Analysis of Manufacturing Systems

using Chi 3.0

I.J.B.F. Adan, A.T. Hofkamp, J.E. Rooda and J. Vervoort

Oct 2012

Technische Universiteit Eindhoven, Department of Mechanical Engineering

Systems Engineering Group, P.O. Box 513, 5600 MB Eindhoven, The Netherlands
http://se.wtb.tue.nl/
[email protected]

ii

Contents
Acknowledgements

vii

1 Introduction

1.1

Manufacturing Systems (MSs) . . . . . . . . . . . . . . . . . . . . . . . . . . .

1.2

Analysis of Manufacturing Systems (AMS)

. . . . . . . . . . . . . . . . . . .

2 Analytical models of deterministic manufacturing systems

2.1

Manufacturing system principles . . . . . . . . . . . . . . . . . . . . . . . . .

2.2

Analysis of deterministic manufacturing systems . . . . . . . . . . . . . . . . 13

2.3

Exercises

2.4

Additional exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24

3 Analytical models of stochastic manufacturing systems

31

3.1

Classification of manufacturing queueing models . . . . . . . . . . . . . . . . 31

3.2

Queueing relations for single-machine workstation . . . . . . . . . . . . . . . . 32

3.3

Queueing relations for a flowline . . . . . . . . . . . . . . . . . . . . . . . . . 35

3.4

Identical parallel machines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37

3.5

Batch machines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39

3.6

Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41

3.7

Exercises

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43

4 Single workstation

47

4.1

Single machine model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47

4.2

Single machine with varying process time . . . . . . . . . . . . . . . . . . . . 49

4.3

Single machine with buffer

4.4

Displaying simulated behaviour . . . . . . . . . . . . . . . . . . . . . . . . . . 54

4.5

Exercises

. . . . . . . . . . . . . . . . . . . . . . . . . . . . 51

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56

iii

iv

Contents

5 Measuring performance

61

5.1

Mean throughput . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61

5.2

Mean flowtime . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62

5.3

Mean utilization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64

5.4

Mean wip-level . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66

5.5

Variance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68

5.6

Measuring other quantities

5.7

Simulation vs Approximation (1): single machine workstation . . . . . . . . . 69

5.8

Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71

5.9

Exercises

. . . . . . . . . . . . . . . . . . . . . . . . . . . . 69

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71

6 Machines in series

73

6.1

Deterministic flowline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73

6.2

Unbalanced flowline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77

6.3

Machines in series without buffers

6.4

Simulation vs Approximation (2): flowline . . . . . . . . . . . . . . . . . . . . 79

6.5

Exercises

6.6

Case . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88

. . . . . . . . . . . . . . . . . . . . . . . . 78

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83

7 Machines in parallel

91

7.1

Deterministic machines in parallel . . . . . . . . . . . . . . . . . . . . . . . . 91

7.2

Analytical vs Simulation (3): Parallel machines . . . . . . . . . . . . . . . . . 93

7.3

Parallel machines for different lot types . . . . . . . . . . . . . . . . . . . . . 94

7.4

Exercises

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97

8 Other configurations

99

8.1

Bypassing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100

8.2

Backtracking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103

8.3

Assembly line . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106

8.4

Jobshop . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108

8.5

Exercises

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111

9 Extending machine models

117

9.1

Representing process time by a distribution . . . . . . . . . . . . . . . . . . . 117

9.2

Machine failure and repair . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127

Contents

10 Case: (Part of ) a waferfab

135

10.1 The manufacturing of Integrated Circuits (ICs) . . . . . . . . . . . . . . . . . 136

10.2 Wafer processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136
10.3 The oxide furnace and lithographic equipment . . . . . . . . . . . . . . . . . . 138
10.4 Analytical approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139
10.5 Simulation-based analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
Bibliography

145

vi

Contents

Acknowledgements
This book is the result of years of research within the Systems Engineering Group of the
Department of Mechanical Engineering of the Eindhoven University of Technology. Credits
for the design of the language go to all members of the -club. In particular, we would like
to mention the following contributions. J.M. van de Mortel-Fronczak stood at the base of
the language [vdMFR95]. N.W.A. Arends was the first to describe the formalism [Are96].
W. Alberts and G.A. Naumoski wrote the first compiler (0.3) [AN98]. A.T. Hofkamp wrote
the real-time compiling system [Hof01] and the latest compiler (Chi 3). V. Bos and J.J.T.
Kleijn helped in providing a formal framework for [BK00]. D.A. van Beek gave suggestions
for improving the language. Finally, close cooperation of the Systems Engineering Group
with the Parallel Programming Group (formerly lead by M. Rem) and the Formal Methods
Group (lead by J.C. Baeten) has improved the quality of the language.
In addition, we would like to thank all (former) students and (former) members of the
Systems Engineering Group that have performed several projects involving the modelling
and analysis of manufacturing systems using Chi in research and engineering environments.
Regarding this book, we thank L.F.P. Etman for his contribution to Chapter 3 and for
preparing a number of exercises. We thank E.J.J. van Campen for providing us with photographs of the Crolles2 waferfab.

Chi 3
This edition is based on the language Chi 3. The original text has been modified, with
minimal effort, to make the text suitable for Chi 3.
Some chapters have been deleted from this edition.

vii

viii

Acknowledgements

Chapter 1

Introduction
The title of this book is analysis of manufacturing systems. In this chapter we mark out
the main focus of the book.

1.1

Manufacturing Systems (MSs)

Manufacturing stems from the Latin words manus (hand) and factus (make). Nowadays,
by manufacturing we mean the process of converting raw material into a physical product
(we do not consider services to be manufacturable). A system is a collection of elements
forming a unified whole. Thus, a manufacturing system refers to any collection of elements
that converts raw material into a product. We can identify manufacturing systems at four
different levels.
At factory level the manufacturing system is the factory (also referred to as plant,
fabricator, or shortly fab). The elements of the system are areas and (groups of)
machines.
At area level the manufacturing system is an area of the factory with several machines
or groups of machines (cells). The elements of the system are individual machines.
At cell level the manufacturing system is a group of machines, that are typically
scheduled as one entity.
At machine level, the manufacturing system is the individual machine (also referred
to as equipment or tools). The elements of the system are machine components.
These levels constitute a hierarchical breakdown of the factory, see Figure 1.1.
In this book we will focus on manufacturing systems at area and cell level. Though, many
parts of this book are also applicable to analysis of entire factories or individual machines.
For instance, the analysis of the output (for instance the number of products processed per
hour) of a factory shows many similarities to analyzing the output of an area, or the output
of machine. Scheduling tasks at machine level is in many ways analogous to scheduling
orders at factory level or scheduling jobs at area level.

Chapter 1. Introduction

Level
4 factory

3 area

manufacturing
cell

1 machine

Figure 1.1: Manufacturing systems regarded at different levels

The types of manufacturing systems that can be analyzed using the approach described in
this book are as diverse as there are products. Some examples are:
a wafer fab (factory level), the lithography area (area level), a waferstepper (machine),
see Figures 1.2 and 1.3,
a CD-R fab (factory level), the CD-R mastering area (area level), a CD-R printing
machine(machine level), see Figure 1.4,
an automotive assembly factory (factory level), the bodywork assembly line (area
level), the welding robots (cell level), a single welding robot (machine level), see Figure 1.5.

Figure 1.2: Lithographic area of the Crolles2 waferfab in Grenoble, France

Figure 1.3: Waferinspection machine and waferstepper in the lithographic area of the
Crolles2 waferfab

1.2. Analysis of Manufacturing Systems (AMS)

Figure 1.4: CD optical mastering area, CD printing machine, CD packaging machine (source:
http://www.sonopress.de)

Figure 1.5: Automotive assembly line (source: http://www.mazda.com)

In this book we consider different types of manufacturing systems. Depending on the specific
industry we are in, we see for example shoes, a pod of wafers, or a printed-circuit board
moving through the factory. Throughout this book we use the term lot to refer to the
common transfer unit in the manufacturing system. If a machine processes multiple lots at
the same time, we say the machine processes batches. A batch is a set of lots.

1.2

Analysis of Manufacturing Systems (AMS)

Some typical design questions when designing or optimizing a manufacturing system are:
What machines do I choose? Do I choose a cheap machine that breaks down often, or
a more expensive machine, that breaks down less often?
How many machines do I use? Do I choose a single high-capacity machine, do I choose
multiple low-capacity machines in parallel, or a combination?
What configuration of workstations do I choose? Do I choose a flowline, a jobshop, or
something in between?
Do I want buffers and if so of what size?
What control strategy do I use for my plant? Do I release new orders at a fixed rate
and if so at what rate? Do I release orders according to a precalculated schedule or
do I release orders based on the plant status?

Chapter 1. Introduction

This book does not go into much detail on the best answers, but presents a way to model and
analyze a manufacturing system to help you answer these questions yourself. We analyze
manufacturing systems with one objective: to evaluate and compare alternative manufacturing system designs and decide on the best alternative.
Question is: on what criteria do we evaluate and compare manufacturing systems? Many
different performance criteria exist. Some commonly used performance indicators are:
throughput : the number of lots per time-unit that leave the manufacturing system,
flowtime : the time it takes a lot to travel through the system,
coefficient of variation on the flowtime (c ): the amount of variability on the flowtime
of a lot,
utilisation u: the fraction of time a machine is processing lots, and
work-in-process or wip-level w: the number of lots in the manufacturing system.
More on these and other performance indicators can be found in [GH85], [Tij94], [HS01]
and in the remainder of this book.
In order to analyze a (design of a) manufacturing system, we need a way to model it and to
determine its performance. We can make many different types of models. Often it is possible
to make a rough estimation of the plant performance without requiring advanced modelling,
see for example [HS01]. In the early stages of the design process we often can not do better
than rough estimates, since we do not have enough detailed data. As we collect more data,
it is possible, for simple plants, to do some more accurate calculations with simple queueing
equations, see again [GH85, Tij94]. However, these queueing equations have a limited range
of application and validity. If plants become more complex, we require advanced queuing
theory (as can be found in for example [GH85, BS93, Tij94, AR01]) and stochastic process
theory. However, these models require extensive mathematical knowledge, substantial effort,
and even these models have a limited range of applicability. Here, we can no longer use
analytical models1 and we enter the field of (discrete-event) simulation. Figure 1.6 shows
the stage in the design process, the amount of data needed, and the range of applicability
related to the analysis methods mentioned.
Many commercial simulation packages are available that use computer power to calculate
and predict the performance of the plant. These packages offer ready-to-use parametric
building blocks of plant components such as machines and buffers. Examples are Flexsim
[Fle02], Witness [Lan02], Automod [Bro03], Simscript II.5 [Cac02], and Extend 5.0 [Ima02].
However, when plants become more complex and components become more unconventional
(due to for example batching policies, control strategies, specific routings, uncommonly used
distributions), substantial low-level programming (in for example C or C++) is required
and modelling and simulation become time consuming. Then, also these packages reach the
boundaries of their applicability.
In this book, we propose a way to analyze manufacturing systems using the formalism
Chi. The Chi-language can be used for modelling and simulating simple manufacturing
systems, but is especially valuable for modelling and simulating complex manufacturing
systems. It offers a high degree of freedom, without requiring extensive programming effort.
1 By analytical models we mean mathematical models that can (sometimes only after a significant amount
of effort) be explicitly solved and yield an exact solution.

1.2. Analysis of Manufacturing Systems (AMS)

design stage
early

Rough estimate
Simple Queueing Equations
Queueing Equations with
extensions for:
- batching
- series and setup-time
- failure
Advanced queuing theory and
stochastic process theory
Discrete-event simulation

final
amount of
data needed

range of
applicability

Figure 1.6: Different analysis methods

Chi is a high-level formalism, tailored to describe concurrent behaviour (which typically

is the case in plants). The Chi-language is based on computer science and mathematics
[BW90, Fok00, Bru97]. It has been used in numerous cases to analyze the performance
and behaviour of complex manufacturing systems. Examples are semi-conductor factories,
chemical plants, automobile assembly lines, and factories for lithographic equipment as well
as lithographic equipment itself, wafer furnaces, or a chemical clinical analyzer.

On reading this book

Readers familiar with the Chi-language can simply start at Chapter 2 and go from there.
Readers unfamiliar with the Chi-language can familiarize themselves with it, by working
through the Chi 3 tutorial [AHR12]. In the beginning of each chapter in this book, we
indicate what elements of the Chi-language are required to understand the models that are
presented.
Each chapter contains a number of exercises. Answers to some of these exercises are included
in the back of this book.

Outline
The structure of this book is as follows. In Chapter 2 we discuss easy-to-use analytical
ways to analyze a deterministic manufacturing system. In Chapter 3 we discuss analytical
approaches to analyze stochastic manufacturing systems, for example with stochastic process
times.
In Chapter 4 we start with modelling a single machine flowline. We present a basic model
of a buffer and a machine. Then, in Chapter 5, we present a way to measure performance of
a manufacturing system by simulation, amongst others, how to measure mean throughput,
flowtime and utilisation. In Chapter 6, we continue with analyzing flowlines of multiple
machines with and without buffering. Subsequently, in Chapters 7 and Chapter 8, we discuss the modelling and analysis of networks of machines, for example, parallel machines,
jobshops, and re-entrant flowlines. Subsequently, in Chapter 9, we present ways to extend
and improve machine models. We discuss aspects like more advanced distributions to represent the process time, adding machine failure and repair, and modelling batching and

Chapter 1. Introduction

set-up times. In the final chapters we present a number of industrial cases, in which the
Chi-language was used to analyze manufacturing systems.

Chapter 2

Analytical models of
deterministic manufacturing
systems
In Chapter 1 we saw that there are different ways to analyze a manufacturing system. In
this chapter and the following chapter, we provide a brief view on analytical models that
can be used to analyze manufacturing systems. We distinguish two classes of analytical
models: deterministic and stochastic models.
A variable is deterministic if we know in advance what exact value it has and will have at
every instant of time, that is, the value of the variable is predetermined. A constant deterministic variable is simply a constant, e.g. 1.54. A dynamic deterministic variable changes,
but the changes are known in advance, for example a fixed sequence: {1,5,1,5,1,5,1,5}. A
variable is stochastic if we do not know the exact value of the variable in advance. We only
know its value obeys some probability distribution. An example of stochastic variable is a
sample of a binomial distribution: {1,0,1,0,1,1,1,0,0,1}.
In this chapter we discuss the principles of manufacturing system analysis and present
analytical approaches to analyze deterministic manufacturing systems. In Chapter 3 we
present analytical approaches to analyze stochastic manufacturing systems.
In the subsequent chapters we occasionally compare the results of the analytical methods
discussed in this and the subsequent chapter with simulation results to investigate the (range
of) validity of the analytical approaches in an engineering environment.

2.1

Manufacturing system principles

Before we introduce analytical models of manufacturing systems, we first introduce a few

basic quantities and the main principles for manufacturing system analysis.

Chapter 2. Analytical models of deterministic manufacturing systems

Basic quantities
In this section we present the following basic quantities: raw process time t0 , throughput ,
flowtime , wip (work-in-process) w, and utilisation u. Other quantities exist, but are dealt
with when encountered in the remainder of the book. Figure 2.1 illustrates throughput,
flowtime, wip, and utilisation for manufacturing systems at factory and machine level.
j [hours]
d [lots/hour]

factory
level

w [lots]
j [hours]
d [lots/hour]

buffer
machine
level

t0 [hours]
w [lots]

Figure 2.1: Basic quantities for manufacturing systems

Raw process time t0 denotes the net time a lot needs processing on a machine. This
process time excludes additions such as setup time, breakdown, or other sources that
may increase the time a lot is on the machine. The raw process time is typically
measured in hours or minutes.
Throughput denotes the number of lots per time-unit that leaves the manufacturing
system. At machine level, this denotes the number of lots that leave a machine per
time-unit. At factory level it denotes the number of lots that leave the factory per
time-unit. The unit of throughput is typically lots/hour.
Flowtime denotes the time a lot is in the manufacturing system. At factory level
this is the time from release of the lot in the factory until the finished lot leaves the
factory. At machine level this is the time from entering the machine (or the buffer
in front of the machine) until leaving the machine. Flowtime is typically measured in
days, hours, or minutes.
Wip-level w denotes the total number of lots in the manufacturing system, i.e. in the
factory or in the machine. Wip is measured in lots.
Utilisation u denotes the fraction a machine is not idle. A machine is considered idle if
it could start processing a new lot. Thus processing time as well as downtime, setuptime and preventive maintenance time all contribute to the utilisation. Utilisation has
no dimension. Utilisation can never exceed 1.0.

2.1. Manufacturing system principles

In literature different formulas are presented to compute the utilisation in specific cases.
We present only one formula:
u=

Tnonidle
Ttotal

(2.1)

Herein Tnonidle denotes the time the machine is not idle during a total timeframe Ttotal . In
the following examples we calculate the utilisation for a number of cases.
Example 2.1 Utilisation (1)
A machine has raw process time t0 of 0.15 hours and processes 5 lots/hour. Calculate the
utilisation.
Processing 5 lots takes 5 0.15 = 0.75 hours. Every hour the machine is idle for 0.25 hours,
hence utilisation u = 0.75.

Example 2.2 Utilisation (2)
A machine has raw process time t0 = 0.15 hours. Lots arrive at a fixed rate of 5.0 lots/hour.
After every fourth lot, the machine needs recalibrating. Recalibration takes 0.10 hour.
Calculate the utilisation.
To process four lots, the machine is occupied for 4 0.15 + 0.10 = 0.70 hours. These
lots are processed over a total time of 4/5.0 = 0.80 hours. Hence the mean utilisation is
0.7/0.8 = 0.875.

Example 2.3 Utilisation for batch processing
A machine processes lots in fixed batches1 . Processing a batch takes 3 hours. After every
batch the machine needs to be cleaned and prepared for the next batch, which takes 1 hour.
Lots arrive at a rate of 2 lots/hour. Calculate the utilisation for a batch size of 10 and 12
lots respectively.
2 lots/hour

batch
machine

Figure 2.2: Batch machine processing in batches of 10 lots

In case of a batch size of 10 lots, the machine processes a batch every 5 hours. For every
batch the machine is occupied for 3 + 1 = 4 hours. The machine is idle for 1 hour every
5 hours, the utilization u = 4/5 = 0.80. In case of a batch size of 12 lots, the machine
processes a batch every 6 hours. The utilisation is now u = 4/6 = 0.67.


First principle: Conservation of mass

The first principle for manufacturing systems is conservation of mass. Similar as for fluid
flow systems, conservation of mass applies to manufacturing systems. Figure 2.3(a) shows
a workstation (buffer and machine). Herein 1 denotes the arrival rate in lots/hour, 1
1A

batch is a set of lots processed together.

10

Chapter 2. Analytical models of deterministic manufacturing systems

denotes the throughput in lots/hour and w1 denotes the number of lots in the workstation. In
Figure 2.3(b), the total throughput 1 of workstation 1 splits into two flows, with throughput
rates 1,1 and 1,2 respectively. In Figure 2.3(c), 2,1 and 2,2 together form the total arrival
rate of workstation 2.
w1
l1

w1
d1

l1

d1,1

l2,1

d2

d1,2
(b)

(a)

w2

l2,2
(c)

Figure 2.3: Three workstations

Conservation of mass for Figure 2.3(a) yields:

dw1
= 1 1
dt

(2.2)

This differential form of mass conservation states that the change in the number of lots
dw/dt in workstation 1 is equal to the number of lots entering workstation 1 minus the
the number of lots leaving workstation 1 per unit of time. On many occasions we consider
systems in steady state. We say a manufacturing system is in steady state when the mean
number of lots in the system is constant, that is, there are no lots piling up in the system,
or dw/dt = 0. In this case, the mean number of lots entering the manufacturing system
equals the mean number of lots leaving the system. Equation 2.2 becomes:
1 1 = 0

(2.3)

Herein 1 and 1 represent the mean arrival rate and throughput in steady state. For the
workstations in Figure 2.3(b) and (c), mass conservation in steady state yields:
1 1,1 1,2 = 0

2,1 + 2,2 2 = 0

(2.4)

We denote the total throughput of workstation 1 as 1 = 1,1 + 1,2 . Moreover, we have

considered workstations here, but we can just as well consider factories or areas.
Example 2.4 Manufacturing areas
Figure 2.4 shows a factory consisting of 6 areas and the raw processing time of every area.
Area 1 receives 3 lots/hour. After being processed in Area 1, 40% of all lots are processed
in Area 2 and 3, and 60% of all lots are processed in Area 4 and 5. Finally all lots are
processed in Area 6. Consider the factory in steady state and calculate the throughput of
Area 2 and 4 and the utilisation for all 6 areas.
The throughput rates for Areas 2 and 4 follow directly from the given probability data:
2 = 0.4 1 = 0.4 3 = 1.2 lots/hour
4 = 0.6 1 = 0.6 3 = 1.8 lots/hour
Area 1 processes 3 lots/hour. Processing 3 lots per hour in Area 1 takes 3 0.2 = 0.6 hours.
Every hour, the area is occupied for 0.6 hours, hence utilisation u1 = 0.6.

2.1. Manufacturing system principles

11

Area
2

0.4
l1 = 3 lots/hour

d6

6
4

0.6

1
2
3
4
5
6

t0 [hrs]
0.2
0.6
0.4
0.3
0.4
0.3

Figure 2.4: Material flow in six manufacturing areas

Area 2 processes 1.2 lots/hour. Processing 1.2 lots in Area 2 takes 1.2 0.6 = 0.72 hours,
hence utilisation u2 = 0.72.
In a similar way, we calculate the utilisation for the remaining areas. u3 = 0.48, u4 = 0.54,
u5 = 0.72, and u6 = 0.9. As all utilisations are smaller than 1, lots do not pile up in this
system.

Example 2.5 Rework
Consider the manufacturing system in Figure 2.5(a). The manufacturing system consists of
two infinite buffers (B1 and B2 ) and two machines (M1 and M2 ). 50% of the lots processed
by machine 1 need to be reprocessed by machine 1. 25% of the lots processed by machine
2, have to be reprocessed by machine 1 and 2 again. Lots arrive at machine 1 with a fixed
arrival rate of 3 lots/hour. Calculate (a) the number of lots that machine 1 processes per
hour and (b) the number of lots that leave the system via machine 2. Assume that the
system is in steady state.
dM2B1

25%

3 lots/hr

B1

M1

50%

B2

M2

75%

lB1 = 3 lots/hr

B1

dB1

M1

dM1B2

B2

dB2

M2

dM2E

dM1B1

50%

(a)

(b)

Figure 2.5: Manufacturing line with rework

The total throughput of machine 1 splits into two flows. We introduce throughput M1 B1
and M1 B2 , denoting the number of lots/hour that are sent from machine M1 back to the
first buffer B1 and from machine M1 forward to the second buffer B2 respectively. We
introduce additional throughput rates, see Figure 2.5(b). Throughput M2 E denotes the
number of lots per hour that exit the system after being processed by machine M2 . From
mass conservation and the probability data, we obtain the following equations:
B1
M1
B2
M2

B1 + M 1 B1 + M 2 B1 B1 = 0
M1 B2 = M1 B1 = 0.5B1
B2 = M 1 B2
M2 E = 0.75B2
M2 B1 = 0.25B2

Filling in (2), (3), and (4) in (1) leads to:

B1 + 0.5B1 + 0.25B2 B1 = 0
B1 + 0.5B1 + 0.25M1 B2 B1 = 0

(1)
(2)
(3)
(4a)
(4b)

12

Chapter 2. Analytical models of deterministic manufacturing systems

B1 + 0.5B1 + 0.125B1 B1 = 0
B1 = 83 B1 = 8 lots/hour
Thus 8 lots/hour enter machine M1 . In steady state, M1 processes 8 lots/hour. Of these 8
lots/hour, 4 lots/hour proceed to buffer B2 and 4 lots/hour are lead back to buffer B1 .
The number of finished lots that leave the system after being processed on machine M2 is:
M2 E

= 0.75B2
= 0.75M1 B2
= 83 B1
= 3 lots/hour

The throughput of this line is 3 lots/hour. Note that the number of lots entering the line is
equal to the number of lots leaving the line. Due to rework, machine M1 has to process 8
lots/hour.


Second principle: Littles law

The second principle for manufacturing systems is Littles law. Littles law states that the
mean wip-level (number of lots in a manufacturing system) w is equal to the product of the
mean throughput and the mean flowtime [Lit61], provided the system is in steady state.
w =

(2.5)

This can be intuitively interpreted by looking at Figure 2.6, which treats a manufacturing
system as a pipeline where lots pass through. The total amount of fluid in the pipeline in
[m3 ] is equal to the flowrate in [m3 /s] multiplied by the time it takes an imaginary fluid
element to flow through the pipe in [s].

j
Figure 2.6: Manufacturing system as pipeline

Analogously, the mean number of lots in a manufacturing system (wip-level w in [lots]) is

equal to the mean number of lots leaving the system per unit of time (throughput in
[lots/timeunit]) multiplied by the mean time a lot remains in the system (flowtime in
[timeunits]). Littles law only applies in steady state.
Example 2.6 Automotive factory
We study the automotive factory in Figure 2.7. By simple measurements we determine that

2.2. Analysis of deterministic manufacturing systems

13

25 days

200 cars/day

Figure 2.7: Automotive factory

200 cars per day leave the factory. On average a car (or its parts) stay for 25 days in the
factory. Calculate the mean number of cars in the factory in steady state.
By Littles law we estimate that the factory contains w = = 200 25 = 5000 cars (or
parts for 5000 cars).

Example 2.7 Areas
A factory has four areas. 50% of the lots are processed in Area 1, 2, and 3. The remaining
50% is processed in Area 1 and 4. Through the year, we count the number of (semi-finished)
lots in each area. The mean number of lots in each area are shown in Figure 2.8. Moreover,
new lots arrive at Area 1 with a an arrival rate = 10 lots/hour. Calculate the mean
flowtime per area.

10 lots/hour

5 lots/hour
2

5 lots/hour

Area
1
2
3
4

w
100
120
100
240

Figure 2.8: Factory with four areas

With Littles law we calculate the mean flowtime per area.

1 = w1 /1 = 100/10 = 10 hours
2 = w2 /2 = 120/5 = 24 hours
3 = w3 /3 = 100/5 = 20 hours
4 = w4 /4 = 240/5 = 48 hours


2.2

Analysis of deterministic manufacturing systems

In this section we explore the possibility to estimate the performance of a factory (or area),
without detailed data and/or extensive calculations. We analyze problems where:
1. the machines have deterministic process times,
2. lots arrive according to a deterministic arrival pattern,
3. the buffers have infinite capacity, and
4. the manufacturing system is in steady state.

14

Chapter 2. Analytical models of deterministic manufacturing systems

Calculating maximum throughput

We discuss an example, in which we determine the maximum throughput for the manufacturing line from Example 2.5. The arrival rate is to be chosen. Machines M1 and M2
process lots with a raw process time of 0.25 and 0.40 hours respectively. Buffers B1 and B2
have infinite capacity. As in Example 2.5, 50% of all lots processed by machine M1 need
rework on machine M1 , and 25% of all lots processed by machine M2 need rework on M1
and M2 .
t0 = 0.25 hours

t0 = 0.40 hours

25%

B1

M1

50%

B2

M2

50%

Figure 2.9: Manufacturing line with rework

We express the total throughput of machine 1 and 2 in terms of the arrival rate . Using
mass conservation and the probability data we determine that machine 1 processes 38 lots
per hour (see Example 2.5). Machine 2 processes 34 lots per hour.
Next, we express the utilisation per machine in terms of the arrival rate. Machine 1 processes
8
8
2
2
3 lots per hour, which takes 3 0.25 = 3 hours. Utilisation uM 1 = 3 . The utilisation
1.6
of machine 2 is uM 2 = 3 .
We conclude that machine 1 is the critical resource, or the bottleneck. The maximum
throughput is obtained when machine 1 is fully utilized, that is, at an arrival rate of =
3/2 = 1.5 lots/hour. The line has a maximum throughput max = 1.5 lots/hour. The
utilisation of machine 2 is then uM 2 = 1.6
3 1.5 = 0.8. Further increasing the release rate
does not increase the line throughput. Instead, the number of lots in buffer B1 will keep
increasing. Then the system is no longer in steady-state.
Basically we followed the following steps to determine the maximum throughput for a manufacturing system:
1. Assume deterministic process times, deterministic arrival rate, infinite buffers and a
steady state system;
2. Express the throughput of each machine in terms of the release rate ;
3. Express the utilisation of each machine in terms of the release rate ;
4. Determine the bottleneck, the maximum allowable release rate, and the maximum
throughput of the line.
Example 2.8 Maximum throughput of a manufacturing line
Consider the manufacturing line in Figure 2.10. The line consists of machine M1 with a
raw process time of 1.8 hours and 2 parallel machines M2 with each a raw process time of
4 hours. 5% of the lots processed on machine M1 need rework on machine M1 . Calculate
the maximum throughput of this line.
Using mass conservation we derive the number of lots that machine M1 has to process:

2.2. Analysis of deterministic manufacturing systems

15

t0 = 4 hours
t0 = 1.8 hours
M2

B1

M1

95%

B2
M2

5%

Figure 2.10: Manufacturing line with rework and parallel machines

+ 0.05 B1 M1 B1 M1
B1 M 1 = M 1

=0
1

= 0.95

Each machine M2 has to process 0.5 lots/hour. The utilisation for machines M1 and M2
is:
u M1
u M2

= 1/0.95 1.8 = 1.89

= 0.5 4 = 2

The two machines M2 together form the bottleneck. The maximum throughput is attained,
if both machines M2 are fully utilized. This is the case for = 0.5 lots/hour. This
manufacturing line has a maximum throughput max = 0.5 lots/hour.

If we plot the throughput and the utilisation u of a deterministic manufacturing system
against the arrival rate , we obtain diagrams very similar to the ones shown in Figure 2.11.
d0[lots/hour]

u [-]
ub [-]

dmax

1
unb [-]
0.5

lmax

l [lots/hour]

l [lots/hour]

Figure 2.11: Throughput and utilisation u against release rate

If release rate is lower than maximum throughput max , the throughput is equal to the
release rate (the system is in steady state). If the release rate is increased beyond max
(=max ), the throughput remains equal to max . The utilisation of the bottleneck ub increases linearly with the release rate up to a maximum of 1. The utilisation of a nonbottleneck station (unb ) also increases linearly with the release rate, but never reaches 1.

16

Chapter 2. Analytical models of deterministic manufacturing systems

Calculating flowtime
Let us consider the manufacturing line in Figure 2.12. We have a two workstation2 line.
Machine M1 has a raw process time of 2 hours, machine M2 has a raw process time of 3
hours.
t0 = 2 hours

B1

t0 = 3 hours

M1

B2

M2

Figure 2.12: Manufacturing line

We assume that the processing times of the machines are deterministic, that the buffers
have infinite capacity, and that the system is in steady state. The line in Figure 2.12 has a
maximum throughput of 1/3 lots/hour (machine M2 is the bottleneck).
The basic approach for determining the flowtime of a lot is virtually following a lot from the
moment it enters the system until it leaves the system again. In Figure 2.13 we display the
state of the system and the lots in it at different time instants. In this way we can follow
different lots as they travel through the line. We release lots at the maximum rate of 1/3
lots/hour.
t

t0 = 2 hrs

t0 = 3 hrs

lot number
6

j = 5 hrs

j = 5 hrs

Figure 2.13: State of the manufacturing line at different time instants

From Figure 2.13 we determine the flowtime at 5 hours. Another way to follow lots as they
travel trough the line, is with a lot-time-diagram. Herein, we plot the state of individual lots
against time. Figure 2.14(a) and (b) show the lot-time-diagram for the line with a release
rate of 1/3 and 1/2 lots/hour respectively.
For a release rate of 1/3 lots/hour we determine from Figure 2.14(a) that the flowtime is 5
hours. For a release rate of 1/2 lots/hour we see in Figure 2.14(b) that the flowtime keeps
increasing. For any release rate under the maximum throughput the flowtime is 5 hours,
for any release rate above the maximum throughput the flowtime grows to infinity.
2 The term workstation is used here to denote a combination of a buffer and a machine or a buffer and a
number of identical machines in parallel.

