This article was downloaded by: [Temple University Libraries]

On: 07 January 2015, At: 03:27

Publisher: Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House,
37-41 Mortimer Street, London W1T 3JH, UK

International Journal of Production Research

Publication details, including instructions for authors and subscription information:
http://www.tandfonline.com/loi/tprs20

A new heuristic for integrated process planning and

scheduling in reconfigurable manufacturing systems
a a b
A. Bensmaine , M. Dahane & L. Benyoucef
a
LGIPM Research Laboratory - National School of Engineering of Metz (ENIM), Metz, France.
b
Aix-Marseille University, LSIS UMR 7296, Marseille, France.
Published online: 22 Jan 2014.

Click for updates

To cite this article: A. Bensmaine, M. Dahane & L. Benyoucef (2014) A new heuristic for integrated process planning and
scheduling in reconfigurable manufacturing systems, International Journal of Production Research, 52:12, 3583-3594, DOI:
10.1080/00207543.2013.878056

To link to this article: http://dx.doi.org/10.1080/00207543.2013.878056

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained
in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no
representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the
Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and
are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and
should be independently verified with primary sources of information. Taylor and Francis shall not be liable for
any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever
or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of
the Content.

This article may be used for research, teaching, and private study purposes. Any substantial or systematic
reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any
form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://
www.tandfonline.com/page/terms-and-conditions
 International Journal of Production Research, 2014
Vol. 52, No. 12, 3583–3594, http://dx.doi.org/10.1080/00207543.2013.878056

A new heuristic for integrated process planning and scheduling in reconfigurable

manufacturing systems
A. Bensmainea∗ , M. Dahanea and L. Benyoucefb
a LGIPM Research Laboratory - National School of Engineering of Metz (ENIM), Metz, France; b Aix-Marseille University,
LSIS UMR 7296, Marseille, France
(Received 22 Februar y 2013; accepted 7 December 2013)

Integrated process planning and scheduling (IPPS) is a manufacturing strategy that considers process planning and scheduling
as an integrated function rather than two separated functions performed sequentially. In this paper, we propose a new heuristic
to IPPS problem for reconfigurable manufacturing systems (RMS). An RMS consists mainly of reconfigurable machine tools
Downloaded by [Temple University Libraries] at 03:27 07 January 2015

(RMTs), each with multiple configurations, and can perform different operations with different capacities. The proposed
heuristic takes into account the multi-configuration nature of machines to integrate both process planning and scheduling. To
illustrate the applicability and the efficiency of the proposed heuristic, a numerical example is presented where the heuristic
is compared to a classical sequential process planning and scheduling strategy using a discrete-event simulation framework.
The results show an advantage of the proposed heuristic over the sequential process planning and scheduling strategy.
Keywords: reconfigurable manufacturing systems; integrated process planning and scheduling; genetic algorithms;
discrete-event simulation

1. Introduction
Recent years have seen the appearance of a recent manufacturing paradigm where production devices, such as machines and
material handling, can be added, removed, modified or interchanged as needed to increase the reactivity toward the real-
world variation (Koren et al. 1999). This new generation of manufacturing systems, called reconfigurable manufacturing
system (RMS), addresses some issues associated with the nature of other existing manufacturing paradigms, more specifically
dedicated manufacturing systems (DMSs) and flexible manufacturing systems (FMSs). Although the fixed automation and
the single product focus of DMS ensure a high volume throughput, resources are underutilised in case when demands volume
decreases, which creates lost. On the other hand, FMSs consist mainly of flexible computer numerically controlled (CNC)
machines, with all possible functionalities built-in, which heightens the system cost.
One of the primary components of RMS is the reconfigurable machine, also called reconfigurable manufacturing tool
(RMT). An RMT is a machine designed for a customised range of operation requirements. RMTs are characterised by their
cost-effective conversion when the contextual requirements change.
Product-realisation process covers a set activities, ranging from the design phase to the finished product, that are
responsible of transforming raw material to a product ready to be delivered. One of the core activities in any product-
realisation process is the process planning function. It consists to bridging the gap between design and manufacturing of a
part by translating the design data into the best manufacturing method. A process plan specifies what components are needed
to produce a product, and the necessary operations to transform the raw materials into a finished product (Salvendy 2001).
Scheduling is another function in the product-realisation process which consists to assigning manufacturing resources
to the operations indicated in the process plans with respect to some relevant criteria, such as due dates and costs. Since
the scheduling function is generally performed after the process planning, it is bounded by the restrictions dictated by the
process plans and by the availability of the production resources.
In manufacturing, process planning and scheduling are both responsible for the effective resource allocation and utilisation.
A process plan is usually generated with no regard for the scheduling objectives and with the assumption that all the resources
are available. Its preparation without the consideration of the shop floor status may become ineffective due to changes of
resources availability. Moreover, process planning and scheduling functions may consider different objectives, which make
it difficult to obtain satisfactory results with a simple sequential execution of the two functions.
Even though the process plan is optimally generated, the conflict between process planning and scheduling at the
scheduling phase causes the modification of the process plan in order to fit with the updated shop floor state. These
∗ Corresponding author. Email: a.bensmaine@enim.fr

© 2014 Taylor & Francis

 3584 A. Bensmaine et al.

modifications are generally carried by non-expert planners which will inevitably lead to loosely followed process plans,
resulting in an ineffective collaboration between process planning and scheduling.
To avoid this conflict problem, an integration approach between process planning and scheduling is considered to create a
realistic process plans that can be executed without alterations. A good integration may result in better product delivery time
and resources utilisation. Thus, integration of process planning and scheduling is essential to achieve integrated manufacturing
and to dismiss the conventional sequential manufacturing approach.
In this paper, we propose a new heuristic for an integrated process planning and scheduling (IPPS) strategy in a
reconfigurable environment. The proposed heuristic takes into account the RMS characteristics such as the reconfigurable
nature of the system and machines, at the logical and physical levels, as well as the customisable capacity of RMT.
The rest of the paper is organised as follows: Section 2 reviews the existing works dedicated to IPPS problem. Section 3
describes the problem under consideration. Section 4 presents the developed heuristic. Section 5 presents an illustrative
numerical example and discusses the obtained results. Section 6 concludes the paper with some remarks and future research
works directions.

