Computers & Industrial Engineering 61 (2011) 663675

Contents lists available at ScienceDirect

Computers & Industrial Engineering

journal homepage: www.elsevier.com/locate/caie

The Operations-Time Chart: A graphical tool to evaluate the performance

of production systems From batch-and-queue to lean manufacturing
Lluis Cuatrecasas-Arbos a,b,1, Jordi Fortuny-Santos a,c,, Carla Vintro-Sanchez a,c,2
a

Department of Management, Technical University of Catalonia (UPC), Spain

ETSETB School of Engineering at Campus Nord, 1-3, Jordi Girona, 08034 Barcelona, Spain
c
EPSEM School of Engineering at Manresa, 61, Bases de Manresa Ave., 08242 Manresa, Barcelona, Spain
b

a r t i c l e

i n f o

Article history:
Received 10 May 2010
Received in revised form 12 April 2011
Accepted 30 April 2011
Available online 7 May 2011
Keywords:
Operations-Time Chart
Efciency
Work-in-process
Lean manufacturing
Modeling

a b s t r a c t
In this paper, we describe the spreadsheet modeling of manufacturing systems by means of the OperationsTime Chart (hereafter, OT-Chart), a graphical tool for an automatic time-phased representation and measurement of the operation of production systems, developed by the authors. In order to improve the design
of a production system, it is necessary to know the key performance metrics of the system (productivity,
lead-time, inventories, downtimes and wait times) and identify the effects of design parameters on system
performance. Calculating some of these magnitudes can be very complicated, especially for production systems involving multiple and conuent processes, with different cycle times and lot sizes. The OT-Chart permits a visual tracking of the aforementioned parameters throughout each process, and like a simulation
tool, the program calculates and displays the effects of changing input parameters. A special version of
the Chart has been designed for lean manufacturing environments, where visual tools are much appreciated. The OT-Chart provides tracking of different types of waste and supports inventory supermarkets
and pull scheduling. The paper includes a case study: a plant is redesigned from a conventional batchand-queue production system into a lean manufacturing system with the help of the OT-Chart (it is used
to test the performance of each layout) allowing managers to evaluate and rene their designs.
2011 Elsevier Ltd. All rights reserved.

1. Introduction
For many companies, the key aspects of current competitiveness
in order to raise customer satisfaction focus on delivery time, quality
and cost. To attain and sustain competitive advantage, some companies have implemented manufacturing practices such as mass
customization or lean manufacturing in search of more exibility
or agility (Fullerton, Cheryl, & McWatters, 2001). Lean manufacturing is known as the production of the necessary items in the necessary quantities at the necessary time (Womack, Jones, & Ross, 1990).
Shah and Ward (2003) associate lean philosophy to improvements
in operational performance measures that encompass the key aspects of competitiveness.
Ben Naylor, Naim, and Berry (1999) state that leanness means
developing a value chain to eliminate all kinds of waste (activities
that do not add value to the product). In consequence, lean

Corresponding author at: EPSEM School of Engineering at Manresa, 61, Bases de

Manresa Ave., 08242 Manresa, Barcelona, Spain. Tel.: +34 93 877 72 81; fax: +34 93
877 72 02.
E-mail addresses: lluis.cuatrecasas@upc.edu (L. Cuatrecasas-Arbos), Jordi.Fortuny@upc.edu (J. Fortuny-Santos), carla.vintro@upc.edu (C. Vintro-Sanchez).
1
Tel.: +34 93 401 68 13.
2
Tel.: +34 93 877 72 81.
0360-8352/$ - see front matter 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.cie.2011.04.022

thinking leads to the minimization of inefciencies or muda as they

are known in Japanese.
According to the philosophy of lean manufacturing, a major
source of waste is work-in-process (WIP). Therefore the primary
goal of a lean production system is to reduce manufacturing costs
by cutting down the in-process inventory (Watanabe & Hiraki,
1997). In conventional production processes, inventories serve as
a protective shield against the effects of many types of failures
(machine breakdown, lack of quality, long machine setup time,
etc.), and therefore inventories conceal such problems.
White and Prybutok (2000) state that although the high levels
of inventory present in traditional production plants (as a result
of large-lot manufacturing methods) allow greater exibility for a
wide array of products, this comes at the cost of efciency because
inventory covers up problems but does not solve them. Inventories
imply an additional consumption of resources because they have to
be managed, stored and manufactured. For this reason, lean manufacturing aims at efcient, well designed processes with no
imbalance or waste, with a minimized cycle-time, that operate
with minimal transfer batches in order to achieve minimized
inventories (Oliver, Delbridge, & Lowe, 1996). Besides, when inefcient activities are removed from the production process, WIP
inventories originated by such activities are signicantly reduced
(Fullerton et al., 2001).

664

L. Cuatrecasas-Arbos et al. / Computers & Industrial Engineering 61 (2011) 663675

2. The Operations-Time Chart: a methodology for the evaluation

of process design
In the initial stage of manufacturing planning, different alternatives are examined. Manufacturing system modeling enables managers to predict the performance of the system and test how
sensitive it is to design parameters (Koo, Moodie, & Talavage,
1994). According to Huettner and Steudel (1992) there are three
possible approaches: Simulation, analytical methods (i.e. queuing
models) and spreadsheet models. In this paper we describe a
spreadsheet based graphical model.
The Operation-Time Chart developed by the authors (Fig. 1) is a
graphical evaluation tool used in the design and improvement of
production systems and related processes that allows companies
to ask what-if questions about their systems without actually
implementing, changing or disrupting the systems themselves.
Thus, due to the complexity of most production systems, simulation analysis becomes a valuable tool for studying the system
dynamics and its behavior (Denzig & Akbay, 2000).
The main objective of simulation tools is to observe the behavior of temporary entities (elements of a modeled system i.e. manufactured parts in a production plant) and collect information on
them. The authors designed the OT-Chart to represent the operation over time of complete production systems made up of parallel
and conuent processes as a sort of simulation tool with a different
approach than those of usual techniques such as discrete event
simulation.
The OT-Chart displays the key parameters of a repetitive single
product manufacturing system in response to a specic schedule
under a push strategy (an MRP or EOQ-based system, where
products are produced according to a schedule derived from estimated demand).
Since excessive inventory is considered to be a waste, work-inprocess inventories are related to the efciency of the entire

