Optimal Membrane-Process Design (OMPD): A software product for optimal design of membrane gas separation processes

https://doi.org/10.1016/j.compchemeng.2020.106724Get rights and content

Highlights

  • Software for optimal design of membrane gas-separation processes is presented.

  • It generates several potential process design configurations.

  • It finds optimal design specifications and operating conditions.

  • It can handle any type or number of objective functions.

  • It optimizes membrane units arranged in multi-step and/or multi-stage configurations.

Abstract

The optimal design of a membrane gas separation process requires minimizing several objective functions subject to nonlinear relationships among the optimizing variables. This article describes a novel software product, named Optimal Membrane-Process Design (OMPD), for the optimal design of membrane gas separation processes. The product generates several potential process design configurations and then searches the process design parameters and operating conditions spaces to arrive at optimal design specifications and operating conditions. It is able to consider every type and any number of operational, compositional, and economical objective functions in a computationally cost-effective manner. It calculates all Pareto optimal solutions in a single trial. It can optimize any number of membrane units arranged in multi-step and/or multi-stage configurations. It optimally places pairs of adjacent membrane units, either two-step or two-stage, while simultaneously considering several membrane types.

Introduction

Membrane gas separation processes are used increasingly to separate gases and generate high purity streams for industrial applications. Compared to other conventional separation technologies such as cryogenic distillation and swing adsorption, membrane technologies are easier to implement and scale up, require less manual handling and lower operational costs, and require lower footprints and energy levels (Gallucci et al., 2013; Baker and Low, 2014; Sanders et al., 2013; Koros and Fleming, 1993; Pandey and Chauhan, 2001). For the past decades, membrane technologies have been used in hydrogen recovery (David et al., 2011), oxygen enrichment of air (Belaissaoui et al., 2014; Haider et al., 2018), CO2 capture (Yong et al., 2018; Baker et al., 2017; Liu et al., 2016), air dehumidification (Qu et al., 2018; Bui et al., 2015), natural gas purification (sweetening, dehydration, removal of impurities), and helium recovery from natural gas (Vaughn and Koros, 2014; Alders et al., 2017; Soleimany et al., 2017), among others. Membrane separations have also been applied to upgrade biogas to pure methane (Chen et al., 2015; Esposito et al., 2019), to produce high purity nitrogen (Spagnolo et al., 1995), recover olefins (Yu et al., 2017; Faiz and Li, 2012; Ramu et al., 2018; Zhilyaeva et al., 2018), and adjust syngas ratios (Peer et al., 2009).

The economic viability of a membrane-based gas separation process depends on the permeability/selectivity of the membrane material and the optimality of the design and operation of the membrane process. A membrane gas separation process should be able to compete with conventional gas separation processes in terms of capital and operating costs, as well as the purity of the gaseous products generated (Lababidi et al., 1996; Lin et al., 2015). In order to design optimal membrane processes, a great number of variables should be considered and optimized. The axes of the design space include membrane types, permeabilities and selectivities of membranes, pressure ratios, flow rates, number and arrangement of membrane units, and the total membrane area of each unit (Low et al., 2013; White et al., 2015). An optimal membrane gas separation process design is usually a multi-objective optimization problem, leading to complex relationships.

An important step in a membrane gas separation process design is the selection of an optimal process configuration; i.e., finding an optimal arrangement of membrane units and determining how the units should be connected to each other (Tessendorf et al., 1999; Kim and Kim, 2018). The most common and simplest arrangement in membrane gas separation is a single membrane unit configuration with no recirculation/recycle stream. In general, this configuration is far from optimality since membranes typically exhibit low selectivities, and higher separation efficiency can be achieved only through co-regeneration and circulation across multiple modules.

To generate a membrane process configuration consisting of multiple membrane units, one can consider multi-stage and/or multi-step configurations (Lokhandwala et al., 2010). In the multi-stage configuration, the permeate of the first unit becomes the feed of the second unit, while in the multi-step configuration, the feed to the second unit is the retentate gas stream from the first unit. Multi-stage and/or multi-step configurations are usually designed with two, three, or four membrane units (Lababidi et al., 1996). Other innovative arrangements similar to that of distillation columns have also been developed and put into practice (Hwang and Ghalchi, 1982; Ohno et al., 1978; Bhide and Stern, 1993). These designs not only use recycle streams while compressing a fraction of either permeate or retentate streams, but also determine optimal locations for the feed streams inputs in modules layouts. The large dimension of the design space described above suggests a need for an efficient software product able to find optimal design specifications and operating conditions of membrane gas separation processes. Efforts have been made to design optimal membrane gas separation processes by applying classical single- and multi-objective optimization techniques (Tessendorf et al., 1999; Kim and Kim, 2018; Ramírez-Santos et al., 2018; Gabrielli et al., 2017; Chang and Hou, 2006; Akinlabi et al., 2007).

Unlike single-objective optimization problems, multi-objective ones in principle have multiple solutions, known as Pareto optimal solutions. Classical multi-objective optimizers, such as weighted sum and ε-constraint, transform a multi-objective problem into a simplified single-objective optimization problem with specific constraint(s). In addition, classical multi-objective optimizers mostly employ deterministic transition rules and scalarizing functions iteratively to solve vector optimizations and to detect a set of Pareto optimal solutions. Rarely, all Pareto optimal solutions can be found in a single trial when traditional optimizers are used. Hence, a traditional optimizer should be utilized several times, and each time the predefined adjustable parameters of the algorithm should be meticulously regulated to obtain a portion of the optimal Pareto front (Shukla et al., 2005).

