International Journal of Logistics Research and

Applications

ISSN: 1367-5567 (Print) 1469-848X (Online) Journal homepage: http://www.tandfonline.com/loi/cjol20

Bifurcation analysis and synchronisation issues in

a three-echelon supply chain

K. R. Anne , J. C. Chedjou & K. Kyamakya

To cite this article: K. R. Anne , J. C. Chedjou & K. Kyamakya (2009) Bifurcation analysis and
synchronisation issues in a three-echelon supply chain, International Journal of Logistics Research
and Applications, 12:5, 347-362

To link to this article: http://dx.doi.org/10.1080/13675560903181527

Published online: 30 Sep 2009.

Submit your article to this journal

Article views: 123

View related articles

Citing articles: 5 View citing articles

Full Terms & Conditions of access and use can be found at

http://www.tandfonline.com/action/journalInformation?journalCode=cjol20

Download by: [Tsinghua University] Date: 25 April 2017, At: 09:46

International Journal of Logistics: Research and Applications
Vol. 12, No. 5, October 2009, 347–362

Bifurcation analysis and synchronisation issues

in a three-echelon supply chain
K.R. Anne*, J.C. Chedjou and K. Kyamakya
Institute for Smart Systems Technologies, University of Klagenfurt, Klagenfurt, Austria

(Received 12 December 2008; final version received 25 April 2009 )

In today’s global marketplace, supply chains are dynamic and volatile. This dynamic and volatile nature
produces various types of uncertainties along the supply chain, for example, demand uncertainty, supply
uncertainty, delivery uncertainty and forecasting uncertainty. These uncertainties make supply chains com-
plex and nonlinear systems as they propagate both upstream and downstream. This work investigates the
dynamic behaviour of a three-echelon supply chain. Modelling of this structure is carried out to display its
nonlinear dynamic behaviour. It is shown that the dynamics (stability) of the supply chain are very sensitive
to external uncertainties. Specifically, a supply chain subjected to these uncertainties can exhibit strange
and undesired states such as saturation and chaos. An adaptive algorithm for the automatic cancellation of
these strange dynamics due to uncertainties is developed by re-adjusting the internal parameters of the sup-
ply chain to achieve its synchronisation. A bifurcation analysis is carried out. This analysis is essential and
useful for strategic decision-makers as it allows both the visualisation and control of the states/dynamics
of the entire supply chain. The internal parameters of the supply chain are used as control parameters, and
various remarkable states are discovered towards the achievement of synchronisation.

Keywords: supply chain synchronisation; supply chain control; bullwhip effect; chaos in supply chains;
bifurcation analysis in supply chains.

Introduction

Supply chain network (SCN) management has been defined as the management of upstream
and downstream relationships with suppliers and customers to create enhanced value in the final
marketplace at less cost to the supply chain as a whole (Jüttner et al. 2007). Today’s global
marketplace is increasingly dynamic and volatile. This dynamic and volatile nature produces
various types of uncertainties along the supply chain, for example, demand uncertainty, supply
uncertainty, delivery uncertainty and forecasting uncertainty. In addition to these uncertainties
caused by external sources, uncertainties observed on a daily basis (e.g. machine breakdowns,
wrong supplies and supply shortages) make supply chains much more complex systems. The
uncertainties propagate along the upstream and downstream of the entire supply chain leading
to the production of various nonlinear dynamic effects. In addition to these uncertainties, the

*Corresponding author. Email: [email protected]

ISSN 1367-5567 print/ISSN 1469-848X online

© 2009 Taylor & Francis
DOI: 10.1080/13675560903181527
http://www.informaworld.com
348 K.R. Anne et al.

relationships between the various players in the supply chain are often characterised by mistrust
and competition.
Moreover, inventory is generally used as insurance against the uncertainties. In the case of
a single enterprise-based supply chain it is relatively easy to overcome the uncertainties with
properly sized inventories at each stage such as raw materials, work in process and finished goods
inventories. The present-day statistical tools and forecasting methods can satisfactorily aid in
determining how much must be held to satisfy the customer demand for the particular product
despite the uncertainties (Davis 1993). However, the problem is much more complicated when
considering the whole network consisting of different players distributed globally. In practical
terms, each player holds some inventory to protect against uncertainties, but the real difficulty
is in determining how much must be held and where to hold it. To date, there has been no clear
analytical way to calculate the propagation of uncertainties up and/or down the supply chain.
Traditionally, firms have relied on experience and intuition in facing uncertainties.
Decisions made with intuition can make the supply chain exhibit various chaotic states. Chaos
is defined as an aperiodic, unpredictable and bounded dynamic in a deterministic system with
sensitivity dependence on initial conditions (Crawford 1991). Chaos is a disorderly, long-term
evolution occurring in deterministic nonlinear systems. The beer game developed at MIT to
introduce students and industrialists to the concepts of economic dynamics sheds further light on
supply chain dynamics (Sterman 1989). It has been found that one in four management teams in
the supply chain creates deterministic chaos in the ordering pattern and inventory levels (Wilding
1998). This clearly demonstrated in practice the occurrence of chaos in supply chains.
The objectives of this paper are threefold: (i) to show the potential of the concept of nonlinear
dynamics and modelling in SCNs; (ii) to show the challenge of synchronisation in SCNs; and (iii)
to offer to strategic decision-makers an approach or technique to control and stabilise the states
of their SCNs when subjected to uncertainties.

