Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Relative accuracy”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 1,351 records · Page 75

Importance of self-interaction-error removal in density functional calculations on water cluster anions

Accurate description of the excess charge in water cluster anions is challenging for standard semi-local and (global) hybrid density functional approximations (DFAs). Using the recent unitary invariant implementation of the Perdew–Zunger self-interaction correction (SIC) method using Fermi-Löwdin orbitals, we assess the effect of self-interaction error on the vertical detachment energies of water cluster anions with the local spin density approximation (LSDA), Perdew–Burke–Ernzerhof (PBE) generalized gradient approximation, and the strongly constrained and appropriately normed (SCAN) meta-GGA functionals. Our results show that for the relative energies of isomers with respect to reference CCSD(T) values, the uncorrected SCAN functional has the smallest deviation of 21 meV, better than that for the MP2 method. The performance of SIC-SCAN is comparable to that of MP2 and is better than SIC-LSDA and SIC-PBE, but it reverses the ordering of the two lowest isomers for water hexamer anions. Removing self interaction error (SIE) corrects the tendency of LSDA, PBE, and SCAN to over-bind the extra electron. The vertical detachment energies (VDEs) of water cluster anions, obtained from the total energy differences of corresponding anion and neutral clusters, are significantly improved by removing self-interaction and are better than the hybrid B3LYP functional, but fall short of MP2 accuracy. Removing SIE results in substantial improvement in the position of the eigenvalue of the extra electron. The negative of the highest occupied eigenvalue after SIC provides an excellent approximation to the VDE, especially for SIC-PBE where the mean absolute error with respect to CCSD(T) is only 17 meV, the best among all approximations compared in this work.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Seismic Tomographic Modeling of the Crust and Upper Mantle beneath Israel and the Middle East: Improved Resolution through Optimized Model Parameterization

Accurate regional seismic travel-time (RSTT) predictions rely on regional phases (e.g., Pg, Lg, Pn, Sn ) to account for 3D effects in the crust and upper mantle that are not captured by 1D models traditionally used for real-time location. The RSTT prediction model accounts for regional-scale crust and upper mantle structure globally by incorporating regional seismic phases into its travel-time calculations. Previous versions of the RSTT model have used a constant grid cell size of 1°. To improve the tomographic accuracy of recovering velocity structure at regional scales, we, in this study, perform data-driven grid refinement on the RSTT model down to a 0.125° grid (~14 km) in pursuit of two main goals: (1) to test the limits of RSTT capability and accuracy of determined velocity structure through variable grid refinement and (2) to image smaller structures in Israel and the Middle East and illuminate upper mantle dynamics operating in this complex tectonic area. We investigate the effects of model parameterization as grid cell size decreases and the trade-offs between recovered velocity structures. Our final dataset includes 4751 events and 499 stations that recorded 79,344 Pn and 7489 Pg . The variable grid refinement method allows recovery of finer-scale velocity structures and reduces travel-time residuals in areas with the highest data coverage. At smaller grid cell sizes, longer paths need to be upweighted to stabilize the inversion. Results illuminate tectonic features undefined in coarser grid-size models; in particular, we observe mantle perturbations related to the subduction zone around the Cyprian arc and crustal anomalies near the Dead Sea fault and throughout the Anatolian plate.

58 GEOSCIENCES↗

A Software/Hardware Framework for Efficient and Safe Emergency Response in Post-Crash Scenarios of Battery Electric Vehicles

The adoption rate of battery electric vehicles (EVs) is rapidly increasing. Electric vehicles differ significantly from conventional internal combustion engine vehicles and vary widely across different manufacturers. Emergency responders (ERs) and recovery personnel may have less experience with EVs and lack timely access to critical information such as the extent of the stranded energy present, high-voltage safety hazards, and post-crash handling procedures in a user-friendly manner. This paper presents a software/hardware interactive tool named Electric Vehicle Information for Incident Response Solutions (EVIRS) to aid in the quick access to emergency response and recovery information. The current prototype of EVIRS identifies EVs using the VIN or Make, Model, and Year, and offers several useful features for ERs and recovery personnel. These features include integration and easy access to emergency response procedures tailored to an identified EV, vehicle structural schematics, the quick identification of battery pack specifications, and more. For EVs that are not severely damaged, EVIRS can perform calculations to estimate stranded energy in the EV’s battery and discharge time for various power loads using either EV dashboard information or operational data accessed through the CAN interface. Knowledge of this information may be helpful in the post-crash handling, management, and storage of an EV. The functionality and accuracy of EVIRS were demonstrated through laboratory tests using a 2021 Ford Mach-E and associated data acquisition system. The results indicated that when the remaining driving range was used as an input, EVIRS was able to estimate the pack voltage with an error of less than 3 V. Conversely, when pack voltage was used as an input, the estimated state of charge (SOC) error was less than 5% within the range of 30–90% SOC. Additionally, other features, such as retrieving emergency response guides for identified EVs and accessing lessons learned from archived incidents, have been successfully demonstrated through EVIRS for quick access. EVIRS can be a valuable tool for emergency responders and recovery personnel, both in action and during offline training, by providing crucial information related to assessing EV/battery safety risks, appropriate handling, de-energizing, transport, and storage in an integrated and user-friendly manner.

