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 379 records · Page 21
All sourcesNASA NTRSDOE OSTI
NASA NTRS2020

Observed Modulation of the Tropical Radiation Budget by Deep Convective Organization and Lower-Tropospheric Stability

This study analyses the observed monthly deseasonalized and detrended variability of the tropical radiation budget, and suggests that variations of the lower-tropospheric stability and of the spatial organization of deep convection both strongly contribute to this variability. Satellite observations show that on average over the tropical belt, when deep convection is more aggregated, the free troposphere is drier, the deep-convective cloud coverage is less extensive, and the emission of heat to space is increased; an enhanced aggregation of deep convection is thus associated with a radiative cooling of the tropics. An increase of the tropical-mean lower-tropospheric stability is also coincident with a radiative cooling of the tropics, primarily because it is associated with more marine low clouds and an enhanced reflection of solar radiation, although the free-tropospheric drying also contributes to the cooling. The contributions of convective aggregation and lower-tropospheric stability to the modulation of the radiation budget are complementary, largely independent of each other, and equally strong. Together, they account for more than sixty percent of the variance of the tropical radiation budget. Satellite observations are thus consistent with the suggestion from modeling studies that the spatial organization of deep convection substantially influences the radiative balance of the Earth. This emphasizes the importance of understanding the factors that control convective organization and lower-tropospheric stability variations, and the need to monitor their changes as the climate warms.

convective aggregation↗
NASA NTRS2017

Deep-Space Navigation Using Optical Communications Systems

Optical communication links using lasers can potentially deliver data rates much higher than those possible using radio frequencies. If optical communications equipment is going to be carried by future deep-space missions, this equipment, with some adaptations, could also be used to perform tracking for trajectory determination. A number of experiments have been performed in Earth orbit and in lunar orbit using optical data links, while other missions have demonstrated optical links over interplanetary distances. Laser ranging using corner cube retroreflectors is a well-established technique that has been used for orbit determination of Earth orbiting spacecraft, for geodesy, and for lunar research, achieving centimeter-level precisions, but it is not a practical method for deep-space distances. There are two main optical tracking types that are being considered for deep-space navigation. The first is optical astrometry of spacecraft: a telescope on the ground images the laser beam coming from a spacecraft against the star background, determining its plane-of-sky position as seen from the observatory. This type will greatly benefit from the release of the high-accuracy star catalog produced by ESA’s Gaia mission, allowing for the generation of plane-of-sky measurements with an accuracy similar to that obtained today using VLBI tracking techniques. The second is optical ranging using active optical systems at both ends of the link, requiring a more careful design of the spacecraft optical communications system. One of the advantages of using optical frequencies is that they are not affected by charged particles in the signal path the way that radio frequencies are, eliminating solar plasma and ionospheric effects from the light-time calculation and the corresponding noise. On the other hand, clouds would preclude any type of optical communication, and daytime light scattering precludes astrometric measurements. This paper presents our analysis so far of the performance that could be achieved using optical data types in a number of deep-space scenarios. One of the questions that we are trying to answer is whether spacecraft equipped with optical communications terminals would also need to carry radio-frequency equipment for navigational purposes. We also want to understand how accurately we will be able to navigate spacecraft in different mission types and phases, and what would be the constraints, advantages, and disadvantages of using optical communications systems for deep-space navigation.

Karimi, Reza↗
NASA NTRS2018

Nature of and Lessons Learned from Lunar Ice Cube and the First Deep Space Cubesat 'Cluster'

