Professional Documents
Culture Documents
ITS Paper 30082 2509100033 Paper
ITS Paper 30082 2509100033 Paper
Abstract – In practice, maintenance for manufacture is Several companies, particularly the manufacturing
complex, either for plant even for every system. It requires companies has been attempting to optimize the maintenance
the proper proposed maintenance tasks. However, some activities in order to maintain their production activities.
companies have a problem for designing and implementing This aims to fulfill the determined target production. One of
maintenance management. In this case, Phonska I Plant PT the manufacturing companies implementing the
Petrokimia Gresik is selected as the observation object of maintenance activity with the purpose to meet the
this research because PT Petrokimia Gresik has not determined production target is PT Petrokimia Gresik.
implemented maintenance based on the reliability. Over the Over the years, PT Petrokimia Gresik has not been
years, Phonska I Plant had high number of unexpected executing the reliability calculation during the certain time
downtime days, and it caused the inability of Phonska I [R (t)] to all the machines and equipments in the plants. In
Plant to meet its production target. If this happens, there practice, reliability calculation should have been done by the
would be high amount of lost sale or profit loss. These Engineering Inspection Department. Nevertheless, reliability
problems have been solved by using the design of proposed calculation in PT Petrokimia Gresik has still been the
maintenance task, which is referred to the seven questions planning. Therefore, it can be concluded that PT Petrokimia
entailed by Reliability Centered Maintenance II Process and Gresik has not been executing the reliability system
Failure Mode & Effect Analysis Identification, relatively to each plant. The absence of reliability
Conventional Preventive Maintenance Scheduling with calculation in plant may cause the over-action and less-
trade-off between reliability and utility scenarios, and action maintenance or inefficiency in [2].
Maintenance Interval for Finding Failure. By using the One of many methods in maintenance task design is
Conventional Preventive Maintenance, the scenario with Reliability Centered Maintenance (RCM) II. RCM II is a
original MTTF results in 34.2 days of repair; the scenario process used to determine what to do in order each asset
with utility + 20% results in 29.33 days of repair; the may perform its function as its operational context. The
scenario with utility + 40% results in 25.08 days of repair; implementation of RCM II requires a big effort, which is the
and the scenario with utility + 50% results in 23.91 days of analysis of the components, machines, and failure
repair. With scenario original MTTF and utility + 20%, identification must be executed by using the Failure Mode
Phonska I Plant cannot meet the production target. With Effect Analysis (FMEA) structurally to maintain the
scenario utility + 40% and 50%, Phonska I Plant can meet performance of the system [3].
the production target. The Maintenance Interval for Finding Based on current maintenance activity, PT Petrokima
Failure Method is determined to get the early analysis of the Gresik needs an improvement of its maintenance system by
potential failure of a machine, so that the operator will get implementing RCM II to assist PT Petrokimia Gresik in
the precaution and direct preventive tasks to reduce the classifying the equipments and designing the maintenance
potential failure and get the plan for maintenance execution task of all the production unit in PT Petrokimia Gresik at
according to preventive maintenance schedule. once. Besides, PT Petrokimia Gresik needs to conduct the
reliability calculation of each equipments, machines, and
Keywords : Proposed Maintenance Task, Reliability components in order PT Petrokimia Gresik might have an
Centered Maintenance II, Conventional Preventive effective maintenance plan to assist PT Petrokimia Gresik in
Maintenance Scheduling, Trade-off Reliability & Utility, fulfilling the production target. In this case, the author select
Maintenance Interval for Finding Failure the Phonska I Plant of II A Department Production because
this plant has been operating since 2000, which means that
I. INTRODUCTION this plant has been operating for not too long time so that the
From the upgrading of production capacity of Phonska I, under the production target, except in 2009. The inability of
PT Petrokimia Gresik ought to maintain the function of Phonska I Plant to meet the production target will cause the
every production unit and machine in the plant to avoid loss or it is known as the lost sale cost or profit loss. If the
downtime. By restarining the downtime, Phonska I may lost sale is converted into rupiah, the lost production should
increase the availabilty of machines in order the production be calculated by calculating the difference number of
capacity can be upgraded. This downtime of system is most targeted production and actual production.
