Pub Date : 2025-04-21DOI: 10.1016/j.compchemeng.2025.109143
Stefan B. Lindström , Rita Ferritsius , Johan E. Carlson , Johan Persson , Fritjof Nilsson
This study focuses on the development of a compact model with improved interpretability compared to similar approaches, relating thermomechanical pulp (TMP) properties, quantified using a fiber analyzer, to Canadian standard freeness and handsheet properties. The data used in this study are obtained from TMP produced by a conical disc refiner. Utilizing the LASSO-regularized Latent Variable Regression (LASSO-LVR) model, we identified three key latent variables – representing shives content, fibrillation, and slender fines content – that accurately predict eight distinct handsheet properties. In a subsequent analysis, we investigated the linkage between refiner settings and Specific Refining Energy (SRE) to these key analyzer readings and, consequently, to handsheet properties. The inclusion of SRE as an internal state variable in the model significantly enhanced predictive accuracy, providing a foundation for more precise and energy-efficient control strategies in refining processes.
{"title":"Predicting handsheet properties and enhancing refiner control using fiber analyzer data and latent variable modeling","authors":"Stefan B. Lindström , Rita Ferritsius , Johan E. Carlson , Johan Persson , Fritjof Nilsson","doi":"10.1016/j.compchemeng.2025.109143","DOIUrl":"10.1016/j.compchemeng.2025.109143","url":null,"abstract":"<div><div>This study focuses on the development of a compact model with improved interpretability compared to similar approaches, relating thermomechanical pulp (TMP) properties, quantified using a fiber analyzer, to Canadian standard freeness and handsheet properties. The data used in this study are obtained from TMP produced by a conical disc refiner. Utilizing the LASSO-regularized Latent Variable Regression (LASSO-LVR) model, we identified three key latent variables – representing shives content, fibrillation, and slender fines content – that accurately predict eight distinct handsheet properties. In a subsequent analysis, we investigated the linkage between refiner settings and Specific Refining Energy (SRE) to these key analyzer readings and, consequently, to handsheet properties. The inclusion of SRE as an internal state variable in the model significantly enhanced predictive accuracy, providing a foundation for more precise and energy-efficient control strategies in refining processes.</div></div>","PeriodicalId":286,"journal":{"name":"Computers & Chemical Engineering","volume":"199 ","pages":"Article 109143"},"PeriodicalIF":3.9,"publicationDate":"2025-04-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143858928","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}
Pub Date : 2025-04-11DOI: 10.1016/j.compchemeng.2025.109139
Maria Kanta, Christos N. Dimitriadis, Evangelos G. Tsimopoulos, Michael C. Georgiadis
The transition to Renewable Energy (RE) is essential for addressing the growing energy demand and meeting the global sustainability goals. Following market trends, this work simultaneously investigates two key aspects: the strategic investment and bidding decisions of an RE producer, and the hourly coordination of Green Certificates Market (GCM) and Electricity Market (EM). To address these aspects a bilevel optimization model is developed. The upper-level problem seeks to maximize the strategic investor's profits, while the lower-level problems sequentially clear the EM and GCM. The model links electricity demand with green certificates demand, and the share of RE in the energy mix with the availability of green certificates. Employing Karush-Kuhn-Tucker conditions, binary expansion, and duality theory, makes the model solvable by commercial solvers. Applied to a modified Pennsylvania-New Jersey-Maryland (PJM) 5-bus system and the IEEE 24-bus test system; the model shows that GCM encourages new RE investments. Strategic bidding in EM enhances these investments by driving down EM prices, securing a growing market share for the RE producer. This price reduction is combined with capacity withholding when needed to prevent zero-price scenarios. Moreover, higher Renewable Portfolio Standard (RPS) targets or increased rival offering prices boost GCM and EM profitability, thereby positively impacting investment decisions. Contrarily, lower wind capacity factors negatively impact new investments as they lead to higher EM and GCM prices.
