SEARCH · Engineering Papers

Results for “gradient methods”

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 901 records · Page 50

All sources NASA NTRS DOE OSTI

DOE OSTI2023

hIPPYlib-MUQ: A Bayesian Inference Software Framework for Integration of Data with Complex Predictive Models under Uncertainty

Bayesian inference provides a systematic framework for integration of data with mathematical models to quantify the uncertainty in the solution of the inverse problem. However, the solution of Bayesian inverse problems governed by complex forward models described by partial differential equations (PDEs) remains prohibitive with black-box Markov chain Monte Carlo (MCMC) methods. We present hIPPYlib-MUQ, an extensible and scalable software framework that contains implementations of state-of-the art algorithms aimed to overcome the challenges of high-dimensional, PDE-constrained Bayesian inverse problems. These algorithms accelerate MCMC sampling by exploiting the geometry and intrinsic low-dimensionality of parameter space via derivative information and low rank approximation. The software integrates two complementary open-source software packages, hIPPYlib and MUQ. hIPPYlib solves PDE-constrained inverse problems using automatically-generated adjoint-based derivatives, but it lacks full Bayesian capabilities. MUQ provides a spectrum of powerful Bayesian inversion models and algorithms, but expects forward models to come equipped with gradients and Hessians to permit large-scale solution. By combining these two complementary libraries, we created a robust, scalable, and efficient software framework that realizes the benefits of each and allows us to tackle complex large-scale Bayesian inverse problems across a broad spectrum of scientific and engineering disciplines. To illustrate the capabilities of hIPPYlib-MUQ, we present a comparison of a number of MCMC methods available in the integrated software on several high-dimensional Bayesian inverse problems. These include problems characterized by both linear and nonlinear PDEs, various noise models, and different parameter dimensions. The results demonstrate that large (~ 50×) speedups over conventional black box and gradient-based MCMC algorithms can be obtained by exploiting Hessian information (from the log-posterior), underscoring the power of the integrated hIPPYlib-MUQ framework.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2023

NSGA-PINN: A Multi-Objective Optimization Method for Physics-Informed Neural Network Training

This paper presents NSGA-PINN, a multi-objective optimization framework for the effective training of physics-informed neural networks (PINNs). The proposed framework uses the non-dominated sorting genetic algorithm (NSGA-II) to enable traditional stochastic gradient optimization algorithms (e.g., ADAM) to escape local minima effectively. Additionally, the NSGA-II algorithm enables satisfying the initial and boundary conditions encoded into the loss function during physics-informed training precisely. We demonstrate the effectiveness of our framework by applying NSGA-PINN to several ordinary and partial differential equation problems. In particular, we show that the proposed framework can handle challenging inverse problems with noisy data.

Lu, Binghang (ORCID:0009000160016632)↗

DOE OSTI2023

Optimization Stability in Excited-State-Specific Variational Monte Carlo

Here, we investigate the issue of optimization stability in variance-based state-specific variational Monte Carlo, discussing the roles of the objective function, the complexity of wave function ansatz, the amount of sampling effort, and the choice of minimization algorithm. Using a small cyanine dye molecule as a test case, we systematically perform minimizations using variants of the linear method as both a standalone algorithm and in a hybrid combination with accelerated descent. We demonstrate that adaptive step control is crucial for maintaining the linear method's stability when optimizing complicated wave functions and that the hybrid method enjoys both greater stability and minimization performance.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

DOE OSTI2025

Multi-scale computational screening and mechanistic insights of cyclic amines as solvents for improved lignocellulosic biomass processing

