SEARCH · Engineering Papers
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.
The PARETO training workshop is a two-hour, interactive experience intended to teach participants how to: install PARETO software, input data into PARETO; run PARETO optimization; model a variety of complex network scenarios; and analyze, interpret, and compare optimization results.
The Austin SHINES project and solution is a software management platform, for an electric grid with a high penetration of dispersed photovoltaic (PV) solar generation sites, which maintains the traditional power quality and reliability associated with grid service. This project developed and deployed the platform as a Distributed Energy Resource Management System (DERMS), engaging multiple advanced controls, to evaluate operation and optimization of a fleet of diverse DER assets, installed at several locations among Austin Energy’s customers and distribution system. The project also produced a methodology to create a replicable DERMS template, adaptable to other regions and market structures. Last, Austin SHINES aimed to demonstrate the solution’s methodology would enable the DER grid ecosystem to serve load at a technical cost (System Levelized Cost of Electricity, or System LCOE) of less than the U.S. Department of Energy SHINES program metric of $0.14/kWh, in a defined boundary, while enabling a high penetration of distributed PV. Research was categorized in 6 reports (Final Deliverables = FD) listed below, with titles and descriptions indicating which area of understanding was investigated: FD-1: System Levelized Cost of Electricity (System LCOE) Methodology The creation and use of the System LCOE to Serve Load metric that encompasses the holistic, system-level costs and benefits of all resources, and enables them to be evaluated based on their ability to support an efficient and low-cost integrated grid ecosystem. FD-2: Software Platform Product Description The creation of new DER control methodologies deployable within a utility-grade software platform that enable DER's to maximize their benefit within a grid, that is capable of serving load enabling a high penetration of distributed PV generation. FD-3: Optimal Design Methodology Optimal design methodologies for individual DER installations that enable utilities to determine the optimal combinations and sizing for individual DER sites. FD-4: Austin SHINES Ownership and Operation Models for DER System Performance A comparison of multiple DER aggregation and ownership methodologies including direct utility control, third-party aggregator, and autonomous. FD-5: Economic Modeling & Optimization A comparison of multiple DER technology mixes and configurations within the distribution system, providing insight into an optimal blend of technologies that best enable the distribution system to serve load at the lowest cost at high penetrations of solar. FD-6: Fielded Assets Deployed DER assets within the Austin Energy SHINES circuits. Austin SHINES provided an opening for state-of-the-art technology products to be deployed, providing a rich opportunity for improving how each of the products perform as stand-alone products, and in concert with other complementary products. The Austin SHINES project comprised of two key metrics for System LCOE: SystemLCOE_SHINES<$0.14/kWh Modeled ΔSystemLCOE_SHINES/ΔSystemLCOE_Base≥20% at same solar penetration The System LCOE calculation uses the costs of the utility-owned infrastructure as it exists today, the cost of the DERs that exist in the system today, and the cost of the purchase of energy from ERCOT wholesale markets over the course of the calendar year. All costs are on an annualized basis. The capital and operating costs are derived from the rate case, which produces a yearly cost. The net cost of energy and services imported to the system is integrated over the test year, as is the load served and solar penetration. The first metric was easily achieved by every scenario considered. The goal was set when the Department of Energy’s SHINES Funding Opportunity Announcement was written in 2015 and was a more difficult target at the time. Due mostly to rapidly declining costs for DERs and the significant decrease in the Electric Reliability Council of Texas (ERCOT) energy market prices, which results in lower net cost of energy purchases, the System LCOE is well below this target for all scenarios considered. A fleet of DERs can assume different mixtures, each of which serves the load at a different LCOE. The optimal mixture of DERs serves load at the smallest System LCOE. The second metric (hereinafter %delta metric) asks that the holistic DERMS controls reduce the incremental cost above the baseline of going to a high solar penetration future by at least 20% as compared to the case of a DER deployment with no sophisticated controls (autonomous). Many comparison sets were created throughout this project. Physical technology was installed for informing utility engineering and testing several types of operational control schemes, through the DERMS. The types of operational control which were compared for valuation of the System LCOE Metric were: Holistic control = using the full suite of the DERMS platform to decide and optimize how/why the systems operate depending on weather, market, and reliability signal input. Autonomous control = a local mode at the asset site, wherein a schedule operates the asset, with visibility into performance only No control = the baseline for comparing value against the other two types of control The types of ownership control included: Direct Utility control = the utility dispatches a signal to each asset Third-Party Aggregator = a third party aggregates a