Step 6 - Maintenance Policy Determination

Modern maintenance philosophy stems from the premise that successful maintenance programs have more to do with the consequences of failures than the asset itself.

In this step, each failure mode is analysed using Reliability Centred Maintenance (RCM) principles. This step establishes new or revised maintenance policies. During this step the following become evident:

• The elements of the current maintenance program that are cost effective, and those that are not (and need to be eliminated),
• What tasks would be more effective and less costly if they were condition based rather than overhaul based and vice versa,
• What tasks serve no purpose and need to be removed from the program,
• What tasks would be more effective if they were done at different frequencies,
• What failures would be better managed by using simpler or more advanced technology,
• What data should be collected to be able to predict equipment life more accurately, and
• What defects should be eliminated by root cause analysis.

Figure 9 illustrates Step 6.

Step 7 - Grouping and Review

Once task analysis has been completed, the team establishes the most efficient and effective method for managing the maintenance of the asset given local production factors and other constraints. In this step it is likely that tasks will be transferred between trades and operations people for efficiency and productivity gains.

Step 8 - Approval and Implementation

In Step 8, the analysis is communicated to local stakeholders for review and comment. The group often does this via a presentation and an automatic report generated from the PM Optimisation software. This software details all the changes and the justification for each.

Following approval, the most important aspect of PMO 2000 then commences with implementation. Implementation is the step that is most time consuming and most likely to face difficulties. Strong leadership and attention to detail are required to be successful in this step. The difficulty of this step increases markedly with more shifts and also with organisations that have not experienced much change.

Step 9 - Living Program

Through Steps 1 to 9, the PM Optimisation process has established a framework of rational and cost effective PM. In the "Living Program", the PM program is consolidated and the plant is brought under control. This occurs as reactive maintenance is replaced by planned maintenance. From this point improvement can be easily accelerated as resources are freed to focus on plant design defects or inherent operational limitations.

During this step, several vital processes for the efficient management of assets can be devised or fine tuned as the rate of improvement accelerates.

• These processes include the following:
• Production / maintenance strategy,
• Performance measurement,
• Failure history reporting and defect elimination,
• Planning and scheduling,
• Spares assessing, and
• Workshop and maintenance practices.

In this step it is the intention to create an organisation that constantly seeks to improve its methods by continued appraisal of every task it undertakes and every unplanned failure that occurs. To achieve this requires a program where the workforce is adequately trained in analysis techniques and is encouraged to change practices to improve their own job satisfaction and to reduce the reduce the unit cost of production.

Implementing a Successful PM Optimisation Program

Selling Maintenance as a Process rather than a Department

Change programs are not easy to implement particularly when an organisation has entered the vicious circle of maintenance.

The author's experience is that in most cases there needs to be some fundamental shifts in behaviour and motives at all levels across the organisation. This almost invariably involves modifying the behaviour and decision-making priorities for middle managers too. Above all, there needs to be a commitment to the long term and if there needs to be some short-term losses, then these will often be well worth it if returns can be generated quickly from the investment in the future.

The most important aspects of managing a PM Optimisation change program to break the vicious cycle, are described in the following paragraphs:

Choose projects that do not focus solely on one aspect.

There needs to be a combination of projects that are likely to result in:

• increased uptime, and
• reduced labour requirements.

In many cases, this means tackling reliability problems in the process bottleneck as well as looking at maintenance intensive item categories 7 that are prolific on site.

The reasons behind tackling projects that carry labour productivity rather than machine uptime is
that

• First line supervisors will contribute to the program if they see that there is a return on investment in labour terms. A target of returning five days labour pa for every one invested should be the lowest acceptable limit.
• Returns on labour productivity are compounding (they can be reinvested in more productivity) whereas uptime improvements are finite.
• Collect data about the before and after case

Collecting data about plant reliability achieves many things. The two major benefits are as follows:

• Steering the analysis in the area of opportunity, and
• Providing the basis for the project teams to demonstrate the value of the work that has been done.

Create cross functional teams from the shop floor

PM Optimisation is not a back office function of statistical perfection. It is an empirically based process of considering preventive maintenance options, and task rationalisation. Involving the people who will be left to do the work is constructive in gaining commitment to make the changes happen. Leaving them out of the analysis creates barriers to implementation.

