Int J Adv Manuf Technol (2009) 44:1219–1229

DOI 10.1007/s00170-008-1916-4

ORIGINAL ARTICLE

Performance analysis of flexible material handling systems

for the apparel industry
James B. Dai & Neville K. S. Lee & W. S. Cheung

Received: 17 July 2008 / Accepted: 22 December 2008 / Published online: 14 February 2009
# Springer-Verlag London Limited 2009

Abstract Flexible material handling systems (FMHS) have of the FMS [2, 3]. According to Tompkins et al. [4], about
been widely used to enhance productivity involved with 20 to 50% of the total production cost is spent on material
product proliferation, and thus far, only fixed-track material handling. This makes the subject of material handling
handling systems such as Eton systems in the apparel increasingly important. In addition, all the complexity of
industry are commonly used. This paper explores the manufacturing is passed on to the material handling system.
potential advantages of a FMHS using free-ranging Therefore, the FMHS has been vital for improving the FMS
automated-guided vehicles with a local positioning system to fulfill the requirements of mass customization, especially
for the apparel industry. First, the free-ranging FMHS in the apparel industry whose demands are fairly various
(FRMHS) for the apparel industry has been designed. Then, and fast changing.
through Monte Carlo simulation and analytical models, the The apparel industry generated a total revenue of 1.5
performance in terms of manufacturing system effective- trillion US dollars in 2006 [5] and was also a major
ness, workstation utilization, and the total transportation manufacturing sector in Hong Kong with the highest gross
distance of the FRMHS are compared with those of the output in 2007 [6, 7]. However, due to the intensified
fixed-track system. Based on our analysis, the current challenge of mass customization and increasing labor costs,
proposed FRMHS can have significant advantages over the the apparel industry in advanced countries or areas has
fixed-track system. faced a steady decline recently [8]. In order to streamline
their production cycle to better respond to consumer
Keywords Free-ranging automated guided vehicle . demand and at the same time save costs and improve
Local positioning system . Unit production system . quality, apparel manufacturers have begun to seek for new
Free-ranging material handling system business and manufacturing practices and strategies, among
which the improvement of the material handling systems
ranks first.
1 Introduction There is abundant intensive research in automatic
handling and manipulation of textile products in the apparel
Flexible manufacturing systems (FMS) are crucial for industry. Sara Lee Knit Products uses automated storage/
modern manufacturing to enhance productivity involved retrieval systems and automated electrified monorail to
with product proliferation [1]. As one of the critical improve their material handling systems [9]. Grippers have
components of the FMS, the flexible material handling been developed and analyzed to grasp flexible materials to
system (FMHS) plays a strategic role in the implementation enhance material handling [10–12]. A robotic system has
been developed for textile-like material handling in [1]
from the perspective of handling operations based on
version and force/torque sensing. All the literature above
J. B. Dai : N. K. S. Lee (*) : W. S. Cheung
only focuses on the operations, but research has also been
The Hong Kong University of Science and Technology,
Sai Kung, Hong Kong, China conducted on designing new efficient material handling
e-mail: [email protected] systems. A FMHS with wired automated-guided vehicle

