Condition Monitoring Example

Title:
Delivered:
Condition Monitoring and Opportunity

Maintenance in Offshore Operations
Student:
11.6.2010
Availability:
Open
Number of pages:
109
Jon Inge Kristoffersen

Abstract:
This master thesis includes a discussion on technical and organisational aspects that should be considered if a
shift to an OBM scheme is desired. Organisational aspects include agility and resilience. The technical aspects
include Onshore Support Centre (OSC), prepositioning of spares and tools and standard operating procedures.
A Failure Mode Effects and Criticality Analysis (FMECA) have been performed on a simplified oil and gas process
line. Based on this FMECA critical equipment was identified and CM methods have been suggested.
With the help of Genetic Algorithms (GA) six different OBM schemes have been proposed. The length of the
opportunity was varied in the six different scenarios. The OBM model produced the lowest costs when the
opportunity was long. The minimum costs increased as the length of the opportunity decreased.
Lastly areas where OBM may influence the logistics planning have been discussed and ways to mitigate these
areas of impact have been suggested. Investment in a fully automated warehouse is one such measure.
Keyword:
Condition Monitoring, Opportunity Based
Maintenance
Advisor:
Magnus Rasmussen
Faculty of Engineering Science and Technology

Department of Marine Technology
MASTER THESIS
for
M.Sc. student Jon Inge Kristoffersen
Department of Marine Technology
Spring 2010
Condition Monitoring and Opportunity Based Maintenance in Offshore

Operations.
(Tilstandskontroll og opportunity basert vedlikehold for operasjoner offshore.)
Offshore operations are often organised with a division of responsibility between three major
entities; drilling and well construction, reservoir and production management, and operations and
maintenance. Offshore operations are dependent on a common logistics supply support, and
utilisation of Condition Monitoring (CM) for critical equipment together with Opportunity Based
Maintenance will enable more efficient utilisation of already limited logistics resources. Within the
Center for Integrated Operations in the Petroleum Industry (IOCenter) there is an interest for
developing CM methods together with collaborative maintenance and logistics planning schemes to
improve the economic and HSE (Health, Safety and Environment) levels associated with offshore
operations.
The M.Sc. thesis therefore includes the following tasks:
1. CM information and Opportunity Based Maintenance (OBM):
a. Discuss technical and organisational aspects that need to be considered before
equipment can be a candidate for OBM.
b. Identify/propose ways to prioritise equipment considered for OBM.
c. Describe how Genetic Algorithms (GA) can be used for optimal organisation of OBM.
2. CM methods:
a. Describe functionality and layout of a topside O&G processing plant.
b. Choose one main equipment or part of a system as a case, and perform a FMECA and
RCM to identify critical components that should be considered for Condition
Monitoring.
c. Describe the applicable CM methods for the chosen equipment.
d. Based on Point 1, propose an OBM scheme for the most critical components for the
chosen equipment.
ii
3. OBM and logistics planning:

a. Describe how logistics planning relates to maintenance of O&G equipment today.
b. Discuss how OBM will influence logistics and how OBM should be related to logistics
planning.
The work should be carried out in close cooperation with MARINTEK and the IO Center
program. Contact person at MARINTEK is Torgeir Brurok
The thesis must be written like a research report, with an abstract, conclusions, contents list,
reference list, etc.
During preparation of the thesis it is important that the candidate emphasizes easily understood and
well written text. For ease of reading, the thesis should contain adequate references at appropriate
places to related text, tables and figures. On evaluation, a lot of weight is put on thorough
preparation of results, their clear presentation in the form of tables and/or graphs, and on
comprehensive discussion.
Three paper copies of the thesis are required. A CD with complete report should also be delivered to
the department. One of the paper copies and a CD should be delivered to MARINTEK by the
candidate.
Starting date: 18th January 2010
Completion date: 11th June 2010
Handed in:
Trondheim 18th January 2010.
Magnus Rasmussen
Professor
iii
Master Thesis
Condition Monitoring and Opportunity Based Maintenance

in Offshore Operations
6/11/2010
NTNU
Preface
This report is the accumulated results of the work put into the Master Thesis course TMR 4905 at
Norwegian University of Science and Technology (NTNU) in the spring of 2010. Working on this thesis
has been challenging as well as rewarding.
People that deserve a thank you for their help, patience and time include Kenneth Juul and Tor Ole
Bang Steinsvik at ABB Oil, Gas and Petrochemicals department for Integrated Operations for
information regarding HXAM and DriveMonitor. Research scientist at SINTEF MARINTEK and IO
Centre contact Torgeir Brurok for his time and help. My advisor Professor Magnus Rasmussen at
NTNU for his time, input and help when it was needed. Professor II Tom Anders Thorstensen at NTNU
for his help and assistance with Manifer. Ralph Hansen and Ida Kastrud at TOTAL E&P Norge for their
help and input regarding equipment types in analyses performed in this thesis. Lastly I would like to
thank my fellow students for their help, patience and taking the time from their own thesiss to listen
to some of my ideas.
Trondheim, 11.6.2010
_____________________________
iii
Summary
In this M.Sc. thesis organisational and technical aspects related to the implementation of an
Opportunity Based Maintenance (OBM) scheme for Oil and Gas facilities have been discussed.
Organisational aspects of importance include agility and resilience. Agility can be understood as the
ability of organisation to benefit from unexpected events. Resilience can be understood as damage
control. Logistics support elements from the space industry can also be beneficial to implement in
the logistics support of Oil and Gas facilities. Elements from the space industry include the
implementation of Onshore Support Centre (OSC), prepositioning of spares and tools and standard
procedures for critical equipment.
A simplified oil and gas production system has been used as a basis for a Failure Mode Effects and
Criticality Analysis (FMECA), identifying critical components that should be considered for Condition
Monitoring (CM). The results of the FMECA were that centrifugal compressors and shell and tube
heat exchangers were ranked as critical equipment. Gate valves were also ranked as critical but have
been neglected in the further analysis due to heat exchangers and compressors being more critical.
Several different CM methods have been described, emphasis have been put on detection method,
application and words of warning when implementing the method. CM methods relevant to the
critical equipment have also been discussed. For compressors the most relevant CM methods
includes Vibration monitoring, Power and Load monitoring. In addition to these general methods
special software applications are also available through different distributors. For monitoring heat
exchangers internal inspection is most commonly used. The inspection interval can be determined by
a Risk Based Inspection (RBI) scheme. When an inspection is due Eddy Current Testing (ECT) or gas
leak tests can be used to determine diminishing pipe thickness in the tube section of the heat
exchanger. HXAM supplied by ABB is also an interesting solution offering real time measurement of
the performance of the heat exchangers.
Lastly how logistics planning is related to maintenance of oil and gas facilities today have been
discussed. How a shift to OBM will impact the logistics planning have also been discussed. Means
that can reduce any negative impact that a shift to OBM may have has also been discussed. One such
measure is the investment in a fully automated warehouse. Also lessons that can be learnt from the
logistics support in the space industry have been emphasised again.
Contents
Preface.....................................................................................................................................................iii
Summary ..................................................................................................................................................v
Table of Figures ..................................................................................................................................... viii
Table of Tables......................................................................................................................................... ix
1
Introduction..................................................................................................................................... 1
CM information and Opportunity Based Maintenance (OBM) ....................................................... 3
2.1
OBM Considerations................................................................................................................ 3
2.2
Prioritising equipment for OBM .............................................................................................. 4
2.3
Organising OBM with GA ......................................................................................................... 7
Oil and gas production and CM methods ...................................................................................... 11

3.1
System layout ........................................................................................................................ 11
3.1.1
Separators ..................................................................................................................... 11
3.1.2
Coalescer ....................................................................................................................... 12
3.1.3
Heat exchangers ............................................................................................................ 12
3.1.4
Scrubbers ....................................................................................................................... 13
3.1.5
Compressors .................................................................................................................. 14
3.1.6
Valves and other control equipment............................................................................. 14
3.2
FMECA ................................................................................................................................... 15
3.2.1
Functional hierarchy ...................................................................................................... 16
3.2.2
Definitions ..................................................................................................................... 17
3.2.3
Failure rates and types .................................................................................................. 18
3.2.4
Results and comments .................................................................................................. 19
3.3
CM methods .......................................................................................................................... 23
3.3.1
Acoustic emissions and Ultrasonics monitoring ............................................................ 24
3.3.2
Colour monitoring ......................................................................................................... 25
3.3.3
Corrosion monitoring .................................................................................................... 25
3.3.4
Electromagnetic testing................................................................................................. 26
3.3.5
Level, leakage & flow monitoring .................................................................................. 26
3.3.6
Load monitoring ............................................................................................................ 27
3.3.7
Magnetic Particle Inspection ......................................................................................... 27
3.3.8
Noise & acoustics monitoring........................................................................................ 28

vii
3.3.9
Oil analysis ..................................................................................................................... 28
3.3.10
Position monitoring ....................................................................................................... 29
3.3.11
Power monitoring.......................................................................................................... 29
3.3.12
Smell & taste monitoring............................................................................................... 31
3.3.13
Temperature monitoring ............................................................................................... 31
3.3.14
Thermography ............................................................................................................... 32
3.3.15
Vibration monitoring ..................................................................................................... 32
3.3.16
Visual Inspection............................................................................................................ 33
3.3.17
Wear debris analysis...................................................................................................... 34
3.3.18
Applicable CM methods ................................................................................................ 36
3.4
OBM simulations using GA .................................................................................................... 38
3.4.1
Costs included in the model .......................................................................................... 38
3.4.2
Developing the model ................................................................................................... 40
3.4.3
Results and comments .................................................................................................. 41
OBM and logistics planning ........................................................................................................... 57

4.1
Logistics planning and maintenance today ........................................................................... 57
4.2
OBM influence on logistics planning ..................................................................................... 61
Conclusions.................................................................................................................................... 65
Future work ................................................................................................................................... 67
Works Cited ................................................................................................................................... 69
Appendix................................................................................................................................................... I
Appendix 1: Failure database ............................................................................................................... I
Appendix 2: FMECA Report ................................................................................................................. V
Appendix 3: Risk matrices for sub functions ................................................................................... XI
Appendix 4: CD with OBM model. .................................................................................................... XVI
Appendix 5: OBM model ................................................................................................................. XVII
Appendix 6: OBM answer reports .................................................................................................. XVIII
Table of Figures
Figure 1 - Example of RUL predictions .................................................................................................... 6
Figure 2 - OBM flowchart ........................................................................................................................ 7
Figure 3 - Crossover ................................................................................................................................. 8
Figure 4 - GA flow .................................................................................................................................... 8
viii
Figure 5 - System layout ........................................................................................................................ 11

Figure 6 Separator (Devold, 2009)...................................................................................................... 12
Figure 7 - Plate heat exchanger (Devold, 2009) .................................................................................... 13
Figure 8 - Shell and tube heat exchanger (Devold, 2009) ..................................................................... 13
Figure 9 - TEG scrubber (Devold, 2009)................................................................................................. 14
Figure 10 - Ball valve.............................................................................................................................. 15
Figure 11 - RCM process ........................................................................................................................ 16
Figure 12 - Functional hierarchy ............................................................................................................ 17
Figure 13 - Functional hierarchy from Manifer ..................................................................................... 20
Figure 14 - Risk matrix: Costs ................................................................................................................ 21
Figure 15 - Risk matrix: Environment .................................................................................................... 22
Figure 16 - Risk matrix: Safety ............................................................................................................... 22
Figure 17 - ALARP principle ................................................................................................................... 23
Figure 18 - Visual spectrum in nm ......................................................................................................... 25
Figure 19 - Corrosion monitor by (Roxar).............................................................................................. 26
Figure 20 - Handheld sound meter........................................................................................................ 28
Figure 21 - Example of performance monitoring .................................................................................. 30
Figure 22 - Temperature meter TF202 by ABB...................................................................................... 31
Figure 23 - Handheld IR camera ............................................................................................................ 32
Figure 24 - Example of vibration monitoring ........................................................................................ 33
Figure 25 - Example of flexible borescope ............................................................................................ 34
Figure 26 Key policy rate development and prediction by Norges Bank (Norges Bank).................... 40
Figure 27 - CPI development and prediction by Norges Bank (Norges Bank) ....................................... 41
Figure 28 - Units included in scenario 1 run 2 ....................................................................................... 44
Figure 30 - Units included in scenario 2 ................................................................................................ 46
Figure 33 - Units included in scenario 4 ................................................................................................ 49
Figure 36 - Units selected in scenario 6................................................................................................. 52
Figure 37 - Screenshot from parameter analysis of total costs later .................................................... 54
Figure 38 - Screenshot from parameter analysis of total costs now .................................................... 54
Figure 39 - SA cycle ................................................................................................................................ 57
Figure 40 - Elements in logistics support............................................................................................... 59
Table of Tables
Table 1: Risk matrix _______________________________________________________________
Table 2: Frequency classes _________________________________________________________
Table 3: Consequence classes _______________________________________________________
Table 4: Failure modes ____________________________________________________________
17
18
18
19
ix
Table 5: CM methods summary _____________________________________________________

Table 6: Investment costs __________________________________________________________
Table 7: Summary of scenarios ______________________________________________________
Table 8: Relationship between equipment name, failure mode, type and cell name ____________
Table 9: OBM result summary _______________________________________________________
34
39
42
42
53
1 Introduction
This master thesis is the continuation of work done in the project thesis (Kristoffersen, 2009). The
recommended areas for further study in the project thesis will form the basis for this master thesis.
This master thesis is written as a part of project 3.2 within the Centre for Integrated Operations in
the Petroleum Industry (IOCentre). Project 3.2 deals with condition monitoring of oil and gas
facilities. The work has been carried out in cooperation with contact person at MARINTEK Torgeir
Brurok.
Offshore operations are split between three major entities: drilling and well construction, reservoir
and production management and operations and maintenance. All these entities rely on a common
logistical supply support chain. Therefore better utilisation of Condition Monitoring (CM) for critical
equipment along with Opportunity Based Maintenance (OBM) will enable a more efficient use of
limited logistical recourses. This master thesis will focus on how CM can be used in OBM and logistics
planning. For OBM planning and organisation Genetic Algorithms (GA) will be used. In order to
identify critical equipment and CM methods a Failure Mode Effects and Criticality Analysis (FMECA)
will be performed using the software tool Manifer. Areas where OBM may have an effect or impact
will be discussed along with possible measures to reduce any negative impacts OBM may have on the
regular operations.
2 CM information and Opportunity Based Maintenance (OBM)

Before equipment can be considered for Opportunity Based Maintenance (OBM) there are several
technical and organisational aspects that need to be considered before OBM can successfully be
implemented. In this section some of these aspects will be discussed, as well as how equipment can
be prioritised in an OBM scheme. Lastly how Genetic Algorithms (GA) can be used to organise OBM
will be discussed.
2.1 OBM Considerations

