Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Predictive analytics”

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 829 records · Page 46

Experimental Verification of Predicted Oscillations Near a Spin Resonance

The Chao matrix formalism allows analytic calculations of a beam's polarization behavior inside a spin resonance. We recently tested its prediction of polarization oscillations occurring in a stored beam of polarized particles near a spin resonance. Using a 1.85?GeV/c polarized deuteron beam stored in COSY, we swept a new rf solenoid's frequency rather rapidly through 400 Hz during 100 ms, while varying the distance between the sweep's end frequency and the central frequency of an rf-induced spin resonance. Our measurements of the deuteron's polarization for sweeps ending near and inside the resonance agree with the Chao formalism's predicted oscillations.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

PARIS: Predicting application resilience using machine learning

The traditional method to study application resilience to errors in HPC applications uses fault injection (FI), a time-consuming approach. Furthermore, while analytical models have been built to overcome the inefficiencies of FI, they lack accuracy. In this paper, we present PARIS, a machine-learning method to predict application resilience that avoids the time-consuming process of random FI and provides higher prediction accuracy than analytical models. PARIS captures the implicit relationship between application characteristics and application resilience, which is difficult to capture using most analytical models. We overcome many technical challenges for feature construction, extraction, and selection to use machine learning in our prediction approach. Our evaluation on 16 HPC benchmarks shows that PARIS achieves high prediction accuracy. PARIS is up to 450x faster than random FI (49x on average). Compared to the state-of-the-art analytical model, PARIS is at least 63% better in terms of accuracy and has comparable execution time on average.

97 MATHEMATICS AND COMPUTING↗

Application of an Erosion Wear Model to Predict Wear of Hammer Milling Components

A workable analytical erosion model that relates critical hammermill process parameters (hammermill geometry and rotational speed) to critical material attributes of inorganic mineral species in feedstock (density, size, and aspect ratio) and substrate (hardness, toughness, and fatigue ductility) was applied to model wear (more specifically, change in the shape of hammers) in hammer milling operations. Results of the model were compared to experimental observations of the shape of hammers used in the INL Stage 1 Vermeer hammermill. Results showed good agreement between predicted and measured shape of the hammer, and that a quality-by-design (QbD) approach can be developed to predict component reliability based on scientific engineering principles in lieu of trial-and-error approaches.

36 MATERIALS SCIENCE↗

Modeling and Analysis of Coupled Flow Through Packed Beds as a Diagnostic Tool for Packed Bed Reactor Health - 17172

