Search NASA⌕ Search

SEARCH · Search NASA

Results for “Distributionally robust optimization”

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

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

At least 559 records · Page 31

A remote sensing algorithm for vertically resolved cloud condensation nuclei number concentrations from airborne and spaceborne lidar observations

Cloud condensation nuclei (CCN) are mediators of aerosol–cloud interactions (ACIs), contributing to the largest uncertainties in the understandings of global climate change. We present a novel remote-sensing-based algorithm that quantifies the vertically resolved CCN number concentrations (N CCN ) using aerosol optical properties measured by a multiwavelength lidar. The algorithm considers five distinct aerosol subtypes with bimodal size distributions. The inversion used the lookup tables developed in this study, based on the observations from the Aerosol Robotic Network, to efficiently retrieve optimal particle size distributions from lidar measurements. The method derives dry aerosol optical properties by implementing hygroscopic enhancement factors in lidar measurements. The retrieved optically equivalent particle size distributions and aerosol-type-dependent particle composition are utilized to calculate critical diameters using κ-Köhler theory and N CCN at six supersaturations ranging from 0.07 % to 1.0 %. Sensitivity analyses indicate that uncertainties in extinction coefficients and relative humidity greatly influence the retrieval error in N CCN . The potential of this algorithm is further evaluated by retrieving N CCN using airborne lidar from the NASA ObseRvations of Aerosols above CLouds and their intEractionS (ORACLES) campaign and is validated against simultaneous measurements from the CCN counter. The independent validation with robust correlation demonstrates promising results. Furthermore, the N CCN has been retrieved for the first time using a proposed algorithm from spaceborne lidar – Cloud-Aerosol Lidar with Orthogonal Polarization (CALIOP) – measurements. The application of this new capability demonstrates the potential for constructing a 3D CCN climatology at a global scale, which helps to better quantify ACI effects and thus reduce the uncertainty in aerosol climate forcing.

54 ENVIRONMENTAL SCIENCES↗

BETO 2021 Peer Review - FCIC Task 6: High Temperature Conversion

The impacts of feedstock variability on pyrolysis processes are significant but poorly defined. Current engineering designs are based on empirical guidelines, useful only over a narrow range of feedstock properties. The objectives of this project are to (1) Develop science-based knowledge of how feedstock attributes and operational parameters impact pyrolysis process reliability and product quality; and (2) Build an experimental and computational toolset that predicts these outcomes, enabling processes to optimize reliability and product quality. Biomass is a complex feedstock. Controlling for and testing the effects of individual attributes is very challenging. This project couples multiscale experimentation and modeling to accurately capture the fundamental physics and chemistry of biomass flow and conversion behavior in feeding and pyrolysis reactor operations. Our focus is on pine residue attributes – anatomical fraction (bark, needles, wood), particle morphology (size/shape distribution, density, porosity), and chemical composition (extractives, biopolymers, alkali metals) – that impact product quality for downstream catalytic upgrading. Because detailed pyrolysis product characterization is limited, cutting-edge analytical techniques are being developed to reveal impactful product attributes. The tools and knowledge developed here will enable integrated pyrolysis-based processes that are more robust, flexible, and market-responsive with respect to feedstock variability.

