Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Fixed Base”

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 91 records · Page 5
All sourcesNASA NTRSDOE OSTI
DOE OSTI2025

Multiphysics Meshfree Degradation Modeling of Energy Storage Materials with Kernel Enrichment

Energy storage materials exhibit strong electro-chemo-mechanical coupling and highly anisotropic material properties, contributing to the formation and propagation of micro-cracking during charge/discharge cycling and ultimately diminishing performance and service life. With microstructural images supplied by the National Renewable Energy Laboratory (NREL), pixel-based meshfree model construction by the reproducing kernel particle method (RKPM) is used to represent the complex material microstructures that dictate the coupled physics of these systems. Traditional electro-chemo-mechanical models rely on mesh-based finite element methods, which can lead to difficulties in meshing such complex geometries and capturing crack propagation due to mesh dependency. The first kernel enrichment discussed will be the interface modified reproducing kernel (IM-RK) [1, 2], constructed by scaling a smooth kernel function with an interface-distance function to achieve strategic discontinuity types (i.e. weak discontinuities for strain discontinuities and strong discontinuities for cracks) and alleviate Gibbs oscillations near these transition zones. The IM-RK is especially useful for areas in which a known discontinuity-type is expected a priori. The second kernel enrichment to be discussed is a neural network-enhanced reproducing kernel (NN-RK) [3, 4], which is introduced to effectively model non-obvious damage and crack propagation in the material microstructures; the location, orientation, and solution transition near a localization are automatically captured by superimposed block-level NN optimizations. This NN enrichment approach allows for effective modeling of localizations via a fixed background discretization, relieving tedious efforts for adaptive refinement in traditional mesh-based methods. Applications to the heterogeneous microstructures of Li-ion battery cathodes will be presented to demonstrate the effectiveness of the proposed methods. NN-RK is additionally used to inform how crack opening and closure in turn affect the electro-chemo-mechanical responses in the material microstructure. Reference: [1] Wang, Y., Baek, J., Tang, Y. et al. "Support vector machine guided reproducing kernel particle method for image-based modeling of microstructures," Comput Mech 73, 907-942 (2024). https://doi.org/10.1007/s00466-023-02394-9. [2] Susuki, K., Allen, J. & Chen, J. S.. "Image-based modeling of coupled electro-chemo-mechanical behavior of Li-ion battery cathode using an interface-modified reproducing kernel particle method," Engineering with Computers (2024). https://doi.org/10.1007/s00366-024-02016-9. [3] Baek, J., Chen, J. S., Susuki, K., "Neural Network enhanced Reproducing Kernel Particle Method for Modeling Localizations," International Journal for Numerical Methods in Engineering, Vol. 123, 4422-4454 (2022). https://doi.org/10.1002/nme.7040.

25 ENERGY STORAGE↗
DOE OSTI2026

Kernel Enriched Meshfree Multiphysics Degradation Modeling of Energy Storage Materials

Energy storage materials exhibit strong electro-chemo-mechanical coupling and highly anisotropic material properties, contributing to the formation and propagation of micro-cracking during charge/discharge cycling and ultimately diminishing performance and service life. With microstructural images supplied by the National Laboratory of the Rockies (NLR), pixel-based meshfree model construction by the reproducing kernel particle method (RKPM) is used to represent the complex material microstructures that dictate the coupled physics of these systems. Traditional electro-chemo-mechanical models rely on mesh-based finite element methods, which can lead to difficulties in meshing such complex geometries and capturing crack propagation due to mesh dependency. The first kernel enrichment discussed will be the interface modified reproducing kernel (IM-RK) [1, 2], constructed by scaling a smooth kernel function with an interface-distance function to achieve strategic discontinuity types (i.e. weak discontinuities for strain discontinuities and strong discontinuities for cracks) and alleviate Gibbs oscillations near these transition zones. The IM-RK is especially useful for areas in which a known discontinuity-type is expected a priori. The second kernel enrichment to be discussed is a neural network-enhanced reproducing kernel (NN-RK) [3, 4], which is introduced to effectively model non-obvious damage and crack propagation in the material microstructures; the location, orientation, and solution transition near a localization are automatically captured by superimposed block-level NN optimizations. This NN enrichment approach allows for effective modeling of localizations via a fixed background discretization, relieving tedious efforts for adaptive refinement in traditional mesh-based methods. Applications to the heterogeneous microstructures of Li-ion battery cathodes will be presented to demonstrate the effectiveness of the proposed methods. NN-RK is additionally used to inform how crack opening and closure in turn affect the electro-chemo-mechanical responses in the material microstructure. References: [1] Wang, Y., Baek, J., Tang, Y. et al. "Support vector machine guided reproducing kernel particle method for image-based modeling of microstructures," Comput Mech 73, 907-942 (2024). https://doi.org/10.1007/s00466-023-02394-9. [2] Susuki, K., Allen, J. & Chen, J. S.. "Image-based modeling of coupled electro-chemo-mechanical behavior of Li-ion battery cathode using an interface-modified reproducing kernel particle method," Engineering with Computers (2024). https://doi.org/10.1007/s00366-024-02016-9. [3] Baek, J., Chen, J. S., Susuki, K., "Neural Network enhanced Reproducing Kernel Particle Method for Modeling Localizations," International Journal for Numerical Methods in Engineering, Vol. 123, 4422-4454 (2022). https://doi.org/10.1002/nme.7040.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2024

