8th IAEA-TM SSO, 2015, May 26-29 Nara

A possible breakthrough of power handling by plasma shaping in tokamak

M. Kikuchi1, A. Fasoli2, T. Takizuka3, P. Diamond4, S. Medvedev5, Y.Wu6, X. Duan7, Y. Kishimoto8, K. Hanada9, M.J. Pueschel10, D. Told11, M. Furukawa12, L. Villard2, O. Sauter2, S. Coda2, B. Duval2, S. Brunner2, H. Reimerdes2, G. Merlo2, J. Jiang6, M. Wang6, M. Ni6, D. Chen6, H. Du6, W. Duan6, Y Hou6, L. Yan7, X. Song7, G. Zheng7, J. Liu7, A. Ivanov5, A. Martynov5, Y. Poshekhonov5, K. Mishra9, A. Fujisawa9, K. Nakamura9, H. Zushi9, K. Nagasaki8, K. Imadera8, Y. Ueda3, K. Kawashima1, K. Shimizu1, T. Ozeki1, H. Urano1, M. Honda1, T. Ando13, M. Kuriyama13, X. Xu14, P. Zhu15, S. Woodruff16 1Japan Atomic Energy Agency, Japan , 2CRPP-EPFL, Switzerland , 3Osaka University, Japan , 4UCSD, USA , 5Keldysh Institute of Applied Mathematics, Russia , 6Southwestern Institute of Physics, China , 6Institute of Nuclear Energy Safety Technology, CAS,China , 7Southwestern Institute of Physics, China , 8Kyoto University, Japan , 9Kyushu University, Japan , 10University of Wisconsin-Madison, USA , 11UCSD, USA , 12Tottori University, Japan , 12Max Planck Institute fur Plasma Physics, Germany , 13Retired , 14LLNL, US , 15University of Science and Technology of China, China , 16Woodruff S. Inc

Our contributions to this IAEA TM SSO, others are posters 1st NTT WS held just before IAEA TM Demo at Hefei 1

IAEA  TM  SSO  (2010)  in  Vienna

(Heuristic drift-based model of the power scrape-off width in H-mode tokamaks) alerting ELM is not only a big problem but also inter-ELM heat is quite serious. -> Since then, I have been thinking how to answer his question. Do we have solution? M. Kikuchi, WCI symposium (invited) 2012 M. Kikuchi, US-EU TTF plenary2013 M. Kikuchi, APTWG plenary 2013 S. Medvedev, IAEA FEC St Petersburg Post dead line M. Kikuchi, Festiva de Theory 2015 (Invited) M. Kikuchi, IAEA TM SSO (2015) Invited D. Chen, This coference G. Merlo, This conference

This has been published in NF 2012 and has highest citations among papers published in 2012.

R Goldston gave an interesting talk. He is a person who can smell “something”

2

1. Power handling issue in standard tokamak

7 order difference

M. Kikuchi, M. Azumi

4

Joffrin

Seki

0.1

1.0

10

100

1000

10-3s 1 year

Fossil

Fission

Fusion Divertor (even with RRC)

Fusion 1st wall He

at

Flu

x (

MW

/m2)  

~1MW/m2

~0.3MW/m2

Surface / Volume ratio is small in Fusion but large in Fission

Present Fusion power handling scenario is very challenging

RRC=Remote  RadiaFve  Cooling

Duration

w/o RRC

High  thermal  efficiency    may  be  possible  only  at  low  heat  flux!!

RadiaFon  heat  flux  on  the  Sun

5

Any energy system (Fusion) must have reliable heat exhaust scenario

•  Tokamak configuration is optimized for good confinement, but not for power handling.

[1] D-shape is good (MHD) for high pedestal pressure with H-mode (ETB), leading to large ΔW loss during ELM. Temporary measure : RMP, Pellet pacing/SMBI [2] D-shape leads to X-point toward small R region. This makes power handling more difficult. Temporary measure : Snow flake, Super X

6

1.1 DEMO power exhaust scenario (>20 years before)

660 MW heat out of plasma center

600MW to be radiated in Core and SOL/divertor

60MW to divertor plates

q=70MW/m2

q=7MW/m2

RELIABILITY/Robustness is a strong concern!! Transient excursion can easily change by large factor!! 7MW/m2 is still too large!!

Q/S=600MW/700m2 ~1MW/m2

For  3GW,  Q=50  system

7

Divertor  Plasma  Control  (Fluid  simula4on)

Albedo=0.96

ParFcle  balance  

Ion  force  balance  

Ion  energy  balance  

Electron  energy  balance  

Imp.  force  balance  

Ueda, Kikuchi, et al. NF1992 Bohm diffusion is assumed for SOL particle transport perpendicular to flux surface.