2.2. Analysis of deterministic manufacturing systems

(a)

0
1
2
3
lot

(b)

lot

10

2
1

5
5
5
5

2
1

l = 1/3 lots/hour
4

j [hrs]

2
1

t [ hrs]

12

j2

2
1

2
0
1
2
3

17

10

12

j [hrs]

14

2
1

2
1

q
1

j3

2
q
1

l = 1/2 lots/hour

5
6
7
8

lot is
1
2

processed on M1
processed on M2
waiting in queue

Figure 2.14: Lot-time-diagram for release rate of (a) 1/3 lots/hour and (b) 1/2 lots/hour

We can use the lot-time-diagram also to determine the mean wip-level. In Figure 2.15 we
derive the wip-level over time from the lot-time-diagram. For instance, for a release rate of
1/3 lots/hour, at t = 7 there are two lots in the system (lot 1 and 2).
2
0
1
2
3
lot

10

t [ hrs]

12

2
1

0
1
2
3

w(t=7)

2
1

2
1

2
1

l = 1/3 lots/hour

lot

w
[lots]
2
1
0

(a)

10

12

w
[lots]
3
2
1
0
t [ hrs]

10

12

w(t=7)
2
q
1

10

12

t [ hrs]

2
1

2
1

q
1

l = 1/2 lots/hour

t [ hrs]

(b)

Figure 2.15: Lot-time-diagram and w-t-diagram for release rate of (a) 1/3 lots/hour and (b)
1/2 lots/hour
For a release rate of 1/3 lots/hour, the behaviour becomes periodic for t > 3 hours with
a period of 3 hours, see Figure 2.15(a). We regard this periodic behaviour as steady state
behaviour3 . The mean wip-level is w = 31 1 + 23 2 = 53 lots. (Try! Is this result in
correspondence with Littles law?) For Figure 2.15(b) the wip-level keeps increasing. For a
release rate higher than the maximum throughput, the mean wip-level grows to infinity.

3 We speak of steady state when the mean number of lots in the system is constant. If the number of
lots in the system evolves periodically, the momentous number of lots in the system changes, but the mean
number of lots in the system is constant.

18

Chapter 2. Analytical models of deterministic manufacturing systems

We now extend the second workstation with a second machine M2 , see Figure 2.16. Workstation 2 is no longer the bottleneck: workstation 1 is now the bottleneck. The line has a
maximum throughput of 1/2 lots/hour.

t0 = 3 hours
M2

t0 = 2 hours

B1

M1

B2
M2

Figure 2.16: Manufacturing line with two parallel machines M2

To determine the flowtime of lots, we again draw lot-time-diagrams. Figures 2.17(a) and
(b) show the lot-time-diagram for a release rate of 1/3 and 1/2 lots/hour respectively.

(a)

0
1
2
3
lot

(b)

lot

2,2
1

10

12

2,1
1

4
2,1
1

6
2,2
1

l = 1/2 lots/hour

2,1
1

10

2,2
1

t [ hrs]

12

2,2

j [hrs]
5
5
5
5

2,2
1

l = 1/3 lots/hour
2

0
1
2
3

4
2,1
1

14

j [hrs]
5
5
5
5
5

lot is
1 processed on M1
2,1 processed on M21
2,2 processed on M22

Figure 2.17: Lot-time-diagram for release rate of (a) 1/3 lots/hour and (b) 1/2 lots/hour

We see that for both release rates the flowtime is 5 hours. Doubling the capacity at workstation 2 increases the maximum throughput but does not influence the flowtime!

We can also use the lot-time-diagrams to determine the utilisation of the individual machines. From the lot-time-diagram we derive the state (idle or not idle) of the individual
machines, see Figure 2.18.

2.2. Analysis of deterministic manufacturing systems

2
0
1
2
3
lot

4
2,1
1

2,2
1

10

12

2,1
1

t [ hrs]

2
0
1
2
3

2,2
1

l = 1/3 lots/hour

lot
u [-]
0.67
0.50
0.50

M1
M21
M22
2

10

12

19

4
2,1
1

2,2
1

2,1
1

l = 1/2 lots/hour

10

2,2
1

12

t [ hrs]

2,1

u [-]
1.0
0.75
0.75

M1
M21
M22
2

t [ hrs]

(a)

10

12

t [ hrs]

(b)

Figure 2.18: Lot-time-diagram and u-t-diagram for release rate of (a) 1/3 lots/hour and (b)
1/2 lots/hour

(Try! From Figure 2.17 derive a wip-time-diagram and determine the mean wip-level for
both release rates. Verify the results by using Littles law.)
Example 2.9 Flowtime of a manufacturing line
Consider the manufacturing line in Figure 2.19. The manufacturing line consists of three
workstations with 1, 4, and 2 machines respectively. The machines have raw process times
of 0.15, 0.8 and 0.35 hours respectively.
workstation 1

workstation 2

workstation 3

t0 = 0.15 hours
t0 = 0.35 hours
t0 = 0.8 hours

Figure 2.19: Three workstation flowline

(i) Calculate the maximum throughput. (ii) Draw a lot-time-diagram and determine the
flowtime at = max . (iii) From the lot-time-diagram derive a wip-time-diagram and
determine the mean wip-level in steady state. Verify that the results are in correspondence
with Littles law.
(i) We calculate the utilisation for each machine in terms of release rate . At release rate
lots/hour, machine 1 is processing 0.15 hours every hour, u1 = 0.15. Each machine in
workstation 2 has to process 41 lots/hour, thus each machine is processing for 41 0.8 = 0.2
hours every hour, u2 = 0.2. Similarly, u3 = 12 0.35 = 0.175. Workstation 2 is the
bottleneck. The maximum throughput is attained when u2 = 1. For u2 = 0.2 = 1, we get
max = max = 1/0.2 = 5 lots/hour.
(ii) Figure 2.20(a) shows the lot-time-diagram.
From the lot-time-diagram we determine = 0.15 + 0.8 + 0.35 = 1.3 hours.
(iii) Figure 2.20(b) shows the wip-time-diagram derived from the lot-time diagram. From

20

Chapter 2. Analytical models of deterministic manufacturing systems

(a)

0
1
2
3
lot 4

(b)

w
7
[lots] 6
5
4
3
2
1
0

t [ hrs]
0.4 0.8 1.2 1.6 2.0
1
2,1
3,1
l = 5 lots/hour
1
2,2
3,2
1
2,3
3,1
1
2,4
3,2
1
2,1
3,1

0.4

0.8

1.2

1.6

2.0

j [hrs]
1.3
1.3
1.3
1.3
1.3

lot is processed on
1 workstation 1
2,1 first machine of workstation 2
3,2 second machine of workstation 3

t [ hrs]

Figure 2.20: Lot-time-diagram and w-t-diagram for release rate of 5 lots/hour

the wip-time-diagram we see that the wip changes periodically for t > 1.2 hours. For t > 1.2
hours, the system is in steady state. The mean wip level in steady state is w = 12 6+ 12 7 = 6.5
lots. Using Littles law we calculate the mean wip-level at w = = 5 1.3 = 6.5 lots.
The results correspond.

For simple deterministic manufacturing lines, like the ones in Figure 2.12, Figure 2.16, and
Example 2.9 the flowtime is simply determined. For release rates under the maximum
throughput the flowtime is equal to the sum of the process times the lot encounters, and
for release rates higher than the maximum throughput, the flowtime grows to infinity, see
Figure 2.21.
j [hours]
jmin

lmax

l [lots/hour]

Figure 2.21: Flowtime against release rate

In these simple cases, following lots, or constructing a lot-time-diagram is not necessary to
determine the flowtime. However, for manufacturing systems with a little more complexity,
a lot-time-diagram may be a very useful approach to determine the flowtime. In the sequel,
we regard two more complex examples in which we determine the flowtime of a lot by
constructing a lot-time-diagram.
Example 2.10 Flowtime for a batch machine
Consider the batch machine in Figure 2.22 which processes batches with a fixed batchsize
of k = 6 lots. Lots arrive with an arrival rate of lots/hour. Processing a batch takes 3

2.2. Analysis of deterministic manufacturing systems

21

hours. (i) Determine maximum throughput max of the batch machine. (ii) Determine the
(mean) flowtime of a lot for two release rates under max . Is the flowtime independent of
the release rate? (iii) Find an expression for the mean flowtime of a lot for a deterministic
batch machine with fixed batch size k, batch processing time tbatch and release rate .
k=6

tbatch = 3 hrs

l lots/hour

batch
machine

Figure 2.22: Batch machine

(i) The batch machine has to process a batch every 6 hours. Processing a batch takes 3
3
= 12 . The utilisation becomes 1 for = max = 2
hours. The utilisation is thus 6/
lots/hour.
(ii) To investigate the flowtime we again follow individual lots as they enter and leave the
system. Figures 2.23(a) shows the lot-time-diagram for a release rate of = 2 lots/hour.
1
0
2
4
6
8
10

lot

lot

t [ hrs]

processing
batch 1
processing
batch 2

l = 2 lots/hour
1

0
2
4
6
8
10

j [hrs]
5.5
5
4.5
4
3.5
3
5.5
5
4.5
4
3.5
3
t [ hrs]

processing
batch 1
processing
batch 2

l = 4/3 lots/hour

??
[hrs]
6.75
6
5.25
4.5
3.75
3
6.75
6
5.25
4.5
3.75
3

Figure 2.23: Lot-time-diagram for (a) = 2 lots/hour and (b) = 4/3 lots/hour
Note that the flowtime is not the same for all lots. The first lot in a batch has to wait
for 5 subsequent lots before it is processed, while the last lot does not have to wait at
all. The mean flowtime is (5.5+5+4.5+4+3.5+3)/6 = 4.25 hours. Figure 2.23(b) shows
the lot-time-diagram for a release rate of = 4/3 lots/hour. The mean flowtime is now
(6.75+6+5.25+4.5+3.75+3)/6 = 4.88 hours. Apparently, for a batch machine the flowtime
is no longer independent of the release rate!
(iii) We now consider a batch machine with batch processing time tbatch and batch size k.
Lots are released at rate lots/hour. The flowtime of the first and the last lot in a batch
of k lots is respectively:
first = wait for batch + tbatch =
last = tbatch

k1
+ tbatch

22

Chapter 2. Analytical models of deterministic manufacturing systems

As long as < max , mean flowtime is:

=

k1
(k 1)/ + tbatch + tbatch
=
+ tbatch
2
2

For release rates > max , the mean flowtime grows to infinity.

(2.6)


2.2. Analysis of deterministic manufacturing systems

23

Example 2.11 Machine with alternating process times

Consider a two-workstation flowline. Each workstation consists of a single machine and
an infinite buffer. Machine 1 has a constant process time of 3.0 hours. Machine 2 has
an alternating process time of 1.0 and 5.0 hours, that is, the process time follows the
deterministic sequence 1.0, 5.0, 1.0, 5.0, 1.0, ....
t0 = 3.0 hours

B1

t0 = 1.0 or 5.0 hours

M1

B2

M2

Figure 2.24: Two-workstation flowline with alternating process times for second machine
(i) What is the maximum throughput of this line? (ii) Determine the (mean) flowtime of a
lot. (iii) Construct a wip-time-diagram and determine the mean wip-level in steady state.
Verify your result with Littles law.
(i) If lots are released at a rate lots/hour, machine 1 and 2 have to process 2 lots per 2
hours. Processing two lots requires 2 3.0 = 6.0 hours on machine 1, and 1.0 + 5.0 = 6.0
6.0
hours on machine 2. The utilisation for both machines is u1 = u2 = 2/
= 3. Both
machines are bottlenecks. The maximum throughput is max = max = 1/3 lots/hour.
(ii) We construct a lot-time-diagram for a release rate of 1/3 lots/hour, see Figure 2.25(a).
2

(a)

0
1
2
3
4
lot

10

12

14

t [ hrs]
16

2
1

2
1

2
1

2
1

2
1

l = 1/3 lots/hour

w
[lots]

j [hrs]
4
8
6
8
6

lot is
1
2
q

processed on M1
processed on M2
waiting in queue

period

(b) 2
1
0

10

12

14

t [ hrs]

Figure 2.25: Lot-time-diagram and wip-time-diagram for = 1/3 lots/hour

From the lot-time-diagram we determine the mean flowtime at (8+6)/2 = 7 hours. Note
that we determine the mean flowtime only after the system has reached steady state, that
is, behaves periodically. Note that the flowtime again depends on the release rate.
(iii) Figure 2.25(b) shows the wip-time-diagram. For t > 6 hours the wip-level evolves
7
periodically with a period of 6 hours. The mean wip level is w = 26 3 + 46 2 = 14
6 = 3 lots.
1
7
With Littles law: w = = 3 7 = 3 lots.

Please note, that determining the flowtime in the way we discussed here only works for
deterministic systems!
For deterministic system systems the process times (or other properties) need not be constant, as long as the changes are deterministic! Recall that the second machine in Example 2.11 did not have a constant process times, but the changes were predetermined and

24

Chapter 2. Analytical models of deterministic manufacturing systems

therefore the process time was deterministic. If the process time was stochastic, for example
the process time is either 1.0 or 5.0, both with 50% probability, we can only construct a lottime-diagram that represents one of the many possible outcomes. For example it could happen that the process time of machine 2 follows the sequence {1.0,1.0,5.0,5.0,5.0,1.0,1.0,5.0}.
For this sequence the flowtime is significantly higher. (Try! Determine the flowtime for
this sequence.) But in case of stochastic process times, any other sequence is possible. We
can no longer use the approach presented in this section. The same holds for stochastic
arrival times, stochastic chance on rework, stochastic break-down, or any other source of
stochasticity.
In Chapter 3 we consider mathematical models to calculate the flowtime for stochastic
process and inter-arrival times.

Keyword overview
Modelling manufacturing systems

2.3

Analyzing manufacturing systems

mass conservation
utilisation u, bottleneck,
maximum throughput
Littles law
throughput and flowtime calculation
for deterministic manufacturing systems

Exercises

1. Consider the manufacturing system with rework and bypassing in Figure 2.26. The
manufacturing system consists of three buffers and four machines.

0.9

B1

0.1

M1

0.2

B2

M2

0.8
M3

B3

M4

machine

t0 [hrs]

M1
M2
M3
M4

2.0
6.0
1.8
1.6

Figure 2.26: Manufacturing system with rework and bypassing

Lots are released at a rate of lots/hour. The numbers near the arrows indicate the
fraction of the lots that follow that route. For instance, of the lots leaving buffer B1
90% goes to machine M1 and 10% goes to buffer B3 . The process time of each machine
is listed in the table in Figure 2.26.
(a) Express the throughput of machine M1 , M3 , and M4 in terms of .
(b) Express the throughput of machine M2 in terms of .
(c) Express the utilisation of each machine in terms of .
(d) What machine is the bottleneck? Determine the maximum throughput of this
system.

2.3. Exercises

25

2. Consider a three-workstation flowline. The workstations each have an infinite buffer

and contain 1, 3, and 2 machines respectively. The process time of the machines in
workstation 1,2 and 3 is 0.9, 3.0, and 1.9 hours respectively.
(a) Calculate the maximum throughput of the line.
(b) Use a lot-time-diagram to determine the flowtime of a lot for release rates under
the maximum throughput.
(c) Determine the mean wip-level in the line for the maximum throughput.
3. We have a three-workstation flowline. Each workstation consists of an infinite buffer
and a number of machines. The number of machines has still to be determined. The
machines in workstation 1,2, and 3 have a process time of 0.10, 0.15, and 0.06 hours
respectively. We want to establish a throughput of 60 lots/hour.
(a) Determine the flowtime of a lot for release rates under the maximum throughput.
(b) Determine the number of machines in workstation 1,2, and 3 required to attain
the desired throughput of 60 lots/hour.
(c) Which workstation (what workstations) is (are) the bottleneck?
4. Consider a two-workstation flowline with two machines. The two machines have a
process time of 2.0 and 3.0 hours respectively. The machines may be placed in any
order. The two alternatives are shown in Figure 2.27.
t0 = 2.0 hr

(a)

B1

(b)

B1

t0 = 3.0 hr

M1

B2

M2

t0 = 3.0 hr

t0 = 2.0 hr

M2

B2

M1

Figure 2.27: Two flowline configurations

(a) What is max for the line in Figure 2.27(a)?

(b) What happens when we set > max ?
(c) What is max for the line in Figure 2.27(b)?
(d) What happens when we now set > max ?
5. Consider the re-entrant flowline in Figure 2.28. Lots are released in the line by generator G at a rate of lots/hour. Each lot passes through the flowline twice and has
the same fixed route, namely M1 , M2 , M1 , M2 . If a newly released lot and a re-entrant
lot arrive at a machine at the same time, the re-entrant lot is processed first.
t0 = 2.0 hr

l
G

B1

M1

t0 = 2.0 hr
B2

M2

Figure 2.28: Re-entrant flowline

26

Chapter 2. Analytical models of deterministic manufacturing systems

(a) Express the utilisation of machine 1 in terms of release rate and determine the
maximum throughput.
(b) Construct a lot-time-diagram for =
flowtime of a lot.

1
5

lots/hour and determine the (mean)

(c) From the lot-time-diagram, derive a wip-time-diagram, and determine the mean
wip-level in steady state.
(d) Verify your result using Littles law.
6. We have a two workstation flowline. Workstation 1 consists of an infinite buffer and a
single-lot4 machine. Workstation 2 consists of an infinite buffer and a batch machine.
The single-lot machine has a process time of 2.0 hours. The batch machine processes
in fixed batches of 5 lots. The process time of a batch is 8.0 hours.
l

tbatch = 8.0 hrs

t0 = 2.0 hrs

k = 5 lots

Figure 2.29: Single-lot and batch machine in series

(a) Express the utilisation of the single-lot machine and the batch machine in terms
of release rate and determine the maximum throughput.
(b) Construct a lot-time-diagram for =
time of a lot.

1
2

lots/hour. Determine the (mean) flow-

(c) Calculate the mean flowtime of a lot without using the lot-time-diagram. You
can use the equation found in Example 2.10(iii).
(d) Determine the mean wip-level.
(e) Can we decrease the mean wip-level by increasing the release rate?
We replace the batch machine by a cheaper one. This machine processes batches of
size 4. Processing a batch takes 10 hours.
l

t0 = 2.0 hrs

tbatch = 10.0 hrs

k = 4 lots

Figure 2.30: Single-lot and batch machine in series (2)

(a) Determine the maximum throughput of the line and determine the utilisation of
the single-lot machine when the line attains the maximum throughput.
(b) Construct a lot-time-diagram for this line, for the maximum release rate. Determine the (mean) flowtime of a lot.
(c) Calculate the mean flowtime of a lot without using the lot-time-diagram. You
can use the equation found in Example 2.10(iii).
4A

single-lot machine processes one lot at a time.

2.4. Additional exercises

27

7. We have a two-workstation flowline. Each workstation consists of an infinite buffer

and a single machine. The flowline is used to process two types of products: type A
and type B. Table 2.1 shows the process time on each machine per type.
machine 1 machine 2
type A
3.0
2.5
type B
2.0
2.5
Table 2.1: Processing times of different product types in [hours]
New products are released in the flowline with a rate of prods/hour. The product
types are released in a fixed sequence: {A,A,B,B,B,A,A,B,B,B,A,A,B,B,B,...}.
(a) Express the utilisation of machine 1 and 2 in terms of release rate . Determine
the maximum throughput of this line.
(b) What is the maximum throughput specified per product type, that is, how many
products of each type are processed per hour on average?
(c) Construct a lot-time-diagram for a release rate of = 0.4 prods/hour.
(d) Determine the mean flowtime per product in steady state.
(e) Calculate the average wip level in steady state.
(f) Calculate the average wip level in steady state, specified per product type.

2.4

Additional exercises

1. The manufacturing system in Figure 2.31 consists of 6 workstations. 20% of the lots
processed by workstation W4 need rework. The lot is first stripped in workstation W5
and then reprocessed by workstation W2 and W3 .
l

W1

W2

W5

W3

0.2

W4

0.8

W6

Figure 2.31: Manufacturing line

The numbers at the arrows indicate the fraction of the lots that follow that route.
(a) Calculate the total throughput of workstation 2,3, and 5 in terms of release rate
.
(b) Calculate the throughput W4 W5 and W4 W6 for workstation 4.
(c) Verify that conservation of mass holds for system W2 W3 W4 W5 .

28

Chapter 2. Analytical models of deterministic manufacturing systems

2. Figure 2.32 shows a flowline with bypassing loops going from workstation 1 to workstation 2, 3, and 4. The numbers at the arrows indicate the fraction of the lots that
follow that route.
l

W1

0.7

W2

W3

W4

0.1
0.1
0.1

Figure 2.32: Manufacturing line with bypassing

(a) Use mass conservation to determine the throughput of workstations 1,2,3, and 4.
(b) Verify that conservation of mass holds for system W1 W2 W3 W4 .
3. Figure 2.33 shows a flowline of workstations with rework loops. 20% of the lots processed by workstation 2 need rework on workstation 1 and 2. 20% of the processed by
workstation 3 need rework on workstation 2 and 3. Finally, 30% of the lots processed
by workstation 4 need rework on 3 and 4.
0.2
l

W1

0.2
W2

0.8

0.3
W3

0.8

W4

d
0.7

Figure 2.33: Manufacturing line with rework

(a) Intuitively, what workstation will have to process the most lots?
(b) Write down the mass conservation equations for workstation 1 through 4.
Mass conservation on system W1 W2 W3 W4 yields that = . For the total throughput
of workstation 4 we have W4 = + W4 W3 .
(a) Calculate W4 in terms of .
(b) Calculate the throughput of each workstation in terms of .
(c) Assuming that the raw process time t0 for each workstation is 1.0 hours, what
workstation is the bottleneck? (Does this correspond to your initial guess in a)?)
(d) What is the maximum throughput of this line?
4. Figure 2.34 shows a so-called job shop. In a job shop lots can each have very different
routings. The from-to-matrix is used to indicate the fraction of the lots that go from
a specific workstation to another workstation. For example, 10% of lots that finish
processing on workstation 1 go to workstation 2, 80% goes to workstation 3, and 10%
goes to workstation 4. All lots enter the job shop via workstation 1 and leave the job
shop via workstation 4.

2.4. Additional exercises

29

l
in

W1

in

W2
from

d
W4

out

to
3

0.1

0.8

0.1

0.7

0.1

out

1.0

W3

0.2

0.1

0.2

0.0

0.2

0.7
0.3

0.5

Figure 2.34: (a)Job shop, (b) From-to-matrix

(a) Intuitively, what workstation processes the most lots in this job shop?
(b) Write down the mass conservation equations for each workstation.
(c) Show that

1
0.1

0.8
0.1

we can write these equations as the following matrix equation.

0.2 0.1 0
W1

W2
0
1 0.2 0.2

=
0
0.7 1 0.3 W3
0.1 0.7 1
W4
0

(d) Solve the matrix equation (for example using Matlab).

(e) How can you easily verify that the throughput W4 is indeed 2? Do the results
correspond to your intuition built in part a)?
(f) The process time t0 for workstation 1,2,3, and 4 is 1.0, 1.3, 0.7, and 0.8 hours
respectively. Which workstation is the bottleneck?

30

Chapter 2. Analytical models of deterministic manufacturing systems

Chapter 3

Analytical models of stochastic

manufacturing systems
In the previous chapter we presented analytical approaches to determine the (maximum)
throughput, the flowtime and the wip-level for deterministic manufacturing systems, that is,
systems with deterministic process times, deterministic arrival rate, deterministic routings,
and no other sources of stochasticity. At the end of the previous chapter, we argued that
these analytical approaches are no longer useful when stochasticity is involved. In this
chapter we present an analytical approach to analyze manufacturing systems, that feature
stochastic process times, stochastic inter-arrival times, or other stochastic phenomena. Not
being a book about queueing theory or stochastic process theory we do not go into details
and derivations. Instead we refer to other books for derivations. We restrict ourselves to the
basic queueing relations for a single workstation (Section 3.2), for a flowline (Section 3.3)
and some extensions on these queueing relations for batching (Section 3.5).

3.1

Classification of manufacturing queueing models

The basic manufacturing queueing model is shown in Figure 3.1.

m = 1/t0
N

...

l = 1/ta

buffer

d = 1/td

machines

Figure 3.1: Basic queueing model

The basic model is a workstation that consists of a buffer and a number of identical parallel
machines. The workstation is characterized by the following quantities.

31

32

Chapter 3. Analytical models of stochastic manufacturing systems

Inter-arrival time ta (with arrival rate = 1/ta ) denotes the time between two subsequent lots entering the system. It is characterized by a distribution, with mean ta
and coefficient of variation1 ca .
Process time t0 (with service rate = 1/t0 ) is the time it takes the machine to
process a lot. Stochastic process time is characterized by a distribution with mean t0
and coefficient of variation c0 .
Buffer size N (or maximum queue length) is the maximum number of lots that can
be stored in the buffer.
The number m of parallel identical machines in the workstation.
Inter-departure time td (with throughput = 1/td ) denotes the time between two
subsequent lots leaving the system.
Kendall [Ken53] introduced a shorthand notation for the characteristics of a workstation
of the form A/B/m/N . Herein A and B represent the distribution of the inter-arrival
time and the distribution of the process time respectively. We distinguish three classes of
distributions: an M denotes an exponential (or Markovian, or memoryless) distribution,
a D denotes a deterministic distribution, and a G denotes any (or a general) distribution.
The m in the Kendall notation refers to the number of parallel machines in the workstation
and the N denotes the maximum number of lots in the buffer.
For example, an M/M/1/ workstation is a workstation with an infinite buffer where lots
arrive with exponentially distributed inter-arrival times and one machine that has exponential processing times. In case of infinite buffers, the symbol is often omitted from the
Kendall notation. So, a G/D/1 workstation is single-machine workstation with an infinite
buffer, the machine has a deterministic process time and the inter-arrival time of the lots
can be any distribution.

3.2

Queueing relations for single-machine workstation

Consider a workstation that consists of infinite buffer B and machine M , see Figure 3.2.
Lots arrive at the buffer with a stochastic inter-arrival time. The inter-arrival time distribution has mean ta and squared coefficient of variation c2a . The coefficient of variation
relates the degree of variability in a distribution to the mean of the distribution. In queueing theory, the coefficient of variation is more often used than the standard deviation. The
machine has stochastic process times, with mean process time t0 and squared coefficient of
variation c20 . Finished lots leave the machine with a stochastic inter-departure time, with
mean td and squared coefficient of variation c2d .
Two preconditions are required: (1) the system is stable, that is, reaches steady-state, and
(2) no lots are lost in the system. The first precondition implies that the mean interarrival time ta should be greater than the mean process time t0 . If ta < t0 , the machine
has insufficient capacity and an ever-increasing queue in the buffer occurs. The second
1 The coefficient of variation c of a distribution is defined as the quotient of its standard deviation and
its mean : c = /.

3.2. Queueing relations for single-machine workstation

B
ta, ca2

33

t0, c02
M

td

Figure 3.2: Single machine workstation

precondition implies that the mean inter-departure time td is equal to the mean interarrival time ta . The approximation of Kingman [Kin61, Tij94] states that the mean waiting
time B in buffer B can be approximated by:
B =

c2a + c20
u

t0
2
1u

(3.1)

Every ta hours, the machine has to process a lot, which takes t0 hours, thus the utilisation
u = t0 /ta .
The Kingman relation in Equation 3.1 is composed of three factors: a variability factor
(c2a + c20 )/2, a utilisation factor u/(1 u) and a term with the process time t0 .
The variability term indicates that if both inter-arrival and process times are deterministic (constant), the waiting time in the buffer is 0. The higher the variability in
the inter-arrival time or in the process time, the longer the mean waiting time in the
buffer becomes.
The utilisation factor shows that the average waiting time of a lot in the buffer increases in a non-linear fashion for increasing utilisation. As an extremum the system
becomes unstable if u = 1. Utilisations of 80% and higher give rise to long waiting
times in systems with variability.
The final factor is the mean process time. The mean waiting time is proportional with
the mean process time.
Equation 3.1 is exact for an M/G/1 system (as already shown by Pollaczek-Khinchine [Pol30,
Khi32]), that is, a single machine workstation with exponentially distributed inter-arrival
times and any distribution for the process time. For other single machine workstations it is
an approximation. As an example, let us consider an M/M/1 system.
Example 3.1 M/M/1 system
An M/M/1 system consists of an infinite buffer and a single machine. The inter-arrival
times and process times are distributed according to an exponential distribution. For an
exponential distribution, the mean is equal to the standard deviation. The mean interarrival time ta is 4.0 hours, the mean process time t0 is 3.0 hours. Determine the mean
flowtime per lot.
For an exponential distribution holds: c = / = 1, so c2a = 1 and c20 = 1. Applying
Equation 3.1 yields:
B =

c2a + c20
u
1+1
u
u

t0 =

t0 =
t0
2
1u
2
1u
1u

34

Chapter 3. Analytical models of stochastic manufacturing systems

The total flowtime becomes:

= B + t0 = (

u
t0
3
+ 1) t0 =
=
= 12 hours
1u
1u
0.25

As the M/M/1 system is an element of the class of M/G/1 systems, the calculation of the
flowtime is exact.

An interesting parameter is the flowtime factor /t0 , which relates the total flowtime to the
raw process time (minimal flowtime). For Example 3.1 the flowtime factor is 4, indicating
that on average 75% of the flowtime is due to waiting in the buffer.
Example 3.2 Flowtime for a workstation with high and low process times
Consider a single machine workstation with an infinite buffer and a machine. The machine
has a process time of either 1.0 or 5.0. The inter-arrival times are distributed according
to an exponential distribution, with mean inter-arrival time ta = 4.0 hours. Determine the
mean flowtime and the mean number of lots waiting in the buffer.
Since, the inter-arrival times are distributed according to an exponential distribution, c2a = 1.
Next, we calculate the standard deviation s0 on the process time.
n

s20 =

1X
(1.0 3.0)2 + (5.0 3.0)2
= 4.0
(xi x)2 =
n i=1
2

The squared coefficient of variation becomes:

c20 =

s20
4.0
4
=
=
t20
3.02
9

Applying Equation 3.1 yields:

B = (

c2a + c20
u
1 + 4/9
0.75

) t0 = (

) 3.0 = 6.5 hours

2
1u
2
1 0.75

Using Littles law, we determine the mean number of lots in the buffer:
wB = B = 6.5 1/4 = 1.63 lots
This manufacturing system is an M/G/1 system, so the calculated flowtime (and thus wiplevel) are exact.

