Master Thesis

Maintenance optimisation of centrifugal pumps in a

European refinery: A case study

KTH Royal Institute of Technology

School of Industrial Engineering and Management

Author: Andreï LAQUET

Thesis responsible: Jerzy MIKLER
 Master Thesis

Abstract
Maintenance has gained credit over the past decades. The oil and gas industry requires
efficient maintenance programs due to the hazardousness surrounding the industry. Crude
oil margins are also dropping and maintenance yields high controllable costs. Therefore,
safety and economy are the driving forces of maintenance optimisation. Refineries have to
operate when margins are the most profitable so reliability is crucial.

Centrifugal pumps are essential features of the refining process. Due to limited resources,
maintenance work is prioritised according to the operating context and risks. Criticality
analysis is a widespread tool in the industry which supports the resource allocation decision.
The criticality level has, thus, to be highly accurate and adapted to the plant situation since it
influences the overall maintenance system efficiency. Maintenance plans must be properly
designed and implemented to enhance reliability.

This paper states maintenance organisation around centrifugal pumps maintenance in a

refinery. First, the current criticality assessment methodology has been found to be
inadequate to the prioritisation of work so a new method has been developed. Then, a
tailored spare parts strategy has been implemented and discussed. Finally, preventive
maintenance plan has been reviewed using pump specialists findings and maintenance
optimisation methodologies which has led to several improvement hints proposals.

Andreï LAQUET
 Master Thesis

Acknowledgment
I would like to thank my supervisor Jerzy Mikler, for his help and advices given
during the thesis.

A particular thank also to my supervisor within the company, Dominique Guedon,

head of the mechanical maintenance department in Total Donges refinery. He
provided me with all the information and resources that led to the project success.
His trust has highly contributed to my integration within the refinery which was
essential for the study.

I take this opportunity to express gratitude to all maintenance department members

who helped me during the project by sharing their personal skills and knowledge
with me.

Donges, FRANCE

2015-07-07

Andreï LAQUET

Andreï LAQUET
Content Master Thesis

Content
1 Introduction ....................................................................................... 1
1.1 Presentation ............................................................................................................... 1
1.2 Scientific objective and aim ..................................................................................... 2
1.3 Structure and methodology .................................................................................... 2

2 Maintenance evolution and scope............................................... 2

2.1 Maintenance through the years .............................................................................. 3
2.2 Maintenance influence and goals ........................................................................... 4
2.2.1 Targets and goals ............................................................................................... 4

2.2.2 Maintenance methodologies ............................................................................ 6

2.3 The petrochemical industry .................................................................................... 9

2.3.1 Background ........................................................................................................ 9

2.3.2 Economy ............................................................................................................. 9

2.3.3 Economy and safety ........................................................................................ 12

3 Improvement methodologies .................................................... 15

3.1 RCM and its limitations ......................................................................................... 15
3.2 Streamlined RCM (SRCM) .................................................................................... 16
3.3 Preventive maintenance optimisation (PMO) .................................................... 17

4 Case study scope ............................................................................ 18

4.1 Centrifugal pumps in the operating context ...................................................... 18
4.2 Risk assessment ....................................................................................................... 19
4.3 Criticality assessment methodologies.................................................................. 20
4.4 Scientific background ............................................................................................. 21

5 Maintenance of centrifugal pumps: a case study ................ 22

5.1 Centrifugal pumps: a brief description ............................................................... 22
5.2 Description of the operating environment .......................................................... 23
5.3 Maintenance management organisation ............................................................. 23

Andreï LAQUET
Content Master Thesis

6 Criticality analysis ......................................................................... 25

6.1 The current situation .............................................................................................. 25
6.2 The new approach .................................................................................................. 27
6.3 Calculation of the functional availability ............................................................ 29
6.4 Evaluation of maximum impact ........................................................................... 30
6.5 Results ...................................................................................................................... 30
6.6 Limits ........................................................................................................................ 31

7 Corrective maintenance system ............................................... 32

7.1 Importance and issue ............................................................................................. 32
7.2 Spare levels according to criticality and common failure modes .................... 33
7.3 Challenges and limitations .................................................................................... 35

8 Preventive maintenance: review and progress ................... 36

8.1 Methodology ........................................................................................................... 36
8.2 Failure mode analysis ............................................................................................ 36
8.3 Current preventive plan ........................................................................................ 37
8.4 Conclusion and improvement hints .................................................................... 39
8.5 Vibration and temperature continuous monitoring: BEACON sensor .......... 40
8.5.1 Interest............................................................................................................... 40

8.5.2 Description of the technology........................................................................ 40

8.5.3 Implementation challenges ............................................................................ 41

9 Conclusion and further studies ................................................. 42

10 References ........................................................................................ 43
11 Appendixes ...................................................................................... 46
11.1 Appendix 1: Refineries general process description ......................................... 46
11.2 Appendix 2: Maintenance actors and criticality scope diagram ...................... 48
11.3 Appendix 3: Spare parts strategy and economical returns............................... 49
11.4 Appendix 4: Mobley’s root cause failure analysis ............................................. 51
11.5 Appendix 5: List of control for pump, mechanical seal and motor ................. 53

Andreï LAQUET
Content Master Thesis

List of figures and tables

Figure 1 : Growing expectations of maintenance ............................................................... 4

Figure 2: System organisation of a company....................................................................... 5

Figure 3: Planned and unplanned Partial Revision number ........................................... 10

Figure 4: Planned and unplanned General Revision rate................................................ 11

Figure 5: The vicious circle of Maintenance ...................................................................... 11

Figure 6: Overhung single stage centrifugal pump .......................................................... 22

Figure 7: Maintenance task organisation ........................................................................... 24

Figure 8: Distribution of centrifugal pumps according to criticality level .................... 26

Figure 9: Criticality matrix ................................................................................................... 28

Figure 10: Criticality assessment components .................................................................. 29

Figure 11: Extract from criticality analysis ........................................................................ 30

Figure 12: Criticality distribution ........................................................................................ 31

Figure 13: Stock levels after regulation .............................................................................. 34

Figure 14: Vibration and temperature monitoring sensor............................................... 41

Table 1: Preventive and corrective maintenance repartition .......................................... 13

Table 2: Accident cause distribution ................................................................................... 14

Appendixes

Figure 15: Refining processes: a simplified chart ............................................................. 46

Figure 16: Maintenance actors and criticality scope diagram ......................................... 48

Figure 17: Common failure modes of centrifugal pumps ............................................... 51

Figure 18: Common failure modes of mechanical seals................................................... 52

Table 3: Economical impact of downtime .......................................................................... 47

Table 4: Spare part strategy according to criticality ......................................................... 49

Table 5: Economical returns of the strategy ...................................................................... 50

Table 6: Controls carried out for centrifugal pumps maintenance ................................ 54

Andreï LAQUET
Introduction Master Thesis

1 Introduction
1.1 Presentation
Over the past decades, the industry has been constantly evolving. The challenges
emerging from both market and society have matured and depict now an entire
different situation and framework. The concerns and focus of 1950’s which have
constructed the old industrial paradigm are now obsolete or at least irrelevant today.

The evolution of the society and the economical changes that the world has been
experiencing build the cornerstones of the maintenance philosophy change. Indeed,
new trade-offs and new challenges have revolutionised the industry and
maintenance work is now granted with higher merit.

For years, production has been the greatest endeavour of the industry. Nearly every
organisation focused on the adding value part of the processes. In other word
everything was set up in order to improve the product production processes. Thus,
maintenance has for long been seen as a necessary evil [1] that had to be fulfilled in
order to keep its great brother, the production, running. Through the past decade,
this feeling has been evolving. Maintenance has gained credits and is seen as a high
leverage business function which generates large controllable operating costs. Since
these costs are controllable maintenance leaves room for improvements and
industries are now trying to exploit it.

Maintenance utopia relies on perfect equipment reliability. Machines have to fulfil a

function which is clearly identified and has performance standards. The performance
standards are the basis of the maintenance work. If not reached an action has to be
taken in order to tackle the problem. That is the most common vision of the
maintenance system but it has to be narrowed. Nowadays the maintenance is not
that simple and the technological evolutions and improvements lead to an even
higher efficiency and precision.

Most of the industrial part of the preventive maintenance work is done in an

informal manner [2]. In practice it is important to master what is done in order to be
able to optimise the methodology. The overall system has to be reviewed to reduce
and optimise maintenance cost. The traditional couple Preventive Maintenance (PM)
and Predictive Maintenance can provide an acceptable and relatively high reliability
even though it is not the most efficient way to proceed. Indeed, adding a refined

Andreï LAQUET 1
Introduction Master Thesis

strategy to the two maintenance practices is the best way to go deeper into the
optimisation of the maintenance work.

Several maintenance methodologies are formalised and exist to do so. The most
common ones are Reliability Centred Maintenance (RCM), Total Productive
Maintenance (TPM) and other Preventive Maintenance Optimisation (PMO). These
methodologies will be developed later on.

1.2 Scientific objective and aim

As mentioned in the presentation, maintenance generates high controllable costs. The
knowledge of these costs can be very useful for the company. As refining margins
fluctuate equipment reliability must remain high in order to be able to operate when
the conditions are the most profitable. Therefore, the main challenge is to find a
balance between low maintenance cost and high reliability. A review of maintenance
plan is needed to reach this objective. The case study has been carried out in a French
refinery which was in the situation depicted above. The need for evaluation of the
maintenance strategy is crucial for this type of plant to be able to master and
optimise the maintenance program. The paper aims to answer to the following
question:

How theoretical findings and maintenance concepts can contribute to optimise

maintenance programs and enhance reliability in mature industries?

A clear methodology has been developed to conduct the study. This guideline is
described in the following section.

1.3 Structure and methodology

First, an overall review of the maintenance evolution in the industry is needed to
understand the scope and the challenges coming from both history of the plant and
the oil and gas sector. Indeed, the history of both sector and facilities are often
relevant to understand current practices. Then, maintenance improvement
methodologies, such as RCM, and prioritisation strategies used to allocate resources
will be described. They are essential to apprehend how maintenance enhancement
can be carried out and what are the challenges. Finally, the case study will describe
to what extend Total strategies contribute to reliability improvement and what could
be done to maximise the plant’s efficiency by optimising the trade- off between cost
and reliability.

