Research paper

An emergency order allocation model based

on multi-provider in two-echelon logistics
service supply chain
Liu Wei-hua
School of Management, Tianjin University, Tjanjin, China
Xu Xue-cai
Transportation Research Center, University of Nevada, Las Vegas, Nevada, USA, and
Ren Zheng-xu and Peng Yan
School of Management, Tianjin University, Tjanjin, China
Abstract
Purpose On one side, the purpose of this paper is to numerically analyze the emergency order allocation mechanism and help managers to
understand the relationship between the emergency coefcient, uncertainty and emergency cost in two-echelon logistics service supply chain. On the
other side, the purpose of this paper is to help managers understand how to deal with the problem of order allocation in the two-echelon logistics
service supply chain better in the case of emergency.
Design/methodology/approach The paper presents a multi-objective planning model for emergency order allocation and then uses numerical
methods withLINGO8.0softwaretoidentifythemodels properties. Theapplicationof theorder allocationmodel is thenpresentedbymeans of acasestudy.
Findings With the augment of uncertainty, the general cost of logistics service integrator (LSI) is increasing, while the total satisfaction of all
functional logistics service providers (FLSPs) is decreasing, as well as the capacity reliability; at the same time the emergency cost coefcient is closely
correlative with the satisfaction and general penalty intensity of FLSPs; nally, the larger the emergency cost coefcient is, the more satisfaction of
FLSPs, but the capacity reliability goes up rst and down later.
Research limitations/implications Management should note that it is not better when emergency cost coefcient is bigger. The general
satisfaction degree of FLSP increases with the augment of emergency cost coefcient, but there is an upper limit of the value, i.e. it will not increase
indenitely with the augment of emergency cost coefcient. This paper also has some limitations. The optional emergency cost coefcient only adopted
a group of data to analyze while the trend of the reliability of logistics capacity needs to be further discussed. In addition, the algorithm of emergency
order allocation model in the case of multi-objective remains to be solved.
Practical implications Under emergency conditions, LSIs can adopt this kind of model to manage their FLSPs to obtain the higher logistics
performance. But LSIs should be careful selecting emergency cost coefcient. In accordance with different degrees of emergency logistics demand, LSIs
can determine reasonable emergency cost coefcient, but not the bigger, the better, on the premise that LSIs acquire maximum capacity guarantee
degree and overall satisfaction degree of FLSPs. FLSPs can make contract bargaining of reasonable emergency coefcient with LSIs to make both sides
get the best returns and realize the benet balance.
Originality/value Many studies have emphasized the capacity allocation of manufactures, order allocation of manufacturing supply chain and
scheduling model of emergency resources without monographic study of supply chain order allocation of logistics service. Because the satisfaction
degree of FLSPs the cost of integrators needs to be considered in the process of order allocation, and the inventory cost of capacity does not exist, it is
different from the issue of capacity allocation planning of manufacture supply chain. Meanwhile, the match of different kinds of logistics service
capacity must be considered for the reason of the integrated feature of logistics service. Additionally, cost is not the most important decision objective
because of the characteristics of demand uncertainty and weak economy. Accordingly, this paper considers these issues.
Keywords Logistics service supply chain (LSSC), Emergency order allocation, Uncertainty, Emergency cost coefcient, Model,
Distribution management, Uncertainty management
Paper type Research paper
The current issue and full text archive of this journal is available at
www.emeraldinsight.com/1359-8546.htm
Supply Chain Management: An International Journal
16/6 (2011) 391400
q Emerald Group Publishing Limited [ISSN 1359-8546]
[DOI 10.1108/13598541111171101]
The research was supported by the National Science and Technology
Foundation Planning of China (Project Number 2006BAD30B08). The
authors appreciate the comments and help from PhD students Ping Yu
and Di Zhang from Shanghai Jiaotong University, China. The case study
is supported by the operation manager Shen Yi and marketing manager Li
Wen-ming from Tianjin Baoyun Logistics Company in China. The
suggestions of the reviewers are also gratefully acknowledged. This study
is also supported by the National Natural Science Foundation of China
(Grant No.70902044).
391
1. Introduction
During the past ten years, all kinds of unexpected incidents
have occurred, such as the wide range of rain and snow
disaster of South China in 2008, Wenchuan earthquake in
China, Indian Ocean tsunami in 2004, various terror attacks,
regional wars, res, economic crisis, SARS, strikes, etc and all
of these inuenced our world more intensely than ten years
ago. Numerous facts have shown that unexpected incidents
not only lead to huge economic losses, but inuence peoples
working and daily life, even society and politics stability,
therefore emergency response to the incidents is necessary,
especially in supply chain. During the course of response, the
quality of logistics operation directly determines whether the
impact of unexpected events can be rapidly eliminated or not.
It will be helpful to improve the enterprises and public
departments response ability to severe natural disasters and
other unexpected incidents by building up efcient and rapid
responsive logistics system.
The improvement of logistics emergency services relies on
the coordination among the various members of logistics
service supply chain (LSSC). Liu (2007) and Choy et al.
(2007), considered that the main structure of LSCC was
functional logistics service included from providers (FLSPs)
to logistics service integrator (LSI) and to manufacturer or
retailer, in which the FLSPs mean traditional third party
logistics rms, such as transportation enterprises, storage
enterprises, etc, and they are classied as the suppliers by LSI
when domestic or international service network is established
because of the singularity, normality and regionality.
Generally speaking, one LSI is associated with many
FLSPs. After the LSI receives the clients orders, they will
distribute the orders to FLSPs who would fulll the logistics
service orders by providing corresponding logistic capability.
The process is called order allocation, and the model
established is order allocation model.
Under emergency condition, how to allocate the orders is
the key to improving logistics operation efciency when
confronting several logistics service providers. Being different
from single FLSP, the LSI not only needs to consider the
reliability of logistics capacity supply and the reasonable
matching of all kinds logistics capacities in the state of
emergency (Lummus et al., 1998), but also takes into account
of the minimum procurement cost and maximum overall
satisfaction degree of LSP. Therefore, the emergency order
allocation of one LSI and various FLSPs is studied in this
paper, and the model established is analyzed.
2. Literature review
The special study of the supply chain order allocation at home
and abroad is still in the early stage while the issue of capacity
allocation closely related to order allocation, was studied in
the early 1990s. So the literature review will be focused on
these two aspects in this paper respectively.
As for supply chain order allocation, Chan et al. (2001)
studied the order release mechanisms in supply chain
management with a simulation approach. A simulation
model of a typical, single channel logistics network was
developed. A new order release approach is proposed which is
found to be superior to those analyzed previously and should
lead to improved supply chain performance. Menon and
Schrage (2002) investigated the customers order allocation
within paper making equipment, whose goal was to minimize
the total production cost. The problem can be formulated as
dual-angular integer programming with identical machines
inducing symmetry. Although large-scale mathematical
programming can be decomposed into smaller ones to be
solved, decomposing integer programming problem still
remains unsolved, while the symmetry intensies the
difculty. This paper develops an approach to solve
decomposable dual-angular integer programming, which can
be successfully applied to the paper-making industry.
Kawtummachai and Van Hop (2005) considered that there
were less papers focusing on the order allocation problem
after the literature review of supplier selection. Because of this
they studied the structure between one factory and multiple
suppliers. In that paper one method was presented to allocate
the orders and minimize the acceptable procurement costs,
meanwhile to guarantee the necessary order service level.
Moreover the price of order quantity and delivery level was
involved. Chan et al. (2004) proposed a heuristic
methodology for order distribution in a demand driven
collaborative supply chain. The optimization methodology
combines an analytic hierarchy process with genetic
algorithms. Chan and Chan (2006) nd an early order
completion contract approach to minimize the impact of
demand uncertainty on supply chains, the Simulation results
indicate that the proposed contract approach is able to
improve the performance measures of the system. Chan et al.
(2006) focuses on optimization of order due date fulllment
reliability in multi-echelon distribution network problems
with uncertainties present in the production lead time,
transportation lead time, and due date of orders. The
proposed algorithm integrates genetic algorithms with analytic
hierarchy process. A hypothetical three-echelon distribution
network is studied, and the computation results demonstrated
the reliability of the proposed algorithms.
An integrated multi objective decision-making process
between supplier selection and order allocation was studied
by Demirtas and U

