New model for large scale chemical industrial layout optimization

https://doi.org/10.1016/j.cherd.2020.06.026Get rights and content

Highlights

  • An area-wide layout optimization method for real industrial size is proposed.

  • Large number of utility connection is considered and their significance to layout is explored.

  • An efficient optimization framework is proposed to solve the problem in ideal time.

  • Pipe rack and pipe connection are optimized simultaneously.

  • Pressure drop, flow rate, and heat loss are considered in segments for pipe connection optimization.

Abstract

Area wide layout is an essential factor that significantly affects the safety, transportation and land occupation of chemical industrial zones. This research proposes a novel methodology to optimize the large-scale area-wide layout design for industrial zones. In layout design, the piping connection between processes is one of the essential issues. However, the complicated pipeline networks and pipe rack have not been taken into account before. Additionally, the current algorithm cannot solve the practical large-scale problem effectively. In this work, a new model and an optimization framework are proposed to solve the general layout problem. In the model, the piping layout and the pipe rack are designed in segments. The layout, pipe connections and pipe arrangement in the pipe rack are optimized simultaneously. Safety, geography and environmental constraints are considered in the optimization. A case study demonstrates the large-scale optimization can effectively reduce the total cost by 14.35% in less than 50 s. The impacts on the area-wide layout of the consideration of multi-utility and pipe rack are also analyzed.

Introduction

Layout design plays an important role in the early design or retrofit stage of a factory or an industrial zone. Such a layout design problem is also called the facility layout problem (FLP), which is firstly proposed by Koopmans and Beckmann (1957). If the layout design is done well, the capital expansion expense is well invested (CCPS, 2018). The layout design problem is very complex and is an NP-Hard problem (Garey and Johnson, 1979) of engineering optimization. FLP has been studied in many fields, for instance, ship cabin (Luo et al., 2015), hospital (Helber et al., 2016), wind farm (Li et al., 2017), chemical plant (Özyurt and Realff, 1999; Ejeh et al., 2018a), and so on.

For the FLP of chemical industrial enterprises, it can be divided into two main types of design according to the scope of facilities, e.g. determining the location of equipment within a processing unit and determining the location of process units within an industrial zone. They are known as plant layout and general layout problems, respectively. Here, the general layout is also called an area-wide layout.

In an area-wide layout problem, the relative position of plants contributes to the material handling cost, land cost, energy loss and production safety directly. Most researches focus on land costs and safety issues for FLP. Georgiadis et al. (1999) considered land occupied, floor construction and total connected piping length in the optimization of FLP. de Lira-Flores et al. (2019) involved safety instrument systems in FLP optimization. Park et al. (2011) used the TNT equivalency method to describe explosion damage costs in solving multi-floor FLP. Khakzad and Reniers (2015) developed a Bayesian network method to calculate both on-site and off-site risks of chemical plants considering domino effects in the optimization of FLP. Alves et al. (2016) proposed a new algorithm based on Monte Carlo and simulated annealing to optimally locate the process or storage units of the chemical plant. Latifi et al. (2017) studied the FLP with various risk under uncertainty using bat metaheuristic algorithm. Brunoro Ahumada et al. (2018) proposed a framework integrating layout formulation with a quantitative risk assessment method. The method can be throughout the whole project lifecycle. However, the above-mentioned methods do not consider the pipe connection.

One of the main objectives in FLP is to minimize material handling costs (Palomo-Romero and Salas-Morera, 2017), especially for area-wide layout whose distance among plants is long. A layout with a reasonable material handling system can reduce the initial capital investment, operation cost, energy loss and the risk during the transport of toxic materials (Tompkins et al., 2010). In this work, we distinguish different piping systems through two aspects. From the material the pipes transported, the utility pipe system is used to transport utility flow and the material pipe system is used to transport up and downstream material flow. For the pipe topology, a simple pipeline is a pipe without a branch, and the pipeline network is the pipe with branches, e.g. steam system. Currently, few works focus on the optimization of the connection between plants in general layout design. Besides, only simple pipelines are involved in current research. The pipeline network is not involved in the current layout optimization method. The pipeline network accounts for a large percentage of the total pipeline and should not be ignored. Furthermore, as an important part of the consideration of piping connections, the optimization of pipe rack layout has not been studied before. It is usually designed manually by engineers after the optimization of plants and pipe routing. In addition, the pipe routing and pipe rack layout are interactive. Therefore, it is necessary to intelligently implement the optimization to reduce complicated manual calculations.