2. Literature review
Downloaded by [Temple University Libraries] at 03:27 07 January 2015

Literature dealing with RMS is rich and covers many areas, such as designing problem, layout optimisation, reconfigurable
control process planning and production scheduling. In this research work, we are mainly concerned by process planning
and scheduling as dissociated functions in RMS, and then as an integrated function (IPPS).
Process planning (process plans generation) in reconfigurable environment saw a significant progress due to the recent
apparition of RMS, and hence the necessity to develop new methods adapted to its particular characteristics. Takahashi et al.
(2006) proposed a stochastic model that considers simultaneously production orders planning and its corresponding system
configuration that maximise profit. Their main idea is to assign several configurations to a group of products, depending on the
profit (use slower and cheaper machines when there are not too much orders waiting, and vice versa). Azab and ElMaraghy
(2007) considered a reconfigurable process plan, where an existing process plan could be reconfigured in order to meet new
requirements. Azab et al. (2007) proposed a new heuristic based on simulated annealing to sequence a set of machining
operations subject to precedence constraints. The objective is to minimise the total idle time. Musharavati et al. (2008)
adapted a simulated annealing metaheuristic for process planning generation in multipart reconfigurable manufacturing lines.
Shabaka and ElMaraghy (2008) proposed a new genetic algorithm-based approach to perform process plan manufacturing
costs in RMS environment. Oke, Abou-El-Hossein, and Theron (2011) proposed a combination of different techniques from
literature to form the process planning of a reconfigurable manufacturing system for mould and die making. They develop
a technique based on the weight of precedence factors to form a machining precedence order. They considered three main
factors to decide the machining precedence order, namely the technological factor, geometric factor and the economic factor.
Bensmaine, Dahane, and Benyoucef (2011) developed a simulation-based NSGA-II approach to solve the problem of process
plans generation for a multi-unit single-product type observed in RMS. They established a multi-unit macro-level process
plans in order to perform different parts/operations on several machines, and where a process plan is associated to each
unit. Chaube, Benyoucef, and Tiwari (2012) adapted NSGA-II algorithm to generate plans in an RMS. Two objectives are
considered respectively the minimisation of the makespan and the total manufacturing cost. Moreover, Bensmaine, Dahane,
and Benyoucef (2012) addressed the problem of process plans generation in RMS from a multi-objective perspective using
archived multi-objective simulated annealing (AMOSA) approach. They elaborated an experimental comparison based on
the obtained pareto-fronts. The results are classified in a preferential order using TOPSIS (Technique for Order of Preference
by Similarity to Ideal Solution). Musharavati and Hamouda (2012) combined simulated annealing (SA) metaheuristic with
knowledge exploitation and parallel architecture to enhance its performances to generate process plans in RMS. They succeed
in getting a satisfactory improvement of the metaheuristic performances and computational effort. Recently, Bensmaine,
Dahane, and Benyoucef (2013) considered the problem of RMS’s design based on products specifications and reconfigurable
machines capabilities. The problem is related to the selection of candidate reconfigurable machines among an available set,
which will be then used to carry out a certain product based on a single product-type characteristics. The selection of the
machines considers two main objectives respectively, the minimisation of the total cost (production cost, reconfiguration
cost, tool changing cost and tool using cost) and the total completion time. An adapted version of NSGA-II is proposed to
solve the problem.
Unlike process planning, scheduling in RMS has not been a very active area since the first prototypes of RMS. Yet,
some trials have been published recently on scheduling in RMS. Li and Xie (2006) proposed a genetic algorithm embedded
with extended timed-place Petri nets so that all possible behaviours of a reconfigurable production line can be completely
modelled by the modules. They oriented the crossover and mutation operators towards the elements of Petri nets model
instead of the ones in the problem space. Galan (2008) schedule ‘product families’ on RMS. Since RMSs are dedicated
 International Journal of Production Research 3585

generally to product families, authors propose an approach to group products into families and then schedule these families
with the objective of minimising the total cost. Metaheusitics are used to solve the problem. Nehzati et al. (2012) presented a
fuzzy-based scheduling model. Authors deal with the job assignment problem in a logical (soft) RMS. The model selects the
best alternative machine with a multi-criteria method. The use of the fuzzy is justified by the consideration of both qualitative
and quantitative comparison criteria. Yu et al. (2012) considered a practical constraint that is the numbers of pallets/fixtures
in the system are limited, so that a part can be released into the system only when a fixture required for the part is available.
They use a priority rule-based approach to solve the problem.
Integrated process planning and scheduling in RMS did not attract much attention in the literature. Actually, we did not
find works that clearly deal with IPPS in RMS. Nevertheless, to provide a vision on the context of IPPS, and how process
planning and scheduling are integrated, a state of the art dedicated to IPPS in classical manufacturing systems is summarised
hereafter.
The problem of integrating process planning and scheduling in manufacturing systems was addressed first by Khoshnevis
and Chen (1991) and Chen and Khoshnevis (1993). They presented the basic issues involved in the integration of the two
functions, developed heuristic algorithms and identified the potential benefits of the integration. Saygin and Kilic (1999)
developed an integrating framework for flexible non-linear process plans and offline rescheduling in FMS. The framework
consists of four integrated stages respectively machine tool selection, process plan selection, scheduling and rescheduling
Downloaded by [Temple University Libraries] at 03:27 07 January 2015