Figure 3.3 shows the mean total flowtime for an M/G/1 system determined analytically
as a function of the utilisation u, with ca = 1, t0 = 3.0 hours, and c0 = {0, 1, 3}.
No machine can ever be utilized for the full hundred percent, since almost all manufacturing
systems contain some sort of variability, which causes the flowtime (and thus wip-level) to
go to infinity for u = 1. In industry this simple observation is often overlooked. But also for
utilizations smaller than one, we can clearly see that flowtime increases strongly at higher
utilisation levels. The higher the machine is utilized the larger the average waiting time of
lots in the queue is, which results in larger flowtimes. Especially in make-to-order industries
one strives for short flowtimes to keep customers happy.
The only way one can improve the trade-off between utilization and flowtime is to decrease
variability. Figure 3.3 shows that for a low variability level the waiting time stays relatively

3.3. Queueing relations for a flowline

35

100
90
Ca = 1.0

Mean flowtime [hours]

80
70
60
50
40
C0 = 3.0

30
C0 = 1.0

20
10
C0 = 0.0

0
0

0.1

0.2

0.3

0.4

0.5
u [-]

0.6

0.7

0.8

0.9

Figure 3.3: Mean flowtime for a M/G/1 system for different values of c0

small even for high utilization levels. The nonlinear increase of waiting time occurs for
very high utilizations. On the other hand, high variability levels cause the waiting time to
strongly increase for low utilization levels.
In Chapter 5 we investigate the validity of this analytical relations by simulation.

3.3

Queueing relations for a flowline

In practice, we often encounter chains of buffers and machines. As an example, consider a

three workstation flowline, see Figure 3.4.
ta,0
ca,0

B
M0

td,0 = ta,1
cd,0 = ca,1

t0,0, c0,0

B
M1

td,1 = ta,2

cd,1 = ca,2

t0,1, c0,1

M2

td,2
cd,2

t0,2, c0,2

Figure 3.4: Three workstation flowline

We calculate the mean flowtime for workstation 0 by using Equation 3.1. To calculate
the mean flowtime for workstation 1, we need the inter-arrival time distribution for workstation 1. The inter-arrival times of workstation 1 are equal to the inter-departure times
of workstation 0 (mean td,0 , coefficient of variation cd,0 ). For a stable system, we have
ta,0 = td,0 = ta,1 = td,1 = ta,2 = td,2 (conservation of mass). But how do we determine cd,1 ?
We can approximate the coefficient of variation cd by Kuehns linking equation [Kue79]:
c2d = (1 u2 ) c2a + u2 c20

(3.2)

Equation 3.2 indicates that for low utilisations (u 0) the squared coefficient of variation
on the inter-departure times c2d resembles the squared coefficient of variation on the inter-

36

Chapter 3. Analytical models of stochastic manufacturing systems

arrival times c2a (c2d c2a ). For low utilisations, lots seldomly have to wait in the buffer before
being processed. The variability on the departure is inherited from the arrival pattern. For
high utilisations (u 1), c2d resembles the squared coefficient of variation on the process
times c20 (c2d c20 ). For high utilisations, lots usually have to wait in the buffer before they
are processed. The variability of the arrival is levelled out by the large queue in the buffer.
The variability on the departure is dictated by the process time variability.
Example 3.3 Three workstations in line
Consider the three workstation flowline from Figure 3.4. For the inter-arrival time at workstation 0 we have ta,0 = 4.0 hours and c2a,0 = 1. The three workstations are identical with
respect to the process times: t0,i = 3.0 hours and c20,i = 0.5 hours for i = 0, 1, 2. Determine
the mean total flowtime per lot.
Since ta > t0,i for i = 0, 1, 2, we have a stable system and ta,i = td,i = 4.0 hours for
i = 0, 1, 2. Subsequently, the utilisation for each workstation is ui = 3.0/4.0 = 0.75 for
i = 0, 1, 2.
We calculate the mean flowtime for workstation 0, using Equation 3.1.
0 = B + t0,0 =

c2a,0 + c20,0
u
1 + 0.5
0.75

t0,0 + t0,0 =

3.0 + 3.0 = 9.75 hours

2
1u
2
1 0.75

Next, we determine the coefficient of variation on the inter-arrival time ca,1 for workstation
W1 using Equation 3.2.
c2a,1 = c2d,0 = (1 u2 ) c2a,0 + u2 c20,0 = (1 0.752 ) 1 + 0.752 0.5 = 0.719
We calculate the mean flowtime for workstation 1.
1 =

0.719 + 0.5 0.75

3.0 + 3.0 = 8.49 hours
2
1 0.75

In a similar way, we determine that c2a,2 = 0.596 and 2 = 7.93 hours. We then calculate
the mean total flowtime.
tot = 0 + 1 + 2 = 26.2 hours
Note that the minimal flowtime without variability (c2a,0 = c20,i = 0) equals 9.0 hours.
We investigate the validity of Kuehns linking equation in Chapter 6.

3.4. Identical parallel machines

3.4

37

Identical parallel machines

The relations from the previous sections allow us to determine the flowtime for a singlemachine workstation and a flowline of a single-machine workstations. In this section we wish
to determine the flowtime and the coefficient of variation on the inter-departure time for a
workstation with identical parallel machines. Figure 3.5 shows a workstation that consists
of an infinite buffer and m parallel identical machines.
t0, c02

td, cd2

...

ta

Figure 3.5: Workstation with m identical parallel machines

In this workstation, all lots wait in a single queue for the next available machine. For a workstation with m identical parallel machines, the waiting time in the buffer is approximated
by the following equation [HS01].

c2a + c20 u 2(m+1)1

t0
(3.3)
B =
2
m(1 u)
A reasonable approximation for the coefficient of variation on the inter-departure times is
the following linking equation [HS01].
u2
c2d = 1 + (1 u2 )(c2a 1) + (c20 1)
m

(3.4)

Both equations are approximations, which are fairly accurate for high degrees of utilisation
and less accurate for lower degrees of utilisation. Note that for m = 1 Equations 3.3 and
3.4 reduce to Equations 3.1 and 3.2. Be careful to calculate the utilisation in accordance
with the number of machines!
Example 3.4 Parallel machines
Consider the flowline in Figure 3.6.
2,1
B

B
ta , c a 2

2,2

3,1
td, cd2

1
2,3

3,2

2,4
t0 = 3.0 hr
c0 = 0.5 hr

t0 = 14.0 hr
c0 = 1.0 hr

t0 = 6.0 hr
c0 = 0.8 hr

Figure 3.6: Flowline with multiple-machine workstations

38

Chapter 3. Analytical models of stochastic manufacturing systems

The flowline consists of three workstations with 1,4, and 2 identical machine respectively.
The raw process time and the coefficient of variation on the raw process time per machine
is displayed in Figure 3.6. The lots arrive at workstation 1 with a mean inter-arrival time of
4.0 hours and a coefficient of variation of 1.0. Determine the mean flowtime per workstation.
We start with determining the utilisation for each workstation. Workstation 1 processes
a lot every 4.0 hours, which takes 3.0 hours, u1 = 3.0/4.0 = 0.75. Workstation 2 can
process 4 lots simultaneously. Each machine in workstation 2 receives one lot every 16.0
hours, u2 = 14.0/16.0 = 0.875. Similarly, we determine the utilisation for workstation 3,
u3 = 6.0/8.0 = 0.75.
Since the utilisation for all workstation is under 1.0, the system is stable and ta,i = ta.1 = 4.0
hours.
We now calculate the flowtime for workstation 1. Workstation 1 has a single machine, so
we use Equation 3.1.
1

c2a,1 + c20,1
u1

t0,1 + t0,1
2
1 u1
2
1 + 0.5
0.75
=

3.0 + 3.0 = 8.625 hours

2
1 0.75

To calculate the flowtime for workstation 2, we need the coefficient of departure for workstation 1, using Equation 3.2.
c2d,1 = c2a,2 = (1 u21 ) c2a,1 + u21 c20,1 = (1 0.752 ) 1 + 0.752 0.52 = 0.578
We calculate the mean flowtime for workstation 2. Workstation 2 contains 4 parallel machines.

2(m2 +1)1
c2a,2 + c20,2 u2
2 =

t0,2 + t0,2
2
m2 (1 u2 )
101
0.578 + 1.0
0.875
=

14.0 + 14.0 = 30.55 hours

2
4 (1 0.875)
We calculate the coefficient of variation on the inter-departure times.
c2d,2

u2
= c2a,3 = 1 + (1 u22 ) (c2a,2 1) + 2 (c20,2 1)
m2
= 1 + (1 0.8752 ) (0.578 1) + 0 = 0.901

In a similar way, we determine that 3 = 12.09 hours and c2d,3 = 0.8136. The mean total
flowtime is
tot = 1 + 2 + 3 = 51.27 hours
Note that the minimal flowtime without variability (c2a = c20,i = 0) equals 24.0 hours.

3.5. Batch machines

3.5

39

Batch machines

In this chapter we have regarded single-lot machines only. In the previous chapter we
discussed batch machines with deterministic process and inter-arrival times. In this section
we take a look at batch machines with stochastic process and inter-arrival times.
Consider a batch machine that processes lots in batches. The batches have a fixed size
k. The machine starts processing the batch only after the batch is complete. The batch
machine is depicted in Figure 3.7.
batchsize k
ta,l
ca,l2

td,b
cd,b2
t0,b c0,b2

Figure 3.7: Machine processes batches of fixed batch size k

Lots arrive with a mean inter-arrival time ta,l and squared coefficient of variation c2a,l . Processing a batch takes on average t0,b , with a squared coefficient of variation c20,b . Question
now is: what is the mean flowtime?
We take a closer look at the lots in the buffer. We, imaginarily, divide the buffer into two
parts. In the first part of the buffer, lots wait until they have formed a complete batch. In
the second part the batches of lots wait for the machine to become idle. The mean flowtime
for the batch machine is the sum of the wait-to-batch-time Bk , the queueing time of the
batches Bq , and the processing time.
batchsize k

ta,b , ca,b2

ta,l
ca,l2

jBk

td,b
cd,b2

jBq

t0,b c0,b2

Figure 3.8: Lots first form a batch and then wait to be processed

The wait-to-batch-time Bk is not equal for all lots in a batch. The first lot in the batch
has to wait for (k 1) lots to arrive, while the last lot in the batch does not have to wait
at all before the batch is complete. The mean wait-to-batch-time is:
Bk =

(k 1) ta,l + 0
(k 1)
Bk (first) + Bk (last)
=
=
ta,l
2
2
2

Note that this result is identical to the result obtained in Example 2.10. After the batch is
complete, it is possible that the machine is not idle and the batch has to wait. In fact, we
can regard the batch machine and the queue of batches in the same way as a basic single-lot
workstation. From Section 3.2 we know that the waiting time in the queue is:
Bq =

c2a,b + c20,b
u

t0,b
2
1u

40

Chapter 3. Analytical models of stochastic manufacturing systems

Herein ca,b is the coefficient of variation on the inter-arrival time of batch! However, we
only know the variation on the inter-arrival time of a lot. Question is: how do we derive ca,b
from ca,l ? From basic statistics we know that for k independent stochastics the variance of
the sum of these stochastics is equal to the sum of the variances, see for example [MR02].
Var(X1 + X2 + . . . + Xk ) = Var(X1 ) + Var(X2 ) + . . . + Var(Xk )
The inter-arrival time of a batch is the sum of the inter-arrival times of k lots.
ta,b = ta,1 + ta,1 + . . . + ta,k
If the inter-arrival times for the lots are independent samples of the same distribution we
get for the variance:
Var(ta,b )

= Var(ta,l + ta,2 + . . . + ta,k )

= Var(ta,1 ) + Var(ta,2 ) + . . . + Var(ta,k )
= Var(ta,l ) + Var(ta,l ) + . . . + Var(ta,l )
= k Var(ta,l )

For the squared coefficient of variation we get:

c2a,b =

2
2
2
a,b
k a,l
c2a,l
1 a,l
=
=
=
t2a,b
(k ta,l )2
k t2a,l
k

We can now calculate the queueing time of a batch in the buffer. Remember that the
utilisation of a batch machine can be determined as follows. Each ta,b = k ta,l the batch
machine processes a batch, which takes t0,b . Hence, utilisation is u = t0,b /(k ta,l ).
Example 3.5 Single-lot and batch machine in series
We have a flowline of a single-lot machine and a batch machine. Lots arrive at the single lot
machine with a mean inter-arrival time of 4.0 hours and a squared coefficient of variation c2a
of 1. Processing a lot on the single-lot machine takes on average 3.0 hours, with c20 = 0.5.
The batch machine processes lots in fixed batches of 3 lots. Processing a batch takes 8.0
hours on average with a squared coefficient of variation c20,b = 0.1, see Figure 3.9. Calculate
the mean flowtime of a lot.
batchsize k
ta,l = 4.0 hr
2
ca,l
= 1.0
t0 = 3.0 hr
c02 = 0.5

t0,b = 8.0 hr
2
c0,b
= 0.1

Figure 3.9: Single-lot machine and batch machine in series

First we calculate the utilisation for the single-lot machine and the batch machine. These
are uM = 3/4 and ubatch = 8/(3 4) = 2/3. Since both utilisations are under 1 we have a
stable system and we can calculate a mean flowtime. For the queueing time for the single-lot
machine we have:
B1 =

c2a + c20
u
1 + 0.5
0.75

t0 =

3 = 6.75 hours
2
1u
2
1 0.75

3.6. Discussion

41

Next we determine the coefficient of variation of the lots leaving the single-lot machine and
entering the batch machine.
c2d = u2 c20 + (1 u2 )c2a = 0.752 0.5 + (1 0.752 ) 1 = 0.7188
We now calculate the wait-to-batch-time and the queueing time of batches in front of the
batch machine.

Bk =

k1
31
ta,l =
4.0 = 4.0 hours
2
2

For the queueing time we need to calculate the coefficient of variation on the inter-arrival
time of a batch.
c2a,b =

c2a,l
0.7188
=
= 0.2396
k
3

Bq =

c2a,b + c20,b
u
0.2396 + 0.1 2/3

t0,b =

t0,b = 2.717 hours

2
1u
2
1/3

The total mean flowtime of a lot in the line is

tot = B1 + t0 + Bk + Bq + t0,b = 6.75 + 3 + 4 + 2.717 + 8 = 24.5 hours

The lots leaving a batch machine, leave the machine in batches. With Equation 3.2 we can
approximate the coefficient of variation on the inter-departure time. The inter-departure
time of batches that is! If we were to have a single-lot machine behind the batch machine of
Example 3.5, we can use the coefficient of variation on the inter-departure times of batches to
determine the coefficient of variation on the inter-departure times of lots leaving the batch
machine. But keep in mind that these inter-departure times are no longer independent.
If we use Equation 3.1 for the single-lot machine behind the batch machine we can only
approximate the queueing time!

3.6

Discussion

In this chapter we have presented analytical approaches to determine the flowtime for a
flowline of workstations with stochastic inter-arrival times and stochastic processing times.
We have even discussed ways to calculate the flowtime for batch machines. In some cases,
the presented queueing relations allow an exact calculation of the flowtime. In many cases,
the queuing relations allow for a reasonable approximation of the flowtime.
However, in practice, many manufacturing systems exist that still can not be analyzed using
the queueing relations in this chapter. To illustrate this, consider the following examples:
The queueing relations are applicable to simple flowlines of workstations. They can
not be used for flowlines that feature rework or bypassing.

42

Chapter 3. Analytical models of stochastic manufacturing systems

The queueing relations assume infinite buffers that dispatch lots in a FIFO (FirstIn-First-Out) manner. They can not be used for manufacturing systems that have
dispatching policies or that have finite buffer sizes.
The queueing relations assume machines that process one lot (or batch) type at a time,
where the process time for all lots (batches) is according to the same distribution. They
can not be used for machines that process different lots in a different manner or that
require set-up time.
We have presented some simple extensions that allow us to calculate the flowtime for
batch machines with fixed batch sizes. The relations presented there are unapplicable
if we use a batch forming strategy that allows only a maximum wait-in-batch-time or
variable batch size.
In short, there are a large number of examples, encountered in practice, that can not be
analyzed with the relations from this chapter. Using probability theory and advanced
queueing theory, it is possible to derive relations for these specific cases. However, this
requires significant mathematical knowledge and effort and even then we have only an
approximation. Moreover, if the systems become larger, the mathematical relations quickly
become large and complex. The analytical approaches from previous and this chapter are
very valuable in giving us a first impression of the systems performance. It provides us
with a sense of the systems bottle-necks, the role of variability sources, and the order of
magnitude of the flowtime, throughput, and wip-level.
However, for realistically sized problems, with realistic properties such as dispatching rules,
finite buffers, and batching strategies, accurately determining the systems performance is
best done with simulation. In the remainder of this book we present an approach to analyze
different kinds of manufacturing systems using the discrete-event simulation language Chi.
We start with simple systems and extend the models step by step. At the end of this
book, we can virtually analyze any manufacturing system using Chi. At some points in the
subsequent chapters, we discuss the practical useability of the relations presented in this
chapter, by comparing simulation results with analytical results.

Keyword overview
Modelling manufacturing systems

Analysis of manufacturing systems

Kendalls notation
flowtime approximation for a G/G/1 workstation
Pollaczek-Khinchines approximation
squared standard deviation
squared coefficient of variation
Kuehns linking equation
flowtime approximation for a G/G/m workstation
flowtime approximation for batch machines

3.7. Exercises

3.7

43

Exercises

1. We have a manufacturing line with three workstations, see Figure 3.10. From measurements done, we have the following data available for each workstation:
Workstation 1 consists of an infinite buffer and 5 identical machines with t0 =
15.0 min/lot and c20 = 1.0.
Workstation 2 consists of an infinite buffer and 12 identical machines with t0 =
30.0 min/lot and c20 = 1.0.
Workstation 3 consists of an infinite buffer and one machine with t0 = 3.0 min/lot
and c20 = 1.0.

ta = 4.0 min

td, cd2

t0 = 15.0 min
c02 = 1.0
m=5

t0 = 30.0 min
c02 = 1.0
m = 12

t0 = 3.0 min
c02 = 1.0
m=1

Figure 3.10: Manufacturing line with multiple-machine workstations

We assume that lots arrive at workstation 1 with an inter-arrival time that is distributed according to an exponential distribution with mean 4.0 minutes.
(a) Calculate the utilisation for workstation 1,2, and 3.
(b) Calculate the mean waiting time in the buffer for workstation 1.
(c) Calculate the mean inter-arrival time ta,2 and squared coefficient of variation c2a,2
for lots arriving at workstation 2.
(d) Calculate the mean waiting time in the buffer for workstation 2.
(e) Calculate the mean waiting time in the buffer for workstation 3.
(f) Calculate the mean number of products in buffer 1,2, and 3.
(g) Explain the low wip-level in buffer 2.
(h) Calculate the mean flowtime of a lot for the entire manufacturing line.
(i) What is the mean number of products in the line (the wip-level)?
2. Consider a manufacturing line consisting of three workstations. Each workstation
consists of an infinite buffer and a single machine. In addition we have the following
data:
Lots arrive according to a distribution with mean ta and squared coefficient of
variation c2a = 2.0.

44

Chapter 3. Analytical models of stochastic manufacturing systems

Workstations 1,2, and 3 have a mean process time of t0,1 = 1.0 hr, t0,2 = 4.0 hr,
and t0,3 = 1.0 hr respectively and a squared coefficient of variation c20,1 = 4.0,
c20,2 = 1.0, and c20,3 = 1.0.
Figure 3.11 shows a graphical representation of the system.
ca2 = 2.0

t0
c02

W1

W2

W3

1.0
4.0

4.0
1.0

1.0
1.0

[hr]
[-]

Figure 3.11: Three-workstation manufacturing line

(a) Calculate the maximum throughput max .

(b) What is the mean flowtime of a product if we require a throughput = max ?
We wish to obtain a throughput = 0.9max .
(a) Calculate the utilisation of machine 1,2, and 3.
(b) Calculate the mean flowtime of the line.
(c) Calculate the mean number of products in buffer 1,2, and 3.
(d) If you were to improve the flowtime and/or throughput of this system, with which
machine would you start?
The management decides to invest in workstation 2. We have two alternatives. In
alternative one we invest in speed. We speed up the machine in workstation 2 so that
t0,2 = 1.0 hr, however this induces an increase in variability: the squared coefficient of
variation c20,2 becomes 4.0. In alternative two we simply add 3 similar machines to the
line: workstation 2 consists of 4 identical machines with t0,2 = 4.0 hr and c20,2 = 1.0.
Both alternatives are graphically represented in Figure 3.12.
Alternative 1

t0
c02
m

Alternative 2

W1

W2

W3

1.0
4.0
1

1.0
4.0
1

1.0
1.0
1

[hr]
[-]
[-]

t0
c02
m

W1

W2

W3

1.0
4.0
1

4.0
1.0
4

1.0
1.0
1

[hr]
[-]
[-]

Figure 3.12: Two alternatives

(a) What is the maximum throughput for both alternatives?

(b) Calculate the mean flowtime for alternative 1 for = 0.9max .
(c) Calculate the mean flowtime for alternative 2 for = 0.9max .
(d) Calculate the wip-level for both alternatives.
(e) Which alternative do you prefer?
(f) At first sight, the alternatives yield only a small decrease (or even an increase)
in flowtime, despite the investments done. Are these investments futile?

3.7. Exercises

45

3. We have a batch machine that processes batches of fixed size k. Lots arrive with mean
inter-arrival time ta,l = 4.0 hours and squared coefficient of variation c2a,l = 1.0. The
process time for a batch is proportional with the batch size: t0,b = k t0 , the squared
coefficient of variation is inversely proportional with the batch size: c20,b = c20 /k. For
this machine we have t0 = 3.0 hours and c20 = 0.50.
batchsize k
ta,l = 4.0 hr
2
ca,l
= 1.0
t0,b = kt0
2
c0,b
= c02 /k

Figure 3.13: Batch machine process batches of fixed size k

(a) Express the mean total flowtime for this batch machine in batch size k.
(b) For what value of k is the flowtime minimal? (Do not forget to assure a stable
system.)
4. We have three identical batch machines in parallel. The batch machines have a single
buffer in which batches are formed. The machines process batches of a fixed size k = 4
lots. Lots arrive at the buffer with ta,l = 0.50 hours and c2a,l = 1.0. The batch process
time is characterized by t0,b = 5.0 hours and c20,b = 0.5.

ta,l = 0.50 hr
2
= 1.0
ca,l
k=4
t0,b = 5.0 hr
2
c0,b
= 0.50

Figure 3.14: Three batch machines in parallel

(a) Calculate the utilisation of the batch machines.

(b) Calculate the wait-to-batch-time in the buffer.
(c) Calculate the queueing time for a batch in the buffer.
(d) Calculate the mean total flowtime.

46

Chapter 3. Analytical models of stochastic manufacturing systems

Chapter 4

Single workstation
In this chapter we take the first step into simulation-based analysis of a manufacturing
system. By means of a single workstation, we present the first simple Chi-specification of
a manufacturing system, and discuss the most elementary components of a manufacturing
system.

4.1

Single machine model

We consider a single machine. This machine processes lots. To analyze the behavior of this
machine, we define three processes, together forming model GME, see Figure 4.1.

Figure 4.1: Model GME

Generator process G sends unfinished lots to machine M. Machine M processes these lots and
sends finished lots to exit process E. We interpret these three processes as a representation
of an entire factory, where process G represents the part of the factory before machine M and
process E represents the part of the factory after machine M. In the sequel, we frequently
encounter the concept of introducing a generator process and an exit process.
For this first analysis we assume that generator G generates a lot every 4.0 hours. For
machine M this implies an inter-arrival time (time between two lot arrivals) ta of 4.0 hours.
Moreover, we assume that processing a lot on machine M takes also 3.0 hours, i.e. the
process time t0 equals 3.0 hours. Exit process E is always able to receive lots. We specify
the processes G, M, E, and model GME as follows.

47

48

Chapter 4. Single workstation

type lot = int;

proc G(chan! lot a):
lot x;
while true:
a!x;
delay 4.0;
x = x + 1
end
end

Chi-4.1

proc M(chan? lot a; chan! lot b):

lot x;
while true:
a?x;
delay 3.0;
b!x
end
end

Chi-4.2

proc E(chan? lot a):

lot x;
for i in range(7):
a?x;
write("%10.1f\tE\treceived lot %5d\n", time, x)
end
end

Chi-4.3

model GME():
chan lot a, b;
run G(a), M(a,b), E(b)
end

Chi-4.4

Every lot is represented by a natural number, namely its id number. Generator G sends a
lot (a!x) every 4.0 time units (1.0 time unit 1.0 hour). Every lot has a unique id number.
The first lot has id number 0. After machine M has received a lot (a?x), the lot is processed
for 3.0 hours, and then sent to exit E (b!x). Exit E can permanently receive lots. After each
lot, exit E writes the current time (time) and the lot id number (x) to the screen.
In the model specification we connect generator G to machine M and machine M to exit E.
Recall that an experiment environment is to be defined in order to simulate the model.
Compiling and simulating this model (ams31.chi) yields the following output.
3
7
11
15
19
23
27

E
E
E
E
E
E
E

received
received
received
received
received
received
received

lot
lot
lot
lot
lot
lot
lot

0
1
2
3
4
5
6

4.2. Single machine with varying process time

49

From the simulation output we tell that exit E receives finished lot 0 at time = 3.0. Subsequently, lots 1, 2, 3, ... are received at equidistant intervals of 4.0 time units (here:
hours). This corresponds to the expected simulation outcome. The expected behaviour is
as follows. At time = 0.0, generator G sends the first lot 0 to machine M. At time = 3.0,
machine M has finished the lot and sends it to exit E. So, lot 0 arrives at time = 3.0 at
exit E. As soon as M has sent the processed lot to E, it can receive a new lot from G. Only,
at time = 4.0 G is ready to send the next lot (lot 1) to M.

Below, we describe the behaviour of model3 GME over time by an execution scenario.

time
0.0
0.0
3.0
3.0
4.0
4.0
4.0
7.0
7.0
8.0
8.0
...

G(a)
a!x
delay 4.0

x = x + 1
a!x
delay 4.0

x = x + 1
a!x

M(a,b)
a?x
delay 3.0
b!x

a?x
delay 3.0
b!x

E(b)

x
0

b?x
write(...)

1
b?x
write(...)

a?x

Table 4.1: Execution scenario for GME

In the execution scenario of Table 4.1 we see the statements that each of the parallel processes
executes at different time instants. At time = 0.0, G sends the value of x over channel a
to M. After this synchronous communication, variable x in process M has been assigned the
value 0. Note that at time = 4.0 several statements are performed. Here, we see that
communication, output, and assignment statements are timeless. Time proceeds only when
a time delay (delay statement) is encountered.

4.2

Single machine with varying process time

In practice, processing times are seldom deterministic. In most cases, we have variability
on the process time, i.e. the process time is distributed according to some distribution. For
this example we assume that the processing time is either 1.0 or 5.0 hours. Both possibilities
have 50% chance. The specification of machine M becomes:

50

Chapter 4. Single workstation

proc M(chan? lot a; chan! lot b):

dist bool d = bernoulli(0.5);
lot x;
while true:
a?x;
if sample d:
delay 1.0
else:
delay 5.0
end;
b!x
end
end

Chi-4.5

On average it still takes machine M (1.0 + 5.0)/2 = 3.0 hours to process a lot. Simulation of
model GME with this new machine M, yields the following results:
5
6
10
18
23
28

E
E
E
E
E
E

received
received
received
received
received
received

lot
lot
lot
lot
lot
lot

0
1
2
3
4
5

Though, on average process M still processes one lot every 3.0 hours, and generator G tries
to generate a lot every 4.0 hours, on average process E receives a lot only every 4.5 hours
instead of one lot every 4.0 hours as for the former example. We determine the average of
one lot per 4.5 hour, by running the simulation for a longer time interval, for example for
10000 hours and divide the total simulated time by the total number of products.
Closer investigation of the behavior of generator G and machine M is required to explain this
drop in performance. Hereto, we use the execution scenario in Table 4.2 that belongs to the
simulation results shown.
Read the following explanation, while studying the execution scenario. At time = 0.0 machine M receives lot 0 from generator G. In this simulation, the first sample from bernoulli
distribution d yields false, resulting in a process time of 5.0 hours. At time = 4.0 generator G is willing to send lot 1, but machine M is still processing lot 0. At time = 5.0
machine M finishes lot 0 and sends it to exit E. Exit E receives the first lot at time = 5.0.
Immediately, machine M receives lot 1 from G. The next sample yields true, resulting in a
process time of 1.0 hour. At time = 6.0 M sends lot 1 to exit E. E receives the second lot at
time = 6.0. Machine M is willing to receive the next lot, however, generator G provides the
next lot at time = 5.0 + 4.0 = 9.0. Only at time = 9.0, machine M can start processing
on lot 2. That lot is finished at time = 10.0.
In the first situation, where generator G cannot send the next lot, we speak of blocking, i.e.
machine M blocks the manufacturing system. In the second situation, where machine M has
to wait for generator G to send the next lot, we speak of starving, i.e. machine M is starving.
The effect of variability on the process time is that the average output is lower than the
average maximum output of 0.25 lots per hour. The effects of variability are even worse,

4.3. Single machine with buffer

0.0
0.0
4.0
5.0
5.0
5.0
5.0
6.0
6.0
9.0
9.0
9.0
10.0
10.0
10.0
...

51

G(a)
a!x
delay 4.0
x= x + 1

a!x
delay 4.0

x= x + 1
a!x
delay 4.0

M (a, b)
a?x
delay 5.0

E(b)

x
0

b!x

b?x
write(time, ...)

a?x
delay 1.0
b!x

a?x
delay 1.0
b!x

1
b?x
write(time, ...)

2
b?x
write(time, ...)

x= x + 1
Table 4.2: Execution scenario 2 for GME

if we apply a same bernoulli distribution with 1.0 and 5.0 hours on the inter-arrival times
(Try!). Moreover, note that individual simulations do not yield identical results!
To cope with variability on process times as well as with variability on inter-arrival times
we use buffers.

4.3

Single machine with buffer

We continue to investigate the model of the previous section, where machine M has process
times of either 1.0 or 5.0 hours. We saw that the average output in case of stochastic process
times was less than in case of deterministic process times, whilst the average process time
for both cases was identical. We identified the cause: blocking and starving. To reduce the
effect of blocking and starving we introduce a buffer with infinite capacity before machine M.
We obtain model GBME, see Figure 4.2.

Figure 4.2: Model GBME

The specification of generator G, machine M, and exit E remains unchanged and is found in
Specifications Chi-4.1, Chi-4.5, and Chi-4.3 respectively. We now present the specification
of buffer B.

52

Chapter 4. Single workstation

proc B(chan? lot a; chan! lot b):

list lot xs; lot x;
while true:
select
a?x:
xs = xs + [x]
alt
not empty(xs), b!xs[0]:
xs = xs[1:]
end
end
end

Chi-4.6

Buffer B is always able to receive lots via channel a. If a lot is received, it is added at the
end of list xs. As long as there are lots in the buffer (not empty(xs)), buffer B is willing
to send the first lot of the list (b!xs[0]). After sending the lot, the buffer removes the first
lot from the list (xs = xs[1:]).
The model specification becomes:

model GME():
chan lot a, b, c;
run G(a), B(a, b), M(b, c), E(c)
end

Chi-4.7

Compilation and simulation yields the following output.