Andreï LAQUET 2
Maintenance evolution and scope Master Thesis

2 Maintenance evolution and scope

2.1 Maintenance through the years
Through the past 80 years, scientists agreed to categorise the evolution of
maintenance through 3 main generations. John Moubray used the evolution of
maintenance to introduce and contextualise Reliability Centred Maintenance (RCM)
in the current industrial framework. [3] It is important to understand the reason why
RCM and other maintenance methodologies have been developed in order to be able
to use these tools efficiently.

The first generation is the symbol of old industrial manners, it covers the period up
to World War II during which maintenance was not be carried out until the event of
failure [4]. The run-to-failure strategy was the only way to proceed. It might seem
archaic nowadays but it suited perfectly equipment lifecycle. At that time there was
no high technology or mechanised process and the equipments were simple and
usually over-designed. Downtime was not important and so was the failure
prevention. In other words, maintenance seemed to be set apart in this generation.
However, it is interesting to understand that before 1940’s industrials did not invest
in maintenance because there was no need to do so.

The second generation was driven by the industrial changes that World War II had
brought. The balance between supply and demand drastically changed during the
war. Workforce was scarce while demand was increasing. This led to a need for
mechanisation which entailed higher plant availability. Industries had to develop
quickly new solutions in order to be able to meet the demand. These new challenges
had a huge impact on how maintenance was seen by industrial managers. Indeed,
preventive maintenance concept made its first appearance in the sixties to tackle
plant downtime.

The strategy switched during the second generation from systematic run-to-failure
maintenance plans to preventive maintenance plans based on fixed intervals
overhauls. After some years, cost of maintenance began to be too high to keep
maintaining equipments with that strategy. Nevertheless, the trend was not to give
up and go back to a run-to-failure methodology but quite the opposite. Plants needed
and still need even higher availability due to lower margins and a fierce competition.

The third and current generation began in the mid-seventies. Moubray characterises
the changes to this third generation as changes driven by “new expectations, new
Andreï LAQUET 3
Maintenance evolution and scope Master Thesis

researches and new techniques”. These adjectives summarise the main spirit
conveyed by the current society. New expectations triggered by growing concerns in
the society. Indeed, safety and environmental standards are ones of the driven forces
which motivate, not to say force, managers to improve their maintenance efficiency.

Source: RCM II by John Moubray (1997) [3]

Figure 1 : Growing expectations of maintenance

The three maintenance generations depicted by John Moubray have been driven
directly by industrial needs. This evolution has, of course, optimised production
efficiency in several plants thanks to these new practices. Originally maintenance
aimed to reduce the risks that higher technologies and processes introduced. The gap
has been first identified in potentially hazardous sector. Indeed, petrochemical
industry and aviation sector were ahead of the others because they both have risk
management as a core feature of the organisation. The prime objective was originally
only safety-driven. Nevertheless it has been proven that maintenance achievements
affect global system performances. [5]

2.2 Maintenance influence and goals

2.2.1 Targets and goals
It is important to consider a range of key performances of a company in order to
envision the context. According to Noyes and Peres [6] system performances are
characterised by:

 The production cost which includes maintenance cost but also production
profits
𝑈𝑝𝑡𝑖𝑚𝑒
 System availability defined by: 𝐴𝑣𝑎𝑙𝑖𝑎𝑏𝑙𝑖𝑡𝑦 = 𝑈𝑝𝑡𝑖𝑚𝑒+𝐷𝑜𝑤𝑛𝑡𝑖𝑚𝑒
So breakdowns and unplanned maintenance operations have an important
impact on the global availability which emphasises the role for an efficient
maintenance program.

Andreï LAQUET 4
Maintenance evolution and scope Master Thesis

 Safety level and potential risks for workers, machines and environment.

Thus, maintenance objective is to ensure a high level of performances to improve the

overall system reliability. Maintenance goals are identified in four main assertions
[7]:

 Reduce unavailability
 Ensure product and services quality
 Master the costs linked to the maintenance program
 Protect the key actors of the maintenance action: persons, environment and
the machines involved in the scope

In other words, maintenance aims to enhance reliability which is linked to three of

the four topics of the assertions above. Reliability covers the concepts of availability,
safety and quality of products and services. The art of maintenance consists in
improving equipment reliability with limited budgets. The main challenge is to fulfil
the operational requirements, through a reliability level target, at optimal cost [8].
This trade-off is the symbol of the third maintenance generation and also the image
off the current society. Actually even the biggest companies pay attention to reduce
their expenses so maintenance plans have to be adapted and optimised to the
industrial estate. In order to be able to do so, maintenance must be seen in the global
frame of a company highlighting its inputs and outputs. The following diagram,
Figure 2, summarises maintenance position in the operating context.

Source: 2002, Tsang [9]

Figure 2: System organisation of a company

Particular attention has to be put on the word system to properly understand this
chart. Maintenance system is included in the production system because it is a

Andreï LAQUET 5
Maintenance evolution and scope Master Thesis

prerequisite for the production to have a maintenance system around. However, this
does not mean that maintenance department in terms of hierarchy is encompassed in
the production department. The trend conveyed by the third generation of
maintenance is to put on an equal footing maintenance and production departments
in order to maximise the enterprise system efficiency.

2.2.2 Maintenance methodologies

The growing expectations of maintenance go with new managerial approaches.
Several well thought-out methodologies have been developed to optimise, review
and design suitable maintenance plans. Reliability-Centred Maintenance (RCM),
Total Productive Maintenance (TPM) with Six-sigma are the most popular ones in the
industry. They are different maintenance development methodologies with their
own degree of improvements and outcomes. Jardine and Tsang [10] express the
differences between RCM and TPM by the following definitions:

 Reliability-Centred Maintenance (RCM): an asset-centred methodology

 Total Productive Maintenance (TPM): a people-centred methodology

Nevertheless, TPM and RCM are not antinomic concepts. Indeed, RCM conducts a
solid way to improve and master preventive maintenance which can enhance TPM
implementation. [11]

RCM methodology takes its origin in the aviation sector with Nowlan and Heap [12]
in the 1960’s. Their work takes place in the second maintenance generation when the
Boeing 747, amongst others, has been introduced. In the 1960’s, periodic overhauls
aimed to guarantee the required safety and availability level. Actually, their research
emphasises quite the opposite of what was targeted. In most cases, periodic
overhauls had no effect on reliability and safety of the equipments even if the
periodicity of the maintenance work was changed. RCM is defined by its creators as
“a scheduled-maintenance program designed to realise the inherent reliability
capabilities of equipment” [12]. John Moubray refined RCM into RCMII based on the
principles of the two American engineers. He defined its version of RCM as “a
process used to determine the maintenance requirements of any physical asset in its
operating context” [3]. His work focuses on the potential consequences more than on
the failure itself. In that way Moubray modernised the concept and opened it to new
horizons wider than the initial aviation scope. Through 7 basic questions the purpose
of RCM is to mitigate failure consequences in order to ensure reliability of the
equipments.

Andreï LAQUET 6
Maintenance evolution and scope Master Thesis

1. What are the functions of the asset?

2. In what way can the asset fail to fulfil its functions?

3. What causes each functional failure?

4. What happens when each failure occurs?

5. What are the consequences of each failure?

6. What should be done to prevent or predict the failure?

7. What should be done if a suitable proactive task cannot be found?

RCM is nowadays one of the most popular methodologies used to implement and
develop maintenance plans. One of the key features of RCM is that it focuses on the
functional requirements, on what the user wants it to do. It is a highly valuable
methodology which could contribute to develop a cost-effective maintenance
program. Nevertheless applying RCM is a complex and time-consuming labour and
industrials might be reluctant to opt for this technique. To tackle this problem,
several attempts have been carried on to build RCM framework in order to simplify
the use of RCM. They will be discussed later on in this paper.

Total Productive Maintenance (TPM) is a people-centred methodology [10] because

every actor from the top managers to the operators has a role to play. It is a
“productive maintenance involving total participation” according to Nakajima’s
definition. [13] TPM aims to break the conventional border between production and
maintenance departments which is one of the biggest challenges to solve to have an
efficient maintenance program. The lack of communication brings problems such as
unavailability of the production resources (i.e. the equipment) which results in
unrealised inspections and other preventive tasks that needed to be fulfilled
according to the maintenance plan [14]

Andreï LAQUET 7
Maintenance evolution and scope Master Thesis

In TPM, operators are involved and play an active role in maintenance detections
and small repairs. It focuses on the six following equipment losses [13]:

 Breakdowns
 Setup and alignment
 Idling and minor stoppages
 Reduced speed
 Defects in process
 Reduced yield

Operators are in charge of carrying some controls during the equipment operating
time to prevent one of the losses depicted above to happen.

There are other forms of maintenance development technologies. However,

managers have to be aware that even the best technical solutions can suffer from
poor management involvement. One of the key of any industrial change is the
communication and the actor’s involvement. Total productive maintenance
methodology as all the Japanese production methodologies, 5S for example, put the
person at the centre of the study. It is highly crucial in any method to take the human
factor as a key input of the system. In Tsang’s chart, the human factor is
encompassed in the labour category. This has to count in the methodology itself but
above all during the implementation process. An accurate and highly sophisticated
RCM approach remains useless and meaningless if it stagnates to the research phase.
These tools are powerful and have proven economical outcomes but they must be
applied and used properly. This is too often underestimated and it has to be taken
into account before implementing any change in habits.

It is a real challenge to motivate industries to use RCM, to name only Nowlan and
Heap’s work. To motivate the use of such complex tools, benefits have to be proven
beforehand. Improvements are proven but every sector has its own specific
challenges or at least the priorities are different. Some will emphasise the security
aspect of their equipment while others will put pressure on reducing costs and then
lower the maintenance costs. The framework in which the study takes place is thus
crucial and need to be narrowed down in order to identify the potential benefits of
improving maintenance.

Andreï LAQUET 8
Maintenance evolution and scope Master Thesis

2.3 The petrochemical industry

2.3.1 Background
This paper aims to review maintenance policy of Total Donges refinery so the study
takes place in the petrochemical industry. It is one of the most advanced industries in
terms of maintenance due to the potential risks brought by hazardousness of the
processes. The global industrial situation has been described in the beginning of this
paper but, as announced above, it is important to highlight the specificities of the oil
and gas sector in order to have an overview of the context to understand how and
why improving maintenance through the methodologies can bring great benefits.

The following case study takes place in a European refinery. The maintenance
management organisation differs from the one of our case study. However, the
outcomes can be spread to a wider range than only the plant studied because
European refineries are similar and have similar challenges to overcome.

