Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Model Predictive Control”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 1,513 records · Page 84
All sourcesNASA NTRSDOE OSTI
DOE OSTI2022

Development of Methane Emissions Model to Assess Fuel Recovery Potential at Gas Well Sites Using On Site Compression

The U.S. natural gas production and consumption has increased 85.5% since 2005 primarily due to the unconventional production methods of horizontal drilling and hydraulic fracturing. Natural gas used as a fuel has a lower greenhouse gas (GHG) footprint than coal and petroleum due to lower Carbon Dioxide (CO2) emissions when combusted. However, the “greener” benefit to natural gas may be negated by leaks in production and transmission systems. Methane (CH4), the primary hydrocarbon in natural gas, has an estimated Global Warming Potential (GWP) of 28-36 over 100 years, meaning it can absorb 28-36 more energy than CO2 which has a GWP of 1.0. Natural gas well sites are prone to methane emissions, or leaks and irregular gas releases, vented to atmosphere throughout production and transmission. The U.S. Department of Energy (DOE) and the National Energy Technology Laboratory (NETL) has recently granted West Virginia University (WVU) funding under agreement DE-FOA-0002005, to “Advance technologies to mitigate methane emissions and increase the efficiency of the natural gas transportation infrastructure”. As part of this funding WVU was tasked with identifying and quantifying sources of methane emissions at unconventional well sites, processing this data, and developing a system to recapture these emissions. A 0-D Simulink model was developed, utilizing standardized methodologies, data from previously conducted studies, as well as collected data from well sites in the Marcellus shale play region. The model was developed to predict emission rates from various components at natural gas well sites as well as the potential to utilize these emissions as fuel for the natural gas powered compressor engines on-site. This model was utilized to run high, medium, and low cases for four identified emission sources, engine size, pneumatic controller count, liquid level production which dictates tank emissions, and compressor packing vent emissions. Due to discrepancies in transient tank emission data, a high and low emission factor for tanks was used, resulting in two sets of 81 executed cases, and 162 unique cases of total site emissions and potential for fuel consumption. Each of the cases were run over 86,400 seconds at a 1 Hz, representative of a full 24 hour day of operation. The fuel consumption offset an average of 557% of fuel consumption on an energy density basis across all 81 cases with the high tank emission factor with a maximum offset of 2334%. The fuel consumption offset was an average of 82.9% for all 81 cases with the low tank emission factor with a maximum offset of 337%. This study highlights flaws in the use of publicly available methane number calculations to determine natural gas’s suitability as an engine fuel as well as the lack of public data for transient liquid storage tank emissions.

03 NATURAL GAS↗
DOE OSTI2023

Development of a High-Pressure Fuel Injection System for Use with Propane-DME

Paper details the design approach and performance of a high-pressure common rail fuel injection system used with Propane-DME mixtures targeting high injection pressures on a light duty 2.2L inline-4 compression ignition engine. The study estimates the bulk modulus of elasticity based on the hardware geometry and operating pressures to assess the compressibility of Propane and DME across a range of pressures typical of the LD engine application. The compressibility factor ranges from 500 to 3000 bar, significantly lower than the theoretical values for the conditions tested. The high-pressure pump performance is optimized via the implementation of an inlet metering valve operated on a crank-angle open/close sequence to control pressure at the common rail. The application of a model-based controller and the use of high-speed sampling of rail pressure indicate that pressure at the rail can be attained with the pump metering valve, and without use of the pressure relieve valve present in the rail. Results show a significant reduction on pumping work parasitics and of fuel heating, on the order of 40°C, at typical working conditions with the proposed approach. Injector rates of injection with Propane-DME are very similar to those of Diesel. Paper compares bench data of rates of injection versus simulation predictions across operating pressures of 200 to 1000bar and from injection commands spanning 0.20 to 2.0ms. Furthermore, the accuracy of the model predictions are attributed in part by to the updated fuel compressibility modulus descriptions attained during the testing.

