Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “deep”

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 1,423 records · Page 79

Drop impact into a deep pool: vortex shedding and jet formation

One of the simplest splashing scenarios results from the impact of a single drop on a deep pool. The traditional understanding of this process is that the impact generates an axisymmetric sheet-like jet that later breaks up into secondary droplets. Recently it was shown that even this simplest of scenarios is more complicated than expected because multiple jets can be generated from a single impact event and there are transitions in the multiplicity of jets as the experimental parameters are varied. Here, we use experiments and numerical simulations of a single drop impacting on a deep pool to examine the transition from impacts that produce a single jet to those that produce two jets. Using high-speed X-ray imaging methods we show that vortex separation within the drop leads to the formation of a second jet long after the formation of the ejecta sheet. Using numerical simulations we develop a phase diagram for this transition and show that the capillary number is the most appropriate order parameter for the transition.

breakup/coalescence↗

Deep learning closure models for large-eddy simulation of flows around bluff bodies

Near-wall flow simulation remains a central challenge in aerodynamics modelling: Reynolds-averaged Navier–Stokes predictions of separated flows are often inaccurate, and large-eddy simulation (LES) can require prohibitively small near-wall mesh sizes. A deep learning (DL) closure model for LES is developed by introducing untrained neural networks into the governing equations and training in situ for incompressible flows around rectangular prisms at moderate Reynolds numbers. The DL-LES models are trained using adjoint partial differential equation (PDE) optimization methods to match, as closely as possible, direct numerical simulation (DNS) data. They are then evaluated out-of-sample – for aspect ratios, Reynolds numbers and bluff-body geometries not included in the training data – and compared with standard LES models. The DL-LES models outperform these models and are able to achieve accurate LES predictions on a relatively coarse mesh (downsampled from the DNS mesh by factors of four or eight in each Cartesian direction). We study the accuracy of the DL-LES model for predicting the drag coefficient, near-wall and far-field mean flow, and resolved Reynolds stress. A crucial challenge is that the LES quantities of interest are the steady-state flow statistics; for example, a time-averaged velocity component $\langle {u}_i\rangle (x) = \lim _{t \rightarrow \infty } ({1}/{t}) \int _0^t u_i(s,x)\, {\rm d}s$ . Calculating the steady-state flow statistics therefore requires simulating the DL-LES equations over a large number of flow times through the domain. It is a non-trivial question whether an unsteady PDE model with a functional form defined by a deep neural network can remain stable and accurate on $t \in [0, \infty )$ , especially when trained over comparatively short time intervals. Our results demonstrate that the DL-LES models are accurate and stable over long time horizons, which enables the estimation of the steady-state mean velocity, fluctuations and drag coefficient of turbulent flows around bluff bodies relevant to aerodynamics applications.

Mechanics↗

Deep learning for morphological identification of extended radio galaxies using weak labels

Abstract The present work discusses the use of a weakly-supervised deep learning algorithm that reduces the cost of labelling pixel-level masks for complex radio galaxies with multiple components. The algorithm is trained on weak class-level labels of radio galaxies to get class activation maps (CAMs). The CAMs are further refined using an inter-pixel relations network (IRNet) to get instance segmentation masks over radio galaxies and the positions of their infrared hosts. We use data from the Australian Square Kilometre Array Pathfinder (ASKAP) telescope, specifically the Evolutionary Map of the Universe (EMU) Pilot Survey, which covered a sky area of 270 square degrees with an RMS sensitivity of 25–35 $\mu$ Jy beam $^{-1}$ . We demonstrate that weakly-supervised deep learning algorithms can achieve high accuracy in predicting pixel-level information, including masks for the extended radio emission encapsulating all galaxy components and the positions of the infrared host galaxies. We evaluate the performance of our method using mean Average Precision (mAP) across multiple classes at a standard intersection over union (IoU) threshold of 0.5. We show that the model achieves a mAP $_{50}$ of 67.5% and 76.8% for radio masks and infrared host positions, respectively. The network architecture can be found at the following link: https://github.com/Nikhel1/Gal-CAM

Astronomy & Astrophysics↗

Coupling Microdroplet-Based Sample Preparation, Multiplexed Isobaric Labeling, and Nanoflow Peptide Fractionation for Deep Proteome Profiling of the Tissue Microenvironment

