Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Differentiability”

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 1,243 records · Page 69
All sourcesNASA NTRSDOE OSTI
DOE OSTI2022

Dynamic calibration of differential equations using machine learning, with application to turbulence models

We present a methodology for calibration of parametric ordinary and partial differential equation models, using off-the-shelf software for back-propagation in Neural Networks (NN). As a prototypical example, we consider calibration of a Reynolds-averaged Navier-Stokes (RANS) turbulence closure model, against ground truth data from direct numerical simulations (DNS) of two different turbulent flows. Numerical time integration is represented as a custom NN, where only the RANS model parameters are trainable. A loss function is defined to quantify the mismatch between the NN prediction and the ground truth over a predefined, finite time integration window. This loss function is then minimized using a gradient descent method utilizing the back-propagation algorithm. Furthermore, this dynamic approach to training is to be contrasted with a static approach, wherein a least square regression estimate for parameters is obtained in the limit of an infinitesimal time integration window. In a first test of static and dynamic approaches against ground truth data generated by the model, the former proves to be significantly faster and more accurate than the latter at recovering the parameters. When both calibration approaches are tested against DNS data, for which it is known that the model cannot achieve a perfect fit, the static approach yields a good prediction only for short times, while the dynamic approach results in physical and stable predictions over the entire integration window. After optimization of the dynamic approach for time step, spatial resolution, stability, and physics-based constraints, we obtain a 50% improvement of outcomes over those obtained from the existing, manually calibrated set of parameters, demonstrating the merits of this systematic and automated procedure.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2025

Solving high-dimensional partial integral differential equations: The finite expression method

Partial integro-differential equations (PIDEs) have broad applications in the sciences, from electro-magnetism to options pricing. Here, in this paper, we introduce a new finite expression method (FEX) to solve PIDEs. This approach builds upon the original FEX and its inherent advantages with new advances: 1) A novel method of parameter grouping is proposed to reduce the number of coefficients in high-dimensional function approximation; 2) A Taylor series approximation method is implemented to significantly improve the computational efficiency and accuracy of the evaluation of the integral terms of PIDEs. The new FEX based method, denoted FEX-PG to indicate the addition of the parameter grouping (PG) step to the algorithm, provides both high accuracy and interpretable numerical solutions, with the outcome being an explicit equation that facilitates intuitive understanding of the underlying solution structures. These features are often absent in traditional methods, such as finite element methods (FEM) and finite difference methods, as well as in deep learning-based approaches. To benchmark our method against recent advances, we apply the new FEX-PG to solve benchmark PIDEs in the literature. In high-dimensional settings, FEX-PG exhibits strong and robust performance, achieving relative errors on the order of single precision machine epsilon, significantly outperforming existing approaches based on neural networks.

Combinatorial optimization↗
DOE OSTI2026

Active operator learning with predictive uncertainty quantification for partial differential equations

With the increased prevalence of neural operators being used to provide rapid solutions to partial differential equations (PDEs), understanding the accuracy of model predictions and the associated error levels is necessary for deploying reliable surrogate models in scientific applications. Existing uncertainty quantification (UQ) frameworks employ ensembles or Bayesian methods, which can incur substantial computational costs during both training and inference. Here, we propose a lightweight predictive UQ method tailored for Deep operator networks (DeepONets) that also generalizes to other operator networks. Numerical experiments on linear and nonlinear PDEs demonstrate that the framework’s uncertainty estimates are unbiased and provide accurate out-of-distribution uncertainty predictions with a sufficiently large training dataset. Our framework provides fast inference and uncertainty estimates that can efficiently drive outer-loop analyses that would be prohibitively expensive with conventional solvers. We demonstrate how predictive uncertainties can be used in the context of Bayesian optimization and active learning problems to yield improvements in accuracy and data-efficiency for outer-loop optimization procedures. In the active learning setup, we extend the framework to Fourier Neural Operators (FNO) and describe a generalized method for other operator networks. To enable real-time deployment, we introduce an inference strategy based on precomputed trunk outputs and a sparse placement matrix, reducing evaluation time by more than a factor of five. Our method provides a practical route to uncertainty-aware operator learning in time-sensitive settings.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2025