1
9
10
17
22
27
32

E
E
E
E
E
E
E

received
received
received
received
received
received
received

lot
lot
lot
lot
lot
lot
lot

0
1
2
3
4
5
6

After running the simulation for a longer time interval, we determine that on average exit E
now receives a lot every 4.0 hours. The buffer has eliminated the effect of blocking and
starving on the long term average output. We investigate the behaviour of model GBME
by means of the execution scenario in Tables 4.3 and 4.4 that belongs to the simulation
results shown.

4.3. Single machine with buffer

0.0
0.0
0.0
0.0
1.0
1.0
4.0
4.0
4.0
4.0
4.0
8.0
8.0
8.0
9.0
9.0
9.0
9.0
10.0
10.0
...

G(a)
a!x
delay 4.0

x= x + 1
a!x
delay 4.0

x= x + 1
a!x
delay 4.0

B(a, b)
a?x
xs = xs + [x]
b!xs[0])
xs=xs[1:]

a?x
xs = xs + [x]
b!xs[0]
xs = xs[1:]

53

M (b, c)

E(c)

b?x
delay 1.0
c!x

c?x
write(time, ...)

b?x
not sample d: delay 5.0

a?x
xs = xs + [x]
c!x
b!xs[0]
xs = xs[1:]

c?x
write(time, ...)

b?x
sample d: delay 1.0
c!x

c?x
write(time, ...)

Table 4.3: Execution scenario for GBME

...
12.0
12.0
12.0
12.0
12.0
...
16.0
16.0
16.0
17.0
17.0
17.0
17.0
20.0
20.0
...

G(a)
x= x + 1
a!x
delay 4.0

x= x + 1
a!x
delay 4.0

B(a, b)

M (b, c)

a?x
xs = xs + [x]
b!xs[0]
xs = xs[1:]

b?x
not sample d:

E(c)

delay 5.0

a?x
xs = xs + [x]
c!x

a!x
delay 4.0

b!xs[0]
xs = xs[1:])
a?x
xs = xs + [x]

b?x
not sample d:

c?x
write(time, ...)
delay 5.0

Table 4.4: Execution scenario for GBME (continued)

Read the following explanation, while studying the execution scenario. At time = 0.0
generator G sends a lot to the buffer. Machine M is available, and at the same time instant
this lot is sent from the buffer to the machine. Sampling from distribution d yields true;
processing the first lot takes 1.0 hour. At time = 1.0, the exit receives the first finished

54

Chapter 4. Single workstation

lot. At time = 1.0, buffer B is empty: machine M has to wait, until the buffer receives a
lot. Machine M is starving. At time = 4.0 the buffer receives lot 1. The buffer immediately
sends this lot to machine M. Processing this lot takes 5 hours. At time = 8.0 generator G
wants to sent lot 2. Though machine M is still in process, the buffer allows the lot to be sent.
Buffer B prevents the generator from being blocked by machine M. After machine M finished
processing lot 1, at time = 9.0, it can immediately continue with lot 2 from the buffer.
Here, buffer B is always able to receive new lots. It has an infinite number of buffer places.
We speak of an infinite buffer . In practice, buffers have a limited number of buffer places.
A finite buffer with N buffer places is specified as follows:
proc B(chan? lot a; chan! lot b; int N):
list lot xs; lot x;
while true:
select
size(xs) < N, a?x:
xs = xs + [x]
alt
not empty(xs), b!xs[0]:
xs = xs[1:]
end
end
end

Chi-4.8

As long as there are less than N lots in the buffer (size(xs) < N), the buffer is able to
receive new lots via channel a.

4.4

Displaying simulated behaviour

There are multiple ways to display simulation results. Apart from textual output, we
presented an execution scenario. The execution scenario provides a complete and accurate
way to describe the behaviour of the simulation model. However, for quick insight on the
behaviour of the simulation model, we prefer to visualize the simulated behaviour. Here we
present two approaches to graphically display the simulation behaviour: the lot-time and
the resource-time diagram. In a lot-time-diagram, which we encountered in Chapter 2, we
plot the state of individual lots against time. In a resource-time-diagram we plot the state
of the individual resources against time.
Figure 4.3 shows a lot-time-diagram distilled from the execution scenario in Tables 4.3
and 4.4.
In a resource-time diagram we record per resource what job or lot is processed. These
diagrams are often referred to as Gantt charts [Gan19]. Figure 4.4 shows a resource-timediagram corresponding to the execution scenario in Tables 4.3 and 4.4.
Besides the lot-time-diagram in which we follow lots and the resource-time-diagram in which
we follow resources, we can plot any quantity or state against time. In Figure 4.5 the number
of lots in the buffer and the state of the machine (idle or busy) are plotted against time.
All these visualizations provide easy insight on the simulated behaviour. They are valuable in verifying your model, that is, checking whether your simulation model behaves as
intended. For deterministic systems we used a lot-time-diagram to determine the mean
flow-time and wip-time-diagram to determine the mean wip-level, see Section 2.2. For a
simulated stochastic system, such a diagram visualizes only (a short period of) one of the
many possible simulation outcomes. Where it does provide insight on the behaviour of the

4.4. Displaying simulated behaviour

lot
0
1
2
3
4
5

55

12

10

16

14

18

20

time [hours]

M
M
B M

lot is

M
B

M
B

processed on M
waiting in buffer

Figure 4.3: Lot-time-diagram of the simulation in Section 4.3

resource
M

8
1

10

12

14

16

18

20

time [hours]
resource is
processing lot i
i

Figure 4.4: Resource-time-diagram of the simulation in Section 4.3

state machine M

s
busy (1)

idle (0)

# [lots]

10

12

14

16

18

16

18

t [hours]

number of lots in B

10

12

14

t [hours]

Figure 4.5: (a) State of machine M (idle or busy), (b) Number of lots in buffer B

simulated manufacturing system, it is of little use in determining the mean flowtime, wiplevel or any quantity. In the next chapter, we discuss how to determine the most important
performance indices for manufacturing systems.

Keyword overview
Modelling manufacturing systems
generator
exit
machine with deterministic process time
machine with stochastic process time
finite buffer
infinite buffer
execution scenario

Analysing manufacturing systems

starving and blocking
process time variability

56

Chapter 4. Single workstation

4.5

Exercises

1. Machine N processes at two different speeds. In fifty percent of the cases the process
time is 0.3 hour, and the other fifty percent the process time is 0.1 hour. We have the
following Chi-specification:
type lot = int;
proc G(chan! lot a):
lot n;
while true:
a!n; n = n + 1
end
end
proc N(chan? lot a; chan lot b):
dist int u = uniform(0,2);
lot x;
while true:
a?x;
if ...:
...
else:
...
end;
b!x
end
end

Chi-4.9

proc E(chan? lot a):

int i; lot x;
while true:
a?x; i = i + 1;
write("%10.1f\tE\treceived lot %5d\n", time, x)
end
end
model GME():
chan lot a, b;
run G(a), M(a,b), E(b)
end
(a) Make a graphical representation of this system.
(b) Predict the mean number of lots that leave the system per hour.
(c) Finish the specification and simulate the model.
(d) Determine the mean number of lots that leave the system per hour by simulation.
(e) Compare the outcome determined by simulation with your predicted outcome.
2. Here is a specification of model S that consists of generator G, 4-place buffer B, and
exit process E. The generator ten times generates a lot every 10 seconds, and the exit

4.5. Exercises

57

process removes lots from the buffer each 5 seconds. If a lot is added or removed, the
new buffer contents are printed.
type lot = int;
proc G(chan! lot a):
for i in range(10):
a!i;
delay 10.0
end
end
proc B(chan? lot a; chan! lot b; int N):
list int xs; int x;
while true:
select
size(xs) < N, a?x:
xs = xs + [x]
alt
size(xs) > 0, b!xs[0]:
xs = xs[1:]
end;
write("%10.1f %s\n", time, xs)
end
end

Chi-4.10

proc E(chan? lot a):

int x;
while true:
a?x; delay 5.0
end
end
model S():
chan lot a,b;
run G(a), B(a, b, 4), E(b)
end
(a) Compile the specification, simulate and explain the results.
(b) What happens if the buffer size is set to two instead of four?
(c) Change the specification such that the generator needs 5 seconds to generate a
lot, and the exit needs 10 seconds to consume a lot. Study again a 4-place and
a 2-place buffer.
3. Consider a workstation consisting of infinite buffer B and machine M. The machine can
process three types of products. The product type is denoted by a natural (0,1,2).
The processing time depends on the product type. The processing time is 1.0, 3.0,
and 6.0 minutes for product type 0,1, and 2 respectively, see Figure 4.6.
The type of the products arriving at B is uniformly distributed. Products arrive at
buffer B with a constant inter-arrival time of 4 minutes. Finished products immediately
can leave the workstation.

58

Chapter 4. Single workstation

machine M
buffer B

finished
products

process time: 1.0 3.0 6.0 min.

Figure 4.6: Work station processes with three product types

(a) Complete the specification below that represents the workstation and its environment.
type lot = int;
proc G(chan! int a):
dist int d = uniform(...);
while true:
a!sample d; delay ...
end
end
proc M(chan? lot a; chan! lot b):
lot x;
...
end
proc B(chan? lot a; chan! lot b):
...
end

Chi-4.11

proc E(chan? lot a):

lot x;
while true:
a?x
end
end
model GBME():
chan lot a, b, c;
run G(a), B(a, b), M(b, c), E(c)
end
(b) Adapt the specification of exit E, so that it prints the number of lots that have
been processed by the manufacturing system to the screen. How many products
are produced by the system per hour?
(c) Determine the number of products that are produced per hour, specified per
product type.
(d) We wish to change the first-in-first-out behavior of the buffer. Each product
type has a different priority. Priority increases with the product type number,

4.5. Exercises

59

i.e. product type 2 has the highest and product type 0 has the lowest priority.
So products of type 2 are processed first, then products of type 1 and then
products of type 0. Adapt the specification and determine by simulation how
many products are processed per hour, specified per product type. (Hint: use (a
tuple of) three lists in buffer B or use list sorting.)

60

Chapter 4. Single workstation

Chapter 5

Measuring performance
In the examples of the previous chapter we have investigated the behavior of a simple manufacturing line. Amongst others, we determined the number of lots that the manufacturing
line processes per hour. In Chapter 2 we presented a number of quantities that characterize the performance of a manufacturing system. In this chapter, we present an approach
to determine the four most frequently used performance indicators by simulation. These
performance indicators are: throughput , flowtime , utilization u, and work-in-process or
wip-level w.
Typically, the throughput, utilization, and wip-level change over time. The flowtime varies
per lot. Let us reconsider the simulation from Section 4.3. From Figure 4.3 we determine
that the flowtime for lot 0,1,2, and 3 is 1.0, 5.0, 2.0, and 5.0 hours respectively. We see that
wip-level w is 1 at for instance = 1.0 and 0 at = 3. For 0.0 20.0 the throughput is
4/20 = 0.2 lots/hour. From Figure 4.4 we determine that the utilization for 0.0 20.0
is 15/20 = 0.8.
In general, we are not interested in the momentous value of performance indicators, but in
the mean throughput , mean flowtime , mean utilization u and mean wip-level w. Herein,
throughput, utilization, and wip-level are averaged over time, and flowtime is averaged
over the lots! In the sequel, we present an approach to calculate the mean values of the
throughput, flowtime, utilization and wip-level. We propose a way to adapt your Chispecification to determine these performance indicators during simulation and apply it on
the manufacturing system from Section 4.3.

5.1

Mean throughput

The mean throughput at time instant is calculated by dividing the number of lots n
that have exited the system up to time instant by the current time . In formula:
=

(5.1)

To determine the mean throughput by simulation, we adapt the specification of exit E.

61

62

Chapter 5. Measuring performance

proc E(chan? lot a):

lot x;
for i in range(1, 1000):
Chi-5.1
a?x;
write("%10.2f\tE\tmean throuhgput =\t%10.3f\n", time, i / time)
end
end
We use i to count the number of lots that has left the system. The mean throughput is
calculated by dividing the number of lots i by the current (simulation) time . Note that
in the previous specifications, a lot is represented by an increasing id number, starting with
0. So, in this case, we could also calculate the mean throughput by (x + 1)/ .
Simulation of model GBME from Section 4.3 with this adapted exit E yields:
5
10
11
17
...
85
93
94
...
9985
9990
9995
10000

E
E
E
E

mean
mean
mean
mean

throughput
throughput
throughput
throughput

=
=
=
=

0.2
0.2
0.272727
0.235294

E
E
E

mean throughput =
mean throughput =
mean throughput =

0.258824
0.247312
0.255319

E
E
E
E

mean
mean
mean
mean

0.249975
0.24995
0.249925
0.2499

throughput
throughput
throughput
throughput

=
=
=
=

After simulating for a large time interval the mean throughput converges to 0.25 lots/hour
(1 time unit in this specification is equivalent to 1 hour). This throughput corresponds to
the results we found at the end of Section 4.3, where exit E received one lot every 4.0 hours.

5.2

Mean flowtime

To calculate the mean flowtime per lot, we record the time at which a lot enters the system.
Hereto, we represent a lot by a record tuple of type (nat,real), where the first element
denotes the id number and the second element the time at which the lot entered the system.
The type declaration and generator G become:
type lot = real;
proc G(chan! lot a):
while true:
a!time; delay 4.0
end
end

Chi-5.2

5.2. Mean flowtime

63

When exit E receives a lot it calculates flowtime of that lot by x.1. One way of
calculating mean flowtime of n lots is adding up the flowtime of all individual lots and
dividing it by the number of lots.
n

n =

1X
i
n i=1

(5.2)

Herein n is the mean flowtime of n lots and i the flowtime of the i-th lot. We prefer to
use the following iterative (and numerically more accurate) approach to calculate the mean
flowtime:

0
for i = 0
(5.3)
i = i 1
i

i1 +
for i = 1, 2, 3, . . .
i
i
Herein i is the mean flowtime of i lots. In exit E we implement this iterative calculation
as follows:
proc E(chan? lot a):
real mphi; lot x;
for i in range(1, 100000):
a?x;
Chi-5.3
mphi = (i - 1) / i * mphi + (time - x) / i;
write("%10.2f\tE\tmean flow time =\t%10.3f\n", time, mphi)
end
end
Note that in this Chi-specification i is represented by . Moreover, is initialized for
i = 0 to 0.0. Simulation yields:
1
9
14
15
21
26
...
89
90
97
98
...
49989
49990
49997
49998

E
E
E
E
E
E

mean
mean
mean
mean
mean
mean

flowtime
flowtime
flowtime
flowtime
flowtime
flowtime

=
=
=
=
=
=

1
3
4
3.75
4
4.33333

E
E
E
E

mean
mean
mean
mean

flowtime
flowtime
flowtime
flowtime

=
=
=
=

2.54545
2.52174
2.625
2.6

E
E
E
E

mean
mean
mean
mean

flowtime
flowtime
flowtime
flowtime

=
=
=
=

4.15468
4.1545
4.15457
4.1544

After simulating for a sufficient long time interval the mean flowtime converges to 4.15
hours. On average, processing a lot takes 3.0 hours, thus on average a lot waits for 1.15
hour in the buffer.

64

5.3

Chapter 5. Measuring performance

Mean utilization

The mean utilization of a machine is defined as the fraction of time that a machine is busy.
In Figure 5.1 the state s (idle or busy) of machine M is plotted against time.

s
busy (1)

idle (0)

ti1 ti ti+1

ti+2
ti+3

ti+4

t [hours]

Figure 5.1: State (idle or busy) of machine M

From this data we can calculate mean utilization un after n events as follows:

un =

n
1 X
(ti ti1 ) si1
tn i=1

(5.4)

Herein un is the mean utilization after the n-th event, ti is the time instant of the i-th event
and si is the state after the i-th event. si = 0 if the machine is idle after the i-th event, and
si = 1 if the machine is busy after the i-th event. We prefer the following iterative (and
numerically accurate) way of calculating the mean utilization:

0
ui = ti1
ti ti1

ui1 +
si1
ti
ti

for i = 0
for i = 1, 2, 3, . . .

(5.5)

Herein, ui is the mean utilization after i events. We have to recalculate mean utilization
ui after each event that changes the state si . To this end, we adapt the specification of
machine M as follows:

5.3. Mean utilization

65

proc M(chan? lot a; chan! lot b):

real tim1, u;
dist bool d = bernoulli(0.5);
lot x;
while true:
a?x;
tim1, u = tAver(tim1, time, u, 0);
write("%10.2f\tM\tmean utilisation =\t%10.2f\n", time, u);
if sample d:
Chi-5.4
delay 1.0
else:
delay 5.0
end;
b!x;
tim1, u = tAver(tim1, time, u, 1);
write("%10.2f\tM\tmean utilisation =\t%10.2f\n", time, u)
end
end
Variable u represents mean utilization ui . Every time the machine receives or sends a lot,
state si changes and mean utilization ui has to be updated. The time instant ti of that
update is the current simulation time . The time instant of the previous update is ti1 .
Recalculation of mean utilization u and updating the time instant ti1 of the most recent
update is performed by function tAver.
func tuple(real t; real u) tAver(real tim1, ti, uim1; int sim1):
if ti == 0.0:
return (ti, uim1)
Chi-5.5
else:
return (ti, tim1 / ti * uim1 + ( ti - tim1 ) / ti * sim1)
end
end
As long as the time instant of the update is zero, the new average and last time instant
remain unchanged. Otherwise, the new average is calculated conform Equation 5.5. If
machine M receives a lot, the machine goes from idle to busy; the previous state si1 = 0.
After the machine has sent the lot, the machine state switches from busy to idle; the previous
state si1 = 1.
Compiling and simulating yields:
5
6
13
14
...
85
89
93
...

M
M
M
M

mean
mean
mean
mean

utilization
utilization
utilization
utilization

=
=
=
=

1
1
0.846154
0.857143

M
M
M

mean utilization =
mean utilization =
mean utilization =

0.588235
0.573034
0.55914

66

Chapter 5. Measuring performance

49993
49994
49997

M
M
M

mean utilization =
mean utilization =
mean utilization =

0.753746
0.75375
0.753725

Running the simulation for a sufficient long time interval causes the mean utilization to
converge to 0.75. This corresponds to the expected behavior. The manufacturing system
receives a lot every 4.0 hours. On average it takes the machine 3.0 hours to process a lot.
Thus, on average the machine is busy for 75% of the time.

5.4

Mean wip-level

Figure 5.2 shows how the wip-level of the manufacturing line can change over time.

w [lots]

ti1

ti

ti+1

t [hours]

Figure 5.2: Work-in-process level over time

The time averaged wip-level wn after n events is given by:

n
1 X
wn =
(ti ti1 ) wi1
tn i=1

(5.6)

Herein, wn is the mean wip-level after n events, ti the time instant of the i-th event, and
wi the wip-level after the i-th event. Iteratively:

0
for i = 0
(5.7)
wi = ti1
ti ti1

wi1 for i = 1, 2, 3, ...

wi1 +
ti
ti
Here, wi is the mean wip-level after the i-th event. We have to recalculate mean wip-level
wi after every event that affects the current wip-level wi . To determine the wip-level during
simulation we need to adapt both generator G and exit E. We choose to let exit E calculate
the wip-level. E is aware of lots leaving the manufacturing system, however E is unaware of
lots entering the manufacturing system. Therefore, we let generator G send a signal to E to
notify E a lot has entered the system. We introduce channel z from G to E. The graphical
representation of model GBME is shown in Figure 5.3.
The specification of generator G and exit E become:

5.4. Mean wip-level

67

Figure 5.3: Model GBME with wip-update channel

proc G(chan! lot a; chan! void z):

while true:
a!time; z!; delay 4.0
end
end

Chi-5.6

proc E(chan? lot a; chan? void z):

real tim1, wmean; int w;
lot x;
for i in range(100000):
select
a?x:
tim1, wmean = tAver(tim1, time, wmean, w);
write("%10.2f\tM\tmean wip =\t%10.2f\n", time, wmean);
Chi-5.7
w = w - 1
alt
z?:
tim1, wmean = tAver(tim1, time, wmean, w);
write("%10.2f\tM\tmean wip =\t%10.2f\n", time, wmean);
w = w + 1
end
end
end
Every time a lot enters the system, generator G sends a signal via channel z to exit E.
In exit E, variable w represents current wip-level wi , and w represents mean wip-level wi .
Exit E is always able to receive lots via channel a and signals from G via channel z. After
either communication, mean wip-level w is recalculated and current wip-level w is updated.
Note that the recalculation of mean wip-level w occurs before updating current wip-level
w. (Try!: Look at Figure 5.2 and Equation 5.7 and argue why this is true.) For this
recalculation we use the same function tAver as we did for calculating the mean utilization,
see Specification Chi-5.5.
Simulation yields:
0
4
10
11
...
94

E
E
E
E

mean
mean
mean
mean

wip
wip
wip
wip

=
=
=
=

0
1
1.3
1.27273

mean wip =

1.08511

68

Chapter 5. Measuring performance

96
100
...
49996
49999
50000

E
E

mean wip =
mean wip =

1.0625
1.06

E
E
E

mean wip =
mean wip =
mean wip =

1.04572
1.04578
1.04578

After running the simulation for a sufficient long time we see that the mean wip-level
converges to 1.05.
From mean throughput determined in Section 5.1 and mean flowtime determined in
Section 5.2, we can calculate mean wip w by applying Littles law: w = = 0.25 4.15 =
1.04 lots. This corresponds to mean wip-level w determined by simulation.

5.5

Variance

Apart from the mean of a quantity we are sometimes interested in its variance (squared
standard deviation) or coefficient of variation (which we can determine from the variance
and the mean). For instance, the variance on the flowtime allows us to predict delivery
reliability.
We iteratively determine the squared standard deviation with the following relation [Lef03]:

s2i = i 2 s2i1 + (xi xi1 )

for i > 1

i1
i
(5.8)

2
for i = 0, 1
si = 0
Herein s2i is the estimated variance of i drawings xi , and xi1 is the estimated mean value of
(i 1) drawings xi . The squared coefficient of variation c2i is obtained by dividing squared
standard deviation s2i by squared mean x2i .
In exit process E below, we use this iterative formula to determine the variance on the
flowtime.
proc E(chan? lot a):
real phi, phi2, mphi, s2phi;
lot x;
for i in range(1, 100000):
a?x; phi = time -x; phi2 = phi * phi;
if i == 1:
s2phi = 0.0
Chi-5.8
else:
s2phi = (i - 2) / (i - 1) * s2phi + phi2 / i;
mphi = (i - 1) / i * mphi + phi / i
end;
write("%10.2f\tE\tmean flow time =\t%10.3f\t", time, mphi);
write("var flow time =\t%10.3f\n", mphi);
end
end

5.6. Measuring other quantities

69

Note that we first recalculate the squared standard deviation, before we recalculate the
mean! From Equation 5.8, we see that in the calculation of s2i for i lots, we use the mean
xi1 of (i 1) lots.

5.6

Measuring other quantities

It is possible to measure other quantities than the ones discussed in this chapter, such as
the mean number of lots in a buffer, the mean number of operations on a lot, the mean
inter-arrival time at a specific machine. Hereto, you can use an approach analogous to the
ones discussed. With respect to determining the mean of a quantity, carefully consider the
following:
Is the quantity to be averaged over time (for example in case of the mean number of
lots in a buffer) or over the lots (for example in case of the mean number of operations
per lot)?
If the quantity is to be averaged over time, use an approach analogous to the one
discussed for the mean utilization or mean wip-level.
If the quantity is to be averaged over the lots, use an approach analogous to the
one discussed for the mean flowtime.
Recalculate the mean for every event that affects the momentous value of the quantity.
This recalculation should be performed just before updating the momentous value of
that quantity.

5.7

Simulation vs Approximation (1): single machine

workstation

In this section we compare analytical results with simulation results. In Chapter 3 we

presented the Pollaczek-Khintchine approximation for the waiting time in a buffer. We
argued that it was exact for a M/G/1 queueing system. In this section, we consider two
manufacturing systems to investigate this.
Example 5.1 D/G/1 system
First, we investigate the manufacturing system of Section 4.3, consisting of a single workstation with an infinite buffer and a machine with a process times of 1.0 or 5.0 hours, both
with 50% probability. The inter-arrival time is fixed, at 4.0 hours. We are thus dealing with
a D/G/1 system.
In Section 5.2 we determined that the flowtime for this system was 4.15 hours. Now let
us see what the analytical approach from Chapter 3 yields. We calculate the coefficient of
variation on the process times.
n

s20 =
c20 =

1X
(1.0 3.0)2 + (5.0 3.0)2
(xi x)2 =
= 4.0
n i=1
2
s20
2
t0

4.0
4
=
3.02
9

70

Chapter 5. Measuring performance

The inter-arrival times are fixed and thus the coefficient of variation on the inter-arrival
times is 0. The mean utilization is 0.75. We calculate the flowtime:
=(

c2a + c20
u
0 + 4/9
0.75

) t0 + t0 = (

) 3.0 + 3.0 = 5.0 hours

2
1u
2
1 0.75

This analytically determined flowtime deviates by about 20% from the simulation results.
For this D/G/1 system, the Pollaczek-Khintchine formula is nothing more than an approximation!

Example 5.2 M/G/1 system
Once more we look at the workstation from Section 4.3, but we let the inter-arrival times be
distributed exponentially with mean 4.0 hours. The process times remain 1.0 or 5.0 hours,
both with 50% chance. We now have an M/G/1 system and the Pollaczek-Khintchine
approximation should be exact.
In Example 3.2 we calculated the flowtime for this workstation:
=(

c2a + c20
u
1 + 4/9
0.75

) t0 + t0 = (

) 3.0 + 3.0 = 9.5 hours

2
1u
2
1 0.75

Now let us see what simulation yields. We use the model from Sections 4.2 and 4.3. We
adapt generator G so that it releases lots with exponentially distributed inter-arrival times.
proc G(chan! lot a):
dist real d = exponential(4.0);
while true:
a!time; delay sample d
end
end

Chi-5.9

We use the exit process E from Section 5.2 to determine the mean flowtime. Simulation
yields:
[user@host chi]./ams37
41109.3 E
mean flowtime = 9.08576
280679 E
mean flowtime = 9.32406
441615 E
mean flowtime = 9.29845
521513 E
mean flowtime = 9.37252
...
8.03191e+06
E
mean flowtime = 9.48781
8.07225e+06
E
mean flowtime = 9.489
8.1125e+06
E
mean flowtime = 9.489
We see that after simulating for a sufficient long time, after over 2 million lots1 , the mean
flowtime starts to converge to 9.5 hours. For this M/G/1 system the Pollaczek-Khintchine
approximation is exact!

1 (Try!:

How can you estimate the number of lots from the simulation output shown above?)

5.8. Discussion

5.8

71

Discussion

In the sequel, we frequently mention the performance indicators throughput , flowtime ,

utilization u and wip-level w. Unless explicitly stated otherwise, we are always referring
to the mean throughput, mean flowtime, mean utilization and mean wip-level. For ease of
reading, we omit the bar above the performance indicator symbols. We use i to refer to
the flowtime of the i-th lot, and we use to refer to the mean flowtime. Similarly, we use
wi or w(t) to refer to the wip-level at a certain point in time, and we use w to refer to the
mean wip-level.

Keyword overview
Modelling manufacturing systems

5.9

Analyzing manufacturing systems

by simulation, determining
mean flowtime ,
mean throughput ,
mean work-in-process w,
mean utilization u, and
variance
approximation vs simulation

Exercises

1. By hand, prove that Equation 5.3 is equal to Equation 5.2 for i = 1, 2, 3, 4.

2. For a workstation system in a factory, lot movement is recorded in a logfile. The
workstation consists of a buffer and a machine. In Figure 5.4 a part of the logfile is
shown.
The data contains a time stamp (time instant in hours), a lot id, and a description
of the event that occurs. From this data estimate the mean throughput, the mean
flowtime, the mean wip-level, and the mean utilization.
3. Verify the results found for the M/G/1 system in Section 5.7 by simulation.
4. By simulation determine the mean throughput, mean flowtime, mean utilization and
mean wip-level of the manufacturing system from Exercise 4.1.
5. Reconsider the workstation from Exercise 4.3 consisting of infinite buffer B and machine M . The machine can process three types of products. The product type is
denoted by a natural (0,1,2). The processing time depends on the product type. The
processing time is 1.0, 3.0, and 6.0 minutes for product type 0,1, and 2 respectively,
see Figure 5.5.
You can use the specification from Exercise 4.3(a).
(a) Adapt the specification of exit E and determine the mean throughput of the
workstation.
(b) Adapt the specification of exit E and determine the mean flowtime per product.
(c) Adapt the specification of exit E and determine the mean flowtime per product,
specified per product type.

72

Chapter 5. Measuring performance

875
875
876.049
876.049
876.049
876.049
880
880
881.396
881.396
881.396
881.396
885
885
887.024
887.024
887.024
887.024
890
890
892.638
892.638
892.638
892.638
895
895
897.496
897.496
897.496
897.496
900
900
901.62
901.62
901.62

LOT0174
LOT0174
LOT0173
LOT0173
LOT0174
LOT0174
LOT0175
LOT0175
LOT0174
LOT0174
LOT0175
LOT0175
LOT0176
LOT0176
LOT0175
LOT0175
LOT0176
LOT0176
LOT0177
LOT0177
LOT0176
LOT0176
LOT0177
LOT0177
LOT0178
LOT0178
LOT0177
LOT0177
LOT0178
LOT0178
LOT0179
LOT0179
LOT0178
LOT0178
LOT0179

ENTERED SYSTEM
ENTERED BUFFER
LEAVES MACHINE
LEAVES SYSTEM
LEAVES BUFFER
ENTERED MACHINE
ENTERED SYSTEM
ENTERED BUFFER
LEAVES MACHINE
LEAVES SYSTEM
LEAVES BUFFER
ENTERED MACHINE
ENTERED SYSTEM
ENTERED BUFFER
LEAVES MACHINE
LEAVES SYSTEM
LEAVES BUFFER
ENTERED MACHINE
ENTERED SYSTEM
ENTERED BUFFER
LEAVES MACHINE
LEAVES SYSTEM
LEAVES BUFFER
ENTERED MACHINE
ENTERED SYSTEM
ENTERED BUFFER
LEAVES MACHINE
LEAVES SYSTEM
LEAVES BUFFER
ENTERED MACHINE
ENTERED SYSTEM
ENTERED BUFFER
LEAVES MACHINE
LEAVES SYSTEM
LEAVES BUFFER

901.62
905
905
906.938
906.938
906.938
906.938
910
910
911.051
911.051
911.051
911.051
915
915
915.345
915.345
915.345
915.345
919.669
919.669
920
920
920
920
924.686
924.686
925
925
925
925
930
930
930.777
930.777

LOT0179
LOT0180
LOT0180
LOT0179
LOT0179
LOT0180
LOT0180
LOT0181
LOT0181
LOT0180
LOT0180
LOT0181
LOT0181
LOT0182
LOT0182
LOT0181
LOT0181
LOT0182
LOT0182
LOT0182
LOT0182
LOT0183
LOT0183
LOT0183
LOT0183
LOT0183
LOT0183
LOT0184
LOT0184
LOT0184
LOT0184
LOT0185
LOT0185
LOT0184
LOT0184

ENTERED MACHINE
ENTERED SYSTEM
ENTERED BUFFER
LEAVES MACHINE
LEAVES SYSTEM
LEAVES BUFFER
ENTERED MACHINE
ENTERED SYSTEM
ENTERED BUFFER
LEAVES MACHINE
LEAVES SYSTEM
LEAVES BUFFER
ENTERED MACHINE
ENTERED SYSTEM
ENTERED BUFFER
LEAVES MACHINE
LEAVES SYSTEM
LEAVES BUFFER
ENTERED MACHINE
LEAVES MACHINE
LEAVES SYSTEM
ENTERED SYSTEM
ENTERED BUFFER
LEAVES BUFFER
ENTERED MACHINE
LEAVES MACHINE
LEAVES SYSTEM
ENTERED SYSTEM
ENTERED BUFFER
LEAVES BUFFER
ENTERED MACHINE
ENTERED SYSTEM
ENTERED BUFFER
LEAVES MACHINE
LEAVES SYSTEM

Figure 5.4: Logdata for single machine workstation

machine M
buffer B

finished
products

process time: 1.0 3.0 6.0 min.