fleet of assets and the utility dispatches one signal for all Autonomous = a local mode is set for operation at the asset site, wherein a schedule operates the asset, with visibility into performance only The types of control methodologies deployable within a utility-grade software platform included: Utility Peak Load Reduction = Lower transmission cost obligation Day-Ahead Energy Arbitrage = Realize economic value through price differential Real-Time Price Dispatch = Realize economic value from real-time price spikes Voltage support = Reduce losses and increase solar generation Distribution Congestion Management = Increase local grid reliability Demand Charge Reduction = Lower customer bills and realize system benefit The fielded assets deployed for the project were: Utility Scale Kingsbery Energy Storage System: 1.5 MW / 3 MWh Li-Ion battery storage Mueller Energy Storage System: 1.75 MW / 3.2 MWh Li-Ion battery storage, 7 Energy Storage Units (250 kW each) La Loma Community Solar: 2.6 MW Commercial Scale Aggregated storage installations at 3 sites, with existing solar (300+ kW): One 18 kW / 36 kWh Li-Ion battery storage Two 72 kW / 144 kWh Li-Ion battery storage Residential Scale Aggregated storage installations: -Six stationary battery storage systems (10 kWh each) at homes with existing solar -One Electric Vehicle installed as Vehicle-to-Grid (V2G) Utility-Controlled Solar via Smart Inverters at 12 homes Autonomously-Controlled Smart Inverters at 6 homes Over the course of the project, Austin SHINES undertook installing more than 3 MW of distributed battery energy storage, smart PV inverters, a DER control platform, and other enabling technologies utilizing customer and utility locations and aggregation models. All of these resources were to be integrated and optimized at the utility level. DER assets and control methodologies were designed to achieve a credible pathway to a System LCOE for energy delivered to load of $0.14//kWh or less by 2020, while maximizing distributed solar generation and maintaining acceptable standards of power quality. The project also established a template for other regions to follow, to maximize the adoption of distributed solar PV in support of an economic and efficient grid. In total, the Austin SHINES project added value to the DER subject area in each layer of integration. From utility, to commercial to residential scales, the sheer hierarchy of communication and coordination was a significant accomplishment in addition to learnings from what these communications revealed was unique to each. Economically, the most effective method demonstrated was the criticality of planning phases. Contingencies and multiple projection scenarios helped guide the project to deploy optimal design as close as feasible, in real world conditions. The project and reports will serve public benefit by outlining specific areas of DER strategy and installation where many stakeholders and needs can be addressed with improved efficiency. Overall, communities and utilities should use the results to guide the increasing options available for powering the grid with DER, renewables, and carbon considerate energy.
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.
This report summarizes the research and capabilities developed as part of the project “Analysis and Validation of Critical Infrastructure Models using Model Order Reduction” (CI-MOR) LDRD project. CI-MOR research enables the solution of large, complex optimization models that naturally arise in national security challenges involving critical infrastructures. Specifically, CI-MOR researchers developed methods to (1) rigorously approximate complex, nonlinear optimization formulations, (2) identify alternative near-optimal solutions, (3) accelerate optimization workflows used for complex applications, and (4) rigorously integrate domain knowledge in stochastic-process models. This report provides an overview of the research done in CI-MOR, and we describe application exemplars used to illustrate CI-MOR capabilities. Furthermore, we describe the software developed by CI-MOR that researchers can leverage to analyze new applications.
Hydropower generation may play an increasingly important role in the power grid under increasing contribution of variable renewable sources such as wind and solar. An improved understanding of the changes to hydropower dispatch under future higher VRE grid conditions reveals research gap that should be informed power grid planning and reservoir water releases policies considering multiple other water uses and varying hydrologic condition. Here this study aims to understand the role of hydropower in a changing power grid by employing a production cost model, PLEXOS, across future power system scenarios, planning horizons, and regions. We explore optimized hydropower dispatch to understand its potential role in minimizing the system cost and renewable curtailment. We also examine the sensitivity of hydropower revenue under various grid scenarios of the Eastern U.S. and hydrology conditions. Results indicate hydropower generation follows net load and compensates for the variability of solar and wind generation. Although energy prices are lower during some periods in the future grid scenarios, there is a potential for higher revenue for hydropower by providing both energy and ancillary services during times of stress. Additionally, hydropower revenue is sensitive to hydrology in the SERC region, which we considered as an example. The feasibility of hydropower dispatching with higher ramps between low and high hourly-capacity factors, as indicated in the optimization model, requires further study to consider other water use and ecology constraints.
