Pub Date : 2024-04-18DOI: 10.1016/j.compchemeng.2024.108703
Sanjula Kammammettu, Shu-Bo Yang, Zukui Li
Conventional chance-constrained programming methods suffer from the inexactness of the estimated probability distribution of the underlying uncertainty from data. To this end, a distributionally robust approach to the problem allows for a level of ambiguity considered around a reference distribution. In this work, we propose a novel formulation for the distributionally robust chance-constrained programming problem using an ambiguity set constructed from a variant of optimal transport distance that was developed for Gaussian Mixture Models. We show that for multimodal process uncertainty, our proposed method provides an effective way to incorporate statistical moment information into the ambiguity set construction step, thus leading to improved optimal solutions. We illustrate the performance of our method on a numerical example as well as a chemical process case study. We show that our proposed methodology leverages the multimodal characteristics from the uncertainty data to give superior performance over the traditional Wasserstein distance-based method.
{"title":"Distributionally robust chance-constrained optimization with Gaussian mixture ambiguity set","authors":"Sanjula Kammammettu, Shu-Bo Yang, Zukui Li","doi":"10.1016/j.compchemeng.2024.108703","DOIUrl":"https://doi.org/10.1016/j.compchemeng.2024.108703","url":null,"abstract":"<div><p>Conventional chance-constrained programming methods suffer from the inexactness of the estimated probability distribution of the underlying uncertainty from data. To this end, a distributionally robust approach to the problem allows for a level of ambiguity considered around a reference distribution. In this work, we propose a novel formulation for the distributionally robust chance-constrained programming problem using an ambiguity set constructed from a variant of optimal transport distance that was developed for Gaussian Mixture Models. We show that for multimodal process uncertainty, our proposed method provides an effective way to incorporate statistical moment information into the ambiguity set construction step, thus leading to improved optimal solutions. We illustrate the performance of our method on a numerical example as well as a chemical process case study. We show that our proposed methodology leverages the multimodal characteristics from the uncertainty data to give superior performance over the traditional Wasserstein distance-based method.</p></div>","PeriodicalId":286,"journal":{"name":"Computers & Chemical Engineering","volume":null,"pages":null},"PeriodicalIF":4.3,"publicationDate":"2024-04-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140645576","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 : 2024-04-17DOI: 10.1016/j.compchemeng.2024.108700
Sumin Hwangbo , J. Jay Liu , Jun-Hyung Ryu , Ho Jae Lee , Jonggeol Na
Nervousness-aware rescheduling is essential in maximizing the profitability and stability of processes in manufacturing industries. It involves re-optimization to meet scheduling goals while minimizing deviations from the base schedule. However, conventional mathematical optimization becomes impractical due to high computational costs and the inability to handle real-time rescheduling. Here, we propose an online rescheduling agent trained by explorative reinforcement learning that autonomously optimizes schedules while considering schedule nervousness. In a static scheduling environment, our model consistently achieves over 90% of the cost objective with scalability and flexibility. A computational time comparison proves that the reinforcement learning methodology makes near-optimal decisions rapidly, irrespective of the complexity of the scheduling problem. Furthermore, we present several realistic rescheduling scenarios that demonstrate the capability of our methodology. Our study illustrates the significant potential of reinforcement learning methodology in expediting digital transformation and process automation within real-world manufacturing systems.