29 ENERGY PLANNING, POLICY, AND ECONOMY↗

NASA Tech Briefs, July 2009

Topics covered include: Dual Cryogenic Capacitive Density Sensor; Hail Monitor Sensor; Miniature Six-Axis Load Sensor for Robotic Fingertip; Improved Blackbody Temperature Sensors for a Vacuum Furnace; Wrap-Around Out-the-Window Sensor Fusion System; Wide-Range Temperature Sensors with High-Level Pulse Train Output; Terminal Descent Sensor Simulation; A Robust Mechanical Sensing System for Unmanned Sea Surface Vehicles; Additive for Low-Temperature Operation of Li-(CF)n Cells; Li/CFx Cells Optimized for Low-Temperature Operation; Number Codes Readable by Magnetic-Field-Response Recorders; Determining Locations by Use of Networks of Passive Beacons; Superconducting Hot-Electron Submillimeter-Wave Detector; Large-Aperture Membrane Active Phased-Array Antennas; Optical Injection Locking of a VCSEL in an OEO; Measuring Multiple Resistances Using Single-Point Excitation; Improved-Bandwidth Transimpedance Amplifier; Inter-Symbol Guard Time for Synchronizing Optical PPM; Novel Materials Containing Single-Wall Carbon Nanotubes Wrapped in Polymer Molecules; Light-Curing Adhesive Repair Tapes; Thin-Film Solid Oxide Fuel Cells; Zinc Alloys for the Fabrication of Semiconductor Devices; Small, Lightweight, Collapsible Glove Box; Radial Halbach Magnetic Bearings; Aerial Deployment and Inflation System for Mars Helium Balloons; Steel Primer Chamber Assemblies for Dual Initiated Pyrovalves; Voice Coil Percussive Mechanism Concept for Hammer Drill; Inherently Ducted Propfans and Bi-Props; Silicon Nanowire Growth at Chosen Positions and Orientations; Detecting Airborne Mercury by Use of Gold Nanowires; Detecting Airborne Mercury by Use of Palladium Chloride; Micro Electron MicroProbe and Sample Analyzer; Nanowire Electron Scattering Spectroscopy; Electron-Spin Filters Would Offer Spin Polarization Greater than 1; Subcritical-Water Extraction of Organics from Solid Matrices; A Model for Predicting Thermoelectric Properties of Bi2Te3; Integrated Miniature Arrays of Optical Biomolecule Detectors; A Software Rejuvenation Framework for Distributed Computing; Kurtosis Approach to Solution of a Nonlinear ICA Problem; Robust Software Architecture for Robots; R4SA for Controlling Robots; Bio-Inspired Neural Model for Learning Dynamic Models; Evolutionary Computing Methods for Spectral Retrieval; Monitoring Disasters by Use of Instrumented Robotic Aircraft; Complexity for Survival of Living Systems; Using Drained Spacecraft Propellant Tanks for Habitation; Connecting Node; and Electrolytes for Low-Temperature Operation of Li-CFx Cells.

Source record↗

Day-Ahead Optimal Operation for Multi-Energy Residential Systems with Renewables

The intermittency and stochasticity nature of distributed renewable energy sources has introduced great challenges to the efficiency and security of energy distribution system operations. To address the negative impacts of intermittent renewable energy sources, this paper proposes here a day-ahead optimal operation strategy utilizing distributed energy resources based on the framework of the interconnected multi-energy system. First, a framework and mathematical models of multi-energy residential systems (MERS) are proposed. Based on the characteristics of residential energy distribution networks, the complex MERS models are reformulated to relieve the computational burden. Furthermore, the uncertainty factors such as renewable energy generation fluctuations and demand variations are handled by a reformulated chance con-strained programming technique. The feasibility and effectiveness of the proposed method are validated through a combined electric power and natural gas test system. Compared to similar models and methods in the existing literature, the proposed method performs better in terms of solution time, model scalability, and robustness in handling uncertainties.

24 POWER TRANSMISSION AND DISTRIBUTION↗

DGaaS: GPU as a Service on Distributed Computing System

In the rapidly evolving landscape of scientific computing, Graphics Processing Units (GPUs) have become indispensable for their unparalleled ability to handle parallel tasks in complex calculations, simulations, and data analysis. Their utility is further magnified in machine learning and AI applications, where they significantly accelerate model training and predictive analytics. Within this context, the Triton Inference Server emerges as a pivotal open-source tool, specializing in AI inferencing and optimizing GPU utilization across various platforms and frameworks. This paper presents an in-depth study on distributed High Throughput Computing (HTC), specifically focusing on the HTCondor framework and its resource provisioning tools, GlideinWMS and HEPCloud. These systems enable large-scale scientific experiments like CMS and DUNE to efficiently access and utilize vast computational resources. The paper explores the core architectural components of GlideinWMS, including jobs, user pools, and worker nodes, and discusses their integration with GPUs and the Triton server. The primary aim of this research is to develop a solution that optimizes GPU utilization by leveraging Glideins and containers. This approach allows computational jobs, particularly those involving AI models, to use GPUs only when essential, thereby facilitating efficient sharing of limited GPU resources. To validate this architecture, the study conducted three key tests involving custom scripts, container-based servers, and Triton server deployments. However, the study faces challenges, notably in locating the Triton server and ensuring secure remote access. To address these issues, future work will focus on developing a proxy mechanism and enhancing security protocols. In conclusion, this study offers a comprehensive roadmap for effective and efficient GPU utilization in distributed High Throughput Computing. It aims to contribute significantly to the scientific community by solving pressing problems and implementing robust solutions in collaboration with the GlideinWMS and HEPCloud teams. The research sets the stage for a more efficient, scalable, and cost-effective paradigm in scientific computing.