affected by the major machine in the plant. The system fails Consequently, in this research, the author attempts to
when one or more major machine stops performing its develop or implement Reliability Centered Maintenance II,
function [5]. which describes and determines the required maintenance
Based on the Annual Production Report of Production tasks and activities in order every equipment and physical
Department 2A PT Petrokimia Gresik over 5 years from asset might maintain its operational function as expected.
2008 to 2012, the downtime of the Phonska I Plant is shown For the success of RCM II implementation, the author
by the graph below [6]. attempts to define the finction and functional failure of
system at first, then to define the mode and effects of its
Scheduled & Unscheduled DT of failure which is commonly known as Failure Mode Effect
Phonska I PT Petrokimia Gresik 2008 - 2012 Analysis (FMEA). At the end, the author attempts to define
61.21
50
41
34.14
31.56
33
33
40
Phonska I. In order to reach the effective and efficient RCM
Days
20.38
24
16.26
15.72
30
22
450000
430000
429000
425914
423507
500000
353423
400000
operational context [3] [9].
Through the RCM application, each failure of
300000 an asset can be observed and traced by the interrelated seven
questions entailed by the RCM Process:
200000 - What are the functions and associated performance
100000
standards of the asset in its present operating context?
- How or in what ways does it fail to fulfill its functions?
0 - What are the modes or the causes of each functional
2008 2009 2010 2011 2012 failure?
Year - What will happen when each failure occurs?
Production (Target) Production (Actual)
- How or in what ways does each failure matter?
Figure 2 Target & Actual Annual Production of Phonska I PT Petrokimia - What can be done to predict or prevent each failure?
Gresik 2008-2012
- What should be done if a suitable proactive task cannot
From the Figure 2, it can be seen that the actual be identified?
production of Phonska I PT Petrokimia Gresik was always
JURNAL TEKNIK, (2013) 1-6 3
By implementing RCM II, the information of a production MTTF (when a component has reached its MTTF, then
unit can be summarized in the seven questions entailed by the value of its MTTR should be appeared in the
RCM Process. From that information, the evaluation of the column ‘Repair’ and the other components are not
function and functional failure can be executed so the repaired). In the stage 2, the formula of the ‘Start’ =
Failure Modes & Effects Analysis can be composed as well ‘Stop (Stage 1)’ + ‘Repair’ and the value of the ‘Repair’
to design the proposed maintenance task [3]. is filled by the same way as the stage 1. Each column of
the stage 3 until the end of the stage is calculated with
2) Reliability Distribution (Failure Rate Model) the same way as before,
Reliability Mathematical model starts by defining the 4. Checking the maintenance period for one year, and note
reliability quantitatively in terms of probability (PDF) and in which stage each component stops,
cumulative density function (CDF) for time-to-failure and 5. Adjusting the remaining MTTF by creating the
defining the failure rate and mean time to failure (MTTF). adjustment table below,
According to Lewis (1994), Reliability is the probability that 6. If any two or more components stop at the same time,
the system survives for a specified period. The specified which means that the reparation should be applied to
time or period may be expressed in terms of random those components, then the longest period of repair
variable x (time to failure) [5]. among those components will be used to update the
The time to failure of equipment is a random variable. In ‘Start’ column in the next stage,
advance of calculating the reliabilty of a machine or 7. The scheduling or staging will have to be discontinued
equipment, the failure distribution of each equipment need when all the components reach their MTTF at once at
to be unraveled statistically. There are two type of failure the same time because the pattern of this staging will be
rate models, where each type of failure rate models the identical to the first staging.
