Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
ISSN 00207543 print/ISSN 1366588X online # 2005 Taylor & Francis Group Ltd
http://www.tandf.co.uk/journals
DOI: 10.1080/00207540500076530
International Journal of Production Research,
Vol. 43, No. 15, 1 August 2005, 31313162
Flexibility-enabled lead-time reduction in exible systems
S. WADHWA*y, K. S. RAOy and F. T. S. CHANz
yDepartment of Mechanical Engineering, Indian Institute of TechnologyDelhi (IIT-D),
Hauz Khas, New Delhi, India
zDepartment of Industrial and Manufacturing Systems Engineering,
The University of Kong Hong, Pokfulam Road, Hong Kong
(Revision received January 2005)
This paper presents the results of a conceptual study and simulation exper-
imentation aimed at understanding the underlying mechanisms of sequencing
exibility-enabled manufacturing lead-time reduction. In spite of a large
body of literature on exibility, the exact mechanism that enables exibility to
reduce the lead time is not fully understood. As a part of our research eorts
on the proactive application of exibility for the performance enhancement of
manufacturing systems, we are motivated to study how exibility can be
employed in a proactive manner to reduce the manufacturing lead time and to
develop an understanding of the underlying mechanisms. Towards this end, we
have developed simulation models of simple exible manufacturing systems
and studied the eect of sequencing exibility on the lead-time performance
under dierent conditions of part load and machine load balancing. The studies
indicated that sequencing exibility has a signicant eect on the lead-time
performance of the manufacturing system, and the eect of exibility varies
under dierent conditions of load balancing. Further studies indicate the
existence of complex interactions between the sequencing exibility, process
concurrency, processor load balancing, and manufacturing lead time. This
paper intends to discuss some of the interesting results of these studies with
a focus on the inherent mechanisms of sequencing exibility-enabled lead-time
reduction.
Keywords: Theory of flexibility-enabled lead-time reduction; Sequencing
flexibility; Concurrency; Load balancing; Manufacturing lead time
1. Introduction
Manufacturing enterprises throughout the world are under great pressure to reduce
time to market, reduce the cost, and meet the global quality standards, while coping
with variety or customization pressures. The rapid advancement of technology is
also resulting in shortened life cycles and putting great pressure on organizations to
achieve shorter and shorter manufacturing lead times. Hence, one of the important
focuses of manufacturing enterprises is to nd ways and means of reducing the
manufacturing lead times to be able to respond quickly for changing market
demands. Towards this end, exibility has emerged as a critical dimension of
*Corresponding author. Email: swadhwa@mech.iitd.ernet.in
competition. Traditionally exibility is viewed as a hedge against uncertainty. It is
our research endeavour to promote the proactive application of exibility for the
performance enhancement of the manufacturing systems. However, this requires an
understanding of the exact mechanisms that enable sequencing exibility to reduce
the manufacturing lead time. This paper addresses the above issue in three parts: the
rst part focuses on the description of a proposed theory towards exibility-enabled
lead-time reduction and its underlying concepts, while the second part presents the
results of the simulation experimentation aimed at understanding the underlying
mechanisms. The third part discusses the implications of the ndings for the
manufacturing systems. It is important to work towards developing a theory of
FLR to share useful exibility ideas using simple frameworks and to explain the
underlying mechanisms. This will add value to both practitioners and researchers,
and may oer a platform for coherent exchange of experiences. Our attempt towards
this direction is an initiation to encourage further evolution in this direction.
2. Part 1: towards a theory of exibility-enabled lead-time reduction (FLR)
The theory of exibility-enabled lead-time reduction (FLR) proposes that, in order
to reduce lead time, we need to deal simultaneously with three factors, viz. the
level of exibility, the level of processor load imbalance, and the level of process
concurrency, in such a way that, for a given level of exibility, the processor load
imbalance is decreased, and the process concurrency is increased.
Before we describe the theory in detail, the four key concepts underlying
the theory of FLR, viz. exibility, lead time, processor load imbalance, and process
concurrency, may be understood as follows.
2.1 What is exibility?
Several attempts have been made in the literature to dene, model, and measure the
exibility with a view to understand its true nature and its eect on the performance
of the manufacturing system. Many denitions of exibility can be found in the
literature. For instance, Carlsson (1989), cites exibility as: (1) those attributes of
a production technology which accommodate greater output variation, as the rms
response to uncertainty, especially in the form of uctuations in demand, but also
market imperfections; (2) a property of initial positions as it refers to the cost, or
possibility of moving to various second period positions. One position is more
exible than another if it makes available a larger set of future positions at any
given level of cost. De Groote (1994) denes exibility as a hedge against the
diversity of the environment; Nilsson and Nordahl (1995) cite a denition of
exibility as the ability to respond eectively to changing circumstances. Upton
(1995) denes exibility as the ability to change with little penalty in time, eort,
cost or performance. Benjaafar and Ramakrishnan (1996) cite the following
denitions from literature: (1) the capacity of a system to assume dierent positions
or to assume a certain number of dierent states; (2) the ability of a manufacturing
plant of being usable for dierent production tasks; (3) the ability to recongure
manufacturing resources so as to produce eciently dierent products of
3132 S. Wadhwa et al.
acceptable quality; (4) the ability to respond eectively to changing circumstances;
and (5) a measure of its capacity to adopt to changing environmental conditions.
