Thesishdjshuhsgdoowuu 8 WJWH

Impact of Skill: Seru vs Classical Assembly Line
A thesis presented to
the faculty of
the Russ College of Engineering and Technology of Ohio University
In partial fulfillment
of the requirements for the degree
Master of Science
Md Abdullah
May 2018
© 2018 Md Abdullah. All Rights Reserved.

2
This thesis titled
by
MD ABDULLAH
has been approved for
the Department of Industrial and Systems Engineering
and the Russ College of Engineering and Technology by
Gursel A. Suer
Professor of Industrial and Systems Engineering
Dennis Irwin
Dean, Russ College of Engineering and Technology

3
ABSTRACT
ABDULLAH, MD, M.S., May 2018, Industrial and Systems Engineering
Director of Thesis: Gursel A. Suer
Seru manufacturing system is the modern version of manufacturing strategy
evolved from Toyota Production System (TPS), Just In Time (JIT) and Lean
Manufacturing. Seru researchers and practitioners claim that, using Seru is always a better
strategy where there are simple equipment involved in an assembly environment. In
contrast with the claim of Seru researchers and practitioners, this thesis aims to compare
Classical Assembly Line with Seru in terms of worker’s skill on different tasks. In this
study, seven skill levels for workers has been considered and it is assumed that skill levels
are normally distributed. Based on the output of Seru and classical assembly line, it is found
that when the skill variation is high for different tasks among workers, Classical Assembly
line is a better choice. Results found from multi-product case with and without skill
consideration suggest that, when skill is not considered Seru is a better choice most of the
time. On the other hand, when skill is considered a hybrid system of Seru and Classical
Assembly line can co-exist. It is also found that when learning and forgetting take place
over time, learning can lead to multi-skilled workers and hence the system become more
suitable for Seru.

4
DEDICATION
This thesis is dedicated to my beloved parents. Without their continuous support, I
wouldn’t have been able to accomplish this.

5
ACKNOWLEDGMENTS
It is my immense pleasure to express my sincere gratitude to all the people who
helped me achieving this. At first, I will thank Department of Industrial and Systems
Engineering, Ohio University for providing me with this precious opportunity of pursuing
my Master’s degree with financial aid.
I also want to express my gratefulness to my adviser Dr. Gursel Suer, who guided
me from the beginning to the end of this thesis. The touch of his wisdom and witty
comments will always be my guidance in my future life and career.
I am grateful to my committee members Dr. Dusan Sormaz, Dr. Ashley Metcalf
and Dr. Todd Myers. I thank them all to contribute and for their valuable comment and
time. Without their support, it wouldn’t have been a success.
It would be an honor for me to mention about Bryan D. Jordan whose support made
it possible to finish my study without any trouble. I am also thankful to my friend Md.
Mehadi Hasan Farhad for his technical support.

6
TABLE OF CONTENTS
Page
Abstract ............................................................................................................................... 3
Dedication ........................................................................................................................... 4
Acknowledgments............................................................................................................... 5
List of Tables .................................................................................................................... 10
List of Figures ................................................................................................................... 16
Chapter 1: Introduction ..................................................................................................... 18
1.1 Assembly Line ................................................................................................... 18
1.1.1 Different Assembly Line Types ................................................................ 19
1.2 Different Assembly Line Layouts ...................................................................... 21
1.3 Related Terms in Line Balancing ...................................................................... 23
1.4 Assembly Line Balancing .................................................................................. 27
1.4.1 Assembly Line Balancing Problem Type ................................................. 30
1.5 Seru Production System ..................................................................................... 31
1.5.1 Divisional Seru.......................................................................................... 32
1.5.2 Rotating Seru ............................................................................................ 32
1.5.3 Yatai .......................................................................................................... 33
1.6 Skills in Assembly Line and Seru System ......................................................... 33

7
1.7 Seru vs Classical Assembly Line ....................................................................... 34
1.8 Mathematical Modeling ..................................................................................... 35
1.9 Objectives of the Thesis ..................................................................................... 36
1.10 Organization of the Thesis ................................................................................. 36
Chapter 2: Literature Review ............................................................................................ 37
Chapter 3: Problem Statement .......................................................................................... 48
3.1 Single Product Case Analysis Based on Standard Time Only.............................. 48
3.2 Single Product Case Analysis Based on Skill ....................................................... 51
3.3 General Classification of Cases ............................................................................ 55
Chapter 4: Methodology and Results for Single Product Case......................................... 56
4.1 Methodology for Single Product Case .................................................................. 57
4.1.1 Mathematical Model 1 ................................................................................. 57
4.2 Results Based on Skill for Single Product (Product 1) ......................................... 58
4.2.1 When Skill Levels 1 to 7 All Are Considered ............................................. 59
4.2.2 When Skill Levels 3, 4, 5 Are Considered................................................... 61
4.2.3 When Skill Levels 2, 4, 6 and 1, 4, 7 Are Considered ................................. 62
4.2.4 Cases Where All Workers Are Good at One Task and One Workers Is Good
at Each Task ……………………………………………………………………………64
4.2.5 When Skill Levels All 4 and All 7 Are Considered..................................... 66

8
4.3 Results Based on Skill for Single Product (Product 2 and Product 3) ............... 69
4.3.1 Results for Product 2................................................................................. 69
4.3.2 Results for Product 3................................................................................. 72
4.4 Impact of Different Standard Deviation ............................................................ 74
4.5 Indexing ............................................................................................................. 78
Chapter 5: Multi-Product Multi-Period Case without Skill .............................................. 81
5.1 Methodology for Multi-Product Multi-Period Case (without Skill) .................. 82
5.1.1 Mathematical Model 2 .............................................................................. 82
5.2 Results for Multi-Product Multi-Period without Skill ....................................... 83
5.3 Methodology for Multi-Product Multi-Period to Maximize Output.................. 92
5.3.1 Mathematical Model 3 .............................................................................. 93
5.4 Results for Output Maximization (Multi-Product Multi-Period without Skill
Case) ………………………………………………………………………………….94
Chapter 6: Multi-Product Multi-Worker Assignment Considering Skills and Learning,
Forgetting Rates ................................................................................................................ 98
6.1 Methodology for Multi-Product Worker Assignment Considering Skills
………………………………………………………………………………….98
6.1.1 Mathematical Model: 4 ............................................................................. 99
6.2 Results for Multi-Product Worker Assignment Considering Skill .................. 100
9
6.2.1 Results from Math Model of Section 6.2 ................................................ 107
6.3 Maximizing Output for Multi-Product Case with Skill ................................... 110
6.3.1 Mathematical Model: 5 ........................................................................... 110
6.4 Results for Output Maximization Model for Multi-Product with Skill Case
……….………………………………………………………………………..112
6.5 Implementing Multi-Product and Multi-Worker Assignment with Skill in
Cellular Manufacturing Environment ............................................................................. 114
6.6 Learning and Forgetting Effect ........................................................................ 115
Chapter 7: Conclusions and Future Works ..................................................................... 124
7.1 Conclusions ...................................................................................................... 124
7.2 Future Works ................................................................................................... 127
References ....................................................................................................................... 128
Appendix A: Mathematical Model 1 .............................................................................. 132
Appendix B: Mathematical Model 2 and 3 ..................................................................... 142
Appenxid C: Mathematical Model 4 and 5 ..................................................................... 144
Appendix D: Learning and Forgetting Effect ................................................................. 147

10
LIST OF TABLES
Page
Table 1: Task and processing time of different tasks of product X .................................. 24
Table 2: Cycle time for assembly line .............................................................................. 50
Table 3: Seru production system....................................................................................... 51
Table 4: Skill level and distribution of operation time (Süer and Tummaluri, 2008) ...... 52
Table 5: Standard operation time and σ value for 15% standard deviation ...................... 52
Table 6: Skill level matrix................................................................................................. 53
Table 7: Operation times based on skill level ................................................................... 53
Table 8: Task distribution as in assembly line .................................................................. 53
Table 9: Summary of output ............................................................................................. 54
Table 10: List of the sections in chapter (chapter four) .................................................... 56
Table 11: Data set 1-mixed skills...................................................................................... 59
Table 12: Operation time based on skill level (data set 1-mixed skills) ........................... 59
Table 13: Worker and task assignment among workstations (data set 1-mixed skills) .... 59
Table 14: Skill levels (data set 2-mixed skills) ................................................................. 60
Table 15: Operation times based on skill levels (data set 2-mixed skills) ........................ 60
Table 16: Worker and task assignment among workstations (data set 2-mixed skills) .... 60
Table 17: Data set 3-skills 3, 4 and 5 ................................................................................ 61
Table 18: Operation times based on skill levels (data set 3-skills 3, 4 and 5) .................. 61
Table 19: Worker and task assignment among workstations (data set 3-skills 3, 4 and 5)
........................................................................................................................................... 61
11
Table 21: Operation times based on skill levels (data set 4-skills 2, 4 and 6) .................. 62
........................................................................................................................................... 62
Table 24: Operation times based on skill levels (data set 5-skills 1,4 and 7) ................... 63
Table 25: Worker and task assignment among workstations (data set 5-skills 1,4 and 7) 63
Table 26: Data set 6-skill for operation 5 is 7................................................................... 64
Table 27: Operation times based on skill levels (data set 6-skill for operation 5 is 7) ..... 64
Table 28: Worker and task assignment among workstations (data set 6-skill for operation)
5 is 7)................................................................................................................................. 65
Table 29: Data set 7-skill for operation 1 is 7................................................................... 65
Table 30: Operation times based on skill levels (data set 7-skill for operation 1 is 7) ..... 65
Table 31: Worker and task assignment among workstations (data set 7-skill for operation
1 is 7)................................................................................................................................. 65
Table 32: Data set 8-W1 highly skilled, W3 poorly skilled ............................................. 66
Table 33: Operation times based on skill levels (data set 8-W1 highly skilled, W3 poorly
skilled)............................................................................................................................... 66
Table 34: Worker and task assignment among workstations (data set 8-W1 highly skilled,
W3 poorly skilled) ............................................................................................................ 66
Table 35: Data set 9-all skills are 7................................................................................... 67
Table 36: Operation times based on skill levels (data set 9-all skills are 7) ..................... 67
12
Table 37: Worker and task assignment among workstations (data set 9-all skills are 7) . 67
Table 38: Data set 10-all skills are 4................................................................................. 67
Table 39: Operation times based on skill levels (data set 10-all skills are 4) ................... 68
Table 40: Worker and task assignment among workstations skill levels (data set 10-all
skills are 4) ........................................................................................................................ 68
Table 41: Summary of output for Seru and assembly line (product 1) ............................. 68
Table 42: Skill levels (all average) ................................................................................... 70
Table 43: Processing time based on skill levels (all average) .......................................... 70
Table 44: Worker and task assignment among workstations (all average) ...................... 71
Table 46: Skill level 1 to 7 normally distributed (data set 1)............................................ 72
Table 47: Processing time based on worker (skill level 1 to 7 normally distributed for data
set 1) .................................................................................................................................. 73
Table 48: Worker and task assignment among workstations (skill level 1 to 7 normally
distributed for data set 1) .................................................................................................. 73
Table 50: Operation times based on skill level with σ = 5% ............................................ 74
Table 51: Task distribution as in assembly line ................................................................ 75
Table 52: Operation times based on skill level with σ = 10% .......................................... 75
Table 54: Operation times based on skill level with σ = 20% .......................................... 76

13
Table 56: Summary of output of different standard deviation (product 1) ....................... 77
Table 59: Range of skill levels for tasks ........................................................................... 78
Table 60: Standard deviation of skill levels...................................................................... 79
Table 61: Summary of output and sum of ranges for product 1 ....................................... 79
Table 64: Cycle time for different number of workers for product A .............................. 84
Table 65: Cycle time for different number of workers for product B .............................. 85
Table 66: Cycle time for different number of workers for product C .............................. 86
Table 67: Cycle time for different number of workers for product D .............................. 87
Table 68: Summary for cycle time.................................................................................... 87
Table 69: Production rate from cycle time ....................................................................... 87
Table 70: Manpower distribution for multi-period (demand data set 1) .......................... 89
Table 73: Summary table for different f values of data set 1............................................ 91
Table 76: Demand data set 1............................................................................................. 95

14
Table 77: Manpower distribution for multi-period from mathematical model 3 (demand
data set 1) with f = 20% .................................................................................................... 95
Table 79: Manpower distribution for multi-period from mathematical model 3 (Demand
data set 2) with f = 20% .................................................................................................... 96
data set 3) f=20% .............................................................................................................. 99
Table 82: Skill matrix for product A............................................................................... 102
Table 83: Skill matrix for product B ............................................................................... 102
Table 84: Skill matrix for Product C............................................................................... 102
Table 85: Product A, cycle time for 1-worker option (Seru Unit) .................................. 103
Table 86: Product A, cycle time for combinations for 2-worker options ....................... 103
Table 87: Product A, cycle time for combinations of 3-worker option .......................... 104
Table 88: Product B, cycle time for 1-worker option (Seru unit) ................................... 104
Table 89: Product B, cycle time for combinations of 2-worker option .......................... 105
Table 90: Product B, cycle time for combinations of 3-worker option .......................... 105
Table 91: Product C, cycle time for 1-worker option (Seru Unit) .................................. 106
Table 92: Product C, cycle time for combinations of 2-worker option .......................... 106
Table 93: Product C, cycle time for combinations of 3-worker option .......................... 107
Table 94: Demand data set 1........................................................................................... 108

15
Table 99: Result for demand data set 1 ........................................................................... 112
Table 100: Summary of results for output maximization ............................................... 113
Table 101: The learning and forgetting rates .................................................................. 116
Table 102: Worker assignment at period 1 ..................................................................... 118
Table 103: Check sheet of worker’s task performance after period 1 ............................ 119
Table 104: Worker assignment at period 2 ..................................................................... 120
Table 105: Check sheet of worker’s task performance after period 2 ............................ 121
Table 106: Revised skill matrix after learning and forgetting take place for product A after
period 2 ........................................................................................................................... 122
Table 107: Revised skill matrix after learning and forgetting take place for product B after
period 2 ........................................................................................................................... 122
Table 108: Revised skill matrix after learning and forgetting take place for product C after
period 2 ........................................................................................................................... 122
Table 109: Revised skill matrix after learning and forgetting take place for product A after
period 8 ........................................................................................................................... 122
Table 110: Revised skill matrix after learning and forgetting take place for product B after
period 8 ........................................................................................................................... 123
Table 111: Revised skill matrix after learning and forgetting take place for product C after
period 8 ........................................................................................................................... 123

16
LIST OF FIGURES
Page
Figure 1: Single model assembly line ............................................................................... 19
Figure 2: Multi-product assembly line .............................................................................. 20
Figure 3: Mixed-product assembly line ............................................................................ 21
Figure 4: Basic straight assembly line layout ................................................................... 21
Figure 5: Straight line with multiple parallel stations....................................................... 22
Figure 6: U-shaped line ..................................................................................................... 22
Figure 7: Precedence relation of product X ...................................................................... 23
Figure 8: Workstations for product X ............................................................................... 24
Figure 9: Distribution of tasks of product X among workstations.................................... 24
Figure 10: Example precedence diagram .......................................................................... 28
Figure 11: Distribution of tasks among workstations ....................................................... 29
Figure 12: Workstations after balancing ........................................................................... 30
Figure 13: A divisional Seru ............................................................................................. 32
Figure 14: A rotating Seru ................................................................................................ 33
Figure 15: Three different Yatais from a rotating Seru .................................................... 33
Figure 16: Precedence diagram for the example problem (product 1) ............................. 48
Figure 17: Each worker is doing all the tasks independently (Seru) ................................ 49
Figure 18: Tasks are assigned to workers according to their skill .................................... 49
Figure 19: Precedence diagram for product 2 ................................................................... 69
Figure 20: Precedence relation of product 3 ..................................................................... 72

17
Figure 21: Precedence relation for product A ................................................................... 83
Figure 22: Precedence relation for product B ................................................................... 84
Figure 23: Precedence relation for product C ................................................................... 85
Figure 24: Precedence Relation of Product D................................................................... 86
Figure 25: Precedence relations for product A, B and C ................................................ 101
Figure 26: A cellular manufacturing environment.......................................................... 115
Figure 27: Selection of manufacturing strategy type based on skill variation ................ 126
18
CHAPTER 1: INTRODUCTION
In this chapter, a discussion of general concepts of assembly line and different terms
related to assembly line such as assembly line layouts, cycle time, station time, lead time,
line efficiency, station efficiency precedence diagram, workstation, and workplace are
presented. Later, it has been followed by brief discussion of assembly line balancing, Seru,
difference between Seru and Classical assembly line, skills in assembly lines, mathematical
programming modeling and objective of the thesis.
1.1 Assembly Line
Assembly work is vital in manufacturing world. Generally, a finished product
cannot be obtained with a single operation or from a single part. Therefore, it is common
that different parts are produced separately by performing different operations. After
producing parts and/or purchasing parts it is required to assemble them together in order to
obtain finished product (Suer 1998). According to Boysen et al., workstations are arranged
serially one after another so that precedence relation is maintained and parts are moved
forward as in a flow-shop environment (Boysen et al., 2007). In other words, an assembly
line is the collection of workstations arranged serially one after another to perform tasks
on a product in order to obtain a finished good.
Some examples of assembly-oriented industry are automobile, aircraft, laptop,
mobile phone and apparel industry. In a computer assembly line, the assembly steps are as
follows: Open the case, install the power supply, attach the components to the mother board
and install the mother board, install CPU and heat sink, install the RAM, install the internal
drives such as Hard Disk Drive (HDD), install video adapter and install Network Interface
19
Card (NIC). In an apparel industry sewing operations where different parts of the garments
are sewed are considered as the assembly operations.
1.1.1 Different Assembly Line Types
This section briefly discusses the classification of assembly lines based on the
number of products handled in an assembly line, such as, i) single-product assembly line,
ii) multi-product assembly line and iii) mixed product assembly line.
i. Single-Product Assembly line
An assembly line which is dedicated to only one product is defined as single
product assembly line. In this type of assembly line, only one product is carried along a
line using conveyor belt or manual system which passes through different stations and at
each station preselected parts are assembled to the product until the finished product is
obtained. For example, a manufacturer produces two different models of vehicle: model A
and model B. So, if only A is assigned in an assembly line alone then this is a single product
assembly line as shown in Figure 1. Depending on the demand of the product the line can
be changed for a different model after a certain period. Normally, this type of assembly
line is used when the demand is very high but the variety of products is not much. Figure
1 shows product arrival in a single model assembly line.
Figure 1: Single model assembly line

20
ii. Multi-product assembly line:
In this type of assembly line, two or more products are assembled in batch in the
same line (Liao, 2014). Suppose, a manufacturer produces three different products: product
A, product B product C. These three different products will be assigned in the same line in
batches where the batch size is preselected based on the customer demand. Since, in this
type of assembly line, products are assembled in different batches whenever a different
batch is to be processed a setup occurs (Figure 2). Multi-product assembly line is also
known as batch assembly.
Figure 2: Multi-product assembly line
iii. Mixed-product assembly line
In mixed-product assembly line, different products are assembled arbitrarily
(Yagmahan 2011). This type of assembly line is flexible in terms of product change which
is helpful for reducing final product inventory. If we consider the example of three different
products (A, B and C) in this assembly line type, product A product B and product C are
assigned in the same line in arbitrary fashion. In other words, the type of assembly line in
which workers work in different products on the same assembly line arbitrarily is called
mixed-product assembly line (Figure 3).

21
Figure 3: Mixed-product assembly line
1.2 Different Assembly Line Layouts
The layout is the way of arranging different workstations in an assembly line.
Several different assembly line layouts are found in the literature. The choice of assembly
line layout depends on the objective and also on the nature/design of the product. In this
section, basic straight line, straight line with multiple parallel stations and U-shaped
assembly lines are discussed.
i. Basic Straight Line:
In this type of layout, workstations are arranged in a straight line one after another
according to the serial of the operation (Figure 4). The job follows the sequence based on
the design of the product (Battaïa et al., 2013).
Figure 4: Basic straight assembly line layout
ii. Straight line with multiple parallel stations:

22
Authors defined this type of line as a set of workstations arranged serially, parallel
or combination of both (Battaïa et al., 2013). Figure 5 shows a straight line with multiple
parallel stations.
Figure 5: Straight line with multiple parallel stations
iii. U-shaped line:
In a U-shaped line, workstations are arranged like a U where the flow of work is
unidirectional (Figure 6). In this type of line, a worker can access both the beginning and
ending of the operation in a flow (Bagher et al., 2011).
Figure 6: U-shaped line

23
1.3 Related Terms in Line Balancing
There are various terms used in assembly line balancing such as Cycle time, Lead
time, Takt time, Station time, Station Efficiency, Line Efficiency, Precedence and
Smoothness index. These are discussed as follows:
Cycle Time: Cycle time is the maximum station time of an assembly line. Suppose,
an assembly line has 8 tasks that need to be done in order to obtain product X. The actual
time for each task is given in Table 1.
Figure 7: Precedence relation of product X
Also assume that, the assembly line has three workers and three workstations
as shown in Figure 8. The tasks are assigned to each workstation by the management
as shown in Figure 9
Station times of all three stations are calculated and are shown in Figure 9.
Therefore, Cycle time = Maximum {station time 1, station time 2, station time 3}
= Maximum {90, 100, 75}
= 100 seconds/ unit
= 1.6 min/unit
24
Table 1: Task and processing time of different tasks of product X

Task Processing time (in
Seconds)
a 30
b 45
c 25
d 15
e 40
f 35
g 50
h 25
Total processing 265
time
Figure 8: Workstations for product X
Figure 9: Distribution of tasks of product X among workstations
Cycle time is also determined based on the demand of the product. This can be
referred as given cycle time in order to meet the customer demand.