Figure 5.5: Work station process three product types

(d) Adapt the specification of buffer B and determine the mean number of products
in the buffer.
6. We reconsider the workstation of the previous exercise. The previous exercise showed
that the mean number of products in the infinite buffer B is 0.59. We replace the
infinite buffer by a buffer with finite capacity N = 1.
(a) Adapt the specification of the previous exercise and determine the mean throughput for this system.
(b) How can the throughput be less than 1/4.0 = 0.25 products/minute, whilst the
generator is set to generate a lot every 4.0 minutes?
(c) Increase buffer capacity N to 2 products. Again, determine the mean throughput.

Chapter 6

Machines in series
Chapter 4 discusses a single machine with a buffer. In this chapter we take a look at
manufacturing lines containing multiple machines in series.

6.1

Deterministic flowline

The first flowline we consider consists of three workstations. Each workstation consists of
an infinite buffer and one machine, see Figure 6.1 and 6.2.

W0

W1

W2

Figure 6.1: Three workstation flowline

Figure 6.2: Workstation

At first, we choose deterministic inter-arrival and processing times. We take inter-arrival

time ta = 4.0 hours. All machines have the same process time t0 = 3.0 hours.
For this simple deterministic system we are not confined to simulation. The analytical
approaches from Chapter 2 are perfectly suitable to analyze this flowline. Figure 6.3 shows
a lot-time and resource-time diagram for a release rate of = 1/ta = 0.25 lots/hour.
We see that the mean flowtime is 9.0 hours, the mean throughput is 0.25 lots/hour and
the mean utilisation is 0.75. In general, for M deterministic identical workstations in series,

73

74

Chapter 6. Machines in series

lot
0
1
2
3

2
M1

resource

4
M2

10
8
12
M3
M2
M3
M1
M2

6
M1

time [hours]
lot is
M1
B

M1
M2
M3

10

12

time [hours]

1
0

processed on M1
waiting in buffer

resource is
processing lot i
i

2
1

Figure 6.3: Analytically constructed (a) lot-time diagram and (b) resource-time diagram

consisting each of an infinite buffer and a single machine, we can deduce:

= max

u = t0

m
X

t0,i = mt0

if max

i=1

u=1

(6.1)
if > max

Or similarly:

ta

u=

= 1
t0

u=1

t0
ta

m
X

t0,i

if ta t0

i=1

(6.2)

if ta < t0

Now let us reproduce these results by simulation. We specify generator G, machine M, infinite
buffer B, and exit E as follows.

6.1. Deterministic flowline

75

type lot = real;

proc G(chan! lot a):
while true:
a!time; delay 4.0
end
end
proc B(chan? lot a; chan! lot b):
list lot xs; lot x;
while true:
select
a?x:
xs = xs + [x]
alt
not empty(xs), b!xs[0]:
xs = xs[1:]
end
end
end

Chi-6.1

proc M(chan? lot a; chan! lot b):

lot x;
while true:
a?x;
delay 3.0;
b!x
end
end
proc E(chan? lot a):
real mphi; lot x;
for i in range(1, 100000):
a?x;
mphi = (i - 1) / i * mphi + (time - x) / i;
write("%10.2f\tE\treceived lot %5d\tmean flow time =\t%10.3f\n", time, i, mphi)
end
end

We specify process W and model GWWWE as follows.

76

Chapter 6. Machines in series

proc W(chan? lot a; chan! lot c):

chan lot b;
run B(a, b), M(b, c)
end
Chi-6.2

model GWWWE():
chan lot a, b, c, d;
run G(a),
W(a, b), W(b, c), W(c, d),
E(d)
end
Simulation yields the following output:
9
13
17
21
25
29

E
E
E
E
E
E

Received
Received
Received
Received
Received
Received

lot
lot
lot
lot
lot
lot

0
1
2
3
4
5

By simulation, we establish that the mean throughput equals 0.25 lots/hour, the mean
flowtime equals 9 hours, and the mean utilisation u is 0.75. This corresponds to the
analytical results.
Figure 6.4 shows how the wip-level changes over time. After 8 hours, the wip-level shows
a repetitive pattern. The first 8 hours show a start-up effect, since the line started empty.
From Figure 6.4 we determine that, after the start-up behaviour ends, mean wip-level
w = 41 3 + 34 2 = 2.25 lots. This satisfies Littles law: w = = 0.25 9 = 2.25.
w [lots]
3
1
2

10

12

14

16

18

t [hours]

Figure 6.4: Work-in-process level over time

By simulation we investigate how mean flowtime , and mean wip-level w respond to different inter-arrival times. The results are shown in Table 6.1.
Note that all results are in accordance with Littles Law (Try!). For ta 3.0 hours, the
utilisation equals t0 /ta and the throughput equals 1/ta . For ta < 3.0 hours, the line capacity
is insufficient to process all lots: the utilisation equals 1 and the throughput equals 1/t0 . The
number of lots in the system and thus the mean flowtime go to infinity. This corresponds
to the analytical results.

6.2. Unbalanced flowline

ta [hrs] u [] [lots/hour] [hour]

12.0 0.25 1/12 = 0.08 9
9.0 0.33 1/9 = 0.11
9
6.0 0.5
1/6 = 0.17
9
5.0 0.6
1/5 = 0.2
9
4.0 0.75 1/4 = 0.25
9
3.5 0.86 1/3.5 = 0.29 9
3.0 1.0
1/3 = 0.33
9
2.0 1.0
1/3 = 0.33

Table 6.1: u, , , and w for t0 = 3.0 hours and different ta ,

6.2

77

w [lots]
0.75
1.0
1.5
1.8
2.25
2.57
3.0

determined by simulation

Unbalanced flowline

We now explore how the behaviour of the three workstation manufacturing line from Section 6.1 is affected if not all machines have the same (but all still deterministic) process
time. As an experiment, we increase the process time for the second machine from 3.0 hour
to 3.5 hour.
With Chapter 2 we easily deduce that the flowtime is 9.5 hours for release rates under
the maximum release rate. The maximum release rate is 1/3.5 lots/hour, or similarly, the
minimum inter-arrival time is 3.5 hours. For inter-arrival times less than 3.5 hours, the
flowtime grows to infinity.
We now use simulation to reproduce these results. To simulate this unbalanced line, we
specify machine M by the following parametric process.
proc M(chan? lot a; chan! lot b; real t0):
lot x;
while true:
a?x;
delay t0;
b!x
end
end

Chi-6.3

Process W and model GWWWE become:

proc W(chan? lot a; chan! lot c; real t0):
chan lot b;
run B(a, b), M(b, c, t0)
end
model GWWWE():
chan lot a, b, c, d;
run G(a),
W(a, b, 3.0), W(b, c, 3.5), W(c, d, 3.0),
E(d)
end

Chi-6.4

Simulation shows that for inter-arrival times greater than 3.5, the flowtime is indeed 9.5
hours.

78

6.3

Chapter 6. Machines in series

Machines in series without buffers

In Sections 4.2 and 4.3 we discussed the influence of buffers in a workstation for a machine
with varying process times. In this section we explore the influence of buffers in a flowline.
We start with the flowline from Section 6.1 and replace the three deterministic machines,
by three machines that each have a process time that is either 1.0 or 5.0 hours. We keep the
inter-arrival time fixed at 4.0 hours. We remove the infinite buffers. We specify machine M
as follows.
proc M(chan? lot a; chan! lot b; real t0, s0):
dist bool d = bernoulli(0.5);
lot x;
while true:
a?x;
if sample d:
delay t0 - s0
else:
delay t0 + s0
end;
b!x
end
end

Chi-6.5

Here t0 is the mean process time on the machine and s0 is the standard deviation on the
process time. Applying statistics to determine the standard deviation of this distribution,
shows that s0 is indeed the standard deviation. (Try!)
Model GMMME becomes:
model GMMME():
chan lot a,
run G(a),
M(a, b,
M(b, c,
M(c, d,
E(d)
end

b, c, d;
3.0, 2.0),
3.0, 2.0),
3.0, 2.0),

Chi-6.6

Simulation shows that the mean throughput is no longer 0.25 lots/hour, but drops down to
0.21 lots/hour. The mean flowtime is no longer 9.0 hours, but increases to 10.96 hour. Just
as for the single machine, see Section 4.2, the three machine flowline suffers from starving
and blocking. Figure 6.5 shows a resource-time-diagram for a single simulation.
At time = 0.0, lot 0 is released in the system. It is processed by machine 1, 2, and 3 in 1.0
hour. Lot 1 is released at time = 4.0. It is processed on machine 1 for 5.0 hours. At time
= 8.0 the generator tries to release the next lot, but it is blocked. Lot 2 is released at time
= 9.0. Lot 2 takes 1.0 hour processing on machine 1, but when it is finished machine 2 is
still processing lot 1. Machine 1 is blocked. And even though, at time = 13.0, machine
1 is not processing a lot, it is still blocked and can not start processing the next lot. In
the mean time, machine 3 has not processed any lot from time = 3.0 till time = 14.0:
machine 3 is starving.

6.4. Simulation vs Approximation (2): flowline

resource
G
M1
M2
M3

8
1

10

12

2
1

14

16

79

blocked
18

20

3
2

22

time [hours]
resource is
processing lot i
i

3
1

blocked

starving

Figure 6.5: Resource-time-diagram for machines in series without buffers

Introducing finite buffers between the machines reduces the effect of starving and blocking.
For infinite buffers the mean throughput is 0.25 lots/hour and the mean flowtime is 14.3
lots/hour. In the exercises, the reader is invited to investigate how the mean throughput
and mean flowtime depend on the buffer size.

6.4

Simulation vs Approximation (2): flowline

Example 6.1 Validity of Kuehns linking equation

To closer investigate the validity of Kuehns linking equation, we consider a single workstation, consisting of an infinite buffer and one machine. The process time of the machine is
distributed according to a gamma distribution with t0 = 3.0 and c20 = 0.5. We consider two
different inter-arrival time distributions. First, we let the inter-arrival time be distributed
according to an exponential distribution (an M/G/1 system) with ta = {4.0, 6.0, 12.0} and
c2a = 1. Then, we let the inter-arrival time be distributed according to a gamma distribution
(a G/G/1 system) with ta = {4.0, 6.0, 12.0} and c2a = 2.0.
We let exit E determine the mean inter-arrival time at the exit (thus the mean inter-departure
time for the machine) and the squared standard deviation. In Section 5.5 we presented an
approach to iteratively determine the variance.
proc E (chan? lot a):
real tp, ta, tai, s2a;
lot x;
tp = time;
for i in range(1, 100000):
a?x; tai = time - tp; tp = time;
if i <= 1:
s2a = 0.0
Chi-6.7
else:
s2a = (i - 2) / (i - 1) * s2a + (tai - ta) * (tai - ta) / i
end;
ta = (i - 1) / i * ta + tai / i;
write("%10.2f\tE\tmean flow time%10.4f\tvar flow time%10.4f\n",
time, ta, s2a / ta / ta)
end
end
Every time a lot arrives at the exit process, the inter-arrival time ta,i is determined, by
subtracting the time instant tp that the previous lot arrived at the exit from the current

80

Chapter 6. Machines in series

time. Subsequently, squared standard deviation s2a on the inter-arrival time and mean interarrival time ta are calculated. The squared standard deviation can only be calculated for
i > 1 and is calculated before calculating the mean, see Equation 5.8.
In Table 6.2 the analytically determined c2d is compared with the c2d determined by simulation.
ta
4.0
6.0
12.0

exponentially distr. inter-arrival time

c2d (analytical) c2d (simulation)
0.719
0.719
0.875
0.875
0.968
0.968

gamma distr. inter-arrival time

c2d (analytical) c2d (simulation)
1.156
1.050
1.625
1.497
1.906
1.828

Table 6.2: Comparing c2d from analytical results with simulation results

We see that Kuehns linking equation is accurate for the M/G/1 system and less accurate
for the G/G/1 system.

Example 6.2 Flowline from Example 3.3 revisited
In Chapter 3 we discussed how to analytically determine the flowtime for a flowline, using
Pollaczek-Khintchines queueing relation and Kuehns linking equation. In Example 3.3 we
calculated the mean flowtime for a three-workstation flowline, with exponentially distributed
inter-arrival times with mean ta = 4.0 hours and a process time distribution with t0 = 3.0
hours and c20 = 0.5. The mean flowtime was calculated at 26.2 hours. Let us try and
reproduce these results by simulation.
Lots arrive at the flowline with an inter-arrival time that is exponentially distributed, with
mean inter-arrival time ta . A lot is represented by an identification number and a time
stamp.
type lot = real;
proc G(chan! lot a; real ta):
dist real d = exponential(ta);
while true:
a!time; delay sample d
end
end

Chi-6.8

We specify the machines with mean process time t0 and a squared coefficient of variation c20 .
We choose the gamma distribution to represent the process times1 . In the Chi Reference
Manual [HR12] we find that the gamma distribution (a, b) has parameters a and B, with
mean = ab and variance 2 = ab2 . If we have mean process time t0 and squared coefficient
of variation c20 we obtain the following parameters (Try!):
(a, b)

with a =

1
c20

and b = c20 t0

We specify machine M as follows:

1 This is an arbitrary choice. Chapter 9 goes into more detail on choosing an appropriate distribution for
process times.

6.4. Simulation vs Approximation (2): flowline

proc M(chan? lot a; chan! lot b; real t0, c20):

dist real d = gamma(1 / c20, c20 * t0);
lot x;
while true:
a?x; delay sample d; b!x
end
end

81

Chi-6.9

We let the buffer process record the mean flowtime in the previous workstation and the
coefficient of variation on the inter-arrival time of lots arriving in that buffer. In Section 5.5
we presented an iterative way to determine the variance. We specify buffer B as follows.
proc B(chan? lot a; chan! lot b):
int i; real ta, tp, tai, s2a, phi;
list lot xs; lot x;
while true:
select
a?x:
i = i + 1; tai = time - tp; tp = time;
if i <= 1:
s2a = 0.0
else:
s2a = (i - 2) / (i - 1) * s2a
+ (tai - ta) * (ta - ta) / i
end;
ta = (i - 1) / i * ta + tai / i;
phi = (i - 1) / i * phi + (time - x) / i;
xs = xs + [time]
alt
not empty(xs), b!xs[0]:
xs = xs[1:]
end
end
end

Chi-6.10

Every time a lot arrives at the buffer, the variance on the inter-arrival time and the mean
inter-arrival time are recalculated. Subsequently, the mean flowtime for the previous
workstation is determined. The flowtime of the lot in the previous workstation is equal
to the current time minus the time instant of entering the previous buffer (x.1). After these
calculations, the time stamp of the lot (x.1) is updated to the current time, so that the next
buffer can determine the flowtime of its preceding workstation.
To determine the mean flowtime in the last workstation, we use the following exit process.
proc E (chan? lot a):
lot x; real mphi;
for i in range(1, 10000):
a?x;
mphi = (i - 1) / mphi + (time - x) / i
end
end

Chi-6.11

82

Chapter 6. Machines in series

Finally we have the following process and model:

proc W(chan? lot a; chan! lot c):
chan lot b;
run B(a, b), M(b , c, 3.0, 5.0)
end
Chi-6.12

model GWWWE():
chan lot a, b, c, d;
run G(a, 4.0),
W(a, b), W(b, c), W(c, d),
E(d)
end

After adding output statements to the buffer and the exit specification, and running the
simulation for a sufficient long time, we obtain the following simulation output:
39771.8
39787.6
39792.3
39829
...
8.45643e+07
8.45644e+07
8.45644e+07
8.45644e+07
...

B0
B1
B2
E

phi
phi
phi
phi

0
9.24992
8.87956
8.25412

ca2
ca2
ca2

1.10149
0.74992
0.68941

B0
B1
B2
E

phi
phi
phi
phi

0
9.74992
8.37956
8.05412

ca2
ca2
ca2

1.00149
0.71992
0.65941

Modelling and simulation of Example 3.3 yields a mean total flowtime = 26.2 hours. At
first sight, the results seem to exactly correspond with the analytical results, but let us
closer compare the analytical and simulation results. Table 6.3 shows the flowtime and
the coefficient of variation on the inter-arrival time ca for each workstation, determined
analytically and by simulation.

W0
W1
W2

Analytical approximation
0 = 9.75 hrs c2a,0 = 1
1 = 8.49 hrs c2a,1 = 0.719
2 = 7.92 hrs c2a,2 = 0.596

Simulation
0 = 9.75 hrs c2a,0 = 1.00
1 = 8.38 hrs c2a,1 = 0.719
2 = 8.05 hrs c2a,2 = 0.659

total = 26.2 hrs

= 26.2 hrs
Table 6.3: Analytical approximation results versus simulation results

In Table 6.3, we see that for workstation W0 the analytical approximation of the mean
flowtime is equal to the mean flow time obtained by simulation. This is correct, since we
are dealing with an M/G/1 system. Moreover, we see that the analytical approximation
of c2d,0 = c2a,1 is also equal to the simulation results. We saw in the previous example
that the for an M/G/1 system, the linking equation is fairly accurate, however, it is only

6.5. Exercises

83

exact for an M/M/1 system2 . The subsequent analytical approximations deviate from the
simulation results. For workstation W1 and subsequent workstations the inter-arrival times
are no longer exponentially distributed. As a consequence, the analytical approximations
no longer yield the exact flowtime and the coefficient of variation on the inter-arrival times
deviates stronger. In this particular case, the deviations cancel each other out, and the total
flowtime determined analytically corresponds (up to three significant numbers) to the one
determined by simulation. In general, this is not the case!


Keyword overview
Modelling manufacturing systems
flowline
parametric machine model
gamma-distributed process time

6.5

Analysis of manufacturing systems

starving and blocking
throughput and flowtime for a flowline
influence of buffers on throughput and
flowtime
analytical vs simulation (2)

Exercises

1. Consider a flowline with three workstations. Each workstation consists of a finite

buffer with N places and a machine with mean process time t0 = 3 hours and standard
deviation 0 = 2 hours on the process time. Lots arrive at the system with an interarrival time of ta = 4 hours. We measure the mean throughput and the mean
flowtime .
G

ta

t0, c0

t0, c0

t0 , c0

Figure 6.6: Three workstation flowline

Are the following statements true or false? (Try to determine your answer by reason.
If desired, use simulation to verify your answer.)
(a) If buffer capacity N increases, the influence of blocking and starving decreases.
(b) If buffer capacity N increases, mean throughput increases.
(c) If buffer capacity N is zero, mean flowtime = 9 hours.
(d) If buffer capacity N goes to infinity, mean throughput goes to 0.25 lots/hour.
(e) If buffer capacity N goes to infinity, the throughput and flowtime are equal to
the throughput and flowtime in case of deterministic process times of 3.0 hours.
2. Reconsider the flowline from Exercise 1. Are the following statements true or false?
(Try to determine your answer by reason. If desired, use simulation to verify your
answer.)
2 For

an M/M/1 system the linking equation always yields the trivial result that c2a = c2d = 1.

84

Chapter 6. Machines in series

(a) For N = , increasing 0 has no effect on mean throughput .

(b) For N = 0, increasing 0 has no influence on mean throughput .

(c) For N = , increasing 0 has no effect on mean flowtime .

(d) For N = 0, increasing 0 has no influence on mean flowtime .

3. Reconsider the flowline from Exercise 1. We now also measure the mean wip-level w.
Are the following statements true or false? (Try to determine your answer by reason.
If desired, use simulation to verify your answer.)

(a) If N = , the minimum flowtime that occurs for a single lot is 9 hours.

(b) If N = 0, the maximum flowtime that occurs for a single lot is 9 hours.

(c) If 0 = 0, mean flowtime = 9 for any buffer capacity N .

(d) For ta > t0 and N = , mean flowtime and mean wip-level w go to infinity.

(e) For t0 > ta and N = 0, mean wip-level w = 3.

(f) For ta > t0 and N = , mean throughput = 1/ta .

4. We have a manufacturing system that consists of two machines M. Both machines are
identical except for the process times: the first one has a mean process time of 0.2
hour with a variation of plus or minus 0.1 hour, whereas the second one has a mean
process time of 0.4 hour with a variation of plus or minus 0.2 hour. The manufacturing
system is modelled by the following specification:

6.5. Exercises

85

type lot = real;

proc G(chan! lot a):
while true:
a!time
end
end
proc M(chan? lot a; chan! lot b; real m, s):
dist int d = uniform(0, 2);
lot x;
list(2) real pt = [m - s, m + s];
while true:
a?x; delay pt[sample d]; b!x
end
end;

Chi-6.13

proc E(chan? lot a):

lot x;
for i in range (1, 10000):
a?x;
write("%10.2f\tE\tMean output\t%10.5f\n", time, i / time)
end
end
model S():
chan lot a, b, c;
run G(a),
M(a, b, 0.2, 0.1), M(b, c, 0.4, 0.2),
E(c)
end
Notice that machine M is specified as a process with parameters M and s, denoting
the mean and standard deviation on the process times. In this way the mean process
times and standard variations can easily be changed.
(a) Draw a graphical representation of the model and simulate the specification.
(b) Determine the mean number of lots processed per hour and explain the results.
(c) Insert a finite buffer with N places between both machines M. Determine the output in [lots/hour] for different buffer sizes (0,1,2,...). Plot the output in [lots/hour]
against the buffer size N . Explain the results.
(d) What buffer size would you advice in this case?
5. Figure 6.7 shows a part of a printed circuit board (PCB) assembly line. This part
contains two component placement robots (CPRs) that place electrical components
on printed circuit boards (PCBs). A PCB is positioned under the placement robot
with a conveyor belt.
Placing one component on a PCB takes 0.8 to 1.1 seconds. The placement time is
uniformly distributed. The number of components that have to be placed by a robot
varies from 10 to 14 components (in Chi: uniform(10,15)) for each robot and is also

86

Chapter 6. Machines in series

conveyor belt
system

Printed Circuit
Board (PCB)

components
to be placed
Component
Placement
Robot (CPR)

Figure 6.7: PCB assembly line

uniformly distributed. We neglect the conveyor transport time. In the first configuration there are no buffer places before and after the placement robots, see Figure 6.7.
We use a Chi-specification to analyze the performance of the PCB-assembly line. A
part of the specification is shown below.

6.5. Exercises

87

type pcb = real;

proc G(chan! pcb a):
while true:
a!time; delay ...
end
end
proc CPR(chan? pcb a; chan! pcb b):
dist real pt = uniform(0.8, 1.1);
dist int nc = uniform( 10, 15);
pcb x;
int i, n;
while true:
a?x; n = ...; i = 0;
while ...:
delay ...;
...
end;
b!x
end
end

Chi-6.14

proc E(chan? ... a):

pcb x;
for i in range(1, 100000):
a?x;
write("%10.2f\tE\treceived PCB\t5d\tflow time%10.2f\n", time, i, time - x)
end
end
model PcbAssembly():
chan ... a, b, c;
run G(a), CPR(a, b), CPR(b, c), E(c)
end
(a) What is the time unit of this specification?
(b) Estimate the mean process time per PCB by hand.
(c) We assume that a new PCB arrives at the PCB assembly line every 15 seconds.
Finish the specification of the PCB assembly line. Determine the mean throughput in PCBs/seconds and the mean flowtime in seconds.
(d) Investigate whether increasing the arrival rate results in a higher throughput.
What maximum throughput can be attained? Is the mean flowtime affected by
increasing the arrival rate?
6. As an alternative for the PCB assembly line of Exercise 5 we have a configuration that
has a buffer place for one PCB between the placement robots, see Figure 6.8.
(a) By simulation, determine the mean throughput and mean flowtime for the PCB
assembly line if the arrival rate is 4 PCBs/minute.

88

Chapter 6. Machines in series

Printed Circuit
Board (PCB)

components
to be placed

buffer place
Component
Placement
Robot (CPR)

Figure 6.8: PCB assembly line with buffers

(b) What is the maximum attainable throughput if the arrival rate can be freely
chosen? In case of maximum throughput, what is the mean flowtime?

6.6

Case

There is something completely wrong, the plant manager pointed out, You said you would
buy three machines of equal capacity! But the line seems to be completely unbalanced.
But I did buy three machines of equal capacity, the design engineer replied, I really did.
Why then is the buffer in front of the first machine generally almost three times as full as
the buffer of the third machine?, the manager continued. Well, sir, to be honest, I dont
know. The lots arrive in the first buffer at a constant rate of 4 lots/hour. The capacity of
each of the machines is much larger: 5 lots/hour. So that seems okay to me. Okay? , a
utilization of 80% you call that okay? Those machines are expensive, I want the full 100%.
Why dont you increase the release rate to 5 lots/hour? Well, Im not sure whether that
is a good thing to do. I feel there is some trade-off between throughput and flowtime, and
currently the flowtime is already near your maximum allowed flowtime of 3 hour. What
trade-off? I want a maximum throughput with a minimum flowtime. And if all lots are
waiting in the buffer of the first machine, I do understand that that does no good for the
flowtime. I demand an explanation. Sir, I really dont understand it, but Ill try to find
out why the buffers are not equally filled and whether or not a trade-off between throughput
and flowtime exists. I hope we can further increase the release rate without surpassing the
flowtime of 3 hour. Report me at the end of this week. I hope you understand we cannot
afford additional machines. The design engineer left the office of the manager and walked
to the production line. I really do not understand, he thought, such a simple production
line. A line with just three machines and a buffer in front of each machine. The vendor of
the machines guaranteed that each of the three machines has a mean effective process time
of 0.2 hour per lot. Wait, he also shortly mentioned something about variation coefficients.
Ah, yes, I remember. He told me that the first machine has a coefficient of variation of about
1.5, and the second and third machine a variation coefficient of 0.5. But what difference
does that make? The first machine is simply much less reliable due to its sensible and
highly complex production process. On average the available capacity is still equal for all
machines. So why bother and spend money to reduce the variability? And it is not for
nothing I have put buffers in front of the machines.

6.6. Case

89

1. We would like to make a Chi-model of the three machine production line to evaluate
the hypothesis of the design engineer that a trade-off is present between throughput
and flowtime. Part of the specification of the production line is given below:
type lot = real;
proc G(chan! lot a; real m):
while true:
delay m; a!time
end
end
proc B(chan? lot a; chan! lot b):
...
end
proc M(chan? lot a; chan! lot b; real m, s):
...
end

Chi-6.15

proc E(chan? lot a):

...
end
model S():
chan lot
run G(a,
B(a,
B(c,
B(e,
E(g)
end

a, b, c,
0.25),
b), M(b,
d), M(d,
f), M(f,

d, e, f, g;
c, 0.2, 0.3),
e, 0.2, 0.1),
g, 0.2, 0.1),

(a) Make a graphical representation of the model.

(b) Complete the specification of processes B and M. Note that we deal with an infinite
buffer. Use a gamma-distribution for the process time of machine M, with M the
mean and s the standard deviation.
(c) Complete the specification of process E such that the mean throughput is computed, as well as the mean flowtime from generator to exit.
(d) Run the program and compute the throughput and flowtime. Try to obtain about
two significant digits. Run the program for several other release levels as well,
and investigate whether there is a trade-off between throughput and flowtime.
For which release level (throughput) is the maximum allowed flowtime reached?
2. To evaluate the observation of the plant manager with respect to the buffer contents,
we must slightly extend our Chi-model. Besides the flowtime from start to exit, we
must compute the average waiting times in the buffers as well. The buffer with the
longest waiting time has the largest number of the lots in its queue (note that we can
compare the waiting times of the buffers since the throughput of the machines are
identical, see Littles Law).

90

Chapter 6. Machines in series

To compute the mean waiting times, the buffers record the lot arrival time and compute the waiting time in the buffer when the lot leaves the buffer.
(a) Change the specification of buffer B such that, when a lot has been received, it
records the time of arrival of that lot, and when a lot is sent, it computes the
waiting time of that lot and the mean waiting time.
(b) Run the program for the original release rate of 4 lots/hour. Let it run some
time until the large fluctuations have died out.
(c) What do your observe with respect to the mean waiting times in the three buffers?
Do you agree with the observation of the plant manager?
(d) Change the coefficient of variation of the first machine into 0.5, and instead
replace the coefficient of variation of the third machine by a value of 1.5. Rerun
the program. What happens to the overall flowtime and the waiting times in the
buffers?
(e) How is the average number of lots waiting in the queue of a machine affected?
3. The Chi simulation model is a powerful tool to analyze the three machine production
line. However, for this simple production line, we can also use analytical approximations to estimate mean throughput and flowtime. These relations enable to better
understand the actual physical behavior.
Consider again the three machine production line where the first machine has a process
time coefficient of variation of 1.5, and the other two machines a coefficient of 0.5.
(a) What is the utilization of each of the machines, given the constant arrival rate
of 4 lots/hour?
(b) Estimate the average waiting time of lots in the buffer of the first machine, and
compute the corresponding WIP-level (i.e. average number of lots waiting in the
queue of the first machine).
(c) Estimate the departure coefficient of variation of the first machine.
(d) Compute the mean waiting times and WIP-levels for the second and third machine as well.
(e) Compare the estimated waiting times with the waiting times obtained via simulation. Is the approximation reasonable?
(f) Can you explain why the lots are not equally distributed over the three buffers?
What will happen when the process time coefficient of variation of 1.5 applies for
the third machine instead of the first machine?

Chapter 7

Machines in parallel
In the previous chapter we considered modelling machines in series. In this chapter we
model machines in parallel.

7.1

Deterministic machines in parallel

Consider three identical machines in parallel with a single (infinite) buffer, see Figure 7.1.
Each machine has a deterministic process time of 9.0 hours. The generator releases a new
lot in the line every 4.0 hours.

b
G

c
c

c
M

Figure 7.1: Three parallel machines

With the analytical techniques from Chapter 2 we expect a mean throughput of = 1/4
lot/hour, an utilisation of 0.75, and a mean flowtime = 9.0 hours. We wish to verify our
expectations by simulation.
The specifications of generator G and machine M are more or less identical to the ones
presented in Chapter 4.