stu n (2008). In this paper an analytic

network process (ANP) and multi objective mixed integer
linear programming (MOMILP) model was constructed,
which aimed at maximizing the total value of purchasing and
minimizing the budget and deciency probability. By
adopting ANP method the priority of suppliers was
calculated after comprehensive consideration of the prot,
opportunity, cost and risk, and then the solution of the
MOMILP model was given.
In recent years, Chinese researchers also have focused on
the order allocation. Ji (2005) established a mixed integer
linear programming model of order allocation oriented to
multi-product, multi-order and multi-period. In order to
shorten the solving time and avoid the illegal chromosome
occurrence, a mixed genetic algorithm based on genetic
algorithm with heuristic rules was designed. Aiming at the
dynamic game problem with incomplete information, Zhang
et al. (2006) established a dynamic game model with
incomplete information, and mechanism of reward and
punishment in the process of supplier selection according to
the data provided by the suppliers, such as the quality,
delivery timing, price, etc. By identifying the data the order
distribution of suppliers was discussed and the game model
and mechanism were validated by an example. Focusing on
maximizing the group prots, Xiang (2006) put forward an
order allocation model of single 0-1 programming with
An emergency order allocation model based on multi-provider
Liu Wei-hua, Xu Xue-cai, Ren Zheng-xu and Peng Yan
Supply Chain Management: An International Journal
Volume 16 Number 6 2011 391400
392
optimization theory. In the model, the factors, such as stock
share, capital cost, xed operation cost, etc, were taken into
account, and the optimal allocation of the market orders, can
be obtained by solving the 0-1 programming model, thus
compiling the production and transportation plans for the
member enterprises.
Since the order allocation process is closely related to the
allocation of capacity, it is necessary to review the relevant
research. Since the 1990s, the allocation of supply chain
capacities has been concentrated on, whose research mainly
emphasized on the coordination mechanism, the
mathematical programming model and mathematical
algorithm of capacity allocation.
Hung and Wang (1997) proposed a linear programming
model to solve the production plan and resource allocation
problem. Bard et al. (1999) studied the capacity planning
problem of semiconductor manufacturing facilities by the
method of queuing and simulation. Swaminathan (2000)
developed a heuristic method to solve the problem of capacity
planning and equipment procurement, and also proposed
integer programming model of random resource, which
considered demand uncertainty and make-to-order
production environment. Furthermore, Tiwari and Vidyarth
(2000) proposed a genetic algorithm to allocate the orders
and facilities. Wang and Lin (2002) applied genetic algorithm
to solve capacity expansion and allocation for a
semiconductor testing facility with a constrained budget.
Korpela et al. (2002) proposed that conventional supply chain
capacity allocation only considered cost and prot
optimization, which is obviously not enough under the
competitive business environment nowadays. They thought
supply chain allocation should not only consider customer
demand, but also the strategic importance and risks. Then a
solution framework was proposed by using Analytic Hierarchy
Process (AHP) and Mixed Integer Programming (MIP),
which included ve steps: step 1 is the denition of the
problem; step 2 is to analyze the strategy importance of
customers; step 3 is to analyze the risks of relationship
between suppliers and customers, step 4 is to analyze the
preference of supply chain structure from customers, and the
last step is the optimization employing the method of MIP.
Among these steps AHP is used to identify the importance
and risks of suppliers, while MIP is for the selection of the
supply chain capacity allocation. Sheu (2006) brought out a
novel dynamic resource allocation model for demand-
responsive city logistics distribution operations. The
proposed methodology is developed based on the following
ve developmental procedures, including:
1 classication of demand attributes;
2 customer grouping;
3 customer group ranking;
4 container assignment; and
5 vehicle assignment.
The numerical results indicated that the model can be
permitted dynamically to manage both the time-varying
customer order data and logistics resources allocation.
Considering the long-term investment of production
facilities and the speediness of demand variety Wang et al.
(2007) thought it is crucial for the decision of capacity
allocation. Then they solved the problem of resource
allocation of make-to-stock manufactures by means of
Constraint Programming Genetic Algorithm (CPGA).
As for the emergency order allocation, Dai and Da (2000)
presented a mathematic model of multi-resource emergency
problem. By introducing the denition of continuous feasible
scheme, the problem can be solved by using the existing
results of single resource model. Al and Gong (2007) built up
multi-objective supplier selection model on the basis of
delivery timing, quality and cost, but the solution algorithm
was not given. Li and Zhao (2007) constructed an emergency
material storage double scale limited location model based on
time-satisfaction. Integrating strategic emergency stock with
real options supply, Liu and Ji (2007) set up a resilient supply
model based on contingency supply. Li et al. (2007)
established an evaluation model of emergency logistics
warranty capacity.
To sum up, the researches above mainly emphasized on the
capacity allocation of manufactures, long-term investment of
capacities, order allocation of manufacturing supply chain and
scheduling model of emergency resources without
monographic study of supply chain order allocation of
logistics service. Because the satisfaction degree of FLSPs
the cost of integrators needs to be considered in the process of
order allocation, and the inventory cost of capacity does not
exist, it is different from the issue of capacity allocation
planning of manufacture supply chain. Meanwhile, the match
of different kinds of logistics service capacity must be
considered for the reason of the integrated feature of
logistics service. Additionally, cost is not the most important
decision objective because of the characteristics of demand
uncertainty and weak economy. What is more, when the
emergency order is allocated, the speed of service is very
important. So the emergency order is strongly required to be
completed on schedule. Accordingly, the emergency order
allocation of two-echelon LSSC is mainly studied in this
paper, and then a model is constructed, nally, the effects of
demand uncertainty and emergency cost coefcient on order
allocation are discussed in detail.
3. Emergency Order Allocation Model and its
solution method in two-echelon LSSC
Under emergency situation, assume that emergency logistics
orders are assigned to each FLSP by LSI and various
capacities provided by FLSPs match with each other possibly.
For example, if logistics service provider B offers 200
transport vehicles, he should give the warehouse of 1,000
square meters in accordance with the matching requirement
between transport and storage capacity, but if B cannot
provide such a warehouse, he might only provide the 500
square meters warehouse, or even not, in this way the
transport capacity and storage capability of B constitutes of a
incomplete matching binding relationship.
In the case of emergency logistics demand, two aspects
need to be considered specially: one is to guarantee the
quantity of logistics capacity required by the emergency order,
the other one is that within the limited time, the faster the
service speed, the better. Assume there are M kinds of
logistics service orders, and the total demand of the jth
logistics service is R
j
. Let the interval u
2
ij
; u

