DISCRETE
SIMULATION APPLIED TO A GAS APPLIANCE COMPANY
Márcio Rodrigues
Clementino
Instituto
Federal de São Paulo – Campus Suzano
E-mail:
clementino_marcio@hotmail.com
Thais Taba da
Silva
Instituto
Federal de São Paulo – Campus Suzano
E-mail:
thaistaba.s@gmail.com
Adriano Maniçoba
da Silva
Instituto
Federal de São Paulo – Campus Suzano
E-mail:
adrianoms@ifsp.edu.br
Wilson Yoshio
Tanaka
Instituto
Federal de São Paulo – Campus Suzano
E-mail:
w.tanaka@ifsp.edu.br
Eugenio de
Felice Zampini
Instituto Federal
de São Paulo – Campus Suzano
E-mail:
eugenio.zampini@ifsp.edu.br
Submission: 29/03/2018
Accept: 29/03/2018
ABSTRACT
The
proper management of the operations is fundamental to the companies' results,
and, particularly, the continuous improvement in production processes is
extremely important to raise the level of efficiency. The present study aimed
to simulate the process of welding of a component called heat exchanger that
makes up a gas water heater. Then, this simulation has the objective of optimizing
the production process. The data collected were obtained from a company located
in Mogi das Cruzes, São Paulo. The results of the simulation model point to
opportunities for improvement, mainly in the direction of reducing work shifts.
Keywords: Discrete simulation, welding process, productive efficiency
1. INTRODUCTION
The market is in
an accelerated process of change, whose intensity increases as new ideas for
technological and logistical improvement emerge. This fact can be clearly seen
in recent years. As the market's demand for quality, price and customization
grows, systems are challenged, being required continual improvements to exceed
their own limits in order to achieve better performance levels with a view to
ensuring that they remain in the market.
In this
scenario, there is a continuous need to obtain advantages that distinguish a
company from its competitors in order to achieve a prominent position in the
market. Through operational research it is possible to develop innovative
technologies to obtain a competitive advantage and thus gain market leadership.
The use of
operational research and simulation can improve the production line processes,
particularly in the segment object of this paper that is a gas water heater
factory, using the study of times and methods. The process simulation technique
is useful for analyzing the behavior of queues and the relationship between
several components of a system, considering the flow of information and
physical elements.
The study in
this segment is justified because, in the analysis undertaken in the company,
the heat exchanger is the piece that represents greater value in the heater
assembly; then, the focus in the continuous improvement in its productive
process is mandatory, due to the profit margin and in light of the current
market with the expansion of competition in the sector and imports from Asia.
In this way, the
present study aimed to simulate the process of welding a component called a
heat exchanger that makes up a gas water heater with the purpose of optimizing
the production process. From the simulation performed, improvement scenarios
were created in order to obtain greater productive efficiency in the work cell
and consequent elimination of existing bottlenecks.
2. LITERATURE REVIEW
2.1.
Production
capacity and row organization
According
to Slack (2002), the operation of processes in an organization represents the
collection of resources destined to the production of its goods and services.
These resources form a set of activities and tasks, with the objective of
unifying them becoming the final product that will be produced by the company
for commercialization.
"For
a good production planning, it is not enough to make the materials available at
the right moment of its use, it must consider the existence of sufficient
capacity to carry out this production" (CALIXTO, 2011). Therefore, the
production is delimited to the factors of use of the product and its production
capacity in the system, thus having a productive flow.
The
production capacity is the maximum output that a factory can deliver. Martins
(2005) states that capacity can be defined as the maximum level that can be
achieved from value-added work considering normal operating conditions for a
certain period of time.
According
to Slack (2002) "The inefficient use of time will be transformed into
extra operating cost" and, following the logic of this statement, the
author makes an analogy of the cost of time with other fixed costs of a
company, for example, factory rent or lighting costs, which tend to remain the
same, or similar, over time, even the company producing large volumes of its
goods, or even if it cannot produce the minimum needed to cover these fixed
costs.
Conforming
to Cox and Spencer (2002), "the strength of a single resource limits the
output of the system, preventing performance improvement as a whole."
