th

Proceeding, 6 International Seminar on Industrial Engineering and Management

Harris Hotel, Batam, Indonesia, February 12th-14th, 2013 ISSN : 1978-774X

ANALYSIS OF FACTORS AFFECTING THROUGHPUT RATE IN FLEXIBLE

MANUFACTURING SYSTEM WITH AUTOMATED GUIDED VEHICLE
SYSTEM

Teuku Yuri M. Zagloel1 , Romadhani Ardi2, Lusyane Eko Tantri3.

1,2,3
Industrial Engineering Dept., Universitas Indonesia, Depok, Indonesia.
E-mail : yuri@ie.ui.ac.id, romadhani@ie.ui.ac.id, lusyane.eko@ui.ac.id

ABSTRACT
Flexibility is known as one of the competitive keys for manufacturing industries. Flexible
Manufacturing System is a complex and expensive system that require an accurate
designing phase. By simulation, it is possible to carefully examine the behavior of FMS
components to predict the performance of a manufacturing system. Buffer size, mean time
to repair, and number of automated guided vehicles are proven as critical factors in FMS
which affecting throughput rate significantly. Furthermore, it is found that there are
significant interactions between buffer size and MTTR and also between MTTR and
number of AGVs.
Key words: Flexible Manufacturing System, buffer size, mean time to repair, automated
guided vehicle.

1. INTRODUCTION management, and development of a modern

multi-stand machining, such as Flexible
As the economy grows the demand for Manufacturing System (Yucel, 2005).
goods and services continues to increase.
To meet the high demand for goods,
manufacturing industries are required to 2. THEORETICAL BACKGROUND
operate on a large scale. Flow shop
manufacturing system can accommodate the FMS is a manufacturing technology and also
production of large quantities but in limited a philosophy (Shivanand, 2006). (Anglani
variability. Meanwhile, job shop can et.al, 2002) and (Yucel, 2005) define FMS as
accommodate consumer demand for product an automated manufacturing system
varieties, but resulted in a relatively low- composed of multi-function machines that
volume production. Flexible Manufacturing are connected by a material handling system
System (FMS) is a way to combine the and controlled by a computer system.
efficiency of flow shop with the flexibility of a Throughput rate is one of the most important
job shop to meet the diverse demand for parameter in measuring the effectiveness of
goods at a lower cost (Basnet, 1994). a manufacturing system, FMS is not an
exception, as shown in (Smith, 1986). The
Flexibility has been known as one of the key definition of the throughput rate is the
competitive for manufacturing companies. number of output processes per time unit
Today, consumer involvement in determining (Chase, 2006). Thus the throughput rate can
the quality and variety of products is be used as a parameter of success that
becoming increasingly important for must be carefully considered in designing an
manufacturing industries. Market trends are FMS.
subject to change in a short time led to a
shorter product life cycles and a competitive There are many factors that affect the
marketplace is forcing manufacturing throughput rate in a manufacturing system.
companies to explore new markets to sell Some authors claim that the flexibility of the
their products. Such market condition raises route as the most influential factor of an
the urgency for change in the organization of FMS. Several other authors include a tool-
production processes, through automation, slots of workstations in their studies. The
computer aided design and manufacturing number of pallets, buffer capacity, and

Analysis of Factors Affecting Throughput Rate in FMS

Teuku Yuri M Zagloel PS-33
 th
Proceeding, 6 International Seminar on Industrial Engineering and Management
ISSN : 1978-774X

machine failure are also discussed by many expected to describe the likely behavior of
authors as factors that have significant the components of FMS in the real world.
impact on FMS. In 2009, I. Um et al (I.Um, The model consists of:
2009) specifically examined the factors that  Six Machining Centers (MC) with input
affect the FMS with AGVs. The number of and output buffer
AGV, AGV speed, AGV deceleration, and  Automated Guided Vehicle system
pickup time proved to be a factor that greatly (AGVs) with a fixed path
affect the FMS. Among the four factors, it  Incoming and outgoing conveyor
appeared that the number of AGV is the
factor that most affects the throughput rate.

