Int J Adv Manuf Technol (2016) 84:1573–1586

DOI 10.1007/s00170-015-7805-8

ORIGINAL ARTICLE

A value stream mapping and simulation hybrid approach:

application to glass industry
Anas M. Atieh 1 & Hazem Kaylani 1 & Ahmad Almuhtady 2 & Omar Al-Tamimi 1

Received: 14 July 2015 / Accepted: 2 September 2015 / Published online: 19 September 2015
# Springer-Verlag London 2015

Abstract This paper introduces a condensed version of simulation model. Not only did the implementation of
lean manufacturing implementation toward improving the prescribed scheme in the glass fabrication company
manufacturing processes in a small to midsize glass fab- revealed a hidden bottleneck, but also the suggested
rication company in Jordan. While the company suffers changes reduced the manufacturing lead time by 6 %
from long manufacturing lead times and unbalanced pro- and increased the performance of the primary bottleneck
duction line, it is almost impossible to convince the tradi- by 32 %. Finally, the computerized relative allocation of
tional administration of this small to midsize manufactur- facilities technique (CRAFT) was used to identify the op-
ing plant to adopt the completely revolutionizing, costly, timal layout of the expanded system.
and risky (at such level) lean manufacturing approach.
Therefore, a much lower cost and verified scheme of en- Keywords Value stream map . Simulation . Process
hancement is provided. The scheme starts with value flowchart . Bottleneck detection . Glass fabrication case study
stream mapping (VSM) to visualize the process flow as
well as identify production status and any potential alerts.
Then, the scheme employs a discrete event simulation 1 Introduction
model to accurately estimate the impact of these potential
alerts. The results of the simulation is then used to iden- Surrounded by energy-rich countries and located in a country
tify all the possible primary and secondary bottlenecks that adopts near open market policy, Jordanian industries suf-
using multiple approaches such as utilization method, fers severe competition and thus are asked to rise over it through
waiting time method, and scenario-based method. An im- improved and sustainable manufacturing plants. In order to be
proving algorithm is then utilized to systematically sug- sustainable and competitive, companies are required to identify
gest changes aiming to relieve the system and produce a every potential area of improvement. Minimizing waste and
better manufacturing lead time and/or increase its increasing efficiency is the only way for a company to move
throughput taking into consideration feasibility. These forward. However, identifying sources of waste in companies
changes are evaluated each time using a modified built on traditional systems is not a trivial task. Lean thinking
concepts have gained a lot of attention in terms of identifying
and removing waste. Researchers’ interest in lean manufactur-
* Anas M. Atieh
ing has even grown in the last two decades [1–3].
[email protected] Lean manufacturing is a revolutionizing philosophy that
defines every non-value adding activity as waste and ultimate-
1
ly aims to eliminate such activities from all processes [4].
School of Applied Technical Sciences, Industrial Engineering
Department, German Jordanian University, Amman Madaba Street,
Many tools have been used to achieve lean manufacturing,
Amman, Jordan such as cellular manufacturing [5], just-in-time [6], total pro-
2
School of Applied Technical Sciences, Mechanical and Maintenance
ductive maintenance (TPM) [7], and 5S [8]. Despite their ef-
Engineering, German Jordanian University, Amman Madaba Street, fectiveness, the implementation of such tools usually requires
Amman, Jordan making drastic changes to current setups, which proves costly.
1574 Int J Adv Manuf Technol (2016) 84:1573–1586

