Search NASA⌕ Search

SEARCH · Search NASA

Results for “extrapolation”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

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

At least 1,765 records · Page 98

tLaSDI: Thermodynamics-informed latent space dynamics identification

Here we propose a latent space dynamics identification method, namely tLaSDI, that embeds the first and second principles of thermodynamics. The latent variables are learned through an autoencoder as a nonlinear dimension reduction model. The latent dynamics are constructed by a neural network-based model that precisely preserves certain structures for the thermodynamic laws through the GENERIC formalism. An abstract error estimate is established, which provides a new loss formulation involving the Jacobian computation of autoencoder. The autoencoder and the latent dynamics are simultaneously trained to minimize the new loss. Computational examples demonstrate the effectiveness of tLaSDI, which exhibits robust generalization ability, even in extrapolation. In addition, an intriguing correlation is empirically observed between a quantity from tLaSDI in the latent space and the behaviors of the full-state solution.

97 MATHEMATICS AND COMPUTING↗

Mesh-based super-resolution of fluid flows with multiscale graph neural networks

A graph neural network (GNN) approach is introduced in this work which enables mesh-based three-dimensional super-resolution of fluid flows. In this framework, the GNN is designed to operate not on the full mesh-based field at once, but on localized meshes of elements (or cells) directly. To facilitate mesh-based GNN representations in a manner similar to spectral (or finite) element discretizations, a baseline GNN layer (termed a message passing layer, which updates local node properties) is modified to account for synchronization of coincident graph nodes, rendering compatibility with commonly used element-based mesh connectivities. Furthermore, the architecture is multiscale in nature, and is comprised of a combination of coarse-scale and fine-scale message passing layer sequences (termed processors) separated by a graph unpooling layer. The coarse-scale processor embeds a query element (alongside a set number of neighboring coarse elements) into a single latent graph representation using coarse-scale synchronized message passing over the element neighborhood, and the fine-scale processor leverages additional message passing operations on this latent graph to correct for interpolation errors. Demonstration studies are performed using hexahedral mesh-based data from Taylor–Green Vortex and backward-facing step flow simulations at Reynolds numbers of 1600 and 3200. Through analysis of both global and local errors, the results ultimately show how the GNN is able to produce accurate super-resolved fields compared to targets in both coarse-scale and multiscale model configurations. Reconstruction errors for fixed architectures were found to increase in proportion to the Reynolds number. Geometry extrapolation studies on a separate cavity flow configuration show promising cross-mesh capabilities of the super-resolution strategy.

Backward-facing step↗

Studies of premixed and non-premixed hydrogen flames

The hydrogen oxidation chemistry constitutes the foundation of the kinetics of all carbon- and hydrogen-containing fuels. The validation of rate constants of hydrogen-related reactions can be complicated by uncertainties associated with experimental data caused by the high reactivity and diffusivity of hydrogen.Here, accurate experimental data on flame propagation and extinction were determined for premixed and non-premixed hydrogen flames at pressures between p = 1 and 7 atm. The experiments were designed to sensitize the three-body H + O 2 + M → HO 2 + M reaction, whose rate is subject to notable uncertainty. This was achieved by increasing the pressure and by adding to the reactants H 2 O and CO 2 whose collision efficiencies are high compared to other species. In the present study, directly measured flame properties were compared against computed ones, in order to eliminate uncertainties associated with extrapolations, as is the case for laminar flame speeds. The measured extinction strain rates exhibit both a positive and negative dependence on pressure with and without weighting with the density, and this non-monotonic behavior is caused by the competition between the H + O 2 → O + OH and H + O 2 + M → HO 2 + M reactions as well as HO 2 kinetic pathways as pressure increases. The various kinetic models considered in this investigation did not reproduce equally well the non-premixed flame extinction data with added H 2 O. Yet, the predicted extinction strain rates were consistent between the various models in the case of added CO 2 . Finally, it was shown that the formulation of binary diffusion coefficient pairs including H–N 2 and H 2 –N 2 has a first order effect on the prediction of extinction strain rates of non-premixed H 2 flames.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Quantitative acetylene measurements in laminar and turbulent flames using 1D Raman/Rayleigh scattering