{"title":"Production rescheduling via explorative reinforcement learning while considering nervousness","authors":"Sumin Hwangbo , J. Jay Liu , Jun-Hyung Ryu , Ho Jae Lee , Jonggeol Na","doi":"10.1016/j.compchemeng.2024.108700","DOIUrl":"https://doi.org/10.1016/j.compchemeng.2024.108700","url":null,"abstract":"<div><p>Nervousness-aware rescheduling is essential in maximizing the profitability and stability of processes in manufacturing industries. It involves re-optimization to meet scheduling goals while minimizing deviations from the base schedule. However, conventional mathematical optimization becomes impractical due to high computational costs and the inability to handle real-time rescheduling. Here, we propose an online rescheduling agent trained by explorative reinforcement learning that autonomously optimizes schedules while considering schedule nervousness. In a static scheduling environment, our model consistently achieves over 90% of the cost objective with scalability and flexibility. A computational time comparison proves that the reinforcement learning methodology makes near-optimal decisions rapidly, irrespective of the complexity of the scheduling problem. Furthermore, we present several realistic rescheduling scenarios that demonstrate the capability of our methodology. Our study illustrates the significant potential of reinforcement learning methodology in expediting digital transformation and process automation within real-world manufacturing systems.</p></div>","PeriodicalId":286,"journal":{"name":"Computers & Chemical Engineering","volume":null,"pages":null},"PeriodicalIF":4.3,"publicationDate":"2024-04-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140621841","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 : 2024-04-16DOI: 10.1016/j.compchemeng.2024.108688
Ilias Mitrai, Prodromos Daoutidis
In this paper, we propose a graph classification approach for automatically determining whether to use a monolithic or a decomposition-based solution method. In this approach, an optimization problem is represented as a graph that captures the structural and functional coupling among the variables and constraints of the problem via an appropriate set of features. Given this representation, a graph classifier can be built to assist a solver in selecting the best solution strategy for a given problem with respect to some metric of choice. The proposed approach is used to develop a classifier that determines whether a convex Mixed Integer Nonlinear Programming problem should be solved using branch and bound or the outer approximation algorithm. Finally, it is shown how the learned classifier can be incorporated into existing mixed integer optimization solvers.
{"title":"Taking the human out of decomposition-based optimization via artificial intelligence, Part I: Learning when to decompose","authors":"Ilias Mitrai, Prodromos Daoutidis","doi":"10.1016/j.compchemeng.2024.108688","DOIUrl":"https://doi.org/10.1016/j.compchemeng.2024.108688","url":null,"abstract":"<div><p>In this paper, we propose a graph classification approach for automatically determining whether to use a monolithic or a decomposition-based solution method. In this approach, an optimization problem is represented as a graph that captures the structural and functional coupling among the variables and constraints of the problem via an appropriate set of features. Given this representation, a graph classifier can be built to assist a solver in selecting the best solution strategy for a given problem with respect to some metric of choice. The proposed approach is used to develop a classifier that determines whether a convex Mixed Integer Nonlinear Programming problem should be solved using branch and bound or the outer approximation algorithm. Finally, it is shown how the learned classifier can be incorporated into existing mixed integer optimization solvers.</p></div>","PeriodicalId":286,"journal":{"name":"Computers & Chemical Engineering","volume":null,"pages":null},"PeriodicalIF":4.3,"publicationDate":"2024-04-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140607238","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 : 2024-04-16DOI: 10.1016/j.compchemeng.2024.108689
Wenlong Wang , Yujia Wang , Yuhe Tian , Zhe Wu
Machine learning-based model predictive control (ML-MPC) has been developed to control nonlinear processes with unknown first-principles models. While ML models can capture nonlinear dynamics of complex systems, the complexity of ML models leads to increased computation time for real-time implementation of ML-MPC. To address this issue, in this work, we propose an explicit ML-MPC framework for nonlinear processes using multi-parametric programming. Specifically, a self-adaptive approximation algorithm is first developed to obtain a piecewise linear affine function that approximates the behaviors of ML models. Then, multi-parametric quadratic programming (mpQP) problems are formulated to generate the solution map for states in discretized state–space. Furthermore, to accelerate the implementation of explicit ML-MPC, a neighbor-first search algorithm is developed. Finally, an example of a chemical reactor is used to demonstrate the effectiveness of the explicit ML-MPC.