If OBM is to be implemented an organisation needs to be agile and resilient. But what does agile and
resilient mean? Different sources have different points of view when it comes to defining agility and
resilience when they talk about organisations. For more information about what different sources
think agility and resilience mean for them I recommend (Vrnes, 2008), in this report different
points of view are presented. However no single definition of what agility and resilience is can be
found, instead what agility and resilience mean in different contexts is presented. Therefore the
definition of agility and resilience presented in (Wahl, Sleire, Brurok, & Asbjrnslett, 2008) will be
used instead. In (Wahl, Sleire, Brurok, & Asbjrnslett, 2008) agility is said to be the ability to respond
to changing environment and be able to benefit from the change. Resilience is described as the
ability of an organisation to respond, monitor and anticipate threats to normal operations. These two
organisational aspects are crucial if OBM is to be utilised by the organisation. The organisation needs
to be agile enough to be able to utilise an opportunity to its benefit and thus generate added value.
In addition to agility the organisation needs to be resilient. Resilience can be understood as the
organisations ability to not be crippled when something changes. An unexpected event should not
mean that the other activities of the organisation are affected. Resilience can therefore be seen as a
form of damage control.
In space operations the culture is focused on maximising the use of scarce resources e.g. logistics
availability. While in oil and gas operations the focus is on fixing failures. Should OBM be
implemented in oil and gas operations a shift towards the view inhabited by space operations seems
necessary. Another lesson that can be learnt from space operation is the development of standard
operating procedures for critical equipment.
Another element that will help the organisation when implementing OBM is the establishment of
Operations Support Centres (OSC). The objective of the OSC is to manage established plans and
handle any plan deviations. In the OSC relevant information is traded across disciplines such as
drilling, production management, maintenance and logistics. The goal of the OSC is to support the
offshore team in problem solving when deviations occur. For the OSC to function properly it needs to
be composed of people from all the relevant disciplines e.g. logistics, maintenance, drilling etc. either
by being there in person or through the use of ICT (Information and Communication Technology).
The development of standard operating procedures for critical equipment along with prepositioned
spares and general crew training will also help in the implementation of OBM. If storage capacity is
limited at the facility or at the supply base then supplier retention can be used to guarantee spare
availability. The idea behind supplier retention is that the supplier guarantees access to spares within
a defined response time.
On the technical side there are several different aspects that need to be considered before
equipment can be a candidate for OBM. One such factor is the availability of spare parts or tools
needed to perform whatever maintenance task is considered for execution during an opportunity. If
spares and tools are not readily available other maintenance tasks should be considered instead, as it
is uncertain how long the opportunity will be and how much time it will take to transport spares and
tools to the facility. Should specialists be required to perform maintenance the equipment should not
be considered for OBM. This is because opportunities will for the most part occur at random and
therefore specialists are not likely readily available at the facility or at a nearby facility either offshore
or onshore. If specialists can be located in short time and the maintenance job is minor the specialists
can guide offshore personnel from the OSC. Offshore personnel should be able to handle most minor
maintenance jobs, it is therefore convenient to only consider minor maintenance jobs when talking
about OBM.
Should there be a need for spare parts, tools or people to do maintenance on the equipment and
these are not present at the facility then the need for transport arise. The need for transport capacity
is a third factor to consider before equipment should be considered a candidate for OBM. Important
logistical aspects to consider is the availability of transport capacity and the logistics lead time.
If scaffolding is needed to reach the equipment or the equipment is in hard to reach places other
equipment should be considered for short opportunity windows. If the window of opportunity is
longer or expected to be longer in duration equipment that requires scaffolding or similar rigging
may be a candidate for maintenance.
If a failure is the cause of a maintenance opportunity and an OBM strategy have been implemented
there is a risk that a lot of time will be spent on identifying ways to exploit the opportunity, and that
the event that triggered the opportunity will be neglected leading to more downtime. To counter this
it would be beneficial to divide the maintenance planners into two teams. The first team should work
on fixing the failure that triggered the opportunity while the second team works on finding ways to
exploit the opportunity. This will ensure that the maintenance planners keep their priorities on both
sides and that downtime costs are kept as low as possible.
2.2 Prioritising equipment for OBM

In the aircraft industry components are changed after a set number of running hours. This strategy
makes it easy to know when a certain component should be changed next. This replacement strategy
also makes it easy to organise and prioritise equipment when it comes to OBM. This is not the case
when it comes to equipment used in oil and gas facilities. Therefore there are several challenges
when it comes to prioritising equipment considered for OBM.
One of those challenges is that knowing the length of the opportunity is almost impossible. How long
the window of opportunity is will also affect what can be maintained during the opportunity.
Because of this fact most of the literature dealing with opportunity based maintenance only
considers smaller maintenance jobs for opportunity maintenance, while allocating larger overhauls
to large maintenance campaigns and annual shutdowns. In theory there is no obstacle in scheduling
large maintenance jobs to opportunities, but it practice it may be difficult to do. If the opportunities
to do maintenance are short large and time consuming maintenance tasks will most likely be
deferred to a later time. If this is repeated enough times the item will eventually fail causing
expensive downtime. It is possible to schedule large maintenance jobs to periods in time when it is
4
known that the facility will not be operational such as well intervention and maintenance. During
such a period it should also be possible to do maintenance on topside equipment and not only
downhole maintenance. Major maintenance jobs on equipment should therefore be given a low
priority if it is expected that the window of opportunity is short. Such maintenance jobs and
equipment requiring long time for maintenance should be given a high priority if a long window of
opportunity is expected.
Equipment located in places that require scaffolding should also not be considered for small windows
of opportunity. This is due to the fact that assembling and disassembling the scaffolding is a time
consuming task and will eat away the total amount of time available during the opportunity. Time
spent on rigging could be spent performing other minor maintenance jobs elsewhere in the facility.
Equipment located in places that proves to be hard to reach may be considered for longer
opportunities. The same reasoning can be used for prioritising equipment in hard to reach places as it
was for minor and major maintenance jobs. That is equipment in hard to reach places is given a low
priority if the opportunity is expected to be short and a high priority if the opportunity is expected to
be longer.
The failure rate of the equipment will play a role in how often maintenance has to be executed.
Therefore it would be prudent to use failure data from OREDA (SINTEF Industrial Management, 2002)
if own failure data is unavailable. In some cases the data presented in OREDA may be on the
conservative side, therefore some corrections can be made to the failure data. If data from OREDA or
own failure data is used the same principle of prioritising equipment applies. Equipment with a high
Mean Time to Failure (MTTR) should be given a low priority if expected length of opportunity is short
and high priority if opportunity is expected to be long. The Mean Time Between Maintenance
(MTBM) and Mean Time Between Failure (MTBF) of different equipment can also be used to
prioritise equipment considered for OBM. If records are kept on when the last replacement or
maintenance action was completed these records can be used as an indication of when the
equipment is likely to fail next. Equipment that has run for a long time will have a shorter Remaining
Useful Life (RUL) than equipment that was just installed. Therefore RUL can be used for prioritising
equipment for OBM. Equipment that have little RUL left should be considered for maintenance
during an opportunity compared to equipment that have a lot of RUL left. Estimating RUL is a large
field on its own and has been omitted from this thesis. For more information about how RUL can be
estimated I recommend using the internet a lot of information on the field is available there. Figure 1
illustrates three possible RUL predictions for one unit nearing the end of its life.
Remaining life
Time
Scenario 1
Scenario 2
Scenario 3
Figure 1 - Example of RUL predictions
From OREDA (SINTEF Industrial Management, 2002) the definitions of different failure modes are
obtained. A Critical failure will cause the immediate and complete loss of the systems capability to
provide its output. A Degraded failure is not critical, but it prevents the system from provided the
specified output. Given enough time a degraded failure may develop into a critical failure. Degraded
failures are usually gradual or partial. An Incipient failure is a failure that does not immediately cause
loss of system output. If an incipient failure is left unattended it may develop into either a degraded
or critical failure. The last failure type defined in OREDA is the Unknown failure. An Unknown failure
is a failure where the severity could not be determined or was not recorded. In prioritising
equipment for OBM the degraded and incipient failures are the most interesting, since the critical
failure triggers a corrective maintenance action. Unknown failures are also not considered because
their impact could not be determined. It can be expected that unknown failures if left unattended
will develop to one of the other failure modes.
To summarise a flowchart has been developed outlining the aspects needed to be considered before
equipment can be considered for OBM as well as prioritising equipment for OBM. The flowchart can
be seen in Figure 2. Consider an equipment of type A that is considered for maintenance the next
time an opportunity arises. The first thing that should be considered is the RUL of A. If it is expected
that A has a long RUL left then it would be better to consider equipment that have a short RUL left
e.g. equipment of type B. If A has short RUL then the location of A should be considered. If A is
located in a place where no scaffolding or only minor scaffolding is required then the size of the
planned maintenance activity will play a role. Minor maintenance jobs that require no or only minor
scaffolding are ok to consider for OBM. If the maintenance job is a major overhaul, but limited
scaffolding is required then the length of the opportunity will play a role. With a long window of
opportunity a major maintenance job can be considered for OBM. A short window of opportunity
combined with a major maintenance job and no scaffolding would disqualify the equipment from
OBM. If major scaffolding is required for the maintenance job the size of the maintenance job needs
to be considered. A minor job will be ok for OBM while a major job will require the length of the
opportunity to be considered as well. If the opportunity is expected to be long then the equipment
can be a candidate for OBM. A short opportunity will disqualify the equipment from OBM.
6
Figure 2 - OBM flowchart
2.3 Organising OBM with GA

In (Saranga, 2004) GA is suggested for OBM planning. In this section ideas presented in this article
will be further developed and adapted for oil and gas facilities and equipment. First an introduction
to GA will be given. For more detailed information about GA I recommend (Wikipedia, 2010).
GA is a subset of Evolutionary Algorithms, and is a search technique used to find exact or
approximate solutions to optimisation problems. In order to arrive at a new solution GA use
inheritance, mutation, selection and crossover. All these techniques are inspired by evolutionary
biology. The first step in setting up an optimisation problem using GA is to create a first generation
consisting of chromosomes. Each chromosome corresponds to a variable in the solution space. The
next step is to pass a random selection of the initial population through the fitness function. The
fitness function can be seen as the objective function in other optimisation problems. The individuals
with the best fitness are selected for reproduction according to the concept of survival of the fittest,
to see if a better solution can be obtained. To create new generations GA use three processes:
1) Reproduction, here selected individuals of one generation are randomly selected for mating.
The selection is biased towards best fitted, only decided by the fitness function i.e. survival of
the fittest.
2) Crossover, here individuals are mated randomly. Not to be confused with reproduction
where characteristics are copied. This mating process creates new characteristics in the next
generation. Information is exchanged in blocks. See Figure 3.
3) Mutation plays only a secondary role in the process. Using mutation new characteristics can
be introduced in the next generation or lost characteristics can be reintroduced. Will avoid
being stuck in local optima.
Figure 3 - Crossover
The GA process is illustrated in Figure 4. It starts with a parent generation. This generation is then
evaluated against the fitness function, those individuals with the best fitness is selected for breeding.
The offspring from the first generation replaces the individuals of the first generation who had the
worst fit. Then the second generation is evaluated and the process repeats. The process can be set to
end after a set number of generations, when the algorithm does not produce offspring with a better
fit than the parent generation, when budgeted time/money have been reached or by manual
inspection. A combination of the above mentioned conditions is also possible.
Figure 4 - GA flow
In OBM planning GA can be used to find the optimal organisation. Depending on the chosen fitness
function different optimisations can be found. According to (Rasmussen, Driftsteknikk Grunnkurs,
2003) any maintenance strategy must satisfy two basic demands:
Applicability
Cost effectiveness
8
Applicability means that preventive maintenance prevents all failures or reduces the amount of
failures. Cost effectiveness relates to the consequences of failures and the costs associated with
doing maintenance to prevent failure. Any OBM scheme that is produced by GA simulations should
therefore satisfy these two demands. Based on the prioritisation suggested in chapter 2.2 All in all 17
parameters have been identified that will influence the OBM organisation. These parameters can be
divided into two different groups. The first group is associated with characteristics of the equipment
such as name, failure mode, fault type, criticality of failure, MTTR etc. The second group are the costs
related to both the equipment and the equipment characteristics. These costs include costs of
logistic, work, downtime, risk, RUL, scaffolding etc. What is included in the costs will be discussed
more thoroughly in section 3.4.1.
3 Oil and gas production and CM methods

The flow from the well head may contain crude oil, gas, condensates, water as well as other
contaminants. Before the hydrocarbons can be sold they need to be processed. This processing is
done by the topside processing equipment. The layout of one such topside processing plant along
with a Failure Mode Effect and Criticality Analysis (FMECA) and some simulations using GA will be
done in this section.
3.1 System layout