42 ENGINEERING↗
DOE OSTI2020

Toward the Prediction and Control of Glass Transition Temperature for Donor–Acceptor Polymers

Semiconducting donor–acceptor (D–A) polymers have attracted considerable attention toward the application of organic electronic and optoelectronic devices. However, a rational design rule for making semiconducting polymers with desired thermal and mechanical properties is currently lacking, which greatly limits the development of new polymers for advanced applications. Here, polydiketopyrrolopyrrole (PDPP)-based D–A polymers with varied alkyl side-chain lengths and backbone moieties are systematically designed, followed by investigating their thermal and thin film mechanical responses. The experimental results show a reduction in both elastic modulus and glass transition temperature (Tg) with increasing side-chain length, which is further verified through coarse-grained molecular dynamics simulations. Informed from experimental results, a mass-per-flexible bond model is developed to capture such observation through a linear correlation between Tg and polymer chain flexibility. Using this model, a wide range of backbone Tg over 80 °C and elastic modulus over 400 MPa can be predicted for PDPP-based polymers. This study highlights the important role of side-chain structure in influencing the thermomechanical performance of conjugated polymers, and provides an effective strategy to design and predict Tg and elastic modulus of future new D–A polymers.

36 MATERIALS SCIENCE↗
DOE OSTI2025

Enhancing Turbulent Mixing and Microphysical Uniformity in a Tall Convection‐Cloud Chamber Through Idealized Heterogeneity of Boundaries

A large convection cloud chamber has been proposed for exploring aerosol–cloud–drizzle interactions under well‐controlled turbulent conditions. Recent theoretical and numerical studies suggest that a convection cloud chamber with two heated and two cooled sidewalls can significantly enhance the liquid water content and thus benefit drizzle initiation. However, a chamber with such a sidewall configuration develops stable stratification and extremely weak turbulence therein. In this study, we conduct large‐eddy simulations of a tall convection chamber with five different sidewall configurations consisting of alternating warm and cold patches. For each configuration, the total surface area of warm patches equals that of cold patches, resulting in the same expected cloud‐free supersaturation based on a flux budget model. Results show that changing the sidewall configuration, while keeping all other factors constant, can substantially enhance turbulent mixing and improve the uniformity of thermodynamic and cloud microphysical properties in the bulk region of the chamber. In addition, turbulence strength is positively correlated with liquid water content and negatively correlated with cloud droplet number concentration, consistent with theoretical predictions. Our results highlight the advantage of building a large cloud chamber using modular patches with individually controllable temperature and humidity to achieve well‐mixed conditions.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2025

ComStock Measure Documentation: Lighting Control for Load Shedding

This report describes the modeling methodology for a single end-use savings shape measure - lighting control for load shedding - and briefly introduces key results. The lighting control for load shedding measure applies lighting dimming control to reduce the lighting load during the building's electricity peak window every weekday. The measure takes daily peak load schedule inputs generated by the method "Dispatch Schedule Generation" described in the "Supplemental Documentation: Dispatch Schedule Generation for Demand Flexibility Measures" to determine the start and end times of the predicted peak window, and then adjusts the lighting dimming level by a percentage reduction from the original schedules during the peak window to reduce the peak demand. The measure is applicable to (large, medium and small) offices, warehouses, and primary and secondary schools, which correspond to approximately 68% of the stock floor area of commercial buildings in ComStock analysis. The measure demonstrates 2%-7% daily peak demand reduction performance for applicable buildings, and 0.43% total site energy savings (0 trillion British thermal units [TBtu]) for the U.S. commercial building stock modeled in ComStock.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗
DOE OSTI2023

Physics-Based Real-time Analysis and Control to Achieve Transient-Free Operations for the ITER Era (Final Report)