A computational screening workflow for the efficient deconstruction of cellulose, lignin and hemicellulose fractions of lignocellulosic biomass using cyclic amines as solvents. Lignocellulosic biomass is a promising feedstock for production of affordable fuels and chemicals from renewable resources. Effective solubilization and subsequent deconstruction of its cellulose, hemicellulose, and lignin fractions is essential for the viability of future biorefineries. This study used quantum chemistry-based equilibrium thermodynamics methods to evaluate the potential of 650 cyclic amines to solubilize cellulose, hemicellulose, and lignin. The activity coefficients of solvent - biopolymer interactions were predicted using the COSMO-RS (COnductor-like Screening MOdel for Real Solvents) method and used to identify cyclic amines that can efficiently dissolve and extract selective fractions of biopolymers during biomass pretreatment. Among the 650 cyclic amines, 1-piperazineethanmaine was predicted to be an effective solvent for extracting all three polymers and was experimentally shown to achieve the highest lignin removal (97.1%). Non-covalent interaction, reduced density gradient and quantum chemical calculations were performed to elucidate the dissolution mechanism of lignin, cellulose and hemicellulose and gain further molecular level insights into the interactions between the cyclic amines and biomass polymers that promote efficient solubilization and extraction. These analyses indicated that 1-piperazineethanmaine and 1-methylimidazole make noncovalent van der Waals, electrostatic interactions and hydrogen bonding with lignin, leading to enhanced lignin removal, while the strong intramolecular hydrogen bonding interactions in cellulose and hemicellulose result in weaker solvent-biopolymer interactions. Overall, the computational approach provided an efficient method for identifying cyclic amines tailored for optimal biomass pretreatment and resulted in the identification of a potential new class of solvents for effective biomass pretreatment.

Kumar, Nikhil↗

DOE OSTI2021

Randomized Sketching Algorithms for Low-Memory Dynamic Optimization

This paper develops a novel limited-memory method to solve dynamic optimization problems. The memory requirements for such problems often present a major obstacle, particularly for problems with PDE constraints such as optimal flow control, full waveform inversion, and optical tomography. In these problems, PDE constraints uniquely determine the state of a physical system for a given control; the goal is to find the value of the control that minimizes an objective. While the control is often low dimensional, the state is typically more expensive to store. This paper suggests using randomized matrix approximation to compress the state as it is generated and shows how to use the compressed state to reliably solve the original dynamic optimization problem. Concretely, the compressed state is used to compute approximate gradients and to apply the Hessian to vectors. The approximation error in these quantities is controlled by the target rank of the sketch. This approximate first- and second-order information can readily be used in any optimization algorithm. As an example, we develop a sketched trust-region method that adaptively chooses the target rank using a posteriori error information and provably converges to a stationary point of the original problem. Numerical experiments with the sketched trust-region method show promising performance on challenging problems such as the optimal control of an advection-reaction-diffusion equation and the optimal control of fluid flow past a cylinder.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2023

A review on experimentally observed mechanical and microstructural characteristics of interfaces in multi-material laser powder bed fusion

Additive manufacturing (AM) is a revolutionary technology. One of the key AM categories, metal powder-based fusion processes, has many advantages compared to conventional methods for fabricating structural materials, such as permitting increased geometric complexity. While single material metal powder AM has advanced significantly in the past decade, multi-material AM is gradually attracting more attention owing to the recent breakthrough in multi-material feedstock delivery and the growing interest of fabricating functionally graded components. Multi-material AM offers an alternative route for applications that require location dependent material properties and high geometrical complexity. The AM community has invented several ways to achieve compositional gradients and discrete boundaries in two and three dimensions using mechanical spreading, nozzle-based, electrophotographic, and hybrid techniques. This article reviews the current state of laser powder bed fusion based multi-material AM of metals with focuses on the characteristics of the material interface as well as the properties and performance of the AM built functionally graded materials. We show the common challenges and issues related to material transitions, such as defects, segregation, phase separation, and the efficacy of some potential solutions including material and process optimizations. Additionally, this study evaluates the applicability and limitations of the existing testing standards and methods for measuring mechanical performance of functionally graded materials. Finally, we discuss mechanical testing development opportunities, which can help multi-material AM move towards higher technological maturity. In general, we find that the link between gradient microstructure and mechanical properties is not well understood or studied and suggest several mechanical tests that may better inform this knowledge gap.