基于机器学习的模型预测控制(ML-MPC)是为控制第一原理模型未知的非线性过程而开发的。虽然 ML 模型可以捕捉复杂系统的非线性动态,但 ML 模型的复杂性导致 ML-MPC 实时实施的计算时间增加。为解决这一问题,我们在本研究中提出了一种使用多参数编程的非线性过程显式 ML-MPC 框架。具体来说,我们首先开发了一种自适应近似算法,以获得可近似 ML 模型行为的片断线性仿射函数。然后,制定多参数二次编程(mpQP)问题,为离散状态空间中的状态生成解图。此外,为了加速显式 ML-MPC 的实现,还开发了一种邻域优先搜索算法。最后,以化学反应器为例演示了显式 ML-MPC 的有效性。
{"title":"Explicit machine learning-based model predictive control of nonlinear processes via multi-parametric programming","authors":"Wenlong Wang , Yujia Wang , Yuhe Tian , Zhe Wu","doi":"10.1016/j.compchemeng.2024.108689","DOIUrl":"https://doi.org/10.1016/j.compchemeng.2024.108689","url":null,"abstract":"<div><p>Machine learning-based model predictive control (ML-MPC) has been developed to control nonlinear processes with unknown first-principles models. While ML models can capture nonlinear dynamics of complex systems, the complexity of ML models leads to increased computation time for real-time implementation of ML-MPC. To address this issue, in this work, we propose an explicit ML-MPC framework for nonlinear processes using multi-parametric programming. Specifically, a self-adaptive approximation algorithm is first developed to obtain a piecewise linear affine function that approximates the behaviors of ML models. Then, multi-parametric quadratic programming (mpQP) problems are formulated to generate the solution map for states in discretized state–space. Furthermore, to accelerate the implementation of explicit ML-MPC, a neighbor-first search algorithm is developed. Finally, an example of a chemical reactor is used to demonstrate the effectiveness of the explicit ML-MPC.</p></div>","PeriodicalId":286,"journal":{"name":"Computers & Chemical Engineering","volume":null,"pages":null},"PeriodicalIF":4.3,"publicationDate":"2024-04-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140618941","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 : 2024-04-16DOI: 10.1016/j.compchemeng.2024.108690
Ilias Bouchkira , Abderrazak M. Latifi , Brahim Benyahia
The development of accurate and reliable mathematical models is the cornerstone for the modeling and optimization of processes. However, most of the existing models suffer from weak prediction capabilities due to poor data information content and poor parameter estimation methodologies. Several estimability approaches have been developed and increasingly implemented to address some of these issues. However, the wider adoption of these methods is still hampered by the lack of standardized and robust methodologies. In this paper, we present a Matlab Toolbox, called ESTAN, designed and developed to make estimability analysis accessible to non-specialist users. It uses a Quasi-Monte Carlo sequence to sample the main unknown parameters within their variation spaces. Then, depending on whether the studied model is computationally cheap or expensive, sensitivity indices are calculated either using the Sobol method or Fourier Amplitude Sensitivity Test (FAST). The calculated sensitivities are finally used within an orthogonalization algorithm to rank the parameters from the most to less estimable ones and to determine the estimable and non-estimable ones based on an estimability cut-off criterion. Various case studies are used to validate the toolbox and guide the users. The first one deals with the non-dynamic Toth adsorption model, while the second one deals with a dynamic batch cooling crystallization model. The main challenge with these two case studies is to show the importance of estimability analysis in the interpolation/extrapolation of model prediction capabilities. The last case addresses a computationally expensive thermodynamic model. The results for all the case studies are found to be promising, showing how the presented toolbox simplifies the investigation of the estimability analysis.
{"title":"ESTAN—A toolbox for standardized and effective global sensitivity-based estimability analysis","authors":"Ilias Bouchkira , Abderrazak M. Latifi , Brahim Benyahia","doi":"10.1016/j.compchemeng.2024.108690","DOIUrl":"https://doi.org/10.1016/j.compchemeng.2024.108690","url":null,"abstract":"<div><p>The development of accurate and reliable mathematical models is the cornerstone for the modeling and optimization of processes. However, most of the existing models suffer from weak prediction capabilities due to poor data information content and poor parameter estimation methodologies. Several estimability approaches have been developed and increasingly implemented to address some of these issues. However, the wider adoption of these methods is still hampered by the lack of standardized and robust methodologies. In this paper, we present a Matlab Toolbox, called ESTAN, designed and developed to make estimability analysis accessible to non-specialist users. It uses a Quasi-Monte Carlo sequence to sample the main unknown parameters within their variation spaces. Then, depending on whether the studied model is computationally cheap or expensive, sensitivity indices are calculated either using the Sobol method or Fourier Amplitude Sensitivity Test (FAST). The calculated sensitivities are finally used within an orthogonalization algorithm to rank the parameters from the most to less estimable ones and to determine the estimable and non-estimable ones based on an estimability cut-off criterion. Various case studies are used to validate the toolbox and guide the users. The first one deals with the non-dynamic Toth adsorption model, while the second one deals with a dynamic batch cooling crystallization model. The main challenge with these two case studies is to show the importance of estimability analysis in the interpolation/extrapolation of model prediction capabilities. The last case addresses a computationally expensive thermodynamic model. The results for all the case studies are found to be promising, showing how the presented toolbox simplifies the investigation of the estimability analysis.