The US Department of Energy, Office of River Protection, is building the Waste Treatment and Immobilization Plant (WTP) in Richland, Washington, to immobilize the hundreds of millions of liters of radioactive waste stored at the Hanford Site in Washington state. The offgas generated during vitrification of these wastes requires treatment to remove and/or destroy chemical and radioactive hazardous components before release to the atmosphere. In particular, volatile and semi-volatile hydrocarbon components are to be catalytically oxidized to carbon dioxide and water, whereas nitrogen oxides (NOx) are to be catalytically reduced to nitrogen and water. These tasks are accomplished in a series of packed bed reactors. Achieving the requisite Destruction/Removal Efficiency (DRE) for each of the tracked components is highly dependent on the flow characteristics of the offgas through the packed beds. The flow kinetics may be monitored and characterized during the initial warm-up of the system. The gases, initially at 76 deg. C (169 deg. F), (a) flow through one side of a countercurrent heat exchanger to recover heat from the exiting gases (which have already passed through the catalytic beds), (b) are then further heated electrically to approximately 400 deg. C (750 deg. F; the preferred operating temperature of the Thermo Catalytic Oxidizer (TCO)), (c) pass through the TCO where the reduced components of the offgas are oxidized over a supported Pt catalyst bed, (d) are partially cooled to approximately 315 deg. C (600 deg. F) by addition of approximately 25% (by weight) diluted ammonia/air mixture, (e) pass through the Selective Catalytic Reducer (SCR) where the NOx components are converted to N{sub 2} and H{sub 2}O over a supported V2O{sub 5}/TiO{sub 2} catalyst bed, and (f) exit through the hot side of the countercurrent heat exchanger. The electrical heater is not powerful enough to raise the temperature of the offgas to the desired 400 deg. C without the contribution of the heat recovered in the heat exchanger. Moreover, the thermal mass of the beds and housings of the TCO and SCR require a significant input of thermal energy to reach the desired operating temperatures; the better part of a day is required to bring the unit from ambient temperatures to operating temperatures. Therefore, the temperature of the stream during the initial warm-up of the system at each point in the system is dependent upon the thermal mass, flow dispersion, and permeability of the components, quantified as characteristic/representative volumes and residence times. By monitoring the temperature of the offgas as it enters and exits each portion of the system during this warm-up period one may model the system as a series of small Continuously Stirred Tank Reactors (CSTR)s (referred to as 'CSTRlets') to explain the flow kinetics through the catalyst beds. These component CSTRlets are adjustable in size and number, that one may model the flow profile of the gases through the catalyst beds. For example, a single large CSTR will present an effluent temperature profile with a simple exponential decay shape. Concatenating more CSTRlets generate more complex sigmoidal responses. Concatenating one CSTRlet significantly larger than the others in the train also alters the shape of the effluent temperature profile, in a fashion not replicable by a train of equal-sized CSTRlets. The ultimate goal is to accurately emulate the entire system as a representative CSTRlet train. Once accomplished, this characterization may be used in subsequent warm-ups (following shutdown for catalyst maintenance, for example) or by an intentional step change in inlet offgas temperature to detect/diagnose anomalies in flow through the beds (e.g., channeling or partial blockage.) This work presents development of this model through analytical, calculational, and empirical means. - The analytic model is accomplished by Laplace transform analysis of each component to recover component process transfer functions. - The calculational portion is performed via brute force finite element calculation. - Fitting empirical data from the warm-up of the TCO skid allows one to select the hypothetical CSTRlet trains which best fit the TCO and SCR beds for the analytical and finite element calculations. The first portion of this paper presents the development of the analytical (Laplace space) model of the coupled system. Estimated values for the initial temperature, mass, specific heat, electrical heat, and relative quantities of the offgas, catalyst beds, stainless steel housings, and ammonia dilution stream are used to generate expected temperature profiles at several selected junctures of the system. The number and size of the component hypothetical CSTRlets are calculated from the analytical model and are used to predict temperature profiles at the exit of the system and at intermediate junctions between system components (where thermocouples are present in the actual unit.) The finite element calculations employ a series of CSTRlet volumes to produce a similar suite of temperature profiles, for comparison with the analytical model results. Finally, empirical data from the Factory Acceptance Test system warm-up are used to select and compare the results from each modeling strategy to select the CSTRlet train(s) which best model the behavior of the actual system. This work presents a novel method of system component characterization, of value to the WTP plant and similar vitrification facilities as well as chemical process industries unrelated to waste management. (authors)

12 MANAGEMENT OF RADIOACTIVE AND NON-RADIOACTIVE W↗

Versatile shaping of a relativistic laser pulse from a nonuniform overdense plasma

We studied the versatile shaping of a petawatt laser pulse using its relativistic transparency in a thin overdense plasma slab. The novel concept here is to use the nonuniformity of the plasma slab in its density or thickness in the transverse direction to control the pulse shaping in both the longitudinal and transverse directions. From 2-dimensional particle-in-cell simulations, we succeeded in fabricating a front shape concave to the propagation direction, an extreme case of transverse shaping. A 1-dimensional analytic formula was then applied to predict the transverse shape, which showed good agreement with the simulations.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

Analytical error bars and RSD for neutron multiplicity counting

This article derives the standard deviation of the moments of neutron multiplicity distributions for both random and (shift register) triggered counting gates by propagating the multinomial distribution of errors in the counting distribution. Exact and approximate analytical formulas for the singles, doubles, and triples error bars are given along with the corresponding RSD (relative standard deviation). These analytical formulas provide a simple way to analyze and predict error bars without needing empirical data or simulations. The comparison of these analytical error bars to actual measurement errors can reveal systematic errors that are not being accounted for in the modeling of the data.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

Application and Evaluation of Analytic Gaming

We describe an "analytic gaming" framework and methodology, and introduce formal methods for evaluation of the analytic gaming process. This process involves conception, development, and playing of games that are informed by predictive models and driven by players. Evaluation of analytic gaming examines both the process of game development and the results of game play exercises.

97 MATHEMATICS AND COMPUTING↗

MAPredict: Static Analysis Driven Memory Access Prediction Framework for Modern CPUs

Application memory access patterns are crucial in deciding how much traffic is served by the cache and forwarded to the dynamic random-access memory (DRAM). However, predicting such memory traffic is difficult because of the interplay of prefetchers, compilers, parallel execution, and innovations in manufacturer-specific micro-architectures. This research introduced MAPredict, a static analysis-driven framework that addresses these challenges to predict last-level cache (LLC)-DRAM traffic. By exploring and analyzing the behavior of modern Intel processors, MAPredict formulates cache-aware analytical models. MAPredict invokes these models to predict LLC-DRAM traffic by combining the application model, machine model, and user-provided hints to capture dynamic information. MAPredict successfully predicts LLC-DRAM traffic for different regular access patterns and provides the means to combine static and empirical observations for irregular access patterns. Evaluating 130 workloads from six applications on recent Intel micro-architectures, MAPredict yielded an average accuracy of 99% for streaming, 91% for strided, and 92% for stencil patterns. By coupling static and empirical methods, up to 97% average accuracy was obtained for random access patterns on different micro-architectures.