Verification of Adaptive Protection in Hardware in the Loop for Coordination with Solar Variability

As inverter-based resources continue to be installed at all levels of the electric grid, the fixed protection schemes used at the distribution level will continue to be stressed until they no longer ensure the protection of the grid. Adaptive protection has been proposed as a solution with the ability to update the protection schemes in near real-time to ensure reliability and increase the resilience of the grid. However, weather variability poses a significant challenge to the ability of these methods to keep the selectivity and reliability of these schemes coordinated. If the ramp rates, due to solar variability, of the inverter-based resources change faster than the adaptive protection can issue new settings, the protection system could be uncoordinated, with the wrong device responding to a system fault. The proposed adaptive protection method ensures that due to solar variability, communication, and protection calculation latency, it can issu e coordinated protection settings promptly, in one minute or less. The hardware-in-the-loop results show the protection settings being issued and maintaining system coordination in under a minute.

Summers, Adam↗
DOE OSTI2025

Deep Reinforcement Learning-Based Control of Energy Storage for Interarea Oscillation Damping

With the increasing electricity consumption and lack of transmission investment, today's power systems are operated much closer to their limits, raising concerns of inter-area oscillations that deteriorate the system stability. Here, this article presents a novel energy storage placement and control approach for enhanced damping of interarea oscillations. Combining the residual analysis and dominant mode analysis, we are able to identify the advantageous locations for placing energy storage that achieve improved damping performance. To overcome the challenges, such as fixed control parameters and insufficient damping, we propose to use a deep reinforcement learning-based approach for energy storage control. A state-of-the-art guided surrogate-gradient-based evolutionary strategy is used to train a learning agent in a robust, efficient, and reproducible manner. Parallel computing is also adopted to speed up the training process. The proposed strategy has been tested on both medium and large-scale systems. The proposed methods have demonstrated their effectiveness in mitigating various interarea oscillations within a timeframe of 20 s, thereby averting system collapse and enhancing power grid stability effectively.

25 ENERGY STORAGE↗
DOE OSTI2025

Review of top quark mass measurements in CMS

The top quark mass is one of the most intriguing parameters of the standard model (SM). Its value indicates a Yukawa coupling close to unity, and the resulting strong ties to Higgs physics make the top quark mass a crucial ingredient for understanding essential aspects of the electroweak sector of the SM. This review offers the first comprehensive overview of the top quark mass measurements performed by the CMS Collaboration using the data collected at centre-of-mass energies of 7, 8, and 13 TeV. Moreover, a detailed description of the top quark event reconstruction is provided and dedicated studies of the dominant uncertainties in the modelling of the signal processes are discussed. The interpretation of the experimental results on the top quark mass in terms of the SM Lagrangian parameter is challenging and is a focus of an ongoing discussion in the theory community. The CMS Collaboration has performed two main types of top quark mass measurements, addressing this challenge from different perspectives: highly precise ‘direct’ measurements, based on reconstructed top quark decay products and relying exclusively on Monte-Carlo simulations, as well as ‘indirect’ measurements, where the simulations are employed to determine parton-level cross sections that are compared to fixed-order perturbative calculations. Recent mass extractions using Lorentz-boosted top quarks open a new avenue of measurements based on top quark decay products contained in a single particle jet, with promising prospects for accurate theoretical interpretations.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗
DOE OSTI2025

