Journal of Manufacturing Systems 58 (2021) 59–74

Contents lists available at ScienceDirect

Journal of Manufacturing Systems

journal homepage: www.elsevier.com/locate/jmansys

Integration of material handling devices assignment and facility

layout problems
Adem Erik, Yusuf Kuvvetli *
Department of Industrial Engineering, Cukurova University, 01330, Adana, Turkey

A R T I C L E I N F O A B S T R A C T

Keywords: Facility layout problems focus on the assignment of departments into the facility layout by considering the
Dynamic facility layout problem minimization of total costs. A well-designed facility result in efficient material flows in transportation. For this
Flexible bay structure reason, the problem should be considered with the changes in demands that cause different material flows be­
Material handling device assignment problem
tween departments. While the aim of the static facility layout problem is to obtain the optimal facility layout, the
Mathematical modeling
Static facility layout problem
dynamic facility layout problem includes the multi-periods dealing with the changes in material flows between
departments. Furthermore, the material handling devices may affect the facility layout decisions although it is
omitted in general. In this study, three different facility layout problems concern static, dynamic and flexible bay
structure and material handling device assignment problems are considered simultaneously. Three mixed integer
programming models are proposed for the addressed problems with integrated decisions. The proposed models
are compared with a two-stage approach composed of solving the classical facility layout problems first, then
solving material handling devices assignment problems. The comparative results show that the integrated de­
cision affects the facility layout and decreases the total costs. Eleven test instances are considered and the in­
tegrated model provides cost efficiency for these test instances.

1. Introduction obtaining the optimal solution becomes difficult. The problem is

therefore having computational complexity, defined as NP-hard [5].
The layout design is one of the decisions that have a direct impact on Facility layout problems are handled together with different ap­
the production costs of firms in an increasingly competitive environ­ proaches in real life applications. One of these is the flexible bay
ment. The transportation costs associated with the facility arrangement structure, which limits the continuous space settlement problem. In one
correspond to 20 %–50 % of the operating costs, 2 %–10 % of these of the early studies, Tong [6] proposed a flexible bay structure based on
transportation costs can be reduced with good plant planning [1]. A layout, facility base, and the width of the flexible bay that is divided
well-regulated plant results in efficient material use and reduced horizontally or vertically depending on the number of sections in the
transport time [2]. facility layout.
The facility layout problem (FLP) aims to determine the locations of Fig. 1 shows an example representation of the flexible bay structure.
the departments within a supportive for continuous production (or While the flexible bay widths in the same line may differ, their heights
service) [3]. FLPs deal with the physical placement of machines, de­ must be the same [7]. The aim of the flexible bay structure is to divide a
partments, workstations in a layout for a specific aim (such as minimum specific region into sub-regions, ensuring to minimize the total material
transportation amount in the manufacturing sector) [4]. handling costs. The areas of the bays may not be the same, but each bay
Type of FLPs varies on the produced product, sector, facility struc­ must provide aspect ratio constraints or minimum length constraints
ture, demand variability and many other factors. FLP deals with the [8].
placement of the N facilities, which are usually given, in N areas, in a A specific case of the facility layout problems is defined as Dynamic
way to build the lowest cost layout. Therefore, all alternatives (N!) must Facility Layout Problem (DFLP). Various definitions are made in many
be evaluated for the best placement. As the number of facilities in­ studies on DFLP [9–15]. In general, the dynamic facility layout problem
creases, the number of alternatives increases exponentially and refers to creating a facility design that provides multi-period planning,

* Corresponding author at: Department of Industrial Engineering, Cukurova University, Balcalı Campus, 01330, Adana, Turkey.
E-mail address: [email protected] (Y. Kuvvetli).

https://doi.org/10.1016/j.jmsy.2020.11.015
Received 17 March 2020; Received in revised form 21 November 2020; Accepted 21 November 2020
Available online 3 December 2020
0278-6125/© 2020 The Society of Manufacturing Engineers. Published by Elsevier Ltd. All rights reserved.
A. Erik and Y. Kuvvetli Journal of Manufacturing Systems 58 (2021) 59–74

Kothari and Ghosh [22] created a heuristic dispersed search model for
the SFLP. Palubeckis [23] used a simulated annealing algorithm, which
is a meta-heuristic algorithm for the SFLP. The developed algorithm is
very fast and the facility has been more effective in problems up to 300
according to the tabu search algorithm reported in the literature. In the
study of Durmaz and Şahin [3], the single row facility layout problem,
which is a variant of the SFLP, was discussed. In the study, NSGA-II and
goal programming methods were used to minimize the total flow dis­
tance between departments and the closeness between departments. In
the study of Tambunan et al. [24], the SFLP was compared using
computer-aided methods BLOCPLAN and ALDEP for the total material
flow distance resulting from the displacement of departments. Although
the solution found with ALDEP is higher than that found with BLOC­
PLAN, it did not require changing the operating mode and the enterprise
had a straight-line flow. Besides, a heuristic simulation method is pro­
posed for the SFLP in the Armor and Buffa [25]. Bazaraa [26] proposed a
Fig. 1. An example for the flexible bay structure.
computer-based branch-bound algorithm for the SFLP.
Love and Wang [27] deals with the SFLP that includes departments
in which the product demands changes during periods thereby material
of different lengths. Van Camp, Carter and Vannelli [28] proposed
transportations between departments vary extensively from a period to
non-linear programming techniques that minimize the material trans­
another period. The aim of the dynamic facility layout problem is to
portation cost. In Tam’s study [29], a simulated annealing algorithm has
minimize the sum of transportation costs between the departments and
been developed to SFLP slicing tree structure to bay assignment con­
to ensure that the facility is organized more effectively by keeping the
taining departments that unequal areas. A constructive greedy heuristic
displacement costs of the departments at the lowest level during the
model, which aims to minimize the inter-departmental flow cost for
planning horizon.
single-row facility layout, which is a variant of SFLP, is proposed [30]. In
As predicted in Meller and Gau [16], there is a tendency in the
the study of Meller [31], a mixed integer model is proposed that ar­
literature to organize layout and production system designs with
ranges the SFLP with unequal departments using multiple division
simultaneous approaches. Another consideration taken into account is
structure. In the study, firstly departments were assigned to the de­
that when dealing with the facility layout problem is the assignment of
partments, and then the partitions were ordered by considering the
material handling devices. The type of material handling devices affects
material handling systems. In the study of Gau and Meller [32], an al­
the layout to be used for the settlement of the machines [17]. This re­
gorithm that works iteratively between genetic algorithm with slicing
quires considering these two dependent design problems together to find
trees and mixed-integer programming is proposed for SFLP with unequal
a solution for facility layout [18]. Efficient material handling ensures
department areas. In their study, Balakrishnan, Cheng and Wong [33]
lower processing times and cycle times in production processes, on-time
addressed the SFLP and assigned 6 machines with unequal areas using
delivery and better quality [19].
simulated annealing and genetic algorithm. In the study, the assignment
The main contribution of the study is to show the effects of integrated
was made by dividing the unequal settlements into smaller areas with
material handling device assignment decisions in the facility layout
the quadratic assignment problem method.
models. Therefore, three different facility layout problems which are
In the study of Yang and Kuo [34], a static hierarchical analytical
static, static with flexible bays and dynamic with flexible bays are
hierarchy process (AHP) and data envelopment analysis (DEA) approach
evaluated. The problems are considered as consecutive models (solve
has been proposed to address the SFLP with unequal department areas.
facility layout problem first then solve the material handling device
In Deb and Bhattacharyya’s [35] study, a multi-factor fuzzy inference
assignments problem) and integrated models. In this way, the changes in
was applied to the plant layout to minimize total flow cost and dead
the minimum cost facility layouts of the problem with different char­
space in the SFLP with fixed pick/drop points and departments with
acteristics are presented. Besides, scenarios and their effects on the
different areas. Solimanpur et al. [36] used a 0–1 nonlinear program­
problem are discussed.
ming and ant colony algorithm together in flexible production systems
for single-row facility layout with departments with different areas. A
2. Literature review
new model is proposed for the static facility layout problem with un­
equal departments and using a flexible bay structure, and the method
In previous studies, FLP has been studied in many studies. It is seen
with mixed-integer minimizing the distance between departments [37].
that these studies deal with the static type of the problem intensely.
In their study, Chae and Peters [38] used a two-stage solution procedure
However, it is also seen that DFLP has been studied to work more
for facility layout with multiple-bay facilities and achieved good results
frequently in recent years. The previous studies on FLP are summarized
in various test problems with algorithms that determine the bay flexi­
in Table 1. When the previous works are examined, the problem;
bility and assign departments to these bays using mixed integer pro­
different characteristics, different solution methods and different pur­
gramming and local search methods. Azevedo, Crispim and Sousa [39]
poses are seen to vary. In this context, while analyzing the literature,
formulated a quadratic programming problem for two different prob­
static facility layout problems and dynamic facility layout problems are
lems that the location of departments within a group of facilities, and the
discussed, and also the department features used in the problems, other
location of departments inside each facility itself and they solved
problem characteristics (flexible bay structure etc.) and solution
problems with multi-objective optimization. Guan et al. [40] developed
methods are presented.
a multi-objective particle swarm algorithm for the multi-workshop fa­
The static facility layout problem (SFLP) focuses on the layout of a
cility layout problem. A new adjacency method has been proposed with
certain number of departments. In order to provide a lower bound,
mixed-integer programming in Tari and Neghabi [41], which provides
which is an important point for the solution of the problem, Anjos et al.
more flexible layouts for facility layout problems with unequal areas. In
[20] found better lower bounds for problems known in the literature
that study, departments with adjacency scores below a certain value
with semi-programming relaxation in their studies. In their study,
were considered adjacent, but this method was applied without
Kothari and Ghosh [21] tried to achieve the minimum material transport
considering the reliability status. For some test problems in the litera­
distance using a genetic algorithm called GENALGO for the SFLP.
ture, it has been shown that the new approach creates flexible and

