Engineering Papers⌕ Search

SEARCH · Engineering Papers

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 991 records · Page 55

Benders Cut Classification via Support Vector Machines for Solving Two-Stage Stochastic Programs

In this work, we consider Benders decomposition for solving two-stage stochastic programs with complete recourse based on finite samples of the uncertain parameters. We define the Benders cuts binding at the final optimal solution or the ones significantly improving bounds over iterations as valuable cuts. We propose a learning-enhanced Benders decomposition (LearnBD) algorithm, which adds a cut classification step in each iteration to selectively generate cuts that are more likely to be valuable cuts. The LearnBD algorithm includes two phases: (i) sampling cuts and collecting information from training problems and (ii) solving testing problems with a support vector machine (SVM) cut classifier. We run the LearnBD algorithm on instances of capacitated facility location and multicommodity network design under uncertain demand. Our results show that SVM cut classifier works effectively for identifying valuable cuts, and the LearnBD algorithm reduces the total solving time of all instances for different problems with various sizes and complexities.

97 MATHEMATICS AND COMPUTING↗

DPM: A deep learning PDE augmentation method with application to large-eddy simulation

A framework is introduced that leverages known physics to reduce overfitting in machine learning for scientific applications. The partial differential equation (PDE) that expresses the physics is augmented with a neural network that uses available data to learn a description of the corresponding unknown or unrepresented physics. Training within this combined system corrects for missing, unknown, or erroneously represented physics, including discretization errors associated with the PDE's numerical solution. For optimization of the network within the PDE, an adjoint PDE is solved to provide high-dimensional gradients, and a stochastic adjoint method (SAM) further accelerates training. Additionally, the approach is demonstrated for large-eddy simulation (LES) of turbulence. High-fidelity direct numerical simulations (DNS) of decaying isotropic turbulence provide the training data used to learn sub-filter-scale closures for the filtered Navier–Stokes equations. Out-of-sample comparisons show that the deep learning PDE method outperforms widely-used models, even for filter sizes so large that they become qualitatively incorrect. It also significantly outperforms the same neural network when a priori trained based on simple data mismatch, not accounting for the full PDE. Measures of discretization errors, which are well-known to be consequential in LES, point to the importance of the unified training formulation's design, which without modification corrects for them. For comparable accuracy, simulation runtime is significantly reduced. A relaxation of the typical discrete enforcement of the divergence-free constraint in the solver is also successful, instead allowing the DPM to approximately enforce incompressibility physics. Since the training loss function is not restricted to correspond directly to the closure to be learned, training can incorporate diverse data, including experimental data.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Bayesian population analysis of a washin-washout physiologically based pharmacokinetic model for acetone

The aim of this study was to derive improved estimates of population variability and uncertainty of physiologically based pharmacokinetic (PBPK) model parameters, especially of those related to the washin-washout behavior of polar volatile substances. This was done by optimizing a previously published washin-washout PBPK model for acetone in a Bayesian framework using Markov chain Monte Carlo simulation. The sensitivity of the model parameters was investigated by creating four different prior sets, where the uncertainty surrounding the population variability of the physiological model parameters was given values corresponding to coefficients of variation of 1%, 25%, 50%, and 100%, respectively. The PBPK model was calibrated to toxicokinetic data from 2 previous studies where 18 volunteers were exposed to 250-550 ppm of acetone at various levels of workload. The updated PBPK model provided a good description of the concentrations in arterial, venous, and exhaled air. The precision of most of the model parameter estimates was improved. New information was particularly gained on the population distribution of the parameters governing the washin-washout effect. The results presented herein provide a good starting point to estimate the target dose of acetone in the working and general populations for risk assessment purposes.