system and therefore they must be taken into account in order to

improve the design of the system or to compare two alternative
congurations. For this reason, the OT-Chart calculates and displays the instant, average and maximum levels of work-in-process
inventories.
After the basic OT-Chart, we also present a version especially
designed for lean manufacturing systems under pull production
scheduling strategies, since pull production planning is one of
the cornerstones of lean management, where production is
demand driven (Sakakibara, Flynn, Schroeder, & Morris, 1997). As
one of the enablers of lean manufacturing is to decrease the time
needed to change equipment for different production runs, users
of the OT-Chart can choose the one-touch machine setup (OTED)
option.
Finally, the OT-Chart for lean environments includes supermarkets or buffers of limited capacity that supply specic processes in order to control and level production when continuous
ow from raw material to customers is not possible.
Although the OT-Chart is not a discrete event simulation package, as a time-phased analysis tool, it is used to test the performance of a process design, regardless of the complexity of its
parallel and assembly tasks. Users may see and read where inventories build up and where and when workstations are idle. Users
can trace the evolution of process metrics, since the development
of the process is on the screen. This information helps them to
evaluate new scenarios and implement processes that are more
efcient.
Methodologies to optimize design, balancing and scheduling of
production systems have been widely explored from the point of
view of operational research by means of both analytical and heuristic procedures (Becker & Scholl, 2006; Ghosh & Gagnon, 1989;
Linn & Zhang, 1999; Rekiek, Dolgui, Delchambre, & Bratcu, 2002;
Scholl & Becker, 2006). The OT-Chart, like a simulation tool, may
help researchers evaluate the goodness of their analytical/heuristic

Fig. 1. OT chart applied to a set of simultaneous, sequential processes.

L. Cuatrecasas-Arbos et al. / Computers & Industrial Engineering 61 (2011) 663675

methods by testing the behavior of the manufacturing system under alternative operating policies.
Many commercial graphical simulation packages can be found
in the market but an electronic spreadsheet can also be used as a
simulation tool (Albright, Wayne, & Zappe, 1999), as we have done
in our research.
The OT-Chart is intended for continuous deterministic simulation models. It uses time slicing techniques instead of the most
usual discrete event mechanism and thus allows a continuous representation of the evolution of the production run.
The most visible output of the OT-Chart is similar to a Gantt
chart because it displays the ow of the process over time, but
its logic is more complex that a mere graph. In Section 3, we describe the conceptual basis behind the OT-Chart and the worksheets it is implemented upon. In Section 4, we analyze a
complete case (including scheduling in lean manufacturing) and
in Section 5 we compare the OT-Chart against other pieces of
software.
3. The logic behind the Operations-Time Chart
The Operations-Time Chart is implemented on an Excel workbook containing several sheets. A rst spreadsheet gives information to the user on how to use the application. On a second
spreadsheet (The Process Sheet), the analyst types a description
of the processes and terminal operations of the manufacturing system. That spreadsheet is designed to handle up to three different
variants of the same product.
On a third spreadsheet (The Implement sheet) the analyst
lists, for a specic product, how many workstations there are in
the system; which work elements (from the previous spreadsheet)
are performed on every workstation and the precedence relation
between operations.
On a fourth spreadsheet (The Diagram sheet) the application
automatically constructs and displays the OT-Chart. Besides displaying a Gantt chart, the OT-Chart calculates and shows the instant values of the WIP accumulated at every activity. The
software computes and shows the key parameters of the process.
Finally, efciency and waste statistics are displayed on a fth
spreadsheet.
In Section 4, a full example is displayed in Tables 1 and 2 and in
Figs. 25.
3.1. The Process sheet
This sheet is devoted to the manual input of the attributes of
operations and processes. It includes the denition of the manufacturing system that will be used to manufacture the products. Even
though the main time unit is the hour, in cell C2, users key in the
unit of time that will be used for the processes (generally, seconds).
In cell C3, it is necessary to write the relation between the desired
unit of time and the hour (in our case, it is 3600).
Each row represents an operation or simple process (i). Column
F is used for the frequency of each operation (fi) -the number of
times that the operation i is performed for every unit of nal product-. Column I is the manual work or time (in seconds) (tsi) that the
operator needs to perform operation i. Column J is the machine
time or time (in seconds) (tmi) that the automatic machine needs
to perform operation i without worker involvement. Times in columns I and J come from time studies or estimations from operators
and foremen, etc.
In column N, users detail quality control activities that will be
performed in each operation. Finally, hidden column S is the time
(TSi) that the operator ideally devotes to operation i per unit of nal
product (Eq. (1)) and hidden column T is the time that the machine

665

(TMi) ideally devotes to operation i per unit of nal product (Eq.

(2)). Time unit for columns N and S is the second, as specied by
the user

TSi tsi  fi

TMi tmi  fi

3.2. The Implement sheet

Some cells must be completed manually. They keep information
about the production system that depend on the designer or that
have been imposed by third parties. Cell AB2 is the scheduled production per day (in pieces); Cell AB4 is the production run length
(PR) (in pieces); Cell AG2 is the working day (hours) which estimates the maximum capacity of the factory; Cell AG3 is the duration of the daily planned non-working time (Planned maintenance,
beginning of day set-up, end of day clean-up and other scheduled
stoppages). Cell AG4 gives the available time (AvT) or practical
capacity and it is calculated as Cell AG2 less Cell AG3.
Cell AK2 does the calculation of the takt time (in seconds). The
takt time is dened as the reciprocal of the average customer demand rate or the working time available over the scheduled daily
production (Abdulmalek & Rajgopal, 2007)

Takt time

Working time available

Av T

Scheduled daily production

SP

Cell AK3 calculates the theoretical minimum number of workers (MNW) required to satisfy the takt time constraint (Eq. (4)). It
is the necessary time to complete the manual work in the production run divided by the takt time. This value is rounded up in cell
AK4.

MNW

P
PR  CTW j
Takt time

In this sheet, each row is a workstation (j).

Columns EN are used to key in the operations in sheet processes (i) that are allocated to workstation j. This allocation is manually performed by the user but the sheet helps the user by giving
information on the unbalance for each workstation, the possible
production and so on. In column P, users give information on the
precedence between workstations. Colum R is the time devoted,
in each workstation, to operations that add value to the product
(they transform it in a way that the customer is willing to pay
for) VAj (Eq. (5)). According to previous calculations, this time is expressed in seconds.

VAj

TSijj

In column S, the user keys in the time devoted, in each workstation, to operations that do not add value to the product NVAj (such
as rework due to defects). That amount of time may come from an
estimation of the real process or may be a target value or even a
safety margin to tackle process variability. Again, this time is expressed in seconds. Column T is the value add time for the machine
(VAMj) (Eq. (6)) and column U gives detail on the non-scheduled
stoppages (estimation per workstation) (NSSj).

VAMj

TMijj

Cell R30 shows the summation of the value add time in column
R. The same applies to the non-value add time in cell S30. Cell R31
is the summation of the add value time and the non-add value
time, thus it is the manual process time. Cell T30 is the summation
of the machine time in column T, and cell U30 is the summation of
the machine stoppages in column U. Cell T31 is the summation of
cells T30 and U30, giving the total machine time.