{"title":"Optimal investment and bidding strategies for wind power in electricity and green certificates markets","authors":"Maria Kanta, Christos N. Dimitriadis, Evangelos G. Tsimopoulos, Michael C. Georgiadis","doi":"10.1016/j.compchemeng.2025.109139","DOIUrl":"10.1016/j.compchemeng.2025.109139","url":null,"abstract":"<div><div>The transition to Renewable Energy (RE) is essential for addressing the growing energy demand and meeting the global sustainability goals. Following market trends, this work simultaneously investigates two key aspects: the strategic investment and bidding decisions of an RE producer, and the hourly coordination of Green Certificates Market (GCM) and Electricity Market (EM). To address these aspects a bilevel optimization model is developed. The upper-level problem seeks to maximize the strategic investor's profits, while the lower-level problems sequentially clear the EM and GCM. The model links electricity demand with green certificates demand, and the share of RE in the energy mix with the availability of green certificates. Employing Karush-Kuhn-Tucker conditions, binary expansion, and duality theory, makes the model solvable by commercial solvers. Applied to a modified Pennsylvania-New Jersey-Maryland (PJM) 5-bus system and the IEEE 24-bus test system; the model shows that GCM encourages new RE investments. Strategic bidding in EM enhances these investments by driving down EM prices, securing a growing market share for the RE producer. This price reduction is combined with capacity withholding when needed to prevent zero-price scenarios. Moreover, higher Renewable Portfolio Standard (RPS) targets or increased rival offering prices boost GCM and EM profitability, thereby positively impacting investment decisions. Contrarily, lower wind capacity factors negatively impact new investments as they lead to higher EM and GCM prices.</div></div>","PeriodicalId":286,"journal":{"name":"Computers & Chemical Engineering","volume":"199 ","pages":"Article 109139"},"PeriodicalIF":3.9,"publicationDate":"2025-04-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143828337","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}
Pub Date : 2025-04-08DOI: 10.1016/j.compchemeng.2025.109142
Beril Tümer , Deniz Şanlı Yıldız , Yaman Arkun
This paper presents a real-time optimization strategy for power allocation between two fuel cell stacks, maximizing overall efficiency while minimizing hydrogen consumption. The proposed method accounts for stack degradation, characterized by a time-varying electron transfer coefficient (α), estimated in real-time using RLS-Kalman filtering from voltage measurements. The strategy also considers hydrogen crossover effects, which impact fuel efficiency and utilization. The optimization approach was evaluated against two conventional strategies—equal distribution and daisy chain—demonstrating superior performance across various operating scenarios. A new efficiency-based daisy chain algorithm was introduced and compared with the classical power-based method, further highlighting the benefits of the optimization framework. The real-time formulation enables on-the-fly parameter estimation and model updates, making it adaptable to multiple stacks and various objective functions. This approach provides a robust and scalable solution for fuel cell power management under degradation, aging, and other adverse conditions.
{"title":"Real-time optimal power sharing in multi-stack fuel cells","authors":"Beril Tümer , Deniz Şanlı Yıldız , Yaman Arkun","doi":"10.1016/j.compchemeng.2025.109142","DOIUrl":"10.1016/j.compchemeng.2025.109142","url":null,"abstract":"<div><div>This paper presents a real-time optimization strategy for power allocation between two fuel cell stacks, maximizing overall efficiency while minimizing hydrogen consumption. The proposed method accounts for stack degradation, characterized by a time-varying electron transfer coefficient (α), estimated in real-time using RLS-Kalman filtering from voltage measurements. The strategy also considers hydrogen crossover effects, which impact fuel efficiency and utilization. The optimization approach was evaluated against two conventional strategies—equal distribution and daisy chain—demonstrating superior performance across various operating scenarios. A new efficiency-based daisy chain algorithm was introduced and compared with the classical power-based method, further highlighting the benefits of the optimization framework. The real-time formulation enables on-the-fly parameter estimation and model updates, making it adaptable to multiple stacks and various objective functions. This approach provides a robust and scalable solution for fuel cell power management under degradation, aging, and other adverse conditions.</div></div>","PeriodicalId":286,"journal":{"name":"Computers & Chemical Engineering","volume":"199 ","pages":"Article 109142"},"PeriodicalIF":3.9,"publicationDate":"2025-04-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143824469","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}
Pub Date : 2025-04-08DOI: 10.1016/j.compchemeng.2025.109144
Chengtian Cui , Qing Li , William L. Luyben , Anton A. Kiss
This study investigates the dynamics and control of discretely heat integrated distillation columns, focusing on two configurations: one utilizing a liquid pumparound loop and the other employing liquid injection for waste heat recovery in a multi-stage vapor recompression cycle. These innovative designs eliminate the need for vapor splitters, simplifying operation and enhancing control robustness. As case study, the methanol/water separation process was modelled to achieve 99.99 mol % purity for both products. Dynamic simulations were conducted in Aspen Dynamics to evaluate the control performance for ± 20 % throughput and composition disturbances. Results demonstrated that the proposed control structures, which rely on inferential temperature-based strategies, effectively maintain product specifications and ensure stable operation. This work provides valuable insights into the practical implementation of discretely heat integrated distillation columns, offering a pathway toward energy-efficient and operationally flexible distillation systems.
