Digital Chemical Engineering最新文献

{"title":"Physics-structured graph neural networks for interpretable fault detection in semiconductor manufacturing processes","authors":"Chun-Pei Lin, Feiyang Ou, Yifei Wang, Chao Zhang, Sthitie Bom, James F. Davis, Panagiotis D. Christofides","doi":"10.1016/j.dche.2026.100337","DOIUrl":"10.1016/j.dche.2026.100337","url":null,"abstract":"<div><div>Smart manufacturing has transformed semiconductor fabrication into a highly data-driven environment where large numbers of in-situ sensors continuously generate process monitoring data across multiple production tools. Leveraging these data streams to support yield improvement and fault detection has become an important challenge, particularly because conventional physical metrology is often costly, time-consuming, or destructive. Machine learning-based soft sensing has therefore emerged as a practical alternative for predicting process outcomes using routine sensor measurements and contextual process information. Previous studies have demonstrated that classical machine learning models such as gradient boosting decision trees (GBDTs) and feedforward neural networks (FNN) can improve pass/fail classification performance when applied to individual tools or aggregated intra-tool data (i.e., tools of the same class). In this work, we extend these approaches to a broader range of semiconductor manufacturing processes (i.e., data from tools belonging in different classes as well data from the same tool when making different products), and propose a physics-structured graph neural network (GNN) framework for PASS/FAIL prediction. In addition, modern tabular deep-learning models, including TabM and TabR, are evaluated as complementary benchmarks to determine whether strong neural tabular learning can improve fault-detection performance beyond conventional GBDT models and physics-structured graph learning. Instead of modeling individual sensors directly, sensor measurements are organized into physically meaningful process subsystems, each representing a group of related sensors describing process dynamics. These subsystem representations are combined with categorical contextual variables and processed through a graph attention architecture to capture interactions among process modules. TabM is used as a parameter-efficient ensemble-style neural tabular model, while TabR introduces retrieval-augmented prediction by comparing each production run with similar historical training runs. The models are evaluated using production data collected over an 18-month period. For each tool, the most recent 20% of the timeline is reserved for testing while the earlier data are used for training and validation. To ensure statistical robustness, only tools with more than 8000 observations within the study period are included. Results demonstrate that the proposed physics-structured GNN improves predictive performance on several tools compared with conventional machine learning approaches while also providing interpretable attention weights that highlight the relative contribution of different process subsystems to predicted failures. The TabM and TabR results further demonstrate that modern tabular deep-learning models can be useful complementary predictors, especially on tools with richer failure information, although they do not consistently replace the optim","PeriodicalId":72815,"journal":{"name":"Digital Chemical Engineering","volume":"20 ","pages":"Article 100337"},"PeriodicalIF":5.4,"publicationDate":"2026-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148853943","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Extrapolative machine learning for resource-efficient dyeing recipe design in recycled PET/PCT microfiber","authors":"Hyeokjun Cho,&nbsp;Seung Geol Lee","doi":"10.1016/j.dche.2026.100335","DOIUrl":"10.1016/j.dche.2026.100335","url":null,"abstract":"<div><div>Textile dyeing is resource-intensive, consuming large amounts of water, chemicals, and energy due to repeated trial dyeing for color matching. Data-driven recipe prediction offers a promising route to reduce these dyeing trials, but most studies have focused on forward prediction within interpolation ranges and have rarely addressed inverse design, particularly in extrapolation regions. Here, we propose an extrapolative machine learning framework for sustainable dyeing recipe design using Gaussian process regression (GPR). While forward prediction estimates color coordinates from a given dye recipe, the inverse problem addressed here predicts dye concentrations