Evaluating switch lifetime in soft-switched single-stage differential-mode SST

The reliability of semiconductor switches in single-stage differential-mode solid-state transformers (DM-SSTs) has not been systematically evaluated under soft-switching operation and realistic grid conditions. This paper presents a switch-level reliability analysis for soft-switched and hard-switched DM-SST configurations by integrating converter-specific power loss modeling with empirical lifetime prediction. Analytical derivation of device current profiles specific to the DM-SST is used to characterize electrothermal stress, which is then mapped to lifetime using degradation models obtained from power cycling tests (PCTs). Applied to realistic SST load profiles and grid voltage variations, this approach provides a probabilistic prediction of switch lifetime for the DM-SST. Lifetime estimates for both SiC MOSFETs and Si IGBTs are presented, offering insight into device degradation under converter operating conditions. The results quantify the reliability benefits of soft switching in single-stage SSTs, highlighting how switching dynamics influence long-term switch degradation.

14 SOLAR ENERGY↗
DOE OSTI2017

A power compensated differential scanning calorimeter for protein stability characterization

This study presented a power compensated MEMS differential scanning calorimeter (DSC) for protein stability characterization. In this microfabricated sensor, PDMS (Polydimethylsiloxane) and polyimide were used to construct the adiabatic chamber (1 μL) and temperature sensitive vanadium oxide was used as the thermistor material. A power compensation system was implemented to maintain the sample and reference at the same temperature. The resolution study and step response characterization indicated the high sensitivity (6 V/W) and low noise level (60 μk) of the device. The test with IgG1 antibody (mAb1) samples showed clear phase transitions and the data was confirmed to be reasonable by comparing it with the results of commercial DSC’s test. Finally, this device used ~1uL sample amount and could complete the scanning process in 4 min, significantly increasing the throughput of the bimolecular thermodynamics study like drug formulation process.

36 MATERIALS SCIENCE↗
DOE OSTI2025

Investigating Fast Scanning Calorimetry and Differential Scanning Calorimetry as Screening Tools for Thermoset Polymer Material Compatibility with Laser-Based Powder Bed Fusion

As additive manufacturing (AM) technology has developed and progressed, a constant topic of research in the area is expanding the library of materials to be used with these techniques. Among AM methods that utilize polymers, laser-based powder bed fusion (PBF-LB) has preferentially used thermoplastic polymers as its starting materials, but the deposition and material joining method employed in PBF-LB may also be compatible with powdered thermoset polymer precursors as feedstocks. To assess the compatibility of candidate thermosetting polymers and PBF-LB, characterization techniques and protocols that link fundamental material behavior to material behavior in the processing environment are needed. Therefore, the objectives of this work are to compare the curing behavior measured with two different calorimetry techniques that can operate in different heating rate regimes, differential scanning calorimetry (DSC) and fast scanning calorimetry (FSC), and to assess the capabilities of these techniques to act as materials screening tools for PBF-LB. A commercial polyester powder coating is used as a model material to evaluate the potential of obtaining complimentary information for material screening through a combination of calorimetry methods, and its non-isothermal curing behavior is measured at heating rates between 5 °C/min and 7500 °C/min. Curing exotherms are observed with both calorimetry techniques, and comparing the enthalpy associated with curing shows that incomplete curing occurs at higher heating rates, with relative conversion values of approximately 30%. The curing data are fit with two isoconversional models, Friedman and Starink, which show a reduced activation energy at higher heating rates as well, signifying a lower barrier to curing at the conditions used in the FSC experiments. Overall, the results of this work indicate that using these two calorimetry techniques as tiered screening tools can provide valuable information about how curing may proceed in PBF-LB and inform materials selection and design activities for additive manufacturing.