Processes and sub processes

666

Table 1
Operations to manufacture a USB ash drive.
Operation
#

Frequency per unit of nal product

Description

Time per operation (s)

16 GB memory Person

16 GB memory Machine

(Number is automatically assigned)

A: Manufacturing of printed circuit board
A1: Manufacturing of Rom memories

A2: Manufacturing of other parts of the board

B2: USB connector inserted and soldered

Change components and reload PCB

Components mounted on board (AUTOMATIC)
PCB unload from charger rack
PCB loaded onto soldering machine
Soldering (AUTOMATIC)
PCB unload from soldering machine

0.02
1
0.02
0.02
1
0.02

600

7
8
9
10
11
12

Change components and reload PCB

Components mounted on board (AUTOMATIC)
PCB unload from charger rack
PCB loaded onto soldering machine
Soldering (AUTOMATIC)
PCB unload from soldering machine

0.02
1
0.02
0.02
1
0.02

120
200

13
14
15

Take a reel of solder

Fit components on board
Soldering

0.02
1
1

10
12
25

16
17
18

Take a reel of solder

Adjust connector
Soldering

0.02
1
1

10
20
40

19
20
21

Methacrylate is loaded
Mould is made
Moulds are unloaded

0.05
1
0.05

30

22
23

Communication software copied

Software is tested

1
1

15
35

24
25
26

Board is adjusted to mould

Make methacrylate melt to adjust
Visual inspection

1
1
1

20
20
25

27
28

Place on control device

Connections and performance are tested

1
1

25
30

29

Reprocessing of defective parts

0.24

120

30
31
32

Flash drive placed on shaped plastic sheet

Place instruction sheet, accessories and printed cardboard
Case is thermo soldered

1
1
1

12
25
10

55
90
180
30
100
600
60

30
150

20

C: Manuf. of plastic moulds

60
45

D: Control Software
15
35

E: Final assembly of the USB ash drive

F: Final test
30

G: Rework on defect
H: Packaging

10

L. Cuatrecasas-Arbos et al. / Computers & Industrial Engineering 61 (2011) 663675

B: Components joined together

B1: Fitting and soldering of two boards

1
2
3
4
5
6

667

L. Cuatrecasas-Arbos et al. / Computers & Industrial Engineering 61 (2011) 663675

Table 2
Comparison between lean and batch and queue designs.
Key design parameters

Lean values

Batch & queue values

Difference lean vs. B&Q

Waste:
Production run size (Units)
Process: total work (VA + NVA) time (process + stoppages) (s)
Process: total machine time (process + stoppages) (s)
Average WIP inventory (whole process) (Units)
Max. WIP inventory (Units)
Waiting time: operators waiting in workstations (hours)
Waiting time: Setup time (for the same production) (s)
Motion and transportation (s)
Average quantity of defective parts (%)

200
470
384
21
35
4.2
1235
119
0.5

2,500
525
389
1588
2845
70.1
3875
312
3.5

2300
55
5
1567
2810
66
11,563
193
3

92.0%
10.5%
1.3%
98.7%
98.8%
94.0%
298.4%
61.9%
85.2%

Efciency:
Number of workstations (nal design, after grouping them)
Total Daily overtime (Man-Hours)
Productivity (average product output per working hour)
Lead time for production run (hours)
Lead time rst transfer batch (hours)

10
17.5
6.3
4.5
0.6

11
23.0
5.9
65.0
15.3

1
6
0
60
15

9.1%
23.9%
6.3%
93.0%
95.9%

Fig. 2. Task grouping and their values (time and transfer batch size) for every workstation (The Implement sheet).

Cell U32 gives information on the portion of work that does not
add value to the product (and therefore it is waste)

NVA P

P
P
NVAj NSSj
P
P
P
VAj VAM j NVAj NSSj

Columns W, X and Y deal with additional time. In column W it is

necessary to indicate the scrap rate (percentage of failed assemblies
non-recoverable through rework). It is expressed as a percentage
(defj) of the production order, as usual in the industry. The process
will have to manufacture these additional pieces in order to complete the production order. Cell W32 calculates the average portion
of defective products as the summation of the defectives in each
workstation, divided by the number of workstations.
In column X, users may enter the time (per piece) devoted to
transportation (Trj) if the operator of the workstation has to devote
time to transportation between workstations. Cell X32 calculates
the average time devoted to transportation as the summation of
the transport times in each workstation (cell X30), divided by the
number of workstations.

Column Y is the setup time (Setj) per production run. Cell Y32
calculates the average setup time per workstation as the summation of the setup time (cell Y30), divided by the number of
workstations.
Column AA is the number of workers in workstation j (Wj) and
column AB is the number of parallel machines (Mj) in workstation
j. These allocations are performed by the user and the automatic
calculations in the workbook help the user determine which
arrangement is better. Column AC is the transfer batch size (number of pieces) (batchj). Column AD is the allowed overtime (in
hours) per day (Overj). Colum AE is the time (in seconds) that a
workstation will stop after completing a transfer batch in order
to avoid overproduction (SBj). This concept is not used in push production systems, where productivity at each station is achieved by
maximizing its output, but it is an important tool in lean manufacturing and even in the Theory of Constraints.
Cell AA30 adds up the number of workers in the workstations,
and cell AB30 calculates the total number of machines. Cell AB30
is the summation of the daily overtime in each post. Cell AB32
computes the daily overtime as the equivalent number of working

668

L. Cuatrecasas-Arbos et al. / Computers & Industrial Engineering 61 (2011) 663675

Fig. 3. Part of the Operations-Time Chart for conventional production of USB ash drives (The Diagram sheet).

Fig. 4. Workstation data for the lean implementation (The Implement sheet).

time (in days). Cell AC31 locates the biggest transfer batch between
workstations and cell AC32 computes the average transfer batch
size as the summation of sizes of the transfer batches between
workstations, divided by the number of workstations.
Previously, the operations have been assigned to workstations,
and now workstations can be grouped in order to improve the design of the plant. For example, we decide that two different workstations will be combined into a single station because their
manual cycles are lower than their automatic cycles. In column
AG, a number is assigned to each one of the previous workstations.
Stations with the same number will be the same. This allocation is
done by the user and tested by the diagram. Column AH gives
information on the (new) number of operators (Wk) assigned at
each workstation k.

Cell AG30 nds the number of grouped workstations and cell

AH32 gives the number of operators (W) after grouping workstations. The closer our design (W) is to MNW (Eq. (4)), the more efcient it will be.
Column AJ is the cycle time (the average time between two contiguous completed pieces coming at the end of the station) for each
grouped station (Ck) (Eq. (8)), per operator. CTMk (in hidden column
AS) will be analyzed in a following paragraph. If the user has specied, in this same sheet, that One-touch exchange of die techniques are in use, then the cycle time will be corrected by a
factor (f) according to Eq. (9), where PR is the production run
length (number of pieces). It means that changing tooling and xtures is reduced to a single step which takes no more than a cycle
time

L. Cuatrecasas-Arbos et al. / Computers & Industrial Engineering 61 (2011) 663675

669

Fig. 5. The Operations-Time Chart for the production of 200 USB ash drive under a lean manufacturing strategy (The Diagram sheet).

