Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “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 1,585 records · Page 88

rlmolecule: A library for general-purpose material and molecular optimization using AlphaZero-style reinforcement learning

This library includes in-progress code for the optimization of materials and molecules using reinforcement learning. This project seeks to develop a generalized machine learning approach for optimizing targeted, and often complex, functional properties over the space of possible material structures, allowing faster identification of promising candidates. Electronic and transport properties of materials are determined by their molecular structure (for molecules) or their crystal structure (for in-organic crystalline materials). The atomic composition, bonding, and 3-D-spatial arrangement in these systems therefore represent a vast, discontinuous search space for optimal candidates. This library implements AlphaZero style reinforcement learning applied to two worked examples: the optimization of stable organic radicals for redox flow batteries, and the determination of stable electrolyte species for solid state batteries. The current code contains proof-of-concept results run on the Eagle HPC system, while the completed package will be a flexible tool for material optimization across of number of energy-relevant fields, which leverages cloud compute resources for the majority of the computational heavy lifting.

Biagioni, David↗

DERMS-RT (Distributed Energy Resource Management Solution using Real-Time Optimization) [SWR-20-46]

DERMS-RT provides a distributed energy resource management solution using real-time optimization to control different types of distributed energy resources (DERs) — such as rooftop PVs, battery energy storage systems, HVAC loads, electric water heater loads, and EV charging loads, and use these DERs to provide distribution voltage regulation and peak demand management services. The control approach implemented by DERM-RT is developed based on the real-time optimal power flow and distributed control previously developed by NREL researchers. The software codes are designed to provide modular DER management solution that can be easily applied to control heterogeneous DERs in distribution grids in coordination with centralized utility management systems. Also, DERMS-RT provides application programming interface (API) to extract useful grid model information from an open-source power system simulation software, use the information to solve the optimization problem, and apply the optimal set-points to the controlled DERs. In this way, DERMS-RT provides a plug-and-play function for implementing the real-time DER control on different utility system models, and it can be used as a simulation testbed to demonstrate the impact of real-time DER optimization on improving grid operations.

Ding, Fei↗

Tools for analysis of optimization models

The software is code for analyzing, debugging, and solving optimization models. The code implements several algorithms and provides convenient APIs to apply these algorithms to optimization models. The algorithms are primarily based on graph theory. They compute well-known partitions of graphs, and use these partitions to provide the user information about their optimization model, including diagnosing certain types of structural modeling errors. Some algorithms may be used as a subroutine to solve an optimization problem, and may call an optimization solver to facilitate this. Well-known linear algebra routines are called to provide further diagnostics. The software contains no data and no models other than toy models used for testing.

Parker, Robert↗

BEopt™ (Building Energy Optimization Tool) version 3.0.1 3/16/2023 [SWR-05-41]

The BEopt™ (Building Energy Optimization Tool) software provides capabilities to evaluate residential building designs and identify cost-optimal efficiency packages at various levels of whole-house energy savings along the path to zero net energy. The BEopt software can be used to analyze both new construction and existing home retrofits, as well as single-family detached and multi-family buildings, through evaluation of single building designs, parametric sweeps, and cost-based optimizations. The BEopt software provides detailed simulation-based analysis based on specific house characteristics, such as size, architecture, occupancy, vintage, location, and utility rates. Discrete envelope and equipment options, reflecting realistic construction materials and practices, are evaluated. The BEopt software uses EnergyPlus®, the Department of Energy's flagship simulation engine. Simulation assumptions are based on ANSI/RESNET/ICC Standards and the Building America Housing Simulation Protocols. The sequential search optimization technique used by the BEopt software: Finds minimum-cost building designs at different target energy-savings levels Identifies multiple near-optimal designs along the path, allowing for equivalent solutions based on builder or contractor preference. Download A BEopt™ Version 3.0.1 Beta release (Mar. 16, 2023) is now available. Download the BEopt 3.0.1 Beta setup file.

