PROACTIVE VERSUS REACTIVE
MAINTENANCE
MEASUREMENT/IMPROVEMENT
B Bigdeli & S Safi

Covaris Pty Ltd
Summary: successful maintenance improvement projects in large organisations require a
systematic and well-founded approach to ensure tangible technical outcomes. A structured
maintenance data analysis has been developed to address key maintenance objectives known as
proactive maintenance, backlog management and failure mode analysis.
Maintenance data (i.e. work orders) for one fiscal year 2002/2003 in a large Alumina refinery is
investigated in an attempt to established short, medium and long-term maintenance improvement
strategies. Data set included more than 61,000 individual work orders with a total amount of actual
cost recorded as high as $63m with more than 330,000 hours of actual work registered in the
CMMS.
Such a structured analysis has led to specific set of recommendations for improvements in planning
and scheduling of works, PM strategies for number of critical assets, new backlog management
strategy and engineering investigations into major failure modes. Since delivering the full report,
several of these improvement initiatives have been successfully implemented, and few others are
being considered for implementation.
Keywords: Maintenance Improvement, PM Strategies, Backlog Management, FMEA
1 INTRODUCTION
Cost cutting and efficiency improvement initiatives in maintenance departments are persistent concerns for maintenance and
engineering managers. The effect of these decisions in larger organisations could amount to the tens of millions of dollars
savings in yearly budgets if defined and implemented well. Therefore, it is of prime importance to be able to finely define
improvement targets with stepwise strategies to achieve specific goals.
There are different approaches to tackle this issue. One approach is to go through organizational change and define a path from
“As it is” condition to “As it should be” situation. Although we do not endorse such exercise, we do not comment on this
approach either. This paper, however, presents a more technically oriented approach with its associated developed
methodology. This has enabled us to define specific, technically viable and achievable solution strategies based on a systematic
interrogation of the Computerised Maintenance Management System (CMMS).
The methodology, process and its technical outcomes have been presented and future improvement initiatives arising from this
investigation are explained.
2 METHODOLOGY
Alumina refineries, designed and operated based on Bayer process (1), are consisted of four distinct process areas, being
Digestion, Clarification, Precipitation and Calcination (apart from its Bauxite mine and Powerhouse). In another classification,
Digestion and Clarification areas are commonly referred to as “Red side” and Precipitation and Calcination areas are called
“White side”. This is due to the colour of the product in each part of the plant, which transforms from red to white as the
process flow passes through facility.
The investigation, therefore, was divided into four sections to represent the true nature of operation and its associated failures.
For each area, following points were investigated:
• Reactive vs. Proactive maintenance
Print Menu Go Back Next Page
ICOMS-2005 Paper 042 Page 2
• Criticality analysis
• Backlog management
• Maintenance improvement initiatives
A data file containing work order history for 2003/2004 fiscal years was prepared and downloaded from the CMMS. This file
contained more than 61,000 individual work orders for both Red side and White side with more than 330,000 actual working
hours recorded against all work orders.
3 REACTIVE VERSUS PROACTIVE MAINTENANCE
The first step in assessing the technical compliance of a Preventive Maintenance (PM) strategy is to measure the relative
magnitude of Reactive works (driven by the plant) versus Proactive works (driven by human management of the plant). The
Covaris (2) classifications for these types of works are:
Reactive work
Breakdown and/or unplanned jobs– no control of timing of the work
Corrective – no control of the timing of the work, and driven by a non-scheduled event
Proactive work
Scheduled – control of the timing and initiated by either calendar time or metered time, typically used for inspections
Condition based – control of the timing and initiated by a scheduled event such as an inspection result
It is worth to mention that a work is also considered to be reactive if initiated as a result of a condition monitoring activity, but
cannot be scheduled (i.e. urgent action required).
There are different work types associated with work orders in different CMMSs. In SAP (3) terminology there are two major
work order types:
PM01 – non-routine maintenance
PM02 – System generated maintenance work orders
Although PM01 work orders are defined as non-routine maintenance jobs, they are not a true indicator of reactive maintenance
work type. Table 1 presents the definitions used to categorise work types in a SAP environment. All PM01 work orders with
priorities 4, P and S are allocated to proactive maintenance due to the ability to plan these types of PM01 work orders in
advance. PM02 work orders contribute the major proportion of the proactive work type.
Priority Required action Reactive/Proactive
1 Immediate action Reactive
2 Must be done within 24 hours Reactive
3 Must be done in current week Reactive
4 Low Priority work Proactive
P To be planned Proactive
S For shut down plan Proactive
Table 1. PM01 work order classification
Therefore, Reactive and Proactive work orders are defined as follows:
Proactive work orders = All PM02 + PM01 (with priorities 4, P and S)
Reactive work orders = PM01 (with priorities 1, 2 and 3)
This interpretation of work orders produced following results in terms of the % of actual time spent on work orders in four
major process areas as shown in Figure 1.
