240 Int. J. Business Excellence, Vol. 14, No.

2, 2018

Performance analysis of vehicle assembly line using

discrete event simulation modelling

Akshay Sarda and Abhijeet K. Digalwar*

Mechanical Engineering Department,
Birla Institute of Technology and Science Pilani,
Pilani, Rajasthan – 333031, India
Email: akshaysarda@gmail.com
Email: akd@pilani.bits-pilani.ac.in
*Corresponding author

Abstract: The purpose of this paper is to utilise simulation as a decision

making tool in a complex manufacturing setup. A vehicle assembly line at an
automobile company in India is modelled and analysed to help managers to
identify the criticality of different parameters. Conveyor speed, operator fatigue
and incoming material quality were selected from a pool of parameters which
affect the line output. On the basis of design of experiments (DOE),
experiments were carried out to capture the effect of input parameters on the
line output. The results from these were further analysed using response surface
methodology (RSM) and analysis of variance (ANOVA). The line output was
found to be most sensitive to operator fatigue followed by incoming material
quality and conveyor speed. This paper provides a structured approach to
analyse the vehicle assembly line in an automobile company and provides a
suitable tool to the management to analyse the complex functioning of a
manufacturing system.

Keywords: discrete event simulation; DES; vehicle assembly line; ARENA®;

response surface methodology; RSM; analysis of variance; ANOVA.

Reference to this paper should be made as follows: Sarda, A. and

Digalwar, A.K. (2018) ‘Performance analysis of vehicle assembly line using
discrete event simulation modelling’, Int. J. Business Excellence, Vol. 14,
No. 2, pp.240–255.

Biographical notes: Akshay Sarda is a graduate student in Mechanical

Engineering at Birla Institute of Technology and Science (BITS), Pilani. His
areas of interest include discrete event simulation of manufacturing systems,
lean manufacturing and sustainable manufacturing. He has completed his
Bachelor’s in Technology from Rajasthan Technical University, Kota. He has
also served as a Senior Engineer, Quality Management & Methods Department
at Bosch Limited, Jaipur for two years.

Abhijeet K. Digalwar received his PhD from BITS Pilani, India. Currently, he
is working as an Associate Professor in Mechanical Engineering Dept. of BITS
Pilani. He has over 20 years of teaching and research experience at graduate
and post-graduate levels. His areas of interest includes performance
measurement systems, world class manufacturing, total quality management,
lean and sustainable/green manufacturing. He has published more than 70
papers in national and international journals and conferences in the area of his
research interest. He is a reviewer of many prestigious national and
international journals. He is a life member of Indian Society of Technical

Copyright © 2018 Inderscience Enterprises Ltd.

 Performance analysis of vehicle assembly line 241

Education, Indian Institutions of Industrial Engineering, Fellow of Institutions

of Engineers. Also, currently he is working as a President-Elect for ISDSI, the
regional body of DSI, USA.

1 Introduction

The manufacturing sector plays a pivotal role in the economic growth of developing and
developed countries. Emerging countries like India have also intensified focus to the
manufacturing sector as a roadmap to achieve growth. In a recent initiative, the
government of India has also launched the ‘Make in India’ program to reinforce the focus
on the manufacturing sector. Hence subsequently as the manufacturing sector flourishes,
industries and manufacturing units will be inadvertently on the path of optimisation. This
presents a challenge to production managers to utilise the given resources optimally.
Taking decisions on a variety of issues can be a daunting task in the shop-floor since
systems are getting more and more complex. The interactive nature of the effective
factors in such a complex production logistics system, for example stochastic parameters,
such as process times, market demand and resource failure, makes it difficult to study the
problems analytically (Darayi et al., 2013).The interaction between different resources,
processes and products leads to multiple possible scenarios which cannot be evaluated
without a simulation technique. Trial and error experimentation with a real system is
expensive and inapplicable. So, simulation modelling methodology facilitates a what-if
analysis study of the current system be pursued without disruptions on the real case. The
need is to use a tool that aids in decision making and also saves on time. Jayaraman and
Agarwal (1996) suggested common application objectives for which simulation can be
used in the automotive/manufacturing industry. These were system throughput
determination, bottleneck detection, manpower allocation and optimisation, comparing
operating philosophies, logistics systems design and analysis, analysis of materials
storage issues, optimising shift patterns, materials handling systems design. Manzini et al.
(2006) demonstrated an innovative approach based on simulation tools and heuristic
techniques to optimise a manufacturing system from the Italian automobile industry.
Simulation has transformed from being a highly specialised and expensive tool in the
1950s to an easy to use, readily available tool in the present generation. It is being used in
the design phase of processes itself to identify and eliminate problems in the nascent
stage (Kelton et al., 2010). Discrete event simulation (DES) is an efficient tool for
solving production planning and control (PPC) problems. This is primarily attributed to
the ability of DES to model and depict process dynamics such as queue build-up and
resource behaviour. It is used to study problems at operational and tactical level (Jeon
and Kim, 2016). Hence DES is a popular choice for solving problems in the
manufacturing domain where system performance and uncertainty are important factors.
For the automotive industry specifically, Jayaraman and Gunal (1997) stated that DES is
now a standard tool used in the design and implementation of different automotive
manufacturing systems ranging from a connecting rod machining sub-system all the way
up to the much more complex automotive assembly systems. They also highlighted that
discrete-event simulation tools available today provide built-in constructs to
242 A. Sarda and A.K. Digalwar