DO01916; No of Pages
1220 Int J Adv Manuf Technol (2009) 44:1219–1229

(AGV), which transports garments from the silkscreen 2.1 System structure
process to the fold-and-pack area, and the conveyor belt,
which delivers the boxed goods from fold-and-pack area to The main concept of the FRMHS is that it can support free-
the shipping area, has been designed to increase the ranging material handling rather than fixed path material
throughput and product quality [13]. The Walking Floor, handling. In order to achieve the free-ranging property, it
which is a sequentially operated reciprocating floor slat requires the following structure and subsystems:
conveyor with typical actuation through three hydraulic
cylinders, provides an opportunity to improve the material – Local positioning system (LPS): To support the
handling throughput [14]. The unit production system function of the FRAGV, an indoor LPS is required to
(UPS), which transports the material by a hanger-like estimate the absolute position information for the
carrier, increases the efficiency and reduces the work in FRAGV. A potentially cost-effective and accurate
process (WIP) level of apparel manufacturing from that of ultrasonic positioning system has been proposed for
traditional bundling systems [15]. There are two classical the AGV [20]. In the ultrasonic positioning system, the
UPS in the market: one is the TUKAtrack Information emitters of ultrasound and radio frequency are placed
Tracking System from the USA, and the other is the Eton on the ceiling of the plant, while the receivers are
system from Eton Systems in Sweden. Other material placed on the FRAGV. Since the radio frequency
handling solutions in the apparel industry include Toyota propagates much faster than the ultrasound, the
System Style quick response methods with garments passed synchronously transmitted signals from the same
by hand, the manual overhead sewing production line in the emitter will arrive at the receiver at different times.
UK-based Peter Ward, and Magic Tube for garment Based on the time difference of propagation for the
production, handling, warehousing, and transportation radio frequency and the ultrasound, one can determine
systems in Salpomec Ltd. [16]. However, Eton Systems the distance between the emitter and the receiver.
from Sweden remains the market leader in the modern Multilateration method can be used to figure out the
apparel industry [17]. With the advancement of automation position of the FRAGV with multiple transmitters
technology, AGV systems, which were first introduced in placed at different locations. Many algorithms such as
1955 [18], have become popular material handling devices Karman filter and particle filter also can be applied to
in the modern factory [19] due to their flexibility, especially improve the tracking and navigation performance.
that of the free-ranging AGV (FRAGV). Recently, a FMHS – Central controller: This is widely used in the manufac-
using FRAGV with a local positioning system [20] has turing industry. In the FRMHS, the central controller is
been proposed in warehouse and some other manufacturing designed for several purposes. Firstly, it can monitor
applications with promising results. Therefore, it is inter- and control the movement of the FRAGV and the entire
esting to investigate whether such a free-ranging FMHS has manufacturing system. Secondly, it can be used to
any advantages over the fixed-track system for the apparel identify failures or problems as well as to optimize the
industry, and this is also the objective of our study. production system. Thirdly, it gives orders to the
The structure of this paper is as follows. In Section 2, the loading module in workstations by radio frequency to
conceptual design of the free-ranging FMHS (FRMHS) load the materials and at the same dispatch jobs to
is proposed; in Section 3, the performance measures in workstations. Fourthly, it stores the information of the
terms of manufacturing system effectiveness, workstation product or the material, which is collected by the radio
utilization, and total transportation distance are formu- frequency identification (RFID).
lated and proposed to compare the FRMHS with the – FRAGV: Basically, the function of the AGV is similar
fixed-track system. In Section 4, Monte Carlo simulation to that of a truck. However, due to the limited space of
is conducted, and the promising results are presented. paths in the MHS, it is vital for the FRAGV to have the
Finally, the conclusion and further discussion are given capability of turning 90° to change the orientation in
in Section 5. the path without changing its position. Therefore, a
special design should be adopted. One of the easy ways
to provide this tight quarter turning is to use two
2 The design of the FRMHS independent motors for the left and right wheels of the
FRAGV. Furthermore, this vehicle is controlled by the
In this section, we address some design considerations that central controller discussed above. This can be accom-
can substantially affect the material flow and system plished by first determining the location of the FRAGV
performance. In particular, we will discuss the design of by LPS. Second, the FRAGV transmits this position
system structure, operating methodology, and the system information to the central controller. Finally, the central
layout. controller controls the speed and direction of the
Int J Adv Manuf Technol (2009) 44:1219–1229 1221

FRAGV. The power supply of the motors is provided – Routing: Central controller determines the optimal
by the rechargeable battery. routing for the FRAGV based on the location of the
– Workstation: In this system, the workstation should be FRAGV and the destination of the FRAGV as well as
equipped with a loading and unloading system for the the traffic condition.
FRAGV to bring the material in and out of the – FRAGV movement control: Aided by the LPS, the
workstation. Often, the tray would be used to contain central controller would be able to track the movement
the parts. To track the material flow, RFID or barcode of the FRAGV. Then, the central controller would
can be used. To enhance the throughput, a buffer that choose the optimal velocity and the direction of the
can hold several trays is used. FRAGV. To reach the planned velocity and direction,
– Battery charging/changing station: A supportive station the central controller would control the input currents
for the FRAGV to charge batteries and exchange to the motors of left and right wheels.
charged batteries with the empty ones. To facilitate – Traffic Control: To avoid the congestion and collision,
the automated charging or exchanging of batteries for the central controller has to coordinate the movement
the FRAGV, the latter are required to stop near the of the FRAGV. LPS and scheduling algorithms play a
battery charging/changing station quickly and accurate- vital role in this step. The control is realized by the
ly; therefore, a specially designed mechanical track is wireless communication with radio frequency.
designed and placed near the station. – Part loading and unloading: Once the FRAGV reaches
the designated workstation, the operation of loading
and unloading would take place. This operation is
2.2 Operating methodology
controlled and monitored by the central controller aided
by the LPS and the FRID technology.
The previous section describes the main structure of the
– Material tracking: RFID can be used to track the
FRMHS. In this section, we will discuss the operating
material flow.
methodology of the FRMHS.
– Rerouting operation: There is a potential advantage of
– Order loading: When an order has been placed in the the FRMHS. Sometimes the production line would
production line, the central controller would generate a have unexpected change of the status of workstation or
production plan based on production process, the bill of the FRAGV, for example, the breakdown of the
materials, the size of the order, and the status of the workstation and the FRAGV. The central controller
production line. Then, from this production plan, the can modify the dispatching and routing order for the
material flow requirements will be generated. FRAGV. The failed FRAGV will be pulled back to the
– FRAGV dispatching: Central controller will select the AGV charging and storage station to avoid traffic
FRAGV based on the material flow requirements and the congestion. The schematic system flow diagram is
status of the FRAGV such as availability and location. shown in Fig. 1.