Modeling distributed power generation systems often requires complicated mathematical expressions that present challenges for commercial optimization solvers. Here, this paper presents a matheuristic to solve a mixed-integer optimization model that informs decisions regarding the design and dispatch of a utility-connected microgrid. We deploy a genetic algorithm to search the system design space and a linear program to solve the economic dispatch problem. The model is a component of a web tool that requires solutions within a few minutes. Our method yields objective function values within 5% of an exogenously produced optimal in fewer than 30 seconds for 90% of our test cases compared to only 10% of our test cases by a traditional optimization solver in the same amount of time.
A collection of analytical studies is presented related to unconstrained and constrained aircraft (a/c) energy-state modeling and to spacecraft (s/c) motion under continuous thrust. With regard to a/c unconstrained energy-state modeling, the physical origin of the singular perturbation parameter that accounts for the observed 2-time-scale behavior of a/c during energy climbs is identified and explained. With regard to the constrained energy-state modeling, optimal control problems are studied involving active state-variable inequality constraints. Departing from the practical deficiencies of the control programs for such problems that result from the traditional formulations, a complete reformulation is proposed for these problems which, in contrast to the old formulation, will presumably lead to practically useful controllers that can track an inequality constraint boundary asymptotically, and even in the presence of 2-sided perturbations about it. Finally, with regard to s/c motion under continuous thrust, a thrust program is proposed for which the equations of 2-dimensional motion of a space vehicle in orbit, viewed as a point mass, afford an exact analytic solution. The thrust program arises under the assumption of tangential thrust from the costate system corresponding to minimum-fuel, power-limited, coplanar transfers between two arbitrary conics. The thrust program can be used not only with power-limited propulsion systems, but also with any propulsion system capable of generating continuous thrust of controllable magnitude, and, for propulsion types and classes of transfers for which it is sufficiently optimal the results of this report suggest a method of maneuvering during planetocentric or heliocentric orbital operations, requiring a minimum amount of computation; thus uniquely suitable for real-time feedback guidance implementations.
Accurate solar Photovoltaic (PV) power forecasting is important for enhancing both the performance and economic feasibility of PV systems. This study evaluates several deep learning models, including Dense Neural Networks (DNN), Long Short-Term Memory (LSTM), Convolutional Neural Networks (CNN), and a hybrid LSTMCNN model, for predicting PV power production one day in advance. Prior to optimization, the models exhibited relatively high errors, with the best model (DNN) achieving a Root Mean Square Error (RMSE) of 31.13 kW and a coefficient of determination (R 2 ) of 62.15 %. After employing Bayesian optimization, the LSTM-CNN model demonstrated the best performance, with the RMSE reduced to 9.79 kW and R 2 improved to 97.62 %, showcasing significant enhancement in predictive accuracy. Here, the economic evaluation considered three cases: rewards for underestimation (0.08 USD/kWh), no rewards, and penalties for both over-and underestimation (120 % of the utility tariff). In the rewards scenario, the LSTM-CNN model reduced the Levelized Cost of Electricity (LCOE) by 4 %, while in the penalty scenario, a backup diesel generator would have increased the LCOE by 49 %. Additionally, the LSTM-CNN model minimized financial losses, achieving the lowest penalties and maximizing net cash flow compared to other models, demonstrating its overall technical and economic superiority.
Waveform filtering is a standard step in traditional seismic phase picking but often receives little attention in deep learning workflows, where models are typically trained on raw or minimally processed waveforms. Although this strategy performs well for local events, we show that performance can degrade substantially at regional distances. To address this limitation, we introduce two ways to incorporate multiband-filtered waveforms into deep learning phase pickers. The stacking approach concatenates filtered inputs along the channel dimension, while the branching approach processes each frequency band through a dedicated network branch before feature fusion. Both approaches can substantially improve performance across epicentral distances of 0° to 20°, but their effectiveness depends strongly on the selected frequency bands. Tests with multiple filter banks show that filter-bank design should be treated as part of model optimization rather than as a fixed preprocessing choice. Grad-CAM analysis of the branching model indicates that band importance varies among waveform samples and across training realizations, with only a weak overall preference for the 0.25 to 0.5 Hz band. These results show that no single filter band is consistently optimal and demonstrate that explicit feature engineering remains valuable for robust deep learning-based seismic phase picking.