Therefore, convincing management to commit to lean thinking processes [20]. By visualizing the flow of materials and infor-
concepts proves to be a major roadblock. To provide enough mation, it provides companies with a “blueprint” for strategic
evidence for management to commit to changes, we take into planning to facilitate companies’ transformation into lean en-
account using tools that are capable of monitoring and quanti- terprises [21]. It has proved very effective in identifying and
fying potential gains once changes are implemented. eliminating waste [22]. Moreover, it can reduce the inventory
In this paper, we approach a glass manufacturing company pipeline and lead to reduction in throughput time [23]. In
in Jordan that suffers from long manufacturing lead times and addition, VSM can be applied practically in any business ac-
constant delays in customer orders delivery. To analyze the tivity and expanded upstream or downstream [24].
operations, we infuse value stream mapping (VSM) with sim-
ulation. Value stream mapping gives a process-oriented view 1.2 Simulation supporting VSM overview
of the company that can help evaluate the current situation and
identify areas of improvement. In addition, we use a simula- One of VSM’s fundamental limitations is its manual nature,
tion model and bottleneck detection methods to further en- which produces a static model that makes the observation and
hance our view of the company’s processes. Building on the evaluation processes of the map difficult [9]. Therefore, it
findings of the VSM and simulation model, we suggest chang- cannot be used—being a static tool—to study a dynamic prob-
es to improve the productivity of the plant. Using the simula- lem [25]. It also suffers from only being capable on analyzing
tion model, we can evaluate the potential gains of the sug- one product or product family image [26]. In addition, lean
gested changes. Afterwards, we build a future state as an extra manufacturing often requires making dramatic changes to the
evaluation of the changes suggested to the system. organization [27]. Therefore, evaluation of the changes sug-
gested by VSM is essential before implementation. This leads
1.1 Value stream mapping overview to the need of a complementary tool with VSM to quantify
gains during the early planning and assessment phase [28]. A
Value stream mapping became popular since the publication complementary tool is also required alongside VSM to handle
of the book Learning to See—Value Stream Mapping to uncertainty and model the dynamics between system compo-
Create Value and Eliminate Muda [9]. It was initially devel- nents for different future state maps [29].
oped with an underlying rationale for the collection and use of Researchers have been trying to overcome its limitations
the suite of tools as being “to help researchers or practitioners and enrich it using additional tools and techniques. For exam-
to identify waste in individual value streams and, hence, find ple, Gurumurthy and Kodali [30] used benchmarking to aid the
an appropriate route to its removal” [10]. It was used primarily VSM. However, many researchers used simulation to enhance
in several industries, including automotive [11], pharmaceuti- VSM. Solding and Gullander [26] pointed out that using a
cal [12], healthcare [13], and service industries [14]. simulation-based VSM can overcome the limitation of analyz-
The traditional supply chain refers only to the specific parts of ing one product and can provide a real-time view of the system
the firm that added value to the product or service under consid- instead of a snapshot. Lian and Van Landheghem [31] trans-
eration while the value stream includes the complete activities of formed data obtained from simulation results into current and
all companies involved [15]. VSM provides a very effective future state maps in order to make VSM dynamic. Braglia et al.
method for visualization, analysis, and redesign of the production [32] built simulation models for current and future states as an
and supply chain processes [16, 17]. The result is a single-page evaluation for VSM. Swallmeh et al. [33] integrated simulation
map that graphically documents the processes in the current state. with VSM to improve an emergency department. Parthanadee
The map contains data such as cycle time, changeover time, and Buddhakulsomsiri [34] used VSM and simulation to im-
work-in-process (WIP) levels, and equipment reliability data. prove batch production in the coffee industry.
After analyzing the current state map and identifying its sources Simulation is capable of enhancing VSM by providing in-
of waste, the process of constructing the future state map starts. formation about the nature of the production process, thus,
The goal is to build a chain of production where the individual overcoming its static nature [35]. Using simulation would
processes are linked to their customer(s) either by continuous provide a way to explore the various opportunities of process
flow or by pull, and each process gets as close as possible to improvement and the impact of the proposed changes before
producing only what its customer(s) need when they need it [9]. implementation [36]. Using recent simulation software can
This is achieved by answering a set of questions to further un- improve the quality of solutions in a relatively short period
derstand the processes and construct a system that minimizes [37–40]. Simulation is not just capable of assessing the rela-
WIP inventory and consequently, manufacturing lead time. tive difference between the process in its current form and the
Value stream mapping is an excellent tool for any enter- suggested form but it can also confirm and strengthen our
prise to become lean [18]. It provides a method for analysis view of all areas that need improvement. This occurs due to
and design of production processes which will address these being able to obtain information that VSM simply cannot
rapid changes [19]. It can serve as a guide to standardize provide, such as resources utilization and queuing times.
Int J Adv Manuf Technol (2016) 84:1573–1586 1575

Simulation is also capable of assessing more than one product

family at the same time, a feature that VSM cannot achieve
[26]. Another advantage of using simulation is the ability of
evaluating different scenarios and measuring the improve-
ments in each one under different circumstances. This gives
it the edge over using VSM only because VSM is a static tool
in nature.
In this paper, we will use a simulation model to identify the
bottleneck(s) in the production line. Then, using both findings
from the simulation model and VSM, we will suggest changes
that can balance processes and decrease the manufacturing
lead time. At the end, we use the simulation model to evaluate
the changes and build a future state map to quantify the gains.

1.3 Problem solving approach

The problem solving approach is illustrated in Fig. 1. The

VSM will help visualize process flow for a product or family
of products as well as identify production status and any po-
tential alerts (Andons) that might be causing problems to the
production system. These Andons might include potential bot-
tleneck resources, cycle time problems, capacity limitations,
inability to meet schedule, etc.
A simulation model that represents the as-is situation is
created using the discrete event simulation software
(ARENA-Rockwell Automation© Wexford, PA, USA); the
simulation model will provide more accurate estimation of
the impact of these Andons on the production system. In ad-
dition to other performance measures, the simulation model
will mainly evaluate bottlenecks in the production. Three bot-
tleneck detection methods are used in this paper: utilization
method, waiting time method, and scenario-based method.
Based on the initial results obtained from the model, a number
of scenarios will be tested. The algorithm starts with the re-
source that is maximally utilized and considers improving
them by reducing process time and/or adding more resources.
Then, the next highest utilization resource will be considered
and new scenarios will be created accordingly. All resources
identified as Andons in the VSM will be considered in a sim-
ilar manner. Each scenario will be run on the simulation model Fig. 1 Problem solving approach diagram
and the overall system performance will be evaluated.
Performance measures including resources utilization, prod-
uct cycle time, and ability to meet deadline, are used to eval- processes different glass grades and mirrors into several
uate system performance. Based on the results obtained, a products based on a highly customizable customer order.
proposed solution will be provided to decision makers. The It serves the industrial, commercial, and residential mar-
last stage would be to examine the impact of the proposed kets in Jordan and the Middle East. It carries its oper-
solutions on the facility plan, to generate a new layout with ations across two plants that share most of the processes
improved flow, and finally, to create VSM future state. with a few processes only carried out in one plant and
not the other. The raw materials used are flat
1.4 Company and its processes untampered glass panels. They come from several sup-
pliers in several dimensions where the most common
Mansour is one of the oldest glass fabrication plants in dimensions are 3.66 m ×2.44 m. Currently, the company
Jordan. It was established in 1944. The company is only capable of addressing 56 % of the potential
1576 Int J Adv Manuf Technol (2016) 84:1573–1586

demand. Following is a summary of the processes car- the procurement process and production plant selection. In the
ried out at the company: analysis below, we only consider one plant.