Fig. 1 Schematic system flow

diagram for the FRMHS

FRAGV position
(X, Y, Z)

Product information feedback to the

central controller by RFID system

Workstation 2
Workstation 1

FRAGV
1222 Int J Adv Manuf Technol (2009) 44:1219–1229

2.3 System layout FRAGV is broken down, the other FRAGV can cross the
path to ensure the continuous material handling and avoid
The facility layout design problem has been a very active congestion due to the specially designed FRAGV, which
research area in the past four decades, and many optimiza- can flexibly turn 90°. The corridors between the subgroups
tion models are reviewed in [21, 22]. However, all the can only be accessed by the workers. In this case, this
models assume that we have known the information design separates the people and the FRAGV for safety. Due
regarding production quantity and routing path of different to the free path property, the machines with similar
products in advance. In the apparel industry, the demand is functions can be arranged in function, product, or hybrid
changing quickly and is very difficult to forecast, so in this layout for easy maintenance, better resources sharing, and
paper, we only focus on constructing the conceptual layout technical support for operators.
of the FRMHS mainly from the perspectives of approxi-
mated system performance and safety. Based on the space
dominated by the fixed-track system as described in the 3 Performance analysis
Appendix, we design the layout of the FRMHS. Figure 2
shows the schematic layout of the FRMHS. In order for our To evaluate the effectiveness of the FRMHS in the apparel
system using the central controller, LPS, and the FRAGV to industry, several performance measures are proposed. The
operate properly, we need a special consideration on the formulation of these measures for both the FRMHS and the
system layout. To avoid the interference of human traffic in Eton systems are also explored in this section.
our FRMHS, the moving path for the AGV and for the
human is separated in our design. 3.1 Manufacturing system effectiveness
As shown in Fig. 2, the loading and unloading work-
stations are positioned at the top. The AGV charging station 3.1.1 Manufacturing cycle efficiency
is located at the bottom, and the workstations are placed in
the center. Each workstation comprises a loading area, Manufacturing cycle efficiency (MCE) is a traditional measure
which is denoted by a small rectangle and an operating area of the manufacturing process. MCE is defined as the ratio of
which is denoted by a large rectangle. The workstations are the time in actual production and setup process over the total
then grouped into subgroups, and a path for the FRAGV in time in the production area [23]. The higher the ratio, the
the center connects all the subgroups together. The FRAGV higher the percentage of time spent in the workstations. The
can only access the path in the subgroups and the path definition is shown in the following formula.
connecting subgroups to transport the material among
SþR
workstations. The path for the FRAGV can hold two bi- MCE ¼ ð1Þ
SþRþW þM
directional paralleling FRAGV. As a result, once one

