SEARCH · Engineering Papers

Results for “Yield Prediction”

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.

SourceResearch connectionFrom yearThrough year

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

At least 1,171 records · Page 65

All sources NASA NTRS DOE OSTI

NASA NTRS2005

Analyses Of Transient Events In Complex Valve and Feed Systems

Valve systems in rocket propulsion systems and testing facilities are constantly subject to dynamic events resulting from the timing of valve motion leading to unsteady fluctuations in pressure and mass flow. Such events can also be accompanied by cavitation, resonance, system vibration leading to catastrophic failure. High-fidelity dynamic computational simulations of valve operation can yield important information of valve response to varying flow conditions. Prediction of transient behavior related to valve motion can serve as guidelines for valve scheduling, which is of crucial importance in engine operation and testing. In this paper, we present simulations of the diverse unsteady phenomena related to valve and feed systems that include valve stall, valve timing studies as well as cavitation instabilities in components utilized in the test loop.

Ahuja, Vineet↗

NASA NTRS2006

Simulations of Transient Phenomena in Liquid Rocket Feed Systems

Valve systems in rocket propulsion systems and testing facilities are constantly subject to dynamic events resulting from the timing of valve motion leading to unsteady fluctuations in pressure and mass flow. Such events can also be accompanied by cavitation, resonance, system vibration leading to catastrophic failure. High-fidelity dynamic computational simulations of valve operation can yield important information of valve response to varying flow conditions. Prediction of transient behavior related to valve motion can serve as guidelines for valve scheduling, which is of crucial importance in engine operation and testing. Feed components operating in cryogenic regimes can also experience cavitation based instabilities leading to large scale shedding of vapor clouds and pressure oscillations. In this paper, we present simulations of the diverse unsteady phenomena related to valve and feed systems that include valve stall, valve timing studies as well as two different forms of cavitation instabilities in components utilized in the test loop.

Ahuja, V.↗

NASA NTRS2008

A Globally Optimal Particle Tracking Technique for Stereo Imaging Velocimetry Experiments

An important phase of any Stereo Imaging Velocimetry experiment is particle tracking. Particle tracking seeks to identify and characterize the motion of individual particles entrained in a fluid or air experiment. We analyze a cylindrical chamber filled with water and seeded with density-matched particles. In every four-frame sequence, we identify a particle track by assigning a unique track label for each camera image. The conventional approach to particle tracking is to use an exhaustive tree-search method utilizing greedy algorithms to reduce search times. However, these types of algorithms are not optimal due to a cascade effect of incorrect decisions upon adjacent tracks. We examine the use of a guided evolutionary neural net with simulated annealing to arrive at a globally optimal assignment of tracks. The net is guided both by the minimization of the search space through the use of prior limiting assumptions about valid tracks and by a strategy which seeks to avoid high-energy intermediate states which can trap the net in a local minimum. A stochastic search algorithm is used in place of back-propagation of error to further reduce the chance of being trapped in an energy well. Global optimization is achieved by minimizing an objective function, which includes both track smoothness and particle-image utilization parameters. In this paper we describe our model and present our experimental results. We compare our results with a nonoptimizing, predictive tracker and obtain an average increase in valid track yield of 27 percent

McDowell, Mark↗

NASA NTRS2009

Measurement and Characterization of Space Shuttle Solid Rocket Motor Plume Acoustics

Lift-off acoustic environments generated by the future Ares I launch vehicle are assessed by the NASA Marshall Space Flight Center (MSFC) acoustics team using several prediction tools. This acoustic environment is directly caused by the Ares I First Stage booster, powered by the five-segment Reusable Solid Rocket Motor (RSRMV). The RSRMV is a larger-thrust derivative design from the currently used Space Shuttle solid rocket motor, the Reusable Solid Rocket Motor (RSRM). Lift-off acoustics is an integral part of the composite launch vibration environment affecting the Ares launch vehicle and must be assessed to help generate hardware qualification levels and ensure structural integrity of the vehicle during launch and lift-off. Available prediction tools that use free field noise source spectrums as a starting point for generation of lift-off acoustic environments are described in the monograph NASA SP-8072: "Acoustic Loads Generated by the Propulsion System." This monograph uses a reference database for free field noise source spectrums which consist of subscale rocket motor firings, oriented in horizontal static configurations. The phrase "subscale" is appropriate, since the thrust levels of rockets in the reference database are orders of magnitude lower than the current design thrust for the Ares launch family. Thus, extrapolation is needed to extend the various reference curves to match Ares-scale acoustic levels. This extrapolation process yields a subsequent amount of uncertainty added upon the acoustic environment predictions. As the Ares launch vehicle design schedule progresses, it is important to take every opportunity to lower prediction uncertainty and subsequently increase prediction accuracy. Never before in NASA s history has plume acoustics been measured for large scale solid rocket motors. Approximately twice a year, the RSRM prime vendor, ATK Launch Systems, static fires an assembled RSRM motor in a horizontal configuration at their test facility in Utah. The remaining RSRM static firings will take place on elevated terrain, with the nozzle exit plume being mostly undeflected and the landscape allowing placement of microphones within direct line of sight to the exhaust plume. These measurements will help assess the current extrapolation process by direct comparison between subscale and full scale solid rocket motor data.