60 APPLIED LIFE SCIENCES↗

Integrated Strategies for Overcoming Resolution Limits in Electron Beam Lithography of Chemically Amplified Resists

Electron beam lithography (EBL) of chemically amplified resists (CARs) faces fundamental challenges, including stochastic electron scattering and acid diffusion, that limit resolution and reproducibility. Using SU-8 as a model CAR, this study systematically investigated complementary strategies to address these challenges, combining multipass exposure, proximity effect correction (PEC) with midrange correction factors, base quencher incorporation, and post-exposure bake (PEB) suppression. Monte Carlo simulations and calibrated PEC modeling revealed that extending the point spread function to include a midrange scattering component significantly improved critical dimension (CD) control across varying pattern densities, correcting deviations that conventional two-term PEC failed to capture. Multipass exposure, particularly 4-pass writing with a 25% offset, redistributed the dose to average stochastic beam and scattering fluctuations, reducing line-width roughness by more than 50% and yielding more uniform nanoscale features. Photoacid confinement was investigated by adding urea as a base quencher, which successfully reduced acid diffusion but introduced substantial sensitivity penalties without improving ultimate resolution or Z-factor performance, underscoring the trade-offs of chemical versus physical confinement. Suppressing PEB most directly minimized acid diffusion, resulting in improved Z-factors and reproducible 30 nm half-pitch dense line/space patterns. Overall, these results demonstrated that PEC with midrange correction, multipass strategies, quencher additives, and PEB-free processing addresses different aspects of the EBL process window and that their integration provides a comprehensive framework for managing stochastic scattering, diffusion, and chemical amplification effects. This framework advances dense nanoscale patterning in CARs and establishes guiding principles for optimizing resist design and process strategies in high-resolution EBL and potentially other advanced lithographies, such as extreme ultraviolet (EUV) lithography.

36 MATERIALS SCIENCE↗

Active sampling for neural network potentials: Accelerated simulations of shear-induced deformation in Cu–Ni multilayers

Neural network potentials (NNPs) can greatly accelerate atomistic simulations relative to ab initio methods, allowing one to sample a broader range of structural outcomes and transformation pathways. In this work, we demonstrate an active sampling algorithm that trains an NNP that is able to produce microstructural evolutions with accuracy comparable to those obtained by density functional theory, exemplified during structure optimizations for a model Cu–Ni multilayer system. We then use the NNP, in conjunction with a perturbation scheme, to stochastically sample structural and energetic changes caused by shear-induced deformation, demonstrating the range of possible intermixing and vacancy migration pathways that can be obtained as a result of the speedups provided by the NNP. The code to implement our active learning strategy and NNP-driven stochastic shear simulations is openly available at https://github.com/pnnl/Active-Sampling-for-Atomistic-Potentials .

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Optimizing Service Restoration in Distribution Systems With Uncertain Repair Time and Demand

This paper proposes a novel method to co-optimize the distribution system operation and repair crew routing for outage restoration after extreme weather events. A two-stage stochastic mixed integer linear program is developed. The first stage is to dispatch the repair crews to the damaged components. The second stage is distribution system restoration using distributed generators, and reconfiguration. We consider demand uncertainty in terms of a truncated normal forecast error distribution, and model the uncertainty of the repair time using a lognormal distribution. A new decomposition approach, combined with the progressive hedging algorithm, is developed for solving large-scale outage management problems in an effective and timely manner. Furthermore, the proposed method is validated on modified IEEE 34- and 8500-bus distribution test systems.

24 POWER TRANSMISSION AND DISTRIBUTION↗

Manifold Learning-Based Polynomial Chaos Expansions for High-Dimensional Surrogate Models

In this work we introduce a manifold learning-based method for uncertainty quantification (UQ) in systems describing complex spatiotemporal processes. Our first objective is to identify the embedding of a set of high-dimensional data representing quantities of interest of the computational or analytical model. For this purpose, we employ Grassmannian diffusion maps, a two-step nonlinear dimension reduction technique which allows us to reduce the dimensionality of the data and identify meaningful geometric descriptions in a parsimonious and inexpensive manner. Polynomial chaos expansion is then used to construct a mapping between the stochastic input parameters and the diffusion coordinates of the reduced space. An adaptive clustering technique is proposed to identify an optimal number of clusters of points in the latent space. The similarity of points allows us to construct a number of geometric harmonic emulators which are finally utilized as a set of inexpensive pretrained models to perform an inverse map of realizations of latent features to the ambient space and thus perform accurate out-of-sample predictions. Thus, the proposed method acts as an encoder-decoder system which is able to automatically handle very high-dimensional data while simultaneously operating successfully in the small-data regime. The method is demonstrated on two benchmark problems and on a system of advection-diffusion-reaction equations which model a first-order chemical reaction between two species. In all test cases, the proposed method is able to achieve highly accurate approximations which ultimately lead to the significant acceleration of UQ tasks.

