European Journal of Operational Research 164 (2005) 1–11

www.elsevier.com/locate/dsw

Invited Review

Loop based facility planning and material handling

Ardavan Asef-Vaziri a, Gilbert Laporte b,*

a
Systems and Operations Management, College of Business and Economics, California State University, Northridge,
18111 Nordhoff Street, Northridge, CA 91330–8245, USA
b
Canada Research Chair in Distribution Management and GERAD, HEC Montr
eal, 3000, Chemin de la C^ote-Sainte-Catherine,
Montr
eal, Quebec, Canada H3T 2A7
Received 11 March 2003; accepted 8 January 2004
Available online 25 March 2004

Abstract

A sizeable proportion of manufacturing expenses can be attributed to facility layout and material handling. Facility
layout decisions involve designing the arrangement of elements in manufacturing systems. Among the most critical
material handling decisions in this area are the arrangement and design of material flow patterns. This survey article
reviews loop based facility planning and material handling decisions for trip based material handling equipment with an
emphasis on unit load automated guided vehicles. The article examines issues related with facility design, material
handling design, and fleet sizing and operating.
Ó 2004 Elsevier B.V. All rights reserved.

Keywords: Layout; Facility planning; Material handling; Automated guided vehicles; Flexible manufacturing

1. Introduction

Facility layout decisions involve designing the arrangement of elements in manufacturing systems.
Among the most critical decisions in this area are the arrangement of departments, production units,
manufacturing cells, centers, storage areas, etc. As pointed out by Apple (1977) it is crucial to incorporate
material handling decisions in layout design. Tompkins et al. (1996) estimate that between 20% and 50% of
operating expenses in manufacturing can be attributed to facility planning and material handling. Thus any
cost saving in this area can contribute to the overall efficiency of the production system.
There exists an extensive literature on such problems (see, e.g., Francis et al., 1992; Ganesharajah et al.,
1998). In this article we review the layout and material handling decisions in the context of loop based
material flow systems. The loop layout is one of the four well known general types of design used in
production systems (Apple, 1977). It lends itself to both product and production simplicity (Afentakis,
1989). In a basic layout design, each cell is represented by a rectilinear, but not necessarily convex polygon.

*
Corresponding author. Tel.: +1-514-343-6143; fax: +1-514-343-7121.
E-mail addresses: ardavan.asef-vaziri@csun.edu (A. Asef-Vaziri); gilbert.laporte@hec.ca, gilbert@crt.umontreal.ca (G. Laporte).

0377-2217/$ - see front matter Ó 2004 Elsevier B.V. All rights reserved.
doi:10.1016/j.ejor.2004.01.037
2 A. Asef-Vaziri, G. Laporte / European Journal of Operational Research 164 (2005) 1–11

The set of fully packed adjacent polygons is known as a block layout. A block layout is usually integrated
with a material flow system which is defined in terms of material handling equipment, configuration and
directions of the material handling network, and the number and locations of pickup (P) and delivery (D)
stations (Tanchoco and Sinriech, 1992). Here we mainly discuss trip based (Bozer et al., 1994) material
handling equipment with an emphasis on unit load automated guided vehicles (AGVs). AGVs are preferred
to stationary material handling robots for their mobility and to conveyors for their flexibility.
In Fig. 1, four common material handling networks are superimposed on a block layout. Fig. 1a cor-
responds to the conventional configuration (Maxwell and Muckstadt, 1982). It is a unidirectional network
covering all the edges of the block layout. To understand the underlying layout of Fig. 1a and b one must
ignore the arrowheads. We symbolically refer to each of the polygons of the block layout as a manufac-
turing cell. A conventional configuration may result in substantial vehicle blocking and is more difficult to
operate and control. A complex and expensive software is required for the operation of such systems
(Sinriech and Tanchoco, 1992a,b). The operational issues can be classified as: (a) Dispatching: when a
station is calling for a vehicle, determine which of the several idle vehicles should serve it. When several
stations are calling, decide where to assign the single idle vehicle. (b) Vehicle routing: determine the best
origin-destination route. (c) Traffic management: if more than one vehicle arrive at an intersection, select
the vehicle that has right of way. The unidirectional streamlined loop network of Fig. 1b is often preferred
for its simplicity (Tanchoco and Sinriech, 1992). It is a loop covering at least one edge of each manufac-
turing cell. Another example of a simplified system is the tandem configuration (TC) (Fig. 1c) which
decomposes the flow pattern into several bidirectional loops each served by a single vehicle. Two loops are
connected to each other either by a transfer station or by a small conveyor section (Bozer and Srinivasan,
1989). This eliminates blocking at the expense of transshipment between the segments. Note that in Fig. 1c
since the loops are close to each others, there is no need for a conveyor but only for a transfer station. If the
loops are disjoint, a piece of conveyor is installed to connect them. Finally, a segmented loop topology is
depicted in Fig. 1d (Sinriech et al., 1996). It is a bidirectional loop covering all cells and partitioned into
nonoverlapping segments. Each segment is assigned to a single vehicle or to some stationary material
handling equipment. The segmented flow pattern is not necessarily connected.
This survey concentrates on loop based configurations, including segmented loop topology and tandem
configurations. Readers interested in conventional configurations are referred to Sun and Tchernev (1996)
for a comprehensive review. In all the models where the layout of the departments is known, with the
exception of the tandem configuration, the AGVs are not supposed to cross the cells and work centers. The
material handling aisles are constructed on the edges separating the manufacturing cells.

