SEARCH · Search NASA

Results for “Linear Models”

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 325 records · Page 18

All sources NASA NTRS DOE OSTI

DOE OSTI2026

A Data-Driven Method for Modeling Creep-Fatigue Stress- Strain Behavior Using Neural ODEs

In this paper, we introduce a data-driven machine learning approach for modeling one-dimensional stress–strain behavior under cyclic loading, utilizing experimental data from the nickel-based Alloy 617. The study employs uniaxial creep–fatigue test data acquired under various loading histories and compares two distinct neural network-based ODE models. The first model, known as the black-box model, comprehensively describes the strain–stress relationship using a Neural ODE equation. To interpret this black-box model, we apply the Sparse Identification of Nonlinear Dynamical Systems (SINDy) technique, transforming the black-box model into an equation-based model using symbolic regression. The second model, the Neural flow rule model, incorporates Hooke’s Law for the linear elastic component, with the nonlinear part characterized by a Neural ODE. Both models are trained with experimental data to accurately reflect the observed stress–strain behavior. We conduct a detailed comparison with the standard Chaboche model, which includes three back stresses. Our results demonstrate that the neural network-based ODE models precisely capture the experimental creep–fatigue mechanical behavior, exceeding the standard Chaboche model’s accuracy. Furthermore, an interpretable model derived from the black-box neural ODE model through symbolic regression achieves accuracy comparable to the Chaboche model, enhancing its interpretability. The results highlight the potential of neural network-based ODE models to depict complex creep–fatigue behavior, eliminating the necessity for experts to define a specific, material-focused model form.

creep-fatigue↗

DOE OSTI2024

Probing Signal-Based Inertia and Frequency Response Estimation for Power Systems with High Penetration of Inverter-Based Resources: Preprint

Power system inertia is the inherent capability of a power system to resist changes in its frequency during disturbances. Real-time inertia estimation technology has become more important due to the low-inertia issues caused by the increasing integration levels of inverter-based resources (IBRs) from renewable energy; however, existing inertia estimation methods hardly consider multiple frequency response controls that act within the same time frame as conventional inertial response, thus making measured inertia values vary under different testing conditions. To resolve this issue, this paper proposes a novel real-time estimation method to simultaneously estimate a power system's inertia constant and frequency response droop constant using a well-designed probing signal. First, we formulate the inertia and frequency response model of a power system with IBRs. Second, through the integration and manipulation of the developed model, we propose a multivariate linear regression- based estimation method that is resilient to measurement noise. Third, we design a probing signal that can be injected by IBRs to incite the required transients for estimation. Finally, we validate the proposed estimation method through comprehensive power- hardware-in-the-loop experiments using inverter hardware and a realistic island power system model. The results demonstrate that the proposed method can accurately estimate the inertia and droop value of the power system with grid-following IBRs and grid-forming IBRs with virtual synchronous machine control.

frequency response↗

DOE OSTI2024

Probing Signal-Based Inertia and Frequency Response Estimation for Power Systems With High Levels of Inverter-Based Resources

Power system inertia is the inherent capability of a power system to resist changes in its frequency during disturbances. Real-time inertia estimation technology has become more important due to the low-inertia issues caused by the increasing integration levels of inverter-based resources (IBRs) from renewable energy; however, existing inertia estimation methods hardly consider multiple frequency response controls that act within the same time frame as conventional inertial response, thus making measured inertia values vary under different testing conditions. To resolve this issue, this paper proposes a novel real-time estimation method to simultaneously estimate a power system's inertia constant and frequency response droop constant using a well-designed probing signal. First, we formulate the inertia and frequency response model of a power system with IBRs. Second, through the integration and manipulation of the developed model, we propose a multivariate linear regression-based estimation method that is resilient to measurement noise. Third, we design a probing signal that can be injected by IBRs to incite the required transients for estimation. Finally, we validate the proposed estimation method through comprehensive power-hardware-in-the-loop experiments using inverter hardware and a realistic island power system model. The results demonstrate that the proposed method can accurately estimate the inertia and droop value of the power system with grid-following IBRs and grid-forming IBRs with virtual synchronous machine control.