Das (1996) denes exibility as the ability of a system or a facility to adjust to
changes in its internal or external environment. Sushil (2000) denes exibility as
the exercise of free will or freedom of choice on the continuum to synthesis the
dynamic interplay of thesis and antithesis in an interactive and innovative manner,
capturing the ambiguity in systems and expanding the continuum with minimum
time and eorts. Wadhwa and Browne (1989) view exibility as a control on ow of
entities through a system and dene three generic decision points for exercising the
control on ow of various entities. This denition is in the context of systemic
exibility dealing with options, change and freedom of choice. Here, it is important
to highlight that many systems oer decision options without actually referring to
them as a form of exibility. For instance, Chan et al. (2004) deal with the
order-distribution decision in a demand-driven collaborative supply chain, without
referring to this as exibility. On the other hand, Wadhwa and Rao (2003) use many
exibility concepts to discuss multiple entity ows from an enterprise synchroniza-
tion perspective in the context of supply chains. Golden and Powell (2000) dene
exibility as the capacity to adapt across four dimensions; temporal, range, intention
and focus. Wadhwa and Rao (2000) view exibility as the ability to deal with change
by judiciously providing and exploiting controllable options dynamically.
Based on the above denitions, it can be summarized that the role of exibility in
a system is to enable the system to manage the change (certain or uncertain), in an
eective and ecient manner. The change in the environment includes change in
both the internal environment (resource bottlenecks, etc.) and the external environ-
ment (customer preferences, etc.). Eective manner refers to the extent to which the
eect of change has been successfully managed, and eciency refers to the time, cost
and eort required to do this.
The literature on manufacturing exibility discusses several types of exibility
(Tincknell and Radclie 1994). The terms and denitions used by various authors
are not always consistent, the same term often being used in dierent ways, and
several dierent names being given to essentially the same type of exibility. Browne
(1984) proposed eight exibility types to describe the nature of a manufacturing
system that is still one of the most widely used classication of exibility types.
Benjaafar and Ramakrishnan (1996) describe 19 types of exibility. There are several
notions such as, range, mobility, uniformity, etc. associated with exibility, which
may be summarized as shown in gure 1.
From the above description, it can be seen that exibility is a multidimensional
polymorphous concept, which is dicult to comprehend, model and measure. There
are many types of exibilities, each type having several types of measures. In general,
exibility in manufacturing can be classied as being either product-related or
process-related. Product-related exibility refers to the variety of manufacturing
options emanated due to the product characteristics, while process-related exibility
emanates from the characteristics of the manufacturing system. A study by Karl
Ulrich (1995) established that, contrary to conventional belief, the manufacturing
systems ability to create variety mainly resides, not in the exibility of the equipment
in the factory, but in the architecture of the product. Similarly, an empirical study
carried out by Thomke (1997) established that the design exibility could greatly
3133 Lead-time reduction in exible systems
enhance the new product development performance. He dened the design exibility
as the incremental cost and time of modifying a design. Closely related to the idea of
design exibility is the Platform Design approach (Lapid 1997, Gonzalez-Zugasti
and Otto 2000) where a large number of variants are derived from a core platform.
Another important direction is set-based thinking (Liker et al. 1996, Sobek 1996),
which advocates communicating design decisions in sets rather than as points, which
requires an IT intensive environment (Gerwin 1993).
From these studies, it is clear that design carries with it certain specic charac-
teristics, which enables its translation into products/services in a more exible
manner. We call this design exibility. Based on the various elaborations of exibility
types, Wadhwa and Rao (2000) indicate that more emphasis is deserved by this
exibility. Since design has the greatest leverage on the product performance, cost
and quality, design exibility will have the greatest potential for performance
improvement. Among the product-related exibilities, sequencing exibility is an
important type of exibility. Hence, we have focused in our current study on
evaluating the performance of sequencing exibility in reducing the manufacturing
lead time.
