Motion primitives (MPs) provide a fundamental abstraction of movement templates that can be used to guide and navigate a complex environment while simplifying the movement actions. These MPs, when utilized as an action space in reinforcement learning (RL), can allow an agent to learn to select a sequence of simple actions to guide a vehicle towards desired complex mission outcomes. This is particularly useful for missions involving high speed aerospace vehicles (HSAVs) (i.e., Mach 1 to 30) where near real time trajectory generation is needed but the computational cost and timeliness of trajectory generation remains prohibitive. This paperdemonstrates that when MPs are employed in conjunction with RL, the agent can learn to solve a wider range of problems for HSAV missions. To this end, using both a MP and and non-MP approach, RL is employed to solve the problem of an HSAV arriving at a non-maneuvering moving target at a constant altitude and with an arbitrary, but constant, velocity and heading angle. The MPs for HSAV consist of multiple pull (flight path angle) and turn (heading angle) commands that are defined for a specific duration based on mission phases; whereas the non-MP approach uses angle of attack and bank angle as action space for RL. The paper describes details on HSAV problem formulation to include equations of motion, observation space, telescopic reward function, RL algorithm and hyperparameters, RL curriculum, formation of the MPs, and calculation of time to execute the MP used for the problem. Our results demonstrate that the non-MP approach is unable to even train an agent that is successful in the base-case of the RL curriculum. The MP approach, however, can train an agent with success rate of 76.6% inarriving at a target moving with any heading angle with a velocity between 0 and 500 m/s.
A natural clinoptilolite sample near the Nevada National Security Site was obtained to study adsorption and retardation on gas transport. Of interest is understanding the competition for adsorption sites that may reduce tracer gas adsorption relative to single-component measurements, which may be affected by the multi-scale pore structure of clinoptilolite. Clinoptilolite has three distinct domains of pore size distributions ranging from nanometers to micrometers: micropores with 0.4–0.7 nm diameters, measured on powders by CO2 adsorption at 273 K, representing the zeolite cages; mesopores with 4–200 nm diameters, observed using liquid nitrogen adsorption at 77 K; and macropores with 300–1000 nm diameters, measured by mercury injection on rock chips (~ 100 mesh), likely representing the microfractures. These pore size distributions are consistent with X-ray computed tomography (CT) and focused ion beam scanning electron microscope (FIB-SEM) images, which are used to construct the three-dimensional (3D) pore network to be used in future gas transport modeling. To quantify tracer gas adsorption in this multi-scale pore structure and multicomponent gas species environment, natural zeolite samples initially in equilibrium in air were exposed to a mixture of tracer gases. As the tracer gases diffuse and adsorb in the sample, the remaining tracer gases outside the sample fractionate. Using a quadrupole mass spectrometer to quantify this fractionation, the degree of adsorption of tracer gases in the multicomponent gas environment and multi-scale pore structure is assessed. The major finding is that Kr reaches equilibrium much faster than Xe in the presence of ambient air, which leads to more Kr uptake than Xe over limited exposure periods. When the clinoptilolite chips were exposed to humid air, the adsorption capability decreases significantly for both Xe and Kr with relative humidity (RH) as low as 3%. Both Xe and Kr reaches equilibrium faster at higher RH. The different, unexpected, adsorption behavior for Xe and Kr is due to their kinetic diameters similar to the micropores in clinoptilolite which makes it harder for Xe to access compared to Kr.
Earth’s environment can be considered especially harsh due to the cyclic exposure of heat, moisture, oxygen, and ultraviolet (UV) and visible light. Polymer-derived materials subjected to these conditions over time often exhibit symptoms of degradation and deterioration, ultimately leading to accelerated material failure. To combat this, chemical additives known as antioxidants are often used to delay the onset of weathering and oxidative degradation. Phenol-derived antioxidants have been used for decades due to their excellent performance and stability; unfortunately, concerns regarding their toxicity and leaching susceptibility have driven researchers to identify novel solutions to replace phenolic antioxidants. Herein, we report on the antioxidant efficacy of organoborons, which have been known to exhibit antioxidant activity in plants and animals. Four different organoboron molecules were formulated into epoxy materials at various concentrations and subsequently cured into thermoset composites. Their antioxidant performance was subsequently analyzed via thermal, colorimetric, and spectroscopic techniques. Generally, thermal degradation and oxidation studies proved inconclusive and ambiguous. However, aging studies performed under thermal and UV-intensive conditions showed moderate to extreme color changes, suggesting poor antioxidant performance of all organoboron additives. Infrared spectroscopic analysis of the UV aged samples showed evidence of severe material oxidation, while the thermally aged samples showed only slight material oxidation. Solvent extraction experiments showed that even moderately high organoboron concentrations show negligible leaching susceptibility, confirming previously reported results. This finding may have benefits in applications where additive leaching may cause degradation to sensitive materials, such as microelectronics and other materials science related areas.