1. Cutting. This process is required for all orders to achieve

the dimensions requested by the customer. It is done using a 2 VSM: current state map
CNC glass cutting table; however, loading and unloading are
almost done manually. The process starts by accumulating The first step of constructing a current state map is to identify a
orders to enter them into queuing software that aims toward product or a product group. However, since the products are
reducing waste material. highly customizable, we end up with a large amount of poten-
2. Edging. This process is also required by all orders. The tial products. Therefore, we analyzed the products in terms of
aim of edging is to machine the edges of glass for functional processes required to produce them and chose a subset of the
purposes, like preparing it for installation, and/or esthetic pur- processes that were most demanded. Any waste identified in
poses if the edges will be visible when installed. The process is the selected processes will affect the majority of the products,
achieved using glass straight line edging machines that work if not all of them, which pinpoints toward the scope of the
on one side at a time. Workers load each cut piece into the study. After analyzing the company’s production, the follow-
cutter four times to finish the job on all edges. Different edging ing processes were chosen:
types results in different times. For example, arising and flat
edging have a relatively short setup time (5–15 min). Pencil 1. Cutting, since it is required for all orders.
edging incorporates a change in the die according to the di- 2. Edging, since it is required for all orders.
ameter of the edge required; a cumbersome task that can take 3. Drilling, since 82 % of orders require at least one drilling
up to 90 min. job.
3. Drilling. This process is done based on customer orders. It 4. Tempering, since 94 % of orders require tempering.
includes drilling holes or removing certain parts to achieve
functional purposes (e.g., glass doors). Next step in constructing the map was data collection. First,
4. Tempering. Heat treatment of glass to enhance its charac- management provided the average of supplier shipments per
teristics such as strength and brittleness. The latter is important month and customer orders per day. They also provided infor-
for safety in case the glass is broken. Customers do not always mation about how information is handled between the suppli-
request it but an overwhelming majority of orders goes er, customer, and the company. For lack of historical data on
through it. the setup and processing times, estimates were collected from
5. Silk screening. This process draws shapes or drawings experienced operators in every workstation starting from cut-
required by the customer on top of glass. It is an esthetic ting and ending with tempering. To validate these estimates,
process and is only done by customer request. the collected approximations were compared with average
6. Manual painting. An alternative to the automated silk times calculated based on the company’s average production
screening; this process uses paint sprays that are operated rate per day at each workstation over the last 4 months.
manually to paint the glass. This process, while time demand- Next, work in progress (WIP) before each workstation was
ing, is requested in a low number of orders. obtained as a snapshot from a typical day at the company. For
7. Lamination. This process produces another type of consistency across the map, WIP and orders per day were
safety glass. Two or more panels of glass are pressed expressed in terms of surface area units (m2). Reliability data
together with an interlayer that would hold the glass in was extracted from a maintenance log kept by management.
case it breaks. Figure 3 shows the current state map that was constructed.
8. Double glazing. This process produces glass that is later The top part of the map shows a high-level view of the inter-
installed in aluminum frames for windows. Two panels of action between the customers, the company, and the supplier.
glass are attached together and optionally with heat- Each box under the operations management represents a pro-
insulating material between them. cess. The number inside each box is the number of workers.
9. Assembly. This process is very customizable and manual. The information table beneath each process box provides key
The output is a 3D shape of glass that requires attaching mul- information such as the cycle time, changeover time, reliabil-
tiple glass parts together. ity expressed as uptime percentage, and availability. Triangles
10. Sand blasting. This process is done for esthetic purposes. before each process represents WIP inventory. Note that be-
Glass panels are sprayed with sand to achieve a distorted look fore the edging process, there are two inventory triangles. This
that eliminates its transparency. is due to having three types of edging at the company. Flat and
pencil edging are with monitored inventory levels contrary to
A flowchart of all processes in the company is shown in arising edging, as no inventory exists for it mainly because it is
Fig. 2. It shows all operations discussed earlier, in addition to a much faster process with lower demand. The timelines at the
Int J Adv Manuf Technol (2016) 84:1573–1586 1577

