SEARCH · Engineering Papers
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.
The data consists of a slide deck that was presented at the INFORMS 2023 conference. The presentation summarizes our approach to learning how to branch and compares our approach to the popular solver SCIP and a state-of-the-art ML-based branching rule.
Poster to explain app that we are presenting at the conference
Explore the source record for details and available documents.
Explore the source record for details and available documents.
Explore the source record for details and available documents.
Explore the source record for details and available documents.
Powerpoint presentation. Background information provided on power plant uprates. Discussion of the current and proposed approaches to power plant uprates. Explanation of what data is used to draft a LAR. Methods such as retrieval augmented generation (RAG) and fine-tuning are discussed. Use case analysis is performed. Different failure types are examined. Conclusions are drawn from the analysis. Future work is proposed.
Summary Distributed temperature sensing (DTS) is a valuable tool to diagnose multistage hydraulic fracture treatments. When a stage interval is shut in, the clusters that take more fluid during pumping warm up more slowly. Therefore, the fluid volume injected into each cluster can be quantitatively interpreted by numerical inversion of the warm-back temperature behavior. This general concept assumes that the different warm-back behavior is controlled by only the injected fluid volume; however, recent observations of DTS data indicate that completion configurations significantly influence the warm-back behavior. This paper investigates the completion effects on the DTS interpretation. In ideal conditions, when a stage is fractured, the upstream stage intervals should show an almost uniform temperature that is close to the injected fluid temperature. This is due to the high fluid velocity of injected fluid in the wellbore, and the upstream intervals have not been perforated (noncommunicating intervals), so the only heat transfer is heat conduction between the wellbore fluid and the surrounding reservoir. But the field DTS data show considerably irregular variations in temperature along the upstream stage intervals. These variations are caused by the completion effects. The nonuniform temperature profile is caused by different heat transfer behavior induced by completion hardware along the production casing string, such as joints, clamps, and blast protectors, and by the sensing cable location in the cement, as well as the cement quality. Because the varying heat transfer behavior impacts the warm-back behavior as well as the temperature profile, the completion effects need to be considered in DTS interpretation. A method of DTS interpretation considering the completion effects to diagnose multistage fracture treatments was developed. Because the heat transfer between a wellbore and a reservoir depends on the overall heat transfer coefficient describing heat conduction through the completion in a forward model, this parameter needs to be tuned all along the wellbore. To calibrate the completion effect, the temperature inversion is conducted using the temperature measured at a stage interval that is upstream of a stage interval currently being treated. Because the interpreted stage interval is not perforated at that time, the thermal behavior at the noncommunicating interval is governed by only the heat conduction through the completion environment. Once the effective values of the overall heat transfer coefficient are estimated along the interpreted stage interval, they can be assumed to be constant physical parameters. Then, the fluid volume distribution is interpreted by using the effective overall heat transfer coefficient profile along each interval. This study provides a field application of the developed interpretation method. The new interpretation method provides more accurate diagnosis of fracture treatments by DTS interpretation.
A genetic algorithm has been developed to optimize the phases of the strongest reflections in SIR/SAD data. This is shown to facilitate density modification and model building in several test cases. Experimental phasing of diffraction data from macromolecular crystals involves deriving phase probability distributions. These distributions are often bimodal, making their weighted average, the centroid phase, improbable, so that electron-density maps computed using centroid phases are often non-interpretable. Density modification brings in information about the characteristics of electron density in protein crystals. In successful cases, this allows a choice between the modes in the phase probability distributions, and the maps can cross the borderline between non-interpretable and interpretable. Based on the suggestions by Vekhter [Vekhter (2005 ▶), Acta Cryst. D61, 899–902], the impact of identifying optimized phases for a small number of strong reflections prior to the density-modification process was investigated while using the centroid phase as a starting point for the remaining reflections. A genetic algorithm was developed that optimizes the quality of such phases using the skewness of the density map as a target function. Phases optimized in this way are then used in density modification. In most of the tests, the resulting maps were of higher quality than maps generated from the original centroid phases. In one of the test cases, the new method sufficiently improved a marginal set of experimental SAD phases to enable successful map interpretation. A computer program, SISA, has been developed to apply this method for phase improvement in macromolecular crystallography.
