Search NASA⌕ Search

SEARCH · Search NASA

Results for “Stochastic optimization”

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 1,387 records · Page 77

Demonstration of Optimal Benchmark Selection Website and Validation of the q c Coverage Metric Using HEU-SOL-THERM-013-003 Experiment

In the work documented in this interim report, the experiment selection toolkit web site was demonstrated and q C coverage metric methodology was validated for IEU-MET-FAST-002-001, MIX-COMP-THERM 004-004, and HEU-SOL-THERM-013-003 experiments. 𝑞 𝐶 is an information-theoretic measure based on mutual information that quantifies the ability of candidate benchmark experiments to reduce the bias and uncertainty of a target criticality safety application. The metric and an accompanying open-source Python toolkit with a web-based interface were tested against a benchmark set of 425 experiments drawn from the International Criticality Safety Benchmark Evaluation Project Handbook. The interface is hosted at https://edim.covdef.com. It accepts sensitivity data files produced by the TSUNAMI-IP module of the SCALE code system and supports both (i) deterministic analysis using the ENDF/B-VII.0 covariance library and (ii) stochastic analysis based on user-supplied keff samples. Demonstrations on representative applications across a range of material composition, spectrum, and form show that q C -guided benchmark selection achieves greater uncertainty reduction with fewer experiments and yields more stable posterior bias and uncertainty estimates than traditional similarity coefficient ( c k )–based selection, while also capturing valuable low-ck experiments that one-to-one metrics overlook.

Abdel-khalik, Hany S. [Indiana Univ.-Purdue Univ. ↗

An Improved Discrete Least-Squares/Reduced-Basis Method for Parameterized Elliptic PDEs

It is shown that the computational efficiency of the discrete least-squares (DLS) approximation of solutions of stochastic elliptic PDEs is improved by incorporating a reduced-basis method into the DLS framework. In particular, we consider stochastic elliptic PDEs with an affine dependence on the random variable. The goal is to recover the entire solution map from the parameter space to the finite element space. To this end, first, a reduced-basis solution using a weak greedy algorithm is constructed, then a DLS approximation is determined by evaluating the reduced-basis approximation instead of the full finite element approximation. The main advantage of the new approach is that one only need apply the DLS operator to the coefficients of the reduced-basis expansion, resulting in huge savings in both the storage of the DLS coefficients and the online cost of evaluating the DLS approximation. In addition, the recently developed quasi-optimal polynomial space is also adopted in the new approach, resulting in superior convergence rates for a wider class of problems than previous analyzed. Numerical experiments are provided that illustrate the theoretical results.

97 MATHEMATICS AND COMPUTING↗

Development of an Improved Methodology to Assess Potential Unconventional Gas Resources

Considering the important role played today by unconventional gas resources in North America and their enormous potential for the future around the world, it is vital to both policy makers and industry that the volumes of these resources and the impact of technology on these resources be assessed. To provide for optimal decision making regarding energy policy, research funding, and resource development, it is necessary to reliably quantify the uncertainty in these resource assessments. Since the 1970s, studies to assess potential unconventional gas resources have been conducted by various private and governmental agencies, the most rigorous of which was by the United States Geological Survey (USGS). The USGS employed a cell-based, probabilistic methodology which used analytical equations to calculate distributions of the resources assessed. USGS assessments have generally produced distributions for potential unconventional gas resources that, in our judgment, are unrealistically narrow for what are essentially undiscovered, untested resources. In this article, we present an improved methodology to assess potential unconventional gas resources. Our methodology is a stochastic approach that includes Monte Carlo simulation and correlation between input variables. Application of the improved methodology to the Uinta-Piceance province of Utah and Colorado with USGS data validates the means and standard deviations of resource distributions produced by the USGS methodology, but reveals that these distributions are not right skewed, as expected for a natural resource. Our investigation indicates that the unrealistic shape and width of the gas resource distributions are caused by the use of narrow triangular input parameter distributions. The stochastic methodology proposed here is more versatile and robust than the USGS analytic methodology. Adoption of the methodology, along with a careful examination and revision of input distributions, should allow a more realistic assessment of the uncertainty surrounding potential unconventional gas resources.