CTM k
f
wk
PR 1
f
PR

Ck

8
9

Column AK is analogous for machine time (CMk), taking into account that several consecutive machine processes can be performed in the same workstation after grouping (Eq. (10)). Ck and
CMk are the main parameters for the Operations-Time Chart. The
real cycle time of the process is the longest of Ck and CMk.
Users may want to change some design parameters in order to
improve the efciency of the workstation. It is preferable that the
manual time is longer than the machine time in order to avoid having the worker waiting, since this is considered waste while the
idle time of the machine is not considered waste

CM k MaxCMajjk  f

10

Cell AJ30 reports the longest manual cycle time and cell AK30
is the longest automatic cycle time. Cell AJ32 gives information
on the maximum imbalance time, dened as the difference between
the longest and the shortest manual cycle times. Cell AK 32 does the
same for machine times. Users may want to change the arrangement
of the production system in order to achieve a better balance
between workstations that guarantees a better ow of the process.
Some of the necessary calculations to reach the results in the
above described cells are performed in hidden columns (from AO
to AW):
Column AO calculates the operator time per piece (in seconds).
This is the time (CWk) that each operator in each workstation has
to devote to each part: processing time including value add time
and non-value add time, and considering overtime necessary to repeat operations to complete a lot when faulty pieces have been
manufactured and time devoted to transportation (Eq. (11))

CW jjk VAjjk NVAjjk  1 defjjk

tr jjk
batchjjk

11

Column AP calculates the machine time (CMj/k) per part (Eq.

(12)), in seconds

Cmjjk VAMjjk NSSjjk  1 defjjk

12

Column AR is the setup time per unit. It is the setup time per
production run divided by the production run length (Eq. (13)). If
OTED is turned on, setup time is automatically switched to zero

ST jjk

set jjk
PR

13

Column AS is the total operator time (in seconds) per workstation per piece (CTWk) including manual work, setup and workstation halts between batches (per piece)

CTW jjk CW jjk

SBjjk
ST jjk
batchjjk

14

Column AT is the total machine time (in seconds) per workstation per piece TTMjjk

TTMjjk Cmjjk

SBjjk
ST jjk
batchjjk

15

In column AW, the equivalent machine cycle is calculated (Eq.

(16)) (CMaj), taking into account the number of parallel machines
devoted to each process j in the workstation. In next sheet (the
Diagram sheet), SB is set to zero.

CMajjk

TTM jjk
M jjk

16

Finally, column AM gives the maximum daily production output

(Pk) (number of parts per day) for each workstation (Eq. (17)).

Pk

Av T Ov erk  3600
MaxC k ; CM k

17

The lowest Pk (which can be found in cell AM31) must be compared with the scheduled production because the real production
of the system cannot be higher than the lowest Pk.
Users then may want to balance the production in each step of the
process. For example, they can try by eliminating people in workstations with overproduction, or they can indicate that overtime is negative (it means that during a part of the regular work time, the
worker will not be producing in his/her workstation). In a similar
way, users may detect that some workstations need to improve their
pace and therefore it is possible to assess the consequences of reducing the non-value add time, setup time or the scrap rate.
Cell AM34 gives the overall labor productivity, which is dened
as the daily production divided by the daily number of labor-hours

Productiv ity

MinPk
P
W  Av T W j  Ov ertimej

18

3.3. The Diagram sheet

This sheet is the representation of the Operations-Time Chart.
Hidden cells keep the necessary logic (the numbers) to draw the
bars.

670

L. Cuatrecasas-Arbos et al. / Computers & Industrial Engineering 61 (2011) 663675

Cell Z2 copies the production run length. In cell Z3 users may

enter the scale lap (DT). In our example it is set to 1000 s. From
now on, the diagram will be operating in time slicing mode: each
column represents 1000 s in the time scale. Every ve cells, there is
a time label.
Columns BD hold the headings of the diagram. For each workstation they show the name of the workstation, the next task, the
cycle time and the transfer batch size (all this data come from
sheet Implement). The program detects the longest cycle time
and therefore it knows if a specic process is a bottleneck or not.
In addition to the option that each station may start operating
at a given start time (specied by the user in column C, since the
software does not support stochastic simulation), the chart includes scenarios where the rst process begins to operate upon
delivery of external materials (also specied by the user in row
7). This is shown in a line just below the timescale labeled Supply
(if any). These entries can cause delays at the rst station and
other stations connected with the rst one.
For every workstation, a hidden row counts the real production
of the workstation at each time slice, according to Eq. (19) as long
as the station has enough raw materials to operate (delivered by
the precedent workstations) on an earliest start (Push) basis. Precedence rules are controlled in a hidden row for each workstation. To
avoid run time errors, other aspects have to be addressed, such as
the production run length or the initial wait for each workstation

Productiontjk ProductiontDTjk

DT
Cyclek

19

Another row counts for when the transfer batch is met and then
it displays the batch size.
Based on previous calculations, another row draws a horizontal
line () if the station is working, and draws an arrow when the
transfer lot is met. If the station is the bottleneck of the system, instead of a line, the picture is a double line (=).
Another hidden row keeps, lap by lap, the amount of parts
received from precedent workstations (whether it has or it has
not been consumed yet), according to the precedence rules.
Another row calculates the current amount of work-in-process
in the workstation (unprocessed parts received from precedent
workstations and assemblies that have been nished but still do
not have been transferred to the following process). Then, some
rows show the amount of work-in-process in each workstation,
at each time slice, as a pile of horizontal lines. Each line represents
one fth of the maximum level of work in process.
Bar charts for production and inventories permit users to
observe the progress of each task and to spot problems. They also
help size up the relation between variables. When we increase the
batch size in one workstation, the diagram shows how the following stations operate in batches: they work for a certain period of
time (The larger the batch size, the longer the period) and then
wait till the following batch arrives. The larger the batch, the larger
the amount of work-in-process in the system and the lead-time for
each workstation will be, because more time is lost waiting.
Some cells show performance indicators, which are mostly
derived from an automatic scan of the hidden rows that keep track
of the production and inventory. For every workstation the software detects the earliest start ESk and the earliest nal EFk of the
production run. Then, the lead time (time between the beginning
of the process and its completion) is the difference between EFk
and ESk. Another cell keeps the total wait time, calculated by Eq.
(20). It is the lead time minus the necessary time to complete
the production run (PR).

Wait timek EF k  ESk  Cyclek  PR

20

A cell keeps the maximum inventory achieved in each workstation and another calculates the average value (AvWIPk) (Eq. (21)).