Should  be  kine-c  at  SOL  !!

8

Do we see significant progress in these 20 years? DEMO : Strong D and impurity puffs at divertor, shallow pellet at SOL

SOL transport : Sophisticated control is required to reduce q~7MW/m2 even with Bohm diffusion (L-mode)

High  Z  :  sheath  acceleraFon  (important  even  for  He)  Stable  semi-­‐detach  is  challenging        In  reactor  :  one  failure  is  serious  !!

Fe  puff  =  0.01Γp

Ueda, Kikuchi NF1992

Q=600MW Γp=2.5x1023/s

Gas puff 7Γp Imp. puff 0.01Γp

τE=1.4s τp=0.5s

Kajita, NF2009 (Top10) W nano structure 9

2.2  Heat  flux  and  radia4on  heat  flux  (Fluid  simula4on)

Ueda,  Kikuchi,  et  al.  NF1992

[1]  ConducFon  heat  flux  is  ~7MW/m2  but  total  heat  flux  including  radiaFon  heat  flux                is  up  to  12MW/m2.      [2]  RelaFvely  high  radiaFon  heat  flux  comes  from  the  strong  accumulaFon  of  impurity  near  the  divertor  plates.  Namely,  back  flow  due  to  thermal  force  is  suppressed  by  fricFon  force.  But  nFe/ne  is  only  0.5-­‐1%.  [3]  More  precise  control  of  impurity  injecFon  profile  is  necessary  to  reduce  peak  radiaFon  heat  flux.

Impurity  density  profile 10

1.2 Old and New Problems

1.2.1: ELM heat flux of H-mode ΔW ~ 20MJ at low collision – 2000th -  Why it happens : Tokamak MHD is not

designed to have soft beta limit.

1.2.2 : Power e-folding length in H-mode (2nd Goldston scaling) – 2010th - Why it happens : Fast SOL flow quickly exhaust heat within thin width for low H-mode flux.

11

Plasma  pressure

1.2.1 OLD problem : ELM (Edge Localized Mode)

Y. Liang ITER summer school 2010 12

ELM (Edge Localized Mode)

Divertor ELM energy density (MJ/m2)

A. Loarte NF2007

Measured ELM energy deposition time：　0.1-0.7ms

If ELM deposit energy in short pulse, tiles will be damaged.

No

of

EL

Ms

13

ELM ：RMP(Resonant Magnetic Perturbation) coils

T. Evans TTF2013

Many issues: "physics : lobes (homoclinic tangle) non-uniform power deposition in toroidal direction"Technologies: neutron damage of coils, localized fast ion loss etc."-> We need to develop method to eliminate ELM without RMP for DEMO.

Many tokamaks equipped with ELM control coils

14

1.2.2 Inter-ELM heat flux : Goldston 2nd scaling

Figure  (Federici,  NF2001)

Previous estimate for ITER:5mm

Recent estimate for ITER:1mm

Div  heat  flux  e-­‐folding  length  λq-­‐div    is  larger  by  flux  expansion  ra4o  for  aJached  plasma.

R.  Goldston  NF2012.  

Note:  L-­‐mode  is  governed  by  different  physics  ,  empirical  scaling  1cm  for  ITER

SOL heat flux e-folding length λq-­‐SOL  

R

1mm

5mm

λq∼ρp

15

Inter-ELM heat flux : key physics of Goldston scaling

ion electron

 (neo)classical  par-cle  transport  in  H-­‐mode

Assumed as same order <vd> l//

λ 0.5cs

✪  Grad  /curvature  B  driV  into  SOL    PSOL  is  Spitzer  thermal  conduc4on  ?  

2nd Goldston scaling(λ∼ρp )!Fast parallel SOL flow reduces λ to 1mm!!

A. Chankin NF2007: Fast parallel flow ~ 0.5Cs comes not from fluid simulation, unresolved issue.

<vd>

<vd>l// =  0.5cs  λ

16

B.  Lipschultz,  FESAC  mee4ng  July,  2012  “  Goldston  scaling  needs  more  check.”

C-Mod (Bp~BpITER) SOL e-folding

length~1mm

H-mode particle flux from separatrix ~ neoclassical drift flux. Γp

ELM free H-mode ~0.1  ΓpL-­‐mode  

Experimental result seems in agreement with Goldston scaling

17

Grad B away from X: Why SOL flow is so fast as 0.5Cs ?

Takizuka, NF2009 showed SOL flow is accelerated by both Trapped and Passing ions. Good for impurity control. Bad for SOL heat flux.