directly from target CIELAB color coordinates. An improved GPR framework integrating inverse and forward models with target-fitting optimization and physically constrained recipe search was developed to enhance extrapolation stability. The framework was based on 250 experimental dyeing samples of microfiber fabrics dyed with three disperse dyes. To evaluate extrapolation performance, six target colors outside the training data distribution were designed and validated through actual dyeing experiments. The improved framework maintained interpolation accuracy comparable to that of the baseline inverse GPR model, which does not include target-fitting optimization. For the six extrapolation targets, it reduced the average experimental <span><math><mrow><mstyle><mi>Δ</mi></mstyle><msubsup><mi>E</mi><mrow><mi>a</mi><mi>b</mi></mrow><mo>*</mo></msubsup></mrow></math></span> from 2.49 to 1.21, with all target-specific values not exceeding 2.05. These results demonstrate that extrapolative machine learning combined with experimental validation enables reliable dyeing recipe design beyond existing datasets. By reducing trial dyeing and improving prediction reliability, the proposed framework offers a practical pathway toward resource-efficient and sustainable textile dyeing.</div></div>","PeriodicalId":72815,"journal":{"name":"Digital Chemical Engineering","volume":"20 ","pages":"Article 100335"},"PeriodicalIF":5.4,"publicationDate":"2026-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148853942","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Data-driven analysis of CO2 loading behavior in aqueous diisopropanolamine-based amine blends","authors":"Maryam Dehbani,&nbsp;Hamed Eivani,&nbsp;Masoud Rahimi","doi":"10.1016/j.dche.2026.100338","DOIUrl":"10.1016/j.dche.2026.100338","url":null,"abstract":"<div><div>Accurate prediction of CO₂ loading in aqueous amine solvents is essential for the design and optimization of carbon capture processes. In this study, five machine-learning models, including Decision Tree (DT), Support Vector Machine (SVM), Gaussian Process Regression (GPR), Multilayer Perceptron (MLP), and Radial Basis Function (RBF) neural networks, were developed to predict CO₂ loading in aqueous diisopropanolamine (DIPA)-based amine blends. A comprehensive dataset comprising 729 experimental data points was collected from the literature, covering temperatures of 297.8–393.1 K, CO₂ partial pressures of 0.12–5888 kPa, and amine concentrations of 10–45 wt.%. The predictive performance of the developed models was evaluated using the coefficient of determination (R²), average absolute relative deviation (AARD), and root mean square error (RMSE). Among the investigated models, the RBF network exhibited the highest prediction accuracy, achieving an R² of 0.9706, an AARD of 5.74%, and an RMSE of 0.0501. Relevancy factor analysis identified CO₂ partial pressure as the most influential parameter, while temperature showed a negative effect on CO₂ loading. The proposed machine-learning framework provides an accurate and computationally efficient tool for predicting CO₂ loading in aqueous DIPA-based amine blends, offering valuable support for solvent screening and carbon capture process optimization.</div></div>","PeriodicalId":72815,"journal":{"name":"Digital Chemical Engineering","volume":"20 ","pages":"Article 100338"},"PeriodicalIF":5.4,"publicationDate":"2026-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148854441","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Hierarchical coordination and power generation optimization for interacted wind farms of large-scale wind farm cluster","authors":"Yihua Zhu,&nbsp;Yuyan Liu,&nbsp;Qibo Liu,&nbsp;Chao Luo,&nbsp;Haochen Zhu,&nbsp;Shaoyuan Li,&nbsp;Yi Zheng","doi":"10.1016/j.dche.2026.100336","DOIUrl":"10.1016/j.dche.2026.100336","url":null,"abstract":"<div><div>To overcome the real-time computation and scalability issues caused by the high-dimensional decision space, nonlinearity in the generation optimization of large-scale wind-farm clusters, this paper proposes a Two-tier hierarchical framework that exploits inter-farm wake steering. Specifically, at the lower tier, each wind farm follows the power reference issued by the upper level while at the same time trying to maximize the wake-deficit velocity that reaches the inflow of downstream farms by adjusting the yaw angle of each turbine. Deep neural networks are trained to learn farm-level optimization and calculate the inter-farm wake, and act as interfarm wake models for upper tier optimization problem. At the upper tier, the controller manages the wake interactions among neighboring farms through regulating the power set-points of each farm under the given ambient wind conditions, to maximize the total power output of the whole cluster. Numerical experiments performed on a Chinese wind farm cluster confirm that the proposed framework achieves the desired optimization performance while significantly lowering computational cost and enhancing scalability compared with conventional centralized optimization.