Fig. 2 Operations flowchart

1578 Int J Adv Manuf Technol (2016) 84:1573–1586

bottom expresses production-waiting time at the top (non- in Table 1. Since some processes deal with surface area as a
value adding) and processing time at the bottom. The upper metric in contrast to others dealing with length (e.g., edging)
timeline shows times when inventory is insufficient and the or number of drills in drilling, we used the number of panels as
lower one shows the average times. Waiting time is calculated a metric throughout the system. This is done in a straightfor-
according to Little’s law, which asserts that the average num- ward manner in processes that work on the surface area.
ber of entities in a queue, L, is equal to the rate at which However, and for processes like edging and drilling, there
customers arrive and enter the system, λ, multiplied by the was a need to estimate a relation between area and edges
average time an entity spends in the system, ω. Therefore, length or number of drills. This was achieved using averages
waiting time was calculated by dividing the WIP inventory calculated from historical data. Example of these can be found
by daily production rate. For example, 348 m2 before pencil in the last column in Table 1.
edging divided by 170 m2 (daily production) equates to 2 days The simulation model starts with the order dispatch (cur-
of waiting time. The bottom part of the timeline expresses rently at 30 glass panels on a daily basis). The glass is loaded
processing time of each process. In this case, we have onto the cutting machine by workers, the cutting then starts
32.9 days of waiting time when inventory is insufficient and and once finished, the workers unload the glass. All glass is
10.9 days on average while the processing time is 612 s in then routed to the edging station. The edging station includes
both cases per 1 m2. The difference is noticeably big, which the three types of edging described earlier; flat, pencil, and
suggests many activities carried out at the company are con- arising. The glass is split between the three types by a separate
sidered as waste according to the lean concepts. It should be module that sends 45 % of glass to flat edging, 50 % to pencil
noted that the data on the map is only for orders that was sent edging, and the last 5 % are sent to arising. The percentages
to production. The company has more orders that entered the were decided based on the average output from historical data.
system and are yet to commence production, which would As soon as a glass panel departs a station, it enters a separate
increase the effect of increasing the production rate of the module that decides if a panel is sent to either drilling or
plant. washing. The percentage was also decided based on the aver-
The map highlights the inventory control at the start to be a age output. Panels sent for drilling are set up by workers and
big factor in increasing the manufacturing lead time. On the then enter the drilling process. Afterwards, all panels enter a
manufacturing level, the tempering process is causing the lon- separate module that sends 1 % of panels for a rework in the
gest waiting time. In the next section, we build a simulation workshop to amend scratches. All of the panels are then sent
model to further investigate the causes of waiting time and to the washing station. After washing, panels enter a series of
either confirm or deny the findings of the VSM. separate modules to decide which processes are done to which
panels, as shown in Fig. 3.
All separate modules are governed by percentages extract-
3 Simulation model ed from the average output of each process. After each panel is
sent to its respective process, the process is performed and
In order to evaluate different alternatives and to further inves- then panels are transferred to the delivery station and leave
tigate the system, we built a simulation model. A simulation the system. Note that glass panels that exit the tempering pro-
model would provide a dynamic picture of all processes in- cess could go to the lamination station before moving to de-
volved and allow evaluating the performance of each process livery and exiting the system.
based on historical data. The model supports the VSM as it After checking that the logic of material flow in the
would assess its output and confirm the potential sources of simulation model matches the actual flow in the company,
waste. It would be used later on to identify bottlenecks in the the system must be validated such that the simulation
system and test changes and proposed solutions. model mimics real-life performance. To achieve that, a
The model was built using discrete event simulation, where comparison was made for the output of the model per
events occur at discrete points of time on the simulation clock. day to the average output of the company over the last
The flowchart built earlier (Fig. 1) was used as an input with 4 months. Table 2 shows the results in terms of panel out
the historical data collected at the company. For simplification per day for each process. Although the model does not
reasons, we assumed the raw material entering the system is of mimic the real system perfectly, the difference is relative-
one size only (3.66 m×2.44 m). The assumption was backed ly small, specially in relation to the popular processes.
by the fact that over 75 % of the used materials are of the However, the error should be taken into consideration
selected dimensions, in addition to the other variants being when applying changes on the real system after testing
very similar in dimensions. The input to the system was set on the simulation model. For further evidence, the tem-
to a constant flow of 30 panels per day. pering furnace, which was observed to be utilized almost
Using the data collected about the setup and processing 100 % of available time, was shown to be utilized 98 % of
time in each process, a range was set to each one as shown the time in the simulation model.
Int J Adv Manuf Technol (2016) 84:1573–1586 1579

Fig. 3 Current state map

4 Bottleneck detection whole workstation is active because both need to be free for
throughput to be affected.
After the simulation model was built and validated, the output The results shown in Table 3 suggest that the tempering
data can be used to identify the bottleneck in the system. First, workstation is the primary bottleneck and the pencil edging
we define a bottleneck to be the machine whose throughput workstation is the secondary bottleneck. Next, we use the
affects the overall system throughput. waiting time method and compare results.
According to Lemessi et al. [41], bottleneck detection
methods can be studied by numerous methods such as: (1) 4.2 Waiting time method
static-calculation method, (2) coefficient-based methods, and
(3) scenario-based methods. Using the simulation output, we This method uses the waiting time of parts in the queue before
will use two methods based on coefficients, the utilization each process to determine the bottleneck. It defines the
method, and the waiting time method. To confirm the results,
we use a scenario-based method as suggested by Roser et al.
[42]. For further validation, we used the static-calculation Table 1 Setup and processing times per panel
method. The method uses the data input into the simulation Process Setup time (min) Processing time (min) Assumptions
instead of the output data.
Cutting 2–3 0.75–1
Flat edging 0–10 15–25 17 m/panel
4.1 Utilization method Pencil edging 15–60 15–25 17 m/panel
Arising – 7–15 17 m/panel
The utilization method measures the percentage of time a ma- Drilling 5–10 6–10 5 drills/panel
chine is active and defines the machine with the highest active Silk screening 15–20 5–10
percentage as the bottleneck [43]. Tempering 5–10 20–30
The discrete event simulation is capable of providing the Double glazing 10–20 5–10
utilization of all machines as an output. It is worth noting that Lamination 5–8 5–10
utilization is taken for workstation and not individual re- Assembly – 15–90
sources (i.e., machines and workers). This assumes that when Sandblasting 5–15 5–10
a worker is active in a workstation and the machine is free, the
1580 Int J Adv Manuf Technol (2016) 84:1573–1586

Table 2 Simulation model

validation data Process Actual output (panels) Simulation output (panels) Percentage difference (%)