25 ENERGY STORAGE↗

Field validation of dynamic mechanical torque measurements using fiber-optic strain sensors for geared wind turbines

Abstract Accurate knowledge of the mechanical loads of wind turbine gearboxes has become essential in modern, highly loaded gearbox designs, as maintaining or even improving gearbox reliability with increasing torque density demands is proving to be challenging. Unfortunately, the traditional method of measuring dynamic mechanical torque using strain gauges placed on the outer surface of a rotating shaft and transmitting the resulting signal is unsuitable for serial deployment due to technical and economic constraints. An alternative method based on fiber-optic strain sensors placed on the stationary outer surface of the gearbox ring gear has been proposed. Like shaft torsion, the radial deformation of the ring gear is proportionate to the rotor torque. Placing the sensors on a stationary component is a cost-effective alternative for serial implementation because the need for complex and expensive data transfer via wireless transmission or a slip ring is eliminated. In this paper, we present the results of an extensive field experiment conducted to evaluate the torque measurement accuracy of this novel sensing solution installed on the gearbox of a Gamesa G97 2-MW wind turbine at the National Renewable Energy Laboratory’s Flatirons Campus. Torque measurements derived from fiber-optic strain sensors placed on the ring gear of the planetary stage are compared to conventional torque measurements from strain gauges placed on the main shaft. Two different torque estimation data processing methods were evaluated, with the method based on operational deflection shapes providing the most accurate results with an average normalized root mean square error below 0.7% for a load revolution distribution analysis. The effect of operating conditions on the torque estimate was also investigated, and the third planet-passing operational deflection shape was found to be the least sensitive to nontorque load-related effects. The fiber-optic strain sensors’ successful operation during the complete test campaign has demonstrated a robust and accurate solution for fleet-wide enhanced gearbox remaining useful life estimation.

17 WIND ENERGY↗

Vapor–liquid equilibrium of water with the MB-pol many-body potential

Among the many existing molecular models of water, the MB-pol many-body potential has emerged as a remarkably accurate model, capable of reproducing thermodynamic, structural, and dynamic properties across water's solid, liquid, and vapor phases. In this work, we assessed the performance of MB-pol with respect to an important set of properties related to vapor-liquid coexistence and interfacial behavior. Through direct coexistence classical molecular dynamics simulations at temperatures of 400 K < T < 600 K, we calculated properties such as equilibrium coexistence densities, vapor-liquid interfacial tension, vapor pressure, and enthalpy of vaporization and compared the MB-pol results to experimental data. We also compared rigid vs fully flexible variants of the MB-pol model and evaluated system size effects for the properties studied. We found that the MB-pol model predictions are in good agreement with experimental data, even for temperatures approaching the vapor-liquid critical point; this agreement was largely insensitive to system sizes or the rigid vs flexible treatment of the intramolecular degrees of freedom. Furthermore, these results attest to the chemical accuracy of MB-pol and its high degree of transferability, thus enabling MB-pol's application across a large swath of water's phase diagram.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Field Validation of Dynamic Mechanical Torque Measurements for Geared Wind Turbines

