Download Generation of closed magnetic flux surfaces in a large spherical

1
Coupling solenoid-free Coaxial Helicity Injection started discharges to
induction in NSTX
R. Raman 1), B.A. Nelson 1), D. Mueller 2), S.C. Jardin 2), T.R. Jarboe 1), M.G. Bell 2),
H.W. Kugel 2), B. LeBlanc 2), R. Maqueda 3), J. Menard 2), M. Nagata 4), M. Ono 2)
and The NSTX Research Team
1) University of Washington, Seattle, WA, USA
2) Princeton Plasma Physics Laboratory, Princeton, NJ, USA
3) Nova Photonics, Princeton, NJ, USA
4) University of Hyogo, Himeji, Japan
Email address of main author: [email protected]
Abstract
Experiments in NSTX have now demonstrated the coupling of toroidal plasmas produced by the technique of
Coaxial Helicity Injection (CHI) to inductive sustainment and ramp-up of the toroidal plasma current. This is an
important step because an alternate method for plasma startup is essential for developing a fusion reactor based
on the spherical torus concept. Elimination of the central solenoid would also allow greater flexibility in the
choice of the aspect ratio in tokamak designs now being considered. In these new results from NSTX, the
central Ohmic transformer was used to apply an inductive loop voltage to discharges with a toroidal current of
about 100 kA created by CHI. The coupled discharges have ramped up to >700 kA and transitioned into an Hmode demonstrating compatibility of this startup method with conventional inductive operation. The electron
temperature in the coupled discharges reached nearly 1 keV and the resulting plasma had low inductance, which
is preferred for long-pulse high performance discharges. These results from NSTX in combination with the
previously obtained record 160 kA non-inductively-generated closed-flux startup currents in an ST or tokamak
in NSTX demonstrate that CHI is a viable solenoid-free plasma startup method for future STs and Tokamaks.
1. Introduction
A method to start a tokamak plasma without reliance on the central solenoid would provide
greater access to lower aspect ratio configurations and reduce reactor cost as a result of a
simpler design. The benefits on plasma performance improvement as a result of operating at
lower aspect ratio are now well known [1,2]. At extreme low values of the aspect ratio, such
as in a spherical tokamak (ST) based reactor, due to the very restricted space for the central
solenoid, elimination of the central solenoid is necessary [3]. Coaxial Helicity Injection
(CHI), a method originally developed for spheromak formation is a promising candidate for
solenoid-free plasma startup and for edge current drive during the sustained operating phase.
The possibility of using CHI in an ST was first proposed in the late 1980’s [4]. At that time, it
was generally believed that the development of non-axisymmetric perturbations of the plasma
equilibrium was needed for plasma startup using the CHI process in STs. This approach was
initially investigated in NSTX and in several other STs. However, in a significant
development during the past few years, it was shown that for the purpose of plasma startup,
axisymmetric reconnection can produce a high quality startup equilibrium. This method
referred to as transient CHI was first demonstrated on the HIT-II experiment at the
University of Washington [5]. As demonstrated on the HIT-II machine the method is
compatible with tokamak and ST designs that use superconducting poloidal field coils, as
rapidly changing coil currents are not required for generating the initial closed flux
equilibrium. Using this method, solenoid-free plasma startup and subsequent coupling to
2
conventional inductive drive has now been successfully demonstrated on the NSTX device as
well.
2. Transient CHI Startup
NSTX has a major/minor radius of 0.86/0.68 m and a toroidal magnetic field at the nominal
major radius up to 0.55 T. It is equipped with a central solenoid providing up to 0.7 Wb of
inductive flux (double swung) which can generate plasma currents up to 1.5 MA. The outer
poloidal field coils needed for equilibrium control are located about 0.5 m away from the
plasma boundary. The entire plasma facing boundary of NSTX is composed of graphite tiles.
NSTX routinely applies conventional helium and deuterium glow discharge cleaning as wall
conditioning techniques. In the past, boronization of the plasma facing surfaces has been used
but in the last two years, lithium coating of the lower divertor plates has been investigated
and developed as an alternate wall conditioning method.