2.3.2 Economy
The economic impact of maintenance management ineffectiveness in oil and gas
companies has been studied by Aoudia, Belmokhtar and Zwingelstein in 2008 [14]. In
order to prove and highlight the importance of maintenance they chose to
demonstrate the contrary of their theory in order to strengthen their words. By
developing the financial losses brought by maintenance management ineffectiveness,
they underlined the potential strength of maintenance improvements.
Moreover, the oil and gas industry is now suffering from the drop in crude oil price
which thwarted companies’ forecasts. It would be yet inaccurate to affirm that the
increasing need for higher maintenance efficiency comes from that drop. However,
economical savings are topical and optimised-cost maintenance plans are welcomed.

Their demonstration relies on two steps. First, ineffectiveness causes due to poor
maintenance management are identified and then evaluated in terms of financial
losses. As mentioned before, the implementation and the importance that
maintenance is granted in the organisation is one of the key of success. The study has
been conducted in a refinery built more than 50 years ago with a large number of
equipment and manufacturers. This description could nearly describe any European
refinery so the challenges are quite similar to a general case in which European
refineries after World War II are considered. They identified three main facts which
point up maintenance management ineffectiveness: Cancellation of preventive
maintenance programs, delays in the implementation of corrective maintenance
actions and number of accidents.

Andreï LAQUET 9
Maintenance evolution and scope Master Thesis

Cancellation of preventive maintenance tasks

Preventive maintenance programmes are crucial in order to reach a high reliability

level. Industrial processes plants operate continuously so any breakdowns can
potentially affect the production and engender a financial loss. This can be expressed
by a need of high availability of equipment [15], so preventive maintenance plans are
designed in order to ensure such level. The most common preventive actions are
scheduled overhauls, inspections, periodic greasing, etc. Focus is often put on
optimising the interval between preventive actions while management often turns to
be the impediment. Indeed, the optimum interval is calculated and based upon
operating hours, operating conditions and machine type. The programme
effectiveness has been limited due to management errors. The equipment is often not
available for maintenance inspections so planned actions are not carried out. This is
mainly due to a lack of communication between production and maintenance. Figure
3 and Figure 4 show the differences between what was planned and what has been
done in the refinery studied between 2001 and 2005. PR stands for Partial Revision
and GR for General Revision.

Source: Case study 2005 by Aoudia et al. [14]

Figure 3: Planned and unplanned Partial Revision number

Andreï LAQUET 10
Maintenance evolution and scope Master Thesis

Source: Case study 2005 by Aoudia et al. [14]

Figure 4: Planned and unplanned General Revision rate

The deviation is greater for Partial Revision than for General Revision due to the high
number of PR in the maintenance program. However, the difference between
planned and unplanned preventive actions lead to higher risks in terms of
breakdowns so behind the managerial problem lays the operating problem. The
priority is often given to operational purpose because the financial loss due to a
decrease in production rate is direct and immediate. On the opposite, the financial
loss caused by inaccuracies in preventive maintenance plan induces, potentially,
higher cost but postponed. Therefore, preventive tasks are obviously less crucial in
operating staff’s mind. The risk is to fall into the vicious circle of maintenance, Figure
5, depicted by Steve Turner [2]:

Source: PM Optimisation, Steve Turner [2]

Figure 5: The vicious circle of Maintenance

Once PM is missed there is a risk of entering in that circle. It is managers’ job to break
it through strong communication and action plan to highlight maintenance
importance in the organisation
Andreï LAQUET 11
Maintenance evolution and scope Master Thesis

2.3.3 Economy and safety

Their study highlights the fact that top managers are often reluctant to use
maintenance improvements approaches because the benefits are hardly assessed in
terms of financial profit. These assertions have to be narrowed down particularly in
industries like petrochemical or any other hazardous industries. Maintenance
development has not always been profit driven and it was quite the opposite when
preventive maintenance was presented for the first time. Even if we see now
preventive maintenance as a way of preventing breakdowns and master maintenance
cost plans it was not the original aim of it. Preventive maintenance has been
developed to increase the reliability of the equipment to meet the demand without
regarding the potential profits of such programs. It is true that, in times of financial
crisis, profit is a need but safety and security remain the cornerstones of the
organisation and managers are eager to invest to maintain or increase the safety
level.

In order to guarantee or at least maximise safety, petrochemical plants are shutdown

to maintain machines that cannot be properly inspected, repaired, replaced or
overhauled in the normal operating context. This strategy is known as Turnaround
maintenance (TAM) in petrochemical industry. Duffuaa and Ben Daya [16] worked
on formalising TAM practices and improvements. According to them TAM is a
periodic maintenance plan during which three types of maintenance work are
performed:

1. Work on equipment which cannot be done unless the whole plant is

shutdown
2. Work which can be done while operating but requires too much resources
(personnel or time).
3. Other noticed defects which couldn’t be repaired.

Nowadays the third point tends to be avoided during TAM period. The operating
loss generated by the shutdown time is so important that managers try to reduce it. It
comes together with the spirit of maintaining assets efficiently. It would be
nonsensical to reduce cost by optimising maintenance system and at the same time
extend shutdown periods.

TAM cost is, thus, reduced without being deleted. This shows once more that safety
is the most important challenge in that industry. Risks of major catastrophes are not
annihilated because it is impossible to have zero risk, but potential safety exposure is
reduced drastically thanks to TAM. However, as pointed out by Duffuaa and Ben

Andreï LAQUET 12
Maintenance evolution and scope Master Thesis

Daya [16], TAM is “a hazardous event” since the accident risk increases due to the
nature of the work. Therefore, on one hand it helps to enhance plant safety while on
the other hand it might cause accidents. It is management staff’s role to emphasise
safety importance during TAM. This will be developed further later on.

Delays in the implementation of corrective maintenance actions

The time between the equipment interruption and the corrective maintenance
execution is highly important and impacts the production. The figures of Table 1
depict the preventive and corrective maintenance proportion in terms of hours of
work for the maintenance staff and unproductive time. It highlights the idea that
planning is the key for efficiency.

Preventive Corrective
Total
Maintenance Maintenance

Proportion of time
85% 15% 100%
spent by workers

Proportion of
7% 66% 73%
unproductive time

Source: Case study 2005 by Aoudia et al. [14]

Table 1: Preventive and corrective maintenance repartition

In this case, corrective maintenance mobilises only 15% of the workers time but
represent 66% of the total unproductive time. Corrective maintenance produces more
than nine times the unproductive time of preventive maintenance with 5 times less
working hours. It emphasises that corrective maintenance is an expensive action to
perform because the lack of preparation introduces high costs due to urgent spare
parts order and overtime. However, it is important to note that in that case study the
preventive maintenance tasks represent nearly all the maintenance time, 85%. It is
important not to fall in the over preventing spirit which tackle productivity.

Number of accidents

A low accident rate is one of the biggest targets in refineries. Maintenance is one of
the greatest generators of accidents due to the variety of the job and the environment
it takes place. To avoid accidents to occur the most important aspect is to plan and
prepare carefully what has to be done and how it will be done. The preparation of

Andreï LAQUET 13
Maintenance evolution and scope Master Thesis

work is hardly perfect in corrective maintenance and people are often under pressure
due to urgent operating needs. This worsen working conditions and lead to
negligence because people involved can hardly step back and think of the best
method to proceed. In the case study, 51% of accidents of 2005 were due to human
errors. The only way to reduce this negligence is by better planning and less pressure
on the teams which is only possible with preventive maintenance or at least early
failure detection.

Cause Example Percent

Human Error Negligence, bad posture, tiredness... 51
Work authorisation, procedure of safety
equipment,
Procedure of chemical products, Procedure of
dismantling and assembly...

Procedure not respected 22

Protection means not used 12

Inadequate tools 4

Outdated equipments 4

Defect in equipment 4
design

Others 4

Source: Case study 2005 by Aoudia et al. [14]

Table 2: Accident cause distribution

In everyday maintenance work, accident rate is directly linked to pressure on
workers. The accident risk is even higher during shutdown periods induced by
TAM. Indeed, TAM brings an unusually high number of workers which work under
time pressure due to fixed deadline. The schedule must be respected because one
operating day loss is quickly translated into extremely high financial losses. A clear
target has to be settled to a zero incident policy with a safety plan including a
working routine with: Work permits, Task specifications, Material needed, Work
environment specification (substances related), Tools and equipment needed

The safety working routine is highly important in order to protect workers against
potential hazards. This type of safety plan brings a huge amount of administrative
tasks to fulfil and breaks maintenance productivity due to coordination problem,
Andreï LAQUET 14
Improvement methodologies Master Thesis

work stopped due to missing signature, etc. However, managers all agree that even if
the safety plans brings some unproductive points it cannot be infringed and
overstepped. These points accentuate the close relationship that maintenance has
with the cornerstones of a plant: safety, productivity and environment. An effective
maintenance program is needed and it can be designed and implemented through
well-known maintenance improvement methodologies.

3 Improvement methodologies
3.1 RCM and its limitations
The challenges that petrochemical industries face towards an efficient preventive
maintenance plan are now clearly identified. Different maintenance methodologies
have been described and the industry is now disposed to implement a way to
improve the maintenance efficiency. Nevertheless, methods like RCM are time
consuming and the benefits are not directly perceptible. Preventive maintenance
activities provide in essence long term benefits and it can be hard to convince the
actors to opt for a painful method during months without any positive outcome.
Conventional RCM approach has some limitations regarding its implementation in
refineries. These limitations have been pinpointed by Deepak and Jagathy in 2013
[17] to justify the need for a new model for RCM in petroleum refineries. The
limitations do not discredit the methodology. The implementation of the method
could be slightly modified in order to easier meet the industrial requirements.

The RCM approach is developed with a design approach which gives an accurate
picture of the situation but is hard to implement and need long analysis. In a
refinery, similar to the one in which our case study has been performed, the
important number of equipment entails too many time-consuming FMECA to be
performed. Deepak and Jagathy assessed 50000 failure modes to analyse taking into
account 2000 rotating machines. Our case study refinery has 1934 rotating machines,
of which 684 are pumps, so the approximation can be used in our case. Due to
limited manpower and time, refiners are not willing to apply the methodology as it
is. One way to deal with that issue is to rank equipment by criticality in order to
prioritise their study. This methodology is applicable to industries and is known as
Sub-Optimal RCM. However, this methodology has been criticised because it
concentrates on critical equipment which sometimes does not affect global reliability.

Andreï LAQUET 15
Improvement methodologies Master Thesis

The spirit of the methodology is to carry out Failure Mode and Effect Analysis on
critical equipment.