frequency response↗

DOE OSTI2024

Systematic Local Simulations of Fast Neutrino Flavor Conversions with Scattering Effects

We investigate the dynamics of fast neutrino flavor conversions (FFCs) in the one-dimensional (1D) and zero-dimensional (0D) models, in which spatial advection is considered and ignored, respectively. In this study, we employ snapshots obtained by our self-consistent, realistic Boltzmann-neutrino-radiation-hydrodynamics simulations. We show that the FFC growth rate is considerably larger in the 1D model than in the 0D model, as expected from the previous linear analysis results. We find that the momentum space dimension does not significantly influence the neutrino transition probability in 1D models. On the other hand, in the 0D model without collisions, the FFC depends on the momentum space, and the azimuthal angle dependence breaks the periodicity of the FFC. Our study demonstrates that collisional instability can lead to further flavor conversions on a long timescale in 1D models after the asymptotic state of FFC has been reached. Such an effect should be taken into consideration when the fast and collisional flavor instabilities coexist.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

DOE OSTI2025

Symmetry Determining Equations of the Rankine-Hugoniot Equations for Variable Velocity Shock Waves

The “constant velocity piston” problem (Fig. 1), also known as the “piston problem,” is a standard model for a one dimensional, in our case linear, symmetric shock wave moving through an inviscid, perfect gas. The model can be divided into two regions - a perturbed section on the left and an unperturbed section on the right - by a moving shock wave moving left to right. Both the perturbed and unperturbed sections, i.e. the shocked and unshocked regions, respectively, obey the Eulerian conservation equations; however, at the exact location of the shock, there is a mathematical discontinuity not satisfied by the Euler equations. To ensure continuity and conservation of certain quantities when crossing between the unshocked and shocked regions, we evoke a series of equations derived from the Eulerian conservation equations, called the Rankine-Hugoniot equations, or “jump” equations as it is often referred to in the literature on the topic. The classical constant-velocity piston problem assumes the piston features a constant driving velocity (among many other willing suspensions of belief required in the pursuit of a first principles equation model); consequent to this assumption is a constant-velocity shock and a constant-velocity shocked flow state. However, using Lie Group Theory (LGT), also known as symmetry analysis, we can attempt to reinterpret the model with a shock wave of variable velocity in time and space. An extension of the model in this way opens up the possibility for obtaining new analytical solutions to the piston problem for certain shock velocity models. In this report, we use LGT to derive the symmetry determining equations (SDEs), whose solutions are Lie groups, which permit analytical solutions. In the future, we can then use the SDEs to define constraint equations on the shock velocity model and what the successive solutions to the Euler equations might be based off such constraints. This report is structured as follows: Section 2 provides a brief derivation of the Rankine-Hugoniot (“jump”) equations; Section 3 gives an overview of Lie group theory; Section 4 derives the SDEs of the jump equations; Section 5 derives the Euler conservation equations for fluids; and Section 6 presents concluding remarks and opportunities for future studies.

42 ENGINEERING↗

DOE OSTI2024

Modeling gas migration through clay-based buffer material using coupled multiphase fluid flow and geomechanics with stress-dependent gas permeability

A model for gas migration through clay-based buffer material is developed for modeling gas generation and migration associated with deep geologic nuclear waste disposal. The model is based on a multiphase fluid flow and geomechanics simulator that is adapted to consider enhanced gas flow when gas pressure is high enough to approach the confining stress magnitude. A key feature in the model is a direct coupling between gas permeability and stress, through a non-linear stress-dependent permeability function. The model was first tested and calibrated by modelling two different laboratory gas migration tests on Wyoming (MX-80) bentonite samples. The calibrated model was then applied to model gas migration through a bentonite buffer of a large-scale gas injection test (Lasgit) conducted at the Äspö Hard Rock Laboratory in Sweden. Observed preferential gas migration along interfaces (between compacted blocks and along the canister surface) required explicit representation of such interfaces in the model. The model with the stress-dependent gas permeability accurately captured observed experimental responses in terms of gas breakthrough time, peak gas pressure, and cumulative gas flow rates. The calibrated model was finally applied to simulate migration of hydrogen gas generated within a breached nuclear waste canister over 10,000 years, involving migration of much larger gas volumes. For the considered gas generation rate and host rock properties, the generated gas could migrate through the bentonite buffer and released into the surrounding host rock at a maximum gas pressure somewhat higher than the initial total stress, though a significant amount of hydrogen remained within the buffer. This modelling sets the stage for further detailed analysis of the impact of hydrogen gas generation on the long-term performance of nuclear waste repositories.