</div></div>","PeriodicalId":72815,"journal":{"name":"Digital Chemical Engineering","volume":"20 ","pages":"Article 100336"},"PeriodicalIF":5.4,"publicationDate":"2026-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148934992","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Online learning supported surrogate-based flowsheet model maintenance","authors":"Balázs Palotai ,&nbsp;Gábor Kis ,&nbsp;Tibor Chován ,&nbsp;Ágnes Bárkányi","doi":"10.1016/j.dche.2025.100287","DOIUrl":"10.1016/j.dche.2025.100287","url":null,"abstract":"<div><div>Surrogate-based flowsheet model calibration is a critical extension of using flowsheet models in Digital Twin (DT) systems. However, maintaining accurate surrogates over time is increasingly challenging, especially when models are deployed in real-time or near-real-time environments, where continuous changes in the physical systems can lead to model drift. To address this challenge, this study introduces a novel online learning–inspired framework to support the continuous maintenance of surrogate-based model calibration. This methodology bridges the gap between offline surrogate development and adaptive model maintenance. By embedding the surrogate in an online learning loop, the framework enables continuous calibration while minimizing reliance on resource-intensive flowsheet simulations. When applied to an industrial flowsheet calibration case, the approach reduced the number of direct calibration steps by up to 94% while preserving global model accuracy. The proposed method offers a scalable, automated, and resilient solution for maintaining surrogate and flowsheet model performance in dynamic industrial environments.</div></div>","PeriodicalId":72815,"journal":{"name":"Digital Chemical Engineering","volume":"18 ","pages":"Article 100287"},"PeriodicalIF":4.1,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145840617","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Design space determination of freezing processes for human induced pluripotent stem cell-derived spheroids using hybrid modeling","authors":"Masaharu Fujioka ,&nbsp;Yusuke Hayashi ,&nbsp;Yuta Yamaguchi ,&nbsp;Tetsuya Fujii ,&nbsp;Hirokazu Sugiyama","doi":"10.1016/j.dche.2026.100294","DOIUrl":"10.1016/j.dche.2026.100294","url":null,"abstract":"<div><div>This work presents the design space determination of freezing processes for human induced pluripotent stem (hiPS) cell-derived spheroids using hybrid modeling. First, a mechanistic model was developed considering spheroid characteristics and freezing phenomena. The developed model was then extended to calculate the cell recovery ratio after thawing using statistical modeling. The hybrid model parameters were estimated based on freezing experiments using hiPS cell-derived spheroids. The extension enabled the calculation of the cell recovery ratio and the number of living cells in the spheroid after thawing, as a function of the cooling rate and the average radius of spheroids. The application of the hybrid model was demonstrated in a case study. The design space of the freezing process was obtained under given constraints related to both quality and productivity. In this way, we could determine design spaces considering both quality and productivity toward industrial manufacturing of hiPS cell-derived spheroids.</div></div>","PeriodicalId":72815,"journal":{"name":"Digital Chemical Engineering","volume":"18 ","pages":"Article 100294"},"PeriodicalIF":4.1,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147385202","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Data-driven safety analytics for dust explosion prevention","authors":"Mohammad Zaid Kamil ,&nbsp;Dongyang Qiu ,&nbsp;Mohammad Alauddin ,&nbsp;Paul Amyotte","doi":"10.1016/j.dche.2026.100298","DOIUrl":"10.1016/j.dche.2026.100298","url":null,"abstract":"<div><div>This study introduces a two-phase data-driven multi-model framework to automate the holistic understanding of dust explosions prevention through critical safety drivers. The proposed framework