Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “stochastic 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.

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

At least 1,099 records · Page 61
All sourcesNASA NTRSDOE OSTI
DOE OSTI2014

Fast variance reduction for steady-state simulation and sensitivity analysis of stochastic chemical systems using shadow function estimators

We address the problem of estimating steady-state quantities associated to systems of stochastic chemical kinetics. In most cases of interest, these systems are analytically intractable, and one has to resort to computational methods to estimate stationary values of cost functions. In this work, we introduce a novel variance reduction algorithm for stochastic chemical kinetics, inspired by related methods in queueing theory, in particular the use of shadow functions. Using two numerical examples, we demonstrate the efficiency of the method for the calculation of steady-state parametric sensitivities and evaluate its performance in comparison to other estimation methods.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2018

Impact Studies and Cooperative Voltage Control for High PV Penetration

The drastically increasing penetration of photovoltaic (PV) has posed tremendous challenges in operation, control, and protection in distribution networks. The over-voltage issue is one of major concerns for extremely high PV penetration (e.g. 100%). It is critical to identify the most sensitive locations to high PV penetration and apply effective methods to mitigate the over-voltage issue. This paper introduced a novel greedy search algorithm to locate worst-case scenarios of over-voltage in both distributed and centralized PV configurations, as well as a cooperative voltage control method to mitigate over-voltage. The search algorithm has the computational advantage over existing stochastic approach and provides more insights in impact studies. The cooperative control method relies upon reactive power compensation to converge system voltage to a predefined value within standard limits. Moreover, a clustering strategy is applied to divide the network into several clusters, within which all nodes share information with neighboring nodes following a scalable communication architecture. The control algorithms and communication hierarchies presented in this paper have been implemented into OpenDSS to form the Multi Agent (MA) OpenDSS package, which has been made available to the reader. The effectiveness of developed methods are demonstrated through simulation studies on the IEEE 123-node test feeder under four different load levels.

Cooperative control↗
DOE OSTI2023

On unifying randomized methods for inverse problems

This work unifies the analysis of various randomized methods for solving linear and nonlinear inverse problems with Gaussian priors by framing the problem in a stochastic optimization setting. By doing so, we show that many randomized methods are variants of a sample average approximation (SAA). More importantly, we are able to prove a single theoretical result that guarantees the asymptotic convergence for a variety of randomized methods. Additionally, viewing randomized methods as an SAA enables us to prove, for the first time, a single non-asymptotic error result that holds for randomized methods under consideration. Another important consequence of our unified framework is that it allows us to discover new randomization methods. Here, we present various numerical results for linear, nonlinear, algebraic, and PDE-constrained inverse problems that verify the theoretical convergence results and provide a discussion on the apparently different convergence rates and the behavior for various randomized methods.

42 ENGINEERING↗
DOE OSTI2020

Optimization-Driven Scenario Grouping

Scenario decomposition algorithms for stochastic programs compute bounds by dualizing all nonanticipativity constraints and solving individual scenario problems independently. Here, we develop an approach that improves on these bounds by reinforcing a carefully chosen subset of nonanticipativity constraints, effectively placing scenarios into groups. Specifically, we formulate an optimization problem for grouping scenarios that aims to improve the bound by optimizing a proxy metric based on information obtained from evaluating a subset of candidate feasible solutions. We show that the proposed grouping problem is NP-hard in general, identify a polynomially solvable case, and present two formulations for solving the problem: a matching formulation for a special case and a mixed-integer programming formulation for the general case. We use the proposed grouping scheme as a preprocessing step for a particular scenario decomposition algorithm and demonstrate its effectiveness in solving standard test instances of two-stage 0–1 stochastic programs. Using this approach, we are able to prove optimality for all previously unsolved instances of a standard test set. Additionally, we implement this scheme as a preprocessing step for PySP, a publicly available and widely used implementation of progressive hedging, and compare this grouping approach with standard grouping approaches on large-scale stochastic unit commitment instances. Finally, the idea is extended to propose a finitely convergent algorithm for two-stage stochastic programs with a finite feasible region.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2015