12 MANAGEMENT OF RADIOACTIVE AND NON-RADIOACTIVE W↗

DOE OSTI2024

GPU Accelerated Sparse Cholesky Factorization

The solution of sparse symmetric positive definite linear systems is an important computational kernel in large-scale scientific and engineering modeling and simulation. We will solve the linear systems using a direct method, in which a Cholesky factorization of the coefficient matrix is performed using a right-looking approach and the resulting triangular factors are used to compute the solution. Sparse Cholesky factorization is compute intensive. In this work we investigate techniques for reducing the factorization time in sparse Cholesky factorization by offloading some of the dense matrix operations on a GPU. We will describe the techniques we have considered. We achieved up to 4x speedup compared to the CPU-only version.

Karsavuran, M Ozan↗

DOE OSTI2025

Frequency-domain modeling of floating wind arrays with shared mooring lines

A frequency-domain model for floating wind turbine dynamics has been extended to model floating wind farms with couplings from shared mooring systems. The model, called RAFT, could previously calculate the mean offsets and wave-induced response spectra for single floating wind turbines. Now, the model supports multiple floating wind turbines, each with their own properties and responses, along with mooring lines that run directly between floating wind turbines in the array, meaning that shared mooring lines or fully suspended dynamic power cables can be included. This capability is achieved by setting up an array-level solution of the system mean offsets and assembling the full system matrices for solving the dynamic response. The quasi-static mooring model MoorPy is used to linearize the mooring system properties. To compute the floating wind turbine relative motions, phase offsets are applied to each turbine's response as a function of wave frequency based on the wavelength and relative positions in the array. These differential motions are then applied to mooring system tension Jacobians to compute the tension loads in the shared mooring lines. Overall, the capability provides a frequency-domain analog to the modeling capabilities of the floating support structure in FAST.Farm. Mean offsets and power spectral density plots of responses are compared between RAFT and FAST.Farm to verify the implementation. The results indicate good agreement within the expectations of a frequency-domain modeling approach and suggest correct implementation of the shared mooring aspects. Additionally, a unique comb-like frequency response in the shared mooring line tensions has been observed. This phenomenon has a clear physical basis and may be an important design consideration for future shared mooring systems.

17 WIND ENERGY↗

DOE OSTI2024

A Green's Function Wind Turbine Induction Model That Incorporates Complex Inflow Conditions

ABSTRACT In this work, we develop a new analytical turbine induction model that can incorporate complex inflow conditions including cases where the wind velocity and temperature profiles can vary as functions of height. This induction model is derived from the linearized Navier–Stokes and leads to a second‐order ODE that can be solved using a Green's function formulation. The corresponding Green's function for several configurations are found including the infinite domain, semi‐infinite domain with ground plane, and a power law velocity inflow profile. The results of this approach are then compared with simulations of the turbine induction field using the AMR‐Wind CFD solver with a uniformly loaded actuator disk model. These comparisons show that the Green's function approach captures the centerline blockage, three‐dimensional blockage flow field, and streamwise velocity slow down, with very good agreement for lower thrust conditions and at larger distances away from rotor disk. The effects of shear on the turbine blockage were also compared using a power law inflow profile, and we show that this approach matches the CFD predictions for the cases considered.

