U.S. Department of Energy by Lawrence Livermore

Acknowledgements

This work was done in collaboration with: Carl Sovinec, University of Washington Roger Raman, University of Washington and NSTX Fatima Ebrahimi, PPPL and University of New Hampshire Jon Menard, PPPL

2  

Page 3: LLNL-PRES-641496 Resistive-MHD simulations of Coaxial Helicity … · 2013. 8. 16. · This work was performed under the auspices of the U.S. Department of Energy by Lawrence Livermore

Comparing simulations with experiment

There are some differences between the simulation results and experiment: • The injector current decays more slowly in the

simulation than in the experiment • The volume of close flux is large but smaller than

experiment I will discuss the differences at the end of the talk

3  

Page 4: LLNL-PRES-641496 Resistive-MHD simulations of Coaxial Helicity … · 2013. 8. 16. · This work was performed under the auspices of the U.S. Department of Energy by Lawrence Livermore

NSTX: Shot shows expanding flux bubble!

See R. Raman, et al. Phys. Rev. Letters 97, 175002 (2006)"

4  

Page 5: LLNL-PRES-641496 Resistive-MHD simulations of Coaxial Helicity … · 2013. 8. 16. · This work was performed under the auspices of the U.S. Department of Energy by Lawrence Livermore

Boundary conditions for helicity injection!• Rate-of-change of net toroidal flux –– equals Vinj –

Vabs

• Absorber voltage –– determined by requiring the total vacuum toroidal flux to be constant, corresponding to a constant ITF

• Bϕ – Bϕ,vac << Bϕ,vac à Vinj – Vabs << Vinj

• Discharge (injector) current –– measured by the change in RBϕ just above the injector slot

• Toroidal flux –– carried in by ExB flow at the injector and out by ExB flow at the absorber

• Equating flows of vacuum toroidal flux yields

!

Erabs = Er

inj rinj,minrabs,min

drB"r2B2rinj ,min

rinj ,max

#

drB"r2B2rabs,min

rabs,max

#

This generalizes the model used in HIT-II: R.A. Bayliss, C.R. Sovinec, and A.J. Redd, Phys. Plasmas 18, 094502 (2011).

5  

Page 6: LLNL-PRES-641496 Resistive-MHD simulations of Coaxial Helicity … · 2013. 8. 16. · This work was performed under the auspices of the U.S. Department of Energy by Lawrence Livermore

Simulation –– A narrow injection slot narrows the injected current!

Injection slot!

Finite elements generate additional resolution!

A narrow slot was found to be important in the experiment –– guided the simulations!

Current flow in flux bubble!

The narrow slot generates a narrow distribution of injected current"

Vertical current inside radius R at the bottom of the flux conserver!Wide slot (11 cm)! Narrow slot (4 cm)!

6  

Page 7: LLNL-PRES-641496 Resistive-MHD simulations of Coaxial Helicity … · 2013. 8. 16. · This work was performed under the auspices of the U.S. Department of Energy by Lawrence Livermore

Absorber Coils –– not used in these simulations

ISSUE:    The  absorber  slot  boundary  condi7on  requires  an  outward  EXB  flow    

Experiment:    Poloidal-­‐field  coils  shield  absorber  slot  from  plasma    Simula7on:    ExB  out-­‐flow  at  the  extractor  slot  ––  generates  plasma  flows  in  plasma  outside  the  flux  bubble    •  Absorber  coils  energized  ––  Problem  as  the  bubble  approaches  NSTX  top    

–  Flows  forced  to  small  R  by  absorber  magne7c  fields  ––  cross  B-­‐field  to  reach  absorber  

–  Flows  interact  with  bubble  –  Result  –  Current  not  fully  localized  in  surface  layer    

•  Absorber  coils  not  energized  ––  Flow  reaches  absorber  slot  along  B    

–  Negligible  interac7on  with  bubble  –  Injector  and  toroidal  currents  largely  localized  in  the  bubble  –  Experimental  problems  with  breakdown  do  not  occur  in  simula7ons        

Conclusion:    No  absorber  coils  in  present  simula7ons  

7  

Page 8: LLNL-PRES-641496 Resistive-MHD simulations of Coaxial Helicity … · 2013. 8. 16. · This work was performed under the auspices of the U.S. Department of Energy by Lawrence Livermore

Two simulations –– used in the present study

hi-­‐temp    High-­‐temperature,  “high  quality”  discharge  without  radia7on  cooling  and  with  a  low  thermal  conduc7vity  across  the  mag.  field    •  Used  to  show  plasma  evolu7on  under  good  condi7ons  

 low-­‐temp    Low-­‐temperature  discharge  with  radia7on  cooling  

 •  Used  to  study  reconnec7on  and  X-­‐point  forma7on  physics  

8  

Discharge  currents  and  voltages  

hi-­‐temp   low-­‐temp  

Temperature  profiles  at  z  =  0  

Note:    The  glitches  shortly  aVer  9  ms  are  when  the  X-­‐point  forms  

t  =  9  ms  

Page 9: LLNL-PRES-641496 Resistive-MHD simulations of Coaxial Helicity … · 2013. 8. 16. · This work was performed under the auspices of the U.S. Department of Energy by Lawrence Livermore

Flux-surface evolution – hi-temp – similar to exp.