This is the final report for the early career award for Prof. Egemen Kolemen that aimed to develop the physics and technical basis of real-time transient – i.e. disruptions and Edge Localized Modes (ELMs) – avoidance and feedback control at the DIII-D nuclear fusion facility for the ITER era. During this project, we reconstructed high quality real-time DIII-D equilibria using kinetic information from the motional Stark effect (MSE), Thomson Scattering (TS), and Charge Exchange Recombination (CER) spectroscopy. These equilibria were analyzed in real-time to determine the low-n ideal MHD stability and the pedestal stability thresholds, and to obtain plasma response models. These calculations were used to understand which physical process evolves to lead to the onset of disruption in ITER-relevant scenarios at DIII-D, and how disruption can be predicted using stability analysis as opposed to the current methods of correlation of various signals, which are descriptive but lack predictive power. In the disruption studies, we focused on tearing modes, since these constitute the vast majority of disruptions in this regime. With this knowledge, real-time disruption avoidance algorithms were developed and tested. The enhanced situation awareness and physics insight were used in advanced real-time feedback control of pedestal and core profiles, which allowed the pursuit and sustainment of transient-free high-confinement regimes that cannot be achieved via feed-forward design based on in-between shot analysis.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗
DOE OSTI2024

Plant responses to elevated CO 2 under competing hypotheses of nitrogen and phosphorus limitations

Abstract The future ecosystem carbon cycle has important implications for biosphere‐climate feedback. The magnitude of future plant growth and carbon accumulation depends on plant strategies for nutrient uptake under the stresses of nitrogen (N) versus phosphorus (P) limitations. Two archetypal theories have been widely acknowledged in the literature to represent N and P limitations on ecosystem processes: Liebig's Law of the Minimum (LLM) and the Multiple Element Limitation (MEL) approach. LLM states that the more limiting nutrient controls plant growth, and commonly leads to predictions of dramatically dampened ecosystem carbon accumulation over the 21st century. Conversely, the MEL approach recognizes that plants possess multiple pathways to coordinate N and P availability and invest resources to alleviate N or P limitation. We implemented these two contrasting approaches in the E3SM model, and compiled 98 in situ forest N or P fertilization experiments to evaluate how terrestrial ecosystems will respond to N and P limitations. We find that MEL better captured the observed plant responses to nutrient perturbations globally, compared with LLM. Furthermore, LLM and MEL diverged dramatically in responses to elevated CO 2 concentrations, leading to a two‐fold difference in CO 2 fertilization effects on Net Primary Productivity by the end of the 21st century. The larger CO 2 fertilization effects indicated by MEL mainly resulted from plant mediation on N and P resource supplies through N 2 fixation and phosphatase activities. This analysis provides quantitative evidence of how different N and P limitation strategies can diversely affect future carbon and nutrient dynamics.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2026

From MAST-U to STEP: Power exhaust control challenges and opportunities

This work explores the challenges and opportunities for power exhaust control in STEP, informed by dedicated MAST-U experiments. The STEP system primarily relies on hydrogenic and argon gas injection into the divertors, compensating for transients originating in the plasma core. Fast transients (>10 Hz) that evolve too quickly for actuators to compensate must be absorbed passively; the enhanced transient buffering of the foreseen long-legged divertor provides a clear advantage over conventional geometries, as demonstrated in the MAST-U Super-X divertor. Experiments further indicate that STEP’s tight divertor baffling enables near-independent control of the upper and lower divertors. The most challenging transients are expected from power-sharing fluctuations, as MAST-U experiments observed extremely fast dynamics. Fluctuations from core pellet fuelling remain relatively benign due to the small pellet size according to simulations. Exhaust control in a reactor like STEP requires an integrated approach befitting its machine-critical nature. A predictive control element, integrated with core plasma control, allows pre-emptive preparation of the divertor for incoming transients. An observer-based diagnostic approach is foreseen to monitor the divertor in the challenging reactor environment, supported by an extended diagnostic set in the non-nuclear phase to validate the required dynamic models.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗
DOE OSTI2025

Grain2mesh: A Python and cubit mesh generator from unprocessed mesoscale images

Predicting bulk behavior from microscale features constitutes a key objective in multiscale modeling research, often involving numerical models composed of finite elements that capture the diversity of constituent phases, shapes, and orientations within the material. The Grain2mesh toolbox allows the user to input unprocessed mesoscopic images for automatic segmentation, pre-processing, quality control, and numerical mesh generation. The numerical mesh generation incorporates Cubit routines to generate robust multi-phase mesh structure for use in computational mechanics solvers. The python classes developed contain detailed documentation and examples to support standard usage and case-specific alternative options.

