Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “two-stage 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 145 records · Page 8
All sourcesNASA NTRSDOE OSTI
DOE OSTI2022

Toward fully automated UED operation using two-stage machine learning model

To demonstrate the feasibility of automating UED operation and diagnosing the machine performance in real time, a two-stage machine learning (ML) model based on self-consistent start-to-end simulations has been implemented. This model will not only provide the machine parameters with adequate precision, toward the full automation of the UED instrument, but also make real-time electron beam information available as single-shot nondestructive diagnostics. Furthermore, based on a deep understanding of the root connection between the electron beam properties and the features of Bragg-diffraction patterns, we have applied the hidden symmetry as model constraints, successfully improving the accuracy of energy spread prediction by a factor of five and making the beam divergence prediction two times faster. The capability enabled by the global optimization via ML provides us with better opportunities for discoveries using near-parallel, bright, and ultrafast electron beams for single-shot imaging. It also enables directly visualizing the dynamics of defects and nanostructured materials, which is impossible using present electron-beam technologies.

36 MATERIALS SCIENCE↗
DOE OSTI2025

Neural architecture codesign for fast physics applications

We develop a pipeline to streamline neural architecture codesign for physics applications to reduce the need for ML expertise when designing models for novel tasks. Our method employs neural architecture search and network compression in a two-stage approach to discover hardware efficient models. This approach consists of a global search stage that explores a wide range of architectures while considering hardware constraints, followed by a local search stage that fine-tunes and compresses the most promising candidates. We exceed performance on various tasks and show further speedup through model compression techniques such as quantization-aware-training and neural network pruning. We synthesize the optimal models to high level synthesis code for FPGA deployment with the hls4ml library. Additionally, our hierarchical search space provides greater flexibility in optimization, which can easily extend to other tasks and domains. We demonstrate this with two case studies: Bragg peak finding in materials science and jet classification in high energy physics, achieving models with improved accuracy, smaller latencies, or reduced resource utilization relative to the baseline models.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗
DOE OSTI2024

Efficient graph representation framework for chemical molecule similarity tasks

Graph data has emerged in numerous scientific domains and machine learning techniques have been widely used for analysis and learning of diverse data for prediction and decision. Machine learning techniques can readily address complex problems by leveraging their structural information. But graphs cannot be directly used for existing machine learning algorithms unless encoded as vectors. The problem of efficient representation of graphs is a substantial challenge in graph machine learning. In this paper, we propose a novel two-stage framework for the representation of chemical molecule graphs based on the strengths of Graph Isomorphism Networks (GINs) and Siamese autoencoders. In the first stage, the GIN model is constructed and trained using the structural information of chemical molecule graphs. Node attributes, edge attributes, and edge indices are used as input data, while graph attributes are used as labels. The GIN model effectively captures the structural characteristics of graphs and can accurately predict graph attributes, i.e., molecular properties. It also generates Graph Embeddings, represented as vectors that encode the structural information of graphs. In the second stage, Graph Embedding vectors are further optimized for downstream similarity tasks while preserving the graph structural information. The Siamese autoencoder is constructed and trained, which reduces the dimensionality of the Graph Embedding vectors, while maximizing the preservation of structural information in the original high-dimensional vectors. The resulting low-dimensional Graph Embeddings can be effectively utilized for tasks such as approximate nearest neighbor search. The experimental results demonstrate the effectiveness of our proposed framework in accurately predicting graph similarity.

Ma, Jiaji↗
DOE OSTI2022

3.5 kW/in3 Planar Coupled Inductor Design and Optimization for a 50 kW 3-level Four-Switch Buck-Boost (3L-FSBB) Converter

Planar magnetics design has been widely used in power electronics field because of the low profile, ease of manufacturability, and high power density. This paper first proposes a 3-level four-switch buck-boost (3L-FSBB) converter for the dc-dc applications where the zero common-mode (CM) voltage emissions, step-up and step-down are all required. With the consideration of power density, efficiency, and cost, a 15oz heavy copper coupled inductor for 3-level FSBB converter was designed and optimized with the low-cost commercial core and planar windings. According to the soft-switching quadrangle modulation for 3-L FSBB, the turns number and stacked core number are optimized first. And then, a two-board design is proposed to reduce the fringing effect and improve the thermal performance. Finally, a 75 kVA two-stage ac-dc-dc converter including a 3-level FSBB converter was built to verify the electrical design.

