Professional Documents
Culture Documents
1. Introduction
Manufacturing system simulation modelling dates back to at
least the early 1960s (Law and Mccomas, 1998) and is one of
the most popular and powerful tools employed to analyze
complex manufacturing systems (OKane et al, 2000, Banks
et al, 2005). According to OKane et al (2000), one way to
forecast the behaviour of these systems is the use of discreteevent simulation, which consists of modelling a system where
changes occur at discrete-time intervals. This is appropriate
for manufacturing systems as their behaviour changes in
such a way.
Some manufacturing issues addressed by simulation
include specifying the need and quantity of equipment and
personnel, performance evaluation and evaluation of operational procedures (Law and Mccomas, 1998). The objectives
of simulation are classied as performance analysis,
capacity/constraint analysis, conguration comparison, optimization, sensitivity analysis and visualization (Harrell
et al, 2000).
The optimization via simulation deserves special attention.
Harrell et al (2000) dene optimization as the process of
trying different combinations of values for the variables that
can be controlled in order to seek for the combination of
values that provides the most desirable output from the
*Correspondence: JAB Montevechi, Universidade Federal de Itajuba
IEPG, Av. BPS, 1303Caixa Postal: 50, Itajuba, MG, 37500-903,
Brasil.
E-mail: montevechi@unifei.edu.br
2. Discrete-event simulation
Simulation is the process of designing a model of a real
system and conducting experiments with such a model. This
is done with the purpose of understanding the behaviour
of the system and/or evaluating various strategies for the
operation of the system (Shannon, 1998). Some advantages
of simulation are:
One can simulate systems that already exist as well as
those that are capable of being brought into existence.
Simulation allows one to identify bottlenecks in information, material and product ows and to test options for
increasing ow rates.
It allows one to gain insights into how a modelled system
actually works and to understand which variables are
most important to performance.
A signicant advantage of simulation is its ability to let
one experiment with new and unfamiliar situations and to
answer what if questions.
5. System modelling
3. Optimization via simulation
Optimization via simulation is the process of trying different
combinations of values for variables that can be controlled
To accomplish the objective of this paper, two manufacturing cells from different companies have been modelled. The
rst one is an application in a Brazilian weapon factory
Figure 1
Quantity
Quantity
Quantity
Quantity
Quantity
Quantity
Quantity
Quantity
Quantity
machines
machines
machines
machines
machines
machines
machines
machines
machines
type
type
type
type
type
type
type
type
type
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
Figure 2
6. Experimentation
Initially, the experiments are planned to identify the most
statistically signicant model factors. After that, the experiments generate an objective function that is used in the
optimization via Solver. Then, these most statistically
signicant factors are used as input data for optimization
via Simrunner. Also, this model is optimized using all factors
as input data in order to compare the results.
Table 2 Variable factor assignment for the multinational automotive plant application
Variable
Factor
A
B
C
D
E
F
G
H
J
K
Quantity of operators
Mean time of the processing of the operators
Mean time of the total setup to machine_01
Mean time of the partial setup to machine_01
Mean time of the total setup to machine_02
Mean time of the partial setup to machine_02
Mean time of the total setup to machine_03
Mean time of the partial setup to machine_03
Mean time of the total setup to machine_04
Mean time of the partial setup to machine_04
High level ( )
2
2
109
39
154
80
152
55
74
49
4
1
83
21
105
55
115
35
46
28
el lni
ni !