ij
_ _
donate the jth
logistics capacity interval provided by the ith FLSP,
and

N
i1
u
2
ij
# R
j
#

N
i1
u

ij
, furthermore, the capacity of
each logistics service provider is supposed to be independent,
so there is no correlation among them. There exists a time
An emergency order allocation model based on multi-provider
Liu Wei-hua, Xu Xue-cai, Ren Zheng-xu and Peng Yan
Supply Chain Management: An International Journal
Volume 16 Number 6 2011 391400
393
reliability interval T
2
ij
; T

ij
_ _
0 # T
2
ij
# 1; 0 # T

ij
# 1
_ _
of
accomplishing the logistics capacity within the range of
quantitative interval provided by each FLSP. And also it is
assumed that the larger the logistics capacity by the provider
is, the less the time reliability.
Let x
ij
denote the jth logistics order provided by LSI to the
ith FLSP, and the value of x
ij
may either lie within the range of
u
2
ij
; u

ij
_ _
or out of it. Considered the uncertainty of customer
requirements, the demand R
j
is regarded as a stochastic
variable following the normal distribution Nm
j
; s
2
j
, so the
random variable is included in the equation of constraint
condition of planning model chance constraint, i.e. the
equation Pr ob

N
j1
x
ij
$ R
j
a can be expressed as the
chance constraint meeting the customer requirements of the
total capacity of the jth logistics service provided by i FLSP,
while a denotes the satised probability. For example, a 95
percent means the total capacity of the jth (Guan et al., 2005;
Liu and Zhao, 1998) providing the method of
converting uncertainty constraints into certainty ones, the
capacity provided should satisfy the condition of

N
i1
x
ij
m
j
F
21
a s
j
.
When the model is applied actually, LSI should not only
take into account of the quantity guarantee of logistics
capacity and time guarantee of logistics service, but also
consider the matching of different logistics capacity, overall
satisfaction of FLSPs and overall cost of LSI, and all the
factors mentioned will be discussed in the followings.
3.1 Reliability of logistics capacity
Reliability of logistics capacity R
ij
means supply reliability of the
jthlogistics capacity providedby the ithFLSPunder emergency
condition. The reliability can be measured comprehensively
fromtwoaspects: one is numeric reliability, reecting the degree
of meeting the capacity provided by FLSP, the other one is time
reliability, reecting the degree of punctuality. The numeric
reliability r
ij
of the jth logistics capacity provided by the ith
FLSP can be described as
r
ij

u

ij
xij
; x
ij
. u

ij
1; x
ij
# u

ij
_

_
1
The time reliability t
ij
is expressed as
t
ij

u

ij
xij
T
2
ij
; x
ij
. u

ij
T
2
ij

u

ij
2xij
u

ij
2u
2
ij
_ _
T

ij
2T
2
ij
; u
2
ij
# x
ij
# u

ij
T

ij
; x
ij
, u
2
ij
_

_
2
Reliability of logistics capacity R
ij
is total reection of numeric
reliability and time reliability. According to the preference of r
ij
and t
ij
required by emergency orders, weighted l
1
and l
2
can be
given, Then the R
ij
can be
R
ij
l
1
r
ij
l
2
t
ij
3
3.2 Degree of capacity matching
When emergency logistics order is allocated, LSI need to
disassemble clients logistics order, and allocate all kinds of
logistics orders (e.g. transportation service order, storage service
order, processing service order) required by clients among
FLSPs. In this model, when each FLSP conducts the business,
their logistics capacities are required to match with each other.
For example, if FLSP A provides 200 highway transport
vehicles, it should provide 1,000 square meters warehouse
possibly to match its transportation capacity. But if it cannot, it
can accommodate nothing or just 500 square meters warehouse.
Degree of capacity matching is introduced to describe
matching situation of different logistics capacities. Let w
kl
denote the capacity matching degree between the kth logistics
service capacity (e.g. transport capacity) and the lth logistics
service capacity (e.g. storage capacity). If k l , then w
kl
1,
w
kl
1=wlk. In incomplete matching relations d
ikl
denotes the
degree of non-exact matching between the kth logistics service
capacity (e.g. transport capacity) and the lth logistics service
capacity (e.g. storage capacity) provided by ith FLSP. The
equation d
ikl
0 denotes exact match, while the equation
d
ikl
1 denotes they do not match completely.
d
ikl
can be dened as equation (4).
d
ikl
1 2w
kl
x
il
x
ik
_ _
2
4
Considered x
ik
can be equal to 0, equation (4) can be
converted into the following equation (5).
d
ikl

