In many areas of constrained optimization, representing all possible constraints that give rise to an accurate feasible region can be difficult and computationally prohibitive for online use. Satisfying feasibility constraints becomes more challenging in high-dimensional, non-convex regimes which are common in engineering applications. A prominent example that is explored in the manuscript is the security-constrained optimal power flow (SCOPF) problem, which minimizes power generation costs, while enforcing system feasibility under contingency failures in the transmission network. In its full form, this problem has been modeled as a nonlinear two-stage stochastic programming problem. In this work, we propose a hybrid structure that incorporates and takes advantage of both a high-fidelity physical model and fast machine learning surrogates. Neural network (NN) models have been shown to classify highly non-linear functions and can be trained offline but require large training sets. In this work, we present how model-guided sampling can efficiently create datasets that are highly informative to a NN classifier for non-convex functions. We show how the resultant NN surrogates can be integrated into a non-linear program as smooth, continuous functions to simultaneously optimize the objective function and enforce feasibility using existing non-linear solvers. Overall, this allows us to optimize instances of the SCOPF problem with an order of magnitude CPU improvement over existing methods.
Wang, Qian; Guillaume, Joseph H.A.; Jakeman, John D.; Bennett, Frederick R.; Croke, Barry F.W.; Fu, Baihua; Yang, Tao; Jakeman, Anthony J.
Factor Fixing (FF) is a common method for reducing the number of model parameters to lower computational cost. FF typically starts with distinguishing the insensitive parameters from the sensitive and pursues uncertainty quantification (UQ) on the resulting reduced-order model, fixing each insensitive parameter at a fixed value. There is a need, however, to expand such a common approach to consider the effects of decision choices in the FF-UQ procedure on metrics of interest. Therefore, to guide the use of FF and increase confidence in the resulting dimension-reduced model, we propose a new adaptive framework consisting of four principles: (a) re-parameterize the model first to reduce obvious non-identifiable parameter combinations, (b) focus on decision relevance especially with respect to errors in quantities of interest (QoI), (c) conduct adaptive evaluation and robustness assessment of errors in the QoI across FF choices as sample size increases, and (d) reconsider whether fixing is warranted. The framework is demonstrated on a spatially-distributed water quality model. The error in estimates of QoI caused by FF can be estimated using a Polynomial Chaos Expansion (PCE) surrogate model. Built with 70 model runs, the surrogate is computationally inexpensive to evaluate and can provide global sensitivity indices for free. For the selected catchment, just two factors may provide an acceptably accurate estimate of model uncertainty in the average annual load of Total Suspended Solids (TSS), suggesting that reducing the uncertainty in these two parameters is a priority for future work before undertaking further formal uncertainty quantification.
Photocathodes based on GaAs and other III-V semiconductors are capable of producing highly spin-polarized electron beams. GaAs/GaAsP superlattice photocathodes exhibit high spin polarization; however, the quantum efficiency (QE) is limited to 1% or less. To increase the QE, we fabricated a GaAs/GaAsP superlattice photocathode with a Distributed Bragg Reflector (DBR) underneath. This configuration creates a Fabry-Pérot cavity between the DBR and GaAs surface, which enhances the absorption of incident light and, consequently, the QE. These photocathode structures were grown using molecular beam epitaxy and achieved record quantum efficiencies exceeding 15% and electron spin polarization of about 75% when illuminated with near-bandgap photon energies.
Dannemann Dugick, Fransiska K.; Bishop, Jordan W.; Martire, Leo; Iezzi, Alexandra M.; Assink, Jelle D.; Brissaud, Quentin; Arrowsmith, Stephen
This special section of the Bulletin of the Seismological Society of America provides a broad overview on recent advances to the understanding of the seismoacoustic wavefield through 19 articles. Leveraging multiphenomenology datasets is instrumental for the continued success of future planetary missions, nuclear test ban treaty verification, and natural hazard monitoring. Progress in our theoretical understanding of mechanical coupling, advancements in coupled-media wave modeling, and developments of efficient multitechnology inversion procedures are key to fully exploiting geophysical datasets on Earth and beyond. We begin by highlighting papers describing experimental setups and instrumentation, followed by characterization of natural and anthropogenic sources of interest, and ending in new open-access datasets. Finally, we conclude with an overview of challenges that remain as well as some potential directions for future investigation within the growing multidisciplinary field of seismoacoustics.
Threaded fastener behavior can be an important aspect of complex component and system behavior, but there is no one-size-fits-all finite element analysis technique. Proper modeling of threaded fastener joints requires careful consideration of many details, from test setup and data acquisition to constitutive modeling and uncertainty quantification approaches. This report details analysis of a “mini-radax” bolted-joint exemplar where a Discrete-Direct uncertainty quantification approach is employed to evaluate margin of the component. The mini-radax geometry is tested to failure on a drop table, and single-coupon tests of individual fasteners serve as foundational data for the analysis. Analysis predictions complement the test data well and provide additional context for engineering decision-making.
The growth of helium bubbles impacts structural integrity of materials in nuclear applications. Understanding helium bubble nucleation and growth mechanisms is critical for improved material applications and aging predictions. Systematic molecular dynamics simulations have been performed to study helium bubble nucleation and growth mechanisms in Fe70Ni11Cr19 stainless steels. First, helium cluster diffusivities are calculated at a variety of helium cluster sizes and temperatures for systems with and without dislocations. Second, the process of diffusion of helium atoms to join existing helium bubbles is not deterministic and is hence studied using ensemble simulations for systems with and without vacancies, interstitials, and dislocations. We find that bubble nucleation depends on diffusion of not only single helium atoms, but also small helium clusters. Defects such as vacancies and dislocations can significantly impact the diffusion kinetics due to the trapping effects. Vacancies always increase the time for helium atoms to join existing bubbles due to the short-range trapping effect. This promotes bubble nucleation as opposed to bubble growth. Interestingly, dislocations can create a long-range trapping effect that reduces the time for helium atoms to join existing bubbles. This can promote bubble growth within a certain region near dislocations.