Modeling diffusion and depletion in high-aspect-ratio atomic layer deposition processes: Process parameters and manufacturing impacts

Atomic layer deposition (ALD) is a powerful technique for modifying the surface chemistry and properties of substrates with complex and nonplanar topologies. However, achieving uniform and conformal deposition on ultrahigh-aspect-ratio substrates remains challenging, typically requiring large quantities of precursors and long exposure times. Furthermore, process optimization is often performed empirically and involves substantial trial and error. In this work, we perform a combined experimental and computational study of ALD Al 2 O 3 infiltration into silica aerogel monoliths (aspect ratio >10 5 ). A reaction-diffusion model is used to explore the effects of key processing parameters, namely, exposure time per dose, precursor source temperature, number of aerogels in the reactor, and reactor volume. The model is based on quasi-static mode ALD, where the dosed precursor is held in the chamber for a fixed period of time before purging. We analyze the trade-offs between process throughput and precursor utilization for each of these parameters. Furthermore, we investigate the co-optimization and interactions between multiple process parameters, demonstrating the potential for further improvements. Furthermore, this physics-based model can be used to identify a set of process parameters for high-aspect-ratio ALD that meet specific manufacturing objective functions, including throughput, cost, and sustainability.

Aerogel↗
DOE OSTI2024

Semi–Analytical Modeling of Transient Stream Drawdown and Depletion in Response to Aquifer Pumping

Analytical and semi–analytical models for stream depletion with transient stream stage drawdown induced by groundwater pumping are developed to address a deficiency in existing models, namely, the use of a fixed stream stage condition at the stream–aquifer interface. Here field data are presented to demonstrate that stream stage drawdown does indeed occur in response to groundwater pumping near aquifer–connected streams. A model that predicts stream depletion with transient stream drawdown is developed based on stream channel mass conservation and finite stream channel storage. The resulting models are shown to reduce to existing fixed–stage models in the limit as stream channel storage becomes infinitely large, and to the confined aquifer flow with a no–flow boundary at the streambed in the limit as stream storage becomes vanishingly small. The model is applied to field measurements of aquifer and stream drawdown, giving estimates of aquifer hydraulic parameters, streambed conductance, and a measure of stream channel storage. The results of the modeling and data analysis presented herein have implications for sustainable groundwater management.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2025

Solving high-dimensional inverse problems using amortized likelihood-free inference with noisy and incomplete data

Here, we present a likelihood-free probabilistic inversion method based on normalizing flows for high-dimensional inverse problems. The proposed method is composed of two complementary networks: a summary network for data compression and an inference network for parameter estimation. The summary network encodes raw observations into a fixed-size vector of summary features, while the inference network generates samples of the approximate posterior distribution of the model parameters based on these summary features. The posterior samples are produced in a deep generative fashion by sampling from a latent Gaussian distribution and passing these samples through an invertible transformation. We construct this invertible transformation by sequentially alternating conditional invertible neural network and conditional neural spline flow layers. The summary and inference networks are trained simultaneously. We apply the proposed method to an inversion problem in groundwater hydrology to estimate the posterior distribution of the log-conductivity field conditioned on spatially sparse time-series observations of the system’s hydraulic head responses. The conductivity field is represented with 706 degrees of freedom in the considered problem. Comparison with the likelihood-based iterative ensemble smoother PEST-IES method demonstrates that the proposed method accurately estimates the parameter posterior distribution and the observations’ predictive posterior distribution at a fraction of the inference time of PEST-IES.

conditional invertible neural network↗
DOE OSTI2024

Effective optimization of atomic decoration in giant and superstructurally ordered crystals with machine learning