Responding to uncertainty

The objective of any player in a SCN is to achieve maximum profits and give maximum customer
satisfaction. In addition to the stated individual objective all players also have a responsibility
towards the global objective of resilient SCNs. Christopher (1992) emphasises this by stating
competition in the future will not be between individual organizations but between competing
supply chains.
This realisation has made the entities in SCNs look beyond their own boundaries, to assess
how the resources of each other can be utilised to achieve the global objective without com-
promising their own individual objective. As a result, entities have relentlessly restructured and
re-engineered their internal organisational boundaries and policies with an objective of trans-
forming their relations from ‘arm’s-length’ relationships to ‘durable arm’s-length’ relationships
(Skjoett-Larsen et al. 2003).
The definition of a truly efficient supply chain is when all players involved are communicating
correct data – a manufacturer is communicating the correct product information and receiving
accurate purchase orders; a retailer is receiving the specific products that were ordered; and the
product is available to the end consumer at the right time and at the right price. However, commu-
nicating the correct data is not always possible in supply chains as each stakeholder has different
objectives and constraints. Traditional supply chain management has been based on limited infor-
mation sharing restricted to the product in consideration and transaction-oriented towards that
product (Ayers 2006). It is well understood (Davis 1993, Chen et al. 2000, Gunasekaran and Ngai
2004, Hwarng and Xie 2008) that information sharing, demand patterns, ordering policy and lead
 International Journal of Logistics: Research and Applications 349

time have a direct impact on the performance of supply chains. Information sharing can reduce
the lead time. Lead time reduction is found to be highly beneficial and can reduce inventory and
demand variability and improve customer service (Chen et al. 2000).
However, during the process of identifying the ways to mitigate the effects due to uncertainties,
companies within the supply chain have realised that they need to achieve the self-organisation of
the supply chain to which they belong so as to satisfy the stated objective. In accordance with the
greater focus on the self-organisation of the supply chain, companies are increasingly focusing
on the prerequisites such as integration, collaboration and synchronisation between all entities in
the supply chain as shown in Figure 1.
The first step towards self-organised supply chains is the integration stage. A complex corporate
structure demands that the various functional units within a company be integrated first. The intra-
entity integration, also called functional integration, is the basic driver towards the integration of
the entire supply chain. The functional integration of purchasing, manufacturing, transportation
and warehousing activities gives the very much needed visibility to the supply chain. Besides
functional integration, inter-temporal integration, also called hierarchical integration, of these
activities over strategic, tactical and operational levels is also important (Jespersen and Skjoett-
Larsen 2005, Ayers 2006, Anne et al. 2008a). This inter-temporal integration requires consistency
among overlapping supply chain decisions at various levels of planning. However, the major
role of inter-temporal integration is in designing the supply chain for the product. The improved
integration of activities across multiple companies/entities in a supply chain allows the collection
of data from the breadth of the supply chain (Fawcett and Magnan 2002). Information technology
is the key enabler for integration in supply chains. Information systems such as EDI (Electronic
Data Interchange) and ERP (Enterprise Resource Planning) provide the information needed for
integration (Lyons et al. 2005). The information systems make the information available; however,
the effect of integration directly depends on the quality of the information made available.
After the integration of the supply chain, the next step towards self-organisation is the collabora-
tion between entities. The objective of the collaboration is to make the information available when
it is needed. Of course collaboration concepts are not new; they exist from the days of traditional
business in the form of contracts. However, the effectiveness and execution speed of collaborations

Figure 1. Key stages to the evolution of adaptive supply chain networks.

350 K.R. Anne et al.

is greatly increased due to technological advances and integration tools. Collaboration is strong
where business-to-business relationships are strong. The degree of collaboration varies depending
on the strength of the integration (Holweg et al. 2005). However, true collaboration among and
between all the entities in a SCN is more difficult to achieve in practice.
Even though data is made available with the help of integration tools and collaborative agree-
ments, often the collected data paints a false image of an operation due to data entry errors
and inconsistent collection procedures. Furthermore, the upstream/downstream requirements are
sometimes not clearly understood, not accurate enough and are not up-to-date. The breadth of
the supply chain can compound the accuracy problem as the data can be reworked or recreated
in between.
To cope with inaccurate and/or varied data, synchronisation is an important step in dealing
with uncertainties. Synchronisation can be classified into two types, namely complete and partial.
Complete synchronisation is observed when data synchronisation is achieved, leading to the real-
time access to available data by all players at the same time. Complete synchronisation enables
the supply chain to react quickly to changes in demand and in product design. This type of
synchronisation is particularly suitable in just-in-time SCNs (Mondragon and Lyons 2007). To
achieve complete synchronisation, the entire chain should follow the integration and collaboration
methods in true spirit.
Partial synchronisation is achieved through a feedback controller item. Apart from data syn-
chronisation as explained in complete synchronisation, a controller item is developed to mitigate
the effects due to both inaccurate data and uncertainties. In this type of synchronisation the major
effort is placed in quantifying the effects due to uncertainties. The modelling and quantification
of the effects caused by the time lag (i.e. time delay), the information discrepancy and the indi-
vidual objectives help in designing the controller/synchroniser element. This controller item can
be unique for each entity or for the supply chain as a whole. Uncertainties and exceptions are
identified early and the data for intelligent response are immediately available. This greatly min-
imises the bullwhip effect, demand amplification and saves downstream partners and customers
from needless activity (Lei et al. 2006, Anne et al. 2008a,b, Hwarng and Xie 2008). In this paper,
we propose an adaptive controller to achieve synchronisation phenomena to mitigate the effects
due to uncertainties.

