Upload
sydnee
View
19
Download
0
Tags:
Embed Size (px)
DESCRIPTION
Impact of Nonlocal Electron Heat Transport on the High Temperature Plasmas of LHD. N. Tamura , S. Inagaki, T. Tokuzawa, K. Tanaka, C. Michael, T. Shimozuma,S. Kubo, K. Ida, K. Itoh, D. Kalinina, S. Sudo, Y. Nagayama, K. Kawahata, A. Komori and LHD team - PowerPoint PPT Presentation
Citation preview
Impact of Nonlocal Electron Heat Transport on the High
Temperature Plasmas of LHDN. Tamura, S. Inagaki, T. Tokuzawa, K. Tanaka, C. Michael,
T. Shimozuma,S. Kubo, K. Ida, K. Itoh, D. Kalinina, S. Sudo,Y. Nagayama, K. Kawahata, A. Komori and LHD team
NIFS, National Institutes of Natural Sciences, JAPAN
21st IAEA Fusion Energy ConferenceChengdu, China
16-21 October 2006
2/10N. Tamura, FEC2006, EX/5-6, Oct. 19, 2006
Full understanding of electron heat transport is necessaryfor achieving a good predictive capability
Experiments on toroidal plasmas shows nonlocality qe(1) = f(Te(1), Te(1-), Te(1+), …, Te(1), Te(1-), …)in electron heat transport
Phase inversion ofcold pulse polarityProfile resilience Fast plasma response
(non-diffusive, ballistic)
Possible theoretical interpretation is “nonlocality” in turbulence (e.g. turbulence spreading)
Observations in LHD heliotron give new insight into nonlocal transport Because LHD has
– different magnetic configuration (normally negative magnetic shear)
– no tokamak-like stiffness in Te profile
Phase inversion ofcold pulse polarity
Recently observedin LHD
Electron heat transport can be explained only by local transport? local transport assumption:
qe(1) = f(Te(1), Te(1), ne(1), ne(1), …)
3/10N. Tamura, FEC2006, EX/5-6, Oct. 19, 2006
Significant rise of core Te in response to edge cooling in LHD
Edge cooling experiment of LHD showsa significant rise of core Te
No change in low-m MHD modesNo density peaking like PEP, RI-
mode
Electron heating dominates (Te/Ti > 1)
Difference between Te measured and that simulated based on simple diffusion model is pronounced in the core ( < 0.6)little at the edge ( > 0.6)
4/10N. Tamura, FEC2006, EX/5-6, Oct. 19, 2006
Characteristics of Nonlocal Te rise in LHD plasmas1) Condition for nonlocal Te rise
Inverse relationship between increment of core Te due to nonlocal effect and ne observedSame as in tokamaksIn LHD, no differences among
heating methods(but always electron heating)
In the LHD, the nonlocal Te rise observed in various plasmas:ECH plasma (i.e. net-current free plasma)
Toroidal plasma current and high-energy ion are not a factorNBI plasma (still Te/Ti > 1)
High-energy electron is also not a factor
5/10N. Tamura, FEC2006, EX/5-6, Oct. 19, 2006
Characteristics of Nonlocal Te rise in LHD plasmas2) Variety of time response
Larger dTe/dt Stronger edge cooling
simultaneity
Te rise delayed
ne increases
6/10N. Tamura, FEC2006, EX/5-6, Oct. 19, 2006
Characteristics of Nonlocal Te rise in LHD plasmas3) Dependence of delay time Favorable condition for delay of nonlocal Te rise
Higher collisionality b* in the coreShorter Te gradient scale length at the edge
7/10N. Tamura, FEC2006, EX/5-6, Oct. 19, 2006
Transient response analysis reveals complex relationshipbetween heat flux and Te gradient
Reduction of qe/ne is not accompanied by changes in local TeEvidence against “standard transport theory” (local & diffusive)
Turn-back of qe/ne is also independent of local Te
r
eee d
t
trTn
rtrq
0
),(
2
31 ),( Heat flux perturbation
nonlocal
local nonlocal
8/10N. Tamura, FEC2006, EX/5-6, Oct. 19, 2006
Reduction of normalized heat flux due to nonlocal effect takes place in a wider region
Region where reduction of qe/ne prominently appear is far from rapidly cooled region and strongly heated region
9/10N. Tamura, FEC2006, EX/5-6, Oct. 19, 2006
How can we understand nonlocal Te rise in LHD?