Crystals with complicated geometry are often observed with mixed chemical occupancy among Wyckoff sites, presenting a unique challenge for accurate atomic modeling. Similar systems possessing exact occupancy on all the sites can exhibit superstructural ordering, dramatically inflating the unit cell size. In this work, a crystal graph convolutional neural network (CGCNN) is used to predict optimal atomic decorations on fixed crystalline geometries. This is achieved with a site permutation search (SPS) optimization algorithm based on Monte Carlo moves combined with simulated annealing and basin-hopping techniques. Our approach relies on the evidence that, for a given chemical composition, a CGCNN estimates the correct energetic ordering of different atomic decorations, as predicted by electronic structure calculations. This provides a suitable energy landscape that can be optimized according to site occupation, allowing the prediction of chemical decoration in crystals exhibiting mixed or disordered occupancy, or superstructural ordering. Verification of the procedure is carried out on several known compounds, including the superstructurally ordered clathrate compound Rb8Ga27Sb16 and vacancy-ordered perovskite Cs2SnI6, neither of which was previously seen during the neural network training. In addition, the critical temperature of an order–disorder phase transition in solid solution CuZn is probed with our SPS routines by sampling site configuration trajectories in the canonical ensemble. This strategy provides an accurate method for determining favorable decoration in complex crystals and analyzing site occupation at unprecedented speed and scale.

Chemistry↗
DOE OSTI2024

Instantaneous mesh load factor ( K γ ) measurements in a wind turbine gearbox using fiber-optic strain sensors

The mesh load factor, K γ , describes how loads are shared between planet gears and has become one of the key design challenges in modern wind turbine gearboxes. Planet load sharing directly impacts tooth root stresses, a critical driver of torque density and gearbox reliability. Experimental evaluation of K γ is typically performed from sun gear tooth root strain gauge measurements, which are complex. Furthermore, such measurements can only provide an average value of load sharing. The present study describes an alternative method to evaluate the mesh load factor in wind turbine gearboxes based on fiber-optic strain sensors installed on the outer surface of the fixed ring gear. We present the results of an extensive measurement campaign to evaluate this novel sensing solution installed on the input planetary stage of a 2-MW wind turbine gearbox at the National Renewable Energy Laboratory's Flatirons Campus (Colorado, USA). The number of strain sensors on the ring gear was selected as an integer multiple of the number of planets, which has enabled an instantaneous evaluation of the mesh load factor. The effect of operating conditions on the planet load-sharing behavior of the gearbox has been investigated. The mesh load factor measured for operating conditions close to rated was below 1.05, well below IEC 61400-4 standard requirements.

17 WIND ENERGY↗
DOE OSTI2024

50 MW Segmented Ultralight Morphing Rotors for Wind Energy

A multi-institutional team designed a 50 Megawatt (MW) rated wind turbine featuring downwind aeroelastic morphing to reduce blade loads and allow an ultralight segmented rotor (mass reduction of about 25% compared to a conventional upwind rotor). The team used a control co-design approach with state-of-the art simulations including for the rotor and tower design using non-linear fluid-structure interactions and control algorithms, and the team also designed, built, and field-tested an aeroelastically-scaled downwind rotor to demonstrate this novel technology and validate the design tool fidelity. In a follow-on phase, these results were used along with an updated Levelized Cost Of Energy (LCOE) methodology to co-design a more detailed set of 25 MW rated turbine designs (including individual pitch control) based on minimum LCOE with highly flexible blades for an Atlantic Ocean offshore fixed-bottom design targeted towards market technology evaluation. A set of upwind and downwind designs at 25 MW rated scales were found to provide the best LCOE, with strong improvements over all previous offshore reference turbines (NREL 5MW, DTU 10 MW and IEA 15 MW). The optimized design would represent the world’s largest offshore turbine design, which combines several state-of-the-art structural, aerodynamic, and control technologies into a new and optimized system concept.