Cubesats operating in deep space face challenges Earth-orbiting cubesats do not. 15 deep space cubesat 'prototypes' will be launched over the next two years including the two MarCO cubesats, the 2018 demonstration of dual communication system at Mars, and the 13 diverse cubesats being deployed from the SLS EM1 mission within the next two years. Three of the EM1 cubesat missions, including the first deep space cubesat 'cluster', will be lunar orbiters with remote sensing instruments for lunar surface/regolith measurements. These include: Lunar Ice Cube, with its 1-4 micron broadband IR spectrometer, BIRCHES, to determine volatile distribution as a function of time of day; Lunar Flashlight, to confirm the presence of surface ice at the lunar poles, utilizing an active source (laser), and looking for absorption features in the returning signal; and LunaH-Map to characterize ice at or below the surface at the poles with a compact neutron spectrometer. In addition, the BIRCHES instrument on Lunar Ice Cube will provide the first demonstration of a microcryocooler (AIM/IRIS) in deep space. Although not originally required to do so, all will be delivering science data to the Planetary Data System, the first formal archiving effort for cubesats. 4 of the 20 recently NASA-sponsored (PSDS3) study groups for deep space cubesat/smallsat mission concepts were lunar mission concepts, most involving 12U cubesats. NASA SIMPLEX 2/SALMON 3 AO will create ongoing opportunities for low-cost missions as 'rides' on government space program or private sector vehicles as these become available.

Bujold, Emily↗
NASA NTRS2022

GNSS-RO Deep Refraction Signals from Moist Marine Atmospheric Boundary Layer (MABL)

The marine atmospheric boundary layer (MABL) has a profound impact on sensible heat and moisture exchanges between the surface and the free troposphere. The goal of this study is to develop an alternative technique for retrieving MABL-specific humidity (q) using GNSS-RO data in deep-refracted signals. The GNSS-RO signal amplitude (i.e., signal-to-noise ratio or SNR) at the deep straight-line height (H(SL)) was been found to be strongly impacted by water vapor within the MABL. This study presents a statistical analysis to empirically relate the normalized SNR (S(RO)) at deep HSL to the MABL q at 950 hPa (~400 m). When compared to the ERA5 reanalysis data, a good linear q–S(RO) relationship is found with the deep H(SL) S(RO) data, but careful treatments of receiver noise, SNR normalization, and receiver orbital altitude are required. We attribute the good q–S(RO) correlation to the strong refraction from a uniform, horizontally stratiform and dynamically quiet MABL water vapor layer. Ducting and diffraction/interference by this layer help to enhance the S(RO) amplitude at deep H(SL). Potential MABL water vapor retrieval can be further developed to take advantage of a higher number of S(RO) measurements in the MABL compared to the Level-2 products. A better sampled diurnal variation of the MABL q is demonstrated with the S(RO) data over the Southeast Pacific (SEP) and the Northeast Pacific (NEP) regions, which appear to be consistent with the low cloud amount variations reported in previous studies.

diurnal variation↗
NASA NTRSDate not supplied

Neurovascular Responses to Simulated Deep Space Radiation in a Human Organ-on-a-Chip Model

A major health risk for human deep space exploration is central nervous system (CNS) damage by galactic cosmic ray radiation. Simulated galactic cosmic rays or their components, especially the high-linear energy transfer (LET) particles such as 56Fe ions, have been shown to cause CNS damage, neuroinflammation and cognitive dysfunction in rodent models, but their effects on human CNS remain to be investigated. CNS damage from any insult, including ionizing radiation, is partially mediated by the blood-brain barrier (BBB), which regulates interactions between CNS and the rest of the body. The main cellular regulators of BBB permeability are astrocytes, which also modulate neuroinflammation. However, there have been few studies on BBB and astrocyte functions in regulating CNS responses, especially in human tissue analogs. Therefore, we utilized a high-throughput 3D organ-on-a-chip system, seeded with human induced pluripotent stem cell-derived astrocytes and brain endothelial cells, or brain endothelial cells alone, to study human neurovascular responses to simulated deep space radiation. We investigated the permeability and morphology of vascular structures formed by endothelial cells, as well as oxidative stress and secreted cytokines and chemokine levels over 1-7 days after irradiation with 0.25 – 0.5 Gy 5-ion simplified simulated galactic cosmic rays or 0.3 – 0.8 Gy high-LET 600 MeV/n 56Fe particles, and compared the outcomes to low-LET X-ray irradiation. We observed that simulated deep space radiation caused delayed astrocyte activation in a pattern resembling CNS responses to brain injury, caused oxidative stress and the production of inflammatory cytokines, and compromised BBB integrity by damaging tight junctions, thus increasing vascular permeability. Furthermore, our results indicate that astrocytes have a dual role in regulating radiation responses: they exacerbate blood-brain barrier permeability early after irradiation, followed by switching to a more protective scar-like phenotype by reducing oxidative stress and pro-inflammatory cytokine and chemokine secretion. In a follow-up study using the same platform, we investigated the dose-rate effects of ionizing radiation, by exposing our model to chronic, low dose-rate, gamma radiation. Our model was significantly improved by adding additional cell types composing the BBB, modelling immune cell infiltration into the brain, and studying the effect of an antioxidant, to measure more complex outcomes and model more closely the effect of deep space radiation on the human BBB. In summary, our results present a human neurovascular model for space radiation studies and potential future automated payload adaptation, and suggest astrocyte regulatory mechanisms as targets for countermeasures to mitigate human neurovascular impairments during deep space exploration.

