Results for “Model prediction” — Page 40

DOE OSTI2020

Probabilistic modeling and prediction of out-of-plane unidirectional composite lamina properties

Computational simulation provides an efficient means to predict the behavior of customized hybrid material configurations using validated, physics-based models. One limitation to this approach is the quality and quantity of available data to characterize the many constituent input properties. Therefore, a systematic approach to identify the most influential parameters on the hybrid behavior and quantify the corresponding uncertainty in predictive capabilities is required. Here, an approach using Bayesian multimodel inference and imprecise global sensitivity analysis is presented to investigate the effects of sparse constituent data on the prediction of composite material properties. The methodology allows the identification, using quantified uncertainties, of the most influential constituent material parameters for specified homogenized properties. This sensitivity analysis further enables a dimension reduction when assessing the influence of uncertainties on material properties and can be used to inform testing programs of the constituent properties that require additional testing/data collection in order to minimize uncertainty in macro-scale composite properties. The methodology is specifically demonstrated on the prediction and sensitivity analysis of out-of-plane mechanical properties of a unidirectional lamina.

36 MATERIALS SCIENCE↗

DOE OSTI2023

Investigation of the Effect of Injected CO 2 on the Morrow B Sandstone through Laboratory Batch Reaction Experiments: Implications for CO 2 Sequestration in the Farnsworth Unit, Northern Texas, USA

About one million tons of CO 2 have been injected into the Farnsworth unit to date. The target reservoir for CO 2 injection is the Morrow B Sandstone, which is primarily made of quartz with lesser amounts of albite, calcite, chlorite, and clay minerals. The impact of CO 2 injection on the mineralogy, porosity, and pore water composition of the Morrow B Sandstone is a major concern. Although numerical modeling studies suggest that porosity changes will be minimal, significant alterations to mineralogy and pore water composition are expected. Given the implications for CO 2 storage effectiveness and risk assessment, it is crucial to verify the accuracy of theoretical model predictions through laboratory experiments. To this end, batch reaction experiments were conducted to model conditions near an injection well in the Morrow B Sandstone and at locations further away, where the CO 2 has been diluted by formation water. The laboratory experiments involved submerging thin sections of both coarse- and fine-grained facies of the Morrow B Sandstone in formation water samples with varying levels of CO 2 . The experiments were conducted at the reservoir temperature of 75 °C. Two experimental runs were conducted, one lasting for 61 days and the other for 72 days. The initial fluid composition used in the second run was the same as in the first. The mineralogy changes in the thin sections were analyzed using SEM and the Tescan Integrated Mineral Analyzer (TIMA), while changes in the composition of the formation water were determined using ICP-AES. During each experiment, a thin layer of white fine-grained particles consisting mainly of dolomite and silica formed on the surface of the thin sections, leading to significant reductions in Ca, Mg, and Sr in the formation water. This outcome is consistent with numerical model predictions that dolomite would be the primary mineral that would react with injected CO 2 and that silica would be oversaturated in the formation water. Changes in mineral abundance in the thin sections themselves were much less systematic than in the theoretical modeling experiments, perhaps reflecting heterogeneities in the mineral grain size surface area to volume ratios and mineral distributions in the thin sections not considered in the numerical models.

58 GEOSCIENCES↗

DOE OSTI2023

Transient Efficiency, Flexibility, and Reliability Optimization of Coal-Fired Power Plants - Final Report

This program developed an advanced model-based monitoring and model-predictive control algorithms for a coal fired power plant (CFPP), and deployed these algorithms in a real-time platform to demonstrate performance benefits for transient flexibility and plant operation efficiency. More specifically, the objectives were successfully achieved through a combination of (i) developing a high-fidelity transient plant model in Apros, which was used as a high-fidelity plant simulation between $100-50\% TMCR$ where TMCR denotes the turbine maximum continuous rating, i.e., baseload, (ii) developing a very fast physics-based reduced-order model (ROM) of the plant, which ran more than $100\times$ faster than real-time, enabling its use as real-time embedded model for model-based estimation (MBE) and model predictive control (MPC) (iii) implementing a real-time MBE based on ROM using a robust unscented Kalman filter (UKF) to continuously tune the ROM to match the measurements from high-fidelity Apros plant model despite significant plant-model mismatch, and thus, obtain a Digital Twin of the plant (iv) designing and implementing a real-time MPC with dual objectives of transient plant load tracking with high ramp rates and minimizing coal consumption, i.e., improving plant efficiency in the baseload-partload operation range of $100-50\% TMCR$. Each key element above was developed and tested individually, and has been reported in corresponding Topical Reports. Finally, all the individual elements were integrated in an overall closed-loop system, that was successfully tested in desktop Simulink test harness simulations with ROM or high-fidelity model as the plant. Thereafter, the Simulink implementation was used to auto-generate C-code and deploy as real-time Docker microservice containers in Linux, and validate that they can run in real-time in the hardware-in-the loop (HIL) setup and produce the same results as in Simulink. The results of the integrated simulation tests in Simulink as well as the real-time HIL deployment are documented in this final report, showing good load tracking for load ramps at $3-4\%/min$ ramp rates, and achieving up to $5.5\%$ reduction in coal relative to baseline operation at $50\% TMCR$ load. The desktop and HIL simulations show successful performance of the overall model based estimation and control solution and achieve the key objectives of the program for flexible, efficient and reliable operation of subcritical coal fired power plants.