Therefore, identifying exactly the limiting resource of the system output
allows the company to achieve an increase in overall production performance
without the need for a high level of investments.
In
consonance with De Freitas Filho (2008), queuing theory is practically applied
to any process or system that involves, in some way, the possibility of queuing
due to constraints in the supply of resources in relation to demand. Once you
have a resource to use, you may experience congestion if the resource is
limited to the service.
In
agreement with Andrade (2009), queuing theory deals with system congestion
problems, whose main characteristic is the presence of clients requesting
services that are not immediately available, so they should wait for their
availability.
Moreira
(2010) considers that people associate the presence of queues with an excess of
demand of a specific service over the capacity. However, it should be taken
into account that the variability and dynamics of events also contribute to the
formation of queues. Queues are formed not only by the lack of service
capacity, but also due to the variability of the lead times, both in the time
between arrivals and in the time of attendance of these objects.
2.2.
Chrono
analysis
According
to Moreira (2004), it is possible to obtain the times of an operation in, at
least, four different ways, they are:
1st.
Time study with stopwatch;
2nd.
Measurement of historical times;
3rd.
Predetermined default data;
4th.
Sample of work.
Moreira
(2004) explains that the tolerance (T) to be applied in the calculation to
obtain the standard time has values that are foreseen in the table of typical
values for the tolerance, and for the use of the values of this table to
determine the tolerance (T), all the conditions involved in the operation must
be observed, adding the percentages identified in the table to the percentage
of the time previously computed, that is, 100%.
In
other words, it can be said that when the normal time is obtained, this time
will be considered the original value (100%) for the calculation basis, and the
tolerance factors (FT) obtained in the typical values table, when added to the
obviously, will always result in a value greater than 100%, the final value
should be higher than normal time, this is precisely to be able to observe the
effects of the operating conditions of the process studied.
After
understanding how to find the standard time, we observe how the equation will
look to determine the final values to be used in the simulation.
First,
symbols are used that will be used to identify each type of time as follows:
RT =
real time;
TN =
normal time;
TP =
standard time;
Thus,
to determine what the normal time will be:
TN =
TR x EF / 100
Where
EF = operator efficiency (%).
At
this point, called operator efficiency (EF), some value must be assigned in
percentage to the analyzed operator, and 100% will be taken as the base if the
analyst evaluates that the operator is inefficient, a certain percentage
assumed by the analyst should be subtracted of 100%. On the other hand, if the
analyst evaluates an operator as above average, it should add up to 100% the
assumed percentage. Thus, to determine what the normal time will be:
Once
the normal time is obtained, the standard time can be calculated, marking the
table of typical values to define the tolerance percentages for the operation,
according to the formula:
TP =
TN x FT / 100
Where
FT = tolerance factor (%), and we also have:
FT =
100 + T
Where
T = tolerance (%) allowed for the operation according to the typical values
table.
2.3.
Simulation
Currently,
the process of simulation is used and accepted more frequently, due to the
increasing computerized power in the workstations, associated with the
improvement of environments for the development of computational friendly
models (FREITAS FILHO, 2008).
When
studying and planning one or more processes, problems arise such as the
capacity dimensioning or the flow of processes that, several times, can have a
complex solution. However, according to Andrade (2009), the experience gained
in constructing the models and performing the simulation can lead to a better
understanding of the system, which makes it possible to improve it. By means of
several forms of model creation and the possibility of performing tests, it is
possible to obtain the solution.
Chwif
and Medina (2013), for the simulation study to be effective, proceed as
follows:
1. Design or formulation of the model: make a clear
modeling of the system to be simulated and the proposed objectives; obtaining
input data through collection; creation of the conceptual model;
2. Conversion of the model: converter the conceptual
model for the computational model; compare between the computational and
conceptual model with objective of evaluation if the operation meets what was
stipulated in the formulation of the initial model;
3. Analysis of the results of the model: apply several simulations of the model; analysis and documentation of results.
4. METHODOLOGY
In order to achieve the objective, a case study was
conducted with a gas appliance company located in the city of Mogi das Cruzes
in São Paulo. There is a production line in the company where the assembly of
the gas water heater takes place. The cell where the welding process of the
part called the heat exchanger is carried out serves to supply this production
line that currently operates in two shifts for six days per week. Depending on
the sales, the two shifts with the same number of employees working are able to
supply the demand of the production line.