Cycle Time =
(Equation 1)
25
If the weekly demand of a product is 500 units then the cycle time for that product
is:
× ×
Cycle Time = = = 4.8 minutes/unit
Station time: The sum of the processing times of the tasks assigned to a station is
known as station time. From
Figure 9, the tasks a, b, d have been assigned to station 1. The sum of the processing
times of those tasks is 30+45+15 = 90 seconds. As a result, station time of station 1 is 90
seconds. Similarly, station times of station 2 and station 3 are 100 seconds and 75 seconds,
respectively.
Lead Time: Lead time is the total time that a process takes to complete a finished
product from the beginning of the process to the end. It is the average length of time for an
input to move all the way along the operations to the end of the operations. In summary, it
is the sum of production times along the assembly line. Cycle time and lead time are related.
Equation 2 shows the relation between Lead time and Cycle time. It is also known as the
Little’s Law.
Lead time = Cycle time × Work in Progress (WIP) (Equation 2)
Here, WIP refers to inputs that are still in the operation. For assembly line, WIP is
equals to the number of workstations assuming that no buffers or waiting between
workstations.
Line Efficiency: It is the ratio of input used for the whole line and the output
obtained from the line. In other words, it is the ratio of total station time of all station to
the product of Cycle time and number of stations.

26
∑
= ×
∗ 100% (Equation 3)
Efficiency depends on several parameters for example worker skill, equipment used
and job method.
Station Efficiency: Station efficiency is the ratio of station time to cycle time. It
measures how well a station is performing compared to the whole line.

= ∗ 100% (Equation 4)

Precedence: In assembly line, there are operations that must be done before other
operation/s. For example, for a desktop or laptop computer, a mother board must be
assembled before a hard disk can be assembled, for a shirt manufacturing cutting must be
done before sewing. Precedence diagram (Figure 10) is a network of nodes which shows
the precedence relations. In a precedence diagram, there are circles and arrows where
number inside the circle represents task and the arrows represent the relation. From Figure
10 it is clear that task 1 must be done before tasks 2 and 3 and tasks 4 and 5 must be done
before task 8.
If task i must precede task j then the relation is expressed by i < j.
Workstation: A workstation is a specific location where some pre-specified tasks
are performed. In other words, workstation is a location in assembly line along the flow of
work where one or more element/s of work is performed by one or more worker/s using
one or more equipment or machines. Generally, a workstation is comprised of a single
worker only but there might cases where a workstation is manned by more than one worker.
For example, in fixed position layout as in aircraft manufacturing line.

27
Smoothness Index: This index indicates how evenly the tasks have been distributed
among workstations in the assembly line. The smaller the value of smoothness index the
better the line is and a zero smoothness index indicates implies perfect balance. Equation
5 shows how to calculate smoothness index (SI).
SI = ∑ ( − )2 (Equation 5)
STmax = Maximum station time (which is the cycle time in most cases)
STi = station time of station i.
Balancing Efficiency: It indicates the percentage utilization of all the stations put
together. The formula for calculating Balancing Efficiency in percentage is given below.

BE = ∗ 100 (Equation 6)
∗
1.4 Assembly Line Balancing
In a general assembly line problem, there are several tasks which are required to be
performed in order to produce the final product where some tasks must be accomplished
before the starting of other tasks. The tasks follow a precedence relation where preceding
task/s must be performed before its successor. When assembly line is designed, first, the
precedence relation is considered and then tasks are assigned to workstations such that
either i) the number of workstations are minimized for a given cycle time or ii) the cycle
time is minimized for a given number of workstations. The combination of tasks grouped
among workstations that fulfils any of the chosen objectives is called balancing.
In other words, Assembly Line Balancing is defined as assigning a number of tasks
to various workstations so as to maximize Balancing Efficiency (BE) or to minimize
Number of workstations (N) or to accomplish any other given objective function for a given
28
volume of output without violating the precedence relationship. So, the objective of line
balancing problem can be of two different types; find the minimum number of workstations
subject to known cycle time (i.e. given demand) or find the minimum cycle time subject to
predetermined number of workstations (Suer, 1998). Each task has its own processing time
and the sum of processing times in each station is station time. Maximum station time of
the assembly line is the cycle time. Therefore, cycle time can lead to assembly rate (AR)
directly (AR= 1/ CT) (Suer, 1998).
In an assembly line precedence graph, there might be multiple starting nodes but
normally it ends with one single node (Rekiek et al., 2002). An example of assembly line
balancing is illustrated using Figure 10.
Figure 10: Example precedence diagram
From the precedence diagram in Figure 10, it is clear that task 1 must be done before
tasks 2 and 3 and task 5 and 6 cannot be performed until task 2 is done. This way, the
finished product is obtained once all 11 tasks are accomplished.

29
In this example, the total processing time of the product is
6+4+8+4+8+2+6+9+5+7+6 = 65 minutes.
Suppose, the demand of the product is 4500 units per year and available time is
40×50×60 = 120,000 min/year. Here, 50 weeks in a year and 40 hours in a week, 60 minutes
in an hour.
Cycle time = available time/ demand = 120,000/4500 = 26.66 minutes per unit.
Since total time is 65 minutes, number of stations required is ≥ (65/26.66) ≥ 2.43 ⩯
3 stations. So, the objective is to assign these 11 tasks to three stations in a way such that
no station time is greater than 26.66 minutes. It is assumed that all stations are equally
capable. For three stations, one possible distribution of tasks is shown in Figure 11.
Figure 11: Distribution of tasks among workstations
Though the minimum station time is 17 minutes, the cycle time of this configuration
is still 26 minutes because the maximum station time is the cycle time which implies that
the product cannot be produced earlier than 26 minutes. Therefore, one might want to
reconfigure the tasks so that the difference between the highest station time and the lowest
station time is minimized.

30
New distribution of tasks is shown in Figure 12 where the line is said to be balanced.
Figure 12 shows that the cycle time is 22 now, which is 4 unit time less than previous
combination.
Objective functions used in literature other than minimizing cycle time and
minimizing workstations are: maximize line efficiency, maximize system utilization and
minimize line smoothness index (Battaïa et al., 2013).
Figure 12: Workstations after balancing
1.4.1 Assembly Line Balancing Problem Type
Based on the objective function, there are two different types of Assembly Line
Balancing (ALB) problems namely Type 1 ALB and Type 2 ALB.
i. Type 1 ALB:
In these types of assembly line balancing problem, the objective is to minimize the
number of workstations given that cycle time is pre-specified (Koltai, 2013).
ii. Type 2 ALB:
The problem of minimizing cycle time when the number of workstation is fixed is
known as type 2 ALB (Koltai, 2013).

31
1.5 Seru Production System
Seru is a Japanese word which means cell (Zhang et. al., 2017). It is a relatively
new production system developed in Japan in 1992 (Sakazume, 2005) well known in Asia
only till now which integrates both lean and agile production system. Japanese electronics
companies for example Canon, Panasonic, Sony, Fujitsu NEC and Hitachi are practicing
Seru (Iwamuro, 2004). There is a consensus among the practitioners of Seru that Seru
production system works better where the demand volume is low but the variety is high
(Villa et. al., 2013). The basis of Seru is that ‘One worker carries all the operations in an
assembly line’ (Zhang et. al., 2017). A Seru is dedicated to one or several similar product
types which have one or more worker who is/are multi-skilled so that they can perform all
the operations that are required to produce the finished product. In other words, long
assembly lines are broken into several small complete lines where simple equipment are
used and only one worker performs the tasks from the beginning to the end of the line.
According to Zhang et. al., (2017) and Yu et al. (2013) Seru has several advantages
over classical cellular manufacturing system such as it is highly flexible for high variety
low volume product, it not only can reduce the workspace, workforce, WIP inventories and
lead time but also it provides improved quality and most importantly it leads to very high
satisfaction of workers. Johnson et. al., (2004) reported that Seru also results in better
ergonomics of work environment. According to Liu et al. (2014), Seru combines the
advantages of job shop, mass production and sustainable manufacturing. From literature it
is found that Seru has three different classifications namely divisional Seru, rotating Seru
and yatai.
32
1.5.1 Divisional Seru
According to Yu et al. (2012), a divisional seru is “a short line staffed with several
partially cross-trained workers where the tasks are divided into different sections and each
section is operated by one or more worker” Yu et al. (2012). Divisional Seru is the
modification of conveyor assembly line where a worker performs more task than in
conveyor assembly line as shown in Figure 13. In divisional Seru, workers are multi-
skilled as compared to the conventional assembly line.
Figure 13: A divisional Seru
1.5.2 Rotating Seru
In rotating Seru there are multiple multi-skilled workers who move from one
workstation to another to perform all the tasks based on a fixed order (Figure 14). In this
type of Seru workers move from one workstation to another with the product. One worker
is responsible for assembling an entire product from beginning to the end. Generally, this
type of Seru has a U-shaped structure (Yu et al. 2013).

33
Figure 14: A rotating Seru
1.5.3 Yatai
Yatai is the special form of rotating Seru where there is only one worker and he/she
doesn’t need to rotate from workstation to workstation rather all the equipment are
accessible. Generally, a Yatai owner is the one who carry out all the operations from
operational to managerial (Yu et al. 2013). Figure 15 shows three different yatais.
Figure 15: Three different Yatais from a rotating Seru
1.6 Skills in Assembly Line and Seru System
In a classical assembly line, it is enough that one worker is skilled in only a few
tasks because that particular worker is doing the same task again and again. On the other
hand, in Seru multi-skilled worker is vital because one worker is doing all or many the
tasks of a line (Liu et al., 2013).

34
Skill level can be represented as categories as in Suer and Tummaluri (2008) where
the authors divided skills into 9 different levels. Level 5 implies that the skill is average,
1 means the best level of skill and 9 is the worst level of skill. Here, it is assumed that
operator’s skill level follows normal distribution and the operation times were determined
based on skill level.
Kuo and Yang (2007) proposed binary representation of skill categories. For
example, if there are two workstations a worker can fall into any of the following
categories: (1, 0), (0, 1) or (1, 1) where 1 means the operator can attend that workstation
and 0 means he/she doesn’t have required skill to attend that workstation. A worker’s skill
category (0, 1) means that he/she cannot attend workstation 1 but can attend workstation
2.
1.7 Seru vs Classical Assembly Line
Sakazume (2004) depicts the difference between Seru and Cellular manufacturing
system in a broad context. He called Seru as Japanese Cell Manufacturing (JCM) and as a
basis of comparison he considers four different points namely i) implementation changes
ii) cell feature iii) advantages and disadvantages of implementation and 4) mechanism
behind the advantages and disadvantages. In terms of implementation change, Sakazume
(2004) noted that Cellular Manufacturing (CM) is the result of evolution of job shop layout
towards a flow shop layout whereas JCM is the result of the change in the distribution of
worker and production equipment. Besides, in JCM there are more cells and few workers
and heavy machineries are replaced by simple equipment. One major advantage of Seru is
35
since Seru uses only one worker for a complete line the utilization of worker is very high
in this case because there is no balancing delay in this case.
1.8 Mathematical Modeling
A mathematical model is set of equations comprises of variables where one
equation is known as the objective function and other equations are known as constraints.
The objective function can be of two types either maximization type or minimization type.
The variables can be integer, binary or real numbers. There are some variables which can
only have value 0 or 1. These are known as decision variables. There are many different
types of mathematical programming models such as linear programming model, integer
programming model, mixed-integer programming model and nonlinear programming.
Generally, a mathematical model has five general steps such as
1. Formulating the model.
2. Gathering the data.
3. Obtaining an optimal solution.
4. Applying sensitivity analysis.
5. Testing and implementing the solution.
Linear programming is used for different problem formulations in many different
sectors such as transportation, manufacturing system, economics, business management,
etc. “A Linear programming problem is a problem of minimizing or maximizing a linear
function in the presence of linear constraints of the inequality and/or equality type”
(Bazaraa & Jarvis, 1943).

36
1.9 Objectives of the Thesis
Seru is comparatively a new approach. There are few papers which studied the Seru
in terms of workers skill and compare Seru and Assembly line. So, this thesis is aimed to
carry the following analysis considering worker skill level:
i. To compare Seru and Classical assembly line in terms of workers skill level.
ii. To gain knowledge about selection of Seru and assembly line by Indexing
obtained from skill levels matrix.
iii. To analyze Seru and Classical Assembly line in terms of multi-product and
multi-period with workers skill consideration where learning and forgetting has
significant effect.
1.10 Organization of the Thesis
The first chapter of the thesis includes the introduction of basic concepts of Seru
and assembly line balancing which is followed by related literature in chapter two. Chapter
three discusses the problem statement. Chapter four contains the methodology and results
for Single product single period case. An analysis of multi-product multi-period case
without skill consideration is presented in chapter five. Chapter six covers multi-product
multi-worker assignment with skill consideration and an analysis of learning and forgetting
effect on worker skill level.

37
CHAPTER 2: LITERATURE REVIEW
Literature regarding assembly line balancing is very rich because the issue is being
discussed for a long time. On the other hand, literature on Seru is limited since this concept
is recent and only practiced in Asia primarily in Japan and China.
Inspired by the real-life problems in jewelry, electronic components, apparel and
sewing industry assembly lines, Süer and Tummaluri (2008) considered a problem of
operator assignment for a labor-intensive cell. The authors considered a manufacturing
process where five operations are required to obtain a finished product and there are 10
products in total. It is natural that if a worker does a specific task for a long time his output
will be more because the worker learned well and if a skilled worker does not do that task
for a significant amount of time his performance of that task will be poor because he
forgets. So, while approaching the problem of assigning operators the authors considered
skill level of workers, learning rate and forgetting rate in the context of the multi-period
environment. A hierarchical approach where there are three phases was proposed. In the
first phase, a mixed integer programming model was developed to determine alternative
manpower for products. In this case, standard time of operations were used. The second
phase is designed to identify cell size and worker level of every cell simultaneously. In
third phase skill of operators was considered as a basis for assigning them in different cells
and it was followed by consideration of output rate and completion time. Two different
approaches for operator assignment were considered namely Max and MaxMin approach.
In Max approach workers are assigned in a way so that the objective of maximizing the
total skill is obtained. In this way, the scope of obtaining multi-skilled workers gets lower
38
and lower because one specific operator will be assigned again and again for the same
operation (forgetting effect) since he has the highest skill for that operation. On the other
hand, MaxMin approach assigns the low skill operator to the operation which is a
bottleneck to rotate the operators to different operations so that they can learn. In other
words, MaxMin approach nurtures the learning effect. Lower makespan value was obtained
using Max approach and MaxMin approach improved skill level on a regular basis.
Suksawat et al. (2005) addressed the issue of evaluating the skill level of workers
considering five factors of both job and worker such as quickness, difficulty, complexity,
thoroughness and reliability. Skill level evaluation was conducted through a Genetic
Algorithm implementation where an initial chromosome has genes which includes machine
type, machine model and skilled technician. The scheduling performance was evaluated
using three different indexs namely free time index (FTI), skill level index (SLI) and
working time index (TWI).
Fitzpatrick et al. (2005) studied the formation of group of workers based on natural
tendencies and interpersonal skills and skill of task. In a cellular manufacturing
environment, each cell needs a group of workers who has different skill set. In order to get
maximum output, it is required to form the team in the best possible way. In their work the
authors assumed that the entire pool of labor can be divided into several skill categories.
They developed a math model and used IPL to solve the problem. They also used a heuristic
method to solve and compare the solution with math model. It was proved that
interpersonal skill, technical competences have a greater impact on the performance of the
team.
39
Ramezanian and Abdullah (2015) presented a study where they considered the
problem of line balancing and worker assignment problem for a mixed-model assembly
line. According to Ramezanian and Abdullah (2015), a mixed model assembly is an
assembly line where it is possible to assemble a set of products at the same time. In their
study, the authors considered two different objectives simultaneously such as minimizing
cycle time and minimizing total operating cost. The authors considered several
assumptions for their model: for example, one precedence diagram is enough to delineate
the precedence diagram for all models, operators lies in same skill level has very subtle
different in their skill and hence it is ignored, the operating time and cost are both
dependent on the skill of operator and task condition, there is only one operator in each
station and the task cannot be shared. An integer linear programming model was developed
to approach this problem. Their problem consists of 40 assembly tasks, 3 different products
and two different skill levels where the number of predetermined stations is 7. An
evolutionary algorithm namely Imperialist competitive algorithm (ICA) was proposed and
the result was compared with Genetic Algorithm (GA). The authors claim that the result
obtained from ICA is better than GA in terms of runtime.
Corominas et al. (2008) conducted a research on a motor-cycle assembly plant
which is located in Spain where different models of motor-cycles are produced. They
attempted to reassemble the line in order to meet seasonal demand. The problem authors
studied has some features such as the company appointed short-term workers during peak
season (just before summer) where these workers take more time to accomplish a task than
a full-time worker, a full-time employee must work with a part time employee and there
40
are different task groups which are incompatible with each other. The assembly line
consists of 103 tasks and the authors set their objective as to minimize the number of short-
term workers for a given cycle time. They developed a mixed linear programming model
and the model was solved by using CPLEX 9.0 optimizer. According to the authors, they
found an optimal solution where they have been able to reduce the number of temporary
workers by two and number of skilled worker by one.
Koltai (2013) presented a study where he formulates workforce skill constraints in
assembly line balancing. He considered three different levels of skill constraints such as
low skill constraints, high skill constraints and exclusive skill constraints. In low skill
constraints, the objective is to determine workstations for workers who can perform only
simple tasks whereas in the high skill constraints it is assumed that the workers need more
skill than low skill workers. At last the third level is exclusive skill constraints where a
group of worker should have the expertise to accomplish a special set of tasks.
Yang and Gao (2016) proposed adjacent cross-training for a mixed model assembly
line (MAL) in order to improve the skill of workers working in adjacent workstations so
that the assembly line can be balanced easily. To meet the variety and uncertainty in
product demand the authors proposed that the workers should share the tasks of adjacent
workstation so that they can be multi-skilled and at the same time the company doesn’t
need to worry about mass training program and associated costs. As a result, when there is
a change in demand of the product the adjacent workers can be used to handle the increased
demand. In their work, they made several assumptions such as model demands are
uncertain, there are some common tasks and each common task represents one special skill,
41
only one worker in each station, task processing time is constant. They proposed a
mathematical model and solved it for 450 different problems from the literature using
BB&R algorithm. They found the optimal solution for 388 problems and for the reaming
problems the lower bound between objective and their solution are very negligible.
Based on a real-life problem of Thin Film Transistor Liquid Crystal Display (TFT-
LCD) inspection and packaging assembly line, Kuo and Yang (2007) studied the problem
of multi-skilled operator assignment problem for a multi-line manufacturing environment.
They developed a mixed integer programming model where the objective function is to
minimize the multiplication of staffing and skill levels. They defined skill categories based
on the operation(s) that a worker can perform as a basis of assigning workers in different
workstations and every worker belong to one of these skill categories. In (TFT-LCD)
Inspection and Packaging (I/P) has three operations wherein first two operations, display
quality and surface defects are inspected sequentially which later followed by packaging
stage. Assuming there are N workstations and L lines in an assembly area the authors
defined skill categories to design their model. They conducted experimentation on real life
data found from a TFT-LCD assembly in Spain. The number of different product groups
is 5 and there are 10 I/P lines.
Liu et al. (2013) presented a study where they studied the problem of training the
workers to obtain multi-skilled worker and assigning the multi-skilled workers in order to
implement seru. They considered the conversion of assembly line into seru, the type of
seru considered here is divisional seru or yatai. The objective of this study is to minimize
the total training cost and balance the total processing time in each seru. The assumptions
42
considered in this study are: for one product type there is only one seru, a worker is capable
of processing multiple tasks of a product or all tasks, the processing time of each task is
standard time, at least one workers must be assigned to each seru. The authors developed
a multi-objective mathematical model where the two objectives are to minimize task-to-
worker training cost and optimal assignment of workers-to-seru. According to authors
since the model is a multi-objective optimization problem they can only obtain best-so-far
result rather than the optimal solution. Therefore, they proposed a three-stage nine steps
heuristic algorithm to trade-off between the targets. Initially, they solved a problem where
there are four different product types and each product has six tasks and those products
were running on a conveyor assembly line with eight skilled workers. They converted the
conveyor assembly line into four seru with two workers at each seru where each seru is
assigned for a specific product type. Later they checked the effectiveness of their model
and heuristics for 30 workers and 13 product types, 40 workers and 14 product types.
Yu et al. (2013) discussed the conversion of assembly line into Seru or pure cell
system. Mentioning that deciding the number of cells to be converted and assignment of
worker is a complicated decision-making problem, they designed a multi-objective model
and proved that it is a NP hard problem. The objectives are to reduce worker for a converted
cell and to increase productivity, in other words, reducing total throughput time (TTPT) at
the same time. In their study, the authors considered only rotating Seru and yatai. Several
assumptions that they considered are the types and batch of products are known, cost of
equipment is negligible since the equipment are simple, one specific batch can only be
assembled in specific assembly line, a worker assigned to a cell needs to perform all the
43
tasks from the beginning to the end, each cell cannot have more worker than total number
of workers and setup time is considered zero when product is similar. Using the improved
exact algorithm they found 540 total feasible solutions for a problem of 5 workers. They
concluded that TTPT value can be kept low using fewer workers but not less than the TTPT
value when there are more workers. But their model is able to find the number of operators
that can be reduced for a given TTPT.
Kaku et al. (2009) presented a study where they defined and approached a line-cell
conversion problem for a single product type. They reported three achievements so far in
their study which are number of cells to be formatted, number of workers required to be
assigned in each cell and reduction of workers when the assembly line is converted into
cells. Three different assembly systems were considered in this paper a pure assembly line,
a pure cell and a mixed cells + assembly line. Total throughput and total labor hours were
taken into account as two different criteria to evaluate the system overall performance.
When it is determined that conveyor assembly line should be converted into Serus
then there are some issues that must be studied such as number of Serus to be generated
and assignment of workers on each seru. Liu et al. (2012) conducted a study where they
studied the problem considering both issues where they proposed a comprehensive
mathematical model. They considered several assumption while designing the
mathematical model such as the long assembly line can be converted into shortened
assembly line and Seru, products must be completed either in a Seru or in the shortened
assembly line completely and lot splitting is not allowed. Finally, they checked the model
44
using real life data from an industry and compared the result with the model of Kaku
(2008). The authors claimed that the proposed model is better than that of kauk’s model.
Liu et al. (2014) suggested a comprehensive guideline about implementing Seru.
The authors suggested some basic steps that should be followed if someone is
implementing Seru for the first time. The steps are identifying product design and process
requirement, Selection of seru, material delivery method, production balancing, cross
training of workers and evaluation of Seru production System.
Yang et al. (2014) studied the problem of line cell conversion considering two
different objectives such as total throughput time (TTPT) and total labor hour (THL). For
this multi-objective problem they first attempted mathematical model and later switched to
Genetic Algorithm (GA) since the problem became NP-hard. They assumed that an
assembly line is being converted into pure cell only unlike the assumption made in Kaku
(2008) where he considered a pure assembly line, a pure cell system and a hybrid system
of assembly line and Seru for line cell conversion problem. In many cases suggested GA
was able to generate ‘reliable’ solution, authors claim.
Stecke et al. (2012) presented a study where the authors discussed the evolution of
Seru from Toyota Production System (TPS). According to authors TPS is not fit for a
volatile market where the product has a very short life cycle. The authors also discuss how
Seru is better than TPS in dealing with volatile product type. Considering the results found
from Samsung, Sony, Canon Panasonic, LG and Fujitsu they found that Seru reduces
workforce, space and lead time. The authors discuss in details why implementing Seru can
be beneficiary for organizations.

