Synthesis framework for distillation sequence with sidestream columns: Application in reaction-separation-recycle system

https://doi.org/10.1016/j.cherd.2020.12.005Get rights and content

Highlights

  • State-Task Network representation for distillation sequence with sidestream column

  • Nonsharp distillation design through modified Fenske-Underwood-Gilliland method

  • Four-component case with mass interaction between reaction and separation systems

  • 14.11% and 5.70% reduction in operational cost and TAC for reference case

Abstract

In this work, a comprehensive design method for three-section sidestream columns along with a novel State Task Network representation including both conventional and sidestream distillation sequence configurations is present. The distillation network is embedded in the superstructure-based framework for reaction-separation-recycle process synthesis to explore the mass interaction in different sub-systems. In the superstructure, complete interconnections among reactor modules as well as necessary connections between reactor modules and distillation columns are involved to guarantee opportune coupled degree. Based on this, component distribution in distillation through Underwood equations is introduced to provide desired products with required purity and multiple component mixtures as feedback to the reactor network, achieving the optimization of mass interaction in the whole flowsheet. Furthermore, Fenske-Underwood-Gilliland shortcut method to design columns with and without sidestream, and finite element with orthogonal collocation approach to design PFRs are employed, so that reasonable tradeoff between model accuracy and solution difficulty is achieved. Finally, two literature examples are performed to demonstrate the feasibility and validity of the proposed methodology.

Introduction

Process synthesis has been providing promising novel solutions to assembly and interconnection of various unit operations into process flowsheets which can convert material and energy inputs into target outputs during the past five decades (Hendry et al., 1973). Great progress has been made for tools and methodology of process synthesis in the areas of both fundamental and application (Westerberg, 2004; Skiborowski, 2018; Martín and Adams, 2019). Methods for performing process synthesis can be classified as heuristic (Smith, 2005), mathematical optimization (Chen and Grossmann, 2017) and hybrid approaches, which can complement each other in solving industrial problems. In the early stage of process synthesis, the process flowsheets for chemical production were hierarchically decomposed into a series of sub-systems (Douglas, 1985) due to its combinatorial explosion nature, resulting in synthesis problems of reactor network, separation sequence, heat exchanger network, and other supporting processes (Cremaschi, 2015).

As the main separation method, distillation handles more than 90% of all the separation and purification tasks and usually accounts for no less than 40% of the total energy consumption in chemical or refinery plants (Caballero and Grossmann, 2014). As a result of its prominence and expense, chemical engineers have been long seeking solutions to reduce the energy consumption and capital cost of distillation process. A huge number of potentially energy-saving alternatives have been proposed, where a sidestream column is one of the most widely applied configurations (Lin et al., 2020). It is well established that the use of sidestream columns can frequently lead to significant savings of capital cost and/or energy consumption (Douglas, 1995; Watzdorf et al., 1999; Gutierrez-Antonio and Jimenez-Gutierrez, 2007). Distillation columns with sidestreams can be applied for separation of any zeotropic multicomponent mixture into three or more products. The energy input for the primary separation is further used to operate the secondary separation in addition (Kraller et al., 2016). Therefore, a column with a sidestream can often be used to replace a sequence of two columns. However, the purity of the sidestream product is limited by thermodynamics and the nature of the distillation process (Wang and Smith, 2005). These columns are normally appropriate when the sidestream product is not required with a high purity, or when there are no strict requirements on the composition of recycle streams (Alatiqi and Luyben, 1985), or when two of the components of the ternary mixture dominate the feed composition (Tedder and Rudd, 1978; Elaahl and Luyben, 1983). Besides, columns can also be designed to make very high-purity sidestream products, through large reflux ratios and a large number of stages, which will often be advantageous when the distillate or bottoms purity requirements are quite high (Kim and Wankat, 2004). Moreover, if the difference in volatilities between components is sufficient, it should be cost-efficient to achieve a sidestream product with high purity of the intermediate component(s) (Timoshenko et al., 2011). Generally, there are three types of approaches for the design of sidestream columns (Kong and Maravelias, 2019), including shortcut (Glinos and Malone, 1985; Adiche and Aissa, 2016), rigorous (Yeomans and Grossmann, 2000) and graphical methods (Rooks et al., 1996; Beneke and Linninger, 2011).