Monil, M. A. H.↗

Prediction of hemiwicking dynamics in micropillar arrays

Dynamic hemiwicking behavior is observable in both nature and a wide range of industrial applications ranging from biomedical devices to thermal management. We present a semi-analytical modeling framework (without empirical fitting coefficients) to predict transient capillary-driven hemiwicking behavior of a liquid through a nano/microstructured surface, specifically a micropillar array. In our model framework, the liquid domain is discretized into micropillar unit cells to enable the time marching of the hemiwicking front. A simplified linear pressure drop is assumed along the hemiwicking length such that the local meniscus curvature, contact angle, and effective liquid height are determined at each time step in our transient model. This semi-analytical model is validated with experimental data from our own experiments and from published literature for different fluids. Our model predicts hemiwicking dynamics with <20% error over a broad range of micropillar geometries with height-to-pitch ratio ranging between ≈0.34 and 6.7 and diameter-to-pitch ratio in the range of ≈0.25–0.7 and without any fitting parameters. For lower diameter-to-pitch ratio data points related to sparse micropillar array arrangements, we suggest modifications to the semi-analytical model. This work sheds light on complex and dynamic solid–liquid–vapor interfacial interactions which could serve as a guide for the design of textured surfaces for wicking enhancement in multi-phase thermal and mass transport technologies and applications.

Mechanics↗

Machine Learning-enabled Scalable Performance Prediction of Scientific Codes

Hardware architectures become increasingly complex as the compute capabilities grow to exascale. Here, we present the Analytical Memory Model with Pipelines (AMMP) of the Performance Prediction Toolkit (PPT). PPT-AMMP takes high-level source code and hardware architecture parameters as input and predicts runtime of that code on the target hardware platform, which is defined in the input parameters. PPT-AMMP transforms the code to an (architecture-independent) intermediate representation, then (i) analyzes the basic block structure of the code, (ii) processes architecture-independent virtual memory access patterns that it uses to build memory reuse distance distribution models for each basic block, and (iii) runs detailed basic-block level simulations to determine hardware pipeline usage. PPT-AMMP uses machine learning and regression techniques to build the prediction models based on small instances of the input code, then integrates into a higher-order discrete-event simulation model of PPT running on Simian PDES engine. We validate PPT-AMMP on four standard computational physics benchmarks and present a use case of hardware parameter sensitivity analysis to identify bottleneck hardware resources on different code inputs. We further extend PPT-AMMP to predict the performance of a scientific application code, namely, the radiation transport mini-app SNAP. To this end, we analyze multi-variate regression models that accurately predict the reuse profiles and the basic block counts. We validate predicted SNAP runtimes against actual measured times.

97 MATHEMATICS AND COMPUTING↗

Review—Effects of Solution and Alloy Composition on Critical Crevice Temperature

Critical temperature for localized corrosion can be a good design parameter because localized corrosion is not likely to occur below that temperature. The critical temperature depends on alloy composition, microstructure, and environment chemistry (including its redox potential). This paper reviews the literature on critical temperature for localized corrosion, expressed either as Critical Pitting Temperature (CPT) or Critical Crevice Temperature (CCT). A history of various testing methods is presented. Different approaches for modeling the temperature of transition to active pit growth are reviewed, including probabilistic aspects of critical temperature. A semi-empirical, electrolyte-based, model is described that can be useful in predicting CCT in service environments that differ from standard laboratory test environments. The model predictions are compared to experimental data for various alloys. The effect of solvent on CCT/CPT is described briefly and future avenues of research are recommended.