</p></div>","PeriodicalId":286,"journal":{"name":"Computers & Chemical Engineering","volume":null,"pages":null},"PeriodicalIF":4.3,"publicationDate":"2024-04-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.sciencedirect.com/science/article/pii/S009813542400108X/pdfft?md5=3fe7ba8d0271ede50721f411b44eb81f&pid=1-s2.0-S009813542400108X-main.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140649978","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-04-16DOI: 10.1016/j.compchemeng.2024.108685
João Bernardo Aranha Ribeiro , José Dolores Vergara Dietrich , Julio Elias Normey-Rico
This paper presents a comparative study of different control strategies to solve the gas-lift optimization (GLO) problem of offshore rigs. GLO consists of distributing the compressed gas between the wells to maximize oil production, considering several operational and process aspects such as the cost of flaring, price fluctuations, measurable noise, external disturbances, and plant-model mismatches. We compare and evaluate the performance of economic nonlinear model predictive control (ENMPC), Modifier-based EMPC (EMPC-Mod), EMPC with Local Linearization on Trajectory (EMPC-LLT), the static Real-Time Optimizer with Parameter Adaptation (ROPA), and the Active Constraint Control (ACC) based on feedback controllers. The study points out the advantages and drawbacks of each approach being useful for engineers to choose the most appropriate strategy. Moreover, the results show that the linear EMPCs and ROPA have similar performance to the theoretical optimal while maintaining minimal computational burden, and also that ACC is satisfactory for this case study.
{"title":"Comparison of economic model predictive controllers for gas-lift optimization in offshore oil and gas rigs","authors":"João Bernardo Aranha Ribeiro , José Dolores Vergara Dietrich , Julio Elias Normey-Rico","doi":"10.1016/j.compchemeng.2024.108685","DOIUrl":"https://doi.org/10.1016/j.compchemeng.2024.108685","url":null,"abstract":"<div><p>This paper presents a comparative study of different control strategies to solve the gas-lift optimization (GLO) problem of offshore rigs. GLO consists of distributing the compressed gas between the wells to maximize oil production, considering several operational and process aspects such as the cost of flaring, price fluctuations, measurable noise, external disturbances, and plant-model mismatches. We compare and evaluate the performance of economic nonlinear model predictive control (ENMPC), Modifier-based EMPC (EMPC-Mod), EMPC with Local Linearization on Trajectory (EMPC-LLT), the static Real-Time Optimizer with Parameter Adaptation (ROPA), and the Active Constraint Control (ACC) based on feedback controllers. The study points out the advantages and drawbacks of each approach being useful for engineers to choose the most appropriate strategy. Moreover, the results show that the linear EMPCs and ROPA have similar performance to the theoretical optimal while maintaining minimal computational burden, and also that ACC is satisfactory for this case study.</p></div>","PeriodicalId":286,"journal":{"name":"Computers & Chemical Engineering","volume":null,"pages":null},"PeriodicalIF":4.3,"publicationDate":"2024-04-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140607237","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 : 2024-04-16DOI: 10.1016/j.compchemeng.2024.108699
Qinming Liu , Fengze Yun , Ming Dong , Wenyuan Lv , Yuhong Liu
In order to facilitate efficient maintenance of batch production systems under the increasingly complex demands of present day, an approach is proposed. This approach addresses maintenance issues of equipment in multi-equipment batch production systems under stochastic demand. First, the method considers distinct equipment degradation characteristics and imperfect maintenance at both the system and equipment levels. It simultaneously employs the mechanism of advancing or delaying maintenance, along with a dual time window opportunity maintenance strategy, to minimize the costs associated with opportunistic maintenance. Then, different models are developed to cater to various scenarios. At the system level, maintenance is conducted through component grouping based on production transition opportunities. At the equipment level, the optimal preventive maintenance cycle duration is determined by calculating the current minimal maintenance cost rate, thus, determining the optimal preventive maintenance timing. The solution methodology employs the Monte Carlo method to simulate the production system across different batches, calculating the actual preventive maintenance timings and total maintenance costs. Finally, by illustrative cases and the optimization of the total cost over the production cycle, the effectiveness of the proposed maintenance strategy for multi-equipment batch production systems under stochastic demand is demonstrated by measuring cost against the frequency of failures.
