Proceedigs of the 15th IFAC Symposium on

Proceedigs
Information of
Proceedigs the
the 15th
Control
of IFAC
IFAC Symposium
Problems
15th on
in Manufacturing
Symposium on
Proceedigs
Information of theOttawa,
Control15th IFAC
Problems Available
Symposium
in on online at www.sciencedirect.com
May 11-13, 2015.
Information Control Problems in Manufacturing
Canada Manufacturing
Information
May 11-13, Control
2015. Problems
Ottawa, in
Canada Manufacturing
May 11-13, 2015. Ottawa, Canada
May 11-13, 2015. Ottawa, Canada
ScienceDirect
IFAC-PapersOnLine 48-3 (2015) 1210–1215
Supply Chain Complexity in the Semiconductor Industry: Assessment from System
Supply
Supply Chain Complexity in the Semiconductor Industry: Assessment from System
Supply Chain
Chain Complexity in
in the
ComplexityViewthe Semiconductor
the Impact ofIndustry:
Semiconductor
and Changes Assessment
Industry: Assessment from
from System
System
View
View and
and the
the Impact
Impact of
of Changes
Changes
View and the Impact of Changes
Can Sun* Thomas Rose**
Can
Can Sun* Thomas Rose**
Can Sun*
Sun* Thomas
Thomas Rose**
Rose**
*Infineon Technologies AG, Am Campeon 1-12, 85579 Neubiberg, Germany (e-mail: can.sun@infineon.com); RWTH Aachen
*Infineon
*Infineon Technologies AG, Am Campeon 1-12, 85579 Neubiberg, Germany (e-mail: can.sun@infineon.com); RWTH Aachen
*Infineon Technologies
Technologies AG,
AG, Am
Am Campeon
Campeon 1-12,
University,
1-12,
University,
85579
85579 Neubiberg,
Templergraben
Templergraben
Germany
55, 52062
Neubiberg,
55, 52062
(e-mail:
Aachen,
Germany can.sun@infineon.com);
can.sun@infineon.com); RWTH
Germany
(e-mail:
Aachen, Germany RWTH Aachen
Aachen
University,
** Fraunhofer FIT, Schloss Birlinghoven,
University, Templergraben
53754 Sankt
Templergraben 55,
Augustin,
55, 52062 Aachen,
Germany
52062 Germany
(e-mail:
Aachen, thomas.rose@fit.fraunhofer.de);
Germany RWTH
**
** Fraunhofer FIT, Schloss Birlinghoven, 53754 Sankt Augustin, Germany
52062(e-mail:
Aachen,thomas.rose@fit.fraunhofer.de); RWTH
** Fraunhofer
Fraunhofer FIT,
FIT, Schloss
Schloss Birlinghoven,
Birlinghoven,
Aachen
53754
53754 Sankt
Aachen University,
University, Sankt Augustin,
Templergraben
Templergraben
Germany
Augustin, 55,
Germany
55, 52062
(e-mail:
(e-mail:
Aachen,
thomas.rose@fit.fraunhofer.de);
Germany
thomas.rose@fit.fraunhofer.de);
Germany
RWTH
RWTH
Aachen University, Templergraben 55, 52062 Aachen, Germany
Aachen University, Templergraben 55, 52062 Aachen, Germany
Abstract: The semiconductor industry is considered to be one of the most complex industries, not only
Abstract: The
The semiconductor
semiconductor industry
industry is
is consideredin theto be one of the most complex industries,
but not
alsoonly
due to more
Abstract:
Abstract:
due to moreThethan
than
500 processing
semiconductor
500 processing steps is considered
steps involved
industry considered
involved in the
to be
be one
one of
of the
manufacturing
tomanufacturing the most
most
and
complex
and various
complex
various
industries,
products,
industries,
products, but
not
not
also
only
the
only
the
due
harsh
due to more than
environment
to more than 500 500
it processing
faces, e.g.,
processing steps
the involved
volatile
steps involved in
electronicthe manufacturing
market and
in the manufacturing the and various
unpredictable products,
demand.
and various products, but
On
but also
the
also the
one
the
harsh
hand, environment
harsh environment
companies have it
it faces,
faces, e.g.,
e.g.,
to adapt the
theirvolatile
the volatile
behaviorselectronic
to the market
electronic market
changing and
and the
the unpredictable
unpredictable
environment; on thedemand.
demand. On
On
other hand, the
the one
one
many
harsh
hand, environment
companies it
have faces,
to e.g.,
adapt the volatile electronic market and the unpredictable demand. On the one
hand,
changes
hand, companies
inevitably
companies have
have to
bring in thetheir
to adapt
adapt their behaviors
behaviors
complexities
their behaviors and
to
to
and the
to the
changing
changing
challenges
the changing to toenvironment;
the supply on
environment;
environment; on the
the
chain
on
other
other hand, many
other hand,
hand, many
management.
the management. Our
many
changes
changes
research inevitably
inevitably
strategy is bring
bring
to build in
in a the
the complexities
complexities
general model for and
complex challenges
challenges
systems to the
the
first andsupply
supply
then chain
chain
generatemanagement.
the supply Our
Our
chain
changes
research inevitably
strategy is bring
to build ina general
the complexities
model for and challenges
complex systems tofirst
theandsupply
then chain management.
generate the supply Our
chain
research
research strategy
instance model from
strategy is
is to build
it.
toit. aa general
A complex
build model
system
generalsystem
modelcan for
for complex
can be decomposed
complex systems first
into
systemsinto and
and then
three
firstthree generate
levels:
then the
the supply
subsystem,
generate chain
component
supply chain
instance
instance model
part. model from
from it. A
isA complex
complex
lowestsystem can be decomposed levels: subsystem, component
and
instance
and part.