17 WIND ENERGY↗

DOE OSTI2026

Robust calibration of non-perturbative models with history matching

We apply, for the first time, Bayes Linear Emulation and History Matching to the calibration of non-perturbative models in Monte Carlo event generators. In contrast to the usual approach of "Monte Carlo tuning", History Matching does not result in best-fit plus ellipsoidal parameter uncertainty estimates but instead identifies all parameter space regions that are consistent with data. This approach leads to a systematic and robust quantification of parametric uncertainties in the models, especially in those challenging cases where different, possibly disjoint, regions of parameter space deliver similar results, which are usually not properly treated with current methodology. We highlight the power of this method with the hadronisation models available through Sherpa : the built-in cluster fragmentation Ahadic and string fragmentation through an interface to Pythia .

Iskauskas, Andrew [Durham U. (main)] (ORCID:000000↗

DOE OSTI2025

Heterogeneous Mixtures of Dictionary Functions to Approximate Subspace Invariance in Koopman Operators: Why Deep Koopman Operators Work

Abstract Koopman operators model nonlinear dynamics as a linear dynamic system acting on a nonlinear function as the state. This nonstandard state is often called a Koopman observable and is usually approximated numerically by a superposition of functions drawn from a dictionary . In a widely used algorithm, extended dynamic mode decomposition (EDMD), the dictionary functions are drawn from a fixed class of functions. Deep learning combined with EDMD has been used to learn novel dictionary functions in an algorithm called deep dynamic mode decomposition (deepDMD). The learned representation both (1) accurately models and (2) scales well with the dimension of the original nonlinear system. In this paper, we analyze the learned dictionaries from deepDMD and explore the theoretical basis for their strong performance. We explore State-Inclusive Logistic Lifting (SILL) dictionary functions to approximate Koopman observables. Error analysis of these dictionary functions show they satisfy a property of subspace approximation, which we define as uniform finite approximate closure. Typically, a Koopman dictionary’s nonlinear functions are homogeneous. In this paper, we discover that structured mixing of heterogeneous dictionary functions drawn from different classes of nonlinear functions achieve the same accuracy and dimensional scaling as the deep-learning-based deepDMD algorithm Yeung et al. ( In: 2019 American Control Conference (ACC), 2019). We specifically show this by building a heterogeneous dictionary comprised of SILL functions and conjunctive radial basis functions (RBFs). This mixed dictionary achieves similar accuracy and dimensional scaling to deepDMD with an order of magnitude reduction in parameters, while maintaining geometric interpretability. These results strengthen the viability of dictionary-based Koopman models to solving high-dimensional nonlinear learning problems.

Johnson, Charles A.↗

DOE OSTI2027

A multi-region approach for the analysis of porous materials and gas-solid reactions using USAXS-SAXS-WAXS: application to CaO carbonation

The reaction rate of gas-solid non-catalytic reactions is typically investigated using reactant conversion data over time and ex-situ measurements of the porous solid reactant textural properties; these data enable the experimental estimation of the initial intrinsic reaction rate, whereas the evolution of the textural properties and the reaction rate over time are evaluated theoretically using reaction models. In this work, a different methodology is presented, based on: a) in-situ time-resolved USAXS-SAXS-WAXS measurements of the solid sample, and b) a multi-region modeling approach; this methodology allows for the estimation of the textural properties and the intrinsic reaction rate at any time during the reaction. Referring to a reaction in which a solid product is obtained from a solid reactant and a gaseous reactant, the porous particle is described as consisting of several distinct regions with different microstructural properties, and both a gas-solid reaction model and a SAXS model are derived and applied to the carbonation of porous CaO. The proposed reaction model highlights the role of the inaccessible reactant and accurately predicts the solid reactant conversion versus time profile. The SAXS model accurately predicts the measured linear trends of the Porod invariant and of the pre-factor of the power law scattering profile versus CaO mass fraction; in the absence of chemical reactions, the proposed equations extend classical SAXS theory to porous materials containing macropores and nonporous solid phases in addition to standard nanoscale inhomogeneous regions.