A simplified model of a topside processing plant is shown in Figure 5 in this figure all of the metering
equipment and much of the control equipment have been omitted. Some equipment that may be
present in other processing facilities has been omitted from this overview. One such example is the
hydrocyclone used for water treatment. Differentiating between different types of valves in the
process is also something that has not been done to a large degree. In Figure 5 a red line represents
gas flow, a yellow line represents two phase flow, the black line is oil flow and the blue line is
produced water.
Figure 5 - System layout
The process shown in Figure 5 does not include the water treatment plant. The process only exports
oil and gas.
3.1.1 Separators
The purpose of the separator is to split the two phase flow from the well into more desirable parts
such as oil, gas and condensates. The production choke reduces the pressure of the flow coming into
the separator to about 3 - 5 MPa (Devold, 2009). Pressure is often reduced in several stages to avoid
flash vaporisation that may cause instability and safety hazards. Figure 6 show a cross-sectional view
of a separator. The retention period is typically 5 minutes. This allows the gas to bubble out, water to
sink to the bottom and oil to be extracted in the middle. At the inlet the separator has a slug catcher
used to catch large gas bubbles or liquid plugs. The slug catcher should not make the flow through
the separator laminar, as some turbulence is desired to speed up the separation of gas. At the end of
11
the separator there is a level barrier to keep water and oil from mixing. Vortex breakers are installed
at the outlets to prevent disturbances in the liquid. The vortex breaker is basically a flange trap that
prevents vortices from dragging oil into the water outlet and vice versa. A demister is installed before
the gas outlet to ensure that liquid droplets are not tapped off at the gas outlet. First stage
separators or high pressure separators can depending on the water cut in the incoming flow reduce
the water content downstream of the separator to about 5 %.
The second stage separator or low pressure separator operates around a pressure of about 1 MPa,
and is similar in design to the first stage separator. The second stage separator may also be
connected to low pressure manifolds. In Figure 5 an oil heater is located between the first and
second separator. Heating the flow from the first stage separator will help in extracting water from
the flow if the water cut is high and the temperature is low. The oil heater is typically a shell and tube
heat exchanger where the oil passes through the tubes and the heating medium is in the shell. Heat
exchangers and heat exchanger types will be discussed later.
Figure 6 Separator (Devold, 2009)
3.1.2 Coalescer
A coalescer is used to remove the final remains of water/liquid in the oil and gas. The colaescer uses
internal electrodes to create an electric field that breaks the surface bonds between the oil/gas and
the water/liquid in the oil water emulsion. With a coalescer the water/liquid content can be brought
down to below 0.1% according to (Devold, 2009). The grid layout and field intensity of the coalescer
varies depending on the manufacturer and oil types.
3.1.3 Heat exchangers
In order for the compressors to operate as efficiently as possible the temperature should be as low
as possible. Lowering the temperature of the gas is done using heat exchangers. In heat exchangers a
hot and a cold medium passes each other separated by plates. There are two types of heat
exchangers that are most commonly used in the oil and gas industry. These are the shell and tube
heat exchangers and the plate heat exchangers. Plate exchangers consist of several plates where the
gas and cooling medium pass between alternating plates in opposing directions. A principal view is
given in Figure 7. In a shell and tube exchanger the gas passes through tubes and the cooling medium
12
is contained by the outer shell. The cooling medium is commonly water with chemical additives that
inhibits corrosion. Figure 8 shows an example of a shell and tube heat exchanger.
Figure 7 - Plate heat exchanger (Devold, 2009)
Figure 8 - Shell and tube heat exchanger (Devold, 2009)
3.1.4 Scrubbers
Should droplets of oil or other liquids enter the compressor the fast rotating blades will be damaged
by erosion. To prevent this gas is passed through a scrubber that removes the droplets. Many
different designs are available but the most common type uses Triethylene Glycol (TEG). A principal
drawing of a TEG scrubber is shown in Figure 9. The gas enters the scrubber at the bottom and
passes through several layers of glycol before exiting at the top. The detail in Figure 9 shows the gas
trap that forces the gas to bubble through each level of the scrubber. Glycol is pumped from the
holding tank and into the top of the scrubber. As the glycol overflows the gas trap it spills down to
the next level absorbing liquid from the gas in the process. Rich glycol is then recycled in the reboiler.
13
Figure 9 - TEG scrubber (Devold, 2009)
3.1.5 Compressors
In an oil and gas facility compressors are used for compressing gas. Depending on the usage different
types of compressors can be used. For high pressure gas injection and where a low capacity is
needed reciprocating compressors is mostly used. For gathering of natural gas screw compressors
are mostly used. Axial compressors are used for air compression and LNG (Liquefied Natural Gas)
cooling. On large installations centrifugal compressors with between three and ten radial wheels are
used according to (Devold, 2009).
On an oil and gas facility one compressor type and size will not be capable of covering the full
pressure range effectively. Pressures required on oil and gas facilities will range from atmospheric
pressure to 20 MPa and higher. Therefore compression is divided into several stages to improve
maintainability as well as availability.
3.1.6 Valves and other control equipment
In order to avoid undesired events and production shutdowns valves and other control equipment is
used to monitor and control the flow of hydrocarbons through the facility. Valves are used to restrict
or stop the flow of hydrocarbons from one part of the facility to another. Different measuring
equipment is located throughout the plant to monitor temperature, pressure, vibrations and flow.
These attributes when monitored and presented to operators will help in decision making and alert
operators if deviations from normal operation is detected.
14
Figure 10 - Ball valve
Devices that are installed on oil and gas facilities need to be protected from becoming ignition
sources if a hydrocarbon leak should occur. Devices are classified accordingly to what kind of
protection it has from becoming an ignition source e.g. Ex.p. if the device is safe by pressurisation.
Also all the different areas of the facility are mapped into different zones depending on the explosion
hazard. The zone ranking ranges from zone 0 to zone 2 and safe area. Zone 0 is defined as an area
where hydrocarbons will be present such as inside pipes and vessels. Zone 1 is an area where there is
a high risk of hydrocarbons being present. While zone 2 is an area where there is a low risk of
hydrocarbons being present. The safe zone is typically the living quarters where hydrocarbons are
not present.
3.2 FMECA
The Failure Mode Effects and Criticality Analysis (FMECA) is a systematic approach used to analyse
systems, how they fail and the consequences these failures will have on the system on a component
level. The first step of any risk analysis technique is describing the system that is going to be
analysed. The following six steps summarises the general approach in performing a FMECA:
1)
2)
3)
4)
5)
6)
Describe components
Describe possible failures and failure modes
Describe failure effects of each failure mode
Ranking failure effects in terms of frequency, severity and specifying reliability data.
Specifying methods for detection.
Describe how unwanted effects can be reduced and eliminated
The FMECA is a part of the Reliability Centred Maintenance (RCM) concept. The RCM concept is a
method used for establishing cost effective maintenance plans. The RCM concept can be deployed in
any phase of the life cycle of a system or facility. In the design phase RCM analysis will for example
help in removing design alternatives that require a lot of preventive maintenance. In the operational
phase a RCM analysis can be used to review the maintenance plans that already in place. The RCM
process is illustrated in Figure 11. As Figure 11 illustrates there are a lot of different inputs and
outputs form the different phases. The end result from the process is maintenance packages
consisting of jobs grouped in a cost effective manner.
15
Figure 11 - RCM process
3.2.1 Functional hierarchy

When describing the components it is useful to create a functional hierarchy to get a better idea of
what function a component serves and how it interacts with its environment as well as what might
happen should the component fail. The functional hierarchy is a logical representation of the system
as a whole where equipment is tied to a specific function. It clearly shows what functions that are
dependent on certain equipment. A failure in any component may in turn result in reduced
performance of the associated function or in a worst case scenario total loss of the function. A
functional hierarchy for the oil and gas production system shown in Figure 5 has been developed and
can be seen in Figure 12. In Figure 12 the main function can be seen in green, while 1st stage sub
functions are in blue and 2nd stage sub functions are in yellow. The associated equipment is shown in
pink.
16
Figure 12 - Functional hierarchy
Once the functional hierarchy had been established the hierarchy needed to be transferred to the
software application Manifer where the FMECA will be performed.
3.2.2 Definitions
To rank the effects of the different failure modes in a systematic manner a risk matrix is needed. The
risk matrix used in this thesis is presented in Table 1. The risk matrix is based on a set of definitions
shown in Table 3 for the different areas of interest. The scale used here ranges from an insignificant
failure (0) to a catastrophic failure (5). Usually a limit of what is acceptable is chosen e.g. 4. This
would indicate that any failure mode that leads to a consequence ranked 4 or above is unacceptable.
For some applications failures ranked as 3 or medium criticality can also be ranked as unacceptable.
Table 1: Risk matrix
Catastrophic
Critical
Major
Minor
Frequent
Probable
5
4
4
4
3
3
2
2
Occasional
Remote
Very unlikely
3
3
3
2
2
1
1
0
In order to precede a definition of the frequency term is needed. In (Kristiansen, 2004) the frequency
is defined in a once per X years manner while the OREDA handbook (SINTEF Industrial
Management, 2002) uses a failure rate given in failures per 106 running hours. In this thesis a
modification of the two will be used. However the definition can be chosen arbitrarily as long as it is
maintained throughout the analysis. The definitions of frequency classes can be seen in Table 2.
17
Table 2: Frequency classes
Failure rate grouping
In x per year equivalent
Frequent
Probable
Occasional
Remote
Very unlikely
100 0.87 or more faults pr year

50 100
0.44 0.87 faults per year
20 50
10 20
0 10
0 0.08 faults per year
The consequences of a failure will be judged against three classes. These categories have been
chosen are assumed to be the most important for an oil and gas facility. The definitions might seem
strict but it is assumed that the facility I question is operated in the North Sea where the public
tolerance for safety and environmental mishaps is low, so it seems like a fair definition. The
consequences with respect to production have been given that division because downtime and
production loss is so expensive. Table 3 shows the definitions of consequence classes used in this
thesis.
Table 3: Consequence classes
Consequence
Catastrophic
Safety
Death of
personnel
Severe personnel
injury
Environment
Large spill >
100m3
Medium spill
<100 m3
Major
Personnel injury
Small spill <10 m3
Minor
No injury
No spill
Critical
Costs/Production
Complete
shutdown
Risk of down
time, reduced
production
No downtime,
reduced
production
No downtime, no
reduction in
capacity
3.2.3 Failure rates and types

All failure data has been taken from the OREDA handbook, (SINTEF Industrial Management, 2002).
Other elements that have been taken from OREDA include the system boundaries and failure modes.
Only the most common failure modes have been used, and failure modes from all the different
criticality categories have been included. Where three or more failure modes are common a
subjective choice has been made on which failure mode to include in the analysis. Since OREDA have
three failure rates listed the mean value has been used in this thesis, both calendar time and
operational time is listed in OREDA and only the operational time has been used here. Several
different equipment types are listed in the OREDA handbook in this thesis only failure data from
centrifugal compressors have been used. Failure data from both plate and shell and tube heat
exchangers have been included. Only the general failure data for separators, scrubbers, contactors
and coalescers have been used since the capacity of these equipment types are not known. General
fault data for different sensor types have also been used. In total four different sensor types have
been included. Both ball and gate valves have been included in the analysis. The failure database can
be found in Appendix 1: Failure database. When it comes to relating failure mode to maintainable
items e.g. actuating device for valves and failure mechanisms e.g. wear the most common type listed
for that failure mode is used.
18
3.2.4 Results and comments

As stated earlier the FMECA has been performed using the software application Manifer. Figure 13
show how the functional hierarchy was implemented into Manifer. In Figure 13 the first stage sub
function is numbered with two numbers e.g. 1.1 Separation, while second stage sub functions are
numbered with three numbers e.g. 1.1.2 Gas Separation. Functional errors can be seen as yellow
triangles. On the level below functional errors equipment is tied to different failure modes. It should
be noted that failure modes taken from the Degraded and Incipient failure modes in OREDA have
been assigned to the insufficient failure mode while Critical failure modes from OREDA have been
assigned to the total loss failure mode. The failure modes used in the analysis are shown in Table 4.
Table 4: Failure modes
Code
Name
AR
DO
ELP
ELU
EO
FCD
FD
FOD
FS
IL
IP
LCP
LO
O
PC
PD
SD
SS
V
Abnormal instrument reading

Delayed operation
External leakage - Process
External leakage - Utility
Erratic output
Fail to close on demand
Fail to function on demand
Fail to open on demand
Fail to start on demand
Internal leakage
Minor in-service problem
Leakage in closed position
Low output
Other
Plugged/Chocked
Parameter deviation
Structural deficiency
Spurious stop
Vibration
19
Figure 13 - Functional hierarchy from Manifer
20
In Figure 14, Figure 15 and Figure 16the total number of registered failure modes for the whole
process shown in Figure 5 is summarised. If an acceptance level of 4 is chosen then the risk matrices
show that for all criticality categories there are failure modes that have been ranked as unacceptable.
To prevent these failure modes form happening or mitigating the consequences of these failure
modes happening further investigation is needed. The risk matrices for all sub functions are
presented with all criticality categories included can be found in Appendix 3: Risk matrices for sub
functions. Due to some difficulties with Manifer, it was impossible to export the full analysis to Excel,
only a smaller version of the FMECA report was exported to .pdf format. The .pdf version contains
most of the information, but alas some of the information has been left out. The complete .pdf
version of the FMECA report can be found in Appendix 2: FMECA Report.
Figure 14 - Risk matrix: Costs
21
Figure 15 - Risk matrix: Environment
Figure 16 - Risk matrix: Safety
Form Appendix 3: Risk matrices for sub functions the most critical components can be found to be
the compressor with 35 failures ranked as unacceptable and the heat exchangers with a total of 3
failure modes ranked as unacceptable. Valves also had failures ranked as unacceptable but due to
time constraints and engineering judgement valves have been omitted from further investigation in
this thesis.
In order to reduce the chance of a failure occurring CM can be applied to the equipment that has
been defined as critical. If any of the failures that represent a medium risk are found to be
unacceptable the ALARP (As Low As Reasonably Practicable) principle can be applied to these to
reduce the risk. The idea behind the ALARP principle is to obtain a balance between risk reducing
22
measures and the costs of implementing them. The ALARP principle is illustrated in Figure 17. The
black circles in Figure 17 represent identified risks. Risks located in the red region are only tolerable
in extraordinary cases, but normally risk reducing measures are implemented to reduce the risk level
to tolerable. In the orange region risk reducing measures are only implemented if the benefit of
implementing the measure is larger than the costs associated with implementing the measure. The
orange region is also called the ALARP region illustrating that it is in this region that risk reducing
measures are implemented if it is practicable. In the green region risk reducing measures are
generally not implemented unless the benefit of implementing the risk reducing measure greatly
overshadows the costs of implementing the measure.
Figure 17 - ALARP principle
When it comes to CM the challenge is to find the right monitoring method to apply from the wide
spectrum of methods available. In section 3.3 different CM methods will be described.
3.3 CM methods
In (Hunt, 2006) 15 different methods for Condition Monitoring (CM) are presented. All the different
CM methods are presented in a similar manner with emphasis put on the purpose, measurements
taken, applicability, what kind of sensors and instrumentation used and words of warning regarding
23
each method. In addition to the CM methods presented in (Hunt, 2006) Electromagnetic Testing,
Magnetic Particle Inspection and Visual Inspection from (DNV, 2007) will also be included. The
complete list of CM methods included in this thesis therefore encompasses:
Acoustic emissions and Ultrasonics