The part
model
The part
level
from
levelit. is the
Athecomplex
lowest system
level can be
level which
which be decomposed
defines four types
decomposed
defines four
into ofthree
intoof
types three
key
levels:
levels: subsystem,
key elements
elements subsystem,
(process,
component
(process, role, object
component
role, object
and
and part.
its The
states) part
and level is
different the lowest level
relationships which
among defines
each four
other. types
Various of key
types elements
of (process,
complexity arerole, object
discussed,
and part.
and its
its
The part
states)
states)
changes and
and
and
level is therelationships
different
different
their potential
lowest level among
relationships
impact onamong
which defines
each
each
the system
four Various
other.
other.
types of types
Various
complexity
key elements
are types of
of
(process,
complexity
complexity
also analyzed. We use are
are
role, object
andiscussed,
discussed,
instance
and
and its states)
changes and
and different
their potential relationships
impact on among
the each
system other. Various
complexity are types
also of complexity
analyzed. We are
use andiscussed,
instance
and
modelchanges
from
and changes and
the their potential
semiconductor
and semiconductor
their potential impact impact
supply on the
chain
on system
to
the to complexity
illustrate
system the
complexity are
system also analyzed.
modeling,
are also We
specify
analyzed. We useuse
the an instance
features
an instanceof
model
model
each from
from
element the
the
on semiconductor
the part level supply
supply
and chain
chain
propose to
the illustrate
illustrate
metrics of the
the system
system
complexity modeling,
modeling,
measurement.specify
specify the
the features
features
Consequently, of
of
two
model
each from the
element on semiconductor
the part supply chaintheto metrics
illustrate the system modeling, specify the featurestwo of
each element
scenarios
each on the
are compared
element on the part
part bylevel
level and
and
calculating
level and
propose
propose the
the variation
propose the metrics of
of
of their
metrics of
complexity
complexity measurement.
measurement.
overall complexity.
complexity
Consequently,
Consequently,
The intermediate
measurement. Consequently, two
results
two
scenarios
of doctoralare
scenarios are compared
compared by
by calculating the variation of their overall complexity. The intermediate results
scenarios
of doctoral
research
are compared
research are by calculating
are presented
calculating
presented and
the
and the variation
future
thefuture
the variation of
of their
research
research their
is
overall
is also
overall
also
complexity.
complexity. The
planned.
planned. The intermediate
intermediate results
results
of doctoral research are presented and the future research is also planned.
© doctoral
of 2015, IFAC
Keywords: (International
research
supply chain, Federation
are presented
complex and of Automatic
the
system, future Control)
research
complexity Hosting
is also
measurement, by
planned. Elsevier
change Ltd. All rights
management, reserved.
semiconductor
Keywords:
Keywords: supply
supply chain,
chain, complex system, complexity measurement, change management, semiconductor
industry,
Keywords:
industry,
system chain, complex
engineering
supplyengineering
system complex system,
system, complexity
complexity measurement,
measurement, changechange management,
management, semiconductor
semiconductor
industry,
industry, system
system engineering
engineering 
 more sophisticated, integrated and specialized components
1. INTRODUCTION  more sophisticated, integrated and specialized
1. INTRODUCTION more
more sophisticated,
and services. For example,
sophisticated, integrated
integrated and
the mobile phone has components
and specialized
specialized components
many more
components
1.
1. INTRODUCTION
INTRODUCTION and services. For example, the mobile phone has many could
more
and services.
functions than
and services. For 20example,
years
For20example, the
ago mobile
and
the mobile carphone has
has many
manufactures
phone more
many could
more
1.1 Motivation functions
functions
double thethan
thannumber20 years
years
of ago
ago and
and
electronic car
car manufactures
manufactures
components in 5 could
years.
1.1 Motivation functionsthethan
double number20 years of ago and components
electronic car manufactures in 5 could
years.