CO2 solid sorbents↗

DOE OSTI2025

Hydrodynamic characterization of the coastal pioneer array ocean observing system

Ocean observation buoys require relatively small amounts of power, yet traditionally necessitate costly resupply trips for battery replacement. With the offshore location of the buoys and small power requirements, wave energy may be an effective solution for providing consistent and reliable power to support the buoy instrumentation. The US National Science Foundation Ocean Observatories Initiative (OOI) includes arrays of point absorber-like buoy systems used for ocean observation that have been deployed at multiple locations including the Southern Mid-Atlantic Bight. A study is currently underway to design a pitch resonator wave energy converter to supplement existing renewable energy generation for powering observation instrumentation. This paper details field measurements from surface moorings of the OOI Coastal Pioneer Array, which informs the subsequent development of a numerical model for the moored observation system. The model is developed in Wave Energy Converter Simulator (WEC-Sim), which leverages the Simscape multibody solver within the MATLAB/Simulink framework and linear potential flow theory to simulate the hydrodynamic interactions and multibody dynamics in 6 degrees of freedom. Multiple tuning variables are considered to produce a model for the system that matches well with empirical data (about 8% error). In conclusion, the WEC-Sim model will serve as a platform for integrating the pitch resonator wave energy converter concept and deployment preparation (detailed design including power take-off and control systems, response evaluation, etc.).

hydrodynamic modeling↗

DOE OSTI2025

Modeling the effects of active wake mixing on wake behavior through large-scale coherent structures

The use of active wake mixing (AWM) to mitigate downstream turbine wakes has created new opportunities for reducing power losses in wind farms. However, many current analytical or semi-empirical wake models do not capture the flow instabilities that are excited through the blade pitch actuation. In this work, we develop a framework, which accounts for the impacts of the large-scale coherent structures and turbulence on the mean flow, for modeling AWM. The framework uses a triple-decomposition approach for the unsteady flow field and models the mean flow and fine-scale turbulence with a parabolized Reynolds-averaged Navier–Stokes (RANS) system. The wave components are modeled using a simplified spatial linear stability formulation that captures the growth and evolution of the coherent structures. Comparisons with high-fidelity large eddy simulations (LESs) of the turbine wakes showed that this framework was able to capture the additional wake mixing and faster wake recovery in the far-wake regions for both the pulse and helix AWM strategies with minimal computational expense. In the near-wake region, some differences are observed in both the RANS velocity profiles and initial growth of the large-scale structures, which may be due to some simplifying assumptions used in the model.

17 WIND ENERGY↗

DOE OSTI2024

WEC-Sim Modeling of Laminar Scientific's Patented Seesaw Wave Energy Converter: Preprint

Laminar Scientific's patented seesaw wave energy converter was modeled in WEC-Sim to predict performance. The device operates by utilizing ocean surface waves to rotate a truss in pitch about a pivot. The pivot is located at the top of two pylons, which are embedded in the seafloor. The seesaw has a float on either end, and the buoyancy forces from each float cause the system to rise or fall with passing waves. Device performance relies upon seesaw length and ocean wavelength creating an antiphase effect. The seesaw truss has an adjustable length intended to achieve this effect. The operation method enforces a narrow band of wavelengths which induce the largest rotational motion from the device. The hydrodynamic analysis of the device was performed using Capytaine, and the results confirmed that the device operates best in a narrow frequency band. Four float-to-float spacing cases and three pylon radii were examined. The hydrodynamic results indicate a match between the model and the physical expectations for the device, and that varying the pylon radii by 0.1-m increments for three instances creates minimal changes in hydrodynamic properties. Power matrices for three float spacing cases of the device were simulated with Joint North Sea Wave Project spectra waves and optimal power take-off damping in WEC-Sim. The maximum average power production for the 15-m spacing case was 14.1 kW with a 5.0-s peak wave period and 4-m significant wave heights. Plots of capture-width ratios indicated that the device performance was linear and confirmed that the device is optimal in a narrow frequency band. The maximum percentage of the available wave power produced by the 15-m device was approximately 16%. Simulations of the device in regular waves were used to produce plots of average power compared to a ratio of float spacing to wavelength. These plots indicate that the power production is maximized at a ratio of 0.5, and further confirm that the device has a narrow frequency response. The device was simulated at an example field location, where the device produced an annual average power rating of 1.6 kW given an average omnidirectional wave climate of 10.3 kW m-1 and an optimal, linearized power take-off model. While the maximum predicted device performance is reliant upon a narrow band of wave frequencies, the conducted analysis provides an opportunity to improve device design prior to prototyping and testing. Modifying the design to respond to a broader frequency range would improve device performance.

