You are on page 1of 23

Int. J. Business Performance and Supply Chain Modelling, Vol. 3, No.

1, 2011

Intermodal transportation within the green supply


chain: an approach based on ELECTRE method
Marie Sawadogo and Didier Anciaux*
Laboratory of Industrial Engineering and Production of Metz,
University Paul Verlaine of Metz,
le du Saulcy, 57045 Metz Cedex 01, France
Fax: +33387315455
E-mail: sawadogo@univ-metz.fr
E-mail: anciaux@univ-metz.fr
*Corresponding author
Abstract: Considering the great environmental awakenings at the planetary
level and the discounted growth of the carriage of goods per transportation
mode in the world, the distribution and haulage companies must start to
take measures to limit their impacts. The goal of this paper is to model the
performance of an intermodal transportation system within the green supply
chain, taking into account the economic, environmental and societal criteria in
order to help decision-makers in choosing the path with the best compromised
benefit/impacts in an intermodal transportation system. We have to trade with
many conflicting criteria, it is thus appropriate to use multicriteria decision
support system to find the best compromised solution. The path selection in a
multimodal transportation network is therefore a multicriteria decision-making
problem. In this paper, we will use an approach based on the elimination and
choice expressing the reality (ELECTRE) method to solve our multicriteria
model.
Keywords: intermodal transport; modelling; multicriteria decision-making;
environmental impacts; supply chain.
Reference to this paper should be made as follows: Sawadogo, M. and
Anciaux, D. (2011) Intermodal transportation within the green supply chain:
an approach based on ELECTRE method, Int. J. Business Performance and
Supply Chain Modelling, Vol. 3, No. 1, pp.4365.
Biographical notes: Marie Sawadogo is obtained her Masters from the
Industrial Engineering School of the Polytechnic Institute of Grenoble (INP
Grenoble) in 2008. She is a PhD student at Laboratory of Industrial
Engineering and Production of Metz (LGIPM) in University Paul Verlaine
since October 2008. Her research focuses on transportation systems
environmental and societal impacts modelling and optimisation. Her PhD
subject is about how to integrate and optimise the environmental, societal and
economic impacts of intermodal transportation within the green supply chain.
She is investigating on using multi-objective ant colony optimisation to achieve
this goal.
Didier Anciaux received his PhD at the University Paul Verlaine of Metz in
1990 and is an Associate Professor since 2006. He works in the Production
System team of the Laboratory of Industrial Engineering and Production of
Metz (LGIPM). His main research activities include software architecture
design using multi-agent systems dedicated to production systems and green
supply chain. His other topics focus on methods to control discrete systems,
Copyright 2011 Inderscience Enterprises Ltd.

43

44

M. Sawadogo and D. Anciaux


such as supply chain, shop-floor or educational organisations. He was involved
in three European projects with major industries like Ford, Renault or Aprilia
and is already a co-author of seven book chapter and many articles in
international revues.

Introduction

The introduction of the Kyoto Protocol in 1997 and the announcement by the European
Commission in 2007 on a commitment to reduce CO2 emissions by 20% by 2020, and
50% by 2050 are some of the steps of the environmental awakenings at the planetary
level. Thus, for environmental and energy reasons, intermodal transport and its impacts
are recently received attention among the transport planners in governments and industry.
Freight transports have been multiplied by 1,000 to the globe (Transport White Paper,
2006). Moreover, customers are making decisions to favour companies and products that
are environmentally and socially responsible. As a result, studies have been conducted to
increase our understanding of pollutant emissions along supply chain with their
consequences, in order to develop schemes for impacts reduction. In addition, researches
have been conducted for the purpose of including sustainability in a general framework to
guide logistics planning and to move towards modes with lower environmental impacts,
such as rail and waterways.
The purpose of this paper is to help decision-makers to choose a path in a intermodal
network, taking into account not only the transportation cost and the transportation
time criteria, but also environmental and societal criteria and thus to help in choosing
among various alternatives, a path with the best compromise benefits/impacts. To this
end, we will use elimination and choice expressing the reality (ELECTRE) method to
find the best compromise solution among several alternatives paths within an intermodal
network.
After situating the framework of our study, we will present the theoretical model
developed to analyse the performance of a transportation system in a green supply chain,
which takes into account the criteria set out above. Afterwards, a general modelling of the
problem with the ELECTRE TRI method is operated. Then, the model is applied to an
example of intermodal transport between Paris and Marseille. Finally we conclude and
we give the prospects for future works.

Green supply chain and intermodal transportation

The logistical activities comprise freight transport, storage, inventory management,


materials handling and all the related information processing. Logistics is the integrated
management of all the activities required to move products through the supply chain. A
supply chain can be defined as an integrated process in which a number of different
actors (suppliers, manufacturers, distributors and retailers) work together to:
1

acquire raw materials

convert these raw materials into finished products

Intermodal transportation within the green supply chain


3

45

deliver these final products to retailers.

This chain is traditionally characterised by material flows and information flows


(Beamon, 1998). For very long time, the distribution of products with various modes of
transportation is not taken into account in the management of supply chain (Erenguc
et al., 1999). On the contrary, it is the external service provider of the supply chain who
always manages it, but it does not support the measurement of the performance to control
the cost. Our study focuses on the analysis of an intermodal freight system. We aim to
propose a solution to reduce the impacts of such a system, by modelling environmental
and social costs in order to propose a solution to solve this model.

2.1 Models integrating production and transport


To choose transportation modes in the integration of production and transport, it is
necessary to take into account the external environmental constraints, the technical
constraints, the commercial constraints, as well as the impacts on costs, qualities and
securities of all the means of transport contributor in shipment. The difficulty lies in the
fact that different constraints cannot be taken sequentially. Each decision depends and
influences, in varying degrees, other choices all along the supply chain. And finally, the
logistic choice of transport constitutes the essential outcome of the market policy of the
company (Hu et al., 2001). Many models have been suggested for the integration of
production and transportation; however, they aim either at proposing the shortest path
between the initial and the final terminals or at reducing the transportation costs while
ensuring acceptable delivery time. In Barnhart and Donald Ratliff (1993), the authors
propose an evaluation model for minimising the routing cost for each shipment with
respect to total transportation and inventory costs. Dullaert et al. (2005) introduce an
evolutionary algorithm for determining the optimal mix of transport alternatives to
minimise total logistics costs, including order costs, transportation costs, and inventory
costs. Janic (2007) develops a model for calculating comparable combined internal and
external costs of intermodal and route freight networks. There are also some jobs
focusing on the minimisation of handling times to guarantee the safety of the transported
goods and people (Boussedjra et al., 2003). The topic of integrating environment
concerns in the transportation systems is a recent research area and various studies have
been conducted in recent years to better take into account the environmental aspects of
transportation in the supply chain management.