Cao, Yuliang↗
DOE OSTI2022

A Fairness-based Distributed Energy Coordination for Voltage Regulation in Distribution Systems

The rapid adoption of photovoltaic (PV) systems combined with falling prices of energy storage is paving the way for a future in which customers could locally supply their energy needs and export surplus power to the grid. However, reverse power flow from multiple sites in a network can cause overvoltage that decreases system reliability and the utilization level of PV system. In this paper, we present a two-stage approach to solve the energy coordination problem in distribution networks. In the first stage, a household with PV and energy storage is optimized to allow customers to adopt different control strategies such as load following and tariff arbitrage. We show that such customer-led control strategies can cause reverse power flows during peak PV generation periods, thus resulting in overvoltage. In such cases, the second stage of the proposed approach is called upon to solve a fairness-based energy coordination problem that can maintain voltages within limits while maximizing the PV generation while fairly distributing the actions among all the PV systems. As a result, each of the resources in the network has equal opportunity to export power and shares the responsibility of voltage regulation. Simulation studies have been carried out on the IEEE 13-bus test feeder to verify the effectiveness of the proposed approach.

Distributed energy resources, fairness, overvoltag↗
DOE OSTI2022

Reinforcement Learning for Volt- Var Control: A Novel Two-stage Progressive Training Strategy

This paper develops a reinforcement learning (RL) approach to solve a cooperative, multi-agent Volt-Var Control (VVC) problem for high solar penetration distribution systems. The ingenuity of our RL method lies in a novel two-stage progressive training strategy that can effectively improve training speed and convergence of the machine learning algorithm. In Stage 1 (individual training), while holding all the other agents inactive, we separately train each agent to obtain its own optimal VVC actions in the action space: fconsume, generate, do-nothingg. In Stage 2 (cooperative training), all agents are trained again coordinatively to share VVC responsibility. Rewards and costs in our RL scheme include (i) a system-level reward (for taking an action), (ii) an agent-level reward (for doing-nothing), and (iii) an agent-level action cost function. This new framework allows rewards to be dynamically allocated to each agent based on their contribution while accounting for the trade-off between control effectiveness and action cost. The proposed methodology is tested and validated in a modified IEEE 123-bus system using realistic PV and load profiles. Simulation results confirm that the proposed approach is robust and computationally efficient; and it achieves desirable volt-var control performance under a wide range of operation conditions.

24 POWER TRANSMISSION AND DISTRIBUTION↗
DOE OSTI2023

Stochastic Look-Ahead Commitment: A Case Study in MISO

This paper introduces the Stochastic Look Ahead Commitment (SLAC) software prototyped and tested for the Midcontinent Independent System Operator (MISO) look ahead commitment process. SLAC can incorporate hundreds of wind, load, and net scheduled interchange (NSI) uncertainty scenarios. It uses a progressive hedging method to solve a novel two-stage stochastic unit commitment. The first stage commitment decisions, made only for those generators whose decision to commit or not in each time period cannot be deferred, can cover the uncertainties within the next three hours. The second stage includes both the dispatch for each of the scenarios and the commitment decisions that can be deferred. Study results on 15 MISO production days show that SLAC may bring economic and reliability benefits under uncertainty.