Christensen, Craig↗

Dispatch Optimization Variable Engine

The Dispatch Optimization Variable Engine (DOVE) is software tool written in python, developed at Idaho National Laboratory (INL) that provides an easily accessible application-programming-interface (API) to performing resource dispatch optimization analysis for integrated energy system (IES) configurations. DOVE is an integral part of the Framework for Optimization of Resources and Economics (FORCE) software suite and is leveraged by codes such as the Holistic Energy Resource Optimization Network (HERON) and the Optimization of Real-Time Capacity Allocation (ORCA). The philosophy behind DOVE is to provide a modular software solution to IES planning and operation by utilizing state-of-the-art algorithms and machine learning. The goal is to accurately capture the dispatching behavior of a complex energy system given varying time-dependent signals for demand and commodity pricing.

McDowell, DylanJ. [Idaho National Laboratory (INL)↗

arco (Assembled Resource-Constrained Optimization) [SWR-26-030]

Arco (Assembled Resource-Constrained Optimization) is a memory-smart optimization DSL and solver for LP and MIP problems on constrained hardware. The software is an optimization framework built around a KDL-based domain-specific language and a CLI compiler/solver. You write optimization models in .kdl files, and the arco CLI compiles, validates, inspects, and solves them. Language bindings (Python today, more planned) provide programmatic access to the same engine. Built for harder optimization problems on constrained resources, Arco is intentional about every allocation, careful with stack and heap behavior, and relentless about minimizing memory usage so more systems can run real workloads. Arco is built primarily for internal use within our organization. You are welcome to try it, but we make no guarantees about API stability or robustness at this stage

Sanchez Perez, Pedro Andres [National Laboratory o↗

Optimization Layers for Pyomo [SWR-25-132]

Optimization Layers for Pyomo solves an optimization problem using Pyomo and IPOPT during the forward pass. It computes the gradient of the optimal solution with respect to the parameters based on the KKT conditions in the backward pass. It is a Python library for constructing differentiable optimization layers in PyTorch from Pyomo optimization models.

Chen, Kejun [National Laboratory of the Rockies (N↗

Machine Learning Approach for Spatiotemporal Multivariate Optimization of Environmental Monitoring Sensor Locations

Abstract Long-term environmental monitoring is critical for managing the soil and groundwater at contaminated sites. Recent improvements in state-of-the-art sensor technology, communication networks, and artificial intelligence have created opportunities to modernize this monitoring activity for automated, fast, robust, and predictive monitoring. In such modernization, it is required that sensor locations be optimized to capture the spatiotemporal dynamics of all monitoring variables as well as to make it cost-effective. The legacy monitoring datasets of the target area are important to perform this optimization. In this study, we have developed a machine-learning approach to optimize sensor locations for soil and groundwater monitoring based on ensemble supervised learning and majority voting. For spatial optimization, Gaussian process regression (GPR) is used for spatial interpolation, while the majority voting is applied to accommodate the multivariate temporal dimension. Results show that the algorithms significantly outperform the random selection of the sensor locations for predictive spatiotemporal interpolation. While the method has been applied to a four-dimensional dataset (with two-dimensional space, time, and multiple contaminants), we anticipate that it can be generalizable to higher-dimensional datasets for environmental monitoring sensor location optimization.

Siddiquee, Masudur R.↗

Optimization of thermal barrier coating performance and durability over a drive cycle