Cutting 31.4 30 4
Flat edging 13 13 0
Pencil edging 17 15 12
Arising 0 2 –
Drilling 8 10 25
Silk screening 3 4.5 50
Tempering 18 19 5
Double glazing 6 5 17
Lamination 5 6 20
Assembly ** 0.4 –
Sandblasting ** 0.1 –

**No recorded data

bottleneck to be the workstation before which parts wait the independently and run the simulation model to analyze the
longest time [44]. We use the output of the simulation model effect each improvement has on the overall system output
to assess the waiting time before all processes in the system [45]. In this case, we improve the setup and cycle time of both
(Table 4). processes by 30 %. Table 5 shows the results of each improve-
According to this method, the pencil edging workstation is ment on the overall system.
the primary bottleneck and the tempering workstation is the Since the improvement to both was with the same percent-
secondary bottleneck. The contrast in results between the two age, and tempering improved the system more, we can con-
methods can be attributed to the difference in processing time; clude that the tempering is the primary bottleneck and pencil
tempering is causing a bottleneck due to long processing times edging is the secondary bottleneck.
while the pencil edging machine is causing a bottleneck due to
long setup times. However, we use the scenario-based method 4.4 Static-calculation method
to confirm which one is the primary bottleneck.
This method compares the processing time of each worksta-
4.3 Scenario-based method tion to the takt time [43]. The takt time is the average unit
production time needed to meet customer demand. Any work-
Previous methods returned different results; however, both station whose processing time exceeds takt time is considered
found that tempering and pencil edging are the top two bot- to limit the pace of overall production, hence, is a bottleneck.
tlenecks in the system. To identify which of the two is the In order to make this comparison, each workstation’s process-
primary bottleneck, we use the scenario-based method. In this ing time needs to be representative of both setup and actual
method, we theoretically improve each workstation processing time. Afterwards, the takt time is calculated based

Table 3 Utilization data for all processes Table 4 Average waiting time for all processes

Process Utilization percentage (%) Process Average waiting time (min)

Cutting 23 Cutting 0
Flat edging 43 Flat edging 55
Pencil edging 82 Pencil edging 1030
Arising 4 Arising 2
Drilling 35 Drilling 9
Silk screening 23 Silk screening 7
Tempering 94 Tempering 160
Double glazing 17 Double glazing 24
Lamination 21 Lamination 0
Assembly 1 Assembly 0
Sandblasting 1 Sandblasting 55
Int J Adv Manuf Technol (2016) 84:1573–1586 1581

Table 5 System improvement in response to improving processes 5.2 Drilling

Process Percentage System System System
improvement output before output after percentage Drilling is shown in the VSM to be the second biggest bottle-
improvement improvement improvement neck point. However, using the simulation model and bottle-
(%) neck detection methods, the process was shown to not affect
Pencil 30 22.5 22.9 1.78
the overall throughput. It has an average of 9-min waiting
edging time. The utilization of workers in the drilling workstation is
Tempering 30 22.5 23 2.23 also relatively low (35 %).
This could be attributed to only 35 % of orders requiring a
drilling job of five or more drills according to historic data.
on the average demand of 30 panels per day. The resulting takt Furthermore, the station has five workers that are also respon-
time was 16 min. Table 6 shows the results of the comparison. sible for washing glass. Those two factors result in high WIP
The results show that assembly, pencil edging, and temper- in terms of volume; however, the queue moves relatively fast
ing are the highest bottlenecks. However, due to assembly’s and does not cause delay even if the demand increased two
low customer demand, we cannot conclude that it is a bottle- times.
neck. Therefore, the method confirms that tempering and pen-
cil edging are the primary and secondary bottlenecks. 5.3 Tempering
Furthermore, if we consider customer demand, tempering
ends up having the highest effect on the throughput due to VSM showed tempering to be the biggest necking point in the
its higher demand (94 %) compared to pencil edging (45 %). line with the highest WIP levels. This fact was confirmed by
bottleneck detection method. The workstation was found to be
the primary bottleneck. It has a utilization of 94 % and an
5 Assessment and validation average waiting time of 160 min. The high utilization can be
attributed to 94 % of customers’ orders requiring a tempering
In this section, we revisit the processes visualized in the cur- job, in addition to having the longest processing time. Adding
rent state map. Each process is analyzed in a perspective of the another furnace is considered an expensive solution.
findings extracted from the simulation model and bottleneck Therefore, we suggested increasing the availability of the fur-
detection. nace each day with overtime to increase its output compared
to other processes. Currently, the tempering furnace operates
for 8 h a day. The increase will be done in increments until
5.1 Edging improvement is achieved. The suggestion will be evaluated
using the simulation model and the future state map in the
The current state map showed edging to have the least WIP following sections.
inventory compared to tempering and drilling. It also showed
that pencil edging had more than two times higher WIP than
flat edging, although processing time for both types of edging 6 Potential gains evaluation
is the same. In addition, customers’ orders are almost evenly
split between flat and pencil edging. The simulation model 6.1 Evaluation via simulation
confirmed the difference in waiting time before each station.
The average waiting time before the pencil edging is 17 times As discussed in the section above, it is suggested that the
more than the waiting time before flat edging. The difference following changes to the system occur in order to relieve the
between them could be attributed to two factors. First, the bottlenecks (pencil edging and tempering):
company has two machines for flat edging compared to only
one pencil edging machine. The other factor is the setup time 1. Adding a second pencil edging machine.
for pencil edging machine, which is higher than the flat edging 2. Increasing availability time of the tempering furnace by
machine. After using bottleneck detection methods, pencil increments of 20 %.
edging was identified as a secondary bottleneck behind tem-
pering. Therefore, pencil edging is a station that requires im- The changes were applied to the simulation model built
provement to achieve better production line balance and de- before to evaluate the results of the changes while holding
crease average waiting time. It is suggested to add another all other parameters constant. Tempering availability was first
pencil edging machine to the workstation. This suggestion increased from 8 to 10 h. While this showed increased output
will be evaluated using the simulation model and the future (16 %), this change increased the utilization and waiting time,
state map in the following sections. which would amplify the effect of the bottleneck on the
1582 Int J Adv Manuf Technol (2016) 84:1573–1586