Modernization of energy systems has led to in- creased interactions among multiple critical infrastructures and diverse stakeholders making the challenge of operational decision making more complex and at times beyond cognitive capabilities of human operators. The state-of-the-art machine learning and deep learning approaches show promise of supporting users with complex decision-making challenges, such as those occurring in our rapidly transforming cyber-physical energy systems. However, successful adoption of data-driven decision support technology for critical infrastructure will be dependent on the ability of these technologies to be trustworthy and contextually interpretable. In this paper, we investigate the feasibility of implementing XAI for interpretable detection of cyberattacks in the energy system. Leveraging a proof-of-concept simulation use case of detection of a data falsification attack on a photovoltaic system using XGBoost algorithm, we demonstrate how Local Interpretable Model-Agnostic Explanations (LIME), a flavor XAI approach, can help provide contextual and actionable interpretation of cyberattack detection.
Deep learning models enable the prediction of clinical endpoints from whole-slide images (WSIs), but many such models function as “black boxes”, lacking transparency about whether and which histomorphological patterns drive their predictions, hindering interpretability and clinical adoption. Here we propose a human-in-the-loop explanation framework, MorphoXAI, which provides both local and global interpretability for deep learning models by incorporating human-expert interpretations. At the global level, it reveals the histomorphological patterns on which the model consistently relies to distinguish between classes of WSIs, as well as the patterns associated with confusion between classes. At the local level, it indicates which of these patterns are used in the prediction of an individual WSI and which regions within the slide correspond to such patterns. We validated our method across multiple deep learning–based WSI analysis tasks spanning different tissue types. The results show that our framework generates explanations that accurately reflect the histomorphology underlying the model’s predictions at both global and local levels. For interpretability and clinical utility in diagnostic contexts, human evaluation results showed that our explanations were easy to interpret, rich in diagnostic features, and directly helpful for diagnostic decision-making, thereby enhancing pathologist-AI collaboration. Our work highlights that unifying global and local explanations and grounding them in expert-interpreted morphology enhances the interpretability and verifiability of deep learning models, thereby facilitating the transparent deployment of such models in clinical practice.
Methods for model explainability have become increasingly critical for testing the fairness and soundness of deep learning. Concept-based interpretability techniques, which use a small set of human-interpretable concept exemplars in order to measure the influence of a concept on a model's internal representation of input, are an important thread in this line of research. In this work we show that these explainability methods can suffer the same vulnerability to adversarial attacks as the models they are meant to analyze. We demonstrate this phenomenon on two well-known concept-based interpretability methods: TCAV and faceted feature visualization. We show that by leveraging the geometry of the problem and carefully perturbing the examples of the concept that is being investigated, we can radically change the output of the interpretability method. The attacks that we propose can either induce positive interpretations (polka dots are an important concept for a model when classifying zebras) or negative interpretations (stripes are not an important factor in identifying images of a zebra). Our work highlights the fact that in safety-critical applications, there is need for security around not only the machine learning pipeline but also the model interpretation process.
We extend and correct a recently proposed maximum-likelihood halo-independent method to analyze unbinned direct dark matter detection data. Instead of the recoil energy as independent variable we use the minimum speed a dark matter particle must have to impart a given recoil energy to a nucleus. This has the advantage of allowing us to apply the method to any type of target composition and interaction, e.g. with general momentum and velocity dependence, and with elastic or inelastic scattering. We prove the method and provide a rigorous statistical interpretation of the results. As first applications, we find that for dark matter particles with elastic spin-independent interactions and neutron to proton coupling ratio f{sub n}/f{sub p}=−0.7, the WIMP interpretation of the signal observed by CDMS-II-Si is compatible with the constraints imposed by all other experiments with null results. We also find a similar compatibility for exothermic inelastic spin-independent interactions with f{sub n}/f{sub p}=−0.8.