On both the HIT-II and NSTX devices, CHI is implemented by driving current along
externally produced field lines that connect the inner and outer vacuum vessel components in
the presence of externally generated toroidal and poloidal magnetic fields. In both machines,
the inner vessel components are electrically insulated from the outer vessel components using
two toroidal ceramic breaks, one at the top and the second at the bottom of the machine.
During CHI operation, the lower divertor plate region is referred to as the injector and the
upper divertor plate region is referred to as the absorber region because the direction of the E
x B drift is away from the injector and into the absorber region.
As shown in Figure 1 a CHI discharge is initiated by first energizing the toroidal field coil
and then the poloidal field coils located near the injector region to generate poloidal flux that
connects the injector electrodes. On NSTX, two main coils near the injector region provide
the required injector flux that connects the inner and outer divertor plates. Gas is then injected
into a 100 Liter cavity below the lower divertor plates. The injected gas emerges through the
gap between the inner and outer divertor plates. This is followed by discharging a small
capacitor bank across the injector electrodes, which are the lower divertor plates. NSTX uses
a 5 to 45 mF (typically 10 – 15 mF) capacitor bank charged up to 1.7 kV. At applied voltages
of 1.6 kV or higher, the applied voltage is sufficient to cause the gas to breakdown and to
produce a plasma discharge. At lower voltages, NSTX also benefitted from the injection of
10 kW of ECH at 18 GHz, which was also injected into the cavity below the lower divertor
plates. This results in an injector current, which is the current provided by the discharging
capacitor that flows through the plasma load. Typical injector currents in NSTX, during the
absence of a condition known as an absorber arc, vary from 1.5 to 5 kA. The absorber arc is a
condition during which part of the injected current flows along the surface of the absorber
insulator. These currents are in the range required to meet a condition known as the ‘bubble
burst’ condition. This condition requires that the injector current exceed a threshold value
needed to overcome the poloidal field line tension of the injector flux. As this condition is
satisfied the injector flux expands into the vessel and fills the vessel as shown by the fish-eye
camera images in Figure 1. The size of the capacitor bank is chosen so as to provide adequate
energy for the injected current to be maintained for the duration needed for the growing CHI
discharge to fill the vessel. At about the time the CHI plasma fully fills the vessel, most of the
capacitor bank energy is depleted, which causes both the voltage appearing across the injector
and the injector current to rapidly diminish. This reduction in the injector current is further
improved by using an actively switched ignitron that short-circuits the capacitor through a
small resistive load. If the injector flux footprints are sufficiently close together in the injector
region, then because the injector current is no loner able to drive the CHI plasma load, the
3
expanded CHI plasma detaches from the injector region through a process of axisymmetric
reconnection near the injector region producing a closed flux equilibrium that now decays on
a resistive L/R time scale.
Fig.1: On the left, is a line drawing showing the main components in NSTX required for plasma
startup using the CHI method. Top-right: CHI injector current and plasma current for a CHI-only
started discharge and bottom-right: fast camera fish-eye images of the plasma at 7.2, 8 and 11 ms
during the discharge. Note that after 10 ms the injector current is zero, so the CHI power supply is
no longer driving the plasma load. The remaining toroidal plasma current flows in closed flux
equilibrium, as seen in the fast camera image at 11 ms, which then decays resistively.
A feature of CHI plasma generation using this method is that flux closure can be
demonstrated unambiguously by the persistence of plasma current after the injector current
has been reduced to zero [6]. This can be seen in the injector and plasma current traces shown
in Figure 1. During recent experiments in NSTX, induction was applied to this closed-flux
plasma equilibrium by ramping the current in the central solenoid to demonstrate coupling of
the CHI produced current to conventional inductive operation.