Colour
Corrosion
Environment
Electromagnetic testing
Level, Leakage & Flow
Load
Magnetic Particle Inspection
Noise
Oil analysis
Position
Power
Smell and Taste
Temperature
Thermography
Vibration
Visual Inspection
Wear and debris analysis
In this thesis environment monitoring has been omitted, due to most of it being covered in several of
the other CM methods. In the following sections a brief summary of each of the CM methods will be
given. Not all of the CM methods presented in the book may be appropriate for oil and gas facilities.
3.3.1 Acoustic emissions and Ultrasonics monitoring
The purpose of acoustic emissions and ultrasonic is to detect changes in a machines condition using
high frequencies. The frequencies typically range from 25 kHz up to 1 MHz. These frequencies are not
detectable by the human ear which can hear up to about 20 kHz (Wikipedia, 2008). Since the human
ear comes up short in detecting these frequencies a piezoelectric transducer is usually used. Since all
machines produce some acoustic emissions it is the type of signal produced that is of interest. A
deteriorating machine condition can typically be associated with an increase in the continuous signal
level. Friction, impacts and cavitations will also change the acoustic emissions levels. In (DNV, 2007)
several sub types of ultrasonic imaging is presented. The different sub groups are based on separate
methods for generating and collecting the ultrasonic signals and as a consequence the different
methods have different applicability and different depths of penetration.
For analysing the signals different methods can be used. But trending is very common for many
different CM methods. The simplest method of trending is to plot the logarithmic acoustic level and
look for a change.
Acoustic emissions can be applied to most equipment that gives of sound either by rotating or
changing flow through the equipment.
24
Special attention should be paid to make sure that frequency range of the sensors is matched with
the required performance of the component. Also transducers should be placed in the region where
the signal strength is the highest. Sensors that detect peak values should be oriented into the best
direction to make sure that the instantaneous value of bursts can be recorded.
3.3.2 Colour monitoring
Colour monitoring is a form of visual inspection where the change of colour in the component
surface or liquid will indicate a developing fault. The change in colour may be caused by either
internal or external effects. Internal effects that may change the surface colour include temperature
and variations in loading or vibrations. External effects that may alter the surface colour include
temperature, corrosion, leakage and physical damage.
The hue of a colour can be measured in two different ways. The first way is to detect the presence of
red, blue and green in the colour. This is related to the sensitivity of the human eye. The second
option is spectral content in which the individual frequencies in the visual spectrum are recorded. In
addition a specific colour sample needs to be defined in terms of saturation and intensity. Saturation
means the vividness of the colour. Terms like bold and insipid can be used to describe saturation.
Intensity can be understood as brightness with white and black as the two extremes of the
intensity scale.
Figure 18 - Visual spectrum in nm
If absolute values of the two features are used slight changes in the colour may go unnoticed as
atmospheric conditions affect the results. Therefore trending is by far a much better option.
Colour monitoring can be used to detect oil degradation due to overheating. System overheat may
also be detected using colour. Industries that may use colour as a CM method includes any industry
that works with oils, all industries where internal temperatures may affect surface temperatures and
industries where gas is involved.
The basic equipment used for collecting data include colorimeter, spectrophotometer and gloss
meter.
If colour monitoring is to be used care should be taken to ensure that the light conditions are the
same every time measurements are taken. Artificial light sources are better than daylight.
Atmospheric changes may cause the spectre to change, humidity may introduce new spectres and
particles may filter certain frequencies.
3.3.3 Corrosion monitoring
Corrosion is material degradation caused by the environment through chemical, electrochemical or
biological actions or reactions. Corrosion monitoring is used to detect the early stages of surface
degradation caused by faults in the system or failure of protective mechanisms.
25
The techniques employed to detect corrosion are numerous. Every detection method has its own
strengths and weaknesses. Most of the analyses based on corrosion measurements are based on an
estimated corrosion rate e.g. 5 micron per year.
Faults that can be detected using corrosion monitoring include leakage, failing barriers and
protective coatings.
Corrosion monitoring is employed in numerous
industries spanning from oil and gas production,
transportation and process industry. Corrosion
monitoring is applied to two major areas of the
industry today, namely pipes and vessels and
reinforced concrete.
The most common sensors used in corrosion
monitoring today is the corrosion coupon and
probe. The corrosion coupon and probe is
designed to be sensitive to what is causing the
corrosion and provide a sample of the real
situation.
When choosing a corrosion monitoring technique
care should be taken to ensure that the technique
is capable of detecting the corrosion and what
Figure 19 - Corrosion monitor by (Roxar)
surface is suffering from the corrosion. A matching
of corrosion type and corrosion measurement technique is necessary. Another thing to be aware of is
that corrosion is a slow process. Data collection and analysis may take a long time.
3.3.4 Electromagnetic testing
Electromagnetic testing uses electromagnetic induction to detect flaws in conductive materials. The
most common type of electromagnetic testing is Eddy Current Testing (ECT). Conventional ECT has
limited penetration depth and is therefore only suited to test for cracks in surfaces. However more
advanced ECT methods have been developed and are capable of deeper penetration (DNV, 2007).
Among the more advanced ECT methods Pulsed Eddy Current Testing and Saturated Low Frequency
Eddy Current (SLOFEC) are interesting.
Pulsed ECT can be used to detect corrosion and erosion over large areas. It does not detect smaller
pits and the test results may be influenced by the presence of large metal masses. Pulsed ECT can be
used to detect faults in wall thicknesses up to 100 mm.
Saturated Low Frequency Eddy Current is based on ECT where direct currents are used for
magnetisation. SLOFEC is according to (DNV, 2007) capable of penetrating wall thicknesses of about
30 35 mm and is therefore best suited for inspecting pipes and corrosion in vessel shells.
3.3.5 Level, leakage & flow monitoring
These three types of monitors are used to determine the condition of a machine or system by
analysing the fluid movements and positions. The condition of the system will then be expressed by
the levels, leakages and flows in the system. The level gives an indication of the amount of fluid in
26
the system. The leakage indicates at what rate the level declines. Trace gas can also be used in leak
testing. The flow indicates the efficiency of movement through the system.
Measurements of level, leakage and flow can be done by a variety of different meters. All the
different meter types have their own strengths and weaknesses.
3.3.6 Load monitoring
Load monitoring encompasses the measuring of six different aspects of loads. These six aspects
include load, force, pressure, strain, torque and weight. Each aspect can be used to detect a
deteriorating condition in a machine or system.
Faults can be detected by measuring one or more of the six aspects. Load can be measured by a load
cell. Force can be measured by a strain gauge, which measures the deformation of a structural
element. Pressure can either be measured by a pressure gauge or by a Bourdon tube connected to a
dial. Strain can be measured by a strain gauge. Torque is measured by a rotational strain gauge or by
angular deformation. Weight can be measured by load cells.
The analysis method is very straight forwards: Simply look for unacceptable changes in the
parameters when the running conditions are identical.
The faults detectable by this kind of CM depend on the parameters that are being monitored. In
many cases the basic sensors used are strain gauges.
When using load monitoring great care should be taken to ensure that the sensors employed can
operate in the expected load range of normal operations. A sensor with a too large range would not
be able to pick up minor changes, while a sensor with too small range would not tolerate
overloading. Strain gauges need to be bonded to the surface with the proper adhesive, otherwise
slipping may occur. Load cells need to be aligned in the direction of the expected load.
An industrial example of electrical load monitoring is the DriveMonitor system offered by ABB.
DriveMonitor is intended to monitor compressor drives (ABB Oil, Gas and Petrochemicals). The basic
functionality of DriveMonitor is to watch the frequency converter of the drive system as well as it
collects and analyses selected drive signals. But it can also be configured to monitor other
components such as the circuit breaker, transformer and driven machine i.e. the compressor
according to (ABB, 2008).
3.3.7 Magnetic Particle Inspection
This form of testing is used for detecting surface and near surface cracks in ferromagnetic materials.
This method is one of the most extensively used methods in the industry today. The test is
performed by magnetising the component that is to be tested. The magnetisation is achieved by
applying a permanent magnet, electromagnet or an electric current. A magnetic field is produced
inside the component and the magnetic field is distorted by the presence of flaws in the material.
Flaws are detected by applying magnetic particles either in the form of a powder or as a liquid. The
magnetic particles accumulate in and around the area of flaws thereby making it visible.
When this kind of testing is applied the orientation of the magnetic field and the imperfection is
important. The best result is achieved when the angle between field lines and imperfection is 30
27
90. Therefore it is best to apply the magnetic field in two directions to ensure that all flaws are
detected. Magnetic Particle Inspection (MPI) can only be applied to surface cracks.
3.3.8 Noise & acoustics monitoring
Noise and acoustics monitoring covers the frequency ranges that the human ear can hear typically
from 20 Hz to 20 kHz. It is important to include acoustics because the surroundings will have an
impact on what is heard.
Faults are detected using a microphone that receives sound pressure waves, a weighting network
that gives acceptable emphasis to each frequency and a display showing the noise signature so that a
comparison can be made against the signature of a good machine.
Four ways of analysing noise signatures are mentioned. These include
sound pressure over time, sound level versus time, sound level versus
frequency and noise indices versus speed or time.
This kind of monitoring can be used to detect faults in bearings, gears,
pumps, rollers and lack of lubrication. Industries or systems where
noise and acoustics monitoring can be used include hydraulically and
mechanical systems as well as production industry, rail and marine
transport.
When implementing noise and acoustics monitoring it is important that
appropriate meters are used. Special attention should be paid to the
type of analysis, the time weighting, noise and frequency level,
accuracy and output. When using this kind of monitoring the
positioning of the meter is of great importance. If a comparative
Figure 20 - Handheld sound meter
analysis is used the meter needs to be placed in the exact same spot
with the same alignment each time and the machine needs to be in the
same operating condition.

3.3.9 Oil analysis
Monitoring the condition of a machine or system using the fluids of the system is called oil analysis. It
should be noted that it is only chemical analysis of the fluid and not analysing any solids that may be
present in the fluid. Analysis of the solids present in the fluid will be covered in wear and debris
analysis later. Calling this kind of monitoring for oil analysis may be somewhat misleading as a
chemical analysis of any fluid in the system may be performed to give an indication of the systems
condition.
Several basic chemical properties that can be measured to give an indication of the condition of a
machine or system include temperature, viscosity and density. However the main aspect of oil
monitoring is the measurement of the additives. The additives are present in the fluid to nullify
errors in the operation. An example is antifoam that reduces the formation of foam generated from
either serious machine problems, or sloshing in a production separator. If a fault is developing and
the additives do their job the amount of additives present in the fluid should diminish.
28
A spectrometric chemical analysis of the fluid will give a breakdown of the different elements present
in the fluid. When using oil analysis it is of paramount importance that the sample containers are
cleaned prior to use. Contaminants entering the sample prior to analysis are also important to avoid.
3.3.10 Position monitoring
This kind of CM is one of the most basic that exists today. It can be performed by asking two simple
questions:
1) Is anything out of place?
2) Is anything moving in the wrong direction or at the wrong speed?
Large deviations can be detected by a human operator, but the small gradual changes needs to be
detected by sensors. The purpose of position monitoring is the detection of abnormal movements
that have happened, is happening or may happen due to a developing fault in a machine or system.
This kind of monitoring can be used by all industries. The measuring of different developing faults is
what will differ from one industry to another.
The analysis is typically taken on a trend basis. Where a gradual change form an acceptable position
is recorded. In order to use this kind of monitoring it is necessary to define a reference point so a
deviation from this point can be detected. The reference point can either be a local point such as its
normal position or a global position by using GPS.
There exists a large range of position monitors today all with different accuracies and variety of
application.
When choosing position monitoring it is important that the sensors chosen deliver the required
accuracy. Since the cost range of different sensors is huge selecting the right sensor for the job is a
challenge. The loss of reference point is another issue when using position monitoring. Losing the
reference point can be attributed to vibrations.
3.3.11 Power monitoring
This kind of monitoring may also includes performance and efficiency monitoring. Monitoring all
three attributes will enable a better view of developing fault in the system. In Power monitoring the
input to the machine is monitored while the output is assumed to be constant. In performance
monitoring the output of the machine is monitored while the input is assumed to be constant. While
in efficiency monitoring both input and output has to be monitored to give usable data, for
conventional purposes the same units for input and output would be required but this is not the
case. However for some application only one of the three attributes is relevant for detecting faults.
An example where only one attribute is relevant for the condition of a machine is a pump that is
supplying constant pressure downstream. In this example only the power consumed by the pump
motor is relevant as the pump delivers a constant output. Depending on developing faults in the
pump the power consumed by the motor will increase.
29
Performance
Limit
Figure 21 - Example of performance monitoring
As already explained the input and the output do not need to have the same unit for a comparison to
be made. All that is required is a set of twin features unique for that system. Such a set of twin
features can be electrical power consumed and shaft torque. The analysis then becomes a
comparison between acceptable and unacceptable ratios or differences.
When employing power monitoring it is advisable to try and use existing sensors in the system for
financial reasons. A decision also has to be made with respect to what is going to be monitored, is
both input and output needed or will just one suffice. When using power monitoring it is necessary
to also look at the reason behind changes i.e. a rise in oil viscosity will result in an increase in power
even if no fault is present.
For monitoring heat exchangers ABB is developing HXAM G (generic) and HXAM ST (shell and
tube). What HXAM G and HXAM ST does is monitor the efficiency of the heat exchanger without
regard for the design or structure of the heat exchanger. HXAM is an abbreviation for heat exchanger
asset monitors. The system detects gross changes in the heat exchanger efficiency as opposed to
small minute changes. The way the system works is by monitoring readable process variables around
the heat exchanger, noting trends in these variables can then be used as an indication of declining
performance. The system alerts maintenance personnel of changes in the KPI (Key Performance
Indicators) associated with fouling indicated by drifting T at reference hot and cold flow, increasing
P across the heat exchanger at either hot or cold leg, low flow or low T readings for either leg and
significant changes in operating point. HXAM ST is also capable of detecting process errors
indicated by temperature crossover, low shell side flow, low heat transfer, high or low tube velocity
and low limiting approach temperature. The required input to both versions of HXAM is temperature
of hot and cold flow going into and coming out of the heat exchanger. However for better results
more input information can be supplied including mass and volume flow of both hot and cold side,
incoming and out going pressure for both sides.
A strength of HXAM is that it requires little configuration. Another element that speaks for HXAM is
that it can use existing instrumentation, so there is no need to fit additional sensors. One of the
weaknesses of HXAM is that best results are obtained if both the hot and cold side is liquid. This is an
30
area where improvements can be made in the future. Another weakness is that HXAM treats the
heat exchanger as a black box. The same model is used regardless of the heat exchanger type,
mediums etc. HXAM has been tested as a pilot and early results called for improvements to the
system. In theory HXAM can be installed today, but the efficiency of the system may be debated.
According to ABB they are working on an improved version of HXAM. This improved version is
expected to be on the market within two years i.e. 2012.
3.3.12 Smell & taste monitoring
The purpose of this kind of monitoring is the detection of developing faults using the smell
surrounding the system or from the smell and taste of the product of the system. This kind of
monitoring is mostly used in food and drinks industry or pharmaceutical industry. Therefore it will
only be mentioned here for more information I recommend (Hunt, 2006).
3.3.13 Temperature monitoring
In temperature monitoring the temperature surrounding a machine or system is used to determine
the condition of the component.
There are two features that are important when it comes to temperature monitoring. These features
are the sensor type used in the measuring and the location of the sensor when the temperature was
measured. When it comes to sensor type many different sensor types exists, depending on the
application either a resistance thermometer or a thermocouple type of sensor may be suitable.
Depending on what is being monitored the location of the temperature sensor will be important. A
bearing being monitored will benefit from having the temperature sensor as close to the rolling
element as possible. The same applies to a liquid, the thermometer needs to be submerged in the
liquid to obtain accurate readings. If the overall temperature of a machine says something about the
condition of the machine then several surface measurements needs to be taken to obtain the
machine condition.
When analysing temperature data trending is usually the most appropriate method. On the other
hand if a critical temperature exists for the machine or system then an alarm just below the critical
level will be a better option.
Temperature monitoring is applicable to almost all
industries due to the conversion of energy taking place.
Energy conversion is in turn likely to lead to a temperature
increase.
When choosing temperature monitoring it is important to
consider the wide range of sensors available and what is to
be measured. When using temperature monitoring it is
important to remember that heat can be convected and
conducted as well as radiated. Therefore it is important to
insulate from regions that is not tested.
Figure 22 - Temperature meter TF202 by ABB
31
3.3.14 Thermography
Thermography or thermal imaging is a condition monitoring technique where the surface
temperature of a machine or system is used to determine the internal condition. Thermography can
be applied to any system where a developing fault may lead to an increase in heat. Thermography is
a CM technique that does not require contact with the machine or system. All that is needed is an
infrared camera to take a picture of the component then a comparison with another image of the
same component may reveal a developing fault. The comparison can either be done by eye or by
computer.
Thermography is capable of detecting
faults in mechanical components,
electrical machines, energy systems such
as boilers and heat exchangers and
electronic systems. This kind of CM can be
applied to most industries where this kind
of equipment is present.
Figure 23 - Handheld IR camera
When choosing thermography as a CM

method it is important to specify the
requirements that the user needs fulfilled
by the infrared camera. Special attention
should be put on resolution, temperature
range, weight, battery life, data storage
and transmission and if the camera should
be used inside or outside. When using an
infrared camera it is important to
remember that the optics are vulnerable.