91

92

Chapter 7. Machines in parallel

type lot = tuple(real t; int k);

proc G(chan! lot a):
while true:
a!(time, -1); delay 4.0
end
end

Chi-7.1

proc M(chan? lot a; chan! lot b; int k):

lot x;
while true:
a?x; x.k = k; delay 9.0; b!x
end
end
Buffer B receives lots from the generator and can send lots to one of the three parallel
machines. We specify buffer B as follows.
proc B(chan? lot a; chan! lot b):
list lot xs; lot x;
while true:
select
a?x:
xs = xs + [x]
alt
not empty(xs), b!xs[0]:
xs = xs[1:]
end
end
end

Chi-7.2

As long as the buffer contains lots, the buffer tries to send a lot to one of the machines over
channel B. The buffer sends the lot to the machine that can receive the lot first. If more
than one machine can receive the lot at the same time, the buffer chooses a machine to send
the lot to in a non-deterministic fashion.
Exit E can always receive lots from the machines.
proc E (chan? lot a):
lot x;
for j in range(10):
Chi-7.3
a?x;
write("%10.2f E received lot %10.1f from machine %2d\n", time, x.t, x.k);
end
end
Finally, we have model Parallel.

7.2. Analytical vs Simulation (3): Parallel machines

93

model Parallel():
chan lot a, b; chan lot c;
run G(a), B(a, b),
unwind j in range(3):
M(b, c, j)
end,
E(c)
end

Chi-7.4

Simulation yields:
9.00
13.00
17.00
21.00
25.00
29.00
33.00
37.00
41.00
45.00

E
E
E
E
E
E
E
E
E
E

received
received
received
received
received
received
received
received
received
received

lot
lot
lot
lot
lot
lot
lot
lot
lot
lot

0.0
4.0
8.0
12.0
16.0
20.0
24.0
28.0
32.0
36.0

from
from
from
from
from
from
from
from
from
from

machine
machine
machine
machine
machine
machine
machine
machine
machine
machine

0
1
2
0
1
2
0
1
2
0

We extend exit E to measure the mean throughput, mean flowtime, and mean wip level. In
correspondence with our expectations, we establish that = 0.25 lots/hour, = 9.0 hours,
and w = 2.25 lots.
A flowline with a single machine with a mean process time of 3.0 hours has the same capacity
of 1/3 lots/hour. In case of deterministic arrival time and process time, the flowtime for the
single-machine line is 3.0 hours. However, in case of high variability on the arrival time and
process times, the parallel-machine line yields a lower variability on the flowtime and even
may yield a lower flowtime than the single machine flowline with the same capacity due to
shorter waiting times. We investigated this in Chapter 3 Exercise 2.

7.2

Analytical vs Simulation (3): Parallel machines

In Chapter 3 we presented analytical approximations for the flowtime in a workstation

consisting of multiple identical machines in parallel. In this section we wish to compare
this analytical approximation with simulation results. As a test case we consider a single
workstation with three identical parallel machines. Lots arrive with a gamma-distributed
inter-arrival times with mean ta and coefficient of variation ca . The process times of each
machine are distributed according to a gamma distribution with mean t0 and coefficient
of variation c0 . We set c2a = 1.0, t0 = 9.0, and c20 = 0.25. We let ta vary within the
set {3.5, 4.0, 5.0, 6.0, 9.0, 12.0}. In Table 7.1 we compare the mean flowtime determined by
simulation with the mean flowtime determined analytically.
We see that, for this test case, the approximation from Chapter 3 is a fair approximation for
the flowtime; all deviations are 1% in order of magnitude. For higher degrees of variability,
for example c2a = c20 = 2.0, the approximation is less accurate.

94

Chapter 7. Machines in parallel

c2a = 1.0, c20 = 2.0

ta
u
(analytical) (simulation)
3.5
0.86
18.90
18.93
4.0
0.75
13.43
13.37
5.0
0.60
10.84
10.75
6.0
0.50
10.06
9.94
9.0
0.33
9.38
9.28
12.0 0.25
9.20
9.12
Table 7.1: Comparing from analytical results with simulation results

7.3

Parallel machines for different lot types

Again, consider three machines in parallel with separate buffers. The generator now releases
lots of three different types in the line. Lot type 0 can be processed on machine M0 , lot
type 1 can be processed on machine M1 , and lot type 2 can be processed on machine M2 .
Lot type 0,1,2 are released in a ratio of 4:3:3, with a release rate of 1 lot every 4.0 hours.
Each machine takes 9.0 hours processing. Each buffer stores one lot type only. We wish to
determine the mean throughput and flowtime per lot type and the utilisation per machine.
Figure 7.2 shows a graphical representation of the flowline.

a0
G

a1

b0

b1

b2

M0
M1

a2

c
c

c
M2

Figure 7.2: Three parallel machines processing different lot types

The type lot is now of type tuple(real t; int k) denoting the lot type {0,1,2}, and the
time of release. We specify generator G as follows.

7.3. Parallel machines for different lot types

95

type lot = tuple(real t; int k);

proc G(list(3) chan! lot a):
dist int d = uniform(0, 10);
list(10) int ka = [0, 0, 0, 0, 1, 1, 1, 2, 2, 2];
int k;
while true:
k = ka[sample d];
a[k]!(time, k);
delay 4.0
end
end

Chi-7.5

Every 4.0 time units, generator G generates a lot. Vector ja and uniform distribution d are
used to represent the creation of different lot types. In this way, lot types 0,1, and 2 are
generated in the ratio of 4:3:3. Lot type i is sent over channel ai . Buffers Bi are standard
infinite buffers. A specification can be found in Chapter 4.
The specification of machines Mi and exit E are identical to the ones presented in Section 7.1.
We specify model Multtype as follows.
model Parallel():
list(3) chan lot a, b;
chan lot c;
run G(a),
unwind j in range(3):
B(a[j], b[j]), M(b[j], c)
end,
E(c)
end

Chi-7.6

By simulation we determine the mean throughput per lot type and the mean flowtime per
lot type. The results are shown in Table 7.2.
lot type

w
0
0.1
31.1 3.11
1
0.075 14.3 1.08
2
0.075 14.3 1.08
all
0.25 21.02 5.25
Table 7.2: Performance for three-type flowline
Note that, though we still have deterministic process times, the mean flowtime of a lot is no
longer 9.0 hours. This is because the lot type is stochastic. Chances are 9/100 + 9/100 +
16/100 = 0.34 that two lots of the same type are generated consecutive. Generating two lots
of the same type in a row, causes an increase in flowtime for that lot. The mean utilisation
for machine M0 , M1 , and M2 is 0.9, 0.68, and 0.68.

Alternative model
In the specification of the multi-type flowline we used separate buffers for each lot type. A
different modelling approach is to use a single buffer process that stores the lot types in

96

Chapter 7. Machines in parallel

separate lists. The graphical representation of the model is depicted in Figure 7.3.

b0
G

B3

b1

M0
M1

b2

c
c

c
M2

Figure 7.3: Three parallel machines processing different lot types, alternative model

The specification of generator G becomes.

proc G(chan! lot a):
dist int d = uniform(0, 10);
list(10) int ka = [0, 0, 0, 0, 1, 1, 1, 2, 2, 2];
int k;
while true:
k = ka[sample d];
a!(time, k);
delay 4.0
end
end

Chi-7.7

We specify multi-type buffer B3 as follows.

proc B3(chan? lot a; list(3) chan! lot b):
list(3) list lot xss; lot x;
while true:
select
a?x:
xss[x.k] = xss[x.k] + [x]
alt
unwind j in range(3):
not empty(xss[j]), b[j]!xss[j][0]:
xss[j] = xss[j][1:]
end
end
end
end

Chi-7.8

Buffer B uses vector xs of lists. Each list in the vector stores a different lot type. The buffer
can always receive a lot over channel a. The lot is added to the list of the corresponding
type. If there are lots of lot type 0 (len(xs.0) > 0), buffer B tries to send the first element
of the list over channel b0 to machine M0 . If multiple lot types are available, B sends a lot

7.4. Exercises

97

to the machine that can receive a lot first. If multiple lot types are available, and multiple
machines are willing to receive a lot at the same time, buffer B chooses in a non-deterministic
fashion.
The specifications of machine M and exit E are identical to the ones presented earlier. The
model specification becomes:
model Parallel():
chan lot a;
list(3) chan lot b;
chan lot c;
run G(a), B3(a, b),
unwind j in range(3):
M(b[j], c)
end,
E(c)
end

Chi-7.9

By simulation we obtain the same results as for Specification Chi-7.5.

Keyword overview
Modelling manufacturing systems
identical parallel machines
conveyor model for parallel machines
parallel machines for different lot types

7.4

Analysis of manufacturing systems

analytical vs simulation (3)

Exercises

1. Reconsider the flowline from Chapter 3 Exercise 1. Determine the mean flowtime for
this flowline by simulation. Compare simulation results with the analytical results.
2. Consider a multi-product flowline with three workstations. Each workstation has
three machines that can perform different operations. Figure 7.4 shows a graphical
representation of the multi-product flowline.

b0
G

b1

M0
M1

b2

c1
c2

M2

d0

c0
B

d1

M3
M4

d2

e0
e1

M6

f0
B

e2
M5

f1

M7

g0
g1

g2

f2
M8

Figure 7.4: Multi-product flowline

The flowline is used to process six different lot types. Each lot type requires processing
on different machines. However, each lot type only requires processing on one machine

98

Chapter 7. Machines in parallel

per workstation. Table 7.3 shows the required process steps per lot type. Moreover,
the table shows what percentage of the lots is of that specific type.
lot type percentage of assortment
routing
0
10%
M0 , M3 , M6
1
10%
M0 , M3 , M7
2
20%
M0 , M4 , M6
3
20%
M0 , M4 , M7
4
20%
M1 , M5 , M8
5
20%
M2 , M5 , M8
Table 7.3: Product assortment and required processes
We release new lots at a rate of 1.0 lot/hour.
(a) By hand, estimate the mean number of lots that arrive at each machine per hour.
(b) By hand, determine the required process time per machine, to attain a mean
utilisation of 0.9.
(c) By hand, determine the minimal flowtime per lot type.
(d) Build a model of the multi-product flowline. Lots are of type (nat, nat, real),
representing the identification number, the lot type, and the time of release in
the line. Assume that the machines have a fixed process time. You can use
fragments of the specifications discussed in Section 7.3.
(e) By simulation, determine the mean flowtime per lot type. Compare these flowtimes with the minimal flowtime per lot type. Explain the increase in flowtime,
while all machines have deterministic process times.

Chapter 8

Other configurations
In the previous chapter we considered the elementary configurations of buffers and machines:
machines in series and machines in parallel. In practice, many other configurations of buffers
and machines exist, see for example [RT98, CAJ01, HS01]. Figure 8.1 shows a number of
possible configurations of machines. Table 8.1 provides a description for each configuration.

(a)

(e)

AP

(f)

(c)
M

(b)

(d)

(g)

Figure 8.1: Different networks of machines

Depending on the desired properties of the production process (e.g. production volume,
flexibility in production volume, flexibility in product type) a different configuration is
desired. In theory, an unlimited number of possible configurations exist. In the previous

99

100

Chapter 8. Other configurations

Configuration
(a) Flowline
(b) Flowline with bypassing

Description
Machines and buffers are placed in series.
Some lots skip an operation. One may think of
an automotive assembly line, where some autos
skip the installation of an airbag.
(c) Flowline with backtracking
(Some) lots revisit a previous operation. One
may think of lots that need rework, of lots that
repeatingly require an operation (e.g. painting
multiple layers with intermediate drying), or of
a re-entrant flowline for the production of ICs.
(d) Parallel machines
From a single buffer lots can be processed on
each of the parallel machines. Parallel machines
increase the capacity.
(e) Converging flowlines
Parallel flowlines converge into a single buffer.
(f) Single machine, multiple buffers Machine processes lots from multiple buffers.
One may also think of an assembly process,
where the machine assembles a number of parts.
(g) Jobshop
We have several workstations, consisting of one
or more machines and buffers. Each lot can have
a different routing.
Table 8.1: Different configurations of machines and buffers

chapters we discussed configurations (a), (d), and (e). In the sequel, we discuss the remaining
four configurations. We present a way to specify these configurations and perform a basic
analysis. Here, we are not interested in conducting a thorough comparison study. Instead,
we provide a number of model components together with an approach to analyze different
configurations. For more on different configurations and lay-outs, we refer to [CAJ01].

8.1

Bypassing

Consider a flowline consisting of three machines and three infinite buffers. The flowline
processes two lot types. The generator releases a new lot every 4.0 hours. The chance that
the new lot is of type 0 is 50%. Lots of type 0 require processing on all three machines
subsequently, lots of type 1 only require processing on machine M0 and M2 . Processing a
lot takes 3.0 hours on machine M0 and M2 and 6.0 hours on M1. The minimal flowtime for
lots of type 0 is 12.0 hours, for lots of type 1 it is 6.0 hours. Thus, on average the minimal
flowtime is 9.0 hours. We wish to determine the mean flowtime by simulation. Figure 8.2
shows a graphical representation of the model.
G

B0

M0

BS

M1

BM

M2

Figure 8.2: Three-workstation flowline with bypassing

8.1. Bypassing

101

A lot is of type tuple(int num; real t; int k), denoting the identification number, the
time of release, and the lot type 0 or 1. We specify generator G as follows.
type lot = tuple(int num; real t; int k);
proc G(chan! lot a):
int i;
dist int d = uniform(0, 2);
while true:
a!(i, time, sample d);
i = i + 1;
delay 4.0
end
end

Chi-8.1

Buffer B0 in front of machine M0 is a standard buffer, in which the lots are all stored in a
single list in order of arrival. Buffer BS in front of machine M1 is a splitting buffer. Buffer
BS receives lots and sends lots of type 0 via channel b to machine M1 and lots of type 1 via
channel c to the buffer in front of machine M2: lots of type 1 bypass machine M1 and are
directly sent to buffer BM. We can specify buffer BS in different ways. Here, we choose to
store all lots in a single list.
proc Bs(chan? lot a; chan! lot b, c):
list lot xs; lot x;
while true:
select
a?x:
xs = xs + [x]
alt
checktype(xs, 0), b!xs[0]:
xs = xs[1:]
alt
checktype(xs, 1), c!xs[0]:
xs = xs[1:]
end
end
end

Chi-8.2

func bool checktype(list lot xs; int n):

if empty(xs):
return false
else:
return xs[0].k == n
end
end
Buffer BS is always willing to receive lots. The function checktype(xs, 0) checks whether
the first lot in list xs is of type 0. If the first lot in xs is of type 0, the function call
checktype(xs, 0) returns true. If the first element in list xs is of type 0, BS tries to send
the head of the list via channel b. If the first element in list xs is of type 1, BS tries to

102

Chapter 8. Other configurations

send the head of the list via channel c. Note that we use the function checktype(xs, 0)
as guard in the selective waiting in BS instead of the guard size(xs) > 0 and xs[0].k =
n. In this way, we avoid taking the head of an empty list 1 , when size(xs) = 0.
Buffer BM is a merging buffer and has one incoming channel, both for lots that have been
processed on machine M1 and a for lots that bypass machine M1. We specify buffer BM as a
standard buffer:
proc B(chan? lot a; chan! lot b):
list lot xs; lot x;
while true:
select
a?x:
xs = xs + [x]
alt
not empty(xs), b!xs[0]:
xs = xs[1:]
end
end
end

Chi-8.3

Buffer BM is always willing to receive lots via channel a, which in the model definition is
connected to Bs and M1.
We specify machine M with a parametric process time.
proc M(chan? lot a; chan! lot b; real t):
lot x;
while true:
a?x; delay t; b!x
end
end

Chi-8.4

Finally, we specify model Bypass.

model Bypass():
chan lot a, b, c, d,
run
G(a),
B(a, b),
M(b,
Bs(c, d, e), M(d,
B(e, f),
M(f,
E(g)
end

e, f, g;

c, 3.0),
e, 6.0),
g, 3.0),

Chi-8.5

After adding output statements to E, simulation yields.

1 In the implementation of Chi3.0, we can use this boolean expression, as the and operator is a conditional
and. This means that for a boolean expression bool-expr-1 and bool-expr-2 the second boolean expression
is only evaluated if the first boolean expression evaluates true.

8.2. Backtracking

10.00
13.00
20.00
23.00
26.00
29.00
32.00
35.00
44.00
50.00

103

E
E
E
E
E
E
E
E
E
E

received
received
received
received
received
received
received
received
received
received

lot
lot
lot
lot
lot
lot
lot
lot
lot
lot

1
0
2
4
3
5
6
7
8
9

We see that lot 1 has passed lot 0. Lot 1 is of type 1 and does not require processing at
machine M1. The flowtime of lot 1 is therefore 6.0 hours. The flowtime of lot 0 is 13.0 hours,
of which 1.0 hour is due to waiting induced by lot 1. We establish a mean throughput of
0.25 lots/hour. The mean flowtime equals 12.5 hours. Part of this flowtime is caused by
the fact that buffer BS only regards the first lot in the buffer. Suppose buffer BS contains
three lots. The first lot is of type 0, the second and third lot are of type 1. If machine M1
is busy, all lots in the buffer have to wait, even when machine M2 is idle. To let buffer Bs
forward lots of type 1, regardless the type of the first lot in line, we use a separate list for
each lot type.
proc Bs(chan? lot a; list(2) chan! lot b):
list(2) list lot xs; lot x;
while true:
select
a?x:
xs[x.k] = xs[x.k] + [x]
alt
unwind j in range(2):
size(xs[j]) > 0, b[j]!xs[j][0]:
xs[j] = xs[j][1:]
end
end
end
end

Chi-8.6

Simulation now yields a mean flowtime of 11.6 hours. However, in this way the mean
flowtime of lots of type 0 increases.

8.2

Backtracking

As an example of a flowline with backtracking, we consider a three-workstation flowline

with possible rework on the middle machine. Machine M1 performs an operation, which has
90% chance of being successful. In buffer BT behind machine M1 lots are tested. Lots that
require rework are sent back to the buffer in front of machine M1. The generator releases
new lots in the line at a rate of 0.25 lot/hour. On every machine, processing a lot takes
3.0 hours. We wish to determine the mean utilisation of the middle machine by simulation.
Figure 8.3 shows a graphical representation of the flowline.

104

Chapter 8. Other configurations

c
G

B0

M0

BM

M1

BT

M2

Figure 8.3: Three-workstation flowline with rework

Lots are of type (nat, real), representing the identification number and the time of release.
Machine M1 sends a tuple of a lot and a bool, in which the bool represents the result of
the operation on machine M1. The last element is true, if the operation on machine M1 was
successful, otherwise it is false. We specify generator G and machine M1 as follows.

type lot = tuple(int num; real t),

bot = tuple(lot lt; bool ok);
proc G(chan! lot a):
int i;
while true:
a!(i, time);
i = i + 1;
delay 4.0
end
end

Chi-8.7

proc M1(chan? lot a; chan! bot b):

dist bool d = bernoulli(0.3);
lot x;
while true:
a?x;
delay 3.5;
b!(x, sample d)
end
end

Testing buffer BT tests the lots that have been processed on machine M1. If the result is
satisfactory, the lot is added to the list of waiting lots for machine M2 , otherwise it is sent
back to merging buffer BM . We specify buffer BT as follows.

8.2. Backtracking

proc BT(chan? bot a; chan! lot b, c):

list lot xs, ys; lot x; bool ok;
while true:
select
a?(x, ok):
if ok:
xs = xs + [x]
else:
ys = ys + [x]
end
alt
not empty(xs), b!xs[0]:
xs = xs[1:]
alt
not empty(ys), c!ys[0]:
ys = ys[1:]
end
end
end

105

Chi-8.8

We use a standard buffer specification (for instance Specification Chi-8.3 ) to represent the
merging buffer. We connect it with the correct channels in accordance with Figure 8.3 by
specifying model Rework as follows.
model Rework():
chan lot a, b, c, d;
chan bot e;
chan lot f, g;
run
G(a),
B(a, b),
M (b, c),
B(c, d),
M1(d, e),
BT(e, f, c), M (f, g),
E(g)
end

Chi-8.9

By simulation, we determine a mean throughput of 0.25 lots/hour and a mean utilisation of

0.833 for machine M1. By hand, we can verify this result. Machine M0 sends 0.25 lots/hour
to machine M1. Of these 0.25 lots/hour 10% need rework. Of this 10%, 10% needs rework
again, and so on. In total, 0.25 + 0.1 0.25 + 0.01 0.25 + ... = 1.111 0.25 = 0.2778 lots/hour
arrive at the buffer for machine M1. This leads to a mean utilisation of 0.2778/0.3333 =
0.833.
We can also deduce this using mass conservation from Chapter 2. Mass conservation on
workstation BM M1 yields:
M0 + 0.1M1 = M1
With M0 = we get:
= 0.9M1 M1 = /0.9

106

Chapter 8. Other configurations

The utilisation of machine M1 then is:

uM1 = M1 t0,M1 = /0.9 t0,M1 = 0.25/0.9 3.0 = 0.833

8.3

Assembly line

Consider an assembly process AP that assembles three different parts p0 , p1 , and p2 . Each
part is stored in a separate buffer. We use separate generator processes to represent the
arrival pattern of the parts (from other areas of the factory). We assume that parts arrive
with a fixed inter-arrival time of 4.0 hours. The assembly process takes one part from each
buffer. After the process has all the required parts, the assembly is started. The assembly
takes 3.0 hours. We wish to investigate the performance by simulation. Figure 8.4 shows a
graphical representation of the assembly line.

G0
G1

a0
a1

b0
b1

AP

b2
G2

a2

Figure 8.4: Line for assembling three parts

We specify process G as follows.

type lot = real,

asy = list(3) real;
proc G(chan! lot a):
while true:
a!time;
delay 4.0
end
end

Chi-8.10

Buffers B are standard buffers storing parts. We specify assembly process AP as follows.

8.3. Assembly line

107

proc AP(list(3) chan? real a; chan! asy b):

asy ps; lot x;
set int cond;
while true:
cond = {0, 1, 2};
select
unwind j in range(3):
j in cond, a[j]?x:
ps[j] = x; cond = cond - {j}
end
end;
b!ps
end
end

Chi-8.11

Process AP uses variable rcv to monitor what parts have been received. If part pi is not
yet received (rcv.i = false), it tries to take a part from the buffer over channel ai . After all
parts have been received, the parts are assembled. The assembly is sent over channel b. We
specify model Assembly as follows.
model Assembly():
list(3) chan real a, b;
chan asy c;
run
unwind j in range(3):
G(a[j]), B(a[j], b[j])
end,
AP(b, c), E(c)
end

Chi-8.12

After changing exit E to log the assemblies that leave the system, simulation yields:
3.000000
7.000000
11.000000
15.000000
19.000000
23.000000
27.000000
31.000000

received
received
received
received
received
received
received
received

assembly
assembly
assembly
assembly
assembly
assembly
assembly
assembly

<(0,0.000000),(0,0.000000),(0,0.000000)>
<(1,4.000000),(1,4.000000),(1,4.000000)>
<(2,8.000000),(2,8.000000),(2,8.000000)>
<(3,12.000000),(3,12.000000),(3,12.000000)>
<(4,16.000000),(4,16.000000),(4,16.000000)>
<(5,20.000000),(5,20.000000),(5,20.000000)>
<(6,24.000000),(6,24.000000),(6,24.000000)>
<(7,28.000000),(7,28.000000),(7,28.000000)>

The i-th element of the assembly represents part pi . We attain a mean throughput of 0.25
assembly/hour. On average, a part is for 3.0 hours in the assembly line. Note that, because
of the deterministic arrival times of the parts, the assembly process receives all parts at the
same time.
We change the arrival times of the parts to become stochastic. We let the lots arrive according to a negative exponential distribution with a mean of 4.0 hours. A single simulation
yields:
3.0000

received assembly <(0,0.00000),(0,0.00000),(0,0.0000)>

108

Chapter 8. Other configurations

6.0000
9.3672
22.2245
25.2245
29.4133
34.1696

received
received
received
received
received
received

assembly
assembly
assembly
assembly
assembly
assembly

<(1,1.78935),(1,2.1233),(1,0.919814)>
<(2,2.84998),(2,4.6275),(2,6.3672)>
<(3,3.54575),(3,19.2245),(3,10.9058)>
<(4,5.14347),(4,19.2386),(4,20.3083)>
<(5,5.19210),(5,19.5010),(5,26.4133)>
<(6,8.51579),(6,23.7321),(6,31.1696)>

We can see that at = 22.2245 for assembly number 3 part p1 (that arrives at = 19.2245)
causes a substantial delay for the assembly process. By simulation, we establish that the
mean throughput is still 0.25 assembly/hour. However, the mean time a part is in the
system increases significantly. (Try! Can you tell from the simulation output whether the
assembly time is still deterministic?)

8.4

Jobshop

The last configuration we discuss, is the jobshop. In a jobshop, we have multiple workstations. Each lot can have a different routing. In principle, a lot could go from every
workstation to any other workstation. As an example, we consider a four workstation jobshop. Each job requires 6 operations. The routing varies per job. The first operation is
always performed on workstation 0, the last operation is always performed on workstation
3. The remaining four operations are evenly distributed over all workstations, for instance,
the chance that the second operation is to be performed on workstation 1 is 25%, just as
the chance that it is to be performed on any other workstation.
W0

W1

W0

W2

W3

W1

W2

(a)

W3

(b)

Figure 8.5: 4-workstation jobshop

Each workstation is connected to every other workstation. Generator G is connected to

workstation 0 and workstation 3 is connected to exit E, see Figure 8.5(a). Each workstation
consists of an infinite buffer and a machine with parametric process time t.
We let the generator release new jobs at a rate of 0.25 lots/hour. We wish to determine
process time t for each workstation, so that the utilisation is 0.80 for each workstation.
It is possible to make a Chi-model of the jobshop in the way depicted in Figure 8.5(a).
However, this model requires a large number of channels and a rather extensive model
specification. Therefore, we propose an alternative model. In this model, we introduce

8.4. Jobshop

109

a central process that connects all workstations. We call this process the transportation
process. The graphical representation of this model is shown in Figure 8.5(b).
A job is of type [nat, [nat], real], representing the identification number, the job routing (e.g.
[0, 2, 1, 2, 2, 3]), and the time of release in the system. We specify generator G as follows:
type job = tuple(int num; list(6) int rout; real t);
proc G(chan! job a):
int i;
dist int d = uniform(0, 4);
list(5) int js;
while true:
js = [0];
for j in range(4):
js = js + [sample d]
end;
js = js + [3];
a!(i, js, time);
i = i + 1;
delay 4.0
end
end

Chi-8.13

The generator generates a routing js, where the first and last element are 0 and 3 and the
middle four elements are each drawn from a separate discrete uniform distribution di2 .
Next, we specify transportation process T . We assume that the transportation takes no
time.
proc T(chan? job a; list(4) chan! job b; chan! job be):
job x;
while true:
a?x;
if empty(x.rout):
be!x
else:
b[x.rout[0]]!x
end
end
end

Chi-8.14

The transportation process receives a job from the generator or from any workstation via
channel a. Then, T checks how many operations are still to be performed. If no operations
are left (len(x.1) = 0), the job is sent to the exit process. If there are operations left, the
job is sent to the corresponding workstation.
Each workstation consists of a machine with deterministic processing time t and a standard
infinite buffer. We specify machine M , bufferB, and process W as follows.
2 Numerical random generators are never really random. They are pseudo-random generators. This
delicate discussion is beyond the scope of this book. As a rule of thumb: use a separate distribution for
every quantity.

110

Chapter 8. Other configurations

proc B(chan? job a; chan! job b):

list job xs; job x;
while true:
select
a?x:
xs = xs + [x]
alt
not empty(xs), b!xs[0]:
xs = xs[1:]
end
end
end
proc M(chan? job a; chan! job b; real t):
job x;
while true:
a?x;
delay t;
x.rout = x.rout[1:];
b!x
end
end

Chi-8.15

proc W(chan? job a, c; real t):

chan job b;
run B(a, b), M(b, c, t)
end
Exit E is always able to receive finished jobs (not shown here). Finally we specify model
Jobshop.
model Jobshop(real t0, t1, t2, t3):
chan job a, be;
list(4) chan job b;
list(4) real t = [t0, t1, t2, t3];
run
G(a),
T(a, b, be),
unwind j in range(4):
W(b[j], a, t[j])
end,
E(be)
end

Chi-8.16

We add output statements to the generator, the machines, and the exit. If we let an
operation on each workstation take 3.0 hours, simulation of a single job yields:
0.00000000000000000
3.00000000000000000
6.00000000000000000

G releases job
M0 processes job
M2 processes job

(0,[0,2,1,0,0,3],0)
(0,[0,2,1,0,0,3],0)
(0,[2,1,0,0,3],0)

8.5. Exercises

9.00000000000000000
12.00000000000000000
15.00000000000000000
18.00000000000000000
18.00000000000000000

111

M1 processes job
M0 processes job
M0 processes job
M3 processes job
E received job

(0,[1,0,0,3],0)
(0,[0,0,3],0)
(0,[0,3],0)
(0,[3],0)
(0,[],0)

Now, we wish to obtain an utilisation of 0.80 for each machine. First, we make an analytical
estimation. On average, workstation W0 receives 1 + 4 P (W0 ) = 1 + 4 0.25 = 2 lots per
job, that is 0.5 lots/hour. W3 also receives 0.5 lots/hour. Workstations W1 and W2 receive
4 0.25 = 1 lot per job, that is 0.25 lots/hour. In order to obtain an utilisation of 0.80, we
estimate that we require a process time of 1.6 hour on machines M0 and M3 and a process
time of 3.2 hours on machines M1 and M2 .
Simulating the jobshop with these process times yields a throughput of 0.25 lots/hour, a
mean flowtime of 26.6 hours and an utilisation of 0.80 on each machine. Note that the mean
2 +t3
flowtime is significantly higher than the mean process time per job (t0 +t3 +4 t0 +t1 +t
=
4
1.6+1.6+42.4 = 12.8 hours). Though the process times are deterministic, the stochastically
generated jobs cause waiting times in the buffers!

Keyword overview
Modelling manufacturing systems
flowline with backtracking (rework)
flowline with bypassing
assembly line
jobshop
bundles

8.5

Analysis of manufacturing systems

Exercises

1. Consider an assembly process that assembles four different parts. Each part has a
different inter-arrival time. We assume that we are dealing with fixed inter-arrival
times. Table 8.2 shows the number of parts required per assembly and the interarrival time of that part.
Each part is stored in a separate buffer. After the assembly processes has collected
all required parts, the assembly commences. Assembling the parts takes 0.8 hours.
The assembly is sent to exit E. Figure 8.6 shows a graphical representation of the
assembly line.
We represent an assembly by an vector of four lists. The i-th list in the assembly
stores the parts of type i. We have the following fragments of specification.