Syntheses of distillation sequence involving sidestream columns have also been carried out. Tedder and Shoaei (1990) provided the solution space of distillation network with up to fifteen final products. Shah and Kokossis (2002) proposed a mixed integer linear programming (MILP) framework for synthesis of distillation sequences containing sidestream columns through combination of pre-optimized separation tasks. Rong et al. (Errico and Rong, 2012; Rong, 2014) discussed the process intensification feature for intensified simple column configurations using less than N-1 columns and provided a strategy-wise method for generation of intensified nonsharp distillation systems with fewer columns. Shenvi et al. (2012) described a matrix-based method for synthesis of distillation configurations with less than N-1 columns, which can separate any zeotropic n-component feed into n products. Cui et al. (2019) proposed an enumeration-based approach for optimizing sidestream distillation configurations with only liquid sidestreams through a stochastic algorithm.

However, conventional sequential decomposition methods are insufficient to deal with interactions among different decision layers of process flowsheet synthesis, leading to unreasonable purity requirement of intermediate streams and inflexible feed condition for distillation system, which should be codetermined by the whole process. Thus, the final design cannot guarantee to be the best alternative for the given objective function. On the other hand, continual advances in flowsheet synthesis area have been made dependent on the rapid evolutions in optimization algorithms and computing performance.

Grossmann and co-authors (Grossmann, 1985; Duran and Grossmann, 1986) proposed a mixed-integer programming approach for a general synthesis framework of processing systems for a chemical plant. Henceforth, superstructure-based mathematical optimization has been adopted by many researchers as one of the most promising tools (Quirante et al., 2018; Mencarelli et al., 2020). For superstructure based mathematical optimization, an initial superstructure with appropriate solution space must be predefined, whose mathematical model is usually formulated into a nonlinear programming (NLP), a mixed integer nonlinear programming (MINLP) or a generalized disjunctive programming (GDP) (Wu et al., 2016). Since a reactor is the core in producing desired products and a distillation separator takes the main proportion of energy consumption in a chemical process, the reaction-separation system synthesis becomes an important branch in process flowsheet synthesis.

For reaction-separation system synthesis, researches on multifarious concerns have been reported. Zhang et al. (2018) presented a simultaneous synthesis method for the rigorous design of reaction-separation processes through GDP. Yin and Liu (2019) proposed an automatic approach to identify the optimal distillation sequence in a reaction-separation system. A simultaneous optimization framework of reactor and distillation network considering multiple reaction paths was provided by Kong and Maravelias (2020a) with the use of matrix presentation for distillation sequence superstructure, which is a comprehensive research for reaction-separation process synthesis with pure recycles. In this research, the reaction and separation systems can be considered in a unidirectional association.

However, constantly exploration for improvement to fulfill ever-evolving demands is urgently needed in chemical industry such as minimizing energy consumption, waste generation and other performance metric (Sitter et al., 2018). Thus, intensified mass and/or energy interaction in reaction-separation system emerged to provide long-term solutions for the aforementioned challenges. Kokossis and Floudas (1991) provided a simultaneous synthesis approach for isothermal reactor-separator-recycle systems with near complete interconnections from separators to reactors. Based on this work, Ye et al. (2020) further enhanced the coupled degree between reaction and separation systems through sloppy splits and component distribution in distillation columns. At the same time, great progress has been made in both stochastic and deterministic optimization algorithms. Neveux (2018) reported an ab-inito process synthesis method with evolutionary programming.

To address the limitation of systems, research on mass interaction among units of reactors and separators has also been carried out. In an early work, the design and control for a chemical process with two reaction steps, three distillation columns, and two recycle streams was presented by Luyben and Luyben (1995). Recker et al. (2015) proposed an unifying framework for integrated reaction-separation process design through stepwise optimization strategy, in which arbitrary interconnections among reactor and separators are allowed. Lately, a MILP framework to explore the maximum economic potential of process consisting of reaction and separation systems as well as heat integration was provided by Kong and Shah (2016, 2017). Nezhadfard et al. (2018) developed a reaction/distillation matrix to systematically generate sequences with only simple two component reactions. Skiborowski et al. (2018) presented a shortcut method for distillation-based process synthesis through a reformulation of the feed angle method, which can provide high accuracy while decreasing solution difficulty. In this research, to achieve more intensive mass interactions between reaction and separation systems, coupling of reaction and separation at unit level is studied. However, to deal with the dramatic increase in solution space, either multi-stage optimization or simplification of process description is usually forced to be used.

