Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “operability”

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 253 records · Page 14
All sourcesNASA NTRSDOE OSTI
DOE OSTI2025

The Effect of Operational Temperature on the Performance and Durability of Solid Oxide Fuel Cells and Solid Oxide Electrolysis Cells

Solid oxide fuel cells (SOFC) and solid oxide electrolysis cells (SOEC) have received great interest due to their highly effective reversibility as power generation and H2 production system without releasing any greenhouse gases into environment. The LSCF electrode exhibits a higher structural and performance stability under both SOFC and SOEC operation due to its mixed ionic and electronic conductivity, and there is no immediate delamination taking place during the initial several hundred hours operation. However, the LSCF based air electrode still presents significant performance degradation (with the increased resistance) over the prolonged operation, such as over 1000 hours of operation under SOFC and SOEC. The influence factors for the cell’s performance and stability need to be optimized to improve the power generation for SOFC and H2 production for SOEC. The effects of operational temperature on the performance and durability for both SOFC and SOEC are electrochemical operation dependent. The performance and performance durability for the first 1500h were currently studied under optimized operational temperature for reversible SOFC/SOEC.

Fan, Yueying [NETL Site Support Contractor, Nation↗
DOE OSTI2023

Engine Operating Conditions, Fuel Property Effects, and Associated Fuel–Wall Interaction Dependencies of Stochastic Preignition

This work for the Coordinating Research Council (CRC) explores dependencies on the opportunity for fuel to impinge on internal engine surfaces (i.e., fuel–wall impingement) as a function of fuel properties and engine operating conditions and correlates these data with measurements of stochastic preignition (SPI) propensity. SPI rates are directly coupled with laser–induced florescence measurements of dye-doped fuel dilution measurements of the engine lubricant, which provides a surrogate for fuel–wall impingement. Literature suggests that SPI may have several dependencies, one being fuel–wall impingement. However, it remains unknown if fuel-wall impingement is a fundamental predictor and source of SPI or is simply a causational factor of SPI. In this study, these relationships on SPI and fuel-wall impingement are explored using 4 fuels at 8 operating conditions per fuel, for 32 total test points. The fuels were directly injected at two different injection timings: an earlier injection timing that initially targets the piston crown and a later injection timing that targets the cylinder liner. At each injection timing, the engine was operated at both 90°C and 70°C coolant and lubricant temperatures, and 185 and 200 kPa absolute intake manifold pressure. This work serves as an exploratory effort to down select conditions and provide initial fuel properties of interest for a secondary study to explore fuel property specific effects on fuel-wall interaction and SPI propensity. Significant findings from this initial operating condition and fuel property exploratory work are: 1. reduced engine operating coolant and lubricant temperatures, along with 2. retarded injection timings were required to increase SPI propensity. Moreover, at these conditions some fuel specific effects were also observed; specifically, increased ethanol content increased measured dye–wall (i.e., fuel–wall) interaction. However, despite increased dye–wall interaction, the increased volatility of the ethanol containing fuels also reduced the estimated fuel retention in the top-ring zone and associated measured SPI propensity. Thus, the findings of this unique approach to explore relationships between fuel-wall impingement and SPI highlight that SPI propensity is more directly proportional to retained fuel, and not simply fuel–wall impingement. Furthermore, fuel retention was found to be directly influenced by complex fuel property and engine operating condition relationships. Either retarded injection timings and/or increased fuel volatility increased fuel wall-impingement, while less volatile fuels and/or reduced coolant temperatures increased fuel retention. Therefore, for a given operating condition, the data highlights that greater volatile fuels exhibit increased fuel wall impingement without increased fuel retention or SPI propensity, while less volatile fuels could exhibit reduced fuel-wall impingement but increased fuel retention and SPI propensity rates.

