You are on page 1of 19

Journal of Manufacturing Systems 32 (2013) 119

Contents lists available at SciVerse ScienceDirect

Journal of Manufacturing Systems


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

Review

A review and critique on integrated productiondistribution planning models


and techniques
Behnam Fahimnia a, , Reza Zanjirani Farahani b , Romeo Marian c , Lee Luong d
a

Room J1-11, Mawson Lakes Campus, University of South Australia, School of Advanced Manufacturing and Mechanical Engineering, Mawson Lakes, SA 5095, Australia
Room 330, Department of Informatics and Operations Management, Kingston Business School, Kingston University, Kingston Hill, Kingston Upon Thames, Surrey KT2 7LB, UK
Room J2-09, Mawson Lakes Campus, University of South Australia, School of Advanced Manufacturing and Mechanical Engineering, Mawson Lakes, SA 5095, Australia
d
Room J1-14B, Mawson Lakes Campus, University of South Australia, School of Advanced Manufacturing and Mechanical Engineering, Mawson Lakes, SA 5095, Australia
b
c

a r t i c l e

i n f o

Article history:
Received 18 June 2010
Received in revised form
14 December 2011
Accepted 10 July 2012
Available online 11 August 2012
Keywords:
Productiondistribution planning
Supply chain management
Integration, Optimisation
Survey

a b s t r a c t
Optimisation modelling of integrated productiondistribution (PD) plans has raised signicant interest
among both researchers and practitioners over the past two decades. This paper provides the readers
with a comprehensive review and critique on the current PD planning and optimisation literature. We
classify the published PD planning models into seven categories based on their degree of complexity
and hence capability in addressing real-life scenarios. Summary tables highlight the main characteristics
of the selected models at each category. Next, the paper reclassies and evaluates the proposed models
based on the solution techniques used. Lastly, the unaddressed areas in the current literature are highlighted, important managerial implications are proposed and directions for future research in the area
are suggested.
2012 The Society of Manufacturing Engineers. Published by Elsevier Ltd. All rights reserved.

Contents
1.
2.
3.

4.

5.

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Integrated PD planning problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Complexity-based classication of literature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.1.
Single-product models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.2.
Multiple-product, single-plant models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.3.
Multiple-product, multiple-plant, single or no warehouse models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.4.
Multiple-product, multiple-plant, multiple-warehouse, single/no end-user models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.5.
Multiple-product, multiple-plant, multiple-warehouse, multiple-end user, single-transport path models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.6.
Multiple-product, multiple-plant, multiple-warehouse, multiple-end user, multiple-transport path, no-time period models . . . . . . . . . . .
3.7.
Multiple-product, multiple-plant, multiple-warehouse, multiple-end user, multiple-transport path, no-time period models . . . . . . . . . . .
Solution-based classication. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.1.
Mathematical techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.2.
Heuristic techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.3.
Simulation modelling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.4.
Genetic algorithms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Implications for the future of PD planning and optimisation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.1.
Category-based modelling implications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.1.1.
Category-based observations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.1.2.
Identied research gaps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.1.3.
Future research trends . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.2.
General PD modelling implications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.2.1.
General observations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Corresponding author. Tel.: +61 8 83023552; fax: +61 8 83023380.


E-mail address: behnam.fahimnia@unisa.edu.au (B. Fahimnia).
0278-6125/$ see front matter 2012 The Society of Manufacturing Engineers. Published by Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.jmsy.2012.07.005

2
2
2
3
4
6
8
9
11
11
11
11
14
14
14
15
15
15
15
15
15
16

B. Fahimnia et al. / Journal of Manufacturing Systems 32 (2013) 119

6.

5.2.2.
Identied research gaps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.2.3.
Future research trends . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Conclusions and directions for future research . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1. Introduction
A supply chain (SC) can be dened as an integrated system synchronising a series of interrelated business processes in order to: (1)
acquire raw materials and parts, (2) transform these raw materials
and parts into nished products, and (3) distribute these products
to either retailers or customers. A SC facilitates information ow
among various business entities such as suppliers, manufacturers,
distributors, third party logistic providers, retailers and customers
[1]. The two core optimisation problems in a SC are production and
distribution planning. In production planning, decisions regarding
hiring and ring of labour, regular time and overtime production,
subcontracting, and machine capacity levels are made for a denite
planning horizon (i.e. usually a one year period). Distribution planning decisions, on the other hand, pertain to determining which
facility(ies) would cater to the demands of which market(s).
In a conventional SC, independent manufacturers, wholesalers,
and retailers are separate business entities seeking to maximise
their own prots, although this goal is known to eventually produce
prot for the system as a whole [2]. It is now demonstrated that production and distribution decisions are mutually related problems
and need to be dealt with simultaneously in an integrated manner [3]. Developing integrated SC models with centralised planning
naturally leads to complex models which are difcult to solve optimally. For this reason alternative solution techniques developed in
the literature are only used to provide near optimal solutions for
small to medium-size integrated SC planning models [2].
Productiondistribution (PD) planning and optimisation in the
context of supply chain management (SCM) has raised signicant
interest for both researchers and practitioners over the past few
years. There might be two primary reasons behind this trend: (1)
positively affecting the protability of the SC through the global
integration of production and distribution activities and (2) reducing the lead-times and offering quicker response to market changes
and hence reducing the propagation of unexpected and undesirable events through the network [4,5]. These are the drivers for
the extensive literature addressing the modelling and optimisation of integrated PD plans in SCs. This paper aims to provide a
comprehensive review and critique on the current literature of PD
planning with special emphasis placed on those targeting the global
optimisation of production and distribution activities. The paper
will highlight the unaddressed areas in the current literature, proposes important managerial implications and suggests directions
for future research in the area.
2. Integrated PD planning problem
A vast amount of research has addressed the issue of optimisation of production plans in the SC context [615]. Many others
have investigated the problems exclusively in the area of distribution planning [1621]. However, a new approach to the analysis of
production and distribution operations has been proposed based
on the integration of decisions of different functions in production and distribution networks into a single optimisation model.
A PD system (depicted in Fig. 1) is often composed of factories
producing the goods and a hierarchy of warehouses or distribution centres (DCs) stocking goods for distribution to retail stores
where the demand for these goods originates [22]. A PD planning
problem is the problem of simultaneously optimising the decision

16
16
16
17

variables of different functions that have traditionally been optimised sequentially in the sense that the optimised outputs of the
production stage have become the input to the distribution stage.
A complex integrated PD plan, illustrated in Fig. 1, deals with
the following problems within the context of SC planning:
Quantity of each product produced in regular-time in each plant
at each period.
Quantity of each product produced in overtime in each plant at
each period.
Quantity of each product outsourced by each manufacturing plant
at each period.
Work-in-Progress (WIP) inventory amount in each plant at the
end of each period.
Inventory amount of nished products temporarily stored in the
stack buffers in each plant at the end of each period.
Quantity of each product shipped from stack buffers to warehouses during each period.
Quantity of each product shipped from warehouses to end-users
during each period.
Quantity of each product shipped from stack buffers directly to
end-users during each period.
Inventory of nished products stored in warehouses at each
period.
Quantity of each product backordered (i.e. shortage or backlogging amount) in each end-user location at the end of each period.
Due to the high number of decision variables, the problem
presented by the PD systems analysis is so complex that optimal solutions are very hard to obtain. The difculties associated
with this type of decision-making can be further amplied by
the complex maze of the network, geographical span of the SC,
and involvement of varied entities with conicting objectives
[23,24]. Indeed, simplication of a real-life scenario in developing a SC model has become unavoidable as most of the complex
PD planning problems are classied under the category of NP-hard
problems [25,26]. Such simplications have led to the development
of a variety of PD planning models in the literature.
3. Complexity-based classication of literature
It is not an easy task to classify the existing literature of integrated PD planning and optimisation. There are three reasons
supporting this difculty [27]:
1. All the developed models consider cost minimisation, prot
maximisation or a combination of both as their objective function.
2. There is a wide variety of assumptions and considerations that
can be made when proposing PD planning models.
3. The literature in the eld is so extensive and a unied body does
not exist.
Few literature surveys on the proposed PD models have
been published [1,25,2744]. Despite the variety of the published
reviews, our survey on the current literature indicates that there is
no specic review on comparing the actual capabilities of the proposed PD models based on their degree of complexity as well as
the solution approaches applied. In this paper, we present a review

B. Fahimnia et al. / Journal of Manufacturing Systems 32 (2013) 119

Fig. 1. A complex real-life integrated PD model.

of recent work on integrated analysis of PD systems based on the


degree of complexity of the models proposed indicating the level
of simplication used. We target the published models integrating the decisions in production and distribution sub-systems in
order to achieve the global optimisation and highlight the benets achieved from the integration of decisions in production and
distribution sub-systems.
We divide the published models into the seven categories (summarised in Fig. 2):
Category 1: Single-product PD models.
Category 2: Multiple-product, single-plant PD models.
Category 3: Multiple-product, multiple-plant, single or no warehouse PD models.
Category 4: Multiple-product, multiple-plant, multiplewarehouse, single or no end-user PD models.
Category 5: Multiple-product, multiple-plant, multiplewarehouse, multiple-end user, single-transport path PD
models.
Category 6: Multiple-product, multiple-plant, multiplewarehouse, multiple-end user, multiple-transport path, no-time
period PD models.
Category 7: Multiple-product, multiple-plant, multiplewarehouse, multiple-end user, multiple-transport path, multiple
time-period PD models.
Fig. 2 illustrates the classication of previous works in the area
of PD planning. Each level of this gure contains the related published works (for that certain level of complexity) which will be
further discussed and analysed in the following sections. This analysis later allows us to identify the important areas where further
research is needed.
To evaluate the degree of complexity for each model, the following dimensions are used in the following sections to examine
each published model:

Type of optimisation (i.e. cost minimisation, prot maximisation,


or multiple objective functions).
Multiplicity of products to be produced at each manufacturing
plant.
Multiplicity of geographically dispersed manufacturing plants.
Multiplicity of machine centres for the production of multiple
products at each manufacturing plant.
The existence of stack buffers at manufacturing plants temporarily storing products prior to their shipment to the warehouses.
Multiplicity of transport paths from plants to end-users
(direct/indirect shipment).
Multiplicity of time-periods.
Regular-time/overtime production and outsourcing alternatives
at manufacturing plants.
Detailed production cost elements in aggregate level for each
plant.
Inventory costs (i.e. WIP inventory costs in manufacturing plants
and inventory of nished products at stack buffers and warehouses).
Shortage/penalty costs of not meeting demand forecast at the end
of each period (also known as backordering or backlogging costs).
Methods applied for modelling and optimisation.

3.1. Single-product models


Haq et al. [22] presented the application of an integrated
productioninventorydistribution planning model in a large fertiliser industry operating in North India incorporating many
realistic conditions such as set-up time and cost, lead times, losses
during production and distribution and recycling of production
losses as well as backlogging. The study limits the manufacturing
of the items to the production of a single product in a set of production sites and disregards production and distribution alternatives
and detailed cost components.

