Professional Documents
Culture Documents
Series Editor
Rajkumar Roy
123
Editors
Hing Kai Chan Olatunde Amoo Durowoju
University of East Anglia University of East Anglia
Norwich Business School Norwich Business School
Norwich NR4 7TJ Norwich NR4 7TJ
Norfolk, UK Norfolk, UK
Fiona Lettice
University of East Anglia
Norwich Business School
Norwich NR4 7TJ
Norfolk, UK
ISSN 1619-5736
ISBN 978-1-4471-4032-0 ISBN 978-1-4471-4033-7 (eBook)
DOI 10.1007/978-1-4471-4033-7
Springer London Heidelberg New York Dordrecht
The aim of this book is to provide a collection of chapters which show the latest
developments in decision-making for supply chain integration. The chapters
together highlight the key problems in managing and integrating supply chains.
The authors then present concepts, tools and methodologies to help to solve these
problems and to improve supply chain integration, decision making and
performance.
In recent years, integrated supply chain problems have been linked quite closely
to the artificial intelligence (AI) community. This is mainly because of the com-
plexity and uncertainty present in supply chains. These techniques can help to
lower the operating costs by allocating the resources within a supply chain wisely.
Among many AI techniques, genetic algorithms (GAs) and multi-agent systems
(MASs) are by far the most common approaches used. This book presents two
pieces of contemporary research in these two areas.
Yeung and Lee in Chap. 1 apply a genetic algorithm in a warehouse cross-
docking operation, which was made famous by Wal-Mart. This is a highly
uncertain operation as the success of this operation relies on the ability of con-
solidating different less-than-truck (LTL) orders in a just-in-time manner. Arrival
of these orders cannot always be accurate due to unforeseen delays. In other words,
poor scheduling of these deliveries would result in higher operating costs. Yeung
and Lee define the GA operations (mutation, crossover and so on) for the cross-
docking problem and illustrate the proposed method through a numerical study.
Soft time window penalty, multiple products and multiple cross-dock doors are all
considered in the model. Results indicate that the GA-based approach can solve
the problem in a reasonable computational time.
MAS is a collection of autonomous agents working in a system independently
towards the system goal. Operations of an MAS is analogous to business dynamics
and hence MAS is a tool to model supply chain systems. Tounsi, Boissiere, Habchi
and Cung, in Chap. 2, propose an agent-based metamodel to address an unan-
swered issuehow to reuse knowledge generated particularly in MASs. Their
study clearly presents the definitions and interactions of agents, and the agentifi-
cation process for real-life applications. Their approach essentially breaks down
v
vi Preface
use of RFID technologies in two different contexts: the fresh meat supply chain
and the healthcare sector.
Butcher and Grants Chap. 5 considers the importance of food quality and
safety. They explore how to use radio frequency identification (RFID) technology
combined with distributed decision making (DDM) to add value in food retailing.
Butcher and Grant interview actors within a fresh meat supply chain to map the
current state process and to identify areas that could be improved. The current state
was then used to identify where RFID and DDM could be used in a future state of
the supply chain. They show that by employing these technologies, many benefits
can potentially be achieved, including food traceability, reduced human inter-
vention, improved process effectiveness, improved collaboration between retailer
and suppliers, as well as other cost saving and effectiveness outcomes. They argue
that it is the synergy of RFID and DDM that can achieve the most benefits in the
supply chain, rather than implementing them separately.
Chong and Chan use and extend the Technology Acceptance Model (TAM) in
their Chap. 6. They use this to help to understand the factors that affect the
adoption of RFID in the healthcare sector. There are many examples of the benefits
of the RFID technologies, such as to better track medicines and their expiry dates,
to monitor when equipment was last maintained and sterilized and to track patient
information in real time. Nonetheless, the healthcare industry has remained rela-
tively reluctant to adopt it. Chong and Chans study of 183 healthcare organisa-
tions finds that perceived usefulness and perceived ease of use are important in the
intention to adopt RFID technologies. This means that the potential users need to
see the technology as relevant to their job, will perform tasks well and will give
tangible results. In addition, the user needs confidence that they can use RFID and
should not be fearful or anxious of the technology. However, surprisingly, security
and privacy concerns, such as unauthorised access to the data, are not a major
concern and will not hinder the adoption of RFID.
Disruptions and disturbances within the supply chain can have consequences
that affect business continuity and competitiveness. Two chapters within this book
consider how more transparent approaches to understanding the risks and costs of
disruptions and disturbances can be anticipated to enable decision makers to
identify suitable alternatives and better manage the uncertainties inherent in
logistics systems and supply chains.
Breuer, Siestrup and Haasis Chap. 7 considers how organisations can rapidly
recover and maintain business continuity when faced with a disruption to their
supply chains. They use the case study examples of two German freight villages,
which are sensitive logistics nodes, to illustrate the concepts. In these systems, a
damaging event can affect multiple supply chains. Although increased integration
brings great benefits, it also carries more risk. In this chapter, a model is presented
which enables risks to be assessed and integrated. This approach enables action
plans to be developed to respond to attacks or disasters and also for alternative
decision paths to be visualised for the identified risks.
Rogers, Pawar and Braziotiss Chap. 8 considers how to assess the cost of
supply chain disturbances, that is risks and uncertainties, to enable firms to make
viii Preface
Artificial Intelligence
Analytical
Approaches
Approaches
(Chapters 1 & 2)
(Chapters 3 & 4)
Technology
Service Operations Supply Chain Adoption
(Chapters 11 & 12) Integration (Chapters 5 & 6)
Lightfoot, Baines and Smarts Chap. 11 explores how, in the face of increased
competition, some manufacturers have started to consider integrated products and
services as a key part of a value added offering to their customers. This serviti-
zation of manufacturing requires an investment in technologies that can monitor
and manage products in use by customers. It also requires a shift in focus from the
more traditional measures of cost, quality and delivery to performance, avail-
ability, and reliability. In addition, the integration of technology with supporting
management processes is needed to ensure the cost effective delivery of product-
centric services. The concepts in this chapter are supported with case study
examples from the aerospace, railway and road transport sectors.
Stumm and Kidds Chap. 12 describes a large research project which aims to
study the issues around the delivery of goods, fuel consumption and pollution in
Paris, France. The chapter describes a project which has been conceived to opti-
mise Parisian urban logistics and minimise operational pollution. Some of the
problems include that vehicle trips are made with empty cargo areas, warehouses
have unused storage capacity, rail and waterways are underutilized at night and
intermodal transport is weak. The project aims to develop a software platform to
enable the pooling of data, to optimise collaboration in real time across the
logistics system and to simulate scenarios and plan alternatives. This platform will
help the different actors in the system to pool resources and consolidate flows,
achieving better optimisation of resources, better service guarantees and reduced
costs and pollution.
The chapter topics are shown in Fig. 1.
We would like to take this opportunity to thank Professor Rajkumar Roy, Editor
of the Decision Engineering Series, for agreeing to publish this edited book. The
publishing team from Springer Verlag have provided excellent guidance
throughout the process, so thank you to Anthony Doyle and Grace Quinn. This
book would not have been published without their support. In addition, we are
x Preface
grateful to all of the reviewers for their valuable time and effort throughout the
review process. Their timely feedback further improved the quality of the chapters
published in this book. Finally, we would like to thank all of the authors for their
excellent chapter contributions.
We hope you enjoy reading this book as much as we have enjoyed editing it!
xi
xii Contents
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 253
Chapter 1
Adoption of Genetic Algorithm
for Cross-Docking Scheduling
with Time Window
Keywords Cross-docking
Genetic algorithm Warehousing logistics
Product delivery Inventory costs
L. Yeung
Division of Systems and Engineering Management,
School of Mechanical and Aerospace Engineering,
Nanyang Technological University, Singapore, Singapore
CKM Lee (&)
Department of Industrial and Systems Engineering,
The Hong Kong Polytechnic University, Hung Hom, Kowloong, Hong Kong
e-mail: carman.lee@gmail.com
1.1 Introduction
The outline of this chapter is shown as followings. Section 1.1 presents the
definition of cross-docking and discusses the main motivations and key concepts
behind this alternative option for tradition logistic systems. The introduction
chapter provides the background information of cross-docking and states the
research objectives subsequently. Literature review studies and distinguishes dif-
ferent types of cross-docking models developed by researchers and how the
optimisation methodologies applied in cross-docking problems. Section 1.3 serves
to illustrate the proposed cross-docking problem that the author will investigate on.
Section 1.4 discusses on the methodology adopted to solve the model and the steps
involved. Section 1.5 describes how experiments are designed to study the influ-
ence of the important parameters of the model based on the obtained results. The
conclusion summarises the highlights of this study and reiterates the important
results drawn from the findings.
there are only a few researches about adopting GA for cross-docking scheduling
with time window. Thus, the authors will investigate how computational intelli-
gence techniques can be adopted for cross-docking to enhance the accuracy and
efficiency of material handling. The proposed model will be designed based on GA
optimisation methodology which produces decision with reasonably low cost to
assign deliveries to specific cross-dock and subsequently from these cross-docks to
the respective customers at the downstream of the supply chain. The authors
designed the networking model with real time consideration and soft time
constraints to simulate the realism of a cross-docking network. In order to model
the nature of cross-docks, deliveries and customer demands have different types of
products with various quantities.
The author designed the network model to decide the execution of deliveries from
suppliers to all the customers through the implementation of allocated cross-docks.
The model consists of realistic parameters such as time horizon planning, cross-
docks with multiple dock doors, multiple product types, transportation cost,
inventory holding cost and soft time window penalties.
In this cross-docking network model, there are m number of possible deliveries,
n number of demanded pickups and h number of available cross-docks. These
anticipated deliveries and demanded pickups have to be executed within respective
time window period. Otherwise, penalty cost will incur. Also, all activities must be
completed within the defined time horizon, that is, between Tmin and Tmax. Each
delivery and pickup consists of multiple product types of different quantities. Each
cross-dock k, with multiple inbounds/outbounds, has a maximum capacity, Qk, for
inventory holding at anytime and the inventory holding cost is measured based on
per unit product and per unit time basis. Considering the variations in the inventory
handling efficiencies, the holding cost for each cross-dock varies as well. The
variation in distance between suppliers, cross-docks and customers are translated
into respective transportation costs. The main objective for this model is to minimise
total cost including the transportation costs, inventory costs and time penalty costs.
The model is based on the following assumptions:
1. Information such as the quantities, types of products and the relevant costs of
the deliveries and pickups are known beforehand. As long as all pickups are
fulfilled, deliveries may or may not be all materialized.
2. The time window in the model for delivery refers to the time interval allowed
for the delivery to reach the allocated cross-dock. Thus, the transportation time
needed for any delivery to its intended cross-dock has been considered
beforehand. Likewise, the time window specified for pickup is the allowable
time for any pickup to leave the cross-dock.
3. Deliveries and pickups have soft time window constraints which means that penalty
will be incurred if deliveries and pickups are executed out of the time window.
6 L. Yeung and CKM Lee
4. Each cross-dock has multiple inbounds and outbounds, and has a maximum
inventory capacity. Each cross-dock may or may not be assigned with deliv-
eries or pickups depending on the cost optimisation.
5. The entire time horizon may not represent the actual real time units such as
minutes and hours. The modeled time unit and time horizon are solely specified
for ease of computations.
Figure 1.1 represents a possible plan for 2 deliveries and 3 pickups via 3 possible
cross-docks. On the 2 time horizons (reach/leave cross-dock k), each coloured strip
indicates the soft time window periods where penalties are not incurred. For each
of the deliveries and pickups, there are 2 corresponding arrows to indicate the
completion of network allocation. For example, the first dashed arrow for Delivery
1 points to time 2 and the second dashed arrow points from the time 2 to cross-
dock 3. This means that Delivery 1 has been planned to reach cross-dock 3 at time
2. Likewise, pickup 1 (dashed arrows) is assigned to cross-dock 3 which will leave
at time 6.
1 Adoption of Genetic Algorithm for Cross-Docking Scheduling 7
The above illustration was simplified for the purpose of understanding the
overview and below is the detail description of the cross-docking network model:
Input
D Set of m deliveries, denoted by i
P Set of n pickups, denoted by j
H Set of h cross-docks, denoted by k
G Set of d products, denoted by r
Tmin Minimum time defining the time horizon
Tmax Maximum time defining the time horizon
Parameters
A i,r i 9 r matrix to represent the amount of product r from delivery i
0 1
a11 . . . a1r
B .. .. . C
@ . . .. A
ai1 air i2D;r2G
0 1
f11 f1r
B .. .. .. C
@ . . . A
fk1 fkr k2H;r2G
E1;i 1 9 i matrix to represent the soft time penalty cost (per unit product, per
unit time) of delivery i
1 ii2D
E1;j 1 9 j matrix to represent the soft time penalty cost (per unit product, per
unit time) of pickup j
1 jj2P
D
Tmin the earliest of delivery time window
D
Tmax the latest of delivery time window
P
Tmin the earliest of pickup time window
P
Tmax the latest of pickup time window
Decision Variables
1 if delivery i is bound for crossdock k at time t;
xi;k;t
0 otherwise;
1 if pickup j is bound from crossdock k at time t
yj;k;t
0 otherwise;
Objective Functions
Minimize CT CI CE where,
1 Adoption of Genetic Algorithm for Cross-Docking Scheduling 9
m X
X h TX
max n X
X h TX
max
CT xi;k;t Ti;k yj;k;t Tj;k
i1 k1 Tmin j1 k1 Tmin
P
h P
d P
Tmax
CI Fk;r zr;k;t Ti;k
k1 r1 Tmin
P
m P P
h Tmax D D
CE xi;k;t E1;i Tmin t for t \ Tmin
i1 k1 Tmin
P
m P P
h Tmax D
D
xi;k;t E1;i t Tmax for t [ Tmax
i1 k1 Tmin
P
n P P
h Tmax
p p
yj;k;t E1;j Tmin t for t \ Tmin
j1 k1 Tmin
P
n P P
h Tmax p
p
yj;k;t E1;j t Tmax for t [ Tmax
j1 k1 Tmin
Subject to:
P
h
(1) xi;k;t 1 8i
k1
Ph
(2) yj;k;t 1 8j
k1
(3) zr;k;t 0 8r; k and Tmin t Tmax
Pd
(4) zr;k;t Qk 8k and Tmin t Tmax
r1
(5) zr;k;Tmin1 0 8r; k
m
P P
n
(6) zr;k;t zr;k;t1 xi;k;t Ai;r yj;k;t Aj;r
i1 j1
8r; k and Tmin t Tmax
The above states all the important parameters required to define this cross-
docking network model. Most of the parameters are rather straightforward other
than the computation for the time penalty cost. There are generally 2 possible cases
to incur this penalty. These 2 cases include execution time for either delivery or
pickup being earlier than the intended time window or being later than that. The
penalty cost is computed by the amount of time violation out of the time window
interval. The objective of the function is to minimize the transportation cost,
inventory cost and time penalty.
Constraint (1) states that all deliveries can only be executed at most once or less
but as long as all demands are fulfilled which is stated in constraint (2). Constraint
(3) is the non-negativity constraint which states that the amount of products in any
cross-dock at any time can never be less than zero. Constraint (4) defines the
maximum capacities for all respective cross-docks. Constraint (5) states the zero
10 L. Yeung and CKM Lee
No
Yes
Store as initial
population
Cross-over Mutation
Constraint fulfillment?
Yes
Rank-based selection No
Stopping criteria?
Yes
Final solution
initial quantity for each cross-dock and constraint (6) ensures that the conservation
of product flow into and out of any cross-dock is upheld.
The binary decision variable xi,k,t corresponds to whether a particular delivery
i is bounded to cross-dock k at time t. Similarly, yj,k,t indicates whether pickup j is
assigned from cross-dock k at time t. And the integer decision variable zr,k,t reflects
the quantity of product r at cross-dock k during time t.
MatLab was used and empty chromosomes with elements filled with zeros are
created. Note that the representation does not exactly resemble a string of chro-
mosome because of the 3D nature of the decision variables. The first set of 6 X
matrixes represents the delivery decision variables where each row refers to the
delivery number and each column refers to the cross-dock number, at 6 different
timing. Similarly, the second set of 6 Y matrixes represents the pickup decision
variables where row refers to the pickup number and each column refers to the
cross-dock number, at 6 different timing shown in Fig. 1.4.
The initial population is generated by randomly assigning binary decisions, 1s
to empty solution as illustrated below. In order to reduce the randomness, this
assignment process is done with respect to the delivery and pickup constraints at
the same time. For example, for the delivery decision variables shown in Fig. 1.3,
delivery 1 (highlighted in yellow) can only happen once or less in that time
interval of 6. Thus, the assignment of 1 can only be made to 1 of 12 highlighted
elements or none, since it is not compulsory to fulfill all deliveries.
12 L. Yeung and CKM Lee
Pickup results
P1 Picked up at time 5
P2 Picked up at time 5
P3 Picked up at time 6
P4 Picked up at time 5
P5 Picked up at time 6
P6 Picked up at time 5
Cross-dock results
H1 Held 5 units at time 5 Held 20 units at time 5 and time 6 Held 5 units at time 5
H2 Not utilized Not utilized Not utilized
This enhanced method creates feasible solutions with higher speed which
eliminates obvious and extreme infeasible solutions. This process is also applied to
the pickup decision variables. At the end of this stage, solutions are feasible to
fulfill the first 2 constraints as mentioned in the problem formulation section.
Thirdly, these solutions will be subjected to the rest of the constraints as stated in
the problem formation section. If any of these constraints are violated, the solution
will be deemed as infeasible and another solution will be generated for the test
again. A complete feasible solution is found if none of the constraints are violated.
Assuming the case of having 3 different product types, 1 complete sample solution
together with the inventory decision variables, is illustrated in Fig. 1.4.
The interpretation of the initial population is shown in Table 1.1. There are 3
deliveries and 6 pickups at different time period. Three types of products are
dispatched and consolidated at 2 cross-dock centres.
The first step of GA begins with the selection of mating parents from the initial
population to reproduce new solutions. This model uses the Ranking-based
Selection to conduct the selection which selects the parents based on the total cost
or ranking of the fitness value. MatLab will compute the total cost of each chro-
mosome and place them in a ranking system from the lowest total cost to
1 Adoption of Genetic Algorithm for Cross-Docking Scheduling 13
the highest total cost. The top ranked E% chromosomes in the ranking system,
also known as elites, will be selected to proceed to the next mating procedures.
Those solutions that are not deemed as elites will not be involved in the subsequent
mating procedures. Since the elites are supposed to have better genes, this
ranking system ensures that the offspring reproduced are of higher quality.
1.4.3 Crossover
X(1)= X(1)=
0000 0000
1000 0000
0010 0010
0000 0000
X(2)= X(2)=
0001 0001
0000 0100
0000 0000
0000 0000
X(3)= X(3)=
0000 0000
0000 0000
0000 0000
1000 1000
selects genes from elite parents, offspring may be infeasible when subjected to all
the constraints. Therefore, these offspring will be eliminated from the pool and
only the feasible solution will be remained.
1.4.4 Mutation
to be mutated. In the third step, the mutation operator will select a new delivery
time from the 2 possible delivery time 2 (Denoted in dashed box). In the last step, a
new cross-dock will be assigned to the delivery to complete the mutation opera-
tion. In this case, the solution is mutated in the xi,k,t decision variables, where the
delivery time of delivery 2 is mutated from time 1 to time 2 and the cross-dock is
mutated from cross-dock 1 to cross-dock 2 (Second column with shaded color).
The experiment for this section aims to investigate the quality of the GA solution
and computational behavior for different available time horizon. The initial pop-
ulation is set as 50 and mutation probability is 0.2. The medium-sized case is
adopted for the experiment and the essential information such as number of
deliveries, number of pickups, available time horizon, number of cross-dock and
number of product types are shown in Table 1.3.
The three cases of available time horizon are 6, 8 and 12 respectively. And the
rest of the parameters such as number of deliveries, pickups, cross-docks and
product types are kept constant throughout. Table 1.4 shows the delivery and pick
up time window.
The summarised results for the experiment are shown in Table 1.5. It is found
that as T increases, the computational time and the total cost increase at decreasing
gradient. As the time horizon increases, the possibility of violation of time win-
dows increases. Thus higher time penalty cost is likely to incur and it can be found
in total cost.
The increase in computational time taken as time horizon increases can be
explained by the introduction of more variables in the problem. This is similar to
the explanation for the sample experiments. For example, the number of variables
in the case when T = 6, is equals to :(D 9 H 9 T) ? (P 9 H 9 T) ?
(H 9 G 9 T) = (6 9 3 9 6) ? (12 9 3 9 6) ? (3 9 3 9 6) = 378 variables
The numbers of variables for the 2 other cases are tabulated in Fig. 1.7.
In fact, the number of variables is directly proportional to the time horizontal
since all other parameters are constant, the term T can be factorized out:
No: of Variables D H T P H T H G T
T D H P H H G
The second experiment aims to investigate the computational behavior for dif-
ferent number of product types. The medium-sized case is also adopted for the
experiment and the essential information is shown as Table 1.6. The initial pop-
ulation is set as 50 and mutation probability is 0.2.
1 Adoption of Genetic Algorithm for Cross-Docking Scheduling 17
1 6 12 6 3 3
2 6 12 8 3 3
3 6 12 12 3 3
1 1, 3 2, 8 2, 8
2 1, 4 3, 6 3, 6
3 2, 5 2, 5 2, 5
4 1, 4 2, 4 2, 4
5 2, 3 2, 5 2, 9
6 2, 4 2, 6 2, 6
1 2, 3 3, 5 3, 5
2 2, 4 4, 8 4, 8
3 3, 5 3, 5 3, 5
4 2, 5 3, 5 3, 5
5 3, 5 4, 8 4, 10
6 2, 5 5, 7 9, 12
7 2, 5 4, 8 8, 10
8 3, 5 3, 6 3, 6
9 2, 5 6, 8 6, 12
10 2, 4 2, 5 2, 5
11 4, 5 4, 6 4, 11
12 2, 3 2, 8 2, 8
In this experiment, 3 cases of product types for any delivery or pickup are 3, 5
and 10.
The respective quantities of each product for the deliveries and pickups in the 3
different cases are summarised in the Tables 1.7 and 1.8.
The results obtained from GA are summarised in Table 1.9. It can be
realized that as the number of product types increases, the total cost increases
18 L. Yeung and CKM Lee
1 6 12 6 3 3
2 6 12 6 3 5
3 6 12 6 3 10
No of variables D H T P H T H G T
D P H H T
Since the other parameters are constant, G can be factorized out from the third
term and the linear relationship is illustrated in Fig. 1.8.
As product type (G) increases, the number of variables increases linearly. Thus,
the increase in the computational time is approximately linear as the time horizon
which is consistent with the results in the time elapsed results. All in all, when
product type G is large, one will expect GA to spend more time in searching for
feasible solutions since the number of variable is large.
1 Adoption of Genetic Algorithm for Cross-Docking Scheduling 19
to the linear relationship between the number of product types and the number of
variables involved.
To sum up, the key challenge in this project is to speed up the GA iterations for
large problem size. As mentioned throughout this book chapter, GA randomly
creates solutions and test for their feasibilities against the 6 constraints set in the
model. In fact, the amount of randomness can be greatly reduced by ensuring the
creation of solutions that obey the first 3 constraints using filters. This means that
GA has the ability to create solutions that at least, obey the deliveries, pickup and
non-negativity constraints. These solutions will then be tested against the rest of 3
other constraints for feasibility, thus reducing the randomness of the GA operation
by about half. However, the model still has room for improvement especially for
large scale problem as huge numbers of variables are involved.
1.6 Conclusion
logistics cost and the deterministic model is subjected to multiple dock doors, soft
time window and multiple products. Genetic Algorithm was adopted to search for
the optimal decision within reasonable computational time for the cross-docking
network problem. The times taken for, medium scale problem (i.e. 6 possible
deliveries, 12 demanded pickups, 3 available cross-docks and the 3 types of
products) are approximately 68 min.
The experiments were conducted to investigate the behavior of the parameters
such as the numbers of deliveries, pickups, cross-docks, time horizon and type of
products. From the result, the complexity of model affects the computational time
while the complexity is due to the increase in the number of variables and the GA
stochastic iterations.
Since the model includes soft time window penalty, multiple products and
multiple cross-dock doors, the proposed method can tackle real life logistic
problem in the turbulent business environment. The future work includes
collecting real data from third party logistics providers to test the robustness of the
model. To show the computational efficiency of genetic algorithm, further work is
to modify the model and compare the performance of the proposed approach with
other methodologies such as simulated annealing or other meta-heuristic methods.
Furthermore, to obtain more feasible solutions, repair process can be conducted to
mend the infeasible solution that violates the constraints. For the constraints setup,
maximum acceptable due dates for product delivery, earliness and tardiness arrival
time of the inbound and outbound truck can be set so as to avoid congestion of
cross-docking.
References
Agustina, D., Lee, C. K. M., & Piplani, R. (2010). A review: mathematical models for cross
docking planning. International Journal of Engineering Business Management, 2, 4754.
Bartholdi, J. J., & Gue, K. R. (2000). Reducing labor costs in an LTL crossdocking terminal.
Operation Research, 48, 823832.
Bhaskaran, S. (1992). Identification of transshipment centre locations. European Journal of
Operational Research, 63, 141150.
Boloori, A. A., Zandieh, M., & Ghomi, S. M. T. F. (2011). Multi-objective genetic-
based algorithms for a cross-docking scheduling problem. Applied Software Computing, 11,
49544970.
Boysen, N. (2010). Truck scheduling at zero-inventory cross docking terminals. Computers and
Operations Research, 37, 3241.
Chen, P., Guo, Y., Lim, A., & Rodrigues, B. (2006). Multiple crossdocks with inventory and time
windows. Computers and Operations Research, 33, 4363.
Douglas, L. M. (2009). Genetic-based scheduling to solve the parcel hub scheduling problem.
Computers and Industrial Engineering, 56, 16071616.
Jayaraman, V., & Ross, A., (2003). A simulated annealing methodology to distribution network
design and management. European Journal of Operational Research, 144, 629645.
Larbi, R., Alpan, G., Baptiste, P., & Penz, B. (2007). Scheduling of transshipment operations in a
single strip and stack doors crossdock. In Proceedings of the 19th International Conference on
Production Research.
22 L. Yeung and CKM Lee
Larbi, R., Alpan, G., Baptiste, P., Penz, B. (2009). Scheduling transshipment operations in a
multiple inbound and outbound door crossdock. In Proceedings of the International
Conference on, Computers and Industrial Engineering (pp. 227232).
Lee, Y. H., Jung, J. W., & Lee, K. M. (2006). Vehicle routing scheduling for cross-docking in the
supply chain. Computers and Industrial Engineering, 51, 247256.
Ley, S.E., Elfayoumy, S., & Jacksonville, I. (2007). Cross dock scheduling using genetic
algorithms. In Proceedings of the IEEE International Symposium on Computational
Intelligence in Robotics and Automation (pp. 416420). Jacksonville: FI.
Li, Y., Lim, A., & Rodrigues, B. (2004). Crossdocking JIT scheduling with time windows.
Journal of Operation Research Society, 55, 13421351.
Li, Z., Sim, C.H., Low, M.Y.H., & Lim, Y.G. (2008). Optimal product allocation for
crossdocking and warehousing operations in FMCG supply chain. In Proceedings of the
IEEE International Conference on Service Operations and Logistics, and Informatics (pp.
29632968).
Li, Z., Low, M. Y. H., & Lim, R. Y. G. (2009). Optimal decision-making on product allocation
for cross-docking and warehousing operations. International Journal of Services Operations
and Informatics, 4, 352365.
Lim, A., Miao, Z., & Rodrigues, B. (2005). Transshipment through crossdocks with inventory
and time windows. Naval Research Logistics, 52, 724733.
Ma, H., Miao, Z., & Lim, A. (2011). Crossdocking distribution networks with setup cost and time
window constraint. Omega, 39, 6472.
Miao, Z., Fu, K., Fei, Q., & Wang, F. (2008). Metaheuristic algorithm for the transshipment
problem with fixed transportation schedules. Lecture Notes in Computer Science, 5027,
601610.
Musa, R., Arnaout, J. P., & Jung, H. (2010). Ant colony optimisation algorithm to solve for the
transportation problem of cross-docking network. Computers and Industrial Engineering, 59,
8592.
Napolitano, M. (2000). Making the move to cross docking: A practical guide to planning,
designing, and implementing a cross dock operation. Oak Brook: Warehousing Education &
Research Council.
Ross, A., & Jayaraman, V. (2008). An evaluation of new heuristics for the location of cross-docks
distribution centres in supply chain network design. Computers and Industrial Engineering,
55, 6479.
Sung, C. S., & Song, S. H. (2003). Integrated service network design for a cross-docking supply
chain network. The Journal of the Operational Research Society, 54, 12831295.
Tsui, L. Y., & Chang, C. H. (1992). An optimal solution to dock door assignment problem.
Computer and Industrial Engineering, 23, 283286.
Wang, H.Y., Lee, C.K.M., & Lee, E.W. (2009). An Investigation of logistic operation in a cross-
docking distribution system with RFID. In Proceedings of the IEEE The 16th International
Conference on Industrial Engineering and Engineering Management (pp. 969974).
Wen, M., Larsen, J., & Clausen, J. (2009). Vehicle routing with cross-docking. Journal of the
Operational Research Society, 60, 17081718.
Chapter 2
A Generic Knowledge Model for SME
Supply Chain Based on Multiagent
Paradigm
J. Tounsi (&)
University of Tunis, SOIE, 41, Rue de la Libert cit Bouchoucha,
2000 Le Bardo-Tunis, Tunisia
e-mail: tounsi.jihene@yahoo.fr
J. Boissire
LISTIC-PolytechAnnecy, University of Savoie, 12, Chemin de Bellevue,
80439 74940 Annecy-le-Vieux, France
e-mail: julien.boissiere@univ-savoie.fr
G. Habchi
SYMME-PolytechAnnecy, University of Savoie, 12, Chemin de Bellevue,
80439 74940 Annecy-le-Vieux, France
e-mail: georges.habchi@univ-savoie.fr
V.-D. Cung
Grenoble INP/UJF/CNRS, Laboratory G-SCOP, 46, Avenue Felix Viallet,
38031 Grenoble Cedex 01 Grenoble, France
e-mail: van-dat.cung@grenoble-inp.fr
Keywords Knowledge model Multi-agent system Decentralised supply chain
Small and medium enterprises (SMEs)
2.1 Introduction
The supply chain concept was born in the 90s when management techniques in
the business world were evolving from independent to collaborative logistics. It is
well known that the supply chain is a complex macro system. This complexity is
firstly due to the variety of the involved organizations and the diversity of rela-
tionships between them, and secondly it results from the decision-making mech-
anisms between these companies. Thus, the success and subsistence of a company
in the economic market rely not only on its ability to integrate managerial pro-
cesses but also on coordinating all the related actors (Drucker 1998; Lambert and
Cooper 2000). Our work in this chapter is mainly focused on Small and Medium
Enterprises (SME). These companies evolve in an unstable and complex network.
In order to guarantee its role in a supply chain, SME must be able to support the
inherent requirements of the supply chain (low lead times, high level consumer
satisfaction, etc.) and the external requirements due to the environment (unpre-
dictable mutation, competition, etc.). Consequently, SME have to collaborate
together in order to achieve their goals without losing their autonomy and identity
(Julien 1997; Villarreal et al. 2005).
According to some field investigations, three major features of the supply chain
which integrates SME clusters arise. Firstly, a supply chain is a complex system, in
particular in the SME context. This complexity is due to the number of autono-
mous actors and the number of SME networks which are linked and work together
to achieve given processes and goals. Secondly, the SME are not often located in
the same geographical area as it could be in a more classical supply chain. Finally,
they face a lack of visibility over the entire supply chain as a result of the two
previous characteristics. Indeed, sites only have local visibility but are coordinated
with other sites through the flow of products. Due to this decentralized organi-
zation and limited view over the overall supply chain, studying the structure and
the behaviour of the SME supply chain is a challenging task and even more so if
sustainable considerations have to be taken into account.
Besides, the study of and the experimentation with the overall supply chain
integration of SME clusters are difficult to implement on actual industrial systems
without heavy investments from all the actors of the supply chain. Thus, in order to
facilitate the analysis of the supply chain network, it is necessary to propose a
modelling solution which reflects the actual system and is able to simulate its
behaviour. In light of this perspective, this chapter proposes a knowledge model
based on the Architecture Model Driven Engineering [ArchMDE (Azaiez 2007)]
development process that aims to identify and model the domain concepts using
the multiagent system. Hence, the work described here is a combination of two
research areas. The first one (industrial engineering scope) proposes a modelling
2 A Generic Knowledge Model for SME Supply Chain 25
approach using different layers that represent different views of the system (i.e. the
system refers to a supply chain). The representation of the domain concepts within
the models allows one to capitalize on the know-how and then facilitates the re-use
of the supply chain concepts in different contexts. The second one (the computer
engineering scope) outlines the transition from the identification step of the domain
structure to the study step for the dynamics behaviour of the domain concepts.
This work aims to combine the domain concepts with the multiagent ones.
In this chapter, we highlight the research work through these two areas. Hence,
the contents of this chapter are organized around four main issues:
How can we build a reusable model?
Which methodology to adopt for domain conceptualization and what are the
concepts?
What are the steps to follow in order to agentify the domain metamodel?
How to move from the modelling stage to the implementation?
To answer these questions, Sect. 2.2 introduces the ArchMDE development
process and its contribution to this work. In the Sect. 2.3, we present the different
steps applied to generate the conceptual metamodel. Section 2.4 highlights the
agentification process and the dynamic behaviour integration of the different
agentified concepts based on multiagent theory. Finally, Sect. 2.5 describes the
transition from the modelling phase to the implementation one.
The heart of this research work is the modelling of SMEs supply chain by using
multi-agent systems in order to build a reusable, flexible and secure knowledge
model. To reach this aim, advances in the field of computer engineering, especially
those dealing with multi-agent paradigm appear to be a promising approach.
To this end, we have adopted a modelling approach called ArchMDE and
proposed a PhD thesis (Azaiez 2007). This approach is based on Model Driven
Engineering (MDE) (Kent 2002) which founds its developing process on pro-
ducing several interrelated models. In the ArchMDE approach, two types of
metamodels are identified: a domain metamodel that describes functional concepts
and properties related to a particular domain (e.g. a SME supply chain) and a
computer modelling metamodel (e.g. a multiagent system). The combination of
both metamodels will generate an agentified metamodel, that constitutes the
starting point of the conceptual models. From this last metamodel, different
functional models are described in order to introduce the functionalities of the
system (Fig. 2.1). Finally, the use of a platform metamodel is necessary to gen-
erate the program code.
This approach is of great interest to fill in the existing gap between the design
and the implementation phases. The following sections describe the step-by-step
approach through those main axes.
26 J. Tounsi et al.
According to the ArchMDE development process, the first modelling step involves
the definition of the domain conceptual model. This step leads to the identification
of the main concepts of SME supply chains. To achieve this objective, we follow a
methodology based on existing conceptual modelling visions in the literature
(Tounsi et al. 2008). In this methodology, the visions are organized into three
steps. Each step addresses concepts related to supply chains. These concepts and
their relationships will then be gathered within a domain metamodel that will be
expressed using Unified Modelling Language (UML). The following section
presents this methodology.
To identify the properties and concepts of the supply chain domain, an incremental
methodology combining three visions is proposed: product vision, structure vision
2 A Generic Knowledge Model for SME Supply Chain 27
Abstract model
- Defines the environment architecture
- Defines the organizations architecture
Structure Vision
Conceptual Model
and process vision. In each step, a vision is applied to build or to refine the
conceptual model. The result of each step (intermediate model) is the input of the
next one. Therefore, at the end of the three steps, a final architecture of the
conceptual model is generated (Fig. 2.2).
Step 1: Product Vision
This vision considers the supply chain dedicated to a particular product (or a
family of products) from the raw materials through to the final goods. It focuses on
the product flow to define the environment and organizations involved in its
management (Thierry 2003). In the methodology framework, the Product Vision
leads to the construction of a first abstract model of the supply chain involving the
environment and organizations:
The environment is characterized by the physical flows and the different steps of
the product transformation as well as the related disturbances.
The organizations are the entities carrying out one or several product transfor-
mation stages and the physical flow management. The organizations involved
can be a network of firms that collaborate to accomplish one or several trans-
formation stages.
Step 2: Structure Vision
This vision has been proposed by Cooper et al. (1997). It considers the archi-
tecture of the supply chain, made up of: actors (decision-making actors and syn-
chronization actors), network structure (roles in the network and the number of
actors for each role) and relationship characteristics between actors. Thus, on the
basis of the abstract model provided by the previous step, the Structure Vision
details the organizations involved and the physical environment:
28 J. Tounsi et al.
The environment is the part containing the physical flow. Therefore, the product
flow and the resources used to achieve its transformation have to be described.
The organization consists in identifying and prioritizing the actors in the net-
work according to their involvement in the different levels of decision-making
as well as the tasks they will be assigned. The information flow management
that depends on the decision-making level is also considered in this approach.
At this step, a more detailed intermediate model is built.
Step 3: Process Vision
This vision is based on the process classification according to the decision-
making levels (Stevens 1989; Chopra and Meindl 2001): strategic, tactical and
operational.
While applying Process Vision, the various categories of processes are iden-
tified and integrated into the previous intermediate model. This can be done
according to the decision level but it also depends on the relationships between the
actors. These relationships can be classified into two categories:
Synchronization: contains processes for exchanging information and physical
flows according to a process program developed and predefined by the decision-
making layers.
Management and control: contain processes that ensure suitable decision
implementation in the perspective of a continuous improvement of processes in
terms of added value.
This step leads to a refined conceptual model of the supply chain.
This section presents the concepts that constitute the domain model. By applying
the methodology described in the preceding section, several concepts, processes
and the architecture of the model were identified. Based on these concepts, a
metamodel of supply chain is proposed.
Step 1: Applying Product Vision
By applying the Product Vision, a first abstract model of the supply chain is
built. It is composed of (Fig. 2.3):
Environment: represent the part allocated to the product flow and management
through the internal resources as well as the external elements able to influence
supply chain activities.
Sub Supply Chain (SSC): represents a group of SME which collaborate to
achieve an internal aim and/or the overall objective of the supply chain. The
SSC is responsible for the management of the product flow in a certain stage of
its life cycle.
2 A Generic Knowledge Model for SME Supply Chain 29
Interaction
SSC(1) SSC(n)
Visibility Visibility
Perimeter Perimeter
of influence of influence
Environment
1 1..n
Environment SSC
Perimeter of influence: represents the visible part of the environment to the SSC
on which it can act by internal conferring (if the action does not disturb the
environment located outside its visibility) or by conferring with another SSC.
Shared perimeter of influence: represents the area of the flow transfer between
two SSCs. It is a shared zone where SSC coordinates their activities to allow the
flow transfer.
Figure 2.4 shows the domain metamodel which reflects this conceptual abstract
model using UML.
Step 2: Applying Structure Vision
By applying Structure Vision, the previous abstract model is refined. The internal
architecture of the SSC and the visible part of the environment (the perimeter of
influence) are described. As showed in Fig. 2.5, the SSC model and its environment
are based on three layers representing the different decision-making levels.
Each layer involves particular concepts and plays a specific role in the SSC:
The Monitoring System is the intelligent layer of the SSC. It controls and
monitors the two other layers through the information provided by the Execution
System. Monitoring Actors (MA) modelling the intelligent actors of SSC are the
main elements of this layer. They establish metrics to evaluate the performance
of the group and consequently act on the other two layers. Hence, MA are the
components responsible of controlling and decision-making into a SSC and of
the coordination of the activities for the overall supply chain.
30 J. Tounsi et al.
SSC
Monitoring System
Decision Measurement
Execution System
Action Observation
Physical System
The Execution System is the reactive layer of the SSC. It has two main roles: (1)
it ensures the synchronization of the physical flow according to the information
gathered from the Physical System, (2) it observes and corrects the Physical
System if a disturbance occurs. In abnormal situations, the Execution System
refers to the Monitoring System for coordination and decision-making. Execu-
tive Actors (EA) are the main entities of this layer. An EA mainly models the
reactive actor in the Execution System. However, occasionally MA can appear
in this layer with reactive behaviour.
The Physical System is the visible part of the SSC environment. It corresponds
to the SSCs perimeter of influence. This layer is composed of non-decisional
elements controlled by the other two layers of the SSC. Two main concepts are
identified: the Moving Entity (ME) modelling the product flow and the Resource
modelling production means.
Figure 2.6 shows the first conceptual abstract model refined in a domain
metamodel. On one hand, we have integrated the identified concepts of each layer.
On the other hand, an abstract class Actor is added to the metamodel for
implementation purposes. Indeed, the Actor class defines the structural char-
acteristics and behaviour of a decisional entity. Thus, both the EA and the MA
inherit from this class. However, the EA class defines the specific characteristics
of an executive actor and likewise for the MA class.
Step 3: Applying Process Vision
The object of the last step is to identify and integrate the different kinds of
processes into the model. Table 2.1 gives a classification of the processes iden-
tified according to their role in decision-making. In the Physical System, the
Physical Processes (PhP) have been identified. A PhP describes the sequence of the
processing stages of a product. It is a concept to be integrated within a domain
metamodel in order to define the tasks that can be handled by the Execution
System.
2 A Generic Knowledge Model for SME Supply Chain 31
The processes identified in both Monitoring and Execution Systems are man-
agement processes. Hence, they represent the dynamic behaviour of the SSC. This
behaviour is induced by control and monitoring decisions that come from either
the SSC or the overall supply chain. This behaviour is basically a communication
mechanism (coordination, collaboration or cooperation).
In order to model management processes and communication mechanisms,
more informational elements are needed for EA and MA to ensure their role in the
domain model. Thus, decisional actors of the SSC (EA and MA) need three
conceptual elements that consolidate their internal architecture:
Indicator: is used by actors for two different tasks. Indeed, the EA control and
detect Physical System deviation by comparing the value of an indicator with its
fixed objective. As for MA, they evaluate the internal performance of the SSC
but also in the overall supply chain.
Action: Actors apply actions when facing indicator deviation.
Organizational Knowledge: is an actors database that stores information about
his acquaintances. For example, if an actor A is an acquaintance of an actor
B this means that B owns information about the identity, the behaviour, the
capabilities and the resources of the actor A. Reciprocally, the actor A
owns the same information about the actor B. According to this, each actor
(EA and MA) owns knowledge about resources of all actors in the same SSC.
However, the MA involved in the overall supply chain have additional internal
acquaintances, each MA owns limited knowledge about the other MA of the
overall supply chain. Note that this knowledge requires continuous updating.
In the same way, the intelligent behaviour of the MA requires the definition of
other conceptual components:
Objective: models the strategic goal of the SSC. According to this aim, the SSC
coordinates its activities with other SSCs in the overall supply chain.
Constraint: is a knowledge that an actor must consider to reach the goal of the
overall supply chain or the SSCs one.
Through the Process Vision, the previous metamodel and its concepts
are refined by integrating identified concepts. Figure 2.7 presents a UML
32 J. Tounsi et al.
SC
1 1
1 1..n
Environment SSC
1
1 1
1 1
1..n 1 1 1..n
Physical Execution Monitoring Actor
System 1 System System
1
1 1 1
1..n
1..n 1..n
1..n
Resource ME EA play Role MA
1 coordinate
SC
1 1
1..n
1 1 monitors
Environment SSC
1 1 11
1 1
Indicator Objective
Action
1..n
1..n
consult establish 1..n
1
1..n 1 1..n
Physical Execution Monitoring 1..n
1 1 Actor Realize
System System System
1 1
1
1
1..n
1..n 1..n 1..n
Resource ME EA play Role MA
1..n
1..n 1..n control 1..n 1..n 1..n
perform transforms
1..n 1..n refers to
PhP refers to
1..n
1..n
1..n
Organizational
Knowledges Constraint
knowledge
acts
representation of the final domain metamodel for the supply chain in the SME
context. It corresponds to the final conceptual model with its associated concepts
regardless of computer technologies.