Fig. 2 Schematic layout of the

FRMHS Loading Area Unloading Area

AGV Recharging and Storage Station

AGV Workstation Path for the AGV Path for People

Int J Adv Manuf Technol (2009) 44:1219–1229 1223

where S denotes the total setup time, R denotes the total where W0 denotes the critical WIP, rb denotes the
running time, W denotes the total waiting time, and M bottleneck rate, and T0 denotes the raw process time.
denotes the total material handling time.
3.2 Workstation utilization
3.1.2 Value added efficiency
Workstation utilization is defined as the fraction of actual
Value added efficiency (VAE) measures the percentage of operating time to the total available time [24]. It reflects the
time added to a product during the production process. VAE average efficiency of the workstations being used in the
is defined as the ratio of total run time to the total production line. In the apparel industry, the order size is
manufacturing time [23] as shown in the following formula. relatively small. Different products often require different
R sequence of production processes. Due to the fixed-track
VAE ¼ ð2Þ property of the Eton systems, it is necessary to change the
SþRþW þM
locations of the machine to suit a new product. Relocating
Although VAE looks similar to MCE, when the setup time machines takes time. Therefore, it would decrease the
is relatively large, improving MCE is not always conducive to productivity of the entire system. However, since the
significant productivity improvement. Therefore, the VAE is FRAGV has the property of free path, there is no need to
valuable for measuring the performance of the manufacturing relocate the workstation for launching a new order.
system, particularly the system whose setup time is changed. In this paper, we assume that all the workstations are never
idle and never fail before finishing an order. Furthermore, the
3.1.3 Work in process production line is well balanced. As we want to compare
the performance of the FRMHS and fixed-track systems, the
Work in process (WIP) is defined as the inventory between the formulation below will include the relocation of the worksta-
start and end points of a product routing and is commonly used tion. During the comparison, we will set the relocation time to
as a criteria to assess manufacturing systems [23, 24]. It has zero for the FRMHS. Therefore, in this formulation, before
significant effect on the inventory cost and the capability of launching a new order, it is necessary to clear the production
flexible and quickly responding to customers’ requirements. line and relocate the workstations. It is interesting to note, for
the purpose of improvement of productivity, that the reloca-
3.1.4 Average time in the system tion for launching a new product can be started while some of
the work for the existing product is being finished. In our case,
Average time in the system (AVT) [25] is the long-term we let l to be the number of workstations finishing the work
average time of a part spent in the system from entering the for the existing product. Therefore, the average time of each
loading station to departing the unloading station. This can order spent in production is TC ðQ þ dNPL e
l Þ þ TR þ TS .
be used to measure the speed of the response to a new order. However, the effective time is only QTC. Therefore, the
effective workstation utilization can be formulated as
3.1.5 Throughput quantity follows:
QTC
Throughput quantity (TH), which is often simply referred to as U¼ ð6Þ
throughput or production volume, is the number of jobs TC ðQ þ dNPL e
l Þ þ TR þ TS
completed in a given period of time. This can also be called where PL denotes the percentage of workstations loaded in
production rate [2, 26]. According to Little’s Law, the an order. Then, dNPL e denotes the total number of work-
relationship of TH, WIP, and the cycle time (CT) is defined as: stations required for the new order, where Q denotes the
WIP order size, TC denotes the CT, TS denotes the setup time, and
TH ¼ ð3Þ TR denotes the total machine relocation time.
CT
When comparing the performance of manufacturing 3.3 Total transportation distance
systems, we often need to consider the performance in the
practical worst case. The TH of the practical worst case Total transportation distance is the most frequently used
THPWC of given WIP level w is defined as follows [4]: criteria for evaluating material handling systems [22]. It is
w defined as the weighted sum of material flow distances
THPWC ¼ rb ð4Þ
W0 þ w
1 between different workstations or departments. The mini-
mum material flow distance is valuable for enhancing the
utilization of the entire system, reducing the throughput
W0 ¼ rb T0 ð5Þ time, and the WIP. In the comparative study between the
1224 Int J Adv Manuf Technol (2009) 44:1219–1229

FRMHS and the fixed-track system, we keep the same Therefore, the transportation distance in a branch for each
transportation speed and the same workstation requirements loaded workstation is formulated as
for both systems. Based on the layouts and the operating
dB ¼ 2ðLB þ HB Þ ð10Þ
principal of the FRMHS and the fixed-track system, the
total transportation distance can be formulated. where LB denotes the length of the branch of a workstation
and HB denotes the height of the branch of a workstation.
3.3.1 Total transportation distance of the free-ranging The transportation distance in the headline of both the
flexible material handling systems simply joined Eton (SJ-Eton) system and the joined-Eton
(J-Eton) will be formulated as follows:
Based on Fig. 2, the layout of the FRMHS is symmetric, In the SJ-Eton system shown in Fig. 8a, the transporta-
and each subgroup of the FRMHS has nFR workstations. tion distance in the headline comprises the path length for
Given the total number of workstations N and the workstations, the distance between two adjacent subgroups
percentage of workstations loaded, the number of sub- dBSG, and the length of the loading and unloading station
groups NGL required can be computed by: lLWS. It is defined as:


dNPL e dH ¼ NLWS þ 2ðNGL
1ÞdBSG þ lLWS þ WH ð11Þ
NGL ¼ ð7Þ
nFR
Based on the number of subgroups, the detailed layout dBSG ¼ 2LB þ WC þ WH ð12Þ
and the material flow of the FRMHS can be figured out.
Then, the total transportation distance of the FRMHS  
consisting of distance in the subgroups and distance among dNPL e þ l
NGL ¼ ð13Þ
subgroups is: nE
    