33 ADVANCED PROPULSION SYSTEMS↗
DOE OSTI2019

ARM-Related Unmanned Aerial System (UAS) and Tethered Balloon System (TBS) Operational Requirements and Approval

This procedure addresses the requirements and process for obtaining approval for Unmanned Aerial System (UAS) and Tethered Balloon System (TBS) operations at DOE Office of Science (SC) Atmospheric Radiation Measurement user facility observatories (ARM sites), ARM-supported training flights, atmospheric instrument development flights, and campaigns at non-ARM sites. ARM sites include both ARM observatories and ARM mobile facility deployments. The protection of people, property, and the environment during UAS and TBS operations is the foundation of executing a successful mission using UAS or TBS. This procedure is designed to facilitate safe mission execution. It does not address obtaining approval for ARM-supported research, which is approved through the ARM Field Campaign (AFC) or Engineering Evaluation (EE) processes. Approval of UAS and TBS operations is a prerequisite for final approval of an AFC or EE by the Infrastructure Management Board (IMB) and may be conducted in parallel with preparation and submittal of the AFC or EE proposal to the IMB. This procedure addresses UAS and TBS requirements; mission planning; documentation; maintenance; risk assessment; airworthiness determination; operation approval; training and qualification requirements and responsibilities for UAS remote pilots; TBS operators; visual observers; maintenance technicians; and A&P mechanics. It also addresses the responsibilities of ARM site managers, the Office of Science Aviation Safety Manager, principal investigators, requesting organizations, and Range Safety Officers. This procedure is derived from DOE O 440.2C, Aviation Management and Safety, and Federal Aviation Administration (FAA) Regulations (FARs), 14 CFR 107, and 14 CFR 101. It contains two sections: Section 3.0 addresses UAS operations; Section 4.0 addresses TBS operations.

unmanned aerial systems, tethered balloon systems,↗
DOE OSTI2020

GOOML: Geothermal Operational Optimization with Machine Learning

Geothermal Operational Optimization with Machine Learning (GOOML) is a project focused on maximizing increased availability and capacity from existing industrial-scale geothermal generation assets. The GOOML project will develop a suite of machine learning-based algorithms that analyze historical production datasets and provide predictive setpoints for geothermal field operations. Historical datasets from New Zealand and the US will provide the input to develop digital geothermal system twins which allow prediction of market conditions, maintenance operations and steamfield optimization. The algorithms will identify key parameters within fields and suggest setpoints for components of the system to maintain optimal generation. Set-points can be instructed to follow mass flow restrictions, generation maximization and optimal field/reservoir balance and give field operators a guide by which generation can be optimized. The datasets that will be used to develop GOOML are sourced from operating geothermal fields in New Zealand and the United States with varying degrees of complexity. This will ensure that most geothermal systems can utilize the GOOML tool to assist in optimizing operations. GOOML aims to achieve a step-change in geothermal operations by developing state-of-the-art machine learning algorithms, comprehensive data analytics, and a first-of-its-kind automated, intelligent geothermal system model.

algorithms↗
DOE OSTI2022

Operating Strategies for Dispatchable PEM Electrolyzers that Enable Low-Cost Hydrogen Production