B. Fahimnia et al. / Journal of Manufacturing Systems 32 (2013) 119

Fig. 2. The classication of literature on integrated PD planning.

Chan et al. [45] used a combined hybrid genetic algorithm (HGA)


and analytic hierarchy process (AHP) approach for optimising a
simplied PD problem in multi-factory SCs. Linear programming
was used for the modelling of the problem and AHP for organising and weighting the decision-making criteria. A HGA was then
designed and implemented to determine job allocation for each
plant. A multi-factory production network is investigated in this
study from where products are transported directly to the markets.
Demirli and Yimer [46] proposed a planning model for a buildto-order SC with uncertain cost parameters. The proposed model
was constructed as a mixed integer fuzzy programming (MIFP). This
study considers some key features of a PD model rarely attended
in literature, such as assembly cost components, inventory costs,
and shortage costs. However, it disregards a number of fundamental production characteristics such as the production of multiple
products in multiple time-periods and the distribution of products
from plants to customers through multiple transport paths.
A complex single-product PD model was studied by Yilmaz
and Catay [47]. The major contribution of this study is the capacity expansion characteristic of the developed model. The problem
is formulated as a mixed integer programming (MIP) model and
three LP relaxation-based heuristics are developed to solve the
problem. Considering unlimited inventory holding capacity at manufacturing plants and ignoring the backlogging possibilities and the
transportation of products from DCs to end-users are among the
characteristics which may considerably preclude the model from
functioning effectively in real-life scenarios.
A single-product, single-plant, no-warehouse PD planning
problem was investigated using meta-heuristics to simultaneously
tackle production and routing decisions [48]. The objective function
in the proposed linear integer model minimises the sum of setup,
inventory and transportation costs for determining the optimal
production amount and quantities to be shipped to each customer.
A basic greedy randomised adaptive search procedure (GRASP) was
used as the solution approach improved through a reactive mechanism and a path-relinking process.
Coronado [49] studied a SC optimisation model considering
uncertainty at suppliers nominal capacity. This model tries to
dene the best diversication and safety stock level allocated

to each supplier when minimising the overall expected SC cost.


This model incorporates the cost components for regular-time and
overtime production of goods and proposes some other important characteristics rarely attended in the literature (e.g. all types
of inventory costs). Backlogging issues and the transportation of
multi-products from multi-plants to multi-end-users are disregarded in this model.
Hamedi et al. [50] proposed a hierarchical solution algorithm to
solve a six-level, single-objective, multi-period, and single-product
SC problem in the sense that the solutions of one section are used as
input for the next section. The proposed mixed integer linear programming (MILP) mathematical model was tested for a few small
and medium size samples in real world gas industries.
A two-phase approach centring on reactive tabu search algorithm was developed by Bard and Nananukul [51] to solve a
simplied PD planning problem incorporating a single production plant producing a single product and distributing the nished
products directly to end customers. The objective function minimises the sum of production setup costs, routing costs, holding
costs at the plant, and holding costs at the customer sites.
Table 1 summarises and compares the characteristics of the
proposed models in Category 1. The number of ticks in front of a
model indicates the dimensions/characteristics accommodated in
the concerned PD model.
3.2. Multiple-product, single-plant models
Models classied in this category can only work in the PD scenarios with a single manufacturing plant. This signicantly limits
the applicability of the developed models because the globalisation
and merging processes have encouraged the manufacturing companies to develop multi-plant policies in order to improve their
efciency [52]. Pyke and Cohen [53] examined the performance
characteristics of a simple integrated PD system comprised of
a single work centre at a factory producing multiple products, a
stockpile of nished goods, and a single retailer. The algorithm
is able to compute expedite and replenishment inventory control
policies for the entire chain. The large number of decision variables
precluded the extensive accuracy testing of the algorithm.

Table 1
Classication of literatureCategory 1.
Author(s), year

Haq et al., 1991

Dimensions of the proposed model


TCM: total cost
minimisation
PM: prot
maximisation
MO: multiple
objectives

Multiple
product
types

TCM

. . .a

TCM

...

Demirli and
Yimer, 2006
Yilmaz and
Catay, 2006

TCM

...

TCM

...

Boudia et al.,
2007

TCM

...

Coronado, 2008

TCM

...

Hamedi et al.,
2009

TCM

...

Bard and
Nananukul,
2009

TCM

...

a
b
c

...

...

Multiple
machine
centres

Stack
buffers
in
plants

...

...

Multiple
DCs

Multiple
end-users

...

Multiple
transport
paths

...

...

...

...

...

...

...

...

...

...

...

...

...

...

...

...

...

...

...

...

...

Multiple
time-periods

Production
alternatives

Production
cost
elements

...

...

Inventory
costs

Shortage/
penalty
costs

...

...

...

...

...

...

...

...

...

Xc

...

...

...

...

...

...

...

...

...

...

...

...

...

...

...

...

A mixed 01 integer
programming
Linear programming
and a combined hybrid
genetic algorithm and
analytic hierarchy
process
MIFP model
MIP model solved with
three LP
relaxation-based
heuristics
A GRASP and two
improved versions
using a reactive
mechanism or a
path-relinking process
A nonlinear
programming
formulation and a
heuristic to decompose
the original
formulation into two
easy-to-solve LP
problems
A hierarchical solution
algorithm and a MILP
coded using LINGO
8.00
A two-phase approach
centring on reactive
tabu search algorithm

B. Fahimnia et al. / Journal of Manufacturing Systems 32 (2013) 119

Chan et al., 2005

Multiple
manuf.
plants

Methods applied

The characteristic is NOT considered in the developed model.


The characteristic is considered in the developed model.
The characteristic is PARTIALLY considered in the developed model.

B. Fahimnia et al. / Journal of Manufacturing Systems 32 (2013) 119

A MIP-based PD model was presented in [2] for which the


Lagrangian relaxation was used to accommodate the production
and distribution sub-problems, and sub-gradient optimisation was
implemented to coordinate the information ow in a hierarchical manner. This study extends the scheme by a heuristic method
which modies distribution decisions if capacity shortages occur
at the production stage [54].
A hierarchical PD planning approach was developed for a single multinational factory transporting multiple product families to
multiple warehouses or chain stores [55]. The approach attempts
to solve the problem optimally by aggregating the time periods and
product families. The obtained aggregate optimal solution for the
model is disaggregated for a single period on a rolling horizon basis
to reduce the problem size.
Lee and Kim developed one of the most generic PD models
in the literature and proposed a specic problem-solving procedure using a hybrid approach combining analytic and simulation
methods [56,57]. In the proposed model, the rst shop produces
a number of parts which are used in the production of multiple
products at the second shop. There is only a single plant modelled
with two shops and a single stack point. The production and distribution capacity constraints in the proposed analytic model are
considered as stochastic variables adjusted in accordance with general PD characteristics obtained from a simulation model. Linear
programming was used for the problem formulation minimising the production, distribution, inventory holding, and shortage
costs. An LP solver was used to implement the model and ARENA
was employed as the simulation tool. This study does not investigate a multi-plant scenario and disregards a detailed production
plan.
A PD planning problem between a manufacturing location and
a DC was examined by Rizk et al. [58], in which multiple parallel
machine centers at the manufacturing location, economies of scale
on transortation costs, as well as dynamic demand at plants and distribution centres were taken into consideration. A MIP model of the
production process and three different formulations representing
the general piecewise linear transportation functions were used to
develop three equivalent mathematical programming models. The
research investigates the impact of choosing a suitable mathematical formulation on the problem-solving time.
Nishi et al. [59] studied a distributed decision-making system for
the integrated optimisation of production scheduling and distribution planning. An integrated optimisation model was formulated
using MILP. The developed model was then decomposed into production scheduling and warehouse planning sub-problems using
an augmented Lagrangian approach. A distributed optimisation
system was developed to solve the sub-problems by gradually generating the feasible solutions through repeatedly exchanging data
obtained at each sub-system (data update) to accommodate the
modications caused by unforeseen changes. The study formulates
the production processes at a single plant and considers a single
warehouse comprising a number of storage areas. There are no
end-users involved in this model and production/transportation
alternatives are simplied.
Farahani and Elahipanah [60] solved a MILP bi-objective model
for a just-in-time (JIT) distribution planning of a three-echelon
SC using multi-objective Genetic Algorithms (GAs). Two functions
were considered for optimisation: cost minimisation as well as
minimisation of the sum of backorders and surpluses of products
in all periods. In fact, the second objective function represents the
JIT delivery and minimises the earliness and tardiness of product deliveries. Delivery lead-times and capacity constraints were
considered for a multi-period, multi-product and multi-channel
network. Since the study primarily concerns the distribution of
items from suppliers to retailers, this model replaces the xed and
variable production costs by the purchasing costs and completely

disregards the production issues and the multiplicity of manufacturing plants.


A new solution approach was proposed by Safaei based on the
integration of mathematical and simulation techniques to solve
an integrated multi-product, multi-period, multi-site PD planning problem [61]. A MILP model was developed for formulating
the problem. In this study, to consider the stochastic factors,
a hybrid mathematical-simulation approach was proposed consisting of independent mathematical and simulation models. The
proposed model can be treated as a single-plant model with multiple machine centres and many characteristics of a real world PD
planning problem are disregarded in this model (e.g. detailed production alternatives, production cost elements, backlogging costs,
and inventory management issues).
The characteristics of the proposed models in Category 2 are
outlined in Table 2.

3.3. Multiple-product, multiple-plant, single or no warehouse


models
Mohamed [62] studied the PD planning for multinational
corporations. Production planning and logistics decisions for
multi-national companies operating under varying ination and
exchange rates was analysed in this research. The study models
the production of multiple products in multiple multinational factories and the distribution of produced items directly to the target
markets.
A new approach was proposed by Gen and Syarif [63] called
spanning tree-based hybrid genetic algorithm (hst-GA) to solve
multi-time period production, distribution and inventory problems. The proposed model integrates design and planning decisions
in a SC network. Facility location decisions, distribution costs, and
inventory management issues were taken into consideration. In
order to improve the efciency of the proposed GA, a Fuzzy Logic
Controller (FLC) was hybridised to the evolutionary process for the
auto-tuning of the GA parameters. In this study, however, a real
world PD network is simplied by disregarding production alternatives and cost components as well as the indirect transportation
of products from plants to customers.
An integrated approach for planning steel production in a
major Canadian steel making company was discussed by Chen
and Wang [64]. In the proposed linear programming raw materials are shipped to the central steel plant and transformed into
semi-nished products with different specications corresponding to the types of the end products to be produced. Semi-nished
products are transported from central plant to the other factories,
from where the nished products are transported to customers.
There are no distribution centres and inventory costs (from any
type) involved in this model.
Kanyalkar and Adil [65] developed a single linear programming
model to generate an integrated aggregate and detailed PD plan
in a multi-site production environment. Also in 2007 and later
in 2008, authors proposed a robust optimisation model incorporating demand uncertainty for integrated aggregate planning of
multi-site procurementproductiondistribution system [66,67].
In the proposed complex model, there is a supplier providing
the plants with the required raw material, from where products
are shipped to the manufacturing plants accordingly. Demand is
considered to be stochastic and can be described in different scenarios. In fact, the developed model may not be fully integrated,
because the distribution decisions are only responsible for corrective actions and do not have direct consequences on the production
decisions.
Table 3 summarises and compares the characteristics of the proposed models in this category.

