Collin R. Johnson , Stijn de Vries , Kerstin Wohlgemuth , Sergio Lucia
{"title":"Multi-stage model predictive control for slug flow crystallizers using uncertainty-aware surrogate models","authors":"Collin R. Johnson , Stijn de Vries , Kerstin Wohlgemuth , Sergio Lucia","doi":"10.1016/j.compchemeng.2025.109456","DOIUrl":"10.1016/j.compchemeng.2025.109456","url":null,"abstract":"<div><div>This paper presents a novel dynamic model for slug flow crystallizers that addresses the challenges of spatial distribution without backmixing or diffusion, potentially enabling advanced model-based control. The developed model can accurately describe the main characteristics of slug flow crystallizers, including slug-to-slug variability but leads to a high computational complexity due to the consideration of partial differential equations and population balance equations. For that reason, the model cannot be directly used for process optimization and control. To solve this challenge, we propose two different approaches, conformalized quantile regression and Bayesian last layer neural networks, to develop surrogate models with uncertainty quantification capabilities. These surrogates output a prediction of the system states together with an uncertainty of these predictions to account for process variability and model uncertainty. We use the uncertainty of the predictions to formulate a robust model predictive control approach, enabling robust real-time advanced control of a slug flow crystallizer.</div></div>","PeriodicalId":286,"journal":{"name":"Computers & Chemical Engineering","volume":"205 ","pages":"Article 109456"},"PeriodicalIF":3.9,"publicationDate":"2025-10-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145322886","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"From titer to quality: Exploring reinforcement learning for bioprocess control in silico","authors":"Mariana Monteiro, Konstantinos Flevaris, Cleo Kontoravdi","doi":"10.1016/j.compchemeng.2025.109452","DOIUrl":"10.1016/j.compchemeng.2025.109452","url":null,"abstract":"<div><div>The production of monoclonal antibodies in mammalian cells is a highly complex and nonlinear process. The industry standard for controlling this process fails to capture its complex dynamics, leading to batch-to-batch variability. This inherent complexity makes bioprocesses challenging to model purely mechanistically, while the lack of rich experimental datasets and the need for interpretability in control policies further prevent the use of fully data-driven solutions. We propose a hybrid methodology for optimising the nutrient feeding strategy that leverages Reinforcement Learning (RL) with mechanistic models of cellular metabolism and glycosylation. The RL agent is trained using an off-policy method for data efficiency and is capable of learning from partial observations of the state, which allows for improved generalization. The controller is adaptable to processes with or without additional product quality considerations, such as glycosylation. We demonstrate that accounting for product glycosylation yields different control strategies whereas neglecting it to focus on titer alone can compromise product quality. The continuous learning abilities of the proposed method ensure adaptability in response to process changes, while the inclusion of a mechanistic model in the environment aids in the interpretability of the learned control actions.</div></div>","PeriodicalId":286,"journal":{"name":"Computers & Chemical Engineering","volume":"205 ","pages":"Article 109452"},"PeriodicalIF":3.9,"publicationDate":"2025-10-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145322888","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Ilias Mitrai , Matthew J. Palys , Prodromos Daoutidis
{"title":"A multistage stochastic programming approach for renewable ammonia supply chain network design","authors":"Ilias Mitrai , Matthew J. Palys , Prodromos Daoutidis","doi":"10.1016/j.compchemeng.2025.109443","DOIUrl":"10.1016/j.compchemeng.2025.109443","url":null,"abstract":"<div><div>This paper considers the effect of ammonia market price uncertainty across multiple years on the deployment of renewable ammonia production facilities in existing ammonia supply chain networks. We use an ammonia supply chain transition optimization model to investigate the effect of this uncertainty. Specifically, we formulate a multistage stochastic programming problem to determine the optimal investment policy for new renewable ammonia production over a multi-year transition horizon such that ammonia demand is satisfied and the total supply chain cost is minimized. The proposed approach is used to analyze the transition of the ammonia supply chain for the state of Minnesota. The results show that the trajectory of the price over time determines the degree to which renewable ammonia production facilities are adopted. In a broad sense, considering the possibility of higher-than-average conventional ammonia market prices through a multistage stochastic problem leads to a wider adoption of renewable production relative to a deterministic problem, which only considers the average market price in an economically optimal supply chain transition. Comparison with a two-stage stochastic programming approach from prior work shows that accounting for price uncertainty across time leads to 4.4% reduction in the cost. For a full transition to renewable production, the multistage stochastic framework results, on average, in a slightly slower transition than the deterministic problem due to scenarios which include lower-than-average market prices.