in modules to minimise the total completion time. Tan and Khoshnevis (2000) reviewed research works that have been
carried out in the area of integration of process planning and scheduling. They discussed the extent of applicability of the
various developed approaches and suggested directions for future research. Zhang, Saravanan, and Fuh (2003) elaborated
a new approach to the integration of process planning and scheduling for batch manufacturing of prismatic parts. Two
modules are considered, namely process planning module and scheduling module, linked by an intelligent facilitator. The
facilitator provides intuitive feedback to the process planning module in the form of extra constraints to process planning of
a particular product based on the schedule performance measures. Several objective functions such as machine utilisation
and job tardiness are provided for the evaluation of the generated schedule. Jain, Jain, and Singh (2005) proposed a scheme
for integration of process planning and scheduling that can be implemented in a company with existing process planning and
scheduling departments when multiple process plans (MPP) for each product type are available. The proposed scheme takes
advantage of MPP while following the real-time strategy for scheduling suitable for changing work-shop status. Li et al.
(2008) developed a genetic algorithm-based approach to facilitate the integration and optimisation of process planning and
scheduling functions. A specific coding and operator schemes have been developed to improve the quality of the solution.
Guo et al. (2009) considered the IPPS problem as a combinatorial optimisation model, and applied the particle swarm
optimisation (PSO) algorithm to solve it. Li et al. (2010) proposed a mathematical model for the integrated process planning
and scheduling problem, and used an evolutionary algorithm-based approach to solve it. The mathematical model is based
on the mixed integer programming model of the job shop scheduling problem. A great summary of IPPS works is done by
Phanden, Jain, and Verma (2011), the authors classified the IPPS works into three major classes: non-linear approaches, closed
loop approaches and distributed approaches. Readers can refer to this paper to more details about IPPS in manufacturing
systems.
Recently, Mohammadi, Karampourhaghghi, and Samaei (2012) developed a mixed integer linear programming (MILP)
scheduling model, with a multi-objective vision. The model has been resolved using a hybrid simulated annealing. Li, Gao, and
Li (2012) focused on a multi-objective integrated process planning and scheduling problem. A game theory-based approach
has been used to deal with the minimisation of respectively total completion time also called makespan, machine workload
and total workload of machines. Lian et al. (2012) presented an imperialist competitive algorithm (ICA) to address the IPPS
problem with the objective of minimising the makespan. Mohapatra, Benyoucef, and Tiwari (2013) used an intelligent search
technique called artificial immune system to integrate process planning and scheduling with a particular consideration of
setup planning. Phanden, Jain, and Verma (2013) proposed an iterative cyclic approach to integrate process planning and
scheduling. It consists of four modules: process plan selection module, scheduling module, schedule analysis module and
process plan modification module.
In all the reviewed IPPS works, the use of IPPS is considered more advantageous than process planning and scheduling
as separated functions. These works concern mainly job shops with classical machines, and to our knowledge the IPPS
problem has not been clearly tackled in the reconfigurable environment. Hence, the main goal of this paper is to propose
an integrated process planning and scheduling problem for the reconfigurable manufacturing system in order to explore the
opportunities as well as the issues that may be encountered. The problem under consideration in our work is detailed in next
section.
 3586 A. Bensmaine et al.

3. Problem description
Being reactive to market changes in the current economic context implies that products have be delivers as soon as possible
after a demand is received, and by consequence manufactured as fast as possible. Thus, in this research work, we consider the
case where a set of products are to be manufactured on the same RMS, with the objective of finishing this set as fast as possible.
Since RMSs are typified by multi-configuration machines, we assume that each machine configuration is characterised by its
functionality and providing different degrees of freedom along the three-dimensional axis ±x, ±y and ±z. Thus, machines
can be reconfigured depending upon the required capacity and the design specifications of the product. For each product type,
precedence relationships exist between its operations. They are due to the technological and the geometrical considerations
necessary to produce accurately every feature of the product. A set of operations of different products are to be completed
on a set of machines. RMTs are reconfigurable in nature and offer varied functionalities in their different configurations.
A configuration can be changed depending on the operating requirements.
In a non-integrated process plans generation approach, process planning needs to solve two different problems. The first
deals with the assignment of each operation to a machine, where the second addresses the sequencing of operations. These
problems are solved having in mind the established objective (total time, total cost . . .).
The assignment of operations to machines can be carried out in three steps:
Downloaded by [Temple University Libraries] at 03:27 07 January 2015

