Modeling and Analysis of a Manufacturing Plant Using Discrete Event Simulation

Radha Krishna R et al. Int. Journal of Engineering Research and Application www.ijera.com
ISSN : 2248-9622, Vol. 7, Issue 2, ( Part -3) February 2017, pp.49-54
www.ijera.com DOI: 10.9790/9622- 0702034954 49 | P a g e
Modeling and Analysis of a Manufacturing Plant Using Discrete
Event Simulation
Radha Krishna R., Siva Krishna S, Vijay Bhaskar A, Sriram G, Vamsi P*,
TVSRK Prasad**
*UG Students, Department of Mechanical engineering, VFSTR, Vadlamudi, Guntur Dt, India
**Associate Professor, Department of Mechanical Engineering, VFSTR, Vadlamudi, Guntur Dt, India
ABSTRACT
Today‟s manufacturing systems are characterized by large number of complexities such as random arrival
patterns of jobs, random processing times, random failure rates, random repair times, random rejection of parts,
etc. The analytical models cannot capture all the randomness mentioned above into the models. There is a need
to incorporate them into models to have a practical and real life model. Simulation comes handy in this aspect.
Discrete Event Simulation (DES) is used to model a manufacturing system to predict its performance. The
inputs to this model include arrival rate, batch size, setup time, processing time, machine breakdown rate,
machine breakdown frequency, machines and their capacities, buffers, rejection percentage and inspection time.
The outputs that are estimated are work in process, flow time, utilization and throughput.
Keywords: Analytical Simulation, Discrete Event Simulation, Random numbers, Optimum process
I. INTRODUCTION
1.1 What is Simulation?
It is evident that there are many problems
of real life which cannot be represented
mathematically due to the stochastic nature of the
problem, the complexity in problem formulation, or
the conflicting ideas needed to properly describe the
problem under study. Under such circumstances
simulation is often used when all else fail. This
method is often viewed as a “method of last resort.”
Simulation is the representative model for
real situations. In the laboratories we often perform a
number of experiments on simulated models to
predict the behavior of the real system under true
environments. The environments in a museum of
natural history and in a geological garden are also
good examples of simulation. Actually the idea of
simulating real system for enjoyment purposes is
already known to us. The chess-playing game is a
non-probabilistic simulation of a flight between
black and white armies. The game of snake and
ladders was initially proposed to simulate the moral
progress of the players who moved up ladders when
they were „good‟ and fell down snakes, indicating
temptation, when they were bad. Like in many other
board games, dice are used as random number
generators.
In all these examples, we have tried to
represent the reality to observe- what would happen
under real operating situations. Thus, such
representation to reality, which may be either in
physical form or in a mathematical equations form,
may be called simulation. A simulation model
mainly consists of two basic phases:
Phase 1: Data Generation. Data generation
involves the sample observation of variables and can
be carried out with the help of any of the fallowing
methods :
(i) Using the random tables:
(ii) Restoring to mechanical devices
(iii) Using electronic computers.
Phase 2: Book-Keeping. The book-Keeping
phase of simulation model deals with updating the
system when new event occur.
1.2 What is Excentre?
It is an important part in piston pump, whose job is
to convert rotary motion to reciprocating motion of
piston. As shown below, its surface is grinded to
mirror finish for smooth operation for over years.
Fig 1. Completely finished Excentre
The manufacturing of Excentre starts in
foundry department which includes casting,
chipping, red oxide. This later moves to machining
department which include turning, drilling, grinding,
shaft assembly and inspection. Every stage is
RESEARCH ARTICLE OPEN ACCESS