36 MATERIALS SCIENCE↗
DOE OSTI2024

Weak baselines and reporting biases lead to overoptimism in machine learning for fluid-related partial differential equations

One of the most promising applications of machine learning in computational physics is to accelerate the solution of partial differential equations (PDEs). The key objective of machine-learning-based PDE solvers is to output a sufficiently accurate solution faster than standard numerical methods, which are used as a baseline comparison. Here, we first perform a systematic review of the ML-for-PDE-solving literature. Out of all of the articles that report using ML to solve a fluid-related PDE and claim to outperform a standard numerical method, we determine that 79% (60/76) make a comparison with a weak baseline. Second, we find evidence that reporting biases are widespread, especially outcome reporting and publication biases. We conclude that ML-for-PDE-solving research is overoptimistic: weak baselines lead to overly positive results, while reporting biases lead to under-reporting of negative results. To a large extent, these issues seem to be caused by factors similar to those of past reproducibility crises: researcher degrees of freedom and a bias towards positive results. We call for bottom-up cultural changes to minimize biased reporting as well as top-down structural reforms to reduce perverse incentives for doing so.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2020

Comparative differential cuproproteomes of Rhodobacter capsulatus reveal novel copper homeostasis related proteins

Abstract Copper (Cu) is an essential, but toxic, micronutrient for living organisms and cells have developed sophisticated response mechanisms towards both the lack and the excess of Cu in their environments. In this study, we achieved a global view of Cu-responsive changes in the prokaryotic model organism Rhodobacter capsulatus using label-free quantitative differential proteomics. Semi-aerobically grown cells under heterotrophic conditions in minimal medium (∼0.3 μM Cu) were compared with cells supplemented with either 5 μM Cu or with 5 mM of the Cu-chelator bathocuproine sulfonate. Mass spectrometry based bottom-up proteomics of unfractionated cell lysates identified 2430 of the 3632 putative proteins encoded by the genome, producing a robust proteome dataset for R. capsulatus. Use of biological and technical replicates for each growth condition yielded high reproducibility and reliable quantification for 1926 of the identified proteins. Comparison of cells grown under Cu-excess or Cu-depleted conditions to those grown under minimal Cu-sufficient conditions revealed that 75 proteins exhibited statistically significant (p < 0.05) abundance changes, ranging from 2- to 300-fold. A subset of the highly Cu-responsive proteins was orthogonally probed using molecular genetics, validating that several of them were indeed involved in cellular Cu homeostasis.

59 BASIC BIOLOGICAL SCIENCES↗
DOE OSTI2021

The Princess Project: From Differential to Integral Experiments

Following the shutdown of the CEA Valduc experimental facilities, where, for more than 50 years, IRSN used to perform experiments related to criticality safety, IRSN initiated a new project named PRINCESS (PRoject for IRSN Neutron physics and Criticality Experimental data Supporting Safety). The objective is to continue collecting experimental data necessary for the IRSN missions in nuclear safety. For this purpose, collaborations with various national and international laboratories have been established. The PRINCESS project covers various nuclear physics fields from nuclear data to criticality-safety and reactor physics providing information to both differential and integral data improvements.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗
DOE OSTI2021

Reconstructing magnetic deflections from sets of proton images using differential evolution

Proton imaging is a powerful technique for imaging electromagnetic fields within an experimental volume, in which spatial variations in proton fluence are a result of deflections to proton trajectories due to interaction with the fields. When deflections are large, proton trajectories can overlap, and this nonlinearity creates regions of greatly increased proton fluence on the image, known as caustics. The formation of caustics has been a persistent barrier to reconstructing the underlying fields from proton images. We have developed a new method for reconstructing the path-integrated magnetic fields, which begins to address the problem posed by caustics. Our method uses multiple proton images of the same object, each image at a different energy, to fill in the information gaps and provide some uniqueness when reconstructing caustic features. We use a differential evolution algorithm to iteratively estimate the underlying deflection function, which accurately reproduces the observed proton fluence at multiple proton energies simultaneously. We test this reconstruction method using synthetic proton images generated for three different, cylindrically symmetric field geometries at various field amplitudes and levels of proton statistics and present reconstruction results from a set of experimental images. Here, the method we propose requires no assumption of deflection linearity and can reliably solve for fields underlying linear, nonlinear, and caustic proton image features for the selected geometries and is shown to be fairly robust to noise in the input proton intensity.