Kenny, Jeremy↗

NASA NTRS2020

Global Estimates and Long-Term Trends of Fine Particulate Matter Concentrations (1998-2018)

Exposure to outdoor fine particulate matter (PM2.5) is a leading risk factor for mortality. We develop global estimates of annual PM2.5 concentrations and trends for 1998–2018 using advances in satellite observations, chemical transport modeling, and ground-based monitoring. Aerosol optical depths (AODs) from advanced satellite products including finer resolution, increased global coverage, and improved long-term stability are combined and related to surface PM2.5 concentrations using geophysical relationships between surface PM2.5 and AOD simulated by the GEOS-Chem chemical transport model with updated algorithms. The resultant annual mean geophysical PM2.5 estimates are highly consistent with globally distributed ground monitors (R2 = 0.81; slope = 0.90). Geographically weighted regression is applied to the geophysical PM2.5 estimates to predict and account for the residual bias with PM2.5 monitors, yielding even higher cross validated agreement (R2 = 0.90–0.92; slope = 0.90–0.97) with ground monitors and improved agreement compared to all earlier global estimates. The consistent long-term satellite AOD and simulation enable trend assessment over a 21 year period, identifying significant trends for eastern North America (−0.28 ± 0.03 μg/m3/yr), Europe (−0.15 ± 0.03 μg/m3/yr), India (1.13 ± 0.15 μg/m3/yr), and globally (0.04 ± 0.02 μg/m3/yr). The positive trend (2.44 ± 0.44 μg/m3/yr) for India over 2005–2013 and the negative trend (−3.37 ± 0.38 μg/m3/yr) for China over 2011–2018 are remarkable, with implications for the health of billions of people.

Melanie S. Hammer↗

NASA NTRS2022

Regional Variations of Mercury's Crustal Density and Porosity From MESSENGER Gravity Data

A new solution of Mercury's gravity field to degree and order 160, named HgM009, is retrieved through a reprocessing of MESSENGER radio science measurements. By combining our latest gravity field with topography data, localized spectral admittance analyses are carried out to investigate Mercury's crustal and lithospheric properties across the northern hemisphere. The measured spectra are compared with admittances predicted by lithospheric flexure models. The localized gravity/topography admittance analyses yield key information on the lateral variations of the bulk density of the upper crust. Elastic and crustal thicknesses are also adjusted in our study, but the local admittance spectra allow us to constrain these parameters only over a few regions. The average bulk density across the observed areas in the northern hemisphere is 2540 ±60 kg m −3 . The crustal porosity is then constrained by using an estimate of the pore-free grain density of surface materials with our measured bulk density. Our estimate of the mean porosity is 14.7 ±1.6 %, which is comparable to, but slightly higher than, the average value measured on the Moon. Larger crustal porosities are observed over heavily cratered regions, suggesting that impact bombardment is the main cause of the crustal porosity.