Ionizing radiation↗
NASA NTRS2020

Amino Acid Encoding for Deep Learning Applications

Background: The number of applications of deep learning algorithms in bioinformatics is increasing as they usually achieve superior performance over classical approaches, especially, when bigger training datasets are available. In deep learning applications, discrete data, e.g. words or n-grams in language, or amino acids or nucleotides in bioinformatics, are generally represented as a continuous vector through an embedding matrix. Recently, learning this embedding matrix directly from the data as part of the continuous iteration of the model to optimize the target prediction – a process called ‘end-to-end learning’ – has led to state-ofthe-art results in many fields. Although usage of embeddings is well described in the bioinformatics literature, the potential of end-to-end learning for single amino acids, as compared to more classical manually-curated encoding strategies, has not been systematically addressed. To this end, we compared classical encoding matrices, namely one-hot, VHSE8 and BLOSUM62, to end-to-end learning of amino acid embeddings for two different prediction tasks using three widely used architectures, namely recurrent neural networks (RNN), convolutional neural networks (CNN), and the hybrid CNN-RNN. Results: By using different deep learning architectures, we show that end-to-end learning is on par with classical encodings for embeddings of the same dimension even when limited training data is available, and might allow for a reduction in the embedding dimension without performance loss, which is critical when deploying the models to devices with limited computational capacities. We found that the embedding dimension is a major factor in controlling the model performance. Surprisingly, we observed that deep learning models are capable of learning from random vectors of appropriate dimension. Conclusion: Our study shows that end-to-end learning is a flexible and powerful method for amino acid encoding. Further, due to the flexibility of deep learning systems, amino acid encoding schemes should be benchmarked against random vectors of the same dimension to disentangle the information content provided by the encoding scheme from the distinguishability effect provided by the scheme.

Hesham ElAbd↗
NASA NTRS2022

Deep Space Atomic Clock Technology Demonstration Mission Results

The Deep Space Atomic Clock (DSAC), a NASA Technology Demonstration Mission, was launched into low-Earth orbit on June 25, 2019 as a hosted payload aboard General Atomics’ Orbital Test Bed (OTB) spacecraft. The DSAC mission has been conducting a two-year demonstration of a mercury ion atomic clock to characterize its space-based performance and to validate its utility for deep space navigation and radio science. Analysis of the collected data using JPL’s GIPSY-OASIS software has shown DSAC’s AD at one-day to be near 310-15; much better than required AD of 210-14. Such low spacecraft clock errors will enable one-way radiometric tracking data with precision equivalent to or better than current-day two way tracking data, allowing a shift to a more efficient and flexible one-way deep space navigation architecture. To verify this, an analog deep space navigation experiment was performed using JPL’s operational navigation software (Monte). The experiment recovered orbit solutions with reduced data sets and geometric variations that are more representative of deep space missions, and showed that orbit determination using DSAC derived data is on par with more traditional two-way datatypes. As a technology demonstrator, DSAC’s development focus has been on maturing the mercury ion trap clock technology rather than achieving the smallest size, weight, and power (SWaP). Over the course of DSAC’s development the project has identified numerous improvements that could be made to significantly reduce SWaP for DSAC’s next version. Indeed, DSAC-2 was recently selected by NASA for further demonstration on the VERITAS mission to Venus. This work will review the DSAC technology, mission, and results from its two-year mission.