1.1
1.1 Motivation double the
that number of electronic arecomponents in
Motivation Besides
double
Besides the
that
the product
number
the product of features
electronic
features are
constantly changing
components
constantly in 5
changing5 years.
due
years.
due
The supply chain in the high-technology semiconductor Besides
to the
Besides that
market the product
demand
that thedemand
productand features
and are
customer
features constantly
pressure.
are constantly changing
This
changing due
requires
due
The
The supply
supply chain
chain in the high-technology semiconductor to the market customer pressure. This requires
industry
The supply
industry is chain in
is characterized
characterized in the
the
by
high-technology
by long fabrication cycle
high-technology
long fabrication
semiconductor
times, high
semiconductor
cycle times, high
to
to the
the market
companies
market
companies
to
to
demand
redesign
demand
redesign
and
their
and customer
supply
customer pressure.
network
pressure. This
and
This requires
collaborate
requires
industry
levels of
industry is characterized
stochasticity,
isstochasticity,
characterized andby long fabrication
non-linearity
bynon-linearity
long fabrication in cycle
the
cycle times,
times, high
manufacturing high companies
with numerous
companies to partnerstheir
to redesign
redesign at supply
their
their differentnetwork
supply
supply network
stage to
network
and
and collaborate
collaborate
andachieve this
collaborate
levels
levels
processof
of stochasticity,
(Wang 2008).andand non-linearity
These all lead to in
in the
the manufacturing
manufacturing
complexity to be with
with numerous
numerous
integration, as it partners
partners
is almost at
at different
different
impossible tostage
stage
do to
to achieve
achieve
everything this
this
within
levels
processof stochasticity,
(Wang 2008). and non-linearity
These all lead in
to the manufacturing
complexity to be with numerous
integration, as it partners
is almost at different tostage
impossible do to achieve
everything this
within
process
managed
process (Wang
(Wang 2008).
in the supply These
2008). chain.
TheseAdd all
all tolead
lead to
that, complexity
to recent trends in
complexity to be
to the
be integration,
one company
integration, as
as it is
(Chopra
it is almost
almost2010).impossible
impossible to
to do
do everything
everything within
within
managed
managed in
in
semiconductor the
the supply
supply
supply chain.
chain.
chain Add
Addare to
to that,
that,
towards recent
recent trends
trends
greater in
in the
the
speed, one
one company
company (Chopra
(Chopra 2010).
2010).
managed in
semiconductor the supply
supply chain. Add
chain andare to that,
towards recent trends
greater in the
speed, one
Fromcompany (Choprawe
above analysis, 2010).
can state that global supply chain is
semiconductor
globalization,
semiconductor supply
supply chain
customization chain andare towards
towards greater
are flexibility speed,
(Chien 2007).
greater speed,
globalization, customization flexibility (Chien 2007). From
From above
above analysis,
analysis, we
we can
can statethethat global supply chain is
globalization,
However,
globalization,to customization
reach these
customization goals and flexibility
certain
and degree
flexibility (Chien
of
(Chien 2007).
complexity
2007).
complex
From
complex
and
above
and
we must
analysis,
we must we can state
consider
consider statethethat
that global
global supply
influence
influence supply
of
chain
of changes
changes
is
chain in
is
in
However,
However,be to to reach
reach these
too, goals certain degree of complexity complex
the supply and
chainwe must
as it is consider
a dynamic the influence
system. of changes in
would
However,
would be to reach these
increased
increased these
too,
goals
goals
e.g.,
certain
e.g., the
certain
the
degree
degree of
globalization
globalization of complexity
resulted in
complexity
resulted in
complex
the
the
and
supply chain we must
as it is consider the
a dynamic system. influence of changes in
would
more
would be increased
complex
be increased too,
processes e.g.,
too, e.g.,and the globalization
globalization resulted
organizational
theorganizational in
structures;
resulted in the supply
supply
Supply
chain
chain
chain
as
as it
it is
complexity is aa dynamic
dynamic system.
is awaresystem.
by leading firms. People
more
more complex
complex
customization processes
processes
and and
and
diversification organizational
increased structures;
structures;
the design
more complex processes and organizational structures; Supply
Supply
could chain
chain
somehow complexity
complexity
observe is
is
and aware
aware
perceive by
by leading
leading firms.
firms. People
People
customization
customization and
and diversification increased the design Supply chain complexity is aware by the complexity
leading firms. and its
People
complexity
customization and diversification
of products. diversification increased
increased the the design
design could
could somehow
somehow observe
observe and
and perceive
perceive the
the complexity
complexity and
and its
complexity
complexity of
of products.
products.
orientation,
could somehow
orientation,
but
but
they
observe
they
can
can
hardly
and
hardly
judge
perceive
judge
whether
the the
complexity
whether the and its
increased
its
increased
complexity
These trends of products.
also reflect the volatile characteristics of a orientation,
complexity but
is they
valuable can tohardly judge
the business whether
goals as the it isincreased
difficult
orientation,
complexity but
is they
valuable can to hardly
thethis judge
business whether
goals as the
it isincreased
difficult
These
These trends
trends also reflect the volatile
which characteristics of a complexity
to measure it. is valuable
is People
valuablefollow to
to the business
strategy goals as
as itit is
in principle:is difficult
value-
These trends also
semiconductor
semiconductor also reflect
supply
supply
the
the volatile
reflectchain,
chain, volatile
which
characteristics
faces a consumer
characteristics
faces a
of
of aa
consumer
complexity
to
to measure
measure
adding it.
it.