X 0; 1; 2;
Prot (Y)
Transformed Y
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
280 036
299 418
299 518
280 236
314 818
295 436
315 036
295 654
299 418
280 036
280 136
260 854
295 436
276 054
295 654
315 036
309 109
270 345
270 545
270 445
285 845
285 745
391 945
391 945
270 345
270 345
270 545
270 445
285 945
285 845
398 545
359 781
0.24076
0.56152
0.56448
0.22885
0.94377
0.43683
0.94834
0.44403
0.56152
0.24076
0.23479
2.36766
0.43683
0.50647
0.44403
0.94834
0.81693
1.02157
0.99989
1.01069
0.06364
0.05902
1.93948
1.93948
1.02157
1.02157
0.99989
1.01069
0.06824
0.06364
1.99412
1.62554
Observed
Poisson probability
Expected
Contribution to chi-square
p299 418
299 419299 517
X299 518
23
0
9
0.733295
0.055936
0.210769
23.4654
1.7900
6.7446
0.00923
1.78996
0.75421
Total
32
2.55341
Select a Transformation
Percent
90
50
10
0.66
P-Value for AD test
99
0.20
0.15
Ref P
0.10
0.05
0.00
0.2
0.4
0.6
1
200000
300000
400000
Percent
90
0.8
Z Value
1.0
1.2
N
32
AD
0.485
P-Value 0.212
50
10
1
-2
Figure 3
Term
0.497
C
BC
CE
BE
A
J
F
D
G
AJ
B
H
BH
CD
AH
AC
CJ
AEF
DE
EF
EJ
AE
EG
E
AD
EH
AF
AG
BD
AB
0.2
0.4
0.6
0.8
1.0
1.2
1.4
Effect
Figure 4
0.5
0.0
-0.5
-1
-1
-1
1
F
0.5
0.0
-0.5
-1
-1
-1
1
J
0.5
0.0
-0.5
-1
Figure 5
-1
-1
1,711
C
BC
Term
CE
BE
BCE
B
E
3
Standardized Effect
Pareto Chart for full factorial design with signicance level a 10%.
Figure 6
1
0.5
B
Mean of Johnson
0.0
-1
1
-0.5
-1
-1
0.5
B
-1
1
0.0
C
-1
1
-1
-0.5
-1
Figure 7
Term
0.279
A
AC
E
FH
EG
AD
H
EJ
FG
B
BJ
AGK
BF
CF
ACJ
AH
AGJ
EH
EK
AEF
D
BK
HJ
DG
AF
BC
AEH
CE
FJ
ADG
Factor
A
B
C
D
E
F
G
H
J
K
Name
num_oper
pro_time
TS_01
PS_01
TS_02
PS_02
TS_03
PS_03
TS_04
PS_04
0.2
Figure 8
0.4
0.6
0.8
1.0
Effect
1.2
1.4
1.6
1.8
num_oper
pro_time
TS_01
PS_01
10.4
9.6
8.8
2
TS_02
109
PS_02
83
39
TS_03
21
PS_03
10.4
9.6
8.8
154
105
80
55
152
115
55
35
PS_04
TS_04
10.4
9.6
8.8
74
46
49
Figure 9
28
2.01
Factor
A
B
C
D
A
B
CD
Name
num_oper
TS_02
PS_02
PS_03
D
AD
BD
Term
ABC
AC
ABD
ABCD
C
AB
ACD
BC
BCD
0
Figure 10
6
8
Standardized Effect
10
12
Pareto Chart for full factorial design with signicance level a 5%.
0:1278G 0:1478FG
In order to make a comparison between the current
approach and the genetic algorithms (GAs) approach, an
105
80
55
55
35
TS_02
10.5
11
10
10.0
9
11
9.5
10
TS_02
9
11
10
PS_02
PS_03
num_oper
9.0
2
154
105
PS_03
PS_02
10.5
10.0
9.5
9.0
80
Figure 11
55
55
35
Value
Quantity of operators
Total setup time of the machine_02
Partial setup time of the machine_02
Partial setup time of the machine_03
3
N (153; 22)
N (73; 46)
N (48; 4)
optimization routine was developed using Simrunner package, considering the signicant factors. Also, an optimization using all ten factors is performed to compare their
performances.