dNPL e NGL where nE denotes the number of workstations in each
TRDFR ¼ 2 LWS þ 2 þ 1 WC ð8Þ
2 2 subgroup, WH denotes the width of the headline, and NGL
denotes the number of subgroups required for an order.
where PL denotes the percentage of workstations loaded in
In the J-Eton system shown in Fig. 8b, we can see that as
an order, LWS denotes the length of the workstation, and WC
the number of subgroups increases, the number of work-
denotes the width of the corridor.
stations placed at the bottom also increases. Suppose the
number of subgroups is k, the number of workstations we
3.3.2 Total transportation distance of the fixed-track system
can placed in the bottom is
(j k
The layouts of fixed-track systems are shown in Fig. 8. ðk
1ÞðWH þWC þ2LB ÞþLB
lLWS
k>2
When an order is loaded, the parts and materials will be a ðk Þ ¼ W WS þW C ð14Þ
0 k2
sent to the headline of the fixed-track system. If a
workstation is assigned a task, the parts and materials will where WWS denotes the width of the workstation.
be transferred to the branch of this workstation; otherwise, Now, the total number of workstations becomes nEk+α(k).
they will be sent to the next workstation by the headline When there are more than two subgroups, we create a
directly. Therefore, the total transportation distance for a function β(k) to find the minimum number of subgroups.
part in the fixed-track systems comprises the path length of  dNPL e
aðk Þ
the headline (main track) and the path length of different bðk Þ ¼ nE k þ aðk Þ
dNPL e k  nE ð15Þ
dNPL e Otherwise
branches including the loading station, the assigned work-
stations, and the unloading station. The total transportation The number of subgroups required is calculated by
distance of the fixed-track system can be formulated as:  dNP e
NGL ¼ nE
L
NPL  2nE ð16Þ
TTD ¼ dNPL edB þ dH ð9Þ arg min bðk Þ Otherwise
where dB denotes the transportation distance in a branch, dH
where arg minβ(k)is the solution of k who minimizes β(k).
denotes the transportation distance in the headline, and
We assume that the workstations at the bottom have
dNPL e denotes the total number of workstations required
higher priority to be loaded. The number of subgroups that
for the new order.
the parts have to go through NSG is defined as:
Since the Eton’s track is fixed at a relatively high level,
in order to facilitate the operators, the parts and materials in  
the branch have to be transferred down to the height of the dNPL e
aðNGL Þ
NSG ¼ ð17Þ
workstation and then to be transferred up to the headline. nE
Int J Adv Manuf Technol (2009) 44:1219–1229 1225

Then, the transportation distance in the headline is defined as: by Eqs. 7, 13, and 16. Assign the tasks to the
workstations and then figure out the transportation
dH ¼ NSG nE LWS þ ð3NSG
2ÞdBSG þ lLWS þ 2ðWH þ LB Þ
distance dj from workstation j−1 to workstation j.
ð18Þ 2. Generate the order size with random variables Q and
where dBSG is the same as that in the SJ-Eton system. the service time Sj at workstation j. When launching a
new order, set the starting service time of the first entity
SS1,j as the finishing service time of the last order
4 Simulation and results FSQ;nPL plus the setup time and the relocation time.
3. If the queue length of the workstation j is  larger than
In this section, simulation is conducted to evaluate the the buffer size C, denoted by FS þ d dj v < FSi
C;j,
manufacturing system effectiveness of both the FRMHS  i;j
1 j
then FSi;j
1 ¼ FSi
C;j
d
j v, SS i;j
1 ¼ FS

sj
1 ,
i;j
1
and the fixed-track systems. otherwise, SSi;j ¼ max FSi
1;j ; FSi;j
1 þ dj v , FSi;j ¼
SSi;j þ sj .
4.1 Monte Carlo simulation 4. Collect the waiting time W and the material handling
time M, average time in the production line
AVT and
Monte Carlo simulation approach is adopted due to the the throughput TH ¼ Q FSQ;nPL
SS1;1 as follows
following reasons. Maione et al. [27] assumes that the
1 XX
Q nPL
material handling time, including the traveling and loading/
W ¼ ðSSi;j
FSi;j
1
dj =vÞ ð19Þ
unloading time, is negligible compared to the processing Q i¼1 j¼1
time. However, in apparel manufacturing, the processing time
X
nPL
is relatively short, which makes the proportion of material M¼ dj =v ð20Þ
handling time higher. For example, in many sewing factories, j¼1
80% of the production time is spent on material handling,
1 X