22 GENERAL STUDIES OF NUCLEAR REACTORS↗

Projection pursuit adaptation on polynomial chaos expansions

Here, the present work addresses the issue of accurate stochastic approximations in high-dimensional parametric space using tools from uncertainty quantification (UQ). The basis adaptation method and its accelerated algorithm in polynomial chaos expansions (PCE) were recently proposed to construct low-dimensional approximations adapted to specific quantities of interest (QoI). The present paper addresses one difficulty with these adaptations, namely their reliance on quadrature point sampling, which limits the reusability of potentially expensive samples. Projection pursuit (PP) is a statistical tool to find the “interesting” projections in high-dimensional data and thus bypass the curse-of-dimensionality. In the present work, we combine the fundamental ideas of basis adaptation and projection pursuit regression (PPR) to propose a novel method to simultaneously learn the optimal low-dimensional spaces and PCE representation from given data. While this projection pursuit adaptation (PPA) can be entirely data-driven, the constructed approximation exhibits mean-square convergence to the solution of an underlying governing equation and thus captures the supports and probability distributions associated with the physics constraints. The proposed approach is demonstrated on a borehole problem and a structural dynamics problem, demonstrating the versatility of the method and its ability to discover low-dimensional manifolds with high accuracy with limited data. In addition, the method can learn surrogate models for different quantities of interest while reusing the same data set.

97 MATHEMATICS AND COMPUTING↗

Distributed Generation Investment by a Microgrid under Uncertainty

This paper examines a California-based microgrid?s decision to invest in a distributed generation (DG) unit fuelled by natural gas. While the long-term natural gas generation cost is stochastic, we initially assume that the microgrid may purchase electricity at a fixed retail rate from its utility. Using the real options approach, we find a natural gas generation cost threshold that triggers DG investment. Furthermore, the consideration of operational flexibility by the microgrid increases DG investment, while the option to disconnect from the utility is not attractive. By allowing the electricity price to be stochastic, we next determine an investment threshold boundary and find that high electricity price volatility relative to that of natural gas generation cost delays investment while simultaneously increasing the value of the investment. We conclude by using this result to find the implicit option value of the DG unit when two sources of uncertainty exist.

29 ENERGY PLANNING, POLICY, AND ECONOMY↗

Detective quantum efficiency of photon-counting x-ray detectors

Purpose: Single-photon-counting (SPC) x-ray imaging has the potential to improve image quality and enable novel energy-dependent imaging methods. Similar to conventional detectors, optimizing image SPC quality will require systems that produce the highest possible detective quantum efficiency (DQE). This paper builds on the cascaded-systems analysis (CSA) framework to develop a comprehensive description of the DQE of SPC detectors that implement adaptive binning. Methods: The DQE of SPC systems can be described using the CSA approach by propagating the probability density function (PDF) of the number of image-forming quanta through simple quantum processes. New relationships are developed to describe PDF transfer through serial and parallel cascades to accommodate scatter reabsorption. Results are applied to hypothetical silicon and selenium-based flat-panel SPC detectors including the effects of reabsorption of characteristic/scatter photons from photoelectric and Compton interactions, stochastic conversion of x-ray energy to secondary quanta, depth-dependent charge collection, and electronic noise. Results are compared with a Monte Carlo study. Results: Depth-dependent collection efficiency can result in substantial broadening of photopeaks that in turn may result in reduced DQE at lower x-ray energies (20–45 keV). Double-counting interaction events caused by reabsorption of characteristic/scatter photons may result in falsely inflated image signal-to-noise ratio and potential overestimation of the DQE. Conclusions: The CSA approach is extended to describe signal and noise propagation through photoelectric and Compton interactions in SPC detectors, including the effects of escape and reabsorption of emission/scatter photons. High-performance SPC systems can be achieved but only for certain combinations of secondary conversion gain, depth-dependent collection efficiency, electronic noise, and reabsorption characteristics.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