2.2 Sequencing exibility
Sequencing exibility (SF) refers to the possibility of interchanging the order in
which required manufacturing operations are performed on a job. In general, each
type of part is associated with a single process plan consisting of an ordered sequence
of operations as determined by the technological constraints in manufacturing of the
part type. However, in most cases, a single process plan is followed as a matter of
uncertainty
affects
performance
operation
at optimum
state
elements
of known
certainty
elements
of known &
unknown
uncertainty
evolving
environment
several
types of
uncertainty
several
types of
flexibility
required
different
levels of
uncertainty
different
levels of
flexibility
required
optimum
static
performance
ability to
redefine
uncertainty
flexibility is
required to
counter the
affect of
uncertainty
flexibility delivery
requires
ability to change
from optimum in a
reversible manner
change
degrades
performance
flexibility
minimizes
performance
degradation
due to change
change
entails
transition
penalty
flexibility
minimizes
transition
penalties
optimum
dynamic
performance
mobility
uniformity
range
the transition
must be rapid
enough to ensure
timely response
agility
proactive
application
of flexibility
uncertainty
affects
performance
operation
at optimum
state
elements
of known
certainty
elements
of known &
unknown
uncertainty
evolving
environment
several
types of
uncertainty
several
types of
flexibility
required
different
levels of
uncertainty
different
levels of
flexibility
required
optimum
static
performance
ability to
redefine
uncertainty
flexibility is
required to
counter the
affect of
uncertainty
flexibility delivery
requires
ability to change
from optimum in a
reversible manner
change
degrades
performance
flexibility
minimizes
performance
degradation
due to change
change
entails
transition
penalty
flexibility
minimizes
transition
penalties
optimum
dynamic
performance
mobility
uniformity
range
the transition
must be rapid
enough to ensure
timely response
agility
proactive
application
of flexibility
Figure 1. Summary of exibility notions.
3134 S. Wadhwa et al.
convenience or due to an established practice. Depending upon the design of the
part, it is possible to alter the sequence of operations to be performed. This gives rise
to the sequencing exibility. The importance of the sequencing exibility lies in the
fact that this can be built into the product design, thereby avoiding the much costlier
option of building other types of exibility, like the routing exibility, into the
manufacturing system. This motivated us to study the performance of sequencing
exibility in reducing the manufacturing lead time.
The potential of sequencing exibility in enhancing the manufacturing system
performance has been recognized by the researchers, and eorts are under way to
understand the underlying mechanisms with a view to exploit this type of exibility.
Rachamadugu and Schriber (1990) studied the eects of sequencing exibility on
the performance of various scheduling rules in environments such as job shops and
exible manufacturing systems. They have found that relative dierences in the
performance of various scheduling rules diminish, and the relative rankings also
change as sequencing exibility increases. They have noted that since sequencing
exibility is product-specic, it exists in conventional manufacturing systems as well
as in exible manufacturing systems. In the past, organizational control issues and
the high cost of early generation information systems precluded the exploitation of
sequencing exibility even when it was available in the product structure. However,
recent advances in information technology and the declining cost of information
systems make it possible to use sequencing exibility at the operational level. They
further noted that if the benets of sequencing exibility in terms of reduced ow
times and inventories are suciently great, this has implications for designing
products in such a way as to maximize the potential sequencing exibility in
manufacturing the products. Schriber (1991) proposed an approach for modelling
of perfect sequencing exibility in a scheduling environment. Benjaafar and
Ramakrishnan (1996) introduced several representation and measurement schemes
for sequencing exibility and studied the relationship between sequencing exibility
and system performance under a variety of design assumptions and operating
conditions.
For the purpose of our studies, we have represented sequencing exibility level
(SFL) in terms of the number of operations that can be interchanged in a process
plan, as shown in gure 2(a)(c). For example, one given sequence at a time means
SFL1. Similarly 2, 3 alternative sequences signify, SFL2, SFL3, respectively.
A job may be completed with one or more sequence of processes Pr
1
. . . Pr
n
.
2.3 Lead time
Lead time is dened as the time between the beginning of a process or project and the
appearance of its results. In the context of theory of FLR, lead time may be under-
stood as the time between the arrival of a customer order and the completion of
order fulllment. Our current research on exibility indicates that exibility can be
used not only for eectively managing the changes but also to enhance the lead-time
performance of the manufacturing system in a proactive manner. For instance,
Wadhwa and Bhagwat (1998) indicate that manufacturing exibility in the form
of routing and machine exibility can be judiciously exploited towards lead-time
reduction in multi-product manufacturing systems. It is shown that this is achievable
3135 Lead-time reduction in exible systems
through a dynamic control of the ow of products and resources. Several studies can
be found in the literature that explores the relationship between exibility and lead
time. However, so far the authors have not come across any eort aimed at under-
standing the exact mechanism of exibility-enabled lead-time reduction. This paper
intends to bridge this gap.
2.4 Processor load imbalance
Processor load imbalance indicates the level of uneven distribution of load among
the processors. This can be measured as the standard deviation of all the machine
utilizations over the specied time bucket.
2.5 Process concurrency
2.5.1 What is concurrency?. The term concurrency refers to acting or existing
together simultaneously. In a manufacturing system, concurrency refers to carrying
out multiple processes on multiple processors simultaneously, and as more and more
processes are carried out concurrently, the overall lead time will decrease, as shown
in gure 3(a)(c).