42 ENGINEERING↗

DOE OSTI2022

Electrostatic powder spreading for metal powder bed fusion applications

Typical powder bed fusion processes spread each layer of powder by pushing it across the build area with a spreader bar or roller. While simple, these methods require good flowing powder and do not have the ability to place more than one powder type per layer. This naturally limits the ability to incorporate more than one material onto each powder layer in powder bed additive manufacturing processes. Here we demonstrate electrostatic powder spreading, a method to deposit a powder layer without mechanical contact with the previous layers. Electrostatic powder spreading (ESPS) uses an electric field formed between a powder container and a counter electrode to move powder particles from the container to the powder bed. We show that the rate at which powder is deposited can be controlled by adjusting the strength of the electric field. Correspondingly, the deposited layer thickness is controlled by the strength of the electric field and the recoating speed. To demonstrate the feasibility of ESPS for laser powder bed fusion powder spreading, nearly fully dense (> 99.8%) coupons were built in a commercial L-PBF machine using the ESPS setup. Finally, we achieve 2D patterning of a single layer by using an array of electrodes and switching the applied electric field on and off. ESPS represents a method to produce an arbitrary patterned multi-material powder bed without complicated powder removal steps and easy implementation into existing L-PBF machines. Furthermore, ESPS presents an opportunity to build parts with 3D gradients in material properties.

36 MATERIALS SCIENCE↗

DOE OSTI2021

Controlling distributed energy resources via deep reinforcement learning for load flexibility and energy efficiency

Behind-the-meter distributed energy resources (DERs), including building solar photovoltaic (PV) technology and electric battery storage, are increasingly being considered as solutions to support carbon reduction goals and increase grid reliability and resiliency. However, dynamic control of these resources in concert with traditional building loads, to effect efficiency and demand flexibility, is not yet commonplace in commercial control products. Traditional rule-based control algorithms do not offer integrated closed-loop control to optimize across systems, and most often, PV and battery systems are operated for energy arbitrage and demand charge management, and not for the provision of grid services. More advanced control approaches, such as MPC control have not been widely adopted in industry because they require significant expertise to develop and deploy. Recent advances in deep reinforcement learning (DRL) offer a promising option to optimize the operation of DER systems and building loads with reduced setup effort. However, there are limited studies that evaluate the efficacy of these methods to control multiple building subsystems simultaneously. Additionally, most of the research has been conducted in simulated environments as opposed to real buildings. This paper proposes a DRL approach that uses a deep deterministic policy gradient algorithm for integrated control of HVAC and electric battery storage systems in the presence of on-site PV generation. The DRL algorithm, trained on synthetic data, was deployed in a physical test building and evaluated against a baseline that uses the current best-in-class rule-based control strategies. Performance in delivering energy efficiency, load shift, and load shed was tested using price-based signals. The results showed that the DRL-based controller can produce cost savings of up to 39.6% as compared to the baseline controller, while maintaining similar thermal comfort in the building. The project team has also integrated the simulation components developed during this work as an OpenAIGym environment and made it publicly available so that prospective DRL researchers can leverage this environment to evaluate alternate DRL algorithms.

24 POWER TRANSMISSION AND DISTRIBUTION↗

DOE OSTI2021

Implicit and coupled fluid plasma solver with adaptive Cartesian mesh and its applications to non-equilibrium gas discharges

