TOWARDS EFFECTIVE AND RESPONSIVE MANAGEMENT OF PLANT
Bikash Bhadury Ali Türkyilmaz Mehmet Sevkli
Fatih University, Faculty of Engineering, Department of Industrial Engineering,
Buyukcekmece, 34900, Istanbul, Turkey
An effective and responsive maintenance management system- one that ensures that the right amount of the
necessary resources are available at the right time for both preventive and corrective maintenance tasks- is not
available. The statistical inventory control models and selective approaches which are used for spare parts
planning, are based on the assumption that the demand for spare parts is independent in nature. In reality, however,
the demand for maintenance resources is dependant on the processes of degradation and failures of the constituent
parts of the equipment. It follows therefore that the MRP framework can, and should, be used for management of
maintenance resources. However, for this, the dependence of demand for maintenance resources has to be
established and the component and sub-system failures and other maintenance activities that generate demand for
maintenance resources have to predicted with a fair degree of confidence. In this paper, an MRP based maintenance
resources management has been presented, and explained through a real-life case application in a thermal power
plant. It is further suggested that the proposed maintenance resources management system be incorporated in the
plant maintenance (PM) module of the ERP system, such as SAP R/3.
Although maintenance is an important function, particularly in capital intensive and continuous process
industries, and maintenance activity is dependent on the availability of necessary resources, the management of
maintenance resources is still carried out in an insufficient and seemingly arbitrary manner. Spare parts, skilled
manpower, tools, tackles and instruments, and sometimes even money have all to be available concurrently for the
performance of the maintenance task. Whereas it is obvious that an effective and responsive management system –
one that ensures that the right amount of the necessary resources are available at the right time for both preventive
and corrective maintenance tasks- is indispensable, such a technique or a model is not available. Statistical
inventory control models and selective approaches for control of multi-item inventories, such as ABC, VED and
FSN etc. and combinations there of have been used for spare parts planning. These reorder point techniques are
based on the assumption that the demand for spare parts is independent in nature, whereas, in reality, the demand for
maintenance resources is dependent on the processes of degradation and failures of the various parts, which make up
the equipment. Manufacturing resources planning (MRP) is a planning and control technique for dependent demand
inventory items. It follows, therefore, the MRP framework can, and should, be used for planning and control of
maintenance resources, such as spare parts, maintenance manpower, and tools and instruments.
Accordingly the dependence of demand for maintenance resources can be established and the failures and
other maintenance activities that generate demand for maintenance resources can be predicted with a fair degree of
confidence, the MRP framework can be used for the management of maintenance resources. Such a system will be
much more effective and also responsive to needs and actual conditions.
2. Components and Logic of MRP-Based Maintenance Resources Management System
An MRP-based maintenance resources management system should essentially have the following core
1. A master maintenance schedule
2. Bills of maintenance (maintenance of materials) - one for corrective maintenance and the others for
Routine preventive maintenance
data/ preventive maintenance Historical failure data
Reschedule maintenance maintenance
Bills of MRSRP* Maintenance
maintenance processor inventory
Figure 1 Flow chart of an MRP-based maintenance resources management system.
* Acronym for man resources and spares requirements planning.
3. A maintenance inventory status file, carrying information about stockable parts, manpower skills, facilities
and tools inventory.
4. A logic processor.
5. A capacity planning system.
6. A module for printing reports.
The interaction of the components of an MRP-based maintenance resources management system is shown
in Fig.1. The proposed model starts with the master maintenance scheduling task. Certain parts of the equipment
need to be replaced after a fixed number of hours of use. Such information is normally available from the equipment
manufacturer, and in cases where manufacturer’s recommendations are either not available or not specific, data
based on users experience on same/similar equipment can be used. MMS is meant for parts which fall under the
routine preventive maintenance category. The number of parts of each type to be replaced, the time required to carry
out the action, and the fixed hours after which this action has to be performed are essentially the inputs to this
constituent of the model.
Functionally, the bill of maintenance is analog onto the bill of materials (BOM) in MRP. The proposed
model calls for two bills –one for preventive maintenance and the other for corrective maintenance. The principal
function of the two bills is to relate all the preventive and corrective maintenance activities to the required resources.