There is increasing interest in developing in-depth proteomic approaches for mapping tissue heterogeneity in a cell-type-specific manner to better understand and predict the function of complex biological systems such as human organs. Existing spatially resolved proteomics technologies cannot provide deep proteome coverage due to limited sensitivity and poor sample recovery. Herein, we seamlessly combined laser capture microdissection with a low-volume sample processing technology that includes a microfluidic device named microPOTS (microdroplet processing in one pot for trace samples), multiplexed isobaric labeling, and a nanoflow peptide fractionation approach. The integrated workflow allowed us to maximize proteome coverage of laser-isolated tissue samples containing nanogram levels of proteins. We demonstrated that the deep spatial proteomics platform can quantify more than 5000 unique proteins from a small-sized human pancreatic tissue pixel (∼60,000 μm2) and differentiate unique protein abundance patterns in pancreas. Furthermore, the use of the microPOTS chip eliminated the requirement for advanced microfabrication capabilities and specialized nanoliter liquid handling equipment, making it more accessible to proteomic laboratories.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Probabilistic Deep Learning Approach to Automate the Interpretation of Multi-phase Diffraction Spectra

Autonomous synthesis and characterization of inorganic materials requires the automatic and accurate analysis of X-ray diffraction spectra. For this task, we designed a probabilistic deep learning algorithm to identify complex multi-phase mixtures. At the core of this algorithm lies an ensemble convolutional neural network trained on simulated diffraction spectra, which are systematically augmented with physics-informed perturbations to account for artifacts that can arise during experimental sample preparation and synthesis. In this work, larger perturbations associated with off-stoichiometry are also captured by supplementing the training set with hypothetical solid solutions. Spectra containing mixtures of materials are analyzed with a newly developed branching algorithm that utilizes the probabilistic nature of the neural network to explore suspected mixtures and identify the set of phases that maximize confidence in the prediction. Our model is benchmarked on simulated and experimentally measured diffraction spectra, showing exceptional performance with accuracies exceeding those given by previously reported methods based on profile matching and deep learning. We envision that the algorithm presented here may be integrated in experimental workflows to facilitate the high-throughput and autonomous discovery of inorganic materials.

36 MATERIALS SCIENCE↗

Toward Stable Deep-Blue Luminescent Colloidal Lead Halide Perovskite Nanoplatelets: Systematic Photostability Investigation

Colloidal lead halide perovskite nanocrystals and nanoplatelets have emerged as promising semiconductor nanomaterials because of their spectral tunability, facile processability, and bright emission with high color purity. In particular, strong quantum and dielectric confinement make atomically thin colloidal lead bromide perovskite nanoplatelets a favorable candidate for next-generation deep-blue-emitting (λ max = 437 nm) materials. However, poor photostability poses a critical challenge; colloidal nanoplatelets suffer from photobleaching or transformation into thicker, less-confined nanostructures with red-shifted emission upon UV irradiation. In this study, we synthesize deep-blue-emitting organic–inorganic hybrid perovskite nanoplatelets (formula: L 2 [ABX 3 ]BX 4 , L: butylammonium and octylammonium, A: methylammonium or formamidinium, B: lead, and X: bromide or iodide) with large lateral dimension (~1 μm) by ligand-assisted reprecipitation and systematically investigate the factors that affect the photostability of those nanoplatelets. We find that freshness of the prepared precursor solutions for ligand-assisted reprecipitation is critical to obtain better stability with high photoluminescence quantum yield of perovskite nanoplatelets. Photobleaching is found to result from intrinsic instability of the perovskite lattice against UV irradiation in nanoplatelets, whereas transformation into thicker nanostructures results from extrinsic factors—moisture, primarily. Furthermore, we observe that substitution of the organic cation from formamidinium to methylammonium and addition of excess alkylammonium bromide ligands significantly enhance both the ambient and photostability. Lastly, we demonstrate that the dropcast film of methylammonium lead bromide nanoplatelets with excess alkylammonium bromide ligands shows dramatically improved stability both under UV irradiation and under ambient conditions. Lastly, this study expands our understanding of the factors that affect perovskite nanoplatelet photostability and opens up new possibilities for the fabrication of stable perovskite nanoplatelet-based optoelectronic devices.