Av WIPk

WIPtjk
EF k  ESk

21

Under the bar graph, the sheet controls (in cell N254) the total
production delivered by the production system (which is the addition of the different batches completed in the last workstation,
computed in row 250).
Row 255 counts the total work-in-process in the system (as the
addition of the different workstations) at each lap (WIPt). Rows
256259 show the amount of total work-in-process, at each time
slice, as a pile of horizontal lines. Cell E259 keeps the maximum
inventory in the process and cell W260 computes the average overall amount of work-in-process (Eq. (22)). N is the number of laps
with positive inventory

Av WIP

WIPt
N

22

The analyst can create new scenarios by modifying decision

variables (for instance, lot sizing) and the value of uncontrollable
variables as demand. For every scenario, the Chart depicts the temporal progress of the process until its completion and records the
values of dependent variables or performance measures such as:
 Total process lead time (for the production batch). It is the earliest nal for the last station less the earliest start for the rst
station.
 First-delivery lead time (for the rst transfer batch).
 Operation lead time: The time each workstation is being used or
idle since it starts processing the rst unit in the production run
until it completes the last unit.
 Total waiting time: for every operation and for the whole
process.
 Average waiting time (total waiting time divided by the number
of workstations).
 WIP inventory: average and maximum WIP for every operation.
 Average WIP in the plant and time at which maximum inventory is reached.
 Instant values of WIP for every workstation.
 Hourly production of the production system (in nished products per hour).
 Cycle time per unit of product.
Times are given in seconds and, because of the length of the
process, in hours in order to be meaningful.
3.4. An illustrative example
In this section, we explore the Operations-Time Chart (Fig. 1) for
a manufacturing system that comprises interconnected, sequential
and/or simultaneous processes, with different transfer batch sizes.
The core data for each operation (identication, cycle time,
transfer batch size, and the target operation of the transfer) specied by the user are listed on the left-hand border of the chart.
Operation 1 delivers parts to Operation 2, this in turn delivers
them to Operation 3. Meanwhile, in a parallel process, Operation
4 delivers parts to Operation 5. Finally, Operation 3 and Operation
5 deliver work pieces to a third downstream process (which could
be a nal assembly), denoted here as Operation 6.
The horizontal arrows represent the time each transfer batch of
product is being processed by each workstation. The length of the
arrows is proportional to the processing time, represented according to the timescale.
The sizes of the transfer batches and the cycle times are not
equal for all the operations. For example, Operation 4 transfers a
batch of 100 units to Operation 5, while Operation 5 delivers 200
units to Operation 6. It does not imply that Operation 5 has to

671

L. Cuatrecasas-Arbos et al. / Computers & Industrial Engineering 61 (2011) 663675

receive 200 parts from Operation 4 to start working. As soon as

Operation 5 receives 100 units from Operation 4, these parts can
be transformed.
Once a transfer batch has been completed in a workstation, it is
sent to the following operation, which, if it is idle and ready, will
process it. For example, the rst batch of 100 pieces, which leaves
Operation 1 when t = 500 s (the cycle time for Operation 1 is 5 s),
directly enters Operation 2 for processing (we have used vertical
dotted lines to represent the motion of transfer lots between processes), but, since Operation 2 is longer than Operation 1, the
second batch (which leaves Operation 1 at t = 1000 s) cannot enter
Operation 2 until t = 1500 s. Likewise, the subsequent batches
must queue longer and longer before entering Operation 2 (When
t = 1500 s, up to three transfer batches are waiting to enter Operation 2), because of the imbalance between Operation 1 and
Operation 2.
The material is transferred equally from Operation 2 to Operation 3, but as the latter has a shorter cycle time, each incoming
batch is processed directly upon arrival from the previous workstation. When Operation 3 nishes a transfer batch, that workstation
has to wait until the next batch arrives.
The software calculates and displays the key values for each
operation on the left-hand side of the chart. These values comprise
the average and maximum levels of inventory (which are based on
measurements taken at a certain frequency and which arise from
the accumulation of material queues), the sum of all waiting times
for the process (such as those from external provision or imbalances between operations), and the total lead time for each operation, which is read directly on the horizontal axis.
In Fig. 1, the user may nd at what time a specic transfer lot
moves from one process to another or leaves the production system. For example, the rst 50 units are completed after 3750 s
and the whole production order (500 units) is nished when the
time is 8500 s. The user might also count the total work-in-process
on the shop oor at any moment (the number of parts waiting at
each operation). For example, when the time is 3000 s, the total
WIP is 1000 units. The chart calculates plant productivity in terms
of units of nished product delivered per hour. In consequence, the
OT-Chart makes it possible to see what activities constrain the
development of the process and therefore need to be redesigned
in order to achieve a better result in terms of production lead-time,
productivity and waste.
3.5. The Diagram sheet in pull mode
The authors have developed the computerized Operations-Time
Chart to support lean implementations, with the following features
based on the lean philosophy:
 The option to work in pull mode. In push mode, one station
starts working as soon as the necessary materials arrive from
precedent workstations. In pull mode, one station cannot start
working until the following stations need more material. This

Inv entory lev elxyjt




material, work-in-process, or nished goods is kept as a buffer

to schedule variability or an incapable process while controlling
and limiting inventory between operations (an operation will
not be performed unless there is room to allocate its product
in the supermarket). Users of the OT-Chart can specify the
capacity of the supermarkets they require.
Delivery of the nal product, according to the takt time, in order
to match production to demand.
A graphic report of the ow, which is a key aspect of lean systems. One of the aims of lean manufacturing is moving one
item, or a small and consistent batch of items, at a time through
a series of processes as continuously as possible, with each step
making just what is requested by the next step and when it is
needed.
Control of waste. The software calculates and displays, for each
operation, the average and maximum WIP levels, the sum of all
waiting times and the total lead time for each operation. In lean
manufacturing, these parameters are considered important
types of waste (activities that consume resources but create
no value from the customers point of view) and therefore must
be minimized.
One-touch exchange of dies (OTED). Another important issue of
lean manufacturing is that machines must be easy to set up
quickly (Harmon & Peterson, 1990).

In push mode, the value of the takt time, calculated in sheet

Implement, is ignored, but now in pull mode, the takt time drives
the pace of the process.
In cell BB2, users may enter the value of the takt time they want
to essay. It is possible to keep the takt time calculated in the previous sheet or to try a different value. For greater values, the process
will slow down in order to avoid overproduction. For shorter values the process will do nothing but show the notice Takt time is
not met, because it is impossible to go faster with the current layout. Then, users may want to redene the design till the aim is
reached.
Now, for the last workstation, Eq. (19) becomes Eq. (23), because a product needs only be produced at the same rate that it
is being sold.

Productiontjk ProductiontDTjk

DT
MaxfCyclek ; Takt timeg

Single-piece ow (Li & Rong, 2009; Miltenburg, 2001) is one of

the key concepts of lean manufacturing, but in practice that needs
a similar cycle time for each workstation which is not always possible to achieve. Therefore, different performance times are leveled
by means of supermarkets (Abdulmalek & Rajgopal, 2007) or buffer
inventories, a tightly managed amount of inventory. Upstream process (x) replenishes the supermarket (as long as there is empty
room) while a downstream workstation (y) withdraws parts from
the supermarket (Eq. (24)). If the supermarket becomes full,
previous processes have to stop because production is not
necessary.