97 MATHEMATICS AND COMPUTING↗

Nanostructured Block Polymer Electrolytes: Tailoring Self-Assembly to Unlock the Potential in Lithium-Ion Batteries

Ion-containing solid block polymer (BP) electrolytes can self-assemble into microphase-separated domains to facilitate the independent optimization of ion conduction and mechanical stability; this assembly behavior has the potential to improve the functionality and safety of lithium-ion batteries over liquid electrolytes to meet future demands (e.g., large capacities and long lifetimes) in various applications. However, significant enhancements in the ionic conductivity and processability of BPs must be realized for BP-based electrolytes to become robust alternatives in commercial devices. Toward this end, the controlled modification of BP electrolytes’ intra-domain (nanometer-scale) and multi-grain (micrometer-scale) structure is one viable approach; intra-domain ion transport and segmental compatibility (related to the effective Flory–Huggins parameter, χ eff ) can be increased by tuning the ion and monomer-segment distributions, and the morphology can be selected such that the multi-grain transport is less sensitive to grain size and orientation.

25 ENERGY STORAGE↗

Improving wafer-scale Josephson junction resistance variation in superconducting quantum coherent circuits

Quantum bits, or qubits, are an example of coherent circuits envisioned for next-generation computers and detectors. A robust superconducting qubit with a coherent lifetime of O (100 µs) is the transmon: a Josephson junction functioning as a non-linear inductor shunted with a capacitor to form an anharmonic oscillator. In a complex device with many such transmons, precise control over each qubit frequency is often required, and thus variations of the junction area and tunnel barrier thickness must be sufficiently minimized to achieve optimal performance while avoiding spectral overlap between neighboring circuits. Simply transplanting our recipe optimized for single, stand-alone devices to wafer-scale (producing 64, 1x1 cm dies from a 150 mm wafer) initially resulted in global drifts in room-temperature tunneling resistance of ± 30%. Inferring a critical current variation from this resistance distribution, we present an optimized process developed from a systematic 38 wafer study that results in < 3.5% relative standard deviation (RSD) in critical current ( ) for 3000 Josephson junctions (both single-junctions and asymmetric SQUIDs) across an area of 49 cm 2 . Looking within a 1x1 cm moving window across the substrate gives an estimate of the variation characteristic of a given qubit chip. Our best process, utilizing ultrasonically assisted development, uniform ashing, and dynamic oxidation has shown = 1.8% within 1x1 cm, on average, with a few 1x1 cm areas having < 1.0% (equivalent to < 0.5%). Such stability would drastically improve the yield of multi-junction chips with strict critical current requirements.

75 CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND↗

SU-E-T-130: Are Proton Gantries Needed? An Analysis of 4332 Patient Proton Gantry Treatment Plans From the Past 10 Years