45
Kaku et al. (2008) studied the conversion of assembly line to cellular manufacturing
strategy considering human-task-related parameters. They considered the number of
possible tasks that can be added, skill level and workers cross training while developed
three distinct models. The proposed models are CAL (Conveyor assembly line), CM
(Cellular Manufacturing) and a hybrid system of CAL and CM. They found that 1 worker
CM has better output with respect to number of tasks compared with 2 workers CM and 3
workers CM. The authors suggested that conversion of line cell system is useful when cross
training is practiced and a certain level of skilled worker is obtained.
Yu et al. (2016) conducted a study reflecting the complexity of converting line into
Seru in terms of scheduling. They considered ten different scheduling rules to show the
impact of commonly used scheduling rules on the performance of line-seru conversion.
They also considered two performance measure namely Total Throughput Time (TTPT)
and Total Labor Hour (TLH) to develop a bi-objective model. This model is used to show
the performance of the conversion in terms of TTPT and TLH. To obtain Pareto-optimal
solution for the bi-objective model they developed two exact algorithm. The authors claim
that their model was successful in minimizing time complexity and space complexity and
an improved computational performance by 98%.
Yu et al. (2017a) presented a study where they considered the conversion of a
hybrid system consists of short line and Seru. Claiming that a hybrid system of seru and
short line is more pragmatic than a pure seru system (some equipment might be expensive
and costly to purchase multiple of those), they developed models for conversion of line-
hybrid Seru systems. In a line hybrid system the short line can be positioned either in front
46
of the Seru/s or in between or behind the Seru/s. In this study, the authors assumed that the
short line will be placed behind the Seru/s. They consider two performance measures
makespan and total labor hour and different constraints such as worker allocation, seru
formation and short line scheduling. The authors claim that the significance of this study
is some insights regarding line-hybrid seru system conversion that they found. The authors
also believe that these insights are important for managerial decision. They also discussed
some properties of line-hybrid seru systems.
Considering two different performance measures such as number of worker and
makespan simultaneously, Yu et al. (2017b) presented a study where their objective is to
reduce worker without increasing makespan for a line-Seru conversion case. According to
the authors, a seru system can reduce both worker and makespan compared to classical
assembly line but, when these two objectives are considered at the same time an increased
value of makespan is found from Pareto-optimal solution if it reduces number of workers.
To avoid this situation two exact and meta-heuristics algorithms are developed for small
and medium sized problem and a meta-heuristics algorithm for large size problem.
Mentioning that in order to convert a line into seru it is highly required that workers
should be multi-skilled, Ying & Tsai (2017) conducted a study where their objective is to
minimize cost of training for worker so that the workers can become multi-skilled and Seru
can be implemented using those multi-skilled workers. They also studied the problem of
assigning multi-skilled workers. They developed a mathematical programming model
which is solved by the help of two stage heuristics method.

47
A few papers addressed the skill comparison between Assembly line and Seru.
Fewer literature compared Seru and classical assembly line based on skill for multi-period
multi-product operator assignment problem.

48
CHAPTER 3: PROBLEM STATEMENT
In this chapter, the problems studied in this thesis is introduced using an example.
The main purpose is to show the impact of skill on both Assembly Line and Seru production
Systems. In section 3.1, a case of single product is considered based only on standard time
for both Seru and classical assembly line. Later, in section 3.2 single product case analysis
based on skill is performed. Section 3.3 discusses all the cases included in this thesis.
3.1 Single Product Case Analysis Based on Standard Time Only
In this example problem, there are six tasks, four workstations and four workers.
The objective is to find whether a Seru (a worker is doing all the tasks independently) is
obtaining better throughput or assigning workers as in classical assembly line is having
better throughput.
Figure 16: Precedence diagram for the example problem (product 1)

49
In this study, tasks for different products are considered independently also, skill
for an operator is defined for each task independently. In other words, a worker can have a
skill level 4 for task 4 but only skill level 2 for task 5.
Figure 16 shows the precedence relation among the tasks of product 1. In order to
obtain the finished product, all these six tasks must be accomplished. There are two ways
to assign the tasks:
1. Each worker can do all of the tasks independently (Seru) as in Figure 17.
2. Tasks can be assigned to different workers (Classical assembly line) as in Figure
18.
Figure 17 shows that worker 1, worker 2, worker 3 and worker 4 are doing tasks 1,
2, 3, 4, 5 and 6 independently (Seru). On the other hand, Figure 18 shows that worker 1 is
doing task 1, worker 2 is doing task 2, worker 3 is doing tasks 3 and 4 and worker 4 is
doing tasks 5 and 6 because they are efficient in doing the tasks they are assigned to.
Precedence relation is also maintained in this case.
Figure 17: Each worker is doing all the tasks independently (Seru)
Figure 18: Tasks are assigned to workers according to their skill

50
Suppose, that the tasks are assigned to workstations as shown in Table 2, in order
to minimize cycle time, here the number of workstations is four. Table 2 shows the station
times for all four stations. The maximum of all station times is 12 minutes. Therefore, the
cycle time is 12 minutes.
Table 2: Cycle time for assembly line

Workstation Workstation 2 Workstation 3 Workstation Cycle time
1 4
Task 1 Task 2 Tasks 3, 4 Tasks 5, 6 Max{10, 8, 12, 12}
10 8 5+7 = 12 3+9=12 = 12
Worker 1 Worker 2 Worker 3 Worker 4
Table 3 shows the production time of the product for all four workers when they
work independently on all of the tasks which is when Seru is used. In this case, production
time is identical for all four workers because processing times are based on standard time.
Output Calculation for Assembly line and Seru based on standard time only:
In a regular day, the total working hour is 8 hours. So, available time from a worker
is = 8×60 minutes = 480 minutes.

Output for Assembly line:
= = 40 units per day.
Output from Seru production system: Available time ∗

+

+
+
= ( + +
+ ) = 44 units per day.

51
It is found that output from Seru (44 units) is higher than the output from classical
assembly line (40 units).
Table 3: Seru production system

Worker 1 Worker 2 Worker 3 Worker 4
Task 1, 2, 3, 4, 5, 6 Task 1, 2, 3, 4, 5, 6 Task 1, 2, 3, 4, 5, 6 Task 1, 2, 3, 4, 5, 6
10+8+5+7+3+9 = 10+8+5+7+3+9 = 10+8+5+7+3+9 = 10+8+5+7+3+9 =

42 42 42 42
3.2 Single Product Case Analysis Based on Skill
In section 3.1, cycle time was calculated based only on standard time of each task.
In real life, it is not likely that two different persons will take the same time to accomplish
the same operation.
Time taken to perform a task by a person depends on several factors such as skill,
equipment used, environment, experience, age and training. In this study, it is assumed that
tasks required simple equipment there are no machine limitations. It is also assumed that,
workers will work to their full potential according to their skill level on specific task.
In this problem, seven skill levels have been considered for workers and it is
assumed that skill levels are normally distributed. Skill level 7 means the best, skill level 4
means average and skill level 1 is the worst. In their study Süer and Tummaluri (2008),
used different skill levels. They used skill level 1 is the best and skill level 9 is the worst.
In this study, 7 is assumed to be the best level of skill and 1 is the worst level. Normally
distributed skill levels have a mean μ and a standard deviation σ (15% standard deviation
has been taken as 15% of the mean in this study). If a worker has high skill level for a task
52
his operation time for that task will be low and vice versa. Since, skill level 1 is the worst,
3 standard deviation has been added to the mean time so that it is reflected in operation
time.
Table 4: Skill level and distribution of operation time (Süer and Tummaluri, 2008)
Skill Level Time
1 μ+3σ
2 μ+2σ
3 μ+σ
4 μ
5 μ-σ
6 μ-2σ
7 μ-3σ
Operation times are calculated based on operator’s skill levels and standard time.
Standard operation time and 15% standard deviation of mean is shown in Table 5.
For example, if a worker has a skill level of 6 for task 2 then the operation time of
that task for that worker is 8 – 2×1.2 = 5.6. Table 6 shows skill levels of different workers
generated randomly for different tasks and Table 7 shows operation times of different
tasks based on skill level.
Table 5: Standard operation time and σ value for 15% standard deviation
Task 1 Task 2 Task 3 Task 4 Task 5 Task 6
Product P1 10 8 5 7 3 9
σ (15% Std.
Dev of mean) 1.5 1.2 0.75 1.05 0.45 1.35
53
Table 6: Skill level matrix

Worker 1 5 3 5 4 4 3
Worker 2 4 2 4 3 4 4
Worker 3 4 4 4 2 1 4
Worker 4 4 6 4 7 4 5
Table 7: Operation times based on skill level

Total of
Task 1 Task 2 Task 3 Task 4 Task 5 Task 6 Each
worker
Worker 1 8.5 9.2 4.25 7 3 10.35 42.3
Worker 2 10 10.4 5 8.05 3 9 45.45
Worker 3 10 8 5 9.1 4.35 9 45.45
Worker 4 10 5.6 5 3.85 3 7.65 35.1
Now, output is calculated from the operation times found after skill is considered.
Table 8 shows task distribution for assembly line.
Table 8: Task distribution as in assembly line
Workstation 1 Workstation 2 Workstation 3 Workstation 4

Task 1 Task 2, 4 Task 3, 5 Task 6
W1 W4 W2 W3
5.6+3.85=9.45
8.5 (Cycle time) 5+3=8 9
Output Calculation:
In a day regular working hour is 8 hours. So, available time from a worker is =
8×60 minutes = 480 minutes.

54

Output for assembly line: = = 50.79 units per day
.
Output from Seru: Available time ∗ +

+ + = + +
. .
+ = 46.14 units per day.

. .
Summary of output based on standard time and based on skill for both Seru and
Assembly line is shown in Table 9.
Table 9: Summary of output

Seru Assembly line
Output based on standard time 45.71 40
Output based on skill 46.14 50.79
From Table 9 it is clear that when skill is considered there is a change in output
compared to the output from standard time. When standard time is considered, the
difference between output from seru and output from assembly line is bigger. In this case,
seru output is higher than assembly line output. On the other hand, when skill is considered
assembly line output (50.79) is better than the output from Seru (46.14). Another point to
notice is that, the gap between the output for Seru in both case (45.71 and 46.14) is very
close but output for assembly line (40 and 50.79) have a bigger gap.
55
3.3 General Classification of Cases
In this study, three different factors are considered for experimentation such as
number of products, number of periods and skill. Based on these three different factors five
different cases will be discussed in this thesis, namely:
i. Single product single period without skill.
ii. Single product single period with skill.
iii. Multiple product single period without skill.
iv. Multiple product single period with skill.
v. Multiple product multiple period considering learning and forgetting rates.

56
CHAPTER 4: METHODOLOGY AND RESULTS FOR SINGLE PRODUCT CASE
In this chapter, results for single product case is discussed based on the
methodology discussed in section 4.1. Table 10 shows the sections and content of different
sections of chapter four.
Table 10: List of the sections in chapter (chapter four)

Section Content
4.1 Methodology for single product case.
4.2 Results based on skill for single product (Product 1)
4.2.1 When skill levels 1 to 7 all are considered
4.2.2 When skill levels 3, 4, 5 considered
4.2.3 When skill levels 2, 4, 6 and 1, 4, 7 are considered
4.2.4 Cases where all workers are good at some task and
one worker is good at each task.
4.2.5 When skill levels all 4 and all 7 are considered
4.3 Results based on skill for single product (Product 2 and product 3)
4.3.1 Results for product 2
4.3.2 Result for product 3
4.4 Impact of Different Standard Deviation
4.5 Indexing
57
4.1 Methodology for Single Product Case
In this section, a mathematical model is discussed for single product case. This
model is designed to find the minimum cycle time and task assignment along with worker
assignment in workstations.
4.1.1 Mathematical Model 1
Indices:
i task index
j workstation index
k worker index
Notations:
T cycle time
tik operation time of task i by worker k
n number of tasks
m number of workstations, number of workers.
Ei earliest station for task i that can be assigned to, given the precedence
relations
Li latest station for task i that can be assigned to, given the precedence relations
Decision variables:
Xijk 1, if task i is assigned in workstation j to worker k; 0, otherwise
Ykj 1, if worker k is assigned in workstation j; 0, otherwise
Objective Function:
Minimize, Z = T (Equation 7)
58
Subject to,
∑ ∑ = 1 ( = 1, 2, … , ) (Equation 8)
∑ ∑ ≤ ( = 1, 2, , … ) (Equation 9)
≤ ( = 1, 2, … , ; = 1, 2, … , ) (Equation 10)
= 1 ( = 1, 2, … , ) (Equation 11)
= 1 ( = 1, 2, … , ) (Equation 12)
∑ ≤∑ ( Equation 13)
Equation (7) is the objective function and it makes sure that cycle time is
minimized. Equation (8) confirms that all the tasks are assigned whereas equation (9)
makes sure that station time at each station does not exceed cycle time. Equation (10) is to
ensure that tasks are assigned at workstation with a worker, only when that worker is
assigned at that workstation. Equation (11) confirms that every worker is assigned and
equation (12) ensures that every workstation has a worker. Finally equation (13) maintains
precedence relations where b is an immediate follower of a.
4.2 Results Based on Skill for Single Product (Product 1)
Seven skill levels for workers for the same problem discussed in chapter 3 has been
considered and it is assumed that skill levels are normally distributed. Skill level 7 means
the best, skill level 4 means average and skill level 1 is the worst level of skill.
For the purpose of experimentation, several different data sets were generated
randomly which follow normal distribution and mathematical model was run to generate
results from those data sets. Later, some data sets were generated in order to check the
59
impact of gap in the skill level. For example, data sets based only on skill levels 3, 4, 5 and
2, 4, 6 and 1, 4, 7 were generated to see the impact of gap in skill level.
4.2.1 When Skill Levels 1 to 7 All Are Considered
In this section two different data sets (Table 11 and Table 14) and their results are
shown where skill levels from 1-7 all are considered assuming that skill levels are normally
distributed.
Table 11: Data set 1-mixed skills

Worker 1 5 2 5 4 4 2
Worker 2 4 3 4 3 4 3
Worker 3 3 5 1 4 6 5
Worker 4 1 3 4 6 4 7
Table 12: Operation time based on skill level (data set 1-mixed skills)
Total of each
worker
10 8 5 7 3 9
Worker 1 8.5 10.4 4.25 7 3 11.7 44.85
Worker 2 10 9.2 5 8.05 3 10.35 45.6
Worker 3 11.5 6.8 7.25 7 2.1 7.65 42.3
Worker 4 14.5 9.2 5 4.9 3 4.95 41.55
Table 13: Worker and task assignment among workstations (data set 1-mixed skills)

Task 1 Task 2 Task 3, 4 Task 5, 6
W1 W2 W4 W3
5+4.9=9.9 (Cycle
8.5 9.2 time) 2.1+7.65=9.75
60
Output Calculation:

Output for assembly line:
= .
= 48.48 units per day.
Output from Seru:
Available time ∗ + +

+ = + + + = 44.12 units
. . . .
per day.
Table 14: Skill levels (data set 2-mixed skills)

Worker 1 3 1 6 5 4 5
Worker 2 2 7 4 2 4 4
Worker 3 6 4 2 6 3 2
Worker 4 5 4 4 3 6 4
Table 15: Operation times based on skill levels (data set 2-mixed skills)
Task Total of
Task 1 Task 2 Task 3 Task 5 Task 6 each worker
4
10 8 5 7 3 9
Worker 1 11.5 11.6 3.5 5.95 3 7.65 43.2
Worker 2 13 4.4 5 9.1 3 9 43.5
Worker 3 7 8 6.5 4.9 3.45 11.7 41.55
Worker 4 8.5 8 5 8.05 2.1 9 40.65
Table 16: Worker and task assignment among workstations (data set 2-mixed skills)
W4 W2 W3 W1
4.4+5=9.5 (Cycle
8.5 time) 4.9+3.5=8.4 7.65
61

Output for assembly line : = = 50.52 units per day.
.
Output from Seru : .

+ .
+ .
+ .
4.2.2 When Skill Levels 3, 4, 5 Are Considered
In this section, skill levels only 3, 4, 5 are considered to check how output of Seru
and output of Assembly line varies when skill levels are very compacted. In other words,
what is the impact of very low skill difference on both seru and assembly line.
Table 17: Data set 3-skills 3, 4 and 5

Worker 1 3 4 4 4 5 4
Worker 2 5 3 4 3 4 4
Worker 3 4 5 3 4 4 4
Worker 4 4 5 4 5 4 3
Table 18: Operation times based on skill levels (data set 3-skills 3, 4 and 5)
Task Total of
4
10 8 5 7 3 9
Worker 1 11.5 8 5 7 2.55 9 43.05
Worker 2 8.5 9.2 5 8.05 3 9 42.75
Worker 3 10 6.8 5.75 7 3 9 41.55
Worker 4 10 6.8 5 5.95 3 10.35 41.1
w2 w4 w1 w3
6.8+5=11.8 (Cycle
8.5 time) 7+2.55=9.55 9

Output for assembly line: = = 40.67 units per day.
.
62
Output from Seru: + + + = 45.60 units per day.

. . . .
4.2.3 When Skill Levels 2, 4, 6 and 1, 4, 7 Are Considered
In this section, skill levels are considered in a way such that the gap among skill
levels are bigger. Here, skill levels 2, 4, 6 (Table 20) and 1, 4, 7 (Table 23) are considered
to check how output of seru and output of assembly line varies when skill levels are more
dispersed.

Worker 1 2 6 4 4 6 4
Worker 2 5 4 4 2 6 4
Worker 3 4 2 6 4 4 4
Worker 4 2 6 4 6 4 2
Table 21: Operation times based on skill levels (data set 4-skills 2, 4 and 6)
Task Total of
4
10 8 5 7 3 9
Worker 1 13 5.6 5 7 2.1 9 41.7
Worker 2 8.5 8 5 9.1 2.1 9 41.7
Worker 3 10 10.4 3.5 7 3 9 42.9
Worker 4 13 5.6 5 4.9 3 11.7 43.2
w3 w4 w1 W2
5.6+5=10.6 (Cycle
10 time) 7+2.1=9.1 9

.
63
Output from Seru: + + + = 45.32 units per day.

. . . .

Worker 1 4 4 1 4 7 4
Worker 2 4 7 4 7 1 4
Worker 3 7 4 7 4 4 1
Worker 4 4 7 1 1 4 7
Table 24: Operation times based on skill levels (data set 5-skills 1,4 and 7)
Total of
Task 1 Task 2 Task 3 Task 4 Task 5 Task 6 each
worker
10 8 5 7 3 9
Worker 1 10 8 7.25 7 1.65 9 42.9
Worker 2 10 4.4 5 3.85 4.35 9 36.6
Worker 3 5.5 8 2.75 7 3 13.05 39.3
Worker 4 10 4.4 7.25 10.15 3 4.95 39.75
Table 25: Worker and task assignment among workstations (data set 5-skills 1,4 and 7)
Task 1, 3 Task 2, 4 Task 5 Task 6
w3 w2 w1 w4
4.4+3.85 = 8.25
5.5+2.75= 8.25 (cycle time) 3 4.95

.
Output from Seru : + + + = 48.59 units per day.

. . . .
64
4.2.4 Cases Where All Workers Are Good at One Task and One Workers Is Good at Each
Task
This section shows data set and result for several different cases. For example in
Table 26 is assumed that all workers are best at task 5, poor at task 1 and average at task 4.
On the other hand, in Table 29 it is assumed that all workers are best at task 1, poor at task
3 and average at task 4. In
Table 32 it is assumed that worker 1 is highly skilled in all the tasks and worker 3
is not very good at any task.
Table 26: Data set 6-skill for operation 5 is 7

Worker 1 3 4 1 4 7 4
Worker 2 3 7 4 4 7 4
Worker 3 3 4 7 4 7 1
Worker 4 3 7 1 4 7 7
Table 27: Operation times based on skill levels (data set 6-skill for operation 5 is 7)
Total of
Task 1 Task 2 Task 3 Task 4 Task 5 Task 6 each worker
10 8 5 7 3 9
Worker 1 11.5 8 7.25 7 1.65 9 44.4
Worker 2 11.5 4.4 5 7 1.65 9 38.55
Worker 3 11.5 8 2.75 7 1.65 13.05 43.95
Worker 4 11.5 4.4 7.25 7 1.65 4.95 36.75

.