MATHEMATICS AND COMPUTING↗
DOE OSTI2021

Restoration Strategy for Active Distribution Systems Considering Endogenous Uncertainty in Cold Load Pickup

Cold load pickup (CLPU) phenomenon is identified as the persistent power inrush upon a sudden load pickup after an outage. Under the active distribution system (ADS) paradigm, where distributed energy resources (DERs) are extensively installed, the decreased outage duration can induce a strong interdependence between CLPU pattern and load pickup decisions. In this paper, we propose a novel modelling technique to tractably capture the decision-dependent uncertainty (DDU) inherent in the CLPU process. Subsequently, a two-stage stochastic decision-dependent service restoration (SDDSR) model is constructed, where first stage searches for the optimal switching sequences to decide step-wise network topology, and the second stage optimizes the detailed generation schedule of DERs as well as the energization of switchable loads. Further, to tackle the computational burdens introduced by mixed-integer recourse, the progressive hedging algorithm (PHA) is utilized to decompose the original model into scenario-wise subproblems that can be solved in parallel. The numerical test on modified IEEE 123-node test feeders has verified the efficiency of our proposed SDDSR model and provided fresh insights into the monetary and secure values of DDU quantification.

24 POWER TRANSMISSION AND DISTRIBUTION↗
DOE OSTI2026

Simplex‐based model for nanoparticle grain identification in four‐dimensional scanning transmission electron microscopy data

Grain identification in polycrystalline nanoparticles, for example, determining which crystal phases are present at each spatial location, is fundamental to materials characterisation. This is particularly challenging when grains overlap extensively, as commonly occurs in four-dimensional scanning transmission electron microscopy (4D-STEM) datasets. We propose a simplex-based model (SBM) in which each simplex vertex represents the diffraction pattern (DP) of a pure grain, and the simplex edges and interior represent overlapping grains. Our SBM grain identification algorithm operates on the Bragg disk (BD) data matrix distilled from the 4D-STEM data to identify the grain membership at each scan position, together with a BD feature matrix whose columns represent the DPs for each constituent grain, which is important for identifying the crystal structure of each grain. We solve the model using a two-stage algorithm. In Stage 1, we adapt a linear mixing algorithm to estimate an initial BD feature matrix whose columns represent DPs of potentially overlapping grains. Our Stage 2 algorithm incorporates sparsity considerations to transform the initial BD feature matrix so that its columns represent DPs of pure grains. Using simulated datasets with various grain configurations, we demonstrate that SBM recovers both the BD feature matrix and membership maps more accurately than existing methods, even when a grain lacks any pure region and completely overlaps with other grains.

4D-STEM segmentation↗
DOE OSTI2024

Evaluation of Minimum NOx Emission From Ammonia Combustion

Abstract Ammonia (NH3) is being explored as a hydrogen carrier with no carbon emissions. However, if burned directly as NH3, rather than being completely decomposed back to N2/H2, the fuel-bound nitrogen comes with a potentially significant NOx emissions penalty. Indeed, several existing studies are showing ammonia combustion NOx emissions that exceed current natural gas fueled, DLN technologies by one to two orders of magnitude. Therefore, it is important to establish the theoretical minimum NOx emissions for an ammonia combustor, to determine how much NOx levels can be reduced via further technology development. In other words, the purpose of this work is not to analyze the performance of a specific combustor but, rather, the fundamental limits of what is achievable. This study quantifies this minimum NOx level for a two-stage combustor system for a given combustor exit temperature and residence time, with a constraint on unburned fuel levels. As expected, the optimum configuration is a rich front end combustor to burn and crack ammonia with significant H2 production, followed by an NO relaxation reactor, followed by a lean stage that consumes the remaining H2. The optimum residence time and stoichiometry of each zone are determined in the fast mixing limit, which essentially balances between NOx production in the primary and secondary zones. These results show minimum NOx levels are in 200–400 ppm range at 1 bar, but drop to levels of ∼25 ppm at 20 bar. These NOx emissions are dominated by NOx production in the primary stage which relaxes to equilibrium levels quite slowly. As processes controlling NOx relaxation to equilibrium in the primary stage dominate overall NO emission levels, combustor NOx sensitivities are essentially opposite that of natural gas fired, DLN systems. Specifically, NOx values drop with increased combustor residence time, increased pressure, and increased combustor exit temperature. These results also suggest that the most important strategy for NOx minimization is to provide sufficient relaxation time after the primary zone for NOx to approach equilibrium—this can be done via kinetic means to accelerate this relaxation rate, such as enhancing pressure or temperature, or increasing residence times. Indeed, this work shows that low pressure combustors specifically optimized for ammonia will have residence times that are one to two orders of magnitude larger than current natural gas systems. By doing so, NOx levels below 10 ppm may be achievable. Finally, we discuss the sensitivity of these values to uncertainties in ammonia kinetics.