60
A. Erik and Y. Kuvvetli Journal of Manufacturing Systems 58 (2021) 59–74

Table 1
Review of previous studies.
Studies Static Dynamic Fbs Equal Area Unequal Area Heuristic Deterministic Objectives

Konak et al. [37] X X X MIP TFD (Minimum)

Armour & Buffa, [25] X X Simulation TFC(Minimum)
Kulturel-Konak & Konak, [61] X AS MIP TFC & TRC (Minimum)
Gau & Meller, [32] X X GA, Slicing Tree MIP TFC(Minimum)
Bazaraa, [26] X X BBA TFC(Minimum)
Asl & Wong [53] X X X PSO TFC & TRC (Minimum)
Komarudin and Wong [47] X X Ant System TFC(Minimum)
Liu & Meller, [44] X X GA MIP TFD (Minimum)
Meller, [31] X X MIP, Dynamic P. TFD (Minimum)
Tam, [29] X X AS TFD (Minimum)
Kulturel-Konak & Konak [48] X X X ACO TFC(Minimum)
Van Camp, Carter ve Vannelli [28] X X NLP TFC (Minimum)
Yang & Peters [59] X X CTLDA TFC & TRC (Minimum)
Durmaz & Şahin, [3] X NSGA-II GP TFD (Minimum) & TPBD
(Maximum)
McKendall & Shang, [9] X ACO TFC & TRC (Minimum)
Şahin, [10] X AS TFC & TRC (Minimum)
Mazinani et al. [11] X X X GA MIP TFC & TRC (Minimum)
Ulutas & islier [12] X X CS TFC & TRC (Minimum)
Vitayasak et al. [13] X BSA TFC & TRC (Minimum)
Kulturel Konak. [14] X X AS MIP TFC& TRC (Minimum)
Turanoğlu & Akkaya. [15] X X BFO TFC& TRC (Minimum)
Ertay et al. [60] X X AHP, DEA TFC& TRC (Minimum)
Kheirkhah et al. [56] X X PSO BP TFC& TRC (Minimum)
Wang et al. [54] X AS TFC& TRC (Minimum)
Kheirkhah & Bidgoli [58] X X GT, GA TFC& TRC (Minimum)
Azevedo et al. [39] X X MIP TFC& TRC (Minimum)
TPBD (Maximum)
Hunagund et al. [62] X X X AS TFC & TRC (Minimum)
Moslemipour [63] X X CS, AS MIP TFC & TRC (Minimum)
Love & Wong [27] X X MIP TFC (Minimum)
Kumar et al. [30] X X CGH TFC (Minimum)
Balakrishnan, Cheng & Wong [33] X X AS, GA QAP TFC (Minimum)
Yang & Kuo [34] X X AHP, DEA TFC (Minimum)
Deb & Bhattacharyya [35] X X MFIS TFC (Minimum)
Solimanpur et al. [36] X X ACO NLP TPBD (Maximum)
Amaral [43] X X ILP TPBD (Maximum)
Anjos & Yen [45] X X SDP TPBD (Maximum)
Amaral [46] X X FCPA TFC (Minimum)
Amaral & Letchford [49] X X BBA TFC (Minimum)
Hungerländer & Rendl, [50] X X MIP, SDP TFD (Minimum)
Lenin et al. [51] X X GA TFD, MIC, TNM (Minimum)
Kothari & Ghosh [21] X GA TPBD (Maximum)
Kothari & Ghosh [22] X SA TPBD (Maximum)
Palubeckis [23] X AS TNM (Minimum)
Anjos et al. [20] X AS SDP TFD (Minimum)
Helber et al. [52] X X FO MIP TFC (Minimum)
Pourvaziri & Pierreval [55] X OQN, AS TFC & TRC (Minimum)
Tambunan et al. [24] X ALDEP, TFD (Minimum)
BLOCPLAN

TFD: Total Flow Distance, TFC: Total Flow Cost, TNM: Total number of machine, TPBD: Total proximity between departments, MIC: Machine investment cost, TRC:
Total re-arrangement cost, CTLDA: Construction type layout design algorithm, MIP :Mixed integer programming, QAP:Quadratic assignment problem, CGH:
Constructive greedy heuristic, ACO: Ant colony optimization, GA: Genetic algorithm, SDP: Semidefinite programming, AS: Simulated annealing, FCPA: Fractional
cutting plane algorithm, BBA: Branch-bound algorithm, DEA: Data envelopment analysis, SA:Scatter search, MFIS :Multi-factor fuzzy inference system, AS: Simulated
annealing, CS: Clonal selection, MIP :Mixed integer programming, SDP: Semidefinite programming, BFO: Bacterial foraging optimization, ALDEP: Automated layout
design program, GA: Genetic algorithm, ILP: Integer linear programming, PSO: Particle swarm optimization, NSGA-II: Non-dominated sorting genetic algorithm II, BP:
Bi-level programming, GP: Goal programming, FO: Fix-and-optimize heuristic, BSA: Backtracking search algorithm, OQN: Open queuing networks, TS: Tabu search
algorithm, GT: Game theory.

effective layouts. In Meller and Bozer’s [42] study, single and two-stage Anjos and Yen [45] used semi-defined programming for a 10-dimen­
models for multi-floor facility layout problems were examined and sional single-row facility layout problem and the gaps between the fa­
compared. Since there are no restrictions on the working time of the cilities were reduced to 5 % with a new formulation based on matrix and
models, it is seen that the one-stage approach gives better results than solutions were found for various problem samples up to 100 de­
the two-stage approach. In the two-stage method, firstly the floor opti­ partments. Amaral [46] proposed a new lower bound for the problem by
mization and then the placement of the departments were performed, using some valid inequalities such as static facility layout cut planes. In
and it was naturally seen that better results were obtained because there the study of Komarudin and Wong [47], the ant system was used to solve
are more options in the single-stage method. The study of Amaral [43], the SFLP with unequal departments. In the Kulturel-Konak and Konak
improved Love and Wong’s [27] model by decreasing the number of [48] study, an ant colony optimization approach has been proposed to
continuous variables to formulate a mixed-integer programming model. solve the SFLP with unequal department areas and flexible compart­
Mixed-integer programming method for the SFLP involving the unequal ments. Amaral and Letchford [49] used a multi-center study on the
areas and sequence pair display in Liu and Meller [44]. problem of SFLP, using a primitive heuristic method based on