Decadal Prediction and Stochastic Simulation of Hydroclimate Over Monsoonal Asia

We developed further our advanced methods of time series analysis and empirical model reduction (EMR) and applied them to climatic time series relevant to hydroclimate over Monsoonal Asia. The EMR methodology was both generalized further and laid on a rigorous mathematical basis via multilayered stochastic models (MSMs). We identified easily testable conditions that imply the existence of a global random attractor for MSMs and allow for non-polynomial predictors. This existence, in turn, guarantees the numerical stability of the MSMs so obtained. We showed that, in the presence of low-frequency variability (LFV), EMR prediction can be improved further by including information from selected times in the system’s past. This prediction method, dubbed Past-Noise Forecasting (PNF), was successfully applied to the Madden-Julian Oscillation (MJO). Our time series analysis and forecasting methods, based on singular-spectrum analysis (SSA) and its enhancements, were applied to several multi-centennial proxy records provided by the Lamont team. These included the Palmer Drought Severity Index (PDSI) for 1300–2005 from the Monsoonal Asia Drought Atlas (MADA), and a 300-member ensemble of pseudo-reconstructions of Indus River discharge for 1702–2005. The latter was shown to exhibit a robust 27-yr low-frequency mode, which helped multi-decadal retroactive forecasts with no look-ahead over this 300-year interval.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2018

Gradient-based optimization for regression in the functional tensor-train format

Predictive analysis of complex computational models, such as uncertainty quantification (UQ), must often rely on using an existing database of simulation runs. In this paper we consider the task of performing low-multilinear-rank regression on such a database. Specifically we develop and analyze an efficient gradient computation that enables gradient-based optimization procedures, including stochastic gradient descent and quasi-Newton methods, for learning the parameters of a functional tensor-train (FT). We compare our algorithms with 22 other nonparametric and parametric regression methods on 10 real-world data sets and show that for many physical systems, exploiting low-rank structure facilitates efficient construction of surrogate models. Here, we use a number of synthetic functions to build insight into behavior of our algorithms, including the rank adaptation and group-sparsity regularization procedures that we developed to reduce overfitting. Finally we conclude the paper by building a surrogate of a physical model of a propulsion plant on a naval vessel.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗
DOE OSTI2021

Algorithm advances and applications of time‐dependent first‐principles simulations for ultrafast dynamics

Abstract Far from equilibrium phenomenon is a central theme of contemporary material research. Such phenomenon can exhibit itself in atomic structure and dynamics, but very often it also happens as non‐equilibrium phenomenon in the electronic structure. In ab initio material simulation, density functional theory (DFT) has played an essential role in studying electronic ground state problems. For excited states, besides many‐body perturbation theory, another powerful tool is the time dependent DFT (TDDFT) method. In particular, the real‐time TDDFT (rt‐TDDFT) method can be used to simulate many non‐equilibrium phenomena directly. Here we introduce our works on some algorithm advances based on our recently rt‐TDDFT method. This method uses the plane‐wave basis set, and significantly accelerates its efficiency by increasing the time step from 0.1–1 as in traditional methods to 0.2–0.5 fs. The noncollinear magnetic moments and spin–orbit coupling have also been included in our rt‐TDDFT method. Furthermore, a Boltzmann‐TDDFT algorithm has been developed to solve the hot carrier overheating problem in Ehrenfest dynamics, and a natural orbital branching algorithm has been developed to overcome the mean‐field approximation in Ehrenfest dynamics nuclear trajectory, thus allows stochastic multiple paths in chemical reactions. Utilizing these methods, we have studied the photoinduced ultrafast demagnetization, ultrafast phase transition, energy transfer between plasmon and hot carriers, as well as the high‐energy ion implantation and low‐energy atomic diffusion in semiconductors. We believe the tools as the ones introduced here can enable us to study a wide range of phenomena which are of great interest in modern day material research. This article is categorized under: Structure and Mechanism > Computational Materials Science Electronic Structure Theory > Ab Initio Electronic Structure Methods Electronic Structure Theory > Density Functional Theory