Another drawback expressed by the industry is also to the design-oriented

characteristic of the methodology. RCM is not an optimisation methodology. The
current plan is not taken into account when an analysis of a prevailing maintenance
program is performed. This might have a harmful effect on the staff because they
might feel that their current work is blamed for the insufficient reliability level. The
method can be implemented in a mature operating context but it is hard to make the
actors collaborate in order to build the entire structure from the start. Even if it might
the best technical solution it can be hard to motivate the actors which would lead to
an inaccurate work.

Due to these obstacles to RCM implementation several alternate approaches have

been developed in the oil and gas industry.

3.2 Streamlined RCM (SRCM)

Streamlined RCM is one of the alternate approaches. It is based on the RCM content
but takes the current maintenance program as the basis of the analysis. The method
aims to assign to each maintenance task the failure mode which has to be prevented
from. This method considers that the current maintenance plan together with the
failure database is already an efficient information provider on the potential failure
modes. It is even more valid in refineries which have been running for more than 50
years and operates continuously. The most frequent failure modes have already
occurred at least once in the whole range of equipment. After the failure mode
analysis, the last three questions of the RCM decision process are performed. These
questions are:

5. What are the consequences of each failure?

6. What should be done to prevent or predict the failure?

7. What should be done if a suitable proactive task cannot be found?

This step focuses on the potential consequences of the failure and on the best way to
keep the equipment away from it.

Moubray himself criticised streamlined RCM considering that it “focus on

maintenance workload reduction rather than plant performance improvement”. It is
a cost-optimised process comparing it to RCM but if applied carefully it could

Andreï LAQUET 16
Improvement methodologies Master Thesis

provide high improvements. The main focus that Moubray express is related to the
protection devices such as spare equipment installed. Considering the retroactive
approach it is true that there might be a gap in the maintenance program if every
failure mode is not analysed. Many of current maintenance policies do not consider
spare equipment as maintainable equipment. If so, the new maintenance program
would continue to spread the gap in this area and be either inexistent or, at least,
poorly sized. If used, Streamlined RCM must consider protective devices carefully.
They often result in no planned maintenance tasks for these machines which
discredit the method and make it “completely indefensible” [18]

3.3 Preventive maintenance optimisation (PMO)

Another alternative approach to RCM is Preventive maintenance Optimisation,
PMO.[2] Steve Turner formalised this methodology while working as a consultant for
Shell. PM Optimisation begins with a global review of the current maintenance
program. It “generates a list of failure modes from the current maintenance program,
an assessment of known failures and by scrutiny of technical documentation”. It is
said to proceed faster than RCM method because it does not analyse every failure
modes. RCM by analysing every failure modes sorts out the relevant ones but
analyses also a lot of insignificant failures. However as Moubray says leaving things
out inevitably increases risk and increases the risk of facing unanticipated failures.

Every method has its benefits and its drawbacks. The most important is to master the
subject by a proper review of what exists and what could have been done. As
explained above Streamlined RCM suffers from the lack of maintenance for
equipment and failure modes which do not have direct operational consequences:
the hidden failures. It is explained by the fact that SRCM starts with current
maintenance program. It is actually a problem when reviewing the maintenance
tasks but nothing prevents the person in charge of the analysis from adding these
types of failure modes to the future maintenance program. Moreover, when the
equipment is standard and well-known it is easier to find the gaps of an existing
program. In fact, PMO methodology includes that failure mode gap to the current
practices. The challenge is to fully complete the list in order to develop an accurate
program. And this is not easy to do when the whole range of failures are not covered.
These two methodologies are appealing to companies because they use the concrete
controls and the current program in order to carry out the analysis. It is more about
reviewing and improving the current maintenance program than redesigning it
which contributes to be easily accepted by maintenance staff.

Andreï LAQUET 17
Case study scope Master Thesis

4 Case study scope

4.1 Centrifugal pumps in the operating context
The case study focuses on the centrifugal pumps of the refinery. Centrifugal pumps
are the most common machines in refineries. “Pumping devices are always part of a
more or less complex system where pump failure can lead to severe consequences”.
That sentence extracted from the handbook on centrifugal pumps written by Johann
Friedrich Gülich [19] highlights the operational importance that centrifugal pumps
carry. Indeed, he compares pumping devices importance in an industrial
environment to the human heart because liquid transport function is vital to any
hydraulic system. In petroleum refineries, pumps are used to move a given fluid
which varies from crude oil to final products such as gasoil

The fluids and the operating context have different specificities that need to be taken
into account when choosing the right pump to be implemented. Indeed, the fluid
circulating in the centrifugal pump induces a large range of flow phenomena which
have a “profound impact on design and operation through the achieved efficiency,
the stability of the head-capacity characteristic, vibration, noise, component failure
due to fatigue, as well as material damage caused by cavitation, hydro-abrasive wear
or erosion corrosion.” [19]

These parameters have to be taken into account when the maintenance task selection
is determined. According to Gülich, life cycle costs and operation efficiency of the
machine highly rely on these parameters, their interaction and also on the
understanding of the relation between the pump and its operating context. Within
the same area of expertise, Azadeh, Ebrahimpour and Bavar [20] developed an entire
system to improve pump failure diagnosis through analysis of the parameter
interaction. They created a range of linguistic rules which “approximate human
reasoning” through If-Then rules. Inputs are the variation of the operating
parameters. Their work emphasises the importance of parameters interrelation.
Putting together all variations lead directly to a diagnosis. This is useful to get
directly to the most likely failure causes and might lead to lower delay of action by
providing conclusion faster.

Even if there are many centrifugal pumps in a refinery, they are nearly all different
due to different operating range and conditions. It can be hard to know where to

Andreï LAQUET 18
Case study scope Master Thesis

start with maintenance optimisation in big plants. Indeed, the review process has to
be carried for every single machine in order to be accurate and enhance its reliability.
One way to prioritise the work is commonly done through a criticality analysis.
Before developing the technical methods to evaluate equipment criticality level it is
important to clarify how this parameter influences the maintenance decision process.

Industrials want to maximise their equipment performances in order to be able to

operate as efficiently as possible. As mentioned before machine performances, from
availability to operating parameters, can impact on economical, safety health and
quality achievements of a company. However, resources are always limited and
managers must decide the right level of resources that can be allocated for the
maintenance optimisation. Therefore, the resource allocation choice is one of the
most important decisions that managers need to take. The decision must be built on
actual and quantifiable facts in order to be justified and take doubts away. The
criticality of a machine is an accurate decision guide to perform this choice.

Resources cover both tangible resources such as financial resources and intangible
resources like human resources. Applying RCM concept does not necessarily means
invest a lot of money but it needs significant start-up expenses in terms of human
resources.

Due to the number of pumps in the plant and for economical purpose the criticality
of each rotating machine has to be evaluated. Several approaches have been
developed in order to assess criticality to equipment.

4.2 Risk assessment

Torabi et al. [21] studied quantitative risk assessment in refineries and petrochemical
plants. Quantitative risk assessment is a way to estimate safety risks. It can be used,
for refineries, as a method to rank the economic impact of equipment or units in
order to prioritise maintenance activities. In that sense, the classification is used as a
decision model for the maintenance department. This methodology is already quite
common in aerospace and nuclear industries. This matches with the evolution of
maintenance generations and the fact that these two industries have been the one
boosting the maintenance progression.

The authors use the model developed for nuclear power plant in order to perform
safety assessment. However, the industries are quite different and the challenges and
available information are slightly distinct. Historical information is often not as

Andreï LAQUET 19
Case study scope Master Thesis

accurate in the oil and gas industry as in the nuclear sector. As in Bevilacqua et al.
study [1] fault tree logic is developed to link the production losses to component
failures. This methodology suits the purpose of linking a component failure to an
event. However the level of detail of the tree lies on the criticality of the machine. So
the criticality is implicitly taken into account but not evaluated as a proper ranking
system.

4.3 Criticality assessment methodologies

Other scientists developed criticality assessment method optimised, or at least
focused, on centrifugal pumps and other rotating machines. There are hardly ever
generalised, most of the time they are entirely experience-based which induces a
high subjective part into the analysis.

Qi et al. [22] And Gomez et al. [23] developed two criticality assessment approaches
which answer to the industrial need of ranking machines according to their financial,
safety and environment impact. The multi-criterion of Gomez et al. [23] encompasses
12 criteria ranked from 0 to 4 grouped in the three main categories:

 Potential risk for operators, operation and environment

 Maintenance parameters: failure detection, MTTR, cost, MTBF
 Operational parameter: operational alternative, utilisation, effect on other
elements

After evaluating all these parameters, the criticality is given by weighting criteria
that takes into account the operational importance of the machines. This assignment
of weight evaluation is hardly easy to do because people from production or
maintenance department have different vision of which challenges are the most
important. Weighting factors will, then, seldom coincide since they have different
point of view. The method faces again the same problem between balancing several
points of view in a criticality analysis. Moreover, the high number of criteria slows
down the process which makes it difficult to implement. However, the criterions
cover all the relevant information needed for a criticality level, from the impact that a
failure might have to the maintenance requirements and likelihood of failure.

Andreï LAQUET 20
Case study scope Master Thesis

Qi et al. [22] opted for a method with fewer criterions. Only three criteria, ranked
from 0 to 4, are used with fixed weighting criteria

 EHS: Environment, Health and Safety

 IOB: Impact On Business
 AMC: Annual Maintenance Cost

The impacts can be inaccurate for equipment with redundancy. Indeed, the loss of a
machine will be considered as not critical if it is doubled because the direct impact is
null. So every machine that has a standby pump would be considered as non critical
with this strategy. If the loss is considered for the function which is covered by the
pump and its stand by then they will be highly critical. The analysis requires a
method to take redundancies as an operating parameter and must be expressed and
not only hidden in the impact reflexion. For that purpose Gomez et al.’s
methodology was better, because it is explicitly one of the criteria, but their strategy
is highly time-consuming which may repel industrials.

4.4 Scientific background

The overview from maintenance concept and its evolution to the particular
maintenance strategies used in oil and gas industry gives an accurate framework of
the challenges and trade-offs that maintenance managers face. The petrochemical
industry has some specificity which must be taken into account for the maintenance
plan and organisation. Indeed, safety is the most important cornerstone of the
industry and maintenance, amongst other activities, is organised to lower safety
risks.

Scientists agree that maintenance strategies, for example RCM, are highly valuable
and induce strong improvements if applied correctly. Partial of prioritised strategies
are said to look for economical savings more than global reliability improvements.
However, the current industrial situation forces oil and gas companies to optimise
their strategy in order to allocate resources as clever and accurate as possible.