A methodology to thermo-mechanically optimize a piston thermal barrier coating over a full drive cycle was established. The optimization objective was to minimize the heat transfer to the engine wall while maintaining structural integrity of the coating. Over 800 candidate materials were investigated and the optimization required more than one million non-road transient drive cycle calculations; real materials were investigated to ensure a realizable result and the existence of thermal and mechanical properties. High computational efficiency was achieved using a recently developed analytical heat transfer technique for multilayer engine walls. An uncoupled approach was utilized for the optimization, wherein the gas temperature and heat transfer coefficient profiles from a fully coupled and calibrated baseline model over the 20-min drive cycle were employed. The coating/piston interface temperature was constrained to be below the maximum piston service temperature limit. The durability was assessed using a recently developed analytical coating delamination framework for engine in-cylinder coatings based on the energy release rate when a crack forms. Results are presented for a mechanically unconstrained optimization and for cases constrained to three fixed levels of drive-cycle maximum energy release rate, and also constrained by the individual material’s toughness. The best-performing coating materials identified were verified using the fully coupled system-level model, which compared well to the uncoupled predictions. A study on the effect of adding a sealing layer to some high-performing, but porous, coatings showed a reduction in fuel consumption benefit and an increased exhaust temperature over the cycle, but the system still outperformed the uncoated case. The results of the study elucidate the importance of including engine performance and mechanical failure considerations in thermal barrier coating design.

Engineering↗

Multiobjective and categorical global optimization of photonic structures based on ResNet generative neural networks

Abstract We show that deep generative neural networks, based on global optimization networks (GLOnets), can be configured to perform the multiobjective and categorical global optimization of photonic devices. A residual network scheme enables GLOnets to evolve from a deep architecture, which is required to properly search the full design space early in the optimization process, to a shallow network that generates a narrow distribution of globally optimal devices. As a proof-of-concept demonstration, we adapt our method to design thin-film stacks consisting of multiple material types. Benchmarks with known globally optimized antireflection structures indicate that GLOnets can find the global optimum with orders of magnitude faster speeds compared to conventional algorithms. We also demonstrate the utility of our method in complex design tasks with its application to incandescent light filters. These results indicate that advanced concepts in deep learning can push the capabilities of inverse design algorithms for photonics.

Jiang, Jiaqi↗

Hyper-Differential Sensitivity Analysis of Uncertain Parameters in PDE-Constrained Optimization

Many problems in engineering and sciences require the solution of large scale optimization constrained by partial differential equations (PDEs). Though PDE-constrained optimization is itself challenging, most applications pose additional complexity, namely, uncertain parameters in the PDEs. Uncertainty quantification (UQ) is necessary to characterize, prioritize, and study the influence of these uncertain parameters. Sensitivity analysis, a classical tool in UQ, is frequently used to study the sensitivity of a model to uncertain parameters. In this article, we introduce "hyper-differential sensitivity analysis" which considers the sensitivity of the solution of a PDE-constrained optimization problem to uncertain parameters. Our approach is a goal-oriented analysis which may be viewed as a tool to complement other UQ methods in the service of decision making and robust design. We formally define hyper-differential sensitivity indices and highlight their relationship to the existing optimization and sensitivity analysis literatures. Assuming the presence of low rank structure in the parameter space, computational efficiency is achieved by leveraging a generalized singular value decomposition in conjunction with a randomized solver which converts the computational bottleneck of the algorithm into an embarrassingly parallel loop. Two multi-physics examples, consisting of nonlinear steady state control and transient linear inversion, demonstrate efficient identification of the uncertain parameters which have the greatest influence on the optimal solution.

97 MATHEMATICS AND COMPUTING↗

Xopt and Badger: a machine learning ecosystem for real-time accelerator control and optimization

Machine learning (ML)-based black-box optimization algorithms have demonstrated significant improvements in accelerator optimization speed, often by orders of magnitude. However, deploying these algorithms in real-time facility control remains challenging due to the specialized expertise and infrastructure required. To bridge this gap, we introduce the Xopt ecosystem, a versatile suite of tools designed to make advanced ML-based optimization accessible to the broader accelerator community. This ecosystem includes Xopt, a modular Python framework that facilitates the integration of ML-based optimization algorithms with arbitrary control problems, and Badger, a graphical user interface built on top of Xopt, which enables seamless deployment of ML algorithms in real-time control systems. The Xopt ecosystem has been successfully applied towards solving challenging real-time control problems at leading international accelerator facilities, including SLAC, LBNL, Argonne, Fermilab, BNL, DESY, and ESRF, demonstrating its effectiveness in real-world optimization tasks. In this presentation, we provide an overview of Xopt’s capabilities and illustrate its impact through case studies from SLAC accelerator facilities including LCLS, LCLS-II, and FACET-II.