. . . .
65
Table 28: Worker and task assignment among workstations (data set 6-skill for operation)
5 is 7)
W4 W2 W3 W1
11.5 4.4+5=9.5 (Cycle time) 7+1.65=8.65 9
Table 29: Data set 7-skill for operation 1 is 7

Worker 1 7 4 1 4 3 4
Worker 2 7 7 4 4 3 4
Worker 3 7 4 7 4 3 1
Worker 4 7 7 1 4 3 7
Table 30: Operation times based on skill levels (data set 7-skill for operation 1 is 7)
Total of
10 8 5 7 3 9
Worker 1 5.5 8 7.25 7 3.45 9 40.2
Worker 2 5.5 4.4 5 7 3.45 9 34.35
Worker 3 5.5 8 2.75 7 3.45 13.05 39.75
Worker 4 5.5 4.4 7.25 7 3.45 4.95 32.55
Table 31: Worker and task assignment among workstations (data set 7-skill for operation
1 is 7)
W4 W2 W3 W1
4.4+5=9.5 (cycle
5.5 time) 7 3.45+4.95 = 8.4

Output for assembly line (Data set task 1: level 7):
= .
= 50.52 units
per day
Output from Seru (Data set task 1: level 7): = + + + = 52.73

. . . .
units per day.

66
Table 32: Data set 8-W1 highly skilled, W3 poorly skilled

Worker 1 6 7 7 7 7 6
Worker 2 4 4 4 4 4 4
Worker 3 1 2 1 2 1 1
Worker 4 5 2 4 3 2 3
Table 33: Operation times based on skill levels (data set 8-W1 highly skilled, W3 poorly
skilled)
Total of
10 8 5 7 3 9
Worker 1 7 4.4 2.75 3.85 1.65 6.3 25.95
Worker 2 10 8 5 7 3 9 42
Worker 3 14.5 10.4 7.25 9.1 4.35 13.05 58.65
Worker 4 8.5 10.4 5 8.05 3.9 10.35 46.2
Table 34: Worker and task assignment among workstations (data set 8-W1 highly skilled,
W3 poorly skilled)
Task 1 Task 2, 3, 5 Task 4 Task 6
W4 W1 W3 W2
8.5 4.4+2.75+1.65=8.8 9.1 (Cycle time) 9

= .
Output from Seru : .

+ + .
+ .
4.2.5 When Skill Levels All 4 and All 7 Are Considered
In this section in Table 35, a best case scenario is depicted where it is assumed that
all workers are best in all tasks whereas in Table 38 it is assumed that all workers are
average in all the tasks.

67
Table 35: Data set 9-all skills are 7

Worker 1 7 7 7 7 7 7
Worker 2 7 7 7 7 7 7
Worker 3 7 7 7 7 7 7
Worker 4 7 7 7 7 7 7
Table 36: Operation times based on skill levels (data set 9-all skills are 7)
Total of each
Task 1 Task 2 Task 3 Task 4 Task 5 Task 6 worker
10 8 5 7 3 9
Worker 1 5.5 4.4 2.75 3.85 1.65 4.95 23.1
Worker 2 5.5 4.4 2.75 3.85 1.65 4.95 23.1
Worker 3 5.5 4.4 2.75 3.85 1.65 4.95 23.1
Worker 4 5.5 4.4 2.75 3.85 1.65 4.95 23.1
Table 37: Worker and task assignment among workstations (data set 9-all skills are 7)
W1 W2 W3 W4
2.75+3.85= 6.6 1.65+4.95=6.6
5.5 4.4 (cycle time) (cycle time)

.

. . . .
Table 38: Data set 10-all skills are 4

Worker 1 4 4 4 4 4 4
Worker 2 4 4 4 4 4 4
Worker 3 4 4 4 4 4 4
Worker 4 4 4 4 4 4 4
68
Table 39: Operation times based on skill levels (data set 10-all skills are 4)
Total of
Task 1 Task 2 Task 3 Task 4 Task 5 Task 6 each
worker
10 8 5 7 3 9
Worker 1 10 8 5 7 3 9 42
Worker 2 10 8 5 7 3 9 42
Worker 3 10 8 5 7 3 9 42
Worker 4 10 8 5 7 3 9 42
Table 40: Worker and task assignment among workstations skill levels (data set 10-all
skills are 4)
W1 W2 W3 W4
10 8 5+7=12 3+9=12

Output for assembly line (Data set all 4): = = 40 units per day.

Output from Seru (Data set all 4): = + + + = 45.71 units per day.
Table 41: Summary of output for Seru and assembly line (product 1)
Output Seru Output Assembly line
Skill level (Units per day) (Units per day)
Data set 1-mixed skills 44.12 48.48
Data set 3-skills 3,4 and 5 45.60 40.67
Data set 6-skill for operation 5 is 7 for 47.24 41.73
all workers
all workers
Data set 8-W1 highly skilled, W3 48.49 52.74
poorly skilled
Data set 9-All skills are 7 83.11 72.72
Data set 10-All skills are 4 45.71 40
69
Output for data set 1, skill level combination 1, 4, 7; when worker 1 is highly skilled
in all task but worker 3 is poorly skilled in all tasks and when all skill level 1-7 is used are
better in assembly line than in Seru. But, Seru has better output than assembly line for skill
levels combination 3, 4, 5, when all worker is highly skilled in operation 5 and 7, when all
are highly skilled and when all are average skilled. On the other hand, output for skill levels
2, 4, 6 is almost same for both Seru and assembly line. From summary Table 23, it is clear
that, difference in skill levels has significant impact on both Seru and assembly line
approach.
4.3 Results Based on Skill for Single Product (Product 2 and Product 3)
In this section, the same experiments as in Section 4.2 is shown for two different
products namely product 2 (Figure 19) and product 3 (Figure 20). Product 2 has 12 tasks
as shown in Figure 19. This time 3 workers and 3 workstations have been considered.
4.3.1 Results for Product 2
This section contains the results for different matrix for product 2.
Figure 19: Precedence diagram for product 2

70
At first result for average skill level for all workers is checked (as in Table 42) and
later several other data set for skill are also considered such as skill level 1 to 7; 2, 4, 6 and
1, 4, 7. In this section an example calculation for skill matrix where all workers has average
skill level for tasks is shown and then the results for other skill matrices are summarized in
Table 45.
In Table 42 it is assumed that all workers have average skill in all the tasks. Later
mathematical model of section 3.3 was run to find the cycle time for assembly line and
result of Seru was compared with it.
Table 42: Skill levels (all average)

Tas Tas Tas Tas Tas Tas Tas Tas Tas Tas Tas Tas
k1 k2 k3 k4 k5 k6 k7 k8 k9 k 10 k 11 k 12
W
4 4 4 4 4 4 4 4 4 4 4 4
1
W
4 4 4 4 4 4 4 4 4 4 4 4
2
W
4 4 4 4 4 4 4 4 4 4 4 4
3
Table 43: Processing time based on skill levels (all average)

Tas Tas Tas Tas Tas Tas Tas Tas Tas Tas Tas Tas Total
k1 k2 k3 k4 k5 k6 k7 k8 k9 k k k of
10 11 12 each
work
er
W 8 69
4 6 7 8 7 5 6 9 2 3 4
1
W 8 69
4 6 7 8 7 5 6 9 2 3 4
2
W 8 69
4 6 7 8 7 5 6 9 2 3 4
3
71

Output for assembly line (skill levels all average): = = = 20.86

units per day.
Output from Seru (skill levels all average): = + + = 20.86 units per day.
Table 44: Worker and task assignment among workstations (all average)
Task 1, 2, 3, 7 Task 4, 6, 9 Task 5, 8, 10, 11, 12
W1 W2 W3
8+4+6+5=23 7+7+9=23 8+6+2+3+4=23
Skill level Output Seru Output Assembly line
(Units per day) (Units per day)
all workers
Data set 7-skill for 28.09 25.61
operation 1 is 7 for all workers
poorly skilled
From the summary Table 45 it is found that in all the cases the output from assembly
line is better than the output from Seru. When average skill level is used then the output is
almost same for both Seru and assembly line which means that when the difference in skill
level is not much Seru and Assembly line system both are same in terms of output. The
72
significance of difference in skill level is more vivid when skill levels 1, 4, 7 only is used.
In that case, the output from assembly line is much better than output from seru.
4.3.2 Results for Product 3
In this section, results for different matrices for product 3 is shown along with a
precedence diagram and an example calculation.
Figure 20: Precedence relation of product 3
Table 46: Skill level 1 to 7 normally distributed (data set 1)

Task 1 Task 2 Task 3 Task 4 Task 5 Task 6 Task 7 Task 8
Worker 1 7 5 4 1 4 3 4 5
Worker 2 4 3 4 6 5 4 2 4
Worker 3 3 4 6 2 4 4 3 4
73
Table 47: Processing time based on worker (skill level 1 to 7 normally distributed for
data set 1)
Task Task Task Task Task Task Task Task Total of
1 2 3 4 5 6 7 8 each
worker
5 6 9 11 7 3 8 7
Worker
2.75 5.1 9 15.95 7 3.45 8 5.95 57.2
1
Worker
5 6.9 9 7.7 5.95 3 10.4 7 54.95
2
Worker
5.75 6 6.3 14.3 7 3 9.2 7 58.55
3
Table 48: Worker and task assignment among workstations (skill level 1 to 7 normally
distributed for data set 1)
Tasks 1, 2, 3 Tasks 4, 5, 6 Tasks 7, 8
W1 W2 W3
2.75+5.1+9=16.85 7.7+5.95+3=16.65 3+8+7=18
Output Seru (Units Output Assembly
Skill level per day) line (Units per day)
all workers
all workers
poorly skilled
Data set 10-All skills are 4 25.71 24
After demonstrating a calculation in Table 46, Table 47 and Table 48 for a skill
matrix the results for other skill matrices are summarized in Table 49.
74
Output for assembly line (Skill level 1 to 7 normally distributed (Data set 1)):

= .
Output from Seru (Skill level 1 to 7 normally distributed (Data set 1)): = +
. .
+ = 25.32 units per day.

.
4.4 Impact of Different Standard Deviation
In this section, three different standard deviation values 5%, 10% and 20% have
been experimented. Then the obtained results are compared. A standard deviation of 15%
is considered in experimentation done in section 3.2 and all other experiments in this study.
While doing this calculation same skill matrix as in section 3.2 is used to maintain the
uniformity of results. Table 50, Table 52 and Table 54 show the operation time for 5%
standard deviation, 10% standard deviation and 20% standard deviation, respectively.
Results for 25% standard deviation is listed in the summary tables.
 When 5% Standard Deviation is used:
Table 50: Operation times based on skill level with σ = 5%

Total time
Task
Task 1 Task 2 Task 3 Task 4 Task 5 of Each
6
worker
Worker 1 9.5 8.4 4.75 7 3 9.45 42.1
Worker 2 10 8.8 5 7.35 3 9 43.15
Worker 3 10 8 5 7.7 3.45 9 43.15
Worker 4 10 7.2 5 5.95 3 8.55 39.7

75
Output Calculation:
In a day regular working hour is 8 hours. So, available time from a worker is =
8×60 minutes = 480 minutes.

= .
Output from Seru: Available time ∗

+

+
+
= .
+ .
+
.
+ .

W1 W4 W2 W3
7.2+5.95=13.15
9.5 (Cycle time) 5+3=8 9
 When 10% Standard Deviation is used:

Total time
Task
Task 1 Task 2 Task 3 Task 4 Task 5 of Each
6
worker
Worker 1 9 8.8 4.5 7 3 9.9 42.2
Worker 2 10 9.6 5 7.7 3 9 44.3
Worker 3 10 8 5 8.4 3.9 9 44.3
Worker 4 10 6.4 5 4.9 3 8.1 37.4

76
Output Calculation:

Output for assembly line =
= .
Output from Seru = + + + = 45.87 units per day.

. . . .

W1 W4 W2 W3
6.4+4.9= 11.3
9 (Cycle time) 5+3=8 9
 When 20% Standard Deviation is used.

Total time
Task 1 Task 2 Task 3 Task 4 Task 5 Task 6 of Each
worker
Worker 1 8 9.6 4 7 3 10.8 42.4
Worker 2 10 11.2 5 8.4 3 9 46.6
Worker 3 10 8 5 9.8 4.8 9 46.6
Worker 4 10 4.8 5 2.8 3 7.2 32.8
Output Calculation:

Output for assembly line =
= = 53.33 units per day.
Output from Seru = = .

+ .
+ .
+ .
77

W1 W4 W2 W3
8 4.8+2.8=7.6 5+3=8 9 (Cycle time)
Table 56, Table 57 and Table 58 shows the summary of results when different
standard deviation value is used.
Table 56: Summary of output of different standard deviation (product 1)

Seru Assembly line
5 % Standard Deviation 45.74 36.50
25% Standard Deviation 47.13 53.33

Seru Assembly line
15 % Standard Deviation 23.27 25
25% Standard Deviation 25.23 29.09

Seru Assembly line
25% Standard Deviation 28.16 32
78
From Table 56, Table 57 and Table 58 it is found that when standard deviation
value 5% and 10% are used the output of Seru is higher than the output from Assembly
line. On the other hand, when 15%, 20% and 25% standard deviation values are used then
Assembly line has better output than Seru. That means the more deviation in skill level the
better it is to use Assembly Line.
4.5 Indexing
In this section, an attempt of indexing is depicted. The objective of the indexing is
to find an insight about selection of manufacturing strategy between Seru and Assembly
line based on skill matrix only. At first indexing is done based on range of skill levels of
tasks. Range is obtained by finding the difference between the maximum skill level and
minimum skill level for that task. For example, in Table 59 the range of the first column is
(5-1) = 4. Later, an indexing attempt based on standard deviation of skill levels are shown
in this section. Table 59 shows an example of range of skill levels for each task and the
sum of the ranges and Table 60 shows an example for Standard Deviation.
Table 61, Table 62 and Table 63 summarize the results for output, sum of ranges
and standard deviation for both Seru and Assembly line obtained from different skill
matrices.
Table 59: Range of skill levels for tasks

Worker 1 5 2 5 4 4 2
Worker 2 4 3 4 3 4 3
Worker 3 3 5 1 4 6 5
Worker 4 1 3 4 6 4 7
Range 4 3 4 3 2 5 Total = 21
79
Table 60: Standard deviation of skill levels

Worker 1 5 2 5 4 4 2
Worker 2 4 3 4 3 4 3
Worker 3 3 5 1 4 6 5
Worker 4 1 3 4 6 4 7
Average
Standard Total of
Deviation 1.71 1.26 1.73 1.26 1.00 2.22 9.17 Standard
Deviation
=1.53
From Table 61, Table 62 and Table 63 it is found that when the value of sum of
ranges higher output for assembly line is higher than Seru. It suggests that when skill level
of workers are more scattered it is better to use assembly line rather than Seru. It is also
found that when the value of total standard deviation is higher assembly line has better
output than Seru.
Table 61: Summary of output and sum of ranges for product 1

Output Sum of Sum of
Seru Output Ranges Standard
(Units per Assembly line Deviation
Skill level day) (Units per day)
9.17
Data set 1 44.12 48.48 21 10.25
Data set 2 45.50 50.52 24 4.09
Data set for 3, 4, and 5 45.60 40.67 9 8.20
Data set for 2, 4, and 6 45.32 45.28 17 13.45
Data set for 1, 4, and 7 48.59 58.18 30 7.05
Operation 5 : all 7 47.24 41.73 15 7.05
Operation 1: all 7 52.73 50.52 12 13.86
W1 highly skilled w3 poorly 48.49 52.74 32 0
skilled
Data set all 7 83.11 72.72 0 0
Data set all 4 45.71 40 0 9.17
80

Seru Output Ranges Standard
(Units Assembly line Deviation
Skill level per day) (Units per day)
Data set 1 20.53 22.96 50 25.43
Data set 2 20.57 22.96 39 20
Data set for 3, 4, and 5 21.02 22.20 19 9.88
Data set for 2, 4, and 6 21.41 22.85 27 14
Data set for 1, 4, and 7 21.52 27.20 54 28.39
Operation 5 : all 7 23.40 24.61 22 11.85
Operation 1: all 7 28.09 25.61 38 20.07
W1 highly skilled w3 poorly 24.30 24.06 60 31.15
skilled
Data set all 7 37.94 34.90 0 0
Data set all 4 20.86 20.86 0 0

Seru Output ranges Standard
(Units Assembly line Deviation
Skill level per day) (Units per day)
Data set 1 25.32 28.48 18 9.61
Data set 2 25.59 27.74 19 9.83
Data set for 3, 4 and 5 25.70 25.73 10 5.46
Data set for 2, 4, and 6 25.87 30.37 16 8.61
Data set for 1, 4, and 7 25.74 27.74 18 9.92
Operation 5 : all 7 26.64 25.13 15 8.02
Operation 1: all 7 25.18 25.19 18 9.66
W1 highly skilled w3 poorly 28.71 26.74 40 21.78
skilled
Data set all 7 46.73 43.63 0 0
One drawback of such an attempt is that these results are dependent on precedence
relation and number workstations used.

81
CHAPTER 5: MULTI-PRODUCT MULTI-PERIOD CASE WITHOUT SKILL
In this chapter, four different products have been considered for four weeks period
where there are five workers and three workstations. Three different combinations of
workers such as one worker, two workers and three workers have been considered. One
worker means that it is a seru unit, two workers means that two different workers are
working on the same product and three workers combination means that three workers are
working in the same product in an assembly line. At first, the cycle time was determined
based only on standard operation time for one worker (which is in other words seru), two
workers and three workers. Cycle time for two workers and three workers are obtained
from the mathematical model described in section 4.1. The cycle times obtained were used
to determine the production rate for different combinations of workers. Later, another
mathematical model is developed to find out the operator distribution over four different
periods for four different products so that the demand of each product is met and the
available number of workers are not exceeded. Section 5.1 describes the methodology for
multi-product multi-period case without skill and section 5.2 contains the results. In section
5.3, yet another methodology is given where the objective is to maximize the output using
the same number of workers obtained from the results of section 5.2. In section 5.2, results
for different value of ‘f’ (percentage of demand that must be produced at each period) are
also shown along.

82
5.1 Methodology for Multi-Product Multi-Period Case (without Skill)
In this chapter, a mathematical model for multi-product and multi-period case has
been developed. Later, four different products have been considered for analysis and some
preliminary results have been shown in section 5.2.
Mathematical model for multi-product multi-period case is presented below.
Indices:
i product index
j period index
k worker index
Parameters:
Pik production rate of product i with k workers
a number of products
b number of worker levels
c number of periods
di demand of product i
f percentage of demand that must be produced at each period
Decision variables:
Xijk Integer value, if product i is chosen for period j with k number of workers
mj Workers required for period j
m Number of workers needed during study period
Objective function:
Minimize, z = m (Equation 9)
83
Subject to:
∑ ∑ × = = 1,2, … , (Equation 10)
≤ = 1,2, … , (Equation 11)
∑ × ⩾ × = 1, 2, … , ; = 1,2, … , (Equation 12)
∑ ∑ × ⩾ = 1, 2 … , (Equation 13)
Here, equation (14) is the objective function which minimizes the manpower level.
Equation (15) calculates the required number of workers for each period. Equation (16) is
to make sure that at each period the total number of required workers does not exceed the
total number of workers. Equation (17) confirms that demand is met or exceeded for each
product. Equation (18) makes sure that at least 20% of each product is produced at each
period.
5.2 Results for Multi-Product Multi-Period without Skill
In this section, cycle times for different products for different worker combination
are calculated. After obtaining cycle time production rate is determined from cycle time.
Figure 21: Precedence relation for product A

84
Figure 21, Figure 22, Figure 23 and Figure 24 show the precedence relation for products
A, B, C and D, respectively.
Table 64: Cycle time for different number of workers for product A
Tasks 1, 2, 3, 4, 5 and 6 Cycle time
Number of worker =
1 Station time = 10+8+5+7+3+9 = 42 42
Task 1, 2, 3 Task 4, 5, 6
Number of workers 23
=2 10+8+5 =23 7+3+9 =19
Task 1, 3 Task 2, 4 Task 5, 6

Number of workers 15
=3 10+5 = 15 8 + 7 = 15 9 + 3 = 12
Figure 22: Precedence relation for product B

85
Table 65: Cycle time for different number of workers for product B
Number of worker = Tasks 1, 2, 3, 4 and 5 Cycle time
1
Station time = 7+5+8+10+6 = 36 36
Number of workers Tasks 1, 2, 3 Tasks 4, 5

=2 20
7+5+8=20 10+6=16
Number of workers Tasks 1, 3 Tasks 2, 4 Tasks 5

=3 15
7+8=15 10+5=15 6
Figure 23: Precedence relation for product C

86
Table 66: Cycle time for different number of workers for product C
Number of worker = Tasks 1, 2, 3, 4 and 5 Cycle time
1
Station time = 6+4+3+7+8=28 28
Number of workers Task 1, 2, 3 Task 4, 5

=2 15
6+4+3=13 7+8=15
Number of workers Task 1, 3 Task 2, 4 Task 5

=3 10
6+4=10 7+3=10 8
Figure 24: Precedence Relation of Product D

87
Table 67: Cycle time for different number of workers for product D
Number of worker = Tasks 1, 2, 3 and 4 Cycle time
1
Station time = 5+6+8+7=26 26
Number of workers Task 1, 2 Task 3, 4

=2 15
5+6=11 7+8=15
Number of workers Task 1, 2 Task 3 Task 4

=3 11
5+6=11 8 7
Table 68: Summary for cycle time

Product CT for 1 worker CT for 2 workers CT for 3 workers
A 42 23 15
B 36 20 15
C 28 15 10
D 26 15 11
Table 68 shows the cycle times (CT) found from different combinations of worker
levels for products A, B, C and D.
Table 69: Production rate from cycle time

Product Weekly Weekly Weekly
production rate production rate roduction rate for
for 1 worker for 2 workers 3 workers
A 57.14 104.34 160
B 66.66 120 160
C 85.71 160 240
D 92.30 160 218.18
88
In Table 69, production rates are calculated from cycle times. In a week normal
working time for a single shift schedule is 40 hours. So, there are 40×60 minutes = 2400
minutes. As a result, weekly production rate for product A when 1 worker is used (in other
words, 1 Seru unit) is = = 57.14.
After obtaining production rates for different products for different worker
combinations, methodology of section 5.2 is used to find the number of workers for
different products in different periods. Three different demand data sets with different
value of f (percentage of demand that must be produced at each period) is experimented
with to check the impact on assignment and number of workers for varying demand. At
first f= 20% was applied and later f= 0%, 5%, 10%, 15% and 25% was also checked. This
approach is followed to observe the impact of forced production on workers requirement
and worker assignment for different products on different period. Demand values are
obtained randomly from uniform distribution. For example, demand value in data set 1 for
product A, B, C and D are 500, 600, 800 and 820, respectively. Three different worker
combinations are considered namely 1 worker, 2 workers and 3 workers. 1 worker option
means 1 seru unit, 2 workers option means an assembly line that consists of two workers
and 3 workers option means an assembly line where 3 workers work in the same line. In
the notation (m×n) the first digit indicates which option and second digit indicates how
many of those option are required.