(“good") People
People follow
follow
complexity
thethis
business
this
offersstrategy
strategy
ways
goals
in
intoprinciple:
principle:
meet
difficult
value-
value-
customer
semiconductor
dominant
semiconductormarket. supply
supply chain,
chain, which
Globalization results in
which faces
veryaacompetitive
faces consumer
consumer to measure it. People follow this strategy intoprinciple: value-
dominant market. Globalization results in very competitive adding
adding
demands (“good")
(“good")
and complexity
complexity
creates a real offers
offers ways
ways
competitive to meet
meet
advantage; customer
customer
while
dominant
product
dominant market.
price for
market. Globalization
customers,
Globalization whereasresults
it
results in
pushes
in very
verythe competitive
companies
competitive adding (“good") complexity offers ways to meet customer
product price for customers, whereas it pushes the demands and creates a real competitive advantage; while
product
to price
reduce
product price for
cost customers,
for by relocating
customers, whereas it
it pushes
pushes the
manufacturing
whereas to companies
the companies
low cost
companies
demands
demands and
value-destroying creates
and creates
value-destroying
(“bad")
(“bad")
a real
a real competitive
complexity
competitive
complexity
pushes
pushes
advantage;
the
advantage;
the
while
customer
while
customer
to
to reduce
reduce cost
cost by
by relocating
relocating manufacturing
manufacturing to
to low cost value-destroying
away and sends (“bad")
the complexity
company into pushes
chaos the
and customer
confusion
countries
to reduce and
countries
and keep
cost keep looking
by relocating
looking
for
for
the
manufacturing
the
even
even to low
lower
lower low cost
places.
cost
places.
value-destroying
away
away and
and sends
sends
(“bad")
the
the
complexity
company into pushes
chaos the
and customer
confusion
countries
countries and
Correspondingly,
and keep
the
keep looking
supply
looking for the
organizations
for the even
evenare lower
evolved
lower places.
from
places.
(Scheiter
away and 2007,
(Scheiter 2007, the company
sends Etheredge company
Etheredge
into
2009).
2009).
chaos
chaos and
into After
After and confusion
complexity
confusion
complexity
is
is
Correspondingly,
Correspondingly, the
the supply organizations
and moreare evolved from (Scheiter
classified, 2007,
we can Etheredge
take these 2009).
actions: After
reduce complexity
complexity is
non-value-added
linear
linear
supply “chains”
Correspondingly,
supply the supply
“chains” supply
into
organizations
into broader
organizations
broader and more
are
are evolved
complex
complex
from
evolvedsupply
from
supply
(Scheiter
classified,
classified,
complexity;we
2007,
we can
can
manage
Etheredge
take
take these
these
value-added
2009). reduce
actions:
actions:
After non-value-added
andreduce
necessarynon-value-added
complexity;
is
linear supply
“networks” “chains”
(Marchese
linear supply (Marchese into
“chains” into2014).broader
2014).
broaderThe and
The more
product
and more complex
complex supply
itself also
supply classified, we can take these actions: reduce non-value-added
“networks” product itself also complexity;
complexity; manage
manage value-added
value-added and
and necessary complexity;
“networks”
becomes
“networks” more (Marchese
complex
(Marchese 2014).
than
2014). The
before. It product
is
The It productdesigned itself also
towards
itself also
and avoid non-value
complexity;
and avoid manage added
non-value value-added
added
complexity
complexityand necessary
in the future.
necessary
in the
complexity;
complexity;
future.
becomes
becomes more
more complex
complex than
than before.
before. It is
is designed
designed towards
towards and avoid non-value added complexity
and avoid non-value added complexity in the future. in the future.
becomes more complex than before. It is designed towards
2405-8963 ©
Copyright © 2015, IFAC (International Federation of Automatic Control)
2015 IFAC 1272Hosting by Elsevier Ltd. All rights reserved.
Peer review©under
Copyright 2015 responsibility
IFAC of International Federation of Automatic
1272 Control.
Copyright © 2015 IFAC
10.1016/j.ifacol.2015.06.249 1272
Copyright © 2015 IFAC 1272
INCOM 2015
Can Sun et al. / IFAC-PapersOnLine 48-3 (2015) 1210–1215
May 11-13, 2015. Ottawa, Canada
Besides this strategic way, people are keen to have some
qualitative methods to evaluate complexity comprehensively.
Only with the quantitative indicators, it could be more
convincing for decision making and thus used as an
operational guideline.
1.2 Research Objective and Questions
The main objective of this research is to develop a practical
framework for managing complexity in the semiconductor
supply chain. It should be able to identify and model the
complexities as well as their changes in a systematic way. It
should also provide certain instruments to assess the
complexities and thus apply various strategies (reduce
complexity, avoid complexity, etc.) based on the
classification of value adding complexity. In the end, the
results of complexity measurement should be validated and
support decision-making.
To achieve this research objective, this paper aims to answer
the following main research question: How to assess supply
chain complexity from system view and the impact of
changes in order to support decision-making in the
semiconductor supply chain?
In order to reach the goal step-by-step, this research question
can be broken down into a few sub-questions as below:
a) How can we evaluate a complex supply chain?