accommodate the manufacturing system parameters. The animation of the assembly line
operation can help engineers to visually detect problems or bottlenecks and also to test
out alternate line designs. The only disadvantage of DES is that it requires a lot of data
for modelling (Jeon and Kim, 2016). Despite of that, DES is a preferred tool for solving
problems in production planning as well as production control. Out of the studies in PPC
that employ simulation, almost 50% of the papers in production planning and around
60% of the papers in production and process design have employed DES (Jeon and Kim,
2016). Many DES software are available in the marketplace such as ARENA®, Flexsim,
Anylogic, plant simulation, auto mode etc. We have chosen ARENA® simulation
software for our analysis. ARENA is a simulation, automation software that has been
developed by Rockwell Automation Company. Using the SIMAN processor and its
simulation language, it is widely used to simulate manufacturing or service processes to
analyse the current performance and other alternative modules (Wang et al., 2009).
ARENA has the flexibility of high level simulators as well as basic procedural languages
such as Microsoft® Visual Basic® and C. Users have the option of selecting alternative
and interchangeable templates of graphical simulation modelling and analysis modules
that can be combined to build a wide variety of simulation models (Kelton et al.,
2010).ARENA provides us with the ease of organising process blocks such as create,
decide, separate, batch etc. that can be connected by lines to establish the flow logic of
the entities in the system. The creation of entities and availability of resources in the
model can be controlled by calendar schedule patterns, from which a complex schedule
including exceptions can be precisely defined. We can additionally define attributes on
the entities that simplify modelling and enhance control over the flow logic (Wang et al.,
2009). Additionally, ARENA provides dynamic animation in the same work environment
and integrated support such as graphics for some of the statistical design and analysis
issues which are important for a good simulation study.

2 Literature review

A number of research papers on simulation were referred to where real time processes
have been simulated and optimised using ARENA®. The parameter of optimisation might
vary from one study to another but the essence of the study remains the same. These
studies have harnessed the simulation capabilities of ARENA® and presented solutions to
the respective problems. L Rahman and Ullah Sabuj (2015) carried out a study on a UPS
manufacturing line where the output is not satisfactory as compared to the input. The
goals of the study have been to minimise flow time, improve layout and maximise
production. The manufacturing process has been modelled after analysing the input data
using the ARENA® Input Analyzer module and flow logic has been created according to
the existing process flow. Based on the simulation reports generated, the authors have
identified the bottlenecks in the process and accordingly improvements have been
suggested, mainly pertaining to the process layout and the manpower allocated. Hecker
et al. (2010) have applied computer based simulation technique in the German bakery
industry. The industry utilised an old fashioned approach to production planning where a
single person or a small group of people are responsible for doing the production
planning. This brought along operational problems of in-efficient staff allocation,
 Performance analysis of vehicle assembly line 243