Purpose: To ascertain the necessity of a proton gantry, as compared to the feasibility of using a horizontal fixed proton beam-line for treatment with advanced technology. Methods: To calculate the percentage of patients that can be treated with a horizontal fixed beam-line instead of a gantry, we analyze the distributions of beam orientations of our proton gantry patients treated over the past 10 years. We identify three horizontal fixed beam geometries (FIXED, BEND and MOVE) with the patient in lying and/or sitting positions. The FIXED geometry includes only table/chair rotations and translations. In BEND, the beam can be bent up/down for up to 20 degrees. MOVE allows for patient head/body angle adjustment. Based on the analysis, we select eight patients whose plan involves beams which are still challenging to achieve with a horizontal fixed beam. These beams are removed in the pencil beam scanning (PBS) plan optimized for the fixed beam-line (PBS-fix). We generate non-coplanar PBS-gantry plans for comparison, and perform a robustness analysis. Results: The percentage of patients with head-and-neck/brain tumors that can be treated with horizontal fixed beam is 44% in FIXED, 70% in 20-degrees BEND, and 100% in 90-degrees MOVE. For torso regions, 99% of the patients can be treated in 20-degree BEND. The target coverage is more homogeneous with PBS-fix plans compared to the clinical scattering treatment plans. The PBS-fix plans reduce the mean dose to organs-at-risk by a factor of 1.1–28.5. PBS-gantry plans are as good as PBS-fix plans, sometimes marginally better. Conclusion: The majority of the beam orientations can be realized with a horizontal fixed beam-line. Challenging non-coplanar beams can be eliminated with PBS delivery. Clinical implementation of the proposed fixed beam-line requires use of robotic patient positioning, further developments in immobilization, and image guidance. However, our results suggest that fixed beam-lines can be as effective as gantries.

60 APPLIED LIFE SCIENCES↗

Analysis of Mars Analogue Soil Samples Using Solid-Phase Microextraction, Organic Solvent Extraction and Gas Chromatography/Mass Spectrometry

Polycyclic aromatic hydrocarbons (PAHs) are robust and abundant molecules in extraterrestrial environments. They are found ubiquitously in the interstellar medium and have been identified in extracts of meteorites collected on Earth. PAHs are important target molecules for planetary exploration missions that investigate the organic inventory of planets, moons and small bodies. This study is part of an interdisciplinary preparation phase to search for organic molecules and life on Mars. We have investigated PAH compounds in desert soils to determine their composition, distribution and stability. Soil samples (Mars analogue soils) were collected at desert areas of Utah in the vicinity of the Mars Desert Research Station (MDRS), in the Arequipa region in Peru and from the Jutland region of Denmark. The aim of this study was to optimize the solid-phase microextraction (SPME) method for fast screening and determination of PAHs in soil samples. This method minimizes sample handling and preserves the chemical integrity of the sample. Complementary liquid extraction was used to obtain information on five- and six-ring PAH compounds. The measured concentrations of PAHs are, in general, very low, ranging from 1 to 60 ng g(sup -1). The texture of soils is mostly sandy loam with few samples being 100% silt. Collected soils are moderately basic with pH values of 8-9 except for the Salten Skov soil, which is slightly acidic. Although the diverse and variable microbial populations of the samples at the sample sites might have affected the levels and variety of PAHs detected, SPME appears to be a rapid, viable field sampling technique with implications for use on planetary missions.

Orzechowska, G. E.↗

TIMELY DELIVERY OF LASER INERTIAL FUSION ENERGY (LIFE)