Engineering↗
DOE OSTI2021

Data for Conversion of High-Solids Hydrothermally Pretreated Bioenergy Sorghum to Lipids and Ethanol Using Yeast Cultures

Glucose and xylose are the major sugars present in cellulosic hydrolysates. The cellulosic sugars can be used for the production of platform chemicals. In this study, productions of lipid and ethanol by yeasts were compared for concentrated bioenergy sorghum syrup. Bioenergy sorghum was hydrothermally pretreated at 50% w/w solids in a continuous industrial reactor and sequentially mechanically refined using a burr mill to improve biomass accessibility for hydrolysis. Fed-batch enzymatic hydrolysis was conducted with 50% w/v solids loading and cellulase cocktail (50 FPU/g biomass) to achieve 230 g/L sugar concentration. Various strains of Rhodosporidium toruloides were evaluated for converting sugars into lipids, and strain Y-6987 had the highest lipid titer (9.2 g/L). The lipid titer was improved to 19.0 g/L by implementing a two-stage culture scheme, where the first stage was optimized for yeast growth and the second for lipid production. For ethanol production, the engineered Saccharomyces cerevisiae SR8ΔADH6 was used to coferment glucose and xylose. Ethanol fermentation was optimized for media nutrients (YP, YNB/urea, and urea), cellulosic sugar concentration, and sulfite conditioning to maximize the ethanol concentration from sorghum syrups. Fermentation of 70% v/v concentrated hydrolysate conditioned with sulfite produces 50.1 g/L ethanol from 141 g/L of sugars.

Conversion↗
DOE OSTI2021

Analysis of pump-to-signal noise transfer in two-stage ultra-broadband optical parametric chirped-pulse amplification

In optical parametric chirped-pulse amplification (OPCPA), pump temporal intensity modulation is transferred to the chirped-signal spectrum via instantaneous parametric gain and results in contrast degradation of the recompressed signal. We investigate, for the first time to our knowledge, the pump-to-signal noise transfer in a two-stage ultra-broadband OPCPA pumped by a single laser and show the dependence of pump-induced signal noise, characterized both before and after pulse compression, on the difference in pump-seed delay in the two stages. We demonstrate an up-to-15-dB reduction of the pump-induced contrast degradation via pump-seed delay optimization. Experiments and simulations show that, even when parametric amplifiers are operated in saturation, the pump-seed delay can be used to minimize the pump-induced contrast degradation that is attributed largely to the noises from the unsaturated edges of the pulse and that of the beam.

47 OTHER INSTRUMENTATION↗
DOE OSTI2025

Distributed Online Voltage Control in Distribution Network: A Two-Stage Real-Time Implementation

The increasing integration of renewable-based distributed generation (DG) brings growing challenges to distribution network (DN) voltage control. To address this issue, a two-stage distributed online voltage control framework (TDO-VC) is proposed in this paper. In the proposed method, legacy voltage control devices are controlled in the upper stage on an hourly timescale, while DGs operate autonomously online in the lower stage to remove instantaneous voltage violations. The idea of receding horizon control is applied in the upper stage to comprehensively consider the current and future renewable generation, and the generalized fast dual ascent (Gf-DA) is employed in the lower stage to effectively manage DGs through near real-time optimization. The effectiveness of the proposed TDO-VC is demonstrated by rigorous theoretical analysis and case studies on IEEE-123 bus system.