Integrate operations and maintenance work management systems

In redistributing the workload, it is important that the various systems of maintenance scheduling come from the same origin or database. In most organisations this is not the case as the operations department has a system that works in isolation from the trades groups.

Implement outcomes as quickly as possible

There is a temptation to celebrate project success after the analysis and move on to new projects leaving implementation to drift and become poorly managed. This is a very bad outcome because the project has consumed scarce resources and wasted them. Without successful implementation, the work has created a cost and the workforce expectations have not been met. The workforce will correctly blame the middle management and participation in future projects will be more difficult to obtain.

⁷ Ie; fans gearboxes, conveyors, machine tooling, pumps, electric motors, instrumentation - where there could be hundreds of similar items. Saving labour on one item multiplies across the whole of the site.

Dysfunctional Organisational Structures

The organisational structure of most capital intensive industries can be described as being departmentalised with maintenance and operations having separate budgets, performance measures, and management structures. There are advantages associated with departmentalised organisational structures; however, such structures often lose efficiency through:

• Conflicting goals and objectives of each department that sometimes result in decisions which are not congruent with the overall business goals. The most common being short-term production goals that often clash with the maintenance objective of reducing the overall cost of maintenance.
• Duplication of effort with many departments attempting to achieve the same result but in isolation of each other. Electrical, mechanical and production PM schedules commonly fall into this category with each department checking the same machine for the same failure modes.
• An overly bureaucratic decision making and approval process at all levels. This is often a result of conflicting objectives between managers.
• Excessive demarcation of roles and responsibilities. Though becoming less prevalent, the inability to take responsibility for certain work due to past traditions prevents efficient use of resources at many sites.
• A proliferation of independent systems and databases. The most common instance is where operations and maintenance personnel work through their own logbooks and records that are kept independently of the CMMS.
• The process of defect elimination is seen largely as an engineering pursuit where problems often have multiple contributing factors and must be solved by cross functional teams. Many factors are not necessarily obvious and many are due to shop floor people taking practical measures to combat other problems that have secondary effects elsewhere.

Conclusion

There are a number of contributing factors to the difficulties faced by the modern asset manager. To be effective at making changes to the performance of a maintenance function, the asset manager should understand how these factors have arisen, how they impact on the business performance and how they can be effectively tackled. There is a way out of the vicious cycle of maintenance and the Optimisation or Rationalisation of Maintenance tasks is a fundamental strategy in this process.

To break the vicious cycle of maintenance, asset managers should focus on the areas of preventive maintenance and defect elimination. To improve their preventive maintenanceorganisations, there must be a shift to an environment where there is no duplication of PM effort,every PM task serves a purpose, all PM tasks are completed at the right interval, and with theright mix of condition based maintenance and overhaul.

Organisations that have mature PM programs and are struggling to complete them should take steps to rationalise what they have, rather than embarking on a "green field" approach such as the traditional approaches to RCM.

There are many statistically based maintenance analysis tools available, however, users should be careful in their choice. They should be mindful that they could be spending a lot of money on expensive packages and consuming a lot of time collecting vague data which, after years of effort, produces a meaningless result.