1 2w
kl
xil
xik
_ _
2
; x
ik
0
1; x
ik
0
_
_
_
5
3.3 Degree of FLSPs satisfaction
In the actual process of emergency order allocation, generally
speaking, the larger the emergency order quantity allocated to
FLSP, the more satisfactory of FLSP, because the more the
emergency order to FLSP, the greater the return of FLSP. But
the order allocated cannot be innite, when the order quantity
is beyond the maximum capacity valueu

ij
, the satisfaction of
FLSP decreases, because the extra costs should be paid by
FLSP for the extra orders.
Assume d
ij
is the ith FLSPs satisfaction degree of the jth
logistics task. The model of satisfaction degree is constructed
as follows:
d
ij

u

ij
xij
; x
ij
. u

ij
d
0
ij

xij 2u
2
ij
u

ij
2u
2
ij
*
1 2d
0
ij
; u
2
ij
# x
ij
# u

ij
xij
u
2
ij
*
d
0
ij
; 0 # x
ij
, u
2
ij
_

_
6
where d
0
ij
denotes the ith FLSPs initial satisfaction degree of
the jth logistics capacity value, when the value of the jth
logistics service order assigned to FLSP ith by LSI is equal to
u
2
ij
which is the minimum value of capacity provided.
Since different FLSPs satisfaction of different logistics
capacity has different preference, it can be divided into two
types, which are single capacity satisfaction degree and
general capacity satisfaction degree. The weight of single
satisfaction degree of capacity is assumed to be 1
1
, the weigh
of general satisfaction degree of capacity is 1
2
, and
1
1
1
2
1. The values of 1
1
and 1
2
can be obtained by
questionnaires lled by FLSP.
An emergency order allocation model based on multi-provider
Liu Wei-hua, Xu Xue-cai, Ren Zheng-xu and Peng Yan
Supply Chain Management: An International Journal
Volume 16 Number 6 2011 391400
394
1 Single capacity satisfaction degree: it means order
allocation satisfaction degree of a certain capacity. Let
l
ij
denote the jth logistics capacity preference of FLSP ith,
where

M
j1
l
ij
1. So the total satisfaction degree of i
FLSPs single capacity is 1
1

M
j1
l
ij
d
ij
_ _
.
2 General capacity satisfaction degree: it means order
allocation satisfaction degree of multi-capacity for FLSP.
For example, in many cases, logistics service provider
would like to provide integrated logistics service, which
means not only the transport capacity and storage
capacity, but also distribution processing capacity will be
provided. Its logistics capacity is integrated, the
coordination and balance of various capacities is
emphasized. In general, the more the average quantity of
various emergency order allocations gotten by providers
is, the better the allocation result of integrated logistics
orders will be, then the higher its satisfaction degree of
general capacity, so the general capacity satisfaction
degree of ith FLSP is 1
2
1
M

M
j1
d
ij
_ _
.
3 In this way, the ultimate satisfaction of order allocation of
ith FLSP is d
i
1
1

M
j1
l
ij
d
ij
_ _
1
2
1
M

M
j1
d
ij
_ _
.
3.4 Construction of Emergency Order Allocation Model
Due to the different importance of FLSPs, the distribution
status of FLSPs in the course of emergency orders allocation
is different. Let b
i
denote the importance of the ith FLSP and

N
i1
b
i
1. The jth logistics demands unit operating costs
of the ith FLPS is c
ij
, the emergency order allocation of LSI
should meet the following four requirements:
1 maximizing the reliability of logistics capacity;
2 maximizing the degree of FLSPs satisfaction, which is the
minimum difference between application value and
optimal allocation value of FLSP;
3 minimizing the cost of LSI;
4 maximizing the degree of capacity matching for each
FLSP as much as possible.
In incomplete match relations, the optimal objective function
of this model is
max Z
1

N
i1

M
j1
b
i
R
ij

N
i1

M
j1
b
i
l
1
r
ij
l
2
t
ij
7
max Z
2

N
i1
b
i
1
1

M
j1
l
ij
d
ij
1
2
1
M

M
j1
d
ij
_ _ _ _
8
min Z
3

N
i1

M
j1
c
ij
x
ij
9
minZ
4

N
i1

M
l1

kl
k1
d
ikl
10
s:t:

N
i1
x
ij
m
j
F
21
a s
j
x
ij
$ 0
_
_
_
_
_
i 1; 2; ; N;
j 1; 2; ; M
11
3.5 Solution of Emergency Order Allocation Model
The model above is a multi-objective planning problem. Due
to conicting and incommensurability of each objective in
multi-objective decision problem, it is hard to nd an
absolutely optimal solution, thus the common method used is
to compromise among objectives in order to get Pareto
optimal solution required by the decision-maker (Liu and
Huang, 2001). At present, there are specic solutions to
multi-objective problems (Xu and Li, 2005; Qian, 1990; Liu
et al., 2003), such as evaluation function method (linear
weighting method, reference to objective method, minimum
maximum basis method), objective programming method,
hierarchical sequence method, interactive programming
method, subordination function method, etc. Different from
the methods mentioned above, because there are four aims in
this model, so we could use the thought of the solution, which
is to transfer the multi-objective problem into single objective
problems. The basic steps are as follows.
Step 1: Considered objective functions Z
1,
Z
2
and Z
3
without Z
4
, objective function Z
3
is changed into the
following equation by introducing emergency cost coefcient.
Z
3
# 1 dZ
*
3
12
Where Z
*
3
denotes the minimum of Z
3
.
The objective of this approach is not minimizing the total
cost, but limiting the cost within a reasonable range. In fact,
because of the week economy of emergency logistics, cost
reduction is not a major objective, so this method is feasible.
Let d, a small positive number, be emergency cost coefcient.
If d 0.1, it means LSI can support the emergency cost
10 percent higher than minimum cost to speed up emergency
response time, and also it will be helpful for completing
emergency order.
Then the objectives Z
1
and Z
2
are merged into a new max
type function, which is as follows:
max F a
1
Z
1
a
2
Z
2
13
s:t:
Z
3
# 1 dZ
*
3

N
i1
x
ij
m
j
F
21
a s
j
x
ij
$ 0
_

_
_
_
_
_
_
_
i 1; 2; ; N;
j 1; 2; ; M
14
The optimal F value can be gotten, i.e. F
*
.
Step 2: Since Z
4
is close to zero as much as possible, the
positive deviation u