{"title":"Dynamics and control of discretely heat integrated distillation columns","authors":"Chengtian Cui , Qing Li , William L. Luyben , Anton A. Kiss","doi":"10.1016/j.compchemeng.2025.109144","DOIUrl":"10.1016/j.compchemeng.2025.109144","url":null,"abstract":"<div><div>This study investigates the dynamics and control of discretely heat integrated distillation columns, focusing on two configurations: one utilizing a liquid pumparound loop and the other employing liquid injection for waste heat recovery in a multi-stage vapor recompression cycle. These innovative designs eliminate the need for vapor splitters, simplifying operation and enhancing control robustness. As case study, the methanol/water separation process was modelled to achieve 99.99 mol % purity for both products. Dynamic simulations were conducted in Aspen Dynamics to evaluate the control performance for ± 20 % throughput and composition disturbances. Results demonstrated that the proposed control structures, which rely on inferential temperature-based strategies, effectively maintain product specifications and ensure stable operation. This work provides valuable insights into the practical implementation of discretely heat integrated distillation columns, offering a pathway toward energy-efficient and operationally flexible distillation systems.</div></div>","PeriodicalId":286,"journal":{"name":"Computers & Chemical Engineering","volume":"199 ","pages":"Article 109144"},"PeriodicalIF":3.9,"publicationDate":"2025-04-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143828338","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}
Pub Date : 2025-04-08DOI: 10.1016/j.compchemeng.2025.109141
Yuchao Qiu , Zuhua Xu , Jun Zhao , Chunyue Song , Xiaoping Zhu
Control performance degradation is a common phenomenon in chemical processes. How to achieve online tuning of controller parameters with few experiments while ensuring the stability and safety of the control system is a significant challenge. In this study, a data-driven controller parameters online tuning method based on model-inherited trust region Bayesian optimization is proposed. First, the objective function and safety constraints of Bayesian optimization for controller online tuning problem are excavated using control performance assessment (CPA) based on time series modeling. Second, utilizing the correlation between local modeling tasks, a model-inherited Gaussian process regression (GPR) method is proposed to build the accurate surrogate model between controller parameters and control performance. This surrogate model consists of two parts: one is the inheritance part of historical GPR models and the other is the residual part interpreted as a zero-mean Gaussian process, so that the model accuracy can be guaranteed with a small amount of evaluation data. Third, a constraint acquisition function based on expected improvement is designed to ensure safe exploration during the tuning procedure, in which feasibility probability of constraints such as overshoot and settling time from CPA are incorporated through a weighted approach. Moreover, a shape-adaptation update method of the trust region is developed to improve optimization efficiency and robustness. Finally, the effectiveness of the method is verified through two industrial cases.