Not only that, both mass and energy interactions also has been simultaneously considered, which can be seen as coupling of reaction and separation at element level. Balakrishna and Biegler (1993) presented a unified approach for simultaneous synthesis of reaction, energy, and separation systems with simplified separators. Meanwhile, Lakshmanan and Biegler (1996b) proposed a synthesis methodology for chemical reactor networks with simultaneous mass integration through mass exchangers. Moreover, kinetic bounds exploration on productivity and selectivity for reactor-separator systems was carried out and a series of CFSTR equivalence principles were proposed by Feinberg and Ellison (2001). Recently, Kuhlmann and Skiborowski (2017) and Kuhlmann et al. (2018) proposed a ‘bottom up’ approach for synthesis of reaction-separation processes with phenomena building blocks as basic units, which was considered to be the highest level of aggregation, so that complex systems (e.g. azeotrope) and unit operations (e.g. reactive distillation) can be involved. However, since this research depart from conventional unit operations, solutions may not be realized by the existing techniques leading to solution space discarded during the translation and reformulation of models. Therefore, these methodologies are superior in providing guidelines for development of new techniques and equipment rather than for process synthesis of current chemical industry.

According to the aforementioned discussion, mass interaction is essential for reaction-separation process synthesis. Sidestream columns are widely used in chemical industry and could generate more mass allocations between reaction and separation sub-systems by providing multiple compositions through different operational conditions. However, there is a lack of systematic generation and design approach for this kind of complex configurations in reaction-separation-recycle process synthesis. Besides, component distribution in recycle streams has been proved to be essential for reaction-separation-recycle process synthesis (Zhang et al., 2018; Ye et al., 2020; Luyben and Luyben, 1995; Recker et al., 2015; Lakshmanan and Biegler, 1996b; Apluche-Manrique et al., 2011), which should be codetermined by the whole system.

Hence, this article aims to present a new representation for synthesis of distillation sequence with both single and sidestream columns and apply it to a reaction-separation-recycle system to form a general reaction-separation-recycle synthesis framework consists of a reactor network and a complex distillation sequence. The selection of ideal reactors as well as their interconnections, the configuration of distillation column sequence, and the stream allocation between reactors and distillation sequence will be explored. Within the proposed MINLP model, the component distribution in distillation columns is also considered. The key contribution of this article is that it extends the scope of the mass interaction between the reaction and separation sub-systems by considering non-sharp splits and sidestream column in the distillation network, which might benefit the component distribution in the reaction system and energy consumption of the whole process. The outline of this paper is as follows. Description of the problem is defined in Section 2. The novel State-Task Network (STN) for sidestream column sequence and the reaction-separation-recycle system superstructure is illustrated in Section 3. The design approach for sidestream distillation column is presented in Section 4, and all mathematical formulations and solution strategy for the proposed model are given in Section 5. Finally, the proposed methodology is demonstrated by two illustrative examples in Section 6, and general conclusions are provided at last in Section 7.

Section snippets

Problem statement

In this work, a design method for three-section distillation columns with one feed and one sidestream is present. Based on this, the proposed distillation network is embedded in a general reaction-separation-recycle system framework to achieve further energy savings. In the proposed superstructure, the reactor network is interconnected with a distillation sequence with sidestream columns to enforce the production and purification of target products along with the recycle of unreacted reactants.