bottlenecks, under utilisation of resources and chaos whenever there was a lapse in
human judgment. The conventional approach of planning production in a backward
manner is highlighted and analysed in the study. Sarda et al. (2015) modelled a vehicle
assembly line and analysed the relationship between the line output with parameters such
as conveyor speed, failure downtime and fatigue factor. Marsudi and Shafeek (2013)
studied the effect of batch size on workstation utilisation while keeping the throughput
rate constant. Other factors such as plant layout, human factors, and facilities were
excluded from the study. The study concluded a positive relation between processing
time and parameters such as product in, product out, utilisation and waiting time.
Greasley (2005) utilised the approach of both system dynamics and DES to analyse the
production planning facility at a gas cylinder manufacturing plant. The objective of the
study was to determine a suitable cylinder sequence so that the system performance is
optimised.
Zahraee et al. (2015) analysed a case study on a colour manufacturing line where the
author has applied Taguchi method to assess the effect of controllable and uncontrollable
factors on the production output. In another study (Zahraee et al., 2014), the authors
determined the bottleneck station in the existing production line and achieved 95%
utilisation after making suitable changes. In their research on a high speed paper
manufacturing setup, Mo and Mahmoudi (2008) have mapped the existing process and
analysed different scenarios with an objective to optimise production so that the available
time for preventive maintenance increases. Mohamad et al. (2012) suggested
improvements in the existing manufacturing line using the help of simulation. They
demonstrated that by de-linking the material transportation activity from the assembly
task at every workstation by using a material handling operator (MHO) significantly
improves the production output. In another study regarding a car manufacturing line
belonging to the Pars Khodro Company (Iran), Rahmani et al. (2012) utilised the
OptQuest module of ARENA® after developing the simulation model. The decision
variable selected was the number of workstations after ensuring a suitable trade-off
between cost of setting up workstation and total throughput. As highlighted by the
authors, O’Kane et al. (2000) simulation studies in manufacturing can be used not only at
the justification phase and design phase but also at the operational phase of
manufacturing. The authors modelled the engine manufacturing facility of a large
company and analysed the effect of changing machines, shifts, labours, routing and jigs
on output, WIP, lead time, machine and labour utilisation. The study by Darayi et al.
(2013) shows how a hybrid Kanban and (r,R) production control and inventory
replenishment strategy can be incorporated and evaluated in a simulation model and give
better results. Wirabhuana et al. (2008) pointed out that simulation is very useful in
providing a ‘test drive’ before making capital investments, without disrupting the existing
system. The authors proposed alternatives based on line balancing, parallel operation,
separated material handling system and full synchronised material handling system.
These alternative models were then compared using the performance indicators line
output, finish product cycle time and line efficiency. Teixeira et al. (2014) created a
model of a manufacturing system comprising of parallel machines in ARENA and
studied the behaviour of the system when subject to different scheduling rules. It was
concluded that a combination of shortest queue rule and shortest processing time rule
yielded the best results. Jayachitra and Prasad (2011) analysed different layouts in an
244 A. Sarda and A.K. Digalwar

automotive component manufacturing industry using a simulation software (WITNESS

2006) to identify the superior performance of virtual cellular layout in comparison with
functional layout, in terms of cost and other performance metrics. This enabled the
authors to conduct a comparison without any relocation of resources which will consume
time and money. Similar to the approach adopted in the present study, Singholi et al.
(2013) identified number of parts and routing flexibility as the critical factors that affect a
flexible manufacturing system the most. They utilised simulation in a FMS environment
to analyse system performance.
Simulation has not only been incorporated by the manufacturing industry. There are
several examples where ARENA® has been used to analyse a process apart from
manufacturing. Sharma and Garg (2012) presented a framework based on Taguchi’s
experiment design which studied the effect of decision and system parameters on the
performance of an automobile service centre. Shi et al. (2015) analysed the telephone
response systems of veterans’ affairs (VA) hospitals, where they addressed existing
issues of long waiting times. The objective was also to make the system more resilient to
changes. The authors came up with resource sharing schemes which proved to be a better
alternative than simply increasing manpower. Simulation models assisted the authors to
analyse the performance of their suggested resource sharing schemes. Hailu et al. (2015)
have evaluated different strategies of product distribution in order to reduce the response
time. The strategies have been evaluated on the basis of 2 variables, transportation cost
and time. They have used ARENA® to analyse a dynamic supply chain network subjected
to variability in terms of demand and transportation. After testing different strategies in
the simulation model, the results were used by applying mixed integer linear
programming (MILP) to identify the best strategy out of these. Liong and Loo (2009)
have captured the functioning of a warehouse and analysed the process by splitting it into
two value streams, loading system and unloading system. Results of the simulation, go on
to indicate a deficiency in the loading system where the lorries from the customer are
facing much longer residence times in the system as compared to the supplier lorries.
Allocation of additional manpower and scheduling of the customer lorries result in
achieving the desired objective of optimising the loading system. Memari et al. (2012)
have applied simulation techniques to a supply chain problem for a production
distribution network. Taking transportation cost and transportation time as the evaluation
criteria, results from different scenarios were tabulated and evaluated using MILP. Wan
et al. (2010) made use of the OptQuest package in ARENA® to optimise allocation of
equipments under constraints of equipment utilisation and minimum operating time for
the handling system of a container port. Shahpanah et al. (2014) focused on solving the
queuing problem to reduce waiting times for the ship tugging operations. The different
scenarios implemented indeed showed improvement, but they also highlighted that
increasing the number of resources beyond a specific point does not have an effect on
waiting time for this particular model. In the study by Duguay and Chetouane (2007) the
complexity of a healthcare system has been modelled using DES. The objective was to
reduce waiting times and the approach considered physicians, nurses and examination
rooms as control variables. After suggesting and simulating scenarios with different
values of control variables, the best possible scenario was selected. Hatami et al. (2014)
created an ARENA® simulation model of a bank system and analysed strategies to
improve customer retention. Greasley and Barlow (1998) used simulation as a re-
 Performance analysis of vehicle assembly line 245