In this work, we present a new fluid plasma solver with adaptive Cartesian mesh (ACM) based on a full-Newton (nonlinear, implicit) scheme for non-equilibrium gas discharge plasma. The electrons and ions are described using drift-diffusion approximation coupled to Poisson equation for the electric field. The electron-energy transport equation is solved to account for electron thermal conductivity, Joule heating, and energy loss of electrons in collisions with neutral species. The rate of electron-induced ionization is a function of electron temperature and could also depend on electron density (important for plasma stratification). The ion and gas temperature are kept constant. The transport equations are discretized using a non-isothermal Scharfetter-Gummel scheme to resolve possible large temperature gradients in the sheaths. We demonstrate the new solver for simulations of direct current (DC) and radiofrequency (RF) discharges. The implicit treatment of the coupled equations allows using large time steps. The full-Newton method (FNM) enables fast nonlinear convergence at each time step, offering significantly improved simulation efficiency. We discuss the selection of time steps for solving different plasma problems. The new solver enables solving several problems we could not solve before with existing software: two- and three-dimensional structures of the entire DC discharges including cathode and anode regions, electric field reversals and double-layer formation, the normal cathode spot and an anode ring, moving striations in diffuse and constricted DC discharges, and standing striations in RF discharges. The developed FNM-ACM technique offers many benefits for tackling the disparity of gas discharge plasma systems' time scales and nonlinearity.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

DOE OSTI2026

Accessing the gluon momentum fraction of nucleons through the gradient flow

We calculate the gluon momentum fraction of the nucleon using lattice QCD, with a nonperturbative renormalization technique based on the gradient flow. The gluon momentum fraction is determined on a single Wilson-clover ensemble using 𝑁 𝑓 =2 +1 flavors with pion mass 358 MeV and lattice spacing 0.094 fm. We employ the variational method to reduce excited-state contamination and apply the distillation framework to ensure a large operator basis. To reduce systematic uncertainties, we apply Bayesian model averaging to all fit procedures. We apply matching coefficients to the flow-time dependent lattice results to recover the gluon momentum fraction in the $\overline{MS}$-scheme at 2 GeV. Our final result is ⟨𝑥⟩ 𝑔 ⁢(𝜇 =2 GeV) =0.482⁢(35), where we quote only statistical uncertainties.

Lattice QCD↗

DOE OSTI2024

Non-resonant picosecond three-wave mixing in the gas phase

We report on the experimental observation of non-resonant, second-order optical sum-frequency generation (SFG) in five different atomic and molecular gases. The measured signal is attributed to a SFG process by characterizing its intensity scaling and its polarization behavior. We show that the electric quadrupole mechanism cannot explain the observed trends and suggest a mechanism based on symmetry breaking along the incident beam path arising from laser-induced species ground state number density gradients. Our results demonstrate that the SFG is about four orders of magnitude stronger than the third-harmonic generation (THG) and independent from any externally applied electric fields. In conclusion, these features make this method suitable for gas number density measurements at the picosecond time scale in reactive flows and plasmas.

LaCombe, Grayson↗

DOE OSTI2021

Perspectives on multi-material additive manufacturing

The last two decades have seen enormous gains in industrial adoption of additive manufacturing (AM) technologies. Its layer-wise approach to fabrication offers designers the opportunity to create structures with unique performance advantages over their traditionally manufactured counterparts, and have created new manufacturing business models and supply chains. While today’s AM technologies have enabled the creation of new geometries, future AM systems that offer simultaneous processing of multiple materials in a single build open opportunities for new product functionality that cannot be achieved by traditional manufacturing methods. Advances in multi-material additive manufacturing, which integrate dissimilar material into a complex, three-dimensional object, is emerging but the advances have been sporadic. Moving beyond homogenous materials, adding multi-materials, gradient, functional and responsive materials, and materials with heterogeneous and graded properties means that a single additive process based on either energy delivery or material deposition alone may not be suitable. Finally, this perspective gives a brief overview of the current status, challenges, and future recommendations for multi-material additive manufacturing. The authors aim to expand the notion of multi-material additive manufacturing beyond combining materials with dissimilar properties, to combinations of materials at different length scales, material classes as well as multiple functionalities.