58 GEOSCIENCES↗
DOE OSTI2022

Selenite and Selenate Sequestration during Coprecipitation with Barite: Insights from Mineralization Processes of Adsorption, Nucleation, and Growth

Coprecipitation of selenium oxyanions with barite is a facile way to sequester Se in the environments. However, the chemical composition of Se-barite coprecipitates usually deviates from that predicted from thermodynamic calculations. This discrepancy was resolved by considering variations in nucleation and growth rates controlled by ion–mineral interactions, solubility, and interfacial energy. For homogeneous precipitation, ~10% of sulfate, higher than thermodynamic predictions (<0.3%), was substituted by Se(IV) or Se(VI) oxyanion, which was attributed to adsorption-induced entrapment during crystal growth. For heterogeneous precipitation, thiol- and carboxylic-based organic films, utilized as model interfaces to mimic the natural organic-abundant environments, further enhanced the sequestration of Se(VI) oxyanions (up to 41–92%) with barite. Such enhancement was kinetically driven by increased nucleation rates of selenate-rich barite having a lower interfacial energy than pure barite. In contrast, only small amounts of Se(IV) oxyanions (~1%) were detected in heterogeneous coprecipitates mainly due to a lower saturation index of BaSeO 3 and deprotonation degree of Se(IV) oxyanion at pH 5.6. Finally, these roles of nanoscale mineralization mechanisms observed during composition selection of Se-barite could mark important steps toward the remediation of contaminants through coprecipitation.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2021

A mesoscopic digital twin that bridges length and time scales for control of additively manufactured metal microstructures

We present our recent development of an integrated mesoscale digital twin (DT) framework for relating processing conditions, microstructures, and mechanical responses of additively manufactured (AM) metals. In particular, focusing on the laser powder bed fusion technique, we describe how individual modeling and simulation capabilities are coupled to investigate and control AM microstructural features at multiple length and time scales. We review our prior case studies that demonstrate the integrated modeling schemes, in which high-fidelity melt pool dynamics simulations provide accurate local thermal profiles and histories to subsequent AM microstructure simulations. We also report our new mechanical response modeling results for predicted AM microstructures. In addition, we illustrate how our DT framework has been validated through modeling–experiment integration, as well as how it has been practically utilized to guide and analyze AM experiments. Finally, we share our perspectives on future directions of further development of the DT framework for more efficient, accurate predictions and wider ranges of applications.

36 MATERIALS SCIENCE↗
DOE OSTI2023

Robust Parameter Design on Dual Stochastic Response Models With Constrained Bayesian Optimization

In engineering system design, minimizing the variations of the quality measurements while guaranteeing their overall quality up to certain levels, namely the robust parameter design (RPD), is crucial. Recent works have dealt with the design of a system whose response-control variables relationship is a deterministic function with a complex shape and function evaluation is expensive. In this work, we propose a Bayesian optimization method for the RPD of stochastic functions. Dual stochastic response models are carefully designed for stochastic functions. The heterogeneous variance of the sample mean is addressed by the predictive mean of the log variance surrogate model in a two-step approach. We establish an acquisition function that favors exploration across the feasible and optimality-improvable regions to effectively and efficiently solve the stochastic constrained optimization problem. Further, the performance of our proposed method is demonstrated by the extensive numerical and case studies. Note to Practitioners-Many manufacturing processes involve undesirable variations, which create variations in the final products. For example, many emerging manufacturing processes, such as nanomanufacturing, involve complex physical and chemical dynamics and transformation, creating variations in the manufacturing output. In such processes, it is crucial to design the manufacturing processes or products so that they have minimum variations in their quality. Meanwhile, it is also important to maintain the overall quality of the designed processes or products. Furthermore, acquiring data from many advanced manufacturing processes is often very costly, especially in the designing stage. In this work, we propose a data-driven method that automatically finds the best setting of manufacturing processes or products with the minimum variations of quality and a given constraint on the average quality satisfied. Our proposed method is used before conducting every experiment; It analyzes the historical data from previous experiments and provides a setting to be used in the next experiment. Our proposed method efficiently utilizes the historical data, and thus finds the best robust setting by conducting only a small number of experiments.