engineering tool to redesign the custody process in police custody in the UK. The found
that simple reallocation of tasks and testing the changes using simulation provided a
significant change in resource utilisation from the custody officer to the jailer. Even
though the research in the field of manufacturing simulation has been extensive, there
have not been studies that help us identify the criticality of the various parameters which
influence the productivity. The present study aims at addressing this research gap and
illustrates how we can incorporate response surface methodology (RSM) along with
simulation to rank parameters that are most critical to output/productivity. This aids
engineers/managers in knowing which factors to address first while striving for optimal
results in the fastest possible manner. The research in the field of simulation modelling
essentially projects simulation as a boon to decision making but there are certain
limitations also of this approach which cannot be ignored. Primarily simulation is a data
driven process and relies heavily on correct data. This places a huge responsibility on the
model builder to firstly gather data in the correct manner and secondly apply the collected
data using the correct set of modules, structuring in the simulation environment. Models
built on incorrect data can mislead the organisation towards erroneous results and
strategies. Since simulation is a data intensive process, areas where available data is
insufficient, the model development must be carried out using the correct set of
assumptions. This further emphasises the importance of model validation and verification
techniques which have been addressed in the present research. Since simulation models
aim to mimic the real system, building a simulation model can get complex involving a
number of variables. The simulation software adopted must be capable of handling the
level of complexity being tried to achieve. Simulation software is dependent on the
computer hardware they are being run on and hence this adds to the costs associated.
Most simulation software are not open source editions and require the user to invest in
procuring the license which can be costly depending on the software. Lastly, it needs to
be recognised that simulation is a method of arriving at the solution to a problem and not
the solution itself. It can be an extremely resourceful tool if implemented appropriately
and correctly.

3 Vehicle assembly at ABC motors

ABC motors is one of the leading commercial vehicle manufacturers of India and due to
data confidentiality agreements, the name of the organisation has been kept confidential.
The company started in 1945. It has revenues of more than INR 100 million annually.
There are more than 34,000 employees working for the company that has its presence in
over 70 countries. It has a market share of around 38% in the utility vehicle segment. The
vehicle assembly line at ABC Motors consists of 40 stations out of which 33 stations are
on the main line and seven stations act as feeder stations to the main assembly line. More
than 150 operators of different skills and age work on the stations in three shifts of eight
hour duration each. The assembly sequence begins with the chassis being introduced at
the chassis loading and axle loading station and continuing up to the end of the assembly
line. On an average, the assembly line produces 240 vehicles per shift of different
variants/models. The company was planning to undertake decisions on optimisation at its
facility, for which the need to understand the criticality of different parameters on line
output arose.
246 A. Sarda and A.K. Digalwar

4 Simulation modelling of vehicle assembly line

The modelling was initiated by understanding the functioning of the existing vehicle line
at ABC Motors. This involved close observation of each station of the assembly line as
well as data collection primarily pertaining to cycle time and manpower requirement. The
data acquired was then incorporated into the model along with certain assumptions. This
was followed by model verification and validation which is imperative and crucial for
any simulation study. Once the accuracy of the model was validated, experiments were
conducted using the model. The results from these experiments were then utilised for
further statistical studies.

4.1 Data collection

The input data required for building the model was collected from the actual assembly
line. Data for six shifts each of eight hour duration was collected pertaining mostly to
individual process timings, manpower, cycle time, daily output etc. Due to a limitation on
the data available (six working shifts), the maximum cycle time for each worker/process
was considered as the cycle time for that worker/process. Availability of more data on
process times etc. would have enabled the authors to prepare a stochastic model rather
than a deterministic model.

4.2 Model translation

Using the above data a model which mimics the assembly line was created (Figure 1)
using the various modules provided in ARENA®. Additionally a number of crucial
assumptions were made as follows:

• The space occupied by the part on the conveyor was 15 feet.