(a) (b)

(c) (d)

Fig. 1. Four basic material flow configurations: (a) conventional unidirectional network, (b) unidirectional loop network, (c) tandem
configuration, (d) segmented loop topology.
 A. Asef-Vaziri, G. Laporte / European Journal of Operational Research 164 (2005) 1–11 3

The remainder of this article is organized as follows. In the following section, we examine two design
problems: the station sequencing problem and the more general cell location problem. Section 3 addresses
material flow design problems per se: tandem designs and single loop designs. The problem of fleet sizing
and operating is covered in Section 4. A conclusion follows in Section 5.

2. Facility design

2.1. Sequencing stations

An important problem when designing loop transportation systems is to determine the relative order of
stations on an unidirectional loop, given a predicted from-to chart showing the material flow between
stations, in order to minimize transportation cost. More precisely, all material unit loads are assumed to
enter and leave the system through a specialized station 0. The objective is to minimize the total material
flow through station 0, measured by the total number of unit loads.
Note that in this problem, the size, shape and orientation of cells are not considered. Each cell is simply
represented by a station (node). The station sequencing problem (SSP) was introduced by Afentakis (1989)
and was proved to be NP-hard by Kouvelis and Kim (1992). Kiran and Karabati (1993) have formulated
the SSP as a special case of the quadratic assignment problem (QAP), while Tansel and Bilen (1998) have
proposed an interchange postoptimization heuristic. In practice, however, only relatively small instances of
the QAP can be solved optimally. Known optimal solutions for instances of size exceeding 20 are rare
although significant advances have been made recently (Anstreicher et al., 2002). The QAP is usually
tackled by means of heuristics (see, e.g., Burkard and Bonninger, 1983). For recent comprehensive refer-
ences on the QAP, see Burkard et al. (1998) and Rendl (2002).
Afentakis (1989) provides a direct formulation to the SSP but solves the problem heuristically by means
of an interchange procedure. Dominance relations are developed by Kouvelis and Kim (1992). These help
reduce the solution space and enable the authors to develop quick heuristics based on a decomposition
procedure, and a branch-and-bound algorithm capable of solving instances of size 12. Kiran and Karabati
(1993) have independently proposed dominance relations. In addition, they have identified a polynomially
solvable case of the SSP and developed approximate solution procedures as well as an exact enumerative
algorithm. The largest instance size attained by these authors is also 12. Using graph theory arguments,
Leung (1992) develops a heuristic to sequence the stations starting with the solution obtained by means of
the LP-relaxation of the problem.

2.2. Locating cells

While the SSP ignores cell characteristics, the cell location problem (CLP) takes physical considerations
into account. The problem is to determine the location, dimension and orientation of n cells so that they can
be feasibly served by a loop configuration system under a transport cost minimization objective. If, as is the
case in unrestricted shaped buildings, all cells can be equally spaced around (outside) a loop, then the CLP
reduces to the SSP (see Fig. 2). In a regular shaped building where all stations cannot be equally spaced, the
transport cost is not merely measured by the frequency of parts passing through station 0, but by the total
distance traveled by all parts.
The CLP is in general more difficult as cells can lie inside the loop as well as outside (Fig. 3). Several
variants of the problem have been studied, not all of which are relevant to this article since their flow
pattern is not loop based (see Kusiak and Heragu, 1987; Francis et al., 1992; Meller and Gau, 1996). A
fundamental article in this area is due to Montreuil (1991), where the author proposes mixed integer linear
programming formulations for integrated layout and flow network design. Cell shapes are assumed to be
4 A. Asef-Vaziri, G. Laporte / European Journal of Operational Research 164 (2005) 1–11

Fig. 2. Seven unequal rectangular cells positioned around a loop. The bold line is the loop material handling system. Black circles
correspond to stations.