Liu, Wen‐Hao↗
DOE OSTI2013

Simulation-based optimal Bayesian experimental design for nonlinear systems

The optimal selection of experimental conditions is essential to maximizing the value of data for inference and prediction, particularly in situations where experiments are time-consuming and expensive to conduct. We propose a general mathematical framework and an algorithmic approach for optimal experimental design with nonlinear simulation-based models; in particular, we focus on finding sets of experiments that provide the most information about targeted sets of parameters. Our framework employs a Bayesian statistical setting, which provides a foundation for inference from noisy, indirect, and incomplete data, and a natural mechanism for incorporating heterogeneous sources of information. An objective function is constructed from information theoretic measures, reflecting expected information gain from proposed combinations of experiments. Polynomial chaos approximations and a two-stage Monte Carlo sampling method are used to evaluate the expected information gain. Stochastic approximation algorithms are then used to make optimization feasible in computationally intensive and high-dimensional settings. These algorithms are demonstrated on model problems and on nonlinear parameter inference problems arising in detailed combustion kinetics.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2015

Diagrammatic theories of anharmonic molecular vibrations

In this review, we summarise recent developments in our laboratory in the field of many-body quantum-mechanical calculations of the anharmonic vibrational structure of molecules. Our size-extensive vibrational self-consistent field (XVSCF) and size-extensive second-order many-body perturbation (XVMP2) methods are, unlike their parent methods (VSCF and VMP2), defined in diagram- matic formulations of the energies and Dyson self-energies, leading to manifestly size-consistent expressions for zero-point energies and anharmonic vibrational frequencies calculable with much greater efficiency. The effective one-mode potentials of XVSCF are quadratic and hence the Schrödinger equation for each mode can be solved analytically, unlike VSCF, where a basis-set expansion of wave functions on more complex one-mode potentials need to be performed; VSCF potentials and their minima (anharmonic geometry) are shown to reduce to the quadratic potentials and their minima (also given analytically) of XVSCF in the thermodynamic limit. By self-consistently solving the Dyson equation with frequency-dependent self-energies, XVMP2 has the ability to calculate anharmonic frequencies of fundamentals as well as combinations and overtones in the presence of strong anharmonic resonance without a multireference or quasi-degenerate formulation, which tends to be non-size-consistent. To eliminate the computational bottleneck of XVSCF and XVMP2, which is the high-rank force-constant evaluation, we have developed alternative algorithms in which the diagrammatic equations are recast as a small number of high-dimensional integrals and then evaluated stochastically using a Metropolis Monte Carlo (MC) method. These MC-XVSCF and MC-XVMP2 methods not only remove the need for force-constant evaluation or storage, but also take into account force constants of up to infinite order according to their importance. Furthermore, they are a new branch of quantum Monte Carlo which can calculate frequencies (excitation energies) directly without fixed-node errors.

Dyson equation↗
DOE OSTI2015

Combining molecular dynamics with mesoscopic Green’s function reaction dynamics simulations

In many reaction-diffusion processes, ranging from biochemical networks, catalysis, to complex self-assembly, the spatial distribution of the reactants and the stochastic character of their interactions are crucial for the macroscopic behavior. The recently developed mesoscopic Green’s Function Reaction Dynamics (GFRD) method enables efficient simulation at the particle level provided the microscopic dynamics can be integrated out. Yet, many processes exhibit non-trivial microscopic dynamics that can qualitatively change the macroscopic behavior, calling for an atomistic, microscopic description. We propose a novel approach that combines GFRD for simulating the system at the mesoscopic scale where particles are far apart, with a microscopic technique such as Langevin dynamics or Molecular Dynamics (MD), for simulating the system at the microscopic scale where reactants are in close proximity. This scheme defines the regions where the particles are close together and simulated with high microscopic resolution and those where they are far apart and simulated with lower mesoscopic resolution, adaptively on the fly. The new multi-scale scheme, called MD-GFRD, is generic and can be used to efficiently simulate reaction-diffusion systems at the particle level.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2014