The maintenance bill of ma terials (MBOM) is a systematic division of the end item into systems. These systems are
further divided into sub-systems. The division continues until a particular level beyond which further division is
infeasible. We refer to this level as the component l vel. At this level, failure data is normally available. Close
observation of failure data enables us to establish the critical failure modes of the component. This data can also be
used to determine the probability distributions and the established
Parame ters that govern the failure and repair processes of each of the failure modes of the component. Also
once the component level is reached and the failure modes are established, a list of stockable parts (necessary
resources ) needs to be attached to the particular equipment level. This completes the MBOM structure.
Corrective maintenance module(CMM)makes use of the MBOM and the information generated therein
cars(discussed above) to predict the occurrence of failure and the completion of repairs The maintenance resource
requirements generated by MMS and CMM are then added together to give the gross requirements. This procedure is
carried out for all the periods over the planning horizon. The lot sizing policies can then be applied to determine the
optimal lot sizes for the parts to be procured. The skills requirement predicted by the system can be used to schedule
The prepared model does not make any assumptions regarding the failure and repair time distributions. The
times to failure of components (for defined failure models) can be modeled using any appropriate distribution that
characterizes the failure the models predicts the occurrence of failures that the maintenance resources can be
procured earlier to avoid stockouts . This enables proactive planning of maintenance resources. The following section
describes the application of the prepared model to a reel life case.
3. Case Application
The model was applied to the fuel system of a 210 MW thermal power unit. The fuel system included air
pulverizers five of which needed to be in operation to generate at full capacity. The six was taken as a standby. On
failure of a pulverizer, the standby was switched on, and repair action was initiated on the failed pulverizer
depending on the availability of the maintenance resources. Failure data of the pulverizers were analyzed, and this
established several critical failure models. Description of the failure model is given Table 1. Five repair garga
operated during the daytime (i.e shifts I and II) while three repair garga operated during night shift (i.e shift III).
Routine preventive maintenance action was carried out on the pulverizers after every 500 hours of continuous
operation. It has also observed that this action consumed O.S how.
A simulation program was developed that incorporated the proposed model. The program was coded in
SLAM II and was supported with some FORTRAN 77 subroutine (Shenoy and Bhadury, 1993 and 1998). Fig. 2
shows the flow chart of the simulation program. The values of this random variable which were intended to cover
95% of the requirements (p: 0.95) are given in Table 2.
Failure mode cause Prob. Failure Dist Parameter Repair Dist. Parameters
1.Mill Failure a. Ring breakage 0,08 Exponential 3415.90 Constant 480.0 --
b. Ball breakage 0,08 Log-Normal 0 37.00
c. Stirrup bolt failure 0,08 Log-Normal 94.00 5.1
d. Foreign matter in 0,76 Log-Normal 14.85 4.3
2.Feeder a. Leakage 0,23 Exponential 733.08 Log-Normal 9.41 7.0
b. Trouble 0,23 Log-Normal 17.75 13.3
c .Jamming 0,20 Log-Normal 4.92 3.2
d. Drag link sheared 0,15 Log-Normal 46.3 24.5
e. Gear box coupling 0,13 Log-Normal 20.31 16.7
sheared 8.8 4.7
f. Rod gate leakage 0,06 Log-Normal
3.Coal carrying pipe Exponential 625.00 Log-Normal 5.94 3.6
4.Coal dust leakage Exponential 756.00 Log-Normal 18.6 5.8
5.Loading cylinder Exponential 747.00 Log-Normal 5.4 4.3
6.Pyrite Gate a. Split pin sheared 0.10 Exponential 556.00 Log-Normal 15.0 12.0
b. Gasket to be 0.80 7.2 4.0
replaced 0.10 5.7 4.0
7.Burner Trouble Exponential 1012.00 Log-Normal 5.0 2.0
Table 1 Description of the failure models.