2.5.2 Constraints against concurrency. Two processes A and B are carried out
sequentially for any of the following reasons:
1. as a matter of established practice or habit, or for convenience;
2. due to functional boundaries;
3. B requires certain information from A;
Pr
1
OR
Pr
2
Pr
3
OR
Finish
Finish
Finish
OR OR
1 2 3
Pr
1
OR
Pr
2
Pr
3
OR OR OR
1 2 3
Pr
1
OR
Pr
2
Pr
3
OR
Start
Start
Start
OR OR
1 2 3
0
1
2
(a)
(b)
(c)
Figure 2. Representation of dierent levels of sequencing exibility. (a) Process plan for
sequencing exibility level (SFL) 0. (b) Process plan for sequencing exibility level
(SFL) 1. (c) Process plan for sequencing exibility level (SFL) 2.
3136 S. Wadhwa et al.
4. B requires certain decision from A;
5. B requires certain materials from A;
6. B requires certain resource which are currently being used by A;
7. there is a need for synchronization of certain events of A and B, which may
result in certain lead-lag precedence relationships between A and B.
Lead-time reduction eorts generally involve overcoming some or all of the
above constraints. Proactive application of exibility is an important step in this
direction. To enable an understanding and measurement of concurrency, we have
evolved a method of measuring process concurrency, as shown in gure 4.
t
0
t
1
t
2
t
3
. t
j
. .
.
. . . . t
n
Process - A
Process - B
t
0
t
1
t
2
t
3
. t
j
. .
.
. . . . t
n
Process - A
Process - B
t
0
t
1
t
2
t
3
. t
j
. .
.
. . . . t
n
Process - A
Process - B
(a)
(b)
(c)
Figure 3. (a) Sequential processes (concurrency 0). (b) Overlapped processes (partial
concurrency). (c) Parallel processes (maximum concurrency).
3137 Lead-time reduction in exible systems
A customers order comprises a set of jobs to be completed and delivered by a
certain time (due date). Each of these jobs may require one or more processes to be
performed on one or more processors (machines). Accordingly, the jobs will be
scheduled on the processors depending on the process plans and the routing
exibility available within the system. The resulting schedule is prepared in the
form of a Gantt chart as shown in gure 4. Usually, the scheduling is done for a
given number of processors (processor p
1
to p
m
) in time buckets, and the resulting
schedule for a time bucket from time t
0
to t
n
will be a Gantt chart, as shown in
gure 4. From the Gantt chart, we can identify the state (busy or idle) of each
processor at any given instance of time. We denote these states (s
ij
) as 1 or 0,
respectively, depending on whether the processor is busy or idle. Column summation
of these states (Ss
ij
) will show us how many machines are busy at any given slice of
time. This is used to build a frequency distribution table, as shown in gure 4. The
mean of this frequency distribution indicates the level of concurrency existing within
the system for a given time bucket.
2.5.3 Understanding the theory of FLR. Having described the key concepts under-
lying the theory, the theory of FLR may be understood as follows. Suppose a job J
has to be performed by a set of processors (machines) P
1
. . . P
m
, and the completion
of the job requires processes Pr
1
. . . Pr
n
to be performed on the job. Let us assume
that we have both exibility and real-time control support to build the required
concurrency among these processes. This is relevant to most exible systems that
have adequate IT support. Wadhwa and Bhagwat (1998) have discussed the other
case. Now, for our real-time control case, among several factors that may inuence
the manufacturing lead time, two factors, namely the level of concurrency among the
t
0
P
1
P
2
P
3
.
P
i
.
P
m
t
1
t
2
t
3
. t
j
. . . . . . . t
n
s
ij
s
ij
s
ij
s
ij
= [0,1] = [0,1,2,.m]
0
10
20
30
40
50
60
70
80
1 2 3 4 5 6 7
Value(n
i
) Frequency(f
i
) n
i
x f
i
1 f
1
1 x f
1
2 f
2
2 x f
2
3 f
3
3 x f
3
. . .
i f
i
i x f
i
. . .
m f
m
m x f
m
Concurrency ( ) =
n
i
f
i
f
i
t
0
P
1
P
2
P
3
.
P
i
.
P
m
t
1
t
2
t
3
. t
j
. . . . . . . t
n
s
ij
s
ij
s
ij
s
ij
= [0,1] = [0,1,2,.m]
t
0
P
1
P
2
P
3
.
P
i
.