Housing markets play a critical role in shaping the spatial and demographic evolution of urban areas. Simulating housing price dynamics can enhance projections of future urban development outcomes. However, traditional hedonic regressions for housing prices, which neglect nonlinear interactions among explanatory variables, often exhibit limited predictive performance. While machine learning (ML) methods can provide a more flexible representation of the relationships between predictors, they are often regarded as “black boxes” due to their complexity and lack of transparency. Interpretable ML techniques provide a promising route by combining the flexibility of ML methods with approaches to analyze the relationships between inputs and outputs. In this study, we employ interpretable ML to analyze the patterns driving the housing market in Baltimore, Maryland, USA. We train an Artificial Neural Network (ANN) to predict Baltimore housing prices based on structural characteristics (e.g., home size, number of stories) and locational attributes (e.g., distance to the city center). We then conduct sensitivity and Partial Dependence Plot (PDP) analyses to interpret the fitted ANN model. We find that the ML model achieves higher predictive accuracy and explains 16 % more of housing price variance than a traditional linear regression model. The interpretable ML model also reveals more nuanced and realistic nonlinear relationships between housing sales price and predictors as well as interactive effects underlying Baltimore home price dynamics. For instance, while the linear model indicates a steady housing price increase over time, our interpretable ML model detects a post-2008 decline, with smaller properties experiencing the sharpest drop.
The Curiosity rover's campaign at Pahrump Hills provides the first analyses of lower Mount Sharp strata. We report ChemCam elemental composition of a diverse assemblage of post-depositional features embedded in, or cross-cutting, the host rock. ChemCam results demonstrate their compositional diversity, especially compared to the surrounding host rock: (i) Dendritic aggregates and relief enhanced features, characterized by a magnesium enhancement and sulfur detection, and interpreted as Mg-sulfates; (ii) A localized observation that displays iron enrichment associated with sulfur, interpreted as Fe-sulfate; (iii) Dark raised ridges with varying Mg- and Ca-enriched compositions compared to host rock; (iv) Several dark-toned veins with calcium enhancement associated with fluorine detection, interpreted as fluorite veins. (v) Light-toned veins with enhanced calcium associated with sulfur detection, and interpreted as Ca-sulfates. The diversity of the Pahrump Hills diagenetic assemblage suggests a complex post-depositional history for fine-grained sediments for which the origin has been interpreted as fluvial and lacustrine. Assessment of the spatial and relative temporal distribution of these features shows that the Mg-sulfate features are predominant in the lower part of the section, suggesting local modification of the sediments by early diagenetic fluids. Conversely, light-toned Ca-sulfate veins occur in the whole section and cross-cut all other features. A relatively late stage shift in geochemical conditions could explain this observation. The Pahrump Hills diagenetic features have no equivalent compared to targets analyzed in other locations at Gale crater. Only the light-toned Ca-sulfate veins are present elsewhere, along Curiosity's path, suggesting they formed through a common late-stage process that occurred at over a broad area.
