Elsevier

Expert Systems with Applications

Volume 79, 15 August 2017, Pages 269-281
Expert Systems with Applications

Harmony search for the layout design of an unequal area facility

https://doi.org/10.1016/j.eswa.2017.02.047Get rights and content

Highlights

  • We consider a layout design of unequal area facilities.

  • Harmony search method is applied with modification to improve its performance.

  • An effective slicing tree representation for layout configuration is presented.

  • New penalty scheme for solution selection is introduced.

  • The proposed approach generates the favorable solutions for well-known problem set.

Abstract

An unequal area facility layout problem (UA-FLP) is a typical optimization problem that occurs when constructing an efficient layout within given areas. In this research, a harmony search (HS)-based heuristic algorithm is presented to solve UA-FLPs. In this study, the facility layout is represented as an allocation of blocks with restrictions in terms of an unequal area and rectangular shape. A more effective facility layout representation is proposed. This is done via a slicing tree representation as a form of layout structure, and via the HS-based algorithm, which generates a quality solution. Once the basic HS solution is generated, modifications are introduced to facilitate improvements. Specifically, the structure of the slicing tree representation is modified, and a re-adjustment operation is added to diversify the possible range of solutions. A penalty scheme is also proposed to improve the feasible region searching capabilities. The effects of the alterations are evaluated by testing well-known problems from previous studies. The proposed algorithm generates the solutions as proficiently as the best results provided by previous research. The proposed method is robust in terms of process, and it determines a favorable solution within a short amount of time.

Introduction

Facility layout problems (FLPs) are typical optimization problems. They require an efficient non-overlapping arrangement of facilities and are considered as a rectangular block in a given space. They have been extensively studied, and many variants have appeared in areas such as manufacturing facility layout and distribution center layout design. Meller and Gau (1996) and Singh and Sharma (2006) reviewed these variants and solution approaches. In most variant cases, determining the optimal arrangement of n departments on a floor space is a primary concern. The variants known as unequal area facility layout problems (UA-FLPs) attempt to determine the optimal arrangement by minimizing the total material handling cost between departments while satisfying the unequal area and aspect ratio restrictions of the departments. Minimizing material movement is very important in facility planning since material handling costs account for 20–50% of total operating expenses in manufacturing (Francis, McGinnis, & White, 1992).

UA-FLPs are NP-hard (Gonçalves and Resende, 2015, Paes et al., 2017). This is because the quadratic assignment problem (QAP), which determines a layout within a finite set of department locations, is already NP-hard (Castillo & Westerlund, 2005). The UA-FLP has unequal area requirements with continuous department positions that have varying dimensions and can be anywhere in a rectangular floor space, and it increases its complexity more than QAP (Castillo & Westerlund, 2005). Because the complexity exponentially increases as the size of the problem increases, exact methods such as the branch and bound method run into difficulties when attempting to solve a large-sized problem optimally.

To achieve favorable solutions in large-sized problems, significant efforts have been made to develop meta-heuristic algorithm-based approaches. In terms of developing meta-heuristic algorithms for UA-FLPs, two views exist. First, encoding can be used to represent an FLP layout configuration, and second, a meta-heuristic algorithm can be used to search a layout configuration region.

The slicing tree structure (STS) is an encoding representation that organizes a layout into a tree structure. The STS consecutively divides the floor space either in a horizontal or vertical direction with a given block space (called a guillotine cut). The encoding vector of the STS is relatively long because it contains a variety of information about the consecutive operations of the slicing tree. The STS requires a (3n  2)-sized encoding vector to represent a layout with n departments, whereas the flexible bay structure (FBS), another well-known layout representation, only requires a (n + α)-sized vector (where α is a decision variable related to the number of bays). The STS, however, can represent a layout that cannot be constructed in the flexible bay structure. Notwithstanding this advantage, the STS has not received much attention because of its unwieldiness (due to the long and complicated encoding vector).