2
can be added to it, where u

2
$ 0; As for
the objective function F already solved, negative deviation u
2
1
can also be added, where u
2
1
$ 0. Then let P
1
; P
2
denote its
priority, then the objective function can be changed into:
minP
1
u
2
1
P
2
u

2
15
s:t:
Z
3
u
2
1
# 1 dZ
*
3

N
i1
x
ij
m
j
F
21
a s
j
Z
4
2u

2
$ 0
x
ij
$ 0
_

_
_
_
_
_
_
_
_
_
_
i 1; 2; ; N;
j 1; 2; ; M
16
An emergency order allocation model based on multi-provider
Liu Wei-hua, Xu Xue-cai, Ren Zheng-xu and Peng Yan
Supply Chain Management: An International Journal
Volume 16 Number 6 2011 391400
395
Step 3: Solve the new model gotten from Step 2 by using
multi-objective planning method.
4. The effects of demand uncertainty and
emergency cost coefcient on order allocation in
two-echelon LSSC
In view of the model built in this paper, software LINGO 8.0
is used to examplify.
4.1 Assumption of essential data of the model
Under an emergency situation, there are ve FLSPs, namely, B,
C, D, E, F, providing logistics service to a LSI A. The integrator
A distributes transportation orders to the ve FLSPs. Assume
the transportation capacity that LSI A requires follows normal
distribution D1,N (200, s
2
1
), where a 95 percent; storage
capacity required follows normal distribution D2,N(130, 16),
where a 95. LSI demands that the proportion of
transportation capacity to storage capacity should meet the
ratio of 20 to 13 as nearly as possible. Correlation parameters of
transportation capacity and storage capacity of B, C, D, Eand F
are shown in Table I. The weight l
1
and l
2
of number reliability
and time reliability are equal, and both is 0.5.
4.2 Effects of demand uncertainty on emergency order
allocation
If the mean value of transportation orders is 200, the variance
s
2
1
are 4, 16, 36, 64 respectively, and the mean value of
storage orders subjects to D2,N (130, 16), the values of
effects of demand uncertainty on emergency orders are shown
in Tables IIIII.
Table displays that under emergency circumstances the
demand uncertainty has great impact on total cost of LSI.
The larger the demand uncertainty is, the more the total cost
of LSI is, and the lower the general satisfaction of FLSPs is,
and the capacity reliability.
4.3 Effects of emergency cost coefcient d of LSI on the
results of order allocation
The effects of correlation parameters(total cost, reliability of
logistics capacity and overall satisfaction degree) on the order
allocation results based on the analysis above are shown in
Table IV and Figure 1, when the emergency cost coefcient
are 5 percent, 10 percent, 15 percent and 20 percent
respectively.
As shown in Table IV, emergency cost of LSI is positively
correlated with its total cost. The larger the emergency cost
coefcient is, the more total cost of LSI is. From Figure 1 it
can be seen that emergency cost coefcient has positive
correlation with the general capacity satisfaction degree of
FLSP. The larger the emergency cost coefcient is, the more
total cost of LSP will be. However, as shown in Table IV, the
reliability of logistics capacity does not increase with the
emergency cost coefcient; it rises rst then drops later.
Furthermore, if the range of emergency cost coefcient is
made from 0 to 100%, new results showed in Figure 2 will be
gotten. According to the Figure 2, in the curve of general
reliability of logistics capacity, point A is the highest point and
the corresponding emergency cost coefcient is 5 percent.
Point B is the balance point of general reliability of logistics
capacity, which means the general reliability of logistics
capacity will keep stable with the emergency cost coefcient
increasing, and the emergency cost coefcient of point B is 85
percent. In the curve of general capacity satisfaction degree,
with the increasing of the emergency cost coefcient, the
general capacity satisfaction degree tends to go up, but the
amplitude will become slower. From the point view of
management results of Figure 2 show that it is not better
when emergency cost coefcient is bigger. The general
satisfaction degree of FLSP is increasing with the augment of
emergency cost coefcient, but there is an upper limit of the
value, i.e. it will not increase indenitely with the augment of
emergency cost coefcient.
Table I Correlation parameters of logistics capacity of FLSP B, C, D, E, F
FLSP c
i1
u
2
i1
; u
1
i1
_
T
2
i 1
; T
1
i 1
_
c
i 2
u
2
i 2
; u
1
i 2
_
T
2
i 2
; T
1
i 2
_
l
i1
l
i2
1
1
1
2
b
i
k
i
d
0
i 1
d
0
i 2
B 10 [50, 70] [0.1,0.9] 15 [20, 40] [0.2, 0.9] 0.4 0.6 0.4 0.6 0.22 0.55 0.2 0.3
C 8 [25, 40] [0.3,0.9] 13 [20, 35] [0.3, 1] 0.35 0.65 0.5 0.5 0.13 0.3 0.35 0.35
D 9 [35, 90] [0.2,0.7] 16 [25, 40] [0.5, 0.8] 0.5 0.5 0.3 0.7 0.3 0.6 0.15 0.3
E 12 [40, 60] [0.4,1] 20 [25, 50] [0.4, 0.7] 0.6 0.4 0.6 0.4 0.15 0.4 0.3 0.2
F 11 [30, 80] [0.4,0.8] 18 [10, 40] [0.2, 0.9] 0.45 0.55 0.65 0.35 0.2 0.5 0.25 0.3
Table II Effects of uncertainty demand on emergency order allocation (1)
Distribution value of
transportation Distribution value of storage Total satisfaction degree Reliability of logistics capacity
FLSP s
2
1
54 s
2
1
516 s
2
1
536 s
2
1
564 s
2
1
54 s
2
1
516 s
2
1
536 s
2
1
564 s
2
1
54 s
2
1
516 s
2
1
536 s
2
1
564 s
2
1
54 s
2
1
516 s
2
1
536 s
2
1
564
B 28.55 50.01 43.01 12.12 20 29.96 24.57 21.63 0.195 0.397 0.298 0.135 0.95 0.862 0.91 0.95
C 26.77 28.87 36.6 40.00 48.01 29.27 32.35 34.94 0.555 0.617 0.866 0.469 0.695 0.828 0.715 0.312
D 76.17 84.23 80.41 87.23 28.57 44.58 46.17 40.00 0.393 0.904 0.859 0.161 0.765 0.649 0.646 0.875
E 6.016 13.47 19.84 0 0 0.0 0.00 0 0.021 0.046 0.068 0.042 0.925 0.925 0.925 0.925
F 65.77 30.00 30.00 73.80 40.00 32.76 33.49 40 0.890 0.530 0.538 0.828 0.91 0.933 0.933 0.899
An emergency order allocation model based on multi-provider
Liu Wei-hua, Xu Xue-cai, Ren Zheng-xu and Peng Yan
Supply Chain Management: An International Journal
Volume 16 Number 6 2011 391400
396
4.4 Model results discussion
Based on the model results, we can draw the following
conclusions:
.
It shows a positive correlation between cost coefcient and
FLSPs overall satisfaction; the large the cost coefcient is,
the more the FLSPs overall satisfaction is. But cost
coefcient and reliability of logistics capacity are not
entirely positively correlated, but rst positively
correlated, then negatively correlated.
.
For LSI, in the actual distribution of emergency logistics
orders, increasing the emergency cost coefcient will
reduce their own prots, but can improve FLSPs
satisfaction, enable the emergency logistics mission to be
completed better. So on the premise of obtaining
maximum reliability of logistics capacity and FLSPs
satisfaction, LSI must develop an appropriate emergency
cost coefcient in accordance with the emergency logistics
demands, rather than pursuit the maximum emergency
cost coefcient. To win the game about emergency orders
allocation with FLSP, LSI can set in advance the different
