Sandia Labs FY21 LDRD Annual Report


New cryogenic fuel configurations for magnetically driven inertial confinement fusion targets. There is near-term need within the Inertial Confinement Fusion (ICF) program at Sandia’s Z machine for advanced cryogenic fuel configurations. This LDRD project provided three new cryogenic fuel configurations for novel ICF target concepts: (1) Deuterium ice fibers with diameters ranging from 200 to 500 µm were extruded to lengths (> 1 foot) and lifetimes (> 10 min), which exceed the needs of Z experiments. Next, thin ice layers (2) and thick ice layers (3) were grown via desublimation, where a slow flow of deuterium gas enters the target and freezes to the walls without entering the liquid phase. The team first demonstrated 10-100 µm layers of deuterium ice for metallic-wall mix mitigation. Next, millimeter(s) thick cylindrical ice shells were grown on the inner wall of cylindrical liners for future high- gain applications. The desublimation-based fills have now been used successfully on Z-ICF experiments, and a near term goal is to grow ice layers which include uniformly mixed spectroscopic dopants (e.g., krypton). The first experiments on Z to use the deuterium ice extruder technology are planned for April 2022. If a next generation petawatt-class pulsed power facility (e.g., 60-MA, 100-ns) capable of nearly double the current of Z becomes available, these and other more complex fuel geometries could be used. (PI: Thomas Awe) (Left) Cross-sectioned

CAD rendering of a screw- driven extruder assembly. The extruder is filled via desublimation. When the extruder cavity is filled with ice, the screw is driven downward, closing off the gas-fill line. With the ice cavity isolated, further screw rotation compresses the deuterium ice through a nozzle, extruding a fiber. (Right) Image of a ~500 µm diameter ice fiber. A full description of the extruder design, operation, and performance is available.

Data-driven, radiation-aware, agile modeling approach for rapid nuclear deterrence design assessment.

This LDRD project pioneered a range of data-driven and machine learning (ML) approaches for the accelerated and automated development of fast, accurate, and predictive data-driven compact device models for normal and radiation environments, with Sandia’s academic experts in ML at University of Illinois Urbana-Champaign contributing. The impacts and contributions of this pathfinder effort are manifold. First, the project established the viability of data-driven compact models for both normal and radiation environments. Second, it demonstrated that data-driven approaches can reduce development times from months to days or even hours, and in some cases can enable a fully automatic model development from data. Last, in collaboration with an ASC advanced ML project, this LDRD provided a first-of-its kind demonstration of the data-driven models and “data-to-simulation” pipeline in Sandia’s production circuit simulator Xyce. These efforts paved the way for early adoption of research results by the NA-11’s Aging and Lifetimes Program to develop an automated tool for nuclear deterrence core surveillance. (PI: B. Paskaleva)



Made with FlippingBook Ebook Creator