Supply chain modelling

The theoretical framework for supply chain management underlies the setting, optimisation and
control of the system model. The system model is not unique for all the supply chains (Poirier
2003). The system dynamics change for each type of product, for example food, oil, consumer
goods, etc., depending upon the processes involved. The system dynamics-based approach to
model the business dynamics was first introduced by Forrester (1961). System dynamics has
its origins in control engineering and management. The approach uses a perspective based on
information feedback and delays to understand the dynamic behaviour of complex physical and
social systems. System dynamics is an approach which is actively used to model managerial
behaviour.
Sterman (1989) proposed a model that could be used to analyse supply chains using the industrial
dynamics fundamentals proposed by Forrester. The Sterman model was actively used to analyse
supply chain dynamics. Due to its visual nature and simplicity, after a lengthy ‘floppy’ period,
the system dynamics approach is now gaining momentum in the modelling of inventory manage-
ment processes, policy development and demand amplification (Angerhofer and Angelides 2000).
The system theory-based modelling is also used to develop feedback controllers to mitigate the
 International Journal of Logistics: Research and Applications 351

bullwhip effect (i.e. demand amplification) to some extent (Laurikkala et al. 2003). However, to
address the issues emanating from uncertainties and the problems occurring in real time, system
dynamics modelling might not be suitable (Angerhofer and Angelides 2000, Chen et al. 2000,
Laurikkala et al. 2003).
In recent years, many researchers have also used an agent-based distributed modelling approach
to cope with SCN (Stadtler and Kilger 2000, Nissen 2001, Jespersen and Skjoett-Larsen 2005).
One or several agents can be used to represent each entity in the SCN. Each agent is assigned
with both a local objective and global objective. With the advent of mobile agents which can
run on lighter platforms, the use of agents to both collect information and take decisions has
become popular. Moreover, the agent paradigm is a natural metaphor for network organisations,
as companies prefer to maximise their own profit rather than the profit of the supply chain (Nissen
2001). The multi-agent-based modelling approach offers a way to elaborate the supply chain as
the agents are autonomous and the agent rules can be defined in advance. The distributed decision
nature of multi-agent systems makes it easier to add other entities in the local environment. Entities
leaving the SCN in the middle will not affect the entire SCN to a great extent. The agent rule
framework provides a certain amount of trust among the partners as it eliminates the mistrust and
deception among entities.
The major disadvantages with multi-agent-based SCN modelling are that, as there is no global
view of the system theoretical optimisation, the optimisation of the supply chain cannot be visu-
alised. As per system dynamics theory, any system can be unstable (Forrester 1961); this theorem
implies that the multi-agent system, which is a system with multi-autonomous elements/entities,
again can be unstable. Further, the multi-agent system is also based on the assumption that all
the participating entities in the SCN are truly integrated and collaborating. However, ‘true inte-
gration and collaboration’ is highly difficult (Sterman 1989, Akintoye et al. 2000, Khouja 2003,
Dejonckheere et al. 2003, Lei et al. 2006, Hwarng and Xie 2008).
By considering these issues (in this work), we envisage the complex nonlinear modelling of
a three-echelon supply chain to represent realistic dynamics. A three-level model is envisaged
(Figure 2) to describe a simple scenario in a very complex supply chain. The nonlinear supply
chain models in the literature (Poirier 2003, Giacomo and Patrizi 2006) focus mainly on the spe-
cific tasks, and thus become a transaction-oriented approach. In this work, we focus mainly on
building a nonlinear supply chain model that can exhibit more complexity covering the informa-
tion distortion, retailer order satisfaction and safety stock. An additional criterion is the extreme
sensitivity of the model to both uncertainties and initial conditions. The following notations are
introduced to facilitate the description of the model:

i time period
m rate of customer demand satisfaction at retailer
r rate of information distortion of products demanded by retailer
k safety stock coefficient at manufacturer
xi the quantity demanded by retailer in current period
yi the quantity distributors can supply in current period
zi the quantity produced in the current period depend on the order

The orders they make might not be equal to the orders they receive. The order-out quantity depends
not only on how much inventory you have already, but also how much you want to supply out. The
order-out quantity at the retailer depends on the ratio m at which the demand is satisfied during
the previous order. The distributor needs to take into consideration, among other things, the rate
of information distortion r that can occur in the received orders. The producer needs to note the
safety stock k to avoid small production batches. These scenarios/phenomena are described in
352 K.R. Anne et al.

