Time-dependent modelling of ELMing H-mode at TCV with SOLPS

Barbora Gulejová 1 of 2EPS 2007 material 20/6/2007

Time-dependent modelling of ELMing H-mode at TCV with

SOLPS

Barbora Gulejová, Richard Pitts, Xavier Bonnin, David Coster, Roland Behn, Jan Horáček, Janos Marki

OUTLINE

* SOLPS code package* ELM simulation – theory* Simulation of the ELM* Comparison of SOLPS with experiment


SScrape-crape-OOff ff LLayer ayer PPlasma lasma SSimulationimulation Suite of codes to simulate transport in edge plasma of tokamaks

B2B2 - solves 2D multi-species fluid equations on a grid given from magnetic equilibrium

EIRENE EIRENE - kinetic transport code for neutrals based on

Monte - Carlo algorithm

SOLPS 5SOLPS 5 – coupled EIRENE + B2.5

Main inputs: magnetic equilibrium Psol = Pheat – Prad

core upstream separatrix density ne

Free parameters: cross-field transport coefficients (D┴, ┴, v┴)

B2 plasma background =>recycling fluxes

EIRENE

Sources and sinks due to neutrals and molecules

measured

systematicallyadjusted

Mesh

72 grid cells poloidallyalong separatrix

24 cells radially


Edge localised mode (ELM)Edge localised mode (ELM)H-mode Edge MHD instabilities Periodic bursts of particles and energy into the SOL

- leaves edge pedestal region in the form of a helical filamentary structure localised in the outboard midplane region of the poloidal cross-section

LFSHFS

divertor targets and main walls erosion first wall power deposition ELMing H-mode=baseline ITER scenario

Energy stored in ELMs: TCV 500 J JET 200kJ ITER 8-14 MJ => unacceptable =>

Small ELMs on TCV – same phenomena !=> Used to study SOL transport

W~600J

Dα


Type III Elming H-mode at TCVType III Elming H-mode at TCV[k

A]

ne

Ip

Dα

Vl

Wheat[kW

][V

][a

u][1

018

m3 ]

# 26730

Type III ELMs

# 26730

Pre-ELM phase = steady state

ELM = particles and heat are thrown into SOL ( elevated cross-field transport coefficients)Post-ELM phase

tpre ~ 2 ms

telm ~ 100 μs

tpost ~ 1 ms

Ip = 430 kA, ne = 6 x 1019 m-3 , felm = 230 Hz

ELMs - too rapid (frequency ~ 200 Hz) for comparison on an individual ELM basis => Many similar events are coherently averaged inside the interval with reasonably periodic elms


upstream targets

Jsat [A.m-2]LP, averageSOLPS

outerinner

Te [eV]

ne [m-3]

Perp.heat flux [MW.m-2]

LP, averageSOLPS

outer

inner

IR

r-rsep r-rsep

r-rsep

v [m.s-1]

D [m2.s-1]Χ [m2.s-1]

same solps result!

Vperp=0

Ansatz

D┴,Χ┴ = radially constant in divertor legs

D┴= 3 m2.s-1 in div.legs1m2.s-1 in SOLΧ┴= 5 m2.s-1 in div.legs6 m2.s-1 in SOL

NO DRIFTS here But simulations

with DRIFTs show early promise of

expected effects

**

Simulation of pre-ELM = steady state of H-mode presented at PSI 17, Hefei,China and published in JNM May 2006

Very good overall

match Good basis for time-dependentELM simulation

<=>Pre-ELM must be

time-dependent too!+

equal time-steps for kinetic and fluid

parts of code ~ 10-6s

**


Simulation of ELM Simulation of ELM

Ddr

dnT

dr

dTn

At

E

ELMELM

ELM ..2

5..

..2

* Instantaneous increase of the cross-field transport parameters D┴, ┴, v┴!

1.) for ELM time – from experiment coh.averaged ELM = tELM = 10-4s2.) at poloidal location -> expelled from area AELM at LFS

From the cross-field radial transport can be estimated the combination of transport parameters corresponding to the given expelled energy WELM, tELM and AELM

D┴

Many different approaches possible =>changes in D┴, ┴ only or in v┴ too …

time

W~600J

Dα

AELM= 1.5m2

W = 600 J

Conv.[eV.m-2.s-1]

Diff.[eV.m-2.s-1]

┴

D ┴

*

LFSHFS


Tools to simulate ELM in SOLPS

outer div.leg

main SOL

main SOLInner div.leg

Several options in SOLPS transport inputfiles :