3.3.15 Vibration monitoring
In vibration monitoring the running vibrations of a machine or system is measured in order to
determine if a fault is developing. Vibration monitoring is one of the largest fields of CM, covering a
wide range of techniques ranging form out of balance sensors to multi frequency measurements.
Vibration monitoring can be applied to any rotating or reciprocating machine as these generate
vibrations when they are running normally. Because of this any change or growth in these natural
vibrations will indicate that everything is not as it should be.
32
Vibration level
Vibration
Vibration
LimitTime
High High
Limit High High
Limit High
Limit High
Figure 24 - Example of vibration monitoring
Figure 24 shows an illustration of a possible vibration monitoring scheme with high and high high
limits. The idea is that when operators get the high alarm an inspection is scheduled to determine if
the machine needs to be shutdown to fix the problem or not. When the high high alarm is received
then the machine needs to be shutdown to do maintenance to avoid severe machine damage. The
shutdown can be done automatically by the control system with an alarm that needs to be
acknowledged by the operator.
Any industry that works with rotating machinery can benefit from vibration monitoring. There is also
a wide range of sensors available depending on the sensor type and size.
It is important to be aware of that vibrations may be transferred through solids and that
measurements are taken at the component of interest. When considering vibration monitoring it is
important to consider the complexity of the system. The more complex the system is the harder it
will be to decide on what to monitor.
3.3.16 Visual Inspection
In Visual Inspection (VI) a borescope is used to detect developing failures. VI is as the name implies a
method where faults are detected by looking at the sample. The borescope can either be flexible or
rigid. Figure 25 show a flexible borescope. Rigid borescopes are limited by the need for a straight line
between the observer and the object to be observed. Flexible borescopes are used when there is no
straight line between the observer and the area to be inspected. Both types of borescopes can be
configured to suit the needs whatever they may be. Options for consideration include the size of the
probe and the length of the probe. A choice between video or fibre scope can also be made when it
comes to flexible borescopes.
33
Figure 25 - Example of flexible borescope
3.3.17 Wear debris analysis

This kind of CM analyses the solids present in fluids in order to determine the condition of the
machine or system. A number of features are analysed to determine the origin of the solids.
This kind of analysis can be applied to any oil lubricated system, including hydraulic systems. This
kind of monitoring can detect faults in rubbing surfaces e.g. bearings, gears, etc., wear of
reciprocating machines and others. Wear debris analysis can be applied by all industries using oil for
lubrication.
The analysis of the debris is usually conducted in an offsite laboratory, but instruments for analysing
the samples closer to the system is being developed.
When selecting this kind of monitoring technique it is important to note that not all analysers are
appropriate for all systems. Some analysers only detect ferrous metals while others detect all
metallic particles. Another thing to remember is that the oil can only contain particles from where
the oil has been. When using wear debris analysis great care should be taken to make sure that the
equipment used in collecting the samples is cleaner than the requirements of the normal system.
This will ensure that no foreign contaminants spoil the sample.
In Table 5 a summary of the CM methods presented above is given.
Table 5: CM methods summary
Method
Method used in fault

detection
Measurements
Applicability
Warnings
Acoustic
emissions and
Ultrasonics
Acoustic emissions
from macihne
Acoustic emission
level
Most equipment
that emits sound
Frequency ranges,
sensor alignment
34
Any industry
using oil
Light conditions,
environmental
conditions
Colour
Visual inspection
Hue, spectral,
saturation and
intensity
Corrosion
Corrosion coupon
Corrosion rate
Most industries
Technique capable
of detecting
corrosion type
Electromagnetic
Testing
Electromagnetic field
Visual and display

readouts
Most industries,
vessels and pipes.
Surface thickness
Level, Leakage
& Flow
Unacceptable changes
in parameters
Level, leakage
and flow
Any industry
working with
vessels
Sensor selection
Load
Unacceptable changes
in parameters
Six aspects of
load
Most industries
Sensor selection
Magnetic
Particle
Inspection
Electromagnetic field
in combination with
magnetic particles
Visual look for

gatherings of
particles
Surfaces of
vessles and pipes
Wall thickness
Noise
Microphones
Sounds emitted
from machines
Hydraulic and
mechanical
systems
Microphone
position,
Microphone
selection
Oil analysis
Chemical analysis of
fluid sample
Chemical
properties,
ammount of
addatives
Any industry
working with oils
Clean sampling
equipment
Position
Comparrison of
normal
position/speed and
actual position/speed
Position,
Vibration
Most industries
Sensor accuracy,
reference point
Power
Input, output and

efficiency of machine
Consumption and
production of
machine
Most industries
where sensors
are already fitted
Use existing
sensors, what to
monitor, reasons
for changes
Smell and Taste
Smell and taste of

products and
surrondings
Smell and taste
Food, drinks and

pharmaceutical
Results affected by
environment
Temperature
Temperature
Temperature
Most industries
Sensor selection
Thermography
Surface temperature
using IR camera
Surface
temperature
Most industries
Specifications of
camera
Complex systems,
where to measure
Configuration of
borescope
Vibration
Increasing vibrations
Vibrations
Any industry
working with
rotating
machinery
Visual
Inspection
Visual
Look for flaws in

development
Vessles such as
heat exchangers
35
Wear and
debris analysis
Analysis of debris
from fluids
Fluid samples
Any industry
working with oils
Analyser selection,
clean sampling
equipment
3.3.18 Applicable CM methods

From section 3.2.4 and Appendix 2: FMECA Report it can be seen that heat exchangers and
compressors are equipment of a critical nature. Implementing CM for these equipment types will
provide early warnings for developing failures and hopefully prevent failure and downtime for the
system. From Table 5 CM methods that are best suited for monitoring heat exchangers and
compressors can be chosen. It should be noted that Table 5 only presents the general type of
monitoring method and that many different commercial solutions within one method may exist.
Finding the best solution may be challenging and time consuming, therefore only a recommendation
for where to look further will be given here.
For monitoring compressors the obvious choice is Vibration monitoring presented in 3.3.15, but both
Power monitoring and Load monitoring presented in 3.3.11 and 3.3.6 respectively are also
possibilities. Finding industrial solutions that corresponds to these monitoring techniques should not
be a problem as all three methods are used today. Some industrial solutions include ABBs
DriveMonitor and GEs System 1 software for CM. System 1 can be configured to monitor most assets
and extension packages exists to specialise the software to suit special needs (GE). System 1 was not
covered in the previous sections because it is a CM software package and may therefore be
compromised of several of the different methods presented earlier depending on how the system
has been configured.
When it comes to CM methods applicable to heat exchangers the list of possibilities diminishes.
What is done today is internal visual inspection. This method however has the huge disadvantage of
requiring major dismantling of the heat exchanger, meaning that the whole system needs to be
shutdown and cleaned before the inspection can be done. The inspection interval can depend on
local rules and regulations as well as the company rules of the operator. In some cases the inspection
interval can be decided by a Risk Based Inspection (RBI). RBI is a decision making technique for
inspections based on risk. Both the consequence of failure and the probability of failure are included
(DNV, 2009). The RBI delivers the following to the inspection program:
What to inspect
When to inspect
Where to inspect
How to inspect
What to report
The result of the RBI is an inspection plan. The period between inspections will also be defined. For
static equipment the inspection interval may return as one inspection every 5 or 10 years as an
example. When an internal inspection is due and the heat exchanger has been dismantled either leak
gas testing or ECT can be used to detect developing cracks and leaks in the tubes.
However inspection of the internals of the heat exchanger requires the whole system to be
shutdown, this is only something that is desired to do as rarely as possible due to the costs.
Therefore a monitoring technique where the heat exchanger does not need to be shutdown would
36
be preferable. HXAM is one such solution, although still in development and pilot testing HXAM
seems to be a good alternative in addition or as a replacement for RBI. A prerequisite for
implementing HXAM is that sensors are already fitted to the heat exchanger and that the information
received from these sensors is reliable. In new facilities fitting the required sensors and
implementing HXAM should not be a problem.
37
3.4 OBM simulations using GA