Figure 2. Model of a three-echelon supply chain.

Figure 2. An in-depth explanation is provided next and a corresponding mathematical model is

derived to analyse the dynamics of the SCN.
We consider that the demand information is transmitted within the layers of the supply chain
with a delay of one unit of time. As illustrated in Figure 1, the ordering quantity is not the same
as the requested order quantity at any level. The order quantity at the current period of time at
the retailer is linearly coupled with the distributor and it is influenced by how much demand is
satisfied in the previous period of time. This scenario/phenomenon is modelled by Equation (1).

xi = m(yi−1 − xi−1 ). (1)

Here m is the ratio at which the demand is satisfied. The dependency/coupling between the
distributor, the producer and the retailer (Figure 1) is not linear. Indeed the distributor needs to
take the combined effect of the retailer and producer into consideration before making his order,
that is, quadratic coupling. Apart from this, the distributor also needs to take into consideration
that the order information received from the retailer might be distorted. This scenario is modelled
by Equation (2).
y1 = xi−1 (r − zi−1 ). (2)
Here, r is the information distortion coefficient. The production quantity from the producer
unit typically depends on the distributor’s orders and the safety stock. However, the distributor’s
orders again depend on the retailer’s orders, that is, the producer needs to take the combined
effect of retailer and distributor into account before making production decisions. This scenario
is modelled by Equation (3)
zi = xi−1 yi−1 + kzi−1 . (3)
Equations (1)–(3) represent the quantity demanded by customers [Equation (1)], the inventory
level of distributors [Equation (2)] and the quantity produced by producers [Equation (3)], where:
xi < 0 denotes that the supply is less than customer demand in the previous period,
yi < 0 denotes that the information is severely distorted and no adjustment is necessary at the
inventory level,
zi < 0 denotes the cases of overstock or return and hence no new production.
Equations (1)–(3) are discrete models describing the dynamics of the SCN in Figure 2. Con-
sidering very small time intervals, the continuous model in Equation (4) can be derived from
Equations (1)–(3).

ẋ = my − (m + 1)x
ẏ = rx − y − xz. (4)
ż = xy + (k − 1)z.
 International Journal of Logistics: Research and Applications 353

If the conditions σ = m + 1 and b = 1 − k are satisfied, Equation (4) leads to the Lorenz model
in Equation (5).

ẋ = σ (y − x)
ẏ = rx − y − xz (5)
ż = xy − bz.

From the theory of dynamic systems it can be proved that this model produces a wide variety
of nonlinear features depending on the parameter values. This model is particularly of interest
when dealing with the modelling of scenarios/phenomena which are very sensitive to initial
conditions and to uncertainties as well. In the case of SCN uncertainties, when added at one layer,
effectively propagate both upstream and downstream. This is a common dynamic exhibited by
realistic SCNs. A similar model [Equation (5)] is also proposed by Lei et al. (2006) to exhibit
the supply chain dynamics and mitigate the bullwhip effect. In this work, we are considering the
external uncertainties caused by external perturbations. These perturbations can occur due to the
marketplace dynamics and volatility. We consider the external perturbations to be nonlinear as
the marketplace behaviour is nonlinear in nature. Assuming that the perturbations can affect any
of the three levels of the SCN, a perturbed form of Equation (5) is proposed in Equation (6).
⎧
⎪   
⎨ẋ = σ (y − x ) + d1
ẏ  = rx  − y  − x  z + d2 , (6)
⎪
⎩    
ż = x y − bz + d3

where di (i = 1, 2, 3) represents the external perturbations.

Before considering the effects of external perturbations, let us focus first on discussing the phase
portrait of a given supply chain for illustration. For this illustration, we define an example supply
chain with the following parameter values for the model in Equation (5): σ = 15, r = 29 and
b = 2/3. Before perturbation, we call this system example the ‘reference supply chain model’.
The ‘phase portrait/diagram’ is the set of all possible states of a dynamic system; each state
of the system corresponds to a unique point in the phase space (Terman and Izhikevich 2008).

Figure 3. Phase space representation of the reference supply chain for σ = 15, r = 29 and b = 2/3.
354 K.R. Anne et al.

Considering two system variables, Xand Y , we can plot one variable against the other at a given
point in time on a standard XY graph. This is called a ‘phase portrait/diagram’ of the system. The
form and structure of the phase portrait reveals information about the system behaviour for the
chosen parameter values: in case of the form of a point, the system is in saturation; in case of a
periodic orbit, the system is in a regular state; and in case of discernible patterns, the system is in
a chaotic state (Terman and Izhikevich 2008). Figure 3 displays the phase portrait of the example
supply chain system, whereby Xref , Yref , Zref represent the x, y and z of the three-level supply
chain described in Equation (5); its form reveals that the system is in a regular state.
After defining the reference model we analyse the effects of external perturbations on this
model. Basically we are concerned with the new dynamics exhibited by the reference supply
chain subjected to external perturbations.