Criticality assessment is, thus, criticised for granting an asset more resources than
another. This choice will impact the reliability of other machines. Nevertheless,
industrial managers do not look for the concept of reliability itself. The main goal is
to maximise economical profitability in order to ensure the highest possible safety
level. These two targets are, of course, provided by a strong maintenance plan and
high reliability levels but not necessarily for every asset.

Andreï LAQUET 21
Maintenance of centrifugal pumps: a case study Master Thesis

5 Maintenance of centrifugal pumps: a

case study
5.1 Centrifugal pumps: a brief description
The case study focuses on the maintenance program of Total refinery of Donges in
France for the 684 centrifugal pumps operating. Centrifugal pumps aim to transform
rotational energy to hydraulic energy of the fluid flow. There are several types of
centrifugal pumps: overhung single stage pumps (see Figure 6 below), multi stage
ones, between bearings, etc. The principle remains the same. One impeller, or several
for multistage pumps, rotate and drag the fluid to the discharge. Several gaskets and
rings together with labyrinth and mechanical seals insure the sealing of the pump.
This is highly important due to hazardousness of the product in refineries.
Mechanical seal is a complex system on its own and will be described further.

Figure 6: Overhung single stage centrifugal pump

Andreï LAQUET 22
Maintenance of centrifugal pumps: a case study Master Thesis

The mechanical seal is an external device which prevents leakage between the pump
and the environment. It is an essential part of the pump especially with hazardous
product which, in case of leakage, can engender safety and environmental problems.

5.2 Description of the operating environment

In Appendix 1, a brief functional description of the refining operation is given. A
general framework of how refineries operate is sufficient to understand the
challenges and the environment which will impact the operating conditions.

As mentioned before centrifugal pumps are nearly everywhere in a refinery. The

chemical processes need the fluid circulation in order to generate interaction. It is
obvious that without any movement nothing can happen. Emphasis should be put
on the product variety that circulates through the different processes. The fluid
properties are highly important to select the machine and choose the right
mechanical seal to install.

5.3 Maintenance management organisation

Maintenance department is composed of several units. Every unit correspond to a
specific maintenance trade. Several companies are involved in the maintenance
process. This is highly important because maintenance efficiency relies on a good
communication between the actors. Appendix 2 illustrates the different actors
involved in pump maintenance. The maintenance organisation has to be formalised
and discussed to understand the process and where there is room for improvement.

Figure 7: Maintenance task organisation summarises the different tasks concerned by

the machine maintenance. There are three main maintenance processes: Corrective
maintenance, Proactive Maintenance and Root Cause Failure Analysis (RCFA).
RCFA is performed for recurrent failures in order to find the root cause and eliminate
the failure if possible or redesign the equipment. It is one of the most important
improvement conveyors of the current system. Nevertheless, it is carried as a default
action to prevent the failure from occurring again. As pointed out in the literature
review and in this case study, criticality is central in the maintenance decision
process. Therefore, the criticality ranking must be accurate and updated to provide
an efficient input.

Andreï LAQUET 23
Maintenance of centrifugal pumps: a case study Master Thesis

Corrective
Maintenance

Equipment Failure Prioritisation

RCFA
Criticality

Preventive Predictive
Maintenance Maintenance
Plan Plan
Control list and frequency

Spare parts stock level, overhaul interval, etc.

Figure 7: Maintenance task organisation

The maintenance task organisation provides an overview of what need to be
reviewed to evaluate maintenance performances. The methodology which will be
used in the case study is, among other points, to analyse the criticality ranking of
centrifugal pumps. The problems encountered by corrective maintenance in
refineries have been identified in the literature review and are also relevant in the
case study situation. Therefore, the corrective maintenance will be analysed focusing
on these identified problems. Then, proactive maintenance plan will, also, be
discussed using a combination of the maintenance methodologies depicted in the
literature review. Finally, an improvement hint is proposed for RCFA in order to
improve current RCFA outcomes but also early failure detection and, then, proactive
maintenance.

Andreï LAQUET 24
Criticality analysis Master Thesis

6 Criticality analysis
6.1 The current situation
Evaluate the criticality of the machines is not a new concept in Total organisation.
Actually each pump has its criticality assessed in the business management software,
SAP in that case. As mentioned in the first part of this paper criticality assessment is
an important part for maintenance managers. Indeed, it is often related to the
machine criticality that resources will, or will not, be allocated in order to improve its
reliability. In theory, reliability should not be linked to criticality but in practice
prioritisation is needed and it allocates resources according to criticality levels.
Criticality is central and linked to every maintenance task. The place of criticality
level in the organisation is highlighted in Figure 7 and Appendix 2.

The distribution of centrifugal pumps according to criticality level is represented by

Figure 8. It reports the current distribution of criticality level among centrifugal
pumps in the refinery. This distribution arises some problems concerning its use.
Indeed, on one hand, the criticality serves as a decision support which helps and
guides maintenance task prioritisation in every day work. On the other hand, more
than half of the equipment is in the medium/high criticality level. The distribution is
summarised by the following figures:

 7% high criticality level

 51% medium/high criticality level
 32% medium criticality level
 10% low criticality level

In most cases the machine ends up in the medium/high criticality category and the
prioritisation is impossible among each category, so in that case for more than half of
the machines. Such system must spread out the most critical machines into several
categories to be efficient.

Andreï LAQUET 25
Criticality analysis Master Thesis

SAP distribution
71 pumps 46 pumps
10% 7%

High
220 pumps
32% Medium/High
347 pumps
51%
Medium

Low

Source: Total SAP database

Figure 8: Distribution of centrifugal pumps according to criticality level

This distribution problem comes from the methodology that has been used to set the
criticality. It has been based entirely on personal points of view without any precise
quantification of impact. It explains why 83% of pumping machinery is classified in
medium or medium/high category. When a pump is seen as important the criticality
is set as high. However, high criticality induces great resources and can be criticised
by others so in general the criticality is brought down to the second level, the
medium-high level. By analogy, the low criticality level is unpopulated due to the
same human reasoning. Machines are seen as key features of the process by persons
who operate them and the criticality is, thus, hardly ever assessed as low.

The need for a new criticality assessment is motivated by the ineffectiveness of the
system caused by the distribution as explained above. The second reason why a new
criticality assessment needs to be carried out is the obsolescence of the current
classification. Processes have changed through the years and some machines even
stopped to operate. The update is needed to keep the coherence between reality and
decision making strategies. To reach a useful distribution the methodology has to be
clear and parameters have to be quantified.

Andreï LAQUET 26
Criticality analysis Master Thesis

6.2 The new approach

The new criticality assessment needs to be designed to tackle the problems identified.
The distribution problem has been diagnosed as coming from the assessment
methodology. There must be a logical framework which determines the criticality
upon parameters and not only human intuition.

A criticality level relies on the potential risk that equipment might engender. As
shown in the literature review the three parameters which matter the most in
industrial environment are: economical, safety and environmental. Considering that
quality is covered by the economical risks because quality problems imply sales
decreases or production loss. The fuzzy criticality assessment system gives a method
on how to evaluate the impact according to these criterions.

On one hand, the impact generated by losing the function fulfilled by the group of
machines is crucial to assess the criticality. On the other hand, in a refinery, the worst
case scenario often leads to high risks which would lead to assess a high criticality to
an important number of machines. A second parameter has to be set regarding the
likelihood of the functional loss. This second parameter has to take into account the
operating context which surrounds the equipment. One or a combination of
production system outputs has to be gathered in order to evaluate this parameter.

The criticality levels are defined by Total through the matrix below, Figure 9. The
boundaries have been changed for confidentiality purpose. The use of criticality
matrix is common in chemical industry. A complete inventory of petrochemicals
companies’ policies towards criticality assessment has been carried by the ministry of
sustainable development in 2004. [24] The economical levels are not given due to
confidentiality issues but it does not impact the comprehension of the method used.

Andreï LAQUET 27
Criticality analysis Master Thesis

Figure 9: Criticality matrix

Availability, output rate, maintainability, safety and profits have been identified by
Tsang, Figure 2, as the key parameters. Achieved availability covers downtimes of
the machines which impacts the probability that the undesired event occurs. In the
case study, the task which is fulfilled by the machine has a great importance on
criticality. Therefore, if a standby is installed then the two pumps must have the
same criticality level because the criticality will rely on the capacity of the functional
block, composed by the two machines, to avoid any undesirable event to occur. Key
inputs which must be taken in the criticality analysis are summarised in Figure 10.

Andreï LAQUET 28
Criticality analysis Master Thesis

Figure 10: Criticality assessment components

In Donges, a great number of machines are doubled or tripled to guarantee
production rate or safety. This key parameter needs to be involved in the analysis
with the maintenance historical data, MTTR, MTBF. Therefore, machines must be
conceived as a whole. The group of machines which fulfil a function must be
considered. The criticality is linked to the loss of a function with operational
consequences and not on a single machine breakdown. Thus, the availability of the
desirable function, either fulfilled by one machine or its spare, corresponds to the
spirit of the criticality matrix and is the key parameter which needs to be studied.

6.3 Calculation of the functional availability

The emphasis is put on the operating context and particularly on the machine
redundancies. The standard AFNOR X60-503 [24] based the availability calculation
on Markov chain theory and is adapted to several machine configurations. The
system considered in the availability analysis is not only the pump itself but the
group of pumps that are used to fulfil a certain function. It synthesises the
configuration, MTBF and MTTR of the functional group of machines and translates
these parameters into a statistic functional availability by using the formula.
Particular cases are shown, in Figure 11, to show the method.

Andreï LAQUET 29
Criticality analysis Master Thesis

Figure 11: Extract from criticality analysis

The criteria used to classify the level of availability echoes the criticality matrix of
Figure 9. The failure probability is ranked from 1 to 5. These boundaries are defined
by the production department considering the levels targeted by criticality matrix.
The matrix levels are then translated into availability levels which delimit the ranges.
This method put emphasis on configuration of the machine. It is crucial for this case
study because more than half of pumps do not operate alone. Most common case is
pump with a standby or two pumps sharing one common standby.

6.4 Evaluation of maximum impact

The evaluation of the impact as shown in the criticality matrix cannot be analytically
assessed. The experience gathered from several actors is the basis for a strong and
accurate risk evaluation. Members, with great experience, from process, maintenance
and operating departments meet to evaluate the potential risk. The target is to
evaluate the consequence of the worst case scenario that a functional loss can
engender.