Fig. 3. General block layout with a loop material handling system. Black circles correspond to stations.

rectangular, but their dimensions are decision variables. Eight variants of the problem including rectilinear
and aisle flows are considered. The proposed models are more comprehensive and versatile than the tra-
ditional QAP approach but, as expected, they enable the exact solution of only very small instances.
Defining inter-cell distances as the rectilinear distance between cell centroids, Tompkins et al. (1996) have
solved an eight cell instance optimally using one of the Montreuil formulations. Genetic heuristics capable
of solving larger instances were proposed by Das (1993) and Rajasekharan et al. (1998).
Goetschalckx and Palliyil (1994) solve a special case of the general problem formulated by Montreuil in
which the block layout is given. They develop a mixed integer linear programming formulation in order to
locate a flow network, with stations, covering all cell edges. No a priori configuration is imposed on the
network, and bi-directional flows, multiple-stations as well as combined P/D stations are allowed. The
objective function contains fixed construction costs as well as the cost of moving full and empty vehicles.
Arc and path formulations of the model are examined.
Banerjee and Zhou (1995) have adapted one of Montreuil’s formulations to a loop based layout with one
combined station per cell, where the objective function is the minimization of the total loaded trip distances.
They have developed a procedure to identify a promising sequence of stations to be served along a undi-
rectional loop. However, they formulate their SSP subproblem as a traveling salesman problem (TSP)
which is inadequate because the TSP does not capture inter-cell flows. Finally, Chae and Peters (2000) have
developed a simulated annealing heuristic for a CLP in which cells are rectangular, P/D stations are
combined and located on a cell boundary edges.

3. Material handling system design

3.1. Tandem configuration design

Given a fixed block layout, a from-to chart, and combined P/D locations, the tandem configuration
problem (TCP) consists of partitioning the flow into a set of independent bidirectional loops, each served by
a single vehicle. The location of tandem loops is not restricted to the cell boundary edges. A grid is
superimposed on the block layout and the tandem loops are defined on that grid.
 A. Asef-Vaziri, G. Laporte / European Journal of Operational Research 164 (2005) 1–11 5

Tandem configurations were conceptualized by Bozer and Srinivasan (1989). Their 1991 article describes a
heuristic based on the generation of a set of promising loops and the selection of an optimal subset by means
of a set partitioning algorithm. Mahadevan and Narendran (1992) define a concept referred to as single vehicle
loop layout which is a design similar to tandem configuration. They address the key issues in design and
operations of such a configuration in an FMS. Bozer and Srinivasan (1992) compare the performance of
tandem and conventional configurations and show that as the required number of AGVs increases, tandem
performs better than conventional. Occena and Yakota (1991, 1993) examine operational policies in a tandem
configuration and conclude that the tandem configuration could be easily implemented in an FMS. Choi et al.
(1994) have developed a simulation model to compare single loop and tandem configurations. The single loop
produces a shorter travel time and idle time, but tandem configurations result in a higher job completion rate.
Lin et al. (1994) have measured the impact of load routing on the performance of a tandem configuration.
Using simulations, Bischak and Stevens (1995) have shown that the tandem system has a higher expected
travel time per load than a conventional configuration. This is due to the extra time required for transshipment
across loops. Ross et al. (1996) have compared the conventional configuration, the tandem configuration, and
a tandem configuration enhanced with an internal loop. The key performance measures were utilization, mean
flow time, tardiness, and percent tardy. They conclude that tandem configuration matches the performance of
conventional configuration, while it provides ease of control. Huang (1997) has developed a two stage
optimization model for rehandling reduction in a tandem configuration. The main idea is to minimize the total
number of transfer points, and to assign the loop to loop transfers to AGVs not operating in the tandem loops.
In the first stage, an integer programming model designs a single transfer point for each tandem loop. In the
second stage, a network to connect the transfer points is designed using a minimum spanning tree.

3.2. Single loop design

While the single loop configuration has been used for a long time for fixed conveyor systems (see, e.g.,
Muth, 1972, 1974, 1975), its application to AGV systems is more recent (Tanchoco and Sinriech, 1992).
This article must be regarded as seminal since it has laid the ground for much of the further research in the
area. In it the authors address the design of a loop configuration based on an existing block layout. Two
objectives are considered: the minimization of the loop length, and the minimization of the time (or dis-
tance) required to carry out all material flow movements between cells. The loop configuration problem
(LCP) is to locate a shortest loop covering at least one edge from each cell. The NP-hardness of the problem
was proved by De Guzman et al. (1997). The problem was formulated by Tanchoco and Sinriech (1992) as
an integer linear program including node degree constraints and subtour elimination constraints, as in the
TSP. The loop flow distance problem (LFDP) was then formulated to minimize the total distance traveled
by the loaded vehicles. An enumerative procedure based on the elimination of dominated solutions is
developed and illustrated. Sinriech and Tanchoco (1993) develop a heuristic to find a starting loop covering
at least one edge of each cell and an improved LFDP solution. Sinriech and Tanchoco (1992a) show how to
locate P/D stations on a fixed single loop while taking into account inter-cell and intra-cell movements.
In a different study, Sinriech and Tanchoco (1992b) have conducted a simulation experiment which con-
cludes that the impact of empty vehicle trips on the performance of AGVs operating on a single loop is neg-
ligible. This was done by comparing a loop configuration to a conventional configuration. The authors show
that fleet size in a loop system is rather robust and does not depend on dispatching rules for empty vehicles.
Laporte et al. (1996) have shown that the LCP can be formulated as a generalized traveling salesman
problem (GTSP). These authors later provided an integer linear programming formulation for the problem
(Asef-Vaziri et al., 2000). Using dominance procedures and various simplification rules, the authors show
that most of the subtour elimination constraints of the problem can be disregarded. Instances involving up
to 40 cells are solved optimally within only a few seconds, using CPLEX. Asef-Vaziri et al. (2001) have
developed an alternative formulation to that of Sinriech and Tanchoco (1993). It is used to simultaneously
6 A. Asef-Vaziri, G. Laporte / European Journal of Operational Research 164 (2005) 1–11