Δ

B  drij

ion

18

How about I-mode (MIT) ?

I-mode : Grad B away from X-point and need high power L -> I (H) mode High edge Te (low collisionality). L-mode like τp  but at lower edge ne. Note : Reactor needs high SOL ne.

[ NSTX Li discharge has high Te and low ne]

Trapped  ion  orbit

Takizuka  CPP2010

Whyte  NF2010 I-mode geometry has even faster SOL flow -> leads to lower edge density? 19

1.  Can we increase ΓpH-­‐mode?  

High recycling at main SOL is prohibitive (wall sputter)! Shallow pellet is still OK. 2. Can we reduce SOL flow speed? Drift across flux surface is key! If we keep fast SOL flow for impurity control, only way is to stay L-mode edge. 3.  If not, shall we kill H-mode? L-mode is best but not sufficient Explore improved confinement with L-mode edge (I-mode?). 4. High edge pedestal is good choice? Can we make soft beta limit? (High edge BSC leads to big ELM) If not, shall we reduce edge beta limit for small ELM?

Key questions :

20

2. How to design optimum configuration?

Power handing the first!

Is “ core the first” good design philosophy?

21

2.1 : Issues in present reactor design philosophy

(A) : Optimization of Core plasma

(B) : Divertor design to match (A)

(C) : consistency of (A)& (B)

D-shape/H-mode is thought as optimum for CORE. 1.  D-shape : Rdiv << Rp : bad for power handling ! 2.  H-mode : Large Pedge -> Large ELM energy loss ! 3. H-mode : Low particle flux ! 4. D shape : huge Amp Turn for “snow flake”. 5. D-shape : inboard blanket design not easy.

SSTR1990

Rp

Rdiv Level of problem : D-shaped > H-mode

22

(A) ：Configuration optimization on power handling

(1) Core to match (A)

(2) Divertor to match (A)

(3)  Integration to match (A)

(B)

2. Think different ! ‘Core the first’ is not a good design philosophy

First priority

We have rich knowledge

23

First Step : Divertor priority higher than core!

Stay hungry , Stay foolish !

A choice - negative D Make edge pedestal β limit SOFT not by finite n peeling but by Mercier/n=∞ Ballooning! Stay in L-mode edge? Find new transport reduction physics! Ex.

Reactor core is more collisionless. Optimization of TEM  

- Trapped electron precession Negative delta reduce “ Stiffening”

- S. Jobes -

24

Make power handling easier by a Large Factor: Case I

R=7m, a=2.7m (A=2.6) Standard D shape : Rx=4.3m, Negative D shape : Rx=9.7m

Factor of 2.5 for Rdiv Snow flake at Rx : Factor of 1.5 - 3

Factors : 2.5 x (1.5-3) = 4-7

Note: Outboard is much easier to install Snowflak

Negative D + Snowflake

25

Make power handling easier by a Large Factor: CaseII

R=7m, a=2.7m (A=2.6) Standard D shape : Rx=4.3m, Negative D shape : Rx=9.7m

Factor of 2.5 for Rdiv DN : Factor of 1.5 - 2

Factors : 2.5 x (1.5-2) = 4-5

Note: DN in D-shape is difficult for piping to inboard blanket.

Negative D + Double Null

26

Flux Tube Expansion Divertor (NTT-FTE diveror)

Orange: co-current Green : counter-current Blue: flux swing coil

Internal FTEcoils ~4MAT each

FTE coil pairs produce flux tube expansion by 2.7. Rp=8.5m, ap=2.4m (A~3.5) Ip=15-17MA (ss, hybrid) Bt=6.2T (Bmax~13T) For 17MA, βN=3, fGW=0.85, we have Pf=3GW.

27

Can we find a solution for power handling in Tokamak Fusion Reactor?

Divertor heat flux : w/o control : ~70MW/m2

: w control : ~10MW/m2

New proposal : negative triangularity + Snowflake / Double null or Flux tube ex. Divertor heat flux : w/o control : ~10MW/m2

: w control : ~2-3MW/m2

Quick  summary  1

Takizuka  JNM2015

28

3 MHD design 3.1 Small change in d can ELM softer (TCV) 3.2 Negative triangularity tokamak (NTT) can be stable at βN>3. ( Medvedev, Kikuchi, et al., NF2015 ) 3.3 Stability of NTT-FTE reactor

29

3.1 MHD stability of negative triangular plasma

Negative delta has higher frequency ELM.

Pochelon PFR2012

Small tilting of upper triangularity makes difference!