61
A. Erik and Y. Kuvvetli Journal of Manufacturing Systems 58 (2021) 59–74

multidimensional scaling and using the branch-bound algorithm. Hun­ the center coordinates of the departments were tried to be found.
gerländer and Rendl [50] analyzed studies on single-row facility layout, Mixed-integer programming is applied to the problem, while the solu­
compared the models used and developed a quadratic semi-definite tion time is reasonable for small size problems. Additionally, a genetic
programming. Lenin et al. [51] have developed a genetic algorithm algorithm is developed for large size problems. In the Kulturel-Konak
that tries to minimize total flow distance, machine numbers and ma­ and Konak [61] study, the bay sizes and dimensions for the cyclic fa­
chine investments in the SFLP, whether it works or not. The algorithm cility layout problem, which is a special type of FLP problem, were
tested with some random problems has been found effective on indi­ addressed as decision variables and resolved with a focused and
vidual and combined targets. Helber et al. [52] developed a hierarchical mixed-integer model and simulated annealing approach. Ulutas and
layout planning based on organizational and operational data in a Islier [12] tried to minimize the sum of material flow and rearrangement
hospital with a large number of departments linked to Hannover Uni­ costs by using a clonal selection algorithm for the dynamic facility
versity. Asl and Wong [53] found the solution of static and dynamic editing problem. In the study, the shoe plant was considered and the
facility layout problems for departments with unequal areas using a algorithm was tested using real life data. Kulturel-Konak [14] solved the
particle swarm algorithm. problem of DFLP with unequal departmental areas, including bays that
Along with the static version of the facility layout problem, some of can easily adapt the region-based block layout to different material
the studies focus on the dynamic version which finds optimal facility handling devices, with tabu search algorithms and mixed integer
layout during periods. McKendall and Shang [9] used a hybrid ant sys­ mathematical modeling. The aim of the study is to obtain the horizontal
tem for the dynamic facility location problem. Şahin [10] developed a and vertical coordinates of the departments with a minimum total cost.
simulated annealing algorithm for a standard quadratic assignment Azevedo et al. [39] aimed to minimize inappropriate department set­
problem in which N departments are assigned to N fields and applied it tlements, materials transportations, and re-arrangement costs while
to the test problems in the literature. The applied algorithm has obtained maximizing closeness between departments using a non-linear integer
mostly better results than the best results in the literature. Wang et al. mathematical model for DFLP. Hunagund et al. [62] handled the facility
[54] proposed a double-row layout plan for the dynamic facility layout layout problem together with unequal department areas and a flexible
problem and determined the solution alternatives by a simulated bay structure. The problem addressed in that study determines the
annealing algorithm. Vitayasak et al. [13] has created three different flexible bay structure, the coordinates of the departments, and the di­
algorithms by modifying the trace-back algorithm for stochastic DFLP. mensions of the facility for periods. Moslemipour [63] proposed a
In this study, it was tried to minimize the re-arrangement of the ma­ quadratic assignment-based mathematical model for stochastic dynamic
chines in the facility and the material transportation costs within the facility layout. The proposed model focused on the assigning of de­
facility according to the demands of the products. The best of the partments to areas achieving minimum re-arrangement and material
modified trace-back algorithms gave better results than the classic transportation costs. Also, effective solutions were reached in a short
trace-back and genetic algorithm. Pourvaziri and Pierreval [55] pro­ time by using annealing simulation and clonal selection algorithms.
posed an open queue network theory for the facility location problem, In this study, the integration of facility layout decisions with material
which takes into account the amount of in-process work in a particular handling device assignment decisions is considered. Three different fa­
problem class in the dynamic facility location problem. The proposed cility layout problems with characteristics static, static included flexible
theory is formulated with a multi-objective cloud-based annealing bays, and dynamic included flexible bays are modeled to show the effect
simulation. of integrated decisions.
Kheirkhah et al. [56] extended the DFLP including material handling
devices. In that study, a two-level model for the problem is proposed to 3. Material handling device assignment and facility layout
consider the purchasing and remaining idle cost of material handling problem
devices using a particle swarm algorithm. Ulutas and Islier [57]
addressed the dynamic facility layout problem with the design of de­ In the facility layout problem, departments are assigned to the fa­
partments with equal areas and transportation vehicles with the same cility area. Let N denotes the number of departments that should be
unit costs. The clonal selection algorithm developed for 3 test problems assigned to the facility area. The departments are assigned to the WxH
in the literature reached the best-known solutions and showed a better area which are width and height limitations of the facility. The de­
performance in terms of solution times. In the study of Kheirkhah and partments may have equal or unequal areas regarding the facilities. In
Bidgoli [58], the problem is modeled as a game theory that addresses this study, the areas are assumed to be unequal for each department for
inefficiencies resulting from the contradiction of goals ignored in the all mathematical models. The length and width of the departments are
DFLP literature. Three metaheuristic methods are proposed in the model limited by the aspect ratio. The departments are assigned to flexible bays
to achieve the Nash equilibrium. Turanoğlu and Akkaya [15] used the in some models in which the flexible bays covered vertical or horizontal
bacterial feed search approach and simulated annealing approach for axes. Let B denotes the number of available bays in the system. Another
dynamic facility layout problem. The hybrid algorithm applied to important characteristic of the problem is to determine the number of
various DFLP samples from the literature has achieved competitive re­ periods. Some of the models consider periodic changes on the flows and
sults in short solution time. thereby they try to find a robust layout plan during the planning horizon
In their study, Yang and Peters [59] identified a robust machine which denotes as T. Finally, in the facility layout planning problem, the
layout on a series of planning time windows covering the total planning material handling devices should also be considered with two-stage
horizons, using an intuitive procedure based on the structure layout approach which solves facility layout problem and material handling
design algorithm to develop flexible machine layout with unequal areas. device assignment problem separately or integrated approach. All of
Ertay et al. [60] collected quantitative data by using fuzzy set theory and these characteristics are summarized in Fig. 2. Accordingly, the effects of
qualitative data by the AHP method for the dynamic facility layout the integrated decisions on the material handling device assignments
problem. Data envelopment analysis was used for the selection of and facility layout is aimed to investigate in this study. Therefore,
alternative layout plans that are created by using computer-aided layout different problem characteristics should be evaluated regarding math­
software. Nine alternative layouts from nineteen alternative layouts ematical models as given follows:
were obtained by using data envelopment analysis to handle the cost
and closeness score as input and maximize the aspect ratio, flexibility, • Model A: Static facility layout problem and material handling device
and quality. In the study of Mazinani et al. [11], the dynamic facility assignments separately.
layout problem was handled together with the flexible bay structure. In • Model B: Static facility layout problem and integrated material
that study, different widths and the number of bays are considered when handling device assignments.

62
A. Erik and Y. Kuvvetli Journal of Manufacturing Systems 58 (2021) 59–74

xvi In the model, the long edge of the facility is located on the hor­
izontal (x) axis without general loss and the flexible bay extends
parallel to the vertical (y) axis (for model E and F).
xvii (0,0) is the starting coordinate on (x,y) axes, which is the
southwest corner of the facility.

3.1. Notations

Notations used in the models are presented as follows:

Nomenclature Relevant
Indices Models

t: periods; t = 1, 2, … , T E,F
i, j : departments; i, j = 1, 2, … , N A,B,C,D,
E,F
k: flexible bays; k = 1, 2, … , Bt : E,F
g: material handling devices; g = 1, 2, … , G A,B,C,D,
E,F
Parameters:
W : facility width on the x axis A,B,C,D,
E,F
H : facility height on the y axis A,B,C,D,
E,F
Fig. 2. Problem Definition. B : maximum allowed number of bays C,D
Bt : maximum allowed number of bays at period t E,F
• Model C: Static facility layout problem with flexible bays and ma­ ai (qi ) : required area (aspect ratio)for department i A,B,C,D
ati (qti ) : required area (aspect ratio) for department i at period t E,F
terial handling device assignments separately. (
lmaks :maximum length for department i lmaks = min{W, A,B
• Model D: Static facility layout problem with flexible bays and inte­ i i
)
grated material handling device assignments. (ai *qi )05 }
• Model E: Dynamic facility layout problem with flexible bays and lmaks
(
C,D
i :maximum length for department i lmaks i = min{H,
material handling device assignments separately. )
(ai *qi )0.5 }
• Model F: Dynamic facility layout problem with flexible bays and ( )
integrated material handling device assignments. lmin
i : minimum length for department i lmin = (ai /qi )0.5 A,B,C,D
i

lmaks
ti :maximum length for department i at period t (lmaks = max{ E,F
ti
The assumptions of the models are as follows: W, (ati *qti )0.5 })
i Departments have a square or rectangular shape with a limited lmin
ti
(
: minimum length for department i at period t lmin = E,F
ti
maximum aspect ratio and equal or unequal areas. 0.5
)
ii Distances between departments are taken as rectilinear distances.
(ati /qti )

iii The width and length of the facility are known in advance. Fij :Amount of material flow between departments i and j A,B,C,D
Ftij :Amount of material flow between departments i and j in E,F
iv The position orientation of the departments can vary horizontally
period t
or vertically within the flexible bay changing the width of flexible Cgij :The cost for transporting per unit material a unit distance A,B,C,D
bay width (for models C,D,E and F). between departments i and j using g device
v If one or more departments are assigned to a bay, then that bay Ctgij :The cost for transporting per unit material a unit distance E,F
between departments i and j using g device in period t
must be completely filled (for models C,D,E and F).
Ati :Rearrangement fixed cost of shifting department i at the E,F
vi Department areas are fixed during the period, but it may be re- beginning period t
arranged from period to period (for models E and F). Rti :Rearrangement variable cost of shifting department i at the E,F
vii Re-arrangement may be made at the end of the period (for model beginning period t
E and F). fgij :The capacity of the device g between facilities i and j. A,B,C,D,
E,F
viii No change happens in material flows during the period and ma­
rwg :Average percentage working time of device g. A,B,C,D,
terial flows are dynamic and deterministic as they can vary be­ E,F
tween periods (for model E and F). sg :Average speed of MHD g in meters per minute A,B,C,D,
ix The displacement of all machines and equipment is possible (for E,F
model E and F). Lg
yb :Average loading and offloading time of device g that is A,B,C,D,
transporting at a medium speed (in min). E,F
x There is no limitation to the number of material handling devices O :Total available time for a period A,B,C,D,
assigned to a bay in the facility (for model E and F). E,F
xi Material handling devices do the transportation tasks at a known cstg :Purchasing cost per unit material handling device g at the E,F
constant speed and capacity. beginning of the period t
catg :Non-operating cost per unit material handling device g E,F
xii There is fixed and known operating, purchasing and non-
thorough period t.
operating costs of a material handling device. cutg : Operating cost of the g handling device thorough period t. E,F
xiii The objective functions of the integrated models (B, D, and F) cug : Purchasing and operating cost of the g handling device A,B,C,D
include different terms that have equal weights by considering Decision variables:
the expert choice. wi :Width (the length in the y-axis direction) of department i A,B,C,D
xiv Each transportation tasks can be carried out with a material wtk :Width (the length in the x-axis direction) of bay k in period t E,F
handling device. hi :Height (the length in the y-axis direction) of department i A,B,C,D
htik :Height of department i in bay k in period t C,D
xv The loading and unloading costs of the material handling devices y
li :Height (the length in the y-axis direction) of department i E,F
are included in the unit cost of the material transportations be­ y
E,F
lti
tween the departments.
(continued on next page)