42 ENGINEERING↗
DOE OSTI2020

Center for the Geologic Storage of CO 2 (GSCO2) (Final Report)

The Center for Geologic Storage of CO2 (GSCO2) focuses on basic-science approaches to recognized industry and technology challenges for commercial deployment of carbon dioxide (CO2) storage in clastic reservoirs. With this focus, the GSCO2 strives to generate new conceptual, mathematical, and numerical models applicable to geologic storage systems in specific and strategically identified research areas. Over the four-year research period (plus two-year no-cost extension), the GSCO2 developed its focus on one central research question: What is the mechanism of injection-induced microseismicity, and can we control and predict its occurrence?

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2021

Network Based Estimation of Wind Farm Power and Velocity Data Under Changing Wind Direction

This paper describes an estimation algorithm for velocity and power output signals in a wind farm under changing wind direction. A graph-theoretic definition describes the wind farm as a collection of nodes (turbines) and time-varying weighted edges (inter-turbine wake propagation) that change as a function of incoming wind direction. The velocity at each turbine is determined through a discrete input-output model. Changes in wind direction serve as the input and the output is defined in terms of a time-varying weighted adjacency matrix that depends on the time-delay of information propagation between turbines. These delays, which are defined in terms of the advection speed of the wind and the distance between the turbines, capture the delayed effect of wind direction changes on the inter-connectivity of the graph as the wind conditions at the farm inlet propagate through the turbine array. An event-based update framework is employed to capture time-dependent topology changes due to shifts in wind direction. Simulation results for dynamically changing wind inlet directions to a circular wind farm are compared to predictions from both the static and dynamic versions of the FLOw Redirection and Induction in Steady State (FLORIS) model. The approach is shown to enable real-time tracking of dynamic changes to wind farm power output within a framework that can be easily integrated into real-time, horizon-based, control strategies that typically do not account for wind direction changes.

distributed↗
DOE OSTI2025

Convergent Manufacturing of Large-Scale Components for Nuclear Applications, via Additive Manufacturing and Powder Metallurgy Hot Isostatic Pressing