36 MATERIALS SCIENCE↗

DOE OSTI2023

Deep Neural Network Informed Markov Chain Monte Carlo Methods

In subsurface flow modeling, quantifying the uncertainty of model parameters and the corresponding uncertainly on output quantities is a crucial task for groundwater management. Markov chain Monte Carlo (MCMC) methods can take advantage of observed data to estimate parameters in a Bayesian setting. However, MCMC can be slow to converge and produce highly correlated samples when the dimensions of the parameters is high. Using gradients for the posterior distribution can help samplers explore the parameter space more efficiently, but obtaining gradients can be computationally challenging.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

DOE OSTI2024

Enhancing the cooling performance of thermocouples: a power-constrained topology optimization procedure

Abstract Heat pumping through thermoelectric devices has many advantages over traditional cooling. However, their current efficiency is a limiting factor in their implementation. In this paper, we approach the non-convex topology optimization of thermoelectrical elements for cooling applications through the method of moving asymptotes (MMA) to improve their cooling capabilities per watt usage. The optimization problem is defined for a given power budget, aiming for the minimum temperature with a known heat pumping need. The introduction of power as a constraint justifies the introduction of the voltage gradient across the thermocouple as a design variable to maintain the thermoelectrical device in its optimum power-to-heat extraction ratio. To better understand the convergence of this non-convex problem, we present a two-variable analytical thermoelectric optimization model. This example provides information on how to select the penalty parameters used to scale the three material coefficients involved in the problem to obtain lower objective values and better convergence using MMA. The analytical model shows the non-convexity of the problem and provides the recommendation to use penalization coefficients of the form $$p_k=p_{\sigma }>p_{\alpha }=1$$ p k = p σ > p α = 1 for the thermal conductivity, electrical conductivity, and Seebeck coefficients. We tested these penalization coefficients through optimizations of a model based on the 1MC10-031 commercial thermoelectric-cooler (TEC) using the finite element method (FEM). These penalization coefficients provided local minima without the need for volume constraints. With this procedure, we found designs that provided temperatures close to 10 degrees lower using 60% less semiconductor material volume compared to the initial design.

Gutiérrez, G. Reales↗

DOE OSTI2026

A dual dynamic shutter system for accelerating ion irradiation sample throughput via lateral gas implantation gradients

Ion irradiation for material performance testing is limited due to its serial nature, which allows for only one value of the implantation (appm) versus dose (dpa) parameter space to be explored for each ion and experiment at a time. While ion irradiation can accelerate the process by up to three orders of magnitude compared to neutron irradiation experiments, the sample throughput for ion irradiation remains relatively low. To address these limitations, a novel capability has been developed at the Michigan Ion Beam Laboratory (MIBL), enabling for the creation of single- and two-dimensional lateral ion implantation gradients using recently installed motorized-controlled ion-beam shutters. This advancement can generate a wide scope of the two-dimensional (H+, He2+) implantation parameter space within a single sample. Integration of this new capability now allows for dual- and triple-ion beam experiments to be performed with full user control over not only the ion implantation depth, but also laterally across the sample by imposing ion implantation concentration gradients, thus providing researchers with a high-throughput means for material testing under various irradiation conditions. Furthermore, recent improvements in MIBL's microbeam ion-beam analysis (IBA) target station now allow for probing these concentration gradients in irradiated alloys with exceptional spatial resolution, down to 10 µm. These two approaches promise to significantly improve ion irradiation capabilities and increase the sample throughput by several orders of magnitude. The application of the shutter technique plus the subsequent microbeam characterization of the imposed implantation gradients are showcased by two proof-of-principle ion-irradiated experiments, one performed on single-crystal Si and the other on the fusion-candidate alloy F82H-IEA. These advancements mark a substantial leap in ion-beam technology, offering researchers a robust, high-throughput method to efficiently investigate candidate alloys with high technological readiness for both advanced fission and fusion reactor applications, in a time- and cost-effective manner.