In this section, the domain metamodel is merged with an agent metamodel using
the ArchMDE methodology (Azaiez 2007). Thus, this section is divided into two
parts: the first one outlines the properties of each multiagent concept according to
2 A Generic Knowledge Model for SME Supply Chain 33
The AIEO approach breaks the whole multiagent system down into four views:
Agent view, Interaction view Environment view and Organization view. Figure 2.8
shows multiagent concepts according to each view and the links between them.
The Agent view defines the agent metamodel composed of the following
concepts:
Agent identifies different kinds of agent according to the decision-making
capacity of the agent (reactive agent, cognitive agent and hybrid agent).
Cognitive agent defines an agent with cognitive abilities. The metamodel
highlights the main concepts modelling the BDI agent (Belief, Desire,
Intention = plan).
Reactive agent defines an agent with reactive abilities to respond to unpre-
dictable events.
Hybrid agent defines an agent with hybrid intelligence (cognitive and reactive
abilities).
Goal defines the aim that an agent should achieve.
Knowledge and Norm define all the knowledge and norm necessary for the
agent to achieve its goal.
Plan represents an action plan implemented by the agent. The plan is com-
posed of one or several elementary actions.
Reactive action is an action implemented by the reactive agent.
The Interaction view describes the dynamic relations between the agents. This
interaction is a structured exchange of messages between the agents through a
specific protocol or language. Thus the interaction metamodel highlights the fol-
lowing concepts:
Interaction protocol represents the interaction protocol adopted by the agents.
Communicative action represents an elementary action of communication
that is part of the interaction protocol.
Message is a set of information exchanged between the agents through the
interaction protocol. The agent interprets the message based on the com-
municative action.
The Environment view focuses on all the elements external to the agent
allowing it to reach its goal or activate its behaviour through events. The elements
belonging to the environment metamodel are as follows:
Active resource represents the resources that activate the behaviour of the
agent by generating events or triggers.
Passive resource defines the resources the agent needs to accomplish its task.
34 J. Tounsi et al.
Up to now, the static part of the domain metamodel has been created. In this section,
we define the dynamic behaviour of the concepts based on the multiagent tools and
theory. Indeed, this dynamic is described by the implementation of interaction
protocols according to the process vision and the communication mechanisms.
Firstly, the process vision allows us to define two scenarios: (1) the synchroni-
zation of the physical processes and (2) the monitoring and the control of processes.
Secondly, a communication mechanism is a framework formalizing interac-
tion between different actors in the network according to their managerial rela-
tionship characteristics (Tounsi et al. 2010). The study of the domain identifies
two kinds of communication framework. Indeed, in the overall supply chain, SSC
coordinate their activities in order to achieve the common objective of the overall
supply chain. Within the SSC, the actors collaborate to achieve a local goal.
This section describes the different protocols implemented in the agentified
domain metamodel taking into account the different scenarios of the process vision
and the communication mechanisms.
The SSC is responsible for the synchronization of the Physical System involved to
achieve its task. This activity consists in applying a communication protocol relative
to the nature of the interaction framework. In this section, the collaboration and
coordination processes are described in order to be implemented in the Execution
System and Monitoring System, and then to synchronize the Physical System.
1..n 1..n
2..n Organizational knowledge Knowledges Constraint
Action Communicative Action
1..n
is interpreted
acts
1..n
Message
Fig. 2.9 The agentified SME supply chain metamodel (Tounsi et al. 2009c)
37
38 J. Tounsi et al.
Figure 2.10 describes the Executive Actors (EA) behaviour in its role of
synchronizing the physical flow (ME). An EA receives a request and reacts
according to its type. Three types of request can be distinguished: (1) a ME
request, (2) a collaborative request from another EA or (3) a negative response to a
collaborative action initiated by the agent itself.
The following sequence highlights the EAs behaviour:
If the request is a negative response for the collaborative demand that the EA
initiated, the EA sends a request to the coordinator agent (MA coordinator) of
the monitoring system.
If the request is a synchronization request coming from the Moving Entity (ME)
or a collaborative request coming from another initiator agent, the EA checks
the availability of the resources concerned.
2 A Generic Knowledge Model for SME Supply Chain 39
If the resource is available, the EA carries out its task, updates the state of the
ME and informs the other agents within the executive system and the coordi-
nator agent at the end of the action.
If the resource is unavailable and the EA has been solicited by another executive
agent to achieve the task, it sends a failure request to the initiator.
If the resource is unavailable and the EA is in charge of the task, then it seeks
within its organizational knowledge an agent from the Execution System of the
SSC that might have the necessary resource.
If the agent finds another agent within its organizational knowledge that can handle
the task, it delegates the responsibility of the task. In this case, the collaboration
process of the agent concerned will be activated and follows the same sequence.
If the agent does not find another agent who has the resource necessary to handle
the task, it sends a request to the coordinator agent. This agent is a monitoring
agent (MA) that receives requests from the Execution System. The MA sends
the information to other monitoring agents in the SSC in order to find a solution.
In unusual situations,1 the Executive System refers to the Monitoring System. In this
case, the group of MA evaluates the situation according to the defined objective and
establishes an action plan. If the objective is not reached, the MA needs to consult
other SSCs to find a suitable solution. Thus, the agents adopt the protocol based on
Contract Net Protocol to provide the coordination of the objectives. The synchro-
nization protocol can be described according to the following steps:
In the Monitoring System, a monitoring actor is responsible for checking all the
requests received and sending them to other MA in the layer. Three kinds of
requests can be distinguished: (1) EA request, (2) reply to a help request or (3) a
help request from another SSC in the overall supply chain.
If a MA coordinator receives an EA request then it sends the information to
other MA. In this case, the group evaluates the situation according to the SSCs
objectives. Two cases may arise: (1) the problem has no impact on satisfying the
SSCs objectives or (2) the objective is deviated.
If there is no impact on the objective, the MA tries to find an internal solution
according to its desire, belief and constraints. If a solution can be found, the MA
coordinator sends the actions plan to the Executive System.
If the objective is deviated or an internal solution cannot be reached, the group
of MA sends a Help Request to other SSCs via the MA coordinator and waits for
the responses.
1
Unusual situation is occurred when Executive System cannot propose a solution for a happened
problem in the physical system.
40 J. Tounsi et al.
Help Request
Help Request
negative response
select an offer
offer selected
sends Help
When all the replies have been received, the MA coordinator ranks them according
to the reception date of the request. The list of responses will be sent throughout the
Monitoring System. According to their beliefs, desires and constraints, the group
of MA chooses the most suitable answer and diffuses the action plan to the
Execution System. In this case, the EA updates the state of the ME.
If the request is a help request from another SSC, the MA coordinator sends the
request to the Monitoring System. In this case, the Monitoring System evaluates
the demands according to internal criteria (objective, constraint, belief, and
desire). If the SSC can provide assistance, it makes an offer to the SSC initiator
or it sends a negative response.
The SSC initiator chooses the suitable offer and sends a confirmation to the
selected SSC and a cancellation response to other bids.
The following figure (Fig. 2.11) shows the sequence of messages between the
SSCs. This diagram represents the coordination process in the overall supply chain
in order to synchronize the physical flow in the case of a disruptive case (SSC
cannot reach the internal aim).
2 A Generic Knowledge Model for SME Supply Chain 41
The last phase of the ArchMDE development process deals with the transition
from the modelling phase to the implementation one. This last phase is mainly
divided into two steps. The first one concerns the refinement of the agentified
domain metamodel by integrating the information necessary for the implementa-
tion. The second step introduces the choice of the development platform. In the
42 J. Tounsi et al.
return (list_perception)
consult(id_perception)
return (alertmin, alertmax)
select(id_indicator, perturbation_cause)
[action exist] retrun (action)
Fig. 2.12 Sequence diagram of the monitoring and control protocol (Tounsi et al. 2011)
next section, we focus on the two steps of the implementation phase of the
ArchMDE development process.
The steps of integration protocols identify the abstract patterns that describe the
dynamic behaviour of the agentified domain concepts. The refinement consists of
the integration of the attributes and the methods that define the architectural and
the behavioural proprieties of each concept. This step provides a final class dia-
gram, the Implementation Metamodel. The encoding of this last result leads to a
dedicated simulation platform.
2 A Generic Knowledge Model for SME Supply Chain 43
2.6 Conclusion
while considering the organisational issues and the managerial relationships. The
resulting knowledge model requires three main features: reusability, simplicity and
genericity. Our main theoretical contributions are on this conceptual model and its
building methodology.
To reach this goal, we have adopted ArchMDE as a development process. The
first step of the development process focused on the identification of the major
concepts. These concepts are consistent with both domains: the inherently decen-
tralized supply chain and multi-agent systems. The high level of conceptualization
of these concepts tends to provide genericity and reusability to the proposed model.
In the second step, we have presented a mapping within the proposed concepts. This
mapping translates the supply chain concepts into MAS concepts. This aims at
simulating the dynamic behaviour of the supply chain based on multi-agent tools.
Major communication protocols between actors (or agents) are also drawn to
emphasize the importance of different kinds of exchange mechanisms. To keep the
model as generic as possible only a few generic protocols are presented here.
In terms of implementation, the concepts and the protocols presented in this
chapter have been developed as a generic Supply Chain Multi-Agent System using
JAVA. A prototype of a 2-echelon Supply Chain with one manufacturer and one
logistician including (1) respectively production and inventory planning processes,
and (2) two-actor negotiation and coordination processes has been tested with
success (Ogier et al. 2010).
Our aim is to be able to simulate a real multi-actor Supply Chain with dynamic
behaviour. Thus, more internal optimization processes and coordination protocols
are under study.
References
Azaiez, S. (2007). Approche Dirige par les modles pour le dveloppement de systmes
multi-agents. Annecy le vieux, France: Thse de lUniversit de Savoie Spcialit
Informatique. (11 Dec).
Chopra, S., & Meindl, P. (2001). Supply chain management: Strategy planning and operation.
NJ: Prentice-hall.
Cooper, M., Lambert, D. M., & Pagh, J. D. (1997). Supply chain management: More than a new
name for logistics. International Journal of Logistics Management, 18(2), 113.
Drucker, P. F. (1998). Managements new paradigms (pp. 152177). Forbes Magazine.
Gutknecht, O., Ferber, J., & Michel, F. (2000). MadKit: Une exprience darchitecture de
plate-forme multi-agents gnrique. 8me Journes Francophones Intelligence Artificielle
Distribue Systmes Multi-Agents JFIADSMA 2000. Saint-Jean Le Vtre Octobre.
Julien, P. A. (1997). Les PME bilan et perspectives (2nd ed.)., Economica, Paris: France.
Kent, S. (2002). Model-driven Engineering IFM 2002 (Vol. 2335)., LNCS, Springer-Verlag.
Lambert, D. M., & Cooper, M. C. (2000). Issues in supply chain management. Industrial
Marketing Management, 29(1), 6583.
Ogier, M., Cung, V.-D., Boissire, J., & Mangione, F. (2010). Supply chain performance in the
case of decentralized planning. In Proceedings of the 8th International Conference on Supply
Chain Management and Information Systems (SCMIS2010) (pp. 6370). Hong Kong, China.
Stevens, G. C. (1989). Integrating the supply chain. International Journal of Physical
Distribution and Materials Management, 19(8), 38.
46 J. Tounsi et al.
Thierry, C. (2003). Gestion des chanes logistiques: Modle et mise en uvre pour laide la
dcision moyen terme. Toulouse: University of Toulouse II. (Accreditation to supervise
research).
Tounsi, J., Boissire, J., & Habchi, G., (2008). A conceptual model for SME Mechatronics supply
chain. In Proceedings of the 6th International Industrial Simulation Conference (ISC08)
(pp. 273280). France: Lyon.
Tounsi, J., Boissire, J., & Habchi, G. (2009a). Multiagent decision making for SME supply chain
simulation. In Proceedings of 23rd European Conference on Modeling and Simulation
(ECMS) (pp. 203211). Madrid: Espagne.
Tounsi, J., Habchi, G., & Boissire, J. (2009b). A multiagent system for production
synchronization in SME mechatronic supply chain. In Proceedings of the 10th Middle
Eastern Simulation and Modeling Conference (MESM) (pp. 9197). Beirut: Liban.
Tounsi, J., Azaiez, S., Habchi, G., & Boissire, J. (2009c). A multiagent approach for modelling
SMEs mechatronic supply chains. In Proceedings of the 13th IFAC Symposium on
Information Control Problems in Manufacturing.
Tounsi, J., Habchi, G., & Boissire, J. (2010). A conceptual model for SMEs mechatronics supply
chain. International Journal of Computer Aided Engineering and Technology, 2(4), 371387.
Tounsi, J., Habchi, G., Boissire, J., & Azaiez, S. (2011). A multi-agent knowledge model for
SMEs mechatronic supply chains. Journal of Intelligent Manufacturing, doi: 10.1007/s10845-
011-0537-1.
Villarreal Lizarraga, C.L., Dupont, L., Gourg, D., Pingaud, H. (2005). Contributing to
management of shared projects in SMEs manufacturing clusters. In Proceedings of the
18th International Conference on Production Research (ICPR-18). Salerno: Italy.
Chapter 3
Integrated Supplier Selection, Pricing
and Inventory Decisions in a Multi-level
Supply Chain
Y. Huang
Faculty of Management and Administration,
Macau University of Science and Technology, Macau, P.R. China
G. Q. Huang (&)
Department of Industrial and Manufacturing Systems Engineering,
University of Hong Kong, Hong Kong, P.R. China
e-mail: gqhuang@hku.hk
3.1 Introduction
Supply chain management (SCM) is an area that receives great attention in the
business community. To implement SCM, the coordination and integration of the
activities within an organization and across the supply chain is necessary. Many
firms identify and qualify adequate suppliers to provide the materials and service
needed (Watts and Hahn 1993). Effectively selecting and evaluating these quali-
fied suppliers and managing their involvement in critical supply chain activities
enable manufacturers to achieve the four dimensions of customer satisfaction:
competitive pricing, product quality, product variety and delivery service (Morgan
1996; Poirier 1999; Tracey and Tan 2001). This paper coordinates supplier
selection, pricing and inventory decisions and proposes a cooperative game theory
approach to evaluate the suppliers for an integrated multi-level supply chain.
There is a vast amount of literature in the industrial field on vertical integrated
supply chains. The literature shows that it is in the best interests of the supply chain to
maximize overall system profit and integrate operations among various entities
through coordination. Williamson (1975) looks at the long-term relationship
between a seller and buyer and finds the advantages of integration, for example,
reducing transaction costs. The integration of supply chain mainly focuses on the
pricing model and distribution inventory model. Boyaci and Gallego (2002) analyze
the problem of integrating pricing and inventory replenishment policies in a supply
chain consisting of a wholesaler and one or more geographically dispersed retailers.
They show that optimally coordinated policy could be implemented cooperatively by
an inventory-consignment agreement. Joglekar et al. (2006) present a set of inte-
gration models for optimal pricing and order quantity decisions in a one manufac-
turer and one retailer supply chain. Khouja (2003) develops inventory coordination
mechanism for the supply chain composed of a single supplier, three manufacturers
and two retailers per manufacturer. Jaber and Goyal (2008) consider optimal order
quantity model in a multiple suppliers, a single vendor and multiple buyers supply
chain. Some researchers also consider supplier selection decisions in the vertical
integration model. For example, Haq and Kannan (2006) developed an integrated
model to select best suppliers and optimize inventory level for a multi-level supply
chain to minimize the total supply chain cost. However, none of them integrates the
above three aspects, i.e., pricing, inventory, and supplier selection decisions.
Papers employing cooperative game theory to study supply chain management
are becoming more popular (Nagarajan and Soic 2006). Gerchak and Gupta (1991)
model their inventory problem as a cooperative game and analyze the joint cost
allocation problem. Granot and Soic (2005) study a multi-stage model of a decen-
tralized distribution system with inventory sharing consisting of n retailers. In the
cooperative stage, they show that Shapley value encourages the retailers to share all
of their residuals. Rosenthal (2008) constructs a cooperative supply chain game
for calculating transfer prices for intermediate goods in vertically integrated supply
3 Integrated Supplier Selection, Pricing and Inventory Decisions 49
chains. Leng and Parlar (2009) model a three-level supply chain cooperative game in
characteristic-function form to allocate cost saving among supply chain members.
This paper considers a three-level supply chain composing of multiple suppliers, one
manufacturer and multiple retailers. The manufacturer obtains components from the
potential qualified suppliers and produces final products for retailers in different
markets. In this paper, we formulate a mixed-integer programming model to integrate
the optimal supplier selection, pricing and inventory decisions. We put forward a
cooperative game theoretic approach and use the Shapley value to evaluate the marginal
contributions of the supply chain members, especially the suppliers, for the above
supply chain in integrated process. Their use is tested through a numerical example. The
impacts of the market scale parameter, setup cost and component cost on the optimal
decisions and marginal contributions of all the chain members are also investigated.
The remainder of this paper is organized as follows. Section 3.2 gives the problem
description and some notations. We formulate the mathematical models and solution
procedure in Sect. 3.3. In Sect. 3.4, a numerical study and the influence of market
scale parameter, component cost and setup cost have been presented. Finally, this
paper concludes in Sect. 3.5 with some limitations and suggestions for further work.
In this paper, the authors consider a supply chain, which is adapted from the cases
in Graves and Willems (2005) and Zhang (2006) (Fig. 3.1). This application case
is concerned with a three-echelon supply chain involving multiple suppliers, one
single manufacturer and multiple retailers. The manufacturer, indicated by m ,
designs and customizes a set of products for retailers (rl ; l 1; 2; . . .; L) in
different independent market segments. Each retailer is served by one type of
product. In Fig. 3.1, the manufacturer focuses on two computer platform products,
involving Notebook A and Notebook B, and sells them to the retailers in two
market regions respectively, namely Europe (EU) and North America (NA).
The architecture for the product is finalized which consists of a series of different
functionality elements. The components offering different levels of the same function-
ality are grouped together in the substitutable component set (SCS), indexed
by i 1; 2; . . .; I: The components within the same SCS can be ranked in order of
decreasing functionality, and higher functionality components can substitute ones
with lower functionality completely, but not vice versa. The lowest functionality
components in all SCSs satisfying the products requirement have also been
predefined.
Let Ni be the number of components in SCS i. Lij is used to denote the component
which is the jth element in the ith SCS, where i 1; 2; . . .; I and j 1; 2; . . .; Ni :
In Fig. 3.1, the two products share the same architecture, with SCSs, processor,
LCD display, memory, hard drive, miscellaneous components and metal housing,
in sequence. Among them, processor and LCD display have several component
50 Y. Huang and G. Q. Huang
Intel Core
s1 2 Duo
Intel Pentium
Dual-Core
s2 LCD Display
SXGA+
Notebook A r1 in EU Market
s3 WXGA
XGA
Notebook B r2 in NA Market
s4
Memory
Hard Drive
s5
Miscellaneous
Components
Metal
s6 Housing
Fig. 3.1 Coordination of suppliers and components selection, pricing and replenishment
decisions (CSCSPR): an illustration
customers to maximize the total profit of the integrated supply chain; secondly, for
the optional suppliers, which one is more important to the supply chain system.
Prior to developing the mathematic model for solving the above problems, the
authors make the following underlying assumptions and simplifications:
1. The demand at retailer l is almost invariably a downward sloping and convex
function with respect to the retail price in real world. The linear demand
function in McGuire and Staelin (1983) and Jeuland and Shugan (1988) is
employed: Dprl Arl erl prl :
2. Vendor managed inventory (VMI) system (Simchi-Livi et al. 2000; Tyan and
Wee 2003) is employed between the suppliers and the manufacturer and between
the manufacturer and the retailers to replenish the components and the products.
This inventory system has been adopted by some industries for years (e.g.
Wal-Mart, Procter&Gamble (P&G), Dell, etc.). Under this assumption, the com-
ponents inventories for the manufacturer and the suppliers are at the side of their
upstream suppliers and their corresponding inventory holding costs are also borne
by their upstream suppliers (Mishra and Raghunathan 2004). The inventories of the
final products are on the manufacturers side and the retailers take zero inventory.
3. Single sourcing strategy (Tullous and Utrecht 1992) is adopted between sup-
pliers and manufacturer. Thus, the manufacturer purchases one type of com-
ponent from only one supplier.
4. Either all or none of the demand for a component is replaced by the lower order
component in the same SCS.
5. Shortages are not permitted, hence the annual production capacity is greater
than or equal to the total annual market demand (Esmaeili et al. 2009).
3.2.1 Notations
In this section, the authors will introduce some other relevant parameters and
decision variables used for the mathematical models.
Parameters:
Drl Retailer ls annual demand
Ar l A constant in the demand function of retailer l, which represents his market
scale
erl Coefficient of the products demand elasticity for retailer l
Pl Manufacturers annual production capacity for product l
cml Production cost per unit product l
52 Y. Huang and G. Q. Huang
Decision variables:
pr l Retail price charged to the customer by retailer l
T Decision variable, manufacturers setup time interval
sijk Binary decision variable to indicate whether component Lij has been used to
replace Lik
rsv Binary decision variable to indicate whether supplier v is used
ar l Binary decision variable to indicate whether retailer l is used
tsv Lij Binary decision variable to indicate whether component Lij is supplied by
supplier v
zLij Binary decision variable to indicate whether component Lij is used
Here, the objective function of the integrated supply chain model is concerned with
the maximization of the overall profit of the supply chain. The total cost of the supply
chain is composed of the production costs, fixed costs to contract with the suppliers
and using components, holding costs for the products, setup and ordering costs,
component costs and annual fixed costs for all the chain members. According to the
given VMI policy, only the manufacturer has to pay for inventory for final products.
The behavior of this inventory level for the product l is illustrated as Fig. 3.2. The
hmp Pl Dr TDrl
annual inventory cost for product ls is given by l 2Pl l (suggested by Zhang
(2007)). The setup cost Sm and ordering cost Om occurs at the beginning of each
production. Thus, the integrated supply chain model (A) is formulated as the fol-
lowing constrained mixed-integer programming problem:
3 Integrated Supplier Selection, Pricing and Inventory Decisions 53
Inventory level
inventory level for final
product l
Pl Drl Drl
T Time
X
L X
L X
V I X
X Ni
max Z prl Drl arl cml Drl arl wsv rsv FLij zLij
l1 l1 v1 i1 j1
X
L
hmp Pl Dr TDr Sm O m
l l l
l1
2Pl T
!
V X
X L X
I X
Ni X
Ni
cLij zLij uLij l uLik l sijk Drl tsv Lij
v1 l1 i1 j1 kj1
X
L X
V
Rr l Rm Rsv rsv 3:1
l1 v1
Subject to:
Drl Arl erl prl 3:2
0 Drl Pl 3:3
X
k1
sijk zLik 1;
j1
8uLik l 6 0; 8i 1; 2; . . .; I; j 1; 2; . . .; Ni 1; k j 1; . . .; Ni ; 3:4
X
V
tsv Lij zLij ; 8i 1; 2; . . .; I; j 1; 2; . . .; Ni 3:7
v1
X
tsv Lij Xsv rsv ; 8v 1; 2; . . .; V; 3:8
Lij 2Qsv
8i 1; 2; . . .; I; j 1; 2; . . .; Ni 1; k j 1; . . .; Ni 3:9
54 Y. Huang and G. Q. Huang
prl 0 T [ 0: 8l 1; 2; . . .; L 3:10
Constraint (3.3) gives the bounds of the annual demand, which cannot exceed
the annual production capacity Pl of the product. As indicated in the fourth point
of the assumption in Sect. 3.2, constraint (3.4) ensures that for the component
predefined for the products, it is either used or replaced by higher functionality
one, but not both. Constraint (3.5) makes sure that only procured components can
be used to replace other components. (3.4) and (3.5) together ensure that the
demands of all components are satisfied. Also, they met the one-way substitut-
ability constraint which ensures that a higher functionality component can replace
a lower functionality component but not vice versa. Constraint (3.6) sets the value
of tsv Lij 0 for all components Lij 62 Qsv for each supplier. Constraint (3.7) indi-
cates that a component is procured from exactly one supplier. Constraint (3.8) sets the
value of rsv to 1, if supplier v supplies a component, and ensures that the number of
different types of components supplied by supplier v is no greater than Xsv : The value
ranges of all the variables are set by constraints (3.9) and (3.10).
In this section, the authors use a cooperative game to evaluate the suppliers involved
in the above supply chain. To formulate a cooperative game, the authors must first
identify the set of all players, N, and a value function v, that associate with every
nonempty subset G (a coalition) of all the players. A real number v(G) represents the
worth of G that describes the total expected gain from the coalition G. And v satisfies
the following two conditions: (3.1) v[ 0; and (3.2) If R; T N and R \ T [;
then vR [ T vR vT :
Based on the definition, the authors define that the set of the game players is
composed of the suppliers, the manufacturer and the retailers, i.e., V L 1
supply chain members. The supply chain is an integrated process, wherein the
components are manufactured to final products, then delivered to customers
(Beamon 1999). Thus, to formulate a cooperative game for such supply chains in
an integrated process, the value function, here, is regarded to be,
ZG; if coalition G could transfer components to final products
pG ;
0; otherwise
3:11
where Z G is the integrated model (A) with the constraints that the selection
variables for those supply chain members out of coalition G to be zero. For example,
if supplier 1 is not within coalition G; Z G is the model (A) adding constraint
rs1 0: It can be seen that a coalition G is valid if and only if G could transfer
components to one product at least. Obviously, the value function defined above
satisfies conditions (3.1). For condition (3.2), 8R; T V L 1 and R \ T [;
3 Integrated Supplier Selection, Pricing and Inventory Decisions 55
where /i N; v is the Shapley value of player i; jSj is the players number in the
coalition S; n is all players numbers, and vS is the value function of the coalition S:
The marginal contribution of adding player i to coalition S is vS [ fig vS:
According to the definition, the Shapley value is obtained by averaging the
marginal contributions for all the coalitions of the game players (Shapley 1971).
In this paper, we use the Shapley value to evaluate the supply chain members,
especially the suppliers. Their importance to the integrated supply chain is
classified according to their marginal contributions.
The integrated supply chain model (A) has binary variables and L 1 non-integer
variables. Some researchers solve such problems through genetic algorithms and
conduct the comparison with the results from Lingo software (Peters and
Rajasekharan 1996; Tavakkoli and Rahimi 2006). In this paper, Lingo is employed
to solve the integrated models. To find out the Shapley values for each supply
chain member, the authors use the following procedure:
Step 1. Denote R fr1 ; r2 ; . . .; rL g to be the set of all the retailers; identify all
the subset of R: For each X; X R;, find out all the suppliers combination SX
satisfying constraints (3.4), (3.5), (3.6), (3.7), (3.8), (3.9) in Model (A) and all the
valid coalitions GX SXi ; m; X ; 8SXi 2 SX :
Step 2. Calculate the value function Z GXt ; 8GXt 2 GX ; 8X R; using Lingo.
Step 3. Use (3.12) to calculate the Shapley values for all the supply chain members.
3.3.1 Initialization
The explicit numerical parameters selected for the base case example reflect those
shown in Lu 1995, Woo et al. 2001; Joglekar et al. 2006. For example, the man-
ufacturers setup cost should be much larger than the ordering cost; the compo-
nents purchasing costs are much larger than the fixed costs of using them. For the
two retailers in the European market and the North American market
(i.e., r1 ; r2 ), we set: Ar1 Ar2 2 106 ; er1 1400; er2 1200: For the
manufacturer, (1) setup cost and ordering cost are: Sm 95; Om 5; (2) holding
costs for the two products are: hmp1 0:5; hmp2 1; (3) production capacity are:
P1 5 106 ; P2 3 106 ; (4) production costs are: cm1 10; cm2 5; (5) fixed
costs of using suppliers are set at: ws1 12; ws2 13; ws3 40; ws4 60; ws5 18;
ws1 31; and (6) fixed costs of using components are set at: FL11 3; FL12
4; FL21 6; FL22 7; FL23 8; FL31 9; FL41 12; FL51 2; FL61 10: The
suppliers purchase costs for the components are: cL11 100; cL12 50; cL21
200; cL22 80; cL23 50; cL31 20; cL41 30; cL51 200; cL61 29: The fixed
costs for all the suppliers, manufacturer and retailers are: Rs1 30; Rs2 15;
Rs3 35; Rs4 23; Rs5 20; Rs6 18; Rm 100; Rr1 50; Rr2 61: The rela-
tionship between the components and their suppliers are shown as Fig. 3.1.
We then calculate the Shapley values for all the chain members. In this paper, the
Shapley value is simply employed to study the marginal contribution of the
supply chain members instead of allocating the profit. For the convenience of
comparison, we consider the value of grand coalition to be 1, thus, the
values for the other coalitions are the ratio of the profit of this coalition to
the profit of the grand coalition. Using the solution procedure in Sect. 3.2.4, the
Shapley values can be obtained. For example, for supplier 1, the Shapley value is:
P jSj1!VL1jSj!
/s1 VL1! Z S Z Snfs1 g=Z;where C is the set of all the
S2C
valid coalitions; Z S is the profit of coalition S; Z is the profit of this grand
coalition.
We can see that supplier 4 is not selected for any component and supplier 6 is
selected for component L31 : Component L21 supplied by supplier 1 is not used for
any product. The retail price and demand for product 2 are higher than product 1
for retailer 1.
3 Integrated Supplier Selection, Pricing and Inventory Decisions 57
Supplier 5 has the largest Shapley value. This is because supplier 5 is the one
and only supplier for the indispensable component L51 for the final products.
Although supplier 1 only provides components for retailer 1, his Shapley value is
the second largest one of the suppliers, since the component L11 he suplies cannot
be substituted. Supplier 4 has the lowest Shapley value. Retailer 2 has a larger
Shapley value than retailer 1 for his better market environment.
Sensitive analysis
A higher component cost will result in a higher cost for production using this
component. Considering such effect, it would be worthwhile examining the
interplay between component cost and optimal supply chain decisions. As such,
we experiment on the model with varying costs of component L11 :
58 Y. Huang and G. Q. Huang
When the cost of component L11 rises from 100 to 200, the retail price for
product 1 increases from 924.2871 to 974.2871, by 5.41% and its demand
decreases from 7.05998 9 105 to 6.35998 9 105, by 9.92%. Product 2s price and
demand are almost the same as the base example. The manufacturer sets his
production to be less frequent. In the above case, the setup time interval increases
from 0.0152 to 0.0154, by 1.32%.
Figure 3.3b presents the changes of the Shapley values for the suppliers and the
retailers under different costs of component L11 : The Shapley values for supplier 1
and retailer 1 decrease as the cost of component L11 increases. For supplier 2,
supplier 3 and retailer 2, the results are reversed. That is to say when one products
component cost increases, the component supplier and the corresponding retail
market for this product will become less important to the supply chain system.
Increase of market scale has the opposite result.
3 Integrated Supplier Selection, Pricing and Inventory Decisions 59
(a) (a) Shapley value for suppliers (b) Shapley value for manufacturer and retailers
0. 25 0. 25
base example base example
Ar1:3500000 Ar1:3500000
0. 2 Ar1:5000000 0. 2 Ar1:5000000
Shapley value
Shapley value
0. 15 0. 15
0. 1 0. 1
0. 05 0. 05
0 0
s1 s2 s3 s4 s5 s6 m r1 r2
Supplier Manufacturer & retailers
(b) (a) Shapley value for suppliers (b) Shapley value for manufacturer and retailers
0. 25 0. 25
base example base example
cL11:200 cL11:200
0. 2 cL11:300 0. 2 cL11:300
Shapley value
Shapley value
0. 15 0. 15
0. 1 0. 1
0. 05 0. 05
0 0
s1 s2 s3 s4 s5 s6 m r1 r2
Supplier Manufacturer & retailers
Fig. 3.3 Shapley value a Shapley value for different market scale parameter. b Shapley value for
different component cost
The influence of production setup cost is very minor on the pricing and supplier
and component selection divisions except for the setup time interval for the
manufacturer. For example, when Sm increases from 100 to 150, the setup time is
lengthened to 0.024, by 57.89%. The change of Sm also, has limited influence on
the Shapley values of the supply chain members.
Through the numerical example, we can see that the proposed cooperative game
can be used to evaluate the supply chain members by calculating their Shapley
values. We believe that the cooperative game approach can provide evidence for the
manufacturer and the supply chain to manage their relationship with the suppliers.
Based on the results obtained from the simulation results presented above, in this
subsection, we wrap up our findings and some implications regarding the impacts
of the aforementioned parameters analyzed above.
60 Y. Huang and G. Q. Huang
Firstly, the increase of retailer 1s market scale will increase the optimal retail
price and demand for product 1 without impact on the other retail market. This is
because the entire supply chain could benefit from the higher price and larger
demand from the increase of retailer 1s market, while maintaining the profit level
in retailer 2s market.
Secondly, retailer 1s market scale increases, he as well as the suppliers of
components for product 1 gain an increased importance in the supply. The
marginal contributions of retailer 1 and the suppliers who only provide
components for product 1 become larger when retailer 1s market scale increases.
Thirdly, the increase of component cost leads to longer setup time interval for
the manufacturer. This is because component cost rise will increase the cost of the
final product, thus decreasing the market demand and inventory consumption rate.
So the manufacturer would like to set up his production less frequently.
Lastly, the change of setup cost will significantly influence the setup time
interval for the manufacturer. However, this cost change has relatively little impact
on the pricing decisions and marginal contributions of the supply chain members.
The manufacturer could counteract the variation of setup cost mainly through
changing his setup time interval.
3.5 Conclusion
This paper proposes an integrated supply chain model for optimal supplier selection,
pricing and inventory decisions in a multi-level supply chain. A cooperative game
approach is used to evaluate the supply chain members. A numerical study is
conducted to examine the integrated supply chain model and cooperative game
model. We also analyze the impacts of market scale parameter, component cost and
setup cost. The results of the numerical analysis show that: (1) one retailer market
scale rise increases the retail price and demand for the product in this market, without
influencing the other retail markets; moreover, an increased market scale increases
the importance of this market and the suppliers who provide components for the
product in this market; (2) the increase of component cost lengthens the setup time
interval of the manufacturer; (3) the change of setup cost has relatively little impact
on the pricing decisions and marginal contributions of the supply chain members.
However, this paper has several limitations which can be extended in the further
research. The competition among multiple products and among multiple retailers
is not covered in this paper. Under this competition, the demand of one
product / retailer is not only the function of his own price, but also the other
products/retailers prices. Secondly, vendor managed inventory (VMI) system is
the main inventory policy employed in the integrated model. Future research could
consider some other inventory system, such as integer multipliers mechanism, with
each supply chain member holding their own inventory. Also, we assume that the
production rate is greater than or equal to the demand rate to avoid shortage cost.
Without this assumption, the extra cost should be incorporated into the future
3 Integrated Supplier Selection, Pricing and Inventory Decisions 61
work. Finally, the Shapley value of the proposed cooperative game cannot be
regarded as a fair allocation rule for the total supply chain profit between the chain
members, because it might not be in the core. In the future work, we may focus on
working out a fair allocation rule for the gains of the integrated supply chain.
References
Xiaojun Wang
Abstract Food waste stemmed from inappropriate quality control and excessive
inventories is a major challenge in perishable food supply chains. Improvement of
visibility and traceability in food supply chains facilitated by modern tracking and
tracing technologies has great potential to reduce waste and improve operations
efficiency. This research aims to maximise food retailers profits through a pricing
approach based on dynamically identified food shelf life features. The proposed
model is evaluated through different pricing strategies to exploit the benefits from
utilising accurate product shelf life information captured through innovated
tracking and monitoring technologies. Numerical analysis is conducted in an
illustrative case study.
Keywords Product tracing and tracking Pricing Perishable food Shelf life
Waste reduction
4.1 Introduction
X. Wang (&)
Department of Management, University of Bristol,
8 Woodland Road, Bristol BS8 1TN, UK
e-mail: xiaojun.wang@bristol.ac.uk
due to damage and spoilage (Ferguson and Ketzenberg 2006). The perishable food
waste in a retail supply chain mainly stems from inappropriate quality control and
excessive inventories that have to be either marked down before the sell-by-date
or thrown away after the date. The financial consequence of waste for retailers and
manufacturers is severe. In the European grocery sector, products that are not
purchased before their sell-by date are estimated to cause costs running into
billions of dollars each year (Karkkainen 2003).
Due to the nature of perishable food, its quality can be considered as a dynamic
state that is decreasing continuously until the point when the food is unfit for sale
or consumption. For perishable food, there is a limited length of time during which
the food is held under predefined storage conditions and retains a required level of
quality after production. According to IFT (1974), shelf life can be defined as the
period between manufacture and retail purchase of a food product, during which
the product is of satisfactory quality. In practice, the shelf life information printed
on food labels is coded when products are packed by producers. However, most
perishable foods are temperature sensitive and their shelf life is therefore a
function of product characteristics, conditions under which the product is main-
tained, and time (Sahin et al. 2007). One limitation of current practice is that the
printed shelf life information does not reflect the real temperature variations that
occur through its life cycle. In fact, actual conditions frequently deviate from
specified conditions through the distribution processes. If appropriate temperatures
are not maintained, product quality can be compromised because of the devel-
opment of different kind of bacteria such as botulism, listeriosis and salmonella
(Mini and Labuza 1992). It may affect the characteristics of product quality and
result in a difference between the actual remaining shelf life and the expected shelf
life printed on a product label. Its consequence can be anything from increased
waste to legal actions.
The emergence of advanced product identification and sensory technologies
such as ratio frequency identification technology (RFID) and time temperature
indicator (TTI) provide great opportunities for effective management of perish-
able food. While these technologies are more often adopted (Koutsoumanis et al.
2005; Li et al. 2006a; Kelepouris et al. 2007), it aids to automatically capture the
product information regarding product identity, properties and related data (e.g.
temperature, humidity, and the time period during which products have been
exposed to the temperature in a supply chain process. etc.) in real time. Such
transparency generates the possibility that, as products pass through a supply
chain, decisions on product pricing based on current product quality, and the best
place and quantity that the products should be delivered, can be made dynam-
ically. This research attempts to prove the proposition that pricing decisions
based on dynamically identified food shelf life can reduce the level of waste and
improve retail profit.
4 Optimal Pricing with Dynamic Tracking in the Perishable Food Supply Chain 65
Inventory control for perishable products has been given much attention in the
inventory literature due to its prevalent existence in the industry. Nahmias (1982)
reviewed inventory models that incorporate perishability and/or decay of the good.
66 X. Wang
The models were classified into fixed life perishability models and random lifetime
models. Raafat (1991) provided a comprehensive survey on continuously deteri-
orating inventory models where deterioration is considered to be a function of
on-hand inventory. Goyal and Giri (2001) presented a review of deteriorating
inventory models after Raafats survey. Based on shelf-life characteristics, the
research classified inventory models into three categories: (1) modes for inventory
with a fixed lifetime; (2) models for inventory with a random lifetime; and
(3) models for inventory which decays corresponding to the proportional inventory
decrease in terms of its utility or physical quantity.
Another area of related research considers the issue of combined pricing and
inventory control for perishable products. Rajan et al. (1992) and Abad (1996)
considered the combined pricing and inventory setup in a deterministic setting.
They assumed the demand is affected by the product remaining shelf life. The
demand is deterministic for the product and is a decreasing function of the age of
the product and price. Determining the right price for a product is a complex
task, in which issues such as operating costs, availability of supply, future demand,
and how much customers value the product, have to be considered. In addition,
Rajan et al. (1992) stated that perishable products exhibit two characteristics
differentiate it from the common pricing problem: (1) physical decay or deterio-
ration of inventory; (2) decrease in market value called value drop associated with
each unit of inventory on hand. Perishable food has to be removed from the shelf
when it reaches its expiration date and this leads to an outdating cost. Furthermore,
customers may be sensitive to the quality of the expiring products and, as a result,
give them a lower valuation. In this case, expiring products have to be priced lower
than new products in order to make customers purchase those expiring.
Price discount is often used in practice when perishable products approach their
expiring date. Tersine (1994) presented an inventory model that incorporate the
relationship between a temporary price discount of a perishable product and
its remaining shelf life. One type of the research is maximising business profits
through pricing or allocating perishable products in an operational process
according to their fixed shelf life (Bhattacharjee and Ramesh 2000; Zhao and Zheng
2000). Another type of research employs the concept of product quality or product
value, to represent product utility attributes on which decisions on pricing can be
made (Kopalle et al. 1996; Li et al. 2006b). Kopalle et al. (1996) presented a dynamic
pricing model incorporating the relationship between expected quality and reference
price. Li et al. (2006b) proposed a dynamic pricing model to investigate the potential
benefits of the value tracing process. However, the model did not consider the cost of
price changes. In fact, price changes are costly in traditional retail stores.
For perishable food, product freshness represented as remaining shelf life has been
considered as the main quality attribute of interest. Ferguson and Ketzenberg
4 Optimal Pricing with Dynamic Tracking in the Perishable Food Supply Chain 67
(2006) found from observing consumers that the freshest products were selected
first. For perishable foods, product freshness has become an order winning criteria
for food purveyors. Peneau et al. (2007) also suggested that freshness is a quality
criterion of great importance to the consumer for the acceptance of fruit and
vegetables. In some researches (Li et al. 2006a, b; Blackburn and Scudder 2009),
the term product value is used to represent the remaining product shelf life.
As the temperature fluctuations across time are usually random, it is difficult to
estimate the accurate product shelf life and evaluate the product value.
The kinetic model approach is the most common method for the shelf life
prediction (Taoukis and Labuza 1989; Fu and Labuza 1993; Mizrahi 2000). With
the kinetic model, the shelf life is usually evaluated by the measurement of one or
more characteristic quality parameters. In the case of monitoring the change of a
quality parameter, according to Labuza (1984), the kinetic equation is expressed as:
d A
k A n 4:1
dt
where [A] is the measured value of the chosen key quality parameter, n is the
chemical reaction order and k is the constant rate of the reaction.
To simplify the computation, some researches (Li et al. 2006a, b; Blackburn
and Scudder 2009) use deterioration rates and time periods in a function to
measure the loss of product values as
V t V0 ekt : 4:2
Here V(t) is the product value at time t. V0 is the original product value when
t = 0. The parameter k is temperature dependent, and is assumed unchanged
during the time period T. The product value assessment function uses an
exponential function to describe the deteriorating product value over time. This
exponentially quality decaying weighing function has been adopted in operational
research on perishable food (Sorger 1988; Kopalle et al. 1996; Fibich et al. 2003).
Based on the food kinetics approach, the deterioration rate k can be calculated with
the available information of time and storage temperature to which food products
are exposed and known the kinetic parameters kA and EA.
From being packed and labelled to being placed on the retailers shelves, food
products have to pass through various supply chain processes. It is a chain of
individual deterioration processes. It can be quantitatively modelled through
aggregating the food deterioration processes in each individual time period. By
combining with the quality assessment function in Eq. 4.2, Li et al. (2006b)
proposed a value tracing process expressed as:
P
N
k i Ti
Vs V0 e i1 4:3
Here Vs is the dynamic product value at a decision point. V0 is the product value at
the beginning of the supply chain. N is the number of tracing processes. When a new
time period starts, the deterioration parameter k may change. The approach
68 X. Wang
generates a value tracing chain and enables the downstream supply chain companies
to identify product values (Vs) dynamically at different supply chain stages.