Hydrogen is a pathway to enabling decarbonization across multiple economic sectors that cannot be directly decarbonized with electricity including heat for industrial operations, medium- and heavy-duty transportation applications, long-duration energy storage, and as a feedstock for chemical synthesis. Producing hydrogen at low costs and carbon footprint is likely essential to economically decarbonize these otherwise "hard to decarbonize" sectors. Low-temperature polymer electrolyte membrane (PEM) electrolysis produces hydrogen from water and electricity and is a rapidly developing pathway towards making hydrogen at the scales required for decarbonization applications. The levelized cost of electrolytic hydrogen is dependent on the capital cost, efficiency, and durability of the electrolyzer system as well as the price of electricity supplied to the electrolyzer and the annual utilization of the electrolyzer (capacity factor). Electricity price and capacity factor depend on the source of energy that the system uses and impact other economic factors. Electricity price and capacity factor and the connections between other aspects of hydrogen production via PEM electrolyzers are the focus of this work. PEM electrolyzers have conventionally been operated at high capacity factors using electricity purchased from utilities with a constant price throughout the year. To achieve lower effective electricity prices, recent work has investigated opportunities for electrolyzers to purchase electricity in wholesale markets, where the cost of electricity varies hourly. This option can lead to lower electricity costs when the electrolyzer is a controllable load that ramps up and down rapidly and turns on and off frequently. This configuration and operating strategy capitalizes on times of low wholesale electricity prices and results in lower hydrogen levelized costs than constant operation due to reduced electricity costs even though the reduced capacity factor increases the cost of recovering the capital investment. Cycling on and off frequently also has implications for electrolyzer durability. An electrolyzer configuration where it is directly connected to renewable generation such as wind or solar, only running when the generator is producing energy, has similar implications on operating strategy, hydrogen levelized cost, capacity factor, and durability. This work provides insight into the relationships between dispatchable electrolyzer operating strategies and the cost of producing hydrogen from these systems, outlining strategies and opportunities to minimize production costs while minimizing operations that are likely to negatively impact system durability and efficiency. We find strategies that minimize electrolyzer cycling and the resulting durability impacts while increasing the hydrogen levelized cost only slightly above the minimum. We also find opportunities for batteries to minimize the number of cycles in systems directly connected to renewable generation. These findings outline key opportunities for future electrolyzer deployments and the synergistic benefits between electrolyzers and increased deployment of renewable energy generation like wind and solar. They also inform research and development that is reducing electrolyzer capital cost while managing durability impacts.

electricity markets↗
DOE OSTI2022

GOOML - Finding Optimization Opportunities for Geothermal Operations: Preprint

Geothermal Operational Optimization with Machine Learning (GOOML) is a transferable and extensible component-based geothermal asset modeling framework that considers complex steamfield relationships and identifies optimization prospects using a data-driven approach. We have used this framework to develop digital twins that provide steamfield operators with an operational environment to analyze and understand historical and forecasted power production, explore new steamfield configuration possibilities, and seek optimal asset management for real world applications. The GOOML modeling software is built on a generic component-based systems framework that allows for both historical and forecast analysis. A GOOML model can perform historical data-assimilation using first-principal thermodynamics to create a meaningful data model. Historical production data can then be coupled with a forecast framework to train machine-learning models of steamfield components to predict future outputs. This modeling environment enables digital exploration of steamfield design configurations and operational scenarios. GOOML digital twins have been developed for steamfields in New Zealand and the United States representing differing power generation and field conditions. These digital twins have been validated by comparing hindcast predictions against historical production data. Reinforcement learning experiments were conducted to demonstrate the ability to programmatically explore the operations space using machine learning agents. Our initial results are compelling; two to five percent increases in annual energy production were demonstrated by the GOOML models with no additional infrastructure build required. GOOML offers a new approach to geothermal operations by applying state-of-the-art machine learning algorithms, comprehensive data analytics, and interaction with digital twins. Through application of these tools, operators will realize greater availability and higher net generation which will increase the cost effectiveness of geothermal energy projects.

access↗
DOE OSTI2023

A New Offering for the Seaman Status Labyrinth - Seaman Status for Nuclear Reactor Operators on Floating Nuclear Power Plants