{"title":"Data-driven controller parameters online tuning method based on model-inherited trust region Bayesian optimization","authors":"Yuchao Qiu , Zuhua Xu , Jun Zhao , Chunyue Song , Xiaoping Zhu","doi":"10.1016/j.compchemeng.2025.109141","DOIUrl":"10.1016/j.compchemeng.2025.109141","url":null,"abstract":"<div><div>Control performance degradation is a common phenomenon in chemical processes. How to achieve online tuning of controller parameters with few experiments while ensuring the stability and safety of the control system is a significant challenge. In this study, a data-driven controller parameters online tuning method based on model-inherited trust region Bayesian optimization is proposed. First, the objective function and safety constraints of Bayesian optimization for controller online tuning problem are excavated using control performance assessment (CPA) based on time series modeling. Second, utilizing the correlation between local modeling tasks, a model-inherited Gaussian process regression (GPR) method is proposed to build the accurate surrogate model between controller parameters and control performance. This surrogate model consists of two parts: one is the inheritance part of historical GPR models and the other is the residual part interpreted as a zero-mean Gaussian process, so that the model accuracy can be guaranteed with a small amount of evaluation data. Third, a constraint acquisition function based on expected improvement is designed to ensure safe exploration during the tuning procedure, in which feasibility probability of constraints such as overshoot and settling time from CPA are incorporated through a weighted approach. Moreover, a shape-adaptation update method of the trust region is developed to improve optimization efficiency and robustness. Finally, the effectiveness of the method is verified through two industrial cases.</div></div>","PeriodicalId":286,"journal":{"name":"Computers & Chemical Engineering","volume":"199 ","pages":"Article 109141"},"PeriodicalIF":3.9,"publicationDate":"2025-04-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143854810","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}
Reward design is a key challenge in reinforcement learning (RL) as it directly affects the effectiveness of learned policies. Inverse Reinforcement Learning (IRL) attempts to solve this problem by learning reward functions from expert trajectories. This study utilizes a reward design using Adversarial IRL (AIRL) frameworks using expert trajectories from Model Predictive Control (MPC). On the contrary, there are also instances where a pre-defined reward function works well, indicating a potential trade-off between these two. To achieve this, we propose a twin-agent reinforcement learning framework where the first agent utilizes a pre-defined reward function, while the second agent learns reward in the AIRL setting guided by MPC with Proximal Policy Optimization (PPO) as the backbone (PPO-MPC-AIRL). The performance of the proposed algorithm has been tested using a case study, namely, mAb production in the bioreactor. The simulation results indicate that the proposed algorithm is able to reduce the root mean square error (RMSE) of set-point tracking by 18.38 % compared to the nominal PPO.
{"title":"An adversarial twin-agent inverse proximal policy optimization guided by model predictive control","authors":"Nikita Gupta , Harikumar Kandath , Hariprasad Kodamana","doi":"10.1016/j.compchemeng.2025.109124","DOIUrl":"10.1016/j.compchemeng.2025.109124","url":null,"abstract":"<div><div>Reward design is a key challenge in reinforcement learning (RL) as it directly affects the effectiveness of learned policies. Inverse Reinforcement Learning (IRL) attempts to solve this problem by learning reward functions from expert trajectories. This study utilizes a reward design using Adversarial IRL (AIRL) frameworks using expert trajectories from Model Predictive Control (MPC). On the contrary, there are also instances where a pre-defined reward function works well, indicating a potential trade-off between these two. To achieve this, we propose a twin-agent reinforcement learning framework where the first agent utilizes a pre-defined reward function, while the second agent learns reward in the AIRL setting guided by MPC with Proximal Policy Optimization (PPO) as the backbone (PPO-MPC-AIRL). The performance of the proposed algorithm has been tested using a case study, namely, mAb production in the bioreactor. The simulation results indicate that the proposed algorithm is able to reduce the root mean square error (RMSE) of set-point tracking by 18.38 % compared to the nominal PPO.</div></div>","PeriodicalId":286,"journal":{"name":"Computers & Chemical Engineering","volume":"199 ","pages":"Article 109124"},"PeriodicalIF":3.9,"publicationDate":"2025-04-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143852345","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}
Pub Date : 2025-04-06DOI: 10.1016/j.compchemeng.2025.109113
Argyri Kardamaki , Athanassios Nikolakopoulos , Mihalis Kavousanakis , Philip Doganis , Matt Jellicoe , William Stokes , Vesa Hongisto , Matthew Simmons , Thomas W. Chamberlain , Nikil Kapur , Roland Grafström , Andrew Nelson , Haralambos Sarimveis
The production of nanomaterials (NMs) has gained significant attention due to their unique properties and versatile applications in fields such as medicine, energy, and electronics. However, ensuring the large-scale synthesis of safe and sustainable NMs while maintaining their functionality remains a critical challenge. This study introduces the Safety by Process Control (SbPC) framework, a novel methodology integrating dynamic first-principles modeling, Model Predictive Control (MPC), and real-time safety monitoring. The framework employs a physics-based population balance model with a Method Of Moments (MOM) approximation to predict the evolution of key NM properties. A toxicity inferential sensor, built on experimental data, is integrated to facilitate real-time hazard assessment. The efficiency of the proposed framework is demonstrated using a continuous silver nanoparticle (Ag NP) production system as a case study. The proposed approach ensures the production of high-quality, safe, and sustainable NMs, aligning with Safe and Sustainable by Design (SSbD) principles and addressing gaps in current NM manufacturing processes. The framework’s adaptability to other NM types highlights its potential as a transformative tool for sustainable nanotechnology.