numerical modeling↗

DOE OSTI2024

Data-Efficient Dimensionality Reduction and Surrogate Modeling of High-Dimensional Stress Fields

Tensor datatypes representing field variables like stress, displacement, velocity, etc., have increasingly become a common occurrence in data-driven modeling and analysis of simulations. Numerous methods [such as convolutional neural networks (CNNs)] exist to address the meta-modeling of field data from simulations. As the complexity of the simulation increases, so does the cost of acquisition, leading to limited data scenarios. Modeling of tensor datatypes under limited data scenarios remains a hindrance for engineering applications. Here, in this article, we introduce a direct image-to-image modeling framework of convolutional autoencoders enhanced by information bottleneck loss function to tackle the tensor data types with limited data. The information bottleneck method penalizes the nuisance information in the latent space while maximizing relevant information making it robust for limited data scenarios. The entire neural network framework is further combined with robust hyperparameter optimization. We perform numerical studies to compare the predictive performance of the proposed method with a dimensionality reduction-based surrogate modeling framework on a representative linear elastic ellipsoidal void problem with uniaxial loading. The data structure focuses on the low-data regime (fewer than 100 data points) and includes the parameterized geometry of the ellipsoidal void as the input and the predicted stress field as the output. The results of the numerical studies show that the information bottleneck approach yields improved overall accuracy and more precise prediction of the extremes of the stress field. Additionally, an in-depth analysis is carried out to elucidate the information compression behavior of the proposed framework.

artificial intelligence↗

DOE OSTI2025

Simplifying activations with linear approximations in neural networks

A key step in Neural Networks is activation. Among the different types of activation functions, sigmoid, tanh, and others involve the usage of exponents for calculation. From a hardware perspective, exponential implementation implies the usage of Taylor series or repeated methods involving many addition, multiplication, and division steps, and as a result are power-hungry and consume many clock cycles. We implement a piecewise linear approximation of the sigmoid function as a replacement for standard sigmoid activation libraries. This approach provides a practical alternative by leveraging piecewise segmentation, which simplifies hardware implementation and improves computational efficiency. In this paper, we detail piecewise functions that can be implemented using linear approximations and their implications for overall model accuracy and performance gain. Our results show that for the DenseNet, ResNet, and GoogLeNet architectures, the piecewise linear approximation of the sigmoid function provides faster execution times compared to the standard TensorFlow sigmoid implementation while maintaining comparable accuracy. Specifically, for MNIST with DenseNet, accuracy reaches 99.91% (Piecewise) vs. 99.97% (Base) with up to 1.31x speedup in execution time. For CIFAR-10 with DenseNet, accuracy improves to 98.97% (Piecewise) vs. 99.40% (Base) while achieving 1.24x faster execution. Similarly, for CIFAR-100 with DenseNet, the accuracy is 97.93% (Piecewise) vs. 98.39% (Base), with a 1.18x execution time reduction. These results confirm the proposed method’s capability to efficiently process large-scale datasets and computationally demanding tasks, offering a practical means to accelerate deep learning models, including LSTMs, without compromising accuracy.

Activation function↗

DOE OSTI2026

Near-Optimal Performance of Stochastic Model Predictive Control

Here, this article presents a regret analysis for stochastic model predictive control (SMPC) in linear systems with quadratic performance index and additive and multiplicative uncertainties. Under a finite support assumption, the problem can be cast as a finite-dimensional quadratic program, but the problem becomes quickly intractable as the problem size grows exponentially in the horizon length. SMPC aims to compute approximate solutions by solving a sequence of problems with truncated prediction horizons and committing the solution in a receding-horizon fashion. Although this approach is widely used in practice, its performance relative to the optimal solution is not well understood. This article reports for the first time a rigorous near-optimal performance guarantee of SMPC: under stabilizability and detectability conditions, the regret of SMPC is exponentially small in the prediction horizon length, allowing SMPC to achieve near-optimal performance at a substantially reduced computational expense.

93E20, 93B45↗