3. Experimental Results
In Figure 2, we show traces for the injector current, the plasma current, and the applied
inductive loop voltage for two CHI started discharges that were coupled to induction. In these
discharges approximately 3 kA of injector current produces about 100 kA of toroidal current
initially. During the decay phase of this current, induction is applied from the central solenoid
by ramping its current from zero with a rectifier power supply with an open-circuit voltage of
4 kV. The decrease in the applied loop voltage over time is due to resistance in the central
solenoid circuit. The external poloidal field coil currents needed for equilibrium are preprogrammed during the first 40 ms of the discharge. After that the standard NSTX plasma
control system [7] is used to control the plasma radial and vertical position, using real-time
data from external flux loops and magnetic probes. This causes the decaying plasma current
to ramp-up, reaching a peak value of 700 kA for shot 128401. In these discharges, starting at
about 40 ms, neutral beams are also injected at a power of 3 - 4 MW to heat the plasma.
During the neutral beam heating phase, discharge 128406 transitions into an H-mode at 175
ms. Evidence of the H-mode is the characteristic drop in divertor Dα emission and the
development of a broad electron density (Ne) profile measured by the Thomson scattering
4
diagnostic (Figure 3). Figure 3 also shows the electron temperature (Te) and electron density
from Thomson scattering for discharge 128401 that did not transition to an H-mode. The fact
that a CHI initiated discharge is able to transition to a discharge that has features suitable for
Figure 2: The CHI injector current, plasma
current, applied loop voltage, Dα signal and the
neutral beam power for an L-mode discharge
(128401) and for a discharge that transitioned to
an H-mode at 175 ms (128406). Both discharges
have different gas programming.
Figure 3: Te and density profiles from Thomson
scattering diagnostic for the discharges shown in
Fig. 2. The discharge that transitioned in to a Hmode reached central Te of 1 keV.
a high-performance long-pulse operation
bodes well for the application of this
startup method to future machines.
Initial attempts to add inductive drive to CHI initiated discharges resulted in no increase in
the plasma current. However, in these discharges there was a significant increase in the O-II
emission when the loop voltage was applied. It was not until extensive conditioning was
performed in the form of D2 glow discharge cleaning (GDC) and electrode discharge
conditioning that the plasma current could be increased by induction. Observation in NSTX is
that as the size of the capacitor bank is increased above 10 mF, the radiated power signal
approaches the input Ohmic power. Increase in the radiated power signal is correlated with
large increases in the oxygen and carbon line radiation signals. These signals increase
because with a larger power supply the magnitude of the injector current increases. At a
higher level of injector current more of the electrode surface contaminants are liberated into
the plasma and contribute to increased energy losses during startup. As demonstrated in
experiments on HIT-II [5], a CHI discharge cannot couple to induction if the radiated power
approaches the input Ohmic power. For these plasmas with about 3 – 4 V of applied loop
voltage and at a plasma current of 100 kA, the input Ohmic power is less than 300 – 400 kW
during the inductive coupling phase. Thus it is essential that during the startup phase that
either (a) the radiated power be kept low through injector conditioning discharges or (b) some
form of auxiliary heating system such as Electron Cyclotron Resonance Heating be used to
heat the target plasma.
Recently the possible benefits for CHI discharges of lithium coating applied to the lower
divertor plates were also tested. During these experiments, lithium was evaporated from two
ovens mounted at the top of the vacuum vessel onto the lower divertor region at a rate of
10 mg/minute for 10 minutes prior to the initiation of a CHI discharge. The evaporator
system is described in Reference [8]. The immediate benefit of lithium was to make the CHI
started discharges more reproducible. The application of lithium reduces recycling of
deuterium at the electrode surfaces and it was anticipated that this would lower the density of
the CHI produced plasma. At lower density the plasma should heat up to a higher
5
temperature, which should reduce the L/R decay rate and thereby increase the current at the
time that induction was applied. A similar effect was seen on the HIT-II experiments when
titanium gettering was employed as a wall conditioning technique. However, despite the
benefit of lithium conditioning to the reproducibility of CHI startup in NSTX, the magnitude
of the initial plasma current produced by CHI did not increase with the use of lithium. The
reason may be that the amounts of lithium used in present experiments were too small to
observe the pumping effect on the fast time scale of the plasma initiation. However, on a
longer time scale wall pumping was observed as discharges after Li conditioning reached
higher plasma currents after coupling to induction.