{"title":"The SABYDOMA Safety by Process Control framework for the production of functional, safe and sustainable nanomaterials","authors":"Argyri Kardamaki , Athanassios Nikolakopoulos , Mihalis Kavousanakis , Philip Doganis , Matt Jellicoe , William Stokes , Vesa Hongisto , Matthew Simmons , Thomas W. Chamberlain , Nikil Kapur , Roland Grafström , Andrew Nelson , Haralambos Sarimveis","doi":"10.1016/j.compchemeng.2025.109113","DOIUrl":"10.1016/j.compchemeng.2025.109113","url":null,"abstract":"<div><div>The production of nanomaterials (NMs) has gained significant attention due to their unique properties and versatile applications in fields such as medicine, energy, and electronics. However, ensuring the large-scale synthesis of safe and sustainable NMs while maintaining their functionality remains a critical challenge. This study introduces the Safety by Process Control (SbPC) framework, a novel methodology integrating dynamic first-principles modeling, Model Predictive Control (MPC), and real-time safety monitoring. The framework employs a physics-based population balance model with a Method Of Moments (MOM) approximation to predict the evolution of key NM properties. A toxicity inferential sensor, built on experimental data, is integrated to facilitate real-time hazard assessment. The efficiency of the proposed framework is demonstrated using a continuous silver nanoparticle (Ag NP) production system as a case study. The proposed approach ensures the production of high-quality, safe, and sustainable NMs, aligning with Safe and Sustainable by Design (SSbD) principles and addressing gaps in current NM manufacturing processes. The framework’s adaptability to other NM types highlights its potential as a transformative tool for sustainable nanotechnology.</div></div>","PeriodicalId":286,"journal":{"name":"Computers & Chemical Engineering","volume":"199 ","pages":"Article 109113"},"PeriodicalIF":3.9,"publicationDate":"2025-04-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143838879","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}
Pub Date : 2025-04-04DOI: 10.1016/j.compchemeng.2025.109114
José Rodrigues Torraca Neto , Bruno Didier Olivier Capron , Argimiro Resende Secchi
This work investigates the application of reinforcement learning (RL) for crystallization process control, focusing on robustness against parametric uncertainty and measurement noise. A curriculum learning approach with progressive uncertainty scaling and soft constraint enforcement was developed to enhance RL agent adaptability and performance. Four actor–critic RL algorithms—Deep Deterministic Policy Gradient (DDPG), Twin Delayed Deep Deterministic Policy Gradient (TD3), Soft Actor–Critic (SAC), and Proximal Policy Optimization (PPO)—were trained and evaluated using baseline, domain randomization, and curriculum learning strategies. The performance of each algorithm was assessed based on key control metrics, including setpoint tracking, control smoothness, and constraint satisfaction, with Nonlinear Model Predictive Control (NMPC) serving as an oracle benchmark. The results show that PPO consistently outperformed other algorithms, achieving the lowest mean absolute percentage error (MAPE) for critical process parameters (2.20%) and the lowest violation probability (0.67%) under curriculum learning. This strategy also reduced control variability, with PPO achieving a control variability index (CVI) of 0.008, indicating smooth control actions. While DDPG and TD3 exhibited competitive performance, SAC suffered from high fluctuations and the lowest rewards across all training strategies, highlighting its limitations in stability-critical applications. The findings highlight the effectiveness of curriculum learning with soft constraints in enhancing RL performance for industrial process control, establishing PPO as a reliable solution for robust crystallization control.