The process measures product quality variations at all stages in the basis of the
recorded food timetemperature profile (TTP). Recent development in automatic
tracking and monitoring technologies such as RFID and TTI provides a potential
solution to detect actual product quality deterioration in real time for large
volume product flows and complex supply chain processes. RFID technology can
remotely identify products instead of requiring a visual alignment of each product
with a scanner like barcode system, and can communicate with multiple products
simultaneously. TTI technologies permit of monitoring the freshness of a product
i.e. they allow evaluation of the shelf life in discrete or continuous time that results
from the temperature history experienced by the product. The applicability of TTIs
has been evaluated for various perishable foods including chilled fish, dairy
products, meat and poultry, frozen fruit and vegetables, and frozen meats. Sahin
et al. (2007) classified existing TTI devices according to the type of information
they provided in comparison with the classical open use by date printed on product
units. Among these, they defined a TTI Type 2 system which provides continuous
real-time information on products remaining shelf life by taking into account
temperature conditions under which the product has been maintained.
It is assumed that an RFID and TTI enabled novel traceability system is in place
so that TTP can be accurately recorded at each stage of the food supply chain and
shared in the network. At a decision point, the successive time periods need to be
traced back dynamically for accurate TTP that affect products remaining shelf
life. With the value tracing function, the retail price would be set to represent
actual food shelf life features. In other words, promotion should not only be used
as a passive method to rescue the product life relying on the expiration dates and
visual quality deterioration. Instead, it might be transformed in a more active
manner to manage demands and reduce waste.
In this section, a dynamic tracking and tracing enabled pricing model is presented.
The benefits of proposed model are evaluated with different pricing strategies that
are practically used in current retailer operations.
Notations:
Parameters:
D0 Nonnegative demand parameter dependent on product utility features
a Nonnegative coefficient representing the demand sensitivity to price
b Nonnegative coefficient representing the demand sensitivity to product value
k Nonnegative deterioration rate of product value or shelf life
(continued)
4 Optimal Pricing with Dynamic Tracking in the Perishable Food Supply Chain 69
(continued)
Co Unit operational cost
Cp Unit price change cost
Q Product ordering quantity
t Time during a selling period where 0 t T
T The length of selling period
Te The time at which the perceived product value reaches the point that will not be
considered by consumers
Tm The length of normal price selling period
V0 The original product quality value
Vs The identified product value at the start of retail stage
P The initial price at the start of a selling period
Variables
P(t) Time dependent retail price
V(t) Time dependent product value
ED Expected demand for a selling period (0, T)
QT
Total profit in a selling period
h Price discount rate
Determining the right price to charge customers for a product is a complex task.
It requires consideration of operating cost, available stock, the demand, and how
much customer values the product. The crucial element that influences pricing
decisions is demand and how it reacts to price changes and other factors
(Elmaghraby and Keskinocak 2003). An important fact is that not all consumers
are the same. Their characteristics and circumstances condition their behaviour,
their purchasing decisions and, in particular, their price evaluation (Estelami et al.
2001; Rosa-Daz 2004). Slade (1998) explained that there are two types of
consumers: some are not price sensitive, whereas others consider price funda-
mental in their decision to make a purchase. In this research, consumers are
assumed to be price sensitive. One commonly used price-dependent linear demand
function in the economic research is described as:
l K aP: 4:4
K and a are nonnegative constant. l is the mean of a demand. Although, the linear
demand function has often been criticized as it assumes that the expected demand
is a strictly linear function of the price, it has been employed extensively relating
to pricing and inventory problems (Lau and Lau 1988; Polatoglu 1991; Abad 1996;
Chew et al. 2009).
In addition to price sensitivity, consumers may be sensitive to the quality of
goods. Food quality can be defined as the assemblage of properties which
70 X. Wang
differentiate individual units and influence the degree of acceptability of the food
by the consumer (Kramer and Twigg 1968). For many products, product quality is
observable before purchase. It is an important factor influencing a consumers
purchase decision. Rosa-Daz (2006) argued the influence of price on consumers
decisions depends on the manner in which the price is perceived and evaluated.
For perishable food products, many consumers consider a new product has a
higher quality than an expiring one and give it a lower evaluation. Considering
supermarket customers, they will prefer to buy newly replenished goods instead of
expiring ones if there is a difference in expiration dates. When prices are the same,
they will prefer the new one. Hardie et al. (1993) developed the notion of reference
quality and empirically demonstrated that differences between observed and
reference quality can significantly affect purchase probabilities.
In this research, customers are assumed being aware of price and the possible
quality deterioration. Therefore, there are two characteristics different to common
pricing problems as discussed earlier. Both the price (P) and the product value
(V) that represents the remaining shelf life are taken into account as influential
factors in a deterministic demand function as Eq. 4.5.
f Dt D0 a Pt b Vt 0; 4:5
a and b are the nonnegative coefficients representing the demand sensitivity to a
product price, and the demand sensitivity to the identified product quality value
respectively (a, b [ 0). D0 is a nonnegative demand parameter dependent on
market size (Lau and Lau 2002). While a price increase negatively affects a
demand, product value apparently has positive effects to the demand. The time
variable t is time elapsed since the start of a selling period. To simplify the
problem, it is assumed that other factors that influence a pricing policy such as
retailers overall price image, competitors price, and customers sensitivity to
product value have been considered in coefficients a and b.
The evaluation of food quality loss is normally generated by a group of
professionals with knowledge and resources. However, without knowledge of
product quality evolution in general and of sensory analysis in particular, it may
be better to predict consumer perception of quality (Peneau et al. 2007). Here, the
drop in product value is modelled from the consumers point of view as the
product ages through the selling period. This drop in value is manifested through a
lessened willingness to pay for the product by the consumer. Through the study of
consumers perceptions and behaviour in grocery perishable categories through a
survey, Tsiros and Heilman (2005) found that consumers willingness to pay
decreases linearly through the shelf life for products with relative lower product
quality risk (PQR) (Lettuce, carrots, milk, and yogurt etc.) and exponentially for
products with relatively higher PQR (beef and chicken etc.). In addition, their
research also found that for this product category, consumers check the expiration
dates before the purchase more often than other products such as lettuce, carrots or
milk. Coincidently, in the grocery supermarket, beef and chicken, are the main
perishable products discounted when they approach their expirations dates.
4 Optimal Pricing with Dynamic Tracking in the Perishable Food Supply Chain 71
In this study, we focus on this category as consumers are more sensitive to the
shelf life for products with high PQR and more dynamic pricing strategies are
employed in the current supermarkets operations. Therefore, as customer per-
ceived product value decreases over time against a given expiration date during a
selling period, the demand function can be then extended as
f Dt D0 a Pt b Vekt 0
RT 4:6
ED fc Dt dt:
0
where:
ED expected demand for a selling period (0, T);
D0 the demand parameter (dependent on product utility features) (Lau and
Lau 2002), D0 [ 0;
P(t) time dependent retail price based on the pricing strategy and P(t) is less than
the maximum price at which consumers would stop buying
(0 \ P(t) \ PM);
V the product value which is dependent on the product shelf life features
(0 \ V \ 100%). The value of V is proportional to remaining product shelf
life derived from a given expiration date;
k the nonnegative deterioration rate of product value or shelf life.
The proposed pricing model is evaluated by assessing the profit realized by the
retailer. The overall profit of a monopolistic retailer is modelled in a given selling
period of duration between t = 0 and t = T. It is obtained through calculating the
revenue and costs involved. For simplification, operational costs (costs for
purchasing, stock keeping and logistics, etc.) are incorporated into a unit opera-
tional cost Co. When the pricing strategy contains price change, a unit cost Cp
should be added for each change. Due to the nature of food product uncertain
storage or weather conditions, the real product shelf lives may differ to those
printed on product labels. For some products, the end of product life can be
visualized and retailer staff or customers will stop selling or purchasing expired
products. For others, the end of product life may not be visualized without
scientific testing. The consumption of unsafe food products may cause illness. The
retailer has to face a series consequence costs such as compensation and damaged
reputation. Nevertheless, this scenario will not be explicitly modelled as such
incidents are also rare cases in the grocery retail operations.
In addition, the length of the selling period, T, may also affect overall profits
regarding its correlations with Te. Te is the time at which the perceived product
value reaches the point that will not be considered by consumers. In practice,
72 X. Wang
retailers often impose a certain shelf life requirement on their purchasing contract.
To minimize the potential outdating cost, the required shelf life is longer than the
length of selling period Te T [ 0. It means all products are sold with
acceptable quality condition during the selling period (0, T), if products are
properly maintained under the controlled storage condition. Considering a scenario
where retailers fix the price of a product over a selling period, the profit in a selling
Q
period, T ; can be estimated as the revenue over a selling period, (0, T), minus all
incurring costs
Y
T ZT
Ptf Dt dt QCo 4:7
0
Here Q is the quantity of the product ordered to sell (ED B Q). The difference
between the ordering quantity (Q) and expected demand (ED) is the quantity of the
products that need to be disposed of. If further cost is required to dispose those
products (Cw), the profit function can be expressed as:
2 3
YT ZT ZT
Ptf Dt dt QCo 4Q f Dt dt5Cw 4:8
0 0
For perishable food products, the phenomenon of price changing over the life
cycle of a product is common. The retailers objective is to maximise the profit
over a selling period by setting optimal prices P*(t). The optimisation problem can,
therefore, written as a conventional optimisation problem: find the optimal pricing
strategy P*(t) for 0 t T; such that:
Y
T
P t max PT Pt; 4:9
Pt
Although short-term price changes for fresh vegetables and meat are routine, a
common approach employed in retail stores is not to regularly change the price of
initial offering products but to markdown the price of perishable foods which
approach their expiration dates in order to minimise the waste. Often, the price
reduction decisions are made based on operators own judgements especially in the
situation where product package is damaged and product quality is compromised
obviously. More often, those decisions rely heavily on the shelf life information
printed on product packages which could deviate from its actual remaining shelf
life. Some products expatriation date could move forward as the specified
4 Optimal Pricing with Dynamic Tracking in the Perishable Food Supply Chain 73
Supermarkets all carry some kinds of price markdown policies although they may differ
across goods and vary over time for each good. We start with a simple strategy in which
retailer set up a lower price, to attract customers to purchase products approaching their
expiration date. In result, two different prices are set up in a selling period:
P 0 t\Tm ;
Pt 4:10
P1 h Tm t T:
Here, Tm is the length of normal selling period after which a price discount h is
deployed for a given product. As a price markdown is always applied before a
product reaches its expiration date, the selling period is then divided into two
intervals: (0, Tm) and (Tm, T). Thus, the demand function can then be expressed as:
ZTm ZT
ED f Ddt fd Ddt 4:11
0 Tm
As two prices exist in a selling period, the profit function is described as:
Y
T ZTm ZT
P f Dt dt P hP CP f Dt dt QCo 4:12
0 Tm
By substituting Eq. 4.6 into Eq. 4.12, the profit function is obtained as:
QT
PD0 a P Tm bkV 1 ek Tm P1 h CP
4:13
D0 a P a hP T Tm bkV ek Tm ek T QCo
To optimise the expected profit, the derivatives with respect to the price
discount h are obtained as:
QT
o b V k Tm
P e ek T a P2P CP PD0 T Tm 2a P2 T Tm h
oh k
Q 4:14
o2 T 2
2aP T Tm
o2 h
74 X. Wang
QT
o2
As T [ Tm ; so o2 h \ 0:
The derivative above confirms the convexity of the profit function. Then, the
optimal price discount solution h* is obtained as:
b Vek Tm ek T 2P CP D0
h 4:15
2a kPT Tm 2P 2a P
Subject to:
bV
ED D0 a P ahPT ah PTm 1 ek T Q 4:16
k
bV
Q 1 ek T D0 a PT
0\h k \100%; :
aPT Tm
Through Eq. 4.16, the optimal price discount rate h* and the optimal profit level
can then be determined based on the identified product value V, the initial price P,
product deterioration rate k, price markdown cost CP, products ordering quantity
Q, length of normal selling period Tm, and demand sensitivity features.
Option I analyzed the situation that two different prices are set up in a selling
period. In practice, retailers often employ a more flexible price markdown policy.
For example, in order to reduce the food spoilage waste, Tesco carries three price
reductions on perishable foods in their dairy operations. To provide an insight into
the impact of the value tracing model on a retailers profit, a more generalized
price markdown case is discussed. For simplification, only operational cost (Co)
and price change cost (CP) are considered.
Referring back to Sect. 4.4.1 if one price markdown is employed, the profit
function can be defined as:
P1 P ED0 h1 P Cp EDm1 QC0 4:17
Here h1 P is the discounted price. ED0 is the demand in normal price selling
period. EDm1 is the demand in price discount period. If the price is marked down
twice, then the selling period is divided into three periods and the profit function
therefore becomes:
P2 P ED0 h1 P Cp EDm1 h2 P 2Cp EDm2 QCo 4:18
Accordingly, if n times price markdown occur, the profit function can be
written as
4 Optimal Pricing with Dynamic Tracking in the Perishable Food Supply Chain 75
X
n
Pn P ED0 hi P iCp EDmi QCo 4:19
i1
If each price markdown rate is the same given rate h, it can be expressed as
hi 1 ih ; i 1; . . .; n: 4:20
By substituting Eq. 4.20 into Eq. 4.19, a pricing model for multiple price
markdown strategy that optimise the overall retail profit is formulated as Eq. 4.21.
The constraints of the optimisation problem are described in Eq. 4.22.
Objective:
X
n
Max Pn P ED0 P ihP iCp EDmi QCo 4:21
i1
Subject to:
ZTm1
ED0 f Dt dt;
0
Tmi1
Z
EDmi f Dt dt
Tmi
f Dt D0 a Pt b Vekt 0
Xn
ED0 EDmi Q
i1
1
0\h\ ; n is the number of price markdowns. 4:22
n
Similar to the approach discussed in Sect. 4.4.1, through differentiating the
profit functions with respect to the price discount h, the derivative above confirms
the convexity of the profit function and the optimal price discount solution h* is
obtained as:
n
P k T k Tmi
Pn
bV e e 2ap D0 nT Tmi
h i1 n i1
CP =2 p
P Pn
2a kp n2 t 2n 1Tmi 2ap n2 t 2n 1Tmi
i1 i1
4:23
Here, n is the number of price markdowns in a selling period. If the price is only
marked down once, the optimal solution h*equals to the result in Option I. As the
shelf-life of perishable food can be considered as a dynamic state that is decreasing
continuously until the point when the food is unfit for sale, more regular price
76 X. Wang
By replacing the price function Eq. 4.10 with Eq. 4.24, the profit function for
multiple price markdowns is formulated as
X
n
Pn p ED0 pehTmi iCp EDmi QCo 4:25
i1
Subject to:
f Dt D0 a P eh t b V ekt 0:
ZT
f Dt dt Q
0
Although various pricing models have been intensively studied in the literature,
the proposed pricing approach is developed by utilising the accurate food shelf life
information. The shelf-life monitoring technology underlies the application of
such a dynamic pricing approach, as the tracking information is a prerequisite for
accurately evaluating product shelf life, and consequently estimating the quality-
dependent demands. Nevertheless, the performance of the model is also dependent
4 Optimal Pricing with Dynamic Tracking in the Perishable Food Supply Chain 77
on demand sensitivity, product shelf life feature, ordering quantity and unit price
change cost, which will be further analysed in the illustrative case.
Table 4.1 describes the example of input parameters and simulation results for a
meat product that four selected stores are selling. The demand coefficients, a, b,
and D0, which varies between different stores were acquired according to the sales
record. Using store 1 as an example, for this particular item, the sales volume
varies from day to day, and on average over a typical week works out as 54 packs
per day. In the promotion period (e.g. buy one get one free), the demand reaches
about 120 packs per day. Assuming a roughly linear relationship between price
and demand, the demand can be obtained from Eq. 4.5. However, consumers
evaluate products by comparing the actual price with a reference or expected price
determined from perceived product quality and the price-quality correlation of a
product category (Ordonez 1998). However, there is few empirical evidence of the
effect of perceived product value on demand. It is assumed that its effect on
the demand is the same as the effect of price. Through inputting the sales data in
the normal selling and promotion periods into Eq. 4.5 respectively, the demand
function can be obtained as:
f Dt 9:79 1:83Pt 1:83 Vt 4:27
The derived levels of price and value coefficients, a and b result in price and
value elasticises that are comparable to those obtained in Kopalle et al. (1996) and
Li et al. (2006b).
78
Table 4.1 Example of input parameters and results for a meat product
Input parameters Price discount Optimal discount Profit difference
D0 a b P k Co CP Q T Tm Vs h(%) h*(%) PD(%)
Store 1 9.79 1.83 1.83 4.5 0.0067/h 3.5 0.05/C 740 240 h 168 h 0.95 25 26.9 3.26
Store 2 8.32 1.56 1.56 630 0.97 25 26.3 2.27
Store 3 12.73 2.38 2.38 950 0.88 25 27.9 5.44
Store 4 7.83 1.46 1.46 590 0.90 25 28.3 6.14
X. Wang
4 Optimal Pricing with Dynamic Tracking in the Perishable Food Supply Chain 79
Here, we assume that the product value at the beginning of the supply chain
equals to 1 (100%), i.e. V0 = 1. The value deterioration rate and operation time at
different supply chain stages can be derived through the recorded time and
temperature information. With these parameters, the identified product value Vs
at the beginning of retail stage can be quantitatively estimated (Li et al. 2006b). To
simplify the simulation experiment, the average value deterioration parameter is
obtained through the shelf life information of the item and the assumption of
exponential decay.
In addition, according to the interview with the store managers, the total
operational cost (Co), which includes the purchase price, ordering cost, delivery
cost, and inventory carrying cost, is estimated at 2 per pack. A minimum 10 days
remaining shelf life is required for all the products delivered to the retail stores.
Consumers will stop buying those products when the time passes their expiration
dates. These products have to be removed from the shelf and disposed. The
average value deterioration parameter is estimated to be k = 0.0076 per hour.
Supermarkets mark down the price of the perishable food at discrete points before
it reaches its expiration date. For this particular item, 25% discount is given in the
last three day before reaching its sell by date. By inputting relevant parameters,
the optimal price discount rate, h*, can then be calculated. Then through replacing
current price discount figure and calculated optimal discount rate h* into Eq. 4.13,
the profit difference is obtained as illustrated in Table 4.1.
Table 4.2 shows the second example, where simulation was conducted for a
fresh vegetable item with a relatively shorter shelf-life. The demand coefficients, a,
b, and D0, are derived through the same approach discussed earlier. The value
deterioration parameter is estimated to be k = 0.0216 per hour. The results of both
examples show the improvement in profit from employment of the value tracing
enabled pricing model. However, our analytical and simulation results have also
demonstrated that the demand coefficients, product shelf life features, pricing
policy, and actual product quality control conditions significantly affect the
performance of the proposed pricing model.
Furthermore, the analysis also highlights the importance of the accuracy of the
deterioration rate in the pricing decision. Multiple markdown pricing strategy is
helpful in meeting this criterion. If a number of markdown processes are deployed
in a selling period, more pricing decisions concerning the markdown rate and the
point at which it occurs can be decided on the basis of the dynamically identified
product deterioration rate. However, in many supermarkets, current practice still
requires extensive manual work in enforcing price changes. Consequently, there is
a dilemma. On one hand increasing the number of price markdown may improve
the probability of profits increase as the deterioration rate captured by the tracking
technology in a selling period is more accurate. On the other hand, an increase in
price changes adds on extra price change cost (CP). To investigate how the cost of
changing price and number of times markdown occurs affect model performance,
the cost ratio Rc (Rc = CP/C0) is defined. Different values of Rc are used to
simulate the profits difference under various assumptions for the number of
markdown times. The results shows that when Rc is low which means that the price
80
Table 4.2 Example of input parameters and results for a fresh vegetable product
Input parameters Price discount Optimal discount Profit difference
D0 a b P k Co CP Q T Tm Vs h(%) h*(%) PD(%)
Store 1 7.92 4.86 4.86 1.49 0.0216/h 1.0 0.05/C 280 72 h 48 h 0.95 20 34.8 8.43
Store 2 6.73 4.13 4.13 240 0.97 20 34.6 7.91
Store 3 10.3 6.32 6.32 360 0.88 20 35.5 11.74
Store 4 6.34 3.89 3.89 220 0.90 20 35.3 10.12
X. Wang
4 Optimal Pricing with Dynamic Tracking in the Perishable Food Supply Chain 81
Fig. 4.1 Performance against various deterioration difference indexes k/k with given product
value differences DV
change cost is relatively low compared to the operational cost, profit increases as
the number of markdown events increases. When the value of Rc keeps on
decreasing, its corresponding curves of profit against markdown times tend to the
curve in which price change cost is ignored in the profit function.
In addition to the benefit of the proposed pricing approach shown in the simulation
results over the current practice, it also demonstrates that the demand coefficients,
product shelf life features, pricing policy, and ordering quantity significantly affect
the performance of the proposed pricing model. Here, further analysis is conducted to
examine the effect of those factors on the model performance.
The simulation results indicate that the benefit from the proposed pricing approach
is noticeably influenced by the actual remaining shelf life (Vs) identified at the start
of selling period. For example, the profit increase is greater in the situation where
there is a significant difference comparing to its original product quality value (V0).
To provide further insight of its impact on the proposed model, the benefits of the
proposed approach were simulated with varying quality value differences. The
quality value difference is defined as DV, where
DV V0 Vs =V0 : 4:28
For perishable foods, products incur a quality loss as they go through various
supply chain processes as its freshness deteriorates over time. In addition, the
value of q also depends on quality deterioration rate (k) and operation time at
different supply chain stages. Practically, it is likely that the deterioration rate (k)
captured by the tracking technology is different from the estimated deterioration
rate (k). This discrepancy would have an effect on the model performance.
82 X. Wang
The fresh vegetable item in store 1 was simulated with the parameters given in
Table 4.2.
The same assumption for the demand sensitivities is held (a = b)
The spoilage cost occurs as wastage cost, which stems only from products that
have to be disposed after the sell-by-date.
Overall the simulation results show that the proposed pricing approach deliver
better performance when perishable foods incur a big quality loss through its supply
chain processes. The quality feature, representing as remaining shelf life in this case,
will affect the customer demand along with its price. With a large DV value,
customers are reluctant to purchase those products which may result in unsold stock
and cause wastage problem. The proposed pricing approach attempts to reduce the
price to stimulate the demand to avoid any further price reduction or extra cost for
disposing waste. In addition, the proposed pricing approach performs better when the
identified actual deterioration rate k is greater than the estimated rate (k) derived
from the expiration date. A greater identified deterioration rate means that product
quality actually deteriorates faster than expected due to improper storage condition.
This consequently imposes impact on customer demand. More benefit can be
generated from the proposed approach by initiating an optimal price discount which
reflects its actual remaining shelf life. Although the overall trends of PD curves move
upwards from left to right, the simulation results also show that PD curve for the DDI,
k/k = 0.9, moves downwards first before moving upwards from left to right. It is
caused by the demand constraint as the quantity of sold items should always be
smaller or equal to the ordering quantity.
While the same assumption (a = b) was held in the above simulations, the
demand coefficients could significantly affect the model performance. To
investigate the performance of the proposed approach with different features of the
demand elasticity to perceived product price and product value, sensitivity analysis
is conducted on the demand elasticity ratio, named pV ratio as:.
p V Ratio b=a 4:29
The simulation is performed by simulating the profit differences against given
product quality loss (DV) with various demand elasticity ratios over a selling
period. The following assumptions hold for this simulation:
The meat product in store 1 was simulated with the parameters given in Table 4.1.
The product deterioration rate (k) captured by the automatic tracking technology
is the same as the rate (k) perceived by the printed expiration date on package.
The spoilage cost occurs as wastage cost.
4 Optimal Pricing with Dynamic Tracking in the Perishable Food Supply Chain 83
Fig. 4.2 Performance against varying price-value sensitivity ratio (b/a) with given product
quality differences DV
Through the simulation, the difference in impact of the demand elasticity to the
food price and quality on the retailers performance is observed. Profit difference,
used as the performance indicator, is demonstrated against different DV with
various demand sensitivity ratios. The analysis results in Fig. 4.2 show that the
proposed approach generally performs better than the current practice, while the
magnitude of the profit increase varies between different pV ratios. It also reveals
the fact that the proposed model generates more benefits when the demand
elasticity to the product quality is greater than that to the price. It means that, for
those products that freshness is the key quality indicator, consumers are more
sensitive to the remaining shelf life feature when they make their purchasing
decisions. A price discount decision considering its actual remaining shelf life will
contribute to the improvement of its business performance. In contrast, for those
products that the demand elasticity to the quality is smaller than that to price, the
contribution to profit increase is less significant.
The analysis results have demonstrated the benefit of the proposed approach for
perishable food management. However, the demand sensitivity nature of perish-
able food products, the pricing policy, and the actual food product quality control
conditions would significantly affect the benefits of the proposed pricing strategy.
An appropriate price policy needs to be determined based on accurately identified
food product deterioration and shelf-life features to guarantee the maximised
profit. This confirms the critical importance of dynamically tracing and monitoring
food supply chain processes for sustaining the business competitiveness.
From a technological point of view, the development of RFID technology and
sensory devices has provided an effective and practical solution in tracking
and tracing temperature and other environmental parameters in food production
and distribution. Many trial projects relating to these technologies have been
conducted or are in progress. The benefits of utilising these technologies to
84 X. Wang
improve supply chain management have been reported extensively such as data
capture automation, item-level product tracking, and improved business process
visibility. In addition, the cost of these technologies is also decreasing. This may
provide a more promising return of the investment on appropriate applications.
However, there are still some obstacles. First, globalisation has made the food
supply chain more complex than before. For example, exotic fruits placed on
supermarket shelves, may travel thousands of miles before reaching consumers.
In the case of cooked pastrami beef, although it may be produced by a local
manufacturer, the raw beef could come from another continent (e.g. South
America). The complexity of such a supply chain is determined by the nature of
products, the legal and quality assurance requirements of food safety, and the
distribution facilities available from production to consumption (Smith and Sparks
2004). In addition, the number of product varieties in the perishable food category
has increased significantly in recent years, and has made the implementation of the
proposed approach more difficult. Another obstacle of implementing proposed
application is that it requires having value tracing process in place throughout
whole food supply chains. This does not only require large investment from
retailers but also imposes extra costs on other parties in the food supply chain. This
adds a substantial amount of uncertainty due to different perspectives and
expectations. As the supply chain captain, retailers have the resources and
influence to drive such innovation. However, the success of such innovation
requires collaboration between all parties of food supply chains.
This research attempts to explore the opportunities from using tracking and tracing
functions to improve the performance of perishable food supply chains. The
development of RFID technology and sensory devices has provided an effective
and practical solution in tracking and tracing temperature and other environmental
parameters in food production and distribution. With the technological develop-
ment and the application of food traceability systems, supply chain visibility and
accuracy of product shelf life information can be significantly improved. Conse-
quently, consumer buying intentions could be more dependent on the real-time
product quality or shelf life features, rather than mainly dependent on a given
expiration date. A product value tracing model was discussed and evaluated
combining with different pricing strategies. Numerical simulations based on a
retail chain case were deployed to analyze the model performance. The analysis
results demonstrate the critical importance of dynamically tracing and monitoring
food supply chain processes for sustaining the business competitiveness.
A major contribution of this research is the proposed innovation of using the
tracking and tracing function to support the pricing decisions. The study indicates
that pricing decisions based on the dynamic identified product value can lead to a
significant improvement in profit. However, our analytical and simulation results
4 Optimal Pricing with Dynamic Tracking in the Perishable Food Supply Chain 85
have also demonstrated that product shelf life features, outdating cost, pricing
policy, and actual product quality control conditions significantly affect the per-
formance of the value tracing enabled pricing model. An appropriate price policy
needs to be adopted according to shelf life features and the accurately identified
product deterioration in order to guarantee the maximised profit. Such a trans-
formation would not only improve food safety management, but be a strategic
innovation method for marketing, food quality management, customer service and
supply chain operations.
For simplification, a linear deterministic demand is assumed in this research.
From a different perspective, there have been numerous empirical studies
investigating demand functions for a wide variety of products. These demand
functions can be an iso-elastic or exponential functions of price. In fact different
demand functions could lead to different results. It will be useful to investigate
further research problem using stochastic models. Furthermore, this research only
focuses on evaluating the benefits from the retailers perspective. However, the
success of such innovation requires collaboration between all parties of the food
supply chain. Therefore, it is essential to consider all the costs incurred through the
whole tracing period. Future research could attempt to compare the benefits from
the perspective of supply chain.
References
Abad, P. L. (1996). Optimal pricing and lot-sizing under conditions of perishability and partial
backordering. Management Science, 42(8), 10931104.
Bhattacharjee, S., & Ramesh, R. (2000). A multi-period profit maximizing model for retail supply
chain management: An integration of demand and supply-side mechanisms. European
Journal of Operational Research, 122, 584601.
Blackburn, J., & Scudder, G. (2009). Supply chain strategies for perishable products: The case of
fresh produce. Production and Operations Management, 18(2), 129137.
Chew, E. P., Lee, C., & Liu, R. (2009). Joint inventory allocation and pricing decisions for
perishable products. International Journal of Production Economics, 120(1), 139150.
Elmaghraby, W., & Keskinocak, P. (2003). Dynamic pricing in the presence of inventory
considerations: Research overview, current practices, and future directions. Management
Science, 49(10), 12871309.
Estelami, H., Lehamn, D. R., & Holden, A. C. (2001). Macro-econimic determinants of consumer
price knowledge: Ameta-analysis of four decades of research. International Journal of
Research in Marketing, 18, 341355.
Ferguson, M., & Ketzenberg, M. E. (2006). Information sharing to improve retail product
freshness of perishables. Production and Operations Management, 15(1), 5773.
Fibich, G., Gavious, A., & Lowengart, O. (2003). Explicit solutions of optimization models and
differential games with nonsmooth (asymmetric) reference-price effects. Operations
Research, 51(5), 721734.
Florence, D., & Queree, C. (1993). Traceabilityproblem or opportunity? Logistics Information
Management, 6(4), 36.
Fu, B., & Labuza, T. P. (1993). Shelf-life prediction: Theory and application. Food Control,
4(3), 125133.
Golan, E., Krissoff, B., & Kuchler, F. (2004). Food traceability. Amber Waves, 2(2), 1422.
86 X. Wang
Goyal, S. K., & Giri, B. C. (2001). Recent trends in modelling of deteriorating inventory.
European Journal of Operational Research, 134, 116.
Hardie, B. G. S., Johnson, E. J., & Fader, P. S. (1993). Modelling loss aversion and reference
dependence effect on brand choice. Marketing Science, 12, 378394.
IFT. (1974). Shelf life of foods. Journal of Food Science, 39, 14.
Karkkainen, M. (2003). Increasing efficiency in the supply chain for short shelf life goods using
RFID tagging. International Journal of Retail and Distribution Management, 31(10), 529536.
Kelepouris, T., Pramatari, K., & Doukidis, G. (2007). RFID-enabled traceability in the food
supply chain. Industrial Management and Data Systems, 107(12), 183200.
Kopalle, P. K., Rao, A. G., & Assuncao, J. L. (1996). Asymmetric reference price effects and
dynamic pricing policies. Marketing Science, 15(1), 6085.
Koutsoumanis, K., Taoukis, P. S., & Nychas, G. J. E. (2005). Development of a safety monitoring
and assurance system for chilled food products. International Journal of Food Microbiology,
100(13), 253260.
Kramer, A., & Twigg, B. (1968). Measure of frozen food quality and quality changes. In
D. K. Tressler (Ed.), The freezing perservation of foods (4th ed.). Westport: AVI Pub. Co.
Labuza, T. P. (1984). Application of chemical-kinetics to deterioration of foods. Journal of
Chemical Education, 61(4), 348358.
Lau, A. H., & Lau, H. (1988). The newsboy problem with price-dependent demand distribution.
IIE Transactions, 20, 16817.
Lau, A. H. L., & Lau, H. S. (2002). The effects of reducing demand uncertainty in a
manufacturer-retailer channel for single-period products. Computers and Operations
Research, 29, 15831602.
Li, D., Kehoe, D., & Drake, P. (2006a). Dynamic planning with a wireless product identification
technology in food supply chains. International Journal of Advanced Manufacturing
Technology, 30(910), 938944.
Li, D., Tang, O., OBrien, C., & Wang, X. (2006b). Improve food retail supply chain operations
with dynamic pricing and product tracing. International Journal Service Operations and
Informatics, 1(4), 347362.
Mini, S., & Labuza, T. P. (1992). Consumer perceptions of consumer timetemperature indicators
for use on refrigerated dairy goods. Journal of Dairy Science, 75, 31673176.
Mizrahi, S. (2000). Accelerated shelf-life tests. In D. Kilcast & P. Subramaniam (Eds.), The
stability and shelf-life of food. Cambridge: Woodhead Publishing Limited.
Nahmias, S. (1982). Perishable inventory theory: A review. Operations Research, 30(3), 680708.
Ordonez, L. D. (1998). The effect of correlation between price and quality on consumer choice.
Organizational Behaviour and Human Decision Process, 75(3), 258273.
US Patent (2006). Electronic shelf label systems and methods [online], United States Patent
7040536. Available from: http://www.patentstorm.us/patents/7040536.html. Accessed 10
Oct 2007.
Peneau, S., Brockhoff, P. B., Escher, F., & Nuessli, J. (2007). A comprehensive approach to evaluate
the freshness of strawberries and carrots. Postharvest Biology and Technology, 45(1), 2029.
Polatoglu, L. H. (1991). Optimal order quantity and pricing decisions in single period inventory
systems. International Journal of Production Economics, 23, 175185.
Raafat, F. (1991). Survey of literature on continuously deteriorating inventory model. Journal of
the Operational Research Society, 42, 2737.
Rajan, A., Rakesh, S., & Steinberg, R. (1992). Dynamic pricing and ordering decisions by a
monopolist. Management Science, 38(2), 240262.
Regattieri, A., Gamberi, M., & Manzini, R. (2007). Traceability of food products: General
framework and experimental evidence. Journal of Food Engineering, 81(2), 347356.
Rosa-Daz, I. M. (2004). Price knowledge: Effects of consumers attitudes towards prices,
demographics, and socio-cultural characteristics. Journal of Product and Brand Management,
13(6), 406428.
Rosa-Daz, I. M. (2006). Demand restrictions in price-based decisions: Managers versus
consumers. Journal of Product and Demand Management, 15(3), 214224.
4 Optimal Pricing with Dynamic Tracking in the Perishable Food Supply Chain 87
Sahin, F., Baba, M. Z., Dallery, Y., & Vaillant, R. (2007). Ensuring supply chain safety through
time temperature integrators. International Journal of Logistics Management, 18(1), 102124.
Slade, M. E. (1998). Optimal pricing with costly adjustment: Evidence from retail grocery prices.
Review of Economic Studies, 65, 87107.
Smith, D., & Sparks, L. (2004). Temperature controlled supply chains. In M. A. Bourlakis & P.
W. H. Weightman (Eds.), Food supply chain management. Oxford: Blackwell.
Sorger, G. (1988). Reference price formation and optimal marketing strategies. Optimal Control
Theory and Economic Analysis, 3, 97120.
Taoukis, P. S., & Labuza, T. P. (1989). Applicability of timetemperature indicators as shelf-life
monitors of food-products. Journal of Food Science, 54(4), 783788.
Tersine, R. J. (1994). Principles of inventory and materials management. Englewood Cliffs:
Prentice Hall.
Tsiros, M., & Heilman, C. M. (2005). The effect of expiration dates and perceived risk on
purchasing behaviour in grocery store perishable categories. Journal of Marketing,
69(2), 114129.
Wang, X., Li, D., & Li, L. (2009). Adding value of food traceability to businesses: A supply chain
management approach. International Journal of Service Operations and Informatics,
4(3), 232257.
Zhao, W., & Zheng, Y. S. (2000). Optimal dynamic pricing for perishable assets with
non-homogeneous demand. Management Science, 46(3), 375388.
Chapter 5
Identifying Supply Chain Value
Using RFID-Enabled Distributed
Decision-Making for Food Quality
and Assurance
Abstract This chapter conceptually considers how retailers might integrate radio
frequency identification (RFID) technology and distributed decision making (DDM)
to provide real-time product information visibility to enhance food retail supply chain
decision-making for assurance of quality and safety. A pilot research project in one
UK retailers fresh meat supply chain is used to illustrate these concepts. Value stream
mapping (VSM) and on-site interviews were the primary research methods for the
project. A current state VSM found non-value adding activities in time and labor
processes, which in turn led to increased errors. A future state VSM suggests that
RFID technology would increase information visibility, reduce handling and check-
ing, and increase speed when it is integrated with an appropriate DDM system. There
are potentially large savings to be had by implementing RFID and DDM systems
where information is required for traceability and control of food quality and safety.
T. Butcher
Applied Logistics Research Group, RMIT University,
Melbourne, VIC 3000, Australia
e-mail: tim.butcher@rmit.edu.au
D. B. Grant (&)
Logistics Institute, Business School, University of Hull,
Hull, HU6 7RX, UK
e-mail: d.grant@hull.ac.uk
5.1 Introduction
The food products industry is a sector where traceability and control of food quality
and safety are critical. Surveillance and transparency is increasingly essential across
supply chains to assure consumers and food standards bodies, of the condition of food
products throughout their journey from field-to-fork (Greenwood 2011; Rogers 2011).
Management of fresh food supply chains requires quick response decision-making
because the quality of fresh agricultural products such as fish, meat, poultry or fruit
and vegetables can change rapidly (Regattieri et al. 2007).
Radio frequency identification (RFID) is considered an appropriate techno-
logical solution to deliver the necessary supply chain visibility and product
traceability to deliver the operational accuracy and manipulation required for such
goods. RFID has a distinct advantage over conventional barcoding technologies, as
it eliminates the line of sight requirement; in other words data capture can occur
without human intervention to reduce leadtime, human error and therefore, cost
(Butcher 2007a; Boeck and Wamba 2008). However, while many retailers believe
RFID has a number of benefits to improve overall business performance there are
some that consider that RFID technology has not been entirely proven as regards
technological and business advantages, particularly in the fresh food supply chain
(Vijayaraman and Osyk 2006).
The use of RFID throughout this supply chain would yield vast volumes of
data on a daily basis. Thus, an important issue here is how could retailers make
sense of such data for decision-making purposes? A homily in the management
information systems discipline, notes that data becomes information only when it
is timely and relevant. In that case, how then can retailers transform data into
information? One method is to adopt distributed decision-making (DDM), which
is a cooperation enabler between interacting decision support systems and aims
to improve the effectiveness of decision-making involving both judgment and
analysis.
In this chapter we consider conceptually how retailers may integrate RFID
and DDM into supply chain processes to increase information visibility and
enhance decision-making. We illustrate these notions via a pilot project under-
taken with a leading UK grocery retailer using value stream mapping (VSM)
(Hines and Rich 1997; Rother and Shook 1999) to investigate the scope for
RFID technologies under a DDM system to reduce cost and add value to the
retailers fresh meat supply chain. This chapter discusses the concepts and issues
behind RFID and DDM to offer a critical discussion of their potential for
integration in food retail supply chains. Those constructs are then explored via a
pilot research project depicting the current and proposed future states of pro-
cesses observed along the retailers fresh meat supply chain to identify where
non-value-adding processes could be eliminated and where RFID-enabled DDM
would add value. From those findings we identify potential time and cost savings
to the retailer and evaluate the benefits of real-time DDM in the context of food
retail supply chain management.
5 Identifying Supply Chain Value 91
equipment, retraining and integration, ongoing tag replacement costs are a sig-
nificant constraint (Smith 2005). For example, using a five cent tag on a one dollar
chocolate bar is not feasible. Retailers are therefore focusing on tagging pallet and
case-loads (for high value items) instead of adopting item-level tagging. Yet in this
context, an RFID dilemma has been uncovered. While pallet-level tagging is
widely demonstrated as feasible, it requires significant investment by both retailers
and suppliers; thereby limiting its adoption (Balocco et al. 2011, p. 310).
Notwithstanding, safety and traceability requirements in the fresh food supply
chain make it imperative that RFID adoption moves toward item level in this
category. There have been calls for RFID to be employed to improve food safety
(Roberti 2009); some retailers already recognize this (Greenwood 2011) but
regulatory authorities are now also listening (Rogers 2011) following catastrophic
events such as bovine spongiform encephalopathy (BSE) outbreaks (Wasserman
2010; Greenwood 2011).
Food quality and safety is of paramount concern to consumers and should
supersede the cost concerns of firms. Market pressures however dictate that retailers
and suppliers must minimize supply chain costs to offer goods at competitive
prices. Supply chain members should therefore seek to add value for themselves
and their customers from RFID in order to warrant the associated costs noted above.
One further opportunity to add value is via the potential it offers for DDM. Fresh
food quality can deteriorate rapidly, thus fresh food supply chain management
requires accurate inventory control to enable quick response decision-making which
will prevent food wastage (Jedermann et al. 2009). RFIDs ability to offer real-time
visibility of inventory at disparate points in a supply chain could be leveraged to
enable DDM that ensures the quality and safety of food throughout the supply chain.
the need for DDM system flexibility and organizational autonomy. In interactive
planning, direct participation is necessary: workers should plan for themselves.
Interactive planning should thus be a continuous process and holistic. Hence, dis-
tributed, decentralized business units are able to simultaneously and independently
plan in a coordinated way via communicating decision support systems to produce
group decisions made at the lowest appropriate level of the organisation.
Communication and coordination are keys to the effectiveness of a DDM
information system. The lack of hierarchy and decentralization implies a lack of
control (Rathwell and Burns 1985). Without control, planning is ineffective.
Control comes from the communication and coordination of the decisions made.
Thus, through the transparency of those decisions comes the conflict resolution
Rathwell and Burns (1985) define as being DDMs core function.
DDM enabled by rapidly advancing information and communication technol-
ogies has had a significant impact on organizational decision-making (Schneeweiss
2003a). Supply chain management (SCM) is viewed as a potential melting pot
where DDM theory and practice can be extended via operational research methods
(Schneeweiss 2003a). Indeed, the need to solve master problems with limited
information sharing is highly beneficial in a supply chain context where competi-
tive pressures, time and distance limit the amount, type and frequency of infor-
mation shared between supply chain partners.
The flexibility and autonomy afforded by DDM is critical to contemporary
quick response SCM, where there is no time to refer decisions up through hier-
archical management structures or across multiple echelons (Butcher 2007b).
Hence, with the potential for RFID to add significant value to SCM the benefits
from coupling RFID with DDM would deliver a more robust business case for
RFID technology adoption.
Key challenges to gaining wider RFID adoption in supply chain management are
identified as including the integration of applications into existing business pro-
cesses, relationships with suppliers and distributors, data sharing challenges, and
the ability to find value in the multiple data captured (Jones et al. 2004). From the
foregoing discussion of DDM benefits it would seem that such operational and
organizational issues might be reconciled using DDM.
As discussed by Dutta et al. (2007), there are a number of generic articles
speculating the benefits of RFID. This is also evident in extant literature relating to
RFID applied to distributed decision-making and decentralized supply chains.
Discussions and models of how RFID might offer a more holistic perspective of
inventory are offered (Lin et al. 2006; Heese 2007). In conventional supply chains,
local inventory decisions are based on the local inventory position (i.e. what stock is
on-hand at any point in the supply chain, plus the quantity on-order, and minus any
back-order quantity) because no more data are available. But, when RFID is
94 T. Butcher and D. B. Grant
employed amounts of data are increased from point-of-sale data, stockholding data
from other locations in the supply chain, and transport lead-time data; all of which
are in real-time (Lin et al. 2006). If analyzed quickly and accurately, such combined
data offer greater accuracy and precision in local management decision-making.