* Multiplying of the transport coefficients in the specified poloidal region

* In 3 different radial regions (core, pedestal, SOL) by different multipliers

Added new options: * Poloidal variation of the multiplicator

* Step function

* Gaussian function

* Choosing completely different shape of

radial profile for chosen poloidal region

pedestal wallcore

No TB

preelm

ELM

x M

ELM


Simulation of ELM Experimental data to constrain the code: * Energy expelled by the ELM through

separatrix (DML) ~ 600 J* Time of ELM-rise (D coavelm) ~ 100 μs

* Jsat at the targets * Heat flux at outer target (IR soon)

* Upstream ne,Te (TS) -> just orientation


Simulation of ELM - resultsUpstream profiles

ne [m-3] ne [m-3]

Te [eV] Te [eV]

r-rsep

r-rsep

D,Chi (ELM): Same shape as preELM => TB, Just multiplied D x 100, Chi x 8

D,Chi (ELM): smaller TB+ Gaussian function poloidally

D, Chi increased only locally !


Comparison with TS – R.Behn

TS measurements (R.Behn) =>* Drop in pedestal width and height appears only for ne

SOLPS * bigger pedestal collaps * higher ne and Te in SOL

But the right tendency – pedestal collapse

D┴

ne

R-Rsep

R-Rsep

time

time

Time evolution of D┴ and ne

tELM=100 µs

ne

Te [eV]

+ different ELMing H-mode shots !

TB

TBno TB

no TB


Simulation of ELM - resultsTarget profiles D,Chi (ELM): smaller TB

+ Gaussian function poloidally

“removal” of TB => higher values at the targets

LP coav data

SOLPS

SOLPSinner

outer

targets targets

inner

innerouterouter

D,Chi (ELM): Same shape as preELM => TB, Just multiplied D x 100, Chi x 8

27

1622

30

35

40

Jsat [A.m-2]

Jsat [A.m-2]

preELMELM

preELMELM

preELMELM

Jsat [A.m-2]

Te [eV]

ne [m-3]

Jsat [A.m-2]

Te [eV]

ne [m-3]


Simulation of ELM - results Target profiles – Heat fluxes

D,Chi (ELM) :Same shape as preELM => TB, just multiplied D x 100, Chi x 8

D,Chi (ELM): smaller TB + Gaussian function poloidally

outer outer

inner inner

Time-dep. heat flux estimated from LP measurements at outer target

R.Pitts, Nuc.Fusion 2003

Like Jsat, Te, and ne, heat flux is higher for the case with smaller TB

IR data available soon (J. Marki)

2016

9


ELM-Time evolution of jsat at targets

Inner

outer target

Dα

Jsat [A.m-2]

SOLPS

SOLPS

LP 5

LP 20

post-ELM

ELM

pre-ELM

pre-ELMELMpost-ELM

R.Pitts, Nuc.Fusion 2003


ELM-Time-dependent solution at targets ELM

Jsat [A.m-2]

Te [eV]

Ti [eV]

ne [m-3]

targetsinner outer

time [s]

midplane separatrixTB during ELM TB during ELM

Jsat [A.m-2]

Te [eV]

Ti [eV]

ne [m-3]

Density is fixed at midplane separatrix => can’t change too much at the targets either => Jsat can increase mostly through Te,Ti => Necessary to increase D more then Chi in order to act on increase of ne

time [s] time [s]


ELM-time-evolution of jsat at targetstargets

innerouter

time [s] time [s]

Jsat [A.m-2] Jsat [A.m-2]

ELM rise time

SOLPSSOLPS

LP 20LP 5Compared with LP close to strike point => Not exactly the sameposition


Timedependence of target Te,Ti

Timedependence of simulated Te,Ti (D.Tskhakaya) for 120kJ ELM at JET (sheat heat coeff assumed 8)

R.Pittts, IAEA,2006

outerinner

Te [eV]Te [eV]

Ti [eV]Ti [eV]

SOLPS

Te is rising quicker then Ti

Propagation down to the target from upstream pedestal


Timedependence of target Te,Ti

Ions are arriving to target later then electrons (~16 μs) In reality should be more, but :1.) probably more collisional 2.) SOLPS – equipartition of energy between ions and electrons

=>

reasonable shift of Te,Ti peaks

outerinner

Te [eV]

Ti [eV]

SOLPSELM ‘end’ ELM ‘end’

peak

peak

R.Pittts, IAEA,2006


Thank you for attention


But not at strike point!!

Probably more collisional

Documents

Time-dependent modelling of ELMing H-mode at TCV with SOLPS