P
m
t
1
t
2
t
3
. t
j
. . . . . . . t
n
s
ij
s
ij
s
ij
s
ij
s
ij
s
ij
s
ij
s
ij
= [0,1] = [0,1,2,.m]
0
10
20
30
40
50
60
70
80
1 2 3 4 5 6 7
Value(n
i
) Frequency(f
i
) n
i
x f
i
1 f
1
1 x f
1
2 f
2
2 x f
2
3 f
3
3 x f
3
. . .
i f
i
i x f
i
. . .
m f
m
m x f
m
Concurrency ( ) =
n
i
f
i
f
i
0
10
20
30
40
50
60
70
80
1 2 3 4 5 6 7
Value(n
i
) Frequency(f
i
) n
i
x f
i
1 f
1
1 x f
1
2 f
2
2 x f
2
3 f
3
3 x f
3
. . .
. . .
i f
i
i x f
i
m f
m
m x f
m
Concurrency ( ) =
n
i
f
i
f
i
Concurrency () =
S
S
n
i
f
i
n
i
f
i
n
i
f
i
n
i
f
i
Sn
i
f
i
f
i
f
i
f
i
f
i
S
S
f
i
Figure 4. Method of computation of concurrency.
3138 S. Wadhwa et al.
processes and the level of load unbalancing among the processors, will have much
greater signicance. It is our thesis that, in order to reduce the lead time, we must
simultaneously increase the process concurrency and reduce the processor load
imbalance, as shown in gure 5. The system control should be designed to do this.
In this process, the inexibility in the system acts as a constraint, and thus exibility
helps in reducing the lead time. However, for a given level of exibility, the benet
depends on how well the system control is able to exploit exibility. In the context of
a manufacturing system, the interactions between these underlying concepts may be
as shown in gure 6.
The part characteristics gives rise to a certain amount of sequencing exibility
which, when used with the help of an intelligent control, will enable dynamic
re-routing of the parts within the manufacturing system. The manufacturing
system characteristics, part mix to be manufactured and the corresponding process
plans together determine the static load balance among the processors. During the
dynamic re-routing of the parts using sequencing exibility, the process concurrency
is then altered, which, together with the existing load balance, aects the manufac-
turing lead time. As a whole, ve kinds of interactions may be identied as shown in
gure 6.
The theory of FLR implies that:
1. An increase in the level of sequencing exibility, when exploited through an
appropriate control, will result in the increase in level of process concurrency.
2. An increase in the level of exibility can be used to decrease the processor
load imbalance.
3. Within certain ranges of processor load imbalance and process concurrency,
a decrease in the processor load imbalance is associated with an increase in
the process concurrency.
Lead Time
Load Unbalance
Concurrency
Inflexibility
in System
System
Control
In order to reduce lead time,
must be increased and load unbalance must be decreased,
simultaneously
System
Control
Figure 5. Basic idea behind the theory of FLR.
3139 Lead-time reduction in exible systems
4. When the process concurrency is below a certain level, a decrease in the load
imbalance will result in an increase in the lead time. Also, when the process
concurrency is above a certain level, any amount of decrease in the load
imbalance will not result in an increase in the lead time.
5. When the load imbalance is below a certain level, an increase in the process
concurrency will result in a decrease in the lead time. Also, when the load
balance is above a certain level, any amount of increase in the process con-
currency will not result in a decrease in the lead time.
6. An increase in the level of exibility will simultaneously lead to an increase in
the level of process concurrency and a decrease in the level of processor load
balance, which in turn will result in a decrease in the lead time.
Thus, as per the theory of FLR, we need to simultaneously increase the process
concurrency and reduce the processor load imbalance to reduce the lead time.
Increasing the level of exibility is one way of accomplishing this. It is possible to
accomplish this using other means, for instance, by increasing the level of intelligence
in the control, by controlling the part-mix, etc. Operation of the theory of FLR with
all such options is beyond the scope of this paper. However, part 3 of this paper
discusses some of the important implications of theory of FLR for exible systems.
3. Part 2: understanding the theory of FLR using simulations
For our endeavour towards proposing the theory of FLR in the context of sequen-
cing exibility, we have carried out extensive simulation experimentation and
analysed the results with the help of MANOVA and Path Analysis, to understand
various types of interactions. The following six hypotheses are proposed to support
the theory.
Process
Concurrency
Makespan
Performance
Routing
Flexibility
Processor
Load Balance
Dynamic
Rerouting
Intelligent
Control
Part Mix &
Routings
Mfg. System
Characteristics
1 4
3
2
5
Process
Concurrency
Process
Concurrency
Makespan
Performance
Sequencing
Flexibility
Processor
Load Balance
Processor
Load Balance
Dynamic
Rerouting
Dynamic
Rerouting
Part Mix &
Routings
Mfg. System
Characteristics
11 44
33
22
55
Figure 6. Interactions between exibility, process concurrency and processor load
unbalance.
3140 S. Wadhwa et al.
3.1 Research hypotheses
H1: An increase in the level of exibility, when exploited through an appropriate
control, will result in an increase in the level of process concurrency.
H2: An increase in the level of exibility, when exploited through an appropriate
control, will result in the decrease in processor load imbalance.
H3: Within certain ranges of processor load imbalance and process concurrency,
a decrease in the processor load imbalance is associated with an increase in the
process concurrency.