89
Table 70: Manpower distribution for multi-period (demand data set 1)

Period 1 Period 2 Period 3 Period 4 Demand
of
Products
Product A (1×2) = (1×1), (2×1) (1×2) = 2 (1×2) = 2 500
2 =3
Product B (1×2) = (1×3) = 3 (1×2) = 2 (1×3) = 3 600
2
Product C (1×2) = (1×2) = 2 (1×4) = 4 (1×2) = 2 800
2
Product D (1×3) = (1×2) = 2 (1×2) = 2 (1×1), (2×1) 820
3 =3
Total Manpower, 9 10 10 10
mj (Each period)

of
Products
Product A (1×2) = 2 (1×2) = 2 (1×2) = 2 (1×2) = 2 300
Product B (1×1) = 1 (1×1) = 1 (1×1) = 1 (1×1) = 1 200
Product C (1×1), (2×1) = (1×3) = 3 (1×3) = 3 (2×2) = 3 1000
3
Product D (2×1) = 2 (2×1) = 2 (2×1) = 2 (1×2) = 2 500
Total Manpower 8 8 8 8
mj (Each period)
Thus, notation (1×3) means 3 Seru units. Table 70, Table 71 and Table 72 show the
results found for manpower distribution for different periods for different demand data sets.
90

of
Products
Product A (1×4), (3×1) (1×6) = 6 (1×6), (2×1) (1×4), (2×1) 1500
=7 =8 =6
Product B (1×4) = 4 (1×5) = 5 (1×4) = 4 (1×4), (2×1) 1200
=6
Product C (1×4) = 4 (1×2), (2×1) (1×4) = 4 (1×4) = 4 1400
=4
Product D (1×2), (2×1) (1×4) = 4 (1×1), (2×1) = (1×1), (2×1) 1000
=4 3 =3
Total 19 19 19 19
Manpower
(Each
period) mj
From Table 70 it is it is found that product A needs 2 seru units in period 1, in
period 2 it needs 1 seru unit and 1 assembly line of 2 workers and in both period 3 and 4 it
needs 2 seru units. In Table 71 and Table 72 different demand data is used to see how
worker assignment is affected for different products in a dynamic demand environment.
Table 73, Table 74 and Table 75 show the number of workers required at each period when
different ‘f’ values are used for different demand values. From these results it is found that
when units equivalent to 25% of total demand (equal product at each period since there are
4 periods) is forced to produce at each period number of required worker increases for all
demand data set.

91
Table 73: Summary table for different f values of data set 1

Demand of Demand of Demand of Demand of
Product A Product B Product C Product D
500 600 800 820
Number of Number of Number of Number of

worker worker worker worker
required at required at required at required at
Period 1 Period 2 Period 3 Period 4
f = 0% 10 10 9 10
f = 5% 10 10 10 10
f = 10% 10 10 9 10
f = 15% 10 10 10 10
f = 20% 9 10 10 10
f = 25% 12 12 12 12

300 200 1000 500

f = 0% 7 7 7 7
f = 5% 7 7 7 7
f = 10% 7 7 7 7
f = 15% 7 7 7 7
f = 20% 8 8 8 8
f = 25% 8 8 8 8
92
At this point this is a managerial decision to make whether to produce different
amount at each period by reducing worker or produce equal amount of units at each period
by increasing number of worker without violating demand constraints.

1500 1200 1400 1000

f = 0% 19 19 19 19
f = 5% 19 19 18 19
f = 10% 19 19 19 19
f = 15% 19 19 18 19
f = 20% 19 19 19 19
f = 25% 20 20 20 20
This study also aims to check the results when worker skills are considered for
different tasks.
5.3 Methodology for Multi-Product Multi-Period to Maximize Output
Methodology described in section 5.2 is designed to minimize the number of
workers so that demand is met for each product. After obtaining the minimum required
number of worker (mj) for specific demand now the objective is to maximize the output
utilizing the same number of workers. In this section, a methodology is used to maximize
the output where number of workers are known from the result of section 5.2.
93
Indices:
i product index
j period index
k worker index
Parameters:
Pik production rate of product i when there are k workers
a number of products
b number of workers
c number of periods
di demand of product i
Decision variables:
Xijk Integer value, if product i is chosen for period j with k number of workers;
0, otherwise
mj Workers required for period j
m Total available worker
Objective function:
Maximize, =∑ ∑ ∑ ( × ) (Equation 14)
Subject to:
∑ ∑ × = = 1,2, … , (Equation 15)
∑ ∑ × ⩾ = 1, 2, … (Equation 16)
∑ × ⩾ × = 1, 2, … , ; = 1,2, … , (Equation 17)

94
Here, equation (19) is the objective function which maximizes the output. Equation
(20) calculates required number of workers at each period. Equation (21) makes sure that
demand is met or exceeded for each product. Equation (22) confirms that at least f% unit
of demand must be produced at each period.
5.4 Results for Output Maximization (Multi-Product Multi-Period without Skill Case)
After obtaining minimum worker level for products based on their demand in
section 5.2, methodology in section 5.3 is used to maximize output utilizing the same
number of worker. For example, in Table 70 the total demand for four products is
500+600+800+820=2720 units and at least 10 workers are needed to produce 2720 units.
The mathematical model in section 5.3 is designed such a way so that the output for four
products can be maximized using these 10 workers. For demand data set 1, required
demand of 2720 units can be filled using 10 workers. On the other hand, using the revised
methodology it is found that these 10 workers can produce 3053 units of product which is
3053-2720=333 units more than the actual demand.
This gives the flexibility of choosing whether to produce more products and hence
increase the demand through pricing and promotion or not.

95
Table 76: Demand data set 1

Demand of Unit Fixed Unit Varying
Products produced workforce produced workforce at
at each each period
period
Product 500 514.26 Period 1= 514.26 Period 1= 9

A 10
B 10
C 10
D 10
Total 2720 3053.26 2960.96
Table 77: Manpower distribution for multi-period from mathematical model 3 (demand
data set 1) with f = 20%
of
Products
Product A (1×2)=2 (1×3)=3 (1×2)=2 (1×2)=2 500
Product B (1×3)=3 (1×2)=2 (1×2)=2 (1×3)= 3 600
Product C (1×2)=2 (1×2)=2 (1×3)=3 (1×3)=3 800
Product D (1×3)=3 (1×2) =2 (1×2) =2 (1×1), (2×1) 820
=3
Total 9 10 10 10
Manpower, mj
(Each period)
96

Demand of Products Unit produced Worker utilized at each
period
Product A 300 457.12 Period 1= 8
Product B 200 266.64 Period 2= 8
Product C 1000 1028.52 Period 3= 8
Product D 500 738.40 Period 4= 8
Total 2000 2490
data set 2) with f = 20%
Period 1 Period 2 Period 3 Period 4 Demand of
Products
Product A (1×2) = 2 (1×2) = (1×2)=2 (1×2)=2 300
2
Product B (1×1) = 2 (1×1) = (1×1)=1 (1×1)=1 200
1
Product C (1×3) = 3 (1×3) = (1×3)=3 (1×3)=3 1000
3
Product D (1×2)=2 (1×2)=2 (1×2)=2 (1×2)= 2 500
Total 8 8 8 8
Manpower, mj
(Each period)

Demand of Products Unit produced Worker utilized
at each period
Product A 1500 1542.78 Period 1= 19
Product B 1200 1266.54 Period 2= 19
Product C 1400 1457.07 Period 3= 19
Product D 1000 1199.9 Period 4= 19
Total 5100 5466.29
97
data set 3) f=20%
Period 1 Period 2 Period 3 Period 4 Demand of
Products
Product A (1×6)=6 (1×8)=8 (1×6)=6 (1×7)=7 1500
Product B (1×6)=6 (1×4)=4 (1×5)=5 (1×4)=4 1200
Product C (1×4)=4 (1×4)=4 (1×4)=4 (1×5)=5 1400
Product D (1×3)=3 (1×3)=3 (1×4)=4 (1×3)=3 1000
Total 19 19 19 19
Manpower
(Each
period) mj
Results for these three different demand data are shown in Table 76, Table 78 and
Table 80 obtained after implementing the maximizing output methodology (methodology
5.3). In this experiment, only f=20% value is used. Worker assignment after running
mathematical model 3 are shown in Table 77, Table 79 and Table 81. From these result it
is observed that output can be increased by utilizing same number of worker. Worker
assignment are also known when output is needed to be increased for different products.
98
CHAPTER 6: MULTI-PRODUCT MULTI-WORKER ASSIGNMENT CONSIDERING
SKILLS AND LEARNING, FORGETTING RATES
In this chapter, worker assignment for multi-product case considering skills is
studied for a single period. In previous chapters, it is found that when skill is not considered
most of the time Seru is being selected as the manufacturing strategy. This chapter aims to
check the outcome in terms of manufacturing strategy and worker assignment when skill
is considered. Three products and five different workers are considered for numerical
experimentation. In this chapter, an experimentation is also done for learning and forgetting
for multi-period.
In section 6.1, a methodology is described how skill is considered for multi-product
worker assignment. Then, in section 6.2, results for single period are shown. Section 6.3
contains a methodology on how to maximize output utilizing minimum number of workers
when skill is considered and section 6.4 contains results based on methodology of section
6.3. In section 6.5, a discussion is presented on how to implement the methodologies of
section 6.1 and 6.2 in a cellular manufacturing environment. At the end, in section 6.6 an
experimentation is illustrated based on learning and forgetting rates for eight consecutive
periods.
6.1 Methodology for Multi-Product Worker Assignment Considering Skills
In this section, a mathematical model is described where multi-product multi-
worker case is considered with skill. Options for number of workers are limited to 1-
worker, 2-worker and 3-worker in this chapter.

99
6.1.1 Mathematical Model: 4
Indices:
i product index
j worker combination index
k worker index
Parameters:
n number of products
m1 number of one worker combinations for each product.
m2 number of two worker combinations for each product.
m3 number of three worker combinations for each product.
m number of combinations for each product (m1+m2+m3).
Pij Production rate of product i when assigned with worker combination j
Di Demand of product i
Cik 1, if worker k is available at combination j
Decision variables:
Xij 1, if product i is assigned with worker combination j ; 0, otherwise
Objective Function:
Minimize, = 1× ∑ ∑ + 2× ∑ ∑ +
3 × ∑ ∑ (Equation 18)
Subject to:
× ≥ ( = 1, 2, … , ) (Equation 19)
100
∑ ∑ × = 1 ( = 1, 2, … , ) (Equation 20)
Equation (18) is the objective function. The objective of this model is of
minimization type. This equation ensures that minimum number of workers is utilized.
Equation (19) makes sure that demand of each product is met or exceeded. Equation (20)
confirms that a worker is not assigned in two or more products at the same time.
6.2 Results for Multi-Product Worker Assignment Considering Skill
Three different products namely, A, B and C are considered. There are 5 different
workers and 3 workstations. Skill matrix for all three products are generated randomly
maintaining normal distribution among the skill levels. Table 82, Table 83, and Table 84
show the skill matrix for Product A, Product B and Product C, respectively. Three different
combinations of workers are considered namely, 1-worker option, 2-worker option and 3-
worker option. One worker means either worker 1 or worker 2 or worker 3 or any of the
worker will work alone on a product independently to do all of the tasks of that product
which means a Seru unit. Two workers combination implies that two workers will work
on the same product in an assembly line where the tasks will be assigned between them
based on their skill and three workers means an assembly line consisting of three workers
where the tasks are assigned among them.
At first, skill matrix for all the products are generated. Since there are five workers,
possible combinations for 1 worker option are w1 (Worker 1), w2, w3, w4 and w5; possible
combinations for 2-worker options are w1w2 (worker1 and worker 2 together), w1w3,
w1w4, w1w5 …… w4w5 (10 different combinations) and there are 10 different
combinations possible for 3-worker option for example w1w2w3 (worker 1, worker 2 and
101
worker 3 together), w1w2w4, w1w4w5, w2w3w4 …… w3w4w5. After that, cycle time
for all different combinations are calculated using the methodology of section 4.1. After
obtaining cycle time for different combinations, daily production rate is calculated by
dividing 480 by the cycle time of that combination. For example, daily production rate for
Product A when only worker 2 is considered is 480/41.55 = 11.57 units. All these
production rates are then used as input into the mathematical model given in section 6.2.
The mathematical model gives us the answer of which combination to use for which
product and also which task to be assigned to which worker based on skill. Figure 25 shows
the precedence relation for Products A, B and C.
Figure 25: Precedence relations for product A, B and C

102
Table 82: Skill matrix for product A

Worker 1 7 5 1 4 6 3
Worker 2 4 3 4 6 3 4
Worker 3 3 1 4 4 4 4
Worker 4 4 5 6 2 4 7
Worker 5 2 5 5 4 2 4
Table 83: Skill matrix for product B

Task 1 Task 2 Task 3 Task 4 Task 5
Worker 1 5 4 7 4 2
Worker 2 4 5 2 1 4
Worker 3 4 4 5 3 4
Worker 4 6 4 4 4 3
Worker 5 3 4 1 7 6
Table 84: Skill matrix for Product C

Worker 1 4 5 4 5 4
Worker 2 1 4 2 7 4
Worker 3 4 4 4 3 6
Worker 4 6 7 3 4 4
Worker 5 2 4 1 5 3
Following tables show different combinations and worker assignment along with
cycle time when that combination is chosen.

103
i. Product A
Table 85: Product A, cycle time for 1-worker option (Seru Unit)
1-Worker Option Station 1 Cycle time
w1 w1—1, 2, 4, 4, 5, 6 39
w2 w2—1, 2, 4, 4, 5, 6 41.55
w3 w3—1, 2, 4, 4, 5, 6 47.1
w4 w4—1, 2, 4, 4, 5, 6 37.35
w5 w5—1, 2, 4, 4, 5, 6 43.95
Table 86: Product A, cycle time for combinations for 2-worker options
2-Worker Station 1 Station 2 Cycle time
Combination
w1-w2 w1—1, 2, 4 w2 – 3, 5, 6 19.30
w1-w3 w1—1, 2, 4 w3 – 3, 5, 6 19.30
w1-w4 w1—1, 2, 4 w4 – 3, 5, 6 19.30
w1-w5 w1—1, 2, 4 w5 – 3, 5, 6 19.30
w2-w3 w3 - 1, 2, w2 - 3, 4, 5, 6 23.1
w2-w4 w4 – 1, 2, 3 w2- 4, 5, 6 20.3
w2-w5 w5 - 1, 2, w2 - 3, 4, 5, 6 22.35
w3-w4 w4 – 1, 2, 3 w3 - 4, 5, 6 20.3
w3-w5 w3 - 1, 3, 5 w5 - 2, 4, 6 22.80
w4-w5 w4 - 1, 2, 3 w5 - 4, 5, 6 20.3
104
Table 87: Product A, cycle time for combinations of 3-worker option

3-Worker Station 1 Station 2 Station 3 Cycle time
Combination
w1- 1, 2, w2 - 3,4 w3- 5,6 12.30
w1-w2-w3
w1 – 1,2 w2 – 3,4 w4 – 5, 6 12.30
w1-w2-w4
w1 – 1,2 w5 – 3,4 w2 – 5, 6 12.45
w1-w2-w5
w1 – 1,2 w3 - 4 w4 – 3, 5, 6 12.30
w1-w3-w4
w1 – 1, 2 w5 – 3, 4 w3 – 5, 6 12.30
w1-w3-w5
w1 – 1, 2 w5 – 3, 4 w4 – 5, 6 12.30
w1-w4-w5
w3 – 1 w2 – 2, 4 w4 – 3, 5 6 14.10
w2-w3-w4
w2 – 1, 3 w5 – 2, 4 w3 - 5, 6 15
w2-w3-w5
w5 -1 w2 – 2, 4 w4 - 3, 5, 6 14.10
w2-w4-w5
w3 – 1 w5 - 2, 4 w4 - 3, 5, 6 13.8
w3-w4-w5
ii. Product B
Table 88: Product B, cycle time for 1-worker option (Seru unit)
1-Worker Option Station 1 Cycle time
w1 w1—1, 2, 4, 4, 5 33.15
w2 w2—1, 2, 4, 4, 5 42.15
w3 w3—1, 2, 4, 4, 5 36.3
w4 w4—1, 2, 4, 4, 5 34.8
w5 w5—1, 2, 4, 4, 5 34.35
105
Table 89: Product B, cycle time for combinations of 2-worker option

Combination
w1-w2 w2- 1, 2 w1-3, 4, 5 20.40
w1-w3 w1-1,2,3 w3 - 4,5 17.50
w1-w4 w1 - 1,2,3 w4 - 4,5 16.00
w1-w5 w1 - 1,2,3 w4 - 4,5 15.35
w2-w3 w3- 1, 2, 3 w2- 4, 5 20.5
w2-w4 w2 – 1, 3 w4 – 2, 4 , 5 21
w2-w5 w2 – 1, 3 w5 – 2, 4 , 5 17.40
w3-w4 w4- 1, 2, 3 w3- 4, 5 18.4
w3-w5 w3 – 1, 3 w5 – 2, 4 , 5 16.5
w4-w5 w4 - 1, 3 w5- 2,4, 5 17.5
Table 90: Product B, cycle time for combinations of 3-worker option

3-Worker Station 1 Station 2 Station 3 Cycle time
Combination
w1- 1, 2, 3 w3 - 4 w2 - 5 15.35
w1-w2-w3
w1 – 1,3 w4 – 2,4 w2 – 5 15
w1-w2-w4
w2 – 1 w1– 2, 3 w5 – 4, 5 9.7
w1-w2-w5
w1- 1, 2, 3 w4 - 4 w3 - 5 15.35
w1-w3-w4
w3 – 1 w1 – 2, 3 w5 – 4, 5 9.7
w1-w3-w5
w4 – 1 w1 – 2, 3 w5 – 4, 5 9.7
w1-w4-w5
w3 – 1, 3 w4 – 2, 4 w2 – 5 15
w2-w3-w4
w2 – 1, 2 w3 – 3 w5 – 4, 5 11.25
w2-w3-w5
w4 -1, 2 w2 – 3 w5 – 4, 5 10.4
w2-w4-w5
w4 – 1, 2 w3 - 3 w5 – 4, 5 9.9
w3-w4-w5
106
iii. Product C
Table 91: Product C, cycle time for 1-worker option (Seru Unit)
1-Worker Station 1 Cycle time
Option
w1 w1—1, 2, 4, 4, 5 26.35
w2 w2—1, 2, 4, 4, 5 28.45
w3 w3—1, 2, 4, 4, 5 26.65
w4 w4—1, 2, 4, 4, 5 24.85
w5 w5—1, 2, 4, 4, 5 31.3
Table 92: Product C, cycle time for combinations of 2-worker option

Combination
w1-w2 w1 – 1, 2, 3 w2 – 4, 5 12.40
w1-w3 w3 -1, 2, 3 w1 – 4, 5 13
w1-w4 w1 -1, 2, 3 w4 – 4, 5 13
w1-w5 w1 – 1, 2, 3 w5 – 4, 5 12.4
w2-w3 w3- 1, 2, 3 w1- 4, 5 13
w2-w4 w4- 1, 2, 3 w2- 4, 5 11.85
w2-w5 w5 - 1, 2, 3 w2 – 4,5 16.15
w3-w4 w3- 1, 2, 3 w4 – 4, 5 13
w3-w5 w3 – 1, 2, 3 w5 – 4, 5 13
w4-w5 w4 – 1, 2, 3 w5 – 4, 5 11.95
107
Table 93: Product C, cycle time for combinations of 3-worker option

3-Worker Station 1 Station 2 Station 3 Cycle
Combination time
w1- 1, 2 w2 - 4 w3 – 3, 5 9.4
w1-w2-w3
w4 – 1, 2 w1 – 3, 4 w2 – 5 8.95
w1-w2-w4
w1- 1,2 w2 – 3, 4 w5 – 5 9.4
w1-w2-w5
w4 - 1, 2, 3 w1 - 4 w3 - 5 9.85
w1-w3-w4
w3 – 1, 2 w1 – 3, 4 w5 – 5 10
w1-w3-w5
w4 - 1, 2, 3 w1 - 5 w5 - 4 9.85
w1-w4-w5
w4 – 1, 2 w2 – 3, 4 w3 – 5 7.75
w2-w3-w4
w3 – 1, 2 w2 – 3, 4 w5 - 5 10
w2-w3-w5
w4 -1, 2, 3 w5 – 4 w4 - 5 9.85
w2-w4-w5
w4 – 1, 2, 3 w3 - 4 w5 – 5 9.85
w3-w4-w5
6.2.1 Results from Math Model of Section 6.2
Table 95, Table 96, Table 97 and Table 98 show the production system for different
products for different demand value. For demand data set 1, product A needs an assembly
line of worker 2 and 3. Task distribution for product A is known from previous result which
is shown in Table 94. In this case, workers 3 is doing tasks 1 and 2 and worker 5 is doing
tasks 3, 4, 5, and 6. On the other hand, Product B and Product C needs one seru unit for
each where Worker 2 works on product B and Worker 5 works on product C independently.
Demands values are obtained randomly from uniform distribution [5, 19].
From these results of different demand values, it is observed that seru is selected
for some product and assembly line is selected for other.