Q
and only 20% is spent on sewing [28]; thus, it is necessary to AVT ¼ FSi;nPL
SSi;1 ð21Þ
take the material handling time into the performance analysis. Q i¼1
Therefore analytical models become invalid, and simulation The other measures can be calculated by these parameters
is used to assess the performance of manufacturing systems using the model we defined in the previous section. Repeat
and material handling systems [29, 30]. Furthermore, in the from step 2 to ensure that all the measures are converged.
apparel industry, since the number of workstations required is
usually large, it is unpractical to formulate the simulation
using traditional software such as SIMAN and ARENA;
thus, discrete time Monte Carlo simulations using MATLAB
are formulated to do the comparative study.
To construct the simulation models for the FRMHS and Table 1 Input parameters for the simulation example
the fixed-track system, several assumptions are required to Input parameters Value
facilitate the comparative analysis:
Number of workstations for the order 60
1. The processing times follow identical independent normal Processing time (s) 5+N (20, 5)
distribution, and the production line is well balanced. Order size N (200, 20)
2. The first workstation never staves, and no preemptive Conveyor speed (m/s) 1.2
failures occur in the entire system. FRAGV speed (m/s) 1.2
3. The speed of handling is fixed no matter whether the Loading percentage 80%
FRAGV or carrier is loaded or not. Length of the workstation(m) 2.2
Width of the workstation (m) 1
4. The number of FRAGV and the carrier is enough for
Width of the corridor (m) 1
each order. With of the headline (m) 0.8
5. First in first out (FIFO) rule is used for all workstations. Length of the workstation branch (m) 1.5
6. Workstations with short transportation distance have Height of the workstation branch (m) 0.8
high priority to be loaded. Length of the loading and unloading station(λLWS) 3 LWS
Subgroup size in the Eton systems 21
Discrete time Monte Carlo simulations using MATLAB Subgroup size in FRMHS 13
are formulated to do the comparative study. This entails the Buffer size in the Eton systems 8
following steps: Buffer size in the FRMHS 2
Total setup time (s) 900
1. Given the number of workstations n and the loading
Total relocation time (s) 900
percentage PL, find the minimal number of subgroups
1226 Int J Adv Manuf Technol (2009) 44:1219–1229

Fig. 3 Monte Carlo simulation

results of comparing manufac-
turing system effectiveness

4.2 Simulation results computed by the measures in the FRMHS over the better
measures in both the SJ-Eton and the J-Eton systems. We see
We now illustrate the performance of the FRMHS by that the FRMHS improves the VAE by over 50%, the WIP
numerical experiments. For this numerical experiment, the and the AVT by over 20%, the MCE by over 10%, and the
practical inputs of the simulation are shown in Table 1. Each TH by over 3% in producing small orders. The underlying
simulation has been run for 200 replications with a period of reason is that the FRMHS shortens the setup time and
24 running hours per day to ensure the convergence. This material handling time and therefore the waiting time.
means that for all the performance measurements, the Although the FRMHS only improves the TH slightly, it
coefficient of variation, which is defined as the standard improves the TH in the practical worst case significantly.
deviation divided by the mean, is less than 5%. Based on the simulation results, using Eqs. 4 and 5, we can
Figure 3 compares the manufacturing system effectiveness find the TH in the practical worst case of the SJ-Eton
of the FRMHS and the fixed-track system in flexible system, the J-Eton system, and the FRMHS are 0.0192,
manufacturing of small order size. The improvement is 0.0195, and 0.0227 unit/s, respectively; therefore, the

Fig. 4 Workstation utilization

improvement under different
loading percentages
Int J Adv Manuf Technol (2009) 44:1219–1229 1227

Fig. 5 Total transportation

distance improvement under
different loading percentages

improvement of THPWC is 16.6%. Moreover, the setup time 5 Discussion and conclusion
of the FRMHS is much shorter than that of the Eton
systems, so the FRMHS can produce much faster than the Besides the potential advantages analyzed above, the
Eton systems at significantly lower inventory level. FRMHS also has other advantages:
– Since there is no physical boundary of production
4.3 Analytical formulation results
groups, resources such as idle workstations can be
shared by different production lines.
Based on the data in Table 1, using the analytical models
– The efficiency and effectiveness of production line with
we formulated in Section 3, the performances of the
parallel workstations can be enhanced. The queue for
workstation utilization and the total transportation distance
parallel workstations can be shared, which ensures that
are compared to evaluate the effectiveness of the FRMHS
the parts will be first come first serves. This can help
in addressing product proliferation or customization.
the supervisor quickly identify potential problems.
Results for the workstation utilization comparison are
Moreover, it facilitates the real-time rescheduling of
shown in Fig. 4. For the small order size, the FRMHS
the FRAGV and workstations [31, 32].
improves the workstation utilization by over 10%. The
– Due to the free-path property of the FRAGV, similar
improvement percentage increases as the loading percent-
functions can be grouped together for better resource
age or the order size decreases. This indicates that the
FRMHS can produce more at the steady state than the Eton
systems, and it is extremely effective for addressing product
proliferation in the apparel industry, especially when there
are several orders loaded in the same production system.
Figure 5 compares the total transportation distance of
the FRMHS and the Eton systems at the situation of
different numbers of workstations and loading percen-
tages. From the figure, we can conclude that the FRMHS
shortens the total transportation distance by about 68%.
Therefore, it could shorten the material transportation time
and then the waiting time. There are two underlying
reasons for these results. First, in the Eton systems, the
parts need to pass through the headline in the central
loading section, which induces extra traveling distance
into the system. However, in the FRMHS, the FRAGV can
turn in both directions on the main path. As a result, the
parts do not need to travel the full main path to return to
the loading station. Second, there is no vertical material
flow distance in the FRMHS. Fig. 6 Eton system from Sweden
1228 Int J Adv Manuf Technol (2009) 44:1219–1229