{"title":"A robust deep reinforcement learning approach for the control of crystallization processes","authors":"José Rodrigues Torraca Neto , Bruno Didier Olivier Capron , Argimiro Resende Secchi","doi":"10.1016/j.compchemeng.2025.109114","DOIUrl":"10.1016/j.compchemeng.2025.109114","url":null,"abstract":"<div><div>This work investigates the application of reinforcement learning (RL) for crystallization process control, focusing on robustness against parametric uncertainty and measurement noise. A curriculum learning approach with progressive uncertainty scaling and soft constraint enforcement was developed to enhance RL agent adaptability and performance. Four actor–critic RL algorithms—Deep Deterministic Policy Gradient (DDPG), Twin Delayed Deep Deterministic Policy Gradient (TD3), Soft Actor–Critic (SAC), and Proximal Policy Optimization (PPO)—were trained and evaluated using baseline, domain randomization, and curriculum learning strategies. The performance of each algorithm was assessed based on key control metrics, including setpoint tracking, control smoothness, and constraint satisfaction, with Nonlinear Model Predictive Control (NMPC) serving as an oracle benchmark. The results show that PPO consistently outperformed other algorithms, achieving the lowest mean absolute percentage error (MAPE) for critical process parameters (2.20%) and the lowest violation probability (0.67%) under curriculum learning. This strategy also reduced control variability, with PPO achieving a control variability index (CVI) of 0.008, indicating smooth control actions. While DDPG and TD3 exhibited competitive performance, SAC suffered from high fluctuations and the lowest rewards across all training strategies, highlighting its limitations in stability-critical applications. The findings highlight the effectiveness of curriculum learning with soft constraints in enhancing RL performance for industrial process control, establishing PPO as a reliable solution for robust crystallization control.</div></div>","PeriodicalId":286,"journal":{"name":"Computers & Chemical Engineering","volume":"199 ","pages":"Article 109114"},"PeriodicalIF":3.9,"publicationDate":"2025-04-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143790953","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}
Pub Date : 2025-04-04DOI: 10.1016/j.compchemeng.2025.109119
Peter Jul-Rasmussen , Monesh Kumar , Jóse Pinto , Rui Oliveira , Xiaodong Liang , Jakob Kjøbsted Huusom
With increased data availability in the (bio)chemical processing industries, there is a renewed interest in leveraging data-based methods to improve process operations. While data-based approaches enable the modeling of phenomena that are difficult to model mechanistically, they also have drawbacks, especially when presented with serially correlated process data with low variation typically found in the process industries. The limitations in both mechanistic and data-based modeling can be addressed through hybrid approaches. The combination of mechanistic and data-based models into hybrid semi-parametric models has shown great promise in mitigating such limitations over the last 30 years. More recently, physics-informed learning approaches have been proposed as an alternative method for embedding process knowledge in data-based models. This work provides a comparative study of hybrid semi-parametric modeling and Physics-Informed Recurrent Neural Networks (PIRNNs) applied to a pilot-scale bubble column aeration case study. The developed models are compared based on the ease of training, the models’ adherence to the governing system equations, the prediction accuracy when reducing the measurement frequency, and the model performance when reducing the quantity of training data. For the considered case study, the hybrid semi-parametric modeling approach generally resulted in superior model performance with high prediction accuracy, good adherence to the physics, and good performance when reducing the quantity of training data.