108

Product Worker Combination Demand Output
(Daily) Rate
(Units)
Product A Workers 2, 4 (Assembly line) 16 23.64
W4 – 1, 2, 3 W2 – 4, 5, 6
Product B Workers 3, 5 (Assembly line) 17 29.09

W3 – 1, 3 W5 – 2, 4, 6
Product C Worker 1 (Seru) 19 20.01
Total = 5 workers.

Product Worker Combination Demand Output Rate
(Daily) (Units)
W4 – 1, 2, 3 W5 – 4, 5, 6
Product B Worker 1 (Seru) 10 14.34

Total = 4 workers

(Daily) (Units)
W4 – 1, 2, 3 W3 – 4, 5, 6
Product B Worker 2 (Seru) and Worker 5 (Seru) 17 25.35

Total = 5 workers.
109

(Daily) (Units)

W4 – 1, 2, 3 W5 – 4, 5, 6
Product B Worker 1 (Seru) 10 14.50

Total = 4 workers.

(Daily) Rate
(Units)
Product A Worker 2 (Seru) 8 12.30
Product B Workers 1, 3, 5 (Assembly line) 32 49.48
W3 – 1 W1 – 2, 3 W5 – 4, 5
Total = 5 workers.
From the results of Table 94 it can be observed that product A and B both need
assembly line of two workers and product C needs one unit of Seru. From
Table 95, Table 96, Table 97 and Table 98 it is also found that when skill is
considered then a both Seru and assembly line system can exist in the production system.
From the results, it can also be observed that workers are being selected for the
tasks where their skill levels are comparatively better than other workers for the tasks they
are assigned to. For example, from Table 94 it can be observed that product A needs an
assembly line of worker 2 and worker 4. In this case, worker 4 has skill levels of 4, 5 and
110
6 for tasks 1, 2 and 3, respectively which are better than other workers for these tasks.
Worker 2 has skill levels of 6, 3 and 4 for tasks 4, 5 and 6, respectively. In this case, it is
also one of the best skill sets for those tasks. Similarly, for product B it is found that worker
3 is assigned for tasks 1 and 3 with skill levels 4 and 5, respectively, and worker 5 is
assigned with tasks 2, 4 and 5 with skill levels 4, 7 and 6, respectively. Finally, for product
C worker 1 is alone working on all the tasks where worker 1 has skill levels 4, 5, 4, 5 and
4. This is because worker 1 has almost average level of skill levels for all the tasks. So,
methodology described in section 6.1 not only gives the tasks assignment for multi-product
and multi-worker scenario but also assigns the workers in best possible way to meet the
demand of the products. It can also be said that when worker skill is considered there may
a hybrid system of be assembly line and Seru as production strategy.
6.3 Maximizing Output for Multi-Product Case with Skill
In this section, a methodology is presented for multi-product case with skill where
the objective is to maximize output utilizing same number of workers found from the
results of Section 6.2. Methodology in Section 6.1 minimizes the number of workers while
meeting or exceeding the demand. From that mathematical model we know exactly how
many workers are needed for a given demand data set. The objective of this methodology
is to check if output can be maximized once the minimum number of worker needed is
known for specific demand data set.
6.3.1 Mathematical Model: 5
Indices:
i product index
111
j worker combination index
k worker index
Parameters:
n number of products
m1 number of one worker combinations for each product.
m2 number of two worker combinations for each product.
m3 number of three worker combinations for each product.
m number of combinations for each product (m1+m2+m3).
Pij Production rate of product i when assigned with worker combination j.
Di Demand of product i.
W Minimum number of workers needed to meet the demand.
Cik 1, if worker k is available at combination j.
Decision variables:
Xij 1, if product i is assigned with worker combination j; 0, otherwise.
Objective Function:
Maximize, =∑ ∑ × (Equation 21)
Subject to,
1× ∑ ∑ + 2× ∑ ∑ + 3 × ∑ ∑ ≤
(Equation 22)
× ≥ ( = 1, 2, … , ) (Equation 23)
∑ ∑ × = 1 ( = 1, 2, … , ) (Equation 24)
112
Equation (26) is the objective function. The objective of this model is to maximize
output. Equation (27) ensures that number of workers does not exceed minimum number
of workers. Equation (28) makes sure that demand is met or exceeded and equation (29)
confirms that a worker is not assigned in two or more products at the same time.
6.4 Results for Output Maximization Model for Multi-Product with Skill Case
In this section, results for same demand data set used in section 6.2 are shown. From
Table 99 it is found that using the methodology in of section 6.3 output can be increased
using the same number of workers. In this case total actual demand is 52 and unit produced
is 72, which is 72-52=20 units more than actual demand. Worker assignment has also
changed in this case. In Table 94 workers 2 and 4 worked on product A. But once output
maximization model has run workers 1 and 2 are assigned for product A. Worker 2 is
replaced by worker 4 for product C and worker assignment for product B remained
unchanged. In summary Table 100 four more demand data sets and their results are shown
along with demand data set 1.
Table 99: Result for demand data set 1

(Daily) Rate
(Units)
Product A Workers 1, 2 (Assembly line) 16 24
W1 – 1, 2, 4 W2 – 3, 5, 6
Product B Workers 3, 5 (Assembly line) 17 29

W3 – 1, 3 W5 – 2, 4, 6
Product C Worker 4 (Seru) 19 19

Total = 5 workers. 52 72
113
From summary Table 100, it can be observed that production can be increased using
the same number of workers. This gives the flexibility to the managers of choosing between
two options i. Produce exact amount that they needed and ii. Increase the production using
capacity to take the advantage of pricing.
Table 100: Summary of results for output maximization

Actual Total Units Total Production
Demand Produced increased
(Daily) by
Product 16 24
Data set A 52 72 20
1 Product 17 29
B
Product 19 19
C
Product 13 24
Data set A 39 77 38
2 Product 10 13
B
Product 16 40
C
Product 19 24
Data set A 52 73 21
3 Product 17 31
B
Product 16 18
C
Product 20 24
Data set A 40 79 39
4 Product 10 13
B
Product 10 40
C
Product 8 11
Data set A 49 80 31
5 Product 32 49
B
Product 9 19
C
114
6.5 Implementing Multi-Product and Multi-Worker Assignment with Skill in Cellular
Manufacturing Environment
In a cellular manufacturing environment, at first, products are grouped into families
based on similarities of their operations. Then families are assigned to their cells and
required operations are performed. After that, products are sent to assembly area.
Methodologies described in section 6.1 and 6.3 can be applied for assembly area of a
cellular manufacturing plant given that product family and their corresponding cells are
known. Suppose, a company is producing 10 different products A, B, C, D, E, F, G, H, I
and J as shown in Figure 26. After applying grouping technique, families for these products
are known. For example, Product A, B and C are in family 1, Products D, E, F are in family
2 and Products G, H, I, J are in family 3.
After applying cell loading techniques, families can be assigned to their specific
cells. In this study, it is assumed that product family and their cells are known. Now,
methodologies of section 6.1 and 6.3 can be implemented to assign workers for each
product family separately in their assembly area.

115
Figure 26: A cellular manufacturing environment
6.6 Learning and Forgetting Effect
In real life, when workers are assigned to some specific tasks in an assembly line
and if they keep doing those tasks continuously for certain periods it is likely that their
performance will increase on those tasks. In other words, their skill for those tasks will
increase. This happens because, naturally people learn over time. But, increase of skill
depends on some factors such as age, training, learning rate, environment and training. On
the other hand, forgetting also takes place, when a worker doesn’t do a task for a long
period of time. Meaning that their skill on those tasks reduces. In contrast with assembly
line, Seru has advantage of assigning workers to all tasks. In this case, workers get the
chance to improve their skill for all the tasks. But, since they work on all the tasks of a
product, the frequency of working on a specific task is lower than the frequency of working
on a task by a worker in assembly line. As a result, a worker assigned in an assembly line
can improve skill on a specific task quicker than a worker assigned in Seru. Apart from
116
that, initially, if workers are all assigned to Seru demand cannot be met because from
previous experiments of this study it is found that when standard deviation of skill levels
are high it is better to use assembly line. In this section, an experimentation is done
considering learning and forgetting effect of workers.
Table 101 shows the learning and forgetting rates. Learning rates are determined
based on number of times a worker work on a specific task. Süer and Tummaluri (2008)
suggested that learning rates are period dependent. In this study, it is assumed that learning
rates depend on frequency rather than period, that means, if a worker works on a task so
many times his skill will increase. On the other hand, forgetting rates in this study are
period-dependent as suggested by Süer and Tummaluri (2008).
Table 101: The learning and forgetting rates

Learning Forgetting
Skill Level Number of Probability Skill Level Periods Probability
times a task
needs to be
performed
From To From To
1 2 120 0.70 2 1 4 0.70
2 3 150 0.65 3 2 4 0.65
3 4 150 0.60 4 3 5 0.60
4 5 200 0.55 5 4 5 0.50
5 6 200 0.50 6 5 6 0.40
6 7 250 0.30 7 6 7 0.30
For example, for a skill level to increase from level 1 to 2 (with a 70% probability),
a worker needs to perform a task for 120 times. In other words, he needs to produce 120
units of that product. On the other hand, a skill level decreases from 2 to 1 (with a 70%
117
probability) means that, that worker did not perform that task for 4 consecutive periods.
Moving from higher skill levels takes more work (learning) and more periods (forgetting)
because at some point learning rate saturates and forgetting a task at which a worker is very
good also takes longer time. For the same reason probability values are lower in this case.
Some other assumptions are also considered for this experiment which are listed
below.
i. One period means one week and each week has 5 working days.
ii. If a worker does a task and he produces ‘x’ units but remains unassigned for
that task for four consecutive periods, his learning rate will be equivalent to
‘x/2’ units from next period.
iii. After an increase or decrease in skill level, it is assumed that, that worker will
start from fresh from the next period at the new skill level.
iv. Demand of products is different at different periods. Daily demand is obtained
randomly from uniform distribution [5, 19].
In this study, experimentation over 8 consecutive periods is carried out. After each
period a check list for each worker is done to keep track of their performance. Table 102
shows worker assignment and output calculation at period 1. Daily demand and output is
converted into weekly demand and output by multiplying by 5. Initial skill matrices are the
same as in section 6.2. Worker assignment is found from methodology described in section
6.1.
118
Table 102: Worker assignment at period 1

Demand Task Task Task Task Task Task Number
(Unit) 1 2 3 4 5 6 of times
operations
performed
at the end
of the
period
Worker
1
Product 16×5=80 Worker 20×5=100
A 2
Worker
3
Worker 20×5=100
4
Worker
5
Worker N/A
1
Product 17×5=85 Worker N/A
B 2
Worker N/A 28×5=140
3
Worker N/A
4
5
Worker N/A 19×5=95
1
C 2
Worker N/A
3
Worker N/A
4
Worker N/A
5
119
Table 103: Check sheet of worker’s task performance after period 1

Product Worker Task 1 Task 2 Task 3 Task 4 Task 5 Task 6
Worker 1 - - - - - -
Worker 2 - - - 100 100 100
Product A Worker 3 - - - - - -
Worker 4 100 100 100 - - -
Worker 5 - - - - - -
Worker 1 - - - - - N/A
Worker 2 - - - - - N/A
Product B Worker 3 140 - 140 - - N/A
Worker 4 - - - - - N/A
Worker 5 - 140 - 140 140 N/A
Worker 1 95 95 95 95 95 N/A
Worker 2 - - - - - N/A
Product C Worker 3 - - - - - N/A
Worker 4 - - - - - N/A
Worker 5 - - - - - N/A
From Table 103, it is found that worker 4 performed tasks 1, 2 and 3 of Product A
for 100 times and worker 2 did tasks 4, 5 and 6 of the same product for 100 times. The
same calculations are done for Products B and C. A single (-) sign and a double (--) sign
means that a worker on that task remain unassigned for one period and two periods
consecutively. Since no learning or forgetting take place after period 1, the initial matrices
remain unchanged.
120
Table 104: Worker assignment at period 2

Demand Task Task Task Task Task Task Number of
(Unit) 1 2 3 4 5 6 times
operations
performed
at the end
of the
period
Worker
1
Product 13×5=65 Worker
A 2
Worker
3
Worker 115
4
Worker 115
5
Worker N/A 70
1
B 2
Worker N/A
3
Worker N/A
4
Worker N/A
5
Worker N/A
1
C 2
Worker N/A 90
3
Worker N/A
4
Worker N/A
5
121
Table 105: Check sheet of worker’s task performance after period 2

Worker 1 -- -- -- -- -- --
Worker 2 -- -- -- 100(-) 100(-) 100(-)
Product A Worker 3 -- -- -- -- -- --
Worker 4 215 215 215 -- -- --
Worker 5 -- -- -- 115 115 115
Worker 1 70 70 70 70 70 N/A
Worker 2 -- -- -- -- -- N/A
Product B Worker 3 140(-) -- 140(-) -- -- N/A
Worker 4 -- -- -- -- -- N/A
Worker 5 -- 140 -- 140 140 N/A
Worker 1 95(-) 95(-) 95(-) 95(-) 95(-) N/A
Worker 2 -- -- -- -- -- N/A
Product C Worker 3 90 90 90 90 90 N/A
Worker 4 -- -- -- -- -- N/A
Worker 5 -- -- -- -- -- N/A
At the end of period 2, it is found that worker 4 has worked on tasks 1, 2 and 3 for
215 times. That worker has skill levels 4, 5 and 6 for those tasks. According to learning
and forgetting rates table, now skill levels can be revised for tasks 1 and 2. Two different
random numbers between 0 to 1 has been generated to match probability. Both random
numbers are more than 0.5. Thus, skill level 4 has changed to 5 for task 1 and for task 2
skill level changed from 5 to 6. Skill levels for product B and C remain unchanged after
period 2.
122
Table 106: Revised skill matrix after learning and forgetting take place for product A
after period 2
Worker 1 7 5 1 4 6 3
Worker 2 4 3 4 6 3 4
Worker 3 3 1 4 4 4 4
Worker 4 5 6 6 2 4 7
Worker 5 2 5 5 4 2 4
Table 107: Revised skill matrix after learning and forgetting take place for product B
after period 2
Worker 1 5 4 7 4 2
Worker 2 4 5 2 1 4
Worker 3 4 4 5 3 4
Worker 4 6 4 4 4 3
Worker 5 3 4 1 7 6
Table 108: Revised skill matrix after learning and forgetting take place for product C
after period 2
Worker 1 4 5 4 5 4
Worker 2 1 4 2 7 4
Worker 3 4 4 4 3 6
Worker 4 6 7 3 4 4
Worker 5 2 4 1 5 3
After revising skill matrices, new cycle time and worker assignment is found by
implementing methodology of section 6.1. After that, experiment for next period is ready
to be performed. This way experiment for eight consecutive period is performed.

123
Table 109: Revised skill matrix after learning and forgetting take place for product A
after period 8
Worker 1 7 5 2 5 5 2
Worker 2 4 3 4 6 4 4
Worker 3 2 1 3 3 3 3
Worker 4 6 7 6 1 4 7
Worker 5 1 4 4 5 3 4
Table 110: Revised skill matrix after learning and forgetting take place for product B
after period 8
Worker 1 5 5 7 4 3
Worker 2 3 4 1 1 3
Worker 3 5 4 6 2 4
Worker 4 6 5 5 4 3
Worker 5 2 4 1 7 6
Table 111: Revised skill matrix after learning and forgetting take place for product C
after period 8
Worker 1 4 5 4 5 4
Worker 2 2 4 2 7 4
Worker 3 5 5 5 5 6
Worker 4 5 6 2 3 3
Worker 5 1 3 1 4 2
Table 109, Table 110 and Table 111 show skill levels after period 8. Bold numbers
indicate change in skill levels where a sign indicates a decrease and a sign indicates an
increase. For example, skill levels for worker 4 at tasks 1 and 2 has increased from 5 and
6 to 6 and 7, consecutively.
124
CHAPTER 7: CONCLUSIONS AND FUTURE WORKS
This chapter concludes the thesis based on the results found from experiment. In
this chapter, scope for future research in this field is also listed.
7.1 Conclusions
In this study, comparison between Seru and Assembly line has been done in terms
of output obtained with and without considering skill levels of workers. At first, output is
calculated without considering skill for a single product. Then, for the same product output
is calculated considering skill. It is found that output with skill is better than output without
skill for assembly line as compared to Seru. After developing a math model an extensive
experiment is done for 3 different products considering worker skills. It is found that, out
of 30 different cases, Assembly line has better output in 14 cases, Seru has better output in
12 cases and remaining 4 cases are a tie. These results bring a question on the claim of Seru
experts that using Seru is always better. The results prove that when skill of workers is
considered then assembly line must be considered as manufacturing strategy rather than
automatically choosing Seru. Since neither assembly line nor Seru prevailed over another,
the next question is when to use which strategy. In order to answer the question, an indexing
attempt is done based on sum of range of skill levels and standard deviation of skill levels.
It is found that, when the value of sum of ranges and standard deviation of skill levels are
higher it is better to use assembly line. Lower standard deviation value indicates that
workers are almost equally skilled and hence it is better to use Seru in that case.
In this study, 7 skill levels from 1 to 7 are considered for a worker where skill level
1 is the worst, skill level 7 is the best and skill level 4 is average skill level. It is assumed
125
that skill levels are normally distributed. For the most of the analysis a 15% standard
deviation of mean is used while determining processing times based on skill levels. To see
the impact of different standard deviation values on output 5%, 10%, 20% and 25%
standard deviation values are also used. It is found that the higher the value of standard
deviation the better output is found from assembly line. It suggests that, if skill levels of
workers for different tasks are more scattered, then Assembly line is better option.
A multi-product multi-period without skill case is also considered. It is assumed
that total demand of each product is known. Four different products are considered for four
different periods. The objective of this experiment is to see how workers are assigned for
different products when demand must be met or exceeded. Keeping the objective as to
minimize number of workers, a mathematical model is developed and experimentation for
different demand data set is performed. It is found that, when skill is not considered
workers are assigned as Seru units in most of the cases. As an extension of this approach,
another methodology is developed where the objective is to maximize output using the
same number of worker found from previous method. After doing experimentation for
some demand data set it is found that more units can be produced without increasing the
total number workers. As a result, decision makers have option to choose between
producing exact amount and producing more units using same capacity.
Experimentation done for multi-product multi-worker assignment considering skill
suggests that, when skill is considered a hybrid system of Seru and Assembly line can be
formed as manufacturing strategy. In this case, three different products with five workers
where there are three different combinations of workers are considered. Three different
126
combinations considered are 1 worker alone (Seru), an assembly line of two workers and
an assembly line of three workers. An extension of this method is also made where the
objective is to maximize output using minimum number of workers.
Learning and forgetting is natural for human being. If a worker works on a specific
task for a long period of time then it is likely that his skill on that task will improve and
vice versa. In this study, an analysis considering learning and forgetting rates of workers
is performed for eight different periods. Skill levels for some tasks for some workers are
increased based on number of units that workers produced and skill level for some tasks
decreased based on forgetting rate.
Hybrid
System of Asse
Seru mbly Line
Assembly line and
Seru
Low Skill Variation High
Figure 27: Selection of manufacturing strategy type based on skill variation
Figure 27 summarizes the main finding of this study which is selection of
manufacturing strategy based on skill variation.
When skill variation is low it is better to select Seru and when skill variation is high
assembly line is better option. For a moderate skill variation, a hybrid system of Assembly
line and Seru gives better solution. Using hybrid system workers can be cross trained and
multi skilled and Seru can be implemented once all workers gain significant skill
improvement on all tasks which is the foundation of Seru.