Wang, Rabi↗
NASA NTRSDate not supplied

Biological Research and Space Health Enabled by Machine Learning to Support Deep Space Missions

A key science goal of the NASA “Moon to Mars” campaign is to understand how biology responds to the Lunar, Martian, and deep space environments in order to advance fundamental knowledge, reduce risk, and support safe, productive human space missions. Through the powerful emerging approaches of artificial intelligence (AI) and machine learning (ML), a paradigm shift has begun in biomedical science and engineered astronaut health systems, to enable Earth independence and autonomy of mission operations. Here we present an overview of AI/ML architecture to support deep space mission goals, developed with leaders in the field. First, we focus on the fundamental biological research that supports our understanding of physiological responses to spaceflight, and we describe current efforts to support AI/ML research including data standardization and data engineering through maximally open and FAIR (findable, accessible, interoperable, reusable) databases and the generation of AI-ready datasets for reuse and analysis. We also discuss remote data management frameworks for research data as well as environmental and health data that are generated during deep space missions. We highlight several research projects that leverage data standardization and management for fundamental biological discovery to uncover the complex effects of space travel on living systems. Next, we provide an overview of cutting-edge AI/ML approaches that can be integrated to support remote monitoring and analysis during deep space missions, including generative models and large language models to learn the underlying biomedical patterns and predict outcomes or answer questions during off world medical scenarios. We also describe current AI/ML methods to support this research and monitoring through automated cloud-based labs which enable limited human intervention and closed-loop experimentation in remote settings. These labs could support mission autonomy by analyzing environmental data streams, and would be facilitated through in situ analytics capabilities to avoid sending large raw data files through low bandwidth communications. Finally, in the context of deep space missions with limited communications or access to medical advice from Earth, we describe a solution for integrated, real-time mission biomonitoring across hierarchical levels from continuous environmental monitoring, to wearables and point-of-care devices, to molecular and physiological monitoring. We introduce a precision space health system that will ensure that the future of space health is predictive, preventative, participatory and personalized.

artificial intelligence↗
NASA NTRSDate not supplied

Addressing the High-Rate Deep Space Communications Shortfall in NASA’s Space Technology Mission Directorate's Envisioned Future

NASA’s Space Technology Mission Directorate (STMD) has identified key technologies needed for future crewed and robotic exploration and science missions. STMD is helping to build the civilian technology base by working with other NASA Mission Directorates, other United States government agencies, commercial industry, and academia to identify technology shortfalls and to develop plans to address them. One critical area of shortfalls lies with deep space communications and navigation. While NASA had huge success to date with the Deep Space Network (DSN), recent studies have shown that without enhancements to current systems, the DSN will be unable to support the anticipated increases in the pace of space exploration or the expected higher data rates from deep space needed soon without severely impacting other missions. High-rate communications from the Moon and beyond is needed to enable future exploration and science missions currently being developed or under consideration. For example, a robust communications infrastructure will be needed to support a sustained human presence on the Moon and its eventual industrialization. High data rate trunk lines between the Earth and the Moon are needed to reduce the number of individual links. The human exploration of Mars will also require high-rate communications between Earth and Mars. Return data rates to Earth from Mars for a single link, for example, are anticipated to be greater than 100 Mb/s; forward data rates to Mars, based on experience from the International Space Station, are anticipated to be greater than 20 Mb/s. Future deep space science missions will also require higher data rates than possible with today’s technology and the current capabilities of the DSN. To support future exploration and science needs, it will be necessary to upgrade the DSN to enhance its radio frequency (RF) capabilities. In addition, it is envisioned that NASA will gradually introduce optical communications to augment its RF systems. Optical communications will enable new science and exploration missions by providing high data rates and better navigation over long distances. This paper will briefly describe STMD’s envisioned future for deep space communications in the 2030+ timeframe and the technology roadmaps being developed for both radio frequency and optical systems.