determine the design of a unidirectional loop and the location of P/D stations, with the objective of
minimizing the total loaded vehicle trip distance. The problem is tackled through a tight formulation of the
linear programming subproblem, preprocessing rules, and suitable solution routines.
The objective function in the models developed by Tanchoco and Sinriech (1992), Sinriech and Tan-
choco (1993), and Asef-Vaziri et al. (2001) is to minimize the total loaded vehicle trip distance. Asef-Vaziri
et al. (2003) incorporate both loaded and empty vehicle trips in their formulation. The original objective
function is non-linear. The total loaded and empty vehicle trip distance is defined as the maximum flow
passing through an arc of the loop, multiplied by the length of the loop. By ignoring the flow and mini-
mizing the length of the loop as is the case in Asef-Vaziri et al. (2000), and then minimizing the maximum
flow passing through the shortest loop (length based simplification), they compute an upper bound on the
total loaded and empty flow trip distance. By ignoring the length of the loop, minimizing the maximum flow
passing through an arc of the loop as in the case in the Afentakis (1989) and Kouvelis and Kim (1992)
formulations, and then minimizing the length of a loop with arc capacity set at this maximum value (flow
based simplification), they compute another upper bound. They also introduce a set of constraints to
linearize the objective function and find an optimal solution for total loaded and empty trip distance. Using
around 600 test problems of layouts of sizes 12, 16, and 20 cells, and from-to charts containing 24, 26, 40,
48, and 64 flow values they show that (a) the loaded vehicle trip distance objective function results in a
design far from optimal for both loaded and empty trip distances, and (b) the length based simplification
outperforms the flow based simplification both with respect to solution quality and computing time. One
important conclusion of this study is that models that ignore empty vehicle trips, such as the one developed
by Tanchoco and Sinriech (1992), Sinriech and Tanchoco (1993), and Asef-Vaziri et al. (2001) can yield
suboptimal solutions containing many empty vehicle trips.
Sinriech and Tanchoco (1995) have developed a model for a general version of the segmented flow
topology. Sinriech et al. (1996) have later developed a model for the loop-based segmented flow topology
(SFT). It first determines a bidirectional single loop using the Sinriech and Tanchoco (1993) formulation
and then partitions it by using a heuristic and then a simulation-based improvement procedure. The pri-
mary goal is to minimize the segment-to-segment transfers, while the secondary goal is to balance the
workload over all segments. The authors have conducted a simulation study which reveals that neither the
loop based segmented flow topology nor the tandem configuration dominates the other.
Regarding the setting and assumptions of TC and SFT, the flow path in SFT is restricted to the edges of
the block layout, this restriction is relaxed in TC. In TC there is a single and combined P/D station per cell,
in SFT there could be multiple stations and separate P and D stations. The location of the combined P/D
stations are already fixed in TC, while SFT allows the model to locate P and D stations on any edge
intersection on the boundary of a cell.
Sharp and Liu (1990) have developed a model to add a short-cut and one or more spurs to a unidi-
rectional loop flow path. Station locations and the flow path are fixed. They first developed a mixed integer
programming model to reduce the search space and identify a set of promising alternatives. The objective
function of the model is to minimize the total spur and shortcut construction costs, as well as vehicle travel,
waiting, and blocking costs. A simulation model is then implemented to examine the promising alternatives
and find the trade-off between construction cost and vehicle operating cost.

4. Fleet sizing and operating

4.1. Home location of idle vehicles

Among AGV control issues is the determination of the home or parking locations of idle vehicles. The
objective of the problem is to find home locations in order to minimize a measure of the response time. The
 A. Asef-Vaziri, G. Laporte / European Journal of Operational Research 164 (2005) 1–11 7

