1.5D TRANSPORT CODE JETTO

1
Effect of Ripple-Induced Ion Thermal Transport on
H-mode Performance V. Parail, T. Johnson, T.
Kiviniemi, J. Lonnroth, P. de Vries, D. Howell,
Y. Kamada, S. Konovalov, N. Oyama, G. Saibene, K.
Shinohara and EFDA JET contributors
2
Outlook

• Ripple losses of thermal ions experimental
  evidence and prospect to use it for ELM
  mitigation
• Orbit Following Monte Carlo code ASCOT and
  simulation of thermal ion ripple losses in
  plasmas with JET and JT-60U magnetic coils
• Predictive transport modelling of JET plasmas
  with ripple-induced transport
• Summary.

3
Ripple-induced transport

• Magnetic field in a tokamak varies in both
  poloidal and toroidal directions
  BB01ecos(q)dsin(Nf)
• Accordingly, there are three different group of
  particles
• Passing particles (practically not influenced by
  ripples)
• Banana particles (experience stochastic
  ripple-banana diffusion)
• Ripple trapped particles (experience direct or
  diffusive losses due to uncompensated vertical
  drift)

4
Thermal ion ripple losses- experimental evidence
Ripple-induced losses are usually considered as a
negative feature, leading to

• prompt losses of fast ions (NBI or a-particles),
  which above all can damage vacuum vessel
• large ripples lead to deterioration of edge
  pedestal and loss of H-mode
• reduce co-NBI plasma rotation and can even
  reverse the sign of toroidal rotation
• Recent JET/JT-60U identity experiments reveal
  significant differences in plasma performance and
  ELM behaviour in otherwise identical plasmas,
  which might be attributed to higher ripple
  amplitude in JT-60U

G. Saibene et al. IAEA, 2004
5
Thermal ion ripple losses- experimental evidence
On the other hand, presence of moderate ripples
might bring some essential benefits

• JT-60U type-I ELMs are smaller, more frequent and
  benign than their JET counterpart
• JT-60U is the only big tokamak, which manages to
  keep both ETB and ITB in a steady state stable
  co-existence
• JT-60U recently reported reaching QH-mode with
  co-injection NBI
• Some experiments with moderate ripples (JET,
  1995) or stochastic magnetic limiter (DIII-D,
  2004) reported some improvement in ELMy H-mode
  performance

6
JET RIP-II (1995) H-mode confinement and ELMs

• Sub-threshold (d0) NBI power.
• dfd16, f(Iodd-Ieven)/(IoddIeven)
• Sweep in d shows transition to H-mode at 0.8
  ripple at the outer midplane separatrix at
  constant PNBI
• fELM ? for d/d16 ?(0 to 0.3) and plasma
  parameters improved!
• For d/d16 gt0.3, H-mode plasma performance
  degraded
• Note d/d16 0.3 gives in JET the same ripple
  at outer midplane as that of JT-60U

B. Tubbing et al,, EPS 1995
7
Why ripple transport might be important for
H-mode?

• Since transport within the ETB is small (of the
  order of cneo), strong pressure gradient and
  current develops
• As soon as edge parameters exceed stability
  limit, an ELM develops to remove excessive
  pressure and current
• Even small additional transport within ETB can
  change plasma dynamics

Ripple-induced transport
8
Orbit Following Monte Carlo code ASCOT was used
to simulate thermal ion ripple losses in plasmas
with JET and JT-60U magnetic configurations

• JET shot 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

See also T. Kiviniemi et al., P2.009, Tuesday
9
ASCOT simulations

• We conclude that
• both diffusive and convective (direct) ripple
  losses are higher in JT-60U coil configuration
  even if ripple amplitude is the same at the outer
  mid-plane
• diffusive losses extend deep inside ETB for both
  configurations
• direct losses are very edge-localised and depend
  strongly on ion collisionality

Ion thermal conduction c (m2/s) - solid
lines Direct escape rate ndlnn/dt (500s-1)-
dashed lines
10