6.5 Results
The combination of both criterions linked to the criticality matrix gives directly the
new criticality of the equipments. The new criticality method spreads out the most
critical equipment into several sectors. It is better for the use that the company target.
However, there are an important number of NA (Non Acceptable) criticality levels.

Andreï LAQUET 30
Criticality analysis Master Thesis

Results
39; 6%
73; 10%

321; 44%
166; 23%

125; 17%

Figure 12: Criticality distribution

6.6 Limits
The method of probability assessment has a great output and summarises the
situation clearly. However, MTTR of machines are set according to the power of the
machine. It is not a precise measurement but, as pointed out in the literature study, in
petroleum refineries, databases are not as precise as in nuclear power plants and it
was not possible to evaluate the real MTTR. Due to that problem MTTR are broadly
over evaluated and so is the criticality.

The risk assessment method induces also some limitations. They are the same than
the ones pointed out by Qi et al. [22]. Potential imprecision can be induced by
gathering personal viewpoint on a machine. Moreover, the method is not flexible and
the risk levels are integers which can be a problem if the risk is close to a boundary.
For example, if a breakdown is estimated to generate 0.95 M€ loss then it would be
consider as intermediate level, the same level as if it was 0.15M€ even if it is 7 times
greater. A ranking system with progressive scale and not only integers could be more
accurate to understand what is hidden behind the risk. Moreover, it would be useful
to be able to rank the machines within a same criticality level. By developing a more
flexible model this could also be done.

Andreï LAQUET 31
Corrective maintenance system Master Thesis

The final drawback which needs to be known by maintenance department comes

from the type of methodology itself. As criticised by Deepak and Jagathy [17],
criticality analyses are often static models. This model is also static, so evolutions and
improvements introduced after the evaluation are not taken into account since the
criticality is assessed at one specific point.

All these limits give credit to the persons who disregard the criticality assessment in
terms of reliability improvements. Nevertheless, due to the important number of
equipment it would be hardly possible to equally allocate resources to every
machine. For preventive maintenance purpose the criticality assessment should not
be the principal parameter to be used. However criticality is a powerful tool to
support task prioritisation and, thus, to manage emergencies.

7 Corrective maintenance system

7.1 Importance and issue
The evolution of maintenance depicted in the literature review highlights the
progressive change of status that corrective maintenance has been experiencing. It
passed from the only way of maintaining equipment to the practice that needs to be
avoided as often as possible. However, it is utopian to target a plant in which
unexpected breakdowns do not occur. Therefore, corrective maintenance will always
be part of the picture. Since the company needs to perform corrective tasks it is vital
to have an efficient management system supervising it. The principal problem comes
from delays in implementation of corrective tasks are widespread in the daily work.
Resources unavailability is generally the cause behind these delays. Machines are
often not freed for maintenance purposes and continue to operate even if a failure
has been identified. Unavailability of spare parts, due to high delivery time or
mistakes in warehouse stock level, is the second cause. New strategies must be
implemented to tackle this problem. Therefore, a prioritisation system and a spare
parts strategy have been designed to increase maintenance efficiency.

Task prioritisation is executed every day to analyse the risk and the urgency of the
situation. This component is a pure managerial issue and not technical. It transposes
the resource limitation dilemma to the daily work. The prioritisation methodology
will not be developed in this paper because it has been reviewed within the
company.

Andreï LAQUET 32
Corrective maintenance system Master Thesis

7.2 Spare levels according to criticality and common failure

modes
Another frequent cause of delays is spare parts availability. Indeed, delivery time can
be quite important depending on the part technical specifications. A tailored spare
parts strategy based on pump criticality level and components functions has been
developed to prevent emergency situations from occurring. This strategy must also
keep a cost optimised focus to be realistic and implementable.

The global stock level of the plant is too high to be kept. Indeed, high profits and old
habits led over the years to store an important number of components. The total
amount exclusively linked to pump components stored is 2.5M€. Even if the stock is
high some key components have been identified as missing so this indicates that,
apart from the stock level, there might be a mismatch between what is stored and
what should be stored.

In France, companies pay taxes according to the spare parts stock. So the higher the
stock level the more expensive the annual taxes. The implementation of stock
changes has to take that into account. Indeed, if every missing item is purchased at
once then the stock level would increase which would generate an important tax
increase.

The strategy is based upon experience of technicians and machine criticality. The
table summarising the strategy is given in Appendix 3. During revision (partial or
general), some components are replaced. Wear parts, are systematically changed and
the other ones are inspected and then replaced if necessary. When a piece is replaced
by a spare it is not always scraped. The repairable parts are repaired, if possible and
economically viable, and then stored again in the warehouse.

Wear parts category contains: wearing rings, coupling repair kit and gaskets. For
other machines, the “big” components, such as impeller and complete rotor, are
stored only for the most critical machines. Together with the criticality-based
strategy, an economy of scale strategy is also recommended. For example if there are
ten identical pumps in the refineries the stock levels will not just be summed. The ten
identical machines will be considered as if there were only 3 of them considering the
most critical ones as references. This logic aims to reduce cost without interfering in
the repair delay.

However, as mentioned in the operating context description, nearly every pump is

different in this refinery due to different operating requirements and products. So
Andreï LAQUET 33
Corrective maintenance system Master Thesis

even if the strategy is interesting in a cost perspective it only has an effect on pumps
with redundancies.

Figure 13 shows the stock evolution due to the new criticality levels and the new
strategy. The difference between what is stored and what need to be stored in order
to respect the specifications has to be explained. It is important to highlight, once
again, that the delta is given only to provide a numerical idea of the difference
between the stock levels.

1400

1200

1000

800

600
Current level

400 New level

200

Figure 13: Stock levels after regulation

The example of gaskets stock level is developed to illustrate the stock evolution. In
our case, the targeted stock of gasket is 1308 gaskets and there are 822 items in the
warehouse. However, purchasing 486 gaskets would not correspond to the purpose.
The items are all different and in that case 537 out of 822 should not be stored
whereas 1023 are missing. The items are specific for a machine and the technical
aspects are central even if it is a cost optimisation policy. In Appendix 3, the
distribution of stock evolution according to component type is shown

These results lead to a decrease of the spare parts by 31.1% which represents around
0.9M€. It is a great economical savings but it has to be carefully implemented because

Andreï LAQUET 34
Corrective maintenance system Master Thesis

this strategy might worsen the situation if not implemented carefully. Indeed, one
day of stoppage for the distillation unit generates to 1M€ of loss. Therefore, it is
obvious that savings and improvements can disappear in a flash.

7.3 Challenges and limitations

This strategy is tightly linked to the accuracy of the criticality analysis. The lack of
flexibility of the criticality assessment can have a huge impact on the storage decision
and once more it could have been useful to know where the machine is placed
amongst the machines of the same criticality. Spare parts levels are easy to imagine in
theory but in practice it can be hard and inaccurate to implement it. The importance
number of components treated, 21213 in total, might engender inaccuracies in the
stock level adjusting. Moreover, pumps are old and it is often difficult to find spare
parts because sometimes constructors do not even exist anymore. Thus, the spare
parts levels have to be carefully handled for these types of machines to avoid
removing irreplaceable components.

The stock optimisation needs new maintenance philosophy too. The stock level of
rotor, which includes impeller and shaft, highlights that need. In the current system,
shaft and impeller are nearly always available in warehouse. 65% of parts included in
the rotor category have spare stored in the warehouse. However, the new strategy
states that only levels 1 and 2, 39% of the pumps, must have spare rotor.
Maintenance technicians are used to have spare parts for these components.
Therefore, the working methods need to change. When a wearing is suspected an
order has to be considered because the delivery time is often important.

Components age has not been taken into account when designing this strategy.
Particular attention has to be put on components’ obsolescence. Indeed, some
components are not being produced anymore. Therefore, a parallel review and
criticality analysis of the components has to be carried out in order to identify which
are critical due to the market and not only the operating conditions.

To summarise this new strategy has to be progressively implemented to avoid a

great taxes increase and also to give time to technicians to be familiar with the new
habits. This raise of awareness and identification of components’ obsolescence
cannot be simply done in a single day and needs time, otherwise important
downtime would emerge and sink the benefits of the strategy.

Andreï LAQUET 35
Preventive maintenance: review and progress Master Thesis

8 Preventive maintenance: review and

progress
8.1 Methodology
Preventive maintenance method needs also to be reviewed. The maintenance plan
needs an overall review of what is done within the company in terms of preventive
tasks. The tasks are spread through several actors and that generates confusion and
sometimes holes in the controls that need to be filled in. Tuner in the PMO
methodology characterises that as the informal manners that need to be formalised in
order to master the overall maintenance process.

As depicted in the beginning several methodologies exist to review and design

maintenance programs. RCM and PMO are two common methodologies used in the
oil and gas industry. Due to limited time, an accelerated maintenance optimisation is
required for the case study. The limitations of using SRCM or PMO types of
methodologies have been explained in the theoretical background and will be the
basis for a hybrid method combining RCM outcomes in terms of hidden failure and
PMO efficiency in terms of time. This reasoning is only possible because we study
one type of machine which is well-known. The common failure modes have been
identified by researchers which allow us to complete holes left by PMO approach
especially on hidden failure.

The method is articulated over four major phases. First, a general overview of the
common failure modes identified in the art is gathered. Then, the current preventive
tasks are formalised through a PMO like approach and completing with the missing
failure modes. Finally, the current strategy will be criticised and improvement hints
will be proposed.

8.2 Failure mode analysis

Centrifugal pumps are well-known equipment and failure mode and effect analysis
has been already discussed by several scientists. Figure 17 is a summary of the
principal failure modes and their effect on the machine extracted from Mobley
maintenance handbook [26]. The list is not exhaustive but gives the principal
parameters which influence pump performances. The problems are the functional
failures and the causes are the failure modes. The important information of the table

Andreï LAQUET 36
Preventive maintenance: review and progress Master Thesis

is the variety that a single failure mode can induce. Problems have from 4 possible
causes up to 20 for vibrations the mean being 10 potential causes for a single
problem. It underlines what has been depicted in the literature study, the parameters
interaction is highly important in order to detect the functional failure. The
mechanical seal has also the same type of root cause failure analysis table.

In Mobley’s diagram only common and principal failure modes are listed. Hidden
failures are not mentioned here and the methodology used needs a special treatment
for them.

8.3 Current preventive plan

The controls carried on centrifugal pumps and mechanical seals are summarised in
Appendix 5. It can be noticed that all important parameters are followed however
there still is room for improvements for some measures.