Regarding the engine failure factor, Chen

and Thinphangnga (Chen, 1996) conducted
a study that addresses the effect of engine
failure on the performance of FMS with
emphasis on the comparison of analytical
and simulation methods. In addition, there is
also a research by Vineyard et al (Vineyard,
1999) which describes the characteristics of
failure and repair rate in the FMS, with a
case study of a manufacturing plant in the
western United States. On the other hand, Figure 1. Model from Paper
Selen and Ashayeri (Selen, 2001) used (Source: I. Um et al, 2009)
simulation in terms of design of experiments
to identify an increase in throughput rate Operation of FMS models which simulated in
through the buffer capacity management. the literature is based on the following
assumptions.
FMS is a complex system and spend  MC, AGV, and conveyor never
substantial investment that requires an breakdown; so it is always available for
accurate design phase. In particular, it is processing and transporting
important to check carefully the behavior of  Each MC can only process one operation
FMS components to predict the performance at a time
of production systems (Anglani et.al, 2002).  No parts were rejected due to quality
One of the best ways to do this is by inspection; therefore, rework is not
simulation. Through simulation, researchers allowed
can see the behavior of the model in a  Each type of part, upon entering the
complex system that can not be easily system must be processed to completion;
explained through analytical or mathematical the cancellation order is permitted
models.  The time required to move a part between
the buffer input / output to MC ignored
 Input and output buffer capacity is
3. RESEARCH METHOD assumed unlimited
 Each AGV can carry only 1 part
3.1. Model Building
Hypothetical model of FMS which designed This model was rebuild using Tecnomatix
by Insup Um et al (I.Um, 2009) is shown in Plant Simulation 9.0 as shown in Figure 2. In
Figure 1. this study, validation is performed by
This model is chosen because it is comparing the resulting throughput with the
comprehensive enough in accommodating throughput models made in the author
machinery flexibility, material handling written on the paper, in the following
flexibility, and routes flexibility. This model is conditions:
also found to be more complex than models  The number of AGV 10 units
of other FMS hypothesis which generally  AGV speed 4.8 m / s
consist of 3 to 4 machining centers, so it is
 AGV 1 m/s2 acceleration

Analysis of Factors Affecting Throughput Rate in FMS

PS-34 Teuku Yuri M Zagloel
 th
Proceeding, 6 International Seminar on Industrial Engineering and Management
ISSN : 1978-774X

 Pickup time 3 s
 The simulation is run for 8 hours

Table 1. Model Validation

Model Simulation Result 342
In the Paper 326

3.2. Experimental Design

This experimental design is using the
General Factorial Design consisting of three
factors: the buffer capacity, MTTR, and the
number of AGV with each of 3 levels, 2
levels, and 3 levels. Arrival time, processing Figure 3. Main Effects Plot
time, and mean time between failure are
stochastic so it needs quite a lot replication From the ANOVA table we can that the
in order to obtain the actual mean value, for buffer capacity factor has P-value <0.001
it to replicate as much as 30 times. So the which means that this factor has a significant
total observation was 540 (30 x 3 x 2 x 3) effect even at 99% confidence level (α =
times with 8-hour simulation each run and 1- 0.1). Positive slope indicates that a larger
hour interval for each replication. buffer capacity will result in greater
throughput.