This report summarizes the fiscal year 2023 (FY23) status of the second phase of a series of borehole heater tests in salt at the Waste Isolation Pilot Plant (WIPP) funded by the Disposal Research and Development (R&D) program of the Spent Fuel & Waste Science and Technology (SFWST) office at the US Department of Energy’s Office of Nuclear Energy’s (DOE-NE) Office in the Spent Fuel and Waste Disposition (SFWD) program.
An additive manufacturing approach combining aerosol jet printing (AJP) and electrodeposition opens a new pathway to the production of lightweight coreless flyback transformer devices for power electronics. AJP of seed layers with resolution on the order of 30μm is combined with electrodeposition of Cu and Ni for decreased resistance. This combined approach addresses known shortcomings of AJP and electrodeposition. Nanoparticle inks used in AJP of metals have low conductivity versus bulk materials due to their high grain boundary resistance. There is a lack of readily available high-resolution patterning techniques for electrodeposition outside of expensive clean-room-based lithography techniques. Combining these two techniques enables the patterning of high-resolution, high-conductivity components. In this manuscript, we report on the construction of coreless flyback transformers consisting of two-layer primary and two-layer secondary spiral inductors separated by layers of a printed UV-curable dielectric. An input voltage of 17 V at 400 kHz was amplified to an output of 1250 V corresponding to a gain of 73.5. COMSOL modeling at the individual inductor level and at the transformer level was used to compare expected inductance, equivalent series resistance (ESR), and coupling with experimentally measured values.
Earthquakes have repeatedly been shown to produce inaudible acoustic signals (< 20 Hz), otherwise known as infrasound. These signals can propagate hundreds to thousands of kilometers and still be detected by ground-based infrasound arrays depending on the source strength, distance between source and receiver, and atmospheric conditions. Another type of signal arrival at infrasound arrays is the seismic induced motion of the sensor itself, or ground-motion-induced sensor noise. Measured acoustic and seismic waves produced by earthquakes can provide insight into properties of the earthquake such as magnitude, depth, and focal mechanism, as well as information about the local lithology and atmospheric conditions. Large earthquakes that produce strong acoustic signals detected at distances greater than 100 km are the most commonly studied; however, more recent studies have found that smaller magnitude earthquakes (Mw < 2:0) can be detected at short ranges. In that vein, this study will investigate the ability for a long-term deployment of infrasound sensors (deployed as part of the Source Physics Experiments [SPE] from 2014 to 2020) to detect both seismic and infrasonic signals from earthquakes at local ranges (< 50 km). Methods used include a combination of spectral analysis and automated array processing, supported by U.S. Geological Survey earthquake bulletins. This investigation revealed no clear acoustic detections for short range earthquakes. However, secondary infrasound from an Mw 7.1 earthquake over 200 km away was detected. Important insights were also made regarding the performance of the SPE networks including detections of other acoustic sources such as bolides and rocket launches. Finally, evaluation of the infrasound arrays is performed to provide insight into optimal deployments for targeting earthquake infrasound.
When high-energy-density materials are subjected to thermal or mechanical insults at extreme conditions (shock loading), a coupled response between the thermo-mechanical and chemical behaviors is systematically induced. Herein we develop a reaction model for the fast chemistry of 1,3,5-triamino-2,4,6-trinitrobenzene (TATB) at the mesoscopic scale, where the chemical behavior is determined by underlying microscopic reactive simulations. The slow carbon cluster formation is not discussed in the present work. All-atom reactive molecular dynamics (MD) simulations are performed with the ReaxFF potential, and a reduced-order chemical kinetics model for TATB is fitted to isothermal and adiabatic simulations of single crystal chemical decomposition. Unsupervised machine learning techniques based on non-negative matrix factorization are applied to MD trajectories to model the decomposition kinetics of TATB in terms of a four-component model. The associated heats of reaction are fit to the temperature evolution from adiabatic decomposition trajectories. Using a chemical species analysis, we show that non-negative matrix factorization captures the main chemical decomposition steps of TATB and provides an accurate estimation of their evolution with temperature. The final analytical formulation, coupled to a diffusion term, is incorporated into a continuum formalism, and simulation results are compared one-to-one against MD simulations of 1D reaction propagation along different crystallographic directions and with different initial temperatures. A good agreement is found for both the temporal and spatial evolution of the temperature field.
We present a graph neural network approach that fully automates the prediction of defect formation enthalpies for any crystallographic site from the ideal crystal structure, without the need to create defected atomic structure models as input. Here we used density functional theory reference data for vacancy defects in oxides, to train a defect graph neural network (dGNN) model that replaces the density functional theory supercell relaxations otherwise required for each symmetrically unique crystal site. Interfaced with thermodynamic calculations of reduction entropies and associated free energies, the dGNN model is applied to the screening of oxides in the Materials Project database, connecting the zero-kelvin defect enthalpies to high-temperature process conditions relevant for solar thermochemical hydrogen production and other energy applications. The dGNN approach is applicable to arbitrary structures with an accuracy limited principally by the amount and diversity of the training data, and it is generalizable to other defect types and advanced graph convolution architectures. It will help to tackle future materials discovery problems in clean energy and beyond.