Floating nuclear power plants present a unique operating environment for land-based nuclear reactor operators. Traditionally located in the control room of a nuclear power plant on land, development of floating nuclear power plants exposes the traditional land-based employees to the marine environment. With the extension of nuclear power generation facilities into the maritime domain, do nuclear reactor operators working on a floating nuclear power plant qualify as seaman under maritime law? Applying existing maritime law, the answer is no, a nuclear reactor operator who operates the nuclear reactor on a floating nuclear power plant does not qualify as a seaman because their work is not in support of the mission of the vessel and the reactor is not connected to a vessel because a floating nuclear power plant is not a vessel. Applying the analysis developed by the Supreme Court in Chandris v. Latsis and the recent Sanchez v. Smart Fabricators of Texas, L.L.C. en banc decision by the Fifth Circuit, a nuclear reactor operator on a floating nuclear power plant does not qualify for seaman status under the Jones Act because their function supports the operation of the reactor and the structure on which the reactor resides does not meet the reasonable person standard established in Lozman v. City of Riviera Beach. Further, existing case law highlights that rendering a structure practically impossible to move eliminates the structure from consideration as a vessel. Because a floating nuclear power plant may be anchored at a seaport or anchored offshore but connected via transmission cables and protected by physical protection barriers, a floating nuclear power plant, with no current means of propulsion is rendered a power plant on water, which is its true function. Recognizing that technological change may alter the conclusion presented in this Article, current designs and structures that exist illustrate the intersection between nuclear and maritime law and the ever-evolving concepts that underpin seamen status in maritime law.

Fialkoff, Marc↗
DOE OSTI2024

An Efficient Annual-Performance Model of a Geothermal Network for Improved System Design, Operation, and Control: Preprint

Geothermal district energy systems (DES), and specifically geothermal networks, provide a viable solution for decarbonizing residential and commercial heating and cooling loads. District energy systems of all kinds enable a thermal resource with a relatively high capital cost (such as a geothermal borehole field) to be shared among a large number of users. While district heating and cooling has been studied for many years, geothermal networks, fifth generation DES that utilize water-source heat pumps and an ambient temperature loop to meet heating and cooling loads, have not been implemented extensively and thus require additional technical and economic optimization to obtain maximum benefits. This paper presents a newly developed reduced-order model that captures the flow of energy within the network, including the commercial and residential users' electrical usage, at an hourly rate over a year. The model includes building loads, heat pumps, borehole fields, and auxiliary heat/cool input, all connected with an ambient-temperature thermal loop model. In the model, operational control is possible for the borehole fields, circulation pump, and auxiliary system. For a given system, the model can output the complete state parameters for each component, the thermal loop, and the collective system, such as flow rate over time, average thermal loop temperature over time, and total electricity usage. The model can also be used to optimize the system control for maximizing system efficiency or minimizing system operational cost. For example, one initial assessment of the borehole controller for an example system showed that a controller with on/off operation of the borehole field reduces annual electrical usage by 33%, compared with continuous operation mode. Hence, the model can assist in optimizing a given system's operation to get the most value out of a geothermal network installation. Future work will consider the model's application to a demonstration project, including the model validation against operational data and system operation optimization.

ambient-temperature loop↗
DOE OSTI2024

AGN-201 Digital Twin Concept of Operations

The AGN-201 is a nuclear reactor at Idaho State University (ISU) and is currently being used in the development of a digital twin (DT) with Idaho National Laboratory (INL). The goal is to create a DT (called the AGN-201 DT) which can monitor an operating nuclear reactor to determine when the reactor is being operated normally; normal operations are any operations that is declared by the ISU staff. To accomplish this, researchers at INL developed a DT ecosystem which can ingest data from the AGN-201 reactor. This data is fed into a series of machine learning (ML) and reactor physics models. The ML and reactor physics models assess the data and determine if the reactor is operating normally. Event(s) that are flagged as anomalies are investigated by the INL staff to determine if any undeclared experiments were conducted. The AGN-201 DT was verified in July/August 2023, when the ISU staff performed undeclared experiments and the INL staff were able to assess the anomalous events and determine what likely caused each event. This project was a steppingstone to promote the use of DTs for international safeguards and marks the first time a DT was able to monitor a nuclear reactor for this purpose. This document provides the concept of operations (CONOPS) for the AGN-201 digital twin (DT). The CONOPS describes how the AGN-201 and AGN-201 DT are expected to operate and details how each are established.

