Maintenance Bureau

Covaris Pty Ltd
www.covaris.com.au
Covaris White Paper - Maintenance Bureau ver 1-0.doc 1
Maintenance Reliability Bureau
Covaris Pty Ltd
PO Box 3456
Bankstown Square NSW 2200
Summary: This paper presents an objective
methodology for maintenance data analysis. The
purposes of these analyses are to identify strengths
and weaknesses in the maintenance management
system, opportunities for improvements, and
benchmark maintenance key elements against
maintenance best practice. A wide range of reports
can be provided from analysing the data from
computerised maintenance management systems.
These reports can be categorised as Performance
Reports, Benchmarking Reports, Optimisation
Reports, and Data Integrity Reports.
Failure Modes and Effect Analysis is one of the main
outcomes of the maintenance engineering bureau
services. This analysis is an easy to use and yet
powerful, proactive maintenance engineering method,
which can identify potential failure modes, determine
their effect on the maintenance costs, and identify
actions to mitigate the failures. The results of the
analyses of the various maintenance system data sets
based on the developed methodology are presented.
1.0 INTRODUCTION
The purpose of the Maintenance Engineering Bureau
Service is to provide a third party objective
maintenance data analysis for companies. This
analysis is expected to be performed by skilled
maintenance engineers on a regular basis, allowing
companies to effectively contract out their
maintenance data reporting, as well as gain the
benefit of benchmarking results against comparable
industries. The reports generated from the
information analysis fall into the following
categories.
Performance Reports: These are process specific
reports that identify process capability and
maintenance performance. Included in this set of
report will be the Maintenance Key Performance
Indicator’s (KPI’s) of a company; however they are
not the entire set of reports. These reports will often
point towards areas of process improvement
opportunities and may be quite specific. The best
presentation of data will require targets for measures
that have been agreed to by the business. These
targets will be obtained during an initial plant audit.
Benchmarking Reports: These reports for an
individual process or manufacturing facility may be
meaningless, however, when compared to other
plants in similar industry, they can show a
requirement for improvement.
Optimisation Reports: These reports differ from
Performance reports as these are used to manipulate
the data that drives the maintenance system. They are
designed to assist in optimising the maintenance
processes for changes in business requirements.
Included in these reports are those that are used to
increase the accuracy of current data (often in
conjunction with a manual review).
Data Integrity Reports: These reports are used to
judge the quality of data being captured. Information
found in the performance reports may only be
considered accurate if the system is being used in a
manner that ensures the data is reliable. These reports
point to not only whether data is being captured
completely (according to business requirements), but
also to assure that accurate data is being recorded (not
just defaults etc.).
It is envisaged for any particular company, large
amount of reports could be available for creation.
Failure Modes and Effect Analyses (FMEA) is the
main outcome of the bureau services. FMEA
presented in this paper is an easy to use and yet
powerful proactive maintenance engineering method.
FMEA is used to identify potential failure modes,
determine their effect on the maintenance costs, and
identify actions to mitigate the failures.
In general, FMEA is a disciplined approach used to
identify every possible failure mode of a process or
product and to determine its effect on other sub-items
and on the required function of the product or
process. The FMEA is also used to rank and prioritize
the possible causes of failures, determine the
frequency and impact of the failure as well as develop
and implement preventative actions. This analysis is a
costly and time-consuming exercise, which is
normally justified for high critical items or equipment
with serious failure consequences.
However, the FMEA presented in this paper is a
quick approach, which analyses the corrective
maintenance work orders from a maintenance system.
It captures work orders from the Computerised
Maintenance Management System (CMMS) and
allocates selected work orders to defined failure
modes. It also specifies, what equipment is affected
by a selected failure mode, what is the total cost of
the work, and what is the total amount of labour
hours spent on work for each particular failure mode.
This exercise is performed by analysing historic data
collected by the CMMS and includes all assets
captured in the system.