Dutta et al. (2007) point out that for such benefits to be realized, inter-
organizational constraints must be addressed first. The common perception is that
only retailers gain a cost benefit from RFID adoption. Retailers mandating sup-
pliers to slap-and-ship RFID onto consignments does not incentivize technology
adoption. On the other hand cost-sharing might. The problem however is that there
is insufficient research on this critical issue (Dutta et al. 2007).
Schneeweiss (2003b) offers the most insight into how DDM might be employed to
solve such inter-organizational research issues in a supply chain context, but does not
explicitly refer to RFID. Seven categories of DDM are identified based on the level of
information sharing between supply chain partners. These range from constructional
DDMs employed in long-term collaborative relationships to ad hoc market coordi-
nation systems commonly used in adversarial buyersupplier negotiations. As
alluded to by Dutta et al. (2007), how retailers choose to implement RFID can affect
their suppliers commitment to it. Hence there should be alignment between the inter-
organizational relationship, the information system, and the data capture system.
We undertook a pilot research study to identify where RFID might add value to
a fresh food retail supply chain. To do so, it was considered that RFID must
provide an input into a DDM system that fits with the current or future inter-
organizational relationship within that supply chain.
A VSM of the current state of the retailers fresh meat supply chain from the meat
supplier to the store is shown as Appendix A. A number of non-value adding
activities caused by waiting time and human intervention were identified. For
example, the seemingly essential human tasks of physical checking, scanning,
verifying and counting of inventory and trays (i.e. case loads) at each stage of the
supply chain were identified as being non-value adding. RFID automated scanning
would eliminate the line of sight requirement of barcode readers and thus eliminate
these human-oriented tasks. Specific constraints at various critical nodes along this
supply chain are offered next.
The fresh meat supplier in Preston, Lancashire supplies fresh lamb meat to the
retailers chilled distribution centre (CDC) in Wakefield, West Yorkshire, at a dis-
tance of about 50 miles or 80 km, on a daily basis with a frequency of one direct
delivery per day. The price level varies depending on the product assortment, which
could include shoulder, leg, chops and various other joints of meat. To maintain
quality and freshness the meat must be chilled down to at least +2C. All meat is
packed in plastic containers and supplied in reusable plastic trays, which vary in
shape and size. The retailer supplies these reusable trays to the meat supplier from
their Wakefield service centre where a tray washing facility is provided.
The information flow between the retailer and their meat supplier is transmitted via
electronic data interchange (EDI) mailboxes. Initially, generic supply orders are
96 T. Butcher and D. B. Grant
raised a number of weeks before based on forecasts, but final more accurate (small
batch) orders are raised a few days before product is required, based on actual demand.
After order picking and packing at the suppliers site, the supplier prints
barcode labels that hold the order number, delivery date, product type, weight,
quantity, etc., and attaches these labels to the trays to identify products in the trays.
One further label is then issued to identify the whole delivery. The trays are then
stacked on pallets or put on dollies (i.e. carts or trolleys) and readied for dispatch
to the retailers CDC.
At this stage the VSM analysis identified labour-intensive, time consuming and
non-value adding activities in the suppliers operations such as excessive label
scanning, verifying and counting of products. Such duplication of effort and
multiple handling, whilst maintaining quality assurance, has an impact on order
lead-time and cost. Further, after the daily order has been dispatched, the supplier
does not gain any information about that shipment from the retailer. Hence the
supplier can only react to orders, and the daily cycle begins again the following day.
The Wakefield CDC is designed for the storage and distribution of fresh and frozen
products and houses three distinct operational areas: frozen products (-25C),
chilled products which are the concern of this research study (+2C) and ambient
products (+13C). This CDC supplies a number of retail stores through Lancashire,
Greater Manchester and Yorkshire. All deliveries from the CDC to stores are via
direct deliveries by in-house distribution services, with each truck containing a
variety of products destined for each particular store.
Before arrival of a meat shipment at the CDC the warehouse operations staff
waiting on delivery from the supplier can only know what is in the truck trailer
based on the order information they have from their information system. Only after
the trucks arrival and opening of the trailer doors can they correlate that infor-
mation with the actual delivery. They then check what products actually arrived,
the quantity, and they confirm the delivery with the information system. Interviews
with staff found that inventory errors are not common but can still occur. This
obviously leads to non-value adding activities to inform the supplier and make
adjustments to the current (i.e. concessions) and possibly subsequent orders.
Upon arrival at the CDC one tray from each pallet or dolly is barcode scanned.
Other trays are physically counted in order to confirm product delivery, to check
the quantity of products and the accuracy of the delivered order. Each inbound
delivery from the meat supplier contains a variety of products for several buyers
(i.e. store) locations. Being fresh meat products with a short shelf life, they stay at
the CDC no more than 23 h. During this time the received products will be cross
docked with store-specific orders being picked, packed and consolidated with other
products newly arrived or held within the CDC to create outbound shipments to the
stores. After the cross docking, the system generates new barcode labels to
associate products with stores. These new barcode labels are then scanned and the
5 Identifying Supply Chain Value 97
deliveries for each store are checked and counted before loading the trucks and
dispatching to the retail outlets.
The VSM analysis identified labour and time-consuming non-value adding
activities here, such as: duplicate barcode scanning, verifying and counting of
products. This is a by-product of potential inbound quality issues discussed above.
There is no product or tray scanning upon arrival at the store. Instead, visual
counting and confirmation against order are performed. Next, the fresh meat is
placed in the back-of-store chiller until front-of-store shelf replenishment is
required. Replenishment is triggered by visually monitoring product availability at
the shelves. When required, operational staff move full trays of meat from the chiller
to the shop floor to replenish the amount required (not necessarily a full tray-load).
Strict standards of food product freshness are observed. Meat products must not be in
an environment above the maximum chilled temperature of +2C for more than
20 min, thereby limiting the replenishment process time. After shelf replenishment,
the trays are returned to the chiller with the remaining meat that was not used.
At the store, ineffective use of labour resources and time-consuming checking
and counting were identified in the VSM analysis. Replenishment is also dependent
on staff and is thus prone to human error.
When the reusable trays are empty, back-of-store personnel collate them to form
pallet-loads. These are then transported to a central tray washing facility at the retai-
lers service centre where the trays are washed and packed onto a truck for redistri-
bution to suppliers. Trays are dispatched to suppliers based on order information they
provide. If extra trays are required the suppliers will request them. There are currently
no data on asset shrinkage (i.e. lost trays) through this process because there is no
application that controls the movement of trays. It is acknowledged that trays do go
missing. This leads to increased handling costs, increased tray recycling lead time,
excessive tray inventory, shrinkage and attrition of trays, and substitute tray costs.
From our current state VSM in Appendix A, we calculated the total lead time to
be 3 h and 17 min, where the total value adding time is 2 h and 55 min (not
including transport lead times between nodes). The non-value adding activities we
identified include duplicate barcode scanning, counting, checking and verifying at
the each point along the supply chain. Each of these activities is prone to human
error. Key deficiencies that lead to non-value adding activities such as duplicate
inventory counting include a:
Lack of information visibility;
Lack of information sharing between partners;
Lack of inventory and reusable assets visibility between various supply chain nodes.
98 T. Butcher and D. B. Grant
5.4 Conclusions
This chapter has conceptualised the potential for integration of RFID and a DDM
information system into a food retail supply chain to improve product traceability,
process effectiveness, efficiencies and cost savings, which in turn should lead to
the added-value of food quality and safety assurance. A pilot research study
identified non-value adding activities in a retailers fresh meat supply chain. The
issues above surrounding these products are paramount considerations for both
retailers and suppliers, and thus operational effectiveness and efficiency are traded-
off in conventional human-oriented distribution.
And yet, we found through our analyses that RFID technologies offer an
opportunity to remove this conventional trade-off. The removal of human inter-
vention in product scanning, counting, checking and verifying can enable
improved delivery performance whilst maintaining product safety and quality
assurance. Further, the integration of the resultant information into a DDM would
enable informed decisions along the supply chain according to appropriate
hierarchical rules to assist in meeting these objectives.
This pilot research study was undertaken as a precursor for a larger initiative to
implement RFID in this supply chain. That subsequent study will later verify
whether or not the time savings we identified are actually achievable and also
determine whether or not a return-on-investment can be achieved. But regardless
of the results, food safety is a primary concern for consumers. This is why this
retailer is considering RFID for this type of product and why we also are likely to
see an increasing number of retailers applying RFID to such high value, safety
critical products whether or not they can achieve a return-on-investment, thereby
negating the current RFID dilemma (Balocco et al. 2011, p. 310).
However, we further argue that simply having RFID tags providing constant
streams of data is insufficient to make proper management decisions across a
decentralized supply chain such as the one discussed here. Thus, the integration of
an appropriate DDM is critical to addressing the three key issues identified in our
empirical research: the lack of information visibility; the lack of information
sharing between partners; and the lack of inventory and reusable assets visibility
between various supply chain nodes. The ability for all actors in this supply chain
to see what others are doing in real-time will enable informed, instantaneous and
accurate decision-making. Such improved decision-making capabilities would, for
example, lead to improved responsiveness by the supplier to daily orders from the
retailer; greater order accuracy and reconciliation at the CDC; more effective store
shelf replenishment; and higher asset utilization at the tray wash facility. We
believe that such an integrated approach to improving supply chain planning and
control systems will stimulate further RFID, DDM and supply chain research to
make the business case in terms of decision-making value-add rather than process
cost reduction, regardless of RFID tag cost.
Acknowledgments We wish to acknowledge the contribution of Yuriy Sergeyev in undertaking
the fieldwork as part of his Masters thesis at the University of Hull Logistics Institute.
100 T. Butcher and D. B. Grant
References
Balocco, R., Miragliotta, G., Perego, A., & Tumino, A. (2011). RFID adoption in the FMCG
supply chain. Supply Chain Management: An International Journal, 16(5), 299315.
Bansal, R. (2004). Now you see it and now you dont [RFID technology]. IEEE Microwave
Magazine, 5(4), 3234.
Boeck, H., & Wamba, S. F. (2008). RFID and buyer-seller relationships in the retail supply chain.
International Journal of Retail and Distribution Management, 36(6), 433460.
Burns, A., Rathwell, M., & Thomas, R. (1987). A distributed decision-making system. Decision
Support Systems, 3(2), 121131.
Butcher, T. (2007a). Radio frequency identification: an enabler of agile supply chain decision-
making. International Journal of Agile Systems and Management, 2(3), 305320.
Butcher, T. (2007b). Supply chain knowledge work: should we restructure the workforce for
improved agility? International Journal of Agile Systems and Management, 2(4), 376392.
Butcher, T., & Lalwani, C. (2007). RFID benefit beyond ROI: lessons from a not-for-profit
cooperative organisation. In Proceedings of the Annual Conference of the Logistics Research
Network, Hull, UK, 57 Sept 2007.
Chappell, G., Durdan, D., Gilbert, G., Ginsburg, L., Smith, J., & Tobolski, J. (2003). The value of
auto-id technology in retail stores. Auto-ID Centre. http://www.autoidlabs.org/single-view/
dir/article/6/143/page.html. Accessed 10 Sept 2009.
Clarke, D., Gosain, S., & Thillairajahn, V. (2005). Realizing the promise of RFID. http://
www.ebizq.net/topics/rfid/features/6165.html?&pp=1. Accessed 9 Sept 2009.
Dutta, A., Lee, H., & Whang, S. (2007). RFID and operations management: technology, value,
and incentives. Production and Operations Management, 16(5), 646655.
Grant, D. B. (2005). The transactionrelationship dichotomy in logistics and supply chain
management. Supply Chain Forum: An International Journal, 6(2), 3848.
Greenwood, D. (2011). NZ farmers need to embrace big brother. ZDNet.http://www.zdnet.com.au/
nz-farmers-need-to-embrace-big-brother-339314563.htm. Accessed 7 May 2011.
Heese, H. S. (2007). Inventory record inaccuracy, double marginalization, and RFID adoption.
Production and Operations Management, 16(5), 542553.
Hines, P., & Rich, N. (1997). The seven value stream mapping tools. International Journal of
Operations and Production Management, 17(1), 4664.
Hingley, M. (2001). Relationship management in the supply chain. International Journal of
Logistics Management, 12(2), 5771.
Hingley, M., Taylor, S., & Ellis, C. (2007). Radio frequency identification tagging: supplier
attitudes to implementation in the grocery retail sector. International Journal of Retail and
Distribution Management, 35(10), 803820.
Hingley, M., Lindgreen, A., Grant, D.B., & Kane, C. (2011). Using fourth party logistics
management to improve horizontal collaboration among grocery retailers. Supply Chain
Management: An International Journal, 16(5), 316327.
Jedermann, R., Ruiz-Garcia, L., & Lang, W. (2009). Spatial temperature profiling by semi-
passive RFID loggers for perishable food transportation. Computers and Electronics in
Agriculture, 65(2), 145154.
Jones, P., Clarke-Hill, C., Shears, P., Comfort, D., & Hillier, D. (2004). Radio frequency
identification in the UK: opportunities and challenge. International Journal of Retail and
Distribution Management, 32(3), 164171.
Lin, D., Barton, R., Bi, H., & Freimer, M. (2006). Challenges in RFID enabled supply chain
management. Quality Progress, 39(11), 2328.
Mller-Seitz, G., Dautzenberg, K., Creusen, U., & Stromereder, C. (2009). Customer acceptance
of RFID technology: evidence from the German electronic retail sector. Journal of Retailing
and Consumer Services, 16(1), 3139.
Poundarikapuram, S., & Veeramani, D. (2004). Distributed decision-making in supply chains and
private e-marketplaces. Production and Operations Management, 13(1), 111121.
5 Identifying Supply Chain Value 103
Rathwell, M., & Burns, A. (1985). Information systems support for group planning and decision-
making activities. MIS Quarterly, 9(3), 255271.
Regattieri, A., Gamberi, M., & Manzini, R. (2007). Traceability of food products: general
framework and experimental evidence. Journal of Food Engineering, 81(2), 347356.
Roberti, M. (2009). Industry must promote RFIDs role in food safety. RFID Journal. http://
www.rfidjournal.com/article/view/5035. Accessed 7 May 2011.
Rogers, C. (2011). Farmers wary of tracing bill. Business Day.http://www.stuff.co.nz/business/
farming/4948434/Farmers-wary-of-Tracing-Bill. Accessed 7 May 2011.
Rother, M., & Shook, J. (1999). Learning to see: value stream mapping to create value and
eliminate muda. Boston: The Lean Enterprise Institute.
Saunders, A. (2005). Can Smart Tags Deliver? Management Today. http://bit.ly/qh6DfB. Accessed
1 Sept 2011.
Schneeweiss, C. (2003a). Distributed decision-making: a unified approach. European Journal of
Operational Research, 150(2), 237252.
Schneeweiss, C. (2003b). Distributed decision making in supply chain management. Interna-
tional Journal of Production Economics, 84(1), 7183.
Smith, A. (2005). Exploring radio frequency identification technology and its impact on business
systems. Information Management and Computer Security, 13(1), 1628.
Swedberg, C. (2008). At Metros new future store, RFID helps assure meat quality. RFID Journal.
http://www.rfidjournal.com/article/articleprint/4109/-1/1/. Accessed 7 May 2011.
Tajima, M. (2007). Strategic value of RFID in supply chain management. Journal of Purchasing
and Supply Chain Management, 13(4), 261273.
Twist, D. (2005). The impact of radio frequency identification of supply chain facilities. Journal
of Facilities Management, 3(3), 226239.
Vijayaraman, B., & Osyk, A. (2006). An empirical study of RFID implementation in the
warehousing industry. International Journal of Logistics Management, 17(1), 620.
Wasserman, E. (2010). Canadian beef processor deploys RFID for food safety. RFID Journal.
http://www.rfidjournal.com/article/purchase/7331. Accessed 7 May 2011.
Wharton, C., Rieman, J., Lewis, C., & Polson, P. (1994). The cognitive walkthrough method: a
practitioners guide. In J. Nielsen & R. Mack (Eds.), Usability inspection methods. New York:
Wiley.
Chapter 6
Understanding the Acceptance of RFID
in the Healthcare Industry: Extending
the TAM Model
Abstract This research aims to investigate the factors that affect the adoption of
Radio Frequency Identification (RFID) in the healthcare industry. Technology
Acceptance Model (TAM) is extended by including the determinants of perceived
usefulness, perceived ease of use, as well as the security and privacy variables. Data
was collected from 183 healthcare organizations, and was analysed using Structural
Equation Modelling (SEM) analysis. The results showed that subjective norm,
image, job relevance and output quality have direct and positive relationships with
perceived usefulness. Facilitating conditions of the healthcare organizations
have positive and direct relationships with the perceived ease of use of RFID.
Both perceived usefulness and perceived ease of use are found to have significant
relationships with the healthcare organizations intentions to adopt RFID. This
research proposed and empirically examined a new technology adoption model that
explains RFID adoption decisions in the healthcare industry. The research findings
could be used to assist decision makers of healthcare organizations to formulate
strategies to improve their RFID system implementation.
6.1 Introduction
Radio Frequency Identification (RFID) technology has gained the attention of both
researchers and practitioners. RFID is the generic name for technologies that use
radio waves to identify and track objects (Jones et al. 2004). RFID enables an object
to be recognized, identified, tracked, and traced automatically without the need for
human intervention. By implementing RFID system, objects can be automatically
recognized, identified, tracked and traced from factories, shipping, warehousing,
hospitals, pharmacies, intermediaries, and customers (Kumar et al. 2009). By
implementing RFID, firms are able to automate their supply chain processes, and
information can be exchanged without physical or geographical location constraints.
One of the major domains with growing interest in RFID applications is the
healthcare industry. Although the healthcare industry has traditionally been a slow
adopter of information technologies (IT), healthcare administrators are now begin-
ning to realize the important role that IT plays in helping them to achieve their
business and performance goals (Lee and Shim 2007). Hospital administrators are
now aware that IT is able help the growth of healthcare organizations in the current
competitive business environment (Chao et al. 2007). One of the main problems
faced by hospitals and healthcare facilities worldwide is the managing and tracking
of assets (Young 2005). Some of the common problems faced by the healthcare
industry include increase in labour costs due to dedicating individuals to search for
equipment when needed; time lost due to unavailable equipment, and increased
inventory costs due to misplaced items (Young 2005). Poor inventory management
in the healthcare industry also often leads to the problems in tracking the availability
and expiry dates of medicines. Hospitals also need to ensure that certain drugs are
only prescribed by appropriate personnel, and that drugs are not being stolen. Other
serious problems faced by the healthcare industry are highlighted by the World
Health Organization such as medical errors and counterfeit drugs (Chao et al. 2007).
RFID is used by hospitals to fight against drugs counterfeiting, and to ensure that
correct medicine dosage are delivered to right patients (Chao et al. 2007).
Despite the potentials of RFID, the adoption of RFID remains low in the
healthcare industry (Lee and Shim 2007). Several studies have examined the factors
that affect the adoption of RFID. These studies have sought to examine the adoptions
of RFID from the perspectives of organizational, environmental, and technological
characteristics (Kim and Garrison 2010; Lee and Shim 2007; Kuo and Chen 2008)
based on technology adoption models such as the Diffusion of Innovation (Rogers
1995) and Technology Acceptance Model (TAM) (Davis 1989). TAM is one of the
most commonly applied models in the study of technology adoption, and has been
used to examine RFID implementation decisions by researchers such as Hossain and
Prybutok (2008), Chen et al. (2007), and Wang et al. (2010). These studies have
adopted the TAM model, and extended it by incorporating relevant, additional
variables to suit the characteristics of RFID (e.g. costs, security). However, recent
researchers such as Benbasat and Barki (2007) have claimed that more research
needs to be conducted in TAM related studies to produce results that will help
6 Understanding the Acceptance of RFID in the Healthcare Industry 107
Although RFID is not a new technology, it has only gained the attention of
practitioners and researchers recently. RFID gained popularity in its applications
in supply chain management due to its ability to track and identify objects without
human intervention at a relatively low cost. Wal-Mart is one of the pioneers in
applying RFID to their supply chain, and it was reported that the implementation
of RFID has saved Wal-Mart over $8 billion per year due to reductions in labour
costs in scanning items, out-of-stock items and item theft, and at the same time has
improved the efficiency of the supply chain (Curtin et al. 2007). The applications
of RFID in the healthcare industry have also gained momentum in recent years.
Hospitals such as the Massachusetts General Hospital, Hospital of the University
of Pennsylvania, University of Pittsburg Medical Center, Carolina Medical Center,
and Washington General Hospital have all implemented RFID in their hospital
information systems (Mehrjerdi 2010).
There are various applications of RFID in the healthcare industry. One of the
main challenges faced by the healthcare industry is the increase in the illicit trading
of medicines (Pandey 2010). Pharmaceutical companies claimed that they have lost
billions of dollars each year as a result of illicit trade, and human lives are lost due to
wrong medications (Pandey 2010). Through RFID, medicines can be given unique
codes which are stored in the RFID tags. Such practice allows the identification of
fake medicines. RFID also helps healthcare organizations to improve their inventory
management. For example, hospitals are able to track and manage equipment such
as wheelchairs, surgical equipment, bed facilities, availability of drugs, and
expiry dates of drugs. Not only will time be saved from tracking equipment and
medicines, RFID tracking will also improve the safety of instruments. For example,
108 A. Y.-L. Chong and F. T. S. Chan
information such as when the equipment was last maintained and last sterilized can
be recorded, thus reducing the possibility of using unclean instruments or faulty
equipment (Pandey 2010; Carr et al. 2010). Other useful applications of RFID
include tracking and monitoring patients health information such as heart rate and
blood pressure in real time without human intervention. The ability to track and
record information in real time will also help eliminate errors such as providing the
wrong site, wrong procedure, and wrong patient operations (Michaels et al. 2007).
Despite the benefits of RFID, the healthcare industry is still reluctant to adopt it
(Angeles 2005; Lee and Shim 2007; Carr et al. 2010). Although high cost is often
cited as the main barrier to technology adoptions, the price of RFID implemen-
tation has reduced significantly in recent years. Furthermore, the benefits gained
from reduced paper transactions, shorter order cycle time, and the subsequent
inventory reduction resulting from speedy transmission of inventory related
information would help justify the investment in RFID (Carr et al. 2010). Fisher
and Monahan (2008) believe that the reasons for low adoption of RFID can
sometimes be beyond financial constraints. According to them, the barriers to
RFID adoption can include mal-adaptation of the RFID system, as well as the
organizational challenges faced by the healthcare industry. According to Vanany
and Shaharoun (2008), some of the technical barriers for RFID adoption include
the complexity involved in RFID implementation. The tag readability of RFID for
example, is strongly dependent on factors such as the dosage form, angle of
rotation, and read distance. Thus the readability can be affected by insufficient read
range, and existence of multiple tagged objects (Yao et al. 2010). Healthcare
practitioners are also concerned with the security and reliability offered by RFID.
By implementing RFID, hospitals increase the risks involved in inappropriate
collection, intentional misuse, or unauthorized disclosure of patients healthcare
records (Yao et al. 2010). Other concerns include the fear of failure in electronic
medical devices due to the high-powered RFID reader (Yao et al. 2010). Although
past studies have investigated the technical, organizational, and environmental
challenges of implementing RFID (Lee and Shim 2007; Hossain and Prybutok
2008), few have focused on the factors that influence the intention to use RFID
technology in the healthcare organizations, in particular from the perspectives of
the healthcare organizations decision makers (Carr et al. 2010). It is therefore the
aim of this research to empirically investigate the factors that contribute to the
intention to adopt RFID in healthcare organizations.
TAM is one of the most widely applied technology adoption models (Venkatesh
2000). TAM posits that both perceived usefulness and perceived ease of use
will determine users behavioural intention to adopt a technology. Perceived ease of
use in turn, will influence the perceived usefulness of a technology (Davis 1989).
TAM has been used to study various types of technologies such as online shopping
6 Understanding the Acceptance of RFID in the Healthcare Industry 109
(Gefen et al. 2003), 3G (Chong et al. 2010, 2011), mobile commerce (Wei et al.
2009), telemedicine technology (Chau and Hu 2002) and email (Straub et al. 1997).
Despite its popularity, TAM is criticized as being over studied by researchers
such as Benbasat and Barki (2007), Venkatesh et al. (2007) and Venkatesh (2006).
They believe that too often studies have just added various additional variables to
TAM, and that different types of technologies have been studied without actually
making new contributions to the field of TAM study. One way to move forward the
field of technology adoption study however is to investigate the determinants of
perceived usefulness and perceived ease of use. According to Venkatesh et al.
(2007), such study will provide useful guidance to managers, designers and trainers
in terms of how to enhance the adoption of technology by influencing users per-
ceptions. Carr et al. (2010) stated that TAM has not been widely applied in studying
RFID adoption the healthcare industry. Researchers such as Lee and Shim (2007) for
example, proposed an RFID technology adoption model using the technology push
and pull factors. However, the study focused strongly on organizational issues in
implementing RFID, and neglects the perceptions of the decision makers. Carr et al.
(2010)s paper is one of the few studies that has applied TAM to examine RFID
adoption decisions. Carr et al. (2010) has incorporated the suggestion of Venkatesh
et al. (2007) by investigating the determinants of both perceived usefulness and
perceived ease of use. However, the determinants focused on by Carr et al. (2010)
involved the technical characteristics of RFID (e.g. compatibility) and the envi-
ronment that the healthcare organizations operate in (e.g. supplier support). The
model did not include the perceptions of the decision makers, such as the social
influence or image gained from using RFID in their organizations, or whether RFID
is relevant to their job, demonstrability of the results from RFID, and output quality
of RFID. Such variables were included by Venkatesh and Davis (2000) in their
research which formulated and tested the determinants of perceived usefulness.
Their study however did not include the determinants of perceived ease of use. The
determinants of perceived ease of use were investigated by Venkatesh (2000) and
included factors such as the users computer self efficacy, computer anxiety, per-
ceived enjoyment etc. There is currently limited research that integrates both
determinants of perceived usefulness and perceived ease of use in one single study.
Furthermore, many TAM studies are based on individual adoptions. Jeyaraj et al.
(2006) stated that one research direction for IT acceptance studies is to include
individual adoption decision in the study of organizational adoption. Another limi-
tation with Carr et al. (2010)s study on RFID is that they neglected concerns per-
taining to privacy and security issues encountered when implementing RFID.
Security and privacy are often considered to be some of the main barriers against the
successful adoption of RFID in the healthcare industry (Xiao et al. 2006).
Based on the above discussions, this research extended the TAM model by
investigating the determinants of perceived usefulness and perceived ease of use, and
incorporated the security/privacy perceptions of the decision makers of the health-
care organizations. This model will be able to provide a practical guide to the
healthcare industry in terms of formulating strategies to enhance the adoption of
RFID by improving the perceptions of decision makers in the healthcare industry.
110 A. Y.-L. Chong and F. T. S. Chan
The determinants of perceived ease of use will be adopted from the study from
Venkatesh et al. (2007), but within the context of RFID adoption. The theoretical
framework from Venkatesh (2000) included two elements: Anchors and Adjust-
ments. Anchors are the general belief about computers and their usage (in this case
RFID), while adjustments are based on beliefs, that are shaped based on direct
experiences with the specific systems. However, the variables from adjustments as
proposed by Venkatesh (2000) such as perceived enjoyment and objective
usability are not appropriate for this research. RFID is a system that is imple-
mented for reasons explained in the earlier section; therefore the perceived
enjoyment from RFID usage should not be able to influence the perceived ease of
use. Objective usability is also not suitable for this research as most of the decision
makers may not have previous RFID experience (Venkatesh 2000). Instead, this
research focuses only on variables such as computer self-efficacy, facilitating
conditions, and computer anxiety of the decision makers in the healthcare
organizations.
Self-efficacy refers to the users perceptions on whether he or she is able to
perform a task using a computer (Compeau and Higgins 1995). In the context of
this research, self-efficacy is referred to as the judgement of the user whether he
or she is able to use RFID. If a person has high confidence on his or her ability
to use RFID, it can serve as a basis for the persons perceptions of how easy
RFID will be to use (Venkatesh 2000). According to Venkatesh (2000), the
relationship between external control and perceived ease of use has been under
studied. External control is able to influence the perceptions on the ease of use of
RFID in the form of technology and resource facilitating context (Venkatesh
2000; Taylor and Todd 1995). The external control used in this study will be
adopted from Venkatesh (2000) (e.g. facilitating condition), and will focus on
whether there are resources available from the organization to support the staff in
overcoming the difficulties of using RFID, especially in the early stages of RFID
learning and usage (Venkatesh 2000). Computer anxiety is adapted from
Venkatesh (2000) whereby it is the users apprehension or fear, when he or she
is faced with the possibility of using RFID. The study from Venkatesh (2000)
found that computer anxiety can have a negative affective reaction toward
computer usage.
Based on the above discussions, the following hypotheses are formulated:
H6: Self-efficacy will have a positive direct effect on perceived ease of use of
RFID
H7: Facilitating condition will have a positive direct effect on ease of use
of RFID
H8: Computer anxiety will have a negative direct effect on ease of use
of RFID
112 A. Y.-L. Chong and F. T. S. Chan
Perceived usefulness and perceived ease of use are both the original variables
proposed in TAM. Perceived usefulness is defined as the degree to which a
person believes that using a particular system would enhance his or her job per-
formance (Davis 1989). Perceived ease of use is the degree to which a person
believes that using a particular system would be free from effort (Davis 1989).
These variables have been studied by past researchers such Pagani (2004), Wei
et al. (2009), Kuo and Yen (2009) and it is generally accepted that both of these
variables have a significant and positive relationship with users intentions to adopt
computer technologies. Furthermore, studies from Liao et al. (2007) and Kuo and
Yen (2009) also confirmed that perceived usefulness of a system can be
strengthened by its perceived ease of use. Based on the TAM model, this research
hypothesizes that:
H9: Perceived usefulness will have a positive direct effect on RFID adoption
H10: Perceived ease of use will have a positive direct effect on RFID adoption
H11: Perceived ease of use will have a positive direct effect on perceived
usefulness
Although RFID can improve patients safety by tracking their information, there
are also concerns that RFID will increase the potential security and privacy
risks. For example, by allowing information to be transmitted automatically,
users of RFID expose themselves to the risks of eavesdropping, spoofing, and
denial of service attacks (Vajda and Buttyn 2003). Furthermore, information
tracked and stored by RFID such as patients records, customers information,
financial data, inventory and equipment information which may have significant
value to the healthcare organizations competitors are potentially exposed to
unauthorized access (Xiao et al. 2006). Therefore despite the benefits and the
affordability of RFID, security and privacy issues might be a major concern to
the decision makers of the healthcare organizations. Therefore this study
hypothesizes that:
H12: Security and privacy concerns will have a negative direct effect on RFID
adoption
Based on the literature review and hypotheses, the research model for this study
is presented in Fig. 6.1.
6 Understanding the Acceptance of RFID in the Healthcare Industry 113
Subjective Norm
Image
Job Relevance
Perceived Usefulness
Output Quality
Result demon-
Intention to adopt
strability
RFID
Self Efficacy
Perceived Ease of
Facilitating Use Security and Privacy
Condition
Computer Anxiety
6.4 Methodology
A survey instrument was used to test the hypotheses proposed in this research.
Using a survey to collect the data and testing the research model is the most
frequently used methodology in information system research (Lee and Shim 2007).
The survey was reviewed with a hospital manager and a clinic operations manager
in Malaysia to ensure that the wording and format were appropriate for the
healthcare industry. Managers, heads of departments, IT managers, or logistic
mangers of the healthcare companies and hospitals were requested to complete the
survey. Web-based survey was used to collect the data. This method is consistent
with the approach employed by Lee and Shim (2007). The survey was emailed to
1,000 organizations listed in the Malaysian Medical and Health care directory,
Klinik2u.com portal, and The Malaysia Ministry of Health web portal. The data
collection process took more than 2 months. One month after the emails were sent
out, a reminder was sent to companies that did not reply to the initial emails. Out
of the 1,000 surveys emailed out, 183 surveys were collected, yielding a response
rate of 18.3%. As the survey asked for self-reported data, there is a risk that a
common method bias might occur. Harmans Single-factor test was applied to
114 A. Y.-L. Chong and F. T. S. Chan
examine this potential bias, and the results suggested that there was no significant
common method bias in the data set (e.g. single factors did not emerge from
un-rotated factor solution).
Out of the 183 respondents, 3.3% (6/182) were Chief Nursing Officers, 10.4%
(19/182) were Chief Executive Officers, 19.8% (36/182) were Chief Information/
Technology Officers, 25.8% (47/182) were General Managers, and 40.7% (74/182)
were Departmental Heads. Demographically, 59.3% (108/182) respondents were
males and 40.7% (74/182) were females. In terms of education, 4.9% (9/182)
were diploma holders, 76.9% (140/182) were degree holders, and 18.1% (33/182)
were postgraduates.
Items used in this study were adopted from previous studies, but were modified to
include RFID as the technology studied. Item sources are found in Table 6.1.
A total of 41 questions were used to measure the independent and dependent
variables used in this study. Responses to the items were made on a five-point
Likert scale format ranging from 1strongly disagree to 5strongly agree.
A two-stage methodology was carried out whereby the measurement model was
developed and evaluated, followed by the evaluation of the full structural equation
model (Lee and Shim 2007). The first step involved testing the convergent and
discriminant validity of the constructs. Cronbachs alpha was used to test the
internal consistency reliability. The results are shown in Table 6.1.
Based on the Cronbachs alpha test, all the constructs have reliability of more
than 0.70 as suggested by Hair et al. (2006) except for Image, Result Demon-
strability, and Security and Privacy. Upon examining the problematic items in
Image, Result Demonstrability, and Security and Privacy, we found that by
dropping the third item of the Image variable, the third item of Result
Demonstrability will result in improved Cronbachs Alpha value. Thus the new
Cronbachs alpha values for Image, Demonstrability, and Security and Privacy
were improved to 0.710, 0.737 and 0.810 respectively. As such, the overall final
items include only 38 items. Convergent validity was examined in this research by
three criteria: all item loadings are more than 0.70; composite reliability is more
than 0.60; and average variance extraction (AVE) is greater than 0.50. All the
items in this research satisfied the criteria.
Table 6.1 Constructs, items and reliability analysis
Factors Items Cronbachs alpha Source
Subjective norm People who influence our behaviour think that we should use RFID 0.738 Venkatesh and Davis (2000)
People who are important to us think that we should use RFID
The senior management of this business has been helpful in the use of RFID
In general, the organization has supported the use of RFID
Image Other organizations in our industry that use RFID have more prestige than 0.507 Venkatesh and Davis (2000)
those who do not
Other organizations in our industry that use RFID have a high profile
Using RFID is a social status in our organizations social circle (0.737)
Job relevance In my job, usage of RFID is important 0.897 Venkatesh and Davis (2000)
In my job, usage of RFID is relevant
Output quality The quality of the output I get from RFID is high 0.731 Venkatesh and Davis (2000)
I have no problem with the quality of RFIDs output
Result demonstrability I have no difficulty in telling others about the results of using RFID 0.570 Venkatesh and Davis (2000)
I believe I could communicate to others the consequences of using RFID
The results of using RFID are apparent to me
I would have no difficulty in explaining why using RFID may or may not be
beneficial
Self efficacy I could complete a job or task using RFID if there is no one around to tell me 0.717 Venkatesh and Davis (2000)
what to do as I go
I could complete a job or task using the RFID if I could call someone for help
6 Understanding the Acceptance of RFID in the Healthcare Industry
if I get stuck
I could complete a job or task using RFID if I had a lot of time to complete
the job for which RFID was provided
I could complete a job or task using the RFID if I had just the built manual or
internet forums facility for assistance
Facilitating condition I have the resources necessary to use RFID 0.770 Venkatesh and Davis (2000)
I have the knowledge necessary to use RFID
RFID is compatible with other systems I use
115
(continued)
Table 6.1 (continued)
116
Subjective
0.235* 2
R = 0.447
Image 0.129*
Perceived Useful-
Job Relevance 0.332** ness
0.461**
0.137*
Output Quality 2
R = 0.44
0.075
Result demon- Intention to adopt
strability 0.214** RFID
0.049 2
R = 0.235
Self Efficacy -0.013
0.461**
Perceived Ease of
Facilitating 0.364*
Use Security and Pri-
Condition vacy
0.064
Computer Anxi-
ety
The measurement model was estimated using the maximum likelihood method
(Lee and Shim 2007). In order to examine the model fit, v2 test statistics/d, CFI and
RMSEA were examined. The result provided a v2 test statistics/d value of less than
3, while CFI was found to be greater than 0.90, and RMSEA value was less than
0.08. The fit indices of the measurement model were within the accepted thresholds.
6.6 Discussion
using the existing systems in order to confirm the usefulness of RFID. Based on the
factors determining the perceived usefulness of RFID, this research has validated
and confirmed the TAM2 model proposed by Venkatesh and Davis (2000),
although result demonstrability is not found to be a significant variable.
Three variables (i.e. self-efficacy, facilitating condition and anxiety) were
proposed to be the determinants of perceived ease of use of RFID. This is a
modification from Venkatesh (2000)s model to suit the context of RFID adoption.
The result suggests that only facilitating condition is an important determinant of
perceived ease of use. Therefore the decision makers self efficacy and anxiety
perceptions on using RFID will not determine whether he or she will perceive
RFID as being easy or difficult to use. This could be due to the decision makers
belief that RFID will only be used by the employees, and therefore their own
computer skills and perceptions will not influence the organizations decisions to
adopt RFID. The result also suggests that the facilitating condition such as having
the necessary technical resources is an important determinant of perceived ease of
use of RFID.
This research empirically confirmed that the two original variables in TAM
(i.e. perceived ease of use and perceived usefulness) are able to influence the
healthcare organizations decisions to adopt RFID. However, the additional vari-
able of security and privacy is found to have no influence on the RFID adoption
intentions. This could be attributed to the fact that in the presence of the two
original TAM variables, security and privacy concerns will not hinder the adoption
of RFID. Healthcare organizations are more concerned with RFIDs usefulness
and ease of use instead of security and privacy issues.
This research investigated the factors that influence the adoption of RFID in
healthcare organizations. This has several implications. In terms of theoretical
contributions, this research extended previous studies on TAM by investigating the
determinants of perceived usefulness and perceived ease of use as suggested by
Benbasat and Barki (2007). Furthermore, this study also applied TAM in the
context of organizational adoption of RFID. As suggested by Venkatesh and Bala
(2008), one way to move forward TAM studies is to focus on supply chain
technologies, which is achieved by this study. Carr et al. (2010) stated that their
study is one of the first studies to have applied TAM in the study of RFID. Our
research extended their studies by incorporating the variables derived from studies
by Venkatesh (2000) and Venkatesh and Davis (2000).
In terms of practical contributions, this study can help healthcare organizations
to understand the factors that will influence the perceived usefulness and perceived
ease of use of RFID adoptions. Strategies can be formulated from the findings to
improve the successful adoption of RFID in the healthcare organizations.
120 A. Y.-L. Chong and F. T. S. Chan
Acknowledgments The work described in this paper was partially supported by a grant from The
Hong Kong Polytechnic University (Project No. G-YX4D). The authors would like to thank The
Hong Kong Polytechnic University Research Committee for the financial and technical support.
References
Davis, F. D. (1989). Perceived usefulness, perceived ease of use, and user acceptance of
information technology. MIS Quarterly, 13(3), 319340.
Fisher, J. A., & Monahan, T. (2008). Tracking the social dimensions of RFID systems in
hospitals. International Journal of Medical Informatics, 77(3), 176183.
Gefen, D., Karahanna, E., & Straub, D. W. (2003). Trust and TAM in online shopping:
An integrated model. MIS Quarterly, 27(1), 5190.
Hair, J. F., Black, W. C., Babin, B. J., Anderson, R. E., & Tatham, R. L. (2006). Multivariate data
analysis. New Jersey: Pearson Prentice Hall.
Hossain, M. M., & Prybutok, V. R. (2008). Consumer acceptance of RFID technology:
An exploratory study. IEEE Transactions on Engineering Management, 55(2), 316328.
Hsu, C. L., & Lu, H. P. (2004). Why do people play on-line games? An extended TAM with
social influences and flow experience. Information and Management, 41(7), 853868.
Hung, S. Y., Ku, C. Y., & Chang, C. M. (2003). Critical factors of WAP services adoption: an
empirical study. Electronic Commerce Research and Applications, 2(1), 4260.
Jeyaraj, A., Rottman, J. W., & Lacity, M. C. (2006). A review of the predictors, linkages, and
biases in IT innovation adoption research. Journal of Information Technology, 21(1), 123.
Jones, P., Clarke-Hill, C., Shears, P., Comfort, D., & Hillier, D. (2004). Radio frequency
identification in the UK: opportunities and challenges. International Journal of Retail and
Distribution Management, 32(3), 164171.
Kim, S., & Garrison, G. (2010). Understanding users behaviors regarding supply chain
technology: Determinants impacting the adoption and implementation of RFID technology in
South Korea. International Journal of Information Management, 30(5), 388398.
Kumar, S., Swanson, E., & Tran, T. (2009). RFID in the healthcare supply chain: usage and
application. International Journal of Health Care Quality Assurance, 22(1), 6781.
Kuo, C. H., Chen, & H. G. (2008). The critical issues about deploying RFID in healthcare
industry by service perspective. In Proceedings of the Annual Hawaii International
Conference on System Sciences, 710 Jan 2008, pp. 110.
Kuo, Y. F., & Yen, S. N. (2009). Towards an understanding of the behavioral intention to use 3G
mobile value-added services. Computers in Human Behavior, 25(1), 103110.
Lee, C. P., & Shim, J. P. (2007). An exploratory study of radio frequency identification (RFID)
adoption in the healthcare industry. European Journal of Information Systems, 16(6), 712724.
Liao, C. H., Tsou, C. W., & Huang, M. F. (2007). Factors influencing the usage of 3G mobile
services in Taiwan. Online Information Review, 31(6), 759774.
Mehrjerdi, Y. Z. (2010). RFID-enabled healthcare systems: risk-benefit analysis. International
Journal of Pharmaceutical and Healthcare Marketing, 4(3), 282300.
Michaels, R. K., Makary, M. A., Dahab, Y., Frassica, F. J., Heitmiller, E., Rowen, L. C., et al.
(2007). Achieving the national quality forums never events: prevention of wrong site,
wrong procedure, and wrong patient operations. Annals of Surgery, 245(4), 526532.
Pagani, M. (2004). Determinants of adoption of third generation mobile multimedia services.
Journal of Interactive Marketing, 18(3), 4659.
Pandey, K. (2010). RFID applications in healthcare. http://www.buzzle.com/articles/rfid-
applications-in-healthcare.html. Accessed 21 Nov 2011.
Rogers, E. M. (1995). Diffusion of innovations. New York: Free Press.
Straub, D., Keil, M., & Brenner, W. (1997). Testing the technology acceptance model across
cultures: A three country study. Information and Management, 33(1), 111.
Taylor, S., & Todd, P. A. (1995). Understanding information technology usage: A test of
competing models. Information Systems Research, 6(2), 144176.
Vajda, I., & Buttyn, L. (2003). Lightweight authentication protocols for low-cost RFID tags. In
Proceedings of the Second Workshop on Security in Ubiquitous Computing, Washington, 12
Oct 2003, pp. 110.
Vanany, I., & Shaharoun, A.B.M. (2008). Barriers and critical success factors towards RFID
technology adoption in South-East Asian healthcare industry. In Proceedings of the 9th Asia
Pacific Industrial Engineering and Management Systems Conference, Indonesian, 35 Dec
2008, pp. 144155.