Mercury's crust↗

NASA NTRS1999

Sonic Boom Prediction and Minimization of the Douglas Reference OPT5 Configuration

Conventional CFD methods and grids do not yield adequate resolution of the complex shock flow pattern generated by a real aircraft geometry. As a result, a unique grid topology and supersonic flow solver was developed at Northrop Grumman based on the characteristic behavior of supersonic wave patterns emanating from the aircraft. Using this approach, it was possible to compute flow fields with adequate resolution several body lengths below the aircraft. In this region, three-dimensional effects are diminished and conventional two-dimensional modified linear theory (MLT) can be applied to estimate ground pressure signatures or sonic booms. To accommodate real aircraft geometries and alleviate the burdensome grid generation task, an implicit marching multi-block, multi-grid finite-volume Euler code was developed as the basis for the sonic boom prediction methodology. The Thomas two-dimensional extrapolation method is built into the Euler code so that ground signatures can be obtained quickly and efficiently with minimum computational effort suitable to the aircraft design environment. The loudness levels of these signatures can then be determined using a NASA generated noise code. Since the Euler code is a three-dimensional flow field solver, the complete circumferential region below the aircraft is computed. The extrapolation of all this field data from a cylinder of constant radius leads to the definition of the entire boom corridor occurring directly below and off to the side of the aircraft's flight path yielding an estimate for the entire noise "annoyance" corridor in miles as well as its magnitude. An automated multidisciplinary sonic boom design optimization software system was developed during the latter part of HSR Phase 1. Using this system, it was found that sonic boom signatures could be reduced through optimization of a variety of geometric aircraft parameters. This system uses a gradient based nonlinear optimizer as the driver in conjunction with a computationally efficient Euler CFD solver (NIIM3DSB) for computing the three-dimensional near-field characteristics of the aircraft. The intent of the design system is to identify and optimize geometric design variables that have a beneficial impact on the ground sonic boom. The system uses a simple wave drag data format to specify the aircraft geometry. The geometry is internally enhanced and analytic methods are used to generate marching grids suitable for the multi-block Euler solver. The Thomas extrapolation method is integrated into this system, and hence, the aircraft's centerline ground sonic boom signature is also automatically computed for a specified cruise altitude and yields the parameters necessary to evaluate the design function. The entire design system has been automated since the gradient based optimization software requires many flow analyses in order to obtain the required sensitivity derivatives for each design variable in order to converge on an optimal solution. Hence, once the problem is defined which includes defining the objective function and geometric and aerodynamic constraints, the system will automatically regenerate the perturbed geometry, the necessary grids, the Euler solution, and finally the ground sonic boom signature at the request of the optimizer.

Siclari, Michael J.↗

DOE OSTI2024

Applicability of 1D site response analysis to shallow sedimentary basins: A critical evaluation through physics-based 3D ground motion simulations

One-dimensional site response analysis (1D SRA) remains the standard practice in considering the effect of local soil deposits and predicting site-specific ground motions, although its range of applicability to realistic seismic wavefields is still in question. In this 1D approach, horizontal and vertical ground shaking are assumed to be induced by vertically propagating shear and compressional waves, respectively. A recent study based on analytical two-dimensional (2D) plane waves and simple point source earthquake simulations has shown two mechanistic limitations in this 1D modelling technique for general inclined seismic waves, that is, systematic over-prediction of the vertical motion and wave trapping in the 1D soil column. In this article, we evaluate in detail the applicability of this 1D modelling approach to realistic three-dimensional (3D) simulated seismic wavefields in shallow sedimentary basins. Linear-viscoelastic 1D SRA predictions using two types of input motions that are commonly used in practice—rock outcrop and in-column motions, are compared with the reference true site response results from 3D earthquake simulations in terms of various measures in the frequency and time domain. It is shown that the horizontal motion in the 3D seismic wavefield exhibits dominant shear wave propagation phenomenon, while the vertical motion is a combined effect of compressional and shear waves and can be over-predicted by the 1D approach when the incident seismic waves are inclined. Direct evidence of the wave refraction process that leads to the vertical motion over-prediction is provided. 1D SRA with in-column inputs can yield motions that have significantly longer duration compared to the true 3D site response solution due to trapped waves, casting in doubt the frequent need for increased soil damping in existing site studies to compensate for wave attenuation due to scattering alone. Sensitivity investigation on the increase of soil profile damping by a multiplier D mul shows D mul values compatible with those found in the literature for both horizontal and vertical motions. It is shown that the level of D mul optimized for a best match of the spectral acceleration is dependent on the characteristic of the input motion and a larger D mul is typically required for the vertical component. In contrast, 1D SRA with outcrop motions predicts motions with shorter significant duration due to its inability to capture the basin-edge generated surface waves. A suite of ground motion simulations was performed to assess the sensitivity of the observations to the basin geologic structure including the velocity gradient, rock-basin impedance contrast and basin depth. The analysis results show that the accuracy of the simplified 1D procedure is dependent on the wavefield composition of both the input motions and the true 3D site response solution. While the horizontal motions in shallow sedimentary basins can, to the first order, be reasonably captured by the simplified 1D approach, 1D SRA for the vertical component is in general not reliable and contributions from inclined shear waves should be accounted for in site-specific evaluation of the vertical design ground motion.