Synchronisation of an externally perturbed supply chain

In this section, we briefly discuss the traditional approach to the investigation of synchronisation
issues. Modern synchronisation tools provide an automation framework but do not concentrate
on what happens if the given data is slightly changed accidentally. Many companies have taken
inspiration from the modern web and wireless technologies to facilitate synchronisation in a
timely manner (Khouja 2003, Anne et al. 2008a). The integration efforts and the collaboration
for the processes within the SCN certainly improve communication by means of EDI and current
internet-based, information exchanges. Better information (point-of-sale data and the collabora-
tive planning, forecasting, and replenishment (CPFR) initiatives), and a general willingness to
work more closely together made the timeliness of information possible to a certain extent. Effi-
ciencies have been gained through improvements that any executive can effect at his or her own
workplace by putting in place the appropriate company-wide initiatives aimed at improving the
internal business process. However, as we have seen, the uncertainties propagate in both directions
(upstream and downstream) along an SCN, so network-wide initiatives are necessary to mitigate
the effects caused by uncertainties.
In this context, we provide different cases of perturbations affecting the data and present tech-
niques/methods to synchronise or stabilise the new states (or perturbed states) exhibited by the
SCN. The causes of instability of the supply chain can be broadly classified into two categories.
The first cause is the dynamic and nonlinear character of the motions (i.e. material/products flow
and information exchanges) between different entities in the supply chains. The second cause
originates from the effects of both external and internal perturbations (Lei et al. 2006) to which
the supply chain is subjected.
An optimal management of the information flows within the supply chains may be of high impor-
tance to alleviate the effects leading to negative consequences on the flows within the chains. This
could be achieved through an adaptive control mechanism which is based on a current comparison
of the dynamical data within the supply chains with the predefined data fixed by the requirements
of the supply chains. Here, an automatic or adaptive control of the flows within the supply chains
should be able to detect changes in the flows and act accordingly/consequently (by undertaking a
given action) to alleviate the undesirable effects and therefore stabilise the system behaviour that
has been perturbed. The achievement of synchronisation is observed when the action undertaken
has allowed the recovery of the original behaviour (eventually thresholds or reference require-
ments) of the supply chain. The schematic description of the adaptive synchronisation of a supply
chain is illustrated in Figure 4.
This research develops an adaptive method (algorithms and/or tools) for the systematic and auto-
matic control of the flows within supply chains. In fact, due to the dynamic changes as discussed
 International Journal of Logistics: Research and Applications 355

Figure 4. An adaptive feedback control model to mitigate the effects due to uncertainties and perturbations.

(in time domain), some predefined settings or requirements within the supply chains (thresholds
such as safety stocks) may be varying accordingly as a consequence of these perturbations. It is
worth mentioning that a combination of the simultaneous effects of both internal and external
perturbations may be responsible for the dynamic motion variations (e.g. flow of materials and
information exchange) within the supply chain. This is a concrete and/or realistic scenario as the
supply chains of many companies are currently exposed to both types of perturbations. However,
the analysis in this work is restricted to the case where the reference model is subjected to external
perturbations due to the fluctuation in the market demand.
Figure 5 shows the phase space structure of the reference supply chain subjected to the external
perturbations with X(ext, int), Y (ext, int), Z(ext, int) representing ẋ  , ẏ  and ż of the supply
chain described in Equation (6). External perturbation di (i = 1, 2, 3), d1 = 0.56 cos(3t), d2 =
20 cos(5t) and d3 = 50 cos(1020 cos(5t)t), is particularly considered as nonlinear perturbation

Figure 5. Chaotic state of the supply chain due to external perturbations.

356 K.R. Anne et al.

as the external uncertainties (e.g. market fluctuation and forecasting demand) exhibit nonlinear
dynamics. The effects of external perturbations on the original (or specific) requirements of the
reference supply chain are clearly shown by the well-known chaotic Lorenz attractors exhibited by
the reference SCN subjected to perturbations. The cancellation process of these effects is achieved
by exploiting the synchronisation controller item shown in Figure 4. The synchronisation process
concerns two different models of the supply chain: (i) the reference model (unperturbed model)
described by Equation (5), and (ii) the externally perturbed model described in Equation (6).
Following the active control approach of Liao (1998), for the purpose of synchronisation we vary
the internal parameters of the perturbed supply chain. To vary the internal parameters, we define
the state errors between the perturbed system and the reference system by Equation (7).