122 A. Y.-L. Chong and F. T. S. Chan
Keywords Supply chain networks Sensitive logistics nodes Supply chain risks
Freight villages Business continuity
G. Siestrup (&)
Business Information Systems, Hochschule Furtwangen University,
Institute for Applied Research, Robert-Gerwig-Platz 1, 78120 Furtwangen, Germany
e-mail: guido.siestrup@hs-furtwangen.de
C. Breuer
Business Information Systems, Hochschule Furtwangen University,
Robert-Gerwig-Platz 1, 78120 Furtwangen, Germany
H.-D. Haasis
Logistics Systems, Institute of Shipping Economics and Logistics,
Universitaetsallee 11-13, 28359 Bremen, Germany
e-mail: haasis@isl.org
7.1 Introduction
sources of risks are identified and integrated into the model. In a third step
(3) standardised strategies and measures are proposed to enable affected compa-
nies to maintain and especially recover business performance in the event of
disruptions. This is exemplified by using the business case of freight villages.
There are numerous definitions of the term risk. Amongst others, these definitions
are the result of different fields of research, variations in the specific settings and
problems being addressed (Ritchie and Brindley 2009). Many of these definitions
include aspects like the likelihood of occurrence, causes and consequences of risks
(e.g. Ritchie and Brindley 2009; Winkler and Kaluza 2007; Pfohl et al. 2010). The
likelihood of occurrence describes the relative frequency of risk occurrences
(Winkler and Kaluza 2007), while the causes of risks are the result of incomplete
information of the decision-maker (Kajter 2007). Effects or consequences of risks
are defined as the negative outcome of a particular event. By contrast, an advanced
comprehension of the term risk may also consider a positive deviation from an
expected goal (e.g. Pfohl et al. 2010; Kersten et al. 2008).
Concerning supply chain risks, different opinions are represented in the field of
research. For example, Ritchie and Brindley (2009) and Jttner et al. (2003) follow
the broad conception and define risk as a danger and a chance, while for instance
Wagner and Bode (2007) and Harland et al. (2003) only take the negative
components of risks into account.
In practice, the term risk is mainly associated with a negative deviation of a
target value (e.g. Kersten et al. 2008; Wagner and Bode 2007).
To avoid dwelling on different definitions, in this chapter the definition of
Kersten et al. (2006, p. 5) is used in which supply chain risk is the damage
assessed by its probability of occurrencethat is caused by an event within a
company, within its supply chain or its environment affecting the business pro-
cesses of at least one company in the supply chain negatively.
For targeted identification and acquisition of supply chain risks, it is advisable to
categorise them. In literature, a variety of risk classifications are suggested.
126 C. Breuer et al.
In literature, there is still no consensus about the definition of supply chain risk
management (Ritchie and Brindley 2009). For instance, Kajter (2003) defines
supply chain risk management as a systematic approach to a cross-company
process of risk management. This process is an integral component of collective
planning and control processes, which are comprised of risk identification, risk
assessment and risk analysis, but also management, monitoring and communica-
tion of risks along the supply chain. In contrast to this definition, Kersten
et al. (2008) consider supply chain risk management as an embedded module of
supply chain management. This embedded module contains all strategies and
measures, all knowledge, all institutions, all processes and all technologies on a
7 Operational Risk Issues and Time-Critical Decision-Making 127
Besides the term logistics nodes, so-called logistics centres (Gourdin 2006;
Braunbeck and Mastria 2007), logistics parks (El Amrani 2007; Leung and Forster
2007) and logistics hubs (Gouvernal et al. 2011) are common in this context. The
analysis and comparison of the different concepts results in a heterogeneous image
motivating the development of a preliminary definition. From an organisational
viewpoint, it can be assumed that logistics nodes consist of multiple business
partners, operating their processes within a certain geographical area. Due to the
concentration of logistics activities, the value added by logistics reaches a high
level, compared to other industrial agglomerations. Value is primarily added by
transportation processes, inventory processes and additional services. This shows
that shared infrastructure is used in order to reduce costs and to intensify
cross-company co-operation. From a functional perspective, logistics nodes play
an important role in intermodal freight transportation. They are connected to
different means of transportation (rail, road, waterborne and air) and involved in
short-distance as well as in long-distance traffic. Other important node-internal
business activities concern logistics activities such as warehousing, packaging,
related services and also logistics-intensive production operations.
Following this argumentation, it can be assumed that the logistics nodes under
consideration are system-relevant components of an economys supply chain
infrastructure. Consequently, disruptions can affect entire supply and demand
processes, leading to significant consequences for business and public. Supply
chain risk strategies and measures have to be defined in order to protect the critical
infrastructure against such negative impacts, or at least to reduce the damage to a
minimum in the case of risk occurrence. In this chapter logistics nodes, which are
part of critical infrastructures (BMI 2005), are referred to as sensitive logistics
nodes. This means that external or internal influences might lead to malfunctions
7 Operational Risk Issues and Time-Critical Decision-Making 129
Freight villages are a special kind of industrial agglomerations. The main features
of such freight villages are multimodality, multi-functionality and supra-
regionality (Aberle 2009). Multimodality means that a freight village combines at least
two modes of transport, normally road and rail, but sometimes also waterborne and air.
For this purpose, a terminal of intermodal transport is located in a freight village.
Multi-functionality describes the variety of upstream, downstream and associated
transport activities within a freight village. In addition to this, synergy effects emerge
by the shared use of warehouse facilities and terminals of intermodal transport. Supra-
regionality characterises freight villages as an interface between local and long-dis-
tance traffic (Welp 2010). This is the primary function of freight villages. Another
function is the formation of efficient, multimodal transportation chains (Rall 2008).
Freight villages also can be characterised by high in- and outgoing transport
volumes combined with turnover and warehouse activities. In a freight village,
different transport and logistics companies are united, for instance forwarding
agencies, logistics service providers, logistics intensive trading and industrial
companies as well as different transport carriers, which retain their legal and
economic autonomy (Rall 2008; Kessler et al. 2009).
The interests of all participants are coordinated by a central business and
development company, which mainly plans, establishes and evolves a freight
village, whereas the operational business of a freight village is performed by the
independent companies located on site (Rall 2008).
Both multiple partners of a single supply chain as well as companies belonging
to different supply chains can be located inside of a freight village (Breuer et al.
2011). If external or internal influences lead to malfunctions or damages in such a
freight village, a number of business partners, supply chains and the regional as
well as the national economy may be affected. Hence, freight villages can be
described as sensitive logistics nodes within a supply chain network.
In Germany, there is currently a network of 35 freight villages. For the
approach presented in this chapter, two of these freight villages were analysed.
A basic requirement for the handling of risks is knowledge of potential risks and
their damage impact. Supply chains can be affected by many risks which are not
necessarily known. For this reason risks may be differentiated into identified and
130 C. Breuer et al.
unidentified risks; the sum of these risks is the total risk supply chains are exposed
to (FAA 2000). Identified risks can be subdivided into acceptable and unaccept-
able risks. Whilst acceptable risks are the part of identified risks that are allowed to
persist once controls are exerted, unacceptable risks are not tolerated and must
either be controlled or eliminated. Residual risks are the part of total risk that
persist after supply chain risk management efforts and comprise acceptable and
unidentified risks (FAA 2000).
If risks are known, supply chains can be managed according to different
strategies and measures for the handling of these risks. It is possible to differentiate
between the two basic strategies, risk prevention and risk mitigation. Besides
these, risk sharing and risk acceptance can be pursued within supply chains
(Kersten et al. 2008). Whilst risk prevention describes the abandonment of
business activities fraught with risk (Henke 2009; Gtze and Mikus 2007), the aim
of risk mitigation lies in diminishing the probability of occurrence and respectively
the dimension of damages. If risks are passed to third parties, for example
insurances, the strategy of risk sharing is applied. When risk acceptance is applied,
risks are deliberately accepted and the consequences are endured (Kummer and
Sudy 2007; Kersten et al. 2008).
According to Gtze and Mikus (2007) measures for achieving the intended aims
of these strategies can be subdivided into measures with regard to the causes of
risks and measures concerning the impacts of risks (see also Wagner and Bode
2007). Measures related to the causes of risks pursue the strategies risk prevention
and risk mitigation. They should eliminate the causes of risks and diminish their
probability of occurrence respectively (Hotwagner 2008).
Measures related to the impacts of risks refer to the consequences of an
occurred risk and aim at the mitigation of the damage caused by this risk.
Therefore, these measures are associated with the strategies risk sharing and risk
acceptance (Czaja 2009). In Fig. 7.1 the coherences are shown.
The approach presented in this chapter focuses on the damage a risk could
cause should it occur, for this reason emphasis is given to measures that minimise
the risks impact.
Fig. 7.1 Coherences between strategies, measures and risks. Adapted from Romeike (2005)
plan. Faced with the same situation, Nokias competitor Ericsson was unable to
develop an adequate strategy and as a result Ericsson suffered great losses
(Sodhi and Tang 2009). Sulzer Medica and Opto Speed even had to declare
bankruptcy after a serious disruption (Brunner and Luthiger 2009).
To support the decision-making process in time-critical situations, emergency
concepts must be developed. These concepts should define standard strategies that
are suitable to reduce the impact of a disruption. Such an emergency concept has
to comprise business continuity and disaster recovery plans, which are prepared in
advance and refer to critical business processes. Business continuity plans have to
ensure, that when emergencies occur, alternative procedures can be rapidly
performed, to maintain business continuity (Thies 2008). Recovery plans enable a
company to return to business within an appropriate time-frame (Thies 2008), thus
reducing the impact of disruptions (Sodhi and Tang 2009).
This is shown in Fig. 7.2 which illustrates different recovery paths for an
affected system.
Once a serious disruption has occurred, the considered logistics core processes
will suffer from a significant loss of performance. The ability to return back into
business depends on the intensity of the disruption impact. The curve of Type III
represents a very time-consuming and cost-intensive recovery path. In case of
Type II, the system is able to recover a high level of performance in the medium
term, whereas Type I represents a less serious situation compared to II and III with
only temporary losses. Furthermore, the recovery path can be influenced by the
organisation itself. From a managerial perspective, being prepared for a crisis
through appropriate ORM measurers, offers significant potentials. Thus, a short-
ened recovery phase is enabled by the implementation of business continuity
concepts (Engel 2009) or by the introduction of emergency and crisis management
(Brunner and Luthiger 2009).
132 C. Breuer et al.
Recovery
curve
Type I
Type III
Ratio
(after Type I with Type II with Type III with
event) ORM ORM ORM
measures measures measures
Fig. 7.2 Alternative performance recovery paths after disruption. Adapted from Engel (2009)
and Brunner and Luthiger (2009)
Today, emergency concepts exist in some areas, for example in the case of nuclear
accidents (Geldermann et al. 2008), in the context of ramp-up management (see for
instance Mller-Dauppert 2005) or in relation to IT-systems (e.g. Thies 2008).
and the customers (Kern and Hartung 2008). Capturing the supply chain processes
creates transparency within them, which allows the business process descriptions
to be used as an information source to manage supply chain risks (Faisal
et al. 2006). This enables the identification of supply chain risk sources and allows
a comprehensive risk assessment (Yzglec et al. 2011). Furthermore, bottleneck
and synergy effects can be detected and incorporated into decision-making
(Haasis 2008). Also, business process descriptions can be used as a foundation for
training and for coordination between the business partners.
In the literature, many publications can be found that also reveal the necessity
of business process descriptions for an effective response in emergency situations,
cf. Wagschal and Huth (2005), von Rssing (2005), Engel (2009) and Zsidisin
et al. (2005).
These publications refer either to single companies or to supply chains.
Business process descriptions in relation to substantial logistics nodes do not exist.
Thus, the aim of the approach presented is to develop business process descrip-
tions for sensitive logistics nodes, for which freight villages are used as an example.
A reference model is applicable for the structured capture of the business
processes. This process-driven reference model can be used as a starting point for
decision-making: if a disruption occurs within a freight village, the business
process description serves as a foundation to search for alternative processes, both
within and outside of a freight village.
7.4 Methodology
Due to the circumstance, that all freight villages show similar processes, a process-
driven reference model for sensitive logistics nodes can be used to create an
improved process transparency. This reference model contains the relevant as-is
processes, based on two selected freight villages. These process descriptions can
be adapted, according to the individual conditions of further freight villages.
In the literature, various types of reference models exist (Fettke and Loos
2003).
From a contextual point of view, the supply chain operations reference model
(SCOR model) is a standard in the logistics area and is concerned with the main
supply chain processes: plan, source, make, deliver and return (Bolstorff et al.
2007). However, a direct transfer of the SCOR-logic to freight villages requires
significant changes, due to the consideration of the following specific characteristics
(Altendeitering et al. 2011):
Freight villages are aggregated logistics areas with supply chains that may either
run in parallel or independently from each other. Also, an individual company
can be located within the freight village.
134 C. Breuer et al.
Table 7.1 Comparison of the DiFOR levels to the organisational structure of freight villages
Level DiFOR Freight villages
1 General planning process General logistics process
2 Adapted planning processes Adapted logistics processes
to a company to a freight village
3 Process map for preliminary Process map for special functional
production planning units in the freight village
4 Digital process map Process map for more specific
functions
5 Modular process unit Modular processes
7.5 Findings
The SCOR and DiFOR model have been adapted in order to create a hierarchical
and modular reference model for sensitive logistics nodes, allowing the processes
to be structured (see for example Bolstorff et al. 2007; Engel et al. 2010). The
model is shown in Fig. 7.3.
The developed reference model focuses on the respective freight village,
illustrated through level one to four. Level zero is used to depict the material and
informational flows, both between different freight villages and between a freight
village and external companies. Therefore, it is possible to easily embed the
particular freight villages into the freight village network.
While the SCOR model categorises the processes in phases that reflect the
satisfaction of a customers demand, the reference model presented uses a different
categorisation: Due to the fact that the value added by logistics processes is high
within freight villages, the processes are divided into the categories: logistical core
and logistical support processes. The business process descriptions themselves are
captured by analysis of the two German freight villages in Bremen and Dresden.
The classification of the reference models hierarchical levels takes place
according to organisational objects.
On level one, a freight village is regarded in its entirety. On this level, the core
and support processes concerning cross-company activities are captured and
depicted in a process map. This process map is presented in Fig. 7.4.
On the underlying level, the business partners in a freight village are clustered,
according to their main business activities. The classification of the cluster is
performed by analysing the companies operations taking place within the existing
35 freight villages in Germany, whereas the set of n clusters is constituted by
N = {1,,n}. Examples of important clusters are the terminal of intermodal
transport, the harbour terminal, general cargo logistics service providers and
courier-express-package service providers. The processes, which take place in the
specific cluster, are depicted in a process map. On this level, the modular structure
7 Operational Risk Issues and Time-Critical Decision-Making 137
K1 K2 Kn
2- Cluster C1 C2 Cn Processes
.. within a
S1 S2 Sn
S1 S2 Sn cluster
3- Sub-cluster Sub-
C1.1 C1.2 C1.c1 . Cn.1 . Cn.c n processes
S1.1 S1.2 S1.c1 Sn.1 Sn.c n
within a sub-
cluster
Fig. 7.4 Example of a process map on level one. Adapted from Altendeitering et al. (2011)
Once the clustering has been performed, the relevant processes within a
freight village can be obtained by linking the processes in the clusters and the
sub-clusters. If a disruption occurs, these interlinked processes form the foundation
for the time-critical decision-making.
The following section describes the proposed procedure to be followed should a
serious disruption occur. The connections are explained on the basis of the cluster
terminal of intermodal transport. The documented process flows, including core
and support processes as shown in the Figs. 7.4 and 7.5, lay the groundwork for the
consecutive steps that must be performed should a serious disruption occur.
Regarding the cluster mentioned above, different sources of risks such as terrorist
attacks, accidents, natural disasters are imaginable and can lead to significant
damages such as:
Disruptions affecting the transport infrastructure (e.g. access roads, railroad
network)
Damages concerning the facilities of the terminal (e.g. crane)
Collapse of external resources (e.g. electrical power supply, information
technology networks)
In order to handle such impacts we propose to establish pre-planned measures
in terms of emergency process descriptions as part of disaster recovery plans.
Depending on type and size of the impact, business objects are routed through the
network structure according to the pre-planned path variants: In the case of a
terminal of intermodal transport, this can mean for example that alternative
resources are targeted within the same cluster (e.g. another crane is selected, if
available) or that the objects (here: container-carrying trucks) are redirected to
another freight village (external to the concerned freight village but internal of the
freight village network). Besides this, further parameters, especially transaction
data, like for instance available capacities or physical inventory have to be
retrieved and considered in the situation of decision-making.
The capability to use alternative capacities within the freight village or within
the freight village network offers potentials and improves the robustness of the
concerned supply chain network. However, these benefits can only be utilised if the
system as a whole is prepared for such disruptions by applicable standard measures.
140 C. Breuer et al.
7.6 Conclusion
Currently, the topic of ORM has remained relatively unregarded, both in practice
and in the relevant literature. This is surprising, considering the significant
importance, not only for the protection of the individual enterprise, but above all,
for the protection of all partners within the bonded supply chains. Even if the
specified underlying risks can be classified as rare events, the efforts described in
this chapter appear acceptable due to the enormous amount of potential damage.
As a starting point for the implementation of ORM, significant knowledge of
the processes is important. Therefore the relevant processes have to be analysed
and documented. If a network is composed of nodes with comparable processes,
the benefits from the development of a reference model can be fully utilised.
Regarding the network of freight villages, this prerequisite is fulfilled, as freight
centres also consist of similar clusters and processes.
After identifying and integrating the risks into the process models, standard
measures have to be derived, defining how to cope with the situation in the event
of a disruption. The result is a comprehensive process model of the considered
network, supporting decision-making by providing adequate action plans and by
visualising alternative decision paths for the identified risks.
Several drawbacks are associated with the presented approach, e.g. high
modelling effort, expenses for keeping the models updated and the high degree of
manual engagement in case of urgent need for action. Concerning this point,
ongoing research aims at a system supported implementation of an event-driven
simulation environment, in combination with a multi-agent framework. The goal is
to accelerate the decision-making process and to include aspects of systems
dynamics. Expected additional opportunities from this approach are for the
automation of time-critical decision-making within sensitive logistics nodes and
their associated logistics networks infrastructures.
Acknowledgments The research underlying this chapter results from the project Prepared-
NET, funded by the German Federal Ministry of Education and Research (BMBF) in the course
of the announcement Securing Supply Chains in the context of the program Research for
Civil Security by the Federal Government. It is part of the Federal Governments Hightech-
Strategy (promotional reference: 13N11136).
References
Hotwagner, B. (2008). Supply chain risk management und dessen systematische Umsetzung in
Unternehmen. Wirtschaft und Management, 8, 2339.
Jttner, U. (2005). Supply chain risk management: Understanding the business requirements from
a practitioner perspective. International Journal of Logistics Management, 16(1), 120141.
Jttner, U., Peck, H., & Christopher, M. (2003). Supply chain risk management: outlining an
agenda for future research. International Journal of Logistics: Research and Applications,
6(4), 197210.
Kajter, P. (2003). Instrumente zum Risikomanagement in der supply chain. In W. Stlzle &
A. Otto (Eds.), Supply chain controlling in der PraxisAktuelle Konzepte und Unterneh-
mensbeispiele (pp. 107135). Gabler: Wiesbaden.
Kajter, P. (2007). Risikomanagement in der supply chain: konomische, regulatorische und
konzeptionelle Grundlagen. In R. Vahrenkamp & C. Siepermann (Eds.), Risikomanagement in
supply chains: Gefahren abwehren, Chancen nutzen, Erfolg generieren (pp. 1327). Berlin:
Schmidt.
Kalwait, F., Meyer, R., & Romeike, F. (2008). Risikomanagement in der Unternehmensfhrung.
Weinheim: Wiley-VCH.
Kern, E.-M., & Hartung, T. (2008). Geschftsprozessorientiertes Risikomanagement am Beispiel
der industriellen Produktion. In F. Himpel, B. Kaluza, & J. Wittmann (Eds.), Spektrum des
Produktions- und Innovationsmanagements. Gabler: Wiesbaden.
Kersten, W., Bger, M., Hohrath, P., & Spth, H. (2006). Supply chain risk management:
development of a theoretical and empirical framework. In W. Kersten & T. Blecker (Eds.),
Managing risks in supply chains: How to build reliable collaboration in logistics. Berlin:
Schmidt.
Kersten, W., Hohrath, P., & Winter, M. (2008). Risikomanagement in Wertschpfungsnetzen
Status Quo und aktuelle Herausforderungen. Wirtschaft und Management, 8, 721.
Kessler, S., Krause, S., & Nagel, L. (2009). Nutzungsmglichkeiten der Workbench zur
Untersttzung des Planungsprozesses von Gterverkehrszentren. In P. Buchholz & U. Clausen
(Eds.), Groe Netze der Logistik: Die Ergebnisse des Sonderforschungsbereichs 559
(pp. 271295). Berlin: Springer.
Kummer, S., & Sudy, I. (2007). Management von transport- und Lagerrisiken in supply chains. In
R. Vahrenkamp & C. Siepermann (Eds.), Risikomanagement in supply chains: Gefahren
abwehren, Chancen nutzen, Erfolg generieren (pp. 255272). Berlin: Schmidt.
Laux, H., & Liermann, F. (2005). Grundlagen der organisation: Die Steuerung von Entscheid-
ungen als Grundproblem der Betriebswirtschaftslehre. Berlin: Springer.
Moder, M. (2008). Supply Frhwarnsysteme. Die Identifikation und Analyse von Risiken in
Einkauf und Supply Management. Wiesbaden: Gabler.
Mller-Dauppert, B. (2005). Logistik-outsourcing: Ausschreibung, vergabe, controlling.
Mnchen: Vogel.
Neiger, D., Rotaru, C., & Churilov, L. (2007). Supply chain risk identification with value-focused
process engineering. Journal of Operations Management, 27, 154168.
OMG (2011). Business process model and notation (BPMN): OMG formally released versions of
BPMN. Accessed May 27, 2011, from http://www.omg.org/spec/BPMN/.
Peck, H. (2005). Drivers of supply chain vulnerability: an integrated framework. International
Journal of Physical Distribution and Logistics Management, 35(4), 210232.
Pfohl, H.-C., Khler, H., & Thomas, D. (2010). State of the art in supply chain risk management
research: empirical and conceptual findings and a roadmap for the implementation in practice.
Logistics Research, 2(1), 3344.
Rajamani, D., Sriskandarajah, C., Pickens, T., & Hameed, S. (2006). A Framework for risk
management in supply chains. Working paper. Accessed May 25, 2011, from http://
som.utdallas.edu/centers/c4isn/documents/c4isn-framework-August06.pdf.
Rall, B. (2008). Auerbetriebliche Logistik. In D. Arnold, H. Isermann, A. Kuhn, et al. (Eds.),
Handbuch logistik (pp. 727788). Berlin: Springer.
7 Operational Risk Issues and Time-Critical Decision-Making 143
Ritchie, B., & Brindley, C. (2009). Effective management of supply chains: risks and
performance. In T. Wu & J. Blackhurst (Eds.), Managing supply chain risk and vulnerability:
Tools and methods for supply chain decision makers (pp. 928). London: Springer.
Romeike, F. (2003). Der Prozess des strategischen und operativen Risikomanagements. In
F. Romeike & R. Finke (Eds.), Erfolgsfaktor risiko-management: Chance fr industrie und
handel. Gabler: Wiesbaden.
Romeike, F. (2005). Modernes Risikomanagement: die Markt- Kredit- und operationellen Risiken
zukunftsorientiert steuern. Weinheim: Wiley-VCH.
Sodhi, M. S., & Tang, C. S. (2009). Managing supply chain disruptions via time-based risk
management. In T. Wu & J. Blackhurst (Eds.), Managing supply chain risk and vulnerability
(pp. 2940). London: Springer.
Servatius, H.-G. (2006). Risikomanagement als umfassende Krisenprvention. Information
Management and Consulting, 21(4), 6977.
Tandler, S., & Eig, M. (2011). Supply chain safety managementKonzeption und
Gestaltungsempfehlungen. In R. Bogaschewsky, M. Eig, R. Lasch, & W. Stlzle (Eds.),
Supply management research (pp. 5791). Gabler: Wiesbaden.
Thies, K. H. W. (2008). Management operationaler IT- und Prozessrisiken: Methoden fr eine
Risikobewltigungsstrategie. Berlin: Springer.
Tuncel, G. (2010). How can risks be managed in logistics networks? In A. Brugnoli (Ed.),
Dangerous materials: control, risk prevention and crisis management. Dordrecht: Springer.
van der Aalst, W. M. P. (2011). Process mining: Discovery, conformance and enhancement of
business processes. Berlin: Springer.
Vancoppenolle, G. (2007). What are we planning for? In A. Hiles (Ed.), The definitive handbook
of business continuity management (pp. 325). Chichester: Wiley.
von Rssing, R. (2005). Betriebliche Kontinuitt als Teil eines integrierten Risikomanagements.
RiskNews, 5, 1419.
Wagner, S., & Bode, C. (2007). Empirische Untersuchung von SC-Risiken und SC-Risikoman-
agement in Deutschland. In R. Vahrenkamp & C. Siepermann (Eds.), Risikomanagement in
supply chains: Gefahren abwehren, Chancen nutzen, Erfolg generieren (pp. 5979). Berlin:
Schmidt.
Wagschal, H., & Huth, M. (2005). Alles im Fluss: Geschftsprozessbasierte Notfallplanung.
RiskNews, 5, 6165.
Welp, H. (2010). Perspektiven von Logistikstandorten im nrdlichen WestfalenEntwicklungs-
varianten fr das Gterverkehrszentrum GVZ Rheine. GeKo Aktuell, 1, 156.
Winkler, H., & Kaluza, B. (2007). berlegungen zu einem integrierten supply chain
performance- und Risikomanagement. In R. Vahrenkamp & C. Siepermann (Eds.),
Risikomanagement in supply chains: Gefahren abwehren, Chancen nutzen, Erfolg generieren
(pp. 319335). Berlin: Schmidt.
Yzglec, G., Hellingrath, B., Wagenitz, A., & Klingebiel, K. (2011). Bestimmung von
Lieferzeitabweichungen unter Bercksichtigung von supply chain-Risiken anhand von
Instrumenten zur Risikobewertung aus der Finanzbranche. In R. Bogaschewsky, M. Eig,
R. Lasch, & W. Stlzle (Eds.), Supply management research (pp. 109138). Gabler:
Wiesbaden.
Zalewski, A., Sztandera, P., & Ludzia, M. (2008). Modeling and analyzing disaster recovery
plans as business processes. In M. D. Harrison & M.-A. Sujan (Eds.), SAFECOMP 2008 (pp.
113125). Berlin: Springer.
Zsidisin, G. A., Ragatz, G. L., & Melnyk, S. A. (2005). The dark side of supply chain
management. Supply Chain Management Review, 9(2), 4652.
Chapter 8
Supply Chain Disturbances:
Contextualising the Cost of Risk
and Uncertainty in Outsourcing
8.1 Introduction
The ever-changing global business context that companies compete in today means
that many companies are under pressure to provide the best quality services for the
customer at a lower price and with a short time to market. Along with lower total
system cost, higher quality and better delivery performance, customers now expect a
shorter product life cycle, reduced time to market and an innovative and customized
H. Rogers (&)
Department of Supply Chain Management, Erlangen-Nuremberg University,
Lange Gasse 20, Erlangen-Nuremberg, 90403 Bavaria, Germany
e-mail: helen.rogers@wiso.uni-erlangen.de
K. Pawar C. Braziotis
Nottingham University Business School, Jubilee Campus,
Nottingham NG8 1BB, UK
product (Pawar and Sharifi 1997; Underdown and Talluri 2002). In response,
manufacturing companies are increasing their global presence (Tang 2006; Bartlett
and Ghoshal 1989), establishing international production networks (Dreyer et al.
2009), entering into joint ventures (Ferdows 1997; De Hek and Mukherjee 2011) and
utilising subsidiaries and operational units in emerging markets (Dyer et al. 1998;
Kostova 1999; Taps and Steger-Jensen 2007). In order to ensure long term survival,
many firms in the West feel that they have a choice to make; either increase
productivity of their existing manufacturing operations or outsource to take
advantage of lower labour costs in countries such as India, China, Romania and
the Czech Republic ( Lalwani et al. 2007b; Taps and Steger-Jensen 2007).
The outsourcing dilemma (as well as the preceding make-or-buy decision) in
manufacturing is probably as old as the manufacturing activity itself, and there is a
constant interest in finding innovative ways of managing it. The roots of this debate
can be traced back to Adam Smiths Wealth of Nations initially published in 1776:
If a foreign country can supply us with a commodity cheaper than we ourselves can make
it, better buy it of them with some part of the produce of our own industry, employed in a
way in which we have some advantage. (Smith 1976, p. 457)
It is generally accepted that cost cutting is the primary driver for outsourcing
operations (Franceschini and Galetto 2003; Heizer and Render 2011; McIvor
2008). However, the whole domain of outsourcing and operations transfer is
wrought with uncertainty and risks, especially when it comes to running operations
smoothly in a new context (Blackhurst et al. 2007; Lalwani et al. 2007b; Harland
et al. 2003; Tapiero and Grando 2006). This contextualisation can include
accounting for a range of factors, such as infrastructure (energy, materials,
transport, communications etc.), education, training, local or national regulations
and culture. Here we examine the causes of uncertainty and risks as major supply
chain disturbances associated with outsourcing products to lower cost locations.
More specifically, we highlight the importance of clearly distinguishing between
uncertainty and risks in seeking to contextualise the holistic cost of outsourcing in
the manufacturing SC context in order to capture both types of disturbances; a
perspective that has been often overlooked.
8.2.1 Outsourcing
Almost two centuries after Adam Smiths comments alluding to where to locate
production, assistance with this strategic decision between make and buy was provided
by several authors including Prahalad and Hamel (1990), Fine and Whitney (1996) and
McIvor et al. (1997). As indicated in Fig. 8.1, simple analysis allows a company to better
position where its products stand against their competitors, thus ensuring appropriate
allocation of resources to exploit its strengths and mitigate any weaknesses.
8 Supply Chain Disturbances 147
hi CONTROL MAKE
non investment
lo
lo business criticality hi
non core core
Outsourcing falls within the make vs. buy decision area and can be defined as
the contractual relationship in which a firm (the client) obtains the services/
products it uses to provide for itself from suppliers (the outsourcing providers)
(Lalwani et al. 2007a). Essentially, all outsourcing decisions follow the funda-
mental make-or-buy decision, and they are the result of dynamic competitive
factors within established industrial sectors. The current globalised and highly
competitive environment has led to the evolution of a so-called outsourcing
economy, which is characterized by an increased focus on core organisational
activities and simultaneous leveraging of external resources, skills, knowledge,
capabilities and competences through outsourcing (Htnen and Eriksson 2009).
Outsourcing has been identified among the key sources of supply chain com-
plexity (Harland et al. 2003). Regarding the questions of what activities to out-
source, there have been contributions such as core activity and supplier influence
risk management models (e.g. Lonsdale 1999), outsourcing decision risk assess-
ment frameworks (e.g. Aubert et al. 2005), and the outsourcing evaluation and
management framework (e.g. McIvor 2008). Subsequently, to address the issue of
where to outsource, location decision models (e.g. Graf and Mudambi 2005) and
relevant key factors (e.g. Bunyaratavej et al. 2007; Lo and Liu 2009) have been
proposed. Finally, the question of how to outsource appears to have attracted the
majority of the contributions with relevant implementation frameworks and
models including Vining and Globerman (1999), Fraering and Prasad (1999), Zhu
et al. (2001), Franceschini and Galetto (2003) and Lalwani et al. (2007b).
The benefits of outsourcing have been well established in the literature, with
typical benefits being cost savings, expertise/technology gains, operations
improvements, and the option to focus on core competencies (Htnen and
148 H. Rogers et al.
Eriksson 2009; McIvor 2008; Kremic et al. 2006; Beaumont and Khan 2005). The
most significant outsourcing decision challenges include recognizing the firms
key competence areas, finding suppliers to provide value to the rest of the
operations, and managing the resulting global network of suppliers and partners
(Quinn 2000). Outsourcing also has well-documented disadvantages. The most
common potential competitive disadvantages attributed to outsourcing are:
1. Unrealistic cost savings: Paradoxically, the cost driver for outsourcing may
act as a boomerang for the companies seeking cost advantages. Rising
costs in outsourcing have been documented in early outsourcing literature
(e.g. Lonsdale 1999). Most costs (e.g. transportation costs or labour costs at
the outsourcing provider) and their impact can be assessed at the beginning
of the agreement. Yet, the relevant cost parameters are subject to constant
change. It has been argued that it is difficult to assess and directly compare
in-house and outsourcing costs (Kremic et al. 2006). For instance, oil prices
constantly fluctuate with an increasing overall trend based on geopolitical
factors. Labour costs in the so-called developing countries (mainly China
and India) also increase as the result of the economic development afforded
by outsourcing (Hopkins 2010).
2. Loss of process and/or product control: More specifically, costs can increase as
outsourced operations prove to be difficult to adequately assess and control
(Heizer and Render 2011). This can in turn lead to negative consequences
regarding product image and/or safety. For example, the melamine scandal that
occurred in 2008, whereby contaminated milk was sourced back to a Chinese
supplier illustrates such an example (Asian Food Worker 2009). It illustrated
how outsourcing to an approved supplier, who in turn outsources parts of their
manufacturing can result in loss of quality control.
3. Loss of knowledge: There are also purely strategic potential disadvantages, in
the form of either over-distilling functions essential for effective competition in
the future or instilling future competition. These relate to the processes that are
actually outsourced (i.e. core or non-core) and demand-specific regarding the
long term view of how the companys industry is going to evolve. A well-
known example in the literature concerns the time that IBM distilled (out-
sourced) the wrong aspects (i.e. the operating system), which led to them
losing ground to their competitors. An example of instilling future competition
is that of Intels decision to outsource chip production to AMD. This allowed
AMD to develop its own capability for the design and manufacture of pro-
cessors, hence emerging as Intels main competitor (Kremic et al. 2006; Heizer
and Render 2011).
4. Negative image: The negative impact of outsourcing on a firms image and
subsequently on its employees is also a danger. Outsourcing can have negative
effect on employees morale in the client firm, and/or for customers due to
ethical reasons. For example, when IBM announced it was to lay off approx-
imately 2,000 employees in 2010, this was directly attributed to outsourcing
(Hasson 2010). Further, companies such as Nike keep a very close eye on
8 Supply Chain Disturbances 149
ensuring that their outsourcing providers comply with the companys code of
conduct and further that they avoid unethical practices such as underpaying
employees and enforced overtime (Locke and Romis 2007).
Our world is characterised by turbulences in the form of uncertainties and risks, with
their scale and magnitude in some countries being considerably higher in comparison
to others (Christopher and Lee 2004). This is difficult to measure directly but is
reflected in indices such as the Corruption Perception Index, published by Trans-
parency International (2011) and the Logistics Performance Index, published by The
World Bank (2010). For example, the earthquake and tsunami-related crisis in Japan
in 2011 resulted in severe shortages of parts for Japanese carmakers (The Economist
2011). Uncertainties and risks in SCs are not only the result of external events such as
wars, floods, industrial action and terrorist activities, but also of internal events such
as changes in business rules or strategies. All of these types of challenges constitute a
complex set of factors that companies need to consider and evaluate in their strategic
plans for outsourcing-related activities. Sources of uncertainties and risks constitute
a major area of concern for all companies in the manufacturing sector, and companies
operating in developing economies are especially vulnerable (Lalwani et al. 2007b).
Further support for this viewpoint comes from Min (1994) and Murphy and Daley
(1994), who pointed out, that country-specific issues affect SCs. This is especially the
case in low-cost countries such as India, where one can experience high transpor-
tation costs, infrastructure issues, variable quality and fragmented third party
logistics (Sahay et al. 2006).
Upon closer examination of the outsourcing literature, it becomes apparent that
the terms risks and uncertainties are often used interchangeably or even as syn-
onymous. The ISO/IEC Guide 73 (Risk managementVocabularyGuidelines
for use in standards 2002) adopted by the British Standards Institute (BSI) defines
risk as the combination of an events probability and its consequence (BSI 2002).
In fact, in this document the term risk is associated with negative consequences.
When the consequences are positive, the term risk is substituted by the term
opportunity. Typically, the options for coping with risks fall within the categories
of accepting it, avoiding it (e.g. by not getting involved in the risk situation),
transferring it (e.g. by sharing it with another party) or reducing it (e.g. by taking
action to reduce its likelihood, impact, or both). The latest edition of Guide 73
Risk Management Vocabulary by ISO defined the term risk as the effect of
uncertainty on objectives (ISO 2009, p. 1), which illustrates some degree of
alteration, or perhaps some development, in the definition of what a risk is. Within
the SC context, the term risk is not fully understood (Zsidisin 2003) and SC risk
management is still an emerging topic (Faisal et al. 2006; Kleindorfer and Saad
2005). It has been suggested that inter-organisational networking generates risks
150 H. Rogers et al.
for both large and small and medium enterprises (Finch 2004), and that these are
essentially the result of supply and demand coordination problems, as well as
disruptions in normal activities (Kleindorfer and Saad 2005). Hence, in line with
previous studies, it has been argued that risks are the result of sources of uncer-
tainty (Cucchiella and Gastaldi 2006), i.e. a function of the level of uncertainty
and the impact of an event (Sinha et al. 2004, p. 155).
On the other hand, an event that can cause disturbance, and for which a proba-
bilistic factor can be assigned, can be classified as either certain or uncertain
(BSI 2002). There is considerable debate in the literature about sources of uncer-
tainty and how to contextualise its dimensions. Wilding (1998) described dynamic
behaviour in SCs as similar to the bullwhip amplification effect (Forrester 1961;
Boute et al. 2008). Geary et al. (2002) added control uncertainty to the previously
identified supply, demand and process uncertainties. They also emphasised the
centrality of information flows as a crucial link between converting customer orders
into targets for manufacturing operations, as well as in linking to the supplier of raw
materials. According to Childerhouse and Towill (2002), the inherent complexity of
material flows in extended SCs is the leading indicator of SC uncertainty. They
further argue that process uncertainties can be minimised or eliminated by applying
certain rules, thus achieving a greater degree of simplification and integration of
material flows in the SC. This in turn leads to a reduction in safety stocks in
incumbent companies (Childerhouse and Towill 2002; Naim et al. 2002). For the
purposes of this research, based on Van der Vorst and Beulens (2002), we define
SC uncertainty as the decision making situation relating to three levels, namely the
SC as a whole, its individual members contributions, and the intra- and extra-SC
environment, which is characterised by lack of deterministic information regarding
state of functions, future actions/behaviour, or impacts/reactions within the three
levels.
In the SC context there appears to be evidence that the terms risk and uncertainty
are used interchangeably. For instance, a framework linking sources of uncertainty
proposed by Mason-Jones and Towill (1998) was subsequently adapted and linked
to risk by Christopher and Peck (2004). We believe there is a clear distinction
between the two, and therefore that they should be treated as different. Although
they are both forms of organisational disturbance, risk has a probabilistic factor
associated with it (Spekman and Davis 2004). Based on the contributions of
Mason-Jones and Towill (1998) and Christopher and Peck (2004), we explore the
main SC disturbances, i.e. risks and uncertainties, collectively in the main cate-
gories of supply-side, demand-side, process, control, environment and society.
Supply-side disturbances can result from poorly performing suppliers who are
unable to meet customer requirements. This can be evaluated by examining sup-
plier delivery performance relating to uncertainty (e.g. planned versus actual lead
times) and risks (e.g. supplier quality performance). Outsourcers who use geo-
graphically distant suppliers are particularly vulnerable to this type of disturbance.
An example of supplier quality issues is provided by Rolls Royces involvement in
the Airbus emergency landing of a Qantas aircraft in Singapore in 2010. Part of the
blame was attributed to defective engine parts (fatigue cracking on the engine feed
8 Supply Chain Disturbances 151
international SCs are also sources of risks and/or uncertainty. Location distur-
bances fall primarily within the category of uncertainty and often entail unpre-
dictable transit transport times, leading to hedging in the form of larger inventory
holdings and therefore higher logistics costs (Arvis et al. 2007).
Societal/political disturbances have been reported in culture-related literature.
For instance, Hofstede (1991) developed an Uncertainty Avoidance Index (UAI)
which he defined as the extent to which a culture feels threatened by ambiguous,
uncertain situations and tries to avoid them by establishing more structure. As
mentioned above, a framework linking sources of uncertainty from a supply,
demand, control and process perspective within the context of environmental
uncertainty was proposed by Mason-Jones and Towill (1998) which was subse-
quently adapted and linked to risk by Christopher and Peck (2004). More recently,
the stability of regime (e.g. Middle Eastern uprisings of 2011) is becoming an
important consideration in outsourcing decisions.
The overall cost of global outsourcing and its downstream impact on logistics and
supply chain implications has been a major theme for both practitioners and
researchers since outsourcing joined the business mainstream (see for example
Ellram and Maltz 1995; McIvor 2008). Puga (2002), for instance, showed that
increasing returns on scale and transport costs encourage firms to locate close to
larger markets, i.e. in areas of high market potential. An example of this is Rolls
Royces decision to build a new manufacturing facility in Singapore. However, as
transport costs fall, proximity becomes less important and peripheral regions with
barriers to entry and lower labour costs gain attractiveness as potential outsourcing
destinations. Compatibility of JIT systems and processes has also emerged as an
important issue. Vickery (1989) investigated the practices of US companies in
reconciling JIT demands with those of global sourcing. He found that the key
success factors for achieving global JIT sourcing include development of long
term buyerseller relationships with effective communication links (including
proactive utilisation of time zone differences) and the implementation of more
accurate buyer-side production planning and scheduling systems. More recently,
Lamming (1996) focused on how to best combine the concept of lean supply with
supply chain management requirements. Pawar et al. (2005) claim that a high
percentage of outsourcing relationships end in failure due to factors such as
selecting outsourcing partners based solely on cost, failing to consider a locations
security and/or risk status, inadequate planning and expectation management,
inappropriate communication channels and poor transition management.
A variety of strategies and approaches have been proposed for mitigating
failure and minimising disruptions and fluctuations to enhance the robustness in
supply chains (Tang 2006). Christopher and Lee (2004) argue that it is confi-
dence-related and using techniques such as six sigma helps create resilient supply
8 Supply Chain Disturbances 153
chains. However, Peck (2005) highlights that there is an absence of any wide-
spread understanding of the scope and dynamic nature of the problem of managing
uncertainty in supply chains. She goes on to propose a conceptual model as an
interactive and adaptive system. Whilst issues surrounding internal and external
benchmarking are seen as relevant to outsourcing decisions, the importance of
total cost of ownership is seen as crucial to mitigating downstream disappoint-
ments in outsourcing (Ellram and Maltz 1995; Fraering and Prasad 1999). This
notion of total or holistic cost of outsourcing is further addressed by McIvors
(2008) outsourcing framework, which seeks to integrate the key concepts in the
outsourcing decision making process and their subsequent impact on a companys
supply base, through four stages: defining core activities, evaluating value chain,
total cost analysis and relationship analysis. These approaches to tackle both the
transitional and long term issues raised by outsourcing in the form of an expert
system allowing managers to adopt a systematic and coherent approach
(Humphreys et al. 2002). However in order to increase their relevance and use-
fulness they must give adequate consideration to disturbances in terms of risks and
uncertainties.
Increasing number of companies rely on their suppliers for the timely delivery of
components and sub-assemblies. As such, many firms are vulnerable due to factors
associated with uncertainties such as supplier quality, order lead-time, order batch
154 H. Rogers et al.
80% of companies consider quality to be the most important factor when selecting
an outsourcing partner. This also proved to be a major issue for one of the case
companies investigated. A senior executive from EngCo1 commented that our
company is known for its high quality and engineering capability, however, some
quality issues still persist especially from externally-supplied parts and compo-
nents. Analysis of the companys record show that the parts and components
defective rate within the supply chain in 2009 exceeded the target by 24%. This is
totally unacceptable and has a severe impact on our profitability and above all our
image in the marketplace. Further analysis reveals that EngCo1 invests in excess of
8 million on supplier development programmes to make sure that the suppliers
are up to speed in terms of meeting its stringent quality standards and requirements.