arrange of emergency cost coefcient according to the
different emergency logistics demands. Once there were
Table III Effects of uncertainty demand on emergency order allocation
(2)
s
2
1
Total cost
Overall satisfaction
degree
Reliability of logistics
capacity
4 4082.232 0.5442759 0.849421
16 4113.816 0.4519563 0.837756
36 4145.4 0.4141338 0.832554
64 4172.487 0.3112 0.8307
Table IV Effects of emergency cost coefcient of LSI on the results of
order allocation
Total cost
General capacity
satisfaction degree
General reliability of
logistics capacity
5 3599.589 0.0528 0.8264
10 3770.998 0.2110 0.8487
15 3942.407 0.47372 0.8635
20 4113.816 0.55196 0.8378
Figure 1 Schematic diagram of overall satisfaction degree reliability of logistics capacity and emergency cost coefcient (a)
Figure 2 Schematic drawing of overall satisfaction degree reliability of logistics capacity and emergency cost coefcient (b)
An emergency order allocation model based on multi-provider
Liu Wei-hua, Xu Xue-cai, Ren Zheng-xu and Peng Yan
Supply Chain Management: An International Journal
Volume 16 Number 6 2011 391400
397
emergency logistics demands, they can make a decision
easily.
.
For FLSP, increasing the emergency cost coefcient may
increase their prots. They can negotiate the appropriate
emergency cost coefcient with LSI so that both sides get
the best returns. For instance, before the emergency
logistics demands occurring, both sides make an
agreement to take the price increasing ratio into
consideration under different situations. When it
happened, they can easily choose the price increasing
ratio.
5 Model extensions: emergency order allocation
problem related to a multi-echelon LSSC
The two-echelon logistics service supply chain is the typical
structure of LSSC, whose order allocation model can be
extended to multi-echelon LSSC. Taking a three-echelon
LSSC as an example, as shown in Figure 3, the FLSP of LSI
A is LSI B,while the FLSP of LSI B is FLSP C. As for the
two-echelon LSSC comprised of A and B, the order allocation
model of this study can be applied, likewise, the two-echelon
LSSC comprised of B and C can be done.
Under the condition of emergency order allocation of
multi-echelon LSSC, LSI A needs to take into account of the
operation quality and performance of FLSP. Because the
logistics service is intangible and the quality cannot be easily
measured, LSI A can take two measures to guarantee the
service performance. On one hand, A should sign detailed
order service standard contract with provider B, if B can not
complete the logistics service due to various reasons
(including the operation mistakes of FLSP C), B will be
published by LSI correspondingly. On the other hand, A can
ask B to provide the relevant information of FLSP C, once
FLSP C makes mistakes or can not complete the
requirements within the standards, A can communicate and
coordinate with B by the relevant information of FLSP
C. These two steps can completely make sure that A can
realize emergency logistics service performance requested by
clients.
6. Application of Emergency Order Allocation
Model: an empirical example from Tianjin Baoyun
Logistics Company in China
The emergency order allocation model proposed in this paper
has been put into practice. Tianjin Baoyun Logistics
Company in China is one of the successful examples, whose
operational experience is worth learning by many logistics
service integrators.
Tianjin Baoyun Logistics Co. Ltd is a professional third
party logistics provider and also is an outstanding logistics
integrator. In 2005, 2006 and 2007, the company was reputed
as the Top 100 logistics enterprises in China and Top Five
logistics enterprises in Tianjin city. Currently, the company
has set up 28 divisions and built good collaboration relations
with 32 large warehousing companies, 20 transportation
companies and 15 professional logistics company. By
integrating these functional logistics service providers,
Baoyun has successfully established widespread business
relations with more than 20 multi-national enterprises, such
as P&G, Siemens, Delphi etc and provided personalized
service according to clients requirements.
According to emergency logistics demand, Baoyun
Logistics Company has much practical experience of dealing
with emergency order allocation, which is worth learning.
With many years operation experience, Baoyun Logistics
Company divided demand of emergency logistics into three
degrees, important emergency event (degree A), big
emergency event (degree B) and general emergency event
(degree C). The emergency cost coefcient increases with the
emergency degree rising, but its not innite and conned to
some extent. Under the condition of general emergency
events (degree C), emergency cost coefcient of the company
is no more than 20 percent. In degree B, emergency cost
coefcient is no more than 50 percent and in degree A, the
emergency cost coefcient is no more than 80 percent.
Before the emergency logistics demand occurs, Baoyun
Logistics Company set up a dedicated le for every FLSP
when they signed cooperation agreements with these FLSP.
These les not only record the type and quantity of service
capacities, but also the time range of service operation and the
service speed. Based on the les, Baoyun Logistics
Companys Operations Management Department judge and
classify the service operation capacity of each FLSP, normally
as three types: Highly Flexible Service, General Flexible
Service and Poor Flexible Service. The classication results
are key references to allocate emergency logistics demands.
In order to improve the capacity reliability of FLSP under
the emergency condition, Baoyun Logistics Company
considers the price increasing ratio in the contract with
FLSP, and the range of the price increasing is conned within
certain limits. For example, when general emergency events
(degree C) happen, the price of transportation outsourced to
FLSPs will increase by 10 per cent to 20 percent. If some
emergency events are beyond the scope of the contract,
Baoyun Logistics Company will coordinate with FLSP to
reach an agreement of the specic price ratio, and the
contracts with FLSP must be approved by general manager.
After the emergency logistics demand occurs, Baoyun
Logistics Company rstly determines the timing requirements
of clients to emergency logistics demands, then classies
emergency logistics demands based on timing requirements.
Baoyun Logistics Company noties each FLSP the timing
requirements, and with the detailed les of FLSP, it chooses
the FLSP who has rapid service speed and highly exible
service as the alternative supplier to join the emergency
logistics orders allocation. Secondly, Baoyun Logistics
Company needs to estimate the demand degree based on
the classication from clients, determines the emergency cost
coefcient, which can be afforded by the company, and then