2.0 ELEMENTS OF MAINTENANCE
SYSTEM
Before analysing data from maintenance systems it is
important to understand the key elements of the
maintenance system. This section presents the key
elements of a well-structured maintenance
management system (2). The first element is a plant
dictionary that has all maintainable equipment listed
in hierarchical form. The next element is the
development of a comprehensive and efficient
maintenance procedure database, which should also
include safety procedures as they relate to
maintenance tasks. The third element is a master
schedule that ensures all registered equipment is
covered by an appropriate maintenance strategy. The
Covaris Pty Ltd
www.covaris.com.au
Covaris White Paper - Maintenance Bureau ver 1-0.doc 2
last element is monitoring maintenance management
system with a reporting system. All this information
is linked together in a Computerised Maintenance
Management System (CMMS). The CMMS will
allow efficient handling of the data and appropriate
planning of scheduled work. The CMMS will also
store equipment data that can be helpful in future
review of current maintenance strategies and help
identify which items can benefit the organization if
further improvements are made to the maintenance
system.
2.1 Plant Dictionary
The level of detail that is necessary for adequate
maintenance system design is quite extensive. This is
often a major problem to the improvement team
given the poor state of documentation and local
knowledge regarding some machines, despite their
criticality to the process.
We recommend the plant dictionary to be structured
according to a parent-child hierarchy. The lowest
level in the hierarchy is a maintainable item that
needs to be checked or inspected – a PM work order
is not usually provided for an item at this level but is
more normally associated with a singular task such as
a corrective maintenance work order or a breakdown.
The second tier is used to logically identify a group
of maintainable items – the PM work order covering
one or more inspections is assigned to an entity at this
level, which normally identifies the equipment types
(e.g. boiler). The top tier is a multiple layer of levels,
which identifies a department.
The plant dictionary includes important equipment
information such as manufacturer, date of supply,
serial number, criticality, equipment and location
identifiers.
2.2 Maintenance Procedures and Routines
This section is concerned with the identification and
specification of Maintenance Strategies. These are
procedures that specify a range of tasks for a specific
equipment type. A maintenance strategy for an
equipment type is a set of procedures, which covers
all preventive maintenance activities for that asset
base. These strategies will be subsequently allocated
to many incidents of the same type of equipment
across the site.
Assigning a procedure to a specific item of
equipment creates a routine. Specifying the timing of
the first release of the routine as a work order is an
important issue to obtain a balanced workload during
the year and avoid the backlog built-up during the
busy period of the year for a particular site. Adjusting
the duration of the routine is required to allow for
either travel or access to the equipment.
2.3 Master Maintenance Schedule
Preventive maintenance work needs to be planned
which sets out the generation of preventive works.
Specifying the timing of the first issue of a routine as
a work order is accomplished by Master Maintenance
Schedule (MMS) worksheets.
In developing the MMS, scheduler defines the first
release of each routine, which will automatically
allocates the subsequent dates when the routine is
due. The master PM scheduler provides a forecast of
all work-orders for the planned PMs. The scheduler
can then check the workload for each trade or the
total workload for the preventive maintenance for the
entire site or a particular department. These will help
the scheduler to examine the effectiveness of the
scheduling process and to avoid congestion in any
particular week, during the busy period of the year or
for a specific work group.
2.4 Reporting Key Performance Indicators
A healthy maintenance management system should
always be monitored with appropriate reporting
system. Key performance indicators (KPIs) form the
basis of the monitoring process for performancebased
maintenance. Use of various KPIs will be
dependant on the capability of the information system
to provide accurate data for calculation purposes.
Most CMMSs are capable of capturing most of the
information necessary for these KPIs. Others should
be developed in collaboration with production
departments and require more involvement from the
management to become alive. The remaining sections
of this paper provide a wide variety of reports which
can be set up for monitoring the performance of a
maintenance management system.
3.0 FAILURE MODES AND EFFECT
ANALYSES
FMEA is an effective engineering tool for analysing
data from maintenance systems. The FMEA is used
to identify failure modes of the maintainable
equipment and the effects of these failures. FMEA
analysis can be used to specify the areas where
improvements can have substantial impact.
