magnetic ripple as a tool for elms mitigation
DESCRIPTION
Magnetic Ripple as a Tool for ELMs Mitigation V. Parail, T. Johnson, J. Lonnroth, T. Kiviniemi, P. de Vries, G. Saibene, Y. Kamada, N. Oyama, K. Shinohara, S. Konovalov, D. Howell. Outlook. Evolution of plasma profiles between ELMs; Ways to maximise top-of-the-barrier plasma pressure; - PowerPoint PPT PresentationTRANSCRIPT
1ITPA Meeting. Kyoto, 18-21 April, 2005
Magnetic Ripple as a Tool for ELMs Mitigation
V. Parail, T. Johnson, J. Lonnroth, T. Kiviniemi, P. de
Vries, G. Saibene, Y. Kamada, N. Oyama, K. Shinohara, S. Konovalov, D. Howell
2ITPA Meeting. Kyoto, 18-21 April, 2005
Outlook Evolution of plasma profiles between ELMs; Ways to maximise top-of-the-barrier plasma
pressure; Ways to avoid (mitigate) ELMs; Analytical theory of ripple transport; Predictive transport modelling of JET plasmas
with ripple transport; OFMC code ASCOT and simulation of thermal ion
ripple losses in plasmas with JET and JT-60U magnetic coils;
Summary.
3ITPA Meeting. Kyoto, 18-21 April, 2005
Evolution of plasma profiles between ELMs (1)
Since
transport
within the ETB
is quite small,
plasma
develops
strong
pressure
gradient to
transmit heat
flux through
the ETB:
nTq
4ITPA Meeting. Kyoto, 18-21 April, 2005
Evolution of plasma profiles between ELMs (2)As soon as edge parameters hit one or the other stability limit, an ELM develops, which throws away excessive pressure or current;
5ITPA Meeting. Kyoto, 18-21 April, 2005
Evolution of plasma profiles between ELMs (3)How can we mitigate ELMs or remove them entirely (without sacrificing performance, which means keeping )?
Reduce the heat flux, which enters ETB (up to but not beyond the limit, which triggers transition to type-III ELMs):
Increase radiated power (extra impurities at the edge);
Increase CX losses (gas puffing?);
Increase the heat flux through the ETB between ELMs by increasing thermal conductivity:
Increase ion density ( );
Increase transport by magnetic ripples or ergodic magnetic limiter;
Induce quasi-continuous benign MHD (EDA, type-II ELMs, washboard modes, pellets ???)
nTq
2/ poliefficlneo
i BTZn
CRITnTnT
6ITPA Meeting. Kyoto, 18-21 April, 2005
Evolution of plasma profiles between ELMs (4) Reduce the heat flux, which enters ETB (up to but not beyond the limit, which triggers transition to type-III ELMs):
Increase radiated power (extra impurities at the edge);impurity accumulation in plasma core; does not affect ions; difficult to optimise pressure within ETB; Increase CX losses (gas puffing?);it simultaneously increase losses between ELMs but it’s difficult to do it in a controlled way;
Increase the heat flux through the ETB between ELMs by increasing thermal conductivity:
Increase ion density ( );difficult to control , often triggers transition to type-III ELMs; pressure and current profiles might be far from optimum; Induce quasi-continuous benign MHD (EDA, type-II ELMs, washboard modes, pellets ???)We need much better understanding of these modes before we could reliably use them in a controllable way;
2/ poliefficlneo
i BTZn
7ITPA Meeting. Kyoto, 18-21 April, 2005
Increase transport by ergodic magnetic limiter (controllable, but increases electron transport only) or by magnetic ripples;
Magnetic ripple as ELM mitigation tool
Follow B. Then |B|~B0[1+cos()+sin(N)] oscillate due to Toridicity Ripple
Particles can be toroidally trapped in magnetic wells caused by the ripple.
The ripple well depth
min
minmax
B
BB
Bmax
Bmin
Ripple well trapping
8ITPA Meeting. Kyoto, 18-21 April, 2005
Ripple well trapping (2) Toroidal symmetry is
broken, so orbits are not confined.