2D↗

Surfactants are Ineffective for Reducing Imbibition of Water-Based Fracturing Fluids in Deep Gas Reservoirs

Minimizing loss of injected hydraulic fracturing fluids into shale along fracture-matrix boundaries is desired because imbibed water restricts gas production and wastes valuable water resources. This problem has motivated the addition of surfactants into water-based hydraulic fracturing fluids in order to reduce the capillary driving force for imbibition. Here, we show that reduction in interfacial tension and wettability alteration has negligible ability to reduce imbibition in deep gas reservoirs. The effectiveness of altering capillary forces acting at the wetting front also depends on the injection pressure acting at the fracture-matrix boundary. The pressure at the interface between the fracture and the shale matrix is constrained between the reservoir pore pressure and formation pressure (rock fracture pressure, also known as breakdown pressure of the rock) and increases with depth to magnitudes that greatly exceed that of capillary pressures. Furthermore, the analyses presented here show that even maximum alteration of interfacial properties that result in strongly hydrophobic interactions between the fracturing fluid and reservoir rock is incapable of significantly reducing imbibition in deep reservoirs. Instead of using surfactants, this analysis points to decreases in wellbore shut-in pressures and shut-in times as practical options for reducing imbibition losses of water-based fluids.

03 NATURAL GAS↗

Ligand- and Structure-Based Analysis of Deep Learning-Generated Potential α2a Adrenoceptor Agonists

The α2a adrenoceptor is a particularly medically relevant subtype of the G protein-couple receptor family. Unfortunately, high-throughput techniques aimed at producing novel drug leads targeted to this receptor have been largely unsuccessful due to the complex pharmacology of adrenergic receptors. As such, cutting-edge in silico ligand- and structure-based assessment and de novo deep learning methods are well-positioned to provide new insights into protein-ligand interactions and potential active compounds. In this work, we (i) collect a dataset of α2a adrenoceptor agonists and provide it as a resource for the drug design community; (ii) use the dataset as a basis to generate candidate active structures via deep learning; and (iii) apply computational ligand- and structure-based analysis techniques to gain new insight into α2a adrenoceptor agonists and assess the quality of the computer-generated compounds. Additionally, we further describe how such assessment techniques can be applied to putative chemical probes with a case study involved proposed medetomidine-based probes.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Coarse-Grained Density Functional Theory Predictions via Deep Kernel Learning

Scalable electronic predictions are critical for soft materials design. Recently, the Electronic Coarse-Graining (ECG) method was introduced to renormalize all-atom quantum chemical (QC) predictions to coarse-grained (CG) resolutions using deep neural networks (DNNs). While DNNs can learn complex representations that prove challenging for kernel-based methods, they are susceptible to overfitting and the overconfidence of uncertainty estimations. Here, we develop ECG within a GPU-accelerated Deep Kernel Learning (DKL) framework to enable CG QC predictions using range-separated hybrid density functional theory (DFT), obtaining a 107 speedup relative to naive all-atom QC. By treating the predicted electronic properties as random Gaussian Processes, DKL incorporates CG mapping degeneracy by learning the distribution of electronic energies as a function of CG configuration. DKL-ECG accurately reproduces molecular orbital energies from range-separated DFT while facilitating efficient training via active learning using the uncertainties provided by DKL. Further, we show that while active learning algorithms enable efficient sampling of a more diverse configurational space relative to random sampling, all explored query methods exhibit comparable performance for the examined system. We attribute this result to the significant overlap of the feature space and output property distributions across multiple temperatures.

97 MATHEMATICS AND COMPUTING↗

Quantum Perturbation Theory Using Tensor Cores and a Deep Neural Network

In this work, time-independent quantum response calculations are performed using Tensor cores. This is achieved by mapping density matrix perturbation theory onto the computational structure of a deep neural network. The main computational cost of each deep layer is dominated by tensor contractions, i.e., dense matrix–matrix multiplications, in mixed-precision arithmetics, which achieves close to peak performance. Quantum response calculations are demonstrated and analyzed using self-consistent charge density-functional tight-binding theory as well as coupled-perturbed Hartree–Fock theory. For linear response calculations, a novel parameter-free convergence criterion is presented that is well-suited for numerically noisy low-precision floating point operations and we demonstrate a peak performance of almost 200 Tflops using the Tensor cores of two Nvidia A100 GPUs.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Adaptive Data-Driven Deep-Learning Surrogate Model for Frontal Polymerization in Dicyclopentadiene