112

Chapter 8. Other configurations

G0

a0

B
b0

G1

a1

b1
AP

G2

a2

G3

a3

part

# in assembly

0
1
2
3

1
2
4
5

b2
b3

Figure 8.6: Assembly line

mean interarrival time [hour]

1.0
0.5
0.25
0.2

Table 8.2: Parts needed in assembly

type prt = real,

asy = list(4) real;
proc G(chan! lot a):
while true:
a!time;
delay 4.0
end
end
func tuple(real t; real m) tAver(real ti_1, ti, mb; int b):
if ti == ti_1:
return (ti, mb)
else:
return (ti, mb * ti_1 / ti + b * (ti - ti_1) / ti)
end
end
proc B(chan? prt a; chan! prt b; int n):
list prt xs; prt x;
real mb; int ti_1;
int i;
ti_1 = time;
while true:
select
a?x:
ti_1, mb = tAver(ti_1, time, mb,size(xs));
xs= xs ++ [x]
alt
not empty(xs), b!xs[0]:
ti_1, mb = tAver(ti_1, time, mb,size(xs));
xs = xs[1:];
i = i + 1;
if i mod 10000 == 0:
write("%10.2f\tB\t%5d\tmb =\t%10.2f\n", time, n, mb)
end
end
end
end

Chi-8.17

8.5. Exercises

113

proc AP(list(4) chan? prt a; chan! asy b; real t0):

list(4) int pqr, rqr = [1, 2, 4, 5];
list(4) prt ptc;
prt p;
while true:
while pqr != rqr:
select
unwind j in range(4):
pqr[j] < rqr[j], a[j]?p:
pqr[j] = pqr[j] + 1;
ptc[j] = ptc[j] + [p]
end
end
end;
delay t0;
b!ptc;
pqr = [0, 0, 0, 0];
ptc = <pts>[ ]
end
end
Chi-8.18
proc E(chan? asy a):
asy x;
while true:
a?x
end
end
model Assembly():
list(4) chan prt a, b;
chan asy c;
list(4) real t = [1.0, 0.5, 0.25, 0.2];
run
unwind j in range(4):
G(a[j], t[j]),
B(a[j], b[j])
end,
AP(...),
E(c)
end

(a) By hand, estimate the utilisation of the assembly process.

(b) Explain, whether the arrival rate of the parts is balanced, that is, does the
assembly process need to wait for the same specific part over and over again?
(c) Finish the fragments of specification into a complete model.
(d) By simulation, determine the average buffer level in each part buffer.
(e) Figure 8.7 shows part of the buffer levels over time.
Finish the diagrams for each part buffer. Verify the simulation results.
(f) Change the arrival rate of the parts to become stochastic. Even after simulating
for a substantial long time, we see that one or more buffer levels keep increasing.
Can you explain this?
2. We have a manufacturing line that processes two lot types. The manufacturing line
consists of 3 workstations. The first workstation has three identical machines M0 .
The second and third workstation each have one machine. Each workstation has an
infinite buffer. Lots of type 0 require processing on all three workstation. Lots of type

114

Chapter 8. Other configurations

B3

B2

B1

B0
1.0

2.0

3.0 t

Figure 8.7: Number of parts in buffer over time

1 require processing on M0 and M2 only. Processing a lot on machine M0 , M1, and M2

takes 9.0, 5.0, and 3.0 hours respectively. Figure 8.8 shows a graphical representation
of this manufacturing line.
b0
G

B0

b1

M0
M0

b2

c0
c1
c2

B1

M1

B2

M2

M0

Figure 8.8: Manufacturing line

Lots are of type (nat, nat, real), representing the identification number, the lot type,
and the time of release in the line.
(a) By hand, estimate the utilisation for each machine if we release lots in the line
at a rate of 0.25 lots/hour.
(b) Build a model of this manufacturing line. Use fragments of the specifications
discussed in Sections 8.1, 8.2, and 7.1.
(c) By simulation, determine the mean throughput and flowtime of this line for a
release rate of 0.25 lots/hour.
(d) By hand, calculate the mean wip level of this line.
(e) By simulation, determine the mean utilisation of each machine. Compare your
results with your estimations from part 2a.
3. In re-entrant flowlines, a lot runs through the flowline multiple times. In this way,
we can perform a sequence of process steps more than once on a single lot. This is
often encountered in the semi-conductor industry, where wafers are processed in layers.
Here, we study the behaviour of a simple re-entrant flowline. The flowline has two
machines and three buffers. The last buffer checks whether the lot needs to re-enter
the flowline or whether it is finished. Figure 8.9 shows a graphical representation of
the re-entrant flowline.

8.5. Exercises

115

B0

M0

B1

M1

B2

Figure 8.9: Re-entrant flowline

Processing takes 0.2 hours on each machine. There are three lot types. The lot types
need 1,2, and 3 loops through the flowline respectively. The generator releases lots at
a fixed arrival rate. On average, each lot type is generated in the same quantity.
(a) By hand, estimate the release rate for which the utilisation of both machines is
0.8.
(b) Let lots be of type (nat, real, nat), where the elements represent the identification
number, the time of release, and the remaining number of loops through the line.
Make a model of this manufacturing line. Below, the specification of buffer B2
is shown.
proc B2(chan? lot a; chan! lot b, c):
list lot xs, ys; lot x;
while true:
select
a?x:
x.rl = x.rl - 1;
if x.rl > 0:
xs = xs + [x]
else:
ys = ys + [x]
Chi-8.19
end
alt
not empty(xs), b!xs[0]:
xs = xs[1:]
alt
not empty(ys), b!ys[0]:
ys = ys[1:]
end
end
end
(c) By simulation, determine the mean utilisation per machine for different release
rates. For which release rate is the utilisation 0.8? Does this correspond to your
estimation?

116

Chapter 8. Other configurations

Chapter 9

Extending machine models

In the previous chapters we have encountered several models of machines. On some occasions, we assumed fixed (constant) process times. On others, we treated process times as
being stochastic. We then proposed a distribution to describe the process time. In this
chapter we address in more detail how to accurately model machines. We will discuss the
following aspects:
using a distribution to represent process time,
modelling the failure and repair of machines, and
using effective process time to represent all sources of variability.

9.1

Representing process time by a distribution

In this section we present a brief stepwise approach to represent process times by distributions. This approach can also, at least to a large extent, be applied to describe the arrival
pattern of raw materials or demand pattern of finished goods. One of the main tasks is
to find and properly fit a distribution to the measured process time data. There is a large
quantity of literature available on fitting a distribution to a set of data. We present the main
techniques involved in fitting distributions, without addressing the statistical background
in detail. More on the statistical background can be found in [LK00, MR02]. We discuss
using distributions to represent process times in six steps.
1. Collect data.
2. Determine whether the observations are independent and identically distributed.
3. Plot a histogram and calculate basic statistical quantities.
4. Choose candidate distributions.
5. Determine the distribution parameters by fitting the chosen distribution to the sample
data and determine how well the fitted distribution fits the data.

117

118

Chapter 9. Extending machine models

6. Implementing the distribution in your Chi-model.

Many packages support statistical analysis of data, amongst others Matlab [Mat03], SAS
[SAS03], SPSS [SPS03], and Statgraphics [Man03]. In this chapter, we use Statgraphics.
Statgraphics is more accessible than the more extensive packages SAS and SPSS, while
having about the same functionality with respect to fitting distributions.

Step 1: Collecting data

The first step is collecting process time data. In modern high-tech fabs, times at which a
machine starts and finishes processing is often recorded in log files. In many fabs we still
need to collect data manually, by means of clocking. There is much that can go wrong in
collecting data. One may think of censored data (the operator reports only fast (or only
slow) jobs or one may think of observations that are not independent (see next step). More
on incorrect data can be found in [BFS87].
Another delicate issue regards the number of observations that is needed to reasonably fit a
distribution. No hard guidelines for determining this number are available. In general, more
observations allow more accurate fitting of the distribution and more profound judgements
regarding the validity of the fitted distribution.
Throughout this section we consider the sample data in Table 9.1 adopted from [Lee99].
105.84
28.92 98.64 55.56 128.04 45.60
67.80 105.12 48.48 51.84 173.40 51.96
54.12
68.64 93.12 68.88
84.12 68.64
41.52 127.92 42.12 17.88
33.00
Table 9.1: Measured process times in seconds

Step 2: Checking data independency

If we wish to fit a distribution to process time data, the individual observations need to
be independent and identically distributed (idd ), that is, the observations are independent
samples of a single distribution. Situations in which the idd assumption is not valid include:
A new operator has been hired and the process time strongly depends on the operator
experience, that is the process time is subject to a steep learning curve. Process time
measurements are negatively correlated.
The machine processes in shifts, and the machine has to warm up in the beginning of
each shift. Process time measurements in the beginning of each shift are negatively
correlated.
We measure the flowtime of lots at a single workstation. If a lot has a large flowtime, the next lot is likely to have a large flowtime too. The flowtimes are positively
correlated.

9.1. Representing process time by a distribution

119

Many processes are non-stationary, that is, distributions change over time. For assessing
time-dependency we can plot the observations against time or against sequence number.
When we observe a pattern, we can use simple regression to analyze the degree of timedependency. Figure 9.1 shows the time-series plot of the data from Table 9.1. The data
shows no clear pattern.

process time

[s]

180
150
120
90
60
30
0
0

12

16
20
24
observation number

Figure 9.1: Time-series plot

From Figure 9.1, we assume that the observations are time-independent.

We present two graphical approaches to assess (auto)correlation1 between subsequent observations. The first one is making a scatter diagram for observations X1 , X2 , ..., Xn in which
we plot the pairs (Xi , Xi+1 ). If the observations are independent, the points are scattered
randomly. The nature of the random scattering depends on the underlying distribution.
If correlation exists, the points lie on a line, in case of positive correlation on an inclining
line and in case of negative correlation on a declining line. The second graphical approach
is the correlation plot. The correlation plot is a graph of the estimated autocorrelations
j for j = 1, 2, ..., l. The estimated autocorrelation j is an estimation of the true correlation j between two observations that are j observations apart in time. If the observations
X1 , X2 , ...Xn are independent, correlations j are zero. When the estimations j deviate
significantly from zero, this is strong evidence that the observations are not independent.
For calculating the autocorrelation j we refer to [MR02]. Here, we used Statgraphics to
draw a scatter diagram and calculate the autocorrelation, see Figure 9.2.
Figure 9.2(a) shows the scatter diagram and Figure 9.2(b) shows the correlation plot for the
observations from Table 9.1. Figure 9.2(b) shows the estimated autocorrelation for different
lags ({0,1,2,...,7}). Also shown are 95.0% probability limits around 0.0. If the probability
limits at a particular lag do not contain the estimated coefficient, there is a statistically
significant correlation at that lag at the 95.0% confidence level. In this case, there is no
evidence that autocorrelation exists.
Apart from these graphical methods, several statistical tests exist. For these, one is referred
to [LK00, MR02]. Judging from the diagrams in Figure 9.2, we assume that the observations
are independent and identically distributed.
1 Statistically, we prefer speaking of autocorrelation, since the observations stem from the same source
distribution.

120

Chapter 9. Extending machine models

Xi +1
180
Estimated Autocorrelations for process time
150

120

0.6

90

j 0.2
-0.2

60

-0.6

30
-1

0
0

30

60

90

120 150 180

Xi

lag j

Figure 9.2: (a) Scatter diagram, (b) Estimated autocorrelation

Step 3: Basic statistical analysis

The next step is to perform a basic statistical analysis. We start by plotting a histogram
of the data. Hereto, we divide the range of data into a number of intervals, and count the
number of observations within each interval. The number of intervals can strongly influence
the appearance of the histogram. A rule of thumb is to take a number of intervals equal to
the square root of the number of observations. Figure 9.3 shows the histogram for the data
of Table 9.1, drawn by Statgraphics.
8

6
frequency

[#]
4

0
0

40

80
120
process time

160
[s]

Figure 9.3: Histogram of process time measurements

Please note that in histograms, the upper limit of each interval is excluded from that interval.
In this case, the second interval includes observations in the range 20 t < 40.

9.1. Representing process time by a distribution

121

Next, we calculate some basic statistic quantities, like the range R, mean x, standard
deviation s, and standardized skewness2 . They are calculated as follows:
n

R
s2

= max(x) min(x)
=

1 Pn
(xi x)2
n 1 i=1

x=

1X
xi
n i=1

n
X
1
(xi x)3
(n 1)s3 i=1

The mean x and standard deviation s are commonly used statistics. Skewness is an
indication for the distribution symmetry. A skewness close to 0 indicates a symmetrical
distribution. Positive values of skewness indicate that the upper tail of the curve is longer
than the lower tail; negative values indicate a longer lower tail. Standardized skewness
values outside the range of 2 to +2 indicate significant deviation from normality.
For the data of Table 9.1 we determine the following:
R = 155.52,

x = 72.22,

s = 37.5,

= 1.97

Thus, we are dealing with a distribution that has a mean of 72.22 seconds, a standard
deviation of 37.5 seconds, and a small positive skewness, indicating a small deviation from
symmetry, with the upper tail being a little longer.

Step 4: Selecting candidate distributions

Many different types of theoretical distributions exist. Here, we discuss the distributions
listed in Table 9.2. These distributions are some of the most often encountered distributions
in modelling manufacturing systems.
discrete distributions
uniform
bernoulli

continuous distributions
uniform
triangular
normal
exponential/gamma
lognormal
Table 9.2: Commonly used distributions

In the sequel, we present the typical shape and the parameters of each distribution, along
with its range of common application. This overview is based on [LK00]. Additionally,
we discuss how the distribution can be used in Chi. Next to its name and parameters, we
indicate the type of parameters and the type of the distribution samples. For example dist
bool bernoulli(real p) indicates that the Bernoulli distribution has one parameter of
type real and that samples from the distribution are of type bool.
Other distributions include the binomial, the Erlang, the Weibull, the beta, the Pearson,
the Johnson, and the log-logistic distribution. More on these distributions can be found in
2 In literature, more than one formula is encountered to calculate skewness. Statgraphics calculates the
so-called standardized skewness, which is calculated as presented here.

122

Chapter 9. Extending machine models

Bernoulli
1

Discrete distribution that has two possible outcomes: false and true.

bernoulli(real p)

0.8

0.6

Function
dist bool bernoulli(real p)

bernoulli(0.69)
0.4

Parameters
p: Chance on sampling true

0.2

0
false

true

mean
variance

p
p(1 p)

Discrete uniform
0.25

Discrete distribution that has several equally

likely outcomes, the numbers {a, a + 1, a +
2, . . . , b2, b1}. Note that b is not included.

uniform(1, 6)

uniform(int a, b)

0.2

0.15

Function
dist int uniform(int a, b)

0.1

0.05

0
1

Parameters
a: Lower bound
b: Upper bound (exclusive!)
mean
(a + b 1)/2
variance ((b a)2 1)/12

Continuous uniform
0.25

Continuous distribution with equal chance of

sampling each value in the range [a, b). Note
that b is not included.

0.2

uniform(real a, b)

uniform(1.0, 6.0)
0.15

Function
dist real uniform(real a, b)

0.1

0.05

0
1

4
x

Parameters
a: Lower bound
b: Upper bound (exclusive!)
mean
(a + b)/2
variance (b a)2 /12

[LK00]. Next to using these theoretical distributions we can also use empirical or tracedriven models. A trace-driven model simply reproduces the collected data (in the same
order). In case of an empirical model, we generate random values from the collected data.
We can draw randomly from the set of data or we can use some sort of interpolation. More
on empirical distributions can be found in [Ban98].

9.1. Representing process time by a distribution

123

Triangular
0.9

Continuous distribution producing samples in

the range [a, c] (a < b < c).

0.8

triangle(real a, b, c)

0.7

Function
dist real triangle(real a, b, c)

0.6
triangle(1.0, 2.0, 4.0)
0.5
0.4
0.3
0.2
0.1
0
0

Parameters
a: Lower bound
b: Mode
c: Upper bound
mean
(a + b + c)/3
variance (a2 + c2 + b2 a b a c b c)/18

Normal
0.5

normal(real m, v2)

0.4

Bell-shaped curve around a with width 2 b.

0.3

Function
dist real normal(real a, b)

normal(3.0, 1.0)
0.2
normal(5.0, 2.0)

0.1

0
0

10

Parameters
a: Mean value
b: Variance
mean
a
variance b

Gamma
0.5

gamma(1.0, 2.0)

gamma(real a, b)

0.4

Distribution which has either a decreasing probability function, or a peak.

0.3

Function
dist real gamma(real a, b)

gamma(6.0, 0.5)

0.2

gamma(3.0, 2.0)
0.1

gamma(6.0, 2.0)

0
0

10
x

12

14

16

18

20

Parameters
a: Shape parameter
b: Scale parameter
mean
ab
variance ab2

Instead of fitting the data to all possible distributions, we first select possible candidate
distributions that reasonably fit the collected data. Often, we can rule out distributions
based on prior knowledge. For example, normal distributions are unlikely to be a good
candidate for process times, since a sample from a normal distribution can be negative.
Next, we compare the histogram from step 3 to the typical shapes of the distribution and

124

Chapter 9. Extending machine models

Lognormal
1.8
1.6

Takes on shapes similar to gamma(p, q) with

p > 1. It can have large peaks close to x = 0.

lognormal(real m, v2)

1.4
1.2

lognormal(0.0, 0.25)

Function
dist real lognormal(real a, b)

1
0.8
0.6
0.4

Parameters
a: Shape parameter
b: Scale parameter
mean
ea+b/2
variance e2a+b (eb 1)

lognormal(0.0, 0.5)

0.2

lognormal(0.0, 1.0)

0
0

0.5

1.5

2.5

3.5

select possible candidates. Finally, we use Statgraphics (or another statistical program)
to draw probability plots. In a probability plot we use specific probability paper for each
distribution. We order the observations in increasing order. We then plot the ordered list
of observations against there cumulative frequency. Figure 9.4 shows the probability plot
for the data from Table 9.1 for a uniform, normal, exponential, and lognormal distribution.

80
60
40
20

(a) Uniform Probability Plot

cum. frequency [%]

30

99.9
99
95
80
50
20
5
1
0.1

60

90

120

150

30

60

90

120

150

process time

99.5
99
95
90
80
70
50
0.1

180

(b) Normal Probability Plot

0

cum. frequency [%]

99.9

180

[s]

(c) Exponential Probability Plot

0

cum. frequency [%]

cum. frequency [%]

100

99.9
99
95
80
50
20
5
1
0.1

100

200

300

400

(d) Lognormal Probability Plot

10

100

process time

1000

[s]

Figure 9.4: Probability plot for (a) uniform, (b) normal, (c) exponential, and (d) lognormal
distribution

Note that the x- and y-axis can be different for each distribution type. If the plotted points
fall approximately along a straight line, the corresponding distribution adequately describes
the data. In this example, we see that the exponential and lognormal distribution are good
candidates.

9.1. Representing process time by a distribution

125

Step 5: Estimation of parameters

The next step is to estimate the parameters of the hypothesized distribution. Most statistical
packages have built-in functionality to estimate these parameters. For more statistical
background on how these parameters are estimated, we refer to [LK00].
We use Statgraphics to estimate the parameters for an exponential, a lognormal, and a
gamma distribution. The fitted distributions are displayed in Figure 9.5, along with a
histogram of the original observations.
8

Lognormal
= 73.16 s
= 41.97 s

2 = 0.74

frequency

frequency

P = 0.86

Exponential
= 72.22 s

2 = 14.27
P = 0.006

2
0

0
0

40

80

120

160

process time

200

40

[s]

80

120

160

process time

200

[s]

frequency

8
6

Gamma
a = 4.025
b = 0.056

2 = 2.304
P = 0.512

2
0
0

40

80

120

160

process time

200

[s]

Figure 9.5: Fitted distributions

To assess the goodness-of-fit, two basic approaches exist. In the first approach we compare
the frequency plot of the fitted distribution with the frequency plot from the observations.
This can be done graphically by plotting the frequency plot of the fitted distribution together
with the histogram of the original observations, as has been done in Figure 9.5.
Quantitatively comparing the histogram with the fitted distribution can be done using a
2 -test [Pea00]. In this test we divide the range of the fitted distribution into k adjacent
intervals. Then, we determine the expected frequency Ej of samples in the j-th interval if
we were sampling from the fitted distribution. The -square test compares this expected
frequency Ej with the actually measured frequency Oj . The test statistic is given by:
2 =

k
X
(Oj Ej )2
Ej
j=1

(9.1)

For small values of 2 , the fitted distribution adequately represents the data. Statgraphics
calculates the P -value associated with the calculated value of 2 , indicating the chance that

126

Chapter 9. Extending machine models

the set of observations are a set of samples from the hypothesized distribution. If P is small
enough, we reject the hypothesis that the observations are from the fitted distributions with
(1 P ) confidence. Generally, if P is greater than 0.10, there is insufficient evidence for
rejecting the hypothesized distribution. In Figure 9.5, the values of 2 and the associated
P -value are calculated for each distribution. Based on the 2 -test, we reject the exponential
distribution with 99.4% confidence. We do not reject the gamma and lognormal distribution.
The 2 values indicate that the lognormal distribution is a better fit.
Please note, that the 2 -test results are strongly influenced by how we choose the intervals.
Different intervals can yield totally different results. [LK00] advises intervals that each
include at least 5 observations.
The second approach to assess goodness-of-fit is comparing the cumulative probability of
the fitted distribution with the cumulative frequency of the original observations. This can
be done graphically or by using a test statistic. Figure 9.6 shows the cumulative probability
for the fitted distributions, along with the cumulative frequency of the observations.
1
cumulative probability

cumulative probability

1
0.8
0.6

Lognormal

0.4
Dmax = 0.09
PKS 0.10

0.2
0
0

30

60

90

120

150

cumulative probability

process time

180

[s]

0.8
0.6

Exponential

0.4

0.2

Dmax = 0.30
PKS 0.01

0
0

30

60

90

120

150

process time

180

[s]

1
0.8
0.6
Gamma

0.4

Dmax = 0.12
PKS 0.10

0.2
0
0

30

60

90

120

150

process time

180

[s]

Figure 9.6: Cumulative probability plot with cumulative frequency of observations

A commonly used test is the Kolmogorov-Smirnov test [Kol33, Smi33]. This test calculates
the maximum vertical distance Dmax of the observations to the cumulative frequency diagram, as indicated for the exponential distribution in Figure 9.6. Using Statgraphics we
have determined the maximum vertical distance Dmax for each distribution. The maximum
distance that is allowed, depends on the number of observations and the specific distribution
type. Statgraphics calculates a PKS -value that indicates the confidence with which the fitted
distribution is adequate. If PKS is sufficiently small, we reject the hypothesized distribution
with (1PKS ) confidence. Generally, if PKS is greater than 0.10, there is insufficient evidence
for rejecting the hypothesized distribution. The values for Dmax and PKS are displayed in
Figure 9.6. Based on the Kolmogorov-Smirnov test, we reject the exponential distribution
with at least 99% confidence. We can not reject the lognormal and gamma distribution. As
for the 2 -test, we again see that the lognormal distribution is the best fit.

9.2. Machine failure and repair

127

For the Kolmogorov-Smirnov test we do not need to define intervals and we can use the test
for any sample size n. However, unlike the 2 -test, which can be used for testing against
both continuous and discrete distributions, the Kolmogorov-Smirnov test is only appropriate
for testing data against a continuous distribution.
Other goodness-of-fit tests exist, such as the Anderson-Darling, Kuiper, and Watson test.
More details on goodness-of-fit tests can be found in [Ste74].

Step 6: Implementation
The final step is to implement the fitted distribution to represent the process time in the Chispecification. In many cases, the parameters calculated by statistical packages correspond
to the parameters with which the distribution is instantiated in Chi. In some cases, the
calculated parameters have to be converted. In case of the lognormal distribution fitted
to the data of Table 9.1, Statgraphics calculates mean = 73.16 and standard deviation
= 41.97, see Figure 9.5. However, in Chi we require the shape and scale parameters a en
b, see the distribution overview from Step 4. Below, we present the conversion from and
to shape and scale parameters a and b for a gamma and a lognormal distribution.
2
2

gamma(a,b)

a=

lognormal(a,b)

a = ln 0.5 ln (1 +

b=

(9.2)

2
2
)
b
=
ln
(1
+
)
2
2

For the fitted lognormal distribution, the resulting machine model is shown below.
proc M(chan? lot inp; chan! lot outp):
real t0 = 73.16,
t02 = t0 * t0,
s0 = 41.97,
s02 = s0 * s0,
a = ln(t0) - ln(1 + s02 / t02) / 2,
b = ln(1 + s02 / t02);
dist real d = lognormal(a, b);
lot x;
while true:
inp?x; delay sample d; outp!x
end
end

9.2

Chi-9.1

Machine failure and repair

In the previous section, we discussed ways to use a distribution to represent the process
time of a lot. In practice, machines may break down and require repair, causing a lot to
longer occupy the machine. In this section we discuss ways to model machine failure and
repair. We distinguish two types of failure [Ger94]:
Operation Independent Failure (OIF) and

128

Chapter 9. Extending machine models

Operation Dependent Failure (ODF).

In case of OIF, the machine breaks down after being up for a certain amount of time,
independent of the amount of time that the machine has been processing lots. By the
amount of time a machine is up, we mean the time that a machine is turned on and has
not broken down. When a machine is up and not processing, the machine is blocked (the
subsequent buffer is full), the machine is starving (the preceding buffer is empty), or the
machine is idle due to for example the control strategy. In case of ODF, the machine breaks
down after processing for a certain amount of time, or after processing a certain number of
lots. In Figure 9.7, the different types of failure are schematically depicted.
0

OIF

10

1 2

15

3 4

t
lot number

30

tfu

6
tfu

tfp
0

25

3 4 5

tfu

ODF

20

tfp
1

3
Nf

5 6

Nf

Figure 9.7: Operation Independent Failure (OIF) and Operation Dependent Failure (ODF)

In the upper case, the machine breaks down after being up for a fixed amount of time tfu
(OIF). In the middle case, the machine breaks down after processing for a fixed amount of
time tfp (ODF). In the bottom case, the machine breaks down after processing Nf = 4 lots
(ODF).
Adopted from [Lew95], we define the following quantities for characterizing failure and repair
behaviour:
Actual time between failure (ATBF): the time between two subsequent failures, denoted by tfa .
Processing time between failure (PTBF): the amount of processing time between two
subsequent failures, denoted by tfp .
Up time between failure (UTBF): the amount of up time between two subsequent
failures, denoted by tfu .
Cycles between failure (CBF): number of cycles (number of lots) processed between
two subsequent failures, denoted by Nf .
Time to repair (TTR): amount of time required for repair, denoted by tr .
These quantities are schematically depicted in the state diagram of Figure 9.8.
All quantities can be stochastic. In that case, we have mean process time between failure
tfp , mean uptime between failure tfu , mean number of cycles between failure N f and mean
time to repair tr .

9.2. Machine failure and repair

0
tfa
tfp
tfu
Nf
tr

129

10

1 2

15

20

3 4 5

25

5
2

8
3

11
5

t
lot number

30

1
3

Figure 9.8: Quantities characterizing failure

In the sequel we present ways to model machine failure. We consider a single workstation,
consisting of a infinite buffer and a machine. The process time t0 of the machine is distributed according to a gamma distribution with mean 3.0 and coefficient of variation of
1.0. We consider four cases of different failure behaviour.

1. The machine fails after processing a certain number of lots.

2. The machine fails after processing for a certain amount of time. After the machine is
repaired, the machine continues processing where it left off at the moment the machine
broke down.

3. The machine fails after processing for a certain amount of time. After repair, the
machine starts processing the lot from the beginning.

4. The machine fails after being up for a certain amount of time. After the machine is
repaired, the machine continues processing where it left off at the moment the machine
broke down.

In all cases, we let lots arrive at the buffer with a fixed arrival rate of 0.25 lots/hour. The
exit can always receive lots.

1. Failure after processing a number of lots

On average, the machine fails every 100/3 = 33.3th lot. The number of lots between
two failures is exponentially distributed and is a whole number. The repair time is also
exponentially distributed with an average of 10.0 hours. We use a standard specification of
generator G, buffer B, and exit E. The machine is specified as follows.

130

Chapter 9. Extending machine models

proc M(chan? lot a; chan! lot b):

dist real dt0 = gamma(1.0 / (1.0 * 1.0), 3.0 / (1.0 * 1.0)),
dNf = exponential(100.0 / 3.0),
dtr = exponential( 10.0);
int Nf;
lot x;
Nf = round(sample dNf);
while true:
if Nf == 0:
Chi-9.2
delay sample dtr;
Nf = round(sample dNf)
else:
a?x;
delay sample dt0;
b!x;
Nf = Nf - 1
end
end
end
The machine specification contains three distributions. Distribution dt0 represents the process time, and distributions dNf and dtr represent the number of lots between failure and
the repair time respectively. We use variable Nf to record the number of lots that can be
processed before failure. We draw Nf from an exponential distribution. Since the exponential distribution yields real samples, and we want Nf to be a whole number, we round the
sample from the distribution to its nearest integer3 If the number of lots before failure Nf
is zero, the machine is repaired, which takes dtr hours, and a new value for Nf is drawn.
If the number of lots before failure Nf is greater than 0, the machine is able to receive a lot,
process it, and send it to the next process. The number of lots before failure is decreased
by 1.
By simulation we establish a mean actual time between failure of 135 hours and a downtime
percentage of 7.4%. Adding the machine failure, causes the mean flowtime of a lot to increase
from 6.5 hours to 12.5 hours. This increase is only for a small part ascribed to the repair
time (on average every 33th lot stays for 3.0 + 10.0 = 13.0 hours on the machine). The
largest part of the flow time increase is due to the increase in variability on the occupation
time of a lot on the machine, which increases the average waiting time in the buffer.

2. Failure after processing for an amount of time

On average, the machine breaks down after processing for 100.0 hours. On average, repair
takes 10.0 hours. The processing time between failure and the time to repair are both
exponentially distributed. We use standard specifications for the generator G, buffer B,
and exit E. The machine with failure is specified as follows.
3 Since all samples taken from a exponential distribution are greater than zero, the rounded sample is a
natural. However, the built-in round function yields an integer. We use the built-in i2n function to convert
the integer into a natural.

9.2. Machine failure and repair

131

proc M(chan? lot a; chan! lot b):

dist real dt0 = gamma(1.0 / (1.0 * 1.0), 3.0 / (1.0 * 1.0)),
dtf = exponential(100.0),
dtr = exponential( 10.0);
timer t0, tf;
real remaining_f, remaining_t0;
lot x;
remaining_f = sample dtf;
while true:
a?x;
tf = timer(remaining_f);
t0 = timer(sample dt0);
while true:
select
ready(t0): break
alt
ready(tf):
remaining_t0 = real(t0);
delay sample dtr;
t0 = timer(remaining_t0);
tf = timer(sample dtf)
end
end;
remaining_f = real(tf);
b!x
end
end

Chi-9.3

The distributions dt0 , dtfp , dtr represent the process time, the process time between failure,
and the repair time respectively. The machine can only fail while processing. After receiving
a lot, we take a sample from the process time distribution for that lot. The variable t0
represents the remaining process time for the lot in process. The variable tfp represents the
remaining process time till failure. As long as t0 is greater than 0.0, the machine finishes
processing (t0 ) or fails (tfp ), depending on what happens first. If the machine finishes
processing, the remaining processing time till failure is decreased by t0 and the remaining
process time is nulled. If the machine fails, the remaining process time is decreased by tfp ,
the machine is repaired, taking dtr , and a new sample from distribution dtfp is drawn to
determine the next moment of failure. Subsequently, the processing in continued. After the
lot is finished processing (t0 = 0.0) the lot is send via channel b.
By simulation, we determine that the machine is down for 7.5% of the time. The mean
actual time between failure is 132 hours. Note that, though the mean process time between
failure is 100.0 hours and the mean repair time is 10.0 hours, the percentage down is not
10% and the mean actual time between failure is not 110 hours. Note that this caused by
the machines being idle some of the time4 !
Without machine failure, the mean flowtime is 6.5 hours. With machine failure, the mean
flowtime is 14.2 hours. Again the increase in variability is the main contributor to the
4 In case of exponential failure and repair times, we can relatively easy determine the flowtime and
percentage downtime by probability calculations. For such calculations, one is referred to [Lew95] or [Ros00]

132

Chapter 9. Extending machine models

increase in flowtime.