21 SPECIFIC NUCLEAR REACTORS AND ASSOCIATED PLANTS↗
DOE OSTI2025

A Theoretical Operational Model for Complex Experiments and its Invariance Theorems

We develop and systematize the Theoretical–Operational Model (TOM), a framework that treats preparation and measurement —including their operational residues— as intrinsic structures of physical theory. The central contribution is a principled geometric–algebraic organization of admissible operational deformations, formulated using quantum channels, renormalization-style flows, and information-geometric tools. Within this structure, operational residues and background processes are represented as effective morphisms attached to these operational components, whose invariants yield constraints on how theoretical parameters vary under specified classes of deformations. Illustrations drawn from muon–electron conversion, long-baseline neutrino oscillations, and quark–gluon-plasma phenomenology show how TOM maps operational effects into inferences about theoretical parameters, enables systematic cross-experimental comparisons, and stabilizes parameter estimation against defined deformation families. By embedding the operational layer—together with its residues—within a structured theoretical setting, TOM supports both theory testing and theory development, clarifying the conceptual relation between experimental realization and the physical quantities represented by the theory.

Pronskikh, Vitaly [Fermilab] (ORCID:00000002518174↗
DOE OSTI2023

Solar Uncertainty Management and Mitigation for Exceptional Reliability in Grid Operations (SUMMER-GO): Project Final Report

The Solar Uncertainty Management and Mitigation for Exceptional Reliability in Grid Operation (SUMMER-GO) project was recently completed through a collaboration among the National Renewable Energy Laboratory, Maxar, the Electric Reliability Council of Texas (ERCOT), the University of Texas at Dallas, the University of California Berkeley, and the University of Colorado Boulder. The project made significant advances in probabilistic solar power forecasting, both through the development of Bayesian model averaging methods for ensemble forecasting and in bringing these and other advancements into practice with Maxar's delivery of operational forecasts to ERCOT. In addition to creating more reliable solar power forecasts, the project developed methods for their utilization in power system operations. These include the development of risk-aware unit commitment and economic dispatch algorithms and methods to reformulate probabilistic forecasts to be used in these power system operational models. Dynamic power system reserve methods were also developed, which have been shown in silico to create economic savings and reliability improvements on an ERCOT-like system as well as financial savings in the ERCOT system through more granular consideration of the uncertainty associated with solar power forecasts. Finally, a situational awareness tool to help grid operators better understand solar power forecast uncertainty in daily operations was developed and extensively vetted.

14 SOLAR ENERGY↗
DOE OSTI2021

A forecast-driven decision-making model for long-term operation of a hydro-wind-photovoltaic hybrid system

Hydro-wind-photovoltaic (PV) hybrid system has the potential to increase the integration of renewable energy sources into an existing grid. For the long-term operation of the system, due to the non-storable nature of wind and PV power, it is essentially to decide the optimal long-term carryover storage of cascade reservoirs. However, it remains a challenge due to high uncertainties of long-term forecasts and complicated hydraulic/electrical relationships between cascade reservoirs. Here in this study, a forecast-driven decision-making model is proposed for the hybrid system, which converts the multi-stage long-term operation process into a two-stage operation problem (including current stage and carryover stage) to avoid using longer-horizon forecast information with lower accuracy. First, the carryover stage energy surfaces (CESs) considering the forecast uncertainties of wind, solar and hydro resources are proposed to characterize carryover stage benefit quantitatively. Then a CESs-based forward decision-making optimization model is developed to guide the long-term operation of a hydro-wind-photovoltaic hybrid system. The applications in a hydro-wind-PV hybrid system of Yalong River basin results show that: compared with conventional operation, 1) power generation increases 9.03%; 2) in terms of the carryover storages control, the reservoir impounding and drawdown timing are delayed, and the drawdown depth is increased, which can be used to formulate better reservoir operation rules.

