Transcript Document

Neutral beam ion loss simulation and scintillator-based
loss diagnostic for NSTX
D. S. Darrow
Princeton Plasma Physics Laboratory
8th IAEA TCM on Energetic Particles in Magnetic Confinement
Systems
San Diego, CA, October 6-8, 2003
This work supported by US DoE contract number DE-AC02-76CH03073
ST fast ion confinement could
differ from conventional tokamak
• m not necessarily conserved (LB~rfi)
• MHD-induced fast ion radial transport may be
stronger in absence of m conservation
• Losses due to large rfi may generate
significant Er and plasma flows (MA~0.25 seen
in NSTX)
Several approaches to evaluate
beam ion confinement
• Energetic Neutral Particle Analyzer (Medley P12)
• Neutron collimator diagnostic (Roquemore P9)
• Multi-sightline solid state NPA (Shinohara)
• New scintillator fast ion loss probe
• Detailed modeling of loss probe signals
New beam ion loss probe
being installed in NSTX
• Enhances substantially
existing Faraday cup probe
• Scintillator-based, so will
resolve energy and pitch
angle of lost beam ions
Scintillator detector:
principle of operation
Scintillator probe assembly
Plasma
Bay J
Aperture
Graphite
armor
Light shield
Vacuum
window
Incident
ions
Scintillator
(inside)
Base &
Heat sink
Apertures designed to resolve all 3
energy components of beam
Gyroradius distributions on
scintillator plate at 0.3 T
• 80 keV D NBI =>
components at 80,
40, & 27 keV
• Probe can resolve
components even
at maximum BT
(0.6 T): r=9.6, 6.8,
& 5.6 cm
• Also covers ~20°90° in pitch angle
with 5° resolution
Scintillator plate also contains
embedded Faraday cups
+0
R 0.28
6.25-0.02
2.94
2.80
0.12
R 0.14
72¡
Seperation between
Angled Grid Lines
(scale is 10x)
1 mm (typ)
• Cups formed by
undercoated aluminum
layer
• Allows rapid absolute
calibration & gives fast
time response
• Cups matched to ~10°
bins in pitch angle
0.40
Layer Arrangement
(dwg is representative)
0.80
Phoshper Layer
Conductive Layer (aka "Grid")
Insulating Layer
Substrate (aka Tantulum Plate)
0.25-0.28 chamfer or radius
(Phosphor Layer - see Note 4)
0.53
Not to be Phosphor-coated.
Electrical Contact Area to
the underlying Grid.
Want to calculate classical beam
ion loss rate to detector at wall
• Comparison of calculation with measurement
should allow determination of which
components of loss are new features of
spherical tokamak geometry vs classicallyexpected losses
Faraday cup loss probe sees
signals from NBI
Detailed model needed since loss
varies strongly with outer gap
• Outer gap is
distance from
separatrix to
limiter at outer
midplane
Detector signal calculation
(isotropic source)
• Follow full gyro-orbit backward from detector
through plasma; integrate source strength along
orbit path & normalize by total source rate*:
TFTR

plas m a e dge
45º
60º
83º
 S(x,)d dA dl
 S(x,)ddV
aper
aper
• For isotropically-emitted
CL
90º
de te ctor

I=2.0 MA, R=2.52 m (#73268)
lim ite r
fast ions (e.g. as), integral
over  is trivial (S() =
constant)
*Follows Chrien ‘80, Heidbrink ‘84
orbit
Detector signal calculation
(directional source)
• For neutral beam ions, use S(x, ) =
Sx(x)Sv(), with Sv() = exp(-q2/qb2), q=angle
between particle velocity vector & beam
injection direction, qb=beam divergence angle
• q=1.5° for NSTX beams, so contributing
portion of velocity space is quite small (0.002
sr)
Spatial part of beam source
function is also well localized
• 1/e width of
beam:
 12 cm horizontal
 42 cm vertical
Typical orbit to detector
• Commonly, only a few steps
contribute in each orbit
Difference between TFTR & NSTX
geometries vastly changes model
gaper
Lorbit
Dg
Aperture
Lsource
Source
• TFTR: La ~10 cm, Lorb~1000 cm, so can
0.1 x 1.3 cm
resolve a source profile with Dg~0.01 rads.
Aperture extent is 1-D, so need only
compute gaper/Dg~100 orbits
• NSTX: LNB~6 cm, Lorb~10,000 cm => need
0.6 cm diam
Dg~0.0006 rads. Aperture extent is 2-D, so
need to follow (gaper/Dg)2~2,800,000 orbits to
resolve source distribution accurately(!)
Aperture y
Finely-resolved sampling confirms
extremely localized source
Aperture x
• Aperture divided into 1000 x 1000
• This level of resolution clearly insufficient
Further developments required
• Confirm that integration error remains small
enough to not affect modeled signal
• Parallelize calculation
• Apply adaptive mesh to focus computation on
regions where signal contribution is most
significant
Summary
• New fast ion loss detector will be available on
NSTX for coming experimental campaigns
– Resolves E & c of loss
• Developing model to compute classical orbit
loss to detectors in NSTX
– Geometry & beam localization in v-space make
this computationally much more intensive than
previous similar models