</div></div>","PeriodicalId":286,"journal":{"name":"Computers & Chemical Engineering","volume":"205 ","pages":"Article 109443"},"PeriodicalIF":3.9,"publicationDate":"2025-10-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145322885","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Applications of machine learning for decision support in biomass supply chains: A systematic review","authors":"Shayan Razmi, Hossein Mirzaee, Gaurav Kumar, Taraneh Sowlati","doi":"10.1016/j.compchemeng.2025.109451","DOIUrl":"10.1016/j.compchemeng.2025.109451","url":null,"abstract":"<div><div>Effective planning of biomass supply chains (BSC), which involve collection, transportation, pre-processing, storage, conversion, and delivery of bioproducts, is essential to ensure efficiency and sustainability. Recently, machine learning (ML) has been adopted to address the supply chain’s complexities for effective planning. ML provides dynamic and data-driven solutions that enhance decision-making. It has been applied for predicting biomass yields, forecasting supply and demand, optimizing logistics and facility location, and improving the efficiency of conversion processes. This review paper highlights the role of ML in BSC planning. This study considers biomass sources such as food processing residues, animal waste (e.g., manure), in addition to forest-based and agricultural-based biomass, examining processes across all stages of a supply chain from upstream to downstream. We examine ML models in previous studies based on their learning paradigms: supervised, unsupervised, and reinforcement learning, and the type of performed analytics: predictive, and both predictive and prescriptive analytics. Challenges related to data availability, computational requirements, and model generalization limit ML applications in BSCs. Future research could focus on scalable and adaptable models for preprocessing, transportation, and harvesting activities by addressing the uncertainty. Integrating advanced ML could significantly enhance the resiliency, sustainability, and efficiency of BSCs, supporting bioeconomy advancement and the achievement of sustainability goals.</div></div>","PeriodicalId":286,"journal":{"name":"Computers & Chemical Engineering","volume":"205 ","pages":"Article 109451"},"PeriodicalIF":3.9,"publicationDate":"2025-10-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145322887","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Leonardo M. De Marco , Jorge Otávio Trierweiler , Fabio Cesar Diehl , Marcelo Farenzena
{"title":"Assessing, diagnosing, and benchmarking control loops using the input-output cross autocorrelation diagram (IO-CAD)","authors":"Leonardo M. De Marco , Jorge Otávio Trierweiler , Fabio Cesar Diehl , Marcelo Farenzena","doi":"10.1016/j.compchemeng.2025.109438","DOIUrl":"10.1016/j.compchemeng.2025.109438","url":null,"abstract":"<div><div>Monitoring the control loop performance is crucial for operation efficiency and safety in industrial processes. This study proposes a new methodology for control loop performance assessment based on the Input-Output Cross Autocorrelation Diagram (IO<img>CAD), a technique already established in the literature. In this work, two novel indicators based on a polar representation of IO<img>CAD are introduced, complementing four existing indicators previously developed using a Cartesian formulation. By analyzing the autocorrelation between the process variable (PV) and manipulated variable (MV), these indicators enable performance evaluation using only routine plant data. Compared to traditional approaches such as the Minimum Variance Control (MVC), the IO<img>CAD-based method shows greater robustness to noise and setpoint changes, while also providing diagnostic insights into the root causes of performance degradation, such as tuning issues or changes in process dynamics. A Control Performance Indicator (CPI) was also proposed. Simulations involving various control loops, including an offshore oil production control loop, confirmed the method’s effectiveness and applicability for real-time monitoring in diverse operational scenarios.