42 ENGINEERING↗

Robust Scheduling of Networked Microgrids for Economics and Resilience Improvement

The benefits of networked microgrids in terms of economics and resilience are investigated and validated in this work. Considering the stochastic unintentional islanding conditions and conventional forecast errors of both renewable generation and loads, a two-stage adaptive robust optimization is proposed to minimize the total operating cost of networked microgrids in the worst scenario of the modeled uncertainties. By coordinating the dispatch of distributed energy resources (DERs) and responsive demand among networked microgrids, the total operating cost is minimized, which includes the start-up and shut-down cost of distributed generators (DGs), the operation and maintenance (O&M) cost of DGs, the cost of buying/selling power from/to the utility grid, the degradation cost of energy storage systems (ESSs), and the cost associated with load shedding. The proposed optimization is solved with the column and constraint generation (C&CG) algorithm. The results of case studies demonstrate the advantages of networked microgrids over independent microgrids in terms of reducing total operating cost and improving the resilience of power supply.

29 ENERGY PLANNING, POLICY, AND ECONOMY↗

Rapid Optimization of Total Variation with Applications in Imaging, Additive Manufacturing, and Qualification

Total Variation optimization penalizes the gradient of a control variable or state. While this work focuses on image processing in particular, it has also found applications in inverse problems and topology optimization. In image processing, the goal is to maintain faithfulness to the original image while denoising and/or deblurring. Additionally, bilevel optimization over the spatially varying regularization weights can illuminate interfaces such as damage regions and other anomalies. We will address two fundamental challenges with TV-optimization: (i) the typical slow convergence of existing TV-optimization methods, and (ii) the selection of spatially varying TV parameters to promote interface detection. Additionally, we will apply such techniques to image data collected in additive manufacturing. In said context, stochasticity in build events induces flaws in the manufactured piece, compromising the integrity of said part. There is a critical need for in-situ monitoring to spot anomalies once they form, and in this setting we apply our total variation and hyperparameter solvers. We will develop a customized algorithm based on for extreme-scale TV-optimization that achieves super-linear or quadratic-convergence, a critical property for real-time, image-by-image analysis. A worst-case outcome is a preprocessing step that enhances image quality in-situ, specifically for out-of-focus and noisy images.

36 MATERIALS SCIENCE↗

Three-Phase Solid Oxide Fuel Cell Anode Microstructure Realization Using Two-Point Correlation Functions

A Monte Carlo methodology is developed as a means for three-dimensional (3D) reconstruction of the microstructure of a three-phase anode used in solid oxide fuel cells, based on two-point statistical functions. The salient feature of the presented reconstruction methodology is the ability to realize the 3D microstructure from its 2D SEM image for a three-phase medium extendable to n-phase media. In the realization procedure, different phases of the heterogeneous medium are represented by different cells which are allowed to grow. The growth of cells, however, are controlled via several optimization parameters related to rotation, shrinkage, translation, distribution and growth rates of the cells. Indeed, the proposed realization algorithm can be categorized as a member of dynamic programming methods and is designed so comprehensive that can realize any desired microstructure. To be more specific, at first the initial 2D image is successfully reconstructed and then the final optimization parameters are used as the initial values for the initiation of the 3D reconstruction algorithm. This paper presents a novel hybrid stochastic methodology based on the colony and kinetic algorithm for the simulation of the virtual microstructure. The simulation procedure involves repeated realizations where each realization in turn consists of nucleation and growth of cells. For each of the subsequent realizations, the controlling parameters get updated by minimization of an objective function at the end of the preceding realization. Here, the objective function is defined based on the two-point correlation functions from the simulated and real microstructures. The kinetic growth algorithm is established on the cellular automata approach which facilitates the simulation procedure. Comparison of the two-point correlation functions from different sections of the final 3D reconstructed microstructure with the initial real microstructure shows a satisfactory agreement which confirms the proposed methodology.