17 WIND ENERGY↗
DOE OSTI2026

CoURAGE KAZR b1 Processing: Corrections, Calibrations, and Processing Report

The U.S. Department of Energy (DOE) Atmospheric Radiation Measurement (ARM) User Facility supports atmospheric and earth system research through a comprehensive network of fixed and mobile observatories. These facilities provide long-term and intensive campaign-based observations of clouds, aerosols, precipitation, radiation, and meteorological state variables. ARM observations are designed to improve the physical understanding and numerical representation of atmospheric processes in earth system models, with particular emphasis on cloud-radiation interactions and precipitation processes. The Coast-Urban-Rural Atmospheric Gradient Experiment (CoURAGE) deploys one of the ARM Mobile Facilities (AMF) to the Mid-Atlantic region surrounding Baltimore, Maryland, for the period 1 December 2024 through 30 November 2025. This deployment focuses on characterizing atmospheric structure, cloud properties, and precipitation processes across strong land-use and surface heterogeneity gradients associated with urban, rural, and coastal (Chesapeake Bay) environments. The CoURAGE deployment complements the Baltimore Social-Environmental Collaborative (BSEC), a DOE Urban Integrated Field Laboratory (UIFL), by providing high-quality atmospheric observations needed to connect urban surface processes, emissions, and meteorology to cloud and precipitation responses. In addition to the central urban site, ancillary observing sites were deployed to rural Maryland northwest of Baltimore and to an island site in Chesapeake Bay. These measurements further complement a long-term atmospheric observatory operated in Beltsville, Maryland, by Howard University in collaboration with the Maryland Department of the Environment. Together, these assets form a four-node regional atmospheric observatory network representing Baltimore and its three primary surrounding environments—urban, rural, and coastal/bay. This coordinated observational strategy enables investigation of spatial gradients in boundary-layer structure, cloud occurrence, precipitation evolution, and aerosol-cloud interactions across complex surface regimes. Within this network, vertically pointing cloud radars play a critical role by providing continuous, high-resolution measurements of cloud and precipitation vertical structure.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2025

On the numerical sensitivity of cellular automata grain structure predictions to large thermal gradients and cooling rates

Cellular automata (CA) models of as-solidified grain structure, originally developed and applied to casting, have become a common means of predicting grain structure resulting from Additive Manufacturing (AM) processes. The majority of these models are based on the decentered octahedron approach, which attempts to correct for the effect of grid anisotropy on the prediction of competitive solidification of dendritic grains. However, AM solidification occurs under cooling rates ($\dot{T}$) and thermal gradients (G) that are orders of magnitude larger than those encountered in casting, and no systematic investigation on the effect of the CA model cell size (Δx) and time step (Δt) on AM microstructure predictions has been performed. Here, in this study, such an investigation is first performed via simulation of individual grains of various crystallographic orientations with a fixed, unidirectional G, showing that CA prediction of the steady-state undercooling matched the expected values based on the interfacial response function at small G and deviated from the expected values at large G. Simulation of competitive growth of multiple grains showed a weakening of the predicted texture as G and Δx became large. Simulation of solidification under AM conditions, where G and $\dot{T}$ vary spatially across the melt pools, showed that not only does grain selection weaken and deviate from expectations at large Δx, but grains with crystallographic $\langle$100$\rangle$ aligned with the grid directions are more adversely affected by the temperature field discontinuities than grains with other crystallographic orientations. Despite the fact that the exact grain competition results depended on Δt, the overall texture development was notably less sensitive to Δt than Δx, provided that a reasonable value of Δt is selected based on the ratio of Δx to the maximum local solidification velocity in the simulation domain. Finally, from the directional solidification and AM simulation results, an analysis of computational cost compared to simulation resolution is performed based on an equation derived to quantify the relatively inaccuracy in grain selection based on the model and temperature field inputs. From this analysis, it is concluded that there is a need for algorithmic improvements to improve CA grain competition accuracy for large G processing conditions as sufficiently small Δx to resolve the necessary competition is intractable for many AM processing conditions.

36 MATERIALS SCIENCE↗
DOE OSTI2024

Cost of Wind Energy Review: 2024 Edition [Slides]

The primary elements of this analysis include: Estimated LCOE for (1) a representative land-based wind energy project installed in a moderate wind resource in the United States, (2) a representative fixed-bottom offshore wind energy project installed in the U.S. North Atlantic, and (3) a representative floating offshore wind energy project installed off the U.S. Pacific Coast. It also updates the LCOE estimates for representative residential-, commercial-, and large-scale distributed wind projects installed in a moderate wind resource in the United States. A sensitivity analyses is included that shows the range of effects that basic LCOE variables could have on the cost of wind energy for land-based and offshore wind projects and provides updated Fiscal Year 2024 values for land-based and offshore wind energy used for Government Performance and Results Act (GPRA) reporting and illustrated progress toward established GPRA targets.