1D assumption↗

DOE OSTI2026

Evidence of Z Z γ Production and Observation of 4 ℓ γ in Proton-Proton Collisions at s = 13 TeV

Evidence of the production of two Z bosons and a photon in proton-proton collisions is reported for the first time in CMS. The analysis uses data collected by the CMS experiment between 2016 and 2018 at a center-of-mass energy of 13 TeV, corresponding to an integrated luminosity of 138 fb - 1 . Evidence for the process p p → Z Z γ → 4 ℓ γ ( ℓ = e , μ ), with an observed (expected) significance of 3.7 (3.1) standard deviations in a fiducial region defined by p T γ > 20 GeV , | η γ | < 2.4 , Δ R ( ℓ , γ ) > 0.5 , m Z between 60 and 120 GeV, and the invariant mass of either of the two Z bosons combined with the photon ( m Z γ ) larger than 100 GeV, is reported. The measured (predicted) fiducial cross section is 60 - 22 + 27 ab ( 47.56 ± 0.04 ab ). Additionally, the inclusive production of p p → 4 ℓ γ is studied by removing the m Z γ requirement to include final state radiation where one Z boson decays to 2 ℓ γ , yielding an observed (expected) significance of 5.0 (4.2) standard deviations and a measured (predicted) fiducial cross section of 156 - 35 + 39 ab ( 99.97 ± 0.09 ab ).

Hayrapetyan, Aram [Yerevan Phys. Inst.]↗

NASA NTRS2020

Adaptation in U.S. Corn Belt Increases Resistance to Soil Carbon Loss With Climate Change

Increasing the amount of soil organic carbon (SOC) has agronomic benefits and the potential to mitigate climate change. Previous regional predictions of SOC trends under climate change often ignore or do not explicitly consider the effect of crop adaptation (i.e., changing planting dates and varieties). We used the DayCent biogeochemical model to examine the effect of adaptation on SOC for corn and soybean production in the U.S. Corn Belt using climate data from three models. Without adaptation, yields of both corn and soybean tended to decrease and the decomposition of SOC tended to increase leading to a loss of SOC with climate change compared to a baseline scenario with no climate change. With adaptation, the model predicted a substantially higher crop yield. The increase in yields and associated carbon input to the SOC pool counteracted the increased decomposition in the adaptation scenarios, leading to similar SOC stocks under different climate change scenarios. Consequently, we found that crop management adaptation to changing climatic conditions strengthen agroecosystem resistance to SOC loss. However, there are differences spatially in SOC trends. The northern part of the region is likely to gain SOC while the southern part of the region is predicted to lose SOC.

Soil Organic Carbon (SOC)↗

NASA NTRS2020

A Systematic Review of Local to Regional Yield Forecasting Approaches and Frequently Used Data Resources

Forecasting crop yields, or providing an expectation of ex-ante harvest amounts, is highly relevant to the whole agricultural production chain. Farmers can adapt their management, traders or insurers their pricing schemes, suppliers their stocks, logistic companies their routes, national authorities their food balance sheets to guide import or export and, finally, international aid organizations can mobilize reliefs. Evidence has grown in the literature that such forecasts with a meaningful lead time are possible on various geographic scales and for a broad range of crops. Here, we present a systematic review of the methods applied in end-of-season yield forecasting and three frequently used data sources: weather data, satellite data and crop masks. Our literature database comprises 362 studies (2004–2019) which were evaluated regarding methods, crops, regions, data sources, lead time and performance. Moreover, we present 24 sources of real-time and predictive weather data, 21 sources of remote sensing data and 16 crop masks. Yield forecasting in our literature sample has been performed for 44 crops in 71 countries, also including many non-staple crops, but with an apparent bias in regions and crops. Forecasting performance depends on various factors, including crop, region, method, lead time to harvest and input diversity. Our systematic review supports a broader application of locally successful approaches at larger scales by providing a comprehensive, accessible compendium of necessary information for yield forecasting. We discuss improvement potentials with respect to methodological approaches and available data sources. We additionally suggest standardization procedures for future forecasting studies and encourage studying additional crops and geographic regions. Implications of forecasts for different target groups on different scales and the adaptation towards climate change are also discussed.