H4: (a) When the process concurrency is below certain level, a decrease in the load
imbalance will result in an increase in the lead time. (b) When the process
concurrency is above certain level, any amount of decrease in the load
imbalance will not result in an increase in the lead time.
H5: (a) When the load imbalance is below a certain level, an increase in the process
concurrency will result in a decrease in the lead time. (b) When the load
imbalance is above certain level, any amount of increase in the process
concurrency will not result in a decrease in the lead time.
H6: (a) An increase in the level of exibility will result in a decrease in the lead
time. (b) An increase in the level of routing exibility will simultaneously lead
to an increase in the level of process concurrency and a decrease in the level
of processor load imbalance, which in turn will result in a decrease in the
lead time.
Simulation experiments were carried out under several possible conditions,
including unidirectional ow shop conditions, general job shop conditions, under
the conditions of part load balancing and unbalancing, under the conditions of
machine load balancing and unbalancing, under the conditions of seven levels of
shop load ranging from very low shop load to very high shop load, under six levels
of routing exibility and under six levels of sequencing exibility. A total of 10 080
simulation experiments have been carried using 40 sets of specially prepared data, to
cover all the above possibilities. The results obtained have been analysed using six-way
MANOVA, and all the interactions have been studied. For further conrmation of
theory, the model has been subjected to path analysis, and all possible paths from
routing exibility to lead time have been studied. The results of the above study
validated the proposed theory of FLR. Alarge number of extensive simulation studies
covering several possible operating conditions and data have been analysed in three
stages, viz. visual graphic analysis, MANOVA and path analysis, as shown in table 1.
We now present some of the important results that may help towards motivating
other researchers to learn and contribute to the evolution of the theory of FLR. Such
a theory may help both researchers and practitioners to proactively deploy exibility
in systems. Wadhwa and Rao (2002) have proposed a exibility maturity model for
industry organizations to benet eectively from the use of exibility. The evolution
of FLR theory may help the practitioners to identify where, when and how exibility
may help towards time-based competitiveness. This paper is only initiating some
important leads towards the evolution of this theory. Hence, we have presented
most of the important interactions of sequencing exibility with other parameters
such as process concurrency, nature of job ows (ow shop vs. job shop), part load
balancing, machine load balancing, shop load, processor load imbalance, etc.
3141 Lead-time reduction in exible systems
The results summarized with graphs are used to support various hypotheses.
Further, the underlying mechanisms of how sequencing exibility leads to lead-
time reduction are also described.
Figure 7 shows the main eect of sequencing exibility on process concurrency.
It can be seen that process concurrency increases with increasing level of sequencing
exibility. This supports the hypothesis H1. However, the rate of increase decreases
as we approach higher levels of sequencing exibility. Hence, the relationship follows
the law of diminishing returns. To further illustrate the relationship between the
sequencing exibility and concurrency, we present the three simplied models as
shown in gures 810, corresponding to sequencing exibility levels of zero, one
and two. It can be seen that as the level of sequencing exibility increases, concur-
rency increases, and the lead time decreases. When the sequencing exibility level is 0
(gure 8), the concurrency is 1.56, and the lead time is 22 units of time. When the
sequencing exibility level is increased to 1 (gure 9), the concurrency level has
increased to 2.57, and the lead time is reduced to 14. As the sequencing exibility
Table 1. Summarized results of analysis.
Hypotheses Visual graphic analysis MANOVA Path analysis
H1 Yes Yes
H2 Yes Yes
H3 Yes Yes
H4 Yes Yes
H5 Yes Yes
H6 Yes Yes
Plot of Means (unweighted)
SFL Main Effect
Rao R (15,22256)=346.07; p<0.000
Sequencing Flexibility Level
P
r
o
c
e
s
s
C
o
n
c
u
r
r
e
n
c
y
5.40
5.45
5.50
5.55
5.60
5.65
5.70
1 2 3 4 5 6
Figure 7. Main eect of sequencing exibility on process concurrency.
3142 S. Wadhwa et al.
level is further increased to 2 (gure 10), the concurrency level has gone up to 2.77,
and the lead time is reduced to 13. It is our research endeavour to develop an
understanding of this exibility-enabled concurrency and its role in the reduction
in the manufacturing lead time.