consults with each FLSP. As a logistics service integrator,
Baoyun Logistics Company has many FLSPs. Once a client
requires emergency logistics, specic persons from market
department of Baoyun Logistics company will take charge of
emergency order allocation; consider comprehensively various
factors, such as service speed, capacity reliability, capacity
matching, satisfaction, etc for FLSP, then sign the contract
with allocation result and service implementation standard,
Figure 3 Typical structure of three-echelon LSSC
An emergency order allocation model based on multi-provider
Liu Wei-hua, Xu Xue-cai, Ren Zheng-xu and Peng Yan
Supply Chain Management: An International Journal
Volume 16 Number 6 2011 391400
398
and nally transfer it to the operation management
department for uniform monitoring.
As to emergency order allocation of multi-echelon LSSC
network, Baoyun Logistics Company has its own operation
procedure. For example, Delphi Company, one of the clients,
had one emergency logistics demand in June 2007. Baoyun
Logistics Company outsourced the emergency logistics order
to Qidong Logistics Company, Xingsheng Logistics
Company, Feilong Freight Logistics Company and Wanfang
Logistics Company. Among them, the transport volume to the
Qidong logistics company was the largest, and 100 5T-trucks
were required, which surpassed its available maximum
transport capacity of 80. But at the same time because
Qidong Logistics Company is also a small-scale LSI, the
company provided another 20 5T-trucks from its FLSPs to
satisfy the demand of Baoyun Company. In order to
guarantee that the FLSP can nish the tasks in accordance
with the standards, Baoyun Company and Qidong Company
signed the detailed logistics service standards, meaning that if
the FLSPs of Qidong logistics company cannot reach the
service standards, Qidong Company would be punished
accordingly. Baoyun Company did not participate in the
coordination of Qidong Company and its FLSPs.
The successful practice of Baoyun Logistics Company
dealing with emergency order allocation veries the
effectiveness of the model results in this paper. For
example, when deals with emergency logistics orders,
Baoyun Logistics Company considers service speed rstly.
With les recorded, it can choose high service speed FLSP as
the alternative supplier to join the emergency logistics orders
allocation. Then in the actual allocation of orders, Baoyun
Logistics Company also gives full consideration to: the
FLSPs ability to provide the right quantity of logistics service
in the right time; the degree of overall satisfaction of FLSP;
the total cost of SI. These factors are also considered in the
model built in my paper. So the model is in line with the
actual operation of enterprises, and it is scientic and
reasonable.
In addition, the operation experiences of Baoyun Logistics
Company proof the validity of the model results. For instance,
under the condition of three different emergency demands,
the emergency cost coefcient of Baoyun Company varies
correspondingly, i.e. the conclusion of this study is
reasonable, the bigger the emergency cost coefcient, the
results are not better. Before the emergency logistics demand
happens, Baoyun Company takes the price increasing ratio
into consideration under different situations in the contract
signed with FLSP, which indicates that, it would be helpful
for the overall satisfaction of providers to appoint the
emergency cost coefcient with the clients in advance in
order to realize superior reliability of logistics capacity.
7. Conclusions and future work
The issue of emergency order allocation, under emergency
circumstance, when LSI facing various FLSP is investigated
in this paper. The model of emergency order allocation based
on multi logistics capacity is presented, which considers the
maximization of reliability of logistics capacity, the
minimization of the cost of LSI, the maximization of
general satisfaction degree of FLSP and the match of each
capacity. The solution is given in the model and the effects of
demand uncertainty and emergency cost coefcient on order
allocation in two-echelon LSSC are analyzed by using
practical example. The results show that with the augment
of uncertainty, the total cost of LSI is increasing, while the
general satisfaction of all FLSPs is decreasing, which is the
same as the reliability of logistics capacity of all FLSPs.
Meanwhile, the emergency coefcient is closely correlated to
the general satisfaction degree. The greater of emergency cost
coefcient, the more satisfaction of FLSPs, but the reliability
of logistics capacity does not have the same trend; it rises rst
then drops later.
The model of emergency order allocation in this paper has
some limitations. For example, the paper mainly considers the
emergency order allocation model of single-cycle; the future
study can be extended to multi-cycle situation to investigate
the emergency order allocation tactics. The optional
emergency cost coefcient only adopted a group of data to
analyze while the trend of the reliability of logistics capacity
needs to be further discussed. In addition, the algorithm of
emergency order allocation model in the case of multi-
objective remains to be solved.
References
Al, J. and Gong, D. (2007), Study on emergency materials
suppliers selection based on multi-objective decision-
making, Logistics Sci-Tech, Vol. 30 No. 6, pp. 173-6
(in Chinese).
Bard, J.F., Srinivasan, K. and Tirupati, D. (1999),
An optimization approach to capacity expansion in
semiconductor manufacturing facilities, International
Journal of Production Research, Vol. 37 No. 15, pp. 3359-82.
Chan, F.T.S., Chung, S.H. and Choy, K.L. (2006),
Optimization of order fulllment in distribution network
problems, Journal of Intelligent Manufacturing, Vol. 17,
pp. 307-19.
Chan, F.T.S., Chung, S.H. and Wadhwa, S. (2004),
A heuristic methodology for order distribution in a
demand driven collaborative supply chain, International
Journal of Production Research, Vol. 42 No. 1, pp. 1-19.
Chan, F.T.S., Humphreys, P.K. and Lu, T.H. (2001), Order
release mechanisms in supply chain management:
a simulation approach, International Journal of Physical
Distribution and Logistics Management, Vol. 31 No. 1,
pp. 124-39.
Chan, H.K. and Chan, F.T.S. (2006), Early order
completion contract approach to minimise the impact of
demand uncertainty on supply chains, IEEE Transactions
on Industrial Informatics, Vol. 2 No. 1, pp. 48-58.
Choy, K.L., Li, C.L., So, S.C.K., Lau, H., Kwok, S.K. and
Leung, D.W.K. (2007), Managing uncertainty in logistics
service supply chain, International Journal of Risk
Assessment and Management, Vol. 7 No. 1, pp. 19-30.
Dai, G. and Da, Q. (2000), The study of combinatorial
scheduling problem in emergency systems, Systems
Engineering Theory & Practice, Vol. 20 No. 9, pp. 52-5
(in Chinese).
Demirtas, E.A. and U