Table 2
Classication of literatureCategory 2.
Author(s), year

Dimensions of the proposed model

Pyke and Cohen,


1993, 1994

TCM

Barbarosoglu and
Ozgur, 1999

TCM

Ozdamar and
Yazgac, 1999
Lee and Kim,
2000, 2002

TCM

Rizk et al., 2006

TCM

TCM

Nishi et al., 2007

TCM

Elahipanah and
Farahani, 2008

TCM

Safaei et al., 2009

TCM

Multiple
product
types

Multiple
manuf.
plants

Multiple
machine
centres

...

...

...

...

...

...

...

...

...

...

...

Stack
buffers
in
plants

...

...
...

...

...

...

...

...

...

Multiple
DCs

Multiple
end-users

Multiple
transport
paths

...

...

...

...

...

...

...

...

...

Multiple
time-periods

...

...

...

Production
alternatives

Production
cost
elements

...

...

...

...

...

...

...

...

...

Inventory
costs

Shortage/
penalty
costs

...

...

...

...

...

...

...

...

Using the
approximation of
steady state
distributions of key
random variables
A MIP model solved
using Lagrangian
heuristic and sub
gradient optimisation
A hierarchical PD
planning approach
A hybrid
analytic-simulation
approach using linear
program, General
Algebraic Modelling
System and ARENA
simulation package
A MIP model solved
with CPLEX
A MILP and using a
Lagrangian approach to
decompose the
problem into
sub-problems
A MILP model solved
using a multi-objective
GA
A hybrid
mathematicalsimulation
approach

B. Fahimnia et al. / Journal of Manufacturing Systems 32 (2013) 119

TCM: total cost


minimisation
PM: prot
maximisation
MO: multiple
objectives

Methods applied

B. Fahimnia et al. / Journal of Manufacturing Systems 32 (2013) 119

...
...

...

...
...

TCM
Kanyalkar and
Adil, 2008

TCM
Gen and Syarif,
2005

...

...
X
...
...

...

...
...

...
...
...
...
MO

...

...
...
X
...

...

...
...
...

Chen and Wang,


1997
Mohamed, 1999

PM

Multiple
manuf.
plants
Multiple
product
types
TCM: total cost
minimisation
PM: prot
maximisation
MO: multiple
objectives

Dimensions of the proposed model


Author(s), year

Table 3
Classication of literatureCategory 3.

Multiple
machine
centres

Stack
buffers
in
plants

Multiple
DCs

Multiple
end-users

Multiple
transport
paths

Multiple
time-periods

Production
alternatives

Production
cost
elements

Inventory
costs

Shortage/
penalty
costs

Methods applied

A linear programming
approach
Analysing different
mathematical models
for different scenarios
Solving a PD LPM
using spanning
tree-based hybrid
genetic algorithms and
fuzzy logic controller
(FLC)
A linear mathematical
formulation solved
using GLPK solver

3.4. Multiple-product, multiple-plant, multiple-warehouse,


single/no end-user models
Cohen and Lee [68] presented a strategic modelling framework and a hierarchical decomposition approach to analyse the
interactions between functions in a SC. They consider four linked
approximate sub-modules each representing a part of the overall
SC: material control, production control, product stockpile, and distribution network control. In a hierarchical decomposition, each
sub-model was heuristically optimised and the output of a submodel solution is used as the input data to others. The model,
however, relies on the non-tested approximations to characterise
and solve the random variables describing the linkages between
most locations [53].
An integrated two-layer model was developed by Tang and Yung
[69] in which the rst layer integrates the production assignment
and lot sizing problems and the second layer integrates decisions
of transportation and order quantity determination. A two-layer
decomposition method is developed for solving the proposed
model combining two heuristics: assignment and transportation.
The proposed model is a typical supplier-warehouse model and
the contribution of the paper is built around the proposed solution
approach by which several medium and large-scale test problems
are solved.
Tasan [70] proposed a two-step approach for the developing of
a PD plan. A MIP model is developed for determining the production quantities and allocation of this production to demands. The
author then uses the outputs of the rst step as inputs to the next
and decides how the allocated product quantities are transported to
the DCs with vehicle routing consideration. GA is employed as the
vehicle routing tool and the model is executed in GA Based Routing Application (GABRA). The proposed model is, however, a basic
single-echelon PD plan disregarding many key aspects of production plans as well as the distribution of items from distribution
centres to the customers.
Kanyalkar and Adil [71] proposed a mixed integer linear goal
programming model for an integrated procurement, production
and distribution planning problem comprising a system with multiple parallel production plants supplied from multiple suppliers
and serving multiple DCs. To reduce the computational burden, the
model uses different time-grids and planning horizons (long-term
and short-term) for aggregate and detailed procurement, production and distribution planning. The goal programming formulation
is solved heuristically using weighted and pre-emptive methods,
implemented in Linux based Gnu Linear Programming Kit (GLPK)
solver. This model simplies the detailed production/outsourcing
alternatives and cost components and disregards the complexity of
employing multiple transport options.
A GA was developed for solving an integrated PD planning problem by Park et al. [3]. The problem studied was a
multi-product, multi-supplier, multi-plant, and multi-DC planning problem. Although integration of production and distribution
activities is the main target, the procurement and inventory management issues were also properly accommodated in this model.
The developed model disregards the transportation of items from
DCs to customers and simplies many production planning issues
faced in real world scenarios.
A fuzzy multi-objective linear programming model was developed by Liang [72] to solve integrated multi-product and
multi-time period PD planning problems. Two objectives were
designed to minimise the total system-wide costs and total delivery time. A piecewise linear membership function was adopted to
represent the fuzzy objectives of the decision-maker for the proposed PD problem, and to achieve more exible doctrines via an
interactive decision-making process. However, considering multiple machine centres and stack buffers at plants, multiple end-users,

B. Fahimnia et al. / Journal of Manufacturing Systems 32 (2013) 119

the direct/indirect transportation of products from plants to endusers, and also taking into consideration the detailed production
cost elements for every single product could considerably increase
the complexity of the developed model by Liang.
A fuzzy mathematical programming based approach was proposed by Bilgen [73] to solve an integrated PD planning problem
incorporating multiple production lines, multiple plants and multiple DCs. The model was rst formulated using MILP and three
FMIP models (corresponding to different aggregation operators)
were further developed to facilitate the embodied uncertainty
factors and fuzziness of constraints, objectives and parameters.
Although the proposed model introduces the inclusion of such
realistic factors as production lines in manufacturing plants, many
characteristics of a real world model are disregarded in this study
(refer to summary Table 4).
A fuzzy production planning model integrating procurement
and distribution plans was developed by Torabi and Hassini [74]
incorporating four objectives simultaneously: (1) minimisation of
total cost of logistics, (2) maximisation of total purchasing value,
(3) minimisation of defective items, and (4) minimisation of late
deliveries. The initial model was formulated using a multi-objective
MILP. To capture the inherent fuzziness of the critical data and
the imprecise nature of the objectives aspiration levels, an interactive fuzzy goal programming formulation was proposed which
was then converted into an auxiliary crisp formulation for nding
a compromise solution. This study is the extension of the authors
previous work, in which a fuzzy single-plant bi-objective possibilistic programming model was proposed to formulate a SC master
production schedule [75]. The proposed models, however, aim to
incorporate the decisions at a tactical level (i.e. mid-term SC planning) and do not include the operational decisions at production
stage as appears in an aggregate production plan. Further, endusers were not considered in these models as a SC node and hence
the transportation issues were only considered up to the DC level.
Table 4 compares the characteristics of the models in Category
4.
3.5. Multiple-product, multiple-plant, multiple-warehouse,
multiple-end user, single-transport path models
An industrial PD planning problem was studied by DhaenensFlipo and Finke [52] proposing a MILP for the formulation of the
model and using CPLEX commercial linear programming codes as
the solver. In this study multiple states and lines are considered
in each production facility each dedicated to the production of a
certain product. However, multiple machine centres in the production line of each product, stack buffers at manufacturing plants,
backlogging costs, detailed production cost elements and production/distribution alternatives, all the characteristics of a complex SC
system, were not taken into consideration in the proposed model.
Chen and Lee [76] investigated the simultaneous optimisation
of multiple objectives in a typical SC with uncertain product prices
under market demand uncertainties. Demand uncertainty is modelled as discrete scenarios with given probabilities for different
expected outcomes. Uncertainties are described as fuzzy variables.
The model is constructed using Mixed Integer Nonlinear Programming (MINLP) to achieve conicting objectives. To nd the degree
of satisfaction of the multiple objectives, a fuzzy decision-making
method is proposed, where the nal decision is acquired by fuzzy
aggregation of the fuzzy goals and the fuzzy product prices, and
the best compromised solution is derived by maximising the overall
degree of satisfaction for the decision. Although a multiple-product
model is developed in this study, every plant batch-manufactures
only one single product at one period. Also, detailed production
aspects at each plant (e.g. considering capacity of machine centres
and stack buffers, and outsourcing opportunities) and the direct