{"title":"Condition-based maintenance optimization for multi-equipment batch production system based on stochastic demand","authors":"Qinming Liu , Fengze Yun , Ming Dong , Wenyuan Lv , Yuhong Liu","doi":"10.1016/j.compchemeng.2024.108699","DOIUrl":"https://doi.org/10.1016/j.compchemeng.2024.108699","url":null,"abstract":"<div><p>In order to facilitate efficient maintenance of batch production systems under the increasingly complex demands of present day, an approach is proposed. This approach addresses maintenance issues of equipment in multi-equipment batch production systems under stochastic demand. First, the method considers distinct equipment degradation characteristics and imperfect maintenance at both the system and equipment levels. It simultaneously employs the mechanism of advancing or delaying maintenance, along with a dual time window opportunity maintenance strategy, to minimize the costs associated with opportunistic maintenance. Then, different models are developed to cater to various scenarios. At the system level, maintenance is conducted through component grouping based on production transition opportunities. At the equipment level, the optimal preventive maintenance cycle duration is determined by calculating the current minimal maintenance cost rate, thus, determining the optimal preventive maintenance timing. The solution methodology employs the Monte Carlo method to simulate the production system across different batches, calculating the actual preventive maintenance timings and total maintenance costs. Finally, by illustrative cases and the optimization of the total cost over the production cycle, the effectiveness of the proposed maintenance strategy for multi-equipment batch production systems under stochastic demand is demonstrated by measuring cost against the frequency of failures.</p></div>","PeriodicalId":286,"journal":{"name":"Computers & Chemical Engineering","volume":null,"pages":null},"PeriodicalIF":4.3,"publicationDate":"2024-04-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140618940","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 : 2024-04-12DOI: 10.1016/j.compchemeng.2024.108681
Eike Cramer , Ji Gao
Demand side management (DSM) contributes to the industry’s transition to renewables by shifting electricity consumption in time while maintaining feasible operations. Machine learning is promising for DSM with reasonable computation times and electricity price forecasting (EPF), which is paramount to obtaining the necessary data. Increased usage of machine learning makes production processes susceptible to so-called adversarial attacks. This work proposes a black-box attack on DSM and EPF based on an adversarial surrogate model that intercepts and modifies the data flow of load forecasts and forces the DSM to result in financial losses. Notably, adversaries can design the data modifications without knowledge of the EPF model or the DSM optimization model. The results show how barely noticeable modifications of the input data lead to significant deterioration of the decisions by the optimizer. The results implicate a significant threat, as attackers can design and implement powerful attacks without infiltrating secure company networks.
{"title":"A black-box adversarial attack on demand side management","authors":"Eike Cramer , Ji Gao","doi":"10.1016/j.compchemeng.2024.108681","DOIUrl":"https://doi.org/10.1016/j.compchemeng.2024.108681","url":null,"abstract":"<div><p>Demand side management (DSM) contributes to the industry’s transition to renewables by shifting electricity consumption in time while maintaining feasible operations. Machine learning is promising for DSM with reasonable computation times and electricity price forecasting (EPF), which is paramount to obtaining the necessary data. Increased usage of machine learning makes production processes susceptible to so-called adversarial attacks. This work proposes a black-box attack on DSM and EPF based on an adversarial surrogate model that intercepts and modifies the data flow of load forecasts and forces the DSM to result in financial losses. Notably, adversaries can design the data modifications without knowledge of the EPF model or the DSM optimization model. The results show how barely noticeable modifications of the input data lead to significant deterioration of the decisions by the optimizer. The results implicate a significant threat, as attackers can design and implement powerful attacks without infiltrating secure company networks.</p></div>","PeriodicalId":286,"journal":{"name":"Computers & Chemical Engineering","volume":null,"pages":null},"PeriodicalIF":4.3,"publicationDate":"2024-04-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.sciencedirect.com/science/article/pii/S0098135424000991/pdfft?md5=1daff4517cd47f17c42632dfccaacd9b&pid=1-s2.0-S0098135424000991-main.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140555201","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-04-12DOI: 10.1016/j.compchemeng.2024.108691
Valentina Negri , Daniel Vázquez , Ignacio E. Grossmann , Gonzalo Guillén-Gosálbez
The broad portfolio of negative emissions technologies calls for integrated analyses to explore the synergies between them and the power sector, with which they display strong links. These analyses should be conducted at a regional level, considering system uncertainties, assessing local benefits and the impact on carbon removal potential. This study investigates how uncertainty in electricity demand affects the optimal design of integrated carbon removal and power generation systems using multistage stochastic programming. Given the model complexity, we propose a tailored decomposition algorithm by extending previous work on the shrinking horizon approach that reduces the computational time by 90 %, enabling insights into various European scenarios. A combination of conventional technologies and biomass could satisfy the electricity demand while providing up to 9 Gt of net CO2 removal from the atmosphere. Omitting uncertainties leads to an underestimation of the total cost and the selection of different technologies possibly leading to suboptimal performance.