Simulations using the TSC code
Figure 4: Simulations using the TSC
code (a) Shown are the poloidal flux
contours at three times during the
discharge evolution, (b) the attained
plasma current and (c) the injector
current.
To understand and assess the full potential of the
transient CHI method to NSTX and to future
machines we have used the Tokamak Simulation
Code (TSC) [9] to simulate NSTX CHI
discharges. TSC, is a time-dependent, freeboundary, predictive equilibrium and transport
code. It uses as input the NSTX vessel geometry
and external circuit parameters. Generation of
closed flux in TSC is as a result of an effective
toroidal loop voltage induced by the CHI ejected
poloidal flux that decreases as the injector current
is reduced to zero. The code has been able to show
consistency with earlier theory [4] for the scaling
of CHI produced current with the injector and
toroidal fluxes. These results in conjunction with
experimental work on HIT-II and NSTX, two
machines with different sizes, suggest that the
amount of injector current required to pull the
injector flux into the vessel increases with the
injector flux but decreases with the toroidal field
indicating that the scaling to future machines with
stronger toroidal field is favorable.
In Figure 4, we show results from such a
simulation for a low current CHI produced
discharge. The plasma evolves in much the same
way as observed experimentally for a similar
discharge shown in Figure 1. As in the
experiment, the voltage to the injector is applied at 5 ms. To simulate the conditions in the
experiment, the applied injector voltage is programmed to reduce to zero at about 10 ms. This
causes the injector current, shown in Figure 4c, to reduce. The plasma current (b) increases to
about 70 kA, it then decays over a time longer than that experimentally observed. The decay
rate of the current could be increased by specifying a larger resistivity for the CHI plasma.
The resulting poloidal flux contours are shown in Figure 4a. These show an initial open flux
discharge expanding away from the injector and into the vessel and after about 9.5 ms the
presence of a closed flux region is evident. Because TSC is an axisymmetric code and does
not include reconnection physics, it was speculated that that there must be a toroidal loop
voltage generated during the transient CHI process. To verify this conjecture, toroidal loop
6
voltage sensors were placed at different locations within the vessel boundary in the
simulations. These show that during the time of the CHI plasma growth into the vessel a large
negative loop voltage is induced as a result of the injected flux. However, as this flux begins
to decay and reduce in magnitude, a modest positive loop voltage of about 2-3 Volts is
generated, which is responsible for the generation of the closed flux surfaces. It is useful to
note that in the actual experiment, the presence of oppositely directed field lines in close
proximity to each other, in the presence if finite resistivity should naturally reconnect. The
presence of a positive loop voltage would further facilitate the reconnection process, and
make the generation of closed flux easier.
Figure 5: Shown are results from a TSC simulation in which the toroidal field and the injector
voltage was varied. For case (a) BT = 0.3 T, E-field=18 V/m, the maximum attained injector
current was 6 kA and Ip was 90 kA. As BT was increased to 0.5 T (case b) at the same voltage the
injector current reduced to 3 kA and Ip reduced to 60 kA. In case c, the E-field was increased to
30 V/m, this caused the injector current to increase to 5 kA and Ip to increase to 120 kA. Similar
scans have been conducted on the HIT-II experiment.
To test previous predictions of the role of toroidal field for CHI discharges, starting from a
discharge such as the one shown in Figure 4, the toroidal field was reduced from 0.4 T to 0.3
T and in a subsequent simulation increased to 0.5 T. These are shown in Figure 5.