basic rules that have been introduced to route the vehicles after completion of a task and having no new
task are: (a) staying at the current D station of the completed task; (b) rotating along the loop based flow
pattern and looking for a new job at the consecutive D stations; (c) routing to a central home location; (d)
routing to one of the decentralized home locations. These models are rooted in p-median and p-center
models. The objective functions discussed are those first defined by Hakimi (1964) as minimization of either
the mean response time or the maximum response time. A third objective function of even distribution of
loads among the vehicles is also discussed. The response time is the empty travel time from the home
location to the calling P station.
Bartholdi and Platzman (1989) have proposed a simple first encountered, first served (FEFS) dispatching
rule for AGVs operating in a decentrally controlled loop flow system. The worst case of their simple
decentralized rule is compared with a complex centralized dispatching rule system. Bozer and Srinivasan
(1991) implement FEFS in a tandem configuration and report promising results. Egbelu (1993) develops a
procedures to identify the optimal home location on a loop flow path. The main idea is to transform any
loop into a circle. Models and solution procedures are developed for single and multiple vehicles operating
on unidirectional and bidirectional single loops with the aim of minimizing the mean response time. Kim
(1995) shows that the optimization models for (a) the central home location for the existing fleet of the
vehicles and (b) the new home location for each new vehicle added to the existing fleet can be solved in
polynomial time. The objective function is similar to that used by Egbelu (1993). Chang and Egbelu (1996)
have extended the work of Egbelu (1993) to allow for changes in material flow over time. The intuition is to
show that, as vehicles complete some tasks over a working period, the optimal home location may change
over time and a new location could become the optimal central home location. They develop exact algo-
rithm to solve this problem. The loop is served by a single vehicle and the objective is to minimize the mean
response time. A discrete stationary Markov chain model was proposed by Kim and Kim (1997) to estimate
mean response time, utilization, and cycle time of AGVs on an unidirectional loop. These authors have also
devised a procedure for the determination of the home location of idle vehicles to minimize mean response
time. Rajotia et al. (1998) have developed a heuristic for the estimation of AGV fleet size in a flexible
manufacturing environment. Their prototype example is a tandem configuration composed of three
interrelated loops. Fleet size is determined by first solving a mixed integer linear program that minimizes the
number of empty trips, and then making adjustments to account for estimated waiting time and blocking.
The solution is validated through simulation. The two objectives considered are mean and maximum re-
sponse times. Gademann and Van de Velde (2000) show that the problem of determining the home posi-
tions under any arbitrary cost function can be solved in a polynomial time for any number of vehicles
performing on a unidirectional loop. Ventura and Lee (2000) have conducted a simulation study to analyze
the performance of AGVs in four configurations, under five vehicle location policies. Ventura and Lee
(2001) propose a modified version of the tandem configuration with more than one vehicle per loop. Lee
and Ventura (2001) and Ventura and Lee (2002) have developed a dynamic programming model to
determine optimal empty vehicle locations on a single loop.

4.2. Blocking and collision avoidance

Given a dual track bidirectional loop formed by two single track unidirectional loops, one in each
direction, and a set of interconnections to switch from one track to the other, there are several control
problems to be addressed: (a) dispatching: assigning tasks to vehicles; (b) routing: determining an origin/
destination route for the assigned task; (c) scheduling: determining the timing of the arrival and departure
of each vehicle in each zone.
Langevin et al. (1996) have developed an integrated model for these three operational problems on a
double track bidirectional loop. The material handling network is defined as a set of P, D, and switching
nodes, and the set of the edges connecting them. Conflict is defined as occupancy or claim of a node or of an
8 A. Asef-Vaziri, G. Laporte / European Journal of Operational Research 164 (2005) 1–11

edge by more than one vehicle. The objective function is to minimize the completion time of all tasks by a
given fleet size of vehicles. The authors use dynamic programming to obtain the optimal solution for two
vehicles and propose heuristics for more vehicles. Desaulniers et al. (2003) have developed an exact method
for the same problem. They use a heuristic to obtain an initial solution, a column generation, and a branch
and bound solution procedure to solve the problem optimally for up to four vehicles. Given a unidirec-
tional loop and a set of fixed stations on it, Ho (2000) uses simulated annealing to develop a dynamic zone
strategy to partition the loop into a loop based segmented flow topology. A pair of zone adjustment and
zone assignment procedures are also developed. The zone adjustment procedure redefines the zones as the
vehicles complete some tasks over a working period. The zone assignment procedure allows the vehicles in
adjacent zones to help each others.

4.3. Multiload AGV on a single loop

While unit load AGVs still represent the single largest segment of the AGV markets, some authors have
examined the case of multiple load AGVs operating on a single loop. Sinriech and Palni (1998) propose an
exact formulation for scheduling of multiple load vehicles over a finite planning horizon. Two heuristics are
developed to handle larger problems. The performance of these heuristics which require perfect knowledge
on a certain number of future periods are compared with the simple first encountered, first served dis-
patching policy. Liu and Hung (2001) consider a single AGV with two load capacity operating on a
bidirectional loop. They develop rule based control policies to avoid deadlock of the vehicle in front of
finite queue capacity of the P/D stations. The essence of rule based heuristics is to avoid circular wait states
caused by inappropriate flow of jobs. The procedure does not always yield an optimal solution and needs
full real time information on the shop floor. Sinriech and Kotlarski (2002) describe a scheduling algorithm
for the operation of a fleet of vehicles carrying more than one load along a single loop. A state based
scheduling algorithm is developed. The impact of increasing the vehicle capacity is evaluated based on
variables such as cycle time, work in progress, and utilization.