63
A. Erik and Y. Kuvvetli Journal of Manufacturing Systems 58 (2021) 59–74

(continued )
hi hi
Nomenclature Relevant ≤ yi + ≤ H ∀i (5)
2 2
Indices Models

:Height (the length in the y-axis direction) of department i in lmin

i ≤ wi ≤ lmaks
i ∀i (6)
period t
xi , yi : (x,y) coordinates of the centroid of department i A,B,C,D wi *hi = ai ∀i (7)
xti , yti : (x,y) coordinates of the centroid of department i in period t E,F
dxij :Horizontal distance between the centers of departments i A,B,C,D
and j (
wi , hi , dijx dijy , xi , yi ≥ 0 ∀j, i (8)
dxij =│xi-xj│=d+x − x
ij +dij )
dxtij :Horizontal distance between the centers of departments i E,F mij ∈ {0, 1} ∀j, i (9)
and j in period t (dxtij =│xti-xtj│=d+x − x
tij +dtij )
y
The objective function (1) of the model is to minimize total material
dij :Vertical distance between the centers of departments i and j A,B,C,D
transportations. Constraints (2) and (3) ensures that the two de­
(
y
dij =│yi-yj│=dij +dij )
+y − y partments i an j cannot be overlapped. Constraints (4) and (5) forces the
y
dtij :Vertical distance between the centers of departments i and j E,F center coordinates of the departments within the facility limitations for
y
in period t (dtij =│yti-ytj│=dtij +dtij )
+y − y vertically and horizontally. Constraint (6) is the aspect ratio constraint
pxti :The amount of horizontal movement for department i from E,F that provides the width of the departments that must be in the limita­
period t− 1 to t (pxti =│xti-xt-1,i│=p+x − x
ti +pti )
tions of the department. Constraint (7) ensures the department area.
pyti :The amount of vertical movement for department i from E,F Finally, constraints (8) and (9) is the valid ranges of the decision
period t− 1 to t (pyti =│yti-yt-1,i│=p+y − y
tij +ptij ) variables.
ng :Number of g devices used A,B,C,D When the optimal facility layout is obtained from the first model,
ntg :Number of g devices used in period t E,F distances between departments in the optimal solution is passed to the
nntg :Number of purchased device g in the period t E,F
second model in order to decide material handling device assignments.
natg :Number of g devices not used in period t E,F *y
{
A,B
Where d*x
ij and dij is the optimal values of the distances between de­
1 if department i and j overlapping in the vertical axis
mij =
0 otherwise partments i-j and the formulation of the second model is given as
{
zik =
1 if department i is assigned to bay k C,D follows:
0 otherwise
{ Objective function
1 if department i is assigned to bay k in period t E,F
ztik =
0 otherwise ∑
G ∑
N ∑
N ( ) ∑G
{
1 if bay k is occupied C,D minz = Cgij *ybgij *Fij * dij*x + dij*y + ng *cug (10)
mk =
0 otherwise g=1 i=1 j∕
=i g=1
{
1 if bay k is occupied in period t E,F
mtk =
0 otherwise Constraints
{
1 if department i is above department j in the same bay C,D ⎛ ( ) ⎞
sij = dij*x + dij*y *ybgij
0 otherwise ∑∑ Fij
E,F * Lg *ygij +
⎝ yb b ⎠ ≤ ng *O ∀g, j, i ∕
=j (11)
stij = j i
fgij sg *rgw
{
1 if department i is above department j in the same bay in period t ∑
0 otherwise ybtgij = 1
{
1 if department i is rearranged at the beginning of period t E,F
∀ g, j, i (12)
rti = g
0 otherwise
{
1 if there is material handling between i and j using device g A,B,C,D
b
ygij = ng ≥ 0 and integer ∀j, i (13)
0 otherwise
{
1 if material handled between i and j using device g at period t E,F
ybtgij = ybgij ∈ {0, 1} ∀j, i (14)
0 otherwise

The objective function (10) of the model A is to minimize total

3.2. Model A transportation costs and the total purchasing and operating costs of
material handling devices. Constraint (11) ensures that the total time
In model A, the classical static facility layout problem solved first and used by the gth type device that has a certain capacity (fgij ) for material
the department centers are fixed. Then, the obtained facility layout is fed transportation is less than the total time available during the planning
to the material handling device assignment problem. The formulation of horizon (O) of this device at a period. This constraint is also used to
the first stage is very similar to the problem is proposed by Van Camp determine the number of material handling devices required for mate­
et al. [28] and the mixed integer nonlinear programming model is given rial transport between departments. Constraint (12) ensures that the
as follows transportation between departments i and j could only be provided by a
Objective function single transportation device type. Constraints (13) and (14) are valid
( ) ranges of the decision variables.
∑
N ∑
N
minz = Fij * dijx + dijy (1)
i=1 j∕
=i
i<j
3.3. Model B

Constraints Model B has the same facility layout characteristics as model A. The
wi wj main difference between the two models is related to how they make
+ ≤ dijx + W*mij ∀i, j (i < j) (2) material handling device assignment decisions. Model B takes these
2 2
decisions in an integrated manner, unlike model A. Accordingly, the
hi hj ( )
(3) mathematical formulation of model B is similar to model A. The objec­
+ ≤ dijy + H* 1 − mij ∀i, j (i < j)
2 2 tive function seen in Eq. 15 is similar to the objective function of model
A (10), and the only difference between them is that the distances be­
wi wi
≤ xi + ≤ W ∀i (4) tween departments are decision variables in model B. Similarly,
2 2
although it is similar to the constraint (11) in Constraint (16), only the

64
A. Erik and Y. Kuvvetli Journal of Manufacturing Systems 58 (2021) 59–74

distances between departments are decision variables. Constraints (2)– ( )

yi − 0.5*lyi ≥ yj − 0.5*lyj − H* 1 − sij ∀i, j i ∕
=j (28)
(9) and (12)–(14) in Model A also apply to model B.
∑
G ∑
N ∑
N ( ) ∑G sij + sji ≤ 1 ∀i, j i < j (29)
minz = Cgij *ybgij *Fij * dijx + dijy + ng *cug (15)
g=1 i=1 j∕
=i g=1
sij + sji ≥ zik + zjk − 1 ∀i, j, k i < j (30)
Constraints
⎛ ( ) ⎞ 0.5*lyi ≤ yi ≤ H − 0.5*lyi ∀i (31)
∑∑ Fij dijx + dijy *ybgij
*⎝Lg *ygij +
yb b ⎠ ≤ ng *O ∀g, j, i ∕
=j (16) wk , hik , lyi , dijx dijy , xi , yi ≥ 0 ∀j, i, k (32)
j i
fgij sg *rgw
zik ∈ {0, 1}, mk ∈ {0, 1}, sij ∈ {0, 1} ∀j, i, k (33)
with subject to (2)–(9) and (12)–(14).
The objective function (17) of the model is to minimize total material
transportations. Constraint (18) forces to assign each department to a
3.4. Model C
bay. Constraint (19) ensures the width of each bay regarding assigned
department areas. Constraint (20) forces to lower and upper bounds of
Model C refers to the static facility layout problem with flexible bay
the bay widths. Constraints (21) and (22) ensures the valid range of x
structures. The model differs from model A by the consideration of
centroids of a department. Constraints (23) – (26) ensures the calcula­
flexible bays. In this type of model, the departments are assigned to bays;
tion of the department lengths. Constraint (25) forces that the total
therefore, the area constraints are satisfied by considering the fixed bay
heights of departments assigned to the bay equal to the height of the
length. The model formulation is obtained from the study of Konak et al.
facility. Constraint (26) ensures the limitations of side lengths of de­
[37] as follows:
partments height’s in the bay. Constraint (27) determines the height of
∑
N ∑
N ( ) the department. Constraints (28)–(30) ensures to avoid overlapping and
minz = Fij * dijx + dijy (17) to provide positional relations on the y-axis. Constraint (31) forces to
i=1 j∕
=i
limitations on the y-axis. Finally, Constraints (32) and (33) are the valid
ranges of the decision variables. Similar to model A, model C has the
i<j

Constraints second model (Eq. 10–14) which includes the optimal values of d*x ij and

∑
B d*y
ij .
zik = 1 ∀i (18)
3.5. Model D
k=1

1 ∑
N
wk = * zik *ai ∀k Model D has the same facility layout characteristics as model C. The
H i=1 difference between the model C and D is that material handling device
assignment decisions are obtained from whether separate models or not.
i *zik ≤ wk ≤ li
lmin maks
+ W*(1 − zik ) ∀i, k (19) Model D takes these decisions in an integrated manner, unlike model C.
Accordingly, the objective function of model D is the same with Eq. 15.
∑
B
xi ≥ wv − 0.5*wk −
(
W − lmin
)
*(1 − zik ) ∀i, k The constraints between (18)–(33), (12)–(14) and (16) also apply to
model D.
i
v=1 (20)

3.6. Model E
v≤k

∑
B
( )
xi ≥ wv − 0.5*wk − W − lmin *(1 − zik ) ∀i, k The dynamic version of the facility location problem with a flexible
i
v=1 (21) bay structure is referred to in the Model E. The model differs from model
C regarding the multi-period consideration of the problem. Multi-period
consideration provides to determine the changes in the layout during the
v≤k

planning horizon. Thus, it is possible to re-configure the layout ac­

∑
B
( )
xi ≤ wv − 0.5*wk + W − lmin *(1 − zik ) ∀i, k cording to the changes that will occur in the flows throughout the pe­
i
v=1 (22) riods. The model formulation is obtained from the study of Mazinani
et al. [11] as follows:
Objective function
v≤k

{ }
hik hjk lmaks lmaks ( ) minz =Z1 + Z2 (34)
* 2 − zik − zjk (23)
i j
− ≤ max , ∀i < j, k
ai aj ai aj ( )
Z1 = Ftij * dijx + dijy (35)
{ }
hjk hik lmaks lmaks ( )
* 2 − zik − zjk (24)
i j
− ≤ max , ∀i < j, k T ∑
∑ N T ∑
∑ N
( )
aj ai ai aj
Z2 = Ati *rti + Rti * pxti + pyti (36)
t=2 i=1 t=2 i=1
∑
N
hik = H*mk ∀k (25) Constraint
i=1 ∑
ztik = 1 ∀t, i (37)
i *zik ≤ hik ≤ li
lmin maks
∀i, k (26) k