2.2 Intermodal transportation


Intermodal transportation is defined by the European Conference of Ministers of
Transport (ECMT) as the carriage of goods by at least two different transportation modes
in the same loading unit without stuffing or stripping operations when changing modes
(Rondinelli and Berry, 2000). Intermodal transportation systems offer a wide choice of
transportation mode and several alternative paths, hence, the need for better coordination
of flows and movements in such a system. During the last decade, numerous publications
have focused on different aspects of intermodal transport problems, such as
complementarities of the different transportation modes, changes in pricing policies
in intermodal systems, management of flows between modes and the potential
environmental impacts of such movements.

46

M. Sawadogo and D. Anciaux

Intermodalism is a transportation process through interconnection of networks,


involving various combinations of modes in which all components are perfectly linked to
form an effective coordination of flows. This organisation offers a full range of transport
modes and multiple transportation options, which can coordinate the deliveries, the
manufacturing, stockpiling and the different distribution functions. The intermodal
networks are characterised by dynamic changes and multiple transportation modes,
operating simultaneously between them (Qu and Chen, 2008). In this system, the crossing
of goods from a transportation mode to another is made by operations called
transhipment. The transportation modes takes into account in our study are road,
air, water and rail. Environmental issues concerning freight become more important
nowadays since it is well-known that the transport sector is the major sources of noise
and numerous air pollutants. Intermodal freight is increasingly considered as a major
potential contributor to solving the sustainability problems of the transport sector.

2.3 Environmental impacts of intermodal transportation


The impacts of transport on the health and the environment are well-known today; they
includes global warming, the deterioration of the ozone layer, the dispersion of organic
and inorganic substances including toxic ground-level ozone, the scarcity of oil and other
natural resources, and the degradation of landscapes and soils. The emissions of gaseous
pollutants and greenhouse gases as NO2, CO2, NOx causes adverse health effects,
damage to buildings and materials, effects on crops and agricultural production, and
impacts on natural and semi-natural ecosystems (Schreyer et al., 2005); this is why
European Union directives limit exhaust emissions from new vehicles.
There are also impacts on transport infrastructures such as damages on road
constructions. To these impacts, must be added noise pollution, traffic congestion, social
impacts like accidents, and energy consumption. The energy that is consumed by
transportation is estimated to be one-third of the entire energy consumed in the European
Union. Pollution from vehicles such as heavy trucks includes gaseous pollutants such as
carbon monoxide (CO), nitrogen oxides (NOx), particulate pollutants and volatile organic
compounds (VOCs). Some hydrocarbons (including VOC) emissions from engine
emissions are carcinogenic. Nitrogen oxides are invisible toxic gas that can form fine
aerosol particles or salts which can contribute to acid rain or smog. The fuel engines are
often emitted as particulate pollutants. These particles are often toxic carcinogenic
chemicals, causing enormous damage to the lungs. In addition, ozone and particulate
matter are responsible for respiratory disease, environmental damage and visibility
problems, such as fog. Moreover, approximately 20% of Europeans suffer from problems
due to transportation noise. For more details on impacts, refer to the wheel impacts
(Figure 1) presented in Anciaux (2007).
According to numerous studies conducted on the environmental impacts of transport,
the trend is the exponential increase if no action is taken to mitigate these impacts.
Piecyk and McKinnon (2009) studied three scenarios illustrating this trend for road
freight. Indeed, the first scenario called business-as-usual (BAU) considers a positive
development in modal split, vehicle utilisation; fuel efficiency and carbon intensity are
likely to result in a 10% reduction in CO2 emissions from the current level, decreasing the
carbon footprint of road freight transport to 17.4 million tonnes of CO2 in 2020. In the
second scenario called optimistic scenario, CO2 emissions from road freight would be
47% below the current level (10.3 million tonnes of CO2 in 2020), for example by

Intermodal transportation within the green supply chain

47

shifting of freight away from road transport to alternative modes and by a reduction in the
average length of haul by 15 km relative to the present value. Finally, in a pessimistic
scenario, the carbon footprint of the road freight sector increases to 30.0 million tonnes of
CO2 in 2020 (56% above the present level) if no action is implemented to limit the
phenomenon. The introduction of green supply chain concept is a way for achieving this
goal.
Figure 1

Environmental and impacts wheel (see online version for colours)

2.4 Green supply chain management


The traditional management of supply chain activities is mainly focused on improving
economic performance as the optimisation of costs of logistics flows, minimising travel
time in the chain, facility location problems, coordination of information flows between
actors, optimisation and inventory management.
Green supply chain management recognises the disproportionate environmental
impact of supply chain processes. The definition and scope of green supply chain
management in the literature has ranged from green purchasing to integrated green
supply chain flowing from supplier to manufacturer to customer, and even reverse
logistic. Srivastava (2007) defines the green supply chain management as integrating
environmental thinking into supply chain management, including product design,
material sourcing and selection, manufacturing processes, delivery of the final product to
the consumers as well as end-of-life management of the product after its useful life. In
the literature, several studies exist on the various aspects and methods of the green supply
chain. An interesting literature review is presented in Zhang et al. (1997); it highlights
both the consideration of environmental impacts in product design and the life cycle
analysis of products. This study also highlights the reduction of impacts during the
production process and the products end of life management including recycling, reuse,