Frontal polymerization (FP) is a self-sustaining curing process that enables rapid and energy-efficient manufacturing of thermoset polymers and composites. Computational methods conventionally used to simulate the FP process are time-consuming, and repeating simulations are required for sensitivity analysis, uncertainty quantification, or optimization of the manufacturing process. Here, in this work, we develop an adaptive surrogate deep-learning model for FP of dicyclopentadiene (DCPD), which predicts the evolution of temperature and degree of cure orders of magnitude faster than the finite-element method (FEM). The adaptive algorithm provides a strategy to select training samples efficiently and save computational costs by reducing the redundancy of FEM-based training samples. The adaptive algorithm calculates the residual error of the FP governing equations using automatic differentiation of the deep neural network. A probability density function expressed in terms of the residual error is used to select training samples from the Sobol sequence space. The temperature and degree of cure evolution of each training sample are obtained by a 2D FEM simulation. The adaptive method is more efficient and has a better prediction accuracy than the random sampling method. With the well-trained surrogate neural network, the FP characteristics (front speed, shape, and temperature) can be extracted quickly from the predicted temperature and degree-of-cure fields.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Deep and Shallow Gap States in Reduced and n-Type Doped m -ZrO 2

Monoclinic zirconium dioxide (m-ZrO 2 ) is a wide-band-gap functional oxide with many applications. Understanding the role of intrinsic defects and dopants is important for improving the properties of m-ZrO 2 relevant to its applications. In this work, we characterize the electronic states and energy levels of oxygen vacancies (VO’s) and n-type Nb, Ta, Rb, H, and F dopants in the bulk and at the majority (1̅11) surface of m-ZrO 2 , using accurate dielectric-dependent hybrid functional calculations. Our results show that VO’s generally behave as deep donors with the excess electrons localized at the vacant O sites in the form of F centers, in agreement with previous studies. In contrast, surface VO’s at two-fold coordinated oxygen sites show a rather shallow (1+/0) transition level. Shallow levels are also obtained for adsorbed hydrogen (H ads ) and fluorine substituting an oxygen atom (FO) either in the bulk or at the surface, whereas Nb and Ta form deep levels. Altogether, our results suggest that FO is the best candidate for realizing n-type conductivity in bulk m-ZrO 2 , while surface VO’s and Hads provide an efficient way to control the surface chemistry of m-ZrO 2 .

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Untapped Potential of Fluoride Ions in Maximizing the Electrochemical Stability of Deep Eutectic Solvents

Advancing the batteries is pivotal to propelling our society toward a sustainable, electrified future. The stability of electrolytes forms the backbone of energy storage systems. This is particularly the case for redox flow batteries (RFBs). Their deployability depends on their longevity and dependability. The presence of unstable electrolytes can trigger undesirable reactions, degrade performance, and lead to battery failure. Aqueous electrolytes, with limited electrochemical stability window (ESW), are prone to hydrogen and oxygen evolution. Conversely, non-aqueous electrolytes offer enhanced stability. In this study, we unveil the ESW of a non-aqueous eutectic solvent, comprising choline fluoride and ethylene glycol—a composition that has eluded experimental investigation until now. Our findings show that the stability window, reduction and oxidation potential limits of deep eutectic solvents, are sensitive to variations in the halide component of the ammonium salt. Furthermore, this work not only highlights the benefits of novel deep eutectic solvents but also sets the stage for their strategic use in future battery electrolytes.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Differential Microscopic Mobility of Components within a Deep Eutectic Solvent

From macroscopic measurements of deep eutectic solvents such as glyceline (1:2 molar ratio of choline chloride to glycerol), the long-range translational diffusion of the larger cation (choline) is known to be slower compared to that of the smaller hydrogen bond donor (glycerol). However, when the diffusion dynamics are analyzed on the subnanometer length scale, we discover that the displacements associated with the localized diffusive motions are actually larger for choline. This counterintuitive diffusive behavior can be understood as follows. The localized diffusive motions confined in the transient cage of neighbor particles, which precede the cage-breaking long-range diffusion jumps, are more spatially constrained for glycerol than for choline because of the stronger hydrogen bonds the former makes with chloride anions. The implications of differential localized mobility of the constituents should be especially important for applications where deep eutectic solvents are confined on the nanometer length scale and their long-range translational diffusion is strongly inhibited (e.g., within microporous media).

