scispace - formally typeset
Search or ask a question

Showing papers in "Aiche Journal in 2019"



Journal ArticleDOI
TL;DR: In this paper, a triple column pressure swing distillation for ternary systems with three binary minimum azeotropes is proposed, which involves thermodynamic insights, a two-step optimization method, and effective control strategy.
Abstract: The separation of ternary nonideal systems with multi‐azeotrope is very important because they are often found in the waste of chemical and pharmaceutical industries, which is much more difficult due to the formation of multi‐azeotrope and distillation boundary. We propose a systematic procedure for design and control of a triple‐column pressure‐swing distillation for separating ternary systems with three binary minimum azeotropes. This procedure involves thermodynamic insights, a two‐step optimization method, and effective control strategy. The separation of tetrahydrofuran (THF)/ethanol/water is used to illustrate the capability of the proposed procedure. It is found that the pressure limits in columns can be determined through the analysis of residue curve maps, distillation boundary, and isovolatility curves. The optimal triple‐column pressure‐swing distillation is generated with the minimum total annual cost (TAC) of $2.181 × 106 in sequence A. The operating conditions are well controlled approaching their desired specifications in an acceptable time when disturbances occur.

156 citations



Journal ArticleDOI
TL;DR: Machine learning ensemble regression modeling tools are employed in the formulation of LMPC to improve prediction accuracy of RNN models and overall closed-loop performance while parallel computing is utilized to reduce computation time.
Abstract: Correspondence Panagiotis Christofides, Department of Chemical and Biomolecular Engineering, University of California, Los Angeles, CA. Email: pdc@seas.ucla.edu Abstract This article focuses on the design of model predictive control (MPC) systems for nonlinear processes that utilize an ensemble of recurrent neural network (RNN) models to predict nonlinear dynamics. Specifically, RNN models are initially developed based on a data set generated from extensive open-loop simulations within a desired process operation region to capture process dynamics with a sufficiently small modeling error between the RNN model and the actual nonlinear process model. Subsequently, Lyapunov-based MPC (LMPC) that utilizes RNN models as the prediction model is developed to achieve closed-loop state boundedness and convergence to the origin. Additionally, machine learning ensemble regression modeling tools are employed in the formulation of LMPC to improve prediction accuracy of RNN models and overall closed-loop performance while parallel computing is utilized to reduce computation time. Computational implementation of the method and application to a chemical reactor example is discussed in the second article of this series.

122 citations


Journal ArticleDOI
TL;DR: The DRL controller is proposed is a data-based controller that learns the control policy in real time by merely interacting with the process and is demonstrated through many simulations.
Abstract: Advanced model-based controllers are well established in process industries. However, such controllers require regular maintenance to maintain acceptable performance. It is a common practice to monitor controller performance continuously and to initiate a remedial model re-identification procedure in the event of performance degradation. Such procedures are typically complicated and resource-intensive, and they often cause costly interruptions to normal operations. In this paper, we exploit recent developments in reinforcement learning and deep learning to develop a novel adaptive, model-free controller for general discrete-time processes. The DRL controller we propose is a data-based controller that learns the control policy in real time by merely interacting with the process. The effectiveness and benefits of the DRL controller are demonstrated through many simulations.

99 citations



Journal ArticleDOI
TL;DR: This work proposes an innovative data-driven surrogate modelling framework which considerably reduces computing time from months to days by exploiting state-of-the-art deep learning technology.
Abstract: Identifying optimal photobioreactor configurations and process operating conditions is critical to industrialize microalgae-derived biorenewables. Traditionally, this was addressed by testing numerous design scenarios from integrated physical models coupling computational fluid dynamics and kinetic modelling. However, this approach presents computational intractability and numerical instabilities when simulating large-scale systems, causing time-intensive computing efforts and infeasibility in mathematical optimization. Therefore, we propose an innovative data-driven surrogate modelling framework which considerably reduces computing time from months to days by exploiting state-of-the-art deep learning technology. The framework built upon a few simulated results from the physical model to learn the sophisticated hydrodynamic and biochemical kinetic mechanisms; then adopts a hybrid stochastic optimization algorithm to explore untested processes and find optimal solutions. Through verification, this framework was demonstrated to have comparable accuracy to the physical model. Moreover, multi-objective optimization was incorporated to generate a Pareto-frontier for decision-making, advancing its applications in complex biosystems modelling and optimization.

78 citations




Journal ArticleDOI
TL;DR: A feature selection algorithm based on nonlinear support vector machine (SVM) for fault detection and diagnosis in continuous processes and results for the Tennessee Eastman benchmark process are presented.
Abstract: In this article, we present (1) a feature selection algorithm based on nonlinear support vector machine (SVM) for fault detection and diagnosis in continuous processes and (2) results for the Tennessee Eastman benchmark process. The presented feature selection algorithm is derived from the sensitivity analysis of the dual C-SVM objective function. This enables simultaneous modeling and feature selection paving the way for simultaneous fault detection and diagnosis, where feature ranking guides fault diagnosis. We train fault-specific two-class SVM models to detect faulty operations, while using the feature selection algorithm to improve the accuracy and perform the fault diagnosis. Our results show that the developed SVM models outperform the available ones in the literature both in terms of detection accuracy and latency. Moreover, it is shown that the loss of information is minimized with the use of feature selection techniques compared to feature extraction techniques such as principal component analysis (PCA). This further facilitates a more accurate interpretation of the results.