Figure 11 shows the interactions between the nature of jobs ow through
the manufacturing system and the sequencing exibility. It can be seen that
Part-1
Part-2
Part-3
Process-1 Process-2 Process-3
Pr
11
Pr
21
Pr
31
Pr
32
Pr
33
Pr
22
Pr
12
Pr
13
Pr
23
P
1
(3) P
2
(3) P
3
(1)
P
1
(1) P
2
(3) P
3
(1)
P
1
(1) P
2
(2) P
4
(3)
Process-4
Pr
14 P
5
(2)
Pr
24 P
4
(3)
Pr
34 P
3
(3)
Process-5 Process-6
Pr
15 P
4
(3)
Pr
16 P
6
(1)
Pr
26 P
6
(1)
Pr
36 P
6
(1)
Pr
35 P
5
(1)
Pr
25 P
5
(3)
Value Frequency
1 1
2 4
3 9
4 0
5 0
6 0
0
2
4
6
8
10
1 2 3 4 5 6
Concurrency ( ) = 2.57
t
0
P
1
P
2
P
3
P
4
P
5
P
6
t
1
t
2
t
3
t
4
t
5
t
6
t
7
t
8
t
9
t
10
t
11
t
12
s
ij
2 2 2 3 3 3 3 3 3 2
Pr
22
t
13
t
14
t
15
t
16
t
17
t
18
Pr
13
Pr
14
Pr
15
Pr
16
Pr
23
Pr
24
Pr
25
Pr
26
Pr
32
Pr
33
Pr
34
Pr
35
Pr
36
Pr
11
Pr
12
Pr
21
t
19
t
20
t
21
t
22
t
23
t
24
t
25
3 3 3 1
Pr
31
Pr
1
OR Pr
2
Pr
3
OR Start Finish OR OR Pr
1
OR OR Pr
2
Pr
3
OR OR Start Start Finish Finish OR OR OR OR
Part-1
Part-2
Part-3
Process-1 Process-2 Process-3
Pr
11
Pr
21
Pr
31
Pr
32
Pr
33
Pr
22
Pr
12
Pr
13
Pr
23
P
1
(3) P
2
(3) P
3
(1)
P
1
(1) P
2
(3) P
3
(1)
P
1
(1) P
2
(2) P
4
(3)
Process-4
Pr
14 P
5
(2)
Pr
24 P
4
(3)
Pr
34 P
3
(3)
Process-5 Process-6
Pr
15 P
4
(3)
Pr
16 P
6
(1)
Pr
26 P
6
(1)
Pr
36 P
6
(1)
Pr
35 P
5
(1)
Pr
25 P
5
(3)
Part-1
Part-2
Part-3
Process-1 Process-2 Process-3
Pr
11
Pr
21
Pr
31
Pr
32
Pr
33
Pr
22
Pr
12
Pr
13
Pr
23
P
1
(3) P
2
(3) P
3
(1)
P
1
(1) P
2
(3) P
3
(1)
P
1
(1) P
2
(2) P
4
(3)
Process-4
Pr
14 P
5
(2)
Pr
24 P
4
(3)
Pr
34 P
3
(3)
Process-5 Process-6
Pr
15 P
4
(3)
Pr
16 P
6
(1)
Pr
26 P
6
(1)
Pr
36 P
6
(1)
Pr
35 P
5
(1)
Pr
25 P
5
(3)
Value Frequency
1 1
2 4
3 9
4 0
5 0
6 0
Value Frequency
1 1
2 4
3 9
4 0
5 0
6 0
Concurrency () = 2.57
t
0
P
1
P
2
P
3
P
4
P
5
P
6
t
1
t
2
t
3
t
4
t
5
t
6
t
7
t
8
t
9
t
10
t
11
t
12
s
ij
2 2 2 3 3 3 3 3 3 2
Pr
22
t
13
t
14
t
15
t
16
t
17
t
18
Pr
13
Pr
14
Pr
15
Pr
16
Pr
23
Pr
24
Pr
25
Pr
26
Pr
32
Pr
33
Pr
34
Pr
35
Pr
36
Pr
11
Pr
12
Pr
21
t
19
t
20
t
21
t
22
t
23
t
24
t
25
3 3 3 1
Pr
31
t
0
P
1
P
2
P
3
P
4
P
5
P
6
t
1
t
2
t
3
t
4
t
5
t
6
t
7
t
8
t
9
t
10
t
11
t
12
s
ij
2 2 2 3 3 3 3 3 3 2
Pr
22
t
13
t
14
t
15
t
16
t
17
t
18
Pr
13
Pr
14
Pr
15
Pr
16
Pr
23
Pr
24
Pr
25
Pr
26
Pr
32
Pr
33
Pr
34
Pr
35
Pr
36
Pr
11
Pr
12
Pr
21
t
19
t
20
t
21
t
22
t
23
t
24
t
25
3 3 3 1
Pr
31
Pr
1
OR Pr
2
Pr
3
OR Start Finish OR OR Pr
1
OR OR Pr
2
Pr
3
OR OR Start Start Finish Finish OR OR OR OR
s
ij
Figure 9. Lead time and concurrency for SFL1.