• The conveyor was of a non accumulating type and moved with a constant speed.
• The conveyor belt was divided into cells of 5 feet each. Thus the chassis occupied
three cells on the conveyor.
• The part processing times were considered to be deterministic.
• There was a lunch break of half an hour included in the schedule at the beginning of
the sixth hour.
• There was no limitation on the number of raw materials being created for the feeder
lines neither was there any consideration of storage constraints for inventory at the
stations.
• Since the actual scenario consisted of workers moving along the conveyor the actual
transfer time between stations was much less than that in the model where the
resources are immobile. Thus the conveyor speed was proportionately scaled to
achieve the actual transfer time of 12 seconds (approx) between stations.
 Performance analysis of vehicle assembly line 247

Figure 1 Simulation model (see online version for colours)

4.3 Model verification and validation

Several verification techniques were applied to ensure that the model has represented the
physical system correctly and that it gives correct results. These were: reviewing the logic
repeatedly, checking the output for reasonableness, observing the animation (Figure 2)
during trials and using the debug feature available in ARENA® to trace errors. Validation
was carried out by comparing the results of the simulation runs with the actual data
regarding the line output and cycle time. The difference between the simulation results
and the actual line output (Table 1) was within permissible limits i.e. 10% difference
(Hailu et al., 2015). In one of the multiple validation techniques suggested by Akiya et al.
(2011), sensitivity analysis can be adopted in this regard. The technique is about
changing the values of the input and internal parameters to see the effect on the output
parameters, in this case line output. The values of conveyor speed, material quality and
fatigue factor were altered to observe a linear relationship between these factors and line
output (Sarda et al., 2015) (Figures 3–5). This validated our simulation model logic.

Figure 2 Model animation (see online version for colours)

248 A. Sarda and A.K. Digalwar

Table 1 Comparison of outputs for 1 hour run

Actual line output Simulation line output

34 units 31 units

Figure 3 Variation of line output with conveyor speed (see online version for colours)

Figure 4 Variation of line output with operator fatigue (see online version for colours)

Figure 5 Variation of line output with material quality (see online version for colours)
 Performance analysis of vehicle assembly line 249

4.4 Experiment design

RSM has been adopted to statistically investigate the influence of process parameters on
the line output. RSM is a collection of statistical techniques that are useful for the
modelling and analysis of problems in which one or more responses of interest are
influenced by several variables (Chien et al., 2010). The objective is to find a relationship
between the responses and several variables and optimise the response however
optimisation is not a part of the current study. RSM turns out to be a powerful tool for
analysing productivity and quality. RSM has the objective of establishing a functional
relationship between the responses and input variables in an unknown and complex
multivariable system subjected to noise. The output function is finally expressed as
y = f ( x1 , x2 , x3 , x4 …… xn ) + ε

where ε represents the error observed in the response variable y, and x1, x2, x3, x4 ….xn.are
the independent variables and f is a first order or second order model (Chien et al., 2010).
In our current investigation, the input variables considered were conveyor speed(c),
incoming material quality (q) and fatigue factor (f). The response variable we were
interested in was line output. Taguchi’s L27 orthogonal array was chosen as the preferred
methodology for designing and conducting experiments (Das et al., 2015). The three
levels of the parameters chosen are shown in Table 2. The L27 OA will have 27 rows
corresponding to the 27 experiments and three columns pertaining to each of the factors
with three different levels of values First column being conveyor speed (c), material
quality (q) and fatigue factor (f). The results of the experiments are listed in Table 3.
Table 2 Model parameters and levels

Levels
Parameter Symbol Units
1 2 3
Conveyor speed c Feet/min 30 50 70
Material quality q % 10 30 50
Fatigue factor f Dimensionless 1 1.3 1.5

Table 3 Experimental results

Line output
Test no. c F q
Y
1 70 1 10 258
2 70 1 30 247
3 70 1 50 180
4 70 1.3 10 198
5 70 1.3 30 198
6 70 1.3 50 171
7 70 1.5 10 170
8 70 1.5 30 170
250 A. Sarda and A.K. Digalwar

Table 3 Experimental results (continued)

Line output
Test no. c F q
Y
9 70 1.5 50 162
10 50 1 10 245
11 50 1 30 241
12 50 1 50 173
13 50 1.3 10 189
14 50 1.3 30 190
15 50 1.3 50 168
16 50 1.5 10 163
17 50 1.5 30 163
18 50 1.5 50 157
19 30 1 10 218
20 30 1 30 218
21 30 1 50 174
22 30 1.3 10 171
23 30 1.3 30 172
24 30 1.3 50 152
25 30 1.5 10 149
26 30 1.5 30 149
27 30 1.5 50 147