In this section some simulations using GA will be run to try and find an optimal organisation of
equipment for OBM. All simulations will be done in Excel with the aid of an upgraded version of the
Excel Solver developed by Frontline Systems called Premium Solver. The model can be used as a
decision support tool when considering equipment for OBM, given certain inputs from the user. The
model requires the user to specify the expected length of an opportunity, probability for logistics
delay and length of the delay, need for scaffolding, and the equipment that is to be considered.
3.4.1 Costs included in the model
In this optimisation problem the objective is to organise equipment into an OBM scheme that has the
lowest total cost. The following cost functions where used:
Logistics delay costs (CLD)
o Costs of extra trip with spares, tools and people. Also includes probability for delay
and increased costs if there is a delay.
Work costs (CW)
o Costs related to the work associated with changing the component including wages
and administration etc.
Downtime costs (CDT)
o Costs associated with having to shutdown the facility
Risk costs (CR)
o Costs associated with keeping a unit nearing the end of its life in operation
RUL costs (CRUL)
o Costs associated with changing the unit before its useful life has been exhausted.
Spares and tools (CST)
o Costs of spare parts and specialised tools that may be required.
Miscellaneous costs (CMI)
o Miscellaneous costs include different smaller costs not included in the other costs
categories.
Scaffolding (CS)
o Costs associated with renting scaffolding and putting up scaffolding if required.
These costs where combined to give the total unity cost for each item.
= + + + + + + +
The logistics delay cost was modelled using an if sentence. If logistics delay is not expected the
costs of logistics is set to 20 000 NOK. If logistics delay is expected the costs of the logistics costs
becomes the product of the length of the delay multiplied by a flat rate of 15 000 NOK. The length of
the delay is user defined.
The work costs were based on the MTTR for each unit and multiplied by an hourly rate set to 2 000
NOK. The MTTR has been taken from OREDA, mean value has been used.
Downtime costs was modelled as 100 000 NOK per hour and multiplied by the length of the
opportunity. The length of the opportunity is specified by the user.
38
The risk cost was modelled based on the failure category from OREDA and a flat rate of 2 000 NOK
for all units. If the unity is not included in the OBM scheme then the flat rate is increased to 2 250
NOK. This is expected to reflect the increased risk of failure when postponing maintenance on an
item. The Critical failure category was given a rating of 3, the Degraded failure category was given a
rating of 2 while the Incipient failure category was given a rating of 1. The Unknown failure category
was excluded. The criticality rating was then used as a power for the flat rate. This organising means
that a degraded failure carries more costs and higher risks than an incipient failure.
The RUL costs were modelled using the investment cost of the unity multiplied by the remaining
useful life and then divided by the mean time between failures (MTBF) for that unit. This
representation means that the RUL cost for each unity will be unique and depend on whether the
unit in question is unique to this system e.g. compressor, separator etc. or more common and off the
shelf e.g. sensors. The cost of RUL can be seen in equation 2. In equation 2 is the failure rate.
=
2
1
The investment costs assumed in this thesis can be seen in Table 6. The investment costs are
assumed based on engineering judgement as well as how customised the unit is e.g. a compressor
will be more tailor made to this installation than a temperature sensor will be.
=
Table 6: Investment costs
Investments
Equipment
Coalescer
Compressor
Contactor
Heat
Exchanger
Pump
Scrubber
Sensors
Separator
Valves
kr
kr
kr
Cost
5 000 000
9 000 000
8 000 000
kr
10 000 000
kr
kr
kr
kr
kr
900 000
3 000 000
500 000
10 000 000
750 000
The costs related to spares and tools were assumed to depend on the repair time. The higher the
repair time the higher the costs. It is assumed that the longer it takes to repair a certain failure the
higher the probability that special tools or spares will be needed. The cost is quantified by a flat rate
of 10 000 NOK plus a time rate of 5 000 NOK/hours multiplied by the MTTR.
The miscellaneous costs were set to 1 500 000 NOK for both doing maintenance at the current
opportunity and postponing the maintenance work. It is believed that this is sufficient to cover
anything that may have been forgotten.
If the unit under consideration is located in a place that is hard to reach or dismantling is needed
then there may be a need for scaffolding. The need for scaffolding was determined based on the
failure mode, the fault type and engineering judgment. The amount depends on user input of yes or
no. If the unit requires scaffolding then costs will be higher than if scaffolding is not required. The
39
costs of scaffolding at opportunity basis was set to 750 000 NOK if scaffolding is required and
100 000 NOK if scaffolding is not required. If maintaining the unit is postponed the costs of
scaffolding was set to 600 000 NOK if required and 75 000 NOK if scaffolding is not required. The
differences in costs are believed to cover the sudden need for scaffolding at opportunity basis and
that there is room for more planning later.
To be able to run any simulations it was necessary to define some objectives and constraints. The
objective has already been discussed. The initial constraint used in this problem was that the time of
the selected maintenance jobs could not exceed the length of the opportunity. The decision to
include a unit was set to be binary. If the decision variable returned as zero the unity would not be
included and if it returned as one it would be included.
3.4.2 Developing the model
The model is set up in such a manner that the costs of doing maintenance on opportunity basis will
occur now and the costs of delaying the maintenance action will occur some time in the future. To be
able to compare a cost that occurs now with a cost that may occur in the future it is necessary to use
present values. The equation used in calculating the present value can be seen in equation 3
3
= (1 + )
In equation 3 the P is the present value today, while the F is the amount that occurs some time in the
future. The p is the interest rate corrected for inflation and the n is the time until the cost occurs in
years. The interest rate will change over time depending on several factors. One of these factors is
the Consumer price index (CPI). Norges Bank publishes both the key policy rate and the CPI the
predictions of both can be seen in Figure 26 and Figure 27 respectively. As these figures show the CPI
is expected to be stable at around 2.5 % while the interest rate is more unpredictable. In the
simulations the interest rate was set to 4.7% and the inflation rate was set to 2.5%. These numbers
where kept the same in all of the simulations. The time until the cost F will incur was set to 1.5 years.
This time horizon seems reasonable since no unit in this case have a RUL that is longer than 3 000
hours. It can therefore safely be assumed that all of the units will have failed 1.5 years from now if no
maintenance is carried out.
Figure 26 Key policy rate development and prediction by Norges Bank (Norges Bank)
40
Figure 27 - CPI development and prediction by Norges Bank (Norges Bank)
The objective function that the GA should try and minimise was expressed as the costs of doing
maintenance later subtracted with the units chosen for OBM subtracted by the cost for maintaining
these units later. This ensures that all elements are only counted once. A mathematical expression of
the objective function can be found in equation 4.
= _ _
The expected length of the opportunity can be specified by the user along with the length of the
logistics delay. The time available to do maintenance was set to be the length of the opportunity
subtracted by the logistics delay if a delay in the logistics chain has occurred. It is believed that this
represents a more realistic view of the world than just having the whole opportunity available to do
maintenance. When a delay in the logistics chain has occurred it is unlikely that the crew offshore will
be able to do any maintenance work aside for minor preparations, therefore the opportunity window
is reduced depending on the length of the delay. Both the length of the opportunity, length of delay
and whether there will be a delay or not are all parameters that are subject to user input. A
screenshot of the entire OBM model can be found in Appendix 5: OBM model.
3.4.3 Results and comments
Several issues were discovered during the development of the model. The issues have been mended
as they have been discovered.
When the model had been developed to a satisfactory level some simulations where run. In this
section only the minimum costs will be presented. Several scenarios of interest have been identified.
One scenario should have a long opportunity (+100 hours), the second scenario should have a
medium length opportunity (50 100 hours) and in the final scenario should have a short
opportunity (< 50 hours). All scenarios where run with and without logistics delay obtaining six
different scenarios in total. After each run the model is reset to not doing maintenance on
opportunity basis.
41
Table 7: Summary of scenarios
Length of
Length
Delay
opportunity of delay included
1
120
15
yes
2
120
15
no
3
75
15
yes
4
75
15
no
5
40
15
yes
6
40
15
no
Three runs will be run for each scenario. This is done to check the results after the GA have been
applied. This is important because of the random selection process applied by the GA. The random
selection means that in theory different results can be obtained each time. Running three
simulations is a simple check to see if the results come up the same. Ideally several runs should be
made, but do to time constraints I have decided to limit myself to three runs to check the solutions.
To make the results easier to read the cells containing the decision variables have been given unique
names making it easier to separate them form one another. The relationship between the cell
naming and the actual equipment can be interpreted as follows:
Scenario
The capital letters gives the name and type of the equipment i.e. HEST means Heat
Exchanger Shell and Tube.
The D or I after the name indicates if the failure type is either Degraded or Incipient
The number at the end is used to separate different failure modes.
The relationship can be seen in Table 8.
Table 8: Relationship between equipment name, failure mode, type and cell name
Name
Coalescer
Coalescer
Compressor
Compressor
Compressor
Compressor
Contactor
Contactor
Contactor
Heat Exchanger Plate
Heat Exchanger Plate
Heat Exchanger Shell and tube
Pump
Pump
Pump
42
Failure Mode
Parameter deviation
External Leakage - Utility
Low Output
Internal Leakage
External Leakage - Process
Internal leakage
Parameter deviation
Structural deficiency
Vibration
Failure Type
Degraded
Incipient
Degraded
Degraded
Incipient
Incipient
Degraded
Incipient
Incipient
Degraded
Incipient
Degraded
Degraded
Incipient
Incipient
incipient
Degraded
Degraded
Degraded
Cell Name
COL_D1
COL_I1
COM_D1
COM_D2
COM_I2
COM_I1
CON_I1
CON_I1
CON_I2
HEP_D1
HEP_I1
HEST_D1
HEST_D2
HEST_I1
HEST_I2
HEST_I3
PU_D1
PU_D2
PU_D3
Pump
Incipient
PU_I1
Pump
Incipient
PU_I2
Scrubber
Degraded
SCR_D1
Scrubber
Incipient
SCR_I1
Scrubber
Incipient
SCR_I2
Sensors
Erratic output
Degraded
SEN_D1
Sensors
Incipient
SEN_I1
Sensors
Low output
Degraded
SEN_D2
Sensors
Erratic output
Incipient
SEN_I2
Sensors
Other
Degraded
SEN_I3
Separator
Plugged/Chocked
Degraded
SEP_D1
Separator
Parameter deviation
Degraded
SEP_D2
Separator
Incipient
SEP_I1
Separator
Incipient
SEP_I2
Valves Ball
Delayed operation
Degraded
VALB_D1
Valves Ball
Degraded
VALB_D2
Valves Ball
Incipient
VALB_I1
Valves Ball
Incipient
VALB_I2
Valves Gate
Valve leakage in closed position Degraded
VALG_D1
Valves Gate
Other
Degraded
VALG_D2
Valves Gate
Incipient
VALG_I1
Valves Gate
Incipient
VALG_I2
The first scenario was with a long opportunity i.e. more than 100 hours. The length of the
opportunity was set to 120 hours in this case. This scenario was run without delay in the logistics
chain. The length of the logistics delay was set to 15 hours. The simulation was set to run for 15
minutes. After three runs of the simulation the lowest cost for scenario 1 was 10 272 062 NOK. This
result was obtained in two of the three runs indicating that this is the optimum. Interestingly the
equipment selected in the two runs that obtained the minimum cost where different. But both runs
selected 23 units. The total time used in scenario 1 was 119.9 hours of the 120 hours available. The
results from the two runs that produced the minimum cost can be seen in Figure 28 and Figure 29. In
Figure 28 and Figure 29 a one in the final value column indicates that this unit is included in the OBM
scheme.
43
Figure 28 - Units included in scenario 1 run 2
44
The second scenario was set up with a delay in the logistics chain and the same length in
opportunity. In this scenario the time available to do maintenance was reduced from 120 hours to
105 hours i.e. the logistics delay was 15 hours. After three runs the minimum cost obtained from
scenario 2 was 71 474 244 NOK. In this scenario the model included 21 units in the OBM scheme.
This result was obtained in two of the three simulation runs. The units included were the same in
both cases. In scenario two the time used was 104.9 hours. The units included in this case can be
seen in Figure 30.
45
Figure 30 - Units included in scenario 2
In scenario three the length of the opportunity was set to 75 hours. In scenario three the simulations
were run without logistics delay. So time available for maintenance was 75 hours. After three runs
the lowest costs were obtained in two simulation runs and was 180 785 167 NOK. This result was
obtained using different selection in both scenarios but the number of units included in both cases
was 17. The different selections can be seen in Figure 31 and Figure 32. The time used in this scenario
was 75 hours.
46
47
The fourth scenario was run with the same length of the opportunity as in scenario three, but in this
case the logistics delay came into play, reducing the length of the opportunity to 60 hours. After
three runs the lowest cost that was obtained was 254 351 906 NOK. In this simulation 14 units was
included in total and the time used in this case was 58.1 hours. Figure 33 shows the units included in
scenario 4.
48
Figure 33 - Units included in scenario 4
In the fifth scenario the length of the opportunity was set to 40 hours without logistics delay. After
three simulation runs the lowest cost obtained in two of the simulation runs was 271 909 927 NOK.
The two runs that produced the minimum cost selected different units but both produced the same
minimum cost. The selection can be seen in Figure 34 and Figure 35. The time spent on maintenance
in this scenario was 39.5 hours.
49
50
In the sixth scenario the length of the opportunity was the same as in scenario 5 i.e. 40 hours.
However, in this scenario the logistics delay was included and thus the time available to do
maintenance was reduced to 25 hours. The minimum cost in this scenario was 333 048 877 NOK. This
simulation run selected 7units for OBM. The time used in this case was 24.9 hours. In this case the
minimum cost was obtained by selecting the same unity in the two of the three simulation runs.
Selected units for this scenario can be seen in Figure 36.
51
Figure 36 - Units selected in scenario 6
After all the simulations had been run a table was compiled summarising the results obtained. The
results can be seen in Table 9.
52
Table 9: OBM result summary
Time
Time
available
used
[hours]
[hours]
1
kr
10 272 062
23
120
119,9
2
kr
71 474 244
21
105
104,9
3
kr
180 785 167
17
75
75
4
kr
254 351 906
14
60
58,1
5
kr
271 909 927
10
40
39,5
6
kr
333 048 877
7
25
24,9
As Table 9 shows the costs of doing maintenance on opportunity basis increases as the length of the
opportunity decreases, the number of units included also decreases with the decreasing length of the
opportunity.
Scenario
Minimum cost
Units
included
The corrected interest rate and the time horizon will both have an impact on the present values used
in these simulations. In order to determine how big this impact will be a sensitivity analysis was
performed. During this analysis the inflation was kept constant at 2.5 %. By comparing the sums of
postponing maintenance and doing maintenance at opportunity basis it was determined that an
interest rate of approximately 25 % would be interesting to investigate further. With an interest rate
at 25 % the total costs of doing maintenance on opportunity basis were higher than the costs of
doing maintenance later. A new simulation was run with the interest rate at 25 %. This simulation
was run with the same inputs as scenario number one. This scenario was run for 5 minutes to get a
quick result. It should be noted that an interest rate of 25 % is much higher than the interest rate
predictions published by Norges Bank (Figure 26). The result from this simulation was that the total
cost of OBM became 78 541 140 NOK. This result was not unexpected as the goal of the model is to
minimise the costs associated with OBM. When the costs of postponing the maintenance work to a
later date are lower than those of doing maintenance now and the way that the objective function
have been defined a negative result is not out of the question. This selection included 23 units and
spent 119.9 of the 120 hours available. To get a result comparable to the results obtained earlier an
additional constraint was added. This constraint stated that the objective function had to be larger or
equal to zero. During this simulation the time horizon was kept the same at 1.5 years from now. As
already discussed in section 3.4.2 the time horizon and its impact will not be investigated further.
With the new constraint the simulation minimum cost was 148 772 NOK this selection included 21
units and spent 119.8 of the 120 hours available. In Figure 37 a screenshot of a parameter analysis of
the total costs later can be seen. The figure show how big an impact the value of different cells has
on the final results. In Figure 37 cell B9 is the length of the opportunity, B6 is the interest rate and B5
is the time horizon. Figure 38 show the same analysis but this time for total costs now. In this figure
H3 is the criticality rating and B9 is the length of the opportunity. A red column in Figure 37 and
Figure 38 indicates a negative impact on the final value of the cell and a blue column indicates a
positive impact.
53
Figure 37 - Screenshot from parameter analysis of total costs later
Figure 38 - Screenshot from parameter analysis of total costs now
54
One last simulation was run with a more realistic interest rate of 8 %. After one simulation that ran
for 5 minutes the minimum cost was 13 973 002 NOK. This selection included 22 units and used
118.1 of the 120 hours available.
From Figure 37 it can be seen that the interest rate (B6) is one of the parameters that has the
potential to greatly influence the results in both a positive and a negative direction. The time horizon
(B5) also has the potential to influence the result but as already discussed this effect can be
neglected. As shown above the interest rate needs to be at an unrealistic level (approximately 25%)
before any real impact will materialise.
55
4 OBM and logistics planning

Planning of the logistical needs of any facility or technical system should ideally start in the concept
development phase. It is important to consider logistical aspects in all phases of the life cycle of any
system. A supportability analysis can in this respect be very useful. The supportability analysis is the
foundation on which logistical requirements are founded (Blanchard, 1998). The supportability
analysis will influence the shape of plans later in the design process.
In the operational phase ongoing improvements to support plans and capabilities made or assumed
in earlier life cycle phases are made. The changes are implemented through reporting of operational
and maintenance data. This data is first collected before it is analysed. The results are used to
improve the supportability analysis from an earlier life cycle phase. The updated supportability
analysis is in turn used in the procurement process for additional spares. Major problem areas are
also noted and recommendations for corrective actions or improvements are initiated where
possible. This iterative process is illustrated in Figure 39.
Figure 39 - SA cycle
The maintenance planners have demands for spares, tools and people that need to be satisfied by
the logistics planners. In order to meet these demands the logistics planners require input from the
maintenance department. The input could include priority of certain spares or tools as well as the
size and weight of what is to be shipped to the facility.
4.1 Logistics planning and maintenance today