Complexity metrics should be developed. The
preliminary research is to understand the sources
and characteristics of the complexities and their
manifestations within a supply chain system.
b) How can we assess the complexities induced by
changes of a system? We need to investigate the
changes and their impact to the system complexity.
c) How to develop the implementation tools for
complexity measurement? This research question
aims to develop an assessment tool for the practical
use to measure complexity.
d) How to verify the first three questions with
industrials cases and thus reduce the non-value
added complexity? This research question integrates
the role of complexity evaluation into decision
making.
1.3 Related Work
Our literature research on complexity measurement is mainly
on these aspects: a) The drivers of supply chain complexity
and their impact; b) The methodologies to measure
complexity; c) The dynamic complexity caused by changes.
In the semiconductor industry, supply chain complexity
arises from a number of sources: network complexity, market
complexity, process complexity, product complexity,
organizational complexity and information complexity
(Christopher 2010, Lindemann 2007, Schuh 2008).
1273
1211
There are various measurement approaches available for the
complexity measurement. We summarize several main
approaches here: information theory including statistical and
entropy-based approach (Schuh 2008, ElMaraghy 2012);
single aspect of supply chain such as system, workflow and
product complexity (Qusaibaty 2004, Cardoso 2006,
Götzfried 2013); supply chain network with many nodes
(Kandjani 2012); Agent-Based Modelling (ABM) for
behaviours modelling (Shalizi 2006). The evaluation of
approaches is based on three criteria: quantitative, practical,
integrated. The first two are easy to understand, and
integrated aspect means that all drivers of complexity should
be considered from a holistic view.
The changes of complexity are also explored, many research
studies are about the organization, process and design
changes. For the measurement, most measurements are
limited to the qualitative analysis. A few approaches, e.g.,
simulation techniques can be used to assess the impact of
changes (Fernandes 2013).
In summary, complexity is generated from different parts of
supply chain and they are interrelated to each other.
However, the evaluation for the complexity is more on the
single and static part while the integrated and dynamic
analysis is missing; complexity measurement lacks
practicality and its role for support decision making is not
explicit; and the complexity caused by and changes and their
impact on the whole supply chain are not much explored.
The rest of this paper is organized as follows: in Section 2 we
introduce our research method and thus propose a general
framework to model the complex systems; Section 3 focuses
on the supply chain model and explains the details of
measurements on a practical example. Section 4 presents the
results achieved and future research plan; the expected
contributions are highlighted in Section 5.
2. RESEARCH METHODOLOGY
There are commonalities among different complex models
and our pre-study has showed supply chain can be viewed as
a complex system (Sun 2015b). Therefore, we adopt the
method of general-to-specific modelling and first study the
general complex systems. The primary goal is to design a
framework for the complex system on the abstract level,
including the system modelling method, complexity analysis
and influence of change. Under this framework, we can
define a clear structure for the supply chain systems;
meanwhile, the supply chain features can be integrated into
the instance model at the bottom level.
2.1 A general model of complex system
System engineering methodology can be used to model a
complex system. A preliminary step is required to define the
problem scope. We employ a system delimitation technique
to define the relevant parts of system, which can be
demarcated into three areas: environment, intervention
system, area of solution (Haberfellner 2002). Environment
covers all external factors (outside of the intervention system)
INCOM 2015
1212 Can Sun et al. / IFAC-PapersOnLine 48-3 (2015) 1210–1215
May 11-13, 2015. Ottawa, Canada
which are relevant to the system complexity. Intervention
system includes the area where all interventions and
modifications related to the problem can be made. It acts as
an interface between the internal system and external
environment. Area of solution is the core part to find problem
solutions. This step is helpful to define the working area and
boundary when we detect the impact of changes.
2.1.1 Definition of terms
Some key terminologies are defined in order to avoid
ambiguity in this paper.
A System can be broadly defined as an integrated set of
elements (activities, subsystems) cooperating together in
order to accomplish a defined objective (INCOSE handbook
2000). A subsystem can be viewed as a system only except
it cannot fulfil a goal completely on its own, it must be
integrated with other subsystems as a whole (Beale). A
component is a clearly identified subset of the products being
designed or produced and it contains many parts. Normally,
part is identified as the basic or inseparable item of a system
(Carson 2012).
2.1.2 System decomposition
System decomposition is a common used technique in the
area of software engineering and system engineering. It can
decompose the system into two or three layers or even more.
A widely used two-layer structure is the subsystems-
components (Jennings 2001, Kim 2003); if the system is very
complex a 3-layer architecture, e.g., the subsystems-
components-parts (Carlson 2012) is exploited and even can
be extended to four-layer: subsystem-components-
subcomponents-parts (Beale). Carlson states that the
definition of system hierarchy depends on the objective.
Since we would like to model a complex supply chain, a
three-layer architecture should be appropriate to capture the
details within a complex system. The break-down of System
is shown on Fig. 1.
Fig. 1. System decomposition into three layers (relationships
are not depicted due to visibility) (adapted from Haberfellner
2002 and ETH 2015)
2.1.3 System attributes and relationships
1274
For each level, the attributes of basic elements and their
interfaces are defined. The basic element for the subsystem
layer is the subsystem, similarly for the layer of component
and part. In our model, the interface part can be described as
the relationships among all the elements. These relationships
are illustrated in Fig. 2.