Our research presents the extension of the established detection of seven major species in 1D Raman/Rayleigh measurements to incorporate acetylene (C 2 H 2 ) as an eighth species. Acetylene is an important soot precursor which is generated in hydrocarbon flames as an intermediate species following decomposition. It occurs over a broad temperature range in flames and has been identified as a potentially detectable species in 1D Raman/Rayleigh measurements. In this paper, we discuss the Raman spectral signature of acetylene, its temperature dependence, calibration procedures, and the interference with other Raman-active species, C 2 and broadband interferences, all of which are essential to quantify for accurate data processing. In this regard, Raman measurements in laminar and turbulent dimethyl ether flames were acquired. The data are spectrally analyzed and low-temperature calibrations of the Raman response are combined with an extrapolation model for high temperatures in order to place the acetylene signals on an absolute scale. Single-shot and conditional-mean values are presented as a function of mixture fraction for a laminar partially-premixed DME/air flame. Measurements from the laminar flame are compared to a 1D laminar flame calculation to assess the accuracy of the calibration and analysis procedures. Furthermore, detection limits, signal-to-noise, and signal-to-interference ratio of the C 2 H 2 measurement are discussed, and single-shot measurements of acetylene in a turbulent DME/air jet flame are demonstrated.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Measuring the soot onset temperature in high-pressure n-dodecane spray pyrolysis

In this paper, soot formation in pyrolyzing sprays of n-dodecane is visualized and quantified in a high-pressure, high-temperature, constant-volume spray chamber at 38 bar, 76 bar, and 114 bar. Sprays of n-dodecane are injected at 500 bar from a single-hole, 186-µm orifice diameter fuel injector. We quantify the temporal evolution of the soot optical thickness and the total soot mass formed in the pyrolyzing sprays using a high-speed extinction imaging diagnostic. The vessel ambient temperature and pressure are varied independently to identify the soot onset temperature for n-dodecane pyrolysis. Linear extrapolation of the maximum soot formation rates as a function of ambient temperature reveals a soot onset temperature near 1450 K. Finally, the onset temperature determined here for n-dodecane is within 50 K of those previously measured along the centerline of atmospheric pressure coflow diffusion flames for smaller alkane fuels.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Accounting for uncertainty in RCCE species selection

A framework is presented to quantify, based on Bayesian evidence, the relative plausibility of species se- lection options in rate-controlled constrained equilibrium (RCCE) reduced chemical models, accounting for uncertainty in the kinetic parameters and experimental data used to refine them. This approach bal- ances the joint goals of matching available data and avoiding overfitting, which is well-understood to limit extrapolative capacity for true prediction. The methodology is applied to homogeneous autoigni- tion, where predictions are known to be particularly sensitive to chemical model details, specially at low temperatures. It is first introduced for hydrogen–air autoignition using an established mechanism, then demonstrated in two applications of methane–air autoignition using the larger GRI-1.2 mechanism. This larger mechanism significantly increases the computational cost of model selection (though not of the subsequent application in predictions), which is alleviated with a time-scale-guided pre-sorting strategy. Uses and extensions of this new formulation are discussed.