Powder metallurgy (PM)–hot isostatic pressing (PM-HIP) has long been recognized as a powerful route for producing fully dense, near net shape metallic components. By consolidating powders under high temperature and pressure, HIP provides isotropic properties, uniform microstructures, and scalability to complex geometries that are vital for sectors such as aerospace, energy, and nuclear power. Yet despite these advantages, the technology has remained constrained by costly trial and error canister fabrication, limitations of conventional forging, and incomplete knowledge about how the canister design influences final part properties. Additive manufacturing (AM), by contrast, thrives on design freedom and geometric flexibility but struggles with speed, scalability, and cost when applied to very large structures. The research presented in this report investigated how a convergent manufacturing approach, combining AM with PM-HIP, can merge the strengths of both technologies, leveraging AM’s flexibility for canister design and HIP’s consolidation capability to deliver reliable, large, and complex parts. The work progressed through three case studies that built on one another in scale and complexity. Small cylindrical canisters fabricated by conventional methods, laser powder bed fusion, and directed energy deposition were filled with stainless steel powders and subjected to HIP. The resulting parts demonstrated near-full density and mechanical properties on par with wrought stainless steel, showing for the first time that AM canisters can be a direct substitute for conventional ones without sacrificing quality. The next step involved a medium-scale, noncentrosymmetric T-valve, which is an enclosed, multibranch geometry that tested the limits of AM + PM-HIP integration. The T-valve achieved predictable shrinkage and uniform densification, confirming feasibility for enclosed designs. However, this study also revealed oxide inclusions and interfacial challenges at the AM + HIP boundary, underscoring the critical importance of controlling interface chemistry and employing robust, in situ strategies, such as melt pool monitoring and thermal monitoring, coupled with nondestructive evaluation techniques such as x-ray computed tomography. Finally, the effort culminated in fabricating a large-scale impeller weighing nearly 2000 lb and spanning 5 ft in diameter. Produced via multirobot wire arc AM and hot isostatic pressed to near-full density, the impeller validated industrial-scale feasibility. Predictive models closely matched experimental shrinkage, tensile properties were spatially uniform across the component, and the AM + PM-HIP interface proved mechanically sound despite the presence of oxide-decorated prior particle boundaries. This large-scale demonstration is a major milestone, showing that hybrid AM + PM‑HIP can reliably deliver components at reactor-relevant scales. Collectively, these studies charted a logical pathway: small-scale work built scientific confidence, medium-scale work highlighted opportunities and challenges, and large-scale work proved industrial impact. The overarching conclusion of this report is that AM + PM-HIP should not be seen as a replacement for forging but as a complementary pathway that provides the US with flexibility, resilience, and new options for manufacturing nuclear-grade components. Looking ahead, several directions emerge as critical to sustaining progress. Predictive modeling must become faster, more accessible, and more accurate, with digital twins and machine learning reducing reliance on trial and error. Powders and alloys must be optimized for HIP, with improved cleanliness, reduced oxides, and tailored chemistries that enhance creep, fatigue, and irradiation resistance. Interfaces between AM and HIP regions must be better engineered through coatings, machining strategies, and surface treatments to mitigate oxide formation and ensure reliable bonding to explore opportunities for HIP of targeted compositional parts, as well as multimaterial HIP cladding applications. Monitoring and nondestructive evaluation need to expand, incorporating multimodal sensors, x-ray computed tomography, and real-time data integration through platforms such as Pelican. At the same time, the pathway to industrial adoption requires techno-economic analysis, machinability studies, and qualification frameworks aligned with industry and regulatory standards. Finally, workforce and academic engagement must be strengthened. Programs that train technicians and engineers for US Navy and US Department of Energy manufacturing challenges should be paired with academic partnerships to support fundamental research, with open sharing of non-export-controlled data to accelerate innovation and build the next generation of experts. In conclusion, this report demonstrates that hybrid AM + PM-HIP is scientifically viable and strategically important. By combining the design agility of AM with the consolidation strength of HIP and embedding modeling, monitoring, and workforce development, this approach provided a transformative new capability for US manufacturing. The path forward is clear: hybrid AM + PM-HIP is not just a promising research direction but is also potentially an industrially relevant pathway that can reshape how nuclear-grade components are designed, qualified, and deployed.

36 MATERIALS SCIENCE↗
DOE OSTI2023

Fast and accurate reduced-order modeling of a MOOSE-based additive manufacturing model with operator learning

One predominant challenge in additive manufacturing (AM) is to achieve specific material properties by manipulating manufacturing process parameters during the runtime. Such manipulation tends to increase the computational load imposed on existing simulation tools employed in AM. The goal of the present work is to construct a fast and accurate reduced-order model (ROM) for an AM model developed within the Multiphysics Object-Oriented Simulation Environment (MOOSE) framework, ultimately reducing the time/cost of AM control and optimization processes. Our adoption of the operator learning (OL) approach enabled us to learn a family of differential equations produced by altering process variables in the laser’s Gaussian point heat source. More specifically, we used the Fourier neural operator (FNO) and deep operator network (DeepONet) to develop ROMs for time-dependent responses. Furthermore, we benchmarked the performance of these OL methods against a conventional deep neural network (DNN)-based ROM. Ultimately, we found that OL methods offer comparable performance and, in terms of accuracy and generalizability, even outperform DNN at predicting scalar model responses. The DNN-based ROM afforded the fastest training time. Furthermore, all the ROMs were faster than the original MOOSE model yet still provided accurate predictions. FNO had a smaller mean prediction error than DeepONet, with a larger variance for time-dependent responses. Unlike DNN, both FNO and DeepONet were able to simulate time series data without the need for dimensionality reduction techniques. Finally, the present work can help facilitate the AM optimization process by enabling faster execution of simulation tools while still preserving evaluation accuracy.