</div></div>","PeriodicalId":286,"journal":{"name":"Computers & Chemical Engineering","volume":"205 ","pages":"Article 109438"},"PeriodicalIF":3.9,"publicationDate":"2025-10-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145265072","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Embedding resilience in natural gas monetization exports","authors":"Mohamad AlMoussaoui , Dhabia M. Al-Mohannadi","doi":"10.1016/j.compchemeng.2025.109444","DOIUrl":"10.1016/j.compchemeng.2025.109444","url":null,"abstract":"<div><div>The economies of several countries depend heavily on hydrocarbon exports, which significantly contribute to their gross national income. As this sector is vulnerable to risks and uncertainties, it is necessary to enhance the resilience of these exports to secure their returns and, hence, the financial security of the national economy. This work studies resilient investment planning to secure the financial returns of the hydrocarbon export sector, considering the production and transportation stages. We develop a novel two-step resilient investment planning approach for the hydrocarbon export sector. In the first step, a portfolio optimization framework is formulated based on Modern Portfolio Theory (MPT) to enhance resilience against price fluctuations associated with hydrocarbon supply chains. In the second step, nine hydrocarbon supply chain resilience metrics are employed to develop a degree of resilience indicator (DORI). The indicator evaluates the performance of financially optimal investment portfolios determined from step one against several risks associated with hydrocarbon exports. The proposed methodology is applied to a case study, considering exporting four chemical commodities to three importers from a natural gas-based economy to determine the optimal investment portfolio. The ability of the proposed DORI to predict portfolio resilience is assessed by running several disruption scenarios. Results highlight the importance of considering resilience metrics, as MPT efficient portfolios with the highest financial returns are not necessarily the most resilient to supply chain disruptions. Results also demonstrate that incorporating a supply chain perspective into the portfolio optimization framework provides additional insights into the hydrocarbon export problem.</div></div>","PeriodicalId":286,"journal":{"name":"Computers & Chemical Engineering","volume":"205 ","pages":"Article 109444"},"PeriodicalIF":3.9,"publicationDate":"2025-10-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145323991","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Microkinetic insights into the impact of coking in dry reforming of methane","authors":"Hye Min Choi , Niket S. Kaisare , Jay H. Lee","doi":"10.1016/j.compchemeng.2025.109441","DOIUrl":"10.1016/j.compchemeng.2025.109441","url":null,"abstract":"<div><div>Coking remains one of the most critical challenges in dry reforming of methane (DRM), causing catalyst deactivation and severe performance loss. While microkinetic modeling (MKM) can capture reaction dynamics at the elementary-step level, existing DRM models lack the ability to represent the evolving nature of coke formation and its mechanistic impact on the reaction network. This study introduces a novel coke-inclusive MKM that explicitly incorporates coke formation pathways and is experimentally validated against DRM data. To interpret the complex, time-dependent behavior of coking, we develop a novel phase-based framework that systematically segments coke accumulation into distinct temporal regimes, each characterized by unique rates and patterns of carbon buildup. Phase-specific mechanistic analysis reveals a gradual shift in the dominant reaction pathways as coking progresses. Early-stage coke formation involves a broad set of surface reactions, opening multiple opportunities for targeted intervention, whereas later stages show a concentration of coking influence in a few critical reactions, such as methane decomposition and CO<sub>2</sub> adsorption. To enhance practicality, a reduced-order coke-inclusive MKM is constructed, retaining essential kinetic features while greatly improving computational efficiency. This integrated modeling strategy — the first to combine a coke-inclusive MKM with phase-based analysis — provides a powerful bridge between detailed reaction mechanisms and application-focused catalyst and reactor design, offering new tools to improve catalyst durability and advance the sustainability of DRM systems.</div></div>","PeriodicalId":286,"journal":{"name":"Computers & Chemical Engineering","volume":"205 ","pages":"Article 109441"},"PeriodicalIF":3.9,"publicationDate":"2025-10-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145322889","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Linear dynamic operability analysis with state-space projection for the online construction of achievable output funnels","authors":"San Dinh , Fernando V. Lima","doi":"10.1016/j.compchemeng.2025.109428","DOIUrl":"10.1016/j.compchemeng.2025.109428","url":null,"abstract":"<div><div>This study presents the development of a dynamic operability analysis approach to determine an operable output funnel for linear time-invariant dynamic systems. Traditional operability mapping approaches are computationally expensive, limiting their application for online control. To address this challenge, a novel two-step calculation procedure is proposed in this article. The first step involves offline computation of the nominal funnel through convex hull construction of manipulated variable projections. The second step involves an online update that adapts the nominal funnel to an operable region based on current state information. The proposed method results in a dynamic funnel that can accommodate process disturbances and measurement noises in the form of transient output constraints. The obtained funnel can be effectively used for model predictive control applications. To demonstrate the effectiveness of the proposed framework, the cyber–physical fuel cell-gas turbine hybrid power system in the HYbrid PERformance (HYPER) process from NETL is used as an example in this study. The dynamic operability funnel constructed with the novel method requires a significantly smaller number of dynamic simulations when compared to the conventional operability mapping method, while maintaining similar accuracy. The results obtained using the proposed approach demonstrate its potential for improving the online control of dynamic systems.