3. Failure after processing for an amount of time, restart processing

We reconsider the machine from the previous example. However, in this case the machine
has to restart processing from the beginning after repair, the specification becomes:
proc M(chan? lot a; chan! lot b):
dist real dt0 = gamma(1.0 / (1.0 * 1.0), 3.0 / (1.0 * 1.0)),
dtf = exponential(100.0),
dtr = exponential( 10.0);
timer t0, tf;
real remaining_f, pt0;
lot x;
remaining_f = sample dtf;
while true:
a?x;
pt0 = sample dt0;
tf = timer(remaining_f);
while true:
t0 = timer(pt0);
select
ready(t0): break
alt
ready(tf):
delay sample dtr;
tf = timer(sample dtf)
end
end;
remaining_f = real(tf);
b!x
end
end

Chi-9.4

By simulation, we determine a slight increase in downtime: 7.8%, a slight decrease in the

mean actual time between failure: 129 hours and a substantial increase in mean flowtime:
20.2 hours. Because, the machine has to restart processing after repair, the percentage
idle time slightly decreases, causing a slight increase in the number of failures per time
unit, that is, a slight increase in percentage downtime and a slight decrease in mean actual
time between failure. This all together causes a small increase in utilization, leading to a
substantial increase in mean flowtime.

4. Failure after being up for an amount of time

In this last case, we consider a machine that fails after being up for an amount of time. On
average, the machine fails after being up for 100.0 hours. Repair, on average, takes 10.0
hours. The mean uptime between failure and the repair time are exponentially distributed.
The machine can break down while waiting for a lot, while processing a lot, and while trying
to send the lot to the next process.
A possible specification of machine M is:

9.2. Machine failure and repair

133

proc M(chan? lot a; chan! lot b):

dist real dt0 = gamma(1.0 / (1.0 * 1.0), 3.0 / (1.0 * 1.0)),
dtf = exponential(100.0),
dtr = exponential( 10.0);
timer tf, t0;
real remaining_t0;
lot x;
tf = timer(sample dtf);
while true:
# Receive a product
while true:
select
ready(tf):
delay sample dtr;
tf = timer(sample dtf)
alt
a?x:
break
end
end;
# Process it
t0 = timer(sample dt0);
while true:
select
ready(tf):
remaining_t0 = real(t0);
delay sample dtr;
tf = timer(sample dtf);
t0 = timer(remaining_t0)
alt
ready(t0):
break
end
end
# Send a product
while true:
select
ready(tf):
delay sample dtr;
tf = timer(sample dtf)
alt
b!x:
break
end
end;
end
end

Chi-9.5

134

Chapter 9. Extending machine models

We use the variables f and 0 to represent the time instants at which the machine fails or
finishes processing a lot respectively. As long as the machine is idle, the machine is waiting
for a lot via channel a. While waiting for a lot, the machine may fail. Upon failure, the
machine is repaired, and time instant f of the next failure is determined. After a lot has
been received, time instant 0 is determined at which the machine is finished processing the
lot. As long as the lot is not finished processing (f inshd = false), the machine may break
down. Upon failure, the machine is repaired and the time instant at which the lot is finished
is increased by the repair time (0 := 0 + tr ). After the lot is finished, the machine tries to
send the lot to the next process. While trying to send the lot, the machine may again fail.

Chapter 10

Case: (Part of ) a waferfab

In previous chapters we have presented the main building blocks for modelling and analysis
of manufacturing systems. With these building blocks we can practically model any manufacturing system. In this and the subsequent chapter we illustrate the use of Chi by means
of two industrial cases. Both cases involve the manufacturing of chips. Most electronic
devices contain one or more integrated circuits (chips). Figure 10.1(a) shows a PC ROM
chip. Chips are produced on wafers. Figure 10.1(b) shows such a wafer.

Figure 10.1: (a) ROM chip in PC, (b) Multiple chips (dies) on a wafer

The first case discusses a strongly simplified, but highly illustrative, analysis of a small part
of a wafer fab. The second case discusses a more detailed and extensive model of a larger
part of a wafer fab.
The cases are distilled from actual research conducted by the Systems Engineering group
at Philips and ASML. For reasons of confidentiality, numbers are fictitious.

135

136

10.1

Chapter 10. Case: (Part of) a waferfab

The manufacturing of Integrated Circuits (ICs)

Manufacturing a chip is a process involving hundreds of steps and requiring from a few days
to up to three months of processing time to complete. Figure 10.2 shows the main steps in
making chips.
Preparation of silicon
wafers

Die packaging

single die
Wafer fabrication

Figure 10.2: Main steps in chip making

Raw silicon is first extracted from common sand. A series of chemical steps refine it until
the purity level reaches 99.9999999 percent [Int04]. Silicon ingots are then grown from
the purified molten silicon and sliced into round wafers ranging from 150mm to 300mm.
On this wafer we create multiple chips (dies). Each die is made up of many layers, each
of which contains a detailed pattern. The wafer manufacturing process consists of forming
this sequence of layers very precisely. The wafer fabrication is discussed in more detail
in the next section. When all layers of the wafer are ready, the microscopic circuitry of
each microprocessor is tested. Then the wafer is cut with a diamond saw, separating the
dies. Each die is then inserted into a protective package which allows it to connect to other
devices. In the chip industry, the wafer processing is called the front-end and the die
packaging is called the back-end.

10.2

Wafer processing

The wafer processing takes a large number (typically over 400) of various operations. In
this section we give a brief impression of the main steps. For more detailed information we
refer to [Wol90]. Figure 10.3 shows the first set of operations performed on the wafer.

Figure 10.3: First operations in wafer processing (source: American Naval Acadamy)

10.2. Wafer processing

137

First, a layer of silicon dioxide is grown by exposing it to gas and extreme heat. The wafer is
then coated with a substance called photoresist. Photoresist becomes soluble when exposed
to ultraviolet light. In the photolithography process, ultraviolet light is passed through a
patterned mask, onto the silicon wafer. The mask protects parts of the wafer from the
light. The light turns the exposed areas into a gooey layer of photoresist. Each layer on the
microprocessor uses a mask with a different pattern. The gooey photoresist is completely
dissolved by a solvent. This reveals a pattern of photoresist made by the mask on the silicon
dioxide.

Figure 10.4: Second series of operations in wafer processing (source: American Naval
Acadamy)

The revealed silicon dioxide is etched away with chemicals. Subsequently, the remaining
photoresist is removed. Then a new layer of silicon dioxide is grown and a layer of polysilicon
is applied. With another photolithography process a part of the polysilicon is maintained.
In a similar way several layers with specific properties are added to the wafer. Sometimes
parts of the wafer are doped with ions, sometimes parts of the wafer are exposed to specific
gases, or sometimes metals are deposited on (parts) of the water.
Figure 10.5 shows a simplified flowchart of the processes in the IC manufacturing process.
wafer processing
wafer slicing

Oxidation

Lithography

Etching

Deposition
Doping
Metallization

repeated

single-crystal
growing

Bonding

Packaging

Testing

Figure 10.5: Flowchart of processes in IC manufacturing

The processing of wafers shows a typical re-entrant flow: a single wafer revisits a single
area. Figure 10.6 shows an example wafer fab layout. (The diffusion area contains the oxide
furnaces.) Each area has a stocker. Operators take lots that are to be processed from the
stocker and put lots that are ready processing back into the stocker. An AGV system moves
lots from one stocker to another.

138

Chapter 10. Case: (Part of) a waferfab

Lithograph
Lithography
y

Ion
Implant

Rewor
Rework
k
Production
Production
Equipment
Equipment

Diffusion
Diffusion

Scra
Scrap
p
Inspectio
Inspection
n

Productio
Production
Equipmen
n
Equipment
t

stocker

Inspectio
Inspection
n

Productio
Equipmen
n
Production
t
Equipment

Inspectio
Inspection
n

stocker

Wafer Starts

stocker

Wafer
Moves
AGV wafer
transportation
system

stocker

Inspectio
Inspection
n

Productio
Production
Equipmen
n
Equipment
t

Etch

Wafer Outs

stocker

Productio
Production
Equipmen
n
Equipment
t

Inspectio
Inspection
n

Thin Films

stocker

Productio
Production
Equipmen
n
Equipment
t

Inspectio
Inspection
n

Metallization

Figure 10.6: Example layout of a waferfab

In this case study, we focus on the oxide furnaces in the diffusion area and the lithographic
equipment in the lithography area.

10.3

The oxide furnace and lithographic equipment

Figure 10.7(a) shows the furnace bay of a wafer fab (the Philips SMST facility in Boblingen)
and a single furnace. We see that about 100 wafers are processed simultaneously in the
furnace. Figure 10.7(c) shows the lithography area of a wafer fab (the Crolles 2 facility in
Grenoble).

Figure 10.7: (a) Furnace bay (source: Micron Technologies), (b) Single (horizontal) furnace
(source: Micron Technologies), and (c) Lithographic area (source: Crolles2 waferfab)

Wafers move through the factory in wafer pods. In Figure 10.7(c) a wafer pod is encircled.
From now on, we refer to a wafer pod as a lot, since it is the main transport unit in the
factory. A single lot can contain up to 25 wafers. (The number of wafers can be smaller if a
customer order requires less than 25 wafers.) A lithographic machine processes lot for lot.
The oxide furnace processes multiple lots (typically more than 25 wafers) simultaneously.

10.4. Analytical approach

139

In other words, the lithographic machine is a single-lot machine and the furnace is a batch
machine.
We wish to investigate the following:
What is the effect of different batch sizes on the throughput, flowtime, and
flowtime variability of the lots.
For this case we consider the lithography and furnace area only. Moreover, we assume that
each area contains a single machine. This study does not yield a flowtime estimation for
a real wafer fab, but it does provide us with valuable insight on the influence of the batch
size on the performance of the most critical resource of the wafer fab: the lithography area.
Figure 10.8 shows a schematic representation of the minifab.

ta = 3.0 hours
ca = 1.0

T0,F = 5.5 hours/batch

C0,F = 0.25
BF

BL

L
t0,L = 2.7 hours/lot
c0,L = 1.0

Figure 10.8: Schematic representation of the minifab

We assume an infinite buffer BF for furnace F and an infinite buffer BL for lithographic
machine L. Moreover, we assume the following characteristics (see also Figure 10.8): lots
arrive with a mean inter-arrival time ta = 3.0 hours with coefficient of variation ca = 1.0
hours. The furnace processes batches of fixed size with process time T0,F = 5.5 hours/batch
and coefficient of variation C0,F = 0.25 hours. The lithographic machine processes lots and
has process time t0,L = 2.7 hours/lot and c0,F = 1.0 hours. Note that we use capital letters
for quantities that are related to batches.
In the sequel, we first investigate the flowtime of lots in this minifab analytically, and then
by means of simulation.

10.4

Analytical approach

For analytically determining the flowtime of the minifab we consider the queueing system
in Figure 10.9.
Furnace

Lithography

batchsize k

ta
ca
T0,F C0,F

Td,F
Cd,F

td,L
cd,L
t0,L c0,L

Figure 10.9: Queueing system representation of the minifab

140

Chapter 10. Case: (Part of) a waferfab

At the furnace, lots arrive with a mean inter-arrival time of ta hours. The lots wait to form
a batch of size k and batches queue up before the batch machine. The furnace processes
batches. The mean process time is t0,F hours/batch. Batches leave the furnace with mean
inter-departure time td,F hour and arrive at the lithographic machine. Here a queue of
batches arises. The first batch in line is broken down lot for lot, as the lithographic machine
process one lot at a time. The mean process time is t0,L hours/lot. After processing, lots
leave the machine with mean inter-departure time td,L .
Please note that ta , t0,L , and td,L are related to single lots and that T0,F and Td,F are related
to batches!
For the flowtime calculation of the flowtime in the furnace area we use the equations from
Section 3.5. We get:
F =

2
c2a /k + C0,F
k1
uF
ta +

T0,F + T0,F
2
2
1 uF

with uF =

T0,F
k ta

(10.1)

Here the first term represents the wait-to-batch-time, the second term represents the queueing time of a batch, and the third term represents the process time of the batch.
The coefficient of variation on the inter-departure time Td,F is determined with Kuehns
linking equation.
2
2
2
Cd,F
= u2F C0,F
+ (1 u2F ) Ca2 = u2F C0,F
+ (1 u2F ) c2a /k

(10.2)

Note that this coefficient on the inter-departure time relates to batches. Hence, we require
the coefficient of variation on the inter-arrival time related to batches (Ca2 ). This coefficient
of variation can be determined from the inter-arrival time of lots: Ca2 = c2a /k, see Section 3.5.
For the litho machine, the flowtime also consists of three terms:
2
2
+ C0,L
Cd,F
uL
k1

T0,L +
t0,L + t0,L
2
1 uL
2
2
k Cd,F
+ c20,L
uL
k1

t0,L +
t0,L + t0,L
=
2
1 uL
2
2
Cd,F
+ c20,L /k
k1
uL
k t0,L +
=

t0,L + t0,L
2
1 uL
2

(10.3)

L =

(10.4)
with uL =

t0,L
ta

(10.5)

The first term represents the queueing time of the batches. The coefficient of variation
on the process time for a single lot c0,L can be related to the process time for k lots:
2
C0,L
= c20,L /k. Moreover, the mean process time for k lots T0,L = k t0,L . The second term
represents the breaking down of the batch. When a batch is first in line, not all its lots are
immediately processed. The first lot in the batch is processed right away, the litho machine
starts processing the second lot after t0,L , the third lot after 2 t0,L , and the k-th lot after
(k 1) t0,L . The average wait-for-batch-breakdown time is (k 1)/2 t0,L . The third, and
last term, is the process time for the lot on the litho machine.
For the given specifications (ta = 3.0, ca = 1.0, T0,F = 5.5, C0,F = 0.25, t0,L = 2.7, and c0,L =
1.0) we investigate the influence of batch size k on flowtime and coefficient variation on
the inter-departure time Cd,F for the furnace area. Figure 10.10 shows how the flowtime
and the coefficient of variation on the inter-departure time vary with k.

10.5. Simulation-based analysis

141

50
0.55

[hours]

0.5
0.45

cd,F

40

[hours]

45

0.4

35
0.35
30

0.3
2

4
[lots]

[lots]

Figure 10.10: (a) Mean flowtime and (b) coefficient of variation on inter-departure time for
different batch sizes

We see that for batch size 1 the flowtime grows to infinity: the capacity is insufficient for
this batch size. Small batches have a longer queueing time. Large batches require a larger
wait-to-batch-time. For the given parameters, we see that the flowtime is minimal for 3 lots.
However, we see that the variation on the inter-departure time for k = 3 lots is significantly
higher, leading to less predictable delivery dates.

10.5

Simulation-based analysis

In the previous section we have applied the analytical approaches from Chapter 3 to investigate the influence of the batch size on the flowtime. In this section we will reinvestigate
the same minifab, but with simulation.
Figure 10.11 shows a graphical representation of the minifab system.
G

BF

BL

Figure 10.11: Graphic representation of model Minifab

We define two types: lots and batches.

type lot = real,
bat = list lot;

Chi-10.1

Lots are represented by an id number and the time at which they are released in the line.
Batches are represented by a list of lots.
Process G generates lots according to the predefined arrival pattern. We specify ta and ca
as parameters.

142

Chapter 10. Case: (Part of) a waferfab

proc G(chan! lot a; real ta,ca2):

dist real d = gamma(1.0 / ca2, ca2 * ta);
while true:
a!time; delay sample d
end
end

Chi-10.2

Buffer BF receives lots and sends batches of fixed size k.

proc BF(chan? lot a; chan! bat b; int k):
bat xs; lot x;
while true:
select
a?x:
xs = xs + [x]
alt
size(xs) >= k, b!xs[:k]:
xs = xs[k:]
end
end
end

Chi-10.3

The buffer can always receive new lots and puts them in a list. If this list contains at least
one complete batch (len(xs) >= k), the buffer tries to send a batch. After sending the
batch, the batch is removed from the list of lots.
Furnace F processes batches. The process time is gamma distributed with mean t0 and
coefficient of variation c0 .
proc F(chan? bat a; chan! bat b; real T, C2):
dist real d = gamma(1 / C2, C2 * T);
bat xs;
while true:
a?xs; delay sample d; b!xs
end
end

Chi-10.4

Buffer BL in front of the litho machine, receives batches and sends single lots to the litho
machine.
proc BL(chan? bat a; chan! lot b):
bat xs, ys;
while true:
select
a?ys:
xs = xs + ys
alt
not empty(xs), b!xs[0]:
xs = xs[1:]
end
end
end

Chi-10.5

10.5. Simulation-based analysis

143

Note that adding the batch xs to the total list of lots xss requires no square brackets: both
xs and xss are lists. The litho machine is a regular machine with a gamma distributed
process time.
proc L(chan? lot a; chan! lot b; real t, c2):
dist real d = gamma(1 / c2, c2 * t);
lot x;
while true:
a?x; delay sample d; b!x
end
end

Chi-10.6

Exit E receives finished lots and calculates the mean throughput and mean flowtime. These
measures are printed to screen every 10000-th lot.
proc E (chan? lot a):
int i; real mft; lot x;
while true:
a?x; i = i + 1; mft = (i - 1) / i * mft + (time - x) / i;
Chi-10.7
if i mod 10000 == 0:
write("%10.2f\tmpt = %5.2f\tmft = %5.2f\n", time, i / time, mft)
end
end
end
Finally, we define parametrized model M inif ab. In this way we can easily adapt parameters
of the model.
model Minifab(real ta, ca, T0F, C0F, t0L, c0L; int k):
chan lot a;
chan bat b, c;
chan lot d, e;
run G(a, ta, ca * ca),
BF(a, b, k), F(b, c, T0F, C0F * C0F),
BL(c, d)
, L(d, e, t0L, c0L * c0L),
E(e)
end
Simulating for the given parameters (see previous section) and k = 3 gives:
30185.8 mtp = 0.331282 mft = 41.8421
60203.9 mtp = 0.332204 mft = 38.2607
90001.4 mtp = 0.333328 mft = 38.5445
...
1.37959e+06
mtp = 0.333433 mft =
1.40924e+06
mtp = 0.333514 mft =
...
2.10035e+07
mtp = 0.333278 mft =
2.10336e+07
mtp = 0.333276 mft =

38.0518
38.0234
37.4706
37.4739

Chi-10.8

144

Chapter 10. Case: (Part of) a waferfab

We see that the mean throughput is 0.333 lots/hour, which means we have a stable system,
where all lots entering the system (ta = 3.0 hours) leave the system. The mean flowtime
converges around 37.1 hours. Simulating for the same settings again, yields a slightly
different outcome: 37.5 hours. Figure 10.12 shows how the analytically determined flowtime
and coefficient of variation on the inter-departure time (line marked with os) along with
the outcome of a number of simulation runs (2 million lots) marked with xs.

[hours]

40

cd,F

[hours]

45

50

35

30

4
k

[lots]

0.5
5
0.
5
0.4
5
0.
4
0.3
5
0.
3

4
k

[lots]

Figure 10.12: (a) Mean flowtime and (b) coefficient of variation on inter-departure time for
different batch sizes determined analytically (line marked by os) and by simulation (marked
by xs)

Using Python [Pyt04] we can visualize the simulation output in for example Labview [Nat04,
Ver02]. Figure 10.13 shows a screendump of the visualisation in Labview.

Figure 10.13: Labview visualisation of simulation output

Bibliography
[AHR12]

I.J.B.F. Adan, A.T. Hofkamp, and J.E. Rooda. Chi 3 tutorial, 2012.

[AN98]

W.T.M. Alberts and G. Naumoski. A Discrete-Event Simulator for Systems

Engineering. PhD thesis, Eindhoven University of Technology, Department of
Mechanical Engineering, The Netherlands, 1998.

[AR01]

I. Adan and J. Resing. Queueing theory. Lecture notes, Eindhoven University of

Technology, Department of Mathematics and Computing Science, Eindhoven,
The Netherlands, 2001.

[Are96]

N.W.A. Arends. A Systems Engineering Specification Formalism. PhD thesis,

Eindhoven University of Technology, Department of Mechanical Engineering,
The Netherlands, 1996.

[Ban98]

J. Banks. Handbook of Simulation. John Wiley and Sons, New York, 1998.

[BFS87]

P. Bratley, B.L. Fox, and L.E. Schrage. A guide to simulation. Springer-Verlag,

New York, 1987.

[BK00]

V. Bos and J.J.T. Kleijn. Formalisation of a production systems modelling

language: the operational semantics of core. Fundamenta Informaticae,
41(4):367392, 2000.

[Bro03]

Brooks-PRI Automation. Automod, 2003. http://www.autosim.com.

[Bru97]

G. Bruns. Distributed Systems Analysis with CCS. International series in computer science. Prentice Hall, New York, 1997.

[BS93]

J.A. Buzacott and J.G. Shantikumar. Stochastic models of manufacturing systems. Prentice Hall, New York, 1993.

[BW90]

J.C.M. Baeten and W.P. Weijland. Process Algebra, volume 18 of Cambridge

Tracts in Theoretical Computer Science. Cambridge University Press, Cambridge, 1990.

[Cac02]

Caci Products Company. Simscript, 2002. http://www.caciasl.com.

[CAJ01]

R.B. Chase, N.J. Aquilano, and F.R. Jacobs. Production and Operations Management. Irwin, Chicago, 2001.

[Fle02]

Flexsim. Flexsim, 2002. http://www.flexsim.com.

[Fok00]

W.J. Fokkink. Introduction to Process Algebra. Springer-Verlag, 2000.

145

146

Bibliography

[Gan19]

H.L Gantt. Organizing for work. Industrial Management, 58:89 93, August
1919.

[Ger94]

S.B. Gershwin. Manufacturing Systems Engineering. Prentice Hall, Englewood

Cliffs, 1994.

[GH85]

D. Gross and C.M. Harris. Fundementals of queueing theory. John Wiley and
Sons, New York, second edition, 1985.

[Hof01]

A.T. Hofkamp. Reactive machine control. A simulation approach using . PhD

thesis, Eindhoven University of Technology, Department of Mechanical Engineering, Eindhoven, The Netherlands, 2001.

[HR12]

A.T. Hofkamp and J.E. Rooda.

Chi 3 Reference Manual,
http://chi.se.wtb.tue.nl/reference manual/reference manual.html.

[HS01]

W.J. Hopp and M.L. Spearman. Factory Physics. McGraw-Hill, Boston, second
edition, 2001.

[Ima02]

Imagine that. Extend, 2002. http://www.imaginethatinc.com.

[Int04]

Intel. How chips are made. educational website, 2004.

[Ken53]

D.G. Kendall. Stochastic processes occuring in the theory of queues and their
analysis by the method of the imbedded markov chain. Ann. Math. Stat, pages
338354, 1953.

[Khi32]

A. Khinchine. Mathematical theory of stationary queues. In Mat. Sbornik,

volume 39, pages 7384, 1932.

[Kin61]

J.F.C. Kingman. The single server queue in heavy traffic. In Proc. of the
Cambridge Philosophical Society, volume 57, page 902, 1961.

[Kol33]

A.N. Kolmogorov. Sulla determinazione empirica di una legge di distribuzione

(in italian). In Giornale dell Istituto Italiano degli Attuari, volume 4, pages
8391, 1933.

[Kue79]

P.J. Kuehn. Approximate analysis of general queueing networks by decomposition. IEEE Trans. Comm., 27:113126, 1979.

[Lan02]

Lanner Group. WITNESS, 2002. http://www.lannergroup.com.

[Lee99]

L. Leemis. Simulation input modelling. In P.A. Farrington, H.B. Nembhand,

D.T. Sturrock, and G.W. Evans, editors, Proceedings of the 1999 Winter Simulation Conference, pages 1423, 1999.

[Lef03]

A.A.J. Lefeber. Iteratieve berekening van en (in dutch). Personal communication, 2003.

[Lew95]

E.E. Lewis. Introduction to reliability engineering. John Wiley and Sons, New
York, second edition, 1995.

[Lit61]

J.D.C. Little. A proof of the queueing formula l = w. Operations Research,

9:383387, 1961.

2012.

Bibliography

147

[LK00]

A.M. Law and W.D. Kelton. Simulation Modeling and Analysis. McGraw-Hill,
2000.

[Man03]

Manugistics Inc. Statgraphics, 2003. http://www.statgraphics.com.

[Mat03]

Mathworks inc. Matlab release 13, 2003. http://www.mathworks.com.

[MR02]

D.C. Montgomery and G.C. Runger. Applied Statistics and Probability for
Engineers. Wiley, New York, third edition, 2002.

[Nat04]

National Instruments. Labview, 2004. http://www.ni.com.

[Pea00]

K. Pearson. On a criterion that a given system of deviations from the probable

in the case of a correlated system of variables in such that it can be reasonably
supposed to have arisen in random sampling. In Phil. Mag. (5), volume 50,
pages 157175, 1900.

[Pol30]

F. Pollaczek. Uber eine Aufgabe der Wahrscheinlichkeitstheorie. In I-II Math.

Zeitschrift, volume 32, pages 64100, 729750, 1930. (in German).

[Pyt04]

Python website, 2004. http://www.python.org/.

[Ros00]

S.M. Ross. Introduction to probability models. Academic press, London, 2000.

[RT98]

R.S. Russel and B.W. Taylor. Operations management. Prentice Hall, Upper
Saddle River, 1998.

[SAS03]

SAS. SAS, 2003. http://www.sas.com.

[Smi33]

N. Smirnov. Estimate of deviation between empirical distribution functions

in two independent samples. In Bulletin Moscow University, volume 2, pages
316, 1933.

[SPS03]

SPSS inc. SPSS, 2003. http://www.spss.com.

[Ste74]

M.A. Stephens. Edf statistics for goodness of fit and some comparisons. In
Journal American Statistical Association, volume 69, pages 730737, 1974.

[Tij94]

H.C. Tijms. Stochastic Models, An algorithmic approach. John Wiley & Sons,
Chichester, 1994.

[vdMFR95] J.M. van de Mortel-Fronczak and J.E. Rooda. Application of concurrent programming to specification of industrial systems. In Proceedings of INCOM95,
pages 421426, Beijing, China, October 1995.
[Ver02]

J. Vervoort. Visualisation of chi simulation output using labview. Technical

report, Eindhoven University of Technology, Eindhoven, 2002.

[Wol90]

S. Wolf. Silicon processing. Lattice press, Sunset Beach, California, 1990.

148

Bibliography

Selected answers

d)

Section 2.3
M1 = 0.9, M3 = 0.9, M4 =
M2 = 0.225
u1 = 1.8, u2 = 1.35
u3 = 1.62, u4 = 1.6
max = 0.56 lots/hour

2a)
b)
c)

max = 1.0 lots/hour

= 5.8 hours
w = 5.8 lots

3a)
b)
c)

= 0.31 hours
6,9, and 4 machines
workstation 1 and 2

4a)
b)
c)
d)

max = 31 lots/hour
B2 fills up
max = 13 lots/hour
B1 fills up

5a)
b)
c)

u1 = 4, max = 41 lots/hour
= 28
3 hours
w = 28/15 = 1.87 lots

6a)

uSL = 2, uB = 58
max = 12 lots/hour
= 14 hours
w = 7 lots
No
max = 0.4 lots/hour, uSL = 0.8
= 15.75 hours

1a)
b)
c)

b,c)
d)
e)
f)
g,h)
7a)
b)

b)
2a)

W1 = , W2 = 0.8
W3 = 0.9, W4 =

3c)
d)

W4 = /0.7 = 1.43
W1 = 1.34, W2 = 1.70
W3 = 1.79
max = 0.56 lots/hour

f)
4d)
4f)

1a)
b)
c)
d)
e)
f)
h)
i)

d)
e)
f)

1a)

Section 2.4
W2 = /0.8, W3 = /0.8

W1 = 1.46, W2 = 1.05
W3 = 2.50, W4 = 2.0
uW1 = 1.46, uW2 = 1.37
uW3 = 1.75, uW4 = 1.60
Section 3.7
u1 = u3 = 0.75, u2 = 0.625
B1 = 5.91 min
c2a,2 = 1
B2 = 0.97 min
B3 = 9.0 min
wB1 = 1.48 lots, wB2 = 0.24 lots
wB3 = 2.25 lots
tot = 63.88 min
wtot = 15.97 lots

g)
h)
i)
j)

u1 = u3 = , u2 = 4
u1 = u3 = 0.225, u2 = 0.9
tot = 63.0 hours
wB1 = 0.20 lots, wB2 = 12.56 lots
wB3 = 0.07 lots
max = 1.0 lots/hour
tot = 86.47 hours
tot = 62.63 hours
wI = 77.82 lots, wII = 56.27 lots

3a)
b)

(k) = 5k + 4.75
kmin = 1

4a)

u = 5/6

2a)
c)
d)
e)

u1 = 2.4, u2 = 2.5
max,A = 0.16 prodA/hr
max,B = 0.24 prodB/hr
= 6 hours
w = 2.4 prod
wA = 0.96 prodA, wB = 1.44 prodB

W5 = /4
W4 W5 = /4, W4 W6 =

149

150

Bibliography

b)
c)
d)

bk = 0.75 hours
bq = 2.69 hours
tot = 8.44 hours

1b)
d)

Section 4.5
= 5.0 [lots/hour]
= 5.0 [lots/hour]

3b)
c)
d)

= 15.0 [lots/hour]
0 = 1 = 2 = 5.0 [lots/hour]
0 = 1 = 2 = 5.0 [lots/hour]

5a)
b)
c)

Section 5.9
= 0.2 lots/minute
= 6.24 minute
u = 0.988
w = 1.263 lots
= 5.0 lots/hour
= 0.2 hour
u = 1.0
w = 1.0 lots

d)

= 0.25 lots/min
= 5.69 min
1 = 3.63 min
2 = 5.36 min
3 = 8.35 min
nB = 0.59 lots

6a)
b)

= 0.24 lots/hour
= 0.249 lots/hour

1
2
3

Section 6.5
T, T, F, T, F
T, F, F, F
F, F, T, F, F, T

4b)
c)

n = 2.353 [lots/hour]
n(N=1) = 2.490 [lots/hour]
n(N=2) = 2.499 [lots/hour]
n(N=3) = 2.500 [lots/hour]

5a)
b)

seconds
t0 = 11.4 s
Section 7.4

machine

2a,b)

c)

0
1
2
3
4
5
6
7
8

[lots/hour]
0.6
0.2
0.2
0.2
0.4
0.4
0.3
0.3
0.4

t
[hour]
1.5
4.5
4.5
4.5
2.25
2.25
3.0
3.0
2.25

min,0,1,4,5 = 9.0 hours

min,2,3 = 6.75 hours

1a)
d)

Section 8.5
uAP = 0.8
b0 = 0.6 lots, b1 = 0.7 lots
b2 = 1.05 lots, b3 = 1.2 lots

2a)

u0 = u2 = 0.75, u1 = 0.625

3a)

ra = 2.0 lots/hour