Seasonal crop yield forecasting↗

NASA NTRS2012

Design of a Laboratory Hall Thruster with Magnetically Shielded Channel Walls, Phase III: Comparison of Theory with Experiment

A proof-of-principle effort to demonstrate a technique by which erosion of the acceleration channel in Hall thrusters of the magnetic-layer type can be eliminated has been completed. The first principles of the technique, now known as "magnetic shielding," were derived based on the findings of numerical simulations in 2-D axisymmetric geometry. The simulations, in turn, guided the modification of an existing 6-kW laboratory Hall thruster. This magnetically shielded (MS) thruster was then built and tested. Because neither theory nor experiment alone can validate fully the first principles of the technique, the objective of the 2-yr effort was twofold: (1) to demonstrate in the laboratory that the erosion rates can be reduced by >order of magnitude, and (2) to demonstrate that the near-wall plasma properties can be altered according to the theoretical predictions. This paper concludes the demonstration of magnetic shielding by reporting on a wide range of comparisons between results from numerical simulations and laboratory diagnostics. Collectively, we find that the comparisons validate the theory. Near the walls of the MS thruster, theory and experiment agree: (1) the plasma potential has been sustained at values near the discharge voltage, and (2) the electron temperature has been lowered by at least 2.5-3 times compared to the unshielded (US) thruster. Also, based on carbon deposition measurements, the erosion rates at the inner and outer walls of the MS thruster are found to be lower by at least 2300 and 1875 times, respectively. Erosion was so low along these walls that the rates were below the resolution of the profilometer. Using a sputtering yield model with an energy threshold of 25 V, the simulations predict a reduction of ~600 at the MS inner wall. At the outer wall ion energies are computed to be below 25 V, for which case we set the erosion to zero in the simulations. When a 50-V threshold is used the computed ion energies are below the threshold at both sides of the channel. Uncertainties, sensitivities and differences between theory and experiment are also discussed.

plasma simulations↗

DOE OSTI2024

Predicting Partial Atomic Charges in Metal–Organic Frameworks: An Extension to Ionic MOFs

Molecular simulation is an invaluable tool to predict and understand the usage of metal–organic frameworks (MOFs) for gas storage and separation applications. Accurate partial atomic charges, commonly obtained from density functional theory (DFT) calculations, are often required to model the electrostatic interactions between the MOF and adsorbates, especially when the adsorbates have dipole or quadrupole moments, such as water and CO 2 . Machine learning (ML) models have been previously employed to predict partial charges and avoid the computational cost associated with DFT calculations. However, previous ML models suffer from small training data sets, which limit their scope of application. In this work, we introduce two novel machine learning models, PACMOF2-neutral and PACMOF2-ionic, aimed at predicting the density-derived electrostatic and chemical (DDEC6) partial atomic charges for both neutral and ionic MOFs. These models not only yield DFT-level accuracy at a fraction of the computational cost but also demonstrate a remarkable improvement in prediction of adsorption, as validated with grand canonical Monte Carlo simulations. Furthermore, the robustness and fast computational time of the PACMOF2 models, along with their transferability to other porous materials such as covalent organic frameworks and zeolites, underscores their potential in high-throughput screening of MOFs for diverse applications.

36 MATERIALS SCIENCE↗

NASA NTRS1975

Light aircraft lift, drag, and moment prediction: A review and analysis

The historical development of analytical methods for predicting the lift, drag, and pitching moment of complete light aircraft configurations in cruising flight is reviewed. Theoretical methods, based in part on techniques described in the literature and in part on original work, are developed. These methods form the basis for understanding the computer programs given to: (1) compute the lift, drag, and moment of conventional airfoils, (2) extend these two-dimensional characteristics to three dimensions for moderate-to-high aspect ratio unswept wings, (3) plot complete configurations, (4) convert the fuselage geometric data to the correct input format, (5) compute the fuselage lift and drag, (6) compute the lift and moment of symmetrical airfoils to M = 1.0 by a simplified semi-empirical procedure, and (7) compute, in closed form, the pressure distribution over a prolate spheroid at alpha = 0. Comparisons of the predictions with experiment indicate excellent lift and drag agreement for conventional airfoils and wings. Limited comparisons of body-alone drag characteristics yield reasonable agreement. Also included are discussions for interference effects and techniques for summing the results above to obtain predictions for complete configurations.