The National Ignition Facility (NIF), the world's largest and most energetic laser system, is now operational at Lawrence Livermore National Laboratory. A key goal of the NIF is to demonstrate fusion ignition for the first time in the laboratory. Its flexibility allows multiple target designs (both indirect and direct drive) to be fielded, offering substantial scope for optimization of a robust target design. In this paper we discuss an approach to generating gigawatt levels of electrical power from a laser-driven source of fusion neutrons based on these demonstration experiments. This 'LIFE' concept enables rapid time-to-market for a commercial power plant, assuming success with ignition and a technology demonstration program that links directly to a facility design and construction project. The LIFE design makes use of recent advances in diode-pumped, solid-state laser technology. It adopts the paradigm of Line Replaceable Units utilized on the NIF to provide high levels of availability and maintainability and mitigate the need for advanced materials development. A demonstration LIFE plant based on these design principles is described, along with the areas of technology development required prior to plant construction. A goal-oriented, evidence-based approach has been proposed to allow LIFE power plant rollout on a time scale that meets policy imperatives and is consistent with utility planning horizons. The system-level delivery builds from our prior national investment over many decades and makes full use of the distributed capability in laser technology, the ubiquity of semiconductor diodes, high volume manufacturing markets, and U.S. capability in fusion science and nuclear engineering. The LIFE approach is based on the ignition evidence emerging from NIF and adopts a line-replaceable unit approach to ensure high plant availability and to allow evolution from available technologies and materials. Utilization of a proven physics platform for the ignition scheme is an essential component of an acceptably low-risk solution. The degree of coupling seen on NIF between driver and target performance mandates that little deviation be adopted from the NIF geometry and beamline characteristics. Similarly, the strong coupling between subsystems in an operational power plant mandates that a self-consistent solution be established via an integrated facility delivery project. The benefits of separability of the subsystems within an IFE plant (driver, chamber, targets, etc.) emerge in the operational phase of a power plant rather than in its developmental phase. An optimized roadmap for IFE delivery needs to account for this to avoid nugatory effort and inconsistent solutions. For LIFE, a system design has been established that could lead to an operating power plant by the mid-2020s, drawing from an integrated subsystem development program to demonstrate the required technology readiness on a time scale compatible with the construction plan. Much technical development work still remains, as does alignment of key stakeholder groups to this newly emerging development option. If the required timeline is to be met, then preparation of a viable program is required alongside the demonstration of ignition on NIF. This will enable timely analysis of the technical and economic case and establishment of the appropriate delivery partnership.

42 ENGINEERING↗

Adaptive Reinforcement Learning Control for Power Distribution in Multi-Output Resonant Converters

This paper presents an adaptive reinforcement learning (ARL)-based control framework for efficient power distribution in a multi-output resonant converter for UAV applications. The proposed system is based on a high-frequency isolated resonant architecture, where a single energy source supplies multiple propulsion loads through independently controlled output rectifiers, addressing the need for coordinated multi-motor power management. The ARL framework dynamically allocates output power by learning optimal phase-shift control actions under varying load demands and operating conditions. The agent autonomously determines control parameters that maximize conversion efficiency while ensuring accurate power sharing among multiple outputs. In addition, the proposed approach enables adaptive operation without requiring detailed system modeling or manual tuning. Experimental results demonstrate stable and efficient performance over a wide range of operating conditions, confirming the effectiveness and robustness of the learning-based control strategy for multi-output resonant converter system.

Asa, Erdem [ORNL] (ORCID:0000000190884812)↗

Design of a SiC-Si moving packed-bed particle-to-sCO 2 heat exchanger for high temperature concentrating solar power applications

Particle-based concentrating solar power systems integrated with sCO 2 power cycles offer high thermal efficiencies but require durable heat exchangers to transfer heat from high-temperature particles to the sCO 2 working fluid. Here, this study presents the design and optimization of a silicon carbide-silicon moving packed-bed heat exchanger for fabrication via binder jetting additive manufacturing. The heat exchanger was designed to withstand a 20 MPa sCO 2 pressure and operate at particle inlet temperatures up to 750 °C. The final design features 152 sCO 2 channels distributed across 19 plates, with elliptical corners and a minimum wall thickness of 3 mm. Flow restrictors at the sCO 2 channel inlets significantly improved flow uniformity, reducing thermal stresses and achieving a structural reliability of 99 % under representative operating conditions. The heat exchanger delivers a thermal duty of 9 kW and a volumetric power density of approximately 1 MW/m 3 in the channel region. Sensitivity studies confirmed the heat exchanger’s robustness under varying operating conditions, demonstrating its viability as a high-performance alternative to metallic heat exchangers for particle-based high-temperature concentrating solar power applications.