5 Results and discussion

This study analysed the results of the experiments conducted using analysis of variance
(ANOVA) for identifying parameters that significantly affect the line output using
MINITAB16 software package. From the P values and F tests for each source of
variation given in Table 4-5, we can identify the significant parameters. The parameter is
considered statistically significant when the P value is observed to be less than 0.05 (Das
et al., 2015). Clearly from the results obtained it can see that the parameters with P value
<0.05 have higher contribution than the others. Hence the most significant contribution
comes from the fatigue factor (f) 57.15%, followed by material quality (q) 15.52% and
conveyor speed (c) 8.41%. Besides these, the interaction among fatigue factor and
material quality (f*q) contributed the highest (10.03%) as compared to other interaction
effects that were negligible. The steepness of the curves for fatigue factor and material
quality in Figures 6–7 also reiterate the above mentioned observations. The error
contribution in the results was also small (3.21%) which confirms the correctness of the
simulation model.
 Performance analysis of vehicle assembly line 251

Table 4 Analysis of variance for line output

Source DF Seq SS Adj SS Adj MS F P Contribution (%)

C 1 2,312.0 2,335.3 2,335.3 45.00 < 0.000 8.41
F 1 15,696.1 15,254.2 15,254.2 293.96 < 0.000 57.15
Q 1 4,262.7 4,805.3 4,805.3 92.60 < 0.000 15.52
c*c 1 101.4 101.4 101.4 1.95 0.180 0.36
f*f 1 68.4 68.4 68.4 1.32 0.267 0.24
q*q 1 1,166.7 1,166.7 1,166.7 22.48 < 0.000 4.24
c*f 1 24.3 24.3 24.3 0.47 0.503 0.08
c*q 1 192.0 192.0 192.0 3.70 0.071 0.69
f*q 1 2,756.0 2,756.0 2,756.0 53.11 < 0.000 10.03
Error 17 882.2 882.2 51.9
Total 26 27,461.9

Table 5 ANOVA of line output

Source DF Seq SS Adj SS Adj MS F P Remarks

Regression 9 26,579.7 26,579.7 2,953.3 56.91 < 0.000 Significant
Linear 3 22,270.8 22,394.8 7,464.9 143.86 < 0.000
Square 3 1,336.5 1,336.5 445.5 8.59 0.001
Interaction 3 2,972.4 2,972.4 990.8 19.09 < 0.000
Residual error 17 882.2 882.2 51.9
Total 26 27,461.9
Note: S = 7.20357; R-Sq = 96.79%; R-Sq(adj) = 95.09%

Figure 6 Main effect plot

252 A. Sarda and A.K. Digalwar

Figure 7 Interaction effect plot (see online version for colours)

Further the line output model using RSM was presented using (1).
Y = 449.773 + 2.25c − 334.295 f − 1.991q − 0.0102c 2 + 56.66 f 2
− 0.0348q 2 − 0.282c * f − 0.01c * q + 3.011 f * q (1)
R = 96.79% R (adj ) = 95.09%
2 2

From the line output model ANOVA, since the P-Values are less than 0.05(95%
confidence level) we can conclude that the model is significant. Additionally the R2 and
adjusted R2 values being very close to 0.1 indicate a good level of fit and accuracy. This
shows a good statistical agreement between the predictive model and the experimental
values. The comparison of the predicted and actual values has also been plotted in
Figure 8 for five additional experiments.

Figure 8 Comparison between actual line output v/s predicted line output (see online version
for colours)
 Performance analysis of vehicle assembly line 253

6 Conclusions

The purpose of the study was to demonstrate how simulation can measure the
performance of a vehicle assembly line and help evaluate different scenarios. On the
basis of the data obtained from the above evaluations, RSM and ANOVA were utilised to
rank conveyor speed, incoming material quality and operator fatigue on the basis of their
criticality on line output. This research also exhibits how using the fundamental data
modules in a simulation software such as ARENA®, one can build a model of a
complicated assembly line and its feeder stations. Additionally the functionality and
preciseness of the model can be confirmed using the animation and debug feature of
ARENA® .Data analysis was used to propose a quadratic model for line output which was
based on the operator fatigue, conveyor speed and incoming material quality. The
quadratic model proved its accuracy when it was tested on a fresh set of data.
Nevertheless, the road ahead lies in incorporating other critical parameters in the
simulation model such as failure downtimes, operator skill, setup change time etc which
at present could not be a part of this study. Including these variables would make the
model more realistic and assist decision makers to gain insights and make the right
choices. Simulation also has positive managerial implications which are indicated in this
study. Line managers now have a resourceful tool at their disposal that aids them in
making the right choices. They can analyse scenarios from different perspectives and
extract solutions and data which can provide a variety of insights about the system. All
this can be done in a virtual environment without disturbing the actual system and
decisions can be made at a much faster pace. Most importantly, simulation reinforces the
principle of data driven decision making, which is one of the founding pillars of quality
management. All present day organisations are striving to build systems that rely on data
driven decisions. Managers can use simulation results to present their ideas in a more
effective manner to all stakeholders of the organisation. This also enables them to
implement strategies effectively by taking their superiors and teams onboard. Thus the
study highlights how a structured approach using simulation and data analysis can assist
in bringing our attention to the critical areas and in eliminating the chances of taking an
ill informed decision.

