Preferential crystallization: Multi‐objective optimization framework

A four objective optimization framework for preferential crystallization of D‐L threonine solution is presented. The objectives are maximization of average crystal size and productivity, and minimization of batch time and the coefficient of variation at the desired purity while respecting design and operating constraints. The cooling rate, enantiomeric excess of the preferred enantiomer, and the mass of seeds are used as the decision variables. The optimization problem is solved by using adaptation of the nondominated sorting genetic algorithm. The results obtained clearly distinguish different regimes of interest during preferential crystallization. The multi‐objective analysis presented in this study is generic and gives a simplified picture in terms of three zones of operations obtained because of relative importance of nucleation and growth. Such analysis is of great importance in providing better insight for design and decision making, and improving the performance of the preferential crystallization that is considered as a promising future alternative to chromatographic separation of enantiomers. © 2009 American Institute of Chemical Engineers AIChE J, 2009     

Optimal design of large‐scale chemical processes under uncertainty: A ranking‐based approach

An approach for the optimal design of chemical processes in the presence of uncertainty was presented. The key idea in this work is to approximate the process constraint functions and model outputs using Power Series Expansions (PSE)‐based functions. The PSE functions are used to efficiently identify the variability in the process constraint functions and model outputs due to multiple realizations in the uncertain parameters using Monte Carlo (MC) sampling methods. A ranking‐based approach is adopted here where the user can assign priorities or probabilities of satisfaction for the different process constraints and model outputs considered in the analysis. The methodology was tested on a reactor–heat exchanger system and the Tennessee Eastman process. The results show that the present method is computationally attractive since the optimal process design is accomplished in shorter computational times when compared to the use of the MC method applied to the full plant model. © 2014 American Institute of Chemical Engineers AIChE J, 60: 3243–3257, 2014     

Equation‐oriented optimization of process flowsheets with dividing‐wall columns

We present a new modeling approach for dividing‐wall columns (DWCs) that is amenable to equation‐oriented flowsheet simulation and optimization. The material, equilibrium, summation, and heat (MESH) equations describing a DWC are highly coupled and nonlinear, making DWC‐based process flowsheets challenging to simulate. Design optimization poses further challenges, typically requiring integer variables to select the number of column stages. To address these difficulties, we represent DWCs as networks of pseudo‐transient (differential‐algebraic) subunit models. We show that these networks have the same steady‐state solution as the original (algebraic) MESH equations, but present significant numerical benefits. We then embed these models in a previously developed pseudo‐transient flowsheet modeling and optimization framework. We further reformulate the models to require only continuous decision variables when selecting the optimal number of stages during design optimization. To illustrate these concepts, we discuss the DWC‐based intensification of the dimethyl ether process. © 2015 American Institute of Chemical Engineers AIChE J, 62: 704–716, 2016     

A multi‐objective optimization approach to polygeneration energy systems design

Polygeneration, typically involving co‐production of methanol and electricity, is a promising energy conversion technology which provides opportunities for high energy utilization efficiency and low/zero emissions. The optimal design of such a complex, large‐scale and highly nonlinear process system poses significant challenges. In this article, we present a multiobjective optimization model for the optimal design of a methanol/electricity polygeneration plant. Economic and environmental criteria are simultaneously optimized over a superstructure capturing a number of possible combinations of technologies and types of equipment. Aggregated models are considered, including a detailed methanol synthesis step with chemical kinetics and phase equilibrium considerations. The resulting model is formulated as a non‐convex mixed‐integer nonlinear programming problem. Global optimization and parallel computation techniques are employed to generate an optimal Pareto frontier. © 2009 American Institute of Chemical Engineers AIChE J, 2010     

Multistream heat exchangers: Equation‐oriented modeling and flowsheet optimization

Multistream heat exchangers (MHEXs), typically of the plate‐fin or spiral‐wound type, are a key enabler of heat integration in cryogenic processes. Equation‐oriented modeling of MHEXs for flowsheet optimization purposes is challenging, especially when streams undergo phase transformations. Boolean variables are typically used to capture the effect of phase changes, adding considerable difficulty to solving the flowsheet optimization problem. A novel optimization‐oriented MHEX modeling approach that uses a pseudo‐transient approach to rapidly compute stream temperatures without requiring Boolean variables is presented. The model also computes an approximate required heat exchange area to determine the optimal tradeoff between operating and capital expenses. Subsequently, this model is seamlessly integrated in a previously‐introduced pseudo‐transient process modeling and flowsheet optimization framework. Our developments are illustrated with two optimal design case studies, an MHEX representative of air separation operation and a natural gas liquefaction process. © 2015 American Institute of Chemical Engineers AIChE J, 61: 1856–1866, 2015     