{"title":"Incorporating first-principles information into hybrid modeling structures: Comparing hybrid semi-parametric models with Physics-Informed Recurrent Neural Networks","authors":"Peter Jul-Rasmussen , Monesh Kumar , Jóse Pinto , Rui Oliveira , Xiaodong Liang , Jakob Kjøbsted Huusom","doi":"10.1016/j.compchemeng.2025.109119","DOIUrl":"10.1016/j.compchemeng.2025.109119","url":null,"abstract":"<div><div>With increased data availability in the (bio)chemical processing industries, there is a renewed interest in leveraging data-based methods to improve process operations. While data-based approaches enable the modeling of phenomena that are difficult to model mechanistically, they also have drawbacks, especially when presented with serially correlated process data with low variation typically found in the process industries. The limitations in both mechanistic and data-based modeling can be addressed through hybrid approaches. The combination of mechanistic and data-based models into hybrid semi-parametric models has shown great promise in mitigating such limitations over the last 30 years. More recently, physics-informed learning approaches have been proposed as an alternative method for embedding process knowledge in data-based models. This work provides a comparative study of hybrid semi-parametric modeling and Physics-Informed Recurrent Neural Networks (PIRNNs) applied to a pilot-scale bubble column aeration case study. The developed models are compared based on the ease of training, the models’ adherence to the governing system equations, the prediction accuracy when reducing the measurement frequency, and the model performance when reducing the quantity of training data. For the considered case study, the hybrid semi-parametric modeling approach generally resulted in superior model performance with high prediction accuracy, good adherence to the physics, and good performance when reducing the quantity of training data.</div></div>","PeriodicalId":286,"journal":{"name":"Computers & Chemical Engineering","volume":"199 ","pages":"Article 109119"},"PeriodicalIF":3.9,"publicationDate":"2025-04-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143808229","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}
Pub Date : 2025-04-03DOI: 10.1016/j.compchemeng.2025.109128
Zhifei Yuliu, Yuqing Luo, Marianthi G. Ierapetritou
The selection of mixed plastic waste recycling technologies directly influences product distribution and thus the overall process economics. This study presents an integrated chemical recycling process that combines pyrolysis, hydrogenolysis, and hydrocracking to transform plastic waste directly into high-value fuel products, eliminating the need for further upgrading. A mixed-integer nonlinear programming model is formulated to consider the selection of reaction pathways considering fuel properties to ensure compliance with product specifications. Robust optimization is used to incorporate the effect of fuel property and economic uncertainty on the process design. Under nominal conditions, the optimized chemical recycling process demonstrates strong economic performance with a unit net present value of $366/t plastic waste. Different levels of conservatism are considered to account for uncertainty. The optimized process shows a robust performance by leveraging different combinations of depolymerization technologies and adjusting the product portfolio.
{"title":"Design of plastic waste chemical recycling process considering uncertainty","authors":"Zhifei Yuliu, Yuqing Luo, Marianthi G. Ierapetritou","doi":"10.1016/j.compchemeng.2025.109128","DOIUrl":"10.1016/j.compchemeng.2025.109128","url":null,"abstract":"<div><div>The selection of mixed plastic waste recycling technologies directly influences product distribution and thus the overall process economics. This study presents an integrated chemical recycling process that combines pyrolysis, hydrogenolysis, and hydrocracking to transform plastic waste directly into high-value fuel products, eliminating the need for further upgrading. A mixed-integer nonlinear programming model is formulated to consider the selection of reaction pathways considering fuel properties to ensure compliance with product specifications. Robust optimization is used to incorporate the effect of fuel property and economic uncertainty on the process design. Under nominal conditions, the optimized chemical recycling process demonstrates strong economic performance with a unit net present value of $366/t plastic waste. Different levels of conservatism are considered to account for uncertainty. The optimized process shows a robust performance by leveraging different combinations of depolymerization technologies and adjusting the product portfolio.</div></div>","PeriodicalId":286,"journal":{"name":"Computers & Chemical Engineering","volume":"199 ","pages":"Article 109128"},"PeriodicalIF":3.9,"publicationDate":"2025-04-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143815407","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}
扫码关注我们
求助内容：
应助结果提醒方式：
京公网安备 11010802042870号