integrates the Natural Language Processing (NLP), Self-Organizing Maps (SOM), and Bayesian Networks (BN) to identify and analyze critical safety drivers. The framework analyzes unstructured incident reports from the Chemical Safety and Hazard Investigation Board (CSB) and WorkSafeBC (WBC) databases using NLP, transforming textual data into actionable insights for proactive safety management utilizing SOM and BN models. Eight critical safety drivers—including process safety management, inherently safer design (ISD), ignition source control, and safeguard effectiveness—are identified and prioritized through sensitivity analysis. By embedding these insights into process operation, the methodology supports Safety 5.0, enabling predictive interventions and reducing the likelihood of catastrophic events.</div></div>","PeriodicalId":72815,"journal":{"name":"Digital Chemical Engineering","volume":"18 ","pages":"Article 100298"},"PeriodicalIF":4.1,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147385205","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A dual-route ammonia process: Combining renewable and low-carbon pathways","authors":"Amin Soleimani Mehr ,&nbsp;Günter Scheffknecht ,&nbsp;Reihaneh Zohourian ,&nbsp;Jörg Maier ,&nbsp;Markus Reinmoeller","doi":"10.1016/j.dche.2025.100282","DOIUrl":"10.1016/j.dche.2025.100282","url":null,"abstract":"<div><div>Ammonia is increasingly recognized as a versatile hydrogen carrier and energy vector with the potential to decarbonize industrial processes and global energy trade. Among the emerging production routes, blue ammonia—derived from natural gas with integrated carbon capture and storage (CCS)—offers a transitional pathway toward low-carbon energy systems. This work presents the development and assessment of a hybrid ammonia production scheme capable of generating both blue and green ammonia within a single flexible framework. The proposed configuration couples a conventional methane-based synthesis loop equipped with a high-efficiency CO₂ capture system to a parallel renewable-driven synthesis loop, thereby enabling dynamic operation across fossil-based and renewable feedstocks. Process simulations demonstrate that, under a natural gas feed of 1660 metric tons per day (MTPD), the hybrid plant achieves a fuel-to-feedstock ratio of approximately 49 % and an overall CO₂ capture efficiency of up to 98.5 %. Captured CO₂ is compressed to 60 bar, allowing downstream utilization for enhanced oil recovery (EOR) or export via pipeline infrastructure. Beyond process performance, the study highlights the potential role of hybrid ammonia in supporting large-scale decarbonization strategies, strengthening energy security, and bridging the technological gap between blue hydrogen and renewable hydrogen production. In particular, the approach aligns with ongoing energy transition initiatives such as Germany’s Energiewende while offering a scalable solution for global low-carbon fuel supply chains.</div></div>","PeriodicalId":72815,"journal":{"name":"Digital Chemical Engineering","volume":"18 ","pages":"Article 100282"},"PeriodicalIF":4.1,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145977797","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Simulation of TCM-based ionic liquids density behavior in a mixture with ethanol using machine learning approaches","authors":"Ghada Al Assi ,&nbsp;Ali Raqee Abdulhadi ,&nbsp;Rekha MM ,&nbsp;Shaker Al-Hasnaawei ,&nbsp;Subhashree Ray ,&nbsp;Amrita Pal ,&nbsp;Renu Sharma ,&nbsp;Aashna Sinha ,&nbsp;Mehrdad Mottaghi","doi":"10.1016/j.dche.2025.100284","DOIUrl":"10.1016/j.dche.2025.100284","url":null,"abstract":"<div><div>The density behavior of TCM-based ionic liquids (ILs) mixed with ethanol forms the focus of this investigation, highlighting their distinctive physicochemical characteristics and industrial relevance. This research utilizes a suite of sophisticated machine learning methods, encompassing K-Nearest Neighbors (KNN), Convolutional Neural Networks (CNN), Least Squares Support Vector Machine (LSSVM), Decision Tree (DT), Adaptive Boosting (AdaBoost), Artificial Neural Networks (ANN), Random Forest (RF), and Ensemble Learning (EL) to model and forecast density behavior with high precision. To ensure each predictive model operated at its highest capability, the tuning of their internal hyperparameters was carried out through the Coupled Simulated Annealing (CSA) optimization strategy. This global search method was chosen for its ability to efficiently traverse complex parameter spaces and avoid premature