Roussel, Ryan [SLAC]↗

Bayesian optimization scheme for the design of a nanofibrous high power target

High Power Targetry (HPT) R&D is critical in the context of increasing beam intensity and energy for next generation accelerators. Many target concepts and novel materials are being developed and tested for their ability to withstand extreme beam environments; the HPT R&D Group at Fermilab is developing an electrospun nanofiber material for this purpose. The performance of these nanofiber targets is sensitive to their construction parameters, such as the packing density of the fibers. Lowering the density improves the survival of the target, but reduces the secondary particle yield. Optimizing the lifetime and production efficiency of the target poses an interesting design problem, and in this paper we study the applicability of Bayesian optimization to its solution. We first describe how to encode the nanofiber target design problem as the optimization of an objective function, and how to evaluate that function with computer simulations. We then explain the optimization loop setup. Thereafter, we present the optimal design parameters suggested by the algorithm, and close with discussions of limitations and future refinements.

43 PARTICLE ACCELERATORS↗

Large-scale Stochastic Optimization for Unit Commitment and Economic Dispatch

Our FY 2012 LDRD research on stochastic optimization with its application in power grid planning and operations has advanced the state-of-the-art of this area in the following aspects: (1) Deploying stochastic unit commitment on large real-world applications, (2) Stochastic optimization coupled with real-time production cost simulation (interleaved), and (3) Real-time stochastic optimization with rolling-horizon look-ahead. In conjunction, a number of computational experiments have been performed to evaluate the capability of the latest stochastic optimization software and explore opportunities for future research. In what follows, we discuss each of these contribution areas, including the results of our computational experiments and their implications on the scalability and solvability of large-scale stochastic optimization models.

24 POWER TRANSMISSION AND DISTRIBUTION↗

Incorporating the Min-Max Mesh Optimization Method within the Target-Matrix Paradigm

The Target-Matrix Optimization Paradigm (TMOP) is a method for improving mesh quality via node movement. Through the use of appropriate pre-defined quality metrics and Target-matrices, TMOP provides the ability to define mesh quality in terms of element shape, volume, and edge orientation as a function of position within the mesh, thus facilitating high-level goals such as size-adapted meshing. The Paradigm is developed further in this paper by showing how it can be applied to the problem of improving the mesh in locations where the quality is the worst. Improving the worst quality portions of a mesh was investigated by Freitag, Amenta, and others, who showed that the problem can be formulated as a min-max optimization problem. The Target-paradigm is used to extend the previous work in two important respects. First, by using quality metrics from the Target-paradigm, quality can be user-specified via Target-matrices instead of the quality definition being hard-wired into the quality metric. Second, by making use of the sample point concept in TMOP, one can extend min-max mesh optimization to non-simplicial element types such as quadrilaterals. Numerical experiments were performed to illustrate these extensions and to investigate their utility. A non-gradient solver known as the amoeba was used to numerically compute the optimal solutions. This derivative-free method, in conjunction with the use of barrier metrics from TMOP, enables the solution of the min-max problem without the use of explicit constraints to guarantee that the optimal mesh is non-inverted. The experiments showed that the use of Target-metrics in min-max with triangle (or tetrahedral) element meshes is able to effectively improve the worst quality elements in terms of shape or in terms of size+shape. The experiments also showed that the mesh min-max extension to quadrilateral element meshes can lead to acceptable meshes using barrier metrics from the Target-paradigm, provided that inter-element mesh smoothness is not important. Poor behavior was observed when using non-barrier Target-metrics in min-max with quadrilateral element meshes. For both simplicial and quadrilateral meshes, the min-sum method can be competitive with min-max, however.