followed by inspection ensuring only quality
products leave the plant. Each stage has a rejection
percentage of 1-2%. The monthly production of
Excentre was 800-900. Everyday 50-60 components
are manufactured on an average. These components
move from stage to stage as a batch with size of 50.
II. LITERATURE SURVEY
Simulation is a powerful tool for solving
many problems, particularly in manufacturing [1,2].
Its use in the modeling and analysis of
manufacturing systems are one of its largest
application areas, which have become increasingly
important in the last couple of decades [3, 4].
Quantitative and qualitative benefits have
been attributed to simplifications, and experienced
by organizations [2]. Quantitative benefits typically
include: reduction in operating costs, throughput
time, capital costs, design-to-market time and faster
implementation of plant changes. Qualitative
benefits typically include: reduction in risk, greater
understanding of processes, improvements in
communication, better team integration and better
development of skills within the organization [2,3].
Simulations allow various issues within
manufacturing to be addressed without the
drawbacks of experimenting with a real
manufacturing system. Typical issues addressed are:
the need for and the quantity of equipment and
personnel, performance evaluation, and evaluation
of operational procedures [2,5].
The adoption of simulations have had
limited application within small and medium
Enterprises (SME) for many years due to the cost of
computing power and lack of computer literacy. As
time has passed, improvements in hardware and
availability of affordable user friendly tools have
enabled computer simulation to be frequently used
to address a wide variety of operational problems
[6]. The case study presented in this paper is
representative of an SME with less than 100
employees.
III. ANALYSIS OF EXCENTRE
The below flow chart shows the sequence of
operations performed to manufacture an excentre.
Fig 2. Process flow of Excentre
A sample of data is collected at each stage
of manufacture. The sample size (number of
observations to be taken) at a particular stage
depends on accuracy required.
Types of data: - The following data were collected
at various stages of manufacture
 Processing time
 Setup time
 Machine breakdown probability
 Machine repair time
 Fraction of parts accepted/rejected
1. Casting:-
Setup time- 20sec, 18, 22, 21, 20, 20, 20, 18, 21, 23,
20, 21, 21, 19, 20 and 22
Processing time- 30sec, 32, 29, 30, 27, 30, 32, 33,
31, 32, 29, 28, 33, 30 and 29
2. Turning:-
Setup time- 10sec, 13, 12, 13, 13, 10, 14, 15, 16, 15,
15, 14, 13, 12, 11 and 15
Processing time- 230sec, 231, 233, 236, 237, 239,
243, 245, 246, 249, 230, 233, 237, 243, 249 and 243
Machine repair time- 10800sec, 11000, 11243,
10933, 11450, 11640, 12100, 12010, 12145 and
12000
In the same manner, the data for all the processes is
collected.
3.1 Standard Error:
The standard error gives an idea about the
reliability and precision of a sample. The smaller the
SE, the greater the uniformity of sampling
distribution and hence, greater is the reliability of
sample. Conversely, the greater the SE, the greater
the difference between observed and expected
frequencies. In such a situation, the unreliability of
sample is greater. The size of SE depends upon the
sample size to a greater extent and it varies inversely
with the size of the sample.
SE= √P.Q(1/n1 + 1/n2)
Where, P= (n1p1 + n2p2) / (p1+p2)
Q= 1-p
N1= number of events in sample one = 16
N2= number of events in sample two = 50
For casting,
Mean of setup time from sample data= 20.37 =
20.37/100= 0.203= p1
Mean of setup time from simulated data= 20.56=
20.56/100= 0.205= p2
P= (0.203*16 + 0.205*50) / (50+16)
= 0.2045
Q= 1-P
= 1- 0.2045 = 0.795
Therefore, SE= √ (0.2045*0.795)*(1/16 + 1/50)
= 0.1158 < 0.5
Since the error calculated is small, the number of
observations is sufficient.

3.2 Sample Analysis:
A sample simulation is performed for casting as shown below:
Table 1: Casting setup time
Time Frequency Percentage Cumulative % Range
18
19
20
21
22
23
2
1
6
4
2
1
12
6
38
25
13
6
12
18
56
81
94
100
00-11
12-17
18-55
56-80
81-93
94-99
Table 2: Casting processing time
Time Frequency Percentage Cumulative % Range
27
28
29
30
31
32
33
2
1
3
4
1
3
2
13
6
19
25
6
19
12
13
19
38
63
69
88
100
00-12
13-18
19-37
38-62
63-68
69-87
88-99
Table 3: Casting Simulation
Abbrevations:
AT Arrival Time
I AT Inter Arrival Time
ST Setup Time
PT Processing Time