Flow direction
Branch for the Workstation process, lower the inventory cost, and have the capability of
fast responding to the customer demands and flexibly,

Unload
Load

suiting various products and volumes of orders. Due to

product proliferation, this potential advantage is important
for the apparel industry. However, in order to realize the
potential of the FRMHS, more development work is
Buffer necessary for the LPS.
Workstation
Head Line
Acknowledgment The authors would like to thank the reviewers
Fig. 7 Schematic layout of a basic Eton line and editors for their valuable comments and suggestions.

sharing, and it is convenient to expand production

Appendix
capacity.
– Like the Eton system, the proposed FRMHS also
Eton System The Eton system, designed by Inge Davidson,
can improve the utilization of labor resources sig-
the founder of Eton Systems Inc., is an UPS with
nificantly by replacing manual material handling by
computerized overhead conveyer and individually address-
automated material handling by the FRAGV. Parts
able workstations, which transports the materials by a
are tracked by the attached RFID tags, and
hanger-like carrier to increase the efficiency and reduce
therefore, they can be taken off the production line
WIP level of apparel manufacturing. Figure 6 shows the
anytime without messing up the parts’ information
appearance of the Eton system, and the newest generation
in the central controller.
of Eton Systems is Eton 5000 Syncro.
In conclusion, the detailed planning and design of the The main idea of the Eton system is to change the
FRMHS is presented. To evaluate the effectiveness of the progressive bundling system in the apparel industry to unit
FRMHS, Monte Carlo simulation and analytical models are production by using a hanger-like carrier to transport the
developed to compare the performance with that of the material through the production line. In addition, it replaces
fixed-track systems, which are widely used in the apparel the manual material transportation, which occupies valuable
industry. Our analysis shows that the FRMHS with FRAGV skillful operators’ time by an automated hanger system so
has substantial potential advantages over the fixed-track that the operator can concentrate on their jobs. Figure 7
systems in terms of manufacturing system effectiveness, shows the schematic layout of a basic Eton line. If
workstation utilization, and the total transportation distance. the workstation is assigned a task, the carrier will hand
Therefore, the FRMHS can streamline the manufacturing the material to the branch of the workstation; otherwise, the