However, the heat exchanger welding cell has limited
production, and routinely the production line needs more welded parts than its
productive capacity, so there is an imbalance and the employees work overtime to
supply the demand. Another measure adopted when this type of imbalance occurs
is the provisional transfer of operators made by the production manager, where
a welder of another process is relocated to supply the productive demand of the
heat exchanger welding process, which is the main part of the heater assembly
line.
This transfer generates an inconvenience for two
productive sectors of the company, once the welding operator takes a shift to
supply the demand of the gas water heater production line in another area; it
fails to comply with the productive plan of its original sector. This leads to
delays in production and deliveries, and generates some extra hours that tend
to be eliminated with the completion of this simulation project. Then, the simulation
undertaken allowed suggesting improvements to mitigate these adverse effects.
The results are presented in the next section.
The simulation undertaken allowed to suggest
improvements to mitigate these adverse effects. The results are presented in
the next section.
5. ANALYSIS OF RESULTS
In order to
create the model, it was necessary to analyze the steps of the cell to be
studied, and the following steps were identified: Arrival, Welding Time,
Cooling Time, Test Time, Water Tank Time, and thereafter collecting data from
the times of each step of the welding cell of the heat exchanger in a sample
period. 100 samples of the times in seconds presented in Tables 1 to 4 were
used as the basis.
Table 1: Welding
Time (in seconds)
Source:
the authors, 2017
Table 2: Cooling
Time (in seconds).
Source:
the authors, 2017
Table 3: Test
Time (in seconds).
Source:
the authors, 2017
Table 4: Tank
Test Time (in seconds).
Source:
the authors, 2017
It was also verified the form of probability distribution
associated to the times collected. The histograms of the data are shown in
Figure 1.
Figure 1: Histograms of collected data Source
Source:
the authors, 2017
With
the data obtained, an adhesion test was performed with the Input Analyzer tool
of the Arena simulator, to find the best distribution in the behavior of this
queue. The results of the analysis are shown in Table 5.
Table
5: Results of the adhesion test.
|
Data |
Probability
distribution |
Parameters |
Quadratic
error |
Significance of
the Chi-Square Test |
|
Welding time |
Normal |
Average 115 and standard deviation 10,9 |
0,019 |
< 0,005 |
|
Cooling time |
Normal |
Average 58,4 and standard deviation 5,49 |
0,021 |
0,0824 |
|
Test time in tank |
Gamma |
34,5 added to
Beta of 4,91 and Alpha of 3,11 |
0,042 |
< 0,005 |
|
Test time |
Weibull |
79,5 added Beta
of 4,98 and Alpha of 1,41 |
0,010 |
0,133 |
Source:
the authors, 2017
As
can be seen in Table 5, the non-significant distributions, that is, those that
did not differ from the indicated distributions were the cooling time and that
of the test time. In this way, the simulation will be conducted with the
limitation that the welding time and the test time of the tank will be
simulated with the indicated distributions even though they have not been
validated in the adhesion test.
For
the part entry, a series of timekeeping was performed, where an average
interval of 6,233 seconds between line supply arrivals was observed, and from
the calculation below, a standard time of 7,200 seconds was obtained
considering a total efficiency (EF) of 105 %, and the final result of this
calculation to obtain the standard arrival time was used in the Arena to
simulate the operation.
Figuring out the
standard time of arrival:
TN = TR X EF / 100
TN = 6,233 X 105/100 = 6,545
The
following tolerance values were considered to be applied to the normal time
according to the typical values table:
1st Personal
time - 5%;
2nd Basic
fatigue - 4%;
3º Average
Monotony - 1% .v
A
total of 10% of tolerance factor (FT) obtained should be added as follows:
FT = 100 + T
FT = 100 + 10 =
110%
Finally,
you will find the default time from the calculation below:
TP = TN X FT /
100
TP = 6,545 X
110/100 = 7,200
For
the output of pieces, a series of timekeeping was carried out, in which an
average of 109 seconds was observed to perform the movement of the cart loaded
with finished pieces to the production line, and from the calculation a
standard time of 120 seconds was obtained considering a (EF) of 100%, and the
final result of this calculation to obtain the standard exit time was used in
the Arena to simulate the operation.