production line and increase failure probability. Therefore, an achieved results gave it the edge over other more sophisticated
increase of an additional 20 %, i.e., 12 h per day availability, options in small companies that are reluctant to adopt lean
was tested. After increasing the availability to 12 h per day, the manufacturing fully with all its cost and risks. In addition,
waiting time and utilization decreased. Table 7 shows the out- the combination of the static VSM and the dynamic
put of each process before and after the modifications, assum- Simulation made it a multidimensional approach that can
ing the same input. The changes caused the output of temper- identify multiple sources of waste, which is a unique feature
ing to increase by 32 %. Moreover, although the pencil edging in lean manufacturing. It could also be used as a monitoring
output remains almost unchanged, the waiting time before it tool, which makes it a valuable tool in all continuously im-
decreased 10 times. This decrease brought it closer to flat proving environments.
edging and made the line more balanced. The decrease of
waiting time and utilization decreases the effect of the process
on the overall system output and makes it more capable of 6.3 Facility layout adjustments
handling higher demand. This is further confirmed in Table 8,
which shows the difference in utilization and waiting time. To benefit the most from the added pencil edging machine sug-
gested above, we conduct a facility layout analysis. The analysis
would evaluate the facility layout if the new machine was added
6.2 Evaluation via future state map next to the old pencil edging machine and look for better solu-
tions. The facility layout problem in manufacturing plants is
Using the modified simulation model, we construct a future concerned with finding the most efficient arrangement of a spe-
state map as to further validate the impact of the suggested cific number of departments in a specific number of locations.
modifications introduced to the system. We start with the cur- As shown in Fig. 5, the facility contains 12 departments
rent state map as a base and then edit values based on the and they are:
changes incurred in the modified simulation model. Figure 4
shows the resulting future state map. The WIP levels were 1. Storage 4. Pencil Edging 7. Washing 10. Double Glazing
modified based on the changes that happened to the average 2. Cutting 5. Arising 8. Silk screening 11. Tempering
waiting queues in the modified simulation model, shown in 3. Flat Edging 6. Drilling 9. Lamination 12. Bending
Table 8. The daily uptime of tempering was increased to 12 h
instead of 8 h. The WIP before pencil edging decreased dra- However, not all facilities can be moved due to large size
matically from 348 to 31.6 m2 and the WIP before tempering and integration into the building. Those departments are lam-
has decreased from 635 to 437 m2. The changes decreased the ination, silk screening, tempering, and bending. To solve the
manufacturing lead time when inventory is insufficient to 30.8 problem, we used the computerized relative allocation of fa-
down from 32.9 days; 6 % enhancement. Moreover, the aver- cilities technique (CRAFT) method developed by Buffa,
age lead time decreased from 10.9 to 8.8; a 20 % enhance- Armour, and Vollmann [46]. It calculates the centroids of all
ment. To incorporate this finding, we modify the facility lay- departments and finds the Euclidean distance between each
out to incorporate the added pencil edging machine. one. It then considers all possible department switches and
This approach is a comparatively low-cost technique to chooses the switch, which results in the lowest distance trav-
identify waste and start the process of achieving lean eled. The method stops and reaches the final solution when no
manufacturing. The ease of implementation and quickly other solutions are available. It is capable of considering the

Table 6 Comparison of
processing times to takt time Process Simplified processing time (m) Percentage of takt time (%)

Cutting 3.375 21
Flat edging 24.5 153
Pencil edging 54 338
Arising 9 56
Drilling 14.3 89
Silk screening 25 156
Tempering 32.5 203
Double glazing 22.5 141
Lamination 14 88
Assembly 55 344
Sandblasting 17.5 109
Int J Adv Manuf Technol (2016) 84:1573–1586 1583

Table 7 Modifications effect on

simulation output Process Output before modifications Output after modifications Percentage difference (%)

Cutting 30 30.00 0
Flat edging 13 14.2 9
Pencil edging 15 14.5 −3
Arising 2 1.21 −40
Drilling 10 9.64 −4
Silk screening 4.5 3.76 −16
Tempering 19 25 32
Double glazing 5 4.43 −11
Lamination 6 8.2 37
Assembly 0.4 0.26 −35
Sandblasting 0.1 0.21 110