8
9
Capacityxy
>
>
<
=


Min
Inv entory lev elxyjtDT Tranferences from xtDT  Transferences to ytDT
>
>
: Max
;
0

is the logic that drives the Pull version of the diagram.

 WIP levels adjusted to supermarket guidelines. In lean environments, a supermarket is a shop oor line-side location, where
parts are sorted and stored. A xed maximum amount of raw

23

24

For other workstations, Eq. (19) is still valid but it is necessary

to control two issues: the rst one, as usual, that there are enough
materials to process (now, in the supermarket x-k that feeds station k); and the second one, that there is room in the supermarket

672

L. Cuatrecasas-Arbos et al. / Computers & Industrial Engineering 61 (2011) 663675

k-y that the station replenishes (Eq. (25)). Besides, the calculation
controls whether the production run or the initial wait are met.

Productiontjk ProductiontDTjk


DT
Min
; Empty roomky ; Inv entory lev elxk
Cyclek
25
The capacity of each supermarket is dened by the user (for
example, in cell AB18). The program allows users to dene the
maximum transfer batch size as a percentage of the supermarkets
capacity (cell Y49) in order to avoid transfer batch sizes too large
for the supermarkets capacity. In a pull mode, the desired lead
time can be achieved by controlling the amount of work-in-process
and therefore, the capacity of the supermarkets. It is possible to
change the capacity of the supermarkets, one by one and see the
effect on the lead time.

4. Using the Operations-Time Chart to evaluate and improve the

design and performance of a real process
In this section, we use the Operations-Time Chart to evaluate
the performance and improve the design of the production system
of a USB ash drive manufacturing plant, and to transform it from a
batch-and-queue system into a lean manufacturing plant. By comparing the OT-Chart for each implementation, we foresee the
improvements resulting from the change to a lean system.
We rst explore and evaluate the design of a conventional,
based on a push strategy, batch-and-queue plant for manufacturing a 16 GB USB ash drive, by means of the OT-Chart. Later, the
results are used to redesign and transform the plant (Meral & Erkip,
1991) according to the lean manufacturing philosophy, under a
pull production strategy.
This case study comprises the different operations summarized
in Table 1. Table 1 shows the task frequencies for the nal product
as well as the task times for operators and for machines.
4.1. The Operations-Time Chart for batch-and-queue production of
USB ash drives
Based on the data in Table 1 (information typed by the user into
the computerized OT-Chart), and taking into account the principle
of specialization, in Fig. 2 we group the 32 tasks in 10 processes or
workstations (each workstation can have one or more workers performing the same tasks), and run the OT-Chart to calculate the consequences of this approach. Data from Table 1 are used to calculate
the cycle times for grouped operations.
The column entitled Next process in Fig. 2 contains information about the precedence relations between operations, showing
where each station transfers the product. A workstation can receive
components from more than one workstation (as in process E). This
information is critical for the OT-Chart, in order to construct the
routing of the product that will be used to determine when and,
where parts have to be produced to complete a production order
on time.
Having assigned the tasks to each workstation, the software
calculates and displays, in columns Time per process, both the
work (operator) time and the machine time per unit of product.
Other parameters such as the number of machines at each
workstation, the size of the transfer lots, and the overtime per
day and per station (based on an estimated daily time of 14 net
hours) have been typed in their corresponding cells.
Since the operators in stations 1 (Ai) and 2 (As) are idle most of
the time, the designer decides to combine them into a new station

1, with just one worker in charge of both tasks and operating with
a cycle time that supersedes that of the machine. After grouping,
the software calculates and displays in columns cycle times the
equivalent cycle time for the worker and the machine for each
workstation.
Fig. 3 shows part of the OT-Chart output for the batch-andqueue production system. The timescale is shown at the top of
each gure (although the user can select the units of time for a better visualization of the chart, the horizontal axis comes always in
seconds, since it is the international unit of time).
The chart shows that when working with large batches (500
units), a lot of material is accumulated as WIP in some stations,
such as stations 5 and 6. We also point out here that station 6 accumulates pieces even before beginning to operate as its operation
depends on material received from preceding operations. The average and maximum stock levels for each station are shown in the
left-hand side column of Fig. 3.
The OT-Chart depicts the execution of production processes and
lists the results. Therefore, the chart allows tracking of the process
ow for every station and for every batch. By modifying the input
parameters, analysts can compare the results of alternative production conditions. Such results can be used to evaluate the efciency of the production system and be helpful in the process of
custom designing the factory.
The bottom part of the OT-Chart (in Fig. 3), summarizes the values of productivity, WIP in plant and lead time achieved in the
manufacturing of 2500 USB ash drives. These data might be used
to compare alternative congurations in terms of speed and even
cost (such as the cost of labor and the cost of holding inventories).
1. The total lead time is 252,500 s (70.1 h).
2. The rst delivery of nished products (500 units) is at 55,000 s
(15.3 h).
3. The maximum amount of materials and components in WIP
inventory is 2845 units.
4. WIP reaches its peak at 90,000 s (25 h).
5. The average WIP inventory for the whole process is 1588 units.
6. The total waiting time (the white spaces in between the arrows
in the chart) is 252500 s (70.1 h).
7. The average productivity is 50 units per hour.

4.2. The Operations-Time Chart as an assistant in the design of a lean

production system
In literature, there are many tools and methodologies to redesign a company into a lean plant (Hines & Taylor, 2000; Womack
& Jones, 1996).
In the present chapter, we use the Operations-Time Chart in
order to transform, according to the philosophy of lean manufacturing, the processes described in previous chapters. Since lean
manufacturing aims to make products in less time, with a ow
strategy and without waste, the levels of WIP and the time that
materials spend in storage are crucial data to assess the goodness
of a plant design. The OT-Chart displays and calculates that information. Since it shows where inventories build up and which operations have to wait for materials, users have clues to redesign the
process and run the OT-Chart again to see the changes in the values
of inventories and times. Thus, the design may be changed in terms
of number of workstations and transfer batch size, until it meets
the performance objectives.
The transformation resulted in a plant based on ve workstations, which correspond respectively to the following processes:
Ai-As, B1-B2, D-C-E, F-G and H (Fig. 4). The values for additional
time have been lowered in the lean implementation because it is