36 - MATERIALS SCIENCE↗

DOE OSTI2023

Prevention of Dual-Mode Excitation in 9-Cell Cavities for LCLSII-HE

Dual-Mode Excitation, also referred to as mode-mixing, is a superposition of two fundamental pi modes in an SRF cavity. In 9-cell TESLA cavities used for the LCLSII-HE project, the two modes that are commonly excited are the pi mode (1300.2 MHz), and the 7/9 pi mode (1297.8 MHz). During vertical cavity qualification testing, only the pi mode is driven by the RF system. However, at high gradients, emitted power at the frequency of the 7/9 pi mode grows despite the RF system only powering the pi mode. When this happens, the RF power measurement system is unable to differentiate between the superimposed modes, and any data taken is invalid. The previous method of dealing with mode-mixing was to turn off power to the cavity, let both modes decay fully, then restore power. Because of the time-sensitive nature of this testing, this leads to underestimates of Q0 or any radiant field emission present in the cavity. The novel RF control solution prevents the 7/9 pi mode from being excited while not interfering with measurements of the fundamental pi mode. A second RF control system is connected to the cavity which drives the 7/9 pi mode. The loop phase for driving the 7/9 pi mode in self excited loop mode is determined then shifted by 180 degrees, thus providing a negative feedback to the undesired mode. Because this off-resonance power can be very small, on the order of milliwatts, it does not interfere with the high-power measurements of the fundamental pi mode. At Jefferson Lab, we are now able to test a cavity for the LCLSII-HE project with no problems from mode-mixing, which allows for CW processing of high-gradient multipacting and field emission.

Owen, Peter↗

DOE OSTI2020

Physics-informed semantic inpainting: Application to geostatistical modeling

A fundamental problem in geostatistical modeling is to infer the heterogeneous geological field based on limited measurements and some prior spatial statistics. Semantic inpainting, a technique for image processing using deep generative models, has been recently applied for this purpose, demonstrating its effectiveness in dealing with complex spatial patterns. However, the original semantic inpainting framework incorporates only information from direct measurements, while in geostatistics indirect measurements are often plentiful. In this work, to overcome this limitation, we propose a physics-informed semantic inpainting framework, employing the Wasserstein Generative Adversarial Network with Gradient Penalty (WGAN-GP) and jointly incorporating the direct and indirect measurements by exploiting the underlying physical laws. Our simulation results for a high-dimensional problem with 512 dimensions show that in the new method, the physical conservation laws are satisfied and contribute in enhancing the inpainting performance compared to using only the direct measurements.

54 ENVIRONMENTAL SCIENCES↗

DOE OSTI2021

Understanding fission gas bubble distribution, lanthanide transportation, and thermal conductivity degradation in neutron-irradiated α-U using machine learning

U—10Zr based metallic nuclear fuel is the leading candidate for next-generation sodium-cooled fast reactors in the United States. US research reactors have used and tested this fuel type since the 1960s and accumulated considerable experience and knowledge about the fuel performance. Most of the knowledge, however, remains empirical. The lack of mechanistic understanding of fuel performance puts a large burden on proof through experimental verification for the qualification of U—10Zr fuel for commercial use. Further, this paper proposes an image data-driven machine learning approach, coupled with domain knowledge provided by advanced post irradiation examination, to provide unprecedented quantified insights into the morphology, size, density and the connectivity of fission gas bubbles and their effects on the fission product transportation and thermal conductivity. Specifically, we developed a method to automatically detect, extract statistics, and classify ~19,000 fission gas bubbles into different categories, and quantitatively link the data to lanthanide transportation through connected bubbles and degradation of thermal conductivity along the radial temperature gradient in a neutron irradiated U—10Zr annular fuel. Results indicate the approach can be modified to study other irradiation effects, such as secondary phase redistribution and gaseous fuel swelling in other irradiated nuclear fuels.

11 NUCLEAR FUEL CYCLE AND FUEL MATERIALS↗