Accurate knowledge of the mechanical loads of wind turbine gearboxes has become essential in modern, highly loaded gearbox designs, as maintaining or even improving gearbox reliability with increasing torque density demands is proving to be challenging. Unfortunately, the traditional method of measuring dynamic mechanical torque using strain gauges placed on the outer surface of a rotating shaft and transmitting the resulting signal is unsuitable for serial deployment due to technical and economic constraints. An alternative method based on fiber-optic strain sensors placed on the stationary outer surface of the gearbox ring gear has been proposed. Like shaft torsion, the radial deformation of the ring gear is proportionate to the rotor torque. Placing the sensors on a stationary component is a cost-effective alternative for serial implementation because the need for complex and expensive data transfer via wireless transmission or a slip ring is eliminated. In this paper, we present the results of an extensive field experiment conducted to evaluate the torque measurement accuracy of this novel sensing solution installed on the gearbox of a Gamesa G97 2-MW wind turbine at the National Renewable Energy Laboratory's Flatirons Campus. Torque measurements derived from fiber-optic strain sensors placed on the ring gear of the planetary stage are compared to conventional torque measurements from strain gauges placed on the main shaft. Two different torque estimation data processing methods were evaluated, with the method based on operational deflection shapes providing the most accurate results with an average normalized root mean square error below 0.7% for a load revolution distribution analysis. The effect of operating conditions on the torque estimate was also investigated, and the third planet-passing operational deflection shape was found to be the least sensitive to nontorque load-related effects. The fiber-optic strain sensors' successful operation during the complete test campaign has demonstrated a robust and accurate solution for fleet-wide enhanced gearbox remaining useful life estimation.

17 WIND ENERGY↗

Predictive Modeling of NOx Emissions from Lean Direct Injection of Hydrogen and Hydrogen/Natural Gas Blends Using Flame Imaging and Machine Learning

This research paper explores the use of machine learning to relate images of flame structure and luminosity to measured NOx emissions. Images of reactions produced by 16 aero-engine derived injectors for a ground-based turbine operated on a range of fuel compositions, air pressure drops, preheat temperatures and adiabatic flame temperatures were captured and postprocessed. The experimental investigations were conducted under atmospheric conditions, capturing CO, NO and NOx emissions data and OH* chemiluminescence images from 27 test conditions. The injector geometry and test conditions were based on a statistically designed test plan. These results were first analyzed using the traditional analysis approach of analysis of variance (ANOVA). The statistically based test plan yielded 432 data points, leading to a correlation for NOx emissions as a function of injector geometry, test conditions and imaging responses, with 70.2% accuracy. As an alternative approach to predicting emissions using imaging diagnostics as well as injector geometry and test conditions, a random forest machine learning algorithm was also applied to the data and was able to achieve an accuracy of 82.6%. This study offers insights into the factors influencing emissions in ground-based turbines while emphasizing the potential of machine learning algorithms in constructing predictive models for complex systems.

08 HYDROGEN↗

Vertical meteorological patterns and their impact on the energy demand of tall buildings

In accordance with the developing economy and growing population, an increasing number of tall buildings have been constructed over the last 20 years. In 2017, there were 144 new buildings worldwide that were 200 m or greater in height; 50% of these buildings were in China. Due to the vertical gradient impact of meteorological parameters, the energy performance of tall buildings differs from that of general buildings. Few studies exist on vertical meteorological changes using measured data at different heights. Most studies on dynamic energy simulation simulate meteorological parameters using models. This study explores vertical meteorological patterns using hourly dry-bulb temperature, relative humidity, and wind speed data from 2007 to 2017 for a 325 m meteorological tower in Beijing. The temperature decreased significantly with increasing altitude (~0.9 °C per 100 m), while the daily temperature difference decreased with increasing altitude. The absolute humidity did not change significantly with height. The wind speed increased with altitude at approximately 2 m/s per 100 m. The building simulation showed that the annual heating load at a height of 320 m increased by 85% from that at 8 m; the annual cooling load decreased by 20%. Such significant differences in thermal loads for 300-m-tall buildings confirm the need to consider vertical meteorological differences in building performance simulations for tall buildings. A greater number of measurement points at different heights improve the simulation accuracy. Finally, guidance on selecting the heights for vertical metrological measurements is provided based on the influences of building thermal loads.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗

COVID-19 Exposure Assessment Tool (CEAT): Exposure quantification based on ventilation, infection prevalence, group characteristics, and behavior