48

M. Sawadogo and D. Anciaux

remanufacturing, waste management... To this must be added studies such as those


presented in Min et al. (1998) highlighting the facility location and routing problems
within the green supply chain.
Besides the challenges of energy and climate change, the supply chain is also entering
an unprecedented age of opportunity. In particular, the modern supply chain still enjoys
relatively inexpensive and abundant energy inputs and raw material resources, and is also
more connected than ever before (Beamon, 2008). In recent year, a number of studies
have been published for impacts abatement in transportation systems. Several studies
were conducted on reducing the environmental impacts of transportation systems. Most
of these studies are partly on the analysis of impacts from transportation systems and the
literature review on methods to reduce impacts and, partly on the estimated costs of
environmental impacts of transport as presented in Piecyk and McKinnon (2007) and
Forkenbrock (1999). Some models for computing the environmental and social impacts
of transportation systems can be viewed in Janic (2007) and Peeters et al. (2007). Studies
are also conducted on the implementation of policies to internalise environmental costs of
transport; Ricci and Black (2005) present a review on the different costs involved in an
intermodal transport and the social cost to be considered.
In this paper, we specifically focus on reducing the impacts from the intermodal
transportation systems. Regarding contributions to transportation impacts abatement, the
introduction of new technologies and transport concepts such as alternative fuel, ecodriving, early morning-distribution system, vehicle utilisation and avoidance of empty
trips, environmental management systems (ISO 14000, ISO 14001) may offer reduced
emissions per vehicle, or encourage switching to more environment-friendly modes.
Another way of reducing environmental impacts is to persuade travellers to minimise the
number of vehicle trips by switching to non-motorised modes.
Rondinelli and Berry (2000) proposed an integrated proactive environment
management system which seeks to prevent pollution and eliminate sources of
environmental degradation of intermodal freight. In addition to these methods we must
add that an alternative to reduce environmental and social impacts is to choose a
judicious path and transportation mode for shipments; this technique is increasingly
studied in the literature. In this vein, Qu and Chen (2008) propose a hybrid multicriteria
decision-making method based on Fuzzy AHP and Artificial Neuron Network for route
selection in intermodal network taking into account the total transport time, the total cost,
and the social benefit such as the effects of traffic congestion. Bontekoning et al. (2004)
underlined the importance of choosing the transportation modes but little work is known
on the calculation of transportation cost, taking account the impacts. Tsamboulas and
Mikroudis (2000) proposed an approach that combines the multicriteria analysis and the
cost benefit analysis to evaluate the environmental impacts and costs of transport
initiatives. Quariguasi Frota Neto et al. (2009) proposed a method to find a trade-off
between the criteria preferred solutions for business and the environment, in order to
answer to the question how much do we have to spend in order to improve
environmental quality?.
So a lot of methods exist for transport environmental impacts reduction. Our
contribution is to help in choosing a path with fewer impacts adding to the traditional
criteria that are cost and travel time. To assess all the alternative paths and choose the
best one is a multicriteria optimisation problem, which can be performed first by
the modelling of all the criteria encountered and the implementation by multicriteria
decision-making methods, as presented in Section 3.

Intermodal transportation within the green supply chain

49

Multicriteria decision-making problem modelling

3.1 Criteria modelling


The problem here consists in transporting m tons of goods with a degree of fragility f,
along a path k from an origin O to a final destination F as presented in Figure 2. Each
path k of the network is a possible alternative A (intermodal or not) and may consist of
one or more branches i, with n branches by path and n + 1 feasible transhipments.
Figure 2

Intermodal transport of m tonnes of goods with a degree of fragility f

The modelling presented here is based mainly on studies conducted by IFEU (Knorr,
2005) and UNITE (Link et al., 2000). These studies include and evaluate the external
costs of transportation system. From these studies, we have obtained some parameters
indicating the fixed and variable transportation cost, the pollution cost, the energy
consumption cost... for the entire transportation mode taking into account. Table 1
represents the values of each parameter of our theoretical model defined by INFRASZrich (Schreyer et al., 2005); these parameters are defined according to data in
European Union countries.
Table 1

Theoretical model parameters

Parameters
c1 (fixed cost)
c2 (hourly fees)
c3 (mileage costs)
vt (average speed)

Unit

Aircraft

Boat

Truck

Train

300

200

100

200

/h

33.33

26.67

20

26.67

/km

0.80

0.10

0.40

0.20

km/h

800

10

80

100

g/T.km

665

49/30

71

18/35

NOx (amount of NOx)

mg/T.km

3,216

833/503

542

29/544

SO2 (amount of SO2)

mg/T.km

438

28/17

41

52/20

NMHC (amount of NMHC)

mg/T.km

248

74/44

38

2/54

part.(amount of particles)

mg/T.km

14

24/14

13

13/15

CO2 (amount of CO2)

(energy consumption factor)

kJ/T.km

9,052

665/401

966

392/473

/1,000 T.km

8.9

7.4

3.2

(average transhipment time)

h/ton

Variable

Variable

Variable

Variable

(average damage cost)

kg/ton

Variable

Variable

Variable

Variable

/1,000.T.km

7.6

(average cost of noise)

(average cost of accident)

50

M. Sawadogo and D. Anciaux

The main objective of this model is to compute the values of the criteria that will serve as
the basis for our analysis. The criteria retained for our study are the transportation cost,
the transportation time, the air pollution, the energy consumption, the noise
pollution, the transhipment, the damage due to transhipment and the accidents
respectively represented by C, T, P, E, B, , D and A. These criteria are grouped into five
major groups namely the cost, time, transhipment, environmental impacts and accidents;
details on the model are presented in Ferreira Dutra and Anciaux (2007).
The overall objective is to find a trade-off among the above criteria; hence, the need
to use multicriteria decision-making to find the best compromised solution allowing
economic development, environment preservation associated with a social development.
The performance of alternatives for selecting a transportation system will be divided into
eight factors: cost, time, pollution, energy, noise, transhipment, damage and
accident.
In Figure 2, we present the studied network. Let k be the number of intermodal
(or not) feasible paths from O to F, n is number of intermodal branches (subpath) of
the intermodal path k, i = 1, , k is the index of each path and t represents the means of
transport used.

3.1.1 Transportation cost


To characterise the overall cost of good transportation, it is necessary to model as closely
as possible the real cost charged by the logistic operator or the carrier for goods
transportation. Thus, the model here includes the drayage cost and the line haul cost
during all the routing period. In the cost model, we are using the following parameters:
fixed cost (fee), an hourly cost (hourly pay of the driver) and cost per kilometre
(distance). The transportation cost is then expressed as follows:
n

Ck =

( c
ci =

i =1

1t

i =1

+ ti c2t + di c3t ) it

(1)

where Ck is the transportation cost on a path k expressed in (), c1t is the fixed costs due
to the use of means of transport t also expressed in (), c2t is hourly fees due to the use of
means of transport t in (/h), c3t represents the mileage costs due to the use of means of
transport t in (/km), ti is the transport time on a branch i expressed in (h), di is
the length of a branch i, in (km), and it is the number of means of transport t used on a
branch i.