State Task Network for distillation sequence with sidestream columns

In the synthesis of distillation sequence, various representations have been proposed (Shah and Kokossis, 2002; Caballero and Grossmann, 2004). Among them, STN is a most widely employed representation, which makes good trade-off among simplicity of description, completeness of structures, complexity of computation and expansibility. Therefore, a modified STN is proposed to realize the synthesis of distillation sequences containing three-section side-stream columns and shown in Fig. 1, which is

Underwood equation for sidestream distillation column design

Underwood equations (Glinos and Malone, 1985; Nikolaides and Malone, 1987) and its derivant (Adiche and Aissa, 2016; Monroy-Loperena and Vargas-Villamil, 2001) have been employed for composition estimation of sidestreams and distillate product in distillation column. However, there is still a lack of comprehensive description for the use of Underwood equation in composition estimation of sidestream columns considering various operation conditions. Hence, a detailed discussion on the composition

Stream allocation equations for reaction-distillation flowsheet

The mass flow rate of the total fresh feedstock to the process must be equivalent to the total mass flow rate of products leaving the processcCOMFFc=pPROFPpwhere, FFc is the mass flow rate of component c in fresh feedstock, FPp is the total mass flow rate of final product p, COM is the set of components, PRO is the set of final products. The fresh feedstock of the reaction-separation-recycle system will be distributed to reactor modules through the feed stream splittersrRMFRCr,c=FFccCOM

Example 1

The first example of this work is the chemical process for monochlorobenzene/ dichlorobenzene production with benzene and chlorine, which is taken from the work by Ye et al. (2020). The two chemical reactions involved in this example are all irreversible reactions which are shown in Fig. 5. Reaction 1 is the chlorination of benzene (A) to generate monochlorobenzene (B), and Reaction 2 is the chlorination of monochlorobenzene to generate dichlorobenzene (C). The example is based on the following

Conclusion

In this work, a synthesis superstructure for reaction-separation-recycle system is proposed, which includes complete interconnections among reactor modules as well as necessary connections between reactor modules and distillation columns. In the superstructure, a novel STN representation is introduced to describe the interconnections among distillation columns and column sections so that not only the advantages of it are inherited but also complex columns with sidestream can be included. In the

Declaration of Competing Interest

The authors report no declarations of interest.

Acknowledgments

The authors gratefully acknowledge financial support from the National Natural Science Foundation of China, under Grant no.21276039.

References (73)

  • C. Gutierrez-Antonio et al.

    Design of side-stream azeotropic distillation columns

    Chem. Eng. Res. Des.

    (2007)
  • A.C. Kokossis et al.

    Synthesis of isothermal reactor-separator-recycle systems

    Chem. Eng. Sci.

    (1991)
  • L. Kong et al.

    Expanding the scope of distillation network synthesis using superstructure-based methods

    Comput. Chem. Eng.

    (2020)
  • Q. Kong et al.

    An optimisation-based framework for the conceptual design of reaction-separation processes

    Chem. Eng. Res. Des.

    (2016)
  • M. Martín et al.

    Challenges and future directions for process and product synthesis and design

    Comput. Chem. Eng.

    (2019)
  • L. Mencarelli et al.

    A review on superstructure optimization approaches in process system engineering

    Comput. Chem. Eng.

    (2020)
  • T. Neveux

    Ab-inito process synthesis using evolutionary programming

    Chem. Eng. Sci.

    (2018)
  • M. Nezhadfard et al.

    Development of a reaction/distillation matrix for systematic generation of sequences in a single two component reaction-separation case study

    Comput. Chem. Eng.

    (2018)
  • N. Quirante et al.

    Hybrid simulation-equation based synthesis of chemical processes

    Chem. Eng. Res. Des.

    (2018)
  • A.U. Raghunathan et al.

    Mathematical programs with equilibrium constraints (MPECs) in process engineering

    Comput. Chem. Eng.

    (2003)
  • S. Recker et al.

    A unifying framework for optimization-based design of integrated reaction–separation processes

    Comput. Chem. Eng.

    (2015)
  • B.G. Rong

    A systematic procedure for synthesis of intensified nonsharp distillation systems with fewer columns

    Chem. Eng. Res. Des.

    (2014)
  • R.W.H. Sargent

    A functional approach to process synthesis and its application to distillation systems

    Comput. Chem. Eng.

    (1998)
  • M. Skiborowski

    Process synthesis and design methods for process intensification

    Curr. Opin. Chem. Eng.

    (2018)
  • M. Skiborowski et al.

    Shortcut-based optimization of distillation-based processes by a novel reformulation of the feed angle method

    Chem. Eng. Res. Des.

    (2018)
  • A.W. Westerberg

    A retrospective on design and process synthesis

    Comput. Chem. Eng.

    (2004)
  • C. Yin et al.