Smetana, F. O.↗

DOE OSTI2021

Future changes in isoprene-epoxydiol-derived secondary organic aerosol (IEPOX SOA) under the Shared Socioeconomic Pathways: the importance of physicochemical dependency

Abstract. Secondary organic aerosol (SOA) is a dominant contributor of fine particulate matter in the atmosphere, but the complexity of SOA formation chemistry hinders the accurate representation of SOA in models. Volatility-based SOA parameterizations have been adopted in many recent chemistry modeling studies and have shown a reasonable performance compared to observations. However, assumptions made in these empirical parameterizations can lead to substantial errors when applied to future climatic conditions as they do not include the mechanistic understanding of processes but are rather fitted to laboratory studies of SOA formation. This is particularly the case for SOA derived from isoprene epoxydiols (IEPOX SOA), for which we have a higher level of understanding of the fundamental processes than is currently parameterized in most models. We predict future SOA concentrations using an explicit mechanism and compare the predictions with the empirical parameterization based on the volatility basis set (VBS) approach. We then use the Community Earth System Model 2 (CESM2.1.0) with detailed isoprene chemistry and reactive uptake processes for the middle and end of the 21st century under four Shared Socioeconomic Pathways (SSPs): SSP1–2.6, SSP2–4.5, SSP3–7.0, and SSP5–8.5. With the explicit chemical mechanism, we find that IEPOX SOA is predicted to increase on average under all future SSP scenarios but with some variability in the results depending on regions and the scenario chosen. Isoprene emissions are the main driver of IEPOX SOA changes in the future climate, but the IEPOX SOA yield from isoprene emissions also changes by up to 50 % depending on the SSP scenario, in particular due to different sulfur emissions. We conduct sensitivity simulations with and without CO2 inhibition of isoprene emissions that is highly uncertain, which results in factor of 2 differences in the predicted IEPOX SOA global burden, especially for the high-CO2 scenarios (SSP3–7.0 and SSP5–8.5). Aerosol pH also plays a critical role in the IEPOX SOA formation rate, requiring accurate calculation of aerosol pH in chemistry models. On the other hand, isoprene SOA calculated with the VBS scheme predicts a nearly constant SOA yield from isoprene emissions across all SSP scenarios; as a result, it mostly follows isoprene emissions regardless of region and scenario. This is because the VBS scheme does not consider heterogeneous chemistry; in other words, there is no dependency on aerosol properties. The discrepancy between the explicit mechanism and VBS parameterization in this study is likely to occur for other SOA components as well, which may also have dependencies that cannot be captured by VBS parameterizations. This study highlights the need for more explicit chemistry or for parameterizations that capture the dependence on key physicochemical drivers when predicting SOA concentrations for climate studies.

54 ENVIRONMENTAL SCIENCES↗

DOE OSTI2021

The updated ITPA global H-mode confinement database: description and analysis

The multi-machine ITPA Global H-mode Confinement Database has been upgraded with new data from JET with the ITER-like wall and ASDEX Upgrade with the full tungsten wall. This paper describes the new database and presents results of regression analysis to estimate the global energy confinement scaling in H-mode plasmas using a standard power law. Various subsets of the database are considered, focusing on type of wall and divertor materials, confinement regime (all H-modes, ELMy H or ELM-free) and ITER-like constraints. Apart from ordinary least squares, two other, robust regression techniques are applied, which take into account uncertainty on all variables. Regression on data from individual devices shows that, generally, the confinement dependence on density and the power degradation are weakest in the fully metallic devices. Using the multi-machine scalings, predictions are made of the confinement time in a standard ELMy H-mode scenario in ITER. The uncertainty on the scaling parameters is discussed with a view to practically useful error bars on the parameters and predictions. One of the derived scalings for ELMy H-modes on an ITER-like subset is studied in particular and compared to the IPB98(y,2) confinement scaling in engineering and dimensionless form. Transformation of this new scaling from engineering variables to dimensionless quantities is shown to result in large error bars on the dimensionless scaling. Regression analysis in the space of dimensionless variables is therefore proposed as an alternative, yielding acceptable estimates for the dimensionless scaling. The new scaling, which is dimensionally correct within the uncertainties, suggests that some dependencies of confinement in the multi- machine database can be reconciled with parameter scans in individual devices. This includes vanishingly small dependence of confinement on line-averaged density and normalized plasma pressure (β), as well as a noticeable, positive dependence on effective atomic mass and plasma triangularity. Extrapolation of this scaling to ITER yields a somewhat lower confinement time compared to the IPB98(y, 2) prediction, possibly related to the considerably weaker dependence on major radius in the new scaling (slightly above linear). Further studies are needed to compare more flexible regression models with the power law used here. In addition, data from more devices concerning possible ‘hidden variables’ could help to determine their influence on confinement, while adding data in sparsely populated areas of the parameter space may contribute to further disentangling some of the global confinement dependencies in tokamak plasmas.