Traditional FMEA as explained in MIL-STD-1629A
identifies maintenance plan analysis as an application
for FMEA. Other applications identifies by this
Standard include “The FMEA shall also be used to
define special test considerations, quality inspection
points, preventive maintenance actions… ”.
However, the FMEA presented in this paper is a
quick approach, which analyses the corrective
maintenance work orders from a maintenance system
as explained in this section.
MaintSpeak has been developed by Covaris to
perform FMEA analysis. The FMEA analysis of the
corrective maintenance work orders from a
maintenance system (independent of CMMS in use)
has the following methodology:
• A first cut of work orders is allocated to a set of
primary failure modes derived on inspection
from the work orders – hence this first cut
process achieves the following:
Covaris Pty Ltd
www.covaris.com.au
Covaris White Paper - Maintenance Bureau ver 1-0.doc 3
o Specifies the primary failure modes
o Makes some bulk allocations of work
orders to these failure modes
o Identifies that portion of the work
order data set which will not be
allocated to failure modes since the
work is more to do with housekeeping,
etc
• A second, finer cut is made where a primary
failure mode with a very large grouping of work
orders is broken out into a number of more
specific failure modes, and the accuracy of the
work order allocation is improved by a more
detailed view of each registered failure mode (in
the first cut, the thousands of work orders were
viewed and considered, when making the second
cut the tens to hundreds of work orders allocated
to a failure mode were considered)
• Each primary failure mode is then considered in
detail and a set of very specific “secondary
failure modes” is derived. The logic is that a PM
work order is intended to address the failure
mode and the operations within a PM are
intended to address each of the secondary failure
modes.
• The list of work orders allocated to each failure
mode is then considered, and the effects of the
failure mode are specified by:
o The range of equipment affected by
the failure mode – as determined by
the equipment associated with work
orders allocated to the failure mode
o The sum cost of work orders allocated
to a failure mode - the software also
handles the sum amount of labour
hours.
o The Mean Time To Failures MTTF for
each failure mode – this has been
derived from a Weibull analysis of the
work order data set allocated to each
failure mode.
3.1 FMEA Case Study
The case study presented in this section is the
application of the FMEA analysis explained in this
paper for analysing data from maintenance system of
a food factory.
The failure modes detected for the factory were based
on a study of 21,624 corrective work orders with
17,148 allocated to a failure mode, noting that the
General failure mode which is used as a catch-all
contained 1,068 work orders. Hence 16,080 work
orders were attributed to useful failure modes, which
represent 74% of the data. It would be preferred to
see over 80% so attributed but found it necessary to
dump numbers of work orders into the General
category since they represented highly infrequent
failure modes (i.e. less than 5 work orders for the
failure mode over the period of four years).
Failure modes and their impact on the plant are
tabulated below. A plot of the top 10 failure modes is
shown in Figure 1. N Equip refers to the number of
equipment which are affected by the relevant failure
mode.
0
200
400
600
800
1000
1200
1400
1600
CONVEYOR BELT
GENERAL
DRIVE
BEARING
ROLLER
AIR
PUMP
CHAIN DRIVE
GLUE
VALVE
No WO, N Equip
0
100000
200000
300000
400000
500000
600000
700000
Total Cost $
N work orders N Equip Total Cost
Figure 1 Top 10 Primary Failure Modes
It can be seen that failure modes associated with
rotating items (i.e. drives, bearings, rollers, pumps,
chain drives and drive belts) dominate the list of top
10 failure modes. We also note that conveyor belt
issues tend to be the most dominant failure mode in
most food factories that we have studied so far. In
this instance, 346 conveyors are affected by this
failure mode. The secondary failure modes are
represented by the following work order work
statements are shown in the table below
CONVEYOR BELT
0 Repair infeed conveyor
1 Repair broken flight lug
2 Repair broken flat belt on grister mill
3 Repair belt joins
4 Repair tensioning mechanism
5 Eliminate spills at sides of belt
6 Conveyor belt keeps tracking off
7 Torn belt
8 Bars catching on crossover conveyor
9 Conveyor gate not operating properly
10 Conveyors out of alignment
11 Conveyor jammed reset o/load
3.2 Weibull Analysis
Weibull analysis is used to test the shape factors and
MTTF of various primary failure modes. The rule of
thumb for Weibull results is as follows:
• Beta values (shape factors) of less than 1 are
very rare and would represent a gross design
problem more than anything else. This would be
where an item was continually being replaced
almost as soon as it entered service.