20 FOSSIL-FUELED POWER PLANTS↗

DOE OSTI2023

Transient Efficiency, Flexibility, and Reliability Optimization of Coal-Fired Power Plants - Final Program Review

This program developed an advanced model-based monitoring and model-predictive control algorithms for a coal fired power plant (CFPP), and deployed these algorithms in a real-time platform to demonstrate performance benefits for transient flexibility and plant operation efficiency. More specifically, the objectives were successfully achieved through a combination of (i) developing a high-fidelity transient plant model in Apros, which was used as a high-fidelity plant simulation between $100-50\% TMCR$ where TMCR denotes the turbine maximum continuous rating, i.e., baseload, (ii) developing a very fast physics-based reduced-order model (ROM) of the plant, which ran more than $100\times$ faster than real-time, enabling its use as real-time embedded model for model-based estimation (MBE) and model predictive control (MPC) (iii) implementing a real-time MBE based on ROM using a robust unscented Kalman filter (UKF) to continuously tune the ROM to match the measurements from high-fidelity Apros plant model despite significant plant-model mismatch, and thus, obtain a Digital Twin of the plant (iv) designing and implementing a real-time MPC with dual objectives of transient plant load tracking with high ramp rates and minimizing coal consumption, i.e., improving plant efficiency in the baseload-partload operation range of $100-50\% TMCR$. Each key element above was developed and tested individually, and has been reported in corresponding Topical Reports. Finally, all the individual elements were integrated in an overall closed-loop system, that was successfully tested in desktop Simulink test harness simulations with ROM or high-fidelity model as the plant. Thereafter, the Simulink implementation was used to auto-generate C-code and deploy as real-time Docker microservice containers in Linux, and validate that they can run in real-time in the hardware-in-the loop (HIL) setup and produce the same results as in Simulink. The results of the integrated simulation tests in Simulink as well as the real-time HIL deployment are documented in this final report, showing good load tracking for load ramps at $3-4\%/min$ ramp rates, and achieving up to $5.5\%$ reduction in coal relative to baseline operation at $50\% TMCR$ load. The desktop and HIL simulations show successful performance of the overall model based estimation and control solution and achieve the key objectives of the program for flexible, efficient and reliable operation of sub-critical coal fired power plants.

20 FOSSIL-FUELED POWER PLANTS↗

DOE OSTI2025

Advancing Multiscale Simulation of Plasma-Surface Interfaces

We report the development of an atomistic-informed, surface-state-dependent predictive model for particle exchange in a carbon-tungsten plasma-surface interface. The predictive model uses machine learning (ML) techniques to learn the energy and angular distributions for particle exchange and rate functions for surface state evolution from molecular dynamics simulations of cumulative bombardment of tungsten by energetic carbon ions. Each predictive component is sensitive to the energy and trajectory of incident plasma species and the surface state. The surface state is represented by a set of surface state descriptors, which were derived from the atomistic surface state for each independent carbon bombardment event. These descriptors are representative of the composition and degree of amorphization of the outermost angstrom of surface material and were chosen to optimize predictive performance for particle exchange at the interface. The distributions for particle exchange (reflection/sputtering) are demonstrated to vary with each surface state descriptor, motivating the development of surface-state-dependent particle exchange models for plasma simulations. The performance of various ML methods was compared, including polynomial quantile regression, artificial neural networks, k-nearest neighbors, and random forest algorithms, with polynomial regression performing the best for interpolation and extrapolation of learned relationships. In addition to the particle exchange model, a neutral network was developed and used to identify data sufficiency throughout surface descriptor space, which will enable real-time feedback during future data production to ensure data is produced where it is most needed, and we provide commentary on improvements to the data production workflow for future endeavors.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