A Diagnoser Algorithm for Anomaly Detection in DEDS under Partial Unreliable Observations: Characterization and Inclusion in Sensor Configuration Optimizaton

Complex engineering systems have to be carefully monitored to meet demanding performance requirements, including detecting anomalies in their operations. There are two major monitoring challenges for these systems. The first challenge is that information collected from the monitored system is often partial and/or unreliable, in the sense that some occurred events may not be reported and/or may be reported incorrectly (e.g., reported as another event). The second is that anomalies often consist of sequences of event patterns separated in space and time. This paper introduces and analyzes a diagnoser algorithm that meets these challenges for detecting and counting occurrences of anomalies in engineering systems. The proposed diagnoser algorithm assumes that models are available for characterizing plant operations (via stochastic automata) and sensors (via probabilistic mappings) used for reporting partial and unreliable information. Methods for analyzing the effects of model uncertainties on the diagnoser performance are also discussed. In order to select configurations that reduce sensor costs, while satisfying diagnoser performance requirements, a sensor configuration selection algorithm developed in previous work is then extended for the proposed diagnoser algorithm. The proposed algorithms and methods are then applied to a multi-unit-operation system, which is derived from an actual facility application. Results show that the proposed diagnoser algorithm is able to detect and count occurrences of anomalies accurately and that its performance is robust to model uncertainties. Furthermore, the sensor configuration selection algorithm is able to suggest optimal sensor configurations with significantly reduced costs, while still yielding acceptable performance for counting the occurrences of anomalies.

97 MATHEMATICS AND COMPUTING↗

National Assessment of Energy Storage for Grid Balancing and Arbitrage: Phase 1, WECC

To examine the role that energy storage could play in mitigating the impacts of the stochastic variability of wind generation on regional grid operation, the Pacific Northwest National Laboratory (PNNL) examined a hypothetical 2020 grid scenario in which additional wind generation capacity is built to meet renewable portfolio standard targets in the Western Interconnection. PNNL developed a stochastic model for estimating the balancing requirements using historical wind statistics and forecasting error, a detailed engineering model to analyze the dispatch of energy storage and fast-ramping generation devices for estimating size requirements of energy storage and generation systems for meeting new balancing requirements, and financial models for estimating the life-cycle cost of storage and generation systems in addressing the future balancing requirements for sub-regions in the Western Interconnection. Evaluated technologies include combustion turbines, sodium sulfur (Na-S) batteries, lithium ion batteries, pumped-hydro energy storage, compressed air energy storage, flywheels, redox flow batteries, and demand response. Distinct power and energy capacity requirements were estimated for each technology option, and battery size was optimized to minimize costs. Modeling results indicate that in a future power grid with high-penetration of renewables, the most cost competitive technologies for meeting balancing requirements include Na-S batteries and flywheels.

energy storage↗

An indirect approach to optimize the reaction rates of thermal NO formation for diesel engines