comprise various type of failure distributions. This method can be used to find the maintenance
The first failure rate model is Constant Failure Rate scheduling or interval for series component or parallel
Model. Constant Failure Rate Model is defined by the components. When the components are in the parallel
assumption that the rate at which the system fails is system, the algorithms above cannot be used. The
independent of its age. The constant failure rate algorithms above are used to find the scheduling
approximation is often quite adequate even though a system maintenance for series components of a machine. To find
or some of its components may exhibit moderate early the scheduling for parallel components for many machines,
failures or aging effects. The only one failure distribustion the algorithms number 3 to 7 are not applicable. If the
of the constant failure rate is the Exponential Distributions. components are parallel and they are independent each
Meanwhile, the another model of failure rate is Time- other, the reparation will be applied for each component
Dependent Failure Rate Model. The Time-Dependent independently. By definition, when a component fails and
Failure Rate Model is used when there are variety of the reparation is applied, the other components do not need
situations in which the explicit treatment of early failurer or to be stop. Consequently, the column of ‘Repair’ will be
aging effect, or both applied to the system. The time- filled with the MTTR of each component [8].
dependent falure rate model comprise the Normal Finding failure task is executed when the proactive task
Distributions, the Lognormal Distributions, and the Weibull cannot even be executed to reduce the multiple failure of the
Distributions [7]. related hidden failure up to the tolerated lower limit of
multiple failure. Finding failure task comprises the hidden
3) Scheduling and Maintenance Interval function checks to detect the failure of an equipment. This is
The calculation for obtaining the optimum maintenance known as the functional check [3].
interval can be found by using many methods and many According to Mobray (1997), proposed interval
formulas. The methods for calculating the maintenance for finding failure task is calculated as below:
interval has been developed by many reserachers and FFI = 2 × Utive × Mtive
approximation. The maintenance interval used to apply the Notes:
maintenance activity for the equipments or assets depends FFI : Failure Finding Interval
on the owner or company’s maintenance management. The Utive : Desired Unavailability of protective device
maintenance interval can be calculated by using the Mtive : MTBF of protective device
preventive maintenance scheduling and maintenance
interval for finding failure.
This maintenance scheduling method uses the data of B. Methodology
Mean Time To-Failure and data of the reparation activity or The methodology or steps of conducting this research are
Mean Time To-Repair of each components or machines. following this flowchart:
The algorithms of this method are below:
1. Sorting the components based on the MTTF from the
lowest to the highest,
2. Creating the column under the title ‘Start’, ‘Stop’, and
‘Repair’ in a row, as many as the operational hours
divided by the lowest MTTF (stage),
3. Formulating the relevant equation (excel function) of
each column in each stage. In the stage 1, the
formulation for ‘Stop’ column = ‘Start’ + MTTF; the
value of the ‘Start’ column is started from zero; the
value of the ‘Repair’ column depends on the lowest
JURNAL TEKNIK, (2013) 1-6 4
Functional Problems
Identification Function and Functional Failure
Phase Identification of Phonska 1
48% without Preventive Tasks
12%
Operational Problems with
Preventive Tasks
FMEA Construction
20% Operaional Problems
without Preventive Tasks
Designing Proposed
Maintenance Task
Utility Calculation and its From the diagram above, it can be seen that the downtime
PMT Designing
Phase Trade-off with Reliability of Phonska I Plant is most caused by the functional problem,
but in reality, the operational problems comes up frequently
Proposed Maintenance in the Phonska I Plant. Most of functional problems have the
Interval & Scheduling preventive tasks that can be done to prevent or reduce the
potential failure of the related equipment. Most of
Analysis and Data operational problems do not have the preventive tasks that
Interpretation can be done to prevent or reduce the potential failure. The
operational problems without preventive tasks are the most
Conclusion and
frequent problems that appear in the Phonska I Plant and
Recommendation
Conclusion &
causing downtime to Plant. This may cause the downtime
Recommendation End because most of operational problems are encountered by
Phase the important machines such as granulator, dryer, coater
drum and by material handling such as conveyor & elevator.