Barua, Bipul [Argonne National Laboratory (ANL), A↗

Machine learning method for enforcing variable independence in background estimation with LHC data: ABCDisCoTEC

A novel solution is presented for the problem of estimating the backgrounds of a signal search using observed data while simultaneously maximizing the sensitivity of the search to the signal. The 'ABCD method' provides a reliable framework for background estimation by partitioning events into one signal-enhanced region (A) and three background-enhanced control regions (B, C, and D) via two smoothly varying, statistically independent variables. In practice, even slight correlations between the two variables can significantly undermine the method's performance. Thus, choosing appropriate variables by hand can present a formidable challenge, especially when background and signal differ only subtly. To address this issue, the ABCD with distance correlation (ABCDisCo) method was developed to construct two learned variables via a neural network trained to provide strong signal-background discrimination with small values of the distance correlation (DisCo) measure between the two learned variables. However, relying solely on minimizing the DisCo can result in learned variables that may not have distributions of background events that are smoothly varying and localized at extreme values, as necessary for the validity of the background estimation. The ABCDisCo training enhanced with closure (ABCDisCoTEC) method is introduced to solve this issue by directly minimizing the nonclosure, expressed as a dedicated differentiable loss term. This extended method is applied to a data set of proton-proton collisions at a center-of-mass energy of 13 TeV recorded by the CMS detector at the CERN Large Hadron Collider. Additionally, given the complexity of the minimization problem with constraints on multiple loss terms, the modified differential method of multipliers is applied and shown to greatly improve the stability and robustness of the ABCDisCoTEC method, compared to grid search hyperparameter optimization procedures.

Hayrapetyan, Aram [Yerevan Phys. Inst.]↗

Resilient co-expansion planning between gas and electric distribution networks against natural disasters

Resiliently designed and constructed integrated gas-electric distribution networks (GEDNs) against natural disasters are crucial to social welfare. In this study, a two-stage robust optimisation-based co-expansion planning model is proposed to attain an integrated GEDN with a given resilience level, by optimising the investment strategies of hardening and selective expansion of power distribution feeders and natural gas pipelines, as well as the location and capacity of natural-gas-fired distributed generation. In the first stage, the overall annual investment and operation cost is minimised under normal operation conditions while in the second stage, the feasibility of the investment decisions under the identified worst-case natural disaster scenario is checked with an adjustable load shedding cost criterion. The proposed model is formulated as a mixed integer second-order cone programming problem with the column and constraint generation algorithm employed to seek the optimal solution. Case studies on two integrated GEDNs demonstrate the performance of the proposed methodology.

Zou, Bo↗

Experimental verification of a real-time compensation functionality for dose changes due to target motion in scanned particle therapy

Purpose: Implementation and experimental assessment of a real-time dose compensation system for beam tracking in scanned carbon beam therapy of intrafractionally moving targets. Methods: A real-time dose compensation functionality has been developed and implemented at the experimental branch of the beam tracking system at GSI Helmholtzzentrum fuer Schwerionenforschung (GSI). Treatment plans for different target geometries have been optimized. They have been delivered using scanned carbon ions with beam tracking (BT) and real-time dose compensation combined with beam tracking (RDBT), respectively. Target motion was introduced by a rotating table. Dose distributions were assessed by ionization chamber measurements and dose reconstructions. These distributions have been compared to stationary delivery for BT as well as RDBT. Additionally simulations have been performed to investigate the dependence of delivered dose distributions on varying motion starting phases for BT and RDBT, respectively. Results: Average measured dose differences between static delivery and motion influenced delivery could be reduced from 27-68 mGy when BT was used to 12-37 mGy when RDBT was used. Nominal dose was 1000 mGy. Simulated dose deliveries showed improvements in dose delivery and robustness against varying starting motion phases when RDBT was used. Conclusions: A real-time dose compensation functionality extending the existing beam tracking functionality has been implemented and verified by measurements. Measurements and simulated dose deliveries show that real-time dose compensation can substantially improve delivered dose distributions for large rotational target motion compared to beam tracking alone.

61 RADIATION PROTECTION AND DOSIMETRY↗

Consideration for the application of artificial neural networks to criticality code validation - 14270