As a search algorithm, HS was developed by Geem, Kim, and Loganathan (2001), and was inspired by a music improvisation process that attempted to determine superior states of harmony. The HS algorithm has been applied to various optimization problems, and it has thus far compared favorably with other traditional meta-heuristic approaches such as GAs (Lee, Geem, Lee, & Bae, 2005), PSOs and SAs (Al-Betar et al., 2015, Gao et al., 2015). The basic concept of the HS algorithm is very similar with that of a GA. However, the HS algorithm generates a new solution after considering all of the existing solutions, while the GA only considers two parent solutions. The remarkable advantages of HS are that its concept and steps are relatively simple, and it does not require initial solution settings (Geem et al., 2001). These advantages make the HS flexible for implementation and combination with other meta-heuristic approaches (Fesanghary, Asadi, & Geem, 2012). Until recently, however, only one paper has utilized an HS algorithm to solve a UA-FLP, and the approach proposed in that paper did not directly use the original HS algorithm; rather, it modified and hybridized the algorithm (Chang & Ku, 2013). This hybrid approach incorporated quadratic constraint programming (QCP) and HS, and it caused the algorithm to have a slower searching process. It was only used to test problems with fewer than twenty departments. This motivated the current authors to develop an entirely new HS-on-UA-FLP approach.

Therefore, in this paper, a new approach is proposed that applies HS to solve a UA-FLP. This approach is intended to produce quality solutions while utilizing the STS in a reasonable amount of time. The HS approach is modified with respect to the procedure by which a new solution vector is generated. An additional adjustment occurs after the new solution vector is generated. This enhances the search process and enables the algorithm to generate multiple solutions. The approach's slicing tree encoding is represented in a (3n  2)-sized vector. Note that the previous approach required a (4n  2)-sized vector (Chang & Ku, 2013). The smaller vector allows the algorithm to achieve a relatively faster search process. Furthermore, the possible range of encoding values is also modified to diversify the search range of the layout configurations. Finally, a penalty scheme is implemented to improve the feasible region-searching capabilities. The results from the computational tests indicate that the approach is relatively stronger and faster than the results offered by previous studies.

The outline of this paper is as follows. In Section 2, the literature on the UA-FLP is reviewed. Section 3 defines the objective function and constraints of the UA-FLP. In Section 4, the proposed approach is described in detail. Subsequently, the computational results of the test problems from previous studies are provided in Section 5. Finally, the conclusions and future research considerations are provided.

Section snippets

Literature review

As mentioned, the UA-FLP is a block layout design problem concerned with the optimal arrangement of departments on a floor space within given unequal areas. Armour and Buffa (1963) formally introduced this problem, and they proposed heuristics to solve it. Heuristic algorithms such as CRAFT (Armour & Buffa, 1963), CORELAP (Moore, 1974) and MULTIPLE (Bozer, Meller, & Erlebacher, 1994) have been proposed to solve UA-FLPs, but large-sized unequal area problems were not considered because of the

Layout design of unequal area facility

In this section, the mathematical model of a generic block layout design problem with unequal area restrictions is addressed. Each department needs to be determined with respect to its position on a floor space and with respect to the width and height dimensions of the unequal area.

In Fig. 1, the dimensions of the floor space are denoted by Lx and Ly, and for each department i, the coordinate of its centroid is denoted by (cix,ciy), and the width and height dimensions are lix and liy,

Harmony search approach

In this section, details regarding the solution approach are explained. First, a discussion on the STS as a solution representation is discussed with a focus on its encoding. Afterwards, an explanation is given regarding how the HS algorithm is implemented to solve UA-FLPs, and, in addition, the ways in which this study might contribute to the original HS- and STS-based approaches are listed.