4.4. Other operational issues

A number of other operational problems have been analyzed in a loop context. Malmborg (1994) has
proposed an analytical method for predicting the work in progress storage requirements for a fleet of AGVs
serving a loop layout. A pairwise exchange greedy heuristic and a simulated annealing version of it were
developed to perform sequential search for the minimization of work in progress storage minimizing line
layouts over alternative AGV fleet sizes. Blazewicz et al. (1994) have studied the problem of combining
production scheduling with AGV scheduling. Their work, motivated by a real application, examines the
case where two AGV loops are interconnected. Hall et al. (2001a) have considered combined job scheduling
and material handling in a unidirectional loop: AGVs serve a set of cells each with a combined P/D station
located on the loop, and the objective is the minimization of the steady state cycle time required to produce
a minimal job set. Three load assignment dispatching rules were proposed. The authors have developed a
genetic algorithm and tested it in a layout composed of ten manufacturing cells served by four vehicles. Hall
et al. (2001b) have studied the fleet sizing problem. Given the minimum steady state cycle time required to
produce a minimal job set, the objective of this model is to minimize the AGV fleet size.

5. Conclusion

Loop based material flow systems are commonly used in industry and have been proved by several
researchers to be highly effective both from a cost and from an operational point of view. Research in this
 A. Asef-Vaziri, G. Laporte / European Journal of Operational Research 164 (2005) 1–11 9

area is rich and has really taken off since the work of Tanchoco and Sinriech in the early 1990s. Because
each manufacturing context is in a sense unique and technologies also evolve, we anticipate a continued
interest in the study of such problems. This review paper has highlighted some of the main contributions in
the field of loop configuration systems. We hope it will inspire researchers and stimulate the study and
understanding of new problems.

Acknowledgements

This work was partly supported by a summer fellowship from the Cullen College of Engineering at the
University of Houston, and by the Canadian Natural Sciences and Engineering Research Council under
grant OGP0039682. This support is gratefully acknowledged. This article was partly written while the
second author visited the Centre for Traffic and Transport at the Danish University of Technology. Thanks
are due to the referees for their valuable comments.

References

Afentakis, P., 1989. A loop layout design problem for flexible manufacturing systems. International Journal of Flexible Manufacturing
Systems 1, 175–196.
Anstreicher, K.M., Brixius, N., Linderoth, J., Goux, J.-P., 2002. Solving large quadratic assignment problems on computational grids.
Mathematical Programming Series B 91, 563–588.
Apple, J.M., 1977. Plant Layout and Material Handling. Wiley, New York.
Asef-Vaziri, A., Hall, N.G., George, R., 2003. The importance of empty vehicle travel in circular flow path design, Working Paper,
Department of Systems and Operations Management, College of Business and Economics, California State University, Northridge,
CA.
Asef-Vaziri, A., Dessouky, M., Sriskandarajah, C., 2001. A loop material flow system design for automated guided vehicles.
International Journal of Flexible Manufacturing Systems 13, 33–48.
Asef-Vaziri, A., Laporte, G., Sriskandarajah, C., 2000. The block layout shortest loop problem. IIE Transactions 32, 727–734.
Banerjee, P., Zhou, Y., 1995. Facilities layout design optimization with single loop material flow path configuration. International
Journal of Production Research 33, 183–203.
Bartholdi III, J.J., Platzman, L.K., 1989. Decentralized control of automatic guided vehicles on a simple loop. IIE Transactions 21, 76–
81.
Bischak, D.P., Stevens Jr., K.B., 1995. An evaluation of tandem configuration automated guided vehicle system. Production Planning
& Control 6, 438–444.
Blazewicz, J., Burkard, R.F., Finke, G., Woeginger, G.J., 1994. Vehicle scheduling in two-cycle flexible manufacturing systems.
Mathematical and Computer Modelling 20, 19–31.
Bozer, Y.A., Srinivasan, M.M., 1989. Tandem configurations for automated guided vehicle systems offer simplicity and flexibility.
Industrial Engineering 21, 23–27.
Bozer, Y.A., Srinivasan, M.M., 1991. Tandem configurations for automated guided vehicle systems and the analysis of single vehicle
loops. IIE Transactions 23, 72–82.
Bozer, Y.A., Srinivasan, M.M., 1992. Tandem AGV systems: A partitioning algorithm and performance comparison with
conventional AGV systems. European Journal of Operational Research 63, 173–191.
Bozer, Y.A., Srinivasan, M.M., Cho, M., 1994. Trip-based material handling systems: Throughout capacity analysis. IIE Transactions
26, 70–89.
Burkard, R.E., Bonninger, T, 1983. A heuristic for quadratic Boolean program with applications to quadratic assignment problems.
European Journal of Operational Research 13, 374–386.
Burkard, R.E., C ß ela, E., Pardalos, P.M., Pitsoulis, L.S., 1998. The quadratic assignment problem. In: Du, D.-Z., Pardalos, P.M. (Eds.),
Handbook of Combinatorial Optimization. Kluwer, Boston, pp. 241–337.
Chae, J., Peters, B.A., 2000. A simulated annealing algorithm based on the closed loop layout (SA-CL) for facility layout design in
flexible manufacturing systems, Working Paper, Department of Industrial Engineering, Texas A & M University.
Chang, S.H., Egbelu, P.J., 1996. Dynamic positioning of AGVs in a loop layout to minimize mean system response time. International
Journal of Production Research 34, 1655–1674.
Choi, H., Kwon, H., Lee, J., 1994. Traditional and tandem AGVs layouts––A simulation study. Simulation 63, 85–93.
10 A. Asef-Vaziri, G. Laporte / European Journal of Operational Research 164 (2005) 1–11

