PIC simulations of the propagation of type
Download
Report
Transcript PIC simulations of the propagation of type
PIC simulations of the propagation of type-1
ELM-produced energetic particles on the SOL of
JET
D. Tskhakaya1,*, A. Loarte2, S. Kuhn1, and W. Fundamenski3
and Energy Physics Group, Association Euratom – ÖAW, Department
of Theoretical Physics, University of Innsbruck, Innsbruck, Austria
2EFDA, Close Support Unit Garching, Max-Planck-Institut fuer Plasmaphysik,
D-85748 Garching bei Muenchen, Germany
3UKAEA Fusion, Association Euratom-UKAEA, Culham Science Center,
Abingdon, United Kingdom
*Permanent address: Institute of Physics, Georgian Academy of Sciences,
Tbilisi, Georgia
1Plasma
Outline of the Talk
•
•
•
•
•
Introduction
Characteristics of the codes EDGE2D-NIMBUS and BIT1
Results of test simulations
Results of ELM-free and ELMy SOL simulations
Conclusions
D. Tskhakaya et. al., 9th EU-US Transport Task Force Workshop Córdoba, Spain, (2002)
Introduction
• Investigation of the energy and particle transport inside the SOL
during ELM activity is an extremely important topic, especially for
predicting the heat loads on the divertor plates of next-generation
fusion devices [Loarte et al., 2000, 2001].
• The short time scale of the process and the low collisionality of the
ELM-produced highly energetic particles define the kinetic nature
of ELMy transport.
• Despite its importance, kinetic simulations of the ELMy SOL are
rare. Simulations done up to now use either simplified linear
profiles for neutrals [Tskhakaya, et al., 2001], or do not consider
them at all [Bergmann, 2002]. Hence, they correspond to very
simplified SOL models with low recycling.
2 D Modelling of the Plasma Edge of
Fusion Devices (EDGE2D, B2-Eirene)
• The plasma edge is modelled with 2-D Fluid (plasma) + 2-D
Monte Carlo Codes (neutrals)
2-D Fluid equations
A
|| A Sources Sinks
t
A : Density, Momentum and Energy (e-, D+, Z+).
Particle, momentum, energy of the neutrals computed with
Monte Carlo Codes (Nimbus, Eirene)
Fluid + Monte Carlo are iterated to convergence
Advantages of 2-D Modelling of EDGE Plasmas
• Realistic 2-D geometry
• Fully time-dependent & consistent plasma solution with
sources and sinks
Disadvantages of 2-D Modelling of EDGE Plasmas
• Fluid approximation is not fulfilled in many interesting
edge plasma conditions (ELMs, hot ions in SOL, etc.)
ELMs are modelled by increasing cELM, DELM ~ (10 - 1000) x c0, D0
during tELM in pedestal & SOL
Experiments lp@ELM ~ (1 - 2) lp@between ELMs
ELM simulation for ITER with B2-Eirene [Loarte, 2000]
9
8 10 7
3.00 10
9
7 10 7
9
6 10
7
2.50 10
5 10
7
2.00 10
9
1.50 10
9
4 10 7
3 10 7
5.00 10
8
0.00 10
0
2 10 7
-0.2
1 10 7
0
0
0.2
0.4
0.6
Distance from Separatrix at Divertor (m)
0.8
9
8 10 7
3.00 10
9
7 10 7
2.50 10
9
6 10 7
2.00 10
9
1.50 10
9
5 10 7
4 10 7
3 10 7
1.00 10
9
5.00 10
8
0.00 10
0
2 10 7
-0.2
1 10 7
0
0
0.2
0.4
0.6
Distance from Separatrix at Divertor (m)
0.8
)
9
3.50 10
-2
1.00 10
-2
3.50 10
Parallel Power Flux @ ELM (W m )
-2
Pedestal + SOL
-2
Parallel Power Flux between ELMs (W m )
Parallel Power Flux @ ELM (W m )
Pedestal only
Te
Radiation
Te @ ELM
Radiation @ EL
PIC code BIT1
The 1d3v (one space and three velocity dimensions) code BIT1 was
developed on the basis of the XPDP1 code from Berkeley.
During the simulation the motion of a large number (up to some 106)
of ions and electrons is followed:
d qi
Vi
E Vi B ,
dt
mi
d
X i Vi , i 1, 2, , N,
dt
E ( E x , 0, 0) .
Nonlinear Coulomb and chargedneutral particle collisions
2x
, E x x .
0
Coulomb collisions and charged-neutral particle collisions are
modelled via a binary collision model, so that the total momentum
and the energy is conserved during a collision :
Choosing random pairs
Colliding particles
At present the code does not follow neutrals, and assumes fixed
neutral density and temperature profiles.