Parameter
Neq_Op080
Neq_Op082
Neq_Op120
Value
2
2
2
Table 8 Best solution for optimization using all factors for the
multinational automotive plant application
Parameter
Value
Quantity of operators
Processing time of the operators
Total setup time of the machine_01
Partial setup time of the machine_01
Total setup time of the machine_02
Partial setup time of the machine_02
Total setup time of the machine_03
Partial setup time of the machine_03
Total setup time of the machine_04
Partial setup time of the machine_04
4
N (1; 0.30)
N (102; 42)
N (21; 12)
N (126; 22)
N (55; 46)
N (119; 41)
N (35; 4)
N (54; 33)
N (33; 15)
Table 7 Best solution for optimization using all factors for the
Brazilian weapon factory application
Parameter
Neq_Op050
Neq_Op052
Neq_Op070
Neq_Op080
Neq_Op082
Neq_Op100
Neq_Op110
Neq_Op120
Neq_Op170
Value
1
1
1
2
2
1
1
2
1
Parameter
A
B
C
D
E
F
G
H
J
Result (Prot)
Number of runs
Design of
experiments
Optimization
using three
factors
Optimization
using all
factors
1
1
1
2
2
1
1
2
1
411 327
1 (*)
1 (*)
1 (*)
2
2
1 (*)
1 (*)
2
1 (*)
411 327
1
1
1
2
2
1
1
2
1
411 327
32
98
Note: The parameters identied with (*) were not used as input for
optimization. They were kept at low level (1).
Table 10 Results from the three procedures for the multinational automotive plant application
Parameter
A
B
C
D
E
F
G
H
J
K
Lead Time (hour)
3
2 (*)
109 (*)
39 (*)
153
73
152 (*)
48
74 (*)
49 (*)
8.28
Number of runs
64
213
Computational time 3 0.2 3.2 3 0.3 3.3
(hour)
4
1
102
21
126
55
119
35
54
33
7.42
2571
3.2
Note: The parameters identied with (*) were not used as input for
optimization. They were kept at low level.
7. Conclusions
The objective of this work was to show how a sensitivity
analysis using Factorial Designs can help the simulation
optimization to reach the best solution. Initially, to
accomplish this objective a fractional factorial design was
used to identify the most statistically signicant effects of
two models. As the experimental response variable of the
rst application does not follow a normal distribution, it was
necessary to apply a transformation to make it normal. As
the response variable follows a Poison distribution, the
Johnson transformation was used. The analysis was
performed using the Lenths method because the fractional
factorial design was unreplicated which made the bilateral
t-test or ANOVA not possible to assess the signicance
of the main and interactions effects.
Afterwards, three optimization trials/tests were carried
out: two using the factors previously identied, and the other
using all factors of the model. For these applications, it is
clear the advantage of determining previously the main
factors and then proceeding the optimization using them
(Solver or Simrunner) instead of proceeding with the
optimization using all factors. Beyond this discussion, it
must be pointed out that the DOE approach improves the
manufacturing system understanding, generating further
knowledge about the importance and signicance of each
resource used, which in turn favors the improvement of the
decision-making process. More than how many resources
are needed to optimize a system, improving its productivity,
increasing its prots and reducing costs, despite of the
relative time spent in the construction of the models,
this twofold approach (DOE/Simulation) elucidates how
the resource can be efciently changed and employed.
References
Banks J, Carson JS, Nelson BL and Nicol DM (2005). DiscreteEvent System Simulation 4th edn, Prentice-Hall: New Jersey.
Bisgaard S and Fuller H (1994). Analysis of factorial experiments
with defects or defectives as the response. Quality Eng 7(2):
429443.
Chou Y, Polansky AM and Mason RL (1998). Transforming
nonnormal data to normality in statistical process control.
J Qual Technol 30: 133141.
Fu MC (2002). Optimization for simulation: Theory vs. practice.
INFORMS J Comput 14(3): 192215.
Haldar A and Mahadevan S (2000). Probability, Reliability and
Statistical Methods in Engineering Design 1st edn, John Wiley &
Sons: New York.
Harrell C, Ghosh BK and Bowden R (2000). Simulation Using
ProModel 3rd edn, McGraw-Hill: Boston.