• Beta values of between 1.3 and 1.7 indicate one
of two possibilities: there are multiple failure
modes in the category or more likely, there is a
manageable driver on the defect rate such as
maintenance quality of work or operational
cleanliness.
• Beta values between 1.7 and 2.2 represent
achieving a respectable life but there is scope for
improvement in the PM strategy.
• Beta values greater than 2.5 mean that the PMs
are doing their job and there is no operational
driver to early failure such as cleanliness or
overload.
Covaris Pty Ltd
www.covaris.com.au
Covaris White Paper - Maintenance Bureau ver 1-0.doc 4
These descriptions would not be found in the
textbooks but represent our experience in analysing
the trends in maintenance work orders.
The Weibull plots all tend to look the same but the
key characteristics are:
• The shape factor
• The x intercept – which is the likely minimum
time to failure
The MTTF does not particularly help us since we are
taking a lot of data over the period of four year, and
hence this tends to remain between 150 to 220 days
for just about all failure modes. We are interested to
find out if the failure is being driven by something we
can manage (ie the shape factor), and whether we are
going to see a potential failure every day, every week
or every month. Since we are dealing with high
frequency failure modes in this section (ie the top
10), their potential to occur will be at least once a
week. A typical plot is shown below.
Figure 2 Weibull Plot for the First Failure Mode
This result has a low shape factor (<2) indicating that
it is not a wear out condition – we expect there are
contributing factors such as past maintenance and/or
operational cleanliness driving this failure mode.
Minimum time to failure is about twice a week when
the intercept with the x-axis is considered.
3.3 PM Schedule Analysis
This section, however, presents a systematic
approach in evaluating maintenance performance,
and some types of reports that can be expected for the
bureau service to provide are presented. This is not
the entire suite of reports, but is a large enough
sample to demonstrate the type of required
information that will need to be processed.
Proper scheduling of works is a common issue in
almost all maintenance departments. To check the
PM scheduling program, we generate a forecast of the
whole PM schedules for the next 12 months. The
number of work orders and the workload will be
calculated against the weeks. The report will show
that how effective PM schedules have been setup. If
the works load (Number of work orders if the work
load is not available) is not distributed evenly over
the weeks, the site needs to improve its PM
scheduling process.
0
50
100
150
200
250
W30W31W32W33W34W35W36W37W38W39W40W41W42W43W44W45W46W47W48W49W50W51W52 W1 W2 W3 W4 W5 W6 W7 W8 W9 W10 W11W12W13W14W15W16W17W18W19W20W21W22W23W24W25W26W27W28W29
Forcast PM work Orders W30 (2003) to W29(2004)
No of Work Orders
Works Load (hrs)
Figure 3 Forecast PM Work Order for One year
Period.
4.0 CONCLUSIONS
This paper presents a methodology to analyse
maintenance data sets form a broad range of
computerised maintenance management systems. The
purposes of these analyses are to identify strengths
and weaknesses in the maintenance management
system and opportunities for improvements. The
developed methodology to perform failure modes and
effect analysis was described.
Easy to use and yet powerful software have been
developed to facilitate FMEA. It captures work
orders from the CMMS and allocates selected work
to defined failure modes. It also specifies which
equipment is affected by a selected failure mode,
what is the total cost of the work and what is the total
amount of labour hours spent on work in this mode.
Weibull analysis is also used to test the shape factors
and MTTF of the various failure modes.
The results from various case studies, showing
outcomes from FMEA and Weibull analysis and wide
range of reports, which can be provided from
analysing the computerised maintenance management
system data sets were presented.