Improving the design of depth filters: A model‐based method using optimal control theory

Because there is no general design method for depth filters, especially for layered configurations, this methodological gap is addressed here. Using optimal control theory, paths of the filter coefficient, a measure for local filtration performance, are determined along the filter depth. An analytical optimal control solution is derived and used to validate the numerical algorithm. Two optimal control scenarios are solved numerically: In the first scenario, the goal of constant deposition along the filter depth is addressed. The second scenario aims at maximizing the time until some maximal pressure drop is reached. Furthermore, a computational strategy is presented to derive discrete layers suitable for practical design from the continuous optimal control solutions. All optimized scenarios are compared to one‐layered filter designs and significant improvements are found. As this work is based on strongly validated and widely used filtration models, the presented methods are expected to have broad applicability. © 2017 American Institute of Chemical Engineers AIChE J, 63: 68–76, 2018     

High‐pressure enzymatic hydrolysis of oil

The hydrolysis of sunflower and soybean oil, catalyzed by two enzymes, non‐immobilized Candida rugosa and immobilized Candida antarctica lipase, was performed at atmospheric and high‐pressure. The results showed that at atmospheric pressure between 40 °C and 60 °C initial reaction rates were influenced by the temperature variation, as expected. Due to favorable physico‐chemical properties of dense gases as reaction media, hydrolysis of soybean oil was performed in non‐conventional solvents: in supercritical (SC) CO₂ and near‐critical propane. In SC CO₂ the activity of non‐immobilized Candida rugosa lipase decreased while the reaction rates of hydrolysis catalyzed by immobilized Candida antarctica lipase were 1.5‐fold higher than at atmospheric pressure. However, the reaction rates for the hydrolyses catalyzed by both lipases, were much higher in propane than at atmospheric pressure.     

Optimization‐based support for process design under uncertainty: A case study

The use of two‐stage stochastic optimization for the support of the solution of process design problems in the early phase of process development where the different potential elements of the production process can only be described with significant uncertainty is discussed. The first stage variables are the design decisions which are fixed after the process has been built, while the second stage variables are the operational parameters which can be adapted to the realization of the uncertainties. The application of the approach to the design of a hydroformylation process in a thermomorphic solvent system is demonstrated. The proposed designs which are computed using the software framework FSOpt are analyzed and compared using different graphic representations which provide insight into what the most important design decisions are. Finally, the experience with the proposed formulation and solution techniques and point out where further advances are needed is reviewed. © 2016 American Institute of Chemical Engineers AIChE J, 62: 3404–3419, 2016     

A backoff strategy for model‐based experiment design under parametric uncertainty

Model‐based experiment design techniques are an effective tool for the rapid development and assessment of dynamic deterministic models, yielding the most informative process data to be used for the estimation of the process model parameters. A particular advantage of the model‐based approach is that it permits the definition of a set of constraints on the experiment design variables and on the predicted responses. However, uncertainty in the model parameters can lead the constrained design procedure to predict experiments that turn out to be, in practice, suboptimal, thus decreasing the effectiveness of the experiment design session. Additionally, in the presence of parametric mismatch, the feasibility constraints may well turn out to be violated when that optimally designed experiment is performed, leading in the best case to less informative data sets or, in the worst case, to an infeasible or unsafe experiment. In this article, a general methodology is proposed to formulate and solve the experiment design problem by explicitly taking into account the presence of parametric uncertainty, so as to ensure both feasibility and optimality of the planned experiment. A prediction of the system responses for the given parameter distribution is used to evaluate and update suitable backoffs from the nominal constraints, which are used in the design session to keep the system within a feasible region with specified probability. This approach is particularly useful when designing optimal experiments starting from limited preliminary knowledge of the parameter set, with great improvement in terms of design efficiency and flexibility of the overall iterative model development scheme. The effectiveness of the proposed methodology is demonstrated and discussed by simulation through two illustrative case studies concerning the parameter identification of physiological models related to diabetes and cancer care. © 2009 American Institute of Chemical Engineers AIChE J, 2010     