With stringent emission regulations, it has become more important for modern diesel engine manufacturers to accurately predict engine-out nitrogen oxide (NO x ) emissions across a wide range of operating conditions. Thermal NO is the major source of engine-out NO x in modern diesel engines. For thermal NO formation, several earlier studies have recommended the forward and reverse reaction rate coefficients of the rate-limiting reaction (O + N 2 ⇌ NO + N). However, due to deficiencies in sub-models and inadequacies of reduced chemical mechanisms to represent diesel combustion, these recommended values more often than not need to be adjusted in reduced order combustion models to accurately predict engine-out NO x . Hence, in this work a systematic and computationally efficient approach has been proposed to streamline the process of determining the optimum reaction rate coefficients. Here, to develop the optimization approach, four different production diesel engines with different operating conditions in terms of speed, load, and exhaust-gas recirculation have been considered. Numerical simulations have been performed using a detailed zero-dimensional velocity-composition-frequency transported probability density function (0D-VCF-tPDF) model that uses hundreds of notional particles to capture in-cylinder stratification. Four different combinations of hydrocarbon and NO x chemical mechanisms were used to represent chemistry. It was found that for the rate-limiting reaction, the pre-exponent factors (A f1 , A r1 ) and activation energies (E A,f1 , E A,r1 ) of the forward and reverse reaction rates follow a linear band in A f1 - E A,f1 and A r1 - E A,r1 space where predicted engine-out NO x match the measured values closely. By encompassing such bands from different engines and considering constraints on activation energies, a reduced search domain of pre-exponent factors and activation energies was constructed that is expected to be applicable to any diesel engine. Eventually, computationally efficient three-line and one-line search approaches were proposed to determine the optimum values of the pre-exponent factors and activation energies that led to a minimum error between measured and predicted engine-out NO x . Finally, these three-line and one-line NO x optimization approaches were applied to a fifth production diesel engine for which the 0D-VCF-tPDF model showed a very good predictive performance in terms of predicting peak pressure, 50% burn rate, and engine-out NO x when compared to measured and 3D-CFD values.

33 ADVANCED PROPULSION SYSTEMS↗

An automated approach to the alignment of compound refractive lenses

Compound refractive lenses (CRLs) are established X-ray focusing optics, and are thus a key element of many beamlines at X-ray facilities, used to either to focus a beam or image a sample. While CRLs are quite established, the stack of single lens elements affords a very small numerical aperture because of the thick lens profile, making them far more difficult to align than classical optical lenses that obey the thin-lens approximation. This means that the alignment must be very precise and is highly sensitive to changes to the incident beam; aligning CRLs is thus typically treated as a time-consuming manual process. Many groups circumvent day-to-day realignment by using engineering controls (e.g. mounting optics) that sacrifice some of the beam’s focusing characteristics, i.e. spot size, or resolution. While these choices minimize setup time, there are clear disadvantages. This work presents a new automated approach to align CRLs using a simple alignment apparatus that is easy to adapt and install at different types of X-ray experiments or facilities. Our approach builds on recent CRL modeling efforts, using an approach based on the Stochastic Nelder-Mead (SNM) simplex method. We demonstrate the efficacy of this method by numerical simulation and then extend this test to real experiments conducted at the Advanced Photon Source to confirm its performance with a synchrotron X-ray beam. This work provides an opportunity to automate key instrumentation at X-ray facilities.

97 MATHEMATICS AND COMPUTING↗

Algorithm to solve a chance-constrained network capacity design problem with stochastic demands and finite support

Here, we consider the problem of determining the capacity to assign to each arc in a given network, subject to uncertainty in the supply and/or demand of each node. This design problem underlies many real-world applications, such as the design of power transmission and telecommunications networks. We first consider the case where a set of supply/demand scenarios are provided, and we must determine the minimum-cost set of arc capacities such that a feasible flow exists for each scenario. We briefly review existing theoretical approaches to solving this problem and explore implementation strategies to reduce run times. With this as a foundation, our primary focus is on a chance-constrained version of the problem in which α% of the scenarios must be feasible under the chosen capacity, where α is a user-defined parameter and the specific scenarios to be satisfied are not predetermined. We describe an algorithm which utilizes a separation routine for identifying violated cut-sets which can solve the problem to optimality, and we present computational results. We also present a novel greedy algorithm, our primary contribution, which can be used to solve for a high quality heuristic solution. We present computational analysis to evaluate the performance of our proposed approaches.