Das, S., 1993. A facility layout method for flexible manufacturing systems. International Journal of Production Systems 31, 279–297.
De Guzman, M.C., Prabhu, N., Tanchoco, J.M.A., 1997. Complexity of the AGV shortest path and single-loop guide path layout
problems. International Journal of Production Research 35, 2083–2092.
Desaulniers, G., Langevin, A., Riopel, D., Villeneuve, B., 2003. Dispatching and conflict-free routing of automated guided vehicles.
International Journal of Flexible Manufacturing Systems (forthcoming).
Egbelu, P.J., 1993. Positioning of automated guided vehicles in a loop layout to improve response time. European Journal of
Operational Research 71, 32–44.
Francis, R.L., McGinnis Jr., L.F., White, J.A., 1992. Facility Layout and Location: An Analytical Approach. Prentice-Hall,
Englewood Cliffs, NJ.
Gademann, A.J.R.M., Van de Velde, S.L., 2000. Positioning automated guided vehicles in a loop layout. European Journal of
Operational Research 127, 565–573.
Ganesharajah, T., Hall, N.G., Sriskandarajah, C., 1998. Design and operational issues in AGV-served manufacturing systems. Annals
of Operations Research 76, 109–154.
Goetschalckx, M., Palliyil, G., 1994. A comprehensive model for the concurrent determination of aisle-based material handling
systems. In: Graves, R. et al. (Eds.), Developments in Material Handling Research. Material Handling Institute of America,
Charlotte, NC, pp. 161–188.
Hall, N.G., Sriskandarajah, C., Ganesharajah, T., 2001a. Operational decisions in AGV-served flowshop loops: Scheduling. Annals of
Operations Research 107, 161–188.
Hall, N.G., Sriskandarajah, C., Ganesharajah, T., 2001b. Operational decisions in AGV-served flowshop loops fleet sizing and
decomposition. Annals of Operations Research 107, 189–209.
Ho, Y., 2000. A dynamic-zone strategy for vehicle-collision prevention and load balancing in an AGV system with a single-loop guide
path. Computers in Industry 42, 159–176.
Huang, C., 1997. Design of material transportation system for tandem automated guided vehicle systems. International Journal of
Production Research 35, 943–953.
Kim, K.K., 1995. Positioning of automated guided vehicles in a loop layout to minimize the mean vehicle response-time. International
Journal of Production Economics 39, 201–214.
Kim, K.K., Kim, J., 1997. Estimating mean response time and positioning idle vehicles of AGVs in loop layout. Computer and
Industrial Engineering 33, 669–672.
Kiran, A.S., Karabati, S., 1993. Exact and approximate algorithms for the loop layout problem. Journal of Production Planning and
Control 4, 253–259.
Kouvelis, P., Kim, M.W., 1992. Unidirectional loop network layout problem in automated manufacturing systems. Operations
Research 40, 533–550.
Kusiak, A., Heragu, S., 1987. The facility layout problem. European Journal of Operational Research 29, 229–251.
Langevin, A., Lauzon, D., Riopel, D., 1996. Dispatching routing, and scheduling of two automated guided vehicles in a flexible
manufacturing system. International Journal of Flexible Manufacturing Systems 8, 247–262.
Laporte, G., Asef-Vaziri, A., Sriskandarajah, C., 1996. Some applications of the generalized traveling salesman problem. Journal of
the Operational Research Society 47, 1461–1467.
Lee, C., Ventura, J.A., 2001. Optimal dwell point location of automated guided vehicles to minimize the mean response time in a loop
layout. International Journal of Production Research 39, 4013–4031.
Leung, J., 1992. A graph-theoretic heuristic for designing loop-layout manufacturing systems. European Journal of Operational
Research 57, 242–252.
Lin, J.T., Chang, C.C.K., Liu, W.C., 1994. A load-routing problem in a tandem-configuration automated guided-vehicle system.
International Journal of Production Research 32, 411–427.
Liu, H.-H., Hung, P.C., 2001. Real-time deadlock-free control strategy for single multiload automated guided vehicle on a job shop
manufacturing system. International Journal of Production Research 39, 1323–1342.
Mahadevan, B., Narendran, T.T., 1992. Design of an automated guided vehicle-based material handling system for an FMS.
International Journal of Production Research 30, 909–922.
Malmborg, C.J., 1994. Heuristic, storage space minimization methods for facility layouts served by looped AGV systems. International
Journal of Production Research 32, 2695–2710.
Maxwell, W.L., Muckstadt, J.A., 1982. Design of automated vehicle systems. IIE Transactions 2, 114–124.
Montreuil, B., 1991. A modeling framework for integrating layout design and flow network design. In: Proceedings of the Material
Handling Colloquium. Material Handling Institute of America, Charlotte, NC, pp. 43–58.
Muth, E.J., 1972. Analysis of closed-loop conveyor systems. AIIE Transactions 4, 34–143.
Muth, E.J., 1974. Analysis of closed-loop conveyor systems, the discrete flow case. AIIE Transactions 6, 73–83.
Muth, E.J., 1975. Modeling and system analysis of multistation closed-loop conveyors. International Journal of Production Research
13, 559–566.
 A. Asef-Vaziri, G. Laporte / European Journal of Operational Research 164 (2005) 1–11 11