Part-1
Part-2
Part-3
Process-1 Process-2 Process-3
Pr
11
Pr
21
Pr
31
Pr
32
Pr
33
Pr
22
Pr
12
Pr
13
Pr
23
P
1
(3) P
2
(3) P
3
(1)
P
1
(1) P
2
(3) P
3
(1)
P
1
(1) P
2
(2) P
4
(3)
Process-4
Pr
14 P
5
(2)
Pr
24 P
4
(3)
Pr
34 P
3
(3)
Process-5 Process-6
Pr
15 P
4
(3)
Pr
16 P
6
(1)
Pr
26 P
6
(1)
Pr
36 P
6
(1)
Pr
35 P
5
(1)
Pr
25 P
5
(3)
Value Frequency
1 11
2 11
3 1
4 0
5 0
6 0
0
2
4
6
8
10
12
Concurrency ( ) = 1.56
t
0
P
1
P
2
P
3
P
4
P
5
P
6
t
1
t
2
t
3
t
4
t
5
t
6
t
7
t
8
t
9
t
10
t
11
t
12
s
ij
1 1 1 2 2 1 2 2 2 3
Pr
22
Pr
31
t
13
t
14
t
15
t
16
t
17
t
18
Pr
13
Pr
14
Pr
15
Pr
16
Pr
23
Pr
24
Pr
25
Pr
26
Pr
32
Pr
33
Pr
35
Pr
36
Pr
11
Pr
12
Pr
21
t
19
t
20
t
21
t
22
t
23
t
24
t
25
2 1 2 1 1 2 2 2 2 1 1 1 1
Pr
34
Pr
1
OR Pr
2
Pr
3
OR Start OR OR Finish Pr
1
OR OR Pr
2
Pr
3
OR OR Start Start OR OR OR OR Finish Finish
Part-1
Part-2
Part-3
Process-1 Process-2 Process-3
Pr
11
Pr
21
Pr
31
Pr
32
Pr
33
Pr
22
Pr
12
Pr
13
Pr
23
P
1
(3) P
2
(3) P
3
(1)
P
1
(1) P
2
(3) P
3
(1)
P
1
(1) P
2
(2) P
4
(3)
Process-4
Pr
14 P
5
(2)
Pr
24 P
4
(3)
Pr
34 P
3
(3)
Process-5 Process-6
Pr
15 P
4
(3)
Pr
16 P
6
(1)
Pr
26 P
6
(1)
Pr
36 P
6
(1)
Pr
35 P
5
(1)
Pr
25 P
5
(3)
Part-1
Part-2
Part-3
Process-1 Process-2 Process-3
Pr
11
Pr
21
Pr
31
Pr
32
Pr
33
Pr
22
Pr
12
Pr
13
Pr
23
P
1
(3) P
2
(3) P
3
(1)
P
1
(1) P
2
(3) P
3
(1)
P
1
(1) P
2
(2) P
4
(3)
Process-4
Pr
14 P
5
(2)
Pr
24 P
4
(3)
Pr
34 P
3
(3)
Process-5 Process-6
Pr
15 P
4
(3)
Pr
16 P
6
(1)
Pr
26 P
6
(1)
Pr
36 P
6
(1)
Pr
35 P
5
(1)
Pr
25 P
5
(3)
Value Frequency
1 11
2 11
3 1
4 0
5 0
6 0
Value Frequency
1 11
2 11
3 1
4 0
5 0
6 0
0
2
4
6
8
10
12
1 2 3 4 5 6
Concurrency () = 1.56
t
0
P
1
P
2
P
3
P
4
P
5
P
6
t
1
t
2
t
3
t
4
t
5
t
6
t
7
t
8
t
9
t
10
t
11
t
12
s
ij
1 1 1 2 2 1 2 2 2 3
Pr
22
Pr
31
t
13
t
14
t
15
t
16
t
17
t
18
Pr
13
Pr
14
Pr
15
Pr
16
Pr
23
Pr
24
Pr
25
Pr
26
Pr
32
Pr
33
Pr
35
Pr
36
Pr
11
Pr
12
Pr
21
t
19
t
20
t
21
t
22
t
23
t
24
t
25
2 1 2 1 1 2 2 2 2 1 1 1 1
Pr
34
t
0
P
1
P
2
P
3
P
4
P
5
P
6
t
1
t
2
t
3
t
4
t
5
t
6
t
7
t
8
t
9
t
10
t
11
t
12
s
ij
s
ij
1 1 1 2 2 1 2 2 2 3
Pr
22
Pr
31
t
13
t
14
t
15
t
16
t
17
t
18
Pr
13
Pr
14
Pr
15
Pr
16
Pr
23
Pr
24
Pr
25
Pr
26
Pr
32
Pr
33
Pr
35
Pr
36
Pr
11
Pr
12
Pr
21
t
19
t
20
t
21
t
22
t
23
t
24
t
25
2 1 2 1 1 2 2 2 2 1 1 1 1
Pr
34
Pr
1
OR Pr
2
Pr
3
OR Start OR OR Finish Pr
1
OR OR Pr
2
Pr
3
OR OR Start Start OR OR OR OR Finish Finish