4. RESULT AND DISCUSSION Buffer can be interpreted as a safety stocks

because it reduces the risk of a line stop. If
To know the magnitude of the effect of each there is a failure on one machine, the
factor and also the interactions that take production can continue for a period of time
place, analysis of variance was conducted by taking parts which available on the buffer.
using Minitab 15. The result is shown in From this point of view, a large buffer
Table 2. capacity should be provided, but on the
other side the buffer will use a larger
ANOVA is a technique that allows us to production floor so that would lead to
examine differences in the influence factor of additional costs (Selen, 2001). For this case,
the sample being taken. In the ANOVA table the addition of buffer capacity of 1 unit to 3
obtained from the software Minitab 15, we units can increase throughput by an average
can see which factors are influencing the of 10 units for 8 hours of simulation, and the
parameters via the p-value indicator. If the p- addition of an average of 10 units if its buffer
value less than or equal to alpha (α) the capacity is increased into 5 units.
influence of these factors are statistically
significant or in other words the null 4.1. MTTR Factor
hypothesis can be rejected. MTTR factor also have a P-value <0.001
which means duration of MTTR have an
Charts of the influence of the main factors of effect on system’s throughput. The negative
throughput shown in Figure 3. The graph slope of the graph shows that MTTR is
has been given an idea of how the effects of inversely proportional to the throughput, ie
each factor and which factors are more MTTR 60 minutes produce throughput
influential. To be more in-depth, analysis of MTTR less than 30 minutes on this system.
each of the major factors will be discussed Compared to two other major factors, MTTR
one by one as follows. has the most significant effect, seen from the
slope of the graph which larger than the
others.

MTTR of 60 minutes represents the engine

failure due to human error, such as push a
wrong button, giving the wrong weight of oil,

Analysis of Factors Affecting Throughput Rate in FMS

Teuku Yuri M Zagloel PS-35
 th
Proceeding, 6 International Seminar on Industrial Engineering and Management
ISSN : 1978-774X

do not closing the interlockingdoor, or not factors, seen from the intersection of lines,
tightening a bolt. While MTTR of 30 minutes but there are also graphs that do not show
represents the engine failure due to an interaction, which showed by paralel
electrical equipment failure such as motors, lines.
relays, starters, transformers, and cables.
Engine failure due to human error and failure
of electrical equipment are two of the three
most common types of engine failures in
FMS (Vineyard, 1999). Though the fact that
engine failure may occur due to several
causes at once, it is not addressed in this
study.

Engine failure due to human error may be

reduced by providing training and creating
standard operating procedures. While the
engine failure due to electrical equipment
failure can be avoided by performing Figure 4.Interaction Plot
preventive maintenance or routine
inspection. 4.3.1. Interaction between MTTR and
Buffer Capacity
4.2. Number of AGVs Factor Interaction between these factors has a p-
The number of AGV factors also indicate p- value <0.001 which means there is a
value <0.001, which means it has a significant interaction. Seen that the buffer
significant effect on system throughput. capacity lines have different slopes to the
Positive slope of the graph shows a directly MTTR. Buffer capacity 5 has the greatest
proportional relationship, which is an slope, a decrease of 50 units throughput will
increase of number of AGV then system’s happen when MTTR increase from 30
throughput will be greater too. Although minutes to 60 minutes. Whereas in buffer
significant at 99% confidence interval, these capacity 1, a decrease is only at an average
factors influence the number of AGV at the of 20 units for 8 hours of simulation.
least compared to the two other factors. The
addition of an AGV unit into the system will Significant interaction means that changes in
only provide additional throughput average one factor levels will result in significant
of 3 units for 8 hours of simulation. changes to the yield, the throughput in this
case. On the MTTR of 30 minutes, it is seen
In the FMS with AGVs, material handling that the buffer capacity increase from 3 to 5
costs can include 20-50% of total operating add up to an average of 20 throughput.
costs (Shirazi, 2010). Therefore, the number Though at MTTR of 60 minutes, the buffer
of AGV used in an FMS system usually just capacity increase from 3 to 5 only add an
a little number to reduce costs. At a factory average of 2 throughput.
in the west United States, which used as
case studies by Michael Vineyard (Vineyard, 4.3.2. Interaction between Number of
1999), there is an FMS consists of four CNC AGV and MTTR
machines with three AGVs. While the PT P-value of interactions between number of
Dirgantara Indonesia used an FMS consists AGV and MTTR is 0.006 which means a
of four CNC machines and have only one significant interaction at 99% confidence
AGV. level. The graph shows that when MTTR 30
minutes the line slope is firstly large enough
4.3. Analysis of Factors Interaction at the increase the number of AGV from 1 to
In addition to the influence of each factor, 2, namely an increase in average throughput
presumably the system’s throughput is also of 10 units, then the slope becomes sligther
affected by the interaction between the at increase number of AGV from 2 to 3, the
factors. Figure 4 shows that there are some throughput increased an average of 5 unit.
charts that show the interaction between