The coronavirus disease 2019 (COVID-19) Exposure Assessment Tool (CEAT) allows users to compare respiratory relative risk to severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) for various scenarios, providing understanding of how combinations of protective measures affect risk. CEAT incorporates mechanistic, stochastic, and epidemiological factors including the (i) emission rate of virus, (ii) viral aerosol degradation and removal, (iii) duration of activity/exposure, (iv) inhalation rates, (v) ventilation rates (indoors/outdoors), (vi) volume of indoor space, (vii) filtration, (viii) mask use and effectiveness, (ix) distance between people (taking into account both near-field and far-field effects of proximity), (x) group size, (xi) current infection rates by variant, (xii) prevalence of infection and immunity in the community, (xiii) vaccination rates, and (xiv) implementation of COVID-19 testing procedures. CEAT applied to published studies of COVID-19 transmission events demonstrates the model’s accuracy. We also show how health and safety professionals at NASA Ames Research Center used CEAT to manage potential risks posed by SARS-CoV-2 exposures.

60 APPLIED LIFE SCIENCES↗

Real-time nuclear activation detectors for measuring neutron angular distributions at the National Ignition Facility (invited)

The Real Time Nuclear Activation Detector (RTNAD) array at NIF measures the distribution of 14 MeV neutrons emitted by deuterium-tritium (DT) fueled inertial confinement fusion implosions. The uniformity of the neutron distribution is an important indication of implosion symmetry and DT shell integrity. The array consists of 48 LaBr 3 (Ce) crystal gamma-ray spectrometers mounted outside the NIF target chamber, which continuously monitor the slow decay of the 909 keV gamma-ray line from activated 89 Zr located in Zr cups surrounding each crystal. The measured decay rate dramatically increases during a DT implosion in proportion to the number of 14 MeV neutrons striking each Zr cup. The neutrons produce activated 89 Zr through an (n, 2n) reaction on 90 Zr, which is insensitive to low energy neutrons. The neutron flux along the detector line-of-sight at shot time is determined by extrapolating the fitted 909 keV decay curve back to shot time. Automatic analysis algorithms were developed to handle the non-stop data stream. The large number of detectors and the high statistical accuracy of the array enable the spherical harmonic modes of the neutron angular distribution to be measured up to L ≤ 4 to provide a better understanding of implosion dynamics. In addition, these data combined with measurements of the down-scattered neutrons can be used to derive fuel areal density distributions. This paper will describe the RTNAD hardware and analysis procedures.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

Simulating water adsorption in metal–organic frameworks with open metal sites using the 12-6-4 Lennard–Jones potential

Metal-organic frameworks (MOFs) with coordinatively unsaturated open metal sites (OMSs) are promising sorbent materials in atmospheric water harvesting (AWH) systems at low relative humidity (RH) due to their strong interactions with water molecules. However, accurate computational modelling of water adsorption in those materials are challenging, as standard force fields (FFs) based on the 12-6 Lennard-Jones (L-J) potential cannot properly describe the water-OMS interactions. In biomolecular simulations, the 12-6-4 L-J potential has been successfully used to model metal ion-water interactions in which the extra 1/r 4 -term is for charge-induced dipole electrostatics. In this work, we adopted this strategy and used the 12-6-4 L-J potential to model water-OMS interactions in MOFs for low RH AWH applications. Notably, without any modifications, the parameters used for aqueous metal ions were able to greatly improve the accuracy in predicting water adsorption isotherms in MOF-74, which highlights the simplicity and transferability of this method.

74 ATOMIC AND MOLECULAR PHYSICS↗

A Metallic Fuel Performance Benchmark Problem Based on the IFR-1 Experiment

Metallic nuclear fuels are an active area of research and development for use in advanced reactors. Robust and accurate metallic fuel performance models are necessary for the design, safety analysis, and licensing of such reactors. However, metallic fuel performance models are not as mature as ceramic fuel performance models and require additional development. In order to better gauge the accuracy of existing models, identify high-priority models for development, and quantify the improvements made by future model development, ORNL has developed a benchmark problem based on the IFR-1 experiment. This work gathers all relevant information on the IFR-1 experiment and uses it to develop the benchmark problem. It simulates the problem using the fuel performance code BISON and compare the results to post-irradiation examination data from the experiment. Then a sensitivity analysis is performed on the BISON model to determine the benchmark problem’s sensitivity to uncertainty in the input parameters. The results suggest that BISON’s mechanical models require additional development. Plastic deformation is under-predicted in the cladding and axial swelling is over-predicted in the fuel. These problems may be related. Additionally, there is a bias in the temperature which may be a result of uncertainty in the input parameters rather than an issue with the fuel performance models.