1 ∑
∑
B wtk = * ztik *ati ∀t, k (38)
hik = lyi ∀i (27) H i
k=1

65
A. Erik and Y. Kuvvetli Journal of Manufacturing Systems 58 (2021) 59–74

ti *ztik ≤ wtk ≤ lti

lmin maks
+ W*(1 − ztik ) ∀t, i, k (39) indicate a valid range for decision variables.
Similar to the other two stage approach models A and C, model E has
∑ ( )
wtj − 0.5*wtk − W − lmin *(1 − ztik ) ∀t, i, k a separate material handling device assignment model which ensures
xti ≥ ti (40)
j≤k the material handling device assignment regarding the optimal values of
*y
∑ ( ) d*x
ij and dij in the previous model. The material handling device assign­
xti ≤ wtj − 0.5*wtk + W − lmin
ti *(1 − ztik ) ∀t, i, k (41) ment model for model E is formulated as follows:
j≤k
Objective function
{ }
htik htjk lmaks lmaks ( ) minz =Z1 + Z2 + Z3 (59)
* 2 − ztik − ztij ≤ 0 (42)
ti tj
− − max , ∀t, k, i, j > i
ati atj ati atj
G ∑
∑ T ∑
N ∑
N ( )
{ } Z1 = Ctgij *ybtgij *Ftij * dij*x + dij*y (60)
htik htjk lmaks lmaks ( ) g=1 t=1 i=1
* 2 − ztik − ztij ≥ 0 (43)
ti tj j∕
=i
− + max , ∀t, k, i, j > i
ati atj ati atj ∑T ∑N ∑T ∑N ( *y
)
Z2 = Ati *rti* + Rti * p*x
ti + pti (61)
∑ t=2 i=1 t=2 i=1
htik = H*mtk ∀t, k (44) ∑T ∑G ∑T ∑G ∑T ∑G
nn *cstg na *catg ntg *cutg (62)
i
Z3 = t=1 g=1 tg
+ t=1 g=1 tg
+ t=1 g=1

ti *ztik
lmin ≤ htik ≤ lmaks
ti *ztik ∀t, i, k (45) Constraints
∑ ⎛ ( ) ⎞
htik = lyti ∀t, i (46) ∑∑ Ftij dij*x + dij*y *ybtgij
k *⎝Lyb
g *ytgij +
b ⎠ ≤ ntg *O ∀t, g, j, i ∕
=j (63)
j i
fgij sg *rgw
yti − 0.5*lyti ≥ ytj + 0.5*lytj − H*(1 − stij ) ∀t, i, j ∕
=i (47)
ntg = nnt− 1g + nntg − natg ∀g, t ≥ 2 (64)
stij + stji ≤ 1 ∀t, i, j > i (48)
ntg = nn1g ∀g, t = 1 (65)
stij + stji ≥ ztik + ztjk − 1 ∀t, k, i, j > i (49)
nntg *natg = 0 ∀t, g (66)
0.5*lyti ≤ yti ≤ H − 0.5*lyti ∀t, i (50)
∑
ybtgij = 1 ∀ t, g, j, i (67)
xti − xt− 1,i ≤ W*rti ∀i, t > 1 (51) g

xt− − xti ≤ W*rti ∀i, t > 1 (52)

1,i ntg , natg , nntg ≥ 0 ∀t, g (68)
yti − yt− ≤ H*rti ∀i, t > 1 (53)
(69)
1,i
ybtgij ∈ {0, 1} ∀t, g, j, i
yt− 1,i − yti ≤ H*rti ∀i, t > 1 (54) The objective function (59) of the model is the sum of the three terms
which are total transportation costs between departments (60), total re-
lyti − lyt− 1,i ≤ H*rti ∀i, t > 1 (55) *y
arrangement costs which are obtained from the optimal values of p*x ti , pti
and rti previous model (61), the last term is the cost incurred by the new
*
lyt− − lyti ≤ H*rti ∀i, t > 1 (56)
material handling device purchasing and the cost due to the absence and
1,i

use of material handling devices (62). Constraint (63) ensures that the
xti , yt i, lyti , wtk , htik , dtij
+x − x
, dtij +y
, dtij − y
, dtij , p+x − x +y − y
ti , pti , pti , pti ≥ 0 ∀t, g, j, i
total time used by the gth type device that has a certain capacity (fgij ) for
(57)
material transportation is less than the total time available during the
ztik ∈ {0, 1}, mtik ∈ {0, 1}, stij ∈ {0, 1}, rti ∈ {0, 1} ∀t, g, j, i (58) planning horizon of this device at a period. This constraint is also used to
determine the number of material handling devices required for mate­
In the objective function (34), the first term (35) total inter- rial transport between departments. The constraints between (64) and
departmental material transportation amounts and the second term (66) describe the relationship among the number of required gth type
(36) is the fixed costs required for rearrangement and the variable costs devices, purchased gth type devices, and non-operated gth type device in
for rearrangement. The objective function aims to minimize all costs period t. Constraint (67) is the constraint that enables the transportation
incurred. between two departments during period t to be performed by a gth type
Constraint (37) ensures that each department is assigned to a single device. Finally, constraints (68) and (69) are valid ranges for decision
bay. Thus, the assignment of one department to more than one bay is variables.
avoided. Constraints (38) and (39) calculate the width of each bay on the
horizontal axis, depending on the departments assigned to that bay.
Constraint sets (40) and (41) determine the locations of the de­ 3.7. Model F
partments’ center along the horizontal axis. Constraints (38), (40) and
(41) ensure that the departments are located within the boundaries of The proposed mathematical model F includes the flexible bay
the facility along the horizontal axis. Constraints between (42) and (49) structure and material handling devices assignment decisions in an in­
are used to determine the location of departments’ centroid along the tegrated manner, unlike the classical dynamic facility layout models.
vertical axis. These constraints also ensure that departments do not The aim of the proposed model is to minimize the total cost caused by
overlap in the vertical axis. Constraint (50) ensures that departments are transporting materials between departments, total displacement (re-
within the boundaries of the facility floor along the vertical axis. Con­ arrangement) costs of departments, and total material handling device
straints between (51) and (56) ensure that the department has the same assignment costs which includes the cost of purchasing, operating and
length, width, and centroid in two consecutive periods where the non-operating material handling devices. The nonlinear mixed integer
department has not been rearranged. Finally, constraints (57) and (58) programming formulation for this problem is given as follows:

66
A. Erik and Y. Kuvvetli Journal of Manufacturing Systems 58 (2021) 59–74

Objective function bays in each period), FBS-DFLP-3 (8 departments, 6 periods and 3

maximum bays in each period) and FBS-DFLP-4 (12 departments, 4
MinZ = Z1 + Z2 + Z3 (70)
periods and 5 maximum bays in each period), respectively. Besides, a
G ∑
∑ T ∑
N ∑
N ( ) problem instance namely P6 (6 departments, 6 periods and 6 maximum
Z1 = Ctgij *Ftij * dijx + dijy *ybtgij (71) bays in each period) is obtained from the study of Yang & Peters [59].
g=1 t=1 i=1 j∕
=i All of the mathematical models are solved using the SCIP solver of
∑T ∑N ∑T ∑N GAMS software with a computer having a 3.60 GHz processor, 16 GB
( )
Z2 = Ati *rti + Rti * pxti + pyti (72) memory and Windows 10 operating system. Generally, the solutions are
t=2 i=1 t=2 i=1
obtained within <6 h for some of the problem instances and all of the
∑T ∑G ∑T ∑G ∑T ∑G problem instances are limited 18 h solution time.
Z3 = nn *cstg + na *catg + ntg *cutg (73)
t=1 g=1 tg t=1 g=1 tg t=1 g=1 For the evaluation of the results from the 6 models examined within
⎛ ( ) ⎞ the scope of the study, the 6th model named model F is selected and the
∑∑ Ftij dijx + dijy *ybtgij problem instances results obtained in this model are presented in sub-
* Lg *ytgij +
⎝ yb b ⎠ ≤ ntg *O ∀t, g, j, i ∕
=j (74)
fgij sg *rgw section 4.1. Sub-section 4.2, 4.3 and 4.4 presents the comparison of
j i
model A–B, C–D and E–F results, in order to analyze the effects of in­
with subject to (37)-(58) and (64)-(69) tegrated facility layout and material handling device assignment
In the objective function (70), the first term (71) inter-departmental decisions.
material transportation costs, the second term (72) is the fixed costs
required for rearrangement and the variable costs for rearrangement,
4.1. Results for Model F
the third term (73) is the cost incurred by the new material handling
device purchasing, and the cost due to the absence and use of material
The layout plan obtained for FBS-DFLP-1, FBS-DFLP-2, FBS-DFLP-3,
handling devices. The objective function aims to minimize all costs
FBS-DFLP-4 and P6 can be seen in Figs. 3, 4, 5, 6 and 7 respectively. The
incurred. Constraint (74) is similar to the constraint (63) but the dif­
y facility layouts for all instances differ from the study of Mazinani et al.
ference that dxij and dij are decision variables in (74).
[11] and Yang & Peters [59] for P6. The difference between the two
layouts infers that the material handling devices have an effect on the
4. Computational study facility layout design.
When the assignments of material handling devices in FBS-DFLP-1
In the computational study, all of the models are evaluated by and FBS-DFLP-2 are examined, a device is purchased at the beginning
problem instances obtained from the literature. The material handling of the 1 st period and according to the results, the same device is used
device assignment parameters are not included in these problem in­ during the periods. Only one device is assigned because there are a few
stances; therefore, the material handling device assignment parameters numbers of departments. In FBS-DFLP-3, there are purchases and non-
are generated randomly for all problem instances. operating material handling devices. Accordingly, five material
For models A and B, three problem instances which have 7, 8, and 9 handling devices are purchased at the beginning of the 1 st period, one
departments, from the study of Komarudin & Wong, [47] namely O7, device is assigned as non-operating at the beginning of the 4th period,
O8, and O9 are solved. For models C and D, three problem instances and finally, the necessity to purchase a device at the beginning of the 6th
which have 10, 12, and 14 departments, from the literature namely period has occurred. The material handling devices assignment de­
VC10, BA12, and BA14 are solved. VC10 instance is obtained from the cisions and costs are changed frequently when the number of periods
Van Camp, Carter ve Vannelli [28] while the other three instances are and departments are increased. are also affects the results. The FBS-
obtained from Komarudin & Wong, [47]. The number of bays for the DFLP-4 and P6 results show that the model purchase the material
models is set 5, 7, and 5 respectively as the study of Konak et al. [37]. handling devices at the beginning of the planning horizon.
Finally, the 4 different data sets used to test the models E and F are Transportation costs (Z1 ), rearrangement costs (Z2 ), and material
obtained from the study of Mazinani et al. [11] which has similar handling devices related costs (Z3 ) in the proposed model vary from
problem characteristics. The four generated problem instances are period to period. The changes in costs for all of the problem instances
named as FBS-DFLP-1 (4 departments, 3 periods and 3 maximum bays in throughout the periods are shown in Fig. 8. When the displacement
each period), FBS-DFLP-2 (5 departments, 2 periods and 3 maximum results of the FBS-DFLP-1 are analyzed, none of the departments in the

Fig. 3. Minimum cost facility layout of FBS-DFLP-1.