References
Akiya, N., Bury, S. and Wassick, J.M. (2011)’ Proceedings of the 2011 Winter Simulation
Conference’, in Jain, S., Creasey, R.R., Himmelspach, J., White, K.P. and Fu, M. (Eds.):
Proceedings of the 2011 Winter Simulation Conference, 11–14 December, pp.2194–2205.
Chien, T.H., Jean, M.D., Lu, M.H. and Huang, W. (2010) ‘Application of response surface
methodology for robustness of responses of yttria stabilized zirconia coatings’, Journal of the
Chinese Institute of Industrial Engineers, Vol. 27, No. 2, pp.90–102.
Darayi, M., Eskandari, H. and Geiger, C.D. (2013) ‘Using simulation-based optimization to
improve performance at a tire manufacturing company’, QScience Connect (a Qatar
foundation academic journal), pp.1–12.
Das, S.R., Dhupal, D. and Kumar, A. (2015) ‘Experimental investigation into machinability of
hardened AISI 4140 steel using TiN coated ceramic tool’, Measurement: Journal of the
International Measurement Confederation, Vol. 62, pp.108–126.
Duguay, C. and Chetouane, F. (2007) ‘Modeling and improving emergency department systems
using discrete event simulation’, Simulation, Vol. 83, No. 4, pp.311–320.
254 A. Sarda and A.K. Digalwar

Greasley, A. (2005) ‘Using system dynamics in a discrete-event simulation study of a

manufacturing plant’, International Journal of Operations & Production Management,
Vol. 25, No. 6, pp.534–548.
Greasley, A. and Barlow, S. (1998) ‘Using simulation modelling for BPR: resource allocation in a
police custody process’, International of Operations &Production Management, Vol. 18,
Nos. 9–10, pp.978–988.
Hailu, H., Jilcha, K. and Birhan, E. (2015) ‘Response time reduction in the leather products
manufacturing industry using arena simulation method’, Afro-European Conference for
Industrial Advancemen, pp.271–281, Springer International Publishing.
Hatami, M., Zahraee, S.M., Ahmadi, M., Golroudbary, S.R. and Rohani, J.M. (2014) ‘Improving
productivity in a bank system by using computer simulation’, Applied Mechanics and
Materials, August, Vol. 606, pp.259–263.
Hecker, F., Hussein, W. and Becker, T. (2010) ‘Analysis and optimization of a bakery production
line using ARENA’, International Journal of Simulation Modelling, Vol. 9, No. 4,
pp.208–216.
Jayachitra, R. and Prasad, P.S.S. (2010) ‘Performance analysis of virtual cellular manufacturing: a
simulation study’, International Journal of Services and Operations Management, Vol. 8,
No. 1, pp.92–107.
Jayaraman, A. and Agarwal, A. (1996) ‘Simulating an engine plant’, Manufacturing Engineering,
Vol. 117, No. 5, p.60
Jayaraman, A. and Gunal, A.K. (1997) ‘Applications of simulation in the design of automotive
manufacturing systems’, Proceedings of the 1997 Winter Simulation Conference, pp.758–764.
Jeon, S.M. and Kim, G. (2016) ‘A survey of simulation modeling techniques in production
planning and control (PPC)’, Production Planning & Control, Vol. 27, No. 5, pp.360–377.
Kelton, W.D., Sadowski, R.P. and Swets, N.B. (2010) Simulation with Arena 5th Edition, McGraw
Hill Education (India) Pvt Limited, New Delhi.
L Rahman, C.M. and Ullah Sabuj, S. (2015) ‘Process flow improvement proposal of a batch
manufacturing system using arena simulation modeling’, Review of General Management,
Vol. 21, No. 1, pp.63–77.
Liong, C-Y. and Loo, C.S.E. (2009) ‘A simulation study of warehouse loading and unloading
systems using ARENA’, Journal of Quality Measurement and Analysis (JQMA), Vol. 5,
No. 2, pp.45–56.
Manzini, R., Persona, A. and Regattieri, A. (2006) ‘Framework for designing and controlling a
multicellular flexible manufacturing system’, International Journal of Services and
Operations Management, Vol. 2, No. 1, pp.1–21.
Marsudi, M. and Shafeek, H. (2013) ‘The evaluation of production line performance by using
ARENA – a case study’, International Journal of Mechanical, Aerospace, Industrial and
Mechatronics Engineering, Vol. 7, No. 9, pp.655–659.
Memari, A., Anjomshoae, A., Galankashi, M.R., Rahim, B.A. and Rahman, A. (2012)
‘Scenario-based simulation in production-distribution network under demand uncertainty
using ARENA’, 2012 7th International Conference on Computing and Convergence
Technology (ICCCT), December, pp.1443–1448, IEEE.
Mo, J.P.T. and Mahmoudi, M. (2008) ‘Optimisation and simulation of high speed production
system’, Journal of Achievements in Materials and Manufacturing Engineering, Vol. 31,
No. 2, pp.794–802.
Mohamad, E., Ito, T., Salleh, M.R. and Nordin, N.A. (2012) ‘Simulation study towards
productivity improvement for assembly line’, Journal of Human Capital Development, Vol. 5,
No. 2, pp.59–69.
O’Kane, J.F., Spenceley, J.R. and Taylor, R. (2000) ‘Simulation as an essential tool for advanced
manufacturing technology problems’, Journal of Materials Processing Technology, Vol. 107,
Nos. 1–3, pp.412–424.
 Performance analysis of vehicle assembly line 255