L. Cuatrecasas-Arbos et al. / Computers & Industrial Engineering 61 (2011) 663675

expected that under lean philosophy quality is improved (less

defective parts) and workstations are closer to each other (less
transportation) and equipped with the necessary tools for more
precise work methods (less non-value added work). Finally, the
reduction in batch size requires a more frequent machine/process
set up and therefore it is necessary to achieve a much shorter setup
time by means of SMED (or OTED) techniques and good maintenance (less non-scheduled stoppages). The physical implementation of the lean process will try to guarantee that these
assumptions hold true.
The execution of the OT-Chart (Fig. 5) starting from data in
Fig. 4 provides information on WIP and lead time for each station
and waiting time between stations. As mentioned before, these
values can be inserted in the future state value stream map, if analysts are using one.
Since this release of the Chart has been adjusted to lean production scenarios, it can hold pull production strategies with small and
adjustable transfer batch sizes, limited and adaptable levels of materials in stock in supermarkets, and according to a exible takt time.
Pull supermarkets allow processes to be linked with material
supply and other processes to prevent overproduction. According
to the results of the OT-Chart, users can type the desired maximum
inventory levels for each supermarket on the same sheet (Fig. 5).
Then the program calculates the corresponding batch size and
yields the new values for lead time and WIP on the chart. This process can be iterative until an acceptable result is met.
Although a theoretical takt time is calculated on a previous
worksheet (taking into account the available time and the scheduled daily production), now users can type a different value on
the OT-Chart, taking into account the cycle time for the system that
is being tested. If the takt time is longer than the cycle time of the
last workstation, the Chart slows down the pace of production and
deliveries in order to meet the takt time. Otherwise, the production
and deliveries will have to speed up.
For the present case study, the initial value of the takt time was
50.4 s. After the execution of the chart, the cycle time is 70 s and
therefore, it is not possible to meet the takt time. The program
warns the user of this fact. In conclusion, the production system
lacks capacity and therefore the system should be redesigned.
If the takt time is set to 70 s, then the process cycle time
matches takt time and the process is capable of guaranteeing the
production ow. If the takt time were set to a higher value, then
the program would calculate a new chart whose cycle time would
be adjusted to the takt time.
In Fig. 5, the delivery rate for nal products has been adjusted to
a takt time of 70 s at workstation 5. That station receives materials
from workstation 4 in pull mode, with the help of a supermarket
(with capacity for 5 units) between both stations. Whenever the
supermarket was full, the production at station 4 would stop (this
would happen with a supermarket capacity of 4 units or less).
Transfer batches are automatically set to 2 units. Likewise, station
4 is supplied from another supermarket that receives materials
from station 3, and so on. Only the pace of the terminal station
needs to be programmed in order to adjust the production of all
stations to the capacity of the supermarkets.
Table 2 displays a comparison between the key parameters of
both implementations (the initial batch and queue layout and
the nal lean design). The nal design shows manufacturing
lead-time reduction, work-in-process inventory reduction, wait
time reduction and personnel productivity improvement. According to Table 2, we can state that the lean design is more efcient
than the initial one because the quantity of production per working
hour has increased, the production lead time has decreased, the
need of overtime has decreased and waste (especially wait time
and WIP inventories) has decreased. This nding ts in with other
studies on lean manufacturing (Shah and Ward, 2003).

673

5. The Operations-Time Chart and other pieces of software

When we developed the Operations-Time Chart, our aim was
not to create a commercial piece of software but a tool that could
help us in research, lecturing and in consultancy tasks, mainly in
order to see the advantages of lean manufacturing vs. batch and
queue manufacturing practices. Therefore, we cannot claim that
the OT-Chart is better than other software packages, but it is intended for different applications.
5.1. The Operations-Time Chart vs. project management software
The OT-Chart resembles a traditional Gantt chart. A Gantt chart
is a popular type of bar chart that illustrates a project schedule. The
tasks of the process appear on the y-axis and the time line on the xaxis. The bars show when a task should start and when it will be
nished, the precedence relations between activities, and the start
and nish dates of the project and its activities. The main difference between the OT-Chart and project scheduling software like
MSProject is that the OT-Chart is intended for manufacturing not
for project scheduling. It supports repetitive tasks, different batch
transfer sizes and concepts like takt time or supermarkets. As explained in Section 3, different types of time values are allowed
per each activity (i.e. man, machine, transportation, non-value
add time, . . .). Besides the evolution of manufacturing over time,
the OT-Chart also displays the evolution of work-in-process inventories and gives information on:
 Production time (the time each station is running is represented by horizontal bold black lines).
 Waiting times for stations (the time transfer lots are waiting can be calculated by the user).
 Inventory generated at every station at every moment
(horizontal dashed gray lines) based on values adjusted
to regular intervals, the maximum and minimum values
of inventory.
 Lead time for each station (determined by the start and
stop times on the timescale).
 Productivity.
Project scheduling software does not support nor gives any
information on the previous concepts. A Project scheduling Gantt
Chart could easily represent an assembly process as long as batch
size was equal for every operation, but not if the size changes from
operation to operation, which is common in industrial processes.
Besides, process scheduling software only could show the ow of
the rst transfer batch, but not the following ones.
5.2. The Operations-Time Chart vs. production scheduling software
Process scheduling software balances capacity and demand.
This type of programs focuses on delivery time and resource availability, but do not focus on waste (such as WIP). The sequence
screen resembles a Gantt chart. They display the sequence of operations assigned to each resource but they consider the manufacturing order as a whole, not taking into account the individual pieces
or transfer batches (the order is considered as a single batch). In
consequence, they consider the lead-time for each resource to be
a constant, not considering the interaction between workstations.
Since results are not realistic, users of production scheduling software usually have to overestimate lead times. On the other hand,
the OT-Chart considers all pieces one by one and the constraints
of the system, and computes the real lead-time (which can be used
by operators to feed production scheduling software). The OTChart is intended for only one order, not for several orders competing for the same resources.