Fig. 4. Minimum cost facility layout of FBS-DFLP-2.

67
A. Erik and Y. Kuvvetli Journal of Manufacturing Systems 58 (2021) 59–74

Fig. 5. Minimum cost facility layout of FBS-DFLP-3.

Fig. 6. Minimum cost facility layout of FBS-DFLP-4.

Fig. 7. Minimum cost facility layout of P6.

second period changed, whereas almost all departments except the first (Fig. 8b). The Z2 and Z3 are crossed during the periods and it means
department changed in the third period (Fig. 8a). Changes encountered that the trade-off between decisions affects the problem.
are affected by conditions such as demand fluctuations between periods, When the problem instance increases the re-arrangement decisions
the variability of material handling costs, and the carrying capacity of are frequently used. As the facility layout problem is one of the strategic
material handling devices. For the FBS-DFLP-1 problem, it is seen that decision-making problems, it is obvious that the re-arrangement de­
the material handling costs are small between the periods due to de­ cisions are made within long periods. When the number of periods in­
mand fluctuations are not high. Similarly, the variability between creases the material handling and re-arrangement costs get higher as can
transportation costs is found to low since the problem instances are a be seen on FBS-DFLP-3 (Fig. 8c). It is obvious that the model decides re-
small-scale facility layout problem. arrangement instead of total transportation costs in order to reduce the
Displacement encountered in the second period due to the demand total cost. In the obtained facility layout, only the location of the fourth
fluctuation and the cost changes during the periods for FBS-DFLP-2 department is not changed place in the third period, while the location

68
A. Erik and Y. Kuvvetli Journal of Manufacturing Systems 58 (2021) 59–74

Fig. 8. The changes of the costs during the periods for (a) FBS-DFLP-1, (b) FBS-DFLP-2, (c) FBS-DFLP-3, (d) FBS-DFLP-4, (e) P6.

of all departments is changed in all the remaining periods. The Z2 for Table 2. Considering the costs, it is seen that transportation costs have a
FBS-DFLP-3 balances the increase of Z3 and Z1 and decreases the total portion between 85 % and 92 %. This conclusion reveals the importance
cost. of transportation costs, which are widely discussed in the literature. The
The costs during the periods are shown in Fig. 8d for FBS-DFLP-4. In operating, non-operating and purchasing costs of material handling
order to overcome these transportation costs, the model tries to make re- devices have a portion between 6 % and 8 % within the total cost con­
arrangement. Controversially, the material handling device assignments sisting of the problem size, while the displacement cost varies between
are preferred for P6 in Fig. 8e. In that problem instances, total trans­ 0.5 % and 6 %. It is observed that especially the transportation costs are
portation costs decreased by material handling device assignments in reduced by re-arrangement decisions included in the model.
the third and sixth periods.
The costs related to transportations, rearrangements and material
handling devices assignments for all instances are summarized in

69
A. Erik and Y. Kuvvetli Journal of Manufacturing Systems 58 (2021) 59–74

Table 2 Table 4
Cost shares of the models. Comparison of the model C and D.
Problem Instances TC MHC RAM Total Costs Problem Instances

4097,679 325 32,278 4454,957 Criteria VC10 BA12 BA14

FBS-DFLP-1
91,98 % 7,30 % 0,72 % 100,00 %
Konak et al. [37] 22899.65 8801.33 5004.55
3125,962 275 87,75 3488,712
FBS-DFLP-2 Literature Model C
89,60 % 7,88 % 2,52 % 100,00 %
Comparison (without cost 21463.07 8786 5311.409
32963,195 3120 2348,018 38431,213
FBS-DFLP-3 multiplication)
85,77 % 8,12 % 6,11 % 100,00 %
Z1 38712.67 15121.50 8205.954
65613,306 4565 2415,93 72594,236
FBS-DFLP-4 Z3
90,38 % 6,29 % 3,33 % 100,00 % 37 4 133.000
Model C Total Costs 38749.67 15125.50 8338.954
Average Distances 13.40 11.96 2.531
# of all MHD 1 1 7
Table 3 Z1 37270.89 14540.33 8205.954
Comparison of the model A and B. Z3 37 4 133.000
Problem Instances Model D Total Costs 37307.89 14544.33 8338.954
Average Distances 9.30 12.39 2.531
Study O7 O8 O9 # of all MHD 1 1 7
Komarudin and Wong [47] 131.68 243.12 236.12 Layout Change 328.37 67.83 0
Liu and Meller [44] 131.63 245.41 246.26
Literature Scholz et al [64] 132.00 243.16 239.07
Comparison Model A 4.4. Comparison of Model E and F
(without cost 131.65 242.69 236.98
multiplication)
Table 5 shows the costs obtained by the model E approach and the
Z1 169.43 641.05 345.7
model F approach. When the model F results are compared with the
Z3 18 46 6
Model A
model E results, there is a significant difference in the total cost for all
Total Costs 187.43 687.05 351.7
Average Distances 3.93 4.09 4.43
problem instances. In the model E solution, the transportation costs are
# of all MHD 2 1 3 higher than the proposed approach together with the total costs. This
Z1 159.49 640.21 321.41 reveals that handling the material handling devices assignment with the
Z3 18 46 8 flexible bay structure reduces the cost of facility location. In addition,
Model B Total Costs 177.49 686.21 329.41 the proposed model affects the transportation costs in the direction of
Average Distances 2.66 4.09 4.44 decrease except for P6.
# of all MHD 2 1 4
In some problem instances, the transportation costs decrease and the
Layout Change 32.11 56.02 68.36
re-arrangement decisions change in the proposed approach even if the
material handling devices related costs are not changed. This shows that,
4.2. Comparison of Model A and B in addition to the decreasing costs, the decisions related to the material
handling devices were taken in an integrated manner have a significant
The static facility layout and material handling device assignment impact on the facility layout.
problem (model A) and the integrated version of the problem (model B) According to results obtained from the model E and F for the FBS-
is compared to show the effect of the integrated decisions on the facility DFLP-1 problem instances, total costs are expressed in Fig. 9(a) . The
layout and the results are shown in Table 3. The first stage of model A total cost increases continuously until the second period for model E
tends to find good solutions as given in the literature. According to cost while the total cost decrease for model F. The main reason for the in­
comparisons, integrated decisions decreases the material handling de­ crease is the increase in transportation costs whereas the decrease
vice assignment costs. Thereby, total costs decreased %5 on average for happens because of the decisions taken in an integrated manner. In the
overall problem instances. The average distances and layout changes are model F, taking displacements in the last period increased the total cost,
indicators of the effect of integrated decisions. Most of the instances
have less average distances between departments and layout changes Table 5
between models A and B have always positive values. It shows that in­ Comparison of the model E and F.
tegrated decisions have an influence on the facility layout.
Problem Instances

4.3. Comparison of Model C and D Criteria FBS- FBS- FBS- FBS- P6

DFLP-1 DFLP-2 DFLP-3 DFLP-4

The static facility layout with flexible bay structure and material Z1 4145.24 3222.02 40014 68929.9 25716
handling device assignment problem (model C) and the integrated Z2 0 72.5 1692 2421.38 524
version of this problem (model D) are compared to show the effect of the Z3 665 275 3060 4565 2046.5
Model
integrated decisions on the facility layout and the results are shown in Total Costs 4810.24 3569.52 44765 75916.3 28286
E Average
Table 4. The first stage of model C tends to find good solutions given in 5.13 7.31 5.43 6.22 26.182
Distances
the literature. According to cost comparisons, both models having # of all
similar total cost values. Only the material transfer costs are reduced in 1 1 5 11 12
MHD
the integrated model (model D) except for BA14. The average distances Z1 4097.68 3125.962 32963 65613.3 26248
and layout changes are indicators of the effect of integrated decisions. Z2 325 275 3120 4565 412
All of the problem instances have the same average distances between Z3 32.278 87.75 2348 2415.93 1707
departments; however, layout changes between models C and D have Model Total Costs 4454.96 3488.712 38431 72594.2 28367
F Average
generally positive values. The number of material handling devices are 4.01 7.23 5.54 6.28 26.142
Distances
same with each other in the models.
# of all
1 1 5 11 10
MHD
Layout Change 54,45 54.45 18 321.5 414.86

70
A. Erik and Y. Kuvvetli Journal of Manufacturing Systems 58 (2021) 59–74

Fig. 9. Comparison of the methods for problem instances of (a) FBS-DFLP-1, (b) FBS-DFLP-2, (c) FBS-DFLP-3, (d) FBS-DFLP-4, (e) P6.