FT Flow Time
MRT Material Removal Time
RN Random Number
BD1 RN 1st
machine Breakdown Random Number
BD1 RT RN 1st
machine Repair Time Random Number
BD1 RT 1st
machine Repair Time
BD2 RN 2nd
machine Breakdown Random Number
BD2 RT RN 2nd
machine Repair Time Random Number
BD2 RT 2nd
machine Repair Time
INS RN Inspection Random Number
INS Inspection
3.3 Model Validation:
In many decision-situations, we may be
interested in knowing whether the parameters of two
populations are alike or different. We shall explain
now the technique of hypothesis testing for
differences between means.
Z= (X1-X2) / √ 2
(1/N1)+(1/N2)
Mean of setup time from sample data, X1= 20.37
Mean of setup time from simulated data, X2= 20.56
Sigma= Average of standard deviation of setup time
=(2.59+2.5)/2
= 2.5
N1= number of events in sample one = 16
N2= number of events in sample two = 50
Z= (20.56-20.37) / √ 2
(1/16 + 1/50)
= 0.278 <1.96
Since the value of Z calculated (i.e., 0.278) is
less than table value of Z (i.e., 1.96), it can be
inferred that the model is adequate with 95%
confidence level.
IV. RESULTS AND DISCUSSIONS
The fallowing system outputs are calculated using
simulation.
 Throughput
 Utilization
 Flow time
 Work In Process
The above outputs are an indication of the
systems performance measures. The above
performance measures are calculated for each stage
of manufacture and also for entire system.
For an easy understanding, the fallowing lines will
give an idea to judge the performance of the system.
1. Utilization should be high
2. Throughput should be high
3. Flow time should be small
4. Work In Process should be small.
4.1 Casting Results:
1) Average number of parts in system
= (50*32)+(49*30)+…….+(1*30) †1520
= 26 parts/sec
2) Average flow time
= 32+62+92+121…..+1520 † 50
=776 sec/part
3) Total of the inter arrival times
= 52+51+51+50+……
= 2591sec
4) Mean of the inter arrival times
= 2591 / 50
= 51.8 part/sec
5) Arrival rate = 50 ÷ 2591
= 0.01991 part/sec
6) Total service time= 1520sec
7) Average service time= 1520 ÷ 50
= 30.4 sec/part
8) Service rate= 1 ÷ 30.4
= 0.03289 part/sec
9) Utilization= Arrival rate ÷ Service rate
= 0.5864
10) Throughput= 50*60*60 ÷ 2643
= 68 part/hr
11) Work in process= 50 parts
Like above process, results will be calculated
for chipping, red oxide, turning, etc.,
4.2 Overall Results:
1) Average number of parts in system
= 65952+27937+……+7151÷
= 22 part/sec
2) Over all flowtime
=776+1374+621+……+179
= 15570 sec/part
3) Total of the inter arrival times
= 2591+3018+…= 39250sec= 11hrs
4) Avg arrival rate= 0.0191parts/sec
5) Service time= 30408 sec= 8.4 hrs
6) Service rate= 0.0949 part/sec
7) Utilization= 69.1%
8) Overall Throughput= 69.5 part/hr
9) Avg WIP= 43.8 parts