Rahmani, N., Tavakkoli-Moghaddam, R. and Ahmadi, T. (2012) ‘Designing the retouch line in an
automobile factory using a simulation-based optimization technique; a real case study’,
Procedia – Social and Behavioral Sciences, Vol. 62, No. 3, pp.995–1000 [online]
http://www.sciencedirect.com/science/article/pii/S1877042812036105 (accessed 24 October
2012).
Sarda, A., Rane, A., Digalwar, A., Sunnapwar, V. (2015) ‘Simulation of a vehicle assembly line
using ARENA’, International Conference on Industrial Engineering – ICIE 2015,
pp.857–861.
Shahpanah, A., Hashemi, A., Nouredin, G., Zahraee, S.M. and Helmi, S.A. (2014) ‘Reduction of
ship waiting time at port container terminal through enhancement of the tug/pilot machine
operation’, Jurnal Teknologi, Vol. 68, No. 3, pp.63–66.
Sharma, R. and Garg, S. (2012) ‘Performance optimisation of automobile service centre using
Taguchi approach’, International Journal of Services and Operations Management, Vol. 12,
No. 2, pp.221–243.
Shi, J., Erdem, E. and Peng, Y. (2015) ‘Article information’, International Journal of Operations &
Production Management, Vol. 35, No. 8, pp.1098–1124.
Singholi, A., Ali, M. and Sharma, C. (2013) ‘Evaluating the effect of machine and routing
flexibility on flexible manufacturing system performance’, International Journal of Services
and Operations Management, Vol. 16, No. 2, pp.240–261.
Teixeira, F.D.O., Oliveira, L. and Varela, L.R. (2014) ‘Comparative analysis of scheduling rules
through arena for parallel machines’, Nature and Biologically Inspired Computing (NaBIC),
2014 Sixth World Congress, pp.170–175.
Wan, J., Meng, Q.Y. and Wang, W. (2010) ‘Optimization for container handling equipments
allocation based on ARENA’, 2010 IEEE 17Th International Conference on Industrial
Engineering and Engineering Management (IE&EM), October, pp.1194–1197, IEEE.
Wang, T., Guinet, A., Belaidi, A. and Besombes, B. (2009) ‘Modelling and simulation of
emergency services with ARIS and Arena. Case study: the emergency department of Saint
Joseph and Saint Luc Hospital’, Production Planning and Control, Vol. 20, No. 6,
pp.484–495.
Wirabhuana, A., Haron, H. and Imtihan, M.R. (2008) ‘Simulation and re-engineering of truck
assembly line’, 2008 Second Asia International Conference on Modelling & Simulation
(AMS), pp.783–787.
Zahraee, S.M., Chegeni, A. and Rohani, J.M. (2015) ‘Characterization of manufacturing system
computer simulation using Taguchi method’, Jurnal Teknologi, Vol. 4, No. 72, pp.77–82.
Zahraee, S.M., Golroudbary, S.R., Hashemi, A., Afshar, J. and Haghighi, M. (2014) ‘Simulation of
manufacturing production line based on Arena’, Advanced Materials Research, Vol. 933,
No. 1, pp.744–748.