DOE OSTI2022

When physics-informed data analytics outperforms black-box machine learning: A case study in thickness control for additive manufacturing

Aerosol jet printing (AJP) has emerged as a promising noncontact additive manufacturing method for high-resolution printing for a wide range of material systems. A key challenge limiting the broader adoption of AJP in the material science community is the lack of methods to precisely control thickness. Herein, we develop a model-based design of experiment (MBDoE) framework that integrates physics-informed models, nonlinear regression, and information criteria to postulate, select and calibrate the best model to describe and optimize the AJP manufacturing process. Starting with already available data from system commissioning (e.g., prior single variable sensitivity analysis), four candidate physics-informed models are postulated and trained. MBDoE identifies a single additional optimal experiment to validate these predictive models with quantified uncertainties, which are then used to determine the best experimental conditions to control printed film thickness. As a comparative benchmark, the analysis is repeated using the same dataset with nonparametric Gaussian process regression (GPR) model that does not incorporate physical information. Using MBDoE principles, we find that only five experiments are necessary to calibrate the nonlinear physics-informed parametric model, and with said limited data, this model outperforms the black-box machine learning GPR model. This key result underscores an emerging trend in the data science community: incorporating physical information into predictive models often drastically reduces the data requirements. Leveraging MBDoE further increased the data efficiency. By design, the proposed data science framework is general in nature and can be easily extended to other experimental and additive manufacturing systems beyond AJP.

Aerosol jet printing↗

DOE OSTI2022

Development of a concurrent coupled Atomistic - Continuum model to predict the defect and grain structure for Additive Manufacturing process [Slides]

Development of a coupled atomistic-continuum model for metal AM process. Two-way coupling between atomistic (heat flux) and continuum (temperature) domain. Identification of grain and dislocation structure at the atomistic domain. Conversion of dislocation data from discrete atomistic to density form at the continuum. Obtain the spatial distribution of grain structure and dislocation density in additively manufactured metal polycrystal material.

36 MATERIALS SCIENCE↗

DOE OSTI2024

Modeling of Additively Manufactured Ceramic Heat Exchangers with Semi-Elliptical Cross-Section Flow Channels

An approximate and easily applied analytical model was developed for heat transfer calculations of heat exchangers consisting of multiple rows and columns of heat transfer fluid flow channels with semi-elliptical cross sections. Heat exchangers of this type are being developed by using ceramic material and additive manufacturing for high temperature and pressure-concentrating solar electric power plants. Calculations using the model require only the geometrical dimensions and flow conditions of the heat exchanger. Comparisons of modeling predictions, with both simulation results and experimental data, were conducted to verify the viability of the model. The results showed good agreement where almost all the modeling predictions were within 20% of the simulation results or the experimental data. Finally, the proposed modeling approach is more generally applicable to heat transfer analysis of heat exchangers with similar flow channel configurations to those considered in this study.

Ceramic↗

DOE OSTI2020

Flue Gas Desulfurization Wastewater Composition and Implications for Regulatory and Treatment Train Design

The U.S. Environmental Protection Agency is currently revising its regulations on trace element discharges from flue gas desulfurization (FGD) wastewater.<br>In this work, we expand a predictive model of trace element behavior at coal-fired power plants (CFPPs) to estimate the trace element concentration of FGD wastewater at the plant level. We demonstrate that variation in trace element concentrations in FGD wastewater can span several orders of magnitude and is a function of both coal rank and installed air pollution control devices. This conclusion suggests that the benefits and costs of FGD wastewater treatment for the median plant will poorly describe the actual benefits and costs over the full range of existing CFPPs.