The motion is a sum of: Oscillation between
turning points; Vertical drift
Detrapped by: Collisions; Moving towards smaller ;
Collisionality regimes: High: (these particles
oscillate between banana and ripple trapped state in a diffusive way)
Low: non-diffusive losses
9ITPA Meeting. Kyoto, 18-21 April, 2005
Ripple Perturbations of Banana Orbits
Ripples perturb banana
orbits at their “banana tips”,
moving it across flux
surfaces.
[A.N. Boozer, Physics of
Fluids 23, 2283 (1980)]
If the “unperturbed tip”
appear at a ripple maximum,
then the reflection appear
earlier and vice versa.
This is a diffusive process
10ITPA Meeting. Kyoto, 18-21 April, 2005
Total ripple-induced transport
Total ripple-induced
transport is a combination of:
convective/diffusive ripple losses (depending on collisionality), which are toroidally localised and
stochastic ripple-banana diffusion, which is toroidally uniform;
N. Oyama, K. Shinohara 2005
11ITPA Meeting. Kyoto, 18-21 April, 2005
We use analytical formula for additional ion thermal transport due to ripple-induced thermal ion losses in our predictive simulations with JETTO
P. Yushmanov, Review of Plasma Physics, v. 16, New York, Consultants Bureau (1991).
Flux surface averaged additional ion thermal transport. Implemented into the JETTO transport code.
232/3
,
5.0~
BRe
TNq
i
i
ithi
Nq has to be satisfied for the extent of the region inside local mirrors on the outboard side of the torus with locally trapped particles. Please note that this inequality is actually NOT satisfied for most JT-60U plasmas!
12ITPA Meeting. Kyoto, 18-21 April, 2005
Narrow edge-localised ripple: reduces performance and increases ELM frequency:
JETTO transport simulation.
The ripple-affected region is assumed to be significantly narrower than the pedestal width.
Bohm / gyro-Bohm transport model.
The density is low in this case.
no ripple-induced transportripple-induced transport
13ITPA Meeting. Kyoto, 18-21 April, 2005
Narrow edge-localised ripple: reduces performance and increases ELM frequency (2)
Ripple-induced
edge-localised
transport
Flattening of the
pressure gradient
near the separatrix,
effective narrowing
of the pedestal.
Lowering of the
pedestal height.
Lowering of the
core pressure due
to profile stiffness.
no ripple-induced transportripple-induced transport
Non-optimum pressuregradient
14ITPA Meeting. Kyoto, 18-21 April, 2005
Narrow edge-localised ripple reduces performance and increases ELM frequency (3)The ELM frequency increases
with the introduction of ripple-induced transport because of: profile stiffness, which leads to increased transport inside the top of the pedestal in the case of lower pedestal height and faster replenishment of ETB; less energy loss during the ELM;
Might explain the higher ELM frequency at JT-60U.
Smaller, more benign ELMs.
Ripple losses can be an important tool used for ELM mitigation.
no ripple-induced transport
With ripple
15ITPA Meeting. Kyoto, 18-21 April, 2005
Wide ripple at the edge: improved performance
JETTO transport simulation.
The ripple-affected region is assumed to be significantly wider than the pedestal width.
Bohm / gyro-Bohm transport model.
The density is high in this case.
no ripple-induced transportwith ripple-induced transport
16ITPA Meeting. Kyoto, 18-21 April, 2005
Wide ripple at the edge: improved performance (2)
The ELM frequency decreases due to larger edge losses between ELMs with increased ripple transport.
no ripple-induced transportripple-induced transport
17ITPA Meeting. Kyoto, 18-21 April, 2005
Wide ripple at the edge: improved performance (3)
Decreasing ELM frequency: The time-average
pressure at the top of the pedestal increases (even if max. pressure stays the same).
The time-average core pressure increases due to profile stiffness.
No ripple
less ripple more ripple
18ITPA Meeting. Kyoto, 18-21 April, 2005
Wide ripple at the edge: improved performance (4)
A reduction in the ELM frequency was occasionally seen in JET ripple experiments in 1995.
Resembles the improved performance obtained with a stochastic magnetic boundary in DIII-D [T. Evans, 2004 IAEA Fusion Energy Conference].
This mechanism is most pronounced in high density plasmas, which have the highest level of transport within the pedestal because of the temperature and density dependency of neo-classical transport.