3.1.2 Transportation time


The transportation time Tk on a path k is obtained by dividing the average speed of
each transportation mode by the travelled distance. We assume that the carriage of goods
is operating without intermediate storage, i.e., the storage time is insignificant. In
this equation, dt is the length of a branch i and vt the average speed for transportation
mode t.
n

Tk =

i =1

ti =

dt

i =1

(2)

Intermodal transportation within the green supply chain

51

3.1.3 Transhipment time


Transhipments are the set of operations allowing moving goods from one mode to
another, possibly with a docking means, without storage. Goods are then are physically
transhipped from one vehicle to another vehicle of the same mode or not (e.g., truck to
truck or truck to train); transhipment here then include terminal transfer. Let k be the
transhipment time of m tons of goods at each branch i of a path k. It depends on the
fragility f of the product, the quantity m of product transhipped and an average
transhipment time par per ton of goods transhipped i. It can be calculated by the
equation presented below:

k =

n +1

m f

(3)

i =1

3.1.4 Air pollution


To estimate the impacts of exhaust emissions caused by intermodal freight, it is necessary
to consider the direct and indirect emissions of different transportation modes. This
estimation model is used to quantify the emissions of CO2, NOx and VOCs, the main gas
emitted by vehicles. The function defined here depends on the quantity of goods
transported, the transportation mode used and the travelled distance. The amount of
pollutants Pkj emitted by a gas j is done by:
n

Pkj =

pij =

i =1

md
i

tj

(4)

i =1

where
Pkj

represents the pollution emitted by gas j on the path k, in (kg)

pij

is the pollution emitted by gas j on a branch i, in (kg)

tj

is the amount of pollutant j emitted by the mean of transport t for each kilometre
travelled, in (kg/T.km).

3.1.5 Energy consumption


The total energy consumption of transport is the energy necessary for the haulage of m
quantity of goods by each transportation mode. The main fuels take into account are
petroleum products, electricity and natural gas. The amount of energy consumed by a
conveyance t is calculated here based on the quantity of goods transported m, the
travelled distance di and the energy consumption factor t defined in Schreyer et al.
(2005) for each transportation mode t.
n

Ek =

e = m d
i

i =1

i =1

where
Ek

energy consumption on the path k, in (kJ)

(5)

52

M. Sawadogo and D. Anciaux

ei

energy consumption on the branch i, in (kJ)

energy consumption by the mean of transport t, in (kJ/T.km).

3.1.6 Noise pollution


Noise can be defined as the unwanted or harmful sound. The transport sector is the most
widespread source of noise in all countries and the most prevalent cause of annoyance
and interference. Therefore, traffic noise reduction measures have the highest priority.
The effects of transport noise are not very well-understood. There is no fully satisfactory
measurement of noise and the nuisance it causes. The noise generated by transportation
systems affects people in residential areas; these nuisances are all the more important
since the traffic is important. Here, we will modelling noise cost Bk according the a noise
cost factor t define in Schreyer et al. (2005) for each transportation mode t, the quantity
of goods transported m and the travel distance dt, hence,
n

Bk =

bt =

i =1

i =1

m dt t
1, 000

(6)

where Bk is the cost related to noise impacts on the path k, in (), bt the cost related to
noises impacts on a branch i, in (), and t is the average cost of noise per ton per
kilometre of the mean of transport t expressed in (/1,000.T.km).

3.1.7 Damages caused by transhipment


During transhipment operations, damage may occur during the transportation and the
handling of products. These damages depend on the quantity of goods transhipped m and
the fragility f of the products, indeed a product such as glass or flowers is more likely to
be damaged as furniture for example. At each transhipment, the damages costs Dk also
depend on an average damage cost per ton of goods transhipped i and are represented by
the equation below:
n +1

Dk =

m f

(7)

i =1

3.1.8 Accident cost


All transportation modes occasionally are involved in accidents and mishaps of various
sorts. When this occurs, adverse outcomes often are experienced by people and their
property (Forkenbrock, 1999). The cost of an accident caused by a transportation mode t
is the uncompensated cost of deaths, injuries, and property damage that occurs due to an
additional trip by the mode in question. The accident cost defined here depends on an
average cost of accidents per ton kilometre t defined by Schreyer et al. (2005); it also
depends on the quantity of goods transported and the travelled distance. Thus,
n

Ak =

i =1

i =

i =1

m dt t
1, 000

(8)

Intermodal transportation within the green supply chain

53

3.2 Modelling with ELECTRE


The model defined here involves many conflicting criteria. Multiple criteria are very
important in judgmental decision-making; many real world decisions making are based
on multiple criteria. Several MCDM methods based on weighted averages, priority
setting, outranking, fuzzy principles and their combinations are employed for
environmental management.
Multicriteria decision-making techniques are then gaining popularity in sustainable
supply chain management. In lengin et al. (2010), a decision support framework
is proposed to analyse the impacts of transportation policies on social system,
environmental issues, and energy; they used the technique for order preference
by similarity to ideal solution (TOPSIS) to analyse the transportation-environment
interactions in order to help the policymakers in finding appropriate policies to alleviate
and mitigate external environmental effects linked to transportation. By the same way,
Kainumaa and Tawara (2006) proposed a multiple attribute utility theory method for
assessing a supply chain by including re-use and recycling throughout the life cycle of
products and services. They integrated return on asset (ROA), customer satisfaction
and life cycle assessment (LCA) of the supply chain by using a multi-attribute utility
function. For the purpose of our study, we will use ELECTRE method here to build the
decision support system.
ELECTRE method was developed by Bernard Roy during the 70s and aims to build
relationships called binary outranking relations, to represent the preferences of decisionmakers. ELECTRE method is mainly French and Belgian outranking method for
multicriteria decision support. It allows decision-makers to choose the best action from a
set of actions. Here, it is the best alternative path (intermodal path or not) among a set of
paths. Best alternative path means a path with less social and environmental impacts, as
with less transportation costs and travel time.
There are two main parts to an ELECTRE application: the first step is the
construction of one or several outranking relations, which aims at comparing in
a comprehensive way each pair of actions; second, an exploitation procedure that
elaborates on the recommendations obtained in the first step. The nature of the
recommendation depends on the problem being addressed: choosing, ranking or sorting.
The basis of the decision-making with ELECTRE is pairwise comparison. ELECTRE
has evolved into a family of methods based on different issues (ELECTRE I, II, III, IV,
TRI ) (Figueira et al., 2005), we will use ELECTRE TRI for our study.