Most offshore operations today are divided into three different areas of responsibility:
1) Drilling and well construction
2) Reservoir and production management
3) Operations and maintenance
57
These three entities relays on a common logistic supply support. When it comes to maintenance it is
common to use an overcapacity strategy (Brurok & Sleire, Opportunity Based Maintenance in
offshore operations, 2009). The overcapacity strategy is designed in such a manner that the total
scheduled activity never exceeds a limit of the total available capacity at any time e.g. 70%. This
ensures that unforeseen events such as sudden failures can be handled by the remaining capacity
e.g. 30%. However even with this overcapacity strategy some jobs are still postponed to the next
period. A job can be postponed due to failures requiring immediate attention or delayed logistics
support vessels due to weather conditions. In the end jobs from earlier planning periods accumulate.
It should also be noted that a job cannot be scheduled until it has been technically planned meaning
instructions, safe job analysis, required skills, tools, materials etc. needs to be defined before the job
can be scheduled.
If maintenance tasks are grouped together costs can be shared. Examples of cost that can be shared
includes logistical support cost related to tools and transportation of personnel and spares both
offshore and possibly to the supply base as well, down time costs, costs associated to administration,
coordination and planning as well as completion and documentation. When organising maintenance
tasks into maintenance packages it is beneficial to organise groups that require similar resources
such as tools, competencies, transportation, personnel etc. In addition it is important to look at what
systems that are interconnected. If one system is shutdown for maintenance then it is a good idea to
identify other equipment that can be maintained at the same time without requiring further
maintenance shutdowns. The condition of the equipment is also important when organising
maintenance packages. When is it sensible to do maintenance? Is it possible to continue and risk a
shutdown due to failure or should an opportunity be exploited since the maintenance shutdown is
planned anyway. Other aspects to consider are the maintainability and access. If major components
must be dismantled to do maintenance it is wise to inspect and possibly maintain other components
at the same location. Doing so will reduce the need for major dismantling later. This will help in
avoiding long shutdowns later.
The nine basic elements of logistics support can be seen in Figure 40. Each of these nine elements
needs to be considered in some way when maintenance and logistics plans are made. What each
element in Figure 40 contains will be described in the following section.
58
Figure 40 - Elements in logistics support
Technical data, information systems/databases is a necessary element in logistics support because

this is where information regarding procedures dealing with system installation, check out and
calibration and instructions regarding operations, maintenance, overhauls etc. are stored. Access to
this information through out the life cycle is important. The maintenance and support planning entity
contains all planning and analysis relevant for the establishment of requirements for the system
through all life cycle phases. Supply support includes all spares for repairable equipment and repair
parts for non repairable equipment, consumables such as lubrication etc and special supplies to
maintaining the primary mission related equipment. Supply support also includes procurement
activities and documentation related to material acquisition, handling, recycling and disposal.
Maintenance and support personnel category includes personnel needed for performing
maintenance and support elements related to the systems primary mission. But other elements are
also included such as test equipment. This category includes personnel on all levels. Maintenance
facilities include facilities related to the support of scheduled and unscheduled maintenance
activities. Maintenance activities at all levels need to be considered. Facilities that need to be
considered include physical plant, housing, workshops and calibration laboratories. The
equipment/resources category includes all tools, CM equipment, test and calibration equipment and
equipment for handling all scheduled and unscheduled maintenance activities for the system in
question. Equipment requirements for all levels of maintenance must be addressed. The distribution
category encompasses all materials, special provisions, equipment, containers etc. needed to support
the packing, handling, storing and transport of mission related elements such as personnel, spares,
test equipment etc. In short the distribution category covers the transport requirements for
supporting the system in performing its mission. In the training and training support category
personnel, equipment, facilities, documentation and associated resources needed for training
operational and maintenance personnel is located. This category contains resources for both initial
and replenishment training. Equipment such as simulators, mock ups etc. used in both informal day
59
to day training as well as more formal training can be developed when required. The last category of
basic logistics support elements is computer resources. This category includes computers, software,
interfaces and networks necessary to assist in scheduled and unscheduled maintenance activities.
Software for CM, diagnostic tapes, etc. can also be included in this category.
Today supply chain management is the technical term most used when it comes to logistics planning.
Supply chain management involves managing a network of businesses involved in providing a
product or service package as required by the end customer. Supply chain management spans all
areas from point of origin to point of consumption i.e. the supply chain (Wikipedia, 2010). Supply
chain management must address problem areas such as the configuration of the distribution
network, the distribution strategy, information flow, managing inventory, cash flow and trade offs
in logistics activities. Of particular importance for the interaction between maintenance and logistics
planning today is the distribution network configuration, information flow and inventory
management. The configuration of the distribution network is important because it defines the
number, location and network mission of different suppliers, production facilities, warehouses etc. In
other words the configuration of the distribution network defines how many, where and what.
Information flow is important because sharing of information regarding demand signals, forecasts,
inventory levels etc. will enable better collaboration between different actors along the supply chain.
A challenge is to decide who and how much information different actors in the supply chain should
be privy to. Having an overview of the inventory through inventory management will help in
managing the supply chain through improved information regarding the quantity and location of
inventory such as spares and tools. Inventory management could also be extended to include where
in a repair cycle repairable items are. All these factors are important when planning logistics
activities. Depending on the demand from the maintenance planners different modes of transport
between suppliers, warehouses, base and offshore needs to be considered to minimise the logistics
costs as well as the maintenance downtime.
Trade Offs in logistics activities may be needed and finding an optimal solution can be challenging.
Close coordination of all elements involved is required to obtain the lowest total logistics cost. An
example from road transport illustrates this point. The rates for full truck loads are more economical
on a cost pr. load basis compared to less than full loads. But transporting only full trucks may
increase the costs associated with keeping inventory and warehousing thus full load transport may
not be the option that delivers the lowest total logistics cost. As the hypothetical example illustrates
the whole supply chain needs to be considered when making decisions affecting the whole supply
chain. The same point applies to transportation to and from offshore facilities.
In order to deal with deviations in the supply chain supply chain event management can be used.
Supply chain event management considers all possible events and factors that can disrupt the supply
chain. Different scenarios are created and solutions can be found.
Vulnerability assessment is another method of identifying and mitigating unwanted events that may
influence the supply chain. The vulnerability assessment differs from risk analysis in that the focus is
different. While risk analysis focuses on human, environmental and property impacts of an accident,
Vulnerability analysis is focused on the system mission and the survivability of the system
(Asbjrnslett, 2009).
60
4.2 OBM influence on logistics planning

A transition to a maintenance philosophy where OBM is predominant is challenging because
opportunities occur at a short time horizon and therefore the organisation needs to utilise short term
maintenance planning which in itself is challenging. However there are several benefits of
implementing OBM as a part of the organisations maintenance plan. One of the largest contributors
to operational expenditures is lost production. Implementing OBM will therefore be beneficial as
little or no extra production is lost, aside form what already has been lost in the event that triggered
the opportunity. OBM will also keep logistical support costs at a minimum as OBM efficiently utilises
the same resources. If the goal is to reduce down time costs and increase availability, while keeping
HSE (Health, Safety and Environment) levels, system reliability and functionality at the required level
then OBM is the answer.
The oil and gas industry today only uses OBM to a limited extent (Brurok & Sleire, Opportunity Based
Maintenance in offshore operations, 2009). The reason for this is that the cost of lost production
greatly overshadows the benefit of exploiting an opportunity. This reason is a cultural issue where
fixing a fault is valued higher than exploiting opportunities triggered by mishaps and plan deviations.
The lack of the total picture also restricts the ability to see the benefits from utilising opportunities.
If a maintenance strategy based on OBM is implemented there will be an impact on the logistics
planning. Areas such as availability of spares, tools, personnel and procedures etc. can be impacted if
an OBM strategy is implemented. To reduce the probability of logistical delay due to these factors
the establishment of standard procedures and increased levels of training for offshore personnel can
be beneficial. This will help in making the planning period ahead of maintenance jobs shorter as well
as to a large degree eliminate the need for specialised personnel. Should specialised personnel still
be required it may be enough to have the specialist at the OSC and not offshore. The specialist will
then guide the offshore crew in the maintenance work. Communication between the OSC and
offshore is done using ICT. To ensure that spares and tools are available when they are needed an
increased use of pre positioned spares and tools can be used either offshore or at the supply base.
If space is limited offshore or at base supplier retention guaranteeing that tools and spares will be
available in a short time can be used as an alternative. Predefined procedures will to a large degree
eliminate the need for planning and performing analyses such as safe job analysis before a job can
scheduled. Having procedures in place that clearly defines what is to be done along with what is
required to get the job in place will also make the job of the logistics planners easier as well as giving
them more time to plan and procure the necessary services.
As discussed in section 4.1 maintenance planning influences some areas related to supply chain
management. In an OBM scheme it is beneficial to put even more emphasis on these areas. How the
distribution network is configured will play a larger role when it comes to OBM. The distances
between suppliers, warehouses, supply bases and offshore consumers are even more important if an
OBM scheme is implemented. This is because opportunities for maintenance will occur on a short
time horizon and as such transporting spares, tools and people between suppliers and supply base
becomes a time sensitive matter. The flow of information between the offshore production and
maintenance department and the onshore OSC is important. This is the channel that communicates
information regarding need for tools and spares that needs to be ordered from warehouses or
suppliers. Inventory management is also important in an OBM based maintenance strategy. Having
61
spares and tools readily available in the warehouse inventory means that purchasing of new stock
does not need to be done during the time sensitive opportunity.
Organising maintenance tasks into maintenance packages is expected to contribute to better
utilisation of opportunities. To effectively pick necessary spares and tools required in any given
maintenance package a fully automated warehouse with radio frequency identification (RFID) tags,
identifying single pieces of equipment or an equipment category, can be used. Most RFID tags today
contain two parts the first part is an integrated circuit where information is processed and stored and
radio signals are modulated and demodulated. The second part is an antenna used for receiving and
transmitting the signal (Wikipedia, 2010). Using RFID will help in improving the efficiency of inventory
tracking and management. The idea is that an order for a maintenance package is sent to the
warehouse management software. Once the order is received the automated system reads the
identification tags required to build the package and then starts to pick necessary equipment and
tools from the warehouse. The layout of the warehouse will depend on the supplier of the
automated warehouse. But in general the warehouse will consist of metal racks with narrow aisles
between them and rails running down the centre of the aisles. Along the rail a pole can travel up and
down the aisle. Once the pole is in the right position a carriage travels up or down the pole to the
location of the desired payload. A tool for grabbing the payload extends and either picks up the
payload or delivers a new payload to the stock. The stock level for tools and spares is reported once
the equipment is picked up and a warning if stock is low is sent to the purchasing department.
Some of the benefits of implementing an automated warehouse are that inventory control will
improve. It will become easier to keep accurate records of available inventory. The labour
productivity will improve while strain on the workers will decrease as the automated systems will
take care of all the heavy lifting. Excessive handling of equipment will be reduced along with damage
to tools and spares as well as misplacing tools and spares. An automated warehouse will also
increase the productive capacity of existing space since the automated systems do not require the
same amount of space to move around on as humans do. With an automated warehouse it is also
possible to include more shelves in the height thus increasing floor space utilisation. The automated
system will also contribute to lowering the response time as no one has to walk around looking for
equipment in the warehouse as well as the capability to quickly move equipment from the
warehouse to a packing station.
Factors speaking against an automated warehouse include the high investment costs along with a
very long payback period. Having implemented an automated system to a warehouse makes the
warehouse difficult and expensive to move, remove and modify. An automated warehouse will
contain mechanical and software components of a high complexity and therefore problem solving
and debugging is likely to require mechanics and electricians to fix. Also if a major part of the system
fails the entire system will likely need to be shutdown due to the high degree of integration and lack
of backup systems.
However the benefits of implementing an automated warehouse seem to outweigh the drawbacks
and limitations, but a cost benefit analysis may be required in order to verify this.
With reference to Figure 40 the following categories will be influenced by adopting an OBM strategy:
Maintenance and support planning
62
Maintenance and support personnel