distribution of products from plants to end-users are ignored in


the proposed model.
Lim et al. [77] presented a simulation approach for the PD
planning based on the work of Jang et al. [78] to determine the
capacities of facilities in a PD network considering three replenishment policies. The study presents a mathematical model to decide
the capacity of facilities and uses Microsoft Excel premium solver to
test the developed model. It also develops a simulation model using
a SC analyser to execute a PD plan for higher customer satisfaction and lower total relevant costs. The developed mathematical
and simulation model were applied to a simple test problem. In
this study, the focus is given to replenishment policies and considering different production/transportation alternatives and cost
components are ignored.
Aliev et al. proposed a fuzzy integrated formulation for a
multi-period and multi-product aggregate PD planning problem
[79]. The model was formulated in terms of fuzzy programming
to accommodate the inherent uncertainties in market demands,
imprecise process times and other related factors. The optimal
solution was then sought using GA. Production and distribution
alternatives and production cost elements are disregarded in this
model.
In research by Selim and Ozkarahan [54], collaborative decisionmaking was studied with the objectives of prot maximisation
and cost minimisation in a PD planning problem. A multiobjective linear programming model was developed for this
concern. Decision-makers imprecise aspiration levels for the goals
were incorporated into the model using a fuzzy goal programming
approach. In this article, centralised and decentralised SC scenarios are studied and computational experiments are provided on a
case problem. This model does not take into account a single objective (i.e. cost minimisation or prot maximisation), but instead
uses a weighted additive approach and gives the decision-maker
the opportunity to decide which partner will possess the dominant weight. Different SC participants are considered as different
entities seeking separate objectives. Considering other operational
characteristics in this model could make it far more applicable for
implementation in different SC environments.
A combinational PD planning problem was studied by Kazemi
et al. with the objective to evaluate the performance of a GA-based
multi-agent system compared to a Lagrangian relaxation approach
[80]. The multi-agent system approach was adopted as a solution methodology because of the difculties associated with the
optimisation of a complex integrated PD planning problem. In
the proposed approach, three GA models with different crossover
methods were developed playing the role of three agents. The three
agents work independently with the objective of enhancing the
quality of solutions achieved in the preceding generation by the
other agents. The approach was demonstrated to generate better solutions in large-scale PD planning problems compared to
Lagrangian relaxation.
A two-level mathematical model was proposed by Das and Sengupta [81] for simulating SC strategic and operational planning
addressing various uncertainties that a multinational corporation
may encounter due to changes in government regulations. Uncertainties incorporated in this model include the changes in the
expected cost of input material, border crossing costs, tariffs, tax
policies, as well as the variations in demand and transportation
times. At the strategic level, the model maximises the overall SC
costs through the selection of plants, quantity and type of products to be produced at each plant, the allocation of the plants to
DCs and nally the allocation of DCs to customer locations to satisfy market demands. At the operational level, the model minimises
inventory costs at DCs addressing the transportation time uncertainty. The sensitivity analyses results indicate that the changes in
the selection of plant set and supply quantity occur when the input

10

Table 4
Classication of literatureCategory 4.
Author(s), year

Dimensions of the proposed model


TCM: total cost
minimisation
PM: prot
maximisation
MO: multiple
objectives
TCM

Tang and Yung,


2004

TCM

Tasan, 2006

TCM

Kanyalkar and
Adil, 2005,
2007, 2008

TCM

Park et al., 2007

TCM

Liang, 2008

TCM

Bilgen, 2009

TCM

Torabi and
Hassini, 2009

MO

Multiple
manuf.
plants

Multiple Stack
machine buffers
in plants
centres

Multiple
DCs

...

...

...

...

...

...

...

...

...

...

...

...

Multiple
end-users

Multiple
transport
paths

Multiple
time-periods

Production
alternatives

Production
cost
elements

...

...

...

...

...

...

...

...

...

...

...

...

...

...

...

...

...

...

...

...

...

...

...

...

...

...

...

Inventory
costs

Shortage/
penalty
costs

...

...

...

...

...

...

...
...

...

...

Four linked,
approximate
sub-models
heuristically optimised
using hierarchical
decomposition
approach
A non-linear
mathematical model
solved using a
two-layer
decomposition
solution approach.
A MIP model solved
using GAs
A mixed integer linear
goal programming
model solved
heuristically using
weighted and
pre-emptive methods
A linear mathematical
model solved using
GAs
A fuzzy two-objective
mathematical
programming model
with piecewise linear
membership function
A MILP model
equipped with three
FMIP models
accommodating the
embodied uncertainty
factors
A multi-objective fuzzy
goal programming
formulation solved
through an auxiliary
crisp formulation

B. Fahimnia et al. / Journal of Manufacturing Systems 32 (2013) 119

Cohen and Lee,


1988

Methods applied

Multiple
product
types

B. Fahimnia et al. / Journal of Manufacturing Systems 32 (2013) 119

resource costs change, triggered by the changes in government


regulations.
A two-phase MILP model and a GA-based solution procedure
was proposed for the scheduling of a build-to-order SC [82].
The original complex problem was decomposed into two submodels: the rst model concerns the assembling and distribution
planning of nal products as per customer order, while the second model involves the production scheduling and raw-materials
procurement based on the outputs of the rst model. The implementation of the proposed GA-based approach in solving a set of
test problems was compared with the performance of mathematical programming approach (using LINGO). The results revealed
that the solutions achieved from both approaches are identical
for small-scale problems while LINGO could not converge to optimal solution for medium-size problems and failed to provide any
solution to the large test problems. The proposed model does not
provide a single optimisation algorithm for planning the entire SC
activities, but instead it decomposes the problem into two subsystems and evaluates them sequentially.
The characteristics of the models presented in Category 5 are
highlighted in Table 5.
3.6. Multiple-product, multiple-plant, multiple-warehouse,
multiple-end user, multiple-transport path, no-time period models
There are only few models in the literature which can t in
this category and this indicates the need for the development of
more complex PD models able to accommodate the unaddressed
areas, particularly in Categories 6 and 7. A real-life case study was
studied by Gunnarsson et al. [83] for modelling the entire SC of
a major European pulp mill company. A mathematical model was
proposed for the planning of the entire SC, using the commercial
CPLEX as solver network all the way to the nal destinations. The
major advantages of the proposed model compared to the former
models are the accurate representation of both demand and supply sides and the optimisation of distribution model. This model
was designed for a single time-period SC and does not take into
consideration the inventory management aspect of a complex PD
model.
Ferrio and Wassick [84] presented a MILP capable of redesigning and replanning an existing chemical supply network made
up of production sites, an arbitrary number of echelons of distribution centres, and customer sites. The mathematical model was
implemented by GAMS/CPLEX solver. Although a single-period SC
network was modelled by the authors and such crucial factors as
inventory management, backlogging of products at customer centres, and detailed production cost elements were disregarded, the
logistics costs are considered in more detail compared to those
formerly reported in the literature.
An integrated model was proposed for the collaboration among
fuel reneries manufacturing multiple fuel products at different
locations [85]. Fuel products are transported from the reneries
to multiple distribution centres and from there to different target
markets. A MINLP was proposed to formulate the objective function
maximising the prot for the integrated SC. Three real world case
studies presented in this study demonstrate the potential benets
resulting from collaborative planning for production and distribution of fuel products (19.7%. reduction in overall delivery costs
through the collaborative network planning).
Table 6 compares the characteristics of Category 6 models.
3.7. Multiple-product, multiple-plant, multiple-warehouse,
multiple-end user, multiple-transport path, no-time period models
In terms of multiplicity, this category contains the most complex models in the area of integrated PD planning. To the best of

11

our knowledge, the sole paper that could t in this category is the
study conducted by Fahimnia et al. [86]. Based on the integration of
Aggregate Production Plan and Distribution Plan, the authors developed a MILP for a two echelon SC considering several real world
variables and constraints. GA was used for the optimisation of the
developed model and numerical results were presented for a real
world medium size case problem. The authors failed to test the
proposed approach on solving large-scale case problems. Table 7
summarises the characteristics of the sole model in this category.
4. Solution-based classication
For a complex realistic PD planning problem, the size of the
search space could become extremely large. In a real-life scenario
hundreds of types of products may be manufactured in a set of
manufacturing plants and the nal products distributed to several
end-users via a number of warehouses. In such cases, the search
space for nding the optimal PD plan may contain thousands of
discrete planning options. For this reason, selecting an effective
optimisation technique for solving a complex real life SC optimisation model is so vital and has always been a real challenge in past
research [26].
The survey on the current literature indicates that many techniques have been proposed for the optimisation of an integrated
PD plan. Each of these techniques has its own strengths and weaknesses and can be helpful in solving certain types of PD planning
problems. Table 8 summarises the solution approaches used in
previous models to solve PD planning problems. These proposed
solution techniques can be classied into four categories: mathematical techniques, heuristics techniques, simulation, and GAs.
4.1. Mathematical techniques
Mathematical techniques include linear programming, nonlinear programming, mixed integer programming, and Lagrangian
relaxation [8789]. Different mathematical techniques have
been adopted to solve SC problems. These include Linear
Programming models [6467,72], Mixed Integer Programming models [22,46,50,52,54,58,62,73,76,81,8385,9093], and Lagrangian
Relaxation models [2,59,94,95].
Mathematical programming models have been demonstrated to
be useful analytical tools in optimising decision-making problems
such as those encountered in SC planning [96,97]. Linear programming (LP) is applicable when all of the underlying models of the real
world processes are linear [89,98]. MIP is used when some of the
variables in the model are real values (fractional values) and others
are integer values (0, 1). MILP occurs when objective function and
all the constraints are in linear form, otherwise it is called MINLP
which is much harder to solve [88]. The idea behind the Lagrangian
relaxation methodology is to relax the problem by removing the
constraints that make the problem hard to solve, putting them into
the objective function, and assigning a weight to each constraint
[99]. Each weight represents a penalty which is added to a solution
that does not satisfy the particular constraint. All these techniques
are fully matured and thus guaranteed to produce the optimal solution or near optimal solutions for a certain type of problem [26].
There are two issues that restrict the application of mathematical modelling in solving complex real world SC planning problems:
1. Mathematical equations are not always easy to formulate, and
the associated complexities in the development of mathematical algorithms increase as the number of constraints increases
[26,100]. Since most of the realistic SCs are complex in nature
with a high number of variables and constraints involved, mathematical optimisation methods such as LP and MIP may not be

12

Table 5
Classication of literatureCategory 5.
Author(s), year

Dimensions of the proposed model


TCM: total cost
minimisation
PM: prot
maximisation
MO: multiple
objectives
TCM

MO

Lim et al., 2006

TCM

Aliev et al., 2007

PM

Selim et al., 2008

MO

Kazemi, et al.,
2009

TCM

Das and
Sengupta, 2009

MO

Yimer and
Demirli, 2009

TCM

Multiple
manuf.
plants

Multiple Stack
machine buffers
centres in plants

...

...

...

...

...

...

...

...

...

...

...

...

...

...

...

Multiple
DCs

Multiple
end-users

Multiple
transport
paths

...

...

...

...

...

...

...

...

Multiple
time-periods

Production
alternatives

Production
cost
elements

Inventory
costs

Shortage/
penalty
costs

...

...

...

...

...

...

...

...

...

...

...

...

...

...

...

...

...

...

A MILP formulation
solved with CPLEX
A MINLP and a fuzzy
decision-making
method to nd the
degree of satisfaction
of multiple objectives
A mathematical model
solved using a supply
chain simulation
optimiser (SCA)
A fuzzy programming
model solved using
GAs
A multi-objective
linear programming
model solved with a
Fuzzy mathematical
programming
approach
Comparing the
performance of a
GA-based multi-agent
approach with the
Lagrangian relaxation
app.
A two-level
mathematical model
solved with LINGO
A GA based solution
procedure is compared
with the performance
of mathematical
techniques

B. Fahimnia et al. / Journal of Manufacturing Systems 32 (2013) 119

Dhaenens-Flipo
and Finke,
2001
Chen and Lee,
2004

Multiple
product
types

Methods applied

Table 6
Classication of literatureCategory 6.
Author(s), year

Dimensions of the proposed model


TCM: total cost
minimisation
PM: prot
maximisation
MO: multiple
objectives

Kim et al., 2008

PM
TCM

PM

Multiple
manuf.
plants

Multiple Stack
machine buffers
centres in plants

...