97 MATHEMATICS AND COMPUTING↗

Training neural networks using Metropolis Monte Carlo and an adaptive variant

Abstract We examine the zero-temperature Metropolis Monte Carlo (MC) algorithm as a tool for training a neural network by minimizing a loss function. We find that, as expected on theoretical grounds and shown empirically by other authors, Metropolis MC can train a neural net with an accuracy comparable to that of gradient descent (GD), if not necessarily as quickly. The Metropolis algorithm does not fail automatically when the number of parameters of a neural network is large. It can fail when a neural network’s structure or neuron activations are strongly heterogenous, and we introduce an adaptive Monte Carlo algorithm (aMC) to overcome these limitations. The intrinsic stochasticity and numerical stability of the MC method allow aMC to train deep neural networks and recurrent neural networks in which the gradient is too small or too large to allow training by GD. MC methods offer a complement to gradient-based methods for training neural networks, allowing access to a distinct set of network architectures and principles.

97 MATHEMATICS AND COMPUTING↗

Occupancy-Based HVAC Control with Short-Term Occupancy Prediction Algorithms for Energy-Efficient Buildings

This study aims to develop a concrete occupancy prediction as well as an optimal occupancy-based control solution for improving the efficiency of Heating, Ventilation, and Air-Conditioning (HVAC) systems. Accurate occupancy prediction is a key enabler for demand-based HVAC control so as to ensure HVAC is not run needlessly when when a room/zone is unoccupied. In this paper, we propose simple yet effective algorithms to predict occupancy alongside an algorithm for automatically assigning temperature set-points. Utilizing past occupancy observations, we introduce three different techniques for occupancy prediction. Firstly, we propose an identification-based approach, which identifies the model via Expectation Maximization (EM) algorithm. Secondly, we study a novel finite state automata (FSA) which can be reconstructed by a general systems problem solver (GSPS). Thirdly, we introduce an alternative stochastic model based on uncertain basis functions. The results show that all the proposed occupancy prediction techniques could achieve around 70% accuracy. Then, we have proposed a scheme to adaptively adjust the temperature set-points according to a novel temperature set algorithm with customers’ different discomfort tolerance indexes. By cooperating with the temperature set algorithm, our occupancy-based HVAC control shows 20% energy saving while still maintaining building comfort requirements.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗

A study of differentiation errors in large-eddy simulations based on the EDQNM theory

This paper is concerned with the investigation of numerical errors in large-eddy simulations by means of two-point turbulence modeling. Based on the eddy-damped quasi-normal Markovian (EDQNM) theory, a stochastic model is developed in order to predict the time evolution of the kinetic energy spectrum obtained by a large-eddy simulation (LES), including the effects of the numerics. Using this framework, the influence of the accuracy of the approximate space differencing schemes on LES quality is studied, for decaying homogeneous isotropic incompressible turbulence, with Reynolds numbers Re{sub {lambda}} based on the transverse Taylor scale equal to 780, 2500 and 8000. The results show that the discretization of the filtered Navier-Stokes equations leads to differentiation and aliasing errors. Error spectra are also presented, and indicate that the numerical errors are mainly originating from the approximate differentiation. In addition, increasing the order of accuracy of the differencing schemes or using algorithms optimized in the Fourier space is found to widen the range of well-resolved scales. Unfortunately, for all the schemes, the smaller scales with wavenumbers close to the grid cut-off wavenumber, are badly calculated and generate differentiation errors over the whole energy spectrum. The eventual use of explicit filtering to remove spurious motions with short wavelength is finally shown to significantly improve LES accuracy.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Deep Reinforcement Scheduling of Energy Storage Systems for Real-time Voltage Regulation in Unbalanced LV Networks with High PV Penetration