For the research of material handling, in the fields of oil and gas gathering system, some related works using different algorithms can be found, e.g., simulated annealing genetic hybrid algorithm (He et al., 2019), modified particle swarm optimization algorithm (Zhang et al., 2018, 2017a; Liu et al., 2019), improved ant colony optimization algorithm (Wang et al., 2018a), and mixed-integer linear mathematical programming (MILP models (Zhang et al., 2017b, c, d). Zhang et al. (2017c) studied the optimization of well-gathering pipeline network, stellated pipeline network, cascade dendritic pipeline network and insertion dendritic pipeline network is optimized by solving MILP models considering obstacles and terrain. However, few FLP researches for chemical industrial zone make a further study about material handling. Guirardello and Swaney (2005) only optimize simple pipeline routing Dijkstra algorithm without considering pipeline networks. In our previous work (Wu and Wang, 2017), the problem was modeled as a minimum spanning tree (MST) problem and solved by Kruskal algorithm and arrangement & combination. However, such an algorithm is very time-consuming (Wang et al., 2018b). Moreover, pipe rack is not considered combined with pipeline network optimization. The pipeline network optimization problem can also be solved by rectilinear Steiner minimal tree (RSMT). It is a tree with a minimum total edge length in Manhattan distance to connect a given set of nodes possibly through some extra nodes (Chu and Wong, 2008). The FLUTE (fast lookup table estimation) algorithm is an exact algorithm for computing RSMT for small cardinality terminal sets (Brazil and Zachariasen, 2015). Zachariasen (1999) and Warme (1998) developed another exact algorithm which named GeoSteiner approach. The algorithm is used to calculate RSMT. Their software is the fastest program that can exactly solve the rectilinear Steiner tree problem with more than 1000 terminals (Juhl et al., 2018). But the above methods have not been applied in pipe routing of the chemical industrial zone. In this research, GeoSteiner approach is employed to solve the problem.

FLP is difficult to be solved, and several solving techniques are proposed for this problem. For example, heuristic method (Fortenberry and Cox, 1985), graph-theoretic heuristics (Kim and Kim, 1995), simulated annealing algorithm (Baykasoğlu and Gindy, 2001), ant colony optimization (Guan and Lin, 2016), branch and bound method (Kettani and Oral, 1993), MILP solver (Emhamed et al., 2007; Ejeh et al., 2018b; Klausnitzer and Lasch, 2019), multi-objective mixed-integer linear programming model (Martinez-Gomez et al., 2015), integrated fuzzy simulation-fuzzy data envelopment analysis (Azadeh et al., 2016), hybrid genetic algorithm(GA) (Nasab et al., 2012; Caputo et al., 2015; Ning et al., 2019). In general, GA is one of the most widely used methods of solving FLP because of its good robustness and scalability, but the optimization time is long. Parallel technologies can be used to boost the optimization speed. OpenMP is an application programming interface (API) for platform-independent shared-memory parallel programming in C, C++, and Fortran (Schmidt et al., 2017). Umbarkar (2016) implemented the parallel genetic algorithm (PGA) using OpenMP for continuous nonlinear large-scale problems, which greatly speed up the optimization time compared with serial GA. In this work, PGA based on OpenMP is applied in FLP of the chemical industrial zone.

This paper makes a further study on the large-scale area-wide layout in the chemical industrial zone, taking the sum of piping cost, energy cost (pressure loss cost and heat loss cost) and pipe rack cost as the objective function. The engineering practices are taken into consideration qualitatively. The contributions of this work are as follows:

  • (1)

    The complex pipeline network of multi-utility is taken into account for the optimization, as well as the pipeline network of material flows.

  • (2)

    The design of a multi-floor pipe rack is added into the optimization of area-wide layout, including the determination of floor number and the detailed layout of pipelines on the pipe rack.

  • (3)

    Both the insulation thickness of the pipe and the space between every two pipes are variable according to the physical and chemical properties of fluids in pipes.

  • (4)

    The method can solve a real-life large-scale problem. A real-life large-scale chemical industrial zone is studied to illustrate the effectiveness and applicability of the proposed method.

  • (5)

    A new method combining GeoSteiner algorithm and PGA is proposed to optimize the area-wide layout problem including plant, piping and pipe rack.

Section snippets

Problem statement

In the optimization of area-wide layout, the connection between plants is deemed to be more remarkable because of the long distance. Therefore, the piping system and pipe rack are further optimized. There are two layers of optimization. The outer layer is to obtain the better relative positions of plants. The investment and energy cost (pressure loss cost and heat loss cost) of the optimal piping system and pipe rack cost are used to evaluate the scheme of plant layout. The industrial zone is

Methodology

The objective function includes pipe investment (Cpipe), power cost for compensating pressure loss (Cpressure), hot utility cost for compensating heat loss (Cheat), pipe rack cost (Cpiperack) and the cost of concrete piers for supporting pipe rack (Cconcretepier), as shown in Eq. (1).minTAC=pnsCppipe,simple+pnsCppressure,simple+pnsCpheat,simple+inpjnbCi,jpipe,network+inpjnbCi,jpressure,network+inpjnbCi,jheat,network+knprCkpiperackp=1,2,3...ns;i=1,2,3...np;j=1,2,3...nb;k=1,2,3...npr

Algorithm

In this research, GeoSteiner software package is embedded in the parallel genetic algorithm (PGA). The algorithm can optimize the plant layout and piping layout simultaneously.