but prevented a larger increase in transportation costs. The total costs obtained from the solution of the FBS-DFLP-4 problem
Fig. 9(b) shows the comparison of models E and F for the FBS-DFLP-2 instance with models E and F are shown in Fig. 9(d). In the model E, it is
problem instance. It is seen that transportation costs decrease in the case found that transportation costs are higher than the model F, similar to
of the re-arrangements in the FBS-DFLP-2 for both approaches. Both the previous problem instances. Similarly, displacement costs are lower
approaches show a similar downward trend in total cost. However, than the model F in the model E, it can be found that the effects of
model F overcomes the model E in the design of facility layout with less displacements decisions on the transportation costs are less than others.
cost. Therefore, it makes model E costlier in total costs. Finally, the total costs
In Fig. 9(c), the total costs obtained from the solution of the FBS- of the P6 during periods are depicted in Fig. 9(e). According to the re­
DFLP-3 problem instance with models E and F are shown. Trans­ sults, the costs are close to each other. Only the difference is occurred on
portation costs increase and decrease for different periods for both ap­ period 4 and 6, due to the displacement costs.
proaches when the fluctuations are on demand. Considering the total
transportation costs, model E has a higher cost than the model F. In the
model F, the simultaneous material handling device assignments and 4.5. Scenario analysis
dynamic facility layout decisions have a significant effect on the total
distance because of re- arrangements. The proposed approach for dynamic facility layout problem has
formed different facility layouts by assigning material handling devices

71
A. Erik and Y. Kuvvetli Journal of Manufacturing Systems 58 (2021) 59–74

and using the flexible bay structure together (model F). The model is directly affects the total cost.
tested under various scenarios in order to determine the sensitivity of In addition to the capacity of the material handling device,
the proposed approach to different conditions. increasing fuel and labor costs in the market conditions also increase the
In the first of these scenarios, the capacity of the material handling material transportation costs. The changes in material transportation
device capacity is changed and discussed. Table 6 shows the results in costs within the range of ± 30 % are shown in Table 7. The increase in
the case of the change in the material handling device capacity within material handling costs among departments directly increases the total
the range of ± 30 % on the facility layout. While the increase in the cost. It also causes the non-operating state in material handling devices
material handling device capacity causes an increase in the unused and reduces the number of material handling devices purchased and
material handling device, the number of material handling device used used per period. The decrease in unit transportation cost also decreases
per period has decreased. The decrease in the capacity of the material the total cost while increasing the cost of transportation and displace­
handling devices increases the number of newly purchased material ment costs. It also increases the number of purchased material handling
handling device and the total cost. The increase and decrease of material devices.
handling devices capacity change the displacement amounts and costs, Increasing the amount of material flow in Table 8 leads to an increase
thereby directly affecting the facility layout. Especially, the decrease in in the number of operating and non-operating material handling de­
the capacity causes the need for more material handling device which vices. In addition, these changes have been reflected in the displacement
costs of the departments and have allowed the facility layout differs.
Although the change in material flow between departments is similar to
Table 6 the change in unit transportation costs on total costs, there is an inverse
The scenarios of the material handling device capacity.
effect in material handling devices. This is because the increase in the
Changes of MHD Capacities number of material handling devices is directly related to the amount of
Performance Criteria
+%30 Current -%30 material to be transported.
Transportation Costs 23986 23283 23236
Material Handling Costs 3235 3050 4280 5. Conclusion
Re-arrengement Costs 2499 2512 2559
Total Costs 29720 28845 30075 Today’s competitive conditions have properties that are suitable for
Total Flows 5239 5239 5239
the customers’ demands other than the product quality. These demands
Total Re-arrangements 277 285 289
Number of operating material handling devices per 2,3 2,5 3,5 require the supply of the product when it is needed. In addition to being
period flexible and fast, this situation requires suitable pricing policies with
Number of purchased material handling devices per 0,5 0,4 0,6 competitors in order to compete in market conditions. Firms are trying
period
to make more profit while producing under all these conditions. One of
Number of non-operating material handling devices 0,2 0 0
per period
the most important ways to make more profit is to reduce costs. For this
reason, firms try to eliminate the costs arising from all transactions that
are non-value-added processes.
Table 7 Transportation costs between departments constitute one of the most
The scenarios of the unit transportation costs. important costs within the facility operation. While the transports
Changes of Unit Transportation Costs within the facility do not provide any value-added process to the
Performance Criteria product, it causes an increase in cycle time and inefficient use of the
+%30 +%15 Current -%15 -%30
facility area. One of the most important tools used to reduce trans­
Transportation Costs 30660 25183 23283 20656 17445
portation costs is to ensure having a proper facility layout. The high
Material Handling Costs 3015 3260 3050 3080 3496
Re-arrengement Costs 2564 2654 2512 2516 2574 demand fluctuation requires the use of a dynamic facility layout, which
Total Costs 36239 31096 28845 26252 23515 needs different facility layouts for certain periods. In this study, it is
Total Flows 5239 5239 5239 5239 5239 aimed to obtain minimum cost layouts as the real-life applications by
Total Re-arrangements 309 347 285 285 344 taking into account assigning flexible bay structure and material
Number of operating material 2,3 2,3 2,5 2,5 2,7
handling devices per period
handling devices in different facility layout problems.
Number of purchased material 0,5 0,7 0,4 0,4 0,6 In the study, three different facility layout problems having different
handling devices per period characteristics are considered and material handling device assignment
Number of non-operating 0,1 0 0 0 0,1 decisions are included in those problems. The first problem considers the
material handling devices per
static facility layout problem while the second one adds the flexible bay
period
structure to that model. The last model is the dynamic facility layout
problem with a flexible bay structure. This model combines the flexible
Table 8 bay structure and the use, unused and purchasing decisions of material
The scenarios of the material flows. handling devices which enables more realistic facilities layouts to be
Changes of Material Flows obtained. To test the models, eleven different problem instances based
Performance Criteria on the previous studies are used. The lack of data is obtained by
+%30 +%15 Current -%15 -%30
generating random numbers. As a result of the tests, it was observed that
Transportation Costs 26111 24485 23283 23305 22225 the decisions of the assignment of material handling devices are effec­
Material Handling Costs 4866 3100 3050 2825 2825
Re-arrangement Costs 2527 2400 2512 2500 2516
tive on the facility layout.
Total Costs 33504 29985 28845 28630 27566 Various scenarios are carried out to test the proposed model under
Total Flows 6811 6025 5239 4453 3667 different parameter values. Based on these scenarios, the decrease and
Total Re-arrangements 316 243 285 283 283 increase in the capacity of material handling devices result in an in­
Number of operating material 2,7 2,5 2,5 2,3 2,3
crease in the total cost. The decrease in the capacity of material handling
handling devices per period
Number of purchased material 1,2 0,4 0,4 0,4 0,4 devices caused the purchase, operating and non-operating of more ma­
handling devices per period terial handling devices and this increases the total transportation cost.
Number of non-operating 0,7 0 0 0,1 0,1 Capacity changes of material handling devices, which also increases the
material handling devices per cost of rearrangement, also affect the facility layout. Apart from this
period
scenario, the increase in unit transportation cost and the amount of