3.2.1 The performance matrix


The first step consists in weighting each criterion. In general, ELECTRE method does not
compute the criteria weights, so we will use an AHP hierarchy for weights determination
(Appendix A); we then used the scale of Saaty (1990) to compute the judgement matrix
by pairwise comparison of the criteria and then to establish a performance matrix.
The criteria are conflicting, this makes the assessment of the alternative more
complex; the border between environmental issues and economic profit is often difficult
to determine. Quariguasi Frota Neto et al. (2009) proposed a method to find a trade-off
between the criteria preferred solutions for business and the environment, they found a
trade-off between the economic value and the environmental quality of an economic
activity where one cannot decrease either the environmental pressure without decreasing

54

M. Sawadogo and D. Anciaux

the economic value. They then proposed a programming assessment of the trade-offs
among the environmental impact indicators and the profit of a given logistics network.
For the purpose of our study, we will use scenarios. So, the weights of the criteria are
defined according to three scenarios. The first one is the industrial scenario where the
criterion cost and time are the most important; according to the scale of Saaty, the
criteria time and cost have an intensity of 9 contrary to criteria environment which
have an intensity of 1 (Table 2 and Table 3). The second one, called ecological or
environmental scenario gives priority to the environment criterion. Finally we define a
joint scenario that includes both first ones, called mixed scenario where the criteria have
substantially equal importance.
Table 2

Example of judgement matrix for the industrial scenario

Cost

Cost

Time

Environment

Transhipment

Accident

Priority

0.2615

Time

0.2571

Environment

0.2615

Transhipment

1/3

1/5

1/3

0.1212

Accident

1/2

1/2

1/5

1/2

0.0988

max = 5.13

CI = 0.03

CR = 2.9%

Table 3

Pollution

Example of judgment matrix of environmental criteria for the industrial scenario


Pollution

Energy

Noise

Priority

1/3

0.2654

Energy

0.6716

Noise

1/5

1/9

0.0629

max = 3.03

CI = 0.01

CR = 2.9%

The weight of the environmental criteria is obtained by aggregating the priority of the
subcriteria pollution, noise and energy. An aggregation is also done to obtain the weight
of the criterion transhipment by comparing transhipment time and damage due to
transhipment. The consistency of each matrix is evaluated.
The weights obtained for each criterion, according to each scenario are recorded in
Table 4.
Table 4
Criterion
C
T

D
P
E
B
A

Performance matrix
Scenarios
Industrial

Ecological

Mixed

0.4217
0.3715
0.0766
0.0511
0.0101
0.0255
0.0024
0.0413

0.1229
0.1033
0.0316
0.0211
0.4364
0.1875
0.0484
0.0488

0.2615
0.2571
0.0727
0.0485
0.1257
0.1060
0.0298
0.0988

Intermodal transportation within the green supply chain

55

The ELECTRE TRI method which we use in our study is the problematic
(allocation procedures) (Mareshal, 2000); the problem here is put in terms
assignment of each action at a predefined category. ELECTRE TRI uses three kinds
input data: alternatives, criteria and profiles. We define four categories as shown
Table 5.
Table 5

of
of
in

Sorting categories

Categories
C4

Excellent

C3

Good

C2

Bad

C1

Worse

Each category must be characterised by a lower and an upper reference profile; each
reference profile is therefore limited by two categories, one upper and one lower. Three
reference profiles are used to segment the criteria categories as shown in Table 6.
Table 6

Reference profiles of the categories

Reference profiles
b3

Boundary between C3 and C4

b2

Boundary between C2 and C3

b1

Boundary between C1 and C2

The reference profiles are defined as a percentage (75%, 50%, 25%) the highest value of
the alternative A for the criteria X. For each criterion, X (C, T, P, E, B, , D, A) and for
each alternative A, the reference profiles are defined by:
b1 X = 0.75* Max ( AX )

(9)

b2 X = 0.50 * Max ( AX )

(10)

b3 X = 0.25* Max ( AX )

(11)

According to each reference (b1, b2, b3), and in relation to each criterion X, we have two
distinct sets of parameters: the importance coefficients, the indifference thresholds q, the
preference thresholds p and veto thresholds. We have thus calculated the concordance
(Appendix E), the conflict and outranking indices, as well as credibility index for every
criterion and alternative with regard to the various scenarios defined above. The veto
threshold is considered as insignificant. The other thresholds are expressed as a
percentage of average differences between the values of the criteria X. The average
difference is a measure of dispersion of a set of K data (number of criteria) as shown in
equations below.
qX = a

1
K

X
k =1

(12)

56

M. Sawadogo and D. Anciaux


pX = b

1
K

k =1

(13)

with
X =

1
K

k.

k =1

We considered as basic values, 10% for a and 40% for b, sensitivity analysis can be
conducted by varying the value of a and b.
To this must be added that in ELECTRE TRI method, outranking relationships are
built so as to compare each alternative to each reference profile. This is done by
calculating the partial concordance indices, global concordance c, discordance level d
and credibility index (Appendix C) for each alternative against each criterion and
according to the scenario outlined above [equations (14), (15) and (16)]; wx is the weight
of the criterion x.
c ( a, bh ) =

d x ( a, bh ) =

x ( a, bh ) =

wx cx ( a, bh )

wx

g x (a ) g x ( bh ) px ( bh )
vx ( bh ) px ( bh )

1 d x ( a, bh )

1 c ( a, b )
X F

(14)

(15)

(16)

with
x

F =
< d x ( a, bh )
c
a
b
,
h)
(

To obtain the outranking relations, we use a cutting level , considered as the smallest
value of the credibility index which is compatible with the assertion that the alternative
a outstrips the reference profile bh. You can refer to Figueira et al. (2005) for more
details on this method. Preferences in ELECTRE methods are modelled by using binary
outranking relations, S, whose meaning is at least as good as.

x ( a, bh ) a S bh

(17)

By applying the following outranking relations, where > means that a is strictly
preferred to bh, < for the inverse, I means that a is indifferent to bh and finally, R
means that a is incomparable to bh, we have assigned a category to each alternative
(Appendix D).