{"title":"A tailored decomposition approach for optimization under uncertainty of carbon removal technologies in the EU power system","authors":"Valentina Negri , Daniel Vázquez , Ignacio E. Grossmann , Gonzalo Guillén-Gosálbez","doi":"10.1016/j.compchemeng.2024.108691","DOIUrl":"https://doi.org/10.1016/j.compchemeng.2024.108691","url":null,"abstract":"<div><p>The broad portfolio of negative emissions technologies calls for integrated analyses to explore the synergies between them and the power sector, with which they display strong links. These analyses should be conducted at a regional level, considering system uncertainties, assessing local benefits and the impact on carbon removal potential. This study investigates how uncertainty in electricity demand affects the optimal design of integrated carbon removal and power generation systems using multistage stochastic programming. Given the model complexity, we propose a tailored decomposition algorithm by extending previous work on the shrinking horizon approach that reduces the computational time by 90 %, enabling insights into various European scenarios. A combination of conventional technologies and biomass could satisfy the electricity demand while providing up to 9 Gt of net CO<sub>2</sub> removal from the atmosphere. Omitting uncertainties leads to an underestimation of the total cost and the selection of different technologies possibly leading to suboptimal performance.</p></div>","PeriodicalId":286,"journal":{"name":"Computers & Chemical Engineering","volume":null,"pages":null},"PeriodicalIF":4.3,"publicationDate":"2024-04-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141083798","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 : 2024-04-12DOI: 10.1016/j.compchemeng.2024.108684
Shiqiang Zhang , Juan S. Campos , Christian Feldmann , Frederik Sandfort , Miriam Mathea , Ruth Misener
Computer-aided molecular design (CAMD) studies quantitative structure–property relationships and discovers desired molecules using optimization algorithms. With the emergence of machine learning models, CAMD score functions may be replaced by various surrogates to automatically learn the structure–property relationships. Due to their outstanding performance on graph domains, graph neural networks (GNNs) have recently appeared frequently in CAMD. But using GNNs introduces new optimization challenges. This paper formulates GNNs using mixed-integer programming and then integrates this GNN formulation into the optimization and machine learning toolkit OMLT. To characterize and formulate molecules, we inherit the well-established mixed-integer optimization formulation for CAMD and propose symmetry-breaking constraints to remove symmetric solutions caused by graph isomorphism. In two case studies, we investigate fragment-based odorant molecular design with more practical requirements to test the compatibility and performance of our approaches.
{"title":"Augmenting optimization-based molecular design with graph neural networks","authors":"Shiqiang Zhang , Juan S. Campos , Christian Feldmann , Frederik Sandfort , Miriam Mathea , Ruth Misener","doi":"10.1016/j.compchemeng.2024.108684","DOIUrl":"https://doi.org/10.1016/j.compchemeng.2024.108684","url":null,"abstract":"<div><p>Computer-aided molecular design (CAMD) studies quantitative structure–property relationships and discovers desired molecules using optimization algorithms. With the emergence of machine learning models, CAMD score functions may be replaced by various surrogates to automatically learn the structure–property relationships. Due to their outstanding performance on graph domains, graph neural networks (GNNs) have recently appeared frequently in CAMD. But using GNNs introduces new optimization challenges. This paper formulates GNNs using mixed-integer programming and then integrates this GNN formulation into the optimization and machine learning toolkit OMLT. To characterize and formulate molecules, we inherit the well-established mixed-integer optimization formulation for CAMD and propose symmetry-breaking constraints to remove symmetric solutions caused by graph isomorphism. In two case studies, we investigate fragment-based odorant molecular design with more practical requirements to test the compatibility and performance of our approaches.</p></div>","PeriodicalId":286,"journal":{"name":"Computers & Chemical Engineering","volume":null,"pages":null},"PeriodicalIF":4.3,"publicationDate":"2024-04-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.sciencedirect.com/science/article/pii/S0098135424001029/pdfft?md5=03eba5c2041e40bcbb340c9c30405539&pid=1-s2.0-S0098135424001029-main.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140554289","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}