12 MANAGEMENT OF RADIOACTIVE AND NON-RADIOACTIVE W↗

Shape optimization under uncertainty for rotor blades of horizontal axis wind turbines

We present a computational framework for the shape optimization of a Horizontal-Axis Wind Turbine (HAWT) rotor blade under uncertainty. Our framework integrates aerodynamic simulations based on the blade element method which utilizes reduced order models of the blade structure and wind load with design sensitivity analysis and nonlinear programming. The wind velocity is modeled as a stochastic process to account for variations in time and space. An additional stochastic process accounts for uncertainties in the structural material properties. The uncertainty propagation is based on a non-intrusive polynomial chaos expansion that allows accurate estimation of stochastic performance metrics such as generated power and structural compliance. Sensitivities of cost and constraint functions with respect to shape parameters namely twist angles are computed with an efficient scheme to enable gradient based optimization. To demonstrate the effect of uncertainty, designs obtained from optimization under uncertainty are compared to those obtained from deterministic optimization.

42 ENGINEERING↗

Battery Energy Storage Scheduling for Optimal Load Variance Minimization

Generation portfolio can be significantly altered due to the deployment of distributed energy resources (DER) in distribution networks and the concept of microgrid. Generally,distribution networks can operate in a more resilient and economic fashion through proper coordination of DER. However, due to the partially uncontrollable and stochastic nature of some DER, the variance of net load of distribution systems increases, which raises the operational cost and complicates operation for transmission companies. This motivates peak shaving and valley filling using energy storage units deployed in distribution systems. This paper aims at theoretical formulation of optimal load variance minimization, where the infinity norm of net load is minimized. Then, the problem is reformulated equivalently as a linear program. A case study is performed with capacity limited battery energy storage model and the simplified power flow model of a radial distribution network. The influence of capacity limit and deployment location are studied.

Melin, Alexander↗

Stochastic quasi-steady state approximations for asymptotic solutions of the chemical master equation

In this paper, we propose two methods to carry out the quasi-steady state approximation in stochastic models of enzyme catalytic regulation, based on WKB asymptotics of the chemical master equation or of the corresponding partial differential equation for the generating function. The first of the methods we propose involves the development of multiscale generalisation of a WKB approximation of the solution of the master equation, where the separation of time scales is made explicit which allows us to apply the quasi-steady state approximation in a straightforward manner. To the lowest order, the multi-scale WKB method provides a quasi-steady state, Gaussian approximation of the probability distribution. The second method is based on the Hamilton-Jacobi representation of the stochastic process where, as predicted by large deviation theory, the solution of the partial differential equation for the corresponding characteristic function is given in terms of an effective action functional. The optimal transition paths between two states are then given by those paths that maximise the effective action. Such paths are the solutions of the Hamilton equations for the Hamiltonian associated to the effective action functional. The quasi-steady state approximation is applied to the Hamilton equations thus providing an approximation to the optimal transition paths and the transition time between two states. Using this approximation we predict that, unlike the mean-field quasi-steady approximation result, the rate of enzyme catalysis depends explicitly on the initial number of enzyme molecules. The accuracy and validity of our approximated results as well as that of our predictions regarding the behaviour of the stochastic enzyme catalytic models are verified by direct simulation of the stochastic model using Gillespie stochastic simulation algorithm.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Topology optimization for the design of porous electrodes

Porous electrodes are an integral part of many electrochemical devices since they have high porosity to maximize electrochemical transport and high surface area to maximize activity. Traditional porous electrode materials are typically homogeneous, stochastic collections of small-scale particles and offer few opportunities to engineer higher performance. Fortunately, recent breakthroughs in advanced and additive manufacturing are yielding new methods to structure and pattern porous electrodes across length scales. These architected electrodes are emerging as a promising new technology to continue to drive improvement; however, it is still unclear which structures to employ and few tools are available to guide their design. Here we address this gap by applying topology optimization to the design of porous electrodes. We demonstrate our framework on two applications: a porous electrode driving a steady Faradaic reaction and a transiently operated electrode in a supercapacitor. We present computationally designed electrodes that minimize energy losses in a half-cell. For low-conductivity materials, the optimization algorithm creates electrode designs with a hierarchy of length scales. Further, the designed electrodes are found to outperform undesigned, homogeneous electrodes. Finally, we present three-dimensional porous electrode designs. We thus establish a topology optimization framework for designing porous electrodes.