Simulation,model‐reduction,and state estimation of a two‐component coagulation process

The issue of state estimation of an aggregation process through (1) using model reduction to obtain a tractable approximation of the governing dynamics and (2) designing a fast moving‐horizon estimator for the reduced‐order model is addressed. The method of moments is first used to reduce the governing integro‐differential equation down to a nonlinear ordinary differential equation. This reduced‐order model is then simulated for both batch and continuous processes and the results are shown to agree with constant Number Monte Carlo simulation results of the original model. Next, the states of the reduced order model are estimated in a moving horizon estimation approach. For this purpose, Carleman linearization is first employed and the nonlinear system is represented in a bilinear form. This representation lessens the computation burden of the estimation problem by allowing for analytical solution of the state variables as well as sensitivities with respect to decision variables. © 2016 American Institute of Chemical Engineers AIChE J, 62: 1557–1567, 2016     

A centroid‐based sampling strategy for kriging global modeling and optimization

A new sampling strategy is presented for kriging‐based global modeling. The strategy is used within a kriging/response surface (RSM) algorithm for solving NLP containing black‐box models. Black‐box models describe systems lacking the closed‐form equations necessary for conventional gradient‐based optimization. System optima can be alternatively found by building iteratively updated kriging models, and then refining local solutions using RSM. The application of the new sampling strategy results in accurate global model generation at lower sampling expense relative to a strategy using randomized and heuristic‐based sampling for initial and subsequent model construction, respectively. The new strategy relies on construction of an initial kriging model built using sampling data obtained at the feasible region's convex polytope vertices and centroid. Updated models are constructed using additional sampling information obtained at Delaunay triangulation centroids. The new sampling algorithm is applied within the kriging‐RSM framework to several numerical examples and case studies to demonstrate proof of concept. © 2009 American Institute of Chemical Engineers AIChE J, 2010     

Multi‐rate observer design for process monitoring using asynchronous inter‐sample output predictions

In this article, the problem of observer design in linear multi‐output systems with asynchronous sampling is addressed. The proposed multi‐rate observer is based on a continuous‐time Luenberger observer design coupled with an inter‐sample predictor for each sampled measurement, which generates an estimate of the output in between consecutive measurements. The sampling times are not necessarily uniformly spaced, but there exists a maximum sampling period among all the sensors. Sufficient and explicit conditions are derived to guarantee exponential stability of the multi‐rate observer. The proposed framework of multi‐rate observer design is examined through a mathematical example and a gas‐phase polyethylene reactor. In the latter case, the amount of active catalyst sites is estimated, with a convergence rate that is comparable to the case of continuous measurements. © 2017 American Institute of Chemical Engineers AIChE J, 63: 3384–3394, 2017     

A quadratic approximation‐based algorithm for the solution of multiparametric mixed‐integer nonlinear programming problems

An algorithm for the solution of convex multiparametric mixed‐integer nonlinear programming problems arising in process engineering problems under uncertainty is introduced. The proposed algorithm iterates between a multiparametric nonlinear programming subproblem and a mixed‐integer nonlinear programming subproblem to provide a series of parametric upper and lower bounds. The primal subproblem is formulated by fixing the integer variables and solved through a series of multiparametric quadratic programming (mp‐QP) problems based on quadratic approximations of the objective function, while the deterministic master subproblem is formulated so as to provide feasible integer solutions for the next primal subproblem. To reduce the computational effort when infeasibilities are encountered at the vertices of the critical regions (CRs) generated by the primal subproblem, a simplicial approximation approach is used to obtain CRs that are feasible at each of their vertices. The algorithm terminates when there does not exist an integer solution that is better than the one previously used by the primal problem. Through a series of examples, the proposed algorithm is compared with a multiparametric mixed‐integer outer approximation (mp‐MIOA) algorithm to demonstrate its computational advantages. © 2012 American Institute of Chemical Engineers AIChE J, 59: 483–495, 2013     