72
A. Erik and Y. Kuvvetli Journal of Manufacturing Systems 58 (2021) 59–74

material transported between departments also increased the total cost. [15] Turanoğlu B, Akkaya G. A new hybrid heuristic algorithm based on bacterial
foraging optimization for the dynamic facility layout problem. Expert Syst Appl
In addition, these parameters influenced the decisions related to mate­
2018;98:93–104.
rial handling devices and had an impact on the facility layout. The [16] Meller RD, Gau KY. The facility layout problem: recent and emerging trends and
decrease in transportation costs caused an increase in the number of perspectives. J Manuf Syst 1996;15(5):351–66.
material handling devices, while the increase in costs caused a decrease [17] Devise O, Pierreval H. Indicators for measuring performances of morphology and
material handling systems in flexible manufacturing systems. Int J Prod Econ 2000;
in the number of material handling devices. The increase in the amount 64(1-3):209–18.
of material flows between departments has increased the assignments of [18] Drira A, Pierreval H, Hajri-Gabouj S. Facility layout problems: a survey. Annu Rev
purchasing, operating and non-operating material handling devices. The Control 2007;31(2):255–67.
[19] Lee YH, Lee MH. A shape-based block layout approach to facility layout problems
increase in the number of non-operating devices results from demand using hybrid genetic algorithm. Comput Ind Eng 2002;42(2-4):237–48.
fluctuations between periods. [20] Anjos MF, Kennings A, Vannelli A. A semidefinite optimization approach for the
Generally, problem instances are discussed in order to show the single-row layout problem with unequal dimensions. Discret Optim 2005;2:
113–22.
proposed dynamic facility layout and material handling assignment [21] Kothari R, Ghosh D. An efficient genetic algorithm for single row facility layout.
problem in this study. Solving the problem instances with increased Optim Lett 2014;8:679–90. https://doi.org/10.1007/s11590-012-0605-2.
complexity takes much solution time. Therefore, using heuristic [22] Kothari R, Ghosh D. A scatter search algorithm for the single row facility layout
problem. J Heuristics 2014;20:125–42. https://doi.org/10.1007/s10732-013-
methods for the solution of medium and larger problems is suggested for 9234-x.
further studies. This study can be extended as the material flows are [23] Palubeckis G. Fast simulated annealing for single-row equidistant facility layout.
considered as stochastic and parameters such as carrying capacities and Appl Math Comutation 2015;263:287–301. https://doi.org/10.1016/j.
amc.2015.04.073.
working capacity ratio can be obtained as fuzzy numbers. The terms of
[24] Tambunan M, Ginting E, Sâri RM. Production facility layout by comparing moment
the objective functions can be extended with weights in order to propose displacement using BLOCPLAN and ALDEP algorithms. In IOP Conference Series:
goal-programming model. Finally, the material handling device main­ Materials Science and Engineering, Vol. 309; 2018. p. p. 012032. No. 1.
tenance, leasing etc. conditions can be considered in the models. [25] Armour GC, Buffa ES. A heuristic algorithm and simulation approach to relative
location of facilities. Manage Sci 1963;9(2):294–309.
[26] Bazaraa MS. Computerized layout design: a branch and bound approach. AIIE
Declaration of Competing Interest transactions 1975;7(4):432–8.
[27] Love R, Wong J. On solving a one-dimensional space allocation problem with
integer programming. Infor Inf Syst Oper Res 1976;14(2):139–43.
The authors declare that they have no conflict of interest. [28] Van Camp DJ, Carter MW, Vannelli A. A nonlinear optimization approach for
solving facility layout problems. Eur J Oper Res 1992;57(2):174–89.
[29] Tam KY. A simulated annealing algorithm for allocating space to manufacturing
Acknowledgements
cells. Int J Prod Res 1992;30(1):63–87.
[30] Kumar KR, Hadjinicola GC, Lin TL. A heuristic procedure for the single row facility
Adem Erik would like to extend thanks to the Scientific and Tech­ layout problem. Eur J Oper Res 1995;87:65–73.
nological Research Council of Turkey (TÜBİTAK) for supporting his Ph. [31] Meller RD. The multi-bay manufacturing facility layout problem. Int J Prod Res
1997;35(5):1229–37.
D. studies (BIDEB-2211 Programme with Grant Application [32] Gau KY, Meller RD. An iterative facility layout algorithm. Int J Prod Res 1999;37
Number:1649B031905573). (16):3739–58.
[33] Balakrishnan J, Cheng C, Wong K. FACOPT: a user-friendly FACility layout
OPTimization system. Comput Oper Res 2003;30:1625–41.
Appendix A. Supplementary data [34] Yang T, Kuo C. A hierarchical AHP/DEA methodology for the facilities layout
design problem. Eur J Oper Res 2003;147:128–36.
Supplementary material related to this article can be found, in the [35] Deb SK, Bhattacharyya B. Fuzzy decision support system for manufacturing facility
layout planning. Decis Support Syst 2005;40:305–14.
online version, at doi:https://doi.org/10.1016/j.jmsy.2020.11.015. [36] Solimanpur M, Vrat P, Sahnakar R. An ant algorithm for the single row layout
problem in flexible manufacturing systems. Comput Oper Res 2005;32:583–98.
References [37] Konak A, Kulturel-Konak S, Norman BA, Smith AE. A new mixed integer
programming formulation for facility layout design using flexible bays. Oper Res
Lett 2006;34(6):660–72.
[1] Hosseini-Nasab H, Fereidouni S, Ghomi SMTF, Fakhrzad MB. Classification of
[38] Chae J, Peters BA. Layout design of multi-bay facilities with limited bay flexibility.
facility layout problems: a review study. Int J Adv Manuf Technol 2018;94:957–77.
J Manuf Syst 2006;25(1):1–11.
https://doi.org/10.1007/s00170-017-0895-8.
[39] Azevedo MM, Crispim JA, de Sousa JP. A dynamic multi-objective approach for the
[2] Islier AA. A genetic algorithm approach for multiple criteria facility layout design.
reconfigurable multi-facility layout problem. J Manuf Syst 2017;42:140–52.
Int J Prod Res 1998;36(6):1549–69.
[40] Guan C, Zhang Z, Liu S, Gong J. Multi-objective particle swarm optimization for
[3] Durmaz ED, Şahin R. Çok amaçlı tek sıra tesis düzenleme probleminin çözümü için
multi-workshop facility layout problem. J Manuf Syst 2019;53:32–48.
NSGA-II ve hedef programlama yaklaşımı. J Faculty Eng Architect Gazi Univ 2017;
[41] Tari FG, Neghabi H. A new linear adjacency approach for facility layout problem
32(3):941–55.
with unequal area departments. J Manuf Syst 2015;37:93–103.
[4] Tate DM, Smith AE. A genetic approach to the quadratic assignment problem.
[42] Meller RD, Bozer YA. Alternative approaches to solve the multi-floor facility layout
Comput Oper Res 1995;22(1):73–83.
problem. J Manuf Syst 1997;16(3):192–203.
[5] Sahni S, Gonzalez T. P-complete approximation problems. J ACM (JACM) 1976;23
[43] Amaral ARS. On the exact solution of a facility layout problem. Eur J Oper Res
(3):555–65.
2006;173:508–18.
[6] Tong X. SECOT: a sequential construction technique for facility design. Thesis
[44] Liu Q, Meller RD. A sequence-pair representation and MIP-model-based heuristic
(PhD). University of Pittsburgh; 1991.
for the facility layout problem with rectangular departments. IIE Trans 2007;39(4):
[7] Eklund NH, Embrechts MJ, Goetschalckx M. An efficient chromosome encoding
377–94.
and problem-specific mutation methods for the flexible bay facility layout problem.
[45] Anjos MF, Yen G. Provably near-optimal solutions for very large single-row facility
Proceedings of the 2003 IEEE International Workshop on Soft Computing in
layout problems. Optim Methods Softw 2009;Vol. 24(Nos. 4–5):805–17.
Industrial Applications 2003. 2003. SMCia/03. (pp. 109-113). IEEE.
[46] Amaral ARS. A new lower bound for the single row facility layout problem. Discret
[8] Wong KY. Solving facility layout problems using Flexible Bay Structure
Appl Math 2009;157:183–90.
representation and Ant System algorithm. Expert Syst Appl 2010;37(7):5523–7.
[47] Komarudin, Wong KY. Applying ant system for solving unequal area facility layout
[9] McKendall AR, Shang J. Hybrid ant systems for the dynamic facility layout
problems. Eur J Oper Res 2010;202(3):730–46.
problem. Comput Oper Res 2006;33:790–803.
[48] Kulturel-Konak S, Konak A. Unequal area flexible bay facility layout using ant
[10] Şahin R. Dinamik Tesis Düzenleme Problemi İçin Bir Tavlama Benzetimi Sezgiseli,
colony optimisation. Int J Prod Res 2011;49(7):1877–902.
Gazi Üniv. Müh Mim Fak Der Cilt 2008;23(No 4):863–70.
[49] Amaral ARS, Letchford AN. A polyhedral approach to the single row facility layout
[11] Mazinani M, Abedzadeh M, Mohebali N. Dynamic facility layout problem based on
problem. Math Program Ser A 2013;141:453–77. https://doi.org/10.1007/s10107-
flexible bay structure and solving by genetic algorithm. Int J Adv Manuf Technol
012-0533-z.
2013;65(5–8):929–43.
[50] Hungerländer P, Rendl F. A computational study and survey of methods for the
[12] Ulutas B, İslier AA. Dynamic facility layout problem in footwear industry. J Manuf
single-row facility layout problem. Comput Optim Appl 2013;55:1–20. https://doi.
Syst 2015;36:55–61.
org/10.1007/s10589-012-9505-8.
[13] Vitayasak S, Pongcharoen P, Hicks C. A tool for solving stochastic dynamic facility
[51] Lenin N, Kumar MS, Islam MN, Ravindran D. Multi-objective optimization in
layout problems with stochastic demand using either a Genetic Algorithm or
single-row layout design using a genetic algorithm. Int J Adv Manuf Technol 2013;
modified backtracking search Algorithm. Int J Prod Econ 2017;190:146–57.
67:1777–90. https://doi.org/10.1007/s00170-012-4608-z.
[14] Kulturel-Konak S. A matheuristic approach for solving the dynamic facility layout
problem. Procedia Comput Sci 2017;108:1374–83.

73
A. Erik and Y. Kuvvetli Journal of Manufacturing Systems 58 (2021) 59–74

[52] Helber S, Böhme D, Oucherif F, Lagershausen S, Kasper S. A hierarchical facility [59] Yang T, Peters BA. Flexible machine layout design for dynamic and uncertain
layout planning approach for large and complex hospitals. Flex Serv Manuf J 2016; production environments. Eur J Oper Res 1998;108(1):49–64.
28:5–29. https://doi.org/10.1007/s10696-015-9214-6. [60] Ertay T, Ruan D, Tuzkaya UR. Integrating data envelopment analysis and analytic
[53] Asl AD, Wong KY. Solving unequal-area static and dynamic facility layout hierarchy for the facility layout design in manufacturing systems. Inf Sci 2006;176:
problems using modified particle swarm optimization. J Intell Manuf 2017;28(6): 237–62.
1317–36. MAK 25. [61] Kulturel-Konak S, Konak A. A large-scale hybrid simulated annealing algorithm for
[54] Wang S, Zuo X, Liu X, Zhao X, Li J. Solving dynamic double row layout problem via cyclic facility layout problems. Eng Optim 2015;47(7):963–78. MAK. 20.
combining simulated annealing and mathematical programming. Appl Soft [62] Hunagund IB, Pillai VM, Kempaiah UN. A simulated annealing algorithm for
Comput 2015;37:303–10. unequal area dynamic facility layout problems with flexible bay structure. Int J Ind
[55] Pourvaziri H, Pierreval H. Dynamic facility layout problem based on open queuing Eng Comput 2018;9:307–30.
network theory. Eur J Oper Res 2017;259(2):538–53. [63] Moslemipour G. A hybrid CS-SA intelligent approach to solve uncertain dynamic
[56] Kheirkhah A, Navidi H, Bidgoli MM. Dynamic facility layout problem: a New facility layout problems considering dependency of demands. J Ind Eng Int 2018;
bilevel formulation and some metaheuristic solution methods. IEEE Transactıons 14:429–42.
on Engıneerıng Management, Vol. 62; 2015. No. 3. [64] Scholz D, Petrick A, Domschke W. STaTS: a slicing tree and tabu search based
[57] Ulutas BH, Islier AA. A clonal selection algorithm for dynamic facility layout heuristic for the unequal area facility layout problem. Eur J Oper Res 2009;197(1):
problems. J Manuf Syst 2009;28(4):123–31. 166–78.
[58] Kheirkhah A, Bidgoli MM. Dynamic facility layout problem under competitive
environment: a new formulation and some meta-heuristic solution methods. Prod
Eng Res Devel 2016;10:615–32. https://doi.org/10.1007/s11740-016-0703-6.

74