30

3.2 Can we achieve reactor relevant βN by NTT? Yes.

LHD magnetic hill - Watanabe NF2005 beta above Mercier - Sakakibara PPCF2008 Res. Interchange m/n=1/1 c.f. Heliotron E has problem with resistive interchange. H-J has magnetic well.

Can tokamak OK with magnetic hill? Yes, we can get βN>3 #: Mode coupled with Mercier to be internal.

βN>3 is stable" Medvedev, Kikuchi, et al. NF 2015

Magnetic Hill

31

3.3 How about Single null stability?

Axisymmetric  mode

The n=0 growth rate with aw/a=1.35, (roughly ITER case), βN=3.46":  24 /s (~4 times higher than for standard ITER configuration).

A=3.5, κ=1.75, "δ/δx = -0.5/-0.9, Ip=14.88MA, IN=1

n  \injy      3.46  1                  2.70        aw/a=1.3  -­‐-­‐>  3.11  2                  3.05  3                  3.18  4                  3.24  5                  3.26 32

SN NTT-FTE diveror has more flexible shaping

Negative/negative Zero/ Negative

33

4. Improved confinement with negative delta 4.1 Can we can get improved confinement with L-mode edge? 4.2 TEM stability 4.3 Flow shear suppression

34

4.1 improved confinement with L-mode edge?

Scenario for improved confinement in NTT - Step 1: Shape and βp optimization to stabilize TEM. Balance between MHD and TEM stabilization is important. -  Step 2: Flow shear suppression of turbulence 0-th order force balance equation to produce mean flow shear

(If TEM γ is smaller, pressure curvature may be sufficient to produce necessary E x B shear)

35

B.B.  Kadomtsev,  O.P.  Pogutse,  Rev.  Plasma  Phys.  1967,  NF  1971

4.2 TEM : Precession de-resonance is key (Kikuchi, Azumi, Rev. Mod. Phys. 2012)

Precession drift

B.B.  Kadomtsev  

36

4.2 TEM : How to stabilize TEM

Growth rate

Rosenbluth’s  maximum  J  principle  (PF1968)

Maximum J principle:

Glasser PF1974""Elongation (big)"Triangularity"Toroidal shift"are stabilizing"

37

4.2 TEM : toroidal drift has strong effect Shafranov  shiV  will  stabilize  TEM  (Connor  NF1983)

Shafranov shift can change precession drift

38

4.2 TEM : Negative δ and Shafranov shift Negative triangularity will stabilize TEM (Rewoldt PF1982)

Negative d can reduce TEM growth rate

Charge  neutral,  Ampere  law

39

Dispersion  rela4on  for  TEM/ITG  modes  in  strong  ballooning  limit.

Weiland  textbook,  2000

Wulu Zhong, 2nd APTWG Tore Supra expl.

4.2 TEM : Evidence of TEM/ITG transition

Also, J. Rice, FEC2012 bifurcation of intrinsic rotation TEM/ITG

40

4.2 : TCV negative triangularity experiment

Negative triangularity produces large Shafranov shift, which changes precession drift of trapped electron. This leads to a change in TEM stability.

Camenen NF2007

More tilted

Less tilted

Non-­‐locality  will  be  reduced  in  Reactor

Large tilting in negative delta ! Similar effect like Er’ ?

41

4.3 : Flow shear suppression

Hahm-Burrell condition:

Pressure curvature can produce E x B flow shear

42

4.3 : Shaping effect of Residual Zonal Flow (RZF)

Xiao-­‐Caqo  PoP2006,  2007 Belli,  Hammeq,  Dorland,  PoP2008

ElongaFon  increases  RZF  NegaFve  δ  may  weakly  reduces  RZF.  

Radial  profile  of  δ    -­‐  dδ/dr  is  key  to  RZF  -­‐    

Understanding  of  RZF  in  nega-ve  triangularity  (κ,-­‐δ,  Δ)  is  necessary

Xiao  PoP2007

Xiao  PoP2007

(1)

Key  is  to  reduce  NC  polariza4on

(1) GS2

GS2

NC polarization ~ (Banana width)2

Negative delta : strong outboard Bp -> smaller banana width!!

43

Summary

•  The power system should have reliable power handling but fusion power handling is challenging in divertor.

•  H-mode with D-shaping “Optimize Core choice” seems enhancing its challenge.

•  Tokamak physics is ready for new innovation. Good knowledge in core physics will make innovation possible.

•  Power handling-driven Tokamak optimization needs good core physics innovation.

•  We proposed “Negative D” as a candidate of this challenge.

•  This might be foolish idea (S. Jobes) but we need more foolish ideas to solve critical power handling issue.

44