</div></div>","PeriodicalId":286,"journal":{"name":"Computers & Chemical Engineering","volume":"205 ","pages":"Article 109428"},"PeriodicalIF":3.9,"publicationDate":"2025-10-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145265073","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Bryan Li , Isaac Severinsen , Timothy Walmsley , Wei Yu , Brent Young
{"title":"Physics-informed neural networks for extrapolating press washer unit operations with heuristic physical knowledge and scarce data","authors":"Bryan Li , Isaac Severinsen , Timothy Walmsley , Wei Yu , Brent Young","doi":"10.1016/j.compchemeng.2025.109440","DOIUrl":"10.1016/j.compchemeng.2025.109440","url":null,"abstract":"<div><div>Extrapolation of process models beyond routine operations is challenging because of the complexity of chemical engineering unit operations, especially for which first-principles models may be unavailable or difficult to formulate. This study investigates to what extent a physics-informed neural network model of a twin roll press washer, incorporating only the generalized heuristic proportionality-based physical relationships that are available, can improve predictive accuracy under non-routine conditions compared to “conventional” data-driven neural network models. The methodology is applied to a case study on predicting roll speed in a twin roll press washer used in pulp and paper production, a key fault-indicating variable for which no established mechanistic or empirical correlations currently exist. To enhance model adaptability, meta-learning is used to treat physical parameters as trainable, allowing the model to adjust them during training and better align physics constraints with observed data. This approach eliminates the need for manual calibration of coefficients in parameterized differential equations, a step that is often impractical in industrial settings due to data scarcity and evolving process conditions. The proposed method achieved a mean squared error of 0.092 RPM<sup>2</sup>, a reduction of nearly 90% compared to purely data-driven models and 30% compared to a fixed-parameter physics-informed neural network model, without significantly increasing training time. The results reinforce the value of the physics-informed neural network modeling approach to process engineering applications and confirm the validity of the proposed novel meta-learning, simple relational physics-based approach.</div></div>","PeriodicalId":286,"journal":{"name":"Computers & Chemical Engineering","volume":"204 ","pages":"Article 109440"},"PeriodicalIF":3.9,"publicationDate":"2025-10-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145262344","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Simultaneous outlier-exclusion and distributionally robust learning through partial optimal transport","authors":"Zhongyu Zhang, Biao Huang, Zukui Li","doi":"10.1016/j.compchemeng.2025.109408","DOIUrl":"10.1016/j.compchemeng.2025.109408","url":null,"abstract":"<div><div>Distributionally robust optimization (DRO) is a powerful framework that mitigates the impact of distributional uncertainty. It aims to optimize the worst-case performance over all possible distributions within an ambiguity set, defined around a nominal distribution which is often set as the empirical distribution constructed from data. However, the presence of outliers in the data may distort the construction of the ambiguity set, thereby degrading the performance of DRO. In this work, we propose an integrated approach that combines outlier exclusion and robust model training. Applying partial optimal transport, we identify and retain the subset of samples that contribute to lower model loss, effectively filtering out potential outliers that cause large losses. This retained subset is used to construct the nominal distribution for the Wasserstein DRO formulation, which addresses the residual distributional uncertainty. We derive tractable formulations for both regression and classification problems under this framework and demonstrate its effectiveness through numerical experiments and real-world chemical process datasets. The results demonstrate that the proposed method provides a simple, effective, and implementable solution for robust learning under both outlier contamination and distributional shifts.</div></div>","PeriodicalId":286,"journal":{"name":"Computers & Chemical Engineering","volume":"205 ","pages":"Article 109408"},"PeriodicalIF":3.9,"publicationDate":"2025-10-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145265753","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}