The ever-growing higher penetration of distributed energy resources (DERs) in low-voltage (LV) distribution systems brings both opportunities and challenges to voltage support and regulation. This paper proposes a deep reinforcement learning (DRL)-based scheduling scheme of energy storage systems (ESSs) to mitigate system voltage deviations in unbalanced LV distribution networks. The ESS-based voltage regulation problem is formulated as a multi-stage quadratic stochastic program, with the objective of minimizing the expected total daily voltage regulation cost while satisfying operational constraints. While existing voltage regulation methods are mostly focused on onetime- step control, this paper explores a day-horizon systemwide voltage regulation problem. In other words, the size of action and state spaces are extremely high-dimensional and need to be delicately handled. Furthermore, in order to overcome the difficulty of modeling uncertainties and develop a realtime solution, a learn-to-schedule feedback control framework is proposed by adapting the problem to a model-free DRL setting. The proposed algorithm is tested on a customized 6-bus system and a modified IEEE 34-bus system. Simulation results validate the effectiveness and near-optimality of voltage regulation by ESS in comparison with a deterministic quadratic program solution.

Wang, Shengyi↗

A Hybrid Gradient Method to Designing Bayesian Experiments for Implicit Models

Bayesian experimental design (BED) aims at designing an experiment to maximize the information gathering from the collected data. The optimal design is usually achieved by maximizing the mutual information (MI) between the data and the model parameters. When the analytical expression of the MI is unavailable, e.g.,having implicit models with intractable data distributions, a neural network-based lower bound of the MI was recently proposed and a gradient ascent method was used to maximize the lower bound [1]. However, the approach in [1] requires a pathwise sampling path to compute the gradient of the MI lower bound with respect to the design variables, and such a pathwise sampling path is usually inaccessible for implicit models. In this work, we propose a hybrid gradient approach that leverages recent advances in variational MI estimator and evolution strategies (ES)combined with black-box stochastic gradient ascent (SGA) to maximize the MI lower bound. This allows the design process to be achieved through a unified scalable procedure for implicit models without sampling path gradients. Several experiments demonstrate that our approach significantly improves the scalability of BED for implicit models in high-dimensional design space.

Zhang, Jiaxin↗

Continuous-variable quantum key distribution in non-Markovian channels

We address continuous-variable quantum key distribution (QKD) in non-Markovian lossy channels and show how the non-Markovian features may be exploited to enhance security and/or to detect the presence and the position of an eavesdropper along the transmission line. In particular, we suggest a coherent-state QKD protocol which is secure against Gaussian individual attacks based on optimal 1{yields}2 asymmetric cloning machines for arbitrarily low values of the overall transmission line. The scheme relies on specific non-Markovian properties, and cannot be implemented in ordinary Markovian channels characterized by uniform losses. Our results give a clear indication of the potential impact of non-Markovian effects in QKD.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

EVI-EnSitePy (Electric Vehicle Infrastructure – Energy Estimation and Site Optimization Tool in Python) [EVI-X Modeling Suite] [SWR-25-07]

EVI-EnSitePy is a comprehensive agent-based tool designed for the analysis and design of high-power charging sites, encompassing a wide array of site agents including Electric Vehicles (EVs), chargers, energy storage units (ESS), renewable energy resources (DER), and loads. This versatile tool offers diverse functionalities and a modular modeling approach, allowing detailed configuration of agents based on power ratings, port numbers, energy capacities, demand requirements, charger interfaces, and flexibility to customize the tool for project specific goals. By simulating agent interactions and employing various metrics, EVI-EnSitePy enables the assessment of site performance, exploration of energy management systems (EMS), and implementation of innovative EV charging policies. Utilizing EV charge schedules and arrival states, the tool performs thorough charging site simulations, with outputs consisting of agent and site-level power profiles and statistical metrics. Employing a tree graph structure, EVI-EnSitePy supports nested site structures and power distribution modeling. The tool's ability to generate charging schedules deterministically or via stochastic analysis further enhances its versatility. Through its features and capabilities, EVI-EnSitePy offers a powerful platform for informed decision-making in the realm of high-power charging site design and operation.

Jackson, Derek [National Renewable Energy Laborato↗