Occena, L.G., Yakota, T., 1991. Modeling of an automated guided vehicle system (AGVS) in a just in time (JIT) environment.
International Journal of Production Research 29, 495–511.
Occena, L.G., Yakota, T., 1993. Analysis of the AGV loading capacity in a JIT environment. Journal of Manufacturing Systems 12,
24–35.
Rajasekharan, M., Peters, B.A., Yang, T., 1998. A genetic algorithm for facility layout design in flexible manufacturing systems.
International Journal of Production Research 36, 95–110.
Rajotia, S., Shanker, K., Bhatra, J.L., 1998. Determination of optimal AGV fleet size for an FMS. International Journal of Production
Research 36, 1177–1198.
Rendl, F., 2002. The quadratic assignment problem. In: Drezner, Z., Hamacher, H.W. (Eds.), Facility Location: Applications and
Theory. Springer-Verlag, Berlin, pp. 439–457.
Ross, E.A., Mahmoodi, E., Mosier, C.T., 1996. Tandem configuration automated guided vehicle systems: A comparative study.
Decision Sciences 27, 81–102.
Sharp, G.P., Liu, F.-H.F., 1990. An analytical method for configuring fixed-path closed-loop material handling systems. International
Journal of Production Research 28, 757–783.
Sinriech, D., Kotlarski, J., 2002. A dynamic scheduling algorithm for a multi-load multiple carrier system. International Journal of
Production Research 40, 1065–1080.
Sinriech, D., Palni, L., 1998. Scheduling pickups and deliveries in a multiple-load discrete carrier environment. IIE Transactions 30,
1035–1047.
Sinriech, D., Tanchoco, J.M.A., 1992a. The centroid projection method for locating pick-up and delivery stations in a single loop AGV
system. Journal of Manufacturing Systems 11, 297–307.
Sinriech, D., Tanchoco, J.M.A., 1992b. Impact of empty vehicle flow on performance of single-loop AGV systems. International
Journal of Production Research 30, 2237–2252.
Sinriech, D., Tanchoco, J.M.A., 1993. Solution methods for the mathematical models of single-loop AGV systems. International
Journal of Production Research 31, 705–725.
Sinriech, D., Tanchoco, J.M.A., 1995. An introduction to the segmented flow approach for discrete material flow systems.
International Journal of Production Research 33, 3381–3410.
Sinriech, D., Tanchoco, J.M.A., Herer, Y.T., 1996. The segmented bi-directional single-loop topology for material flow systems. IIE
Transactions 28, 40–54.
Sun, X.-C., Tchernev, N., 1996. Impact of empty vehicle flow on optimal flow path design for unidirectional AGV systems.
International Journal of Production Research 34, 2827–2852.
Tanchoco, J.M.A., Sinriech, D., 1992. OSL-optimal single loop guide paths for AGVs. International Journal of Production Research
30, 665–681.
Tansel, B.C., Bilen, C., 1998. Move based heuristics for the unidirectional loop network layout problem. European Journal of
Operational Research 108, 36–48.
Tompkins, A., White, J.A., Bozer, Y.A., Frazelle, E.H., Tanchoco, J.M.A., Trevino, J., 1996. Facilities Planning. Wiley, New York.
Ventura, J.A., Lee, C., 2000. A simulation study of idle vehicle positioning policies for automated guided vehicle systems. In: Graves,
R.J. et al. (Eds.), Progress in Material Handling Research, Developments in Material Handling Research. Material Handling
Institute of America, Charlotte, NC, pp. 392–403.
Ventura, J.A., Lee, C., 2001. A study of the tandem loop with multiple vehicles configuration for automated guided systems. Journal of
Manufacturing Systems 20, 153–165.
Ventura, J.A., Lee, C., 2002. Optimally locating multiple dwell points in a single loop guide path system. IIE Transactions
(forthcoming).