Clue from LHD experiment
Physical mechanism of long-range correlation is unclear It should have characteristics as follows:
Response delayed with higher b* & shorter Te gradient scale length
Radial extent close to a
1. Nonlocality in e-transport revealed by edge cooling
nonlocal
2. Transitions between “nonlocal” and “local” in e-transport also revealed
local nonlocal
Long-range
Short-range
Long-range
Short-range
Long-range
Short-range
rati
o
On a basis of radial correlation
10/10N. Tamura, FEC2006, EX/5-6, Oct. 19, 2006
Summary
Nonlocal Te rise invoked by edge cooling observed in low density and high temperature plasmas of LHD as well as tokamaks
New aspects of nonlocal Te rise from LHDObservation in net-current free plasmaTime response of nonlocal Te rise is variable
Time response of core Te rise is quicken by larger edge perturbation (larger Te gradient scale length?)
Delay of nonlocal Te rise increased with…increase in collisionality in the coredecrease in Te gradient scale length at the edge
Transient response analysis suggestscomplex relationship between flux and gradienttransitions between “nonlocal” and “local”
11/10N. Tamura, FEC2006, EX/5-6, Oct. 19, 2006
END
12/10N. Tamura, FEC2006, EX/5-6, Oct. 19, 2006
No relationship between density fluctuations and
nonlocal Te rise is observed in LHD
13/10N. Tamura, FEC2006, EX/5-6, Oct. 19, 2006
No relationship between density fluctuations andnonlocal Te rise observed in LHD
No significant change in core density fluctuation (measured by X-mode Reflectometer) observed after the onset of core Te rise
14/10N. Tamura, FEC2006, EX/5-6, Oct. 19, 2006
Comparison with a linear gyro-kinetic
calculation
15/10N. Tamura, FEC2006, EX/5-6, Oct. 19, 2006
Comparison with a linear gyro-kinetic calculation
To esimate growth rate ITG and real frequency r(ITG) of ITG/TEM modes, GOBLIN (GyrOkinetic Ballooning LINnear equation solver) code usedComparison is done for two time slices
1.38s (Before TESPEL injection) 1.40s (Just before nonlocal Te rise)
Profiles of ne, Te, Ti as input parametersne: increase outside ~ 0.6Te: almost not changedTi: Profile shape of Ti is assumed to be sam
e as that of Te (only Ti0 is measured)
In the code, collisionless approximation is usedStabilization of TEM modes due to the incre
ase of collisionality* cannot be expressed by the GOBLIN code*F. Ryter et al., PRL 95, 085001(2005)
16/10N. Tamura, FEC2006, EX/5-6, Oct. 19, 2006
Linear gyro-kinetic calculation shows TEM turbulencedominant at the periphery of the plasma with nonlocal Te rise
ki is fixed at 0.5
Calculation resultsITG/TEM is unstable
outside ~ 0.2TEM-driven component
dominant outside ~ 0.5
Experimental resultsMaximum change of normalized
heat flux takes place around ~ 0.4
17/10N. Tamura, FEC2006, EX/5-6, Oct. 19, 2006
Characteristics ofe-transport in LHDw/o nonlocal Te rise
18/10N. Tamura, FEC2006, EX/5-6, Oct. 19, 2006
A gyro-Bohm like dependence can explain low-power NBIheated plasmas in LHD
Critical gradient scale length is unclear A gyro-Bohm like dependence of e on Te observed tr PB observed e weakly depends on Te
19/10N. Tamura, FEC2006, EX/5-6, Oct. 19, 2006
Difference between Nonlocal Te rise
& CERC
20/10N. Tamura, FEC2006, EX/5-6, Oct. 19, 2006
Nonlocal Te rise and CERC coexist
Nonlocal Te rise in CERC plasma Feature of CERC
Strongly peaked Te profile
Associated withEr bifurcation
Antithetical feature ofnonlocal Te riseHeat flux jump takes
placein a wider region of plasma
NOT associated withEr bifurcation
21/10N. Tamura, FEC2006, EX/5-6, Oct. 19, 2006
Transient response analysis for
stronger edge cooling
22/10N. Tamura, FEC2006, EX/5-6, Oct. 19, 2006
Transient response analysis suggests heat flux jump rateincreased with stronger edge cooling
Larger dTe/dt Stronger edge cooling
Heat flux jump rate increased
23/10N. Tamura, FEC2006, EX/5-6, Oct. 19, 2006
Experimentally observed Interaction
betweenthe core and the edge
24/10N. Tamura, FEC2006, EX/5-6, Oct. 19, 2006
Transient response analysis indicates existence ofinteraction between the core and the edge
qe/ne with > 0.4 decreases prior to reduction of that with < 0.4 (A)
Termination of decrease in qe/ne seems to propagate from core to edge (B)Decrease in qe/ne with
> 0.2 seems to be overshot, and goes back to a metastable level
Turn-back of qe/ne to pre-injection level is started almost simultaneously, seems to propagate from edge to core (C)