Bernard Edwards↗
DOE OSTI2017

Towards Scalable Deep Learning via I/O Analysis and Optimization

Deep learning systems have been growing in prominence as a way to automatically characterize objects, trends, and anomalies. Given the importance of deep learning systems, researchers have been investigating techniques to optimize such systems. An area of particular interest has been using large supercomputing systems to quickly generate effective deep learning networks: a phase often referred to as “training” of the deep learning neural network. As we scale existing deep learning frameworks—such as Caffe—on these large supercomputing systems, we notice that the parallelism can help improve the computation tremendously, leaving data I/O as the major bottleneck limiting the overall system scalability. In this paper, we first present a detailed analysis of the performance bottlenecks of Caffe on large supercomputing systems. Our analysis shows that the I/O subsystem of Caffe—LMDB—relies on memory-mapped I/O to access its database, which can be highly inefficient on large-scale systems because of its interaction with the process scheduling system and the network-based parallel filesystem. Based on this analysis, we then present LMDBIO, our optimized I/O plugin for Caffe that takes into account the data access pattern of Caffe in order to vastly improve I/O performance. Our experimental results show that LMDBIO can improve the overall execution time of Caffe by nearly 20-fold in some cases.

Pumma, Sarunya↗
DOE OSTI2017

Addressing Technical and Policy Challenges for Remediation of Metals and Radionuclides within Complex Deep Vadose Zone Environments - 17028

Deep vadose zone contamination is a significant issue in many regions of the world. Although remediation of deep vadose zone contamination is important, direct contact with the contamination does not threaten human health or the environment due to its depth in the vadose zone. Yet, movement of contamination from the deep vadose zone to the groundwater creates the potential for exposure and risk to receptors. Therefore, the challenge of deep vadose zone remediation is to maintain a minimal rate of contaminant transport such that flux from the contaminated vadose zone is limited to protect groundwater resources and down-gradient receptors. We herein review major processes for viable remediation of deep vadose zone metal and radionuclides that form the practical constraints on remedial actions and provide an overview of how these principles are being practiced to address key contaminants at the U.S. Department of Energy Hanford Site. (authors)

12 MANAGEMENT OF RADIOACTIVE AND NON-RADIOACTIVE W↗
DOE OSTI2017

Conceptual Design of the Package Handling System for the Deep Borehole Engineering Demonstration - 17064

Deep borehole disposal involves drilling one or more boreholes into crystalline basement rock to a depth of about 5 km, emplacing waste packages in the lower 2 km of the borehole, and sealing and plugging the upper 3 km. These depths are several times deeper than for typical mined repositories. A deep borehole field test (DBFT) is proposed to confirm the safety and feasibility of the deep borehole concept. The primary goals are to demonstrate the construction and characterization of deep boreholes, demonstrate the equipment and operations used for safe waste handling and emplacement downhole, study geologic controls on waste form stability and isolation, and evaluate the overall safety and practicality of the concept. The test will be conducted using simulated waste packages; no radioactive wastes will be handled at the deep borehole field test site. A key component of the DBFT is an engineering demonstration of the equipment and operations needed to transport waste packages to the site, move packages between the transportation and emplacement systems, and lower or retrieve packages into or out of the borehole. This paper describes the conceptualization of test package handling, from transportation and receipt through the interface with the borehole. It also describes the use of an integrated test facility to check out equipment prior to deployment at the DBFT site, and performance of the engineering demonstration at the DBFT site. (authors)

12 MANAGEMENT OF RADIOACTIVE AND NON-RADIOACTIVE W↗
DOE OSTI2015

A Preliminary Analysis on the Applicability of Deep Borehole Disposal for Spent Fuels in Korea - 15496