08 HYDROGEN↗

Burning velocities of R-32/O 2 /N 2 mixtures: Experimental measurements and development of a validated detailed chemical kinetic model

This work entails characterizing the flammability of the refrigerant R-32 (CH 2 F 2 ) by both experimental measurements and modeling. Burning velocities S u were measured using a constant-volume spherical-flame method for R-32/O2/N 2 mixtures with O 2 /N 2 ratios ranging from 21% (synthetic air) to 40%, pressures of (1 to 3) bar, and equivalence ratios $\phi$ of (0.8 to 1.3). Based on a critical assessment of available data, and extended by our own calculations, a detailed chemical kinetic model was developed and key reactions determined using reaction path and sensitivity analyses. Initiation and combustion were identified as distinct kinetic regimes and burning velocities were found to be controlled by two primary reactions: unimolecular decomposition of CH 2 F 2 → CHF + HF and the subsequent reaction, CHF + O 2 → CHFO + O, the latter reaction initiating the radical chain propagating and branching by producing O atoms. Sensitive rate constants in the kinetic model were critically adjusted within their uncertainties and current knowledge bounds to best fit the experimental burning velocities. We found that rate constants in the model could be adjusted to match a given experimental S u for specific conditions (O 2 loading, P, T, $\phi$). This, however, then fixes predicted burning velocities for other all conditions within (3 to 4)% if physically realistic rate parameters are maintained. Thus, the entire set of experimental data is fit, not just to particular conditions. Relative random uncertainties in the experimental Su measurements were (4 to 6)%, but assumptions made for thermal radiation lost by the burned gas in the spherical-flame experiments add an additional systematic uncertainty. Systematic differences between the limiting cases of adiabatic (no thermal radiation lost) and optically-thin (all thermal radiation lost) varied significantly with conditions and ranged from (4 to 30)% at high to low velocities, respectively, translating into uncertainties of (2 to 15)% considering the average of two limiting cases. Comparison of experimental and kinetically modeled Su values suggests that the burned gas tends towards the optically-thin limit at the lowest pressures and fuel loadings and toward the adiabatic limit at the highest pressures and loadings. We tested and found support for this conclusion with a detailed analysis as a function of all the conditions (T, P, % O 2 , $\phi$). This behavior appears to transition from optically-thin to adiabatic as the density of the initial fuel increases, which results in increased CO 2 in the burned gas and thus increased absorption of the thermal radiation (consistent with the Beer-Lambert Law). The validated detailed model based on evaluated kinetics is shown to accurately predict burning velocities for R-32 O 2 /N 2 mixtures over a wide range of conditions and provides a reliable basis for extrapolation to other conditions.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Re-evaluation of rate constants for the reaction N 2 H 4 (+ M) $^→_←$ NH 2 + NH 2 (+ M)

Rate constants for the dissociation/recombination reaction N 2 H 4 (+ M) $^→_←$ NH 2 + NH 2 (+ M) are determined by a combination of quantum-chemical calculations and statistical unimolecular rate theory. Between 1100 and 2500 K, limiting low-pressure rate constants for hydrazine dissociation in the bath gas Ar of k 0 = [Ar] 6.1 × 10 20 (T/1000 K) –7.3 exp (-34490 K/T) cm 3 mol –1 s –1 , limiting high-pressure rate constants of k ∞ = 7.6 × 10 16 (T/1000 K) –1.0 exp(-33600 K/T) s –1 , and center broadening factors of the falloff curves (between 1100 and 1600 K in Ar) of F cent = 0.71 exp(- T/1460 K) + 0.29 exp(- T/21 K) + exp(-13400 K/T) were calculated. Using equilibrium constants K C = 1.7 × 10 3 (T/1000 K) –1.5 exp(-33,460 K/T) mol cm –3 , between 300 and 600 K limiting low-pressure rate constants for the reverse recombination of NH 2 radicals in the bath gas N 2 of k rec,0 = [N 2 ] 4.4 × 10 20 (T/400 K) –6.9 exp (-1630 K/T) cm 6 mol –2 s –1 , limiting high-pressure rate constants of k rec,∞ = 4.4 × 10 13 (T/1000 K) 0.44 exp(-140 K/T) cm 3 mol -1 s –1 , and center broadening factors (between 300 and 600 K in N 2 ) of F cent = exp(- T/1130 K) + exp (-10340 K/T) were obtained. A comparison with experimental results for hydrazine dissociation from the literature suggests incomplete falloff extrapolations toward k ∞ and experimental problems in the determination of k 0 at temperatures above about 1600 K. Finally, implications of the present re-evaluated rate constants for the modeling of high temperature ammonia oxidation kinetics are discussed, showing an only small influence of their precise values on the overall properties of the process.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Deep-freeze graph training for latent learning