(1) Identify the tool approach direction (TAD) and the type of tool required to carry out the particular operation.
(2) Identify among the available machines and the various configurations, the set of machines, which can perform that
particular operation. This is done by identifying the TAD offered by the machine and the tools available with it.
(3) Assign the machines and the appropriate configuration to the particular operation of the part.
The sequencing of operations defines the order in which operations are carried out, based on precedence constraints. The
generated process plans are those with a minimum makespan. The makespan provides information about the time required
to perform the last operation of the process plan, it includes:
• Operation processing time: Time required by a machine to process an operation for a particular operation. It depends
on the machine, its configuration and the type of the operation to be performed. The information about the processing
time may be obtained from the past data or by conducting pilot runs;
• Transportation time: Time required for transportation of the job between machines. It is calculated in the case where
a particular job has an operation to be performed on different machine. We designate by ‘job’ every product that is
still in the manufacturing system;
• Configuration changeover time: Time required to change machine configuration. It is accomplished by adding or
removing the modules of the machine to change its functionality. The time involved in this activity depends on the
machine and the previous configuration on which machine was operating;
• Tool changeover time: Time required for a machine to change its tool depending on the type of operation to be
performed on it. The tool changeover is distinguished from the configuration changeover because unlike machine’s
modules, tools are not parts of the RMT. Thus, even if a machine has the satisfying configuration in terms of TADs
to perform a particular operation, it should also possess the right tool in order to be assigned to that operation.
Furthermore, a machine may change the tool and keeps its current configuration, and vice versa.
The key for integrating process planning and scheduling lies in the fact that a part can be manufactured through different
process plannings. Unlike the classical sequential process planning and scheduling, IPPS relaxes the single sequence and the
definitive assignment of operations to machines. Thus, alternative process plans can be generated using two strategies:
(1) Process flexibility: corresponds to the possibility of carrying out an operation on different machines/configurations,
tools and tool approach directions (TADs). In RMS, a reconfigurable manufacturing tool (RMT) can be reconfigured
in order to adapt its functionality and capacity to operating requirements (Koren 2010).
(2) Operations sequencing flexibility: refers to the possibility of interchanging the order in which operations are per-
formed. In fact, from an operations precedence graph, many operation sequences are feasible and satisfy precedence
constraints (Figure 1).

4. Proposed heuristic
Traditional approaches that treat process planning and scheduling separately can result in local solutions for the two functions.
The integration of both functions into one optimisation problem, by considering the constraints of process planning and
scheduling simultaneously, increases significantly the solution space. Actually, if scheduling is considered alone, the solution
space is already huge enough to make it hard for exact methods to generate optimal solutions. The situation is worse with
 International Journal of Production Research 3587
Downloaded by [Temple University Libraries] at 03:27 07 January 2015

Figure 1. Operations precedence graph with its different alternative possible sequences.

IPPS. Metaheuristics may represent an alternative to exact methods with some extra efforts. Firstly, the IPPS is subject to
many constraints such as operations precedence, machine feasibility in terms of TADs and tools, machines availability, etc.
Knowing that most of metaheuristics use a random mechanism to generate a new generation of solutions from an existing
one, these constraints make the task of creating new feasible solutions complex, and most of the generated solutions would
be unfeasible. Consequently, the metaheuristic implementation should provide a correction routine that repairs the unfeasible
solutions. To cope with these difficulties, we propose a new heuristic which allows an efficient integration of process planning
and scheduling in a reconfigurable environment.
The process flexibility implies that every operation can be performed on different machines/configurations. The assign-
ment of operations to machines in an IPPS strategy should consider not only the processing time, which is the case in classical
process planning, but should also consider the availability of machines since they are shared among all the entities present
in the system. To include machines availability in the IPPS process, we define a selection index (S I ) that will be used to
decide both operations selection and operation/machine assignment. The selection index is calculated for every operation op
not scheduled yet, using the availability information about the candidate machines that are able to perform op. Furthermore,
at any time while the manufacturing system is operating, every RMT in the system may have an operation on which it is
performing, as well as waiting jobs in its associated queue. Thus, the availability time (AT ) of an RMT is the sum of the
processing time, configuration time and tool changing time for every job in the machine, i.e. the job being processed as well
as the jobs in the waiting in the queue (Equation 1).
At each step of the IPPS, the S I is calculated for every operation not yet scheduled. The operation which will be scheduled
is the one with the highest selection index. The availability time (AT ) of every machine is calculated at every step of the
IPPS as well. It is the sum of the processing time, configuration time and tool changing time for every entity in the machine,
i.e. the entity being processed and the entities in the waiting queue associated to the considered machine.
Q m +1 
N         
ATm = Pr oc Mmk Cmk [k] + CC T Cmk Cmk+1
 + T C T Tmk Tmk+1
 (1)
k=0

i∈M(op) ATi
Average ATop = (2)
|M(op)|
where k = 0 is the operation processed by the machine, k = 1...N Q m the operations waiting in machine m queue, and
k = N Q m + 1 is the operation op that we intend to assign.
Average ATop − min(ATm )
S Iop = , where m ∈ M(op) (3)
Average ATop
 3588 A. Bensmaine et al.
where:
ATm Availability time for machine m
S Iop Selection index of operation op
N Qm Number of jobs in the queue of machine m
M(op)   Set of machine that are able to perform operation op
Pr oc Mmk k [J ob ]
Cm k in its configuration C k
The processing time on job k using machine Mm
k m
  
k k+1
CC T Cm Cm Configuration changing time. The time required by machine Mm to change its configuration
from Cmk used to perform the operation k to configuration C k+1 used to perform the next job
m
k+1
Average
 AT op  The average value of the machines able to perform the operation op
T C T Tmk Tmk+1 Tool changing time. The time required by machine Mm to change the tool from Tmk used to
perform the operation k to Tmk+1 used to perform the next job k + 1
Downloaded by [Temple University Libraries] at 03:27 07 January 2015

Figure 2. Flowchart of the proposed IPPS heuristic.