The EngCo1 executive further commented: in our company about 7075% of parts
and components are bought from external suppliers, with little or no collaboration or
resource-sharing; the challenge here is that some suppliers also deal with our
competitors and this poses an information security problem. The implications of
poor collaboration and resource-sharing are delay in parts and components delivery
as well as defects in some parts and components. EngCo1 was unable to achieve an
acceptable and workable level of supplier integration (i.e. with its own facilities,
systems and processes), owing to significant software dissimilarities and underlying
data security issues.
Communication/information flow costs: The nature and extent of SC distur-
bances when dealing with multiple suppliers is demonstrated in an example pro-
vided by a manager from EngCo5. The scenario was that the customer is based in
France, suppliers are based in Germany and EngCo5 (who provides the system
integration) is based in the UK. This geographical dispersion had serious follow-on
implications as described by a manager at EngCo5; so we had a specification
from the French, in French. We translated it in English to give it to the Germans,
who in turn translated it into German to speak amongst themselves and translated
it back into English. When it finally reached the French customer, it did not bear
any relation to what they initially wanted. The whole issue was that nobody in this
entire thing owned English. The Germans were very careful in understanding;
they had translated it correctly from English to German. Then there was the
translation from French to English but nobody managed the English in the middle.
The specification went on a long journey from FrenchEnglishGermanEnglish
French. Nobody managed the English in between. And it was no good at all for the
customer, just a load of rubbish.
Lack of a clear specification and agreement of what is expected from the
supplier can lead to huge financial problems and misunderstandings. A manager
from EngCo2, who sources components from a Japanese supplier, highlights the
potential nature of the problem and implications for cost: Although English is
widespread and spoken in most countries, it is not the primary language, leading to
miscommunication and unnecessary errors. An example is the lack of under-
standing of the invoice sent to our Japanese supplier. No currency was listed but it
should have been in US dollars. This was incorrectly interpreted by the supplier
which led to currency errors and incorrect pricing.
156 H. Rogers et al.
to the customer, so any inventory that is held by the supplier is at their cost. So the
supply chain is encouraged to be as lean as possible.
An important part of running an efficient and responsive SC is the integration of
the various stakeholders. This process is incredibly complex, particularly when
several hundred partners are involved in large projects. A logistics executive from
EngCo2 stated I would not underestimate the sheer effort that goes into keeping
whole programme integrated .the cost of integration can run into several million
pounds.
The objectivity and underlying rationale for outsourcing must always be at the
heart of any SC strategy and constant reviews and analyses are important to
ascertain that potential benefits are fully realised. These reviews also need to
consider the make vs. buy matrix (see Fig. 8.1), otherwise one can lose sight and
direction of outsourcing principles. Any deviations can be costly and can lead to
unforeseen consequences. An executive from EngCo1 commented: make a list
right at the start of what not to outsource and include on this, your core compe-
tencies. Boeing fell victim to this when it outsourced the design, as well as
manufacture of some components for one engine product: Unfortunately for
Boeing, the scale of the change, its ability to manage that process, overly opti-
mistic projections from suppliers and other factors led to huge delays with the new
aircrafts expected launch date and $2 billion in extra costs (Gilmore 2008).
Cost reduction with the least potential for uncertainty and risk remains a major
factor when outsourcing. As stated by a SC Executive at EngCo1: The prime
reason for outsourcing for us is to reduce total cost. Therefore, one needs to ensure
that there are no hidden costs and that there is a contingency strategy for exiting
the supplier. As discussed, there are many sources of disturbance (namely,
uncertainty-related and risk-related) costs throughout the manufacturing SC. The
outsourcing SC disturbance contextualisation framework presented in Fig. 8.2
captures both uncertainty-related and risk-related disturbances and seeks to assist
in assessing the major aspects of uncertainty-related and risk-related costs, prior to
the outsourcing decision being made. In line with Mason-Jones and Towill (1998),
the determinants of sources of uncertainties and risks can be categorised as internal
to the firm (process and control related), external to the firm (demand and supply
related) and external to the SC network (environmental related). By proactively
and systematically addressing these uncertainties the resilience of the SC
increases (Christopher and Peck 2004).
Based on our research evidence, supply disturbance is primarily the function of
supplier quality, e.g. defective rate targets in the form of service level agreements
and relevant results achieved, and alignment and integration with suppliers to
enhance the efficiency and responsiveness of the SC. Demand disturbance is
primarily the outcome of demand volatility (resulting in the bullwhip effect) and,
158 H. Rogers et al.
Supply disturbance
Environmental disturbance
Demand disturbance
Process disturbance
Control disturbance
hence, inaccurate demand forecasts, and can impact the whole SC. Further, control
disturbance is associated with communication/information flows, e.g. specifying
requirements and their interpretation. Process disturbance constitutes primarily the
outcome of delivery issues, e.g. poor schedule and delivery adherence. The above
confirm the causes of SC uncertainty and risks proposed by Mason-Jones and
Towill (1998) and Christopher and Peck (2004) respectively. Moreover, the above
have traditionally constituted areas of research in relation to the areas of SC
uncertainties and/or risks. Nevertheless, the two types of determinants of SC
uncertainty and risks that have emerged as very important in determining the cost
of SC outsourcing environmental and societal/political disturbances. Essentially,
these two types of disturbances are the ones each company has the least power (if
indeed any) to influence.
Studying environmental disturbances, Intellectual Property Rights (IPR) issues
and protection in some sectors (e.g. aerospace, pharmaceuticals, software) is of
prime consideration in the outsourcing decision. Playmobil (German toymaker)
brought back manufacturing from China to Germany owing to IPR concerns
(Simon 2010). Counterfeiting and infringement of IPR within the pharmaceutical
8 Supply Chain Disturbances 159
sector is quite common amongst many companies, especially amongst firms within
emerging economies. The counterfeit industry is estimated to be around $19
billion and the use of counterfeit medicines can result in side effects, allergic
reactions (which will result in treatment failure) or even death (World Health
Organization 2010). Similarly, educational issues (the level of education and
language ability of the workforce) in an international partners county can increase
the levels of SC disturbances experienced, which in turn can negate apparent cost
savings. A further subset of environmental disturbances is bureaucracy issues.
An important factor to consider here is that information transparency can vary
widely according to which country one is dealing with. For example, outcomes and
timescales when dealing with government officials and processes in India can be
very variable and rather unpredictable. Furthermore, political issues (such as
the stability of the regime in the outsourcing country) are a disturbance factor, as is
the consideration of whether one is dealing with state owned versus private firms
(this is especially important in China). Finally, infrastructure issues fall into the
category of environmental disturbances. Huge differences exist in terms of infra-
structure capabilities across emerging markets. In India this is still a big challenge,
whereas China now has an increasingly sophisticated transport system in place.
This in turn directly affects the logistics performance.
Concerning societal disturbances, cultural issues (i.e. the mindset of people in
different cultures) can create unexpected SC disturbances. For example, what is
written in a contract can be interpreted in many different ways, according to the
persons cultural background. Even something as simple as timekeeping expec-
tations between Europeans and Indians or Brazilians can cause disruptions.
In China and India, many of our managers have come to realise that yes does not
always mean yes (EngCo3 executive). Similarly, the way that decisions are made
can vary from country to country, resulting in relationship management issues.
People-focused societies, such as India, value relationships more highly than
processes. In China a special form of relationship (guanxi) is of utmost importance
(Hofstede 1991). Loyalty issues are also classified as societal type of SC distur-
bance. As an example, Volkswagen brought back manufacturing from the lower
cost location of the Czech Republic to Germany, following serious problems with
lack of employee loyalty. The company felt that the local employees were using
VW as a training camp and that once they were qualified they switched jobs for
as little as a 50 Euro annual pay rise. A VW executive commented: They had no
sense of loyalty. Theyd leave overnight. I pointed out theyd signed a contract but
they didnt care (Woodhead 2007). Tata Automotive in India also had to
re-consider their entire recruitment practices for trainee engineers. The company
found that the attrition rate amongst their engineers was extremely high with more
than 50% leaving the company within 6 months of successful completion of their
training period. The senior management of the company realised that the attrac-
tiveness and market value of these engineers had been enhanced considerably
following their Tata training. As with VW, the Managing Director felt that the
company had invested heavily to train these engineers and was perplexed by their
lack of company loyalty. Finally, time zones/holiday season issues are societal
160 H. Rogers et al.
factors that may cause SC disturbances. Religious celebrations and time zone
differences cause disruptions in supply and need to be fully accounted for when
assessing lead times, deliveries and meeting times (Christmas, Eid, Chinese New
Year, Diwali, etc.). More recently, the stability of regime (e.g. Middle Eastern
uprisings of 2011) is becoming an important consideration in outsourcing
decisions.
The effects of all of the above-mentioned sources of SC disturbances together
form the focal construct of assessing the holistic cost of supply chain outsourcing,
which in turn determines the likelihood of long term competitiveness. The next
logical step is to provide guidelines for performance measures and practical
implementation methods.
8.6 Conclusion
In order to ensure that outsourcing relationships are successful in the long term and
for cost-benefit predictions to come to fruition, firms must be realistic about the
scale and magnitude of disturbances inherent across the supply chain and factor
this into their outsourcing decision calculations. Our research has highlighted that
cost associated with SC disturbances, namely uncertainties and risks, are often
underestimated and that the reasons for these disturbances can be difficult to
predict. Our research evidence suggests that we need to consider issues and types
of disturbances that include, yet also move beyond the frameworks proposed by
Mason-Jones and Towill (1998) and Christopher and Peck (2004). This is borne
out by the fact that in a globalised world, many of these disturbances are related to
the environment or the specific society, meaning they are outside of the SC net-
work influencing mechanisms (and hence beyond the firms direct control). The
outsourcing SC disturbance assessment framework that we propose seeks to assist
in anticipating and hence managing these risks and uncertainties throughout the
SC. By categorising potential hot spots for disturbances and building these into
cost-benefit projections, a better estimate of SC outsourcing costs will be achieved.
Our research has highlighted that in order to assess the holistic cost of SC out-
sourcing, companies should seek to assign costs to the individual elements of the
framework i.e. how can costs be assigned to environmental-related disturbances?
What kinds of costs would these be? Similarly, how can costs be assigned to
demand, supply, process and control-related disturbances? What kinds of costs
would these be? It is only by increasing the transparency of the disturbance factors
and understanding what lies beneath, that realistic forecasts of cost projections can
be made.
As the framework was developed with companies in the manufacturing sector,
additional testing (in particular empirically-based comparative studies) in other
sectors is desirable to enhance its validity. Furthermore, more specific disturbance
assessment criteria need to be developed and tested. One avenue for achieving this
in a non-commercially sensitive manner is to form a Community of Practice
8 Supply Chain Disturbances 161
(Wenger et al. 2002), with a professional supply chain association such as the
Chartered Institute of Purchasing and Supply. Despite these limitations, this
framework paves the way towards developing practical and realistic expectations
for managing SC disturbances, with a view to making the right outsourcing
decisions and ensuring long term competitiveness.
References
Ahmed, M. (2010). Apple delays global release of iPad. The Times. http://business.timesonline.
co.uk/tol/business/industry_sectors/technology/article7097310.ece. Accessed 10 Mar 2011.
Arvis, J. F., Mustra, M. A., Panzer, J., & Ojala, L. (2007). Connecting to compete: Trade logistics
in the global economy. Washington: World Bank.
Asian food worker (2009). Melamine contamination of the global food chain. http://
asianfoodworker.net/?p=440. Accessed 15 Dec 2010.
Aubert, B. A., Patry, M., & Rivard, S. (2005). A framework for information technology
outsourcing risk management. The DATA BASE for Advances in Information Systems, 36(4),
928.
Bartlett, C., & Ghoshal, S. (1989). Managing across borders: The transnational solution. Boston:
Harvard Business School Press.
Beaumont, N., Khan, Z. (2005). A taxonomy of refereed outsourcing literature. Monash
University Business and Economics Working Paper, 22/05.
Bhatnagar, R., & Sohal, A. (2005). Supply chain competitiveness: Measuring the impact of
location factors, uncertainty and manufacturing practices. Technovation, 25(5), 443456.
Blackhurst, J., Wu, T., & OGrady, P. (2007). A network-based decision tool to model
uncertainty in supply chain operations. Production Planning and Control, 18(6), 526535.
Boute, R., Disney, S., Lambrecht, M., & Van Houdt, B. (2008). A win-win solution for the
bullwhip problem. Production Planning and Control, 19(7), 708711.
BSI (2002). Guide 73: 2002 risk managementvocabularyguidelines for use in standards.
British Standards Institution (BSI).
Bunyaratavej, K., Hahn, E. D., & Doh, J. P. (2007). International offshoring of services: A parity
study. Journal of International Management, 13(1), 721.
Childerhouse, P., & Towill, D. R. (2002). Analysis of the factors affecting real-world value
stream performance. International Journal of Production Research, 40(15), 34993518.
Christopher, M., & Lee, H. (2004). Mitigating supply chain risk through improved confidence.
International Journal of Physical Distribution and Logistics Management, 34(5), 388396.
Christopher, M., & Peck, H. (2004). Building the resilient supply chain. International Journal of
Logistics Management, 15(2), 114.
Cucchiella, F., & Gastaldi, M. (2006). Risk management in supply chain: A real option approach.
Journal of Manufacturing Technology Management, 17(6), 700720.
De Hek, P. A., & Mukherjee, A. (2011). Joint venture buy-outs under uncertainty. The Journal of
Industrial Economics, 59(1), 155176.
Dreyer, H. C., Alfnes, E., Strandhagen, J. O., & Thomassen, M. K. (2009). Global supply chain
control systems: A conceptual framework for the global control centre. Production Planning
and Control, 20(2), 147157.
Dyer, J. H., Cho, D. S., & Chu, W. C. (1998). Strategic supplier segmentation: The next best
practice in supply chain management. California Management Review, 40(2), 5777.
Ellram, L., & Maltz, A. (1995). The use of total cost of ownership concepts to model the
outsourcing decision. International Journal of Logistics Management, 6(2), 5566.
Faisal, M. N., Banwet, D. K., & Shankar, R. (2006). Supply chain risk mitigation: Modeling the
enablers. Business Process Management Journal, 12(4), 535552.
162 H. Rogers et al.
Ferdows, K. (1997). Made in the world: Global spread of production. Production and Operations
Management, 6(2), 102111.
Finch, P. (2004). Supply chain risk management. Supply Chain Management: An International
Journal, 9(2), 183196.
Fine, C.H., Whitney, D.E. (1996). Is the make-buy decision process a core competence? Working
paper Massachusetts Institute of Technology.
Forrester, J. W. (1961). Industrial dynamics. Cambridge: MIT Press.
Fraering, M., & Prasad, S. (1999). International sourcing and logistics: An integrated model.
Logistics Information Management, 12(6), 451460.
Franceschini, F., & Galetto, M. (2003). Outsourcing: Guidelines for a structured approach.
Benchmarking: An International Journal, 10(3), 246260.
Geary, S., Childerhouse, P., & Towill, D. R. (2002). Uncertainty and the seamless supply chain.
Supply Chain Management Review, 6(4), 5261.
Ghauri, P., & Grnhaug, K. (2010). Research methods in business studies (4th ed.). London: FT
Prentice Hall.
Gilmore D (2008) Supply chain comment: Chryslers prototype electric car and outsourcing
product design. http://www.scdigest.com. Accessed 5 Dec 2010.
Graf, M., & Mudambi, S. M. (2005). The outsourcing of IT-enabled business processes:
A conceptual model of the location decision. Journal of International Management, 11(2),
253268.
Harland, C., Brenchley, R., & Walker, H. (2003). Risk in supply networks. Journal of Purchasing
and Supply Management, 9(2), 5162.
Hasson I (2010) Latest IBM Layoffs Blamed on Outsourcing. http://www.fiercecio.com/story/
latest-ibm-layoffs-blamed-outsourcing/2010-03-03. Accessed 10 April 2011.
Htnen, J., & Eriksson, T. (2009). 30 ? years of research and practice of outsourcing
exploring the past and anticipating the future. Journal of International Management, 15(2),
142155.
Heizer, J., & Render, B. (2011). Operations Management (10th ed.). New Jersey: Pearson
Education.
Hofstede, G. (1991). Cultures and organizations: Software of the mind. London: McGraw-Hill.
Hopkins, M. S. (2010). Your next supply chain. MIT Sloan Management Review, 51(2), 1724.
Humphreys, P., Mclvor, R., & Huang, G. (2002). An expert system for evaluating the make or
buy decision. Computers and Industrial Engineering, 42(24), 567585.
ISO (2009) GUIDE 73: 2009(E/F) Risk managementVocabulary. International Organization
for Standardization (ISO).
Kleindorfer, P. R., & Saad, G. H. (2005). Managing disruption risks in supply chains. Production
and Operations Management, 14(1), 5368.
Kostova, T. (1999). Transnational transfer of strategic organizational practices: A contextual
perspective. Academy of Management Review, 24(2), 308324.
Kremic, T., Tukel, O. I., & Rom, W. O. (2006). Outsourcing decision support: A survey of
benefits, risks, and decision factors. Supply Chain Management: An International Journal,
11(6), 467482.
Lalwani, C. S., Pawar, K. S., & Shah, J. (2007a). Contextualisation framework for the
manufacturing supply chain. Nottingham: University of Nottingham Business School.
Lalwani, C. S., Pawar, K. S., & Shah, J. (2007b). Contextualisation of supply chain networks.
International Journal of Production Economics, 106(1), 1.
Lamming, R. (1996). Squaring lean supply with supply chain management. International Journal
of Operations and Production Management, 16(2), 183196.
Lee, H. L. (2002). Aligning supply chain strategies with product uncertainties. California
Management Review, 44(3), 105119.
Lo, C. P., & Liu, B. J. (2009). Why India is mainly engaged in offshore service activities, while China
is disproportionately engaged in manufacturing. China Economic Review, 20(2), 236245.
Locke, R., & Romis, M. (2007). Improving work conditions in a global supply chain. MIT Sloan
Management Review, 48(2), 5462.
8 Supply Chain Disturbances 163
The World Bank (2010). International logistics performance index: Ranking. http://www1.world
bank.org/PREM/LPI/tradesurvey/mode1b.asp#ranking. Accessed 25 Mar 2011.
Transparency International (2011). Transparency international corruption perceptions index.
http://www.transparency.org/policy_research/surveys_indices/cpi. Accessed 15 May 2011.
Underdown, R., & Talluri, S. (2002). Cycle of success: A strategy for becoming agile through
benchmarking. Benchmarking: An International Journal, 9(3), 278292.
Van der Vorst, J. G. A. J., & Beulens, A. J. M. (2002). Identifying sources of uncertainty to
generate supply chain redesign strategies. International Journal of Physical Distribution and
Logistics Management, 32(6), 409430.
Vickery, S. K. (1989). International sourcing: Implications for just-in-time manufacturing.
Production and Inventory Management Journal, 30(3), 6672.
Vining, A., & Globerman, S. (1999). A conceptual framework to understand the outsourcing
decision. European Management Journal, 17(6), 645654.
Wenger, E. C., McDermott, R. & Snyder, W.M., 2002. Cultivating communities of practice:
A practical guide to managing knowledge. Boston: Harvard University Press.
Wilding, R. (1998). The supply chain complexity triangle: Uncertainty generation in the supply
chain. International Journal of Physical Distribution and Logistics Management, 28(8),
599616.
Woodhead, M. (2007). German firms retreat from the Eastern front. The Sunday Times, 25 Nov.
World Health Organization (2010). Medicines: Counterfeit medicines. http://www.who.int/
mediacentre/factsheets/fs275/en. Accessed 25 Nov 2010.
Yamamura N (2009). Exploring manufacturing industry outsourcing relationships from a customer
service perspective. Unpublished dissertation, University Business School, Nottingham.
Zhu, Z., Hsu, K., & Lillie, J. (2001). Outsourcinga strategic move: The process and ingredient
for success. Management Decision, 39(5), 373378.
Zsidisin, G. (2003). A grounded definition of supply risk. Journal of Purchasing and Supply
Management, 9(56), 217224.
Chapter 9
Product Containment Resources
Facilitating Decision-Making in Complex
Supply Networks: A Case Study of Milk
Distribution from Farm to Retail
Per Engelseth
Abstract This study seeks to shed light on the issue of supply chain integration
through highlighting interactions between decision makers with relation to the
flow of goods they manage and are responsible for operating. This approach
combines (1) focus on supply networks inherent complexity, (2) supply opera-
tions focusing on product containment facilities, and (3) Aldersons (Dynamic
marketing behavior. A functionalist theory of marketing, Richard D. Irwin, 1965)
transvection view of product supply to develop a model regarding how deci-
sion-making may be viewed to support product supply. Through combining these
views of product supply a case narrative of milk product supply from farm to retail
provides the basis for modelling this supply network as decision-making events;
interlinked with transformed products, information and knowledge with the
ultimate purpose of value realisation from an end-user perspective. As products
flow towards the end-user facilitating value creation, decision-makers are depen-
dent on knowledge transformation within firms which is dependent on information
transformation between firms as well as within firms to secure value realization.
P. Engelseth (&)
Department of Economics, Social Science and Informatics,
Molde University College, Molde, Norway
e-mail: peen@himolde.no
9.1 Introduction
Fig. 9.1 Supply network integration in relation to decision-making (adapted from Engelseth
2012)
9.2 Discussion
This section provides a brief summary of four components that together influence
the construction of the case narrative and the following analysis and conclusion.
The parts of this section are: (1) how supply network complexity calls for an
adapted approach to studying decision-making in relation to product supply, (2)
what product containment constitutes and how containment resources support the
proposed multiplex aspects of connectivity in supply networks, (3) Aldersons
(1965) transvection approach and how this approach may be adapted to model
product supply from the perspective of product containment facilities, and (4)
methodological considerations concerning the chosen case study research strategy
and for constructing the case narrative.
degree to which these elements are differentiated. Sterman (2001) also discerns
between combinatorial complexity and dynamic complexity. In line with this
thinking, Pathak et al. (2007) discern between complicatedness and complexity.
While complicatedness refers to an intricate collection of parts, the relatedness of
the parts is fixed and well-defined. This is the case in a machine viewed as a
system; a fixed structure that facilitates in the case of a supply network, value-
generating transformations. Complexity in a supply network includes humans
interacting with both machines and the environment that consists of society and
nature; in complexity interrelatedness of supply network components is unclear
and non-fixed as are the borderlines if it is described as a system. This complexity-
oriented view evokes perceptions of an unpredictable supply network substance
matter. Taking complexity into close consideration impacts on how supply chain
management (SCM) may be perceived and used to manage the logistics of product
supply. According to Surana et al. (2005), managers must appropriately bal-
ance how much to control and how much to let emerge. This balancing is local
and event-specific, rendering detailed control based on holistic presuppositions
theoretically irrelevant. Product value is a complex and adaptive construct in a
supply network. Creating value through product supply functions involves prod-
ucts undergoing transformations in transcendent production and logistics contexts.
It is especially evident when attempting to manage a complete supply network in
line with fundamental SCM objectives that features of complexity are intrinsically
evoked. The complete supply network viewed as a system emerges as obviously
unmanageable due to its inherent complexity. Surana et al. (2005) points to that
a major obstacle remains in the deployment of coordination and decision
technologies to achieve complex, adaptive and flexible collective behaviour in the
network. Reality as perceived by decision-makers in supply networks is local.
From a more distant perspective the context of actors in supply networks is
complex and may even seem messy if the aim is to model and thereby optimise a
hopefully simple flow of products. For instance, Pathak et al. (2007) state that
decision-making in supply network context is based on noncomplex assump-
tions (e.g. linearity, a buyersupplier dyad, sparse connectivity, static environ-
ment, fixed and non-adaptive firm behaviour), problems are hidden, leaving plenty
of room for understanding and improving the underlying processes.
efficiencies related e.g. to new product development cycles, order fulfilment lead
times, understanding changing market demands, and developing supply opera-
tions. Connectivity in a supply network context is pictured in literature as a
managerial supply network integration challenge (e.g. Closs et al. 2005; Fawcett
et al. 2007). Fawcett et al. (2007) point to the importance of actor willingness in
achieving quality information sharing in supply chains. Closs et al. apply the term
information connectivity in line with this strategic understanding as an aspect of
the information flow supporting supply chain integration. Connectivity contributes
accordingly to supply network transparency or visibility.
Francis (2008) discusses to how supply chain visibility, involving the
identity, location and status of entities transiting the supply chain, captured in
timely messages about events, along with the planned and actual dates/times for
these events, is dependent on interlinking managerial principles with technical
processes. Sherer (2005) points to how development in information technology
(IT) is still unmatched with organisational developments within SCM; discussing
how IT supports networked and dynamic information flows, while SCM still hold
metrics based on a rather static and linear view of product supply processes. Sezen
(2008) reveals limitations of information sharing and integration in developing
supply chain performance pointing out that more overall supply chain design
issues have the strongest impact. This calls for developing SCM principles when
viewing supply chain connectivity issues. Here it is proposed to view product
containment facilities as the unit of analysis to discover how this resource inter-
links decision-making with value generation through multiple interacting product
flows.
where significant others are actors involved in its more important business rela-
tionships. These significant business actors linked together in a supply network
providing a focal firm with a perception of its immediate supply network context
as well as perceptions of purpose derived from interactions with its significant
others. In addition, applying Meads (1934) framework, these perceptions are
embedded in a form of network culture; the generalized other. An interactionist
view is applied to develop the transvection through modelling the supply network.
This involves taking into account multiple mental models or pictures (Miller
and Berger 2001), to create a knowledge basis for product supply development
(Petrie 1992; Andersen 2000; Vernadat 1996, 2004).
Lockamy and McCormack (2004) provide a model that provides a classification
and description of interdependence between supplier and customer relationships,
and how these relationships may be subject to different degrees of integration.
Skilton and Robinson (2009) model interrelationship between traceability
(intrinsically involving the complete chain), transparency, supply network degree
of tight coupling and supply network complexity. These relatively static models,
however, do not depict in further detail how to in-practice develop supply network
integration; involving the two levels, intra-organisational integration and coupling
functionally integrated firms with other firms to achieve seamless information
flows and coordinated transformations through operations in product flows. In a
study by Frohlich and Westbrook (2001), the metrics of supply network integration
were classified as consisting of (1) access to planning systems, (2) sharing
production plans, (3) joint EDI access/networks, (4) knowledge of inventory mix/
levels, (5) package customization, (6) delivery frequencies, (7) common logistical
equipment/containers, and (8) common use of third party logistics. Choi and
Dooley (2009) propose modelling supply networks composed of three alternative
perspectives: (1) focusing on the visible portion, (2) focusing on the invisible
portion, and (3) focusing on the interface between the visible and invisible
portions. Following Choi and Dooley (2009), this provides guidance regarding the
use of future models in relation to managerial decision-making regarding what
may be controlled, and what cannot be controlled. Frohlich and Westbrook (2001)
classification provides guidance as to the components involved in developing
supply network integration. It is here assumed that knowledge concerning details
of a system or network provides foundation for modelling and thereby developing
logistics processes.
Within SCM literature Lambert and Cooper (2000) point to the early contribu-
tions of Alderson (1965) and Bucklin (1966), how these marketing channels scholars
conceptualized the key factors for why and how channels are created and
structures. From a supply chain standpoint, these researchers were on the right track
in the areas of (1) identifying who should be a member of the marketing channel, (2)
describing the need for channel coordination, and (3) drawing actual marketing
channels. Alderson (1965) pictures product supply as a complete end-to-end flow
of products in what he labelled as a transvection; an end-to-end piecemeal process
in a marketing channel where product transformations are directed by intermittent
decision-making events (termed sorts) as described in Fig. 9.2.
9 Product Containment Resources Facilitating Decision-Making 173
sort sort
Initiating provision as End-user
conglomerate resources sort consumption
sort
Fig. 9.2 The transvection model showing product flows in a marketing channel. Arrows indicate
product transformation
Data was collected in line with Sterns et al. (1998) approach adapted to food
distribution; a phenomenological approach implying successive use of theory and
field work to develop a conceptual framework. The empirical foundation of this
study is a case study of fresh food products in a complete supply chain (Engelseth
2007). From this study, milk supply from farm to retailer has been chosen to
illustrate the role of containment facilities as information resources supporting
decision making in the scope of a complete supply network. Data in the original
study was collected using a semi-structured interview technique by listing the
topics rather than questions for each interview. This initial research project where
this data was sourced from consisted of 63 interviews with members of four
different fresh food products (strawberries, milk, fish and bananas) with focus on
package use to support the information flow through a complete supply chain.
Approximately 25 of these interviews provide relatively direct foundation for the
constructing the following case narrative adapted to the provided analytical
framework. These used interviews can be grouped in five categories: (1) different
actors directly involved in the milk product supply chain, from farm, through dairy
to retailer, (2) supporting actors in the focal chain such as supporting functions
within the dairy corporation (e.g. product testing laboratory), (3) packaging
suppliers, and (4) transport companies, and (5) more generic actors such as
information system suppliers, standardisation organisations (such as GS1), gov-
ernment food quality inspectors, industrial federations and media. In this study
both interviews and observations were open, meaning that the true intention of the
case study was communicated to the informants and persons observed. An inter-
view guide was created and adapted to each informant. Data collection was
directed by an emergent frame of reference written down and successively refined.
This fundamentally iterative method meant that each interview guide exhibited in
practice a particular stage of learning in the research project. The research protocol
directed the formulation of interview guides, each adapted to a specific informant,
and taking into consideration preceding findings. This involved designing the
research process leading to observations generated new questions on which
further interviews could be based and eventually added new dimensions to the
subject, which eventually resulted in a new view of the phenomenon itself
(Dubois and Gadde 2002). This openness was also used as a tool to ensure
credibility of the research as discussed later in this chapter. All interviews were
taped and transcribed.
9 Product Containment Resources Facilitating Decision-Making 175
Production dairies
WILAB Production
(solid products)
control system Production dairies
(liquid products)
Order information
Sales department
at the dairy
Forecasting and planning
input Distribution transport
operations information
Corporate logistics
department TINE (corp.) Distribution centre logistics
department
The studied internal vertically integrated dairy supply network of the Norwegian
producer owned cooperative TINE consists of 53 production facilities organised
into five regions. This number of facilities is being gradually reduced in an effort
to rationalise production and distribution of its products. Products are mainly
classified into one of two groups: solid or liquid. Examples of liquid
products are juices, milk, pudding desserts, and yoghurt; solid products are
mainly cheeses. 23 dairies produce various milk products. These dairies are
commonly termed as liquid-dairies also functioning as distribution centres.
The network also supplies raw materials and goods produced by other companies
to consumers in Norway and exports mainly cheese products to consumers in
other countries.
176 P. Engelseth
9.3.3 Product
Raw-milk is stored in tanks at the farm, on the truck, and in silo tanks. The milk
tanks may be used for any liquid product and have devices to keep the milk at the
required temperature of 4C. Raw-milk is used to manufacture different dairy
9 Product Containment Resources Facilitating Decision-Making 177
products. Milk demands the highest quality sorting of the raw-milk; lower quality
is used for cultured products such as sour milk (buttermilk), cottage cheese, and
yoghurt. Tanks used to store raw-milk have refrigeration devices and are located at
the farm, on the truck, and at the dairy. At the studied regional dairy, seven silo
tanks with a capacity of 100,000 L are used. Silo tanks are filled up gradually and
the volume stored in the tank is registered. In its finished form, milk is poured into
cartons which are packed in plastic trays holding 10 1 L milk cartons. Milk is then
packed into a variety of consumer-level packaging sizes and shapes. The product is
packed into 1, 1, , and L Pure-Pak cartons and 10 L bag-in-box type dis-
penser cartons. This case focuses on the 1 L carton, the most common packaging
type used. This package size accounts for somewhat over 50% of the low-fat milk
production volume produced which in turn accounts for 50% of the studied dairys
total milk product production. The 1 L carton is predominately distributed through
food-retailer outlets to consumers. The low-fat milk carton is printed using pink
and white colouring with text communicating product information and producer
contact information, including a telephone hotline for customer queries. Also, one
side of the milk carton contains text with information, or entertainment that is not
directly related to the product. On top of the arch, numerical codes are stamped
indicating the place and exact time of production. One code indicates which dairy,
including the production line, where the milk has been produced. The best-before
date is the most obvious information stamped and indicates a date 10 days after
production.
Finished milk products are filled into specialised 1 L Pure-Pak drinking cartons.
The arched-top Pure-Pak carton is used because it is believed to symbolise
freshness and high quality, compared with the main competing carton type,
the brick-like TETRA-Pak. This consumer perception of freshness is related to the
practice that the Pure-Pak carton is only used for less durable, liquid food products.
The 1 L carton is stacked on plastic trays containing 10 1 L cartons each and roll-
rack containers holding 160 1 L cartons. Milk cartons on blue plastic trays or
cartons not on trays are loaded into roll-rack containers. These trays and containers
are designed specifically to hold 1 L, standard shaped Pure-Pak cartons of milk or
juice products. The 1 L size, which was designed more recently, is adapted to
also fit onto the trays and roll-rack containers. The roll-rack is a relatively small,
manually handled, grid-like container on wheels and with a brake. It measures
42 cm in width, 67 cm in depth and 124.5 cm in height. The roll-rack is a flexible
container with shelves on both sides that are hinged along the walls of the con-
tainer making it possible to split half the container horizontally, as well as split the
container in a vertical fashion. The roll-rack may also be folded in a way that
allows the racks to be pushed one inside the other horizontally comparable to how
shopping carts are placed together, to save storage space for unused roll-racks.
Roll-rack containers are used to carry a variety of TINE products and to store and
transport these products in the dairy terminal and on trucks to retailer facilities.
Milk cartons may also be placed into roll-rack containers without using plastic
trays when they carry one type of milk product exclusively.
178 P. Engelseth
Dairy: Dairy:
Farm: Roll racks or trays
Truck storage tank
cow
1-litre
Retailer: Roll carton
End-
Sort user
Sort Sort
Sort
Sort
Handling
Sort and outbound
Transport and transport Purchase
receiving milk
Various containment forms of milk illustrated in Fig. 9.4. The figure describes
how milk undergoes production in two phases i.e. when the cow is milked, and at
the dairy. Within the supply chain, the raw-milk is transformed through production
at the dairy into a finished consumer-packed product. The total time from milking
until the best-before date runs out is approximately 1314 days. Each morning, the
cows that are producing must be milked; otherwise they suffer. The milking
equipment is manually attached to the cows udder and the pump is switched on.
Milk is then pumped into a tank located in a room adjacent to where the cows are
held. This tank is refrigerated and has gauges indicating temperature and the
volume of milk contained. Control samples are taken from each individual milking
cow every month by the central quality control. These samples are placed into
small plastic cups and sealed with a lid. On the lid, a label, printed by the
The tank vehicles normally arrive within an interval of two hours some time
before noon three times a week. The dairy farmers need not be present when milk
is collected. The truck driver enters the unlocked production facility and attaches
the pumping equipment on the truck to the storage tank in the barn. When the milk
has been loaded into the truck, it is blended with deliveries from other producers.
The truck driver takes a sample from the milk tank at the farm and places it in a
sealed plastic cup which then is affixed with a label containing identification
information in numerical, textual, and bar-coded form. Then the truck follows its
route, stopping at other farms to collect milk, before arriving at the dairy.
Tank vehicles carrying raw milk arrive at the dairy each day except Sundays,
the first truck arriving at around 11:00 a.m. The dairy owns and operates 340 tank
vehicles; each truck has a capacity of 15,000 L which may be doubled by attaching
a trailer wagon. Samples are taken from the milk in each truck and analysed at the
9 Product Containment Resources Facilitating Decision-Making 179
laboratory before being pumped into one of the seven silo tanks, which may
contain up to 100,000 L each. The milk may be pumped out within an hour if the
quality is accepted. If not, the samples from each of the producers are tested to
trace the farm that is the source of the quality failure. The delivered milk is
assigned to one of the seven silo tanks in accordance with the milks quality and a
FiFo (first-in-first out) storage policy.
Raw-milk is pumped in a continuous manner into the production lines.
The studied dairy runs a dual-shift production schedule and produces about
30.9 million litres of low fat milk annually. 1112 thousand litres of milk are
produced per hour. Milk flows continuously through the production equipment.
No additives are used in the production. During the production process, the raw-
milk is separated and the level of fat regulated and given a heat treatment by
bringing the raw materials to 72C. The milk is pasteurised and homogenised,
cooled down, packed, and then automatically loaded onto roll-rack and/or plastic
tray containers.
Terminal workers collect products ordered by retailers at a designated loca-
tion, and these products are placed at a location dedicated to a specific delivery
route. 2526 thousand individual products are picked up every day at the
liquid dairy which also functions as a distribution centre. Products packed
mainly in distribution packaging are picked from a specific location and loaded,
either on pallets or roll-racks, and then moved into containers at the dedicated
location in the dairy. Some goods are collected on a roll-rack type transport
packaging; a cage with wheels. These products are placed at a special location in
the terminal area of the distribution centre. Here, the racks are placed in aisles,
one for each product. The floor of this terminal location is tilted, allowing the
racks to roll automatically into a convenient position for the truck driver to pick
up. Most retailers in the Oslo area receive unbroken roll-rack containers of milk
products. However, most other products are placed manually onto the roll-rack
containing a mixture of different products. About 30% of the goods picked at the
studied distribution centres are contained in non-mixed roll-rack containers.
Each of these is designated to a single retailer. The containers bound for the
different customers are loaded on two levels into the delivery truck in a manner
that allows for an easy unloading sequence. All delivery trucks are of approx-
imately equal size and have a refrigerated storage compartment designed to carry
roll-rack containers on two levels.
The truck carrying products in roll-racks is received at the back of the store.
Each retailer receives between 1 and 30 roll-rack containers. Smaller stores must
retrieve trays of low-fat milk from a container carrying other products, while the
larger retailers receive a number of entire containers loaded with milk. The
goods are then moved into the store display area, although milk is usually stored
for a few hours until the remaining previous supply is sold out. The retailer does
not keep a buffer inventory of milk products. The consumers pick milk together
with other dairy products, out of a refrigerated area, usually displayed behind a
glass door.
180 P. Engelseth
The long-term level involves planning in relation to the farm and production.
An intermediate form of planning takes place at the dairy. This provides forecasts
that are used as information input in relation to information transformation which
is interlinked with the actual product transformation depicted in the preceding
section. Finally tracking and tracing are described as activities using transformed
information to create new forms of information concerning product location and
product history. In the following descriptions information concerns resources
embedded in information systems, while knowledge is embedded in actors (e.g.
farmers, logistics departments, and retailers).
Each year, the farmers organisations and the Norwegian Government negotiate an
agreement regarding farm subsidies. The amount of these subsidies also regulates
the total volume of raw-milk that is produced. This agreement represents the basis
for calculating raw-milk process and production volumes at each farm producing
raw-milk. At each dairy, the organisation department informs farmers helping
them to produce in accordance with production quotas.
Each farm is visited three times a week by the milk collection truck. This truck has
an onboard computer that registers the milk collected at the farm facilities. The
truck computer is also provided with information to be communicated to the
farmers. It produces a receipt informing about the amount collected, quality
assessment of the previous collection, and other handy reminders from the regional
dairy the farmer is affiliated with. The truck driver has a mobile phone that may be
9 Product Containment Resources Facilitating Decision-Making 181
used to re-route the truck or provide last-minute information to the truck driver.
The truck also has an on board compartment that stores trays of raw-milk samples.
From each delivery, two samples are placed into small cups that have a bar-coded
label attached produced by the truck computer. Once every second month, each
farm also delivers a set of similar cups of samples, one from each producing cow,
attached with labels produced by the truck computer and previously provided to
the farmer. These samples are sent to the livestock control department.
The dairy sales department has representatives who every morning phone
customers according to a phone list provided by the MOVEX ERP system.
The regional dairy supplies milk to various kinds of customers; HORECA (Hotel-
restaurants-catering), small kiosks, and food-product retailers, for instance, are
important customers. The food retailers expect the daily call from the dairy at
approximately the same time each day. Prior to this call, the retailer personnel
must estimate the needed volume of dairy products to be delivered later that same
day. This takes the staff at a large, full assortment supermarket about of an
hour every morning. The retailer has a binder containing a product list that
includes product presentations, names, and numerical product codes used to assist
in the daily registering of the stock of various products. The retailer orders dairy
products through a manual process. Milk is one of about 250 products covered on
the order list provided by the dairy to each customer. On the list, each product is
indicated with a 4-digit product number, a text describing product name and type,
including the size of the consumer packages, and the quantity of consumer
packages in each distribution-level package which at this stage of the flow func-
tions as a logistics unit. Milk is registered in units of blue plastic trays or roll-
racks. During the daily morning phone conversation, the sales representative
promotes products that according to statistics may be selling poorly, attempting to
discuss reasons for this and also trying to sell a few extra units of each product.
The MOVEX ERP system creates all the documents and labels needed to deliver
milk to retailers. When orders from all the customers have been received, the
information system automatically creates picking lists that are used to load the
various trucks. These lists are used by about 40 terminal workers at the dairy in
Oslo to place the various products ordered by each customer at a designated
location assigned to a specific truck. The information system also produces a
transport document that is used by the truck driver. This document contains
information concerning the time and date of the transport, how many roll-rack
containers and other packaging forms are loaded, the destination of each of these
packaging units, and the name of the dairy supplier. The driver uses this freight bill
182 P. Engelseth
A retailer representative receives the goods and signs the order document at the
supermarket. The goods are then moved into the store display area. The truck
driver returns a copy of the signed order to the regional dairy. The retailer has
already registered the ordered amount in the store computer system, and now the
delivery is verified. Later, the retailer receives an invoice for the delivered amount
at regular intervals. When milk is purchased contained in its consumer packaging,
this is registered in the store computer system. Even though both orders and sales
are registered in this system, it is not used to calculate milk product inventory.
The only information provided by the retailer to the customer regarding the
product is a shelf label on or below the door of the display area.
First, milk may be tracked in its state as a raw material in transport between the
producer and the dairy. Second, milk may be tracked when it is in storage and in
production at the dairy facility. Third, milk may be tracked in transport to the
retailer. The truck driver transporting raw-milk to the dairy follows a pre-defined
transport route. Thus, the approximate location of a truck may be estimated. If, for
some reason, it becomes necessary to alter this route while the transport of raw-
milk is underway, the truck driver may be contacted through mobile phone. During
production, milk from one specific silo-tank is always used, and it is therefore
possible to ascertain whether milk from one specific farm is part of a limited
collection of raw-milk supplies from different producers. In the same manner, milk
that has been filled into a carton may also be traced to milk from a specific silo.