Intermodal transportation within the green supply chain

57

Implementation and results

The model defined above is applied to an intermodal transportation problem between


Paris and Marseille as shown in Figure 3.
We suppose that:

One thousand tons of goods are to be delivered from the Peugeot factory in
Aulnay-sous-Bois in Paris to the industrial centre in Marseille. There are
12 possibilities for intermodal combination of transportation modes. And the
distance statistics used in the model are based on the report and research of Air
France, Michelin, SNCF and Google Earth.

The delivered loading units for train and ship are supposed to be 20 feet (a TEU
or 20 feet equivalent unit) as is common in Europe. Each unit has an average gross
weight of 12 tonnes of goods. And the load of freight truck is defined as 30 tons per
vehicle.

The transhipment costs depend largely on the location and capacity of terminals.
We suppose here the transhipment cost is 27/TEU at the rail terminal, 50/TEU
at the seaport and 45 per loading unit on the airport.

The pollution statistics is defined according to the study in Knorr (2005).

Figure 3

Intermodal transportation system studied


PARIS

LYON

MARSEILLE

776
315
357

464

277

776

464

357

277
315
660

391

357

315
315

There are 12 possible paths or alternatives which are intermodal or not to deliver the
goods to their final destination. Appendix B presents the different paths, the succession
transportation mode used and distances of each path in the network.

4.1 Assignment by category


The three reference profiles were assessed by ELECTRE TRI for each criterion,
which defines four categories. The references profiles calculations results are presented in
Table 7.

58

M. Sawadogo and D. Anciaux

Table 7

Reference profiles for each criterion

Reference
profiles

b1

5.14

10.38

5.12

1,494

1.47

150

0.01

1.47

b2

10.27

20.75

10.25

2,987

2.95

300

0.03

2.95

b3

15.41

31.13

15.37

4,481

4.42

450

0.04

4.42

Maximum
value

20.55

41.50

20.49

5,974

5.90

600

0.05

5.90

4.2 Assignment by alternatives


The assignment by category of the alternatives according to each scenario is in relation to
two procedures. Indeed, ELECTRE TRI is the twinning of two procedures (Appendix F):
a pessimistic procedure (conjunctive logic) and an optimistic procedure (disjunctive
logic) (Roy, 1994; Figueira et al., 2005). The assignments are realised for a cutting level
= 0.75 as shown in Table 8. To refine previous solutions, sensitivity analysis is done on
the indifference and preference threshold. A sensitivity analysis is also conducted on the
cutting level with values between 0.5 and 1.0. Generally, if we wish to mark the
disjunctive or conjunctive character of the procedure, the cutting level should be close
to one.
Table 8

Assignment by category for a cutting level = 0.75

Categories

Industrial

Ecological

Mixed

Optimistic Pessimistic

Optimistic Pessimistic

Optimistic Pessimistic

C4

Excellent

A05, A07

A05, A07

A01, A05, A01, A05,


A06
A06

C3

Good

A04, A06, A08, A09,


A08, A09,
A11
A11, A12

A02, A03, A02, A07,


A07, A08
A08

C2

Bad

A01, A02, A02, A10,


A10
A12

A04, A09, A03, A04,


A10, A11, A10, A11,
A12
A12

C1

Worse

A03

A01, A03,
A04, A06

A09

A05

A05

A01, A02, A07, A08


A04, A06,
A07, A08,
A09, A11,
A12
A10

A02, A10,
A11, A12

A03

A01, A03,
A04, A06,
A09

Due to the fact that variations depend on the cutting level, an aggregation of the results of
sensitivity analysis for each procedure is performed to help decision-makers in making a
right choice. This aggregation is to calculate an average value of the numbers referred to
categories, and rounded the final result.
This gives us a more simplified final sorting. Table 9 gives us in each scenario, an
alternative which belong to a category.

Intermodal transportation within the green supply chain


Table 9

59

Final sorting of alternatives by category

Categories
C4

Excellent

C3

Good

C2

Bad

C1

Worst

Industrial scenario

Ecological scenario

Mixed scenario

A05

A05

A05

A07, A08, A09, A11

A01, A02, A03,


A06, A07, A08

A06, A07, A08, A11

A01, A02, A04,


A06, A10, A12

A04, A10, A11, A12

A01, A02, A03, A04,


A09, A10, A12

A03

A09

Depending on the desired scenario, a path as well as a transportation mode is proposed to


the decision-maker, to deliver his goods to the final destination. The sorting of
alternatives in relation to reference categories as shown in Table 9 indicates that the
alternative A05 (train) is classified as an excellent alternative. A07 and A08 are good for
all scenarios. For ecological scenario, alternatives A01, A02, A03, A06 can be selected in
addition to the previous alternatives. The decision-maker has to choose between many
solutions ranging from best to worst. Indeed, for one who wants to reduce his
environmental impacts, by choosing alternative A05 instead of the alternative A09, he
can reduce the energy consumption by 5,607 GJ and the CO2 emission by 411.74 tons.
These results demonstrate that the wise choice of a path and a transportation mode can
contribute on reducing the environmental impacts of the overall intermodal network.

Conclusions and future works

The economic climate changes and the environmental awareness are forcing companies
to reconsider the organisation of their supply chain, with the aim of favour transportation
modes environmentally friendly and insure better coordination of flows to reduce
environmental impacts and traffic congestion. Climate change and CO2 emissions are
clearly becoming significant factors in logistical decision-making. In this vein, the
present paper aims to support the choice of sustainable transportation mode and path in
an intermodal network.
We have then included environmental aspects when choosing a freight transport
system. The performance of the alternatives were modelled taking into account the
transportation cost, the transportation time, time and damage associated with
transhipment, the pollutant emissions, energy consumption, noise and costs related to
accidents. But given the inability to optimise all the criteria simultaneously, we made a
multicriteria analysis of the problem. By applying ELECTRE TRI method we have
obtained a sorting of all the alternatives going from the best to the worst. For the case
study, the alternative A05 (train) is the best choice. When we compare the environmental
impacts of this alternative with the worst alternatives (A09 aircraft) for the ecological
scenario, the carbon footprint calculation gave a CO2 reduction by 411.74 tonne, energy
saving by 5,607 GJ, noise cost reduction by 3.39/1,000.T.km. This study highlights that
logistics decision-makers have to integrate the environmental concerns since the planning
of their operations and particularly in the choice of the transportation mode used for
shipments.
The multicriteria methods such as ELECTRE, AHP, etc., allow determining a
path according to a certain number of criteria but they require enumerating all the

60

M. Sawadogo and D. Anciaux

alternatives; thus, they are not effective for a large-sized network. So, we intend to
use multi-objective optimisation techniques, such as metaheuristics or evolutionary
algorithms to find an optimal solution, in order to give an integrated decision support tool
for decision-makers in transport and logistics. In the future, we would build up another
multicriteria method, which are more adaptable and efficient for this problem and study
the sensibility of parameters, both of the mathematical model but also of the multicriteria
method. We will also modelling more impacts of the impact wheel presented by
including criterion such as congestion, and integrates them in an objective function in
order to optimise it. We aim to use graph theory for our network modelling as a
multiobjective shortest path problem; we will include factors influencing the impacts
such as transportation mode loading factors and the traffic congestion.