For each operation op, the S I is defined as the difference between the Average ATop of the set of all machines that are
able to perform op and the minimum AT of that same set. The S I is higher when the current state of the system ensures
that op will be achieved in a relatively short time if the machine with the minimum AT is used, since its AT is close to the
average AT . Note that the operation processing time is already included in the calculation of AT . Therefore, the AT is not
calculated based on the starting time of an operation but rather on its makespan. Once AT and S I are calculated respectively
for RMTs and operations that are not yet scheduled, the IPPS schedules the operation with the highest S I and calculates
again AT s and S I s. A global view of the heuristic algorithm is given in Figure 2.
The idea behind choosing the operation with the greatest S I is the following: every operation could be performed by at
least one RMT in the system. In the case where multiple machines are capable of the processing the same operation then a
decision should be made in order to choose the most adequate one in term of delay. This selection depends on a dynamic state
 International Journal of Production Research 3589

Table 1. Structure of a process plan.

Operation O1 O3 O4 O2 O6 O7 O5
Machine M1 M1 M2 M1 M2 M3 M2
Configuration C1 C1 C3 C1 C2 C1 C2
Tool T1 T2 T1 T2 T3 T2 T4

Figure 3. Genome representing a schedule.

of the RMS as well as RMTs availability. For every operation among the unscheduled operations, there is a machine that is
Downloaded by [Temple University Libraries] at 03:27 07 January 2015

able to carry it out in the shortest possible time, that is the machine with the minimum AT . When an operation is selected to
be scheduled then that will be on the machine with the minimum AT .
The selection criterion of operations is based on the degree of shortness of the carrying out time. If an operation has the
possibility to be scheduled on the machine with a very low AT compared to other candidate machines, and that the degree
of shortness is the highest compared to other concurrent operations, then it will be immediately scheduled. The mentioned
degree of shortness is the selection index S I (Equation 3). The higher is the index S I , the more important is the degree of
shortness. Actually, index S I is high for a particular operation op when the fastest machine that is able to perform op is quite
faster than the Average ATop . On the other side, if S I is low, that means that other machines can perform the same operation
op with a close operating time to the fastest, so the operation will not lose much if it will not be scheduled on that fastest
machine.

5. Illustrative example
In this section, we consider a numerical example to illustrative the applicability of the proposed heuristic. The numerical
results of the heuristic and the classical sequential approach are compared using a discrete-event simulation framework
adapted to RMS environment. The simulation framework is developed using the Java programming language. It is based
on a set of classes divided according to their foundational nature into two packages, namely simulation package and RMS
package. The simulation package contains the necessary classes to run the simulation (i.e. Simulation class, Scheduler class
and Event class), and is responsible for launching and monitoring the whole simulation. The RMS package gathers the set
of classes required to model the considered RMS (i.e. ReconfigurableMachine class, Link class and Entity class).
Before going into the details of the numerical results, we present in the following the implementation of the used classical
sequential strategy which consists into two main steps, respectively, process plan generation and operations scheduling.
Initially and for each product type, a process plan is generated. Then, based on the generated process plans, all the operations
are scheduled. In our case and to generate process plans, the approach proposed by Bensmaine, Dahane, and Benyoucef
(2012) is used.
A process plan is presented as an M × N matrix (Table 1), where N is the number of operations required to carry out
the product, and M represents the four necessary information for a process plan, which are the concerned operation, the
associated machine, the used configuration as well as the operating tool. The process plan table is interpreted from the left
to the right, column by column. For example, the first column of Table 1 is read as follows: the operation O1 is processed
on machine M1, with the C1 configuration, using the tool T 1. The remaining columns are interpreted using the same way.
Moreover, for the generated process plans, feasible schedules are obtained using a basic simulation-based genetic
algorithm. The simulation models are build using the developed discrete-event simulation framework. The application
of genetic algorithm (GA) to scheduling problems is well known in the literature (Nakano and Yamada 1991; Fang, Ross and
Corne 1993; Cheng, Gen, and Tsujimura 1999; Gonçalves, de Magalhaes Mendes, and Resende 2005; Pezzella, Morganti,
and Ciaschetti 2008). Implementing a genetic algorithm, like almost all other metaheuristics, begins with the coding phase.
In our case, the schedule genome is defined as a sequence of operations (Figure 3).
A specific sequence of operations (genes) represents one genome in the population. Of course, we must ensure that the
precedence constraints are followed. Every gene contains the identifier of the operation with its corresponding product. For
instance, the gene O1P1 represents the operation number one in the process plan of product one. The initial population of
 3590 A. Bensmaine et al.

Table 2. Offered TADs by machine configuration.

Provided axes
Machine Configuration
+x +y +z −x −y −z

C1 × × ×
Machine 1 C2 × × ×
C3 × ×
C1 × × ×
Machine 2 C2 × × × ×
C3 × × ×
C1 × × ×
Machine 3
C2 × × × ×
Downloaded by [Temple University Libraries] at 03:27 07 January 2015

Figure 4. Operations precedence graph of product 1.