Fig. 2. A hierarchical model of various relationships in a
complex system (adapted from Jennings 2001)
The relationship includes dependencies and interactions
among all the elements from different levels. The former
relationship is about the static interdependencies and the
latter one refers to the dynamic connection. In Fig. 2, the
lines between any two elements without arrows stand for
dependencies, while lines with arrows mean the interactions.
The relationships can also be categorized as the local or
external ones according to the position of the two elements. If
they are within the same level or under the same element, the
relationship is local, otherwise it is external. In Fig. 2, the
black line means the local relationship while the red line
means the external one. We can also distinguish the
relationships by strength. For example, the solid line
represents the pair is strong related and the dashed line is the
weak related pair.
Besides the multiple types of relationships, a constraint set
should also be given. E.g., a part within one subsystem is not
allowed to be directly related to another subsystem.
2.1.4 The part layer- PROS analysis
At the lowest level of this abstract model, the part layer, it
defines the basic types of elements (parts) with distinct
features, a group of which under certain rules constitutes the
component on the upper layer.
We employ a conceptual model including four elements:
process, role, object and its states (PROS) to describe a
complex system. It also defines the six types of relationships
among these elements; the constraints are given too (Sun
2015a). For example, the process can be related to any other
elements. A simple PROS model is sketched in Fig. 3.
This model could cover most key features in a system and
keep it simple, thus we consider using these four types of
INCOM 2015
Can Sun et al. / IFAC-PapersOnLine 48-3 (2015) 1210–1215
May 11-13, 2015. Ottawa, Canada
“pieces” on the part layer. The specification of relationships
can also be taken from PROS.
Fig. 3. A simple system described by PROS approach
2.2 Complexity analysis on different types
Complexity analysis is based on the hierarchical model built
in Section 2.1. We can address the complexities from
different views:
a) The element and relationship complexity. The
element complexity is related to the attributes and
type of each element; relationship complexity is
about the dependencies and interactions among all
elements. For the overall complexity, the quantity
and impact factors of elements and relationships
should also be considered (Sun2015b).
b) The structural complexity, functional complexity,
and organizational complexity. This corresponds to
the different layers of our model respectively: part,
component and subsystem layer (Schuh 2013, Salado
2014).
c) The internal and external complexity. The
complexity within the solution area of a system
belongs to the internal complexity, and the external
one means the complexity caused by any
environmental factors (ETH 2015, Sun2015b).
d) The dynamic and static complexity. The static
complexity is related to the inherent
interdependencies while the dynamic one is
influenced by the interactions and changes.
2.3 Changes of system and their impacts
Changes lead to the dynamics of system and complexity is
generated accordingly. There are many types of changes in a
system. Below we list some categories of them and also
analyse their potential impacts on the system complexity.
a) Internal changes and external changes. The changes
could be generated from the local internal system or
stimulated by the environment. External changes
might affect more parts than the internal ones.
1275
1213
b) Passive changes or active changes. Passive changes
refer to the physical components, while the active
changes are more initiated by humans, which have
more uncertainty and thus increase complexity.
c) Changes on the different levels: subsystem,
component or part. Changes occurred at the higher
layer have more influence on the whole system than
at the lower layer.
3. SUPPLY CHAIN INSTANCE MODEL
Based on the framework developed in Section 2, we can build
a supply chain instance model, which inherits the properties
from the general model and highlights the specific features of
each type of elements at the part level and below. In this
section we take a simplified semiconductor supply chain as
an example, to illustrate the system modelling, metrics
formulation and scenarios comparison.
3.1 A model of semiconductor supply chain
A typical semiconductor supply chain has five stages:
customer, Original Equipment Manufacturer (OEM), tier 1,
tier 2 and raw material supplier. Two flows run in the supply
chain, one up and one down. For the order flow, the customer
orders products from the OEM and the OEM sends the order
to its upstream - the tier 1 until the final order is received by
the upstream end of the chain- the raw material supplier. Vice
versa is the delivery flow: the raw material supplier produces
units and delivers them to other downstream stages until the
final products reach the end customer.
This supply chain can be mapped into the three-layer model
introduced in Section 2.1.
 The subsystem layer: it has five subsystems:
customer, OEM, tier 1, tier 2 and raw material
supplier.
 The component layer: it varies according to the
subsystems. It may include the module of
production, order management, delivery service, etc.
 The part layer: it defines the key elements for the
modules on the component layer. It includes
processes such as order request, delivery and their
responsible entities; the objects like orders, products
and their varied states, etc.
To get an easy identification of the elements on each layer,
we employ the reference models for process and data
management in the supply chain management field. One
widely accepted framework is the Supply-Chain Operations
Reference (SCOR) model, with the scope from suppliers’
supplier to customers’ customer, which defines the main
abstract processes in a hierarchical way. For the data model,
Entity-Relationship (ER) model is one popular method to
describe the information structure, which is usually defined
by individual companies. Based on these reference models, it
is easy to extract the elements and their features, especially
helpful for us to get more details on the lower levels.
