Historical research guide · 0973 · 2017
Detection of an anomalous pressure on a magneto-inertial-fusion load current diagnostic
Mark Hess · Brian Hutsel · Christopher Jennings · J. P. VanDevender · Adam Sefkow · Matthew Gomez · Patrick Knapp · George Laity · Daniel Dolan · Derek Lamppa · Kyle Peterson · William Stygar · Daniel Sinars
In one minute
Sandia’s Z machine drives about twenty million amps into a metal tube to crush fusion fuel, and knowing exactly how much current arrives matters. One way to measure it is to let the current’s own magnetic field shove a thin aluminium plate and watch the far side of that plate move with a laser. Mark Hess and twelve Sandia colleagues report that on these shots the plate starts moving too early and too hard. To explain the motion with magnetic pressure alone you would need at least a megaamp more current than every other instrument on the machine says is there. So the push is real but its cause is not the load current, and the authors call it an anomalous pressure. They measure its shape, feed it back into their magnetohydrodynamic code as an extra boundary pressure, and get excellent agreement with the laser. Then they work through what could be doing the pushing — ruling out radiation, and pointing at charged particles crossing the feed gap.
Connection
Chapter 9 is built on the principle that you take a careful, published measurement seriously even when the accounting does not close, and this is a model of how that is done: the anomaly is stated in megaamps and gigapascals per nanosecond, the mundane explanations are tested first, and the measurement that would settle it is named. Chapter 12 gets the setting — the final power feed of the machine that is currently the world’s best magneto-inertial fusion driver.
From the source
The photonic Doppler velocimetry diagnostic measures load current indirectly: the load current’s magnetic field diffuses into a thin flyer plate that replaces a section of the inner magnetically insulated transmission line, the field and the induced current density push the plate, and a laser reflected off the far surface reads that surface’s velocity by interferometry. On shot z2850 the flyer was aluminium-6061, 384 micrometres thick to within one micrometre; the velocity uncertainty of the diagnostic is about 10 metres per second and its timing uncertainty relative to the B-dot signals is 3 nanoseconds.Section II, Experimental Configuration, and Figures 1a and 2; Section III, first paragraph
The early-time flyer motion cannot be produced by the load current. Reproducing it with magnetic pressure alone requires an unfolded load current of roughly 2 megaamps at 2.968 microseconds, while at that same moment the averaged B-dot probes measure about 460 kiloamps at the load and about 700 kiloamps at the insulating stack 166 centimetres out, with quoted B-dot uncertainties of about 1.1 and 2.1 per cent — so the current needed sits well outside the uncertainty bounds of both measurements and of the machine’s circuit model.Section III, Measurement of the Anomalous Pressure; Figures 3 and 4
Treating the discrepancy as an extra mechanical pressure closes it. Unfolding a time-dependent anomalous pressure from the measured velocity and applying it as a boundary condition on the inner flyer surface, in addition to the magnetic pressure of the load current, brings the ALEGRA simulations into excellent agreement with the photonic Doppler velocimetry data. The pressure switches on near 2.94 microseconds, and the two unfolds — one against the B-dot current, one against the circuit model — agree closely at onset before diverging near the peak.Section III, closing paragraphs; Figures 5 and 6
Radiation is ruled out as the source. The ALEGRA simulations put the temperature rise from ohmic heating on the inside of the final feed at less than 10 kelvin at 2.97 microseconds, whereas producing the anomalous pressure by radiation ablation at that moment would require radiation temperatures of at least 9,000 kelvin.Section IV, Discussion and Conclusions, first paragraph
A one-dimensional magnetic-insulation calculation for the power feed gap, including the experiment’s uniform 10 tesla axial field, predicts that electrons are insulated at all times — the critical insulation current is zero, and it would take a feed voltage of 8.5 megavolts to overcome the axial field against the circuit model’s peak of 1.16 megavolts — but that negative ions are not. Hydrogen negative ions become uninsulated from about 2.962 to about 2.988 microseconds, and heavier negative ions such as hydroxide from about 2.953 to about 3.034 microseconds, so uninsulated negative ions crossing the gap and striking the flyer could explain part of the observed pressure.Section IV; Equation 6 and Figures 7 and 8
The accounting is carried through to numbers that can be checked. At 100 kilo-electronvolts a hydrogen negative ion stops within about a micrometre of aluminium, which is also the thermal conduction length over the 10 nanosecond ramp, and an anomalous pressure ramping at 0.013 gigapascals per nanosecond implies a radial ion current loss near 7 amps per square centimetre and an ion density near ten to the seventeenth per cubic metre — which the machine’s background vacuum of about ten to the minus five torr could supply. The equivalent electron estimate is at most 400 amps per square centimetre. The authors name what would settle it: three-dimensional kinetic simulations of electrons and ions in the curved inner transmission line and final feed, and plasma cleaning of the power feed surfaces as a possible mitigation.Section IV, closing paragraphs
The way in
https://doi.org/10.1063/1.4975021How to cite it
Mark Hess, Brian Hutsel, Christopher Jennings, J. P. VanDevender, Adam Sefkow, Matthew Gomez, Patrick Knapp, George Laity, Daniel Dolan, Derek Lamppa, Kyle Peterson, William Stygar, Daniel Sinars (2017) Detection of an anomalous pressure on a magneto-inertial-fusion load current diagnostic doi:10.1063/1.4975021