Electronic excited-state phenomena provide a compelling intersection of fundamental and applied research interests in the chemical sciences. This holds true for coordination chemistry, where harnessing the strong optical absorption and photocatalytic activity of compounds depends on our ability to control fundamental physical and chemical phenomena associated with the nonadiabatic dynamics of electronic excited states. The central events of excited-state chemistry can critically influence the dynamics of electronic excited states, including internal conversion (transitions between distinct electronic states) and intersystem crossing (transitions between electronic states with different spin multiplicities), events governed by nonadiabatic interactions between electronic states in close proximity to conical intersections, as well as solvation and electron transfer. The diversity of electronic and nuclear dynamics also makes the robust interpretation of experimental measurements challenging. Developments in theory, simulation, and experiment can all help address the interpretation and understanding of chemical dynamics in organometallic and coordination chemistry. Synthesis presents the opportunity to chemically engineer the strength and symmetry of the metal–ligand interactions. This chemical control can be exploited to understand the influence of electronic ground state properties on electronic excited-state dynamics. New time-resolved experimental methods and the insightful exploitation of established methods have an important role in understanding, and ideally controlling, the photophysics and photochemistry of transition metal complexes. Techniques that can disentangle the coupled motion of electrons and nuclear dynamics warrant emphasis. In this work, we present a review of electron localization dynamics in charge transfer excited states and the dynamics of photoinitiated spin crossover dynamics. Both electron localization and spin crossover have been investigated by numerous research groups with femtosecond resolution spectroscopy, but challenges in experimental interpretation have left significant uncertainty about the molecular properties that control these phenomena. Our Account will emphasize how tailoring the experimental probe, femtosecond resolution vibrational anisotropy for electron localization, and femtosecond resolution hard X-ray fluorescence for spin crossover can make a significant impact on the interpretability of experimental measurements. The emphasis on thorough and robust interpretation has also led to an emphasis on simpler molecular systems. This enables iteration between experiment and theory, a requirement for the development of a more predictive understanding of electronic excited-state phenomena and an essential step to the development of design rules for solar materials.
The issues associated with long-screened wells (LSWs) (and open boreholes) at contaminated sites are well documented in the groundwater literature but are still not fully appreciated in practice. As established in seminal and review papers going back over three decades, the interpretation of sampling results from LSWs is challenging in the presence of vertical hydraulic gradients and borehole flow; furthermore, LSWs allow for vertical redistribution of contamination between aquifer layers. Acknowledgment of these issues has led to the development of new technologies and well designs to enable discrete-zone monitoring (DZM), yet LSWs remain common for many reasons, for example, as multipurpose wells, for geophysical logging, and (or) as legacy installations. Despite the literature on LSWs and despite the adoption of DZM at many sites, the use of LSWs persists and the challenges of interpreting sampling results from LSWs remain. In this issue paper, we provide a conceptual overview of the problems posed by LSWs and review existing literature and past work to improve the interpretation of sampling in LSWs. We draw on experience from previous studies at the Hanford Site in eastern WA, USA, and use synthetic examples to illustrate key concepts and challenges for interpretation. A recently published analytical modeling framework is used to develop illustrative synthetic examples and demonstrate a workflow for building scientific intuition to understand issues around interpreting samples from LSWs, which is critical to effective characterization and groundwater remediation at sites with LSWs.
Material characterization techniques are widely used to characterize the physical and chemical properties of materials at the nanoscale and, thus, play central roles in material scientific discoveries. However, the large and complex datasets generated by these techniques often require significant human effort to interpret and extract meaningful physicochemical insights. Artificial intelligence (AI) techniques such as machine learning (ML) have the potential to improve the efficiency and accuracy of surface analysis by automating data analysis and interpretation. In this perspective paper, we review the current role of AI in surface analysis and discuss its future potential to accelerate discoveries in surface science, materials science, and interface science. We highlight several applications where AI has already been used to analyze surface analysis data, including the identification of crystal structures from XRD data, analysis of XPS spectra for surface composition, and the interpretation of TEM and SEM images for particle morphology and size. We also discuss the challenges and opportunities associated with the integration of AI into surface analysis workflows. These include the need for large and diverse datasets for training ML models, the importance of feature selection and representation, and the potential for ML to enable new insights and discoveries by identifying patterns and relationships in complex datasets. Most importantly, AI analyzed data must not just find the best mathematical description of the data, but it must find the most physical and chemically meaningful results. In addition, the need for reproducibility in scientific research has become increasingly important in recent years. The advancement of AI, including both conventional and the increasing popular deep learning, is showing promise in addressing those challenges by enabling the execution and verification of scientific progress. By training models on large experimental datasets and providing automated analysis and data interpretation, AI can help to ensure that scientific results are reproducible and reliable. Although integration of knowledge and AI models must be considered for the transparency and interpretability of models, the incorporation of AI into the data collection and processing workflow will significantly enhance the efficiency and accuracy of various surface analysis techniques and deepen our understanding at an accelerated pace.