Scientific and engineering advances are primarily driven by multi-tier conceptual constructs and conditional theoretical frameworks. The theories allow predictions of hypothetical system responses, given a set of approximate conditions (ranges of applicability) imposed on latent parameters that cannot be measured directly. Learning to estimate the latent variables (Latent Learning) helps to pinpoint the anticipated range-edge anomalies and improves the confidence in interpretation, interpolation and extrapolation of limited experimental data. Due to high dimensionality and extreme non-linearity of the materials science problems, very large datasets are typically required for conventional data-driven model development. The vital experimental data collection, particularly on microstructural phases, is very challenging, which makes it difficult to compile a high-quality database. Incorporation of the domain knowledge into the computational graph structure, initialization and optimization processes presents a viable mechanism for developing accurate models, with limited datasets. Furthermore, this study successfully utilized the approach to build the Deep Freeze Graph (DeepFreG) by mapping known causality relationships and by digitizing empirical domain knowledge for Latent Learning (LL), with specific applications in materials science.

36 MATERIALS SCIENCE↗

Developing a robust strength model using physically-informed genetic programming

The strength of materials is influenced by a range of external conditions, such as temperature and deformation rate. Consequently, materials that demonstrate substantial variations in their mechanical behavior due to fluctuations in temperature and strain rate require complex strength models to accurately predict material performance in real-world applications. To predict such complex behavior, a robust and flexible strength model is necessary. In this work, we utilize genetic programming-based symbolic regression (GPSR) to develop data-driven strength models that accurately represent the measured stress–strain responses of tin across a wide range of strain, strain rate and temperature regimes. The GPSR models are constrained by physically-informed conditions, which leads to significant improvement in extrapolation. The best model is integrated into a multi-physics code to perform Taylor impact simulations, validating the model’s accuracy and robustness. In conclusion, the model predictions showed excellent agreement with experimental results, particularly when compared to predictions using traditional strength models.

Genetic programming↗

Graph neural networks for CO 2 solubility predictions in Deep Eutectic Solvents

Deep Eutectic Solvents (DESs) are a promising class of solvents for CO 2 capture. DESs are complex mixtures that can be designed to optimize CO solubility and overall capture process efficiency. However, the vast design landscape of DES mixtures makes experimental investigation prohibitive; as such, there is a need for computational models that can quickly and efficiently navigate the design space and inform data collection efforts. In this work, we propose Graph Neural Network (GNN) models for predicting CO 2 solubility for DESs; the GNN leverages a mixture graph representation that captures the molecular structure of the DES components as well as their intermolecular interactions. Here, we compare the GNN framework against alternative architectures (neural networks, graph convolution networks, and random forests) and data representations (molecular fingerprints, sigma profiles, and graphs). We show that the proposed approach offers superior predictive performance; specifically, we show that solubility can be predicted reliably directly from molecular structure (without the need of using sigma profiles as proposed in previous studies). This result is important, as obtaining sigma profiles requires expensive density functional theory computations. We also explored the ability of GNNs to predict solubility for new DES mixtures and operating conditions. We found that the model extrapolates across temperature reliably. However, we also found deficiencies in the ability of the model to predict solubility for DES mixtures, pressures, and molar ratio not included in the training sets; we show that this is due to an inherent lack of chemical diversity in datasets available in the literature. The proposed computational capabilities can thus help navigate the design space of DES and inform data collection efforts. Our models, data, and benchmarks are shared as Python code implemented in Jupyter notebooks.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

A quantum Monte Carlo study of the molybdenum dimer (Mo 2 )