The production of milk is highly automated and controlled by the WILAB system,
which system registers the volume of milk that is in the production process. In
practice, tracking milk in production involves identifying which silo tank was used
as the source of raw-milk. Tracking milk is carried out predominantly when a
9 Product Containment Resources Facilitating Decision-Making 183
quality discrepancy has been detected. Only a few such incidents were reported in
the past years and involved detecting a fault with a packaged, finished product that
already had been delivered from the terminal facility. Quality control is carried out
at the laboratory at the dairy upon arrival from the farm. In most cases, a faulty
product will still be in storage at the dairy when a test may reveal product quality
discrepancies. In repeated or more serious cases, involving large volume of goods,
a retailer may report product failures to the sales department. Tracking during
transport from the distribution centre to the retailer is done using the MOVEX ERP
system. If a milk product needs to be stopped en-route to its retailer destination, a
form of detective work must be carried out since MOVEX does not automati-
cally provide tracking data. The personnel at the dairy, checking the MOVEX
order, must go through freight bills, picking lists and production schedules, using
the time of production as an indicator to estimate according to pick-up times of the
products, which retailers most likely will be receiving a specific faulty batch of
goods. Then the truck drivers carrying these goods must be notified by mobile
phone to stop delivery. If delivery has taken place the retailers must be notified by
phone.
the date of production, and then finding out which farms this milk was collected
from. Should there be an immediate need to locate a quality discrepancy reported
from a retailer or consumer, the individual samples taken from each farm location
may be used. The contents of the truck and the silo have, however, already been
tested, hence it is only if a mistake has been made in this testing that such
procedures will be used. This way, quality discrepancy from a 1 L carton of milk
may be traced back to an individual cow. This process is time-consuming and
involves several different actors and informational resources. Product tracing is
mainly used to detect the origin of product discrepancies. However, queries may
be posed just to receive detailed product information not provided by the pack-
aging. Consumers and retailers are the two actors in the supply chain that most
likely may demand such information. Retailers have an established communica-
tion link with the sales department, and, occasionally, retailers call by phone
seeking more detailed information about product characteristics. The dairy man-
agement could not account for any actual direct consumer enquiries regarding the
specific features of its milk products.
The preceding case narrative depicts how a relatively simple product transfor-
mation is supported by a far more complex information transformation. Product
containment facilities remain what they were when we started this study, an
operations facility as well as a research tool. These are two distinct realms of
package use applied in this study. Product containment facilities including pack-
ages are evoked as a central logistics resource although not prominent in profes-
sional discourse. It is primarily the researcher that sees the value of package and
goods containment use. However, when considering the informational aspect of
goods containment facilities in binding products with knowledge through trans-
forming information, such as when goods are identified, the role of packages and
goods containment facilities in securing connectivity between technical resources
and actors becomes evident. This study, however, although concerned with
logistics development, is not primarily concerned with developing package design
as highlighted in packaging logistics. The role of the package as a systems com-
ponent in the overall supply network is evoked.
Different decision-making sorts indicate knowledge transformation dependent
on updated information concerning the flow of products. This information may be
concerned with at core different temporal aspects of the product; past (tracing),
present (goods identification and tracking) or future (supply). The case narrative,
structured in accordance with Aldersons (1965) transvection model, also reveals
how product supply takes place in a complex network of multiple actors managing
equally complex flows of information and products. Linkages between actors,
often conceptualized as business relationships, are proposed to be modeled as
constituting two main technical transformed resources (products and information)
9 Product Containment Resources Facilitating Decision-Making 185
Information transformation
Transformed Transformed
Transformed
knowledge knowledge
knowledge
Purpose: Value
Product transformation realization (product
placement in the hands of
an end-user)
References
Choi, T. Y., & Krause, D. R. (2006). The supply base and its complexity: Implications for
transaction costs, risks, responsiveness, and innovation. Journal of Operations Management,
24(5), 637652.
Closs, D. J., Swink, M., & Nair, A. (2005). The role of information connectivity in making
logistics programs successful. International Journal of Physical Distribution and Logistics
Management, 35(4), 258277.
Dubois, A., & Gadde, L.-E. (2002). Systematic Combining: An abductive approach to case
research. Journal of Business Research, 55(7), 553560.
Engelseth, P. (2007). The role of the package as an information resource in the supply chain, a
case study of distributing fresh foods to retailers in Norway. Unpublished doctoral
dissertation, BI Norwegian School of Management, Oslo.
Engelseth, P. (2012). Modelling transformations in a complete fresh food value network.
In H. Jodlbauer, J. Olhager & R. J. Schonberger (Eds.), Modelling Value. Selected Papers of
the 1st International Conference on Value Chain Management. Berlin: Springer.
Fawcett, S. B., Osterhaus, P., Mangan, G. M., Brau, J. C., & McCarter, M. W. (2007).
Information sharing and supply chain performance: The role of connectivity and willingness.
Supply Chain Management: An International Journal, 12(5), 358368.
Forrester, J. W. (1958). Industrial dynamics: A major breakthrough for decision makers. Harvard
Business Review, 36(4), 3766.
Forrester, J. W. (1961). Industrial dynamics. Cambridge: MIT Press.
Francis, V. (2008). Supply chain visibility: Lost in translation? Supply Chain Management:
An International Journal, 13(3), 180184.
Frohlich, M., & Westbrook, R. (2001). Arcs of integration: An international study of supply chain
strategies. Journal of Operations Management, 19(2), 185200.
Garcia-Arca, J., & Prado, J. C. P. (2008). Packaging design from a supply chain approach. Supply
Chain Management: An International Journal, 13(5), 375380.
Gottinger, H. W. (1983). Coping with complexity: Perspectives for economics, management and
social sciences. Dortrecht: D. Reidel.
Jahre, M., & Hatteland, C. J. (2004). Packages and physical distributionimplications for
integration and standardisation. International Journal of Physical Distribution and Logistics
Management, 34(2), 123139.
Lambert, D., & Cooper, M. (2000). Issues in supply chain management. Industrial Marketing
Management, 29(1), 6583.
Lockamy, A., & McCormack, K. (2004). The development of a supply chain process maturity
model using concepts of business process orientation. Supply Chain Management: An
International Journal, 9(4), 272278.
Mead, G. H. (1934). Mind, Self, and Society. In C. W. Morris (Ed.), Chicago: University of
Chicago Press.
Miller, T. E., & Berger, D. W. (2001). Totally integrated enterprises: A framework and
methodology for business and technology improvement. Boca Raton: St. Lucie Press.
Pathak, S. D., Day, J. M., Nair, A., Sawaya, W. J., & Kristal, M. M. (2007). Complexity and
adaptivity in supply networks: Building supply network theory using complex adaptive
systems perspective. Decision Sciences, 38(4), 45474580.
Petrie, C. J. (1992). Enterprise integration modeling in proceedings of the first international
conference, scientific and engineering computation series. Cambridge: MIT Press.
Rosenbloom, B. (1995). Channels of Distribution. In M. Baker (Ed.), Companion encyclopedia of
marketing. London: Routledge.
Sezen, B. (2008). Relative effects of design, integration and information sharing on supply chain
performance. Supply Chain Management: An International Journal, 13(3), 233240.
Sherer, S. A. (2005). From supply-chain management to value network advocacy: Implications
for e-supply chains. Supply Chain Management: An International Journal, 10(2), 7783.
Skilton, P. F., & Robinson, J. L. (2009). Traceability and normal accident theory: How does
supply network complexity influence the traceability of adverse events? Journal of Supply
Chain Management, 45(3), 4053.
188 P. Engelseth
Sterman, J. D. (2001). System Dynamics modelling: Tools for learning in a complex world.
California Management Review, 43(4), 826.
Sterns, J. A., Schweikhardt, D. B., & Peterson, H. C. (1998). Using case studies as an approach
for conducting agribusiness research. International Food and Agribusiness Review, 1(3),
311327.
Surana, A., Kumara, S., Greaves, M., & Raghavan, U. N. (2005). Supply-chain networks: A
complex adaptive systems perspective. International Journal of Production Research, 43(20),
42354265.
Thompson, J. D. (1967). Organizations in action. New York: McGraw Hill.
Twede, D., & Parsons, B. (1997). Distribution Packaging for Logistical Systems. Surrey:
Pira International.
Vernadat, F. B. (1996). Enterprise modelling and integration: Principles and applications.
London: Chapman & Hall.
Vernadat, F. (2004, June). Enterprise Modelling: Objectives, Constructs & Ontologies. Tutorial
held at the EMOI-CAISE Workshop, Riga, Latvia.
Chapter 10
Order Fulfillment: A Key to Supply Chain
Integration
10.1 Introduction
The order fulfillment process can be described as a core supply chain process for
supply chain management (SCM). This is because the order fulfillment process
triggers all the other supply chain processes into action to satisfy the customers
orders. It is often the only process through which members of a supply chain
interact and as such is a key process for not only implementing but improving
supply chain integration and outcomes.
In the first section of this chapter, supply chain integration will be discussed. An
important issue is that in reality there is limited uptake of supply chain integration
amongst firms. Through discussion of the management of order fulfillment, the
Implementing supply chain integration requires firms to make the transition from
a functional focused organisation to one focused on business processes, first inside
the firm and then across firms into the supply chain. Integration of key business
processes across the chain through SCM can add value for customers and to other
stakeholders (Lambert and Pohlen 2001). The reality is that while much attention
has been given to overall SCM and supply chain integration, uptake amongst firms
is often limited. In reality, most integration only occurs through one tier of the
supply chain. Businesses are often struggling to have their own internal processes
understood across internal functions and lack understanding of the internal
processes of their customer or suppliers for the benefit of their own business
success. Branch (2009) states that most companies, 47% of them, are working
to create seamless processes within their own organisation only, 34% of companies
focus on integration with first tier suppliers, 11% of companies focus on
integration with key customers and only 9% are integrating logistics up and down
stream between customers and suppliers as depicted in Fig. 10.1.
Partial integration, rather than holistic integration, is more likely to be obtained
and there is also a risk of tightly integrated supply chains becoming obsolete by the
time they are established due to competition, technological changes and shortened
product life cycles. Consequently, there will be many varieties and degrees
of integration and there will be many combinations and degrees of complexity in
inter firm relationships across supply chain members. Most often what is required
are the incremental steps that can lead to successful integration and more
understanding of the diversity of SCM and managerial approaches and mechanisms
for dealing with integration and performance measurement.
There remains a real lack of quantitative models in this area that are flexible
enough to deal with the complexity of the supply chain environment and robust or
comprehensive enough to ensure supply chain integration is enhanced. In the order
fulfillment model presented supply chain improvement efforts are made through
being tied to the performance measurement system itself. The implications of
having a focused supply chain performance measurement system in place are that
a firm will have an appropriate understanding of their internal and external
10 Order Fulfillment 191
Scope of SCM
Focus on seamless
internal processes
Logistics integrated
between key suppliers
and customers up and
downstream
Fig. 10.1 Scope of SCM under taken by firms (Adapted from Branch 2009, p. 39)
Fig. 10.2 Supply chain configuration showing order fulfillment (OF) internally and externally
supply chain. It is often the only interface a firm has with its customers making it a
core supply chain process for meeting customer requirements as illustrated in
Fig. 10.2. In a nut shell, what customers want is to have their orders handled quickly,
accurately and in a cost effective manner (Shapiro et al. 2004). The specifics of
customer wants will vary from customer to customer and need to be addressed.
On a micro level, having worked for over 15 years in manufacturing and logistics
and SCM environments the author has experienced order fulfillment at a professional
level. Late deliveries of supplies result in delayed production; order back logs, re-
scheduling, increased inventory carrying costs, waste of time and labour, and poor
customer satisfaction and even loss of customers. In a retail environment stock
replenishment managers face frustration with stock outs and stock reports with line
upon line of unavailable product and excess of unwanted stock with ineffective key
performance indicators (KPIs) and often no continuous improvement initiatives
available to address the problems. Whilst the consumer has a seemingly endless
choice of products and services to choose from the experience is often frustrating, full
of compromise and requirements are often not met.
There is increasing recognition of the importance of order fulfillment in
meeting strategic objectives and customers needs and of its potential to improve
supply chain performance through increased integration across the chain.
emphasis from the literature is that they need to be based on meeting the
customers needs and what creates value for the customer.
The majority of metrics that are currently used are qualitative, such as
responsiveness, flexibility and customer responsiveness. The specific process
measures are limited, focusing on order cycle times, fill rate, order completeness or
perfect order, and do not fully reflect the breadth of this important process
(Beamon 1999; Gunasekaran et al. 2004; Forslund 2006). Finding a way to
determine the most appropriate measures for the variety of presentations of the
order fulfillment process would fill a gap in performance measurement not only
for this important core supply chain process, but would also address a severe
deficiency in supply chain performance measurement on the whole.
The perfect order index has become a popular supply chain metric within
industry because of the growing realisation that common supply chain metrics are
not customer centric. A perfect order is defined as:
Being delivered on time
Complete
Damage free and
Having the correct invoice.
The perfect order index is calculated by multiplying the results of each of the
four metrics with each other (Visatek 2007) as in Fig. 10.3.
Factors affecting the perfect order are lack of collaboration between supply
chain members and lack of visibility of demand at the front end of the supply
chain, which are key supply chain integration issues.
The financial implications of the imperfect order are significant due to returns
costs, labour costs for multiple shipments, replacement product costs, costs of
refunding the purchase price and providing credit, costs of processing additional
receipts for multiple shipments, loss of revenue through cost of lost sales and cost
of lost customers.
The benefits of achieving the perfect order include (Lapide 2007):
Less inventory
Earlier payment
Cash-to-cash cycle reduced, time to market improvement with fewer returns
and discrepancies
Reduced stock outs
Reduction in days payable outsourcing and days sales outstanding.
194 Y. Amer and L. Luong
This chapter presents an order fulfillment model which can be used by operations
managers in a practical way to optimise the order fulfilment process within a
supply chain and meet the strategic objectives of meeting the customer require-
ments whilst aiding supply chain integration. As the order fulfillment process can
have different configurations depending on the nature of the firm and position in
the chain the model accounts for this variability and will:
support the supply chain strategy of meeting customer needs
accommodate varying order fulfillment structures
select the most appropriate order fulfillment performance measures for varying
order fulfillment structures
address the internal and external order fulfillment environment and
optimise order fulfillment performance through evaluating, improving, moni-
toring and controlling the order fulfillment process.
Hence the incorporation of the most appropriate tools to develop the order
fulfillment model is a crucial consideration and the first requirement would be its
ability to take into account customer requirements.
196 Y. Amer and L. Luong
DFSS focuses on profit maximising through effective translation of the voice of the
customer (VOC) and designing in quality rather than just improving quality, thus
fitting the requirement of a means to support strategy and to meet customer
requirements.
DFSS also approaches systems design proactively where the focus is placed
on spending more effort in collecting, translating and managing the VOC through
a systematic integration of tools, methods, processes at the onset of a project
and later relating the established critical to customer requirements (CCRs)
to lower level Key Process Input Variables (KPIVs) through a predictive transfer
function (TF). Therefore it is possible to determine the appropriate metrics for
the order fulfillment process and for meeting the customers requirements (Amer
et al. 2008).
Effective translation of VOC from a wide continuum of customers as found in
the supply chain is also possible making DFSS an approach which can deal with
multiple customers and channels.
Additionally DFSS focuses on controlling the supply chain and supply chain
process variation to Six Sigma standards. Contrary to the apparent statistical
implication of DFSS, a Six Sigma performance level is preferred but not the
mandatory target for all situations. However, a step by step systematic approach to
development is stressed upon in any case (Amer et al. 2007, 2008). This provides
a means of designing the supply chains process flow, analysing the systems
structure and adjusting and redesigning it (Amer et al. 2010). Monitoring
performance through the evaluation of Key Process Output Variables (KPOVs)
and KPIVs will identify when there is a failure to meet customer requirements
and signal the need to redesign through a cycle view. Thus the DFSS approach
provides a means to design, monitor, control and improve supply chain processes
as required.
DFSS strategy should provide design principles that directionally lead to good
concepts. In a totality, the DFSS strategy of the model to design right the first
time means it:
is customer driven and focused
measures the effectiveness of the design process
has measures to compare performance versus requirements, aims to achieve key
business objectives
upfront development of robustness and testing for verification
brings paradigm shift from find-fix-test to prevention
uses integrated approach to design from concept to validation
pinpoints where it is easier to implement changes when needed
increases the potential for high reliability and robustness
establishes Six Sigma capability by conceptual means upfront and
uses optimisation to set tolerances when design concept is finalised.
10 Order Fulfillment 197
The tools and activities that were selected for use in the model from under the
umbrella of Six sigma are the Quality Function Deployment (QFD) approach to
effectively translate the VOC and the fishbone tool combined with a process flow
chart as a brainstorming tool. The cause and effect analysis of fishbone analysis
provides a structured way to help think through all possible causes of a problem
helping to carry out a thorough analysis of a situation. Flow charts organise infor-
mation about a process in a graphical manner and makes it clear what is impacted at
every stage. They link the internal processes with the external, giving the designers
insight into where processes intersect with suppliers and other supply chain
members. Brainstorming helps the design team to identify major processes and
causes of problems. These are then connected in the fishbone diagram. Unlike in
product design, performance metrics for a service design like supply chain design are
mostly intangible in nature, which involves an inherent fuzziness in their behavior.
The CCRs for a supply chain requires a predictive transfer function to be developed
to relate the KPOVs with their relevant KPIVs. Fuzzy logic through the use of
the Fuzzy Inference System (FIS) via MATLAB fuzzy logic toolbox was chosen.
The reason for this choice was that fuzzy logic provides a means to deal with
qualitative supply chain measures and to deal with the complex and variable nature of
the supply chain requirements. The MATLAB fuzzy logic tool box makes the fuzzy
logic approach accessible and applicable in an operational setting. Table 10.1 shows
the relationship between the order fulfillment model requirements and the tools
selected to accommodate that requirement.
The order fulfillment model is depicted in Fig. 10.5 and is tabulated following
the Design for six sigma activities. In this model the key, core process order
fulfillment, as described by Amer et al. (2008, 2010), is selected to illustrate
the model. Any order fulfillment configuration could have been applied. The
assumption to be made is that the order fulfillment model is presented as a process
from a companys or applications point of view.
198 Y. Amer and L. Luong
10.5.1 Identify
In the Identify phase the supply chain design team is formed, the team defines the
scope of the project they are undertaking; they establish and translate the VOC to
determine the CCRs.
10 Order Fulfillment 199
The cross functional team is formed with members representative of the various
functions involved in the supply chain. SCM involves nearly all functions
within a firm; therefore it is important to have members from, i.e. logistics,
warehousing, manufacturing, shop floor, operations, finance, packaging etc.,
depending on the firms structure. The important consideration is that it should
be representative of the chain. There should also be representation from the
different organisational hierarchies, i.e. support from top level management,
middle management and shop floor/co workers etc. Training in the methodol-
ogies and tools would need to be undertaken and a team leader who is able to
steer change management needs to be selected prior to implementation.
The idea is to develop a cohesive group who, through experience and time,
initiate and energise improvement across the organisation. This initial step of
developing the team should not be underestimated as it can make or break the
initiative.
200 Y. Amer and L. Luong
The team will initially define the scope, in the first instance brainstorming to
consider any particular operational problems or concerns. Defining the scope will
enable the team to understand the criteria for success. The team will need to
identify the customer(s) and their requirements and clearly define the design
requirements for the supply chain processes through the following steps;
Quantifying the VOC begins by tracking down the customers into their market
segments. The VOC can be initiated from both the internal and external customers
of the supply chain. As an example we could segment customers by how, where,
and when the customer uses the product or service. This information can help
focus the supply chain design teams attention to CCRs. After market segmen-
tation the team obtains information from the customer in a structured way using a
variety of methods. These methods include analysis of customer complaints, focus
groups, surveys, operational reports, through historical data analysis, competitive
intelligence, as well as interviews, and on site customer visits. This allows the
team to understand the major drivers of CCRs which are later analysed under the
Design phase and broken into big Ys representing customer requirements at an
internal operational metric level, that is, the project KPOV level (El-Haik and
Roy 2005).
Using QFD and brainstorming the VOC are used to iteratively derive design
specifications. The team determines the relationship between the CCRs and the
technical requirements and determines the target for each CCR. This last step of
translating the VOC uses QFD to compare customer requirements, in the form of
KPOVs, against current internal order fulfillment system design to satisfy those
requirements (Summers 2007; El-Haik and Al-Aomar 2006).
Performance gaps that may be identified by the team during this process serve
as the basis for Six sigma improvement projects. This is one of the advantages of
this model over other supply chain performance measurement initiatives as it has
an inbuilt continuous improvement focus. The quantified customer requirement,
that is KPOV variables, is compared against base lines representing current
system performance to identify and prioritise Six sigma improvement projects.
As an example, if customers require their deliveries to be on time plus or minus
one day, but current performance reports show an arrival variance of plus or minus
three days, this performance gap could serve as a potential six sigma project.
The business benefits gained from closing this performance gap are not only
increased customer satisfaction but lower supplier delivery costs when non value
adding time elements are eliminated from the process. The identification of
performance gaps should be conducted in a systematic way to ensure the Six sigma
improvement projects are properly prioritised prior to their development.
10 Order Fulfillment 201
HOWs / Outputs
Undamaged deliveries
Accurate invoices
Importance (1-5)
Delivery on time
Order flexibility
No extra costs
Transport
WHATs / Inputs
Direction of Improvement
Order Fulfillment 5 9 9 3 1 3 9 3
Cost 4 3 3 9 1 3 3
Product Quality 3 1 9 1 3
Quality Service 2 3 3 9 3 1
Environment friendly 1 9
Fig. 10.6 A first level QFD for order fulfillment (Amer et al. 2010)
For illustrative purposes, from a first level QFD, as depicted in Fig. 10.6, a
House of Quality is deployed from which it is determined that the CCR is order
fulfillment.
The whats/inputs is used to represent what the customer wants, so whats
are really customer attributes. These whats are general ideas that require more
detailed definition, for example customers say they want their order to be fulfilled
when they purchase a product. The order to be fulfilled may be a very desirable
feature but since it has different meanings to different people it cannot be acted
upon directly. The hows are design features derived from the team to answers
the whats. Each of the initial whats needs operational definitions. The
objective is to determine a set of CCRs that can materialise the whats.
This process translates customer expectations into design criteria. For each
what there should be one or more hows that describe a means of attaining
customer satisfaction. For example, order fulfillment can be achieved through
delivery being on time, undamaged and meeting demand quantity. At this stage
only overall requirements that can be measured and controlled need to be
determined.
The team defines the hows in a solution-neural environment and not be restricted
by listing specific parts and processes. The team is considered as an expert team,
202 Y. Amer and L. Luong
that is why the composition of the team is important. The hows are itemised
whereby the list of whats can be realised. In addition, each how will have some
direction of goodness or improvement. The relationship in every (what, how) cell
in the House of Quality can be displayed by placing a number representing the
cause and effect relationship strength in that cell. The score 9 means a strong
relationship, the score 3, means medium and the score 1 indicates a weak
relationship.
After determining the strength of each what/how relationship and marking it
in the cells, the team will review the relationship matrix. Blank rows or columns
indicate gaps in either the teams understanding or a deficiency in fulfilling cus-
tomer attributes. A blank row shows a need to develop a how for the what in
that row, indicating a potentially unsatisfied customer attribute. When a blank
column exists then one of the hows does not impact any of the whats.
Delivering that how may require a new what that has not been identified or it
may be a waste. The relationship matrix gives the supply chain design team the
opportunity to review their work leading to better planning and therefore better
results (Rath and Strong 2002).
The next step is to determine what extent the CCR in the column contributes to
meeting customer attributes in the row. This requires a subjective weighing of the
possible cause and effect relationship to rank the CCRs and customer features in order,
multiply the numerical value of the symbol representing the relationship by the
customer desirability index where 5 represents extremely important, 4 very
important, 3 somewhat important, 2 a little important and 1 not important. The
customer desirability index is the numerical rating of the relative importance of a given
customer attribute. For example, if the what is order fulfillment, the customer
usually thinks it is very important, so this will have a very high rating. Consequently,
order fulfillment is summed over all the customer features in the what selection which
provides a measure to the relative importance of such CCRs to the team. This is used as
a planning index to allocate resources and efforts by comparing the strengths,
importance, and interactions of these various relationships. This importance rating is
the importance of the hows rating and is a relative measure indicating the importance
of each what or how to the design and is calculated using Eq. 10.1.
The House of Quality in Fig. 10.6 demonstrates the order delivery time has
the highest level of importance (66), followed by undamaged delivers/quality (63)
with the order quantity being the next highest (62), then cost (51), transport (45),
meet urgent demand (36) and finally order flexibility (21). The team would focus
on the highest scoring, such as delivery time, undamaged goods and quantity.
10.5.2 Design
The Design phase has two aspects, the concept development and the preliminary
design. To develop the concept, the systems technical and operational requirements
need to be defined. Once the parameters are understood these must be translated into
10 Order Fulfillment 203
the actual, effective design. Then the CCRs are translated to process requirements.
Essentially the system design and CCR flow down are parallel activities. Design and
optimisation of order fulfillment begins with the creation of a system model map.
A system model implies that the team understands the major inputs xs and outputs
ys of the order fulfillment process as well as its interrelationship with other
processes both internally and externally.
Unlike product design, a service like the supply chain is best represented by the flow
of activities addressing each core process. the order fulfillment process can be
described by a sequence of sub-processes, such as order entry, order processing, start
manufacture, fill order, delivery, customer receipt, and post delivery activities.
These processes will vary from situation to situation. This is an illustrative case based
on the case study of Amer et al. (2010) where analysis was given to order fulfillment
by the supply chain design team through fish bone analysis incorporating
brainstorming and process mapping. Figure 10.7 shows a fishbone analysis for
making an order.
Functional, performance and operating requirements are allocated to order
fulfillment design components. A further analysis of the supply chain structure is
carried out through the critical to customer flow down to link the above variables
204 Y. Amer and L. Luong
to the supply chain system hierarchy. This includes analysis of information flows
as shown in Fig. 10.8, which illustrates the complexity of information in the case
of companies vendor-managed inventory information flow which in the situation
of stock outs would warrant further analysis by the design team.
The CCRs flow down to key input and output variables provides the basis for the
development of transfer functions. An active measurement system is put in place
for both internal and external CCRs metrics, sometimes called the big Y
10 Order Fulfillment 205
Fig. 10.9 Order fulfillment process flow with KPIVs/KPOV (Amer et al. 2010)
10.5.3 Optimise
The optimise stage involves further consideration of the order fulfillment design
to ensure it is deliverable and involves identifying the sources of variability,
minimising performance sensitivity to all sources of variation by ensuring a robust
206 Y. Amer and L. Luong
design, determining the supply chain capability and comparing with the design
specifications. During this phase the supply chain design team ensures the changes
to processes as required are put into place, this could involve reviewing not only
how things are performed but the documentation and support as well i.e. changes
to standard operating procedures etc. The aim is to develop robust subsystems. The
order fulfillment system and sub systems performance is evaluated through the
fuzzification of the perfect order TF.
Fuzzy logic is applied to optimise the order fulfillment TF. The fuzzy logic is
implemented as a part of the model using MATLAB fuzzy logic toolbox. The steps
for measuring the Perfect Order using the FIS are:
1. A list is made of the input variables that affect the Perfect Order, namely DT,
Qty, Qual, and MA.
2. The fuzzy membership values associated with the variables are assigned.
3. The if then rules are determined. The upper and lower specification limits for the
rules are set according to the expert knowledge of the team and from feedback
from customers. Once the rules have been established, the upper and lower limits
can be altered. Rules still stay the same, the over spread of the membership
function may get narrow or it may broaden thereby making the performance
evaluation stricter or looser. For example if the limits for the DT were 2 days,
expanding them to 3 days will keep the overall set of rules the same, just
allowing more room for the suppliers. The delivery time expectation is now more
relaxed. The membership functions can be modified to include more acceptance
and rejection stages. However, this will increase the number of rules drastically.
4. The FIS is used to determine the appropriate output fuzzy membership value
5. The weighted average method is used to determine the crisp value for the
output variable.
6. The Perfect Order is evaluated in relation to the crisp value which can be
translated to a linguistic value as in Table 10.2.
10 Order Fulfillment 207
Any value for order output between the values of 00.29 will cause the order to
be rejected, and values between 0.30 and 0.49 indicate that the order is poor and
needs major attention and action. Although the order value from 0.50 to 0.69
indicates acceptance of the order it shows a high gap from the optimum and the
processes and performance needs significant improvement. Order value between
0.70 and 0.89 indicates a low level of satisfaction aiming to close the gap towards
optimum. An order output value between 0.90 and 1 indicates that the order is
fulfilling the requirements and requires the processes to be monitored and main-
tained as shown in Fig. 10.10 and Table 10.2.
10.6 Validate
This stage includes checks to ensure that the application of the order fulfillment
model is complete, valid and will meet requirements in real life. The system will
be validated to ensure the order fulfillment design meets the requirements and that
established process controls in operations ensure that the CCRs are always pro-
duced to the specifications of the Optimise Phase. The design also needs to be
verified over time to ensure it is achieving the objectives and meeting the
requirements. The fuzzy evaluation rules can be modified as required accordingly.
The response surfaces of the rules exhibit the overall performance of the order
fulfillment at any settings within the limits of the membership functions. These
results are compared against the customer requirements. If this stage suggests that
the design does not meet the requirements then it is necessary to return to either the
Identify or Design stage. Figure 10.11 demonstrates a cycle view of the model
which shows the link between validate and redesign as required or return to
208 Y. Amer and L. Luong
Identify if there is a perception that customer requirements have shifted and need
to be redefined. It is suggested that the cycle view has a regular timetable to ensure
the supply chain design is optimised on a continuous basis. Such time frames
would need to be determined by the organisation itself, one that is in flux may need
to cycle more frequently than one that has a more stable agenda.
the order fulfillment model using DFSS and fuzzy logic has a continuous improve-
ment focus tied to cause and effect analysis, statistical tools, and quality control
mechanisms which aim to achieve Six sigma level.
The model enables the mapping of connections between the internal aspects of
order fulfillment and the external, bringing the focus out across the chain which
facilitates a connection between the internal and external members of the chain
(Amer et al. 2010). Likewise, the KPIVs can be developed to address both the
internal and external aspects of the order fulfillment process. This occurs through
the DFSS Design phase through process flow mapping, fishbone analysis and
brainstorming undertaken by the supply chain design team. This process addresses
a deficiency in SCM that most firms fail to understand their own internal pro-
cesses, and the processes of other chain members and have a tendency to focus on
a single firms processes instead of extended collaborative processes. The
approach of DFSS allows for targeted metrics that are relevant to the process
performed and the order fulfillment model can accommodate varying order ful-
fillment structures thereby making the metrics relevant to that structure.
The model could be used to build a new supply chain as in the case of axiomatic
design, to re design an existing one, or just to analyse, understand and improve the
existing chain (Amer et al. 2010).
The model has the potential for supply chain design innovation that is adaptive to
changing circumstances. It can be applied to different channels, addressing the issue
that supply chains can be obsolete by the time holistic frameworks are in place and
they therefore need to deal with multiple goals and strategies within a supply chain.
The order fulfillment model presented makes a contribution to improving supply
chain integration through an appropriate performance measurement system.
Further research opportunities exist to contribute to the understanding and
validity of the order fulfillment model using DFSS and fuzzy logic presented in
this Chapter by:
Applying the order fulfillment model in a variety of small to medium sized
enterprises
Using the model in different positions within a supply chain and its use be
extended across a chain
Assessing the business outcomes achieved
Other research opportunities exist to develop the order fulfillment model using
DFSS and fuzzy logic presented by:
Adapting the model to other supply chain processes. Given the mounting
research of demand management this is highlighted as an area for extension
Consideration of the use of Design of Experiment to optimise process param-
eters and process structures in the Optimise phase
Using discrete event analysis and simulation in the Verification phase to verify
the final design
Developing suitable IT platforms for the model which could make it more
accessible for commercial application
Enhancement of the model would involve the design of a simulation package.
210 Y. Amer and L. Luong
References
Amer, Y., Luong, L., Lee, S. H., Ashraf, M. A., & Wang, Y. C. (2007). A systems approach
to order fulfillment using design for six sigma methodology. International Journal of Business
and Systems Research, 1(3), 302316.
Amer, Y., Luong, L., Lee, S. H., & Ashraf, M. A. (2008). Optimising order fulfillment using
design for six sigma and fuzzy logic. International Journal of Management Science and
Engineering Management, 3(2), 8399.
Amer, Y., Luong, L., & Lee, S. H. (2010). Case study: Optimising order fulfillment in a global
retail supply chain. International Journal of Production Economics, 127(2), 278291.
Azevedo, R.C., DAmours, M., Ronnqvist, M. (2007). Designing the order fulfillment process: scope
and business variables. In Proceedings of the 2nd International conference on Changeable, Agile,
Reconfigurable and Virtual Production (CARV 2007) Toronto (pp. 2224), Ontario, Canada.
Beamon, B. (1999). Measuring supply chain performance. International Journal of Operations
and Production Management, 19(3), 275292.
Branch, A. E. (2009). Global supply chain management and international logistics. New York:
Routledge.
Croxton, K. (2003). The order fulfillment process. International Journal of Logistics Management,
14(1), 1932.
El-Haik, B., & Roy, D. (2005). Service design for six sigma. New York: Wiley.
El-Haik, B., & Al-Aomar, R. (2006). Simulation- based lean six sigma and design for six sigma.
New York: Wiley.
Forslund, H. (2006). Performance gaps in the dyadic order fulfillment process. International
Journal of Physical Distribution and Logistics Management, 36(8), 580595.
Gunasekaran, A., Patel, C., & McGaughey, R. (2004). A framework for supply chain
performance measurement. International Journal of Production Economics, 87(3), 333347.
Kuzeljevic, J., & Smyrlis, L. (2006). Warehousing perfecting the perfect order requires a much
more customer-centric approach. Canadian Transportation Logistic, 109(11), 6667.
Lambert, D. M., & Pohlen, T. L. (2001). Supply chain metrics. International Journal of Logistics
Management, 12(1), 119.
Lapide, L. (2007). Not-so-perfect order. Supply Chain Management Review, JulyAugust: 78.
Lee, H.L. (2000). Creating value through supply chain integration. Supply Chain Management
Review, SeptemberOctober, pp. 3036.
Lee, H. L. (2004). The triple a supply chain. Harvard Business Review, 82(10), 102112.
Rath and Strong. (2002). Design for six sigma. Massachusetts: Rath and Strong Publishers.
Shapiro, B., Katsuri, R., & Sviokla, J. (2004). Staple yourself to an order. Harvard Business
Review, 82(78), 162171.
Summers, D. C. S. (2007). Six sigma basic tools and techniques. USA: Pearson Prentice Hall.
Visatek, K. (2007). Perfecting the perfect order. www.scvisions.com. Accessed 5 July 2008.
Chapter 11
Emerging Technology and the Service
Delivery Supply Chain
Abstract Most manufacturers offer services, but may not use them as the basis of
their competitive strategy. Researchers in the Services Marketing, Operations
Management and Service Management communities have been studying the
provision of such product-service offerings for over 20 years. In this chapter we
illustrate the context of such service provision by manufacturers and explore how
the application of existing and developing technologies (sensors, signal process-
ing, ICT) can be used to support the effective and efficient delivery of product-
centric services and how the service delivery supply chain is affected. We provide
examples from the aerospace, railway and road transport sectors to consider the
relevant service enabling technologies employed in the delivery of advanced
product-centric services.
11.1.1 Introduction
Our aim has been to understand the differences between the information and
communication technologies used in production operations and those used by
successful servitized manufacturers. We suspected at the outset that these would
be complex processes which would require careful examination and so a case-
based methodology (Voss 1992; Yin 2003) was chosen. As there is little previous
work in this specific aspect of servitization, we adopted a largely inductive
approach to the study, setting out to identify those areas where leading companies
converged in their practices. Our decision to explore this particular topic was
inspired by earlier work of Baines et al. (2009) who postulated that the enabling
technology in servitized operations would differ to those in production and that
these would be complemented by differing technologies across the structure and
infrastructure of operations.
In short, our method has been to carry out in-depth and multi-disciplinary case-
studies of leading manufacturers in the delivery of advanced product-centric ser-
vices by interviewing senior managers/executives from differing operational areas
in these organisations (see Table 11.1). The partner organisations for this study
have been Rolls-Royce, Caterpillar, Alstom, MAN and Xerox. The research
approach has been to carry out semi-structured interviews with personnel ranging
from Customer Services, and Service Marketing, through to Chief Executive
Officers. Overall some 140 h of interviews were conducted. In each case the focus
of the interviews was on the generic structure of the ICTs used to support advanced
product-centric services, along with understanding the factors that have led to their
construction. The collected data was analysed by the research team using mind
214 H. W. Lightfoot et al.
Servitization, the term coined by Vandermerwe and Rada (1988), is now widely
recognised as the process of creating value by adding services to products. Since
the late 1980s its adoption as a competitive manufacturing strategy has been
studied by a range of authors (e.g. Wise and Baumgartner 1999; Oliva and Kal-
lenberg 2003; Slack 2005) who have specifically sought to understand the
development and implications of this concept. This literature indicates a growing
interest in this topic by academia, business and government (Hewitt 2002), much
of which is based on a belief that a move towards servitization is a means to create
additional value adding capabilities for traditional manufacturers.
Manufacturing companies have been selling services for some time. Tradi-
tionally, however, the tendency has been for managers to view services as a
necessary evil in the context of marketing strategies (Wise and Baumgartner 1999;
Gebauer et al. 2006). Here, the main part of total value creation was considered to
stem from physical goods, and services were assumed purely as an add-on to
products (Gebauer and Friedli 2005). From this beginning, there has now been a
dramatic change in the way services are developed and marketed by manufac-
turing companies. The provision of services has now turned into a conscious and
explicit strategy with services becoming a main differentiating factor in a totally
integrated product and service offering. Today, the value proposition often
includes services as fundamental value-added activities (e.g. Vandermerwe and
Rada 1988; Quinn et al. 1990) and in some cases can reduce the product to being
just a part of the offering (Oliva and Kallenberg 2003). Indeed, some companies
have found this to be a most effective way to open the door to future business.
A key feature of servitization strategies is a strong customer centricity. Cus-
tomers are not just provided with products but broader more tailored solutions.
These deliver desired outcomes for specific customers, or types of customer, even if
this requires the incorporation of products from other vendors (Miller et al. 2002;
Davies 2004). This use of multi-vendor products to deliver customer centric
solutions is exemplified by Alstoms maintenance, upgrade and operation of trains
and signaling systems, and similarly Rockwells on-site asset management for
maintenance and repair of automation products. Oliva and Kallenberg (2003)
consider this customer orientation to consist of two separate elements. Firstly, a
shift of the service offering from product-oriented services to users processes
oriented services (i.e.: a shift from a focus on ensuring the proper functioning and/
or customers use of the product, to pursuing efficiency and effectiveness of the end-
users processes related to the product). Secondly, a shift of the nature of customer
interaction from transaction-based to relationship-based (i.e. a shift from selling
products to establishing and maintaining a relationship with the customer).
11 Emerging Technology and the Service Delivery 215
We are now in an era where the services that companies provide must be smarter
and go beyond bundled offerings of product upkeep and upgrades. They must
deliver better value to customers and be more cost effective to deliver. This is
achieved by building in intelligence and connectivity into products and the supply
chain, and using the information provided to better manage the product in use.
Electronic and electromechanical products are in many cases viewed as stand-
alone objects. However, networking and managing these products in use will
enable manufacturers to deliver smarter services through their ability to monitor
status in order to report on product performance and usage in the operational
environment, together with diagnostics to facilitate repair and maintenance and
profiling of the product in use. Smarter services enable the manufacturer to
become pre-emptive based on intelligence from the product in the field and to take
action to avoid product failure in addition to improving the responsiveness of the
overall supply chain. This creates more value for the customer by removing the
element of surprise and impact associated with a product failure in the field. From
the manufacturers perspective smarter services deliver higher visibility of product
performance and customer behaviour. However, for most companies delivering
smarter services, the technology is not a major challenge, since although critical to
the task, the necessary technologies are, for the most part, now well enough
established. Rather, in most companies, the biggest challenge is for management to
adopt a new perspective on the use of the available technologies and the nature and
implications of their service business offering on the wider supply chain.
Case work (Lightfoot et al. 2011) has also illustrated that where a manufacturer
sets out to adopt advanced services two sets of macro business pressures are
incurred that help to explain their practices. One set reflects the direct customer
demands of a service offering and, in the case of more advanced services, is
concerned with measures of performance, availability and reliability. Performance
is concerned with the extent to which the full capability of equipment is accessible,
for example the power delivered by a gas-turbine as a percentage of that specified.
11 Emerging Technology and the Service Delivery 217
. weve got lots of data but its not in the same format, some is paper, some is
electronic, some is mainframe, some in pcs, some is in different businesses
(Baines et al. 2009) The operations planning and control systems used generally
encompass hardware manufacture together with maintenance, service processing
and data storage, and decision making Alonso-Rasgado (2004). A case study
senior manager in a case companys service organisation recognized that being
more joined up between your delivery arm and your commercial arm and the
customer service groups becomes much more imperative (Baines et al. 2009)
when the emphasis is on product availability and the delivery of functionality
to provide capability to the customer. Although, in product-centric servitized
organisations physical product quality tends to be controlled using traditional
production techniques, there is an understanding that (product) reliability, main-
tainability and reparability are important and that service quality is a measure of a
products availability to deliver functional performance for the customer.
A product that is reliable, easy to maintain and quickly repaired is essential to the
delivery of the level of service quality expected by customers.
As an example of this, the engine management system in MAN (UK, Truck and
Bus) vehicles is connected using the CANbus to a data acquisition device that
allows the continuous collection of information concerning the operation and
performance of the vehicle. This data, including GPS location, is transmitted in
real-time to the MAN U.K. control centre using cellular communications. The
data collected provides reports showing how the vehicle is being driven together
with the actual fuel consumption and whether optimum fuel efficiency is being
achieved. Vehicles and drivers can be compared by vehicle type, depot, route or
other parameters. Also Alstom (TLS) provides Virgin Trains UK West Coast
Mainline fleet with a maintenance, repair and overhaul capability against an
availability contract. To effectively and efficiently deliver this service Alstom has
introduced a range of enabling technology solutions to provide dynamic data on
train status and location, underpinned by an on-board Train Management System
(TMS) that monitors a range of 25 train systems that include propulsion, tilt, high
tension, braking, air, etc. using around 15 computers interfaced to various on-
board sensors. Using GPRS the TMS sends fault codes to fleet control centers
staffed by experts, who communicate with the maintenance depot staff regarding
the work needed, including parts, to rectify faults identified prior to the arrival of
the train at maintenance depots where routine maintenance and fault rectification
has to be completed in a narrow time window. The TMS data, expert analysis
220 H. W. Lightfoot et al.
data and work scheduling and parts acquisition is integrated using a sophisticated
IT system based around SAP. To effectively deliver the required level of service
responsiveness and flexibility of labour resource there has been an increase in the
level of in-sourcing of activities formerly provided by the external supply chain.
Alstom now considers remote monitoring of on-board systems as fundamental to
efficient and effective train operating and maintenance.
The use of technology (sensors, on-board processing, telemetry) to monitor an
asset in use, enables the service provider to dynamically monitor its performance
and take any necessary corrective action, which may well be carried out remotely.
For example SNELL provides Routing, Switching Conversion and Automation
products to broadcasters and programme makers and effects preventative product
maintenance via an automated system using on-board log files. Availability is
typically measured as the extent of time that a product or asset is available for use,
as a proportion of the scheduled availability. Here again the enabling technologies
in evidence allow the transmission of fault information to operations centers,
where expert staff could take maintenance and/or repair decisions at an early stage
and hence through advance materials and resource planning reduce any associated
asset down time and so optimize its availability to the customer. For Alstom TLS
the remote monitoring of the Train Management Systems on-board the UK
Pendolino fleet is fundamental to delivering against their contract to ensure train
availability. Reliability is taken as a measure of mean-distance between in-service
failures and again the capability to monitoring the asset in use and to analyse and
record operational performance issues enables feedback into the asset design loop
to improve its designed reliability. Here, for example Caterpillar have analysed
real time load distribution data collected from the suspension systems on their
quarry mining trucks to improve product reliability. The second principal pressure
is from the host manufacturer, and is concerned with the resources required to
deliver the service offering, and is typically measured as contract delivery cost.