INCOM 2015
1214 Can Sun et al. / IFAC-PapersOnLine 48-3 (2015) 1210–1215
May 11-13, 2015. Ottawa, Canada
3.2 Complexity measurement on the supply chain system
Our measurement aims to reflect the features of complex
systems comprehensively. Besides the quantity of elements
and relationships in a system, we are more interested in these
aspects: the variety of elements and their individual
contributions to the system complexity; the collaborationism
of them acting together in order to reach the system goals; the
dynamic dependencies among elements. For example, a
process could match the objective of a system at different
degrees. The better it meets the goal, the less complexity it
contributes to the system.
Hence, we can list all factors which influence the system
complexity, for each of them we assign different values or
value scales. The quantitative measurement can be applied to
each type of complexity mentioned in Section 2.2 after
analysis of all possible influence factors for a system.
3.3 The scenario of changes: from non-collaboration to the
collaboration
The complexity measurement serves for decision-making
support. Using the supply chain in Section 3.1 as an example,
we can view it as a system with the goal that each supply
chain partner has to fulfil the incoming orders and minimize
the inventory cost. Two scenarios are drawn out: the first one
requires that each partner makes decisions separately without
communication with others; while the second one uses the
collaboration mechanism and thus each stage can get better
overview about the whole supply chain.
We assume that the non-collaboration scenario is less
complex than the collaborative one, but has higher cost. This
can be validated with an experiment. By collecting the results
we can calculate the complexities based on the measurement
approaches proposed in Section 3.2. Instead of comparing the
whole system complexity of two scenarios, we can also only
consider their difference, or the changed part from scenario 1
to scenario 2, which is limited to the area of intervention
system. By checking the affected elements and involved
relationships in this area we can evaluate the changes and
thus calculate the complexity and cost.
4. RESULTS ACHIEVED
Towards answering the research questions in Section 1, we
summarize the intermediate results of this research, some of
which have been published.
The first stage research is about the complexity management
theory, which includes: understanding the complex system
with the support of a general framework for the system
modelling; presenting the key drivers in the supply chain
through a conceptual model called PROS; demonstrating the
complexity analysis with an industrial example from the
semiconductor manufacturing (Sun 2015a).
Based on the general model of complex systems, the
measurement methods are investigated; and the key influence
factors of the system complexity and their impact are
1276
identified. A few preliminary measurement metrics and key
parameters have been proposed, e.g., system collaboration,
satisfaction of goals, etc. The system complexity can be
calculated based on these basic formulas (Sun 2015b). These
measures are partly validated by a practical example, for
some input lacking of real data we simply use assumptions.
The change analysis for supported decision making is also
investigated to compare two technology solutions (Sun
2015a), or two business scenarios (Sun 2015b).
5. THE FUTURE RESEARCH PLAN
The future research focus on these aspects: the full definition
of the general complex model including various relationships
should be described in details. Based on this fundamental
model, dynamic changes and their impact on the system
within the space-time domain should be distinguished.
Afterwards, a thorough list of the influence factors with their
potential impacts on the system complexity will also be
provided and thus analysed.
The features for each element on the part level, e.g., the
diversity, coupling, influence should be investigated. Among
these features, it is worth noting that one type of element, the
agents, is a main driver to increase complexity as the
human’s decision making could lead to huge uncertainty. A
state-of-the-art method is the Agent Based Modelling
(ABM), which can be used to analyse the humans’ behaviour
in a Complex Adaptive System (CAS) (Aelker 2013).
Other simulation techniques such as discrete event modelling
are also helpful to analyse the influence factors, e.g. tracking
one or more variables at the same time, or tuning the values
of input parameters, an implementation tool should be
developed too.
The proposed metrics and formulas can be validated and
calibrated through the industrial cases. The required input
values for some parameters can be obtained from the real
data in the semiconductor supply chain.
The established metrics of complexity measurement can not
only be applied to one tier-2 company, but also be able to
benchmark the whole semiconductor supply chain. The
comparison with other tier 2 companies will be also taken
into account.
6. EXPECTED CONTRIBUTIONS
In this doctoral proposal we aim to set up a framework for
complex system modelling and develop metrics for
complexity measurement in a supply chain. The uniqueness
and scientific contribution are highlighted below:
 Analyse the complexity from a holistic system view
instead of on certain aspect
 Build a general complex system model using system
thinking and combine the techniques from software
engineering and system engineering
INCOM 2015
Can Sun et al. / IFAC-PapersOnLine 48-3 (2015) 1210–1215
May 11-13, 2015. Ottawa, Canada
 Consider the overall complexity not only from the
static part but also the changes and their impact on
the system complexity
 Develop practical metrics and formulas to measure
complexity and benchmark industry practices
 Emphasize the uncertainty and complexity oriented
from the humans’ behaviour and model it through
the simulation technique
 Support decision making with complexity metrics
and cost & benefits analysis
REFERENCES
Aelker, J., Bauernhansl, T., and Ehm, H. (2013). Managing
complexity in supply chains: a discussion of current
approaches on the example of the semiconductor
industry. Procedia CIRP, 7, 79-84.