 Discharge pressure control

Operators control it through the manometer they control that it is not null and that it
does not fluctuate “too much”. There is a lack of quantification in two ways. First,
admissible pressure fluctuation needs to be quantified in order to set an alarm level.
Second, operating range is often not known neither by the operator nor the
maintenance method staff. Machines are old and the operating conditions have
evolved and they are might have undergone a change of their working range. An
evaluation of the operating range needs to be implemented in order to control
machines use. Once the range is known a visual marker can be put on every
manometer so that the control is simple and fast.

 Vibration

The vibration can be induced by nearly every failure mode. In Mobley’s table 20 out
of 31 common failure modes provoke vibration. Operator control aims only to report
that a defect is already in the machine. The precision is low and it is more of a default
observation than a preventive control. However this control is easy to perform and
can be useful as it is carried out three times a day.

A finer control is performed monthly or every two months on critical machines. It is

not performed on every pump because the workload is already quite high. It is seen
as pointless to take more machines because the vibration measure would not be
analysed due to lack of resources. This control is finer and can detect a phenomenon

Andreï LAQUET 37
Preventive maintenance: review and progress Master Thesis

long time before it wears every mechanical part in the machine. Controls according
to criticality are useful in that case. Nevertheless, it would be useful to have at least
one vibration route for every machine so that if a problem is detected the vibration
route taken can be compared to one taken in good conditions.

 Temperature, leakage, air suction in the motor, noise level

All these controls are done by the operator thanks to his senses. They are not precise
and in some cases a finer measure can be performed if needed.

Leakage for mechanical seal is a functional feature so there must be one. When a
quench is installed (steam sweeping of the dirt) the flow rate must be 0.2 bar which
corresponds to a drop by drop leakage. Installing a manometer to clearly quantify
and adjust that rate would be easier than train all operators to sense when the
quench needs to be adjusted. This kind of improvement comes from the review of the
preventive controls by balancing experts’ point of view and operators’ manners of
controlling. It is important to give the suitable tools to operators to improve
parameters control.

 Missing controls

As expected by John Moubray theoretical assertions some hidden failures are forgot
in operators’ check list. Mobley’s tables do not list them neither because they are not
identified as key failure modes because they are safety devices which prevent a
common failure mode to happen.

There are two kinds of hidden failures which are identified as missing here, the
functional holes cleaning and the quench setting. The quench setting has already
been discussed in the previous paragraph. Concerning functional holes, in a pump
and mechanical seal design there are three main holes that needs to be clean to
operate properly.

o Pump drain hole to eliminate gas and water before starting the pump
o Air vent on bearing to equilibrate pressure in the bearings to avoid high
pressure to destroy the balls and the
o Mechanical seal drain oil collector which if clogged might stock water in
the cavity and then bring water in the mechanical seal which can induce
grafoil gaskets destruction.

Andreï LAQUET 38
Preventive maintenance: review and progress Master Thesis

The problem comes from the fact that most of the operators are not aware of these
small details that highly lower MTBF. Training and explanation is needed in order to
implement correctly the cleaning of these holes.

Other missing actions are listed in green in Appendix 5. They have lower impact on
pump breakdowns so they are not discussed in the core of the paper even if they are
proposed as improvement controls.

8.4 Conclusion and improvement hints

Current operators check list covers nearly every parameter. As pointed out above,
there are rooms for improvements concerning the accuracy of the measure. It is
hardly possible to early detect a failure thanks to human senses because environment
factors, such as weather, modify human perception. The improvement hints have
been discussed in the previous part for the missing controls and the improvements of
the current system. However more than the parameter itself we have to go back to
what creates this phenomenon: the sporadic change in process.

In order to comprehend and fully optimise the pump functioning all operating
parameter are useful to follow. However the methodology used suffer from one
major gap, it does not take into account the evolution of a machine through the time.
If a parameter has been detected as out of range, for example a drop of discharge
pressure, and then came back to normal then it will be simply ignored and
considered as solved. This is one of the major inconvenient that determines why
FMCA is difficult to carry out and needs a long period of time and several failures in
order to identify the cause. The interaction of parameters, as pointed out in several
pump scientific studies [19] [20], is highly valuable for pump failure diagnosis.
However the problems often come from operational and short phenomenon which
cannot be seen by measuring only once per shift. Therefore continuous monitoring
finds its full interest in these cases but it is impossible to implement continuous
monitoring for more than a thousand machines so another proposal is explained.

Andreï LAQUET 39
Preventive maintenance: review and progress Master Thesis

8.5 Vibration and temperature continuous monitoring:

BEACON sensor
8.5.1 Interest
Continuous monitoring brings potential advantages to the current system. As
expressed above, the process conditions are highly harmful for centrifugal pumps
and the changes in process highly affect pumps reliability. The transient states
induced by production changes and the collateral effects on pumps are not mastered.
They must be known so that machines can be design to tolerate these conditions.
Today, the strategy used for recurrent problems is to develop a cause analysis
through Ishikawa diagram in order to find why the failure appears and how to
reduce the risk. This way of proceeding is not a preventive approach but it has
shown that there is a lack in the follow-up of equipment and it is hardly possible to
determine the origin of a failure. It is after several failures that the cause arises
because parameters vary from a period to another and the common factor is in
general the cause. The challenge is to have a maximum of information about the
time range during which the unexpected has come.

Continuous monitoring is a way to gather a maximum of information. However, it is

not economically viable to apply this strategy to every machine. The operating
condition and the relationship between mechanical problems and process
phenomena do not need precise measures of vibration or temperature. On a first
stage it is only to link which, if one, process episode might have engendered the
problem. The proposal is a sensor that is not a real continuous monitoring but it
provides in our case the desired benefits by taking measures with fixed time interval.
The beacon sensor produced by Flowserve is one of these kinds of sensor.

8.5.2 Description of the technology

The sensor is attachable to any pump it requires only a threaded hole to be
maintained. It measures both temperature and vibration with fixed time interval
between measures. It is not a continuous monitoring but the time interval is set in
order to be able to detect brief operating problem. A removable memory is plugged
in the sensor so that data can be extracted to be analysed. The sensor is adapted to
industrial use because it does not generate overload, data is analysed only when the
LED becomes red which means that temperature or vibration, boundaries have been
overstepped. The light turns red if the pump is out of boundaries and flashes green
and red if the machine operates in normal conditions but has experienced an out of

Andreï LAQUET 40
Preventive maintenance: review and progress Master Thesis

range period. The control can be added to the operator list and then it works with the
same logic than all other control.

Figure 14: Vibration and temperature monitoring sensor

8.5.3 Implementation challenges

The technology suits the purpose of our study but it has to be carefully implemented
in order to be trusted by the users. The LED makes it easy to implement because
there is no need for training operators which is costly and can be incomplete
sometimes due to the great number of operator in the plant. It is in line with lean
management which aims to facilitate and make things logic and normal. The
parameters and the boundaries, which will indicate when the LED has to turn red,
have to be well defined form the beginning. If the light turns red without any reason
the sensor might seem pointless in workers’ minds. It is important to avoid that so
that red lights continue to be identified as a problem and not something normal and
common.

Information provided by the sensor can be added to the current manners of detecting
failure cause in particular cases. However, the real target is to detect the failure
before it occurs. It is possible if the organisation is formalised, mastered and driven
by managers’ commitment because the analysis of a red light event has to be done
quickly after the arrival to avoid losing information on the process conditions.

Andreï LAQUET 41
Conclusion and further studies Master Thesis

9 Conclusion and further studies

The review of the maintenance plan for centrifugal pumps in the refinery has pointed
out some breaches. Corrective maintenance is still an emergency action which has to
be performed. Scientific background and experience have shown that criticality
ranking is necessary to structure the process. Criticality has been used to design a
tailored spare parts strategy which will induce major economical savings. However,
components’ obsolescence must be considered before implementing this strategy. A
further work would be to gather pump models installed in the refinery and
determine, in relation with manufacturers, which components are not manufactured
anymore. It is crucial to anticipate failure which might lead to high delays or even to
pump substitution if the component cannot be replaced. This is a central step to
complete the mastery of the current maintenance plan in terms of cost and efficiency.

Based on scientists’ previous study of pumps common failure modes, current

preventive plans have been reviewed and completed. This study is done for a general
case but every machine has its own specific challenges so, a proper RCM study has to
be performed for each machine to ensure reliability. Action plans must be tailored to
specific working environment. Important operating parameters are followed but the
control precision is too often based on operator senses which bring uncertainty and,
thus, failure observation comes often too late to prevent from spread wear in the
machine. Scientific findings highlighted that the link between production condition
and mechanical wear highly impacts centrifugal pumps lifetime. Putting in parallel
both components is too often impossible. A cost effective sensor has been proposed
to fill in this breach.

Further study could be to carry proper FMECA on critical machines especially on the
one categorised as Non Acceptable. Criticality assessment is static so it must be
updated to be effective. Either the criticality must be reviewed periodically or a
dynamic model has to be developed to solve the problem.

Research findings raised great improvement ideas and this paper highlights the
potential improvement hints possible to put in practice for maintenance plans even
in mature plants. However, it is the economical context and safety policy which drive
companies’ strategy. Equipment reliability itself is not the target. Therefore, money
and time will be allocated for reliability improvement only if the gain is nearly
guaranteed which is difficult to demonstrate in most preventive actions.

Andreï LAQUET 42
References Master Thesis

10 References
[1] Bevilacqua Maurizio, Braglia Marcello, Montanari Roberto. “The classification
and regression tree approach to pump failure rate analysis”; Reliability Engineering
System and Safety, 79, pp. 59-67; 2003

[2] Turner Steve. “PM Optimisation – Maintenance analysis of the future”; ICOMS
Annual conference Melbourne; 2001

[3] Moubray John. “RCM II Reliability-Centered Maintenance”; Industrial Press Inc.,

2001 (Book)

[4] Geraerds W.M.J.. “Towards a Theory of Maintenance”; The English University

Press London. 1972 (Book)

[5] Oke Sunday Ayoola. “An analytical model for the optimisation of maintenance
profitability”; International Journal of Productivity and Performance Management,
Vol.54 No. 2, pp 113-36; 2005

[6] Noyes D., Peres F. « Analyse des systèmes – Sureté de fonctionnement »,

Techniques de l’ingénieur, 2007.