Unlike classical deterministic multi-objective optimization algorithms, artificial intelligence (AI)-based optimizers are capable of finding all Pareto optimal solutions efficiently in an evolutionary single simulation trial (Mohammadi et al., 2018; Kannan et al., 2009; Garshasbi et al., 2016; Russell and Norvig, 2009; Garshasbi et al., 2019). AI-based optimization techniques include swarm intelligence, simulated annealing, particle swarm optimization, and genetic algorithms. Among different Genetic Algorithms, Non-dominated Sorting Genetic Algorithm (NSGA-II), which is suitable for multi-objective optimization, is used in this study.

This article describes a novel software product, named Optimal Membrane-Process Design (OMPD), for the optimal design of membrane gas separation processes. The product first generates several potential process design configurations as an initial population, and then searches the process design parameters and operating conditions spaces to arrive at optimal design specifications and operating conditions. The software product does not have any limitations in terms of the type or the number of optimizing variables, the type or the number of objective functions, or the number of membrane units arranged in multi-step and/or multi-stage configurations. As a case study, the software product is used to design an optimal membrane process for the separation of nitrogen gas (N2) from methane (CH4).

Section snippets

Mathematical model

Fig. 1 schematically shows a membrane unit equipped with an A-selective membrane type and challenged with a feed composed of two components; i.e., Gas A and Gas B. The feed stream enters the membrane unit and gets split into two streams, the permeate stream (the one that passes through the membrane) and the retentate stream (the one that contains the rejected components). QF, QR, and QP are the total flow rates (cm3[STP] s−1) of the feed, retentate, and permeate streams, respectively, and xF, A

Configuration generation

The optimal membrane gas separation process design requires an efficient configuration generator capable of generating all potential membrane gas separation process configurations. An optimizer then interacts with the configuration generator and finds an optimal configuration. The configuration generator should be able to enliven and dial in all potential configurations that satisfy membrane gas separation principles. The development of an appropriate and computationally optimal configuration

Membrane process design optimization

The main objective of the membrane gas separation system design optimization is to minimize the cost of separation while achieving the predefined separation goals. From an economic standpoint, this often becomes a trade-off between the total membrane area and the recompression power needed for one or more recycle streams. On the other hand, the product quality, including both product purity and recovery, are the main preset technical goals for each membrane separation process design. Hence,

Problem statement

To evaluate the performance of the software product, it was used to develop an optimal membrane process design for the separation of nitrogen (N2) from methane (CH4). Process design optimizations were performed for the methane-selective and nitrogen-selective membranes, the permeances and selectivities of which are listed in Table 2. Furthermore, the feed gas stream was assumed to have a nitrogen mole fraction of 10%, a methane mole fraction of 90%, and a flow rate is 10 MMscfd at 500 psia. All

Conclusions

A software product, called OMPD, for the optimal design of membrane gas separation processes was introduced. It first generates a pool of practically-meaningful process configurations and then solves a multi-objective optimal problem to arrive at optimal design specifications, including a membrane process configuration and the design specifications of each membrane unit. The application and performance of the software product was shown by applying it to case studies. The results obtained

References (53)

  • P. Gabrielli et al.

    On the optimal design of membrane-based gas separation processes

    J. Membr. Sci.

    (2017)
  • F. Gallucci et al.

    Recent advances on membranes and membrane reactors for hydrogen production

    Chem. Eng. Sci.

    (2013)
  • S. Garshasbi et al.

    A hybrid genetic algorithm and Monte Carlo simulation approach to predict hourly energy consumption and generation by a cluster of net zero energy buildings

    Appl. Appl. Energy

    (2016)
  • S. Garshasbi et al.

    Optimal learning group formation: A multi-objective heuristic search strategy for enhancing inter-group homogeneity and intra-group heterogeneity

    Expert Syst. Appl.

    (2019)
  • S. Haider et al.

    Carbon membranes for oxygen enriched air - Part I: Synthesis, performance and preventive regeneration.

    (2018)
  • M. Hosseinnezhad et al.

    Realization of manufacturing dye-sensitized solar cells with possible maximum power conversion efficiency and durability

    Sol. Energy

    (2017)
  • S.-T. Hwang et al.

    Methane separation by a continuous membrane column

    J. Membr. Sci.

    (1982)
  • M. Kim et al.

    Optimization model for the design and feasibility analysis of membrane-based gas separation systems for CO2enhanced coal bed methane (CO2-ECBM) applications

    Chem. Eng. Res. Des.

    (2018)
  • W.J. Koros et al.

    Membrane-based gas separation

    J. Membr. Sci.

    (1993)
  • H. Lababidi et al.

    Optimization of module configuration in membrane gas separation

    J. Membr. Sci.

    (1996)
  • H. Lin et al.

    CO2-selective membranes for hydrogen production and CO2 capture - Part II: Techno-economic analysis

    J. Membr. Sci.

    (2015)
  • H. Lin et al.

    CO2-selective membranes for hydrogen production and CO2 capture - Part I: Membrane development

    J. Membr. Sci.

    (2014)
  • L. Liu et al.

    Post-combustion carbon dioxide capture via 6FDA/BPDA-DAM hollow fiber membranes at sub-ambient temperatures.

    (2016)
  • K.A. Lokhandwala et al.

    Membrane separation of nitrogen from natural gas: A case study from membrane synthesis to commercial deployment

    J. Membr. Sci.

    (2010)
  • B.T. Low et al.

    A parametric study of the impact of membrane materials and process operating conditions on carbon capture from humidified flue gas

    J. Membr. Sci.

    (2013)
  • P. Pandey et al.

    Membranes for gas separation

    Prog. Polym. Sci.

    (2001)
  • Cited by (0)

    View full text