Comparing PMO and RCM Methods of Maintenance Analysis

Methods of defining Initial Maintenance Requirements
The common methods for defining the initial maintenance requirements for plant and equipment
are as follows:
RCM,
Streamlined RCM,
Statistical Methods, or
Experience, trial and error.
The origin of the processes is briefly discussed below.
RCM
Nolan and Heap (1978) coined the term Reliability Centred Maintenance (RCM) and developed
the original method. RCM was not designed for use for "in service" assets. However, in the
absence of better methods since 1978, it has been applied retrospectively in many organisations
after the plant has been commissioned. In over 20 year since its derivation, RCM has failed to
become a day to day activity performed by most organisations. Few organisations have applied
RCM to anything other than their most critical assets which indicates there are serious difficulties
associated with applying RCM in organisations with mature plant.
Streamlined RCM
Due to a perception that RCM was a very time consuming and labour intensive activity, a number
of shortened versions of RCM have been devised in an attempt to speed up the analysis or
increase the overall value of the time committed to analysis. Many of these methods have used
the acronym, RCM to describe the process but do not conform to the works of Nolan and Heap
(1978) nor the SAE Standard for RCM. These streamlined approaches are known as streamlined
RCM techniques.
Statistical Methods
There are three main types of statistical maintenance analysis programs known to the author.
1. One of these is based on MILSTD 2173 and works from the premise that no inspection task is
100% effective. The algorithms adjust the interval of "on condition" tasks to account for less
than perfect inspection methods.
2. Another is based on the notion that the more frequent the inspection the higher the cost of
maintenance but the lower the chances of failure. The objective of maintenance under this
algorithm is to determine the lowest overall cost of maintenance. This algorithm is flawed if
inspection is near 100% reliable or is fail safe8 as, providing the inspection is inside the PF
8 Any incorrect sample will suggest there is a failure when in fact there is not. Oil analysis or
vibration analysis are examples where most of the problems are fail-safe.
SECTION 2. COMPARING PMO AND RCM METHODS OF MAINTENANCE ANALYSIS
OMCS Ph 0419397035 www.pmoptimisation.com Page 19
interval, more inspections only add to the cost of maintenance but not reduce the chances of
failure.
3. The third statistical method has uses to Weibull analysis as a basis. This method suffers
mostly from poor data integrity.
The overwhelming problem with statistical methods in the vast majority of industrial plant is that
the failure history data is so unreliable and incomplete that any statistical inferences drawn from
such data are wildly inaccurate and lack any worthwhile statistical confidence. The algorithms
are also reliant on accounting inputs such as the cost of PM, repair and failure. All of these inputs
are subject to the vagaries of the accounting systems deployed.
The second large problem is that statistical methods tend to be used by engineers or contractors
who are not sufficiently familiar with the equipment an the manner that it is used on site. Often
the result is a misguided program which is totally discredited by the tradesmen and operators
because of its low quality and secondly because they were not sufficiently involved in its
derivation.
Some explanation of the first two methods is contained at Section 3 of this paper.
Experience, Trial and Error
In many cases, capital acquisition programs fail to recognise the need to define the maintenance
program prior to the "Operation" stage of the equipment life cycle. Often, the plant is installed
and operated without a formal maintenance program. Over time, the operations and
maintenance staff begin to conduct inspections and perform various maintenance activities
largely at their own initiative. Failures occur and the maintenance program has tasks added to it.
In some organisations, the work is formalised by generating electronic or paper based
maintenance schedules. In other organisations, the work continues to be done in a completely
informal manner. Even though some managers may believe that there is no preventive
maintenance done within their plant, this situation is highly unlikely. The confusion is often that
the preventive maintenance is not appreciated, as there is no documentation.
Methods of Reviewing Maintenance Requirements
PM Optimisation
Regardless of how a maintenance program has been developed, there is a constant need to
review and update the program based on failure history, changing operating circumstances and
the advent of new predictive maintenance technologies. The generic process used to perform
such analyses is known as PM Optimisation (PMO). PMO has been performed, no doubt, since
the world became mechanised and humans realised the benefits of performing preventive
maintenance. PMO as a technique has been refined to reflect the RCM decision logic since the
formulation of RCM in 1978.
There are a number of methods that have been created under the acronym PMO. One of these
has been applied in the US Nuclear power industry for over 8 years and has been recognised as
a major benefit by the North American Nuclear Regulatory Commission (Johnson 1995).