9  

7.0  ms  

8.5  ms  

9.7  ms  

7.0  ms  

10.0  ms  

8.5  ms  

13.1  ms  

Note  X-­‐point  forma7on  at  boXom  of  NSTX  

Page 10: LLNL-PRES-641496 Resistive-MHD simulations of Coaxial Helicity … · 2013. 8. 16. · This work was performed under the auspices of the U.S. Department of Energy by Lawrence Livermore

low-temp –– “simpler” flux configuration at 9 ms Use to analyze X-point formation

10  

hi-­‐temp  has  two  shallow  (<0.1  mWb)  regions  of  closed  flux  •  High  temperature    

–  small  magneNc  diffusion  coefficient    –  current  is  well  localized  to  the  surface  

of  the  flux  bubble  •  Flat  poloidal  flux  ––  allows  some  flux  

surface  closure    low-­‐temp  forms  a  finger  of  flux  above  the  injector  slot  ––  easier  analysis  

Page 11: LLNL-PRES-641496 Resistive-MHD simulations of Coaxial Helicity … · 2013. 8. 16. · This work was performed under the auspices of the U.S. Department of Energy by Lawrence Livermore

X-point forms by 9.120 ms

11  

t = 9.006 ms ! t = 9.060 ms ! t = 9.120 ms ! t = 9.216 ms !

ReconnecNon  is  underway  by  9.060  ms  

Page 12: LLNL-PRES-641496 Resistive-MHD simulations of Coaxial Helicity … · 2013. 8. 16. · This work was performed under the auspices of the U.S. Department of Energy by Lawrence Livermore

Flux-surface formation correlates with flow reduction from injector slot

12  

Poincáre  puncture  plots  

Poloidal  flow  vectors  

9.060  ms  

9.006  ms    

•  No  closed  flux  surfaces    

•  Flow  from  the  injector  slot  supports  the  flux  “bubble”    

9.060  ms    

•  Closed  flux  surfaces  at  large  z  

 

•  Injector  voltage  and  ExB  velocity  are  small  

 

•  Li`le  flow  from  the  injector  slot  

9.006  ms  

Page 13: LLNL-PRES-641496 Resistive-MHD simulations of Coaxial Helicity … · 2013. 8. 16. · This work was performed under the auspices of the U.S. Department of Energy by Lawrence Livermore

Temperatures are low in flux plume –– flux closure heats plasma locally

13  

t = 9.006 ms! t = 9.060 ms!

Magnetic diffusion!!!!!!!Diffusion distance in 50 µs is!!!!Distances are similar to those in the plume!

Dm = 411 Te3/2

! 5 m2 s at 20 eV! 40 m2 /s at 5 eV

!x " 0.01 m at 20 eV" 0.04 m at 5 eV

5.4 eV!

15.3 eV!

4 eV!

12.1 eV!

20.3 eV!

25.5 eV!

Page 14: LLNL-PRES-641496 Resistive-MHD simulations of Coaxial Helicity … · 2013. 8. 16. · This work was performed under the auspices of the U.S. Department of Energy by Lawrence Livermore

Magnetic pressure analysis

14  

! pB2

2" B p #!B p "

B!2

rr̂

$

%&&

'

())= 0

B p !"Br # 0

!!r2B!0"B! +"B!

2 + Br2 + Bz

2( ) ! "22B!0"B! +"B!

2

r

!W ! 2B"0!B" +!B"2 + Br

2 + Bz2 " #4

B"0!B" +!B"2 2

r$ dr + const.

The  evolving  configura7on  is  in  quasi  magne7c-­‐force  balance  

The    magne7c  curvature  in  the  “finger”  is  small  un7l  the  X-­‐point  forms  

Set                                                  ––  The  terms  in              cancel  so  B! = B!0 +"B! B! 02

Integrate  over  r  to  get  twice  the  magne7c  pressure  due  to  the  plasma  

Page 15: LLNL-PRES-641496 Resistive-MHD simulations of Coaxial Helicity … · 2013. 8. 16. · This work was performed under the auspices of the U.S. Department of Energy by Lawrence Livermore

δBϕ at 9.006 ms (before X-point formation starts)

15  

“Background”  from  injected  current  near  the  surface  of  the  stretched  poloidal  flux.    

Scales  approx.  as  1/r      ContribuNon  from  currents  driven  by  the  plasma  flow  in  the  flux-­‐finger  

Page 16: LLNL-PRES-641496 Resistive-MHD simulations of Coaxial Helicity … · 2013. 8. 16. · This work was performed under the auspices of the U.S. Department of Energy by Lawrence Livermore

Plasma-magnetic fields before and during reconnection !