Diffusion and stochastic island generation in the magnetic field line random walk

The cross-field diffusion of field lines in stochastic magnetic fields described by the 2D+slab model is studied using a semi-analytic statistical approach, the decorrelation trajectory method. We show that field line trapping and the associated stochastic magnetic islands strongly influence the diffusion coefficients, leading to dependences on the parameters that are different from the quasilinear and Bohm regimes. A strong amplification of the diffusion is produced by a small slab field in the presence of trapping. The diffusion regimes are determined and the corresponding physical processes are identified.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2013

Comparison of SERPENT and CASMO-5M for pressurized water reactors models

The objective of this work is to perform a preliminary assessment of the capability of SERPENT to generate cross sections for a PWR Beginning-of-Life (BOL) isothermal mini-core by comparing a SERPENT/PARCS solution with the results obtained using a CASMO-5M/PARCS approach. The PARCS code is used instead of the usual SIMULATE-3 to analyze the Swiss Reactors, because interfaces with PARCS already exist to obtain neutronic data from SERPENT. For the PWR configurations, the differences between CASMO-5M and SERPENT solutions are within 200 pcm at the assembly level and thus rather small when considering the deterministic transport method (energy/angular/space discretization) in CASMO-5M versus the stochastic treatment of SERPENT, the statistical uncertainties in the Monte-Carlo approach as well as the eventual differences in nuclear data used by both codes. At the 2D mini-core level, no major difference is observed when comparing PARCS run with CASMO-5M versus SERPENT cross sections. For the generation of kinetic parameters, non trivial differences are observed due both to the methods and the data used. For the relatively limited number of configurations considered, it is hard to make any definitive statement on the benefits of using Monte Carlo codes in terms of nuclear data generation. (authors)

21 SPECIFIC NUCLEAR REACTORS AND ASSOCIATED PLANTS↗
DOE OSTI2014

Array magnetics modal analysis for the DIII-D tokamak based on localized time-series modelling

Here, time-series analysis of magnetics data in tokamaks is typically done using block-based fast Fourier transform methods. This work presents the development and deployment of a new set of algorithms for magnetic probe array analysis. The method is based on an estimation technique known as stochastic subspace identification (SSI). Compared with the standard coherence approach or the direct singular value decomposition approach, the new technique exhibits several beneficial properties. For example, the SSI method does not require that frequencies are orthogonal with respect to the timeframe used in the analysis. Frequencies are obtained directly as parameters of localized time-series models. The parameters are extracted by solving small-scale eigenvalue problems. Applications include maximum-likelihood regularized eigenmode pattern estimation, detection of neoclassical tearing modes, including locked mode precursors, and automatic clustering of modes, and magnetics-pattern characterization of sawtooth pre- and postcursors, edge harmonic oscillations and fishbones.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗
DOE OSTI2012

Optimization of Monte Carlo transport simulations in stochastic media

This paper presents an accurate and efficient approach to optimize radiation transport simulations in a stochastic medium of high heterogeneity, like the Very High Temperature Gas-cooled Reactor (VHTR) configurations packed with TRISO fuel particles. Based on a fast nearest neighbor search algorithm, a modified fast Random Sequential Addition (RSA) method is first developed to speed up the generation of the stochastic media systems packed with both mono-sized and poly-sized spheres. A fast neutron tracking method is then developed to optimize the next sphere boundary search in the radiation transport procedure. In order to investigate their accuracy and efficiency, the developed sphere packing and neutron tracking methods are implemented into an in-house continuous energy Monte Carlo code to solve an eigenvalue problem in VHTR unit cells. Comparison with the MCNP benchmark calculations for the same problem indicates that the new methods show considerably higher computational efficiency. (authors)

21 SPECIFIC NUCLEAR REACTORS AND ASSOCIATED PLANTS↗
DOE OSTI2019

Modelling the Fluctuations of Reactive Shock Waves in Heterogeneous Solid Explosives as Stochastic Processes