674

L. Cuatrecasas-Arbos et al. / Computers & Industrial Engineering 61 (2011) 663675

5.3. The Operations-Time Chart vs. discrete event simulation software

Discrete event simulation packages simulate queuing processes.
They can be used to model manufacturing systems and they
display statistics on cycle time (the time an entity is in the system)
and work-in-process (the number of entities in the model).
Although most of those programs display on-screen statistics
and a nal report, they do not either generate permanent graphical
output data in order to keep a record of instant values (over time),
or give a written representation (like the OT-Chart) of the interaction that may occur between operations and WIP. Lee, Cho, Im, and
Kim (2002) consider that an important limitation as such feature
would help managers evaluate the dynamic behavior of a productive system and decide about modifying operations in search of
better performance.
6. Conclusions
We have described the computerized Operations-Time Chart, a
graphical evaluation and review technique tool for the design and
time-phased analysis of parallel production and assembly systems,
based on multiple processes that operate in complex conditions.
The OT-Chart builds a model that helps the users visualize some
aspects of the system; facilitates otherwise tedious calculations
for process parameters; and allows testing various system designs
like a simulation tool and testing out the impact of planning
decisions.
The major advantage of the OT-Chart is that it has been designed for manufacturing and assembly operations and therefore,
besides the basic work elements for every task, it is capable of handling additional time for transportation, machine setup, stoppages,
or lack of quality that prolong the cycle time. The OT-Chart includes, calculates and displays typical elements from the lean philosophy such as takt time, non-value added activities, waste, OTED,
supermarkets and pull ow.
Other software packages that display traditional Gantt charts are
orientated towards projects, not towards manufacturing, so they do
not include some features present in our OT-Chart (such as transfer
batch size or WIP inventories build up monitoring). Practitioners
used to Gantt charts can easily understand the OT-Chart. Besides,
spreadsheet software is usually available in companies.
Although the OT-Chart is not a next event simulator, it is capable of modeling a production system and depicting the evolution of
work over time. In this sense, it can be used as a simulation tool.
The development over time of production processes is displayed
on the screen (and it can be printed) instead of on-screen movement of entities, and therefore the chart allows tracking of the process ow, for every station and for every batch.
Data manipulation is simple and instant visual (and numerical
as well) feedback allows the user to quickly test the effects of design parameters on the performance of the system. The OT-Chart
does not pretend to substitute next event simulation. These two
tools are compatible and offer different approaches. After an analysis with the OT-Chart, it is possible to test the best design with a
simulation model.
As a drawback, the OT-Chart does not support stochastic input
parameters (demand, cycle times) but, in lean environments, the
deterministic approach is common.
The OT-Chart is currently limited to parallel/merging processes
that manufacture a single product but this structure is usual in many
manufacturing systems. Other extensions are being developed.
In order to test the features of the OT-Chart, it has been used to
improve the production processes in a USB ash drive manufactur-

ing plant, and it has allowed the designer to adapt the plant gradually to lean manufacturing principles. The OT-Chart has been
used, as a simulator, in order to nd out the effects of some
changes (like the batch size or the capacity of supermarkets) required in implementing lean manufacturing.
References
Abdulmalek, F. A., & Rajgopal, J. (2007). Analyzing the benets of lean
manufacturing and value stream mapping via simulation: A process sector
case study. International Journal of Production Economics, 107, 223236.
Albright, S. C., Wayne, L. W., & Zappe, C. J. (1999). Data analysis and decision making with
Micrososft excel. Pacic Grove: Duxbury Press-Brooks/Cole publishing company.
Becker, C., & Scholl, A. (2006). A survey on problems and methods in generalized
assembly line balancing. European Journal of Operational Research, 168, 694715.
Ben Naylor, J., Naim, M. M., & Berry, D. (1999). Leagility: Integrating the lean and
agile manufacturing paradigms in the total supply chain. International Journal Of
Production Economics, 62(12), 107118.
Denzig, B., & Akbay, K. S. (2000). Computer simulation of a PCB production line:
metamodelling approach. International Journal of Production Economics, 63, 195200.
Fullerton, R. R., Cheryl, S., & McWatters (2001). The production performance benets
from JIT implementation. Journal of Operations Management, 19(3), 8196.
Ghosh, S., & Gagnon, R. J. (1989). A comprehensive literature review and analysis of
the design, balancing and scheduling of assembly systems. International Journal
of Production Research, 27, 637670.
Harmon, R. L., & Peterson, L. D. (1990). Reinventing the factory: Productivity
breakthroughs in manufacturing today. New York: Free Press.
Hines, P., & Taylor, D. (2000). Going lean: A guide to implementation. Cardiff Business
School, Cardiff: Lean Enterprise Research Centre.
Huettner, C. M., & Steudel, H. J. (1992). Analysis of a manufacturing system via
spreadsheet analysis, rapid modelling, and manufacturing simulation.
International Journal of Production Research, 30, 16991714.
Lee, Y. H., Cho, M. K., Im, S. J., & Kim, Y. B. (2002). Suply chain simulation with
discrete-continuous combined modeling. Computers and Industrial Engineering,
43, 375392.
Li, S. G., & Rong, Y. L. (2009). The reliable design of one-piece ow production
system using fuzzy ant colony optimization. Computers and Operations Research,
36, 16561663.
Koo, P. H., Moodie, C. L., & Talavage, J. J. (1994). Performance evaluation of
manufacturing systems: A spreadsheet model. Computers and Industrial
Engineering, 26(4), 673688.
Linn, R., & Zhang, W. (1999). Hybrid ow shop scheduling: A survey. Computers and
Industrial Engineering, 37(12), 5761.
Meral, S., & Erkip, N. (1991). Simulation analysis of a JIT production line.
International Journal of production economics, 24, 147156.
Miltenburg, J. (2001). U-shaped production lines: A review of theory and practice.
International Journal of Production Economics, 70, 201214.
Oliver, N., Delbridge, R., & Lowe, J. (1996). Lean production practices: International
comparisons in the auto components industry [Special issue]. British Journal of
Management, 7, 2944.
Rekiek, B., Dolgui, A., Delchambre, A., & Bratcu, A. (2002). State of art of optimization
methods for assembly line design. Annual Reviews in Control, 26, 163174.
Sakakibara, S., Flynn, B. B., Schroeder, R. C., & Morris, W. T. (1997). The impact of
just-in-time manufacturing and its infrastructure on manufacturing
performance. Management Science, 43(9), 12461251.
Scholl, A., & Becker, C. (2006). State-of-the-art exact and heuristic solution
procedures for simple assembly line balancing. European Journal of
Operational Research, 168, 666693.
Shah, R., & Ward, P. T. (2003). Lean manufacturing context, practice bundles, and
performance. Journal of Operations Management, 21(2), 129149.
Watanabe, N., & Hiraki, S. (1997). A modeling approach to a JIT-based ordering
system. Annals of Operations Research, 69, 379403.
White, R. E., & Prybutok, V. (2000). The relationship between JIT practices and type
of production system. Omega, 29(2), 113124.
Womack, J. P., & Jones, D. T. (1996). Lean thinking: Banish waste and create wealth in
your organization. New York: Simon and Schuster.
Womack, J. P., Jones, D. T., & Ross, D. (1990). The machine that changed the world.
New York: MacMillan/Rawson Associates.
Llus Cuatrecasas Arbs is a professor of operations
management. He is an industrial engineer and he
received his PhD from the Technical University of
Catalonia. He has been the head of the Department of
Management. He is the President of the Instituto Lean
Management Espaa (Spanish branch of the Lean
Enterprise Institute). He has authored many books on
manufacturing and his research has appeared in
national and international journals.

L. Cuatrecasas-Arbos et al. / Computers & Industrial Engineering 61 (2011) 663675

Jordi Fortuny Santos is an associate professor of operations management. He is an industrial engineer and he
received his PhD from the University of Lleida (Spain).
He is interested in manufacturing, especially in lean
environments. His research has appeared in national
and international journals.

675

Carla Vintr Snchez is an assistant professor of operations management. She is an industrial engineer and
she is currently completing her PhD at the School of
engineering at Manresa. Her research interests include
manufacturing and management systems. She has
published in national and international journals.