References
Anciaux, D. (2007) Chane logistique verte: un modle intgrant laspect environnemental du
transport intermodal, 7th International Congres of Industrial Engineering, Trois-Rivires,
Qubec, Canada, 710 June.
Barnhart, C. and Donald Ratliff, H. (1993) Modelling intermodal routing, Journal of Business
Logistics, Vol. 14, No. 1, pp.205223.
Beamon, B.M. (1998) Supply chain design and analysis: models and methods, International
Journal of Production Economics, Vol. 55, No. 3, pp.281294.
Beamon, B.M. (2008) Sustainability and the future of supply chain management, Operations and
Supply Chain Management, Vol. 1, No. 1, pp.418.
Bontekoning, Y.M., Macharis, C. and Trip, J.J. (2004) Is a new applied transportation research
field emerging? A review of intermodal rail-truck freight transport literature, Transportation
Research A, Vol. 39, pp.134.
Boussedjra, M., Bloch, C. and El Moudni, A. (2003) Solution optimale pour la recherche du
meilleur chemin intermodal, Proceedings of 4th Conference Francophone de MOdlisation et
SIMulation, Toulouse, France.
Dullaert, W., Maes, B., Cernimmen, B. and Witlox, F. (2005) An evolutionary algorithm for order
splitting with multiple transport alternatives, Expert Systems with Applications, Vol. 28,
pp.201208.
Erenguc, S.S., Simpson, N.C. and Vakharia, A.J. (1999) Integrated production/distribution
planning in supply chains: an invited review, European Journal of Operational Research,
Vol. 115, No. 2, pp.219236.
Ferreira Dutra, L. and Anciaux, D. (2007) Mthodes multicritres daide la dcision: une
approche pour la chane logistique verte intgrant le transport intermodal, Paper presented at
the 2nd Alban Conference, Grenoble, France, 1112 May.
Figueira, J., Mousseau, V. and Roy, B. (2005) ELECTRE methods, in Figueira, J., Greco, S. and
Ehrgott, M. (Eds.): Multiple criteria decision analysis: state of the art surveys, International
Series in Operations Research & Management Science, Vol. 78, Chapter 4, pp.133162,
Springer, New York.
Forkenbrock, D.J. (1999) External costs of intercity truck freight transportation, Transportation
Research Part A, Vol. 33, pp.505526.
Hu, Q., Arun, K. and Zhang, S. (2001) A bidding decision model in multi-agent supply chain
planning, International Journal of Production Research, Vol. 39, No. 15, pp.3293301.
Janic, M. (2007) Modelling the full costs of an intermodal and road freight transport network,
Transportation Research Part D, Vol. 12, pp.3344.

Intermodal transportation within the green supply chain

61

Kainumaa, Y. and Tawara, N. (2006) A multiple attribute utility theory approach to lean and
green supply chain management, International Journal of Production Economics, Vol. 101,
pp.99108.
Knorr, W. (2005) EcoTransIT: Environmental Methodology and Data, IFEU, available at
http://www.ecotransit.org/.
Link, H., Stewart, L., Maibach, M., Sansom, T. and Nellthrop, J. (2000) The Accounts Approach,
UNITE: Unification of Accounts and Marginal Costs, for Transport Efficiency.
Mareshal, B. (2000) Mthodes dAide la Dcision, Universit Libre de Bruxelles.
Min, H., Jayaraman, V. and Srivastava, R. (1998) Combined location-routing problems: a
synthesis and future research directions, European Journal of Operational Research,
Vol. 108, pp.115.
Peeters, P., Szimba, E. and Duijnisveld, M. (2007) Major environmental impacts of European
tourist transport, Journal of Transport Geography, Vol. 15, No. 2, pp.8393.
Piecyk, M. and McKinnon, A. (2007) Internalising the External Costs of Road Freight Transport in
the UK, Logistics Research Centre, School of Management and Languages, Heriot-Watt
University, Edinburgh, UK, available at http://www.greenlogistics.org.
Piecyk, M.I. and McKinnon, A.C. (2009) Forecasting the carbon footprint of road freight transport
in 2020, International Journal of Production Economics, doi: 10.1016/j.ijpe.2009.08.027.
Qu, L. and Chen, Y. (2008) A hybrid MCDM method for route selection of multimodal
transportation network, Part I, LNCS 5263, pp.374383.
Quariguasi Frota Neto, J., Walther, G., Bloemhof, J., van Nunen, J.A.E.E. and Spengler, T. (2009)
A methodology for assessing eco-efficiency in logistics networks, European Journal of
Operational Research, Vol. 193, No. 3, pp.670682.
Ricci, A. and Black, I. (2005) Measuring the marginal social cost of transport, Research in
Transportation Economics, Vol. 14, pp.245285.
Rondinelli, D. and Berry, M. (2000) Multimodal transportation, logistics, and the environment:
managing interactions in a global economy, European Management Journal, Vol. 18,
pp.398410.
Roy, B. (1994) Aide Multicritre la Dcision: Mthodes et Cas, Economica, Paris.
Saaty, T. (1990) How to make a decision: the analytic hierarchy process, European Journal of
Operational Management, Vol. 48, pp.926.
Schreyer, C., Schneider, C., Maibach, M., Rothengatter, W. and Doll, C. (2005) External Cost of
Transport, INFRAS.
Srivastava, S.K. (2007) Green supply-chain management: a state-of-the-art literature review,
International Journal of Management Reviews, Vol. 9, No. 1, pp.5380.
Transport White Paper (2006) Final communication from the commission to the council and the
European parliament: keep Europe moving sustainable mobility for our continent, Mid-term
Review of the European Commissions 2001, Brussels, 22 June.
Tsamboulas, D. and Mikroudis, G. (2000) EFECT evaluation framework of environmental
impacts and costs of transport initiatives, Transportation Research Part D, Vol. 5,
pp.283303.
lengin, F., zgr, K., ule, ., lengin, B. and Akta, E. (2010) A problem-structuring model
for analyzing transportation-environment relationships, European Journal of Operational
Research, Vol. 200, pp.844859.
Zhang, H.C., Kuo, T.C., Lu, H. and Huang, S.H. (1997) Environmentally conscious design
and manufacturing: a state-of-the-art survey, Journal of Manufacturing Systems, Vol. 16,
pp.352371.