Error‐triggered on‐line model identification for model‐based feedback control

In industry, it may be difficult in many applications to obtain a first‐principles model of the process, in which case a linear empirical model constructed using process data may be used in the design of a feedback controller. However, linear empirical models may not capture the nonlinear dynamics over a wide region of state‐space and may also perform poorly when significant plant variations and disturbances occur. In the present work, an error‐triggered on‐line model identification approach is introduced for closed‐loop systems under model‐based feedback control strategies. The linear models are re‐identified on‐line when significant prediction errors occur. A moving horizon error detector is used to quantify the model accuracy and to trigger the model re‐identification on‐line when necessary. The proposed approach is demonstrated through two chemical process examples using a model‐based feedback control strategy termed Lyapunov‐based economic model predictive control (LEMPC). The chemical process examples illustrate that the proposed error‐triggered on‐line model identification strategy can be used to obtain more accurate state predictions to improve process economics while maintaining closed‐loop stability of the process under LEMPC. © 2016 American Institute of Chemical Engineers AIChE J, 63: 949–966, 2017     

Nonlinear robust optimization for process design

A novel robust optimization framework is proposed to address general nonlinear problems in process design. Local linearization is taken with respect to the uncertain parameters around multiple realizations of the uncertainty, and an iterative algorithm is implemented to solve the problem. Furthermore, the proposed methodology can handle different categories of problems according to the complexity of the problems. First, inequality‐only constrained optimization problem as studied in most existing robust optimization methods can be addressed. Second, the proposed framework can deal with problems with equality constraint associated with uncertain parameters. In the final case, we investigate problems with operation variables which can be adjusted according to the realizations of uncertainty. A local affinely adjustable decision rule is adopted for the operation variables (i.e., an affine function of the uncertain parameter). Different applications corresponding to different classes of problems are used to demonstrate the effectiveness of the proposed nonlinear robust optimization framework. © 2017 American Institute of Chemical Engineers AIChE J, 64: 481–494, 2018     

Conceptual design of ammonia‐based energy storage system: System design and time‐invariant performance

Chemicals‐based energy storage is promising for integrating intermittent renewables on the utility scale. High round‐trip efficiency, low cost, and considerable flexibility are desirable. To this end, an ammonia‐based energy storage system is proposed. It utilizes a pressurized reversible solid‐oxide fuel cell for power conversion, coupled with external ammonia synthesis and decomposition processes and a steam power cycle. A coupled refrigeration cycle is utilized to recycle nitrogen completely. Pure oxygen, produced as a side‐product in electrochemical water splitting, is used to drive the fuel cell. A first‐principle process model extended by detailed cost calculation is used for process optimization. In this work, the performance of a 100 MW system under time‐invariant operation is studied. The system can achieve a round‐trip efficiency as high as 72%. The lowest levelized cost of delivered energy is obtained at 0.24 $/kWh, which is comparable to that of pumped hydro and compressed air energy storage systems. © 2016 American Institute of Chemical Engineers AIChE J, 63: 1620–1637, 2017     

Simultaneous solution approach to model‐based experimental design

A model‐based experimental design is formulated and solved as a large‐scale NLP problem. The key idea of the proposed approach is the extension of model equations with sensitivity equations forming an extended sensitivities‐state equation system. The resulting system is then totally discretized and simultaneously solved as constraints of the NLP problem. Thereby, higher derivatives of the parameter sensitivities with respect to the control variables are directly calculated and exact. This is an advantage in comparison with proposed sequential approaches to model‐based experimental design so far, where these derivatives have to be additionally integrated throughout the optimization steps. This can end up in a high‐computational load especially for systems with many control variables. Furthermore, an advanced sampling strategy is proposed which combines the strength of the optimal sampling design and the variation of the collocation element lengths while fully using the entire optimization space of the simultaneous formulation. © 2013 American Institute of Chemical Engineers AIChE J, 59: 4169–4183, 2013     

Learning surrogate models for simulation‐based optimization

A central problem in modeling, namely that of learning an algebraic model from data obtained from simulations or experiments is addressed. A methodology that uses a small number of simulations or experiments to learn models that are as accurate and as simple as possible is proposed. The approach begins by building a low‐complexity surrogate model. The model is built using a best subset technique that leverages an integer programming formulation to allow for the efficient consideration of a large number of possible functional components in the model. The model is then improved systematically through the use of derivative‐free optimization solvers to adaptively sample new simulation or experimental points. Automated learning of algebraic models for optimization (ALAMO), the computational implementation of the proposed methodology, along with examples and extensive computational comparisons between ALAMO and a variety of machine learning techniques, including Latin hypercube sampling, simple least‐squares regression, and the lasso is described. © 2014 American Institute of Chemical Engineers AIChE J, 60: 2211–2227, 2014