All (relevant to the SOL) charged-neutral particle collisions
between hydrogen isotopes are implemented:
Elastic
A+e -A+e
Excitation A + e - A* + e
Ionization A + e - A+ + 2e
10
1
10
0
10
-1
10
-2
10
10 -18 m2
10 -20 m2
2
10
Elastic
A + A+ - A + A+
Charge-exchange A + A+ - A+ + A
A = H, D, T
ionization
elastic collision
excitation
10
1
10
0
elastic
charge exchange
-1
0
10
2
E [eV]
10
4
10 -3
10
10
-1
E [eV]
Charged-neutral particle collision cross-sections
10
1
10
3
Secondary electron emission is implemented in the code.
Secondary electron emission due to electron impact
e 0
E
exp 2 2 E / E0 , 1 / cos , if 60 , else 2
E0
For graphite E0 300 eV ,
0 1.
Secondary electron emission due to ion impact [Diem, 2001]
i V m / s 106 0.18,
if V 1.8 105 m / s ,
i 0,
if V 1.8 105 m / s .
1
e
0.8
0.7
0.6
i
0.5
0.6
0.4
0.4
0.3
0.2
0.2
0.1
0
0
1000
E [eV]
2000
0
0
400
V [km/s]
800
B
Divertor
plate
Particle
source
Divertor
plate
x
Simulation geometry
During the simulation, the Maxwell-distributed electrons and ions
are injected into the source region. Particles reaching the divertor
plates are absorbed.
In the PIC simulation neutral density and temperature profiles are
used from the corresponding fluid simulations.
Fluid simulation
Neutral density and
temperature
PIC simulation
Test simulations
• Sheath effects play an important role not only in the ELMy but
also in the ELM-free SOL
Source effect
5
4
x 10
3
qx [W/m2]
2
1
0
-1
PIC
Spitzer (Sp.)
Sp.+sheath
-2
-3
-4
0
1
2
3
4
x [m]
5
6
7
Electron heat flux profile from PIC, and corresponding SpitzerHärm and the Spitzer-Härm + “sheath” heat fluxes. ne=130.
ELM-free and ELMy SOL simulations
Mismatch between fluid and kinetic (PIC) simulations
It is necessary to shorten the simulated SOL (scaling has to
be conserved).
During the PIC simulation the plasma density and
temperature at the source cannot be controlled directly.
Input parameters are the particle source intensity and the
temperature of injected particles.
Source effect: There are peaks in the density and the
temperature profile.
• the absence of sheath in the fluid simulation results in a
higher plasma density at the wall than in the PIC
simulation. Hence, in order to have a similar chargedneutral collisionality in the PIC simulation it is necessary
to shift the neutral density (obtained from the fluid
model).
18
12
x 10
10
n (fluid)
N
n (fluid)
i
n (PIC)
i
n (PIC)
8
N
6
4
2
0
0
5
10
x [mm]
15
Two sets of simulations have been made for low and
high recycling SOL
3
x 10
Low recycling ELM-free SOL
19
Fluid
PIC
n [m-3]
2.5
Ti Fluid
Te Fluid
Ti PIC
Te PIC
250
T [eV]
200
2
150
1.5
100
1
50
0.5
0
0
15
2
x 10
4
x [m]
6
8
0
0
2
4
x [m]
6
High recycling ELM-free SOL
19
n [m ]
Ti Fluid
Te Fluid
Ti PIC
Te PIC
200
Fluid
PIC
-3
8
T [eV]
150
10
100
5
50
0
0
2
4
x [m]
6
8
0
0
2
4
x [m]
6
8
ELMy SOL for JET-like conditions
• PeEDGE = 2.5 MW, PiEDGE = 6.5 MW
nsepbefore ELM = 0.8 1019 m-3 (low recycling)
= 1.7 1019 m-3 (high recycling)
ce,i before ELM = 0.75 m2/s, D before ELMe = 0.10 m2/s
•
nsepELM = 5 1019 m-3 (low recycling)
= 1020 m-3 (high recycling)
Te,i ELM = 1.5 keV (low recycling)
= 0.75 keV (high recycling)
cELM/cbefore ELM = DELM/D before ELM = 100
tELM = 200 ms , DWELM ~ 100 kJ
Low recycling case
The secondary electrons (SE) do not play any significant role
2000
without SE (t=0 mks)
with SE (t=0 mks)
without SE (t=150 mks)
with SE (t=150 mks)
Potential [V]
T [eV]
800
1500
without SE (t=0 mks)
with SE (t=0 mks)
without SE (t=150 mks)
with SE (t=150 mks)
1000
e
600
1000
400
500
0
0
200
2
4
x [m]
6
8
0
0
2
Potentil and electron temperature profiles in the
SOL, as parameters most „sensitive“ to the SE.