A deep geological disposal system, the disposal depth is about 500 m below ground, is considered as the safest method to isolate the spent fuels (SF) or high-level radioactive waste from the human environment with the best available technology at present time. The disposal safety of this system has been demonstrated with underground research laboratory and some advanced countries such as Finland and Sweden are implementing their disposal project on commercial stage. However, if these high-level radioactive wastes can be disposed of in deeper and more stable rock formation than deep geological disposal depth, it has several advantages. Therefore, as an alternative disposal concept, i.e., deep borehole disposal technology is under consideration in number of countries in terms of its outstanding safety and cost effectiveness. In this paper, the general concept for deep borehole disposal of spent fuels or high level radioactive wastes was reviewed. And the key technologies and challenges in development of this disposal method with the nuclear environment were analyzed. Also, very preliminary deep borehole disposal concept including disposal canister concept was developed according to the nuclear environment. These results will be used as an input for the analyses of applicability for DBD in Korea. (authors)

12 MANAGEMENT OF RADIOACTIVE AND NON-RADIOACTIVE W↗
DOE OSTI2018

Identification and Removal of Contaminant Sequences From Ribosomal Gene Databases: Lessons From the Census of Deep Life

Earth’s subsurface environment is one of the largest, yet least studied, biomes on Earth, and many questions remain regarding what microorganisms are indigenous to the subsurface. Through the activity of the Census of Deep Life (CoDL) and the Deep Carbon Observatory, an open access 16S ribosomal RNA gene sequence database from diverse subsurface environments has been compiled. However, due to low quantities of biomass in the deep subsurface, the potential for incorporation of contaminants from reagents used during sample collection, processing, and/or sequencing is high. Thus, to understand the ecology of subsurface microorganisms (i.e., the distribution, richness, or survival), it is necessary to minimize, identify, and remove contaminant sequences that will skew the relative abundances of all taxa in the sample. In this meta-analysis, we identify putative contaminants associated with the CoDL dataset, recommend best practices for removing contaminants from samples, and propose a series of best practices for subsurface microbiology sampling. The most abundant putative contaminant genera observed, independent of evenness across samples, were Propionibacterium, Aquabacterium, Ralstonia, and Acinetobacter. While the top five most frequently observed genera were Pseudomonas, Propionibacterium, Acinetobacter, Ralstonia, and Sphingomonas. The majority of the most frequently observed genera (high evenness) were associated with reagent or potential human contamination. Additionally, in DNA extraction blanks, we observed potential archaeal contaminants, including methanogens, which have not been discussed in previous contamination studies. Such contaminants would directly affect the interpretation of subsurface molecular studies, as methanogenesis is an important subsurface biogeochemical process. Utilizing previously identified contaminant genera, we found that ~27% of the total dataset were identified as contaminant sequences that likely originate from DNA extraction and DNA cleanup methods. Thus, controls must be taken at every step of the collection and processing procedure when working with low biomass environments such as, but not limited to, portions of Earth’s deep subsurface. Taken together, we stress that the CoDL dataset is an incredible resource for the broader research community interested in subsurface life, and steps to remove contamination derived sequences must be taken prior to using this dataset.

16S rRNA↗
DOE OSTI2023

Improving Protein–Ligand Interaction Modeling with cryo-EM Data, Templates, and Deep Learning in 2021 Ligand Model Challenge

Elucidating protein–ligand interaction is crucial for studying the function of proteins and compounds in an organism and critical for drug discovery and design. The problem of protein–ligand interaction is traditionally tackled by molecular docking and simulation, which is based on physical forces and statistical potentials and cannot effectively leverage cryo-EM data and existing protein structural information in the protein–ligand modeling process. In this work, we developed a deep learning bioinformatics pipeline (DeepProLigand) to predict protein–ligand interactions from cryo-EM density maps of proteins and ligands. DeepProLigand first uses a deep learning method to predict the structure of proteins from cryo-EM maps, which is averaged with a reference (template) structure of the proteins to produce a combined structure to add ligands. The ligands are then identified and added into the structure to generate a protein–ligand complex structure, which is further refined. The method based on the deep learning prediction and template-based modeling was blindly tested in the 2021 EMDataResource Ligand Challenge and was ranked first in fitting ligands to cryo-EM density maps. These results demonstrate that the deep learning bioinformatics approach is a promising direction for modeling protein–ligand interactions on cryo-EM data using prior structural information.