cost of material handling and flow between departments, Constructing the current state map helped identify the
which is the primary concern in this case. It can also take into major products or processes in the system. It also helped
account departments that cannot be moved; which is a major in quantifying the performance of the company based on
constraint in the problem. As input, the method requires an the key performance indicator (KPI) of manufacturing
initial layout and flow matrix between departments. lead time. All other information included in the map make
To apply the method, an MS Excel add-in [47] was used. a significant base that can be used for any improvement
The flow between each department was expressed as percent- efforts. Using simulation as a complementary tool to
ages of 100 to represent the flow in the simulation model. VSM adds a dynamic dimension to the tool. It helped in
Although the tool is capable of considering cost, it was exclud- judging changes for a VSM on several products or pro-
ed because the company uses a forklift for transfer between all duction lines instead of the one-dimensional effort VSM
departments, which makes cost irrelevant. Afterwards, the ini- focuses on. Simulation also serves as a check and evalu-
tial layout was expressed in spreadsheet format. After running ation tool in findings extracted from a VSM, in addition
the tool to optimize for the lowest Euclidean distance traveled to evaluating potential improvements in different scenari-
between departments, a layout was constructed. The suggested os, i.e., incrementally modifying the tempering process
layout decreases distance traveled by 5 %. Figure 6 shows the until waiting time is decreased.
changes suggested on a blueprint. The bottleneck detection methods made extra use of
information extracted from the simulation model to

7 Conclusion and future work recommendations Table 8 Modifications effect on waiting time and utilization

Process Average Average Utilization Utilization

This paper proposes a condensed version of lean manufactur- waiting time waiting time before after
ing, which combines the merits of VSM and simulation in a before after modification modification
systematic scheme aiming toward addressing small to midsize modifications modifications (%) (%)
(min) (min)
manufacturing plant problems. The condensed version targets
relieving the system from its bottlenecks in a feasibility- Cutting 0 0 23 19
conscious improving algorithm. Flat edging 55.2 56.4 43 30
The proposed scheme employing VSM, simulation, and Pencil 1030.8 93.6 82 39
bottleneck detection methods was implemented in a real case edging
study involving glass fabrication in Jordan. The plant, which Arising 2.4 2.4 4 2
is considered relatively small, suffered from traditional Drilling 9 10.8 35 28
manufacturing problems such as long lead times and unbal- Silk 7.2 4.8 23 17
anced production lines. Practical experience revealed one po- screening
Tempering 165.6 114 94 85
tential bottleneck (tempering furnace); for which the solution
Double 2.4 4.8 17 14
was too costly and risky. glazing
The implementation of the proposed scheme was able Lamination 0 0 21 20
to identify another bottleneck, and proved to be efficient Assembly 0 0 1 1
in identifying waste, in addition to evaluating and quanti- Sandblasting 0 0 1 1
fying changes to a system before implementation.
1584 Int J Adv Manuf Technol (2016) 84:1573–1586

Fig. 4 Future state map

accurately identify primary and secondary bottlenecks. quantifying the effect of changes suggested by the use
This paper made use of three approaches to find potential of the manufacturing lead time KPI.
areas of improvement in the company. Identifying bottle- Combining the three tools, and the flowchart described
necks accurately ensures that improvement efforts are not earlier, can provide a building stone for any future improve-
spent in the wrong area. This important aspect helps avoid ment efforts. The flowchart can simplify the process. The
the wasteful use of resources. It also highlights the signif- VSM visualize the processes and rates their performance ac-
icance of certain departments on the output of the system. cording to their effect on lead time. Keeping the VSM updated
The future state map proves to be a great tool in would serve as a good monitoring tool for management in the

Fig. 5 Current facility layout

Int J Adv Manuf Technol (2016) 84:1573–1586 1585

Fig. 6 Suggested facility layout

future. The simulation model can serve as an evaluation tool to 8. Osada T (1991) The 5S’s: five keys to a total quality environment.
Quality Resources
any suggested improvement before actual implementation.
9. Rother M, Shook J (1998) Learning to see: value stream mapping to
The current state map highlights the inventory control and add value and eliminate muda. Lean Enterprise Institute
supplier management as an area that requires consideration. 10. Singh B, Sharma S (2009) Value stream mapping as a versatile tool
The manufacturing lead time depends on the suppliers’ lead for lean implementation: an Indian case study of a manufacturing
firm. Meas Bus Excell 13(3):58–68
time due to the lack of a better inventory control system.
11. Schweizer W (2011) Value stream engineering—four paradigms for
Introducing a better scheduling system for internal processes process design in industrial engineering. In 21st International
could help immensely in reducing setup times. This would Conference on Production Research, Stuttgart, Germany (CD-ROM)
help in providing more accurate delivery times to the cus- 12. Chowdary BV, George D (2011) Improvement of manufacturing
tomers. The furnace is also an area that requires future atten- operations at a pharmaceutical company: a lean manufacturing ap-
proach. J Manuf Technol Manag 23(1):56–75
tion; it would be worth investing in optimizing the time avail- 13. Kaale RL et al (2005) Time value stream mapping as a tool to
able for the furnace. measure patient flow through emergency department triage. Ann
Emerg Med 46(3):108
14. Morlock F, Meier H (2015) Service value stream mapping in indus-
trial product-service system performance management. Procedia
CIRP 30:457–461
15. Hines P, Rich N (1997) The seven value stream mapping tools. Int J
References Oper Prod Manage 17(1):46–64
16. Jones D, Womack J (2002) Seeing the whole. Lean Enterprise
Institute, Brookline
1. Hodge GL et al (2011) Adapting lean manufacturing principles to
17. Matt D (2008) Template based production system design. J Manuf
the textile industry. Prod Plan Contr 22(3):237–247
Technol Manag 19(7):783–797
2. Melton T (2005) The benefits of lean manufacturing: what lean
18. Russell RS, Taylor BW (1999) Operations management, 2nd edn.
thinking has to offer the process industries. Chem Eng Res Des
Prentice-Hall, Upper Saddle River
83(6):662–673
19. Sihn W, Pfeffer M (2013) A method for a comprehensive value
3. Pattanaik L, Sharma B (2009) Implementing lean manufacturing stream evaluation. CIRP Ann Manuf Technol 62(1):427–430
with cellular layout: a case study. Int J Adv Manuf Technol 42(7-
20. Tyagi S, Vadrevu S (2015) Immersive virtual reality to vindicate the
8):772–779
application of value stream mapping in an US-based SME. Int J
4. Bhasin S, Burcher P (2006) Lean viewed as a philosophy. J Manuf Adv Manuf Technol 1-14
Technol Manag 17(1):56–72 21. Khalid NK, Hashim AY, Salleh MR (2014) On value stream map-
5. Choobineh F (1988) A framework for the design of cellular ping and its industrial significance. J Ind Intell Inform 2(2)
manufacturing systems. Int J Prod Res 26(7):1161–1172 22. Vamsi Krishna Jasti N, Sharma A (2014) Lean manufacturing im-
6. Monden Y (2011) Toyota production system: an integrated ap- plementation using value stream mapping as a tool: a case study
proach to just-in-time. CRC Press from auto components industry. Int J Lean Six Sigma 5(1):89–116
7. Nakajima S (1988) Introduction to TPM: total productive mainte- 23. Kalsaas BT (2002) Value stream mapping. An adequate method for
nance. Productivity Press, Inc, Cambridge, p 129 going lean?
1586 Int J Adv Manuf Technol (2016) 84:1573–1586