02 PETROLEUM↗

A Feature-Based Stochastic Permeability of Shale: Part 1—Validation and Two-Phase Permeability in a Utica Shale Sample

Estimate of permeability plays a crucial role in flow-based studies of fractured tight-rocks. It is well known that most of the flow through tight-rocks (e.g., shales) is controlled by permeable features (e.g., fractures, laminations, etc.), and there is negligible flow through the matrix. However, current approaches in the literature to model permeability of tight-rocks do not account for such features present within the rock ranging from micro-scale to field-scale. Current permeability modeling approach assumes a single continuum without considering the presence of permeable features within the matrix (e.g., micro-fractures) or outside the matrix (e.g., natural fractures). Although the laboratory-measured permeability implicitly captures discrete features present in that sample (e.g., fractures, laminations, micro-fractures), most of the permeability models proposed for shale do not account for these features. Fracture permeability in the literature is typically modeled using an ideal slit assumption; however, this highly overestimates its permeability because fractures in real medium are non-ideal in terms of their porosity and tortuosity, which affect their permeability. Additionally, the transition zone between fracture and matrix also affects the permeability of fracture. Here in this study, part of a two-part series, a new method to predict permeability of fractured shale by discretizing the medium into matrix (inorganic and organic) and fractures is presented. New analytical expressions of permeability are derived to account for non-ideal nature of porous medium and two-phase flow in fractures. Rock feature in each cell of the grid is identified as one of the three elements (organic matter, inorganic matter, or fracture), and permeability of that cell is estimated using a suitable analytical expression. This method allows estimating permeability at any scale of interest and more robustly than by a pure analytical approach. The proposed method is validated against local and global-scale measurements on three fractured samples from laboratory. Finally, the method is used to predict two-phase flow permeability of supercritical CO 2 displacing water within a fracture in a Utica shale sample. The proposed two-phase flow permeability equations can be used as a quick analytical tool to predict relative permeability estimates of two-phase flow in fractured shale samples. In Part 2, the proposed method is used to estimate field-scale permeability through an optimization process that uses field-scale production and other readily available information.

58 GEOSCIENCES↗

Quantifying the drivers and predictability of seasonal changes in African fire

Africa contains some of the most vulnerable ecosystems to fires. Successful seasonal prediction of fire activity over these fire-prone regions remains a challenge and relies heavily on in-depth understanding of various driving mechanisms underlying fire evolution. Here, we assess the seasonal environmental drivers and predictability of African fire using the analytical framework of Stepwise Generalized Equilibrium Feedback Assessment (SGEFA) and machine learning techniques (MLTs). The impacts of sea-surface temperature, soil moisture, and leaf area index are quantified and found to dominate the fire seasonal variability by regulating regional burning condition and fuel supply. Compared with previously-identified atmospheric and socioeconomic predictors, these slowly evolving oceanic and terrestrial predictors are further identified to determine the seasonal predictability of fire activity in Africa. Our combined SGEFA-MLT approach achieves skillful prediction of African fire one month in advance and can be generalized to provide seasonal estimates of regional and global fire risk.

54 ENVIRONMENTAL SCIENCES↗

The quenching of galaxies, bulges, and disks since cosmic noon

Here, we present an analysis of the quenching of star formation in galaxies, bulges, and disks throughout the bulk of cosmic history, from z = 2 – 0. We utilise observations from the Sloan Digital Sky Survey and the Mapping Nearby Galaxies at Apache Point Observatory survey at low redshifts. We complement these data with observations from the Cosmic Assembly Near-Infrared Deep Extragalactic Legacy Survey at high redshifts. Additionally, we compare the observations to detailed predictions from the LGalaxies semi-analytic model. To analyse the data, we developed a machine learning approach utilising a Random Forest classifier. We first demonstrate that this technique is extremely effective at extracting causal insight from highly complex and inter-correlated model data, before applying it to various observational surveys. Our primary observational results are as follows: at all redshifts studied in this work, we find bulge mass to be the most predictive parameter of quenching, out of the photometric parameter set (incorporating bulge mass, disk mass, total stellar mass, and B/T structure). Moreover, we also find bulge mass to be the most predictive parameter of quenching in both bulge and disk structures, treated separately. Hence, intrinsic galaxy quenching must be due to a stable mechanism operating over cosmic time, and the same quenching mechanism must be effective in both bulge and disk regions. Despite the success of bulge mass in predicting quenching, we find that central velocity dispersion is even more predictive (when available in spectroscopic data sets). In comparison to the LGalaxies model, we find that all of these observational results may be consistently explained through quenching via preventative ‘radio-mode’ active galactic nucleus feedback. Furthermore, many alternative quenching mechanisms (including virial shocks, supernova feedback, and morphological stabilisation) are found to be inconsistent with our observational results and those from the literature.