The artificial neural network (ANN) is a consistent example for non-parametric parameter estimation and regression, with powerful universal approximation properties, that is distribution independent. ANN application to criticality code validation centers on aligning mathematical relationships between code validation parameters and neural network prediction for those parameter values. Two specific goals are: (1) an appropriate ANN architecture to determine validation parameter values, and (2) bootstrapping individual ANNs as a 'Prediction Interval' (PI) technique to estimate the Calculation Margin and Upper Subcritical Limit. The individual ANN workflow is segregated into Training, Validation (network), and Testing Sets, where the network predictive properties and performance are expressed in the latter set. The Bootstrap ANN replicates a large set of individual ANNs (taken from the original data distribution) where (1-α/2)100% PI estimates can be determined. Results from the ANN method are compared to a variety of validation examples: a 'toy example' from ANSI/ANS-8.24, Appendix D, a variety of methods noted in a Nuclear Criticality Safety Division 2009 paper, and several validation campaigns for highly enriched uranium and plutonium systems. The ANN method is robust and can be applied to single or multivariate inputs. A trained ANN can provide both an average estimate (e.g., single value) and an estimate over a range of independent variable values where the kUSL can be formed in a regression relationship suitable for extrapolation. The input dependence has been tested, ranging from 15 to over 400 sample size distributions. An expanded Shapiro-Wilk test (for sample sizes up to 5000) was implemented to confirm that the ANN method is transparent to distribution function. The Principle Component Analysis method was applied as an input preprocessing step embedded within the ANN in order to reduce dimensionality and correlation for multivariable inputs. This can potentially aid in determining accurate margins over applying arbitrary margins. The automatic optimization of the Bootstrap ANN PI is left for future consideration. (authors)

42 ENGINEERING↗

Simulating competition in the US bioeconomy to produce hard‐to‐electrify transportation fuels using limited biomass resources

This study presents a novel bioeconomy optimization framework, BiOpt, designed to address critical questions regarding the strategic use of limited US biomass resources for biofuel production. By integrating detailed techno-economic analyses, life cycle assessments, and resource assessment data, BiOpt optimizes resource distributions across competing technologies to maximize economic performance and/or minimize greenhouse gas emissions. Using feedstock scenarios from the 2023 Billion Ton Study, the analysis explores optimal biomass allocations across sustainable aviation fuel, diesel, and marine biofuel conversion pathways given varying production targets and policy incentives. Results demonstrate distinct feedstock preferences and pathway utilizations when prioritizing economic returns vs. emissions reductions. For instance, fats, oils, and greases were highly favored in cost-optimized scenarios, while low-carbon feedstocks such as wet waste dominated greenhouse gas-minimized strategies. The findings underscore the pivotal role of policy incentives and technological advances in shaping biofuel supply chains and provide actionable insights for scaling sustainable biofuel production to decarbonize hard-to-electrify sectors. This framework offers a robust tool for policymakers and stakeholders to evaluate biofuel strategies that balance energy output, economic viability, and environmental impact.

09 BIOMASS FUELS↗

When ancient numerical demons meet physics-informed machine learning: adjoint-based gradients for implicit differentiable modeling

Recent advances in differentiable modeling, a genre of physics-informed machine learning that trains neural networks (NNs) together with process-based equations, have shown promise in enhancing hydrological models' accuracy, interpretability, and knowledge-discovery potential. Current differentiable models are efficient for NN-based parameter regionalization, but the simple explicit numerical schemes paired with sequential calculations (operator splitting) can incur numerical errors whose impacts on models' representation power and learned parameters are not clear. Implicit schemes, however, cannot rely on automatic differentiation to calculate gradients due to potential issues of gradient vanishing and memory demand. Here we propose a “discretize-then-optimize” adjoint method to enable differentiable implicit numerical schemes for the first time for large-scale hydrological modeling. The adjoint model demonstrates comprehensively improved performance, with Kling–Gupta efficiency coefficients, peak-flow and low-flow metrics, and evapotranspiration that moderately surpass the already-competitive explicit model. Therefore, the previous sequential-calculation approach had a detrimental impact on the model's ability to represent hydrological dynamics. Furthermore, with a structural update that describes capillary rise, the adjoint model can better describe baseflow in arid regions and also produce low flows that outperform even pure machine learning methods such as long short-term memory networks. The adjoint model rectified some parameter distortions but did not alter spatial parameter distributions, demonstrating the robustness of regionalized parameterization. Despite higher computational expenses and modest improvements, the adjoint model's success removes the barrier for complex implicit schemes to enrich differentiable modeling in hydrology.

58 GEOSCIENCES↗