36 MATERIALS SCIENCE↗
DOE OSTI2024

Enhancing Fluid Flow Pressure and Saturation Prediction Accuracy and Reducing Uncertainty with Committee Machine – Illinois Basin Decatur Project (IBDP) as a Case Study

Presentation at the 17th International Conference on Greenhouse Gas Control Technologies GHGT-17 held in Calgary, Canada, October 20-24, 2024. Carbon capture and storage (CCS) is a way to play a critical role in the global transition to a low-emission economy. Current progress is hampered by a number of factors, among which the lack of risk-informed design tools and decision support frameworks is seen as a major roadblock. Significant interest exists in using artificial intelligence to accelerate CCS site feasibility studies, as well as to facilitate the permit application process. Existing works commonly train a single deep learning model. This work investigates the feasibility of using a conventional ensemble learning (committee machine) technique to further improve prediction accuracy. Ensemble-based algorithms generally improve over individual base learners in terms of robustness and accuracy. Deep ensembles, however, are time-consuming to create and train. A pragmatic question is whether small-sized ensembles may lead to prediction improvement. Here we evaluated the efficacy of an ensemble learning technique using the latent spectral model (LSM), an efficient deep neural operator algorithm, as base learners. Preliminary results, obtained using the Illinois Basin-Decatur Project (IBDP) carbon sequestration data/model, show that small-sized ensembles can improve prediction over the base learners, achieving prediction accuracy of ~1.6 psi root mean square error (RMSE) on pressure (relative the average reservoir pressure of 3150 psi), and less than 1.3% for saturation.

Sun, Alexander↗
DOE OSTI2021

Building Intelligent Cyberinfrastructure to Learn Iteratively from both Observations and Models for Understanding Watershed Dynamics

Focal Area(s): Predictive modeling through the use of AI techniques and AI-derived model components; the use of AI and other tools to design a prediction system comprising of a hierarchy of models (e.g., AI-driven model/component/parameterization selection). Science Challenge: Watershed processes, such as the fate and transport of sediment, carbon and nutrients across landscapes and their fluxes to water bodies (e.g., streams, rivers and lakes), have important implications for global and regional carbon and nutrient dynamics, biogeochemical functioning of terrestrial ecosystems, and soil functions. The magnitude of lateral surface/subsurface transport and fluxes of sediment, carbon and nutrients are key factors controlling the vulnerability of watersheds to climate extremes such as droughts, wildfires, and floods. Recent field observations and other scientific evidence suggest that the magnitudes of lateral transport and fluxes of sediment, carbon and nutrients are governed primarily by the spatial and vertical heterogeneity of landscape and soil properties and by pedogenic processes. However, the current generation of land surface and watershed models do not mechanistically couple the terrestrial and hydrologic systems, nor do they represent sufficiently the spatial and vertical heterogeneity of land surface and subsurface properties. On the other hand, increasing complexity of coupled watershed and land surface models requires more data to parameterize, calibrate and validate. Remote sensing (RS) provides a means to acquire spatial data and characterize their heterogeneity at the watershed scale, overcoming a major limitation associated with conventional point measurements. To improve the representation of land-surface and surface/subsurface process coupling and sub-grid heterogeneity in watershed models, it is essential to build our predictive understanding by learning from both the multi-scale multi-process modeling and diverse multi-scale data while leveraging powerful artificial intelligence (AI) techniques.

54 ENVIRONMENTAL SCIENCES↗