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

A Comment on “Deep Proteogenomics of a Photosynthetic Cyanobacterium”

Proteomic researchers strive to achieve complete annotation of protein-coding DNA sequences to provide a foundational context for their relevant biological data. A recent deep proteogenomic study using a photosynthetic cyanobacterium Synechocystis sp. PCC 6803 by Spät et al. proposed 64 refined open reading frames (ORFs). By searching LC-MS/MS data from affinity chromatography-isolated protein complexes, our laboratory identified that six of these high-abundance ORFs possess Nterminal initiation start sites that differ than those proposed in the alternative models. Our findings are supported by highly confident MS2 data, phylogenetic analysis, chemical labeling, and established data from two independent research groups. Based on these highquality experimental identifications, we subsequently propose a standardized strategy and set of criteria for future deep proteogenomic efforts to ensure accurate and stringent proteogenomic annotation.

cyanobacteria↗

Review: Technical and Policy Challenges in Deep Vadose Zone Remediation of Metals and Radionuclides

Remediation of contamination in the deep vadose zone is complicated by heterogeneous contaminant distribution and the difficulty of accessing the contaminant zone and delivering amendments. It is difficult to predict future contaminant movement with sufficient certainty for regulatory decisions, when contamination is left in place. This paper reviews major processes viable for deep vadose zone remediation. Chemical delivery technologies and hybrids combining several remediation mechanisms are considered valuable areas for research and development.

54 ENVIRONMENTAL SCIENCES↗

Modeling crustal deformation and rupture processes related to upwelling of deep CO2-rich fluids during the 1965-1967 Matsushiro Earthquake Swarm in Japan

In Matsushiro, central Japan, a series of more than 700,000 earthquakes occurred over a 2-year period (1965-1967) associated with a strike-slip faulting sequence. This swarm of earthquakes resulted in ground surface deformations, cracking of the topsoil, and enhanced spring-outflows with changes in chemical compositions as well as carbon dioxide (CO{sub 2}) degassing. Previous investigations of the Matsushiro earthquake swarm have suggested that migration of underground water and/or magma may have had a strong influence on the swarm activity. In this study, employing coupled multiphase flow and geomechanical modelling, we show that observed crustal deformations and seismicity can have been driven by upwelling of deep CO{sub 2}-rich fluids around the intersection of two fault zones - the regional East Nagano earthquake fault and the conjugate Matsushiro fault. We show that the observed spatial evolution of seismicity along the two faults and magnitudes surface uplift, are convincingly explained by a few MPa of pressurization from the upwelling fluid within the critically stressed crust - a crust under a strike-slip stress regime near the frictional strength limit. Our analysis indicates that the most important cause for triggering of seismicity during the Matsushiro swarm was the fluid pressurization with the associated reduction in effective stress and strength in fault segments that were initially near critically stressed for shear failure. Moreover, our analysis indicates that a two order of magnitude permeability enhancement in ruptured fault segments may be necessary to match the observed time evolution of surface uplift. We conclude that our hydromechanical modelling study of the Matsushiro earthquake swarm shows a clear connection between earthquake rupture, deformation, stress, and permeability changes, as well as large-scale fluid flow related to degassing of CO{sub 2} in the shallow seismogenic crust. Thus, our study provides further evidence of the important role of deep fluid sources in earthquake fault dynamics and surface deformations.

54 ENVIRONMENTAL SCIENCES↗

Ferrous Iron Under Oxygen‐Rich Conditions in the Deep Mantle

Abstract Recent experiments have demonstrated the existence of previously unknown iron oxides at high pressure and temperature including newly discovered pyrite‐type FeO 2 and FeO 2 H x phases stable at deep terrestrial lower mantle pressures and temperatures. In the present study, we probed the iron oxidation state in high‐pressure transformation products of Fe 3+ OOH goethite by in situ X‐ray absorption spectroscopy in laser‐heated diamond‐anvil cell. At pressures and temperatures of ~91 GPa and 1,500–2,350 K, respectively, that is, in the previously reported stability field of FeO 2 H x , a measured shift of −3.3 ± 0.1 eV of the Fe K‐edge demonstrates that iron has turned from Fe 3+ to Fe 2+ . We interpret this reductive valence change of iron by a concomitant oxidation of oxygen atoms from O 2− to O − , in agreement with previous suggestions based on the structures of pyrite‐type FeO 2 and FeO 2 H x phases. Such peculiar chemistry could drastically change our view of crystal chemistry in deep planetary interiors.

58 GEOSCIENCES↗