Each of the PMO methods has differences and there is no accepted standard for PMO.
Discussions contained in this paper are therefore, based on the method of PMO known as
PMO2000. Some of the comments and comparisons made between PMO and other methods
may not apply to methods of PMO.
SECTION 2. COMPARING PMO AND RCM METHODS OF MAINTENANCE ANALYSIS
OMCS Ph 0419397035 www.pmoptimisation.com Page 20
The PMO2000 process has been developed over a five-year period by OMCS with the assistance
of several Australian Companies. There are now 12 users of PMO2000 in the Australia Pacific
Region. The PMO2000 process is endorsed by SIRF Roundtables Ltd9 and is the global
maintenance analysis tool of choice for one of the world's largest mining companies. PMO2000
is the intellectual property of OMCS. The methodology is described in detail at Section 1.
Comparing RCM and PMO
What is RCM
According to the standard SAEJA1011, any RCM program should ensure that all of the following
seven questions are answered satisfactorily and are answered in the sequence shown:
1. What are the functions and associated desired standards of performance of the asset in its
present operating context (functions)?
2. In what ways can it fail to fulfil its functions (functional failures)?
3. What causes each functional failure (failure modes)?
4. What happens when each failure occurs (failure effects)?
5. In what way dose each failure matter (failure consequences)?
6. What should be done to predict or prevent each failure (proactive tasks and task intervals)?
7. What should be done if a suitable proactive task cannot be found (default actions)?
What is PM Optimisation
The questions answered in completing a PMO2000 analysis are as follows:
1) What maintenance tasks are being undertaken by the operations and maintenance personnel
(task compilation)?
2) What are the failure modes associated with the plant being examined (failure mode
analysis)?
a) What is (are) the failure mode(s) that each existing task is meant to prevent or detect
b) What other failure modes have occurred in the past that have not been listed or have not
occurred and could give rise to a hazardous situation.
3) What functions would be lost if each failure were to occur unexpectedly (functions)? [optional
question]
9 SIRF Roundtables Ltd was formed on 1 July 2000 out of the Strategic Industry Research
Foundation Ltd as a result of the foundation's withdrawal from its industry shared learning
networks (including the IMRt) and related support activities.
SECTION 2. COMPARING PMO AND RCM METHODS OF MAINTENANCE ANALYSIS
OMCS Ph 0419397035 www.pmoptimisation.com Page 21
4) What happens when each failure occurs (failure effects)?
5) In what way does each failure matter (failure consequences)?
6) What should be done to predict or prevent each failure (proactive tasks and task intervals)?
7) What should be done if a suitable proactive task cannot be found (default actions)?
The complete PMO2000 methodology has nine steps. The seven questions listed above are a
subset of the complete PMO2000 methodology. The additional steps in PMO2000 not listed
above are as follows:
· Grouping and Review
· Approval and Implementation
· Living Program
These final three steps are necessary to implement the analysis outputs and ensure that the
PMO analysis does not stop once the first review has been completed. These steps are not
considered relevant to this paper as it is assumed that RCM analysis must also perform these
steps to ensure a successful outcome. RCM and PMO are considered identical in this regard.
Functional Differences between RCM and PMO
RCM and PMO are completely different products with the same aim; to define the maintenance
requirements of physical assets. Asset managers should be aware however, they have been
designed for use in completely different situations. RCM was designed to develop the initial
maintenance program during the design stages of the asset life cycle (Moubray 1997) whereas
PMO has been designed for use where the asset is in use.
As a result, PMO is a method of review whereas RCM is a process of establishment. Whilst
arriving at the same maintenance program, PMO is far more efficient and flexible in analysis than
RCM where there is a reasonably good maintenance program in place and where there is some
experience with the plant operation and failure characteristics.
Methodology differences between RCM and PMO
The central difference between RCM and PMO is the way in which failure modes are generated.
· RCM generates a list of failure modes from a rigorous assessment of all functions, a
consideration of all functional failures and then an assessment of each of the failure modes
that relate to each functional failure. RCM seeks to analyse every failure mode on every
piece of equipment within the system being analysed.
SECTION 2. COMPARING PMO AND RCM METHODS OF MAINTENANCE ANALYSIS
OMCS Ph 0419397035 www.pmoptimisation.com Page 22
· PMO generates a list of
failure modes from the
current maintenance
program, an assessment
of known failures and by
scrutiny of technical
documentation - primarily
Piping and
Instrumentation
Diagrams (P&IDs).
The differences in the two
approaches mean that PMO
deals with significantly less
failure modes than RCM and
arrives at the failure modes in
a far quicker time frame.
Experience in the US Nuclear
Power Industry was that over
a large number of analyses,
PMO was on average six times faster than RCM (Johnson, 1995). The methodological