17 WIND ENERGY↗
DOE OSTI2025

Monitoring installation of partially occluded subassemblies in modular construction factories using BIM, ray tracing, and computer vision

Modular and offsite construction methods are being increasingly adopted due to the advantages they offer in terms of project completion time, quality, and energy-efficiency. Despite these advantages, the current state of monitoring systems in modular construction factories highly relies on labor-intensive, subjective, and error-prone observational methods. A large body of research has aimed to automate the monitoring process using an array of sensors, such as IMUs and RFIDs, during the past two decades. Recently, computer vision-based methods have gained increasing interest as a non-intrusive technology to monitor the process inside modular construction factories. However, partial occlusion challenges have impeded their practical application on a large scale. This challenge is specifically important for monitoring the installation of subassemblies since they can obstruct the view of the monitoring camera, especially those that enable long-term monitoring like closed-circuit television (CCTV) fixed-view surveillance cameras. Here, this paper aims to address this challenge by proposing a novel computer vision-based method to monitor the installation of new subassemblies inside modular factories in highly occluded scenes. The proposed methodology identifies the subassemblies in the CCTV video footage using computer vision, analyzes the occlusions using BIM and ray casting techniques, and estimates the progress of assembly by comparing the BIM model with the detected subassemblies in the video. The proposed methodology was successfully validated on surveillance videos captured from a volumetric modular construction factory in the U.S., achieving 93% accuracy in identifying the installation of subassemblies. The results from this research show that the integration of BIM and computer vision is a promising method for monitoring the installation processes inside modular factories under severe occlusion.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2024

The Light Dark Matter Experiment (LDMX)

The constituents of dark matter are still unknown, and the viable possibilities span a very large mass range. Specific scenarios for the origin of dark matter sharpen the focus to within about an MeV to 100 TeV. Most of the stable constituents of known matter have masses in this lower range, and a thermal origin for dark matter works in a simple and predictive manner in this mass range as well. If there is an interaction between light DM and ordinary matter, as there must be in the case of a thermal origin, then there necessarily is a production mechanism in accelerator-based experiments. The Light Dark Matter eXperiment (LDMX) is a planned electron-beam fixed-target missing-momentum experiment at SLAC that has unique sensitivity to light DM in the sub-GeV mass range. Although optimized for a missing momentum technique, LDMX is effectively a fully instrumented beam dump experiment, making it possible to search for visibly decaying signatures of dark sector particles. This would provide another outlet for LDMX to probe complementary regions of dark matter phase space for a variety of models. This contribution will give an overview of the theoretical motivation for LDMX, the main experimental challenges and how they are addressed, as well as projected sensitivities in comparison to other experiments.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗
DOE OSTI2024

Capturing O2 Desorption Through Isoconversional Kinetics for CFD Application

The present work examines desorption mechanisms and kinetics of the calcium-doped strontium perovskite at 25% calcium content (i.e., Sr_0.75 Ca_0.25 FeO_(3-δ)). Laboratory-scale, low inventory, fixed bed redox cycle experiments were developed and conducted in both dry and steam-based environments for three isothermal temperatures, 450, 500 and 550C. The temporal evolution of the conversion extent of the material sample over the redox cycles were obtained through the analysis of the gaseous products of the redox reactions. Using subsequently isoconversional differential methods, it was found that the activation energy exhibits a strong dependence on the extent of conversion during the desorption of this perovskite. In addition, the reconstruction of the reaction model showed that the desorption kinetics are controlled by a three-step mechanism. A successful a priori verification of the isoconversional desorption kinetics against the dry and steam-based environment desorption experimental data are shown.

Konan, Ndri A.↗
DOE OSTI2024

Continuous integration data-driven platform of industrial-scale subsurface storage for real-time analytics

This project helped address the growing need for efficient and scalable models to support geological carbon and energy storage, which are crucial for achieving net-zero emissions. Traditionally accurate high-fidelity numerical models have been used to simulate relevant storage processes under a handful of processes, however such models are computationally demanding, making uncertainty quantification impractical. Consequently, we first developed a machine learning framework, based on Graph Neural Operators (GNOs), to improving the accuracy of model predictions for a fixed computational budget. We then developed an Ensemble of Improved Neural Operators (ENO), which uses bagging and Monte Carlo dropout techniques, to further improve prediction accuracy. Lastly, we developed the way to explain progressive transfer learning methods to reduce the amount of training data and computational cost of training (i.e., reduce trainable parameters) when using our models for multiple storage sites. Our numerical investigation, which used real-world case studies, demonstrated that our framework can significantly improve the safety and efficiency of geological storage operations, with potential applications in other domains such as geothermal reservoirs and climate modeling.

54 ENVIRONMENTAL SCIENCES↗