...

...

...

...

Multiple
DCs warehouses)

Multiple
DCs

Multiple
transport
paths

Multiple
time-periods

Production
alternatives

Production
cost
elements

Inventory
costs

Shortage/
penalty
costs

...

...

...

...

...

...

...

...

...

...

...

...

...

...

...

A mathematical model
solved using CPLEX
A MILP model
implemented using
GAMS/CPLEX
A MINLP model solved
using GAMS/DICOPT

Table 7
Classication of literatureCategory 7.
Author, year

Dimensions of the proposed model


TCM: total cost
minimisation
PM: prot
maximisation
MO: multiple
objectives

Fahimnia et al.,
2011

TCM

Methods applied

Multiple
product
types

Multiple
manuf.
plants

Multiple Stack
machine buffers
centres in plants

Multiple
DCs warehouses)

Multiple
DCs

Multiple
transport
paths

Multiple
time-periods

Production
alternatives

Production
cost
elements

Inventory
costs

Shortage/
penalty
costs

B. Fahimnia et al. / Journal of Manufacturing Systems 32 (2013) 119

Gunnarsson
et al., 2007
Ferrio and
Wassick, 2008

Multiple
product
types

Methods applied

A MILP model solved


using genetic
algorithms

13

14

B. Fahimnia et al. / Journal of Manufacturing Systems 32 (2013) 119

Table 8
Techniques and tools used for the development of PD optimisation models.
Techniques/tools
Mathematical techniques
Linear programming

Mixed integer programming

Lagrangian relaxation

Heuristic techniques

Authors
Chen and Wang (1997)
Kanyalkar and Adil (2005, 2007, 2008)
Liang (2008)
Haq et al. (1991)
Mohamed (1999)
Dhaenens-Flipo and Finke (2001)
Yan et al. (2003)
Bhutta et al. (2003)
Chen and Lee (2004)
Demirli and Yimer (2006)
Rizk et al. (2006)
Gunnarsson et al. (2007)
Paksoy et al. (2007)
Tsiakis and Papageorgiou (2008)
Kim et al. (2008)
Ferrio and Wassick (2008)
Selim et al. (2008)
Hamedi et al. (2009b)
Das and Sengupta (2009)
Bilgen (2009)
Barbarosoglu and Ozgur (1999)
Jayaraman and Pirkul (2001)
Syam (2002)
Nishi et al. (2007)
Cohen and Lee (1988)
Pyke and Cohen (1993, 1994)
Ozdamar and Yazgac (1999)
Yeh (2005)
Yilmaz and Catay (2006)
Coronado (2008)
Tang and Yung (2004)

Simulation

Lee and Kim (2000, 2002)


Ritchie-Dunham et al. (2000)
Jain et al. (2001)
Williams et al. (2003)
Yan et al. (2003)
Moyaux et al. (2004)
Sarker et al. (2005)
Lim et al. (2006)
Safaei (2009)

Genetic algorithms

Syarif et al. (2002)


Chan et al. (2005)
Gen and Syarif (2005)
Altiparmak et al. (2005, 2006, 2007)
Yeh (2006)
Tasan (2006)
Aliev et al. (2007)
Park et al. (2007)
Elahipanah and Farahani (2008)
Kazemi et al. (2009)
Yimer and Demirli (2009)
Fahimnia et al. (2011)

very effective in solving real world SC planning problems [26,87].


For this reason, mathematical techniques are only suitable for
solving small to medium size PD planning problems with limited variables and constraints.
2. Even if it is possible to translate an immense and difcult problem into mathematical equations, the problem would become
intractable or NP-Hard due to the exponential growth of the
model size and complexity [3,26]. According to Dantzig [98], the
father of linear programming, high computer memory and long
CPU time is required in order to process complex mathematical
algorithms. This makes it almost impossible using mathematical
techniques to cope with large real-life PD planning problems,
unless oversimplied.

4.2. Heuristic techniques


The limitations of mathematical techniques as outlined in Section 4.1 have forced the use of heuristics in nding feasible solutions
for large-scale SC planning problems. Heuristic methods are generally experience-based techniques that help in problem-solving
and overcome many shortcomings of traditional optimisation techniques. A heuristic method is normally used to rapidly nd a
solution that is hoped to be close to the optimal solution. There have
also been many attempts in literature for solving PD problems
using other heuristic techniques [47,49,53,55,68,101,102]. There
are, however, two reasons why heuristic techniques are not always
considered as an effective method for the optimisation of complex
integrated PD plans:
1. Heuristic techniques do not promise an optimal solution to the
problem. As opposed to mathematical techniques, heuristics do
not guarantee the optimal solutions but are usually (not always)
able to give a good acceptable solution in solving complex problems [103].
2. In complex SC planning problems with vast search spaces,
heuristic techniques (e.g. Simulated Annealing) are not always
effective in locating the global optimal or near optimal solutions.
Although the Simulated Annealing approach can often manage
to make its way through the traps of local optima, its ability and
efciency in exploring the search space is highly limited by its
characteristic of examining only one point of the space at a time
[104].
4.3. Simulation modelling
Simulation modelling in the area of SCM is used to observe how
a real system performs, diagnose problems and predict the effect
of changes in the system, evaluate SC activities, and suggest possible solutions for improvements [105]. Simulation techniques can
be ideal for reproducing the behaviours of complex systems [105].
Many previous studies have analysed the capability of simulation
modelling in SC modelling and optimisation [5,61,105121].
Because of many inuential sources of stochastic variation in
todays SCs, simulation can be a highly effective tool in making
operationally and economically sound business decisions [122].
While this technique is capable of describing various real and
detailed situations, the following can justify the limited application
of this methodology for the optimisation of complex PD planning
problems [3,26]:
1. It is difcult to search for an optimal value using simulation techniques. Like heuristic techniques, the rst and primary drawback
of a simulation modelling is its inability to guarantee optimality
of the developed solution. In many cases, even nding the near
optimal solution cannot be guaranteed by this method.
2. It is costly and takes much time and effort to analyse the obtained
results. Simulation software packages are generally very expensive to purchase and very time-consuming to analyse the outputs
(i.e. auto-generated codes or reports).
4.4. Genetic algorithms
Introduced by Holland [123], GAs are stochastic algorithms categorised in the class of general-purpose search methods which
simulate the processes in a natural evolution system [12,124].
GAs combine directed and stochastic search methods and are able
to achieve a good balance between exploration and exploitation
of the search space [86]. GAs have been proven to be a highly
effective and efcient tool in solving complex engineering and
manufacturing problems and some of their successful applications

B. Fahimnia et al. / Journal of Manufacturing Systems 32 (2013) 119

in the optimisation of SC models have been proposed in literature


[3,45,60,63,70,72,79,80,82,86,125129].
The advantages of using GA techniques for solving large optimisation problems are their robustness, searching exibility and their
evolutionary nature [104,130]. GAs are principally able to search
large, complicated and unpredictable search spaces which facilitates this technique to locate the optimal solution demonstrated
by the convergence of the tness function as the number of evolutions increases [26,123,131133]. A GA produces a large population
of solutions, for each of which the evaluation of the tness function
is sought; hence, a parallel computer is required when running GA
for very large-scale optimisation problems [134,135].
There are, however, a number of challenges when designing a
customised GA procedure to solve a certain PD planning problem
[136]. The rst challenge in developing a GA-based optimisation
technique is to form the chromosome structure which accordingly
affects the whole GA procedure. Based on the size and nature of
the problem, dissimilar PD planning problems require different
chromosome representations and therefore the existing GA procedures cannot be used to solve different problems from various
complexity levels. The second difculty is the construction of customised genetic operators to perform the mating process on the
chromosomes. Lastly, designing a constraint handling mechanism
is generally a complicated task in order to ensure the effective
implementation of the model constraints [86].
5. Implications for the future of PD planning and
optimisation
In this section, based on our ndings, we rst outline the
observations and modelling implications at each category, discover
research gaps, and highlight trajectories and trends which will
enable the identication of future research directions. The general observations and managerial insights for PD planning and
optimisation is next highlighted in this section.
5.1. Category-based modelling implications
The classication of literature presented in this paper has important managerial and modelling implications for SC academics and
practitioners. The category-based implications are discussed in the
following subsections.
5.1.1. Category-based observations
Single-product PD models are not currently drawing much
research attention unless they t the purpose in a sector specic study. For instance, single-product models are broadly used
in modelling gas networks and crude oil SCs. Apart from such
sector specic studies, multi-product PD models may generally
have wider real world applications.
Development of single-plant PD models is still a requirement
in monopolistic markets in which a supplying rm is the sole
provider of a certain type of product (a typical scenario in
telecommunication and media markets). In some other cases,
solitariness of manufacturing plants can be a privilege for modelling purposes in terms of simplifying the complexity of the
proposed solution methodology. In real-life, two possible scenarios arise. The difference between the two is distinguished thus:
(1) Geographically dispersed rms operate under solo ownership
(e.g. military logistics systems and international monopoly
enterprises). Here, a unied objective function is able to
express the overall system requirement(s).
(2) Multiple manufacturers in different geographical locations
are looked at as different entities with separate power centres, for example, this is typically the case in the automotive

15

industry. In such cases multiple objective functions are


required to describe the goals of the network participants.
Advancement of transportation and warehousing technologies
in one hand, and the growing interest in working with external
transport providers, and on the hand have facilitated the effective use of domestic and international warehouses in distribution
networks. For this reason, there is only a small interest in single
or no warehouse PD models, unless developed for a specic case
problem.
The majority of the developed mathematical models contain
binary and integer variables which are easy to formulate but
difcult to solve.
Most of the existing models ignore the dynamic nature of the key
input parameters and for simplication purposes deterministic
values are applied as an alternative. This precludes the developed
models from functioning effectively in volatile environments. The
only few dynamic models found in the literature use discrete time
horizons and discrete approximation of key input data which may
be inapplicable in many real world situations.
5.1.2. Identied research gaps
A SC may be owned by several entities, each aiming to increase
their own protability. In a broad picture, every member of a SC
is considered as a customer of the immediate upstream member. There are, however, end-users or end customers who are
the nal recipients of the goods and services without the need
to supply a subsequent SC participant. In most of the developed
PD models the signicance of addressing customer expectation
is not respected as the ultimate SC objective.
Studying continuous time horizons has received poor attention in
the past research. The volatile nature of the critical input parameters imposing frequent changes in managerial decision-making
may force SC planners to use continuous time models instead of
considering equal time intervals with deterministic parameters.
In a dynamic continuous time model (probably from non-convex
and nonlinear natures which are substantially harder to solver),
changes in key parameters and managerial decisions may occur
at any time during the planning horizon.
5.1.3. Future research trends
As SCs grow larger, more participants (and in particular more
manufacturers) join the existing SCs. Besides newly established
medium and large size enterprises generally have more than
one geographically dispersed production facility. This would call
for the development of more multi-plant models simulating the
characteristics of todays SCs.
Cost-minimisation and service-maximisation solutions have
been the most widely cited attributes in the current literature. The
need for staying continually attuned to customer expectations
has raised the importance of developing customer satisfaction
measures and models. This will require the development of
more multi-attribute (rather than multi-objective) techniques
particularly using fuzzy theory concepts to accommodate various
aspects of customer expectations.
There is a clear growing interest in the development of dynamic,
probabilistic and stochastic PD models. Despite the fact that
dynamic models are more realistic and hence more practicable
in real world scenarios, the solution techniques are considerably
harder to develop requiring substantial efforts and creativity (i.e.
robust optimisation techniques).
5.2. General PD modelling implications
There are also some overall implications for PD planning and
optimisation. These are highlighted in the following subsections.