59 BASIC BIOLOGICAL SCIENCES↗
NASA NTRS2012

Deep Space Communication

ITU defines deep space as the volume of Space at distances from the Earth equal to, or greater than, 2 106 km. Deep Space Spacecraft have to travel tens of millions of km from Earth to reach the nearest object in deep space. Spacecraft mass and power are precious. Large ground-based antennas and very high power transmitters are needed to overcome large space loss and spacecraft's small antennas and low power transmitters. Navigation is complex and highly dependent on measurements from the Earth. Every deep space mission is unique and therefore very costly to develop.

deep space communications↗
DOE OSTI2023

Recurrent Convolutional Deep Neural Networks for Modeling Time-Resolved Wildfire Spread Behavior

The increasing incidence and severity of wildfires underscores the necessity of accurately predicting their behavior. While high-fidelity models derived from first principles offer physical accuracy, they are too computationally expensive for use in real-time fire response. Low-fidelity models sacrifice some physical accuracy and generalizability via the integration of empirical measurements, but enable real-time simulations for operational use in fire response. Machine learning techniques have demonstrated the ability to bridge these objectives by learning first-principles physics while achieving computational speedups. While deep learning approaches have demonstrated the ability to predict wildfire propagation over large time periods, time-resolved fire-spread predictions are needed for active fire management. Here, in this work, we evaluate the ability of deep learning approaches in accurately modeling the time-resolved dynamics of wildfires. We use an autoregressive process in which a convolutional recurrent deep learning model makes predictions that propagate a wildfire over 15 min increments. We apply the model to four simulated datasets of increasing complexity, containing both field fires with homogeneous fuel distribution as well as real-world topologies sampled from the California region of the United States. We show that even after 100 autoregressive predictions representing more than 24 h of simulated fire spread, the resulting models generate stable and realistic propagation dynamics, achieving a Jaccard score between 0.89 and 0.94 when predicting the resulting fire scar. The inference time of the deep learning models are examined and compared, and directions for future work are discussed.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2021

A Physics-Constrained Deep Learning Model for Simulating Multiphase Flow in 3D Heterogeneous Porous Media

Physics-based simulators for multiphase flow in porous media emulate nonlinear processes with coupled physics, and usually require extensive computational resources for software development, maintenance and simulation execution. As a result, a huge demand exists for fast modeling of coupled processes in a wide range of subsurface applications including geological sequestration, hydrocarbon recovery and geothermal energy extraction. In this work, an efficient physics-constrained deep learning model is developed for solving multiphase flow in 3-Dimensional (3D) heterogeneous porous media. The model fully leverages the spatial topology predictive capability of convolutional neural networks, specifically U-Net with successive contracting and expansive steps, and is coupled with an efficient continuity-based smoother to predict flow responses that need spatial continuity. Furthermore, the transient regions are penalized to steer the training process such that the model can accurately capture flow in these regions. The model takes inputs including properties of porous media, fluid properties and well controls, and predicts the temporal-spatial evolution of the state variables (pressure and saturation). While maintaining the continuity of fluid flow, the 3D spatial domain is decomposed into 2D images for reducing training cost, and the decomposition results in an increased number of training data samples and better training efficiency. Additionally, a surrogate model is separately constructed as a postprocessor to calculate well flow rate based on the predictions of state variables from the deep learning model. We use the example of CO 2 injection into saline aquifers, and apply the physics-constrained deep learning model that is trained from physics-based simulation data and emulates the physics process. The model performs prediction with a speedup of ~ 1400 times compared to physics-based simulations, and the average temporal errors of predicted pressure and saturation plumes are 0.27% and 0.099% respectively. Furthermore, water production rate is efficiently predicted by a surrogate model for well flow rate, with a mean error less than 5%. Therefore, with its unique scheme to cope with the fidelity in fluid flow in porous media, the physics-constrained deep learning model can become an efficient predictive model for computationally demanding inverse problems or other coupled processes.

58 GEOSCIENCES↗