4
x [m]
6
8
5
x 10
19
t=0 mks
t=50 mks
t=200 mks
t=250 mks
t=400 mks
n [m-3]
4
e
Potential [V]
2000
3
1500
2
1000
1
500
0
0
2
4
x [m]
6
8
t=0 mks
t=50 mks
t=200 mks
t=250 mks
t=400 mks
1000
T [eV]
e
800
t=0 mks
t=50 mks
t=200 mks
t=250 mks
t=400 mks
2500
0
0
2
4
x [m]
6
t=0 mks
t=50 mks
t=200 mks
t=250 mks
t=400 mks
1400
1200
8
T [eV]
i
1000
600
800
400
600
400
200
200
0
0
2
4 x [m]
6
8
0
0
2
4
x [m]
6
8
From fluid simulation
1.5
19
n [x2.10
T [keV]
e
T [keV]
4
-3
m ]
x 10
From PIC simulation
5
V [m/s]
||i
i
2
1
0
t=0 mks
t=50 mks
t=200 mks
t=250 mks
t=400 mks
0.5
-2
t [mks]
0
0
100
200
300
-4
0
400
2
4
x [m] 6
8
From PIC simulation
From PIC simulation
10
2500
x 10
23
F (outer divertor)
e
F (outer divertor)
i
F (inner divertor)
e
F (inner divertor)
F [1/sm 2]
Potential [V]
2000
8
1500
6
1000
4
500
i
2
t [mks]
0
0
50
100
150
t [mks]
200
0
0
50
100
150
200
From PIC simulation
El. (outer divertor)
Ions (outer divertor)
El. (inner divertor)
Ions (inner divertor)
500
q [MW/m2]
x
400
300
200
100
0
0
500
100
t [mks]
200
From fluid simulation
q MW/m
2
400
300
200
100
t [mks]
0
0
100
200
300
400
High recycling case
The secondary electrons (SE) do not play any significant role
without SE (t=0 mks)
with SE (t=0 mks)
without SE (t=130 mks)
with SE (t=130 mks)
500
400
Potential [eV]
300
without SE (t=0 mks)
with SE (t=0 mks)
without SE (t=130 mks)
with SE (t=130 mks)
350
300
T [eV]
e
250
200
200
100
150
0
100
-100
50
-200
-300
0
2
4
x [m]
6
8
0
0
2
Potential and electron temperature profiles in the
SOL, as parameters most „sensitive“ to the SE.
4
x [m]
6
8
2
x 10
21
800
t=0 mks
t=65 mks
t=130 mks
-3
n [m ]
e
t=0 mks
t=65 mks
t=130 mks
Potential [V]
600
1.5
400
1
200
0.5
0
0
0
2
4
x [m]
6
t=0 mks
t=65 mks
t=130 mks
350
300
8
T [eV]
-200
0
2
4
x [m]
6
1000
t=0 mks
t=65 mks
t=130 mks
T [eV]
i
e
8
800
250
200
600
150
400
100
200
50
0
0
2
4
x [m]
6
8
0
0
2
4
x [m] 6
8
From PIC simulation
5
x 10
4
6
V
||e
t=0 mks
t=65 mks
t=130 mks
[m/s]
3
1
2
0
1
-1
0
-2
1200
2
4
x [m]
6
8
From fluid simulation
17
-3
n [x10 m ]
Te [eV]
Ti [eV]
1000
5
From PIC simulation
V [m/s]
||i
2
3
-1
0
x 10
t=0 mks
t=65 mks
t=130 mks
-3
0
1000
2
4
x [m]
6
8
From fluid simulation
q MW/m
2
800
800
600
600
400
400
200
200
0
0
100
200
300
t [mks]
400
0
0
100
200
300
t [mks]
400
Conclusions
• Sheath effects play an extremely important role in both
the ELM-free and the ELMy SOL:
i. The electron heat flux due to the “cut-off” effect can exceed the
Spitzer-Härm heat flux even in a highly collisional regime.
ii. the potential drop in the sheath affects the time scale of heat loads
on the divertor plates during the ELM.
• The secondary electrons do not play any significant role
in the ELMy SOL
Conclusions
•
During ELM activity the time evolution of the heat load
on the divertor plates exhibits two peaks:
i. The first (small) one appears in an electron time scale after ELM
set-on and corresponds to highly energetic ELM electrons arriving
at the divertor.
ii. The second peak corresponds to the main ELM burst propagating
through the SOL with a high-energy ion speed.
•
For more realistic modelling of the ELMy SOL it is
necessary to consider the neutrals self-consistently