Computational results

Computational experiments are implemented to evaluate the performance of the approach, and the results are compared to those from the previous literature. The algorithm is coded using JAVA, and the tests are conducted on a computer with an Intel Core i5 CPU (3.2 GHz) and 8GB of memory. Datasets of well-known problems with 7–62 departments are used. Table 1 shows the names and setups for the problems. Note that the number in the name of the problem indicates the number of departments required to

Conclusion

An unequal area facility layout problem is a typical optimization problem that occurs while constructing an efficient block layout within given unequal areas. In this research, a harmony search-based heuristic algorithm is presented to solve UA-FLPs. The block layout configuration is represented via a slicing tree representation. Modification methods are proposed to improve the performance of the algorithm. A more effective slicing representation is proposed for the block layout to diversify

Acknowledgment

This research was supported by the MSIP (Ministry of Science, ICT and Future Planning), Korea, under the ITRC (Information Technology Research Center) support program ((IITP-2016-H8601-16-1010)) supervised by the IITP (Institute for Information & communications Technology Promotion).

References (53)

  • R.D. Meller et al.

    The facility layout problem: Recent and emerging trends and perspectives

    Journal of Manufacturing Systems

    (1996)
  • R.D. Meller et al.

    Optimal facility layout design

    Operations Research Letters

    (1998)
  • M.G.H. Omran et al.

    Global-best harmony search

    Applied Mathematics and Computation

    (2008)
  • F.G. Paes et al.

    A hybrid genetic algorithm with decomposition phases for the unequal area facility layout problem

    European Journal of Operational Research

    (2017)
  • J.M. Palomo-Romero et al.

    An island model genetic algorithm for unequal area facility layout problems

    Expert Systems with Applications

    (2017)
  • A. Saraswat et al.

    A framework for multi-objective facility layout design

    Computers and Industrial Engineering

    (2015)
  • D. Scholz et al.

    STaTS: A slicing tree and tabu search based heuristic for the unequal area facility layout problem

    European Journal of Operational Research

    (2009)
  • E. Valian et al.

    An intelligent global harmony search approach to continuous optimization problems

    Applied Mathematics and Computation

    (2014)
  • K.Y. Wong et al.

    Solving facility layout problems using flexible bay structure representation and ant system algorithm

    Expert Systems with Applications

    (2010)
  • O. Abdel-raouf

    A survey of harmony search algorithm

    International Journal of Computer Applications

    (2013)
  • G.C. Armour et al.

    A heuristic algorithm and simulation approach to relative location of facilities

    Management Science

    (1963)
  • F. Azadivar et al.

    Facility layout optimization using simulation and genetic algorithms

    International Journal of Production Research

    (2000)
  • Y.A. Bozer et al.

    An improvement-type layout algorithm for single and multiple-floor facilities

    Management Science

    (1994)
  • M.S. Chang et al.

    A slicing tree representation and QCP-model-based heuristic algorithm for the unequal-area block facility layout problem

    Mathematical Problems in Engineering

    (2013)
  • T. Dunker et al.

    A coevolutionary algorithm for a facility layout problem

    International Journal of Production Research

    (2003)
  • R.L. Francis et al.

    Facility layout and location: an analytical approach

    (1992)
  • Cited by (50)

    • Addressing Unequal Area Facility Layout Problems with the Coral Reef Optimization algorithm with Substrate Layers

      2020, Engineering Applications of Artificial Intelligence
      Citation Excerpt :

      For that matter, these approaches utilized interactive evolutionary computation to solve the UA-FLP. For representing a plant layout, three main structures have been commonly used: Block Layout (Meller and Gau, 1996; Castillo et al., 2005; Gonçalves and Resende, 2015), Slicing Trees (Shayan and Chittilappilly, 2004; Scholz et al., 2009; Azadivar and Wang, 2000; Komarudin and Wong, 2010; Kang and Chae, 2017) and Flexible Bays (Tate and Smith, 1995; Wong and Komarudin, 2010; Kulturel-Konak and Konak, 2011; Kulturel-Konak, 2012; Palomo-Romero et al., 2017; Garcia-Hernandez et al., 2019; Liu and Liu, 2019). The last category has been the most used because it offers the advantage of simplifying the UA-FLP by means of reducing the problem complexity (Wong and Komarudin, 2010).

    View all citing articles on Scopus
    View full text