Distributed generation↗
DOE OSTI2021

Scalable Techniques for Stochastic Power Flow Problems (Final Report)

The proposed research focuses on developing scalable algorithms for two-stage security-constrained OPF problems with AC power flow constraints, a class of problems complicated by (i) scale arising from a scenario representation; and (ii) the presence of nonlinearity, nonconvexity, and possibly second-stage discreteness or complementarity. Unfortunately, most existing solvers cannot contend with both challenges simultaneously; accordingly, the proposed research focuses on developing solution techniques that can both scale with the number of scenarios and contend with nonconvexity and second-stage complementarity. We consider three avenues for addressing such problems: (i) Variable sample-size SQP (VS-SQP) methods that combine sparse Quasi-Newton updates with a scalable variance-reduced stochastic gradient scheme for stochastic QP subproblems, allowing for contending with second-stage complementarity via regularization; (ii) Variable sample-size stochastic Interior-point (VS-sIP) schemes that propose a sampling-based regularized (to allow for contending with complementarity) interior-point schemes in which a Schur-complement technique is employed for decomposing the Newton direction computation step; (iii) Variable sample-size tractable ADMM (VS-tADMM) schemes combine variable sample-sizes with carefully designed techniques for resolving each of the nonconvex updates (by leveraging the QCQP structures). We intend to compare the three schemes using performance profiles in terms of solution quality, scalability, etc. and then select one scheme which will then be developed and further refined in Python for purposes of the GO competition.

42 ENGINEERING↗
DOE OSTI2022

Multi-Fidelity Stochastic Economic Dispatch for Operating Low Carbon Power Grids

Grid operators can address the inherently stochastic nature of renewables by solving a two-stage stochastic programming model that minimizes the cost of dispatch decisions while accounting for the complex grid dynamics. It is common to use a sample average approximation to estimate the expectation of the second stage costs in this model. However, the large sample count needed for numerical accuracy makes effective modeling large-scale electric grids computationally intractable. We introduce a control variate multi-fidelity estimator for the second-stage recourse that enables high quality dispatch decisions in real-time with a reduced computational burden. We obtain a hierarchy of model fidelities by linearizing the AC power flow system representation to DC power flow, and by relaxing transmission and voltage network constraints. We evaluate the performance of our proposed method on a synthetic grid with 73 buses against a deterministic baseline with persistence forecast and a high-fidelity reference. Our analysis shows a computational speed-up of 7.62x with a minimal loss in accuracy. The multi-fidelity method is well suited to fidelity combinations that use a simplified network topology in their lower fidelity model and is an attractive option for applications where accurate grid modeling needed on a limited computational budget.

24 POWER TRANSMISSION AND DISTRIBUTION↗
DOE OSTI2024

Advanced HVAC Humidity Control for Hot-Humid Climates

During this project we develop and validate a cost-effective, integrated control solution to improve humidity control and comfort for energy-efficient homes in hot-humid climates. This study focuses on developing a strategy that is effective, field tested, and practical for builders to install with minimal disruption to standard practices. A successful solution would simplify the transition to high-performance humidity control and be the basis for design and installation guidance. By relying on the central system as a starting point, the strategy employed minimizes system complexity and cost for builders, while improving comfort and operating cost for homeowners. The solution strategy was to coordinate the cooling, dehumidification, and ventilation functions of central, ducted HVAC systems to better control indoor humidity, improve occupant thermal comfort, and capture energy savings. The primary strategic goals were to: (1) optimize dehumidification by the central air-conditioning system, particularly during part-load conditions, using conventional equipment with modified control settings and lower system airflows; (2) maximize ventilation during heating/cooling on-cycles, to “bank” and condition outdoor ventilation air, and minimize ventilation during off-cycles; (3) quantify the effectiveness and energy impact of the dehumidification and ventilation strategies, while identifying a metric that would be useful to evaluate latent effectiveness. For the test houses in our study, located in Richmond Hill, Georgia; Houston, Texas; and Monroe, Louisiana we observed: (1) the indoor humidity did not exceed 60% RH during the monitored cooling season for 99% of the time in Richmond Hill, 96% of the time in Houston, and 90% of the time in Monroe; (2) the dehumidification strategy improved the steady-state latent capacity of the HVAC system at design conditions by 16% to 49% at the Houston test house and by 28% to 71% at the Monroe test house, depending on which mode the system was operating in; and (3) the good results at the test houses were primarily due to the amount of time the air-conditioning system operated in ramping or dehumidification modes, or both, particularly during the early cooling season. This study demonstrates that air conditioners or heat pumps with a single-stage compressor can provide good humidity control without the need for a two-stage or variable-stage compressor system. The airflow and control settings for ramping and dehumidification modes are critical to control indoor humidity in hot-humid climates, particularly during part-load and shoulder season conditions. The dehumidification strategy used in this study did not jeopardize the mechanical reliability of the cooling equipment. The strategies used in this study are applicable across various equipment brands, models, and efficiency levels, and also applicable to a broad range of homes in hot-humid climates. Results will vary by specific equipment, location, and house configuration and construction.