16

B. Fahimnia et al. / Journal of Manufacturing Systems 32 (2013) 119

5.2.1. General observations


A vast amount of research (above 80% of all the published works)
has focused on the development of multi-product models which
characterise the nature of todays SCs.
Total cost minimisation (TCM) has been recognised as the most
respected performance measure for the evaluation of SC performance.
The inclusion of multiple DCs in the proposed published models has been highly regarded for modelling real world logistics
networks.
The use of traditional linear programming is no longer popular for formulating PD planning problems. Linear programming
has shown to be incapable of describing the actual complexity of
realistic SC planning models.
5.2.2. Identied research gaps
Prot maximisation (PM) has rarely been considered a PD objective function.
Despite the urgent need of bi and multi-objective PD models in
real world scenarios, evidence suggests that multi-objective (MO)
models have not received enough attention among academics.
For simplication purposes, considering detailed production cost
components, including the costs incurred at each machine centre,
has not been studied extensively.
Stack buffers (where products are stored in manufacturing plants
before shipment to warehouses and/or end-users in plants) have
rarely been used for modelling purposes (except for particular
case problems where it may t the purpose).
5.2.3. Future research trends
It may no longer be enough to develop a single objective SC
model minimising the overall SC costs or maximising prot.
Future research should pay special attention to quantifying and
formulating multiple objectives which may include traditionally
veried objective functions (e.g. cost, service level, just-in-time,
agility and leanness) and contemporary objective functions for
greenness and social sustainability (e.g. social and environmental
impacts, safety issues and voluntariness). Mathematical modelling of MO models is not always straightforward and may
require substantial time and effort to formulate multiple system requirements when dealing with thousands of independent
variables and constraints.
There is a clear trend towards the development of multi-product
and multi-plant PD models. Indeed, about 65% of all the
published models for the period of 20002011 address the multiplicity of product types and manufacturing plants. Yet there
are sector specic studies in which the PD models may have
particular characteristics that do not apply to other sectors.
Allowing for multiple transport paths (i.e. direct and indirect
transportation) and shipment modes (e.g. the use of trucks, air
transport and sea links) has become a new research trend for the
past 5 years. The key encouragement for this trend is to address
alternative shipment costs as well as economies of scale in transport when developing PD models. The consideration of multiple
transport routes and modes is primarily motivated by the recent
trend of globalisation and the opportunity to operate in multiple
counties/regions, popularly termed as multinational SCs.
One perennial concern in the context of SC modelling and optimisation is the development of appropriate solution approaches
for tackling large real world PD planning problems with large
numbers of continuous, integer and binary variables and constraints. Simulation, heuristics and meta-heuristic techniques
are still the dominant solution techniques in the literature.
GA and its extended version, Memetic Algorithm (MA), have
been recognised by several researchers as the most promising

heuristics techniques for handing real world optimisation problems [60,70,80,86,126,129]. There is still a need to further extend
the effectiveness of the existing solution approaches and to test
the new arrivals such as Ant Colony Optimization (ACO) and Bee
Colony Optimization (BCO) techniques.
The rapidly increasing number of SC participants requires
academics and practitioners to pay special attention to SC ownership and power domination issues. Modern SCs generally
consist of several entities with pre-existing locations, capacity
limitations and intellectual properties. This calls for facilitating inter-organisational collaboration which involves sharing
common goals, prots, information, expertise, resources, and
experience. Although generally there might not be a sole leadership involved in the modern SCs, in some cases, one or more
participant(s) in the chain may have the dominant power over
one or more other participants. Overall, future research needs to
study the increasing complexity of SC ownership and investigate
the impact of power domination issues on the SC strategies at
tactical and operational levels.
6. Conclusions and directions for future research
Over the last two decades, the signicance of PD planning and
optimisation at tactical and operational levels has been recognised
by academics and practitioners as a competitive advantage for the
growing production/distribution rms. An integrated PD plan covers the planning of activities in a vast scope from raw material
suppliers to manufacturers and warehouses through to the endusers. This large planning scope with multiple chain players makes
the PD planning problem a complex problem with several decision variables and constraints. This paper reviewed the state of the
art in optimisation modelling of PD plan. We classied the current
literature into seven categories based on the degree of complexity.
This classication could be of potential value to future researchers
in the eld and is also capable of further renements. A second
classication was also presented in the paper based on the solution
techniques used for tackling the proposed integrated PD planning
problems.
Despite the growing interest in PD planning and optimisation,
there are still several real world planning problems which have
remained unaddressed. We encourage academics to investigate the
SC planning problems which may now only concern practitioners.
Our ndings have some important implications for SC planners and
suggest the following directions for future research in the area.
Apart from some sector specic studies, most of the former PD
planning models are only the oversimplication of real world scenarios (principally due to the actual complexities of SC plans). A
realistic range of variables and constraints needs to be incorporated in future PD models.
One perennial concern is the development of appropriate solution approaches for tackling large real world PD planning
problems. Various solution techniques have been used to deal
with small and medium size PD planning problems ranging
from mathematical models, heuristics, simulation, and knowledge based system, to the latest fuzzy programming and GAs.
However, nding the optimal solution in a large real world PD
planning problem using simplistic techniques is impossible or
subject to heavy computing overheads. Some approaches which
might be able to handle large problems are not able to prove the
optimality of the solutions found or do not have the potential to
nd the optimal solution on their own. Hence, there is a need
to further extend the effectiveness of the existing solution techniques to be capable of handling realistic PD planning problems
with large numbers of variables and constraints.

B. Fahimnia et al. / Journal of Manufacturing Systems 32 (2013) 119

Considering multiple performance measures in developing


PD optimisation models allows the consideration of several
attributes in systematic decision-making [26,137]. The current
literature requires quantifying and formulating multiple PD performance indicators including both traditional and contemporary
objective functions (e.g. cost, service level, social impact, environmental impact, and safety measures).
SC ownership and power domination issues have been overlooked in the past research. Inter-organisational collaboration
needs to be accommodated in developing PD models for sharing common goals, prots, information, expertise, resources,
and experience. The areas requiring more attention by eld
researchers include the impact of SC ownership and power domination issues on the SC strategies/performance at tactical and
operational levels.
The development of dynamic, probabilistic and stochastic PD
models is currently followed by relatively few researchers. While
these researchers should continue holding the course, others
need to begin moving towards addressing the volatility and
uncertainty of the key input parameters in future PD models.
In line with the nancial value of a product comes the environmental and social burden incurred during different SC stages
[138]. This makes companies responsible for their environmental
and social impacts. Now, more than ever, brand-owning companies are coming under pressure from stakeholders to consider
the environmental and social factors in their entire SC [138,139].
The literature on sustainable SCM is still limited in quantity. It is
hoped that more researchers will be attracted to this issue which
requires the development of adequate measures quantifying the
environmental and social impacts of SC activities.

References
[1] Min H, Zhou G. Supply chain modeling: past, present and future. Computers
& Industrial Engineering 2002;43:23149.
[2] Barbarosoglu G, Ozgur D. Hierarchical design of an integrated production
and 2-echelon distribution system. European Journal of Operational Research
1999;118:46484.
[3] Park B, Choi H, Kang M. Integration of production and distribution planning
using a genetic algorithm in supply chain management. In: Melin P, Castillo O,
Ramrez EG, Kacprzyk J, Pedrycz W, editors. Analysis and design of intelligent
systems using soft computing techniques. Berlin: Springer; 2007. p. 41626.
A, Puigjaner L. Multiobjec[4] Guilln G, Mele FD, Bagajewicz MJ, Espuna
tive supply chain design under uncertainty. Chemical Engineering Science
2005;60:153553.
[5] Persson F, Olhager J. Performance simulation of supply chain designs. International Journal of Production Economics 2002;77:23145.
[6] Wang D, Fang S-C. A genetics-based approach for aggregated production
planning in a fuzzy environment. IEEE Transactions on Systems, Man and
Cybernetics 1997;27:63645.
[7] Tavakkoli-Moghaddam R, Rabbani M, Gharehgozli AH, Zaerpour N. A fuzzy
aggregate production planning model for make-to-stock environments. In:
Proceedings of the 2007 IEEE international conference on industrial engineering and engineering management. 2007. p. 160913.
[8] Wang R-C, Fang H-H. Aggregate production planning with multiple objectives in a fuzzy environment. European Journal of Operational Research
2001;133:52136.
[9] Wang R-C, Liang T-F. Application of fuzzy multi-objective linear programming to aggregate production planning. Computers & Industrial Engineering
2004;46:1741.
[10] Chen J-H, Ho S-Y. A novel approach to production planning of exible
manufacturing systems using an efcient multi-objective genetic algorithm.
International Journal of Machine Tools and Manufacture 2005;45:94957.
[11] Kim B, Kim S. Extended model for a hybrid production planning approach.
International Journal of Production Economics 2001;73:16573.
[12] Ganesh K, Punniyamoorthy M. Optimization of continuous-time production
planning using hybrid genetic algorithms-simulated annealing. International
Journal of Advanced Manufacturing Technology 2005;26:14854.
[13] Byrne MD, Bakir MA. Production planning using a hybrid simulationanalytical approach. International Journal of Production Economics
1999;59:30511.
[14] Techawiboonwong A, Yenradee P. Aggregate Production Planning using
Spreadsheet Solver: model and case study. ScienceAsia 2002:291300.
[15] Pradenas L, Penailillo F, Ferland J. Aggregate production planning problem, a
new algorithm. Electronic Notes in Discrete Mathematics 2004;18:1939.