Supply support
Facilities
Test equipment
The planning of maintenance and support activities and required resources will in an OBM strategy
be focused on maximising short windows of time. This is due to opportunities on a general basis
occurring on a short time horizon. Limited time for executing maintenance activities will mean that
the support plans needs to be agile enough to handle sudden demands for capacity while still being
though enough to withstand sudden changes this is also true for the organisation as discussed in
section OBM Considerations2.1. Maintenance and support personnel will be influenced in a similar
manner as the planning activities. An OBM strategy will in general be dominated with long periods of
low demand and short bursts where the demand is high. Savings can potentially be made by having a
smaller permanent staff and hire subcontractors when the demand exceeds what the permanent
staff can handle (Rasmussen, Operation Technology; Maintenance, 2002).
The supply support category will be influenced by the sudden raises in demands during opportunities
before stabilising to a lower level during normal operations. This may lead to higher storage and
warehouse costs during sustained normal operations. An alternative could be extended use of
supplier retention as described earlier to reduce some of the storing and warehousing costs.
In an OBM strategy the facilities need to be designed so that the through put is high and the service
time is as short as possible while still fulfilling requirements to HSE and the quality of work. High
through put and short service time is essential as opportunities to do maintenance are short and the
total downtime of the system as a whole should be kept at a minimum.
It is important that the location of tools and equipment for CM, testing and calibration is kept within
reach. This will ensure a more efficient utilisation of opportunities for either maintenance or
inspections and help in keeping the total downtime of the system as low as possible.
63
5 Conclusions
This master thesis covers the areas of offshore operations dealing with condition monitoring and
opportunity based maintenance. How OBM can and may effect the organisation and what is required
of an organisation that wishes to start utilising OBM have been discussed. Organisational factors that
needs to be considered or in place before OBM can be implemented includes: resilience, agility,
standard operating procedures, OSC and prepositioning of spares at supply base or through supplier
retention. Technical aspects that need to be considered before OBM is implemented includes spare
parts, need for scaffolding, RUL of unit in question etc. The technical aspects have been implemented
into a flowchart (Figure 2). The OBM flowchart can also be used as a general guide when it comes to
prioritising equipment for OBM.
A FMECA has been performed on a simplified process line for oil and gas. The purpose of the FMECA
is to identify critical equipment that should be considered for CM. The critical equipment in this case
included heat exchangers of shell and tube type and compressors of the centrifugal type. Valves also
had the potential to be selected for CM but were neglected due to time constraints as well as heat
exchangers and compressors had more failures ranked as critical.
Based on the equipment found to be most critical in the FMECA general CM methods have been
described with emphasis on detection method, data collected, application and elements that should
be considered when implementing this kind of CM. Specific CM methods that can be used to monitor
the failures found to be most critical in the FMECA have also been discussed. Some industrial
solutions and CM software packages have also been discussed where appropriate. For compressors
Vibration, power and load monitoring are possibilities. For heat exchangers HXAM or a similar real
time performance monitoring scheme is the only option that does not require major dismantling.
Based on the OBM scheme some optimisations have been run using GA. In total six scenarios were
run. The length of the opportunity was varied between short and long. Two scenarios were run with
the same opportunity. One of the scenarios had a delay in the logistics chain meaning that the time
available to do maintenance was decreased. To ensure that the results obtained were the minimum
cost three simulations were run for each scenario. This is due to the random selection process
employed by the GA. Three simulation runs have been judged sufficient to indicate the minimum
costs obtained are in fact the minimum costs. The minimum costs were lowest when the length of
the opportunity was long, with decreasing length of opportunity the costs increased. This has to do
with the defined objective function. A short opportunity means that most maintenance tasks have to
be postponed to a later date. A simple sensitivity analysis determined that the interest rate and thus
the corrected interest rate would have an impact if the interest rate became 25 % or larger. Minor
changes in the interest rate will not have a significant impact on the results.
How logistics planning can be related to maintenance of Oil and Gas facilities have been discussed
along with how OBM will impact logistics planning. Some measures that can help in mitigating the
impact of shifting to an OBM strategy have also been discussed. Some measures that can be
implemented to reduce the impact of OBM include fully automated warehouse with RFID tags
identifying spares and tools. Another element that is expected to help mitigate a shift to OBM is the
implementations of the support strategy used in space operations with the implementation of OSC
and prepositioning of spares and tools.
65
6 Future work
During the spring of 2010 a new OREDA handbook was published. The updated data in this edition
could potentially alter the information in the failure database along with the results of the FMECA
and the OBM organisation. It would be preferable to check the data used in this thesis against the
data in the new edition of the OREDA. Updates and changes should be made were necessary and the
FMECA and OBM organisation should be performed again if changes have been made.
The OBM model could be developed further and made more advanced with implementing failure
distributions for the failure modes. This could give better and more realistic results.
In this thesis most cost figures have been assumed based on engineering judgment. Some of these
assumptions may very well be off the mark. Time and effort should be put into confirming the
assumed figures used in this thesis. I have talked with some of my fellow students who needed
relevant cost figures and they have told me that such figures are hard to come by. Therefore it may
be difficult to get confirmation on the numbers used in this thesis.
Only detailed CM methods from ABB have been used in this thesis, this is because I know people
working with this on a daily basis in ABB. Other providers of CM equipment and methods have not
been contacted for detailed information. Given more time this should be done. In this thesis only
some information available from company websites has been used. Interesting suppliers of CM
equipment includes GE, Siemens, SKF, ValveWatch etc. Company web pages are included below:
www.ge.com
www.siemens.com
www.skf.com
www.valvewatch.com/
67
7 Works Cited
ABB. (2008). DriveMonitor. Retrieved April 8, 2010, from ABB:
http://search.abb.com/library/Download.aspx?DocumentID=3BHT%20490%20532%20R0001&Langu
ageCode=en&DocumentPartID=&Action=Launch&IncludeExternalPublicLimited=True
ABB Oil, Gas and Petrochemicals. Integrated Operations Overview. Oslo: ABB.
Asbjrnslett, B. E. (2009). Assessing the Vulnerability of Supply Chains. In Supply Chain Risk (pp. 15 33). Springer US.
Blanchard, B. S. (1998). Logistics Engineering and Management (5th ed.). Prentice Hall International.
Brurok, T. (2010, May 8). Meeting. Trondheim, Norway.
Brurok, T., & Sleire, H. (2009). Opportunity Based Maintenance in offshore operations. 22nd
International Congress of Condition Monitoring and Diagnostic Engineering Management
(COMADEM 2009).
Devold, H. (2009). Oil and gas production handbook - An introduction to oil and gas production (2
ed.). Oslo: ABB.
DNV. (2009). DNV-RP-G101 Risk Based Inspection of Offhosre Topsides Static Mechanical Equipment.
Oslo: DNV.
DNV. (2007). DNV-RP-G103 Non- Intrusive inspection. Oslo: DNV.
Erikstad, S. O. (2009). TMR4115 - Design Methods Lecture note #5 - Genetic Algorithms. Retrieved
January 2010
GE. (n.d.). System 1 Software. Retrieved May 3, 2010, from GE Energy: http://www.geenergy.com/prod_serv/products/oc/en/system_soft.htm
Hunt, T. M. (2006). The Concise Encyclopaedia of Condition Monitoring (1st ed.). Oxford: Coxmoor
Publishing Company.
IOCenter. (2008, May 14). Integrated Planning. Retrieved April 12, 2010, from IOCenter e-Room:
https://project.sintef.no/eRoomReq/Files/petroleum/IOcenter/0_26d75/20080423.P33%20Lecture.I
ntegrated%20Planning%20Concept.pdf
Kristiansen, S. (2004). Maritime Transportation Safety Management and Risk Analysis. Trondheim:
Butterworth-Heinemann.
Kristoffersen, J. I. (2009). Project Thesis - Use of Condition Monitoring data in maintenance and
logistics planning and optimisation. Trondheim.
Norges Bank. (n.d.). Inflation. Retrieved May 25, 2010, from Norges Bank: http://www.norgesbank.no/templates/article____12346.aspx
69
Norges Bank. (n.d.). Key policy rate. Retrieved May 25, 2010, from Norges Bank: http://www.norgesbank.no/templates/article____41194.aspx
NORSOK STANDARD. (2001). NORSKO STANDARD Z-008: Criticality analysis for maintenance
purposes. Oslo.
Ramstad, L. S., & Wahl, A. M. (2010). Offshore Integrated Planning and Performance Management.
Trondheim: Sintef IOCentre.
Ramstad, L. S., Wahl, A. M., & Halvorsen, K. (2010). Imporved Coordination with Integrated Planning:
Organisational Capabilities. Trondheim: MARINTEK.
Rasmussen, M. (2003). Driftsteknikk Grunnkurs. Trondheim: Tapir.
Rasmussen, M. (2002). Operation Technology; Maintenance. Trondhiem: Tapir.
Rasmussen, M. (2007). TMR 4310 Marin Teknikk 4 Plitelighets- og tilgjengelighets- betraktninger for
design og drift. Trondheim: Tapir.
Roxar. (n.d.). Brochures. Retrieved March 16, 2010, from Roxar:
http://www.roxar.com/getfile.php/Files/Roxar%20Brochures/on-line_communication.pdf
Saranga, H. (2004). Opportunisitic maintenance using genetic algorithms. Journal of Quality in
Maintenance Engineering , 10 (1), 66-74.
SINTEF Industrial Management. (2002). OREDA Offshore Reliability Data Handbook (4th ed.).
Trondheim: OREDA participants.
Thorstensen, T. A. (2001, January 28). Notat kurs TMR4260 Driftsteknikk grunnkurs. RCM - metoden .
Trondheim: Institutt for marint maskineri, NTNU.
Vrnes, R. (2008). Nettverksbaserte Organisasjoner - Kunnskapsstatus og relatert forskning.
Trondheim: MARINTEK.
Wahl, A. M., Sleire, H., Brurok, T., & Asbjrnslett, B. E. (2008). Agility and Resilience in Offshore
Operations. Retrieved February 2010, from IOCenter e-Room:
https://project.sintef.no/eRoomReq/Files/petroleum/IOcenter/0_294f0/20081028%20P33%20Articl
e.Agility%20and%20Resilience%20in%20Offshore%20Operations.pdf
Wikipedia. (2008, June). Ear. Retrieved March 12, 2010, from Wikipedia:
http://en.wikipedia.org/wiki/Ear
Wikipedia. (2010, January 21). Genetic algorithm. Retrieved January 25, 2010, from Wikipedia:
http://en.wikipedia.org/wiki/Genetic_algorithms
Wikipedia. (2010, April 21). Radio-frequency identification. Retrieved April 2010, 2010, from
Wikipedia: http://en.wikipedia.org/wiki/Rfid
Wikipedia. (2010, April 13). Supply chain management. Retrieved April 14, 2010, from Wikipedia:
http://en.wikipedia.org/wiki/Supply_Chain_Management
70
71
Appendix
Appendix 1: Failure database
Item
Type/Use
Failure mode
Failure rate
[per 10^6 h]
MDT
Comment
[manhours]
General
Abnormal
instrument
reading
14,25
6,3
Critical
3.2.1
Parameter
deviation
20,63
92,3
Critical
3.2.1
Parameter
deviation
53,34
48,4
Degraded
3.2.1
Minor in-service
problem
84,28
12,7
Incipient
3.2.1
41,89
44
Critical
1.1.1
44,02
47,4
Critical
1.1.1
External Leakage
- Utility
48,64
24,1
Degraded
1.1.1
Low Output
10,75
14,1
Degraded
1.1.1
Abnormal
instrument
reading
143,25
7,8
Incipient
1.1.1
Internal Leakage
476,62
Incipient
1.1.1
Abnormal
instrument
reading
40,39
31,2
Critical
3.2.2
Structural
deficiency
36,15
19,5
Critical
3.2.2
External Leakage
- Process
18,47
5,7
Degraded
3.2.2
External Leakage
- Process
40,70
13,7
Incipient
3.2.2
Abnormal
instrument
reading
106,37
5,9
Incipient
3.2.2
Source
Coalescer
Compressor
Centrifugal
Fail to start on
demand
Spurious stop
Contactor
General
Heat
Exchanger
Plate
Shell and
Tube
External Leakage
- Process
10,65
30,7
Critical
3.1.3
Internal leakage
4,26
17
Degraded
3.1.3
Abnormal
instrument
reading
8,74
3,5
Incipient
3.1.3
Structural
deficiency
6,17
81,2
Critical
3.1.5
External Leakage
- Utility
4,41
3,3
Degraded
3.1.5
Parameter
deviation
2,88
12
Degraded
3.1.5
Abnormal
instrument
reading
22,99
5,8
Incipient
3.1.5
External Leakage
- Process
8,19
2,4
Incipient
3.1.5
Minor in-service
problem
17,31
18,1
incipient
3.1.5
Spurious stop
22,56
45
Critical
1.3.1
External Leakage
- Process
7,04
42
Critical
1.3.1
Low output
4,62
45,3
Critical
1.3.1
External Leakage
- Utility
57,14
36,3
Degraded
1.3.1
5,75
33,9
Degraded
1.3.1
8,58
78,1
Degraded
1.3.1
Abnormal
instrument
reading
274,18
8,1
Incipient
1.3.1
Minor in-service
problem
391,65
10,4
Incipient
1.3.1
Pump
Centrifugal
Structural
deficiency
Vibration
Scrubber
II
General
Structural
deficiency
5,18
19
Critical
3.2.7
External Leakage
- Process
8,88
7,5
Degraded
3.2.7
Abnormal
instrument
reading
52,12
11,1
Incipient
3.2.7
Minor in-service
problem
12,50
14,5
Incipient
3.2.7
Fail to function
on demand
3,59
Critical
4.2.1
Erratic output
2,04
Degraded
4.2.1
Minor in-service
problem
4,56
3,6
Incipient
4.2.1
1,55
Critical
4.2.2
2,43
3,67
5,55
5,3
7
5,3
Degraded
Incipient
Degraded
4.2.2
4.2.2
4.2.3
Fail to function
on demand
3,10
4,3
Critical
4.2.4
Spurious
operation
3,10
2,7
Critical
4.2.4
Abnormal
instrument
reading
14,03
6,9
Critical
3.2.8
External Leakage
- Process
9,55
6,1
Critical
3.2.8
Plugged/Chocked
29,68
4,3
Degraded
3.2.8
Parameter
deviation
8,92
6,3
Degraded
3.2.8
Abnormal
instrument
reading
23,71
6,8
Incipient
3.2.8
Minor in-service
problem
49,42
Incipient
3.2.8
Sensors
Flow
Level
Pressure
Temperature
Spurious
operation
Low output
Erratic output
Other
Separator
General
III
Valves
Ball
Gate
Fail to close on
demand
5,85
9,9
Critical
4.3.1
Fail to open on
demand
3,46
9,9
Critical
4.3.1
Delayed
operation
2,13
10,9
Degraded
4.3.1
External Leakage
- Utility
2,94
14,7
Degraded
4.3.1
Abnormal
instrument
reading
4,91
10,1
Incipient
4.3.1
Minor in-service
problem
5,57
3,6
Incipient
4.3.1
Fail to close on
demand
9,62
9,7
Critical
4.3.5
Fail to open on
demand
66,05
12
Critical
4.3.5
Valve leakage in
closed position
2,84
9,4
Degraded
4.3.5
4,55
5,4
Degraded
4.3.5
262,36
3,4
Incipient
4.3.5
19,29
2,6
Incipient
4.3.5
Other
Abnormal
instrument
reading
External Leakage
- Utility
IV
Appendix 2: FMECA Report
VI
VII
VIII
IX
Appendix 3: Risk matrices for sub functions

Risk matrices: Compression
XI
Risk matrices: Control
XII
Risk matrices: Cooling
XIII
Risk matrices: Pumping
XIV
Risk matrices: Separation
XV
Appendix 4: CD with OBM model.
XVI
Appendix 5: OBM model
XVII
Appendix 6: OBM answer reports

Answer report from scenario 1:
XVIII
XIX
Answer report form scenario 3:
XX
XXI
XXII
XXIII

Condition Monitoring Example

Uploaded by

Condition Monitoring Example

Uploaded by

Title:

Condition Monitoring and Opportunity

Jon Inge Kristoffersen

Faculty of Engineering Science and Technology

Condition Monitoring and Opportunity Based Maintenance in Offshore

3. OBM and logistics planning:

Trondheim 18th January 2010.

Condition Monitoring and Opportunity Based Maintenance

CM information and Opportunity Based Maintenance (OBM) ....................................................... 3

Prioritising equipment for OBM .............................................................................................. 4

Organising OBM with GA ......................................................................................................... 7

Oil and gas production and CM methods ...................................................................................... 11

System layout ........................................................................................................................ 11

Heat exchangers ............................................................................................................ 12

Valves and other control equipment............................................................................. 14

Functional hierarchy ...................................................................................................... 16

Failure rates and types .................................................................................................. 18

Results and comments .................................................................................................. 19

Acoustic emissions and Ultrasonics monitoring ............................................................ 24

Colour monitoring ......................................................................................................... 25

Corrosion monitoring .................................................................................................... 25

Level, leakage & flow monitoring .................................................................................. 26

Load monitoring ............................................................................................................ 27

Magnetic Particle Inspection ......................................................................................... 27

Noise & acoustics monitoring........................................................................................ 28

Oil analysis ..................................................................................................................... 28

Position monitoring ....................................................................................................... 29

Smell & taste monitoring............................................................................................... 31

Temperature monitoring ............................................................................................... 31

Vibration monitoring ..................................................................................................... 32

Wear debris analysis...................................................................................................... 34

Applicable CM methods ................................................................................................ 36

OBM simulations using GA .................................................................................................... 38

Costs included in the model .......................................................................................... 38

Developing the model ................................................................................................... 40

Results and comments .................................................................................................. 41

OBM and logistics planning ........................................................................................................... 57

Logistics planning and maintenance today ........................................................................... 57

OBM influence on logistics planning ..................................................................................... 61

Future work ................................................................................................................................... 67

Works Cited ................................................................................................................................... 69

Figure 5 - System layout ........................................................................................................................ 11

Table 5: CM methods summary _____________________________________________________

2 CM information and Opportunity Based Maintenance (OBM)

2.1 OBM Considerations

2.2 Prioritising equipment for OBM

Figure 1 - Example of RUL predictions

Figure 2 - OBM flowchart

2.3 Organising OBM with GA

3 Oil and gas production and CM methods

3.1 System layout

Figure 5 - System layout

Figure 6 Separator (Devold, 2009)

Figure 7 - Plate heat exchanger (Devold, 2009)

Figure 8 - Shell and tube heat exchanger (Devold, 2009)

Figure 9 - TEG scrubber (Devold, 2009)

Figure 10 - Ball valve

Figure 11 - RCM process

3.2.1 Functional hierarchy

Figure 12 - Functional hierarchy

Table 2: Frequency classes

Failure rate grouping

In x per year equivalent

100 0.87 or more faults pr year

Small spill <10 m3

3.2.3 Failure rates and types

3.2.4 Results and comments

Abnormal instrument reading

Figure 13 - Functional hierarchy from Manifer

Figure 14 - Risk matrix: Costs

Figure 15 - Risk matrix: Environment

Figure 16 - Risk matrix: Safety

Figure 17 - ALARP principle

Acoustic emissions and Ultrasonics

Figure 18 - Visual spectrum in nm

same operating condition.