12 MANAGEMENT OF RADIOACTIVE AND NON-RADIOACTIVE W↗

Quantum discriminator for binary classification

Abstract Quantum computers have the unique ability to operate relatively quickly in high-dimensional spaces—this is sought to give them a competitive advantage over classical computers. In this work, we propose a novel quantum machine learning model called the Quantum Discriminator, which leverages the ability of quantum computers to operate in the high-dimensional spaces. The quantum discriminator is trained using a quantum-classical hybrid algorithm in $$\mathcal {O}(N\log N)$$ O ( N log N ) time, and inferencing is performed on a universal quantum computer in $$\mathcal {O}(N)$$ O ( N ) time. The quantum discriminator takes as input the binary features extracted from a given datum along with a prediction qubit, and outputs the predicted label. We analyze its performance on the Iris and Bars and Stripes data sets, and show that it can attain 99% accuracy in simulation.

97 MATHEMATICS AND COMPUTING↗

Consistency and reproducibility in atomic layer deposition

Atomic layer deposition (ALD) is a thin film synthesis technique that can provide exquisite accuracy and precision in film thickness and composition even on complex, large area substrates. Based on self-limiting surface chemistry, ALD can be insensitive to process conditions and reactor designs, allowing an ALD process developed in one lab to be easily reproduced in other labs. In practice, however, ALD is sometimes difficult to reproduce or replicate, and the results can vary substantially between ALD reactors and between labs. This is exemplified by large deviations in reports on the growth of, e.g., Al2O3, FeOx, and TiO2 given the same precursors under similar conditions. Furthermore, the problem of irreproducibility seems to be growing as ALD is adopted by more researchers and integrated into new applications. In this article, the authors highlight some of the major sources of variations and errors and common misconceptions related to ALD. In particular, the authors focus on issues related to precursors, substrates, and deposition tools. The authors illustrate these problems through examples from the literature, and they present results from numerical simulations that describe how nonidealities would manifest in thickness profiles in a typical cross-flow reactor. They also describe how reproducibility in ALD is linked to consistent experimental practice and reporting between labs. The authors’ hope is that by educating newcomers to ALD and advocating for consistent reporting of deposition conditions, they can minimize irreproducibility and enable ALD practitioners to realize the full potential afforded by self-limiting surface chemistry.

36 MATERIALS SCIENCE↗

Field Validation of MVA Technology for Offshore CCS: Novel Ultra-High-Resolution 3D Marine Seismic Technology (P-Cable) (Final Report)

The objectives of the proposed study were to deploy and validate a specific monitoring technology, high-resolution 3D marine seismic (HR3D), appropriate for large-demonstration and commercial-scale offshore CCS sites. The project accomplished successful acquisition two HR3D seismic surveys. The first HR3D dataset was over the offshore injection site of the Tomakomai, Japan integrated pilot CCS project, which at the time of survey acquisition was actively injecting CO 2 . The first survey also represented a successful international collaboration between the DOE NETL program and Japan’s national CCS program and was the first successful acquisition and use of HR3D over an active CO 2 injection site (Meckel, Feng et al. 2019). The Tomakomai HR3D survey successfully tested a novel 4-streamer HR3D system array in which, for the first time, no cross-cable (aka “P-Cable”) was utilized and only four GeoEel streamers were used instead of the standard 12-streamer configuration. Consequently, this was not, strictly speaking, a deployment of the “P-Cable” system of (Planke and Berndt 2004) but rather a modified version, thereof, and it is the first known demonstration of the modified system configuration. One very positive outcome from the Japanese collaboration earlier in the project was the ability to learn from the Japanese how they used tail buoys with GPS to determine the position of the seismic source and receivers in time and space. Based on that experience, GCCC designed and built six GPS receivers that could be used to position the streamer receivers and the seismic source via tail buoys. A fundamental advance that was made on the original design, was the ability to directly power the tail buoy GPS units and transfer data through the streamers (i.e., vs. the batteries used at Tomakomai). The bulkiness of the GPS batteries caused drag and episodic surging of the buoys, which affected data quality by lifting up the tail end of the streamers so the receivers were not at the same depth. The units were tested onshore for accuracy and functionality, and the design was subsequently and successfully tested in marine acquisition mode during the SLP survey acquisition. The marine acquisition test and survey satisfied Subtasks 2.2.2, Novel Positioning Technology Selection and Subtask 2.2.3, Novel Positioning Technology Deployment. Results of the novel positioning technology selection (Subtask 2.2.2) were considered successful and will be incorporated in future HR3D seismic acquisition projects to reduce costs, improve deployment safety at sea, and integrate both seismic and data recording via a single data transfer through the streamers to the recording system. The project also established a permitting process through NETL NEPA compliance, which included an Environmental Assessment in a marine setting and is required for conducting these types of surveys using Federal funding. The permitting process charted a “boilerplate,” which can allow future surveys related to other funded projects to move forward more expeditiously. Future improvements that could be considered are more robust seals on the GPS module and stronger materials (especially joints) on tail buoy fabrication. These would increase fixed costs, but would be advisable and probably more economic long-term if multiple HR3D surveys are planned. Project Accomplishments include: • Pre-survey Sensitivity Study • Marine geochemistry methods and data analysis • Successful HR3D seismic dataset acquired @ Tomakomai active CO 2 injection marine site • Developed advanced seismic processing techniques • No NRMS anomalies detected in overburden; Demonstration of containment • Repeatability study • Second survey collected @ San Luis Pass, TX • 4D application using positioning techniques developed in the project for monitoring were successful