stu n, O. (2008), An integrated

multiobjective decision making process for supplier
selection and order allocation, Omega, Vol. 36 No. 1,
pp. 76-90.
Guan, Z., Zhou, B. and Ma, Q. (2005), Research on vendor
selection & order quantity allocation problem under multi-
An emergency order allocation model based on multi-provider
Liu Wei-hua, Xu Xue-cai, Ren Zheng-xu and Peng Yan
Supply Chain Management: An International Journal
Volume 16 Number 6 2011 391400
399
product purchases, Computer Integrated Manufacturing
Systems, Vol. 11 No. 11, pp. 1626-32 (in Chinese).
Hung, Y.F. and Wang, Q.Z. (1997), A new formulation
technique for alternative material planning an approach
for semiconductor bin allocation planning, Computer and
Industrial Engineering, Vol. 32 No. 2, pp. 281-97.
Ji, X. (2005), Order allocation model in supply chain and
hybrid genetic algorithm, Journal of Southwest Jiaotong
University, Vol. 40 No. 6, pp. 811-5 (in Chinese).
Korpela, J., Kylaheiko, K., Lehmusvaara, A. and Tuominen,
M. (2002), An analytic approach to production capacity
allocation and supply chain design, Int. J. Production
Economics, Vol. 78, pp. 187-95.
Kawtummachai, R. and Van Hop, N. (2005), Order
allocation in a multiple-supplier environment, Int.
J. Production Economics, Vol. 94 No. 1, pp. 231-8.
Liu, B. and Zhao, R. (1998), Stochastic Programming and
Fuzzy Programming, Tsinghua University Press, Beijing,
pp. 75-88 (in Chinese).
Liu, B., Zhao, R. and Wang, G. (2003), Uncertain
Programming and Its Application, Tsinghua University
Press, Beijing, pp. 74-5 (in Chinese).
Li, J., Tang, S. and Jiang, Z. (2007), Study on evaluation
model for warrant of emergency logistics, Journal of
Lanzhou Jiaotong University, Vol. 26 No. 6, pp. 64-7
(in Chinese).
Li, J. and Zhao, L. (2007), Emergency material storage
double scale limited location model based on time-
satisfaction, Journal of Southeast University (Natural
Science Edition), Vol. 37, Supplement (in Chinese).
Liu, W. (2007), Research on the coordination of capacity
cooperation in logistics service supply chain, doctor
dissertation of Shanghai Jiaotong University, Shanghai.
Liu, X.W. and Huang, W. (2001), Fuzzy inference based
optimization method for multiobjective decision making,
Journal of Management Sciences in China, Vol. 4 No. 6,
pp. 71-6 (in Chinese).
Liu, X. and Ji, J. (2007), Resilient supply network design
based on contingency supply, Control and Decision, Vol. 22
No. 11, pp. 1223-6 (in Chinese).
Lummus, R.R., Vokurka, R.J. and Alber, K.L. (1998),
Strategic supply chain planning, Production and Inventory
Management Journal, Vol. 39 No. 3, pp. 49-58.
Menon, S. and Schrage, L. (2002), Order allocation for
stock cutting in the paper industry, Operations Research,
Vol. 50 No. 2, pp. 324-32.
Qian, S. (1990), Operation Theory, Tsinghua University Press,
Beijing, pp. 7448-51 (in Chinese).
Sheu, L.-B. (2006), A novel dynamic resource allocation
model for demand-responsive city logistics distribution
operations, Transportation Research Part E, Vol. 42 No. 6,
pp. 445-72.
Swaminathan, J.M. (2000), Tool capacity planning for
semiconductor fabrication facilities under demand
uncertainty, European Journal of Operational Research,
Vol. 120 No. 3, pp. 545-58.
Tiwari, M.K. and Vidyarth, N.K. (2000), Solving machine
loading problems in a exible manufacturing system using a
genetic algorithm based heuristic approach, International
Journal of Production Research, Vol. 38 No. 14, pp. 3357-84.
Wang, K-J. and Lin, S.H. (2002), Capacity expansion and
allocation for a semiconductor testing facility with a
constrained budget, Production Planning and Control,
Vol. 13 No. 5, pp. 429-37.
Wang, S.M., Chena, J.C. and Wang, K.J. (2007), Resource
portfolio planning of make-to-stock products using a
constraint programming-based genetic algorithm, Omega,
Vol. 35 No. 2, pp. 237-46.
Xiang, J., Huang, P. and Wang, Z. (2006), Order allocation
model based on prot maximization of horizontal
conglomerate, Journal of Southwest Jiaotong University,
Vol. 41 No. 2, pp. 241-4 (in Chinese).
Xu, J. and Li, J. (2005), Theory and Method of Multi-Objective
Decision, Tsinghua University Press, Beijing, pp. 80-125 (in
Chinese).
Zhang, X.-M., Li, G.-Q. and Zhang, Y. (2006), Dynamic
game study on the incomplete information about order
distribution of suppliers in supply chains, Chinese Journal
of Management, Vol. 3 No. 5, pp. 519-23 (in Chinese).
Corresponding author
Liu Wei-hua can be contacted at: lwhliu@yahoo.com.cn
An emergency order allocation model based on multi-provider
Liu Wei-hua, Xu Xue-cai, Ren Zheng-xu and Peng Yan
Supply Chain Management: An International Journal
Volume 16 Number 6 2011 391400
400
To purchase reprints of this article please e-mail: reprints@emeraldinsight.com
Or visit our web site for further details: www.emeraldinsight.com/reprints