127
7.2 Future Works
i. In this study demand values are obtained from uniform distribution. Future
research can focus on stochastic demand values.
ii. Production volume has not been considered in this study. A future research can
be done considering production volume.
iii. In learning rates, factors such as age, task difficulty and training can be
incorporated in future works.
iv. Cost and space requirement calculation can also be added in future research.
v. There may be commonalities between tasks of different product. For example,
two different products can have welding work to be done. Commonality among
tasks should also be considered.
vi. Worker preference might affect the decision of worker assignment. Worker
preference can also be considered.
vii. Indexing can be done considering proximity, number of workstation and
precedence relation.
128
REFERENCES
Bagher, M., Zandieh, M., & Farsijani, H. (2011). Balancing of stochastic U-type assembly
lines: an imperialist competitive algorithm. The International Journal of Advanced
Manufacturing Technology, 54(1-4), 271-285.
Battaïa, O., & Dolgui, A. (2013). A taxonomy of line balancing problems and their solution
approaches. International Journal of Production Economics, 142(2), 259-277.
Bazaraa, M., & Jarvis, J. (1943). Linear programming and network flows (2nd ed.).
Canada: John Whily & Sons.
Boysen, N., Fliedner, M., & Scholl, A. (2007). A classification of assembly line balancing
problems. European journal of operational research, 183(2), 674-693.
Corominas, A., Pastor, R., & Plans, J. (2008). Balancing assembly line with skilled and
unskilled workers. Omega, 36(6), 1126-1132.
Fitzpatrick, E. L., & Askin, R. G. (2005). Forming effective worker teams with multi-
functional skill requirements. Computers & Industrial Engineering, 48(3), 593-
608.
HBS Toolkit – Basic Operations Self-Instructional workbook. (2000, April 18). Retrieved
from http://hbswk.hbs.edu/archive/1460.html#1
Mathematical Programming in Practice. (n.d.). Retrieved from
http://web.mit.edu/15.053/www/AMP-Chapter-05.pdf
Iwamuro H (2004) Cellular manufacturing system. Nikkan Kogyo Shimbun, Tokyo. (In
Japanese).
Jonsson, D., Medbo, L., & Engström, T. (2004). Some considerations relating to the
reintroduction of assembly lines in the Swedish automotive industry. International
Journal of Operations & Production Management, 24(8), 754-772.
Kaku, I., Gong, J., Tang, J., & Yin, Y. (2009). Modeling and numerical analysis of line-
cell conversion problems. International Journal of Production Research, 47(8),
2055-2078.
129
Kaku, I., Murase, Y., & Yin, Y. (2008). A study on human-task-related performances in
converting conveyor assembly line to cellular manufacturing. European Journal of
Industrial Engineering, 2(1), 17-34.
Koltai, T. (2013). Formulation of multi-level workforce skill constraints in assembly line
balancing models. IFAC Proceedings Volumes, 46(9), 772-777.
Kuo, Y., & Yang, T. (2007). Optimization of mixed-skill multi-line operator allocation
problem. Computers & Industrial Engineering, 53(3), 386-393.
Kumar, D. M. (2013). Assembly line balancing: a review of developments and trends in
approach to industrial application. Global Journal of Research In
Engineering, 13(2).
Liao, L. M. (2014). Construction and comparison of multi-model and mixed-model
assembly lines balancing problems with bi-objective. Journal of Industrial and
Production Engineering, 31(8), 483-490.
Liu, C., Stecke, K. E., Lian, J., & Yin, Y. (2014). An implementation framework for seru
production. International Transactions in Operational Research, 21(1), 1-19.
Liu, C., Yang, N., Li, W., Lian, J., Evans, S., & Yin, Y. (2013). Training and assignment
of multi-skilled workers for implementing Seru production systems. The
International Journal of Advanced Manufacturing Technology, 69(5-8), 937-959.
Liu, C., Li, W., Lian, J., & Yin, Y. (2012). Reconfiguration of assembly systems: From
conveyor assembly line to serus. Journal of Manufacturing Systems, 31(3), 312-
325.
Liu, C., Stecke, K. E., Lian, J., & Yin, Y. (2014). An implementation framework for Seru
production. International Transactions in Operational Research, 21(1), 1-19.
Panneerselvam, R., & Sankar, C. O. (1993). New heuristics for assembly line balancing
problem. International Journal of Management and Systems, 9(1), 25-36.
Ramezanian, Reza, and Abdullah Ezzatpanah (2015). "Modeling and solving multi-
objective mixed-model assembly line balancing and worker assignment
problem." Computers & Industrial Engineering 87, 74-80.
Rekiek, B., Dolgui, A., Delchambre, A., Bratcu, A., 2002. State of art of assembly lines
design optimization. Annual Reviews in Control 26 (2), 163–174.
130
Salveson, M. E. (1955). The assembly line balancing problem. Journal of industrial

engineering, 6(3), 18-25.
Sivasankaran, P., & Shahabudeen, P. (2013). Modelling hybrid single model assembly line
balancing problem. Udyog Pragati, 37, 26-36.
Stecke, K. E., Yin, Y., Kaku, I., & Murase, Y. (2012). Seru: the organizational extension
of JIT for a super-talent factory. International Journal of Strategic Decision
Sciences (IJSDS), 3(1), 106-119.
Süer, G. A. (1998). Designing parallel assembly lines. Computers & industrial
engineering, 35(3-4), 467-470.
Süer, G. A., & Alhawari, O. (2011). Operator assignment decisions in a highly dynamic
cellular environment. Operations Management Research and Cellular
Manufacturing Systems: Innovative Methods and Approaches: Innovative Methods
and Approaches, 258.
Süer, G. A., & Tummaluri, R. R. (2008). Multi-period operator assignment considering
skills, learning and forgetting in labour-intensive cells. International Journal of
Production Research, 46(2), 469-493.
Suksawat, B., HilraokaI, H., & Ihara,T. (2005). A new approach manufacturing cell
scheduling based on skill-based manufacturing integrated to genetic algorithm.
Retrieved February 13, 2017 from
http://www.springerlink.com/content/q885171535723r5q/fulltext.pdf
Sakazume, Y. (2005). Is Japanese Cell Manufacturing a New System? A Comparative
Study between Japanese Cell Manufacturing and Cellular Manufacturing (< Special
English Issue> Production and Logistics). 日本経営工学会論文誌, 55(6), 341-
349.
Villa, A., & Taurino, T. (2013). From JIT to Seru, for a Production as Lean as
Possible. Procedia Engineering, 63, 956-965.
Yagmahan, B. (2011). Mixed-model assembly line balancing using a multi-objective ant
colony optimization approach. Expert Systems with Applications, 38(10), 12453-
12461.
131
Yang, C., & Gao, J. (2016). Balancing mixed-model assembly lines using adjacent cross-
training in a demand variation environment. Computers & Operations
Research, 65, 139-148.
Yin, Y., Kaku, I., & Stecke, K. (2008). The evolution of seru production systems
throughout Canon. Operations Management Education Review, 2.
Ying, K. C., & Tsai, Y. J. (2017). Minimising total cost for training and assigning
multiskilled workers in seru production systems. International Journal of
Production Research, 55(10), 2978-2989.
Yin, Y., Stecke, K. E., Swink, M., & Kaku, I. (2017). Lessons from seru production on
manufacturing competitively in a high cost environment. Journal of Operations
Management, 49, 67-76.
Yu, Y., Wang, S., Tang, J., Kaku, I., & Sun, W. (2016). Complexity of line-
seru. SpringerPlus, 5(1), 1-26.
Yu, Y., Tang, J., Gong, J., Yin, Y., & Kaku, I. (2014). Mathematical analysis and solutions
for multi-objective line-cell conversion problem. European Journal of Operational
Research, 236(2), 774-786.
Yu, Y., Gong, J., Tang, J., Yin, Y., & Kaku, I. (2012). How to carry out assembly line–cell
conversion? A discussion based on factor analysis of system performance
improvements. International Journal of Production Research, 50(18), 5259-5280.
Yu, Y., Tang, J., Sun, W., Yin, Y., & Kaku, I. (2013). Reducing worker (s) by converting
assembly line into a pure cell system. International Journal of Production
Economics, 145(2), 799-806.
Yu, Y., Sun, W., Tang, J., & Wang, J. (2017a). Line-hybrid seru system conversion:
Models, complexities, properties, solutions and insights. Computers & Industrial
Engineering, 103, 282-299.
Yu, Y., Sun, W., Tang, J., Kaku, I., & Wang, J. (2017b). Line-seru conversion towards
reducing worker (s) without increasing makespan: models, exact and meta-heuristic
solutions. International Journal of Production Research, 55(10), 2990-3007.
Zhang, X., Liu, C., Li, W., Evans, S., & Yin, Y. (2017). Effects of key enabling
technologies for Seru production on sustainable performance. Omega, 66, 290-307.
132
APPENDIX A: MATHEMATICAL MODEL 1
Min = Ct;
! Constraint 1;
x11w1+x11w2+x11w3+x11w4+ x12w1+x12w2+x12w3+x12w4+
x13w1+x13w2+x13w3+x13w4+ x14w1+x14w2+x14w3+x14w4 =1;
! Constraint 2;
!Station 1;
3.85 * x11w1 + 5.95 * x11w2 + 8.05 * x11w3 + 7 *
x11w4 +
6.6 * x21w1 + 12 * x21w2 + 10.2 * x21w3 +
6.6*x21w4 +
9 * x31w1 + 4.95 * x31w2 + 9 * x31w3 + 9 * x31w4
+
8.7 * x41w1 + 3.3* x41w2 + 6 * x41w3 + 6.9*x41w4
+
21.75 * x51w1 + 15 * x51w2 + 10.5* x51w3 +
8.25*51w4 +
8 * x61w1 + 9.2 * x61w2 + 8 * x61w3 + 9.2 *
x61w4 <=Ct;
!Station 2;
3.85 * x12w1 + 4.9 * x12w2 + 8.05 * x12w3 + 7 *
x12w4 +
6.6 * x22w1 + 13.8 * x22w2 + 10.2 * x22w3 + 10.2 *
x22w4 +
9 * x32w1 + 9 * x32w2 + 10.35 * x32w3 + 9 * x32w4
+
8.7 * x42w1 + 6.9 * x42w2 + 7.8 * x42w3 + 5.1 *
x42w4 +
21.75 * x52w1 + 15 * x52w2 + 10.05 * x52w3 + 15 *
x52w4 +
8 * x62w1 + 9.2 * x62w2 + 8 * x62w3 + 9.2 * x62w4
<= Ct;
!Station 3;
3.85 * x13w1 + 4.9 * x13w2 + 8.05 * x13w3 + 7 *
x13w4 +
6.6 * x23w1 + 13.8 * x23w2 + 10.2 * x23w3 + 10.2 *
x23w4 +
133
9 * x33w1 + 9 * x33w2 + 10.35 * x33w3 + 9 * x33w4 +

8.7 * x43w1 + 6.9 * x43w2 + 7.8 * x43w3 + 5.1 *
x43w4 +
21.75 * x53w1 + 15 * x53w2 + 10.05 * x53w3 + 15 *
x53w4 +
8 * x63w1 + 9.2 * x63w2 + 8 * x63w3 + 9.2 * x63w4
<= Ct;
!Station 4;
3.85 * x14w1 + 4.9 * x14w2 + 8.05 * x14w3 + 7 *
x14w4 +
6.6 * x24w1 + 13.8 * x24w2 + 10.2 * x24w3 +
10.2 *x24w4+
9 * x34w1 + 9 * x34w2 + 10.35 * x34w3 + 9 * x34w4+
8.7 * x44w1 + 6.9 * x44w2 + 7.8 * x44w3 + 5.1 *
x44w4+
21.75 * x54w1 + 15 * x54w2 + 10.05 * x54w3 + 15
* x54w4+
8 * x64w1 + 9.2 * x64w2 + 8 * x64w3 + 9.2 * x64w4
<= Ct;
! Constraint 3;
! Precedence Relationship;
!(1--2);
x21w1 <= X11w1;

x21w2 <= X11w2;
x21w3 <= X11w3;
x21w4 <= x11w4;
x22w1 <= x12w1 + x11w2 +x11w3 +x11w4;

x22w2 <= x12w2 + x11w1 +x11w3 +x11w4;
x22w3 <= x12w3 + x11w1 +x11w2 +x11w4;
x22w4 <= x12w4 + x11w1 +x11w2 +x11w3;
x23w1 <= x13w1 + x11w2 + x11w3 + x11w4 + x12w2 + x12w3 + x12w4

;
;
;
;

+ x13w2 + x12w3 + x12w4 ;
+ x13w1 + x13w3 + x13w4 ;
+ x13w1 + x13w2 + x13w4 ;
+ x13w1 + x13w2 + x13w3 ;
134
! (1--3);
x31w1 <= X11w1;

x31w2 <= X11w2;
x31w3 <= X11w3;
x31w4 <= x11w4;
x32w1 <= x12w1 + x11w2 +x11w3 +x11w4;

x32w2 <= x12w2 + x11w1 +x11w3 +x11w4;
x32w3 <= x12w3 + x11w1 +x11w2 +x11w4;
x32w4 <= x12w4 + x11w1 +x11w2 +x11w3;

;
;
;
;

+ x13w2 + x12w3 + x12w4 ;
+ x13w1 + x13w3 + x13w4 ;
+ x13w1 + x13w2 + x13w4 ;
+ x13w1 + x13w2 + x13w3 ;
! (2--4);
x41w1 <= X21w1;

x41w2 <= X21w2;
x41w3 <= X21w3;
x41w4 <= x21w4;
x42w1 <= x22w1 + x21w2 +x21w3 +x21w4;

x42w2 <= x22w2 + x21w1 +x21w3 +x21w4;
x42w3 <= x22w3 + x21w1 +x21w2 +x21w4;
x42w4 <= x22w4 + x21w1 +x21w2 +x21w3;

;
;
;
;

+ x23w2 + x22w3 + x22w4 ;
135

+ x23w1 + x23w3 + x23w4 ;
+ x23w1 + x23w2 + x23w4 ;
+ x23w1 + x23w2 + x23w3 ;
! (3--5);
x51w1 <= X31w1;

x51w2 <= X31w2;
x51w3 <= X31w3;
x51w4 <= x31w4;
x52w1 <= x32w1 + x31w2 +x31w3 +x31w4;

x52w2 <= x32w2 + x31w1 +x31w3 +x31w4;
x52w3 <= x32w3 + x31w1 +x31w2 +x31w4;
x52w4 <= x32w4 + x31w1 +x31w2 +x31w3;

;
;
;
;

+ x33w2 + x32w3 + x32w4 ;
+ x33w1 + x33w3 + x33w4 ;
+ x33w1 + x33w2 + x33w4 ;
+ x33w1 + x33w2 + x33w3 ;
! (4--6);
x61w1 <= X41w1;

x61w2 <= X41w2;
x61w3 <= X41w3;
x61w4 <= x41w4;
x62w1 <= x42w1 + x41w2 +x41w3 +x41w4;

x62w2 <= x42w2 + x41w1 +x41w3 +x41w4;
x62w3 <= x42w3 + x41w1 +x41w2 +x41w4;
x62w4 <= x42w4 + x41w1 +x41w2 +x41w3;

;
;
136

;
;

+ x43w2 + x42w3 + x42w4 ;
+ x43w1 + x43w3 + x43w4 ;
+ x43w1 + x43w2 + x43w4 ;
+ x43w1 + x43w2 + x43w3 ;
! (5--6);
x61w1 <= X51w1;

x61w2 <= X51w2;
x61w3 <= X51w3;
x61w4 <= x51w4;
x62w1 <= x52w1 + x51w2 +x51w3 +x51w4;

x62w2 <= x52w2 + x51w1 +x51w3 +x51w4;
x62w3 <= x52w3 + x51w1 +x51w2 +x51w4;
x62w4 <= x52w4 + x51w1 +x51w2 +x51w3;

;
;
;
;

+ x53w2 + x52w3 + x52w4 ;
+ x53w1 + x53w3 + x53w4 ;
+ x53w1 + x53w2 + x53w4 ;
+ x53w1 + x53w2 + x53w3 ;
! Constraint 4;
x11w1 <= yw11;

x21w1 <= yw11;
x31w1 <= yw11;
x41w1 <= yw11;
x51w1 <= yw11;
x61w1 <= yw11;
x12w1 <= yw12;

137
x22w1 <= yw12;

x32w1 <= yw12;
x42w1 <= yw12;
x52w1 <= yw12;
x62w1 <= yw12;
x13w1 <= yw13;

x23w1 <= yw13;
x33w1 <= yw13;
x43w1 <= yw13;
x53w1 <= yw13;
x63w1 <= yw13;
x14w1 <= yw14;

x24w1 <= yw14;
x34w1 <= yw14;
x44w1 <= yw14;
x54w1 <= yw14;
x64w1 <= yw14;
x11w2 <= yw21;

x21w2 <= yw21;
x31w2 <= yw21;
x41w2 <= yw21;
x51w2 <= yw21;
x61w2 <= yw21;
x12w2 <= yw22;

x22w2 <= yw22;
x32w2 <= yw22;
x42w2 <= yw22;
x52w2 <= yw22;
x62w2 <= yw22;
x13w2 <= yw23;

x23w2 <= yw23;
x33w2 <= yw23;
x43w2 <= yw23;
x53w2 <= yw23;
x63w2 <= yw23;
x14w2 <= yw24;

x24w2 <= yw24;
x34w2 <= yw24;
x44w2 <= yw24;
x54w2 <= yw24;
x64w2 <= yw24;
x11w3 <= yw31;

x21w3 <= yw31;
x31w3 <= yw31;
x41w3 <= yw31;
x51w3 <= yw31;
x61w3 <= yw31;
138
x12w3 <= yw32;

x22w3 <= yw32;
x32w3 <= yw32;
x42w3 <= yw32;
x52w3 <= yw32;
x62w3 <= yw32;
x13w3 <= yw33;

x23w3 <= yw33;
x33w3 <= yw33;
x43w3 <= yw33;
x53w3 <= yw33;
x63w3 <= yw33;
x14w3 <= yw44;

x24w3 <= yw44;
x34w3 <= yw44;
x44w3 <= yw44;
x54w3 <= yw44;
x64w3 <= yw44;
x11w4 <= yw41;

x21w4 <= yw41;
x31w4 <= yw41;
x41w4 <= yw41;
x51w4 <= yw41;
x61w4 <= yw41;
x12w4 <= yw42;
x22w4 <= yw42;
x32w4 <= yw42;
x42w4 <= yw42;
x52w4 <= yw42;
x62w4 <= yw42;
x13w4 <= yw43;

x23w4 <= yw43;
x33w4 <= yw43;
x43w4 <= yw43;
x53w4 <= yw43;
x63w4 <= yw43;
x14w4 <= yw44;

x24w4 <= yw44;
x34w4 <= yw44;
x44w4 <= yw44;
x54w4 <= yw44;
x64w4 <= yw44;
! Constraint 5;
yw11+yw12+yw13+yw14=1;
yw21+yw22+yw23+yw24=1;
yw31+yw32+yw33+yw34=1;
139
yw41+yw42+yw43+yw44=1;
yw11+yw21+yw31+yw41=1;
yw12+yw22+yw32+yw42=1;
yw13+yw23+yw33+yw43=1;
yw14+yw24+yw34+yw44=1;
@BIN (X11W1);
@BIN (X11W2);
@BIN (X11W3);
@BIN (X11W4);
@BIN (X12W1);
@BIN (X12W2);
@BIN (X12W3);
@BIN (X12W4);
! Variables;
@BIN (X13W1);
@BIN (X13W2);
@BIN (X13W3);
@BIN (X13W4);
@BIN (X14W1);
@BIN (X14W2);
@BIN (X14W3);
@BIN (X14W4);
@BIN (X21W1);
@BIN (X21W2);
@BIN (X21W3);
@BIN (X21W4);
@BIN (X22W1);
@BIN (X22W2);
@BIN (X22W3);
@BIN (X22W4);
@BIN (X23W1);
@BIN (X23W2);
@BIN (X23W3);
@BIN (X23W4);
@BIN (X24W1);
@BIN (X24W2);
@BIN (X24W3);
@BIN (X24W4);
@BIN (X31W1);
@BIN (X31W2);
140
@BIN (X31W3);
@BIN (X31W4);
@BIN (X32W1);
@BIN (X32W2);
@BIN (X32W3);
@BIN (X32W4);
@BIN (X33W1);
@BIN (X33W2);
@BIN (X33W3);
@BIN (X33W4);
@BIN (X34W1);
@BIN (X34W2);
@BIN (X34W3);
@BIN (X34W4);
@BIN (X41W1);
@BIN (X41W2);
@BIN (X41W3);
@BIN (X41W4);
@BIN (X42W1);
@BIN (X42W2);
@BIN (X42W3);
@BIN (X42W4);
@BIN (X43W1);
@BIN (X43W2);
@BIN (X43W3);
@BIN (X43W4);
@BIN (X43W1);
@BIN (X43W2);
@BIN (X43W3);
@BIN (X43W4);
@BIN (X51W1);
@BIN (X51W2);
@BIN (X51W3);
@BIN (X51W4);
@BIN (X52W1);
@BIN (X52W2);
@BIN (X52W3);
@BIN (X52W4);
@BIN (X53W1);
@BIN (X53W2);
@BIN (X53W3);
@BIN (X53W4);
141
@BIN (X54W1);
@BIN (X54W2);
@BIN (X54W3);
@BIN (X54W4);
@BIN (X61W1);
@BIN (X61W2);
@BIN (X61W3);
@BIN (X61W4);
@BIN (X62W1);
@BIN (X62W2);
@BIN (X62W3);
@BIN (X62W4);
@BIN (X63W1);
@BIN (X63W2);
@BIN (X63W3);
@BIN (X63W4);
@BIN (X64W1);
@BIN (X64W2);
@BIN (X64W3);
@BIN (X64W4);
@BIN (YW11) ;
@BIN (YW12) ;
@BIN (YW13) ;
@BIN (YW14) ;
@BIN (YW21) ;
@BIN (YW22) ;
@BIN (YW23) ;
@BIN (YW24) ;
@BIN (YW31) ;
@BIN (YW32) ;
@BIN (YW33) ;
@BIN (YW34) ;
@BIN (YW41) ;
@BIN (YW42) ;
@BIN (YW43) ;
@BIN (YW44) ;
142
APPENDIX B: MATHEMATICAL MODEL 2 AND 3
Mathematical Model 2
/*********************************************
* OPL 12.6.0.0 Model
* Author: ma981115
* Creation Date: Nov 30, 2017 at 3:58:43 PM
*********************************************/
range product = 1..4;
range period = 1..4;
range worker = 1..3;
float productionrate [product][worker]=[[57.14, 104.34, 160],

[66.66, 120, 160],
[85.71, 160, 240],
[92.30, 160, 218]];
int demand [product]=[1500, 1200, 1400, 1000];
dvar int+ x[product][period][worker] in 0..1;
int requiredworker[period];
int m;
float f=0.2;
int z;
minimize z;
subject to {
forall (j in period)
sum(i in product)
sum(k in worker)
x[i][j][k]*k == requiredworker[j];
requiredworker[j]<=m;
forall (i in product)
forall(j in period)
sum( k in worker)
productionrate [i][k]*x[i][j][k]>= f* demand[i];
sum(j in period)
sum( k in worker)
productionrate [i][k]*x[i][j][k]>= demand[i];
z==m; }
143
/*********************************************
* Author: ma981115
* Creation Date: Dec 5, 2017 at 7:11:44 PM
*********************************************/
range product = 1..4;

range period = 1..4;
range worker = 1..3;
float productionrate [product][worker]=[[57.14, 104.34, 160],