    Automatic identification of the optimal distillation sequence based on its integration with reactor

    Chem. Eng. Sci.

    (2019)
  • X. Zhang et al.

    Rigorous design of reaction-separation processes using disjunctive programming models

    Comput. Chem. Eng.

    (2018)
  • X. Zou et al.

    Optimal design of complex distillation system for multicomponent zeotropic separations

    Chem. Eng. Sci.

    (2012)
  • I.M. Alatiqi et al.

    Alternative distillation configurations for separating ternary mixtures with small concentrations of intermediate in the feed

    Ind. Eng. Chem. Process Des. Dev.

    (1985)
  • M. Apluche-Manrique et al.

    Steady state analysis of snowball effects for reaction–separation–recycle systems with thermally coupled distillation sequences

    Chem. Eng. Res. Des.

    (2011)
  • S. Balakrishna et al.

    A unified approach for the simultaneous synthesis of reaction, energy, and separation systems

    Ind. Eng. Chem. Res.

    (1993)
  • D.A. Beneke et al.

    Graphical design and analysis of thermally coupled sidestream columns using column profile maps and temperature collocation

    AIChE J.

    (2011)
  • L.T. Biegler

    Nonlinear Programming Concepts, Algorithms, and Applications to Chemical Processes

    (2010)
  • Q. Chen et al.

    Recent developments and challenges in optimization-based process synthesis

    Ann. Rev. Chem. Biomol. Eng.

    (2017)
  • J.M. Douglas

    A hierarchical decision procedure for process synthesis

    AIChE J.

    (1985)
  • Cited by (8)

    • An inherently safer development approach for thermally coupled distillation sequences: Application in hazardous chemical separation

      2022, Process Safety and Environmental Protection
      Citation Excerpt :

      Thus, in order to assess the system from the different perspectives, approaches of multi-objective optimization should be stated (Haimes et al., 1971; Carreras et al., 2016). For optimization approaches, both short-cut (Ye et al., 2021) model and rigorous model (Zou et al., 2012) can be used for distillation column design. However, the optimization of distillation sequences involves the selection of the number of columns and their corresponding separation tasks and the investment conditions which are related to discrete decisions, and the operating conditions and inherent safety indices which are related to continuous decisions, to minimize a performance function, usually the hazard of system and/ or the total annualized cost that involves investment and operating costs.

    • Synthesis of optimal reaction-separation system with sufficient operational flexibility

      2022, Chemical Engineering Science
      Citation Excerpt :

      Zhang et al. (2018) presented a superstructure to design reaction and separation processes rigorously and simultaneously using the generalized disjunctive programming (GDP) models. Based on the State Task Network (STN), Ye et al. (2020) proposed a synthesis and optimization methodology to design reaction and separation flowsheet with the minimum total annualized cost (TAC); all possible distillation sequences and both conventional and complex distillation columns can be considered (Ye et al., 2021). Ma et al. (2021) used rigorous models to synthesize reaction − separation − recycle processes.

    • Integrated design and optimization of reactor-distillation sequence-recycle-heat exchanger network

      2022, Energy
      Citation Excerpt :

      While the optimization of recycling streams is not considered in the integration of reactor and distillation sequence. Ye et al. [39] proposed a synthesis and optimization method to design the reaction and separation process with minimal cost, and extended it with all possible distillation sequences including conventional and complex distillation columns considered [40]. Ma et al. [41] used a highly nonconvex MINLP formulation to model reaction−separation−recycle processes rigorously, and a relaxed MINLP solution strategy is adopted to solve the model to get local optimality.

    • Rigorous NLP distillation models for simultaneous optimization to reduce utility and capital costs

      2021, Cleaner Engineering and Technology
      Citation Excerpt :

      The improved adapted graphical methods and shortcut methods can readily be applied to modeling and optimization of distillation systems due to the nature of their formulation and the small number of equations. Furthermore, most of the models mentioned above can be solved as NLP problems, with the exception of the ones used by Ye et al. (2021) and Kong and Maravelias (2019). However, the product purity or recovery constraints may not be met when the predicted operating parameters such as feed stage location, number of stages and reflux ratio are used in a simulation study for verification.

    View all citing articles on Scopus
    View full text