In this report we have studied the molybdenum dimer (Mo 2 ) system. The binding energy was calculated by means of the fixed-node DMC (FN-DMC) method. The Slater part of the trial wave function was constructed by the Selected-CI method by using the orbitals generated by the KS-DFT method with a hybrid meta-GGA exchange and correlation functional, TPSSh. We also carried out CCSD(T) calculations which were subsequently extrapolated to the complete basis set (CBS) limit. The results are presented.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Evidence of an oxidation induced phase transformation for a delta phase plutonium-gallium alloy

The oxidation of an ~1.5 at% Ga plutonium metal alloy in dry air as a function of time was studied using a combination of FIB-SEM, AES, and XRD. Analysis of SEM images of FIB cross-sections of the oxide allowed for direct measurement of the oxide thickness as well as determination of heterogeneity in the extent of the oxide. AES analysis performed on cross-sections of the oxide allowed for quantitative analysis of the atomic composition and semi-quantitative analysis of the chemical composition. Initial preparation of the sample was found to induce a martensitic transformation to α' - Pu with an average thickness of 233 nm. Electropolishing was found to be sufficient to fully remove the α' layer with creation of only a minimum oxide layer. Growth of the oxide in dry air was found to initially follow parabolic kinetics, with a parabolic rate constant orders of magnitude higher than expected for δ-Pu but in-line with values extrapolated to room temperature for α-Pu. Following complete oxidation of the initial α` layer, further regions of plutonium metal depleted in gallium were observed to form at the oxide / metal interface. However, gallium, as Ga 2 O 3 , was found to be enriched at the surface and present in the oxide in an amount comparable to the relative Ga:Pu ratio of the bulk metal alloy. Finally, the results presented here serve to bridge a diverse array of previous experimental and theoretical data and provide the first comprehensive view of the fate of the δ-stabilizer during formation of an oxide.

36 MATERIALS SCIENCE↗

Revisiting a minimally destructive analytic approach for determining electrochemical kinetic parameters: Measuring aluminum corrosion across a wide pH range based on the Butler-Volmer equation

Here, this study revisits the three-point sampling of the simplified Butler-Volmer equation to address the limitations of strong potentiodynamic polarization, which can introduce irreversible damage and uncertainty in corrosion analysis. The method extracts electrochemical kinetic parameters while minimizing polarization effects, evaluates noise sensitivity relative to overpotential, and accounts for errors from signal noise, OCP drift, ohmic resistance, and mass-transfer constraints. Verified against the Tafel extrapolation method for aluminum corrosion across a wide pH range, this low-polarization approach enables accurate evaluations with specific error estimates, offering a robust alternative to linear polarization resistance methods that assume constant Tafel slopes.

36 MATERIALS SCIENCE↗

Machine learning in materials research: Developments over the last decade and challenges for the future

The number of studies that apply machine learning (ML) to materials science has been growing at a rate of approximately 1.67 times per year over the past decade. In this review, I examine this growth in various contexts. First, I present an analysis of the most commonly used tools (software, databases, materials science methods, and ML methods) used within papers that apply ML to materials science. The analysis demonstrates that despite the growth of deep learning techniques, the use of classical machine learning is still dominant as a whole. It also demonstrates how new research can effectively build upon past research, particular in the domain of ML models trained on density functional theory calculation data. Next, I present the progression of best scores as a function of time on the matbench materials science benchmark for formation enthalpy prediction. In particular, a dramatic improvement of 7 times reduction in error is obtained when progressing from feature-based methods that use conventional ML (random forest, support vector regression, etc.) to the use of graph neural network techniques. Finally, I provide views on future challenges and opportunities, focusing on data size and complexity, extrapolation, interpretation, access, and relevance.