16  

9.006  ms  

9.060  ms  

•  δBϕ  drops  as  the  flow  from  the  slot  ceases    

•  The  width  of  the  “flux-­‐finger”  decreases  as  the  poloidal  field  reconnects  at  the  null  line  

Page 17: LLNL-PRES-641496 Resistive-MHD simulations of Coaxial Helicity … · 2013. 8. 16. · This work was performed under the auspices of the U.S. Department of Energy by Lawrence Livermore

Magnetic pressure before and during reconnection !

17  

9.006  ms  

9.060  ms  

•  Before  reconnecNon  the  magneNc  pressure  varies  smoothly  through  the  flux-­‐finger  

       •  During  reconnecNon  the  magneNc  

pressure  adjacent  to  the  poloidal-­‐field  null  exceeds  pressure  at  the  null,  driving  flow  and  field-­‐lines  into  the  null  

Page 18: LLNL-PRES-641496 Resistive-MHD simulations of Coaxial Helicity … · 2013. 8. 16. · This work was performed under the auspices of the U.S. Department of Energy by Lawrence Livermore

Magnetic fields following poloidal-field reconnection !

18  

9.216  ms  

•  The  radial  “cut”  is  below  the  X-­‐point  and  thus  in  the  private-­‐flux  region  

•  The  fields  vary  smoothly  and  there  is  no  poloidal-­‐field  null  •  The  curvature  is  not  negligible  ager  the  X-­‐point  forms    Conclusions    •  X-­‐point  formaNon  is  “triggered”  by  the  drop  of  toroidal  flux  carried  by  the  

flow  from  the    injector  slot  •  The  reconnecNon  is  axisymmetric  and  thus  characterized  as  “Sweet-­‐Parker”  

Page 19: LLNL-PRES-641496 Resistive-MHD simulations of Coaxial Helicity … · 2013. 8. 16. · This work was performed under the auspices of the U.S. Department of Energy by Lawrence Livermore

Following flux-closure – current evolution in hi-temp!

19  

10.0  ms  

13.1  ms  

•  Following  closure  the  toroidal  current  confined  in  the  closed  flux  is  ≈  13%.    It  increases  with  Nme  =>  100%  at          ≈  13.5  ms  

 

•  With  constant  boundary  flux  the  enclosed  current  remains  constant  ––  the  ohmic  drive  due  to  the  changing  poloidal  flux  is  important  

 

•  The  injector  current  goes  to  zero  at  the  same  Nme  as  the  enclosed  current  =  100%  

 

•  Ager  ≈  13.5  ms  the  closed  flux  is  “leaning”  on  the  central  column  

Page 20: LLNL-PRES-641496 Resistive-MHD simulations of Coaxial Helicity … · 2013. 8. 16. · This work was performed under the auspices of the U.S. Department of Energy by Lawrence Livermore

The enclosed net toroidal flux = net injected toroidal flux at the same time !

20  

•  The  net  toroidal  flux  injecNon  rate  =    Vinj  –  Vabs    <<    Vinj    

 •  The  toroidal  flux  outside  the  closed  flux  =>  0  when  the  injector  current  =  0  

Page 21: LLNL-PRES-641496 Resistive-MHD simulations of Coaxial Helicity … · 2013. 8. 16. · This work was performed under the auspices of the U.S. Department of Energy by Lawrence Livermore

Remaining differences with experiment include!

21  

Injector  current    

•  Experiment:  Iinj  =  0  at  t  =  11.5  ms  (Itor  =  154  kA)  Itor  =  106  kA  at  13.5  ms    •  SimulaNon  Iinj  =  0  at  t  =13.5  ms      Itor  =  118  kA  at  13.5  ms    Closed  volume  at  9  ms    

•  Vol(exp.)  >  Vol(sim.)  •  Experiment:  Rmax  ≈  1.5  m  at  midplane  •  SimulaNon:  Rmax  ≈  1.3  m  at  midplane    Density  spaNal  distribuNon    

•  Experiment:  High  density  in  current  channel,  low  density  inside  flux  bubble  •  SimulaNon:  Uniform  density        ImpuriNes    

•  Experiment:  Important  •  SimulaNon:  Included  but  model  needs  improvement    Absorber  coils    

•  Experiment:  On  •  SimulaNon:  Off  

Page 22: LLNL-PRES-641496 Resistive-MHD simulations of Coaxial Helicity … · 2013. 8. 16. · This work was performed under the auspices of the U.S. Department of Energy by Lawrence Livermore

Summary!

22  

•  Coaxial  Helicity  Injec7on  in  NSTX  reproduces  many  of  features  of  the  experiment  

 –  Flux-­‐bubble  forma7on  and  general  characteris7cs  –  X-­‐point  forma7on  and  flux  surface  closure  soon  aVer  reduc7on  in  injector  voltage  

–  Long-­‐lived  closed  flux  surfaces  suitable  for  start-­‐up  plasmas    

•  Several  differences  remain  which  will  be  addressed  in  the  coming  months  

   The  next  goal  of  the  study  is  to  beXer  reproduce  the  laboratory  results