1000 genomes is randomly generated. The population then evolves using the genetic operators, namely crossover (with a
probability of 0, 8) and mutation (with a probability of 0, 2). The offspring are selected according to their fitness, which
correspond to schedules makespans obtained by the simulation. Moreover, the simulation starts with an initial empty RMS,
and stops when all the operations are realised. More details about process planning in RMS are given (Bensmaine, Dahane,
and Benyoucef 2012), and more general information about scheduling with genetic algorithms are presented in Cheng, Gen,
and Tsujimura (1999).
The rest of the section is devoted to the description of the inputs and the obtained numerical results. RMTs exist in different
configurations, and their input specifications include the set of available configurations and their respective provided tool
approach directions (Table 2). The processing time of each operation depends on the assigned machine and its selected
configuration. Unlike flexible manufacturing, the variability of products that exist at the same time on an RMS is limited.
Actually, RMSs are designed around the part family, and may produce a couple of product types at the same time. Therefore,
we consider a basic case where two product types are being manufactured on the same RMS.
The input data consist of two product types, each of which possesses an operations precedence relationship graph and
the tool approach direction for each operation. To objectively explore the potential of the proposed heuristic, many instances
are randomly generated, by generating the operations precedence graph as well as the required TADs and tools according to
the following steps:
(1) generate the number of operations between 5 and 20 (according to the literature, 5–20 interval is quite representative);
(2) Classify operations into a number of levels ranging from 4 to 15 levels, in order to have a more or less balanced
density on levels.
(3) Precedence relationships are randomly created between operations taking into account the level of each operation,
i.e. links should only be established from a higher to a lower level, in order to avoid cycles in the graph.
These generated pairs of products are then scheduled using the classical approach and the IPPS heuristic. The obtained results
are compared.
The generated data of the first instance are presented hereinafter. Two product types are generated, where the precedence
graphs are illustrated in Figures 4 and 5, and their relative requirements in terms of TADs given in Tables 3 and 4. Using
the classical approach, two process plans are presented in Tables 5 and 6. Once the process plans are generated, products
operations are then ready to be scheduled. The schedules are obtained using above genetic algorithm.
 International Journal of Production Research 3591

Figure 5. Operations precedence graph of product 2.

Downloaded by [Temple University Libraries] at 03:27 07 January 2015

Table 3. Product 1 required TADs.

Required axes
Operation
+x +y +z −x −y −z

op1 ×
op2 ×
op3 × ×
op4 ×
op5 × ×
op6 × ×
op7 × ×
op8 ×
op9 × ×
op10 × ×
op11 ×
op12 × ×

Table 4. Product 2 required TADs.

Required axes
Operation
+x +y +z −x −y −z

op1 × ×
op2 ×
op3 × ×
op4 × ×
op5 ×
op6 ×
op7 × ×
op8 × ×

Our IPPS heuristic is applied on the same instances, and the overall results are summarised in Table 7. The gap between
the makespan of the classical approach and our proposed heuristic is calculated using the formula classical−I
I PPS
PPS
. The gap
rate ranges from 8.7% to 20.6%, maintaining a positive values for all instances, which demonstrates an improvement of the
overall makespan using the proposed IPPS heuristic, and by consequence a better performing manufacturing delay. This is
due to its exploitation of RMT reconfigurability, by the integration between process planning and scheduling, which relaxes
the rigidity of process plans resources assignment, and allows a freer browsing of the search space. The classical approach
 3592 A. Bensmaine et al.

Table 5. Process plan for product 1.

Operation O1 O12 O4 O6 O2 O5 O7 O3 O8 O9 O10 O11

Machine M3 M2 M2 M3 M1 M3 M1 M3 M3 M3 M2 M2
Configuration C1 C2 C2 C1 C1 C1 C3 C1 C1 C2 C2 C3
Tool T6 T6 T3 T7 T7 T2 T8 T2 T9 T10 T1 T5

Table 6. Process plan for product 2.

Operation O4 O1 O3 O5 O6 O8 O2 O7
Machine M3 M3 M3 M1 M1 M3 M3 M3
Configuration C1 C1 C1 C1 C1 C1 C1 C2
Tool T6 T6 T2 T7 T8 T9 T7 T2
Downloaded by [Temple University Libraries] at 03:27 07 January 2015

Table 7. Classical approach vs. IPPS heuristic comparison.

Makespan using a Makespan using Computational effort  Makespan GAP 

Instance classical approach (s) IPPS heuristic (s) of our IPPS heuristic (s) classical−I P P S
I PPS

Instance 1 3032 2602 2.3 16.5%

Instance 2 14290 11864 4.2 20.4%
Instance 3 7387 6481 3.7 14.0%
Instance 4 5521 4702 3.6 17.4%
Instance 5 10362 8754 4.0 18.4%
Instance 6 2941 2670 2.5 10.1%
Instance 7 3799 3149 2.6 20.6%
Instance 8 8814 8107 3.6 8.7%
Instance 9 5586 4852 3.0 15.1%
Instance 10 3546 3117 2.0 13.8%

has the advantage of performing process planning and scheduling in an offline mode, before launching the production, which
free the manufacturing system from any heavy computational activity during the operating phase. IPPS generally does not
offer this characteristic, thus it is important to consider the computational effort when it comes to IPPS. Calculation time of
our IPPS heuristic shows that it is not time consuming regarding the overall makespan (Table 7), which is very necessary to
the applicability in a realistic manufacturing environment.