Predictive modelling of JET plasma with ripple
losses

• Information from ASCOT was used to perform
  predictive transport modelling of ELMy H-mode
  JET/JT-60U plasma with realistic level and
  distribution of ripple losses
• Flux surface averaged additional ion thermal
  transport implemented in the JETTO transport code
  in the following way
• Direct losses assumed to be edge localised and
  are implemented in t-approximation d(niTi)/dt
  -n niTi
• Diffusive losses assumed to have a wide radial
  distribution (wider than ETB)
• Bohm/gyroBohm or Weiland model used for core
  transport with ETB assumed to have neo-classical
  level of transport
• ELMs are triggered by ballooning mode

11
Predictive modelling of JET plasma with ripple
losses

• We first test narrow edge-localised direct
  losses in t-approximation
• Flattening of the pressure gradient near the
  separatrix leads to effective narrowing of the
  pedestal and subsequent reduction of the energy
  stored in the pedestal
• Stiffness of core transport propagates this
  reduction deeper into the core

Red lines - no ripple losses Blue lines -
nmax100s-1 Green lines - nmax300s-1
12
Predictive modelling of JET plasma with ripple
losses

• Narrow edge-localised ripple losses reduces
  performance and increases ELM frequency
• Energy losses during the ELM are reduced, which
  leads to smaller, more benign ELMs

Red lines - no ripple losses Blue lines -
nmax100s-1 Green lines - nmax300s-1
13
Predictive modelling of JET plasma with ripple
losses

• We assume that ripple losses are diffusive with
  wide ripple localisation
• since transport is nearly uniform within ETB,
  pressure profile just before ELM is practically
  the same for all levels of ripple losses
• What is different however its the ELM
  frequency, which goes down when we increase
  ripple transport

Red lines - no ripple losses Blue lines -
Dcmax1m2/s Green lines - Dcmax1.5m2/s
14
Predictive modelling of JET plasma with ripple
losses

• The ELM frequency decreases due to larger edge
  losses between ELMs with increased ripple
  transport
• The time-average pressure and plasma energy
  content increase with increased ripple losses
  (even if max. pressure stays the same)
• A reduction in the ELM frequency and rise in the
  energy content were seen in JET ripple
  experiments in 1995
• This result resembles the improved performance
  obtained with a stochastic magnetic boundary in
  DIII-D (T. Evans et al., 2004 IAEA Fusion Energy
  Conference).

Red lines - no ripple losses Blue lines -
Dcmax1m2/s Green lines - Dcmax1.5m2/s
15

• SUMMARY
• Magnetic ripple is not necessarily a deficient
  feature of the tokamak if carefully controlled,
  it 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
• Present study is limited to transport and MHD
  analyses and should be extended to take into
  account such important effects as plasma
  rotation, radial electric field generation,
  particle diffusion etc.
• Experiments on JET and JT-60U are under
  preparation to elucidate the role of controlled
  magnetic ripple in ELMy H-mode performance

16
Discussion
17
Ripple well trapping
Ripple-induced transport (2)

• Toroidal symmetry is broken for locally trapped
  particles, so orbits are not confined.
• The motion is a sum of
• Oscillation between turning points
• Vertical drift
• Detrapped by
• Collisions
• Moving towards smaller d
• These losses can be either convective or
  diffusive depending on collisionality

18
Ripple Perturbations of Banana Orbits
Ripple-induced transport (3)

• Ripples perturb banana orbits at their banana
  tips, moving them across flux surfaces.
• If the unperturbed tip appear at a ripple
  maximum, then the reflection appear earlier and
  vice versa.
• This is a diffusive process

19
1

• Similar ripple magnitude
• Plasma shapes
• Field line geometry

0.5lt BT ripple lt1 outer midplane But small in
x-point region
20
JET JT60-U shape and 0.3 d/d16
JET 32 coils ripple is 0.1

1
0.1
1
Note for the same ripple, fast ion losses may be
different (NB - ICRF). In JT-60U losses are high
also because NB are ?