13 HYDRO ENERGY↗
DOE OSTI2022

A comprehensive and fair comparison of two neural operators (with practical extensions) based on $\mathrm{FAIR}$ data

Neural operators can learn nonlinear mappings between function spaces and offer a new simulation paradigm for real-time prediction of complex dynamics for realistic diverse applications as well as for system identification in science and engineering. Herein, we investigate the performance of two neural operators, which have shown promising results so far, and we develop new practical extensions that will make them more accurate and robust and importantly more suitable for industrial-complexity applications. The first neural operator, DeepONet, was published in 2019 (Lu et al., 2019), and its original architecture was based on the universal approximation theorem of Chen & Chen (1995). The second one, named Fourier Neural Operator or FNO, was published in 2020, and it is based on parameterizing the integral kernel in the Fourier space. DeepONet is represented by a summation of products of neural networks (NNs), corresponding to the branch NN for the input function and the trunk NN for the output function; both NNs are general architectures, e.g., the branch NN can be replaced with a CNN or a ResNet. According to Kovachki et al. (2021), FNO in its continuous form can be viewed conceptually as a DeepONet with a specific architecture of the branch NN and a trunk NN represented by a trigonometric basis. In order to compare FNO with DeepONet computationally for realistic setups, we develop several extensions of FNO that can deal with complex geometric domains as well as mappings where the input and output function spaces are of different dimensions. We also develop an extended DeepONet with special features that provide inductive bias and accelerate training, and we present a faster implementation of DeepONet with cost comparable to the computational cost of FNO, which is based on the Fast Fourier Transform. Here we consider 16 different benchmarks to demonstrate the relative performance of the two neural operators, including instability wave analysis in hypersonic boundary layers, prediction of the vorticity field of a flapping airfoil, porous media simulations in complex-geometry domains, etc. We follow the guiding principles of FAIR (Findability, Accessibility, Interoperability, and Reusability) for scientific data management and stewardship. The performance of DeepONet and FNO is comparable for relatively simple settings, but for complex geometries the performance of FNO deteriorates greatly. We also compare theoretically the two neural operators and obtain similar error estimates for DeepONet and FNO under the same regularity assumptions.

42 ENGINEERING↗
DOE OSTI2025

A higher-order finite-element implementation of the nonlinear Fokker–Planck collision operator for charged particle collisions in a low density plasma

Collisions between particles in a low density plasma are described by the Fokker–Planck collision operator. In applications, this nonlinear integro-differential operator is often approximated by linearised or ad-hoc model operators due to computational cost and complexity. In this work, we present an implementation of the nonlinear Fokker–Planck collision operator written in terms of Rosenbluth potentials in the Rosenbluth–MacDonald–Judd (RMJ) form. The Rosenbluth potentials may be obtained either by direct integration or by solving partial differential equations (PDEs) similar to Poisson's equation: we optimise for performance and scalability by using sparse matrices to solve the relevant PDEs. We represent the distribution function using a tensor-product continuous-Galerkin finite-element representation and we derive and describe the implementation of the weak form of the collision operator. We present tests demonstrating a successful implementation using an explicit time integrator and we comment on the speed and accuracy of the operator. Finally, we speculate on the potential for applications in the current and next generation of kinetic plasma models.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗
DOE OSTI2021

Strategies for microgrid operation under real-world conditions

Microgrids are an increasingly relevant technology for integrating renewable energy sources into electricity systems. Based on a microgrid implementation in California, in this study we investigate microgrid operation under real-world conditions. These conditions have not yet been considered in combination and encompass energy charges, demand charges, export limits, as well as uncertainty about future electricity demand and generation in the microgrid. Under these conditions, we evaluate the performance of two frequently applied groups of strategies for microgrid operation. The first group is composed of proactive strategies that optimize decisions based on forecasts of future electricity generation and demand. The second group includes reactive strategies that make operational decisions based exclusively on the current state of the microgrid. We evaluate the performance of the strategies under varying operational parameters, forecast accuracies, and microgrid configurations—well beyond our Californian showcase. Our results confirm the expectation that proactive strategies outperform reactive ones in the majority of settings. Yet, reactive strategies can perform better under short control intervals or under moderate prediction errors of PV generation or demand. Furthermore, the interplay between real-world conditions and operational strategies reveals several additional insights for research on microgrid operation. First, we find that demand charges and export limits decisively affect microgrid performance. Second, the impact of forecast errors is highly non-linear and non-monotonous. Third, escalating negative interactions between forecast errors and demand charges make proactive strategies benefit from longer control intervals. This result is contrary to existing best practice, which promotes short control intervals to minimize the impact of uncertainty.