Although current phenomenological burn models are useful for describing the average or bulk reactive flow behaviour of heterogeneous explosives, one fundamental weakness inherent to these models is the loss of detailed microstructural information at the scale of the calculation. In order to include the effects of the microstructure, and in particular the underlying material heterogeneities that influence the build-up to detonation, a new paradigm is put forth for modelling sub-grid, reaction-induced fluctuations (i.e. “hot spots”) at the continuum level. This modelling approach assumes that the reaction rate is stochastic, rather than deterministic, and it uses Langevin-type equations with a mathematical framework built upon Itô calculus and Lambourn's CIM model for shocked heterogeneous explosives. This approach follows directly from our previous letter, Ref. [1], and is inspired by the probability density function (pdf) methods used in turbulent reactive flows. Here, the stochastic burn model is derived, implemented, and exercised far beyond what has been shown in previous work. New hydrocode simulation results demonstrate the role of stochastic fluctuations during shock initiation; these fluctuations are approximated by collections of discrete particles, that evolve with drift (i.e. deterministic) and diffusion (i.e. stochastic) coefficients. Furthermore, the particle values are propagated and averaged to calculate the heat release, yield strength, and material impedance in each computational cell. Hydrocode simulation results further show how the fluctuating hot spot energy may or may not be transmitted to the wave front, and result in a detonation wave-like structure. The fundamental stochastic nature of this model permits simulations to have varying outcomes with the same initial conditions; this allows for go/no-go estimation (e.g., marginal or failed detonations), which might possibly be calibrated using the statistical distributions from real materials. Mesoscale calculations of shocked heterogeneous explosives also show that these fluctuations are physically justified (i.e., Ref. [2]), and it is hypothesized that the pdf functions provide a link between the meso(grain) and continuum scales for practical engineering calculations. Hence, our approach is a paradigm shift for building efficient, reduced order continuum burn models, that represent the sub-grid stochastic behaviour of shocked heterogeneous solid explosives.

42 ENGINEERING↗
DOE OSTI2014

Stochastic Modeling of Overtime Occupancy and Its Application in Building Energy Simulation and Calibration

Overtime is a common phenomenon around the world. Overtime drives both internal heat gains from occupants, lighting and plug-loads, and HVAC operation during overtime periods. Overtime leads to longer occupancy hours and extended operation of building services systems beyond normal working hours, thus overtime impacts total building energy use. Current literature lacks methods to model overtime occupancy because overtime is stochastic in nature and varies by individual occupants and by time. To address this gap in the literature, this study aims to develop a new stochastic model based on the statistical analysis of measured overtime occupancy data from an office building. A binomial distribution is used to represent the total number of occupants working overtime, while an exponential distribution is used to represent the duration of overtime periods. The overtime model is used to generate overtime occupancy schedules as an input to the energy model of a second office building. The measured and simulated cooling energy use during the overtime period is compared in order to validate the overtime model. A hybrid approach to energy model calibration is proposed and tested, which combines ASHRAE Guideline 14 for the calibration of the energy model during normal working hours, and a proposed KS test for the calibration of the energy model during overtime. The developed stochastic overtime model and the hybrid calibration approach can be used in building energy simulations to improve the accuracy of results, and better understand the characteristics of overtime in office buildings.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗
DOE OSTI2020

A review on the integration of probabilistic solar forecasting in power systems

As one of the fastest growing renewable energy sources, the integration of solar power poses great challenges to power systems due to its variable and uncertain nature. As an effective approach to promote the integration of solar power in power systems, the value of probabilistic forecasts is being increasingly recognized in the recent decade. While the current use of probabilistic forecasts in power systems is limited, enormous amount of research has been conducted to promote the adoption of probabilistic forecasts and many methods have been proposed. This paper gives a comprehensive review on how probabilistic solar forecasts are utilized in power systems to address the challenges. Potential methods to deal with uncertainties in power systems are summarized, such as probabilistic load flow models, stochastic optimization, robust optimization, and chance constraints. Additionally, specific areas where these methods can be applied are discussed and state-of-the-art studies are summarized.

14 SOLAR ENERGY↗