Simulations show that as the toroidal field is reduced at fixed injector voltage, the injected
flux more easily expands into the vessel. This is because of a larger increase in the injector
current. The discharge at 0.3 T has a peak injector current at 6 kA, whereas at 0.5 T it
decreases to 3 kA. Because of this, the discharge at 0.5 T at 9.5 ms has still not pulled into the
vessel as much as the discharge at 0.3 T does at 8.8 ms. At higher toroidal field the injector
current decreases because of an increase in the impedance of the plasma load. The higher
toroidal field effectively increases the length of the field lines so that for a fixed applied loop
voltage, the amount of current that can be driven is reduced. Again to test this hypothesis,
starting from the discharge shown in Figure 5b, the applied voltage across the injector was
increased. Now, at this higher value of the toroidal field and higher voltage the injector
current increases to 5 kA, still 1 kA less than at 0.3 T, but this current is now sufficient to
more easily pull the injector flux into the vessel. While more detailed simulations are
required to precisely establish the relationship between the applied voltage, the injector flux
and the toroidal field, the TSC results are consistent with the earlier predictions, from simple
physics arguments [4]. Because the injected flux now more fully fills the vessel, the resulting
CHI produced toroidal current increases above what was achieved at 0.3. The plasma and
injector currents at 0.5 T and at the higher injector voltage are shown in Figure 5c.
c)
7
It is this type of experimental optimization that allowed HIT-II to maximize the CHI
produced plasma current. On HIT-II as the injector flux was increased, both the toroidal field
and the CHI capacitor bank charging voltage both had to be increased to increase the
magnitude of closed flux current.
4. Summary and Conclusions
Using the method of transient CHI in NSTX, 160 kA of closed-flux toroidal current has been
produced. Now, for the first time in NSTX, CHI started discharges have been successfully
coupled to induction to show compatibility between CHI started discharges and the
conventional inductive approach. While results similar to this have previously been
demonstrated on the smaller HIT-II experiment, this is the first such demonstration on a large
ST with a poloidal field configuration more prototypical of a compact ST reactor such as the
Component Test Facility (CTF). Another significant new result is achieving a current
multiplication factor up to 70, which is an order of magnitude larger than achieved in HIT-II
and suggests a favorable scaling of the technique with machine size. The transitioning of the
CHI-produced discharges in NSTX to a high confinement H-mode with low plasma
inductance demonstrates the potential for the application of this method to future machines.
Initial simulations using the TSC code are helping to understand the relationship between the
injector current, injector voltage and the injector and toroidal fluxes, which is necessary for
the extrapolation of CHI to larger machines.
Acknowledgments
We acknowledge the support of the NSTX team for operation of the machine systems and
diagnostics. Special thanks are due to E. Fredd, R. Hatcher, S. Ramakrishnan, C. Neumeyer
for support with CHI related systems, and to Dr. N. Nishino of Hiroshima University for
providing a fast camera that was used in some of the experiments. The fast camera images in
this paper are from a fast camera provided by Nova Photonics, Inc.. This work is supported
by US DOE contract numbers FG03-96ER5436, DE-FG03-99ER54519 and DE-AC0276CH03073.
References
[1] PENG, Y-K M, et al., Plas. Phys. Cont. Fus., 47 B263 (2005)
[2] BELL, M.G., BELL, R.E., GATES, D.A., et al., Nuclear Fusion 46, S565-S572 (2006)
[3] NAJMABADI, F. and the ARIES Team, Fusion Eng. Design, 41, 365 (1998)
[4] JARBOE, T.R., Fusion Technol. 15, 7 (1989)
[5] RAMAN, R., JARBOE, T.R., HAMP, W.T., et al., Physics Plasmas 14, 022504 (2007)
[6] RAMAN, R., NELSON, B.A., BELL, M.G., JARBOE, T.R., MUELLER, D., et al., Phys. Rev. Let., 97,
175002 (2006).
[7] GATES, D., et al., Nuclear Fusion 46, 17 (2006)
[8] KAITA, R., et al., “Plasma performance improvement with Lithium coated plasma-facing components in
NSTX,” Proceedings of the 22nd IAEA Fusion Energy Conference, Geneva, EX/P4-9 (2008
[9] JARDIN, S.C., DeLUCIA, J.L., POMPHREY, N.J., Compt. Phys. 66, 481 (1986)