62

M. Sawadogo and D. Anciaux

Appendix A
AHP hierarchy

Cost

Time

Environment

Pollution

Noise

Accidents
Energy

Criteria

Choose a path with


minimum impact

Goal

Transship ment
Time

Damage

Set of alternatives

Appendix B
Intermodal transportation system studied
Paths

Means of transport

Distances (km)

Total (km)

A01

Truck

776

776

A02

Truck-train

464-315

779

A03

Truck-boat

464-357

821

A04

Truck-aircraft

464-277

741

A05

Train

776

776

A06

Train-boat

464-357

821

A07

Train-aircraft

464-277

721

A08

Train-truck

464-315

779

A09

Aircraft

660

660

A10

Aircraft-boat

391-357

748

A11

Aircraft-train

391-315

706

A12

Aircraft-truck

391-315

706

Intermodal transportation within the green supply chain

63

Appendix C
Example of concordance indices and credibility indexes computed for the
industrial scenario
Industrial
A01

Concordance indices
c(a, b3)

c(b3, a)

c(a, b2)

c(b2, a)

c(a, b1)

c(b1, a)

0.48

0.89

0.48

0.52

0.53

0.47

A02

0.41

0.94

0.48

0.59

0.95

0.15

A03

0.04

0.97

0.11

0.96

0.58

0.84

A04

0.37

0.63

0.41

0.63

0.46

0.55

A05

0.87

0.30

1.00

0.00

1.00

0.00

A06

0.50

0.50

0.50

0.50

0.59

0.42

A07

0.83

0.59

0.87

0.16

0.92

0.08

A08

0.41

0.88

0.95

0.55

0.95

0.05

A09

0.46

0.55

0.89

0.11

0.96

0.04

A10

0.04

0.96

0.46

0.54

0.95

0.45

A11

0.71

0.59

0.83

0.17

1.00

0.08

A12

0.37

0.63

0.41

0.59

1.00

0.55

Credibility index
c(a, b3)

c(b3, a)

c(a, b2)

c(b2, a)

c(a, b1)

c(b1, a)

A01

0.48

0.89

0.48

0.52

0.53

0.47

A02

0.41

0.94

0.48

0.59

0.95

0.13

A03

0.01

0.97

0.09

0.96

0.58

0.84

A04

0.37

0.63

0.41

0.63

0.46

0.55

A05

0.87

0.30

1.00

0.00

1.00

0.00

A06

0.50

0.50

0.50

0.50

0.59

0.42

A07

0.83

0.59

0.87

0.16

0.92

0.05

A08

0.41

0.88

0.95

0.55

0.95

0.03

A09

0.46

0.55

0.89

0.11

0.96

0.02

A10

0.02

0.96

0.46

0.54

0.95

0.45

A11

0.71

0.59

0.83

0.17

1.00

0.05

A12

0.37

0.63

0.41

0.59

1.00

0.55

64

M. Sawadogo and D. Anciaux

Appendix D
Outranking relations for a cutting level = 0.75
Outranking relations

Industrial

Ecological

Mixed

b3

b2

b1

b3

b2

b1

b3

b2

b1

<

A02

<

>

>

>

A03

<

<

<

<

<

<

<

A04

A05

>

>

>

>

>

>

>

>

>

A06

A07

>

>

>

>

>

>

>

A08

<

>

>

>

>

>

>

A09

>

>

A10

<

>

<

>

<

>

A11

>

>

>

>

A12

>

>

>

A01

Appendix E
Example of concordance and discordance indices calculation for the
alternative A01
A01

cj(a1, b3)

0.00

1.00

1.00

1.00

0.00

1.00

0.00

0.00

cj(a1, b2)

0.00

1.00

1.00

1.00

0.00

1.00

0.00

0.00

cj(a1, b1)

0.00

1.00

1.00

1.00

0.00

1.00

1.00

0.00

cj(b3, a1)

1.00

1.00

0.00

0.00

1.00

0.00

1.00

1.00

cj(b2, a1)

1.00

0.00

0.00

0.00

1.00

0.00

1.00

1.00

cj(b1, a1)

1.00

0.00

0.00

0.00

1.00

0.00

0.00

1.00

dj(a1, b3)

0.25

0.00

0.00

0.00

0.36

0.00

0.15

0.23

dj(a1, b2)

0.16

0.00

0.00

0.00

0.22

0.00

0.02

0.14

dj(a1, b1)

0.06

0.00

0.00

0.00

0.09

0.00

0.00

0.05

dj(b3, a1)

0.00

0.00

0.01

0.02

0.00

0.01

0.00

0.00

dj(b2, a1)

0.00

0.06

0.11

0.11

0.00

0.12

0.00

0.00

dj(b1, a1)

0.00

0.15

0.21

0.20

0.00

0.24

0.05

0.00

Intermodal transportation within the green supply chain

65

Appendix F
Exploitation procedures
The objective of the exploitation procedure is to exploit the binary relations. The role of
this exploitation is to propose an assignment of each alternative to a category.

Pessimistic rule. An action a will be assigned to the highest category Ch such that
aS bh1.
a Compare a successively with bh, h = k 1, k 2,0.
b The limit bh is the first encountered profile such that aS bh. Assign a to category
Ch+1.

Optimistic rule. An action will be assigned to the lowest category Ch such that bh > a.
a
b

Compare asuccessively with bh, h = 1, 2,, k 1.


The limit bh is the first encountered profile such that bh > a. Assign a to category
Ch.

You might also like