ex = x  − x, ey = y − y, ez = z − z, (7)

where x  , y  and z are perturbed states and x, y and z are unperturbed states. The synchronisation
problem in this context can be equivalent to the problem of stabilising the system shown in
Equation (7). This is possible through a suitable choice of the internal variables of the perturbed
system. Moreover, the adaptive control algorithm considers the effects of external perturbations
and adjusts the values of the internal parameters (σ, r, b) of the three-echelon supply chain by
tiny variations dσ , dr and db. The variation of each internal parameter is performed in well-
defined ranges (or windows) of variation. Performing the parameter variations in these ranges is
necessary as we cannot vary the parameters beyond the realistic scenario. A threshold error is fixed
(which is less than approximately 0.02) under which full alleviation of the effects due to external
perturbations is supposed to be effective; this leads to the achievement of synchronisation, which
results in the recovery of the behaviour of the reference system.
Figure 5 shows the structure in phase space of the supply chain subjected to external pertur-
bations (d1 = 0.56 cos(3t), d2 = 20 cos(5t) and d3 = 50 cos(10t)). This structure in the phase
space shows the occurrence of the well-known bullwhip and chaotic effects in the three-level
supply chain by displaying discernible patterns. The regulation process has been exploited to
adjust values of the internal parameters to achieve synchronisation, that is, the full cancellation
of the effects due to perturbations. The corresponding values obtained from the adaptive control
processes as illustrated in Figure 4 for the achievement of the regulation process are dσ = 35,
dr = 15 and db = 0.09. The results of this process are shown in Figure 6. Indeed, below the
precision of 2% when compared with the reference model in Figure 3, Figure 6 shows attractors
similar to those of the reference or original supply chain system.
Further investigations have been performed to show that the supply chain subjected to external
perturbations can exhibit the state of saturation which is characterised by non-dynamic (or fixed)
data in each of the three levels. Indeed, for d1 = 10 cos(5t), d2 = 5 cos(10t) and d3 = 10 cos(10t),
and the achievement of the state of saturation is clearly shown in Figure 7. The saturation manifests
itself by a sudden exhibition of fixed or constant values/data along each level of the externally per-
turbed three-echelon supply chain. When the state of saturation is obtained, further changes/flows
in the supply chain are not represented effectively. To alleviate the effect (i.e. saturation) due to
external perturbations, we performed the adaptive regulation process explained previously by
adjusting the internal parameters of the SCN. The appropriate values of the internal parameters
to alleviate the effects due to external perturbations are: dσ = 4, dr = 2 and db = 1. The result
of the regulation process is shown in Figure 8. Indeed, tiny variations of the internal parameters
of the supply chain lead to the achievement of synchronisation. This is manifested by the abrupt
change of the state of the system from the saturation state (Figure 7) to a regular state (Figure 8),
which is similar to the state of the reference supply chain (Figure 3).
The supply chain subjected to external perturbations can exhibit various remarkable states such
as chaos (Figure 5) and saturation (Figure 7), to name but two. The regulation process can be
 International Journal of Logistics: Research and Applications 357

Figure 6. Alleviation of the chaotic effect caused by external perturbation with adaptive synchronisation.

Figure 7. Saturation state of the supply chain due to external perturbations.

performed to cancel or mitigate the effects due to external perturbations. This process is based
on the adjustment of the internal parameters of the supply chain. Various new and interesting
states of the supply chain are discovered towards the achievement of synchronisation which
is characterised by the cancellation or alleviation of the effects due to external perturbations.
Therefore, an interesting and open question is raised about the exploration of appropriate methods
to control the states of the supply chain. Indeed, the stability of the supply chain is not robust (i.e.
it is very sensitive) to external perturbations. The control process might lead to the derivation of
the parameter ranges (windows) within which each of the various states of the supply chain can
358 K.R. Anne et al.

Figure 8. Alleviation of the saturation effect caused by external perturbation with adaptive synchronisation.

Figure 9. Bifurcation plot showing the sensitivity of the supply chain to the internal variable dσ .
 International Journal of Logistics: Research and Applications 359

be found (or defined). Bifurcation analysis is an appropriate method to describe the various states
of the supply chain in well-specified windows of parameter values.

Bifurcation analysis

Bifurcation is a qualitative change observed in the behaviour/state of a system as its parameter

settings vary. Bifurcation is observed if the state of the system suddenly changes qualitatively
upon small/smooth variation of the parameter values. Bifurcation theory (Crawford 1991) is the
study of bifurcation scenarios with the aim of defining/determining the states (equilibrium/fixed
points, periodic or chaotic states) of the system in a given parameter space. Basically, bifurcation
values/points are critical values leading to qualitative changes in the states of the system.
The preceding section has shown that the annihilation or alleviation of the effects due to external
perturbations is possible through the achievement of synchronisation. Nevertheless, during the
regulation process, we found that the perturbed supply chain system was very sensitive to very
small variations of the internal parameters of the supply chain. Indeed, various states of the
supply chain were observed ranging from regular to chaotic states. These states were observed
when monitoring the internal parameters (e.g. 0 ≤ dσ ≤ 50 and 0 ≤ dr ≤ 50) of the supply chain.
Therefore, bifurcation analysis can help to discover the various states towards the achievement of
synchronisation. This analysis can also be used to control and cancel the effects due to external
perturbations. Figures 9 and 10 are bifurcation diagrams showing the states of the perturbed supply
chain. The control parameters dσ and dr are obtained by adjusting the internal parameters σ and

Figure 10. Bifurcation plot showing the sensitivity of the supply chain to the internal variable dr.
360 K.R. Anne et al.