47 OTHER INSTRUMENTATION↗
DOE OSTI2021

Stiff neural ordinary differential equations

Neural Ordinary Differential Equations (ODEs) are a promising approach to learn dynamical models from time-series data in science and engineering applications. This work aims at learning neural ODEs for stiff systems, which are usually raised from chemical kinetic modeling in chemical and biological systems. We first show the challenges of learning neural ODEs in the classical stiff ODE systems of Robertson’s problem and propose techniques to mitigate the challenges associated with scale separations in stiff systems. We then present successful demonstrations in stiff systems of Robertson’s problem and an air pollution problem. The demonstrations show that the usage of deep networks with rectified activations, proper scaling of the network outputs as well as loss functions, and stabilized gradient calculations are the key techniques enabling the learning of stiff neural ODEs. The success of learning stiff neural ODEs opens up possibilities of using neural ODEs in applications with widely varying time-scales, such as chemical dynamics in energy conversion, environmental engineering, and life sciences.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2024

Analysis of differential scanning calorimetry data for aged plutonium

Differential scanning calorimetry data for samples of a 52 year old plutonium alloy with 3.3 at. % Ga that were heated beyond the melting point is analyzed using transition state theory to find activation energies for the δ to ε and ε to liquid phase transitions. A Bayesian statistical method involving a Gaussian process model is used to find mean values and confidence intervals for the activation energies. The activation energy for the δ to ε phase transition increases by 3.3 ± 3.8% per decade, relative to the case when all age related plutonium lattice point defects have been removed through annealing. The corresponding increase in activation energy for the ε to liquid transition is shown to be 7.1 ± 1.8% per decade. It is postulated that the change in activation energy with age for both phase transitions is caused, in part, by the accumulation of the same type of lattice point defects associated with the observed increase in elastic bulk modulus over time.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2017

Stable perturbations of linear differential equations generating a uniformly bounded group

Stability problems for solutions of the differential equation u{sup ′}(t)=Au+εB(t,u) in a Banach space are considered. It is assumed that for ε=0 this equation generates a uniformly bounded group of class C{sub 0}. Sufficient conditions on B and A are found under which the solutions of this equation are bounded for small ε. A linearization principle is proved for this equation under certain conditions on the operator B. Bibliography: 9 titles. (paper)

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2019

Direct Numerical Simulation of Turbulent Flow Inside a Differentially Heated Composite Cavity

In this work, turbulent flow inside a modified differentially heated cavity at high Rayleigh number (Ra ~ 10 9 ) has been studied through fully resolved direct numerical simulation (DNS) using the high-order spectral element method code Nek5000. The flow configuration includes two separate physical phenomena: the natural recirculation itself, and the flow inside a curved channel. Simulations have been carried out using both the Boussinesq approximation and the low-Mach compressible formulation. Significant discrepancies between the two methods inform of the extreme caution that should be exercised when using the Boussinesq approximation in the limits of its applicability. The DNS solutions are analyzed in terms of polynomial-order convergence and Reynolds stress budgets, and the turbulence quantities and velocity profiles are presented as a reference for the validation of turbulence models.

22 GENERAL STUDIES OF NUCLEAR REACTORS↗
DOE OSTI2025

A physics-constrained neural ordinary differential equations approach for robust learning of stiff chemical kinetics