For
this calculation there is no increase by efficiency factor (EF), since in this
operation efficiency was considered within the average, that is, 100% of total
efficiency. In addition, the same tolerance values used for the calculation of
the standard time of arrival are considered, so there is a total increase of
10% per tolerance factor (FT). However, even though these values are the same,
there is a substitution of the average monotony factor for the factor of use of
muscular strength in 10 pounds, which according to the table of typical values
is maintained with the same percentage of 1% to apply to the calculation. After
making the necessary considerations, the calculations to obtain the standard
exit time will be:
Calculation
of the standard output time:
FT = 100 + T
FT = 100 + 10 =
110%
TP = TN X FT /
100
TP = 109 X
110/100 = 120
5.1.
Simulation
model
The
simulation model is presented in Figure 2 after performing the adhesion test
with the objective of modelling in the Arena the current scenario, with the
actual times for the simulation to be performed.
Figure 2: Current Cell Scenario
Source:
The authors (2017
The
cell process begins with the arrival of parts that is estimated according to
the production demand, in the standard time of 7,200 seconds. After the batch
arrives, the exchangers are welded by the welder. Then, the part that was
welded is cooled and after cooling the part goes to the test process.
If
the piece is approved, it goes to the lot of pieces tested where it will form a
single lot that will be moved to the production line by a cart whose default
time is 120 seconds. In case of failure of the part, it goes to the sealing
tank process. After the tightness test is performed on the tank, the helper
identifies the weld failure on the part and returns it to the welder to rework
the weld points.
After
the model was simulated for 7 hours, the simulator generated a report with
information presented in Figure 3.
Figure 3: Graph of Utilization of Source
Source:
The authors (2017
According
to the data in Figure 3, which are presented in percentages, it is possible to
verify that the idleness of the helper resource is at 59.09% of the working
time, while the refuse operator using another Welding resource is in use for 100%
of the working time shift. According to the data collected and treated with the
Arena tool and its presented results, it is concluded that the operator
resource and the solder resource are the bottlenecks of the system.
In
this way, the presented scenario is explicitly unbalanced by the fact of
idleness of a collaborator and in counterpart to the total use of another
collaborator.
5.2.
Improvement
opportunities
Considering the
data presented in the previous analysis, one can identify the system two bottlenecks
to be explored: the Welding process and the idleness of the helper who tests
the exchangers.
5.2.1. Welding
process
Currently
the welding process has its capacity to use the resource in 100%, which limits
the system in its productive capacity, so it is necessary to analyze the
reasons that lead to this occurrence. A hypothesis raised by the fact that the
welding time is the process that demands more time within the lead time of the
production cell.
5.2.2. Helper
According
to the data collected and the process simulated by the Arena tool, it is noted
that the level of productive efficiency of the auxiliary resource that assists
the welding process is approximately 41%, and the processes performed by it to
assist the welding of the exchanger of heat has as a percentage of 36.48% for
test machine, 50, 45% for the heat exchanger solder chiller and 27.43% of the
tank that performs the leakage test performed on the exchangers that are
deprecated on the test machine. Therefore, the welding process of the heat
exchanger is not limited by the performance of the helper assisting the
process, since it is capable of performing all the auxiliary activities of the
welding process and yet it is very idle in the 7-hour shift.
5.2.3. Proposed
scenario
After
the modeling and simulation with the actual times of the work cell, the
proposal is a new scenario where improvements are applied to this work cell,
with the objective of balancing the level of resource utilization and thus
eliminating one of the bottlenecks (Figures 4 and 5).
Figure 4: New proposed scenario (Part 1)
Source:
Authors (2017)
Figure 5: New proposed scenario (Part 2)
Source:
Authors (2017)
Given
this new scenario, we have the following configuration:
-Turn
single working seven hours in six days.
-
Relocation of welder for this shift.
-
Acquisition of new Welding equipment.