r, respectively. Figures 9 and 10 show the extreme sensitivity of the supply chain to the variations
of dσ and dr. Indeed, windows of regular states are shown which alternate with windows of
chaotic states (e.g. period-1, period-3 and chaotic attractors are shown). From Figures 9 and 10,
windows of parameter values can be defined in which each of these states can occur. A control or
cancellation of these states is possible and can be achieved through the synchronisation analysis.
Bifurcation diagrams are therefore an important tool because they can be used to define the ranges
of the internal parameters of the supply chain in which the synchronisation can be achieved.
Two main conditions are important for the achievement of synchronisation. The first condition
is related to the periodicity. The second condition for the achievement of synchronisation is
described in Equation (7). For instance, considering the periodicity of the original supply chain,
that is, period-1 attractor (Figure 3), the synchronisation (or the annihilation/cancellation of the
effects due to external perturbations) to be achieved must lead to the same periodicity. This is
the consequence of the competition between the internal variation of the system values and the
effects of external perturbations. It can be derived from the first condition related to the periodicity
in the following windows 5 ≤ dσ ≤ 10, 30 ≤ dσ ≤ 50, and 0 ≤ dr ≤ 27 (Figures 9 and 10) in
which the synchronisation of the supply chain can be achieved. These windows define the set of
the internal parameter settings in which the computation must be performed to fulfil the second
condition [Equation (7)]. It is worth noting that the bifurcation diagrams were explored to define
the ranges or windows of the internal parameters of the supply chain in which the regulation
process can be performed. A random choice of these windows to perform synchronisation is
also possible. Nevertheless, when computing in random windows it is not possible to know if
synchronisation can be achieved. Therefore, the method based on the bifurcation diagram is a
systematic tool that can be exploited by strategic decision-makers to evaluate how far their supply
chain can be affected if the parameter settings are changed.

Conclusions

The management of supply chains is a complex issue which involves numerous time-varying
dynamic situations. This work has been concerned with the investigation of the effects of uncer-
tainties generated by the dynamic and volatile global marketplace on the stability of a three-level
supply chain. The dynamics of this supply chain were modelled mathematically by the well-
known Lorenz oscillator. The uncertainties were considered as periodic external perturbations.
The mathematical models were derived by exploiting the structure of the three-level supply chain
proposed in this work. We have illustrated how a supply chain can exhibit chaotic modes causing
the Forrester effect and also saturation modes when subjected to external perturbations. These
two modes show the states of instability of the three-level supply chain subjected to perturbations.
A regulation scheme was designed and exploited to cancel or alleviate the effects due to exter-
nal perturbations. It was shown that this cancellation or alleviation leads to the achievement of
synchronisation which is characterised by the re-establishment of the reference data in the sup-
ply chain. Two main criteria were defined for the achievement of synchronisation. These criteria
were exploited to derive some appropriate values of the system parameters for the achievement of
synchronisation. The regulation process was based on the variation of the internal parameters of
the supply chain in well-specified ranges. These are ranges within which the achievement of syn-
chronisation is possible. The challenging issue concerned the method to derive or determine these
ranges of parameters. We have shown that bifurcation analysis is appropriate to determine these
ranges. Bifurcation analysis was carried out where two important internal parameters (i.e. rate of
customer demand satisfaction and rate of information distortion at distributor) were considered
as control parameters. Bifurcation diagrams were obtained showing the extreme sensitivity of the
 International Journal of Logistics: Research and Applications 361

three-echelon supply chain when subjected to both external and internal perturbations. It has been
found through such bifurcation diagrams that the effects due to perturbations can lead to both
chaotic and regular states of the supply chain and that these states alternate when monitoring the
internal parameters of the supply chain. The bifurcation analysis in this work has been shown to be
important as it could help strategic level decision-makers to better understand the performance of
the supply chain over a range of parameter settings. The regulation process exploited in this work
was based on an adaptive algorithm for the automatic cancellation of the effects of the external
perturbations by re-adjusting the internal thresholds. This process is particularly appealing as it
is possible to control or adjust the internal thresholds of the supply chain. The solutions proposed
in this paper offer a new range of possibilities for risk managers and provide a future research
direction with the aim of considering the concept of nonlinear dynamics.
An open question under investigation concerns the design of ‘analogue computing’ based simu-
lators based on cellular neural network (CNN) technology to achieve the adaptive synchronisation
in SCNs. This investigation is of high importance due to both the complexity and dynamic char-
acter of SCNs in practice. These features make supply chains very difficult to simulate by means
of classical simulation tools. It would also be of great interest to consider the case where the
external perturbations to which the SCNs are subjected are non-periodic and stochastic. This is a
realistic scenario which currently manifests itself in commercial SCNs and which can reflect the
evolution of market demand.