Mauter, Meagan↗

DOE OSTI2021

Predicted Reaction Mechanisms, Product Speciation, Kinetics, and Detonation Properties of the Insensitive Explosive 2,6-Diamino-3,5-dinitropyrazine-1-oxide (LLM-105)

2,6-Diamino-3,5-dinitropyrazine-1-oxide (LLM-105) is a relatively new and promising insensitive high-explosive (IHE) material that remains only partially characterized. IHEs are of interest for a range of applications and from a fundamental science standpoint, as the root causes behind insensitivity are poorly understood. In this work, we adopt a multitheory approach based on reactive molecular dynamic simulations performed with density functional theory, density functional tight-binding, and reactive force fields to characterize the reaction pathways, product speciation, reaction kinetics, and detonation performance of LLM-105. We compare and contrast these predictions to 1,3,5-triamino-2,4,6-trinitrobenzene (TATB), a prototypical IHE, and 1,3,5,7-tetranitro-1,3,5,7-tetrazoctane (HMX), a more sensitive and higher performance material. The combination of different predictive models allows access to processes operative on progressively longer timescales while providing benchmarks for assessing uncertainties in the predictions. We find that the early reaction pathways of LLM-105 decomposition are extremely similar to TATB; they involve intra- and intermolecular hydrogen transfer. Additionally, the detonation performance of LLM-105 falls between that of TATB and HMX. We find agreement between predictive models for first-step reaction pathways but significant differences in final product formations. Predictions of detonation performance result in a wide range of values, and one-step kinetic parameters show the similar reaction rates at high temperatures for three out of four models considered.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

DOE OSTI2026

A Decomposition-Based Learn-To-Optimize Approach with Feasibility Layer Assistance for Sub-Hourly Unit Commitment

Sub-hourly unit commitment (UC) with 15-min intervals is gaining significant attention as a way to respond rapidly to the fluctuations in electricity supply and demand introduced by renewable resources. However, the increased temporal resolution and complex inter-temporal dependencies pose substantial computational challenges for traditional optimization methods. To this end, this paper explores a decomposition-based learn-to-optimize approach. Building on recent advances in machine learning, our method revisits the long- overlooked Lagrangian relaxation framework, which is a classical decomposition technique that enables tractable subproblem solving. These smaller subproblems are inherently well-suited for machine learning, as their reduced dimensionality and structural regularity allow predictive models to efficiently learn and generalize solution patterns. We thus propose a generic predictive model, which embeds Gated Recurrent Units (GRUs) and Attention in the encoder-decoder structure, and integrate a rule-based feasibility layer to capture temporal dependencies, reduce training effort, and improve feasibility w.r.t. unit-level constraints. Our method has been validated on the IEEE 118-bus system, demonstrating promising performance in solving sub-hourly UC problems efficiently and feasibly.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2021

Using statistical modeling to predict and understand fusion experiments

Over 300 cryogenic layered direct-drive inertial confinement fusion implosions have been successfully executed on the OMEGA Laser System in the last decade. However, extracting sufficient understanding from these experiments to develop new designs or to identify or mitigate degradation sources remains challenging. Recently, a statistical modeling approach was developed to successfully design and predict improved implosion experiments on OMEGA. Here, we show that one-dimensional simulations can be used to predict the outcomes of systematically perturbed three-dimensional simulations and that this statistical modeling approach can be used to identify or rule out physical mechanisms for some of the degradation sources observed on the OMEGA Laser System for direct-drive cryogenic inertial confinement fusion. In this instance, we investigate the fusion yield dependencies on the ion temperature asymmetries and laser beam size observed in experiments by comparing with trends in three-dimensional synthetic simulation databases. Using the statistical model on these systematically perturbed simulations, we find that the statistically inferred dependency on the measured ion temperature asymmetries is well explained by imposed ℓ =1 modes. However, we find that the dependency on the laser beam size is only dominated by the illumination non-uniformity for some extreme cases.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

DOE OSTI2020

Control of Vibration Systems with Mechanical Motion Rectifier and their Applications to Vehicle Suspension and Ocean Energy Harvester

Vibration control is a large branch in control research, because all moving systems may induce desired or undesired vibration. Due to the limitation of passive system’s adaptability and changing excitation input, vibration control brings the solution to change system dynamic with desired behavior to fulfill control targets. According to preference, vibration control can be separated into two categories: vibration reduction and vibration amplification. Lots of research papers only examine one aspect in vibration control. The thesis investigates the control development for both control targets with two different control applications: vehicle suspension and ocean wave energy converter. It develops control methods for both systems with simplified modeling setup, then followed by the application of a novel mechanical motion rectifier (MMR) gearbox that uses mechanical one-way clutches in both systems. The flow is from the control for common system to the control design for a specifically designed system. In the thesis, active (model predictive control: MPC), semi-active (Skyhook, skyhook-power driven damper: SHPDD, hybrid model predictive control: HMPC), and passive control (Latching Control) methods are developed for different applications or control performance comparison on single system. The thesis also studies about new type of system with switching mechanism, in which other papers do not talk too much and possible control research direction to deal with such complicated system in vibration control. The state-space modeling for both systems are provided in the thesis with detailed model of the MMR gearbox. From the simulation, it can be shown that in the vehicle suspension application, the controlled MMR gearbox can be effective in improving vehicle ride comfort by 29.2% compared to that of the traditional hydraulic suspension. In the ocean wave energy converter, the controlled MMR WEC with simple latching control can improve the power generation by 57% compared to the passive MMR WEC. Besides, the passive MMR WEC also shows its advantage on the passive direct drive WEC in power generation improvement. From the control development flow for the MMR system, the limitation of the MMR gearbox is also identified, which introduces the future work in developing active-MMR gearbox by using an electromagnetic clutch. Some possible control development directions on the active-MMR is also mentioned at the end of the thesis to provide reference for future works.