The V31 containment vessel was procured by the US Army Recovered Chemical Materiel Directorate (RCMD) as a third-generation EDS containment vessel. It is the fifth EDS vessel to be fabricated under Code Case 2564 of the 2019 ASME Boiler and Pressure Vessel Code, which provides rules for the design of impulsively loaded vessels. The explosive rating for the vessel, based on the code case, is twenty-four (24) pounds TNT-equivalent for up to 1092 detonations. This report documents the results of explosive tests that were performed on the vessel at Sandia National Laboratories in Albuquerque, New Mexico to qualify the vessel for field operations use. There were three design basis configurations for qualification testing. Qualification test (1) consisted of a simulated M55 rocket motor and warhead assembly of 24lbs of Composition C-4 (30 lb TNT equivalent). This test was considered the maximum load case, based on modeling and simulation methods performed by Sandia prior to the vessel design phase. Qualification test (2) consisted of a regular, right circular cylinder, unitary charge, located central to the vessel interior of 19.2 lb of Composition C-4 (24 lb TNT equivalent). Qualification test (3) consisted of a 12-pack of regular, right circular cylinders of 2 lb each, distributed evenly inside the vessel (totaling 19.2 lb of C-4, or 24 lb TNT equivalent). All vessel acceptance criteria were met.
This report documents the development of the Blue Canyon Dome (BCD) testbed, including test site selection, development, instrumentation, and logistical considerations. The BCD testbed was designed for small-scale explosive tests (~5 kg TNT equivalence maximum) for the purpose of comparing diagnostic signals from different types of explosives, the assumption being that different chemical explosives would generate different signatures on geophysical and other monitoring tools. The BCD testbed is located at the Energetic Materials Research and Testing Center near Socorro, New Mexico. Instrumentation includes an electrical resistivity tomography array, geophones, distributed acoustic sensing, gas samplers, distributed temperature sensing, pressure transducers, and high-speed cameras. This SAND report is a reference for BCD testbed development that can be cited in future publications.
Neural ordinary differential equations (NODEs) have recently regained popularity as large-depth limits of a large class of neural networks. In particular, residual neural networks (ResNets) are equivalent to an explicit Euler discretization of an underlying NODE, where the transition from one layer to the next is one time step of the discretization. The relationship between continuous and discrete neural networks has been of particular interest. Notably, analysis from the ordinary differential equation viewpoint can potentially lead to new insights for understanding the behavior of neural networks in general. In this work, we take inspiration from differential equations to define the concept of stiffness for a ResNet via the interpretation of a ResNet as the discretization of a NODE. Here, we then examine the effects of stiffness on the ability of a ResNet to generalize, via computational studies on example problems coming from climate and chemistry models. We find that penalizing stiffness does have a unique regularizing effect, but we see no benefit to penalizing stiffness over L2 regularization (penalization of network parameter norms) in terms of predictive performance.
Radiation Portal Monitors (RPMs) were deployed throughout the port and border infrastructure of the United States (U.S.) beginning in 2003 to monitor for the possible presence of uncontrolled radiological and nuclear materials. Since that time, the U.S. Government (USG) has learned much about the operational challenges faced in the field. Principal among the shortcomings has been the lack of flexibility afforded the USG when all Internet Protocol (IP) rights and interfaces of the system are owned by the Original Equipment Manufacturer (OEM).
Criticality Control Overpack (CCO) containers are being considered for the disposal of defense-related nuclear waste at the Waste Isolation Pilot Plant (WIPP). At WIPP, these containers would be placed in underground disposal rooms, which will naturally close and compact the containers closer to one another over several centuries. This report details simulations to predict the final container configuration as an input to nuclear criticality assessments. Each container was discretely modeled, including the plywood and stainless steel pipe inside the 55-gallon drum, in order to capture its complex mechanical behavior. Although these high-fidelity simulations were computationally intensive, several different material models were considered in an attempt to reasonably bound the horizontal and vertical compaction percentages. When exceptionally strong materials were used for the containers, the horizontal and vertical closure respectively stabilized at 43:9 % and 93:7 %. At the other extreme, when the containers completely degraded and the clay seams between the salt layers were glued, the horizontal and vertical closure reached respective final values of 48:6 % and 100 %.
The Big Hill SPR site has a rich data set consisting of multi-arm caliper (MAC) logs collected from the cavern wells. This data set provides insight into the on-going casing deformation at the Big Hill site. This report summarizes the MAC surveys for each well and presents well longevity estimates where possible. Included in the report is an examination of the well twins for each cavern and a discussion on what may or may not be responsible for the different levels of deformation between some of the well twins. The report also takes a systematic view of the MAC data presenting spatial patterns of casing deformation and deformation orientation in an effort to better understand the underlying causes. The conclusions present a hypothesis suggesting the small-scale variations in casing deformation are attributable to similar scale variations in the character of the salt-caprock interface. These variations do not appear directly related to shear zones or faults.
We evaluate neural radiance fields (NeRFs) as a method for reconstructing 3D volumetric scenes from low Earth orbit satellite imagery. We leverage commercial satellite data to reconstruct a scene using existing software tools. In doing so, we identify difficulties in these mapping datasets for NeRF generation. We propose potential applications in geospatial intelligence for context and improved image interpretation.