64 citations


Journal ArticleDOI
TL;DR: In this paper, the integration of Koopman operator methodology with Lyapunov-based model predictive control (LMPC) for stabilization of nonlinear systems is proposed, which results in a standard convex optimization problem which is computationally attractive compared to a nonconvex problem encountered when the original nonlinear model is used.
Abstract: Funding information National Science Foundation, Grant/Award Number: CBET-1804407 Abstract In this work, we propose the integration of Koopman operator methodology with Lyapunov-based model predictive control (LMPC) for stabilization of nonlinear systems. The Koopman operator enables global linear representations of nonlinear dynamical systems. The basic idea is to transform the nonlinear dynamics into a higher dimensional space using a set of observable functions whose evolution is governed by the linear but infinite dimensional Koopman operator. In practice, it is numerically approximated and therefore the tightness of these linear representations cannot be guaranteed which may lead to unstable closed-loop designs. To address this issue, we integrate the Koopman linear predictors in an LMPC framework which guarantees controller feasibility and closed-loop stability. Moreover, the proposed design results in a standard convex optimization problem which is computationally attractive compared to a nonconvex problem encountered when the original nonlinear model is used. We illustrate the application of this methodology on a chemical process example.


Journal ArticleDOI
TL;DR: In this article, a mixed integer linear programming model for the optimal planning of a waste management system in a multi-echelon supply chain network was developed, which aims to find a trade-off between supply chain costs, depletion of waste and efficient use of generated waste while considering the environmental impacts.
Abstract: The growing waste generation, increasing environmental regulations, and limited land area for waste disposal necessitate an effective and efficient waste supply chain management solution in terms of both socioeconomic perspectives and environmental sustainability. Waste management is connected to supply chain decisions as it involves waste generation, collection, separation, transportation, processing, and disposal. Accordingly, this paper develops a mixed integer linear programming model for the optimal planning of a waste management system in a multi-echelon supply chain network, which aims to find a trade-off between supply chain costs, depletion of waste and efficient use of generated waste while considering the environmental impacts. Various recycling and WtE technologies are utilized to convert plastic and mixed waste into value-added products including fuel, electricity, and heat. Though recycling is preferable from an environmental point of view, it is shown that the waste-toenergy option is more economically efficient.









Journal ArticleDOI
TL;DR: The eNMPC applies local optimization methods and achieves profiles similar to the scheduling solved using deterministic global optimization methods over the complete closed-loop simulation time horizon, and enables economic improvements similar to an idealized quasistationary scheduling.
Abstract: Correspondence Alexander Mitsos, AVT Process Systems Engineering, RWTH Aachen University, 52074 Aachen, Germany. Email: amitsos@alum.mit.edu Abstract The penetration of renewable electricity promises an economic advantage for flexible operation of energy-intense processes. One way to achieve flexible operation is economic model predictive control (eNMPC), where an economic dynamic optimization problem is directly solved at controller level taking into account a process model and operational constraints. We apply eNMPC in silico to an air separation process with an integrated liquefier and liquid-assist operation. We use a mechanistic dynamic model as both controller model and plant surrogate. We conduct a closed-loop case study over a time horizon of 2 days with historical electricity prices and input disturbances. We solve the dynamic optimization problems in DyOS. Compared to the optimal steady-state operation, the eNMPC operating strategy gives a significant improvement of 14%. We further show that the eNMPC enables economic improvements similar to an idealized quasistationary scheduling. While the eNMPC provides control profiles qualitatively similar to those obtained from deterministic global optimization of quasistationary scheduling, the eNMPC satisfies the product purity constraints all the time whereas the quasistationary scheduling sometimes fails to do so. The eNMPC applies local optimization methods and achieves profiles similar to the scheduling solved using deterministic global optimization methods over the complete closed-loop simulation time horizon.





Journal ArticleDOI
TL;DR: In this article, a versatile pore network model is used to study deactivation by coking in a single catalyst particle, which allows to gain detailed insights into the progression of deactivation from active site, to pore, and to particle, providing valuable information for catalyst design.
Abstract: A versatile pore network model is used to study deactivation by coking in a single catalyst particle. This approach allows to gain detailed insights into the progression of deactivation from active site, to pore, and to particle – providing valuable information for catalyst design. The model is applied to investigate deactivation by coking during propane dehydrogenation in a Pt‐Sn/Al2O3 catalyst particle. We find that the deactivation process can be separated into two stages when there exist severe diffusion limitation and pore blockage, and the toxicity of coke formed in the later stage is much stronger than of coke formed in the early stage. The reaction temperature and composition change the coking rate and apparent reaction rate, informed by the kinetics, but, remarkably, they do not change the capacity for a catalyst particle to accommodate coke. On the other hand, the pore network structure significantly affects the capacity to contain coke.