16 TIDAL AND WAVE POWER↗

DOE OSTI2025

Redox Poise during Rhodospirillum rubrum Phototrophic Growth Drives Large-scale Changes in Macromolecular Synthesis Pathways

During photoheterotrophic growth on organic substrates, purple nonsulfur photosynthetic bacteria like Rhodospirillum rubrum can acquire electrons by multiple means, including oxidation of organic substrates, oxidation of inorganic electron donors (e.g., H2), and by reverse electron flow from the photosynthetic electron transport chain. These electrons are stored as reduced electron-carrying cofactors (e.g., NAD(P)H and ferredoxin). The overall ratio of oxidized to reduced cofactors (e.g., NAD(P)+:NAD(P)H), or ’redox poise’, is difficult to understand or predict, as are the cellular processes for dissipating these reducing equivalents. Using physics-based models that capture mass action kinetics consistent with the thermodynamics of reactions and pathways, a range of redox conditions for heterophototrophic growth are evaluated, from conditions in which the NADP+/NADPH levels approach thermodynamic equilibrium to conditions in which the NADP+/NADPH ratio is far above the typical physiological values. Modeling predictions together with experimental measurements indicate that the redox poise of the cell results in large-scale changes in the activity of biosynthetic pathways and, thus, changes in cell macromolecule levels (DNA, RNA, proteins, and fatty acids). Furthermore, modeling predictions indicate that during phototrophic growth, reverse electron flow from the quinone pool is a minor contributor to the production of reduced cofactors (e.g., NAD(P)H) compared to other oxidative processes (H2 and carbon substrate oxidation). Instead, the quinone pool primarily operates to aid ATP production. The high level of ATP, in turn, drives reduction processes even when NADPH levels are relatively low compared to NADP+ by coupling ATP hydrolysis to the reductive processes. The model, in agreement with experimental measurements of macromolecule ratios of cells growing on different carbon substrates, indicates that the dynamics of nucleotide versus lipid and protein production is likely a significant mechanism of balancing oxidation and reduction in the cell.

59 BASIC BIOLOGICAL SCIENCES↗

DOE OSTI2025

Redox poise in R. rubrum phototrophic growth drives large-scale changes in macromolecular pathways

During photoheterotrophic growth on organic substrates, purple nonsulfur photosynthetic bacteria like Rhodospirillum rubrum can acquire electrons by multiple means, including oxidation of organic substrates, oxidation of inorganic electron donors (e.g., H 2 ), and by reverse electron flow from the photosynthetic electron transport chain. These electrons are stored as reduced electron-carrying cofactors (e.g., NAD(P)H and ferredoxin). The overall ratio of oxidized to reduced cofactors (e.g., NAD(P)+:NAD(P)H), or ’redox poise’, is difficult to understand or predict, as are the cellular processes for dissipating these reducing equivalents. Using physics-based models that capture mass action kinetics consistent with the thermodynamics of reactions and pathways, a range of redox conditions for heterophototrophic growth are evaluated, from conditions in which the NADP+/NADPH levels approach thermodynamic equilibrium to conditions in which the NADP+/NADPH ratio is far above the typical physiological values. Modeling predictions together with experimental measurements indicate that the redox poise of the cell results in large-scale changes in the activity of biosynthetic pathways and, thus, changes in cell macromolecule levels (DNA, RNA, proteins, and fatty acids). Furthermore, modeling predictions indicate that during phototrophic growth, reverse electron flow from the quinone pool is a minor contributor to the production of reduced cofactors (e.g., NAD(P)H) compared to other oxidative processes (H 2 and carbon substrate oxidation). Instead, the quinone pool primarily operates to aid ATP production. The high level of ATP, in turn, drives reduction processes even when NADPH levels are relatively low compared to NADP+ by coupling ATP hydrolysis to the reductive processes. The model, in agreement with experimental measurements of macromolecule ratios of cells growing on different carbon substrates, indicates that the dynamics of nucleotide versus lipid and protein production is likely a significant mechanism of balancing oxidation and reduction in the cell.