4.3 Results at a glance:
Table 4: Overall Results
Process Parts in
sys
Flow
time
I AT M AT Service
time
Utilization Throughput WIP
Casting
Chipping
Red oxide
Turning
Drilling
Shaft-ass
Grinding
Drilling
Cotter
Inspection
25
24
22
22
22
21
22
21
22
20
775
1374
620
5366
2176
271
2345
1174
1292
179
2591
3018
2108
11340
6496
1043
5866
3249
2797
742
52
63
47
258
151
24
136
80
68
19
1520
2679
1221
10505
4235
540
4584
2299
2466
359
58
86
58
89
65
53
77
70
88
47
68
57
76
14
23
145
25
45
52
190
50
48
45
44
43
43
43
41
41
40
UNITS:
Flow time(sec/part) I AT(sec) M AT(sec/part) Service time(sec) Throughput(Parts/hr) WIP(parts)
4.4 Graphs:
Based on the above results, the following
graphs are drawn to find out the optimum parameters
in each stage. The following are NON-
DOMENATED or PARETO OPTIMAL points i.e.,
each point has equal importance with the other, but
we can‟t make a conclusion which is optimum
among those. If a point is good on one objective, the
other pareto optimal point is good on other
objective.
Fig 3: Utilization vs. WIP
Note: Processes chipping, turning and drilling are
efficient from the above plot. The remaining
processes are not efficient; when the utilization and
WIP are taken as the objectives.
Fig 4: Utilization vs. Flow time
Note: Processes casting, Red oxide, Shaft assembly,
Inspection are efficient from the above plot. The
remaining processes are not efficient; when the
utilization and flow time are taken as the objectives.
Fig 5: Utilization vs. Throughput
Note: Processes casting, chipping and turning are
processes are not efficient; when utilization and
throughput are taken as objectives.
Fig 6: Flow time vs. WIP
Note: Processes Casting, Drilling and chipping are
processes are not efficient; when flow time and WIP
are taken as objectives.

4.5 Suggestions:
Setup time is more in turning. To reduce it
some fixtures can be designed and implemented.
Utilization is less in inspection. To improve
utilization, the inspector may be assigned more tasks
than present.
V. CONCLUSION
Discrete event simulation model was
designed for a manufacturing plant. The model was
validated by collecting data and conducting
hypothesis test. It was found that the model was
valid at 95% confidence level.
Simulation was performed for 50 pieces
and various performance measures were determined.
Finally, the various processes were compared
regarding efficiency. This is known as
benchmarking. Various graphs showing performance
measures were plotted and pareto optimal front was
marked.
VI. SCOPE FOR FUTURE WORK
In the present work, hand simulation was
done for one part called Ex-centre of Reciprocating
pump. The results obtained in this work may be
compared with analytical model like queueing
model. Also the results may be compared with arena
software.
Because of lack of time, we have limited
our work to hand simulation only. In future, this
work can be extended to include all parts of
reciprocating pump and also all other pumps such as
submersible pump, centrifugal pump etc, made by
the company, using arena software. Also in future, it
is proposed to determine the cost of each product
also by using simulation.
REFERENCES
[1] Jeffrey W. Herrmann, Edward Lin, Bala
Ram, Sanjiv Sarin, Adaptable simulation
models for manufacturing, Proceedings of
the 10th International Conference on
Flexible Automation and Intelligent
Manufacturing, Volume 2, pp. 989-995,
College Park, Maryland, June 26-28, 2000
[2] B.W. Hollocks, The impact of simulation in
manufacturing decision making, Control
Eng. Practice, Vol. 3, No. 1, pp. 106-
112,1995
[3] F. Hosseinpour, and H. Hajihosseini,
Importance of Simulation in
Manufacturing; World Academy of
Science, Engineering and Technology 51
2009
[4] Y.G. Sandanayake, C.F. Oduoza, D.G.
Proverbs, A systematic modeling and
simulation approach for JIT performance
optimisation, Robotics and Computer-
Integrated Manufacturing 24 (2008) 735-
743
[5] S. Andradottir, K.J. Healy, D.H. Withers,
and B.L. Nelson , Simulation of
manufacturing systems, Proceedings of the
1997 Winter Simulation Conference
[6] Methodology for rapid identification and
collection of input data in the simulation
manufacturing systems, Simulation Practice
and Theory 7 (2000) 645-656
[7] Simulation modeling and Analysis,Averill
M Law ( TATA Mc Graw Hill, 4th ed)
[8] Discrete event system simulation, Jerry
Banks(pearson,4th ed)

Modeling and Analysis of a Manufacturing Plant Using Discrete Event Simulation

More Related Content

Viewers also liked (13)

Similar to Modeling and Analysis of a Manufacturing Plant Using Discrete Event Simulation (20)

Recently uploaded (15)

Modeling and Analysis of a Manufacturing Plant Using Discrete Event Simulation