Beale, D. and Bonometti, J. ESMD Course Material:
Fundamentals of Lunar and Systems Engineering for
Senior Project Teams, with Application to a Lunar
Excavator.
http://www.eng.auburn.edu/~dbeale/ESMDCourse/Chapt
er2.htm, retrieved on 01.03.2015
Carlson, C. (2012). Effective FMEAs: Achieving Safe,
Reliable, and Economical Products and Processes Using
Failure Mode and Effects Analysis (Vol. 1). John Wiley
& Sons.
Chien, C. F. (2007). Made in Taiwan. Industrial Engineer,
39(2), 47-49.
Chopra, Sanjiv. (2010). Supply Chain Issues in the High-
Technology Semiconductor Industry. Profit Magazine.
http://www.oracle.com/us/corporate/profit/archives/opini
on/060810-chopra-161432.html, retrieved on 01.03.2015
Christopher, M. (2010). Logistics and Supply Chain
Management. 4th Edition. Financial-Times/ Prentice
Hall.
Cardoso, J. (2006). Approaches to compute workflow
complexity. Dagstuhl Seminar Proceedings 06291, The
Role of Business Processes in Service Oriented
Architectures, Germany.
ElMaraghy, W., ElMaraghy, H., Tomiyama, T., and
Monostori, L. (2012). Complexity in engineering design
and manufacturing. CIRP Annals-Manufacturing
Technology, 61(2), 793-814.
ETH Chair of Technology and Innovation Management,
Systems Engineering Methodology (2015),
http://www.timgroup.ethz.ch/en/courses?id=55, retrieved
on 01.03.2015
Etheredge, K. and OKeefe, J. (2009). Getting a Handle on
Complexity. Supply Chain Management Review, A.T.
Kearney, http://www.atkearney.com/documents,
retrieved on 01.03.2015
Fernandes, J., Silva, A., and Henriques, E. (2013). Modeling
the Impact of Requirements Change in the Design of
Complex Systems. In Complex Systems Design &
Management (pp. 151-164). Springer Berlin Heidelberg.
1277
1215
Götzfried, M. (2013). Managing Complexity Induced by
Product Variety in Manufacturing Companies
Complexity Evaluation and Integration in Decision-
Making. PhD dissertation, University of St. Gallen.
Haberfellner, R. (2002). Systems Engineering: Methodik und
Praxis. Daenzer, W.F. (ed.). Verlag Industrielle
Organisation.
INCOSE. Systems Engineering Handbook v2.0. 2000.
Jennings, N. R. (2001). An agent-based approach for building
complex software systems. Communications of the ACM,
44(4), 35-41.
Kandjani, H., Wen, L., and Bernus, P. (2012). Enterprise
Architecture Cybernetics for Collaborative Networks:
Reducing the Structural Complexity and Transaction
Cost via Virtual Brokerage. In Information Control
Problems in Manufacturing, 14(1), 1233-1239.
Kim, H. M., Michelena, N. F., Papalambros, P. Y., and Jiang,
T. (2003). Target cascading in optimal system design.
Journal of mechanical design, 125(3), 474-480.
Lindemann, U. and Maurer, M. (2007). Facing multi-domain
complexity in product development. In The future of
product development (pp. 351-361). Springer Berlin
Heidelberg.
Marchese, K. and Lam, B. (2014). Anticipatory supply chains
Business Trends 2014: Navigating the next wave of
globalization. Deloitte University Press.
Qusaibaty, A., Howard, N., and Rolland, C. (2004). Process
Complexity: Towards a Theory of Intent-oriented
Process Design. Presented at the 2nd International
Conference of Information and Systems (INFOS), Cairo,
Egypt.
Scheiter, S., Scheel, O., and Klink, G. (2007). How Much
Does Complexity Really Cost? A.T.Kearney,
Duesseldorf, http://www.atkearney.com/documents,
retrieved on 01.03.2015
Schuh, G., Potente, T., Varandani, R. M., and Schmitz, T.
(2013). Methodology for the assessment of structural
complexity in global production networks. Procedia
CIRP, 7, 67-72.
Schuh, G., Monostori, L., Csáji, B. C., and Döring, S. (2008).
Complexity-based modeling of reconfigurable
collaborations in production industry. CIRP Annals-
Manufacturing Technology, 57(1), 445-450.
Shalizi, C. R. (2006). Methods and techniques of complex
systems science: An overview. In Complex systems
science in biomedicine (pp. 33-114). Springer US.
Sun, C., Rose, T., Ehm, H., and Heilmayer, S. (2015a).
Complexity Management in the Semiconductor Supply
Chain and Manufacturing Using PROS Analysis, ICISO
2015, IFIP AICT 449, pp. 166–175.
Sun, C., Rose, T., Ehm, H., and Heilmayer, S. (2015b). A
System Framework for Complexity Measurement and
Evaluation on the Example of Supply Chain, BUSTECH
2015, Nice.
Wang, W. and Rivera, D. E. (2008). Model predictive control
for tactical decision-making in semiconductor
manufacturing supply chain management. Control
Systems Technology, IEEE Transactions, 16(5), 841-855.