[66.66, 120, 160],
[85.71, 160, 240],
[92.30, 160, 218]];
int demand [product]=[1500, 1200, 1400, 1000];
dvar int+ x[product][period][worker] in 0..1;
int requiredworker[period]=[9,9,9,9];
int m;
float f=0.2;
int z;
maximize z;
subject to {
sum(i in product)
sum(k in worker)
x[i][j][k]*k == requiredworker[j];
forall(j in period)
sum( k in worker)
productionrate [i][k]*x[i][j][k]>= f* demand[i];
sum(j in period)
sum( k in worker)
productionrate [i][k]*x[i][j][k]>= demand[i];
z== sum (i in product)

sum(j in period)
sum( k in worker)
productionrate [i][k]*x[i][j][k] ; }
144
APPENXID C: MATHEMATICAL MODEL 4 AND 5
/*********************************************
* Author: ma981115
* Creation Date: Nov 29, 2017 at 1:29:18 AM
*********************************************/
int c1 =...;
int c2 =...;
int c3 =...;
int np =...;
int w =...;
range Products = 1..np;

range Combination = 1..c3;
range worker = 1..w;
range m1=1..c1;
range m2=(c1+1)..c2;
float productiontime[Products][Combination]=...;
int demand[Products]=...;
int coefficient[worker][Products][Combination]=...;
dvar int+ x[Products][Combination] in 0..1;
dvar float z;
dvar float z1;
dvar float z2;
dvar float z3;
minimize z;
subject to {
forall (i in Products)
sum(j in Combination)
x[i][j]*productiontime[i][j]>=demand[i];
forall (k in worker)
sum(i in Products)
x[i][j]*coefficient[k][i][j]==1;
z1== sum(i in Products)

sum (j in m1)
x[i][j];
145
z2== 2* (sum(i in Products)

sum (j in m2)
x[i][j]);
z3== 3*(sum(i in Products)

sum (j in m3)
x[i][j]);
z == z1+z2+z3;
/*********************************************
* Author: ma981115
* Creation Date: Dec 5, 2017 at 7:26:57 PM
*********************************************/
int c1 =...;
int c2 =...;
int c3 =...;
int np =...;
int w =...;
range Products = 1..np;

range Combination = 1..c3;
range worker = 1..w;
range m1=1..c1;
float productiontime[Products][Combination]=...;
int demand[Products]=...;
int coefficient[worker][Products][Combination]=...;
dvar int+ x[Products][Combination] in 0..1;
dvar float z;
dvar float z1;
dvar float z2;
dvar float z3;
dvar float p;
maximize p;
subject to {
forall (i in Products)
x[i][j]*productiontime[i][j]>=demand[i];
146
forall (k in worker)
sum(i in Products)
x[i][j]*coefficient[k][i][j]==1;
z1== sum(i in Products)

sum (j in m1)
x[i][j];
z2== 2* (sum(i in Products)

sum (j in m2)
x[i][j]);
z3== 3*(sum(i in Products)

sum (j in m3)
x[i][j]);
z >= w;
p == sum (i in Products)
x[i][j]*productiontime[i][j];
}
147
APPENDIX D: LEARNING AND FORGETTING EFFECT
Learning and forgetting rate

Learning Forgetting
Skill Level Number of Probability Skill Level Periods Probability
times a task
needs to be
performed
From To From To
1 2 120 0.70 2 1 4 0.70

2 3 150 0.65 3 2 4 0.65
3 4 150 0.60 4 3 5 0.60
4 5 200 0.55 5 4 5 0.50
5 6 200 0.50 6 5 6 0.40
6 7 250 0.30 7 6 7 0.30
148
Worker Assignment at period 1

(Unit) 1 2 3 4 5 6 times
operations
performed
at the end of
the period
Worker
1
A 2
Worker
3
Worker 20×5=100
4
Worker
5
Worker N/A
1
B 2
3
Worker N/A
4
5
Worker N/A 19×5=95
1
C 2
Worker N/A
3
Worker N/A
4
Worker N/A
5
149
Check sheet of workers task performance after period 1

Product Worker Task 1 Task 2 Task 3 Task 4 Task 5 Task 6
Worker 1 - - - - - -
Worker 2 - - - 100 100 100
Product A Worker 3 - - - - - -
Worker 4 100 100 100 - - -
Worker 5 - - - - - -
Worker 1 - - - - - N/A
Worker 2 - - - - - N/A
Product B Worker 3 140 - 140 - - N/A
Worker 4 - - - - - N/A
Worker 5 - 140 - 140 140 N/A
Worker 1 95 95 95 95 95 N/A
Worker 2 - - - - - N/A
Product C Worker 3 - - - - - N/A
Worker 4 - - - - - N/A
Worker 5 - - - - - N/A
150

(Unit) 1 2 3 4 5 6 times
operations
performed
at the end
of the
period
Worker
1
A 2
Worker
3
Worker 115
4
Worker 115
5
Worker N/A 70
1
B 2
Worker N/A
3
Worker N/A
4
Worker N/A
5
Worker N/A
1
C 2
Worker N/A 90
3
Worker N/A
4
Worker N/A
5
151
Check sheet of worker’s task performance after period 2

Worker 1 -- -- -- -- -- --
Worker 2 -- -- -- 100(-) 100(-) 100(-)
Product A Worker 3 -- -- -- -- -- --
Worker 4 215 215 215 -- -- --
Worker 5 -- -- -- 115 115 115
Worker 1 70 70 70 70 70 N/A
Worker 2 -- -- -- -- -- N/A
Product B Worker 3 140(-) -- 140(-) -- -- N/A
Worker 4 -- -- -- -- -- N/A
Worker 5 -- 140 -- 140 140 N/A
Worker 1 95(-) 95(-) 95(-) 95(-) 95(-) N/A
Worker 2 -- -- -- -- -- N/A
Worker 4 -- -- -- -- -- N/A
Worker 5 -- -- -- -- -- N/A
Revised skill matrix after learning and forgetting take place for product A after period 2
Worker 1 7 5 1 4 6 3
Worker 2 4 3 4 6 3 4
Worker 3 3 1 4 4 4 4
Worker 4 5 6 6 2 4 7
Worker 5 2 5 5 4 2 4
Revised skill matrix after learning and forgetting take place for product B after period 2
Worker 1 5 4 7 4 2
Worker 2 4 5 2 1 4
Worker 3 4 4 5 3 4
Worker 4 6 4 4 4 3
Worker 5 3 4 1 7 6
152
Revised skill matrix after learning and forgetting take place for product C after period 2
Worker 1 4 5 4 5 4
Worker 2 1 4 2 7 4
Worker 3 4 4 4 3 6
Worker 4 6 7 3 4 4
Worker 5 2 4 1 5 3
153

(Unit) 1 2 3 4 5 6 times
operations
performed at
the end of
the period
Worker
1
A 2
Worker
3
Worker
4
Worker
5
Worker N/A 14×5=70
1
B 2
Worker N/A
3
Worker N/A
4
Worker N/A
5
Worker N/A
1
C 2
Worker N/A 18×5=90
3
Worker N/A
4
Worker N/A
5
154

Worker 1 --- --- --- --- --- ---
Worker 2 55 55 55 155 155 155
Product A Worker 3 --- --- --- --- --- ---
Worker 4 215(1) 215(1) 215(1) --- --- ---
Worker 5 --- --- --- 115 115 115
Worker 1 140 140 140 140 140 N/A
Worker 2 --- --- --- --- --- N/A
Product B Worker 3 140(--) --- 140(--) --- --- N/A
Worker 4 --- --- --- --- --- N/A
Worker 5 --- 140(--) --- 140(--) 140(--) N/A
Worker 1 95(--) 95(--) 95(--) 95(--) 95(--) N/A
Worker 2 --- --- --- --- --- N/A
Worker 4 --- --- --- --- --- N/A
Worker 5 --- --- --- --- --- N/A
Worker 1 7 5 1 4 6 3
Worker 2 4 3 4 6 4 4
Worker 3 3 1 4 4 4 4
Worker 4 5 6 6 2 4 7
Worker 5 2 5 5 4 2 4
Worker 1 5 4 7 4 2
Worker 2 4 5 2 1 4
Worker 3 4 4 5 3 4
Worker 4 6 4 4 4 3
Worker 5 3 4 1 7 6
155
Worker 1 4 5 4 5 4
Worker 2 1 4 2 7 4
Worker 3 4 4 4 4 6
Worker 4 6 7 3 4 4
Worker 5 2 4 1 5 3
Worker task assignment at period 4

Dem Task Tas Tas Tas Tas Tas Number of
and 1 k2 k3 k4 k5 k6 times
(Unit operations
) performed at
the end of the
period
Worker 1 120
Worker 2
Product 18×5 Worker 3
A =90 Worker 4 120
Worker 5
Worker 1 N/A
Worker 2 N/A
Product 17×5 Worker 3 N/A 145
B =85 Worker 4 N/A
Worker 5 N/A 145
Worker 1 N/A
Worker 2 N/A 56
Product 15×5 Worker 3 N/A
C =75 Worker 4 N/A
Worker 5 N/A
156

Worker 1 120 120 ---- 120 ---- ----
Worker 2 55(-) 55(-) 55(-) 155(-) 155(-) 155(-)
Product A Worker 3 ---- ---- ---- ---- ---- ----
Worker 4 215(--) 215(--) 335 ---- 120 120
Worker 5 ---- ---- ---- 115(-) 115(-) 115(-)
Worker 1 140(-) 140(-) 140(-) 140(-) 140(-) N/A
Worker 2 ---- ---- ---- ---- ---- N/A
Product B Worker 3 285 ---- 285 ---- ---- N/A
Worker 4 ---- ---- ---- ---- ---- N/A
Worker 5 ---- 285 ---- 285 285 N/A
Worker 1 95(---) 95(---) 95(---) 95(---) 95(---) N/A
Worker 2 56 56 56 56 56 N/A
Product C Worker 3 180(-) 180(-) 180(-) 180(-) 180(-) N/A
Worker 4 ---- ---- ---- ---- ---- N/A
Worker 5 ---- ---- ---- ---- ---- N/A
Worker 1 7 5 1 4 6 2
Worker 2 4 3 4 6 4 4
Worker 3 2 1 4 4 4 4
Worker 4 5 6 6 1 4 7
Worker 5 1 5 5 4 2 4
Worker 1 5 4 7 4 2
Worker 2 4 5 1 1 4
Worker 3 4 4 5 2 4
Worker 4 6 4 4 4 2
Worker 5 3 4 1 7 6
157
Worker 1 4 5 4 5 4
Worker 2 1 4 2 7 4
Worker 3 4 4 4 4 6
Worker 4 6 7 2 4 4
Worker 5 2 4 1 5 2

(Unit) 1 2 3 4 5 6 times
operations
performed
at the end
of the
period
Worker 1
Worker 2
Product 19×5=95 Worker 3
A Worker 4 120
Worker 5 120
Worker 1 N/A 70
Worker 2 N/A
Product 9×5=45 Worker 3 N/A
B Worker 4 N/A
Worker 5 N/A
Worker 1 N/A
Worker 2 N/A 80
Product 10×5=50 Worker 3 N/A
C Worker 4 N/A
Worker 5 N/A
158

Worker 1 120(-) 120(-) 120() ----- -
Worker 2 55(-) 55(-) 55(-) 155(-) 155(-) 155(-)
Product A Worker 3 - ----- ----- ----- -----
Worker 4 335 335 (335)120 120 120
Worker 5 - ---- ---- 235 235 235
Worker 1 210 210 210 210 210 N/A
Worker 2 ----- ----- ----- N/A
Product B Worker 3 285(-) ---- 285(-) - ---- N/A
Worker 4 ----- ----- ----- ----- - N/A
Worker 5 ----- 285(-) 285(-) 285(-) N/A
Worker 1 47 47 47 47 47 N/A
Worker 2 135 135 135 135 135 N/A
Product C Worker 3 180(-) 180(-) 180(-) 180(-) 180(-) N/A
Worker 4 ----- ----- - ----- ----- N/A
Worker 5 ----- ----- ----- -- N/A
Worker 1 7 5 1 4 6 2
Worker 2 4 3 4 6 4 4
Worker 3 2 1 3 4 3 4
Worker 4 6 6 6 1 4 7
Worker 5 1 5 5 5 3 4
Worker 1 5 5 7 4 3
Worker 2 3 4 1 1 4
Worker 3 4 4 5 2 4
Worker 4 6 4 4 4 2
Worker 5 2 4 1 7 6
159
Worker 1 4 5 4 5 4
Worker 2 2 4 2 7 4
Worker 3 4 4 4 4 6
Worker 4 6 7 2 4 4
Worker 5 2 4 1 5 2
Worker assignment at period 6

Dema Task Task Task Task Task Task Number of
nd 1 2 3 4 5 6 times
(Unit) operations
performed at
the end of
the period
Worker 1
Worker 2
Product 13×5= Worker 3
A 65 Worker 4 65
Worker 5
Worker 1 N/A
Worker 2 N/A
Product 17×5= Worker 3 N/A 145
B 85 Worker 4 N/A
Worker 5 N/A 145
Worker 1 N/A 90
Worker 2 N/A
Product 16×5= Worker 3 N/A
C 80 Worker 4 N/A
Worker 5 N/A
160

Worker 1 120(--) 120(-- 120(-- ------ --
) )
Product A Worker 2 55(--) 55(--) 55(--) 155(-- 155(-- 155(--)
) )
Worker 3 -- - ------ - ------
Worker 4 65 400 (335)185 65 185 185
Worker 5 -- ----- ----- - - 235(-)
Worker 1 210(-) 210(-) 210(-) 210(-) 210(-) N/A
Worker 2 - - ------ N/A
Product B Worker 3 430 ---- 430 - ---- N/A
Worker 4 ----- ----- ----- ----- - N/A
Worker 5 - 430 430 430 N/A
Worker 1 137 137 137 137 137 N/A
Worker 2 -- 135(-) 135(-) 135(-) 135(-) N/A
Product C Worker 3 180(--) 180(-- 180(--) 180(-- 180-(- N/A
) ) )
Worker 4 ------ ------ -- ------ ------ N/A
Worker 5 ------ ------ ------ --- N/A
Worker 1 7 5 1 4 5 2
Worker 2 4 3 4 6 4 4
Worker 3 2 1 3 3 3 4
Worker 4 6 7 6 1 4 7
Worker 5 1 4 4 5 3 4
Worker 1 5 5 7 4 3
Worker 2 3 4 1 1 3
Worker 3 5 4 6 2 4
Worker 4 6 5 4 4 3
Worker 5 2 4 1 7 6
161
Worker 1 4 5 4 5 4
Worker 2 2 4 2 7 4
Worker 3 4 4 4 4 6
Worker 4 5 7 2 4 4
Worker 5 2 4 1 4 2
162

Demand Ta Task Task Task Task Task Number of
(Unit) sk 2 3 4 5 6 times
1 operations
performed at
the end of the
period
Worker 60
1
A 2
Worker
3
Worker
4
Worker
5
Worker N/A
1
B 2
Worker N/A
3
Worker N/A 135
4
Worker N/A 135
5
C 1
Worker N/A
2
Worker N/A 90
3
Worker N/A
4
Worker N/A
5
163

Task 1 Task 2 Task 3 Task 4 Task 5 Task
6
Worker 1 180 180 60 180 60 60
Worker 2 55(---) 55(---) 55(---) 155(-- 155(---) 155(-
Product A -) --)
Worker 3 --- -- - -- ------
-
Worker 4 65(-) - (165)185(- 65(-) 185(-) 185(-
) )
Worker 5 - - -- -- 235(-
-)
Worker 1 210(-- 210(-- 210(--) 210(-- 210(--) N/A
) ) )
Product B Worker 2 -- -- ------- N/A
Worker 3 - ----- - -- ----- N/A
Worker 4 135 ------ 135 ------ -- N/A
Worker 5 -- 135 - 565 135 N/A
Worker 1 137(-) 137(-) 137(-) 137(-) 137(-) N/A
Worker 2 --- 135(-- 135(--) 135(-- 135(--) N/A
Product C ) )
Worker 3 270 270 270 270 270 N/A
Worker 4 - ------- --- ------- ------- N/A
Worker 5 ------ ------ --- N/A
Worker 1 7 5 1 5 5 2
Worker 2 4 3 4 6 4 4
Worker 3 2 1 3 3 3 3
Worker 4 6 7 6 1 4 7
Worker 5 1 4 4 5 3 4
164
Worker 1 5 5 7 4 3
Worker 2 3 4 1 1 2
Worker 3 5 4 6 2 4
Worker 4 6 5 4 4 3
Worker 5 2 4 1 7 6
Worker 1 4 5 4 5 4
Worker 2 2 4 2 7 4
Worker 3 5 4 4 5 6
Worker 4 5 6 2 4 3
Worker 5 1 4 1 4 2
Worker task assignment at Period 8

Dema Tas Tas Tas Tas Tas Tas Number of
nd k1 k2 k3 k4 k5 k6 times
(Unit) operations
performed
at the end of
the period
Worker 1 60
Worker 2
Product 12×5= Worker 3
A 60 Worker 4
Worker 5
Worker 1 N/A
Worker 2 N/A
Product 14×5= Worker 3 N/A
B 70 Worker 4 N/A 70
Worker 5 N/A
Worker 1 N/A
Worker 2 N/A
Product 19×5= Worker 3 N/A 100
C 95 Worker 4 N/A
Worker 5 N/A
165

Task Task 2 Task 3 Task 4 Task 5 Task 6
1
Worker 1 240 240 120 60 120 120
Worker 2 55(--- 55(----) 55(----) 155(--- 155(----) 155(----
Product A -) -) )
Worker 3 ---- --- -- ---
Worker 4 65(--) -- (165)185(- 65(--) 185(--) 185(--)
-)
Worker 5 -- -- --- --- 235(---)
Worker 1 210(-- 210(--- 210(---) 210(--- 210(---) N/A
-) ) )
Product B Worker 2 --- --- - N/A
Worker 3 -- ------ -- --- ------ N/A
Worker 4 205 70 205 70 70 N/A
Worker 5 --- 135(-) -- 565(-) 135(-) N/A
Worker 1 137(-- 137(--) 137(--) 137(--) 137(--) N/A
)
Product C Worker 2 ---- 135(--- 135(---) 135(--- 135(---) N/A
) )
Worker 3 100 370 370 100 370 N/A
Worker 4 -- - ---- -------- - N/A
Worker 5 ------- - ---- N/A
Worker 1 7 5 2 5 5 2
Worker 2 4 3 4 6 4 4
Worker 3 2 1 3 3 3 3
Worker 4 6 7 6 1 4 7
Worker 5 1 4 4 5 3 4
Worker 1 5 5 7 4 3
Worker 2 3 4 1 1 2
Worker 3 5 4 6 2 4
Worker 4 6 5 5 4 3
Worker 5 2 4 1 7 6
166
Worker 1 4 5 4 5 4
Worker 2 2 4 2 7 4
Worker 3 5 5 5 5 6
Worker 4 5 6 2 3 3
Worker 5 1 3 1 4 2
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
Thesis and Dissertation Services

Thesishdjshuhsgdoowuu 8 WJWH

Uploaded by

Copyright:

Available Formats

You might also like

Thesishdjshuhsgdoowuu 8 WJWH

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Thesishdjshuhsgdoowuu 8 WJWH

Uploaded by

Copyright:

Available Formats

Impact of Skill: Seru vs Classical Assembly Line

the Russ College of Engineering and Technology of Ohio University

of the requirements for the degree

© 2018 Md Abdullah. All Rights Reserved.

This thesis titled

Impact of Skill: Seru vs Classical Assembly Line

has been approved for

the Department of Industrial and Systems Engineering

and the Russ College of Engineering and Technology by

Professor of Industrial and Systems Engineering

Dean, Russ College of Engineering and Technology

ABDULLAH, MD, M.S., May 2018, Industrial and Systems Engineering

Impact of Skill: Seru vs Classical Assembly Line

Director of Thesis: Gursel A. Suer

Seru manufacturing system is the modern version of manufacturing strategy

strategy where there are simple equipment involved in an assembly environment. In

suitable for Seru.

This thesis is dedicated to my beloved parents. Without their continuous support, I

wouldn’t have been able to accomplish this.

It is my immense pleasure to express my sincere gratitude to all the people who

my Master’s degree with financial aid.

comments will always be my guidance in my future life and career.

I am grateful to my committee members Dr. Dusan Sormaz, Dr. Ashley Metcalf

time. Without their support, it wouldn’t have been a success.

Mehadi Hasan Farhad for his technical support.

List of Tables .................................................................................................................... 10

List of Figures ................................................................................................................... 16

Chapter 1: Introduction ..................................................................................................... 18

1.1 Assembly Line ................................................................................................... 18

1.1.1 Different Assembly Line Types ................................................................ 19

1.2 Different Assembly Line Layouts ...................................................................... 21

1.3 Related Terms in Line Balancing ...................................................................... 23

1.4 Assembly Line Balancing .................................................................................. 27

1.4.1 Assembly Line Balancing Problem Type ................................................. 30

1.5 Seru Production System ..................................................................................... 31

1.5.1 Divisional Seru.......................................................................................... 32

1.5.2 Rotating Seru ............................................................................................ 32

1.5.3 Yatai .......................................................................................................... 33

1.6 Skills in Assembly Line and Seru System ......................................................... 33

1.7 Seru vs Classical Assembly Line ....................................................................... 34

1.8 Mathematical Modeling ..................................................................................... 35

1.9 Objectives of the Thesis ..................................................................................... 36

1.10 Organization of the Thesis ................................................................................. 36

Chapter 2: Literature Review ............................................................................................ 37

Chapter 3: Problem Statement .......................................................................................... 48

3.1 Single Product Case Analysis Based on Standard Time Only.............................. 48

3.2 Single Product Case Analysis Based on Skill ....................................................... 51

3.3 General Classification of Cases ............................................................................ 55

Chapter 4: Methodology and Results for Single Product Case......................................... 56

4.1 Methodology for Single Product Case .................................................................. 57

4.1.1 Mathematical Model 1 ................................................................................. 57

4.2 Results Based on Skill for Single Product (Product 1) ......................................... 58

4.2.1 When Skill Levels 1 to 7 All Are Considered ............................................. 59

4.2.2 When Skill Levels 3, 4, 5 Are Considered................................................... 61

4.2.3 When Skill Levels 2, 4, 6 and 1, 4, 7 Are Considered ................................. 62

at Each Task ……………………………………………………………………………64

4.2.5 When Skill Levels All 4 and All 7 Are Considered..................................... 66

4.3.1 Results for Product 2................................................................................. 69