With
these settings, the new scenario that presented in its report the results shown
in Figure 6 was simulated in ARENA:
Figure 6: Report of the new scenario
Source:
The authors (2017)
From
the data presented by the Arena simulator report it was possible to notice a
considerable increase in the level of use of the Helper Resource, and thus
having a balance of the line. This was only possible due to the relocation of
the welder who demanded an increase in productivity in a single shift.
Comparing
with the previous scenario, it is noticed that the percentage of use of the
helper resource that was of 40,91% increased to 77,74%, and the processes
executed by him in the new scenario have a good level of occupation, because
where before was 36.48%, now rose to 69.90% for the test machine, and from 50,
45% to 96.95% in the heat exchanger solder chiller, and from 27.43% to 42% ,
51% in the tank that performs the leakage test on the exchangers that are
rejected on the test machine.
A
welder worked 100% of the shift to supply the demand of the production line,
without causing inconveniences like the lack of finished parts. In the new
scenario with two welders the level of occupation of one welder reduced to
95.28% and another welder reduced to 97.76%, both working in a single shift.
For
the company that currently works in two shifts, this new scenario presented
will bring benefits, since it has the reduction of one shift, thus leaving this
time available for possible events such as increased demand, machine breakdown,
absence of employees, among other events.
Another
point to note is that in the current scenario we have 4 employees working in
the cell, that is, 2 employees per shift. And with the new scenario we have only
3 employees acting in a single shift, and this represents a reduction of the
cost with direct labor employed in the heat exchanger welding process.
6. CONCLUSION
The objective of this study was to identify and
analyze the potential of operational research techniques, more specifically
simulation, to perform the modeling of the problem and the application of the
Arena simulator.
Currently, the production line of the company studied,
operates in two shifts to supply the market demand and generate stocks for collective
vacations and eventual setbacks.
The process of welding the heat exchanger, which is
the main part of the gas water heater, takes place in two shifts and its
welding production is matched to the production line to meet the right time.
It can be observed that from the analyses carried out,
it is possible to suggest improvements in the flow of the analyzed operation.
Although the simulation allows the times to be estimated and approximated to
some typical probability distribution, the data used in this study are real and
collected in the work cell. Several phases of the production process were
identified and addressed in the proposed model.
The problem proposed by the study was how to reduce
the idleness of a work cell, to balance the level of utilization of the
resources and, thus, to increase the efficiency of the cell, and based on the
results obtained through the simulation, it was verified that the implantation
of the model will double the capacity of the current welding process, and the
opportunity arises to reduce the welding process to a single turn, thus having
sufficient service of welded parts to supply the demand of the two-shift heater
production line currently.
The company will have the welding equipment available
on a shift with the ability to weld parts to a possible surplus demand, for
rework, for maintenance, or even generate a stock for possible process stops
when needed.
7. REFERENCES
ANDRADE, E. L. (2009). Introdução à
pesquisa operacional: métodos e modelos para análise de decisões. 4ª ed.
LTC editora. Rio de Janeiro, p.124.
CALIXTO, F. (2011). Logística um
enfoque prático. São Paulo: Saraiva, p. 76.
CHWIF, L.; MEDINA, A. C, (2013). Modelagem
e simuação de eventos discretos: teoria & aplicações, 3ª ed. Rio de
Janeiro: Elsevier.
COX III, J. F.; SPENCER, M. S. (2002). Manual
da Teoria das Restrições: Prefácio de Eliyahu M. Goldratt. Porto Alegre:
Bookman, p. 71.
FREITAS FILHO, P. J. (2008). Introdução
à modelagem e simulação de sistemas: com aplicações em arena. 2ª ed.
Florianópolis: Visual Books.
MARTINS, P. G.; LAUGENI, F. P. (2005). Administração
da Produção. 2ª ed. rev. aum. e atual. São Paulo Saraiva, p. 31.
MOREIRA, D. A. (2004). Administração
da produção e operações. São Paulo: Pioneira Thomson Learning, p. 295-296.
SLACK, N.; CHAMBERS, S.; JOHNSTON, R. (2002). Administração da Produção. 2ª Ed. São Paulo: Atlas.