Menu Go Back Next Page
ICOMS-2005 Paper 042 Page 3
26.2
32.7 31.9 33.1
73.8
67.3 68.1 66.9
0
10
20
30
40
50
60
70
80
Digestion Clarification Precipitation Calcination
Production area
% of actual time
% Reactive maintenance % Proactive maintenance
Figure 1. Maintenance planning indicator in 4 major process areas
This figure indicates that the maintenance planning is influenced by unplanned events with magnitudes of 26.2%, 32.7%,
31.9% and 33.1% of the total maintenance time recorded in Digestion, Clarification, Precipitation and Calcination production
areas respectively.
A ratio of 85% planned work to 15% unplanned work as a target value is considered to be a good maintenance practice for this
indicator (4). Therefore, there is a need to reduce the quantity of reactive maintenance by greater than half in most production
areas. However this metric does not portray the full picture. Additional issues need to be considered including:
• We need to ascertain why the quantity of PM02 work is not driving down the PM01 work – this is a matter of
addressing failure modes in the plant and can be either a design issue or an improvement in the repair tasking, both of
which represent engineering improvements that need to be progressed.
• The metric does not portray estimated labour productivity, which can be impacted by wait time. If the actual time on
refurbishment of equipment within the work order is increased (i.e. wrench time) then not only is it likely that reactive
work can be relieved, but total costs can be reduced.
Plant area Reactive works
(%)
Proactive works
(%)
Improvement
Opportunities
(%)
Digestion 26.2 73.8 11.2
Clarification 32.7 67.3 17.7
Precipitation 31.9 68.1 16.9
Calcination 33.1 66.9 18.1
Average 31 69 16
Table 2. Improvement opportunities in planning and scheduling of works
Table 2 shows the improvement opportunities for different areas of the plant, which can be achieved by the following:
• Refinement of the maintenance strategy so that the reactive work is better anticipated, possibly through improved
analysis of trends in inspection results.
• Re-design or improvement of materials – it was noted that at this site practices in assessing effectiveness of materials
are not consistent across the plant types with better assessments conducted for some plant types compared to other. It
was found, for instance, that material selection for Valves had better engineering understanding of the process
requirements than that of Pipes in the same part of the plant.
Total actual cost throughout the refinery for the analysed data set was $63.5m. As a rule of thump, an unplanned maintenance
work is at least 50% more expensive than a planned one (over labour, material and overhead cost components). Therefore,
there is a possibility of maintenance cost reduction by a magnitude of $275K on average for each 1% reduction in unplanned
works from the current level of 31% (see Table 2).
Consequently $4.4m reduction in maintenance costs can be expected by achieving the target value of 85% planned work orders
on average throughout the plant.
Menu Go Back Next Page
ICOMS-2005 Paper 042 Page 4
4 CRITICALITY ANALYSIS
Equipment criticality codes used at this site (consistent with general SAP definitions) are shown in Table 3.
Code ABC indicator text
A Immediate production loss
B H/Potential production loss
C Potential production loss
D No production impact/H cost
E Low production impact/L cost
Table 3. Equipment criticality index
The actual time recorded for PM01 work orders with high priorities (i.e. priorities 1, 2 & 3) indicates the amount of reactive
workload on maintenance crews. This indicator is shown in Figure 2.
PM01 work orders with priorities 1, 2 & 3
0
5000
10000
15000
20000
25000
30000
35000
40000
A B C D E
Equipment criticalities
Actual time (h)
Figure 2. Actual time recorded for PM01 work orders with high priorities (Reactive)
A total amount of 92,469 hours is recorded for PM01 work orders with priorities 1, 2 and 3 (i.e. unplanned works imposed to
the weekly plan). This analysis reveals the following points:
• High criticality plant, which had a reliance on urgent corrective maintenance. This reflects on both client’s
understanding of plant condition as well as the maintenance strategy for these assets.
• Urgent responses in maintenance to low criticality assets are somehow puzzling. In some cases these may be justified
where safety issues have arisen, but such rapid response should be assessed if they are causing unnecessary pressure on
the scheduled work program.
5 BACKLOG MANAGEMENT
At the time of the analysis a total number of 5,205 work orders were in backlog. Figure 3 shows the distribution of work orders
based on equipment criticalities throughout the refinery.
Menu Go Back Next Page
ICOMS-2005 Paper 042 Page 5
Work orders in backlog versus equipment
criticalities
0
200
400
600
800
1000
A B C D E A B C D E
PM01 PM02
Equipment criticalities
No. of WO in backlog
Figure 3. Backlog report against equipment criticalities
Comment on the backlog data indicates that:
• PM02 work is getting done despite the high loading of PM01 work – this may be seen as a cost driver on the labour
and also troublesome insofar as the client is not seeing a good return on its investment on this labour.
• Delays in attending to PM01 work may mean that additional stress is being placed on the plant with two consequences:
o Impact on production rates or quality
o Growth in the size of the fault that will be eventually rectified under a PM01 work order
The existing backlog report, however, does not relate equipment criticalities to work orders priorities to produce an important
indicator known to Covaris as Backlog Management Index (BMI) number.