convergence. The learning algorithms were trained and validated on a curated experimental dataset consisting of 426 density observations, of which approximately eighty percent (341 samples) were dedicated to model training, while the remaining portion (85 samples) served as an independent test set for evaluating generalization performance. The study covers three TCM-based ionic liquids: [C1C2IM][TCM], [C1C2PYR][TCM], and [C1C2MOR][TCM], in mixture with ethanol, across varying mole fractions and temperatures. The modeling leveraged key input features including ionic liquid type, molar mass (g/mol), mole fraction, and temperature (K) which are crucial determinants of the density behavior. The Monte Carlo–based sensitivity evaluation revealed that the ionic liquid mole fraction exerted the strongest effect on the system, with IL type, molar mass, and temperature contributing in descending order of influence. Prior to developing any of the predictive models, the complete dataset was thoroughly examined to verify its consistency, reliability, and overall suitability for machine-learning–based analysis. After this verification stage, the trained models were benchmarked using multiple statistical criteria. Among all evaluated approaches, the convolutional neural network demonstrated the most superior predictive capability, reflected in its minimal RMSE values, near-unity R² scores, and the lowest AARE percentages in both the training and independent testing evaluations. These findings clearly confirm the remarkable ability of machine learning techniques particularly convolutional neural networks to precisely predict the density of mixtures containing TCM-based ionic liquids and ethanol. This method provides a strong, efficient, and economical substitute for conventional experimental methodologies, empowering scientists to predict density characteristics with enhanced assurance and decreased experimental effort.</div></div>","PeriodicalId":72815,"journal":{"name":"Digital Chemical Engineering","volume":"18 ","pages":"Article 100284"},"PeriodicalIF":4.1,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145840551","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Control mode switching for guaranteed detection of false data injection attacks on process control systems","authors":"Shilpa Narasimhan ,&nbsp;Nael H. El-Farra ,&nbsp;Matthew J. Ellis","doi":"10.1016/j.dche.2025.100279","DOIUrl":"10.1016/j.dche.2025.100279","url":null,"abstract":"<div><div>Control-enabled cyberattack detection approaches are necessary for enhancing the cybersecurity of process control systems (PCSs), as evidenced by recent successful cyberattacks against these systems. One type of cyberattack is false data injection attacks (FDIAs), which manipulate data over sensor-controller and/or controller–actuator communication links. This work presents an active detection strategy based on control mode switching, where the control parameters and/or the set-point are adjusted to induce perturbations that reveal stealthy FDIAs which would otherwise go undetected. To guarantee attack detection, the perturbations introduced by the detection method must be “attack-revealing”, a concept formally defined using reachability analysis in this work. Building on this foundation and considering a specific class of FDIAs, a screening algorithm is developed for selecting control modes that guarantee attack-revealing perturbations in the presence of an attack. A theoretical result is established, identifying control modes incapable of guaranteeing attack detection for a subset of these attacks—specifically, non-bias adding attacks, which do not cause a steady-state offset. This result simplifies the screening process by reducing the candidate control mode set and ensuring that only effective control modes are considered. The applicability of the screening algorithm is demonstrated for several FDIAs, including: (1) multiplicative attacks, (2) non-bias adding multiplicative attacks, and (3) replay attacks, where historic process data is injected into communication channels. The simulation results on an illustrative process validate the effectiveness of the modified screening algorithm and the active detection method in detecting non-biased additive and multiplicative replay attacks.</div></div>","PeriodicalId":72815,"journal":{"name":"Digital Chemical Engineering","volume":"18 ","pages":"Article 100279"},"PeriodicalIF":4.1,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145665348","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}