References

Akintoye, A., McIntosh, G., and Fitzgerald, E., 2000. A survey of supply chain collaboration and management in the UK
construction industry. European Journal of Purchasing and Supply Management, 6 (3–4), 159–168.
Angerhofer, B.J. and Angelides, M.C., 2000. System dyanamics modelling in supply chain management: research review.
In: Paul A. Fishwick ed. Proceedings of the 32nd Winter simulation conference, Orlando, Florida, 342–351.
Anne, K.R., Chedjou, J.C., and Kyamakya, K., 2008a. Modelling of a three echelon supply chain: stability analysis and
synchronization issues. International workshop on nonlinear dynamics and synchronization, Klagenfurt, Austria.
Anne, K.R., Chedjou, J.C., and Kyamakya, K., 2008b. Bifurcation analysis and synchronization issues in a three echelon
supply chain network. Logistics Research Network annual conference, Liverpool, UK.
Ayers, J.B. 2006. Handbook of Supply chain management. New York: Auerbach Publications.
Chen, F., Drezner, Z., Ryan, J.K., and Simchi-Levi, D., 2000. Quantifying the Bullwhip effect in a simple supply chain:
the impact of forecasting, lead times, and information. Management Science, 46 (3), 436–443.
Christopher, M., 1992. Logistics and supply chain management: strategies for reducing costs and improving services.
Financial Times, Pitman Publishing, London.
Crawford, J.D., 1991. Introduction to bifurcation theory. Reviews of Modern Physics, 63 (4), 991–1037.
Davis, T., 1993. Effective supply chain management. Sloan Management Review, 34 (4), 35–46.
Dejonckheere, J., Disney, S.M., Lambrecht, M.R., and Towill, D.R., 2003. Measuring and avoiding the bullwhip effect:
a control theoretic approach. European Journal of Operational Research, 147 (3), 567–590.
Fawcett, S.E. and Magnan, G.M., 2002. The rhetoric and reality of supply chain integration. International Journal of
Physical Distribution and Logistics Management, 32 (5), 339–361.
Forrester, J.W., 1961. Industrial dynamics. Portland: Productivity Press.
Giacomo, L.D. and Patrizi, G., 2006. Dynamic nonlinear modelization of operational supply chain. Journal of Global
Optimization, 34 (4), 503–534.
Gunasekaran, A. and Ngai, E.W.T., 2004. Information systems in supply chain integration and management. European
Journal of Operational Research, 159 (2), 269–295.
Holweg, M., Disney, S., Holmström, J., and Smårosa, J., 2005. Supply chain collaboration: making sense of the strategy
continuum. European Management Journal, 23 (2), 170–181.
Hwarng, H.B. and Xie, N., 2008. Understanding supply chain dynamics: a chaos perspective. European Journal of
Operations Research, 184 (3), 1163–1178.
Jespersen, B.D. and Skjoett-Larsen, T., 2005. Supply chain management: in theory and practice. Copenhagen: Copenhagen
Business School Press DK.
Jüttner, U., Christopher, M., and Baker, S., 2007. Demand chain management-integrating marketing and supply chain
management. Industrial Marketing Management, 36 (3), 377–392.
Khouja, M., 2003. Synchronization in supply chains: implications for design and management. Journal of the Operational
Research Society, 54, 984–994.
Laurikkala, H., Vilkman, H., Ek, M., Koivisto, H., and Xiong, G.-Y., 2003. Modelling and control of supply chain with
system theory, Tampere University of Technology, Institute of Machine Design.
362 K.R. Anne et al.

Lei, Z., Yi-Jun, L., and Yao-Qun, X., 2006. Chaos synchronization of bullwhip effect in a supply chain. In: Proceedings
of the international conference on management science and engineering, ICMSE ’06. Lille, 557–560.
Liao, T., 1998. Adaptive synchronization of two Lorenz systems. Chaos, Solitons & Fractals, 9 (9), 1555–1561.
Lyons, A.C., Mondragon, A.E.C., Bremang, A., Kehoe, D.F., and Coleman, J., 2005. Prototyping an information
system’s requirements architecture for customer-driven, supply-chain operations. International Journal of Production
Research, 43 (20), 4289–4319.
Mondragon, A.E.C. and Lyons, A.C., 2007. Investigating the implications of extending synchronized sequencing in
automotive supply chains: the case of suppliers in the European automotive sector. International Journal of Production
Research, 46 (11), 2867–2888.
Nissen, M.E., 2001. Agent-based supply chain integration. Information Technology and Management, 2 (3), 289–312.
Poirier, C.C. 2003. Using models to improve the supply chain. USA: Routledge.
Skjoett-Larsen, T., Thernøe, C., and Andresen, C., 2003. Supply chain collaboration: theoretical perspectives and empirical
evidence. International Journal of Physical Distribution & Logistics Management, 33 (6), 531–549.
Stadtler, H. and Kilger, C., 2000. Supply chain management and advanced planning: concepts, models, software, and
case studies. Berlin, Germany: Springer.
Sterman, J.D., 1989. Modeling managerial behavior: misperceptions of feedback in a dynamic decision making experiment.
Management Science, 35 (3), 321–339.
Terman, D. and Izhikevich, E.M., 2008. State space. Scholarpedia, 3 (3).
Wilding, R., 1998. The supply chain complexity triangle: uncertainty generation in the supply chain. International Journal
of Physical Distribution & Logistics Management, 28 (8), 599–616.