Analysis of Factors Affecting Throughput Rate in FMS

PS-36 Teuku Yuri M Zagloel
 th
Proceeding, 6 International Seminar on Industrial Engineering and Management
ISSN : 1978-774X

While at the time MTTR 60 minutes, a small interaction between MTTR and the
initial slope of the line become larger, so that number of AGV.
when these two lines are near the 4. The interaction between the buffer
intersection will happen. capacity and the number of AGV and also
the interaction between the three factors
So it can be interpreted if an engine failure proved to be not significant, which means
due to human error (MTTR 60 minutes) can changes in the levels of one factor will
not be avoided then the additional number of give the same value of change to the
AGV will not help much in increasing the throughput rate at any level of other
number of system’s throughput. Otherwise in factors.
engine failure caused by electrical failure or
MTTR 30 minutes, adding the number of As a suggestion, subsequent studies should
AGV will be sufficient to affect the increased also examine other parameters in addition to
throughput of the system, especially in throughput rate, such as utilization (either on
increasing the number of AGV from one unit the machine, AGV, or land area), and track
into two units. density. There are many other factors can
also be studied such as tool magazine
4.3.3. Interaction between Buffer capacity on machine, pallet number, and
Capacity and Number of AGV vehicle recharging. This study can also be
The interaction between the buffer capacity enhanced by adding a financial aspect. In
by the number of AGV has a p-value 0.667, conclusion, the Authors wish this study may
which means no interaction between these be useful in expanding the reader's
two factors. In other word, increase the understanding of the flexible manufacturing
amount of buffer capacity of the AGV at any system, as well as hoping this research can
level will give the same result, ie an average be refined in further opportunities.
increase of 3 units of throughput for each
additional unit AGV.
6. REFERENCES
4.3.4. Interaction between the Three
Factors (a) A. Anglani et al. (2002) Object-oriented
Interactions between these three factors modeling and simulation of flexible
produces a p-value 0.414 or accept null manufacturing systems: a rule-based
hypothesis which means these three factors procedure. Simulation Modelling Practice
have no significant interaction. In other and Theory, vol. 10, 209–234.
words the change in the level of one factor (b) Bangsow, Steffen. (2010) Manufacturing
will give the same value of change to Simulation with Plant Simulation and
throughput rate at any level of other factors. SimTalk: Usage and Programming with
Examples and Solutions. Springer-Verlag
Berlin Heidelberg.
5. CONCLUSION (c) Basnet, Chuda and Mize, Joe H. (1994)
Scheduling and Control of Flexible
From the results of data processing and Manufacturing Systems: A Critical
analysis, it can be concluded from this study Review.
that: (d) Chase et al (2006), Operation
1. These three main factors; the buffer Management: for Competitive Advantage
capacity, the mean time to repair, and the with global Cases. McGraw-Hill
number of AGV are shown to have International Edition. Pg. 163-164.
significant influence on the throughput (e) Chen and Thinphangnga, (1996)
rate in this model of FMS. Analytical Modeling and Analysis of
2. The factor that has most influence on the Flexible Manufacturing Systems
throughput rate is the MTTR, followed by Considering System Component
buffer capacity and number of AGV Failure/Repair Rates, Journal of
3. Interaction of factors that proved Manufacturing System, Vol. 15, No.3,
significant are the interaction between the 143-154.
buffer capacity and MTTR and also

Analysis of Factors Affecting Throughput Rate in FMS

Teuku Yuri M Zagloel PS-37
 th
Proceeding, 6 International Seminar on Industrial Engineering and Management
ISSN : 1978-774X