29 ENERGY PLANNING, POLICY, AND ECONOMY↗
DOE OSTI2026

Long-Term Impacts of Constrained Transmission Deployment on the Cost-Reliability Tradeoff

Traditional Resource Adequacy (RA) frameworks in the U.S. undervalue the contributions of inter-regional transmission to resource adequacy during stress periods, focusing on the availability of nameplate capacity instead. However, availability of nameplate capacity does not always translate into electricity delivery, especially during tail events. Moreover, the rapid deployment of energy-limited resources and increasing electricity demand challenge existing resource adequacy frameworks and couple regional electricity demand and availability of supply via transmission. We propose a two-stage framework that goes beyond the existing capacity-centered approaches to reveal the RA contributions of transmission. In the first stage we introduce a multi-objective optimization framework to quantify the merits of transmission expansion via Pareto Frontiers under alternative futures of no transmission investment, primary energy resources availability and demand growth. The second stage focuses on tail events and leverages the results of the first stage to characterize the risk profile of regional consumers across the U.S. under the alternative energy futures. We find that no new transmission can lead to a more expensive and less reliable national grid across scenarios, however, the impact on regional RA can vary. The probabilistic analysis reveals that transmission investments can alleviate the tail risk of consumers, however, the availability of fuel resources does not always alleviate regional tail risks. Our findings inform policymakers and utilities on the prioritization of transmission investments to mitigate the risk of widespread outages, also for tail events, and ensure reliable and affordable electricity delivery to all.

24 POWER TRANSMISSION AND DISTRIBUTION↗
DOE OSTI2024

Understanding the High-Temperature Deformation Behaviors in Additively Manufactured Al6061+TiC Composites via In Situ Neutron Diffraction

Aluminum matrix composites (AMCs) are designed to enhance the performance of conventional aluminum alloys for engineering applications at both room and elevated temperatures. However, the dynamic phase-specific deformation behavior and load-sharing mechanisms of AMCs at elevated temperatures have not been extensively studied and remain unclear. Here, in situ neutron diffraction experiments are employed to reveal the phase-specific structure evolution of additively manufactured Al6061+TiC composites under compressive loading at 250 °C. It is found that the addition of a small amount of nano-size TiC significantly alters the deformation behavior and increases the strength at 250 °C in comparison to the as-printed Al6061. Unlike the two-stage behavior observed in Al6061, the Al6061+TiC composites exhibit three stages during compression triggered by changes in the interphase stress states. Further analysis of Bragg peak intensity and broadening reveals that the presence of TiC alters the dislocation activity during deformation at 250 °C by influencing dislocation slip planes and promoting dislocation accumulation. These findings provide direct experimental observations of the phase-specific dynamic process in AMCs under deformation at an elevated temperature. The revealed mechanisms provide insights for the future design and optimization of high-performance AMCs.

36 MATERIALS SCIENCE↗