59 BASIC BIOLOGICAL SCIENCES↗

DOE OSTI2021

Human-automated vehicle interactions

This dissertation is proposed to answer the question: how can the interactions between human and automated vehicles be used to improve the overall performance of automated driving technology? Multiple different modules in automated vehicles such as the perception, motion plan and motion control modules can potentially be benefitted from human-automated vehicle interactions. For perception module, the self-correction of faulty sensors can be achieved using human demonstration data. For motion plan and motion control modules, the performance of the low-level motion controller can be improved with the help of human demonstration, and the behavior of the motion planner can be improved using human intervention data during automated driving. Moreover, a better model for a human driver could improve the overall efficiency and comfort of vehicles in connected mixed traffic. In this dissertation, the technical research toward these goals has been completed and has resulted in several peer-reviewed publications. Optimization methods and model predictive control are used extensively to improve energy efficiency while maintaining safe and comfort driving. An inverse model predictive control (IMPC) method has been developed and it has been proven to be effective in modeling the motion of human driven vehicles. The proposed method has demonstrated its benefits in both connected automated highway driving and the bilateral adaptation of human driver and automated driving controller in human-in-the loop simulations. The proposed future research seeks to broaden the application of IMPC by considering a more comprehensive cost function design and applying it to more complex driving situations.

Guo, Longxiang↗

DOE OSTI2021

HUMAN-AUTOMATED VEHICLE INTERACTIONS

This dissertation is proposed to answer the question: how can the interactions between human and automated vehicles be used to improve the overall performance of automated driving technology? Multiple different modules in automated vehicles such as the perception, motion plan and motion control modules can potentially be benefitted from human-automated vehicle interactions. For perception module, the self-correction of faulty sensors can be achieved using human demonstration data. For motion plan and motion control modules, the performance of the low-level motion controller can be improved with the help of human demonstration, and the behavior of the motion planner can be improved using human intervention data during automated driving. Moreover, a better model for a human driver could improve the overall efficiency and comfort of vehicles in connected mixed traffic. In this dissertation, the technical research toward these goals has been completed and has resulted in several peer-reviewed publications. Optimization methods and model predictive control are used extensively to improve energy efficiency while maintaining safe and comfort driving. An inverse model predictive control (IMPC) method has been developed and it has been proven to be effective in modeling the motion of human driven vehicles. The proposed method has demonstrated its benefits in both connected automated highway driving and the bilateral adaptation of human driver and automated driving controller in human-in-the loop simulations. The proposed future research seeks to broaden the application of IMPC by considering a more comprehensive cost function design and applying it to more complex driving situations.

Guo, Longxiang↗

DOE OSTI2022

Physics-informed graph neural networks for predicting cetane number with systematic data quality analysis

Designing alternative fuels for advanced compression ignition engines necessitates a predictive model for cetane number (CN). In this study, the physics-informed graph neural networks are introduced for a reliable CN prediction by considering molecular features pertinent to the physical properties of molecules that affect CN. The reliability of measured data is another key factor to consider for improving the predictive model. Various experimental instruments for measuring CN exist, including standard and non-standard methods. In this regard, a systematic data quality analysis was carried out for the total 630 CNs collected from literature and new measurements in this study using Advanced Fuel Ignition Delay Analyzer (AFIDA). The results from this data curation process were reflected in the model by imposing lower sample weights on the data coming from less reliable measurement techniques. This approach effectively maximized the prediction accuracy while incorporating data from all available sources. Using the sample weights decreased the mean absolute error (MAE) up to 0.8 CN units. The accuracy was also improved by introducing the CN-related physical properties (the number of hydrogen bond donors and acceptors); the test set MAE is 5.74 and 7.01 for the model with and without such properties, respectively. Investigating molecular structural effects on CN was also carried out to gain chemical insights into factors used to design new fuel candidates. The dimensionality reduction analysis of feature vectors showed a clear clustering in terms of functional groups and CN and the structural effect derived from the model was consistent with the physicochemical insights. Finally, this physics-informed model and data curation would be helpful for accurate CN prediction and inform rational fuel design.

97 MATHEMATICS AND COMPUTING↗