36 MATERIALS SCIENCE↗

Projective multiscale time-integration for electrostatic particle-in-cell methods

The simulation of problems in kinetic plasma physics is often challenging due to strongly coupled phenomena across multiple scales. In this work, we propose a wavelet-based coarse-grained numerical scheme, based on the framework of Equation-Free Projective Integration, for a kinetic plasma system modeled by the Vlasov–Poisson equations. A kinetic particle-in-cell (PIC) code is used to simulate the meso scale dynamics for short time intervals. This allows the extrapolation over long time-steps of the behavior of a coarse wavelet-based discretization of the system. To validate the approach and the underlying concepts, we perform two 1D1V numerical experiments: nonlinear propagation and steepening of an ion wave, and the expansion of a plasma slab in vacuum. The direct comparisons to resolved PIC simulations show good agreement. We show that the speedup of the projective integration scheme over the full particle scheme scales linearly with the system size, demonstrating efficiency while taking into account fully kinetic, non-Maxwellian effects. This suggests that the approach is potentially interesting for kinetic plasma problems with a large separation of scales.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

ELSI — An open infrastructure for electronic structure solvers

Routine applications of electronic structure theory to molecules and periodic systems need to compute the electron density from given Hamiltonian and, in case of non-orthogonal basis sets, overlap matrices. System sizes can range from few to thousands or, in some examples, millions of atoms. Different discretization schemes (basis sets) and different system geometries (finite non-periodic vs. infinite periodic boundary conditions) yield matrices with different structures. The ELectronic Structure Infrastructure (ELSI) project provides an open-source software interface to facilitate the implementation and optimal use of high-performance solver libraries covering cubic scaling eigensolvers, linear scaling density-matrix-based algorithms, and other reduced scaling methods in between. In this paper, we present recent improvements and developments inside ELSI, mainly covering (1) new solvers connected to the interface, (2) matrix layout and communication adapted for parallel calculations of periodic and/or spin-polarized systems, (3) routines for density matrix extrapolation in geometry optimization and molecular dynamics calculations, and (4) general utilities such as parallel matrix I/O and JSON output. The ELSI interface has been integrated into four electronic structure code projects (DFTB+, DGDFT, FHI-aims, SIESTA), allowing us to rigorously benchmark the performance of the solvers on an equal footing. Based on results of a systematic set of large-scale benchmarks performed with Kohn–Sham density-functional theory and density-functional tight-binding theory, we identify factors that strongly affect the efficiency of the solvers, and propose a decision layer that assists with the solver selection process. As a result, we describe a reverse communication interface encoding matrix-free iterative solver strategies that are amenable, e.g., for use with planewave basis sets.

97 MATHEMATICS AND COMPUTING↗

Local reduced-order modeling for electrostatic plasmas by physics-informed solution manifold decomposition

Despite advancements in high-performance computing and modern numerical algorithms, computational cost remains prohibitive for multi-query kinetic plasma simulations. Here, in this work, we develop data-driven reduced-order models (ROMs) for collisionless electrostatic plasma dynamics, based on the kinetic Vlasov-Poisson equation. Our ROM approach projects the equation onto a linear subspace defined by the proper orthogonal decomposition (POD) modes. We introduce an efficient tensorial method to update the nonlinear term using a precomputed third-order tensor. We capture multiscale behavior with a minimal number of POD modes by decomposing the solution manifold into multiple time windows and creating temporally local ROMs. We consider two strategies for decomposition: one based on the physical time and the other based on the electric field energy. Applied to the 1D1V Vlasov–Poisson simulations, that is, prescribed E-field, Landau damping, and two-stream instability, we demonstrate that our ROMs accurately capture the total energy of the system both for parametric and time extrapolation cases. The temporally local ROMs are more efficient and accurate than the single ROM. In addition, in the two-stream instability case, we show that the energy-windowing reduced-order model (EW-ROM) is more efficient and accurate than the time-windowing reduced-order model (TW-ROM). With the tensorial approach, EW-ROM solves the equation approximately 90 times faster than Eulerian simulations while maintaining a maximum relative error of 7.5% for the training data and 11% for the testing data.

Electrostatic plasmas↗