97 MATHEMATICS AND COMPUTING↗

Large-Scale Stochastic Optimization for Unit Commitment and Economic Dispatch

Our FY 2012 LDRD research on stochastic optimization with its application in power grid planning and operations has advanced the state-of-the-art of this area in the following aspects: (1) Deploying stochastic unit commitment on large real-world applications, (2) Stochastic optimization coupled with real-time production costsimulation (interleaved), and (3) Real-time stochastic optimization with rolling-horizon look-ahead. In conjunction, a number of computational experiments have been performed to evaluate the capability of the latest stochastic optimization software and explore opportunities for future research. In what follows, we discuss each of these contribution areas, including the results of our computational experiments and their implications on the scalability and solvability of large-scale stochastic optimization models.

29 ENERGY PLANNING, POLICY, AND ECONOMY↗

Optimization and Annual Average Power Predictions of a Backward Bent Duct Buoy Oscillating Water Column Device Using the Wells Turbine.

This Technical Report presents work completed by The Applied Research Laboratory at The Pennsylvania State University, in conjunction with Sandia National Labs, on the optimization of the power conversion chain (PCC) design to maximize the Average Annual Electric Power (AAEP) output of an Oscillating Water Column (OWC) device. The design consists of two independent stages. First, the design of a floating OWC, a Backward Bent Duct Buoy (BBDB), and second the design of the PCC. The pneumatic power output of the BBDB in random waves is optimized through the use of a hydrodynamically coupled, linear, frequency-domain, performance model that links the oscillating structure to internal air-pressure fluctuations. The PCC optimization is centered on the selection and sizing of a Wells Turbine and electric power generation equipment. The optimization of the PCC involves the following variables: the type of Wells Turbine (fixed or variable pitched, with and without guide vanes), the radius of the turbine, the optimal vent pressure, the sizing of the power electronics, and number of turbines. Also included in this Technical Report are further details on how rotor thrust and torque are estimated, along with further details on the type of variable frequency drive selected.

13 HYDRO ENERGY↗

Next-generation Algorithms for Assessing Infrastructure Vulnerability and Optimizing System Resilience

This report summarizes the work performed under the project project Next-Generation Algorithms for Assessing Infrastructure Vulnerability and Optimizing System Resilience. The goal of the project was to improve mathematical programming-based optimization technology for infrastructure protection. In general, the owner of a network wishes to design a network a network that can perform well when certain transportation channels are inhibited (e.g. destroyed) by an adversary. These are typically bi-level problems where the owner designs a system, an adversary optimally attacks it, and then the owner can recover by optimally using the remaining network. This project funded three years of Deon Burchett's graduate research. Deon's graduate advisor, Professor Jean-Philippe Richard, and his Sandia advisors, Richard Chen and Cynthia Phillips, supported Deon on other funds or volunteer time. This report is, therefore. essentially a replication of the Ph.D. dissertation it funded in a format required for project documentation. The thesis had some general polyhedral research. This is the study of the structure of the feasible region of mathematical programs, such as integer programs. For example, an integer program optimizes a linear objective function subject to linear constraints, and (nonlinear) integrality constraints on the variables. The feasible region without the integrality constraints is a convex polygon. Careful study of additional valid constraints can significantly improve computational performance. Here is the abstract from the dissertation: We perform a polyhedral study of a multi-commodity generalization of variable upper bound flow models. In particular, we establish some relations between facets of single- and multi- commodity models. We then introduce a new family of inequalities, which generalizes traditional flow cover inequalities to the multi-commodity context. We present encouraging numerical results. We also consider the directed edge-failure resilient network design problem (DRNDP). This problem entails the design of a directed multi-commodity flow network that is capable of fulfilling a specified percentage of demands in the event that any G arcs are destroyed, where G is a constant parameter. We present a formulation of DRNDP and solve it in a branch-column-cut framework. We present computational results.

97 MATHEMATICS AND COMPUTING↗