The high computational cost associated with solving for detailed chemistry poses a significant challenge for predictive computational fluid dynamics (CFD) simulations of turbulent reacting flows. While deep learning techniques have been explored to develop faster surrogate models, they often fail to integrate reliably with CFD solvers. This instability arises because traditional deep learning approaches optimize for training error without ensuring compatibility with ordinary differential equation (ODE) solvers, resulting in accumulation of errors over time. Recently, neuralODE (NODE) based approaches have been shown to be a promising technique to emulate and accelerate detailed chemistry computations. Here, in the present work, we extend this NODE framework for stiff chemical kinetics by incorporating mass conservation constraints directly into the loss function during training. This ensures that the total mass as well as the individual elemental species masses are conserved in an a-posteriori manner. Proof-of-concept studies are performed with the novel physics-constrained NODE (PC-NODE) approach for homogeneous autoignition of hydrogen-air mixture over a range of composition and thermodynamic conditions. It is demonstrated that the PC-NODE framework not only improves the physical consistency of the resulting data-driven model with respect to mass conservation criteria, but also improves training efficiency. PC-NODE is shown to achieve 2–100× speedup relative to the hydrogen-air detailed chemical mechanism depending on the type of the ODE solver (implicit or explicit) used during autoregressive inference tests. Lastly, a-posteriori studies are performed wherein the trained PC-NODE model is coupled with a CFD solver. It is shown that higher accuracy is achieved with PC-NODE relative to the purely data-driven NODE approach. Moreover, PC-NODE also exhibits robustness and generalizability to unseen initial conditions from within (interpolative capability) as well as outside (extrapolative capability) the training regime.

computational combustion↗
DOE OSTI2016

Design and fabrication of a differential scanning nanocalorimeter

This paper describes the design, fabrication, and characterization of a differential scanning nanocalorimeter that significantly reduces the sample volume to microliters and can potentially improve the temperature sensitivity to 10 µK. The nanocalorimeter consists of a polymeric freestanding membrane, four high-sensitive low-noise thermistors based on silicon carbide (SiC), and a platinum heater and temperature sensor. With the integrated heater and sensors, temperature scanning and power compensation can be achieved for calorimetric measurement. Temperature sensing SiC film was prepared by using sintered SiC target and DC magnetron sputtering under different gas pressures and sputtering power. The SiC sensing material is characterized through the measurement of current–voltage curves and noise levels. The thermal performance of a fabricated nanocalorimeter is studied in simulation and experiment. The experiment results show the device has excellent thermal isolation to hold thermal energy. As a result, the noise test together with the simulation show the device is promising for micro 10 µK temperature sensitivity and nanowatt resolution which will lead to low-volume ultra-sensitive nanocalorimetry for biological processes, such as protein folding and ligand binding.

42 ENGINEERING↗
DOE OSTI2019

Negative differential resistance and quantum oscillations in FeSb 2 with embedded antimony

We present a systematical study on single crystalline FeSb 2 using electrical transport and magnetic torque measurements at low temperatures. Nonlinear magnetic field dependence of Hall resistivity demonstrates a multi-carrier transport instinct of the electronic transport. Current-controlled negative differential resistance (CC-NDR) observed in current–voltage characteristics below ~ 7 K is closely associated with the intrinsic transition ~ 5 K of FeSb 2 , which is, however, mediated by extrinsic current-induced Joule heating effect. The antimony crystallized in a preferred orientation within the FeSb 2 lattice in the high-temperature synthesis process leaves its fingerprint in the de Haas-Van Alphen (dHvA) oscillations, and results in the regular angular dependence of the oscillating frequencies. Nevertheless, possible existence of intrinsic non-trivial states cannot be completely ruled out. Our findings call for further theoretical and experimental studies to explore novel physics on flux-free grown FeSb 2 crystals.

36 MATERIALS SCIENCE↗
DOE OSTI2021

Optimized finite-build stellarator coils using automatic differentiation

A new stellarator coil design code is introduced that optimizes the position and winding pack orientation of finite-build coils. The new code, called flexible optimized curves in space using automatic differentiation (AD) and finite build (FOCUSADD), performs gradient-based optimization in a high-dimensional, non-convex space. The derivatives with respect to parameters of finite-build coils are easily and efficiently computed using AD. FOCUSADD parametrizes coil positions in free space using a Fourier series and uses a multi-filament approximation to the coil winding pack. The orientation of the winding pack is parametrized with a Fourier series and can be optimized as well. Optimized finite-build coils for a Wendelstein 7-X (W7-X)-like stellarator are found, and compared with filamentary coil results. The final positions of optimized finite-build W7-X-like coils are shifted, on average, by approximately 2.5 mm relative to optimized filamentary coils. These results suggest that finite-build effects should be accounted for in the optimization of stellarators with low coil tolerances.

43 PARTICLE ACCELERATORS↗