Case study

The case study is established based on a real petrochemical industrial zone in China. The zone is divided into 5 × 5 grids. There are 25 plants. Their names and abbreviations are summarized in Table S5. The geographical and traffic conditions of the zone are as follows.

  • 1)

    The annual minimum frequency wind direction of the area is north.

  • 2)

    Topography increases from northeast to southwest gradually.

  • 3)

    The railway is in the northwest of the industrial area.

  • 4)

    Highway and city are in the northeast of the area.

Conclusion

This research presents an effective method of the area-wide layout optimization which can be employed to guide the design chemical industrial zone. For optimization targets, the piping of multi-utility and material flow pipe with pipeline networks are optimized in chemical plant layout. Besides, the multi-floor pipe rack is taken into consideration. For models, the calculations of pipe insulation and pipe spacing meet different flow characteristics and variable pipe diameters, as well as the

Declaration of interests

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgments

Financial supports from the Science Foundation of China University of Petroleum, Beijing (No. 2462018BJC004) are gratefully acknowledged.

References (57)

  • N. Hallale et al.

    Refinery hydrogen management for clean fuels production

    Adv. Environ. Res.

    (2001)
  • G. He

    A methodology for the optimal design of gathering pipeline system in old oilfield during its phased development process

    Comput. Ind. Eng.

    (2019)
  • N. Khakzad et al.

    Risk-based design of process plants with regard to domino effects and land use planning

    J. Hazard. Mater.

    (2015)
  • J.Y. Kim et al.

    Graph theoretic heuristics for unequal-sized facility layout problems

    Omega

    (1995)
  • A. Klausnitzer et al.

    Optimal facility layout and material handling network design

    Comput. Oper. Res.

    (2019)
  • S.E. Latifi

    Process plant layout optimization with uncertainty and considering risk

    Comput. Chem. Eng.

    (2017)
  • W. Li

    Multi-objective evolutionary algorithms and hyper-heuristics for wind farm layout optimisation

    Renew. Energy

    (2017)
  • Y. Liu

    Layout optimization of large-scale oil–gas gathering system based on combined optimization strategy

    Neurocomputing

    (2019)
  • J. Martinez-Gomez

    Optimization of facility location and reallocation in an industrial plant through a multi-annual framework accounting for economic and safety issues

    J. Loss Prev. Process Ind.

    (2015)
  • X. Ning

    Reducing noise pollution by planning construction site layout via a multi-objective optimization model

    J. Clean. Prod.

    (2019)
  • M.Z. Stijepovic et al.

    Optimal waste heat recovery and reuse in industrial zones

    Energy

    (2011)
  • Y. Wang

    A human-computer cooperation improved ant colony optimization for ship pipe route design

    Ocean. Eng.

    (2018)
  • R. Wang

    An industrial Park layout design method considering pipeline length based on FLUTE algorithm

  • Y. Wu et al.

    A chemical industry area-wide layout design methodology for piping implementation

    Chem. Eng. Res. Des.

    (2017)
  • H. Zhang

    An improved PSO method for optimal design of subsea oil pipelines

    Ocean. Eng.

    (2017)
  • H. Zhang

    Sensitivity analysis and optimal operation control for large-scale waterflooding pipeline network of oilfield

    J. Pet. Sci. Eng.

    (2017)
  • H. Zhang

    A unified MILP model for topological structure of production well gathering pipeline network

    J. Pet. Sci. Eng.

    (2017)
  • H. Zhang

    An MILP method for optimal offshore oilfield gathering system

    Ocean. Eng.

    (2017)
  • Cited by (5)

    • A new multiple-risk map approach to solve process plant layout considering safety and economic aspects

      2021, Journal of Loss Prevention in the Process Industries
      Citation Excerpt :

      Ejeh et al. (2019) proposed a MILP approach for a multi-floor PPL problem considering connectivity costs, pumping costs, installation of safety devices, piping, and additional features. Recently, Wu et al. (2020) used a GBM to optimize the pipe layout on the pipe rack, assuming that the distances between every two adjacent facilities are the same and longer than the minimum safety distance. These models include all or some elements of the quantitative risk analysis (QRA) methodology (CCPS 2009), i.e., hazard identification scenarios, frequency, and consequence analysis.

    View full text