6. Conclusion and future works

In this paper, we proposed a new heuristic to solve the problem of integrated process planning and scheduling (IPPS)
in a reconfigurable manufacturing system (RMS). RMSs are designed to be fast enough and to produce with a sufficient
flexibility to react to the market changes. Thus, in order to take full advantage of their capacities, it is necessary to develop
new manufacturing strategies that take into account the reconfigurability of RMSs.
The proposed heuristic is applied on a simple example using a discrete-event simulator that we developed. The proposed
simulator is implemented to fit the RMS modelling needs in term of RMT. The obtained results are compared with the classical
sequential process planning and scheduling, and the results showed the efficient applicability of the heuristic in term of the
total completion time. Another important advantage of the proposed heuristic lies in the fact that it is an ‘online’ approach.
Indeed, in a real-world manufacturing environment, machines are vulnerable to failures, and sometimes in an unpredictable
moment. The proposed heuristic considers the available machines at every steps, a failed machine could simply ignored
during the operations/machines assignment.
Moreover, due to the complexity of the IPPS problem, we are planning firstly to adapt existing metaheuristics, such
as genetic algorithms (GA), simulated annealing (SA) and particle swarm optimisation (PSO) developed for classical
manufacturing environment, to reconfigurable environment. A natural comparison between the performances of our heuristic
and the metaheuristics is necessary. Secondly, we are expecting to model exactly the IPPS in RMS environment using
 International Journal of Production Research 3593

mathematical programming if possible, and then to solve it to optimality on small instances using optimisation tools and to
establish comparisons between the mathematical model and our heuristic.
As perspectives for future work, it is important to consider RMTs failure to examine the robustness of the proposed
heuristic. Since machines failures occur randomly, the developed simulation framework will perfectly match the context.
Furthermore, other criteria can be considered, such as product quality with preference on machines, and customer demands
with due dates.

Funding
These research works were partially supported by the Region Lorraine, France, through the funding of Mr. Bensmaine’s PhD thesis (Oct.
2010–Nov. 2013). This support is gratefully acknowledged.

References

Azab, A., and H. A. ElMaraghy. 2007. “Mathematical Modeling for Reconfigurable Process Planning.” CIRP Annals – Manufacturing
Downloaded by [Temple University Libraries] at 03:27 07 January 2015

Technology 56 (1): 467–472.

Azab, A., G. Perusi, H. A. ElMaraghy, and J. Urbanic. 2007. “Semi-generative Macro-process Planning for Reconfigurable Manufacturing.”
In Digital Enterprise Technology, edited by P. Cunha and P. Maropoulos, 251–258. New York, NY: Springer.
Bensmaine, A., M. Dahane, and L. Benyoucef. 2011. “Simulation-based NSGA-II Approach for Multi-unit Process Plans Generation in
Reconfigurable Manufacturing System.” In 16th IEEE International Conference on Emerging Technologies and Factory Automation,
Volume v, Toulouse, France, 1–8.
Bensmaine, A., M. Dahane, and L. Benyoucef. 2012. “Process Plan Generation in Recongurable Manufacturing Systems using AMOSA
and TOPSIS.” In 14th IFAC Symposium on Information Control Problems in Manufacturing, Bucharest, 560–565.
Bensmaine, A., M. Dahane, and L. Benyoucef. 2013. “A Non Dominated Sorting Genetic Algorithm Based Approach for Optimal Machines
Selection in Reconfigurable Manufacturing Environment.” Computers & Industrial Engineering 66 (3): 519–524.
Chaube, A., L. Benyoucef, and M. K. Tiwari. 2012. “An Adapted NSGA-2 Algorithm Based Dynamic Process Plan Generation for a
Reconfigurable Manufacturing System.” Journal of Intelligent Manufacturing 23 (4): 1141–1155.
Chen, Q., and B. Khoshnevis. 1993. “Scheduling with Flexible Process Plans.” Production Planning & Control:The Management of
Operations 4 (4): 333–343.
Cheng, R., M. Gen, and Y. Tsujimura. 1999. “A Tutorial Survey of Job-shop Scheduling Problems Using Genetic Algorithms, Part II:
Hybrid Genetic Search Strategies.” Computers & Industrial Engineering 36 (2): 343–364.
Fang, H.-L., P. Ross, and D. Corne. 1993. “A Promising Genetic Algorithm Approach to Job-shop Scheduling, Rescheduling, and Open-
shop Scheduling Problems.” In Proceedings of the Fifth International Conference on Genetic Algorithms, San Mateo, CA: Morgan
Kauffman, 375–382.
Galan, R. 2008. “Hybrid HeuristicApproaches for Scheduling in Reconfigurable Manufacturing Systems.” In Metaheuristics for Scheduling
in Industrial and Manufacturing Applications. Vol. 128 of Studies in Computational Intelligence, edited by F. Xhafa and A. Abraham,
211–253. Berlin: Springer.
Gonçalves, J. F., J. J. de Magalhaes Mendes, and M. G. Resende. 2005. “A Hybrid Genetic Algorithm for the Job Shop Scheduling Problem.”
European Journal of Operational Research 167 (1): 77–95.
Guo, Y. W., W. D. Li, A. R. Mileham, and G. W. Owen. 2009. “Applications of Particle Swarm Optimisation in Integrated Process Planning
and Scheduling.” Robotics and Computer-Integrated Manufacturing 25 (2): 280–288.
Jain, A., P. K. Jain, and I. P. Singh. 2005. “An Integrated Scheme for Process Planning and Scheduling in FMS.” The International Journal
of Advanced Manufacturing Technology 30 (11–12): 1111–1118.
Khoshnevis, B., and Q. M. Chen. 1991. “Integration of Process Planning and Scheduling Functions.” Journal of Intelligent Manufacturing
2 (3): 165–175.
Koren, Y. 2010. The Global Manufacturing Revolution: Product-Process-Business Integration and Reconfigurable Systems. Hoboken, NJ:
John Wiley & Sons.
Koren, Y., U. Heisel, F. Jovane, T. Moriwaki, G. Pritschow, A. G. Ulsoy, and H. Van Brussel. 1999. “Reconfigurable Manufacturing
Systems.” CIRP Annals - Manufacturing Technology 48 (2): 527–540.
Li, X., L. Gao, and W. Li. 2012. “Application of Game Theory Based Hybrid Algorithm for Multi-objective Integrated Process Planning
and Scheduling.” Expert Systems with Applications 39 (1): 288–297.
Li, X., L. Gao, X. Shao, C. Zhang, and C. Wang. 2010. “Mathematical Modeling and Evolutionary Algorithm-based Approach for Integrated
Process Planning and Scheduling.” Computers & Operations Research 37 (4): 656–667.
Li, X., L. Gao, G. Zhang, C. Zhang, and X. Shao. 2008. “A Genetic Algorithm for Integration of Process Planning and Scheduling Problem.”
In Intelligent Robotics and Applications. Vol. 5315 of Lecture Notes in Computer Science, edited by C. Xiong, H. Liu, Y. Huang, and
Y. Xiong, 495–502. Berlin: Springer.
Lian, K., C. Zhang, L. Gao, and X. Li. 2012. “Integrated Process Planning and Scheduling Using an Imperialist Competitive Algorithm.”
International Journal of Production Research 50 (15): 4326–4343.
 3594 A. Bensmaine et al.