Figure 3 Flowchart of Research
degradation of reliability from time to time of these machines is still above 0.1 and so the utility is increased
machines is extremely fast as well. As example, the again to be + 50%.
reliability at time = MTTF = 199 hours of M109 is only The maintenance scenario by using utility + 50% results
0.273, which is very low. This phenomenon may happen in 23.9167 days of repair. The estimated production that can
because M109 is the most frequent machine that be reached by using this scenario is 420513.62 tons, which
encountered the failure during 2012. In the other side, the is more than the target production. At the utility + 50%, no
reliability of Q102 is still 0.4999 at time = MTTF = 825 machines have reached the reliability under 0.1, which
hours, which is still very high. This happens because during indicates that this scenario is still possible to do although the
2012, the Q102 rarely encountered the failure. The probability of failure is getting higher.
reliability characteristic of each machine is most affected by The maintenance scenario is not continued to the increasing
the frequency of failure of the machine itself during its 60% of utility, because at utility + 60%, some machines has
operational life or production. This reliability calculation reached the reliability under 0.1 or excess the maximum
will be base for trade-off together with utility of the plant. TTF that has ever been encountered over the years. That is
why; the maintenance-scheduling scenario by using this
preventive maintenance is not continued to utility 60%.
E. Maintenance Scheduling using Trade-Off
To know the comparison of machines’ reliability in each
The maintenance scheduling of all the 24 major machines maintenance scenario, the graphic in Figure 5 and 6 will
of Phonska I Plant is made according to the Conventional show this comparison. If the reliability of 24 machines is
Preventive Maintenance Method. The maintenance by using recapped in one graphic, the graphic will be so crowded and
this method is done with four scenarios, they are original it is divided into two graphics, the reliability graphic of
utility trade-off, utility + 20%, + 40%, and + 50%. The material handling machines and processing machines
trade-off is stopped when there is one, more machines reach
the maximum TTF, or in other words, the reliability of the
machine is below 0.1, which is very low and the probability Reliability of Processing Machines
0.6
of failure of this machine is very high. B101 Normal
The maintenance scheduling by using original MTTF C107 Normal
0.5
results in 34.2 days of repair and so the plant will stop
C108 Weibull 2
operating as long as 34.2 days if this maintenance-
scheduling scenario is implemented. The consequence of 0.4 C109 Normal
Reliability
using this scenario is the repair action is more frequent and M108 Normal
the target production cannot be reached. The production 0.3
M109 Lognormal
target of the Phonska I Plant in 2013 is 415000 tons. With
the capacity 450000 per year, the capacity per day will be M110 Normal
0.2
1232.88 tons. By using this scenario maintenance, the M117 Weibull 3
estimated production that can be reached is 407825.23 tons, 0.1 P103 Weibull 3
which is less than 415000 tons. By using the maintenance
P107AB Normal
scenario of utility +20%, the MTTF of the machines will
0
increase as well. The original utility is increased 20% first Q101AB Normal
and the result of this utility of each machine is then Q102 Normal
conformed into the time to find the exact value of the MTTF
at utility +20%. With this scenario, the Phonska I Plant will Utility
spend 29.33 days for repair days, which means that the
Phonska I Plant should stop operating as long as 29.33 days Figure 5 Reliability of Processing Machine in Each Utility Scenario
to execute the maintenance. Apparently, the Phonska I Plant
is still unable to meet its production target with the Reliability of Material Handling Machines
maintenance scenario of utility +20%. The estimated
0.6
production that can be reached by the Phonska I Plant by M101 Normal
using this scenario is 413835.52 tons, which is still less than M106 Normal
0.5
450000 tons. The original utility of all the machines is then
M111 Weibull 3
increased 40%. All the utility of machine that has been
increased 40% is then conformed to time to find the exact 0.4 M114 Normal
Reliability