[7] Zille Valérie. « Modélisation et évaluation des stratégies de maintenance

complexes sur des systèmes multi-composants », Thèse de doctorat EDF, 2009

[8] Mikler Jerzy, Frangoudis Constantinos, Lindberg Bengt. « On a systematic

approach to development of maintenance plans for production equipment », Journal
of Machine Engineering, Vol. 11, No 1-2, 2011

[9] Tsang Albert H.C. “Strategic dimensions of maintenance management”, Journal

of Quality in Maintenance Engineering, Vol.8 No. 1, pp. 7-39; 2002

[10] Jardine Andrew K.S., Tsang Albert H.C. “Maintenance, Replacement, and
Reliability: Theory and Applications, Second Edition”; CRC Press; 2005

[11] Ben-Daya Mohamed. “You may need RCM to enhance TPM implementation”;
Journal of Quality in Maintenance Engineering, Vol.6 No.2 pp. 82-85, 2000.

Andreï LAQUET 43
References Master Thesis

[12] Nowlan Stanley, Heap Howard. “Reliability-Centered Maintenance”; United

States Department of Defence; 1978

[13] Nakajima Seiichi. “Introduction to TPM”; Productivity Pr, Eleventh Printing

edition, 1988

[14] Aoudia Mouloud, Belmokhtar Oumhani, Zwingelstein Gilles. “Economic impact

of maintenance management ineffectiveness of an oil and gas company” Journal of
Quality in Maintenance Engineering, Vol.14 No. 3; 2008

[15] Arts R.H.P.M., Knapp Gerald M., Mann Jr Lawrence, "Some aspects of measuring
maintenance performance in the process industry", Journal of Quality in
Maintenance Engineering, Vol. 4, pp. 6 – 11; 1998

[16] Duffuaa Salih O., Ben Daya Mohamed. “Turnaround maintenance in

petrochemical industry” 2004

[17] Deepak Prabhakar P., Jagathy Raj V. P. “A New Model For Reliability Centered
Maintenance In Petroleum Refineries”; International Journal of Scientific &
Technology Research V. 2; 2013

[18] Moubray John. “The Case against Streamlined Reliability Centered

Maintenance”; Aladon, UK; 2000 (Online Article)

[19] Gülich Johann Friedrich. “Centrifugal Pumps”; Springer-Verlag Berlin

Heidelberg; 2014 (Book)

[20] Azadeh A., Ebrahimipour V., Bavar P. “A fuzzy inference system for pump
failure diagnosis to improve maintenance process: The case of a petrochemical
industry”; Elsevier; 2009

[21] Torabi KK, Karimi B., Parmar R., Oliverio M., Dinnie K.“Quantitative risk
assessment for process design modification and maintenance optimization in
refineries and petrochemical plants”; The Canadian journal of chemical engineering,
V. 84; 2006

[22] Qi H.S., Alzaabi R.N., Wood A.S. and Jani M. M. "A fuzzy criticality assessment
system of process equipment for optimised maintenance management"; International
Journal of Computer Integrated Manufacturing, pp. 112-125; 2015

Andreï LAQUET 44
References Master Thesis

[23] Gomez de Leon Hijes Felix C., Cartagena Jose Javier Ruiz. “Maintenance strategy
based on a multicriterion classification of equipments”; Reliability Engineering
System and Safety, 91, pp. 444-451; 2006

[24] Merad M.M. « Analyse de l’état de l’Art sur les grilles de criticité » ; Ministère de
l’Ecologie et du Développement Durable ; 2004

[25] AFNOR Standard X60-503 « Introduction à la disponibilité »

[26] Mobley, R. Keith. “Root Cause Failure Analysis”; Butterworth-Heinemann; 1999

(Book)

Andreï LAQUET 45
Appendix 1: Refineries general process description Master Thesis

11 Appendixes
11.1 Appendix 1: Refineries general process description
The chart below is a simplified chart of refining processes. It is not particularly describing the
case study plant.

Source: ExxonMobil

Figure 15: Refining processes: a simplified chart

Andreï LAQUET 46
Appendix 1: Refineries general process description Master Thesis

 Functional description of the principal units by AFPM (American fuel and

petrochemicals manufacturers):

Distillation: First stage of the reining process. The crude oil is boiled and condensed
again to fractionate the different hydrocarbon components by using the difference
between boiling points. Heavier components will be collected in the lower part of the
column and lighter ones, such as Liquefied Petroleum Gas (LPG), in the lower part.

Fluid Catalytic Cracking Unit: uses heat and catalyst to break or “crack” large gas
oil molecules into a range of smaller ones.

Hydrotreating: Hydrogen is used to remove part of the sulphur.

Reformer Unit: The molecular structures of crude and coker naphthas are
transformed using heat, catalyst and moderate pressure. It aims to produce a high
octane primary gasoline blend stock called reformate.

Alkylation Unit: acid catalyst is combined with small molecules to produce larger
ones collectively called alkylate. Alkylate has a high octane and is the cleanest
burning of the gasoline blendstocks.

 Economical impact are given below for the refinery of Donges

Economical impact

Catastrophic Major Intermediate

Distillation stoppage > 4 days >1 day

FCC stoppage >14 days >2 days

Hydrotreating 1 stoppage >25 days >2.5 days

HD2 >30 days >3 days

Catalytic Reforming
>30 days >3 days
stoppage

Table 3: Economical impact of downtime

Andreï LAQUET 47
Appendix 2: Maintenance actors and criticality scope diagram Master Thesis

11.2 Appendix 2: Maintenance actors and criticality scope diagram

Figure 16: Maintenance actors and criticality scope diagram

Andreï LAQUET 48
Appendix 3: Spare parts strategy and results Master Thesis

11.3 Appendix 3: Spare parts strategy and results

C1 C2A C2B C2C C3

Centrifugal pumps
N/A Criticality 1 Criticality 2 Criticality 3 Criticality 4
Complete spare pump (installed or in warehouse) 1 1 0 0 0
Rotor Shaft + Impeller 0 0 1 0 0
Wearing rings Casing set and Impeller set 2 2 1 1 0
Bearings Antifriction – Radial/thrust set or Sleeve/ Pad type 1 1 1 1 0
Oil Rings (if any) set 1 1 1 1 0

Balancing drum and liner / balancing disk and counter disk (if applicable) 1 1 0 0 0

Shaft sleeve 2 2 1 1 0

Throat bushing(s) (static and rotating, including central bushing if any 1 1 1 1 0

Packing rings – if any 3 3 1 1 0

Gaskets: pump casing, flat, O-ring, Bearing Housing, stuffing box, etc. 2 2 1 1 1

Flexible type (diaphragms and bolts, rubber elements) 1 1 1 1 0

Coupling

Repair kit (flexible elements + screws+ bolts) 2 2 0 0 1

Minimum flow valve (Schroeder / Schroedhal) complete 1 1 1 0 0

Table 4: Spare part strategy according to criticality
Andreï LAQUET 49
Appendix 3: Spare parts strategy and results Master Thesis

Type Missing parts Exceeding parts Current level New level

Wearing rings 564 197 389 756

Gaskets 1053 598 826 1281
Packing rings 122 247 270 145
Bearing 139 67 126 198
Throat Bushing 89 32 69 126
Rotor (Impeller, shaft) 78 390 476 164
Shaft sleeve 23 13 42 52
Coupling-Metastream 51 42 81 90
Coupling-Other 22 20 27 29
Minimum flow valve 29 3 4 30
Coupling-Miniflex 2 0 0 2
Complete Pump 20 57 43 6
Coupling-Flector 0 0 0 0
Coupling-Flexident 1 8 14 7
Coupling-Citroën 20 32 58 46
Coupling-feedback 3 0 0 3
TOTAL 2216 1706 2425 2935
Table 5: Economical returns of the strategy

Andreï LAQUET 50
Appendix 4: Mobley’s root cause failure analysis Master Thesis

11.4 Appendix 4: Mobley’s root cause failure analysis

Figure 17: Common failure modes of centrifugal pumps

Andreï LAQUET 51
Appendix 4: Mobley’s root cause failure analysis Master Thesis

Figure 18: Common failure modes of mechanical seals

Andreï LAQUET 52
Appendix 5: List of control for pump, mechanical seal and motor Master Thesis

11.5 Appendix 5: List of control for pump, mechanical seal and motor

N° Action Interval Responsible Cause

Bended shaft, casing distorted, cavitation, hydraulic instability, suction volume
1 Bearing temperature control Shift Operator too low, mechanical wear or defect, bearing default, misalignment, Total Suction
Head too high or low
Mechanical seal temperature
2 Every shift Operator Mechanical seal defect
control
3 Cooling system control Every shift Operator Clogged cooling pipes
4 Oil leakage Every shift Operator Mechanical seal defect, gasket design problem
5 Vibration (human sense) Every shift Operator Nearly every defect.
4 weeks, 8 weeks
6 Vibration route (sensors) Eiffel Nearly every defect. Analysis of the route to establish cause.
or never (criticality)
7 Bearing greasing 6 months Greaser Avoid bearing dry friction
Colour : if dirty (with particles) important wearing acceleration
8 Oil level and colour control Every shift Operator
Level : lower occurrence of dry friction
9 Oil replenishment Weekly round Greaser Ensure that the machine operate without dry friction
10 Oil change 6 months Eiffel Eliminate potential erosive particles
11 Air suction in the motor Every shift Operator Motor cooling clogged, motor trips
12 Motor running Continuous Control room Motor trips
13 Noise level Every shift Operator Cavitation, Air or gas entrained, hydraulic instability
Clogged impeller, Cavitation, Air or gas entrained , NPSH too low, suction
14 Discharge pressure Every shift Operator
volume problem, internal wear, leakage

Andreï LAQUET 53
Appendix 5: List of control for pump, mechanical seal and motor Master Thesis

If low suction
15 Strainer control Suction volume problem
pressure
16 Gas pressure in accumulator Weekly round Greaser Pressure in envelope seal too low
17 Programmed revision 5 or 10 years / Turnaround maintenance

Missing controls
18 Quench flow control Every shift Operator Seal clogged
To be determined
Unblock or control cooling
19 upon machine Increase of temperature
pipes
context
Cleaning of functional holes Mechanical seal drain: water in the seal (destruction);
Operator/
20 (pump and seal drain, air Every shift/Weekly Pump drain: liquid in the pump (increased wearing or impeller destruction)
Greaser
vent) Air vent, pressure too high in the bearing (ball wear increase)
Fin cleaning for gas
21 Every year Eiffel Defect in the cooling system of pressurised seals
pressurised seals

Table 6: Controls carried out for centrifugal pumps maintenance

Andreï LAQUET 54