differences between RCM and PMO are illustrated at Figure 10.
How and why PMO is faster than RCM
Overview
The main reasons why PMO is faster than RCM are summarised below. The points are
discussed in detail later in the paper.
1. Insignificant failure modes are not
analysed by PMO whereas RCM
analyses all likely failure modes.
2. Using PMO, many failure modes can be
rolled up and analysed together
whereas with RCM, failure modes are
analysed separately.
3. With PMO, a detailed functional
analysis is an optional step. The
function of the equipment is completed
as part of Consequence Evaluation
because a consequence of any failure
is a loss of function by definition.
Figure 10 Comparison of PMO and RCM
RCM
Functions
Functional Failures
RCM - All Failure
Modes to be
reviewed
PMO
Current PM
Failure History
Technical Documentation
Same
PM Program
PMO Pool of Failures
to be reviewed
Preventable
Failure Modes
Conventional
R C M
PM
Optimisation
Costs Time Benefits
Source: US Nuclear Power Industry
Figure 11 Comparison of the costs, time and
benefits of RCM compared with PMO
SECTION 2. COMPARING PMO AND RCM METHODS OF MAINTENANCE ANALYSIS
OMCS Ph 0419397035 www.pmoptimisation.com Page 23
How and why failure mode analysis of insignificant failures is avoided by PMO.
The equipment design and the way it is operated determine the type and likelihood of failure
modes. In the context of maintenance analysis, failure modes can be broken into categories
based on the following:
· their likelihood,
· their consequences, and
· the practicality and feasibility of preventing or predicting them. This point is illustrated in
Figure 11.
The focus of good equipment design
is to ensure high levels of reliability,
maintainability and operability. This
means eliminating high likelihood and
high consequence failures.
It is therefore, not surprising that
when reviewing the complete set of
likely failure modes using RCM
analysis, that by far the greatest
number of outcomes, or
recommendations, are No Scheduled
Maintenance. This is to say that for
the failure modes left in the design in
question, either
· Their likelihood is very low,
· There is no technically feasible predictive or preventive maintenance task known to manage
them, or
· The task that is known costs more to do than the cost of the cost of unexpected failure. The
less critical the equipment is to productive capacity, the more likely that the cost of the
maintenance outweighs the costs of the failure over a given life cycle.
In the author's experience, full RCM analysis of equipment shows that, on average, about 80% of
failure modes result with the policy of No Scheduled Maintenance. This information is presented
in Figure 12. This number rises with electronic equipment such as a PLC and falls with
equipment that has a high number of moving parts such as a conveyor.
It therefore follows that, if the objective of a maintenance analysis workshop is to define the
maintenance program, and all the likely failure modes are analysed, around 80% of the analysis
will be low value adding (or a complete waste of time). This is because the analysis finds that
there is no maintenance solution for 80% of the failure modes. Those failure modes could have
been culled at the start with no loss of analysis quality.
With this same objective in mind it is therefore logical to seek a process which limits the analysis
to those 20% of failure modes that are likely to yield a maintenance solution and no more. In
practice this is not completely feasible, as that pool of failure modes that receive PM is not
defined until the analysis is performed.
Figure 11 Considerations required for maintenance
analysis.
Likelihood Consequences PM Feasibility
High Hazard Feasible
Medium High Cost Not Feasible
Low Low Cost
A PM program is targeted at the conditions listed in the grey areas
of the table. These comprise the minority of failures, as the intent
of design is to engineer out hazards and high costs failures
particularly when their likelihood is high.
SECTION 2. COMPARING PMO AND RCM METHODS OF MAINTENANCE ANALYSIS
OMCS Ph 0419397035 www.pmoptimisation.com Page 24
If the failure modes are low consequence and infrequent, then there is unlikely to be any costeffective
modifications either.
The missing element here is failure modes that have hazardous consequences but have not
happened before. It is accepted that the downside of this approach is that failure modes that
could result in a hazard may be omitted so therefore it is wise to obtain the technical
documentation and perform a "desk top10" FMECA to trap these hazards if they exist. PMO2000
therefore errs on the side of caution and lists the failure modes that have the following attributes:
· Are currently the subject of Preventive Maintenance,
· Have happened before, or
· Are likely to happen and may cause a hazard.
How and why using PMO many failure modes can be rolled up and analysed together
RCM treats each failure mode independently resulting in the same analysis and tasks being
written many times. PMO starts from the maintenance task and therefore many failure modes
can be listed against the one task. This significantly reduces the analysis time by reducing the
records that need to be dealt with. The concept can be best described by reference to the
following example.
10 Review the technical documentation assessing if consequences of any failure would lead to a
hazardous situation.