3D seismic GPS positioning↗

High-Throughput Large-Scale Targeted Proteomics Assays for Quantifying Pathway Proteins in Pseudomonas putida KT2440

Targeted proteomics is a mass spectrometry-based protein quantification technique with high sensitivity, accuracy, and reproducibility. As a key component in the multi-omics toolbox of systems biology, targeted liquid chromatography-selected reaction monitoring (LC-SRM) measurements are critical for enzyme and pathway identification and design in metabolic engineering. To fulfill the increasing need for analyzing large sample sets with faster turnaround time in systems biology, high-throughput LC-SRM is greatly needed. Even though nanoflow LC-SRM has better sensitivity, it lacks the speed offered by microflow LC-SRM. Recent advancements in mass spectrometry instrumentation significantly enhance the scan speed and sensitivity of LC-SRM, thereby creating opportunities for applying the high speed of microflow LC-SRM without losing peptide multiplexing power or sacrificing sensitivity. Here, we studied the performance of microflow LC-SRM relative to nanoflow LC-SRM by monitoring 339 peptides representing 132 enzymes in Pseudomonas putida KT2440 grown on various carbon sources. The results from the two LC-SRM platforms are highly correlated. In addition, the response curve study of 248 peptides demonstrates that microflow LC-SRM has comparable sensitivity for the majority of detected peptides and better mass spectrometry signal and chromatography stability than nanoflow LC-SRM.

59 BASIC BIOLOGICAL SCIENCES↗

On Neural Architectures for Astronomical Time-series Classification with Application to Variable Stars

Despite the utility of neural networks (NNs) for astronomical time-series classification, the proliferation of learning architectures applied to diverse data sets has thus far hampered a direct intercomparison of different approaches. Here we perform the first comprehensive study of variants of NN-based learning and inference for astronomical time series, aiming to provide the community with an overview on relative performance and, hopefully, a set of best-in-class choices for practical implementations. In both supervised and self-supervised contexts, we study the effects of different time-series-compatible layer choices, namely the dilated temporal convolutional neural network (dTCNs), long-short term memory NNs, gated recurrent units and temporal convolutional NNs (tCNNs). Additionally, we also study the efficacy and performance of encoder-decoder (i.e., autoencoder) networks compared to direct classification networks, different pathways to include auxiliary (non-time-series) metadata, and different approaches to incorporate multi-passband data (i.e., multiple time series per source). Performance—applied to a sample of 17,604 variable stars (VSs) from the MAssive Compact Halo Objects (MACHO) survey across 10 imbalanced classes—is measured in training convergence time, classification accuracy, reconstruction error, and generated latent variables. We find that networks with recurrent NNs generally outperform dTCNs and, in many scenarios, yield to similar accuracy as tCNNs. In learning time and memory requirements, convolution-based layers perform better. We conclude by discussing the advantages and limitations of deep architectures for VS classification, with a particular eye toward next-generation surveys such as the Legacy Survey of Space and Time, the Roman Space Telescope, and Zwicky Transient Facility.

79 ASTRONOMY AND ASTROPHYSICS↗