This paper discusses modeling and field studies of controlled EV charging that have been performed with the goal of minimizing requirements for infrastructure upgrades, minimizing building peak demand charges, and maximizing the use of on-site generation. We present a large-scale workplace charging pilot of a demand-controlled scheduled EV charging system with over 250 active daily commuters, successfully demonstrating management of aggregate charging power to avoid new infrastructure investments, mitigate peak demand charges, and provide cost-effective workplace charging to users. In addition to understanding opportunities for demand management, integrating these controllable loads into the energy modeling process for new buildings will also be necessary. This paper then presents an example energy modeling process that evaluates the potential effects of EV charging on building load profiles and infrastructure requirements for a mixed-use community. Finally, we discuss an illustration of how EV charging can be controlled to be synergistic with other building loads and distributed generation.
How can we collect the most useful labels to learn a model selection policy, when presented with arbitrary heterogeneous data streams? In this paper, we formulate this task as a contextual active model selection problem, where at each round the learner receives an unlabeled data point along with a context. The goal is to output the best model for any given context without obtaining an excessive amount of labels. In particular, we focus on the task of selecting pre-trained classifiers, and propose a contextual active model selection algorithm (CAMS), which relies on a novel uncertainty sampling query criterion defined on a given policy class for adaptive model selection. In comparison to prior art, our algorithm does not assume a globally optimal model. We provide rigorous theoretical analysis for the regret and query complexity under both adversarial and stochastic settings. Our experiments on several benchmark classification datasets demonstrate the algorithm’s effectiveness in terms of both regret and query complexity. Notably, to achieve the same accuracy, CAMS incurs less than 10% of the label cost when compared to the best online model selection baselines on CIFAR10.
In conventional creep testing (CCT) a specimen is subject to constant load and temperature for a long-duration until creep rupture occurs. Conventional testing can be costly when considering the number of experiments needed to characterize the creep response of a material over a range of stress and temperature. To predict long-term creep-rupture properties, the time-temperature-stress superposition principle (TTSSP) approach has been employed where stress and/or temperature is applied at an elevated level; the result of which are extrapolated down to low stress and/or temperature conditions. These methods have been successful in predicting minimum-creep-strain-rate (MCSR) and stress-rupture (SR) but suffer from an inability to predict the creep deformation curve or account for changes in deformation mechanisms or aging that occurs at long-duration. An accelerated technique, termed the Stepped isostress method (SSM) allows the accelerated testing of materials to determine their creep deformation response. Unlike TTSSP tests, the SSM test employs a single specimen where the stress is periodically step increased until rupture. The SSM creep deformation curve is processed (time and strain shifted) to produce an accelerated creep deformation curve that represent the creep deformation curve at the initial stress level in SSM. A processing procedure for metals has yet to be developed. The research objective of this study is to develop a processing procedure for SSM test data using a creep-damage constitutive model. Triplicate SSM tests were conducted on Ni-based superalloy Inconel 718 at 650°C with stress being periodically increased until rupture. Triplicate CCT tests were conducted at the initial stress level of the SSM tests. The Sine-hyperbolic (Sinh) creep-damage model was employed in this study. The Sinh creep-damage constitutive model is based on coupled creep strain rate and damage evolution equations; where both rates are dependent on the current state of damage. Calibration is two-step: analytical and numerical optimization. Each stepped creep deformation curve is tackled quasi-analytically to determine MCSR and SR related material constants and accumulated damage. The damage accumulated at the end of each step was then passed onto subsequent steps to calibrate the MCSR, rupture prediction, and damage evolution. Numerical optimization was applied to optimize model constants involved in the creep strain constitutive equations in order to generate best-fitted Sinh creep deformation curves. The Sinh model predictions were compared to the SSM and CCT data. The Sinh model satisfactorily predicts the SSM data and thus the calibrated material constants provides a good estimate of rupture found in the CCT data. Calibration using SSM data reduces the number of tests needed to calibrate a model; significantly reducing costs. A single SSM test replaces numerous creep tests at different stresses.