BMI is an equivalent risk value that shows the combined maintenance priority and equipment criticality. In this analysis the
following definition has been used to define the BMI:
BMI = (Task Priority) * (Asset Criticality)
The value of BMI can vary from 1 to 25. The highest number shows the highest risk to the plant. The following table shows
the recommended BMI numbers.
Asset Criticality
Priorities/Criticalities E D C B A
1 1 2 3 4 5
2 2 4 6 8 10
3 3 6 9 12 15
4 4 8 12 16 20
Task Priority
5 5 10 15 20 25
BMI is used in the backlog KPI report to show the equivalent risk. The acceptable limits for the backlog data is defined and
presented in the table below (as an example for this client only). This policy will be graphically presented in the backlog report
to distinguish between the acceptable performances and the risky performances. The report will also show the specific task that
is a threat to the assets or has the criticality changed over time
Menu Go Back Next Page
ICOMS-2005 Paper 042 Page 6
Level Preferred Maximum Time
in Back Log
Recommended Response to Remove from Backlog
1 2 months Review if task is necessary – delete if not; increase
task criticality if it is
2 1 month Increase task criticality to expedite attention
3 1 week Urgent priority to address task
A typical backlog report considering BMI number is shown below.
Figure 4. Sample backlog report using BMI concept
It should be noted that the broken line is a corporate policy and may slightly differ for in different organisations. All work
orders in the right hand side of the broken line are unacceptable and represent a risky performance.
6 MAINTENANCE IMPROVEMENT INITIATIVES
Following detail analysis of work order histories in four production areas, the distribution of high cost assets was investigated
and following conclusions were made:
1. Maintenance improvement opportunities in mill facility exist. It was found that maintenance
strategies/procedures for four Bauxite mills could mot prevent mill failures and as a consequence mill
reliabilities were lower than expected. A maintenance improvement study with a detailed analysis of high
priority work orders (i.e. PM01 WO with priorities 1,2 and 3) was recommended for mill area.
2. Engineering investigation into fouling mechanism for heat exchanger tubes in Digestion area. Heat exchangers
are originally designed to last for 90 days in operation, however, slurry deposit and subsequent erosion force
operation to take them out of service in less than 60 days (even in 45 days in some cases). Heat exchangers
operation suffers from this issue showing much lower than expected reliability and this issue amounts to almost
$6m extra cost to maintenance each year. An engineering investigation is certainly justified.
3. Pump wearing study in Digestion area was also found important. Cost figures for pump maintenance suggested
that an improved pump design (i.e. material and internal coating) would significantly reduce frequent
unpredicted pump failures.
4. Work order history analysis and subsequent review into NDT inspection regimes throughout refinery indicated
different level of accuracy as well as coverage in the plant. While some part of the refinery had an acceptable
level of NDT plan coverage, the lack of a technically viable NDT plan was a major contributor to pipes and
valves deterioration and unplanned equipment failure. Unpredicted failures due to pipes and valves erosion and
661 Backlog (05/02/2002)
0
5
10
15
20
25
30
1 10 100 1000 10000
Days
Menu Go Back Next Page
ICOMS-2005 Paper 042 Page 7
holes contributed to more than $3m of maintenance cost. Therefore, a complete review and overhaul of the
NDT plan was strongly recommended.
5. It was found that “Cause codes” were not recorded well when closing work orders and information contained in
notifications could not produce a reliable set of failure modes. However, in most cases work order headers
contained reliable information on what is wrong with the asset and contained proper technical data. Covaris has
developed a unique tool to perform a comprehensive Failure Mode and Effect Analysis (FMEA) based on the
work order history data analysis (5). Such study requires thorough engineering examination of corrective work
orders and will identify failure modes, their costs effect on individual asset base and associated risks. Similar
investigation for this site is strongly recommended.
7 CONCLUSIONS
In this paper, a systematic approach to analyse and evaluate maintenance data (i.e. work orders) in a large Alumina refinery is
presented. Proactive maintenance is discussed and a measuring concept independent of CMMS terminology is developed.
Based on this analysis short, medium and long-term maintenance improvement strategies are also presented for the case
studied. Data set included more than 61,000 individual work orders with a total amount of actual cost recorded as high as $63m
with more than 330,000 hours of actual work registered in the CMMS.
A refined backlog report process described in detail, which enables planners to understand and mange risks associated with
works in backlog. Recommended initiatives are in different stages of implementation and improvements in some cases (e.g.
initiatives 1, 2 and 4) are being realised.
8 ACKNOWLEDGEMENTS
The authors acknowledge the input of work colleagues at Covaris Pty Ltd and the many clients and research partners who have
contributed to this work.
9 REFRENCES
1. R. Harris, Production of Alumina, Aluminium Handbook (1999)
2. S. Safi, S. Mozar, From Reactive Maintenance to Proactive Preventive Maintenance System, ICOMS-2004, Sydney pp.
1-8, May 25 – 28, (2004)
3. B Stengl & R Ematinger, SAP R/3 Plant Maintenance, P82-92 (2001)
4. RW Peters, Measuring Overall Craft Effectiveness (OCE), Maintenance Journal, P76-78 (2003)
5. S. Safi, R. A. Platfoot, Report on