17

[16] Amiri A. Designing a distribution network in a supply chain system: formulation and efcient solution procedure. European Journal of Operational
Research 2006;171:56776.
[17] Chopra S. Designing the distribution network in a supply chain. Transportation Research Part E: Logistics and Transportation Review 2003;39:12340.
[18] Liang T-F. Distribution planning decisions using interactive fuzzy multiobjective linear programming. Fuzzy Sets and Systems 2006;157:130316.
[19] Lockett AG, Westwood JB. Distribution planning in a turbulent environment.
European Journal of Operational Research 1985;19:3340.
[20] Raj PS, Nichols WG, Sterling JU, Moynihan GP. SM2ILE: a heuristic simulation
tool for strategic distribution planning. Computers & Industrial Engineering
1992;23:2414.
[21] Jansen DR, Weert AV, Beulens AJM, Huirne RBM. Simulation model of multicompartment distribution in the catering supply chain. European Journal of
Operational Research 2001;133:21024.
[22] Haq AN, Vrat P, Kanda A. An integrated productioninventorydistribution
model for manufacture of urea: a case. International Journal of Production
Economics 1991;25:3949.
[23] Pitty SS, Li W, Adhitya A, Srinivasan R, Karimi IA. Decision support for integrated renery supply chains: part 1. Dynamic simulation. Computers &
Chemical Engineering 2008;32:276786.
[24] Pandey A, Masin M, Prabhu V. Adaptive logistic controller for integrated
design of distributed supply chains. Journal of Manufacturing Systems
2007;26:10815.
[25] Thomas DJ, Grifn PM. Coordinated supply chain management. European
Journal of Operational Research 1996;94:115.
[26] Lair AMN. An integrated model for optimising manufacturing and distribution network scheduling. PhD thesis. School of Advanced Manufacturing and
Mechanical Engineering, University of South Australia, Adelaide; 2008. p. 266.
[27] Sarmiento AM, Nagi R. A review of integrated analysis of
productiondistribution systems. IIE Transactions 1999;31:106174.
[28] Vidal CJ, Goetschalckx M. Strategic productiondistribution models: a critical
review with emphasis on global supply chain models. European Journal of
Operational Research 1997;98:118.
[29] Coronado Mondragon AE, Lyons AC, Michaelides Z, Kehoe DF. Automotive
supply chain models and technologies: a review of some latest developments.
Journal of Enterprise Information Management 2006;19:55162.
[30] Donselaar KV, Kokke K, Allessie M. Performance measurement in the
transportation and distribution sector. International Journal of Physical Distribution & Logistics Management 1998;28:43450.
[31] Van Donselaar K, Sharman G. An innovative survey in the transportation and
distribution sector. International Journal of Operations & Production Management 1997;17:70720.
[32] Sarimveis H, Patrinos P, Tarantilis CD, Kiranoudis CT. Dynamic modeling and
control of supply chain systems: a review. Computers & Operations Research
2008;35:353061.
[33] Meixell MJ, Gargeya VB. Global supply chain design: a literature review and
critique. Transportation Research Part E: Logistics and Transportation Review
2005;41:53150.
[34] Dasci A, Verter V. A continuous model for productiondistribution system
design. European Journal of Operational Research 2001;129:28798.
[35] Erenguc SS, Simpson NC, Vakharia AJ. Integrated production/distribution
planning in supply chains: an invited review. European Journal of Operational
Research 1999;115:21936.
[36] Beamon BM. Supply chain design and analysis: models and methods. International Journal of Production Economics 1998;55:28194.
[37] Terzi S, Cavalieri S. Simulation in the supply chain context: a survey. Computers in Industry 2004;53:316.
[38] Kleijnen JPC. Supply chain simulation tools and techniques: a survey. International Journal of Simulation & Process Modelling 2005;01:829.
[39] Frohlich MT, Westbrook R. Arcs of integration: an international study
of supply chain strategies. Journal of Operations Management 2001;19:
185200.
[40] Holweg M, Bicheno J. Supply chain simulationa tool for education,
enhancement and endeavour. International Journal of Production Economics
2002;78:16375.
[41] Semini M, Hakon F, Jan Ola S. Applications of discrete-event simulation to
support manufacturing logistics decision-making: a survey. In: Proceedings
of the 2006 Winter simulation conference (WSC 06). 2006. p. 194653.

[42] Mula J, Peidro D, Daz-Madronero


M, Vicens E. Mathematical programming
models for supply chain production and transport planning. European Journal
of Operational Research 2010;204:37790.
[43] Peidro D, Mula J, Poler R, Lario F-C. Quantitative models for supply chain
planning under uncertainty: a review. The International Journal of Advanced
Manufacturing Technology 2009;43:40020.
[44] Fahimnia B, Luong L, Marian R. Optimization/simulation modeling of the
integrated productiondistribution plan: an innovative survey. WSEAS Transactions on Business and Economics 2008;5:4457.
[45] Chan FTS, Chung SH, Wadhwa S. A hybrid genetic algorithm for production
and distribution. Omega 2005;33:34555.
[46] Demirli K, Yimer AD. Productiondistribution planning with fuzzy costs. In:
Proceedings of the annual meeting of the North American fuzzy information
processing society, 2006 (NAFIPS 2006). 2006. p. 7027.
[47] Yilmaz P, Catay B. Strategic level three-stage production distribution
planning with capacity expansion. Computers & Industrial Engineering
2006;51:60920.

18

B. Fahimnia et al. / Journal of Manufacturing Systems 32 (2013) 119

[48] Boudia M, Louly MAO, Prins C. A reactive GRASP and path relinking for a combined productiondistribution problem. Computers & Operations Research
2007;34:340219.
[49] Coronado JL. An optimization model for strategic supply chain design under
stochastic capacity disruptions. Texas: College of Engineering, Texas A&M
University; 2008. p. 1110.
[50] Hamedi M, Farahani RZ, Husseini MM, Esmaeilian GR. A distribution planning model for natural gas supply chain: a case study. Energy Policy
2009;37:799812.
[51] Bard JF, Nananukul N. The integrated productioninventory
distributionrouting problem. Journal of Scheduling 2009;12:25780.
[52] Dhaenens-Flipo C, Finke G. An integrated model for an industrial
productiondistribution problem. IIE Transactions 2001;33:70515.
[53] Pyke DF, Cohen MA. Performance characteristics of stochastic integrated
productiondistribution systems. European Journal of Operational Research
1993;68:2348.
[54] Selim H, Araz C, Ozkarahan I. Collaborative productiondistribution planning
in supply chain: a fuzzy goal programming approach. Transportation Research
Part E: Logistics and Transportation Review 2008;44:396419.
[55] Ozdamar L, Yazgac T. A hierarchical planning approach for a
productiondistribution system. International Journal of Production
Research 1999;37:3759.
[56] Lee YH, Kim SH. Productiondistribution planning in supply chain considering
capacity constraints. Computers & Industrial Engineering 2002;43:16990.
[57] Lee YH, Kim SH. Optimal productiondistribution planning in supply chain
management using a hybrid simulation-analytic approach. In: Proceedings
of the 2000 Winter simulation conference. 2000. p. 12529.
[58] Rizk N, Martel A, DAmours S. Multi-item dynamic productiondistribution
planning in process industries with divergent nishing stages. Computers &
Operations Research 2006;33:360023.
[59] Nishi T, Konishi M, Ago M. A distributed decision-making system for integrated optimization of production scheduling and distribution for aluminum
production line. Computers & Chemical Engineering 2007;31:120521.
[60] Elahipanah M, Farahani RZ. A genetic algorithm to optimize the total cost
and service level for just-in-time distribution in a supply chain. International
Journal of Production Economics 2008;111:22943.
[61] Safaei AS, Moattar Husseini SM, Farahani RZ, Jolai F, Ghodsypour SH. Integrated multi-site productiondistribution planning in supply chain by hybrid
modelling. International Journal of Production Research 2010;48:404369.
[62] Mohamed ZM. An integrated productiondistribution model for a multinational company operating under varying exchange rates. International
Journal of Production Economics 1999;58:8192.
[63] Gen M, Syarif A. Hybrid genetic algorithm for multi-time period
production/distribution planning. Computers & Industrial Engineering
2005;48:799809.
[64] Chen M, Wang W. A linear programming model for integrated steel production and distribution planning. International Journal of Operations &
Production Management 1997;17:592610.
[65] Kanyalkar A, Adil G. An integrated aggregate and detailed planning in a multisite production environment using linear programming. International Journal
of Production Research 2005;43:443154.
[66] Kanyalkar A, Adil G. Aggregate and detailed production planning integrating
procurement and distribution plans in a multi-site environment. International Journal of Production Research 2007;45:532953.
[67] Kanyalkar A, Adil G. A robust optimisation model for aggregate and detailed
planning of a multi-site procurementproductiondistribution system. International Journal of Production Research 2010;48:63556.
[68] Cohen MA, Lee HL. Strategic analysis of integrated productiondistribution
systems: models and methods. Operations Research Society of America
1988;36:21628.
[69] Tang J, Yung K-L, Ip AWH. Heuristics-based integrated decisions for logistics
network systems. Journal of Manufacturing Systems 2004;23:113.
[70] Tasan A. A two step approach for the integrated production and distribution
planning of a supply chain. In: Huang D-S, Li K, Irwin GW, editors. Intelligent
computing. Berlin: Springer; 2006. p. 8838.
[71] Kanyalkar AP, Adil GK. Aggregate and detailed production planning integrating procurement and distribution plans in a multi-site environment.
International Journal of Production Research 2007;45:532953.
[72] Liang T-F. Fuzzy multi-objective production/distribution planning decisions
with multi-product and multi-time period in a supply chain. Computers &
Industrial Engineering 2008;55:67694.
[73] Bilgen B. Application of fuzzy mathematical programming approach to the
production allocation and distribution supply chain network problem. Expert
Systems with Applications: An International Journal 2010;37:448895.
[74] Torabi SA, Hassini E. Multi-site production planning integrating procurement
and distribution plans in multi-echelon supply chains: an interactive fuzzy
goal programming approach. International Journal of Production Research
2009;47:547599.
[75] Torabi SA, Hassini E. An interactive possibilistic programming approach for
multiple objective supply chain master planning. Fuzzy Sets and Systems
2008;159:193214.
[76] Chen C-L, Lee W-C. Multi-objective optimization of multi-echelon supply
chain networks with uncertain product demands and prices. Computers &
Chemical Engineering 2004;28:113144.
[77] Lim SJ, Jeong SJ, Kim KS, Park MW. A simulation approach for
productiondistribution planning with consideration given to replenishment

[78]

[79]

[80]

[81]

[82]

[83]

[84]
[85]

[86]

[87]
[88]

[89]
[90]

[91]

[92]

[93]

[94]

[95]
[96]
[97]
[98]
[99]
[100]

[101]
[102]

[103]

[104]

[105]

[106]

[107]