(f) I. Um et al, (2009) The simulation design to The Graduate School of Natural and
and analysis of a Flexible Manufacturing Applied Sciences of Middle East
System with Automated Guided Vehivle Technical University.
system, Journal of Manufacturing (n) Zaini, Emsosfi et al. (2000) Perencanaan
System, Vol. 28, 115-122. Dan Penjadwalan Flexible Manufacturing
(g) Selen, Willem J. and Ashayeri, Jalal Systems (FMS) Di PT. IPTN Bandung
(2001) Manufacturing cell performance Menggunakan Metoda Heuristik Dan
improvement: a simulation study, Algoritma Beam Search. Jurnal
Robotics and Computer Integrated Teknologi Industri, Vol. IV, No. 4, 229 –
Manufacturing, Vol. 17, 169-176. 240.
(h) Shirazi, Babak et al (2010) A six sigma
based multi-objective optimization for
machine grouping control in flexible AUTHORS BIOGRAPHIES
cellular manufacturing systems with
guide-path flexibility, Advances in T. Yuri M. Zagloel is a professor in
Engineering Software, Vol. 41, 865–873. Industrial Engineering Department, Faculty
(i) Shivanand et al. (2006) Flexible of Engineering Universitas Indonesia. He is
Manufacturing System. New Delhi : New a Head of Manufacturing System Laboratory
Age International Publisher. IEUI. Contact him at yuri@ie.ui.ac.id
(j) Smith, M. L. et al (1986) Characteristic of
U.S. Flexible Manufacturing System-A Romadhani Ardi is a lecturer in Industrial
Survey, Flexible Manufacturing Systems: Engineering Department, Faculty of
Operation Research Models and Engineering Universitas Indonesia. He is a
Applications. coordinator of Manufacturing System
(k) Tecnomatix Plant Simulation Step-by- Laboratory IEUI. Contact him at
Step Help. (2008) Siemens Product romadhani@ie.ui.ac.id
Lifecycle Management Software II (DE)
GmbH. Lusyane Eko Tantri is a master degree
(l) Vineyard, Michael et al. (1999) Failure student in Industrial Engineering
rate distributions for Flexible Department, Faculty of Engineering
manufacturing systems: An empirical Universitas Indonesia. She is a member of
study, European Journal of Operational Manufacturing System Laboratory IEUI.
Research, Vol. 116, 139-155. Contact her at lusyane.eko@ui.ac.id
(m) Yucel, Necati Deniz. (2005) Simulation
Of A Flexible Manufacturing System: A
Pilot Implementation. A Thesis Submitted

Attachment

Figure 2. Model Layout

Analysis of Factors Affecting Throughput Rate in FMS

PS-38 Teuku Yuri M Zagloel
 th
Proceeding, 6 International Seminar on Industrial Engineering and Management
ISSN : 1978-774X

Table 2. Data Processing Result

General Linear Model: Throughput versus Kapasitas Buffer; MTTR; Jumlah AGV
Factor Type Levels Values
Kapasitas Buffer fixed 3 1; 3; 5
MTTR fixed 2 30''; 60''
Jumlah AGV fixed 3 1; 2; 3

Source DF Seq SS Adj SS Adj MS F P

Kapasitas Buffer 2 43084 43084 21542 129,67 0,000
MTTR 1 204439 204439 204439 1230,58 0,000
Jumlah AGV 2 4155 4155 2078 12,51 0,000
Kapasitas Buffer*MTTR 2 16510 16510 8255 49,69 0,000
Kapasitas Buffer*Jumlah AGV 4 386 386 97 0,58 0,677
MTTR*Jumlah AGV 2 1740 1740 870 5,24 0,006
Uk Buffer*MTTR*Jumlah AGV 4 656 656 164 0,99 0,414
Error 522 86721 86721 166
Total 539 357692

Analysis of Factors Affecting Throughput Rate in FMS

Teuku Yuri M Zagloel PS-39