The enabling technology and systems practices provide better visibility of the asset
in use in terms of condition, operating characteristics, time in use and location. The
remote access to this information for the service provider facilitates more timely
awareness of faults to provide faster maintenance/repair actions and can lead to
improved equipment design, operator behaviour and less need for manual obser-
vation of the asset in the field. The resulting improved responsiveness of the
service provider has a positive impact on asset performance and availability,
whilst the improvements in design and operator behaviour have a positive impact
on reliability. The increasing sophistication of enabling technologies and systems
has a downside in that it contributes to an increase in service delivery costs.
The relationship between the enabling ICT and business performance is cap-
tured in Fig. 11.2. This illustrates that there are five principal routes through which
11 Emerging Technology and the Service Delivery 221
Fig. 11.2 Illustrating the relationship between enabling technology and systems and key
performance measures for advanced services contracts
intervention and hence lower the overall cost of service delivery. Also, by having
better visibility of how an asset is operated, how long it is in operation and where it
is being used will enable the service provider to provide the most appropriate
support contract that is priced to reflect the customers usage pattern for that asset.
These enabling technology and systems practices, and their implications on
business performance, are moderated by other practices within the broader service
delivery system. For example, design and technical capability and resources are
essential to an effective problem response process, as is the ability to integrate
information systems to plan materials and resources in a timely and accurate
manner throughout the supply chain.
Offering advanced services brings greater risk for the service provider. Hence,
managing risk is critically important to manufacturers. As risk exposure increases,
so too does the scope for reduced revenue/profits. One way manufacturers are
controlling risk is to employ technology to provide real-time visibility of the health
and performance of equipment as it is used by the customer. What is absent from
the extant literature is a proper consideration of the technology used to enable
many product-centric service offerings from manufacturers.
Therefore, the following examples of the provision of advanced product-centric
service offerings in the aerospace, railway and road transport sectors are offered as
indicators of how technology has enabled the effective and efficient delivery of
product-centric services.
identified as the next location for routine checks and maintenance. This infor-
mation is used to identify resource requirements and trigger logistics and supply
chains before the aircrafts arrival at its destination airport or (MRO) facility.
Information is also fed back into the engine design process where design for repair
ability and maintainability play an ever increasing role in effective and efficient
service delivery.
The Alstom (TLS) customer, Virgin trains, operates the UK West Coast Mainline
under a 10 year franchise from the UK government with their focus on driving the
trains, selling the tickets and marketing their offer. It contracts Alstom to provide
47 of 52 Pendolino trains, available for use, by Virgin each day and to repair and
maintain this fleet against this availability contract. To effectively and efficiently
deliver this service Alstom has introduced a range of enabling technology solu-
tions to provide dynamic data on train status and location underpinned by an
on-board Train Management System (TMS) that monitors a range of 25 train
systems that include propulsion, tilt, high tension, braking, air, etc. using around
15 computers interfaced to various on-board sensors. Using GPRS the TMS sends
fault codes to fleet control centres staffed by experts, who communicate with the
maintenance depot staff regarding the work needed, including parts, to rectify
faults identified prior to the arrival of the train at maintenance depots where
routine maintenance and fault rectification has to be completed in a narrow time
window from around 11 p.m. to 5 a.m. The TMS data, expert analysis data and
work scheduling and parts acquisition is integrated using a sophisticated IT system
based around SAP. Alstom now consider remote monitoring of on-board systems
as fundamental to efficient train operating and maintenance.
The connection of the engine management system of the MAN vehicle using the
CANbus, allows the continuous collection of information concerning the operation
and performance of the vehicle. This data, including GPS location, is transmitted
in real-time to the MAN website using GPRS communications. The data collected
provides reports showing how the vehicle is being driven by the driver together
with the actual fuel consumption and whether optimum fuel efficiency is being
achieved. Vehicles and drivers can be compared by vehicle type, depot, route or
other parameters. Owners and operators of heavy goods vehicle fleets incur sig-
nificant costs in maintaining the necessary documents needed to comply with
legislation. Missing, misfiled or incomplete documentation can cause significant
224 H. W. Lightfoot et al.
problems and even lead to financial penalties for offending companies To simplify
and reduce the costs of the management and storage of all required documentation,
the solution offered is a managed web service hosted by MAN, that manages the
storage of all necessary documents and certification relating to the fleet of vehicles,
their usage, maintenance and inspection histories. Input of data is a mixture of
electronic transfer of data from MAN business systems and the automated scan-
ning and input of MAN dealer generated inspection and certification documents.
Operators access this via a simple web browser.
Data from large numbers of sensors, deployed in infrastructure (such as roads and
buildings) can give decision makers a heightened awareness of real-time events.
Some industries, such as chemical production, install masses of sensors to bring
much greater granularity to monitoring with signals being sent to actuators to
adjust processes. Advances in wireless networking technology, greater standardi-
zation of communications protocols enable sensor data collection almost anywhere
at any time. Products embedded with sensors, can be tracked and interactions with
them monitored. The ensuing massive increase in data generation and the need to
store and manipulate the data is facilitated by the declining cost of data storage
capacity and computing power.
Enabling technology and systems practices provide better visibility of the asset
in use in terms of condition, operating characteristics, time in use and location. The
remote access to this information for the service provider facilitates more timely
awareness of faults to provide faster maintenance/repair actions and can lead to
improved equipment design, operator behaviour and less need for manual obser-
vation of the asset in the field. The resulting improved responsiveness of the
service provider has a positive impact on asset performance and availability, whilst
the improvements in design and operator behaviour have a positive impact on
reliability. The increasing sophistication of enabling technologies and systems has
a downside in that it contributes to an increase in service delivery costs. However,
the visibility provided of asset operating characteristics, time in use and location
can help to mitigate some of the risks adopted by the service provider and relieve
contract delivery costs. Also improvements in product reliability and availability
can reduce service delivery cost, and can reduce need for manual observation/
intervention within the delivery system.
These enabling technology and systems practices, and their implications on
business performance, are moderated by other practices within the broader service
delivery system. Key here are design and technical capabilities together with
integrated information systems. This facilitates the in-sourcing of activities for-
merly provided through the external supply chain in order to manage the necessary
over capacity of resources and to off-set poor delivery response from suppliers.
As illustrated in Fig. 11.2, decisions about these other practices impact the
11 Emerging Technology and the Service Delivery 225
References
Alonso-Rasgado, T., Thompson, G., & Elfstrom, B. (2004). The design of functional (total care)
products. Journal of Engineering Design, 15(6), 515540.
Baines, T., Lightfoot, H., Peppard, J., Johnson, M., Tiwari, A., Shehab, E., et al. (2009). Towards
an operations strategy for product-centric servitization. International Journal of Operations
and Production Management, 29(5), 494519.
Baines, T., Lightfoot, H., & Smart, P. (2011). Servitization within manufacturing operations: An
exploration of the impact to facilities practices. Proceedings of the Institution of Mechanical
Engineers Part B: Journal of Engineering Manufacture, in press.
Brax, S. (2005). A manufacturer becoming service providerchallenges and a paradox.
Manufacturing Service Quality, 15(2), 142156.
Byeron, A. (2006). The right service strategies for product companies. McKinsey Quarterly, 1,
4051.
Davies, A., Brady, T., & Hobday, M. (2004). Moving base into high-value integrated solutions:
A value stream approach. Industrial and Corporate Change, 13(5), 727756.
Gebauer, H. (2006). Success factors for achieving high service revenues in manufacturing
companies. Benchmarking, An International Journal, 13(3), 374386.
Gebauer, H., & Friedli, T. (2005). Behavioural implications of the transition process from
products to services. Journal of Business & Industrial Marketing, 20(2), 7080.
Hewitt, P., Secretary of State for Trade and Industry, (2002). The governments manufacturing
strategy, 4 April 2002.
Lightfoot, H., Baines, T., Smart, P. (2011). Examining the information and communication
technologies enabling servitized manufacture. Institution Journal of Mechanical Engineers
Part B 2, 19641968.
Miller, D., & Hartwick, J. (2002). The problem of solutions: Balancing clients and capabilities.
Business Horizons, 4(2), 312. (Mar/Apr).
226 H. W. Lightfoot et al.
Oliva, R., & Kallenberg, R. (2003). Managing the transition from products to services.
International Journal of service Industry Management, 14(2), 110.
Quinn, B., & Paquette, P. (1990). Technology in services: Creating organisational revolutions.
The McKinsey Quarterly, 3, 9198.
Slack, N. (2005). Patterns of servitization: Beyond products and services. Cambridge: Institute
for Manufacturing.
Vandermerwe, S., & Rada, J. (1988). Servitization of business: Adding value by adding services.
European Management Journal, 6(4), 314324.
Voss, C. (1992). Applying service concepts in manufacturing. International Journal of
Operations and Production Management, 12(4), 9399.
Wise, R., & Baumgartner, P. (1999). Go downstream: The new profit imperative in
manufacturing. Harvard Business Review, 77, 133141. (Sept/Oct.).
Yin, R. K. (2003). Case study research: Design and methods. London: Sage Publications.
Chapter 12
Coordinating Parisian Urban Transport
Abstract A large research project has been initiated by the French government
gathering together a consortium of industrial firms (transport companies, inter-
mediaries, software companies) and researchers to study the issues pertaining to
the delivery of goods, fuel consumption, traffic congestion and pollution, focussed
primarily on the City of Paris. Initially this study was to be from a micro-
perspective, but it rapidly encompassed the logistics needs of several sectors.
The project focuses on the sharing of total transport capacity between vendors via
a market portal which will optimise the traffic intensity and so reduce the pollution
caused by the commercial logistics processes thereby increasing the quality
of life for residents, commuters and tourists. This chapter offers an outline of
the project from its conception to the current evolutionas it is still ongoing at the
time of writing in mid-2011.
M. Stumm
IFSTTAR, 2 rue de la Butte Verte, 93166
Noisy le Grand Cedex, France
J. B. Kidd (&)
Aston Business School, Aston University, Aston Street,
Birmingham, B4 7ET, UK
e-mail: j.b.kidd@aston.ac.uk
12.1 Background
From the beginning the project was conceived to optimise Parisian urban logistics.
It is known as LUMD (Logistique Urbaine Mutualise3 Durablei.e. the Sustain-
able Mutualised Urban Logistics project).
It was accepted that well organised urban logistics are essential to the dynamics
and supply of goods to cities. Thus an optimal organisation of logistics services
should serve three types of interests as the project proposals noted:
1
ADEMEAgence de lEnvironnement et de la Matrise de lEnergie (ie the French
Environment and Energy Management Agency).
2
A conference bringing together the government, local authorities, trade unions, business and
voluntary sectors to draw up a plan of concrete measures to tackle the environmental issue. The
name Grenelle comes from the first conference bringing all these players together, which took
place in May 1968 in offices in the Rue de Grenelle, Paris.
3
Mutualizationis not easy to translate exactly to English. It implies a legal connotation of risk
sharing akin to a co-operative society thus connoting joint obligations. It is a wide-spread concept
in France due to the prevalence of [mutual] insurance policies.
12 Coordinating Parisian Urban Transport 229
That said, the partners aimed to lower the kgCO2/ton delivered by 50%,
modally shift 25% logistics from the roadways, achieve 30% better filling of
trucks, and reduce their mileage by 30%.4 During the early part of 2008 the project
was conceptualised and presented to partners as one that would cost little to them
overall (no expensive changes to their warehousing or vehicles) but would offer
benefits through better overall management. The project was initiated in late 2008
to be implemented by Q4 2011 when Version 1.0 was in fact presented and
progressed through an update into Version 2.0. The live implementation will take
place early in 2012 after all the simulations and demonstrations of Version 2.0
have been signed-off by the partners.
At the centre of this study is Paris, a well-known global city: Wikipedia states
The City of Paris is largely unchanged since 1860s when Napoleon III and the Seine
prefect Baron Haussmann re-styled the streets and boulevards of the centre. Now it has an
estimated population of 2,193,031 (January 2007), but the Parisian metropolitan area has a
population of nearly 12 million, and is the largest metropolitan area in the Euro zone
It has been an important settlement for more than two millennia, and today is one of the
worlds leading business and cultural centres and its influence in politics, education,
entertainment, media, fashion, science and the arts all contribute to its status as one of the
worlds major global cities. According to 2005 estimates, the Parisian urban area is
Europes biggest city economy, and is fifth in the worlds list of cities by GDP (from
Wikipedia: Paris, accessed April 20th 2011)
Over millennia Paris was the capital of France except for brief periods when
kings relocated the capital, or when France was occupied by foreign forces.
As might be expected, its architectural and cultural heritage has frequently rede-
veloped and now Paris is defined by its wide boulevards, and also with tiny streets
that interconnect parks and small squares supporting the local communities way
of life that is loved by tourists and residents.
It is an attractive location for tourism with some 45 million visitors per year;
and as the capital city of France it supports many Fortune 500 companies and
government officesthus it has a large flux of commuters. All need to be offered
4
Smart Logistics Grid www.point-one.nl/Nieuws/en/SCS2011-17_Presentation_Poteloin
. French-Dutch innovation seminar Smart Cities Paris, 26th and 27th May 2011.
12 Coordinating Parisian Urban Transport 231
Kunzig and Broecker (2008) refer to the original Keeling Curve which demon-
strated that man had accelerated the concentration of air-born CO2 from the late
1950s onwards. They also discuss many solutions that ought to be investigated
collectively and instigated as mitigation or as adaptation. Hansen et al. (2007)
reported on changes to CO2 using a geological time-frame, and discussed the
potential for man to induce flip changes to all global systems due to the aggregate
effects of climate change. Their analysis and similar others (for instance, from the
UN or OECD) carry implications resulting from the continuance or the increase of
air-born pollutants year-on-year into the futureas the anthropomorphic global
warming (AGW) greenhouse gasses syndromewhich may cause irreversible
climate change. The warnings of The Stern Report (Stern 2007) indicated that if
we collectively do nothing the planet will be in grave danger, but for relatively
little cost to our collective GDPs we can all contribute to a better future; and the
Intergovernmental Panel on Climate Change (IPCC 2007) in their report Climate
Change 2007: The Physical Science Basis assessed the scientific knowledge about
the natural and human drivers of climate change. Further, the UN has organized
meetings under the United Nations Framework Convention on Climate Change
(UNFCCC5) at Rio de Janeiro (1992the Rio Earth Summit), Kyoto (1997), Bali
(2007), Copenhagen (2009) and Cancun (2010) in order to obtain voluntary
agreements by nations heads on reductions in their pollutants. These meetings had
only a moderate success brought about mainly by arguments between the devel-
oped and developing nations, the rich and the poor, and who would carry the costs.
Stern suggested in 2007 all should pay, but pay only about 1% of their GDP; but
now, through doing too little, might cost us 1020% GDP as we muddle through,
too late, with chaotic systems change.
5
Details may be found at http://unfccc.int/2860.php
232 M. Stumm and J. B. Kidd
Turning to national issues we note that each country has a different energy mix
based on its varying use of primary fuels [nuclear, coal, oil, gas, hydro, etc.] that
contribute to its characteristic total carbon footprint. This not a constant but alters
over time as new power generating systems are commissioned and older dirty
systems are decommissioned. For instance, the UK was once heavily dependent on
fossil fuels for its electricity generation but now only about 22% of its generation
comes from coal (the dirtiest fuel). Even so, electricity generation remains the
main source of UK pollution, with road transportation and housing as the second
and third major sources. France at present is almost totally dependent for its
electricity on nuclear fuels (84%) plus hydro-electric generation (6%): even the
household electrical bills of EDF (Electricit de France) proclaim its green cre-
dentials noting only 4% of its electricity is generated from coal thereby yielding a
pollution loading of *50 gmCO2eq/kWhin contrast, the UK creates 891
gmCO2eq/kWh from its electricity generation (WNA 2007). Therefore, in France,
the pollution from road transport is first in line, and housing is second, assuming a
much greater proportional significance than in the UK. There is thus a somewhat
heavy French government pressure on its transport sector, though to be fair, such
pressure has been exercised since the Grenelle of 1968.
Measures have been taken to achieve a 20% reduction overall in French CO2
emissions by 2012 through enacting new regulations carrying a hoped-for substi-
tution from road transport to alternative transport. As an aid, waterway and railway
transport would be further developed (planning for an increase of 25% volume of
freight in each mode) thus hopefully removing many lorry trips from the road by
2020. This too is an aim of the EU-wide Marco Polo II programme.6 However this
may not occur as we note that while road pollution (Le Havre to Paris) versus river
pollution in general terms is double, the road journey is only 195 km versus 335 km
by river, and river traffic is intrinsically slow (ADEME 2008). As with many
government announcements, the French government has made several rules, sug-
gestions, and laws with different actionable datesas a result its avowed goals for the
mitigation of pollutants is unlikely to be met (Rseau Action Climat France, 2001).
This action group maintains a scathing view today7similar in fact to the review of
the national limits versus daily pollution levels in or near Paris as presented by
AIRPARIF (Surveillance de la qualit de lair en Ile-de-France).
Notwithstanding this confusion the French ecology minster Nathalie Kosciusko-
Morizet wishes to ban older vehicles from the centers of cities.
saying the future is in small cars and electric vehicles so she is setting up a Zone
dAction Prioritaire pour lAir (Zapa) which will allow the banning of 4 9 4 s, older cars,
motorbikes, scooters, lorries and buses for a three-year test. She wants to take 10 million
vehicles off the roads including 8 m cars over 13 years old, 1.6 m motorbikes and scooters
and 300,000 lorries and buses which are too old to have particulate filters fitted. So far
6
http://ec.europa.eu/transport/marcopolo/
7
See http://www.rac-f.org/IMG/pdf/Analyse_synthetique_Grenelle_1_apres_1ere_lecture_AN_
et_Senat.pdf.
12 Coordinating Parisian Urban Transport 233
eight cities have agreed to join the test: Paris, Lyon, Bordeaux, Nice, Grenoble, Clermont-
Ferrand, Aix-en-Provence and Saint-Deniswhich will start in July 20128
Italy has also joined this discussionwe note the Case of Vicenza and the
potential implication of the Italian Courts judgment on EU Lawsas that city
implemented very restrictive regulations on logistics providers entering the city
centre (Ville et al. 2010).
Paris is served well by national and international logistics services (road, rail,
waterways and air). And, as is the case of all major cities globally, its economic
growth and the intensity of trade in goods (visible and invisible) are central to the
citys opportunities. However, greater trade is a threat on account of its negative
impact on the environment given it usually creates more pollution and greater
climate change effects.
The City of Paris in reality is similar to a pumped-up village as it rests wholly
within its defining ring-road (the Priphrique) inside of which are the wide
boulevards and a myriad of little roads that are frequently blocked by traffic of
many types. Although many offices are inside the City, some are just outside the
Priphrique (e.g La Dfense): yet all contribute to the commuter stream of Paris,
as do the logistics operations needed to service residents, commuters and busi-
nesses. From a logistics providers standpoint any increase in the congestion of
roadway infrastructures is of concern, particularly in urban areas delivering goods
to the last mile.
When it comes to current practices in the logistics sector, we find that:
many vehicle trips on the French roads are made with an empty cargo area (load
bay),
numerous warehouses have unused storage capacity,
intermodal transport is weak,
the capacity of rail transport RER [Regional Express Railway] and the Metro
subway at night-time (and waterways) are underutilized, and
8
Old cars could be banned from cities. http://www.connexionfrance.com/Old-vehicles-banned-
Kosciusko-Morizet-ecology-pollution-asthma-Zapa-2CV-12651-view-article.html. April 7th, 2011.
9
http://www.politicsforum.org/forum/viewtopic.php?f=14&t=127412
234 M. Stumm and J. B. Kidd
Thus the specific characteristics of urban areas which concentrate the activities
of final distribution require high levels of service and reliabilityhigher than that
required for [simple] optimisation services in inter-regional or inter-city contexts.
In the general urban environment we find that:
the density of logistics flows is very high
all current distribution systems are extremely fragile because the logistics
operators (delivering goods) competitively use infrastructure that is often
deemed more important for the transport of people, as these are optimised for
people-flows
the development of capillary distribution requires logistics space on several
hierarchical levels: logistics zones around the metropolitan area, distribution
centres at the approaches to city centres, and local (micro) receiving points.
None at present have good inter-communication.
Goods optimisation must take into account the following needs:
to have real-time responsiveness, since the availability of capacity varies
greatly, especially in downtown logistics areas
large volumes and many types of information are needed to ensure multimodal
pooling from end-to-end
geographically referenced databases need to be linked to transport capabilities
(duration, cost, real-time hold-ups) over the logistics networks
12 Coordinating Parisian Urban Transport 235
We define:
An Urban Logistics Space (ULS) as a set of facilities at break bulk points,
having resources, and systems intended to optimise the circulation of goods in
cities, both from a functional standpoint and an environmental standpoint. ULSs
are devoted to recomposing the flows that circulate throughout the city by offering
services specific to logistics professionals (storage, handling, accommodation of
their vehicles, etc.). ULSs can assume several aspects and satisfy different
objectives, depending on their degree of coverage.
It is not enough to put in place ULSs to solve only the logistics problems of
goods transportation in citiesthey must be focused on the local need to adapt
logistics structures to a particular urban context. A detailed analysis of local
dysfunction would make it possible to define the needs and characteristics of a
ULS prior to implementation (OECD 2002).
Furthermore we ought to:
Support the use of ULSs by stricter regulations and the involvement of the
necessary governance to ensure compliance.
Guarantee the creation of public subsidies to cover a portion of the ULS service,
at least for its start-up phase. The success of a ULS will be assured along with its
financial equilibrium if the price of the service is consistent with the market.
Govern the project with a neutral regulator to facilitate cooperation.
We note that these three points are the subject of wide discussion in several
environmental sectors because regulation and feed-in tariffs are found to desta-
bilize existing structures and cause unforeseen consequences (cf. photo-voltaic
installation and feed-in tariffs have been too enthusiastically applied in advance of
the decommissioning of older [polluting] electricity generation systems).
It is a tradition that goods arriving near to Paris are transhipped to new carriers at
metropolitan warehouses where the goods are broken-down to smaller pallet
groupings. These warehouses are some 50100 km distance from Paris and are
236 M. Stumm and J. B. Kidd
case is little different to any major capital city. Carriers use a variety of modes to
deliver goods to the end-user or to a safe-deposit point for user pick-up (even using
the cross-Paris RER rail system, or the Metro [underground railways] at nights).
However, the demand for goods is often in conflict with the need to maintain the
service infrastructure: the Parisian roads are often repaired at night time; and
maintenance, safety inspections or the replacement of sections of the urban rail-
ways takes place at night. In these cases the theoretical infrastructure is disrupted
and delivery mode flexibility hampered.
Paris and its metropolitan region combine airways, waterways, railways, and
roadway resourcesthe logistics service offered is therefore complex and highly
heterogeneous, and delivery is often solved piecemeal by individual firms:
The pooling of resources and transfer from roadway mode to alternative transport modes
still rely on local initiatives. Monoprix, for example, has added a multimodal dimension to
its own logistics delivery plan in order to reduce environmental impacts. Their logistics
resources combine waterway, railway, and roadway transport. In 2006, 83% of its goods
arrived in Le Havre (a sea port) then travelled up the river Seine to Paris. In 2008, non-
food products and non-alcoholic beverages used the RER railway network, and were then
carried by their own vehicles over the last kilometre (Dablanc 2009).
According to PREDIT (2007) existing information systems are now unable to keep
up with changes in logistics business models or the intensity of modern data
exchanges. In addition, 70% of information exchanges still do not take place in
real time, and only 9% of logistics specialists are equipped with an Electronic Data
Interchange (EDI) system. This state of affairs has altered little since the period of
19951998 when data was collected across Europe for the COST 330 study (COST
1998). In theory a cooperative information system would allow numerous com-
panies to move towards logistics pooling and integration into collaboration net-
works. Therefore, developing such operational tools is crucial for the future of
10
http://www.logistique-urbaine.com/?page=consortuim
238 M. Stumm and J. B. Kidd
businesses so they may better manage the supply and demand of their logistics
services and optimise allocations. This is the case for Paris, it is not an exception.
A [simple] e-commerce application such as a Freight Exchange Portal allows
shippers to find goods and avoid empty return trips. Freight Exchanges, however,
are limited to establishing connections and they do not allow users to plan pooling
projects long in advance. Nor do they allow one to simulate and choose between
various opportunities: the only evaluation criterion for sharing [roadway] transport
capacity is cost at the instant a comparison is needed. Furthermore, Freight
Exchanges do not give any indication of the CO2 footprint of potential choices.
An important commercial aspect is data privacy, and the project managers of
the LUMD development constantly stress this aspect while also emphasising that it
is only through securely managed optimised [global] data that an operator can
judge a proposed plan. Financial responsibility will also be addressed as many
finely nuanced cash flows will be managed by the projects software. Finally, with
respect to the development of the proposal, the project managers stress that their
governance will have a broad reach including the legal responsibility for the pick-
up and handling of other operators goods using third party vehicles or storage
space. This is in fact a very knotty issue in practice, as well as in terms of the law.
11
Its description may be seen at http://www.corail-software.com.
240 M. Stumm and J. B. Kidd
Fig. 12.2 LUMD core modulesmarket place and business process module
1. Market Place [Place du March (PM)] for pooling and negotiating logistics
capacities.
2. Business Process Management module (BPMM) to optimise planning and
forecasting for winwin cooperation.
Figure 12.2 illustrates the logic of LUMD with its proposed operations marked
from 1 through to 8 as interactions of costing, simulation and transaction interplay
within the system.
Selling Agents: these agents represent logistics players having surplus logistics
capacities for transport and warehousing.
Coordination Agents: these agents manage interactions, cooperation, negotiations,
and the security of transactions between Purchasing Agents and Selling Agents.
Search Agents: will use a learning algorithm (multiple-criteria decision-making
assistance, economic models, Web page labelling mechanisms, etc.) and
will find better pathways to a given piece of information. Each Search Agent
will learn to search for products as best as possible and will launch automated
pre-negotiation tasks. This learning will form a collective memory within the
community of Buyers and Sellers.
Negotiation Agents: Depending on specific communication protocols complying
with the logistics constraints these agents retrieve the list of Seller Agents
offering potential logistic over-capacities in order to conduct negotiations with
several Buyer Agents at the same time. Each agent participating in the nego-
tiations will have its own preferences defined with the help of a preference
template, and negotiations will be conducted with the help of multi-criteria
auction algorithms. In the event of a logistics pooling agreement, the Negotia-
tion Agents finalise the transaction.
After the needs of a logistics player have been identified, the Market Place
module searches for the offers or requests to meet the identified needs. A search is
made both in the announcements available on the LUMD platform and on Web
pages in order to collect as much information as possible on the logistics solutions
meeting the needsi.e. capacities involved and pooling policy (logistics schema,
geographical location, etc.). This step allows the logistics player to compare
various alternatives and the optimum characteristics for transport pooling or goods
storage. Although this seems rather like an instantaneous pooling portal the aim
of the LUMD project is to provide a long look-ahead to allow individual firms to
plan reasonably, and to broker better logistics mechanisms.
The automated negotiation performed by the Market Place module includes
various types of negotiation such as multiple item auctions in which the bidder
negotiates an indivisible set of items which may or may not be identical; and
multiple attribute auctions in which the bidder negotiates according to several
attributes involving not only the price but also other attributes, such as logistics
load-mix characteristics, capacities of vehicles and/or storage, CO2 footprint, and
kilometres travelled.
Many of the preference relationships used in multiple-criteria auctions are
based on an additive aggregation model in which the buyer defines preferences by
choosing a weight for each criterion so representing its importance. The robustness
of such an additive model in this case must be researched: it is intuitively simple to
comprehend, yet more complex relationships may be more valid. For instance, an
approach based on the Tchebychev aggregation model makes it possible to satisfy
the compensation aspect of the additive aggregation model, and it gives the bidder
control over the negotiation process for each attribute (Gonzlez-Pachn and
Romero 2006).
242 M. Stumm and J. B. Kidd
It can be seen that the Market Place module will be used to organise and
manage logistics service pooling exchanges. This module will create value by
consolidating flows and optimising the use of unused capacitiesits functions can:
Consolidate offers and requests for logistics service capacity,
Link pooling opportunities to requests via transaction arbitration in an auto-
mated and optimised fashion,
Negotiate prices in real time in a transparent, equitable, and automated way by
estimating the value of the vacant spaces on the market, but also on other
characteristics such as alternative CO2 footprints. Effectively it will be based on
yield management principles12getting the best satisficing13 solution at the
current instant according to marginal profit opportunities.
12
Yield management http://en.wikipedia.org/wiki/Yield_management#Econometrics.
13
http://www.businessdictionary.com/definition/satisficing.html
12 Coordinating Parisian Urban Transport 243
approaches will be explored. The heuristics will provide a sensible cut-off rule
against over-long computational searches for the tiniest of gains as optima are
approached; and the objective function will be based on multiple criteria (cost
minimisation, CO2 emission reduction, empty-load distance minimisation, quality
improvement, and shorter operational times). The pairing of simulation and
optimisation algorithms will make constraint-based planning and pooling man-
agement more dynamic and adaptive.
Decision-making assistance algorithms and methods will be used and managed
by the multi-agent models. This combination will provide the necessary elements
of expertise so that the cooperative information system and Business Process
Management will generate coherent decisions and relevant arbitrations in real time.
Consequently the Business Process Management module will be dedicated to
manage collaboration and the pooling of data (communication, computerisation of
exchanges, interoperability, etc.). It will also simulate pooling scenarios to aid
intelligent planning, and it will be self-adaptive with real-time optimisation of
processes and flows. The technology and software architecture of the BPMM is
based on Service-Oriented Architecture (SOA), Full-web service, and JAVA/J2EE
(i.e Java 2 Enterprise Edition).
The Business Process Management module will be the first cooperative infor-
mation system dedicated to the pooling of multimodal logistics. The innovation
of the module derives from the coupling of complementary technologies of both
artificial intelligence and operational research: it is a challenge that the consortium
must meet, with its R&D work taking place at the intersection of two comple-
mentary research fields:
Agent Technology: this positions itself as an emergent paradigm for the design
and development of communication systems and cooperative, self-adaptive
applications endowed with intelligence.
Optimisation, Simulation and Decision-Making Assistance: the purpose of this
technology grouping is to develop constraint-based arbitration and simulation
tools for arbitrating between pooling opportunities, evaluating the economic and
ecological performance of different scenarios (relevance of roadway-alternative
modes, CO2 savings, tolls, etc.), managing and optimising collaboration and
pooling in real time (automated planning of multi-player delivery, warehousing
consolidation, synchronisation of processes, modal transfer, etc.) and better
rationalising of process management throughout the supply chain.
12.4.2 In Practice
While all parties accepted the LUMD model proposal, in practice there is bick-
ering. Meetings are held regularly within the main groups of actors (carriers,
software developers and ministry officials) and also between groups. The carriers
244 M. Stumm and J. B. Kidd
think progress is too slow as they wish to proceed to a fuller aggregation mode as
quickly as possible (their costs continue to rise); the software developers might
seem too esoteric (looking too deeply for novelty); and the ministry personnel are
always looking for value for money (requiring constant justification of project
costs). Outside influences also have to be noted, such as the EDI Standards
Committee whose rules govern the logical processes of freight mixing and
matching in vehicles.
14
See http://www.unece.org/cefact/.
12 Coordinating Parisian Urban Transport 245
account new business processes not just for LUMD and Paris, but globally. This dual
presentation (of logic, as well as EDI message notations) circumvented the usual
blustering by different national members as the logical need was clear. Nevertheless,
the Procedures of Standardization had to be followed, with members taking pro-
posals back to their national discussion groups before final ratification in Geneva in
September 2010, and in Washington in March 2011.15 Often it is vacuous discussion
that makes any Standardisation processes so long-winded: in this case the wrangling
was circumvented through the clarity of the ontological argument.
The key words of the LUMD project are end delivery, mutualisation and
sustainability. These were not originally taken into account in the UN/CEFACT
standards. The LUMD project presents an opportunity for the relevant Stan-
dardisation Body to update its standard components to support a real business
environment.
The multi-pooling of urban logistics service provisioning depends upon the cross-
pooling of different optimisation layers wherein there are isolated over/under-
capacities (transport and storage). The optimisation demands the elimination of
under-utilized resources (routes not used by private services), as well as using
public transportation networks and tourism routes during their low or out-of-
service times, and linking these with small-scale entrepreneurial initiatives. Within
the LUMD platform this pooling will be made possible by linking offers and
requests according to pairing rules (qualification, real-time consolidation, and
aggregation of needs).
Optimisation of the urban logistics scheme implies a certain degree of com-
plexity and must therefore be accompanied by a more complex innovation in
service offerings as a complement to the technological innovationeffectively
following Ashbys Law of Requisite Variety (Ashby 1956). Innovation in service
provisioning is based on the creation of an urban logistics integrator. The objec-
tives of this organisation are not only to ensure the operation of the LUMD
platform, but also to provide a rich selection of high-performance services. The
required shifts in optimisation modes are illustrated in Fig. 12.3. A while ago
consolidation of loads might only have been linked by stronger Service Level
Agreements (SLAs) in the #1 aggregation sectorwhich we note is a less strong
version of the LUMD mutualisation goal. Now we hope that the system moves to
sector #2 costs optimisation wherein there is a joint optimisation of logistics and
environmental costs (through the mass consolidation of ULS) and the minimisa-
tion of pollution. Then the LUMD ought to move the optimisation into sector #3
which will achieve better SLAs for all concerned.
15
Changes may be followed at http://www.unece.org/trade/untdid/directories.htm. See Direc-
tories 10A, 10B and 11A.
246 M. Stumm and J. B. Kidd
In the early stages of the project it was not possible to define explicitly the
algorithms to be used by the software teams. However, some will be inclined to
GML (Geographical Mark-up Language) to define routes and special objects along
these routes. In a GML document, spatial objects are classified into three cate-
gories: point objects, line objects and polygon objects. For example, in a given
GML document describing spatial objects, such as houses, shops, hotels, roads,
rivers, lakes, and so onbuildings belong to point objects; roads and rivers, etc.,
belong to line objects; and lakes and residential quarters, etc., belong to polygon
objects (Ng and Han 2002; Guha et al. 2000). These descriptions were un-natural
for logistics providers and might have seemed esoteric (i.e. the fantasy language of
software engineers) but they have been invaluable in simplifying complex map
data.
There were many other hurdles to overcome, not least breaking through the
existing logistics opacity as there is a natural reticence about discussing secure
commercial data openly with competitors. This has always been an issue in
co-opetition environments wherein some aspects of a firms product range is
combined with that of a competitor so they may co-operatively exploit a market
segment while in other product lines they remain competitors (Brandberger and
Nalebutt 1996). In LUMD we foresaw situations where someones delivery
carried by a competitor will involve that competitor being late with his own
delivery (at least according to an original forecast). However, aggregating solu-
tions over time ought to show that far from individual carriers being disadvan-
taged, all in fact achieve winwin.
We accept that situations become convoluted when considering the legal
ramifications of carrying others goods and taking responsibility for them (insur-
ances against damage or loss, guaranteeing pick-up, and timely delivery, etc.).
We imagine that ontological developments such as those desired in software
purchasing might be invoked later between the multiple partners in the Parisian
context. Asfand-e-yar et al. (2009) propose a two-party ontology (in their case, for
end-user as well as the software seller) that obviates the end-user having to read in
detail (or claiming to have read [thus agreeing]) the small print at the time of
software purchase. The ontological identification of needs and offers in LUMD
12 Coordinating Parisian Urban Transport 247
enable matches to be found quickly that take into account secondary or tertiary
conditions. For instance, in the above software case, when a purchased module is
re-engineered to be followed by onwards selling of the redeveloped software in
good faith, it may be found the proposed sale might be forbidden to some nations
users of the first-instance softwarethus banned at the second-stage sale. In the
Parisian real-time logistics activity of LUMD there will be insufficient time to
absorb the legalese of contracts, thus an enabling ontology (such as the above) will
make partnerships more transparent and hopefully more comprehensive at the
same time. This is supported by Hette et al. (2010) who also discuss the gener-
alities of LUMD.
12.5 Discussion
At first sight the LUMD project seems to be concerned only with the last kilo-
metre/mile problem related to local/home purchase patternsfor instance,
Edwards et al. (2009), who found on average, when a customer buys fewer
than 24 items per traditional shopping trip (or fewer than 7 items for bus users) it is
likely that the home delivery [via a consolidated delivery truck] will emit less CO2
per item purchased. They continue both consumers and suppliers need to be
better informed about the environmental implications of their respective pur-
chasing behaviour and distribution methods. Gevaers et al. (2009) also give an
overview of new last mile concepts, including innovative break-boxing of
standard containers (TEU/FEU) to deliver onwards by small truck to local dis-
tribution points, and even to homes (also see Ocado.com); but they did not discuss
consolidation algorithms or the application of optimisation techniques. Others
discuss food miles, but this is too simplistic suggests Saunders et al. as it does not
consider the total energy use in the production and delivery of the product. They
go on to say this report has shown that in the case of [New Zealand] dairy and
sheep meat production NZ is by far more energy efficient (even including the
transport) cost than the UK: twice as efficient in the case of dairy, and four times as
efficient in case of sheep meat (Saunders et al. 2006). They continue:
A number of caveats should be raised when interpreting these results. The most important
of these was the lack of comparable data between the countries and more importantly the
lack of data in particular for the EU countries on production systems and their energy use.
A second important caveat is that the analysis assumes that the EU would be able to meet
the shortage of supply if NZ did not supply the EU market. It also assumes that this can be
done using the same levels of inputs currently used. This may well not be the case
especially for important products where NZ supplies significant parts of the market
especially out of season. To supply these would mean that EU land would have to be
diverted from other uses and this land is unlikely to be of the same quality as existing land
producing the product and therefore may well require greater inputs. Moreover, in some
products such as onions it is doubtful that the technology exists to be able to store the
product to match the same window of supply as NZ produce can. Finally the report has
also assumed that the products to replace NZ imports are of similar quality and type, this is
clearly not always the case (Saunders et al. 2006, p. 93).
248 M. Stumm and J. B. Kidd
The Parisian problem is classical for many businesses in so far as they look
for frequent small deliveries. The foremost issue within the LUMD project is
mutualisationit is about the consolidation of several orders from several sup-
pliers to many independent vendors who, through this process, would share a
warehouse and/or a delivery vehicles space thus reducing their total urban
logistics space. Well before the global economic crisis began in the United States
in 2007 many people in Paris were demanding better services at a lower [eco-
logical] cost. Tourists, commuters, and local residents found that shop keepers had
increased their public floor space to promote more sales: but through these changes
the shop owners found their stockrooms had become tightly constrained. This was
because many buildings could not be modified to increase their physical volume
within the space and aesthetic constraints of the city. Often the buildings, even
whole streets of shops, had become national treasures and as listed buildings
there was little that owners could do to increase their fixed volume of sales-floor
plus stock-space. Shop managers looked to ever more frequent Just-in-Time
deliveries, but these became enmeshed in the grid-lock of Parisian traffic so
lengthening delivery time and increasing urban pollution.
Later, as the financial crisis radically altered purchasing habits the vendors and
logistics forwarders found load compositions had changed; and also they observed,
greatly reduced demand volumes. They looked for new solutions to be of benefit to
themselves and for their client population. There was also a constant awareness of
the increasingly onerous European Community regulations on pollution, modes of
work, and so onin general, upon all aspects of Health and Safety. These regu-
lations have become costly to implement within supply chains, in part because
some are inappropriate in local contexts and have been enacted by an ill-informed
legislature. Similar arguments are clearly presented by McKinnon who quotes
the business mantra if you cant measure it, you cant manage it is especially relevant
in the present context. The quantity and quality of macro-level freight data is very patchy.
Even in those countries, such as the Netherlands, the US, the UK and France, with large
freight datasets, there are significant gaps. In most developing countries there is a serious
dearth of freight statistics. Data deficiencies are even greater at an international level and
this is a particular concern as an increasing proportion of freight movement is cross-
border. Lack of data is therefore impeding macro-level analysis of the carbon intensity of
freight operations (McKinnon 2008, p. 2)
This does not justify a lax approach to the Parisian case. Rather it presumes that
the LUMD project must attempt to optimise carefully with potentially fragile data
(over many [fuzzy] attributes), and must eventually offer a retrospective evaluation
of what was achieved by the project.
To enable the pooling of resources and the consolidation of flows it is essential
to control overall performance with distinctions made between three increasingly
sophisticated levels of service, illustrated in Fig. 12.3, as one moves to better
service guarantees with reduced overall costs:
Consolidation: Pooling of logistics service offers and requests allows the seller/
buyer to organise and optimise its transport of goods (vehicle, route, date, and
12 Coordinating Parisian Urban Transport 249
time). The seller/buyer searches for the offer corresponding to its request on the
platform. Under the consolidation conditions, the owner [of the goods] is
responsible for the goods and transport conditions from end to end though there
are complex legal issues here as multi-carriers will be involved.
Cost optimisation: Pooling logistics of service offers and requests is useful input
for the urban logistics integrator, which offers an optimum route to the logistics
service requester. Under these optimisation conditions, each owner [of the
goods] is responsible for the goods and transport conditions.
Cost and service optimisation: this provides end-to-end management of
responsibilities (administration/maintenance of the technological platform and
embedded field-specific intelligence), and a guarantee of results. Under condi-
tions of full optimisation guaranteed by the urban logistics integrator, the
logistics service requester takes advantage of a turn-key service and can thus
focus better on its needs.
The shift towards better integration and efficiency ought to improve the
quality of life for the Parisian. And if the LUMD programme works well it may
perhaps be exported to other high density logistics regions of France such as Lille
or Lyon; or further afield.
12.6 Conclusion
Within the LUMD project much has been promisedas is usual with all large
scale software projects. Soon however individual modules will be proved and
merged so that simulations can be made with a view to giving early oversight
of potential solutions and interfaces to the actors in the Parisian logistics
system. By planning iterative developments it is hoped that individual differ-
ences can be ironed out early in the development phases, rather than having a
solution imposed later which may be proved to be unworkable. As we noted
above, simulations within the ontological development have led to new data,
and new codes for updating the UN/CEFACT standards requirements which
have been accepted by the UN/ECE committees in September 2010 and in
March 2011.
Until the Parisian actors (carriers, vendors and Ministry officials) see that
logistics transparency (in the LUMD sense) really can work they will remain
sceptical. And, of course, if the managers of the system governance are weak the
whole project will be at risk. The systems managers are not only to be product
champions, but they are also to be governors with teeth who can manage the
rules and ensure that the correct modalities are maintained by the Parisian logistics
community. For the moment all actors are looking forward to a solution being
provided for an increasingly difficult delivery problem.
250 M. Stumm and J. B. Kidd
References
A H
An agentified knowledge, 23 Healthcare, 105113, 118120
Heuristics, 243
Holistic cost, 145, 146, 152, 153, 158, 160
C
Complexity, 168
Containment resources or facilities, 165 I
Cooperative game, 4749, 54, 55, 5961 ICT, 211
Cost of outsourcing, 145, 146, 153 Inventory, 4753, 60
Critical to customer requirements, 196
Cross-docking, 16, 10, 16, 20
K
Key process input variables, 196
D Key process output variables, 196
Design for six sigma, 189, 190
Distributed decision
making (DDM), 89, 90 M
Disturbances, 145, 146, 150154, 156161 Model, 2332, 44, 45
Modeling, 23
Multi-level supply chain, 47, 48, 60
E
Enabling technology, 211, 224
O
Optimization, 45
F Order fulfillment model, 195
Food retail supply chain Order fulfillment process, 189
concepts, 89, 90, 94, 99 Outsourcing, 145149, 151161
Freight optimisation, 227
Freight villages, 123, 129
Fuzzy logic, 197 P
Partial integration, 190
Perfect order index, 193
G Performance measurement, 193
Genetic Algorithm, 1, 10, 20, 21 Perishable food supply chains, 63, 84