Database↗

DOE OSTI2023

Uncertainty quantification of a deep learning fuel property prediction model

Deep learning models are being widely used in the field of combustion. Given the black-box nature of typical neural network based models, uncertainty quantification (UQ) is critical to ensure the reliability of predictions as well as the training datasets, and for a principled quantification of noise and its various sources. Deep learning surrogate models for predicting properties of chemical compounds and mixtures have been recently shown to be promising for enabling data-driven fuel design and optimization, with the ultimate goal of improving efficiency and lowering emissions from combustion engines. In this study, UQ is performed for a multi-task deep learning model that simultaneously predicts the research octane number (RON), Motor Octane Number (MON), and Yield Sooting Index (YSI) of pure components and multicomponent blends. The deep learning model is comprised of three smaller networks: Extractor 1, Extractor 2, and Predictor, and a mixing operator. The molecular fingerprints of individual components are encoded via Extractor 1 and Extractor 2, the mixing operator generates fingerprints for mixtures/blends based on linear mixing operation, and the predictor maps the fingerprint to the target properties. Two different classes of UQ methods, Monte Carlo ensemble methods and Bayesian neural networks (BNNs), are employed for quantifying the epistemic uncertainty. Combinations of Bernoulli and Gaussian distributions with DropConnect and DropOut techniques are explored as ensemble methods. All the DropConnect, DropOut and Bayesian layers are applied to the predictor network. Aleatoric uncertainty is modeled by assuming that each data point has an independent uncertainty associated with it. The results of the UQ study are further analyzed to compare the performance of BNN and ensemble methods. Although this study is confined to UQ of fuel property prediction, the methodologies are applicable to other deep learning frameworks that are being widely used in the combustion community.

33 ADVANCED PROPULSION SYSTEMS↗

DOE OSTI2025

Probabilistic Predictions for Fastener Failure in the Sandia Mechanics Challenge Using the Discrete-Direct Uncertainty Quantification Approach

This paper documents the blind and post-blind analysis predictions for the 2023 Sandia Mechanics Challenge (SMC), which involved predicting the behavior of a threaded fastener joint structure subjected to shock loading. Utilizing repeat sets of fastener calibration data from various experimental configurations including tension, double shear, and joint tension, we developed a library of calibrated models which were propagated through the application model using the Discrete-Direct (DD) uncertainty quantification (UQ) approach. Although the initial blind predictions did not incorporate spare-sample processing to quantify fastener failure probabilities, the analyses yielded reasonable conclusions aligned with experimental results. In the post-blind analysis phase, we focused on enhancing the fidelity of the aluminum constitutive model and innovating the DD approach to obtain probabilistic predictions for fastener failure, particularly when quantities of interest (QoIs) approach their bounds. The improved aluminum model captures the behavior of the cantilever under shock loading more accurately, predicting both partial and complete cracks, although it tends to underpredict failure propagation. The enhanced DD approach facilitates probabilistic predictions that reflect the interdependent failure mechanisms of the fasteners and the cantilever, revealing that while certain fasteners are more likely to fail, the failure does not necessarily follow a progressive pattern. Overall, the post-blind analyses significantly improved the predictive capabilities of the model, providing valuable insights into the SMC application and establishing a robust foundation for informed engineering decisions. The methodology demonstrates a cost-effective and extensible approach suitable for a wide range of applications, highlighting the importance of uncertainty quantification to provide context for engineering decision making.

42 ENGINEERING↗