25 ENERGY STORAGE↗

Operation Optimization using Reinforcement Learning with Integrated Artificial Reasoning Framework

In large and complex systems, operational decision-making requires a systematic analysis with a vast amount of data from both process parameters and component status monitoring. In this paper, we present an integrated artificial reasoning approach for system state transition models that can help operational decision-making with explainable and traceable reasoning. The integrated artificial reasoning framework is a physics-based approach of defining the system structure in a Bayesian network, so we leveraged it in a Markov decision process (MDP) for finding optimal operational solutions. In our proposed framework, the MDP is implemented on a dynamic Bayesian network (DBN), which represents causalities in a system. The multilevel flow modeling was utilized in order to extract these causalities in a more efficient and objective manner. Since multilevel flow modeling is based on the fundamental energy and mass conservation laws, the target system is decomposed into several mass, energy, and information structures, which serve as the basis for a DBN. The MDP consists of the processes of finding a solution for the Bellman equation, which can be derived from the conditional probability equations of the constructed DBN. System operators can capture stochastic system dynamics as multiple subsystem state transitions based on their physical relations and uncertainties coming from the component degradation process or random failures. We analyzed a simplified example system to illustrate finding an optimal operational policy with this approach.

99 GENERAL AND MISCELLANEOUS↗

Evaluating Gaussian process metamodels and sequential designs for noisy level set estimation

Abstract We consider the problem of learning the level set for which a noisy black-box function exceeds a given threshold. To efficiently reconstruct the level set, we investigate Gaussian process (GP) metamodels. Our focus is on strongly stochastic simulators, in particular with heavy-tailed simulation noise and low signal-to-noise ratio. To guard against noise misspecification, we assess the performance of three variants: (i) GPs with Student- t observations; (ii) Student- t processes (TPs); and (iii) classification GPs modeling the sign of the response. In conjunction with these metamodels, we analyze several acquisition functions for guiding the sequential experimental designs, extending existing stepwise uncertainty reduction criteria to the stochastic contour-finding context. This also motivates our development of (approximate) updating formulas to efficiently compute such acquisition functions. Our schemes are benchmarked by using a variety of synthetic experiments in 1–6 dimensions. We also consider an application of level set estimation for determining the optimal exercise policy of Bermudan options in finance.

97 MATHEMATICS AND COMPUTING↗

pypolymix

SAND2026-16700O Pypolymix builds stochastic surrogate models using a lightweight Python library. Researchers working with large-scale scientific models can use it to enhance optimization and uncertainty quantification studies. Pypolymix enables the efficient construction of surrogate models and improves analysis and performance of complex simulations. Sandia National Laboratories is a multimission laboratory managed and operated by National Technology & Engineering Solutions of Sandia, LLC, a wholly owned subsidiary of Honeywell International Inc., for the U.S. Department of Energy’s National Nuclear Security Administration under contract DE-NA0003525.

SciDAC↗

Learning Forecasts of Rare Stratospheric Transitions from Short Simulations

Abstract Rare events arising in nonlinear atmospheric dynamics remain hard to predict and attribute. We address the problem of forecasting rare events in a prototypical example, sudden stratospheric warmings (SSWs). Approximately once every other winter, the boreal stratospheric polar vortex rapidly breaks down, shifting midlatitude surface weather patterns for months. We focus on two key quantities of interest: the probability of an SSW occurring, and the expected lead time if it does occur, as functions of initial condition. These optimal forecasts concretely measure the event’s progress. Direct numerical simulation can estimate them in principle but is prohibitively expensive in practice: each rare event requires a long integration to observe, and the cost of each integration grows with model complexity. We describe an alternative approach using integrations that are short compared to the time scale of the warming event. We compute the probability and lead time efficiently by solving equations involving the transition operator, which encodes all information about the dynamics. We relate these optimal forecasts to a small number of interpretable physical variables, suggesting optimal measurements for forecasting. We illustrate the methodology on a prototype SSW model developed by Holton and Mass and modified by stochastic forcing. While highly idealized, this model captures the essential nonlinear dynamics of SSWs and exhibits the key forecasting challenge: the dramatic separation in time scales between a single event and the return time between successive events. Our methodology is designed to fully exploit high-dimensional data from models and observations, and has the potential to identify detailed predictors of many complex rare events in meteorology.

Finkel, Justin↗