Spare Period Number
part No: 1 2 3 4 5 6 7 8 9 10 11 12
1 0 0 0 0 0 0 0 0 0 0 0 0
2 0 0 0 0 0 0 0 0 0 0 0 0
3 0 0 0 0 0 0 0 0 0 0 0 0
4 3 3 3 3 3 3 2 2 3 2 2 2
5 1 2 1 2 2 1 1 1 1 1 1 1
6 1 1 1 1 1 1 1 1 1 1 1 1
7 1 1 1 1 1 1 1 1 1 1 1 1
8 9 9 9 8 8 8 8 8 7 7 7 7
9 7 8 7 7 7 7 7 7 6 6 6 6
10 8 8 8 7 7 7 7 6 7 7 6 6
11 2 2 2 2 2 2 2 2 2 2 2 1
12 9 8 8 8 7 7 8 7 7 7 7 6
13 6 6 6 6 5 6 5 5 5 5 5 5
14 8 8 8 7 7 8 8 7 7 7 7 7
Table 2 Demand Prediction for spare parts (p: 0.95)
As can be seen from the table the demand predictions for demands for the first three spare parts. i.e. grinding rings,
grinding balls and the stirrup bolts, were very low and these parts were expensive too. The best way to control these
type of items is to make use of the (S-I, S) policy, in which an order for the part is placed whenever a demand occurs
thereby keeping the stock level constant. The lead times for these items were taken from the uniform distribution
with a lower bound of 4 months and an upper bound of 5 months, i.e.U (4, 5). For the rest of the items under study, a
lot-for-lot(L-F-L) ordering policy was used with a lead time of U(2,3).
The following steps were employed to determine the optimal configuration of the safety stocks for rest of the items
Increment time print report
START INTLC STATE EVENT(I)
Inizalite First time Record
simulation to failure Failures
Repair Times END OF
Figure 2 Flow chart for the simulation program.
Step 1: Start with an initial configuration of safety stock, i.e
Spare Part 1 2 3 4 5 6 7 8 9 10 11 12 13 14
Safety Stock 1 1 1 0 0 0 0 0 0 0 0 0 0 0
Step 2: Simulate the system ‘n’ number of times. (The parameter ‘n’ is to be determined using central limit theorem
for a given degree of confidence).
Step 3: During the process if a pulveriser is down for lack of a spare part, a penalty cost equivalent to the time fir
which the part is not in the store is credited against it. Else at the end of any period a holding cost is charged on the
average inventory carried over the period. These costs are accumulated over all ‘n’ runs in the routines PENALTY
and STATE respectively.
Step 4: After ’n’ simulation runs, determine the average penalty and holding cost against all the parts.
Step 5: if the average penalty cost for any part is more than its average holding cost, increment the safety stock for
that part by one unit and obtain new configuration is the optimal configuration for the spare parts.
Step 6: Repeat Steps 2 though 5 and stop when the average penalty cost for all the parts fall below the corresponding
holding costs. The final configuration is the optimal configuration for the spare parts.
Table III shows the optimal configuration for the safety stocks determined using the steps described above.
The model was also used to study the effect of increase or decrease in the number of repair gangs operating in the
system. The first in first out (FIFO) repair policy was utilized. Statistics on the waiting time for pulverisers and the
average number of pulverisers waiting for repair were collected over the 100 simulation runs. After every batch of
100 runs, the number of repair gangs deployed in the system was varied and the response obtained from it.
4. Incorporation of Proposed Model In PM Module Of ERP Package.
The plant maintenance (PM) module of an ERP system, any SAP R/3 Release 3X, console of a large master
data file and package for service management, preventive maintenance, and maintenance order management, are
shown in figure 3. The service management package of the module deals with service contracts and agreements,
including billing, cancellations and warranty processing. Since processing of the service contracts and agreements
that the company enters into with its customers is not covered by the proposed model, since it deals only with the
maintenance-both preventive and corrective- of the company’s own production equipment, the service management
package should quite obviously be retained. However, the preventive maintenance and maintenance management
packages of the PM module can be replaced by the proposed model. Some alternations will be necessary in the
organization.(structure and coding) of the master maintenance schedule (MMS) and the two bills of the maintenance
(two MBOMs) of the proposed model to make them compatible with the master data files of the PM module.
Figure 3 Plant Maintenance (PM) Application Module of an ERP system.
1 Dinesh Shenoy K. and Bikash Bahadury,(1993), “MRSRP- A Tool For Manpower Resources And Spares
Reqruirements Planning”, Computer and Industrial Engineering, Vol. 24, No.3,pp 421-430
2 Dinesh Shenoy K. and Bikash Bahadury,(1998), “Maintenance resources managemenet:Adopting MRP,
Taylor&Francis Ltd. London EC4 3DE, U.K
Notes From SAP R/3 Release 3X