Li, A., and N. Xie. 2006. “A Robust Scheduling for Reconfigurable Manufacturing System Using Petri Nets and Genetic Algorithm.” In
The Sixth World Congress on Intelligent Control and Automation, Dalian, 7302–7306.
Mohammadi, G., A. Karampourhaghghi, and F. Samaei. 2012. “A Multi-objective Optimisation Model to Integrating Flexible Process
Planning and Scheduling Based on Hybrid Multi-objective Simulated Annealing.” International Journal of Production Research 50
(18): 5063–5076.
Mohapatra, P., L. Benyoucef, and M. Tiwari. 2013. “Realising Process Planning and Scheduling Integration Through Adaptive Setup
Planning.” International Journal of Production Research 51 (8): 2301–2323.
Musharavati, F., and A. S. M. Hamouda. 2012. “Enhanced Simulated-annealing-based Algorithms and their Applications to Process
Planning in Reconfigurable Manufacturing Systems.” Advances in Engineering Software 45 (1): 80–90.
Musharavati, F., N. Ismail, A. Majid, S. Hamouda, and A. R. Ramli. 2008. “A Metaheuristic Approach to Manufacturing Process Planning
in Reconfigurable Manufacturing Systems.” Jurnal Teknologi 48: 55–70.
Nakano, R., and T. Yamada. 1991. “Conventional Genetic Algorithm for Job Shop Problems.” In Proceedings of the Fourth International
Conference on Genetic Algorithms, San Mateo, CA, 474–479.
Nehzati, T., N. Ismail, F. A. Aziz, and S. A. Hosseini. 2012. “Research Outline on Reconfigurable Manufacturing System Production
Scheduling Employing Fuzzy Logic.” International Journal of Information and Electronics Engineering 2: 813–816.
Oke, A. O., K. Abou-El-Hossein, and N. J. Theron. 2011. “Process Planning for Reconfigurable Manufacturing System for Mould/Die
Making.” In The 4th Robotics and Mechatronics Conference of South Africa, Pretoria.
Downloaded by [Temple University Libraries] at 03:27 07 January 2015

Pezzella, F., G. Morganti, and G. Ciaschetti. 2008. “A Genetic Algorithm for the Flexible Job-shop Scheduling Problem.” Computers &
Operations Research 35 (10): 3202–3212.
Phanden, R. K., A. Jain, and R. Verma. 2011. “Integration of Process Planning and Scheduling: A State-of-the-Art Review.” International
Journal of Computer Integrated Manufacturing 24 (6): 517–534.
Phanden, R. K., A. Jain, and R. Verma. 2013. “An Approach for Integration of Process Planning and Scheduling.” International Journal
of Computer Integrated Manufacturing 26: 284–302.
Salvendy, G. 2001. Handbook of Indusrtrial Engineering: Technology and Operations Management. 3rd ed. Hoboken, NJ: John Wiley &
Sons.
Saygin, C., and S. E. Kilic. 1999. “Integrating Flexible Process Plans with Scheduling in Flexible Manufacturing Systems.” The
International Journal of Advanced Manufacturing Technology 15 (4): 268–280.
Shabaka, A. I., and H. A. ElMaraghy. 2008. “A Model for Generating Optimal Process Plans in RMS.” International Journal of Computer
Integrated Manufacturing 21 (2): 180–194.
Takahashi, K., K. Morikawa, T. Ohiro, and A. Takubo. 2006. “Selecting an Optimal Production Order and Its Corresponding Configuration
in a Reconfigurable Manufacturing System with Multiple Product.” In Reconfigurable Manufacturing Systems and Transformable
Factories, edited by A.-I. Dashchenko, 609–626. Berlin: Springer.
Tan, W., and B. Khoshnevis. 2000. “Integration of Process Planning and Scheduling – A Review.” Journal of Intelligent Manufacturing
11 (1): 51–63.
Yu, J.-M., H.-H. Doh, J.-S. Kim, D.-H. Lee, and S.-H. Nam. 2012. “Scheduling for a Reconfigurable Manufacturing System with Multiple
Process Plans and Limited Pallets/Fixtures.” World Academy of Science, Engineering and Technology 62: 232–237.
Zhang, Y. F., A. N. Saravanan, and J. Y. H. Fuh. 2003. “Integration of Process Planning and Scheduling by Exploring the Flexibility of
Process Planning.” International Journal of Production Research 41 (3): 611–628.