19ITPA Meeting. Kyoto, 18-21 April, 2005
Numerical simulation of ripple losses in realistic geometry (1)
#60856 belongs to JET/JT-60U identity plasma.
Ripple-induced transport in plasma with JET coils is outboard midplane localised.
Same plasma with JT-60U coils should have much larger ripple transport near x-point;
JT-60U coils JET coils I2/I1=0.5
After drawing general conclusions on what might be expected from magnetic ripples we decide to perform a detailed analysis of realistic plasmas
20ITPA Meeting. Kyoto, 18-21 April, 2005
21ITPA Meeting. Kyoto, 18-21 April, 2005
Numerical simulation of ripple losses in realistic geometry (2)
Orbit Following Monte Carlo code ASCOT was used for the simulations;
Code follows guiding centre orbits of thermal particles ensemble in a torus;
3D JET and JT-60U ripple model has been included as B=B0 +
B1cosN + B2cos2N , where Bn=Bn(R,Z), N- number of toroidal
coils;
Only collisions with a fixed background are considered;
Ensemble of particles can be initialised as:
delta-function in radius or
according to assumed ne(r) and Ti(r);
22ITPA Meeting. Kyoto, 18-21 April, 2005
Numerical simulation of ripple losses in realistic geometry (3)
To speed up the simulations, can be calculated from the spreading of a “pulse” (with the radial electric field turned off):
f(t=0,r, v)=(r-r0)fMaxwell(v) If the process is diffusive the variance will grow at a constant rate
<(r (t,v)-<r (t,v)>)2> =V (t,v)=2Dv(v)t
Since the flow ~Dv, the transport coefficients are given by:
0 - relates pressure gradients and particles transport,
2 - relates temperature gradients and energy transport, and
1 - is the off-diagonal coefficient.
Accuracy ~ [Ntp -1/2 e-E/T Nmeasurements]-1/2 ~ 20%
)(2
5
22
1 23 vf
T
mv
dt
dVvd Maxwell
n
n
23ITPA Meeting. Kyoto, 18-21 April, 2005
Numerical simulation of ripple losses in realistic geometry (4)
Rho=0.98, JET I2/I1=0.5 Rho=0.98, JT-60U coils
24ITPA Meeting. Kyoto, 18-21 April, 2005
JT-60U coils
JET coils, I2/I1=0.5
JET coils, I2/I1=1.0
neo-classical CHI-I
25ITPA Meeting. Kyoto, 18-21 April, 2005
Numerical simulation of ripple losses in realistic geometry (5)
Rho=0.98, JETcoils, I2/I1=0.5
Rho=0.98, JT-60U coils
26ITPA Meeting. Kyoto, 18-21 April, 2005
Numerical simulation of ripple losses in JET geometry (1)
Four level of current imbalance between odd and even JET coils is being considered: I1/I2= 1.0; 0.66; 0.33 and 0.0 JAERI OFMC code is being used for the modelling of fast ion losses
I1/I2= 0.66 I1/I2= 0.33 I1/I2= 0.0
2 2 2
27ITPA Meeting. Kyoto, 18-21 April, 2005
FULL ripple
66% ripple
33% ripple
Pmax~3.6MW/m2
2
2
28ITPA Meeting. Kyoto, 18-21 April, 2005
Full ripple
Half ripple
No ripple
ASCOT code is being used to evaluate the role of magnetic ripple in thermal ion transport;
“Pulse propagation” technique is being used for first level of analysis;
Three levels of current imbalance are used: I1/I2= 0.0; 0.5 and 1.0;
JET shot #60856 (JET/JT-60U
identity plasma with Ip=1.15 MA)
is used as an example;
Numerical simulation of ripple losses in JET geometry (2)
29ITPA Meeting. Kyoto, 18-21 April, 2005
Numerical simulation of ripple losses in JET geometry (3)
30ITPA Meeting. Kyoto, 18-21 April, 2005
SUMMARY
Magnetic ripple, if carefully controlled, might serve as a valuable tool for ELM mitigation;
Magnetic ripple losses increase thermal ion transport only and it might be better to use it in combination with stochastic magnetic limiter;
Experiments on JET and JT-60U are under preparation to elucidate the role of controlled magnetic ripple in ELMy H-mode performance;