24. Seth D et al (2008) Application of value stream mapping (VSM) for 35. McDonald T, Van Aken EM, Rentes AF (2002) Utilising simulation
minimization of wastes in the processing side of supply chain of to enhance value stream mapping: a manufacturing case applica-
cottonseed oil industry in Indian context. J Manuf Technol Manag tion. Int J Logist 5(2):213–232
19(4):529–550 36. Chen C Snowdon J, Charnes J (2002) The current and future status
25. Profozich D (1997) Managing change with business process simu- of simulation software. in Winter Simulation Conference
lation. Simon & Schuster Trade 37. Chan FT (1995) Using simulation to predict system performance: a
26. Solding P, Gullander (2009) Concepts for simulation based value case study of an electro-phoretic deposition plant. Integr Manuf
stream mapping. in Simulation Conference (WSC), Proceedings of Syst 6(5):27–38
the 2009 Winter. IEEE 38. Chan FT, Abhary K (1996) Design and evaluation of automated
27. Sohal AS (1996) Developing a lean production organization: an cellular manufacturing systems with simulation modelling and
Australian case study. Int J Oper Prod Manage 16(2):91–102 AHP approach: a case study. Integr Manuf Syst 7(6):39–52
28. Abdulmalek FA, Rajgopal J (2007) Analyzing the benefits of lean 39. Chan FT, Jiang B (1999) Simulation-aided design of production
manufacturing and value stream mapping via simulation: a process and assembly cells in an automotive company. Integr Manuf Syst
sector case study. Int J Prod Econ 107(1):223–236 10(5):276–283
29. Sigari S, Clark R (2013) Applying lean thinking to improve the 40. Irani Z et al (2000) Re-engineering manufacturing processes
production process of a traditional medium-size British manufactur- through simulation modelling. Logist Inf Manag 13(1):7–13
ing company. Int J Inf Oper Manag Educ 5(2):154–169 41. Lemessi M et al. (2012) Semi-automatic simulation-based bottle-
30. Gurumurthy A, Kodali R (2009) Application of benchmarking for neck detection approach. in Proceedings of the Winter Simulation
assessing the lean manufacturing implementation. Benchmark Int J Conference. Winter Simulation Conference
16(2):274–308 42. Roser C, Nakano M, Tanaka M (2003) Comparison of bottleneck
31. Lian Y-H, Van Landeghem H (2007) Analysing the effects of lean detection methods for AGV systems. in Simulation Conference,
manufacturing using a value stream mapping-based simulation gen- 2003. Proceedings of the 2003 Winter. IEEE
erator. Int J Prod Res 45(13):3037–3058 43. Kelton WD, Law AM (2000) Simulation modeling and analysis.
32. Braglia M, Carmignani G, Zammori F (2006) A new value stream McGraw Hill Boston
mapping approach for complex production systems. Int J Prod Res 44. Roser C Nakano M, Tanaka M (2002) Tracking shifting bottle-
44(18-19):3929–3952 necks. in Japan-USA Symposium on Flexible Automation
33. Swallmeh E et al. (2014) Integrating simulation modelling and val- 45. Sengupta S, Das K, VanTil RP (2008) A new method for bottleneck
ue stream mapping for leaner capacity planning of an emergency detection. in Proceedings of the 40th conference on Winter simula-
department, in Sixth International Conference on Advances in tion. Winter Simulation Conference
System Simulation, SIMUL. Nice – France 46. Buffa ES, Armour GC, Vollmann TE (1964) Allocating facilities
34. Parthanadee P, Buddhakulsomsiri J (2014) Production efficiency with CRAFT. Harvard University Boston, MA
improvement in batch production system using value stream map- 47. Jensen PA (2004) Facility layout add-in Available from: https://
ping and simulation: a case study of the roasted and ground coffee www.me.utexas.edu/∼jensen/ORMM/omie/computation/unit/lay_
industry. Prod Plan Contr 25(5):425–446 add/lay_add.html