[108]

policies. The International Journal of Advanced Manufacturing Technology


2006;27:593603.
Jang Y-J, Jang S-Y, Chang B-M, Park J. A combined model of network design
and production/distribution planning for a supply network. Computers &
Industrial Engineering 2002;43:26381.
Aliev RA, Fazlollahi B, Guirimov BG, Aliev RR. Fuzzy-genetic approach to
aggregate productiondistribution planning in supply chain management.
Information Sciences 2007;177:424155.
Kazemi A, Fazel Zarandi M, Moattar Husseini S. A multi-agent system to solve
the productiondistribution planning problem for a supply chain: a genetic
algorithm approach. The International Journal of Advanced Manufacturing
Technology 2009;44:18093.
Das K, Sengupta S. A hierarchical process industry productiondistribution
planning model. International Journal of Production Economics
2009;117:40219.
Yimer AD, Demirli K. A genetic approach to two-phase optimization of
dynamic supply chain scheduling. Computers & Industrial Engineering
2010;58:41122.
Gunnarsson H, Rnnqvist M, Carlsson D. Integrated production and distribution planning for Sdra cell AB. Journal of Mathematical Modelling and
Algorithms 2007;6:2545.
Ferrio J, Wassick J. Chemical supply chain network optimization. Computers
& Chemical Engineering 2008;32:2481504.
Kim Y, Yun C, Park SB, Park S, Fan LT. An integrated model of supply network
and production planning for multiple fuel products of multi-site reneries.
Computers & Chemical Engineering 2008;32:252935.
Fahimnia B, Luong L, Marian R. Genetic algorithm optimization of an
integrated aggregate productiondistribution plan. International Journal of
Production Research; 116, in press.
Jordan DW, Smith P. Mathematical techniques: an introduction for the engineering, physical, and mathematical sciences. 3rd ed. Oxford: OUP; 2002.
Chinneck JW. Chapter 13: binary and mixed-integer programming. In: Practical optimization: a gentle introduction. Ottawa, Ontario, Canada: Carleton
University; 2004.
Chinneck JW. Chapter 1: introduction. In: Practical optimization: a gentle
introduction. Ottawa, Ontario, Canada: Carleton University; 2004.
Bhutta KS, Huq F, Frazier G, Mohamed Z. An integrated location production,
distribution and investment model for a multinational corporation. International Journal of Production Economics 2003;86:20116.
Paksoy T, Krsat Gules H, Bayraktar D. Design and optimization of a strategic
productiondistribution model for supply chain management: case study of a
plastic prole manufacturer in Turkey. Selcuk Journal of Applied Mathematics
2007;8:8399.
Tsiakis P, Papageorgiou LG. Optimal production allocation and distribution supply chain networks. International Journal of Production Economics
2008;111:46883.
Yan H, Yu Z, Edwin Cheng TC. A strategic model for supply chain design
with logical constraints: formulation and solution. Computers & Operations
Research 2003;30:213555.
Jayaraman V, Pirkul H. Planning and coordination of production and distribution facilities for multiple commodities. European Journal of Operational
Research 2001;133:394408.
Syam SS. A model and methodologies for the location problem with logistical
components. Computers & Operations Research 2002;29:117393.
Xu J, He Y, Gen M. A class of random fuzzy programming and its application
to supply chain design. Computers & Industrial Engineering 2009;56:93750.
Geoffrion AM. A guided tour of recent practical advances in integer linear
programming. Omega 1976;4:4957.
Dantzig GB. Linear programming. Operations Research 2002;50:427.
Fisher ML. An applications oriented guide to Lagrangian relaxation. Interfaces
1985;15:1021.
Mndez CA, Cerd J, Grossmann IE, Harjunkoski I, Fahl M. State-of-the-art
review of optimization methods for short-term scheduling of batch processes.
Computers & Chemical Engineering 2006;30:91346.
Pyke DF, Cohen MA. Multiproduct integrated productiondistribution systems. European Journal of Operational Research 1994;74:1849.
Yeh W-C. A hybrid heuristic algorithm for the multistage supply chain
network problem. The International Journal of Advanced Manufacturing
Technology 2005;26:67585.
Chinneck JW. Chapter 14: heuristics for discrete search: genetic algorithms
and simulated annealing. In: Practical optimization: a gentle introduction.
Ottawa, Ontario, Canada: Carleton University; 2004.
Xing K. Design for upgradability: modelling and optimisation. Adelaide:
School of Advanced Manufacturing and Mechanical Engineering, University
of South Australia; 2006.
Tarokh MJ, Golkar M. Supply chain simulation methods. In: Proceedings of
the IEEE international conference on service operations and logistics, and
informatics, 2006 (SOLI 06). 2006. p. 44854.
Lee YH, Cho MK, Kim SJ, Kim YB. Supply chain simulation with
discretecontinuous combined modeling. Computers & Industrial Engineering 2002;43:37592.
Yan T, Luh PB, Weidong F, Narimatsu K. Supply chain performance evaluation:
a simulation study. In: Proceedings of the IEEE international conference on
robotics and automation, 2003 (ICRA 03), vol. 1742. 2003. p. 174955.
Schunk D, Plott B. Using simulation to analyze supply chains. In: Proceedings
of the 2000 Winter simulation conference, vol. 1092. 2000. p. 1095100.

B. Fahimnia et al. / Journal of Manufacturing Systems 32 (2013) 119


[109] Sarker R, Kara S, Freeman G, Kayis B, Ray T, Abbass H. A multi-agent simulation study for supply chain operation. In: Proceedings of the international
conference on computational intelligence for modelling, control and automation and international conference on intelligent agents, web technologies and
internet commerce. 2005. p. 72833.
[110] Rossetti MD, Chan HT. A prototype object-oriented supply chain simulation
framework. In: Proceedings of the 2003 Winter simulation conference, vol.
1612. 2003. p. 161220.
[111] Ritchie-Dunham J, Morrice DJ, Scott J, Anderson EG. A strategic supply chain
simulation model. In: Proceedings of the 2000 Winter simulation conference,
vol. 1262. 2000. p. 12604.
[112] Moyaux T, Chaib-draa B, DAmours S. Multi-agent simulation of collaborative
strategies in a supply chain. In: Proceedings of the third international joint
conference on autonomous agents and multiagent systems, 2004 (AAMAS
2004). 2004. p. 529.
[113] Minegishi S, Thiel D. System dynamics modeling and simulation of a particular
food supply chain. Simulation Practice and Theory 2000;8:32139.
[114] Jain S, Workman RW, Collins LM, Ervin EC, Lathrop AP. Development of a
high-level supply chain simulation. In: Smith JS, Peters BA, Medeiros DJ,
Rohrer MW, editors. Proceedings of the 2001 Winter simulation conference.
2001.
[115] Jain S, Workman RW, Collins LM, Ervin EC. Development of a high-level supply chain simulation model. In: Proceedings of the 2001 Winter simulation
conference, vol. 1122. 2001. p. 112937.
[116] Jain S, Ervin EC, Lathrop AP, Workman RW, Collins LM. Analyzing the supply
chain for a large logistics operation using simulation. In: Proceedings of the
2001 Winter simulation conference, vol. 1122. 2001. p. 11238.
[117] Higuchi T, Troutt MD. Dynamic simulation of the supply chain for a short life
cycle productLessons from the Tamagotchi case. Computers & Operations
Research 2004;31:1097114.
[118] GoldSim. Dynamic simulation and supply chain management. Washington,
USA: GoldSim Technology Group; 2007.
[119] Changchien SW, Shen H-Y. Supply chain reengineering using a core process
analysis matrix and object-oriented simulation. Information & Management
2002;39:34558.
[120] Armbruster D, Marthaler D, Ringhofer C. Efcient simulations of supply
chains. In: Proceedings of the 2002 Winter simulation conference, vol. 1342.
2002. p. 13458.
[121] Siprelle AJ, Parsons DJ, Clark RJ. Benets of using a supply chain simulation tool
to study inventory allocation. In: Proceedings of the 2003 Winter simulation
conference, vol. 231. 2003. p. 23845.
[122] Williams EJ, Gunal A. Supply chain simulation and analysis with SIMFLEX.
In: Snchez PJ, Chick S, Ferrin D, Morrice DJ, editors. Proceedings of the 2003
Winter simulation conference. 2003.
[123] Holland JH. Adaptation in natural and articial systems: an introductory
analysis with applications to biology, control, and articial intelligence. Ann
Arbor: University of Michigan Press; 1975.

19

[124] Fahimnia B, Luong L, Marian R. Genetic algorithm optimization of a twoechelon supply chain network. In: Proceedings of the international conference
on industrial engineering and systems management (IESM 2009). 2009.
[125] Altiparmak F, Gen M, Lin L. A genetic algorithm for supply chain network
design. In: Proceedings of the 35th international conference on computers
and industrial engineering. 2005. p. 1116.
[126] Altiparmak F, Gen M, Lin L, Karaoglan I. A steady-state genetic algorithm for
multi-product supply chain network design. Computers & Industrial Engineering 2007;56:52137.
[127] Altiparmak F, Gen M, Lin L, Paksoy T. A genetic algorithm approach for multiobjective optimization of supply chain networks. Computers & Industrial
Engineering 2006;51:196215.
[128] Syarif A, Yun Y, Gen M. Study on multi-stage logistic chain network: a
spanning tree-based genetic algorithm approach. Computers & Industrial
Engineering 2002;43:299314.
[129] Yeh W-C. An efcient memetic algorithm for the multi-stage supply chain
network problem. The International Journal of Advanced Manufacturing
Technology 2006;29:80313.
[130] Fahimnia B, Luong L, Marian R. Optimization of a two-echelon supply network
using multi-objective genetic algorithms. In: Proceedings of the 2009 world
congress on computer science and information engineering. 2008.
[131] Fahimnia B. Optimisation of manufacturing lead-time, using genetic algorithm. Adelaide: School of Advanced Manufacturing and Mechanical
Engineering, University of South Australia; 2006. p. 151.
[132] Marian R. Optimisation of assembly sequences using genetic algorithms.
Adelaide: School of Advanced Manufacturing and Mechanical Engineering,
University of South Australia; 2003. p. 287.
[133] Sobhi-Najafabadi B. Optimal design of permanent magnet generators. Adelaide: School of Electrical and Information Engineering, University of South
Australia; 2002. p. 331.
[134] Haupt R, Haupt S. Practical genetic algorithms. In: Discrete applied mathematics. New Jersey: John Wiley & Sons, Inc.; 2004. p. 1261.
[135] Fahimnia B, Luong L, Marian R. An integrated model for the optimization of
a two-echelon supply network. In: Proceedings of the 9th global congress on
manufacturing and management (GCMM2008). 2008.
[136] Fahimnia B, Luong L, Marian R. Optimization and performance evaluation
of an integrated productiondistribution plan in supply chains. In: Proceedings of the ASME international mechanical engineering congress & exposition
(IMECE2009). 2009.
[137] Pishvaee MS, Jolai F, Razmi J. A stochastic optimization model for integrated
forward/reverse logistics network design. Journal of Manufacturing Systems
2009;28:10714.
[138] Seuring S, Mller M. From a literature review to a conceptual framework
for sustainable supply chain management. Journal of Cleaner Production
2008;16:1699710.
[139] Hume J. Business guide to a sustainable supply chain. Auckland: New Zealand
Business Council for Sustainable Development (NZBCSD); 2003.

You might also like