24 POWER TRANSMISSION AND DISTRIBUTION↗
DOE OSTI2024

Co-optimization of nuclear reactor flexible power operation and maintenance scheduling

As flexible power operation of nuclear power plants becomes more attractive due to the reduction in fossil-fueled dispatchable generation on energy grids, finding optimal power production strategies that balance revenue generation with operational concerns becomes more complex. This article presents a general framework to aid operators in designing economically optimal long term dispatch strategies for nuclear power plants. The principal novelty is the linking of estimated system remaining useable life (RUL) to strategic operational decisions. It is shown that, depending on the relationship between the fixed costs from maintenance and the associated lost revenue from an outage, it can be economically optimal in the long term to delay a maintenance outage and not perform this alongside refueling. For a given relationship between power ramping and degradation, optimal strategies were found that discouraged load following in some situations while minimizing unnecessary maintenance. It is shown that heavy load following can cause maintenance and refueling outages to diverge due to their inverse relationships with respect to load following, potentially leading to a significant loss in capacity factor. As a result, this general framework can be applied to specific reactor dispatch allowing operators to adapt operational strategies as future grid conditions change.

21 SPECIFIC NUCLEAR REACTORS AND ASSOCIATED PLANTS↗
DOE OSTI2022

Neural-network based collision operators for the Boltzmann equation

Kinetic gas dynamics in rarefied and moderate-density regimes have complex behavior associated with collisional processes. These processes are generally defined by convolution integrals over a high-dimensional space (as in the Boltzmann operator), or require evaluating complex auxiliary variables (as in Rosenbluth potentials in Fokker-Planck operators) that are challenging to implement and computationally expensive to evaluate. In this work, we develop a data-driven neural network model that augments a simple and inexpensive BGK collision operator with a machine-learned correction term, which improves the fidelity of the simple operator with a small overhead to overall runtime. The composite collision operator has a tunable fidelity and, in this work, is trained using and tested against a direct-simulation Monte-Carlo (DSMC) collision operator.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗
DOE OSTI2026

Active operator learning with predictive uncertainty quantification for partial differential equations

With the increased prevalence of neural operators being used to provide rapid solutions to partial differential equations (PDEs), understanding the accuracy of model predictions and the associated error levels is necessary for deploying reliable surrogate models in scientific applications. Existing uncertainty quantification (UQ) frameworks employ ensembles or Bayesian methods, which can incur substantial computational costs during both training and inference. Here, we propose a lightweight predictive UQ method tailored for Deep operator networks (DeepONets) that also generalizes to other operator networks. Numerical experiments on linear and nonlinear PDEs demonstrate that the framework’s uncertainty estimates are unbiased and provide accurate out-of-distribution uncertainty predictions with a sufficiently large training dataset. Our framework provides fast inference and uncertainty estimates that can efficiently drive outer-loop analyses that would be prohibitively expensive with conventional solvers. We demonstrate how predictive uncertainties can be used in the context of Bayesian optimization and active learning problems to yield improvements in accuracy and data-efficiency for outer-loop optimization procedures. In the active learning setup, we extend the framework to Fourier Neural Operators (FNO) and describe a generalized method for other operator networks. To enable real-time deployment, we introduce an inference strategy based on precomputed trunk outputs and a sparse placement matrix, reducing evaluation time by more than a factor of five. Our method provides a practical route to uncertainty-aware operator learning in time-sensitive settings.

97 MATHEMATICS AND COMPUTING↗