79 ASTRONOMY AND ASTROPHYSICS↗

Uncoupling Electrokinetic Flow Solutions

The continuum-scale electrokinetic porous-media flow and excess charge redistribution equations are uncoupled using eigenvalue decomposition. The uncoupling results in a pair of independent diffusion equations for “intermediate” potentials subject to modified material properties and boundary conditions. The fluid pressure and electrostatic potential are then found by recombining the solutions to the two intermediate uncoupled problems in a matrix-vector multiplication. Expressions for the material properties or source terms in the intermediate uncoupled problem may require extended precision or careful rewriting to avoid numerical cancellation, but the solutions themselves can typically be computed in double precision. The approach works with analytical or gridded numerical solutions and is illustrated through two examples. The solution for flow to a pumping well is manipulated to predict streaming potential and electroosmosis, and a periodic one-dimensional analytical solution is derived and used to predict electroosmosis and streaming potential in a laboratory flow cell subjected to low frequency alternating current and pressure excitation. The examples illustrate the utility of the eigenvalue decoupling approach, repurposing existing analytical solutions or numerical models and leveraging solutions that are simpler to derive for coupled physics.

58 GEOSCIENCES↗

Spatially distributed flame transfer functions for predicting combustion dynamics in lean premixed gas turbine combustors

The present paper describes a methodology to improve the accuracy of prediction of the eigenfrequencies and growth rates of self-induced instabilities and demonstrates its application to a laboratory-scale, swirl-stabilized, lean-premixed, gas turbine combustor. The influence of the spatial heat release distribution is accounted for using local flame transfer function (FTF) measurements. The two-microphone technique and CH{sup *} chemiluminescence intensity measurements are used to determine the input (inlet velocity perturbation) and the output functions (heat release oscillation), respectively, for the local flame transfer functions. The experimentally determined local flame transfer functions are superposed using the flame transfer function superposition principle, and the result is incorporated into an analytic thermoacoustic model, in order to predict the linear stability characteristics of a given system. Results show that when the flame length is not acoustically compact the model prediction calculated using the local flame transfer functions is better than the prediction made using the global flame transfer function. In the case of a flame in the compact flame regime, accurate predictions of eigenfrequencies and growth rates can be obtained using the global flame transfer function. It was also found that the general response characteristics of the local FTF (gain and phase) are qualitatively the same as those of the global FTF. (author)

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Incorporating historical data and past analyses for improved tensile property prediction of 9% Cr steel

Data-driven analytical clustering and visualization techniques were applied to the dataset of 9% Cr experimental alloy data generated through the eXtremeMAT project. Techniques and results were compared with the resulting clusters obtained through similar analytical techniques on previous and reduced versions of the dataset. The principal components were generated in order to reduce the dimensionality of the complex dataset and to visualize the underlying trends in the data. Partitioning around medoids was performed on the resulting principal components to determine relevant clusters. Domain knowledge labels were further applied to the principal components to compare the labels with the trends identified through the clustering methods. The resulting clusters show the addition of the new data into the dataset, and show where alignment between the domain knowledge labels and partitioning around medoids (PAM) clusters show where the labels are appropriate. The clusters can be used to compare the tensile properties of the alloys, and to reduce the variation in the dataset.

Alloy design, Mechanical Properties, data analysis↗

Relaxation drag history of shock accelerated microparticles

Experimental measurements of the displacements of shock accelerated microparticles from shortly after shock interaction to the particle relaxation time show time-dependent drag coefficients ( $C_{D}$ ) that are much higher than those predicted by quasi-steady and unsteady drag models. Nylon particles with mean diameter of $4~\unicode[STIX]{x03BC}\text{m}$ , accelerated by one-dimensional normal shocks (Mach number $M_{s}=1.2$ , 1.3 and 1.4), have measured $C_{D}$ values that follow a power-law behaviour. The drag is a function of the time-dependent Knudsen number, $Kn^{\ast }=M_{s}/Re_{p}$ , where the particle Reynolds number ( $Re_{p}$ ) is calculated using the time-dependent slip velocity. Also, some portion of the drag can be attributed to quasi-steady forces, but the total drag cannot be predicted by current unsteady force models that are based on the Basset–Boussinesq–Oseen equation and pressure drag. The largest contribution to the total drag is the unsteady component ( $C_{D,us}$ ) until the particle attains $Kn^{\ast }\approx 0.5{-}1.0$ , then the unsteady contribution decays. The quasi-steady component ( $C_{D,qs}$ ) increases almost linearly with $Kn^{\ast }$ , intersects the $C_{D,us}$ at $Kn^{\ast }\approx 2$ and becomes the primary contributor to the drag towards the end of the relaxation zone as $Re_{p}\rightarrow 0$ . Finally, there are currently no analytical models that are able to predict the nonlinear behaviour of the shock accelerated particles during the relaxation phase of the flow.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