Fig. 8 Configuration of the

simply joined Eton and the
joined Eton systems

U L

U L

A: Simply Joined Eton system B: Joined Eton system

Int J Adv Manuf Technol (2009) 44:1219–1229 1229

material will be handed to the next workstation directly by 11. Ragunathan S, Karunamoorthy L (2006) Modeling and dynamic
analysis of reconfigurable robotic gripper system for handling
the headline.
fabric materials in garment industries. J Adv Manuf Syst 5
There are three basic types of configurations for Eton (2):233–254. doi:10.1142/S0219686706000820
systems. If the number of workstations for an order is fixed, 12. Tait N (1996) Materials handling in the garment factory. Apparel
when the number of workstations required is small, the Int 27(5):20–23
13. Aldrich J (1995) Flexible materials handling. Appar Ind Mag 56
independent Eton system is often used, while for larger
(5):47–49
number of operations, the simply joined Eton system is 14. Beason M (1999) Here’s a new material handling solution. Textile
more suitable. However, if the number of workstations World 149(2):61–63
required is not fixed, the joined Eton system would be 15. Hill JE (1994) A study of the cost and benefits of a unit
production system versus the progressive bundle system. Clemson
preferable because the system could send the parts to Apparel Research Facility Pendleton SC. Available via DIALOG.
different subgroups, and it is not necessary for them to http://handle.dtic.mil/100.2/ADA299226. Cited 10 May 2008
travel through the whole system. The joined Eton system 16. Tait N (2004) How to reduce: materials handling costs. Apparel
also allows different production groups to operate as 45(10):28–32
17. Tait N (2007) How to cut down on handling time. Fashion
individual lines without sharing the loading and unloading
Business International Apr–May: 52–55
stations. Figure 8 shows the two types of configurations. A 18. Muller T (1983) Automated guided vehicles. IFS Publications,
detailed introduction to Eton systems can be accessed by UK
visiting the homepage of the company Eton Systems (http:// 19. Egbelu PJ (1993) Concurrent specification of unit load sizes and
automated guided vehicle fleet size in manufacturing system. Int J
www.eton.se).
Prod Econ 29:49–64. doi:10.1016/0925-5273(93)90023-E
20. Mahajan A, Figueroa F (1999) Automatic self-installation and
calibration method for a 3D position sensing system using
References ultrasonics. Robot Auton Syst 28(4):281–294. doi:10.1016/
S0921-8890(99)00042-1
21. Chittratanawat S, Noble JS (1999) An integrated approach for
1. Paraschidis K, Fahantidis N, Petridis V, Doulgeri Z, Petrou L, facility layout, P/D locations and material handling system design.
Hasapis G (1994) Robotic system for handling textile and non Int J Prod Res 37(3):683–706. doi:10.1080/002075499191733
rigid flat materials. Comput Ind 26(3):303–313. doi:10.1016/ 22. Kim JG, Goetschalckx M (2005) An integrated approach for the
0166-3615(95)00022-V concurrent determination of the block layout and the input and
2. Beamon BM (1998) Performance, reliability, and performability output point locations based on the contour distance. Int J Prod
of material handling systems. Int J Prod Res 36(2):337–393. Res 43(10):2027–2047. doi:10.1080/00207540412331333432
doi:10.1080/002075498193796 23. Fogarty DW (1992) Work in process: performance measures. Int
3. Jawahar N, Aravindan P, Ponnambalam SG, Suresh PK (1998) J Prod Econ 26(1–3):169–172. doi:10.1016/0925-5273(92)
AGV schedule integrated with production in flexible manufactur- 90059-G
ing systems. Int J Adv Manuf Technol 14(6):428–440. 24. Viswanadham N, Narahari Y (1992) Performance modeling of
doi:10.1007/BF01304622 automated manufacturing systems. Prentice-Hall, Englewood NJ
4. Tompkins JA, White JA, Bozer YA, Frazelle EH, Tanchoco JMA, 25. Saad SM, Byrne MD (1998) Comprehensive simulation analysis
Trevino J (2002) Facilities planning. Wiley, New York of a flexible hybrid assembly system. Integr Manuf Syst 9(3):156.
5. Datamonitor (2007, March) Consumer durables & apparel doi:10.1108/09576069810210394
industry profile: global. Retrieved April 6, 2008, from Business 26. Egbelu PJ, Tanchoco JMA (1984) Characterization of automatic
Source Premier Database guided vehicle dispatching rules. Int J Prod Res 22(3):359–374.
6. Census and Statistics Department, The government of Hong Kong doi:10.1080/00207548408942459
Special Administrative Region (2007) Industry production: prin- 27. Maione B, Semeraro Q, Turchiano B (1986) Closed analytical
cipal statistics for all manufacturing establishments by major formulae for evaluating flexible manufacturing system perfor-
industry group. Available via DIALOG. http://www.censtatd.gov. mance measures. Int J Prod Res 24(3):583–592. doi:10.1080/
hk/hong_kong_statistics/statistical_tables/index.jsp?charsetID= 00207548608919751
1&subjectID=4&tableID=100. Cited 15 May 2008 28. No author name available (2006) Materials handling: adding
7. HKTDC (2008) Industry focus garments: clothing. Available via cost—with no added value. Textile Network 1–2:16–19
DIALOG. http://garments.hktdc.com/content.aspx?data=gar 29. Smith JS (2003) Survey on the use of simulation for manufactur-
ments_content_en&contentid=173975&w_sid=194&w_pid= ing system design and operation. J Manuf Syst 22(2):157–171.
679&w_nid=11802&w_cid=1&w_idt=1900-01-01. Cited 15 May doi:10.1016/S0278-6125(03)90013-6
2008 30. Beamon MB, Chen VCP (1998) Performability-based fleet sizing
8. Chin KS, Pun KF, Lau H, Leung YS (2004) Adoption of in a material handling system. Int J Adv Manuf Technol 14
automation systems and strategy choices for Hong Kong apparel (6):441–449. doi:10.1007/BF01304623
practitioners. Int J Adv Manuf Technol 24(3-4):229–240. 31. Wong WK, Leung SYS, Au KF (2005) Real-time GA-based
doi:10.1007/s00170-003-1592-3 rescheduling approach for the pre-sewing stage of an apparel
9. Witt CE (1995) Automated material handling: breakthrough in manufacturing process. Int J Adv Manuf Technol 25(1–2):180–
textile industry. Mater Handl Eng 50(1):48–52 188. doi:10.1007/s00170-003-1819-3
10. Lin H, Taylor PM, Bull SJ (2007) Modeling of contact 32. Wong WK, Mok PY, Leung SYS (2006) Developing a genetic
deformation for a pinch gripper in automated material handling. optimization approach to balance an apparel assembly line. Int J
Math Comput Model 46(11–12):1453–1467. doi:10.1016/j.mcm. Adv Manuf Technol 28(3–4):387–394. doi:10.1007/s00170-004-
2007.02.019 2350-x