Man machine development cycle and mathematical models for optimal human performance
Abstract Urea is the most used nitrogen fertilizer due to its ease of storage, transportation, and application. It is made by combining ammonia and carbon dioxide (CO 2 ), both of which are produced predominantly from fossil fuels at present. The recent momentum behind ammonia production using renewable-powered electrolysis offers an opportunity to both make urea in a more sustainable way and utilize CO 2 from external sources. In this work, we present a techno-economic optimization model to minimize the cost of making urea in this way. The model allows for time-varying chemical production in response to renewable variability by simultaneously optimizing production facility design and hourly operation. We performed a case study for Minnesota considering the use of byproduct CO 2 from bioethanol production. We found that the present-day levelized cost of renewable urea is between $268 mt −1 and $413 mt −1 at likely implementable production scales up to 250 000 mt yr −1 . This is within the range of historical conventional urea prices while offering at least 78% carbon intensity reduction. Projecting to 2030, there is a clear economic case for renewable urea production with levelized cost as low as $135 mt −1 due to technology improvement and electrolysis manufacturing expansion, facilitating a urea production scale increase to 525 000 mt yr −1 . Optimal facilities use wind energy, with hydrogen and ammonia production operating in a flexible, time-varying way to minimize battery and hydrogen storage capacities. Urea production operates near steady state due to the relatively low cost of intermediate ammonia buffer storage. A mix of imported methane and locally produced hydrogen are used to provide heat for steam consumed in the urea synthesis.
Composite materials are widely used in aircraft structures because of their superior mechanical properties. However, their complex failure modes require sophisticated inspection methods to ensure structural integrity. Ultrasonic testing (UT) is a common non-destructive inspection (NDI) technique for aircraft composites that can detect internal and external defects with high resolution and accuracy. Despite their effectiveness, traditional UT methods rely on the manual interpretation of ultrasonic signals, which is time-consuming, labor-intensive, and subjective. Furthermore, processing such large-scale data, particularly across materials of varying thicknesses, significantly increases the computational demands of deep learning model optimization. To overcome these challenges, we propose an efficient sparse mixture-of-experts (MoE) model with a multi-level loss function and introduce four novel training objectives to improve computational efficiency and accuracy in identifying surface defects in composite aircraft materials. Here, we evaluated our approach on material with multiple thicknesses or domains comprising various defects. Our experimental results demonstrate higher accuracy and F1-Score, with only 10% training epochs compared to baseline MoE.
The Environmental Determinants of the Diabetes in the Young (TEDDY) study has prospectively followed, from birth, children at increased genetic risk of type 1 diabetes. We evaluated the potential of machine learning to identify new biomarkers that predict imminent (within 6 months) development of persistent islet autoantibodies to insulin, GAD or IA-2 in TEDDY participants through integration of time-invariant risk factors with time-varying metabolomics. The predictive modeling was initiated with over 220 potential biomarkers; through ensemble-based feature evaluation, the optimal model included 42 biomarkers, returning a cross-validated receiver operating characteristic area under the curve of 0.74. The model identified a principal set of 20 time-invariant markers, including 16 single nucleotide polymorphisms and two HLA-DR genotypes, gestational age, and exposure to a prebiotic formula. Integration of the metabolome identified 22 high-priority metabolites and lipids, including adipic acid and ceramide d42:0, that predicted development of islet autoantibodies, dependent upon the time horizon. The majority (86%) of metabolites that predicted development of islet autoantibodies belonged to 3 pathways: lipid oxidation, phospholipase A2 signaling, and pentose phosphate pathway. TEDDY data suggest that these metabolic processes may play a role in triggering islet autoimmunity.
I will give an overview talk about flight software systems, robotics technologies and modeling for energy minimization as applied to vehicles and buildings infrastructures. A dominant issue in both design and operations of robotic spacecraft is the minimization of energy use. In the design and building of spacecraft increased power is acquired only at the cost of additional mass and volumes and ultimately cost. Consequently, interplanetary spacecrafts are designed to have the minimum essential power and those designs often incorporate careful timing of all power use. Operationally, the availability of power is the most influential constraint for the use of planetary surface robots, such as the Mars Exploration Rovers. The amount of driving done, the amount of science accomplished and indeed the survivability of the spacecraft itself is determined by the power available for use. For the Mars Exploration Rovers there are four tools which are used: (1) models of the rover and it s thermal and power use (2) predictive environmental models of power input and thermal environment (3) fine grained manipulation of power use (4) optimization modeling and planning tools. In this talk I will discuss possible applications of this methodology to minimizing power use on Earth, especially in buildings.