White Paper RAM

Published on October 2017 | Categories: Documents | Downloads: 20 | Comments: 0 | Views: 388
of x
Download PDF   Embed   Report

Comments

Content

WHITEPAPER

MAROS LITE
Energy from air

AUTHOR: Victor Borges, RAM Product Manager, DNV GL - Software
DATE: 31 May 2015

SAFER, SMARTER, GREENER

SAFEGUARDING
LIFE,
PROPERTY
AND THE
ENVIRONMENT

Date: 31 May 2015

Prepared by DNV GL - Software

© Copyright DNV GL AS 2014. All rights reserved. No use of the material is allowed without the prior
written consent of DNV GL AS.

TABLE OF CONTENTS
TABLE OF CONTENTS ....................................................................................................................I
1

INTRODUCTION .............................................................................................................. 1

2

IMPORTANCE OF WIND TURBINES .................................................................................... 2

3

MAINTENANCE SUPPORT ................................................................................................. 2

3.1

Crew

2

3.2

Workboats

2

3.3

Helicopter support

2

3.4

Spare parts

3

3.5

Safety equipment

3

4

LIFECYCLE COST ANALYSIS (LCC) .................................................................................... 3

4.1

Annual Discount Rate

3

4.2

Capital Expenditure

4

4.3

Operating Expenditure

4

4.4

Product Price

4

5

RAM ANALYSIS ............................................................................................................... 4

6

CASE STUDY .................................................................................................................. 4

6.1

Introduction

4

6.2

Reliability Block Diagrams and Reliability data

6

6.3

Maintenance resources and priority

6.4

Financial aspects

7

RESULTS ..................................................................................................................... 11

8

SENSITIVITY ANALYSIS ................................................................................................. 15

8.1

Planned renewal

15

8.2

Conditional monitoring

19

9

CONCLUSION ............................................................................................................... 21

10

ABOUT THE AUTHOR ..................................................................................................... 21

11

REFERENCES ................................................................................................................ 22

| WHITEPAPER | MAROS LITE | www.dnvgl.com/software

7
10

Page i

1

INTRODUCTION

A fantastic source of energy surrounds the Earth: our atmosphere. Our atmosphere is formed of air
which is a mixture of different gases, liquid and solid particles. Heat energy from the sun warms up the
atmosphere and asymmetrically the Earth.
Warm air is lighter than cold air; cold air is more dense than warm air so it sinks down through warm air.
On the other hand, warm air rises through the atmosphere. When the air rises through the atmosphere,
it creates a low pressure area, when it sinks through the atmosphere, it creates a high pressure areas.
In order to balance this different pressure, air particles move from areas of high pressure (cold air) to
areas of low pressure (warm air). This movement of air is known as the wind.
The wind is influenced by number of factors such as the earth’s movement and its irregular surface. For
instance, where warm land and cool sea meet, the difference in temperature creates thermal effects,
which causes local sea breezes.
A wind turbine is a machine that transforms the kinetic energy of the wind into mechanical and then
electrical energy. Wind turbines consist of a foundation, a tower, a nacelle and a rotor. The foundation
prevents the turbine from falling over. The tower holds up the rotor and a nacelle (or box).
The nacelle contains large primary components such as the main axle, gearbox, generator, transformer
and control system. The rotor is made of the blades and the hub, which holds them in position as they
turn. Most commercial wind turbines have three rotor blades. The length of the blades can be more than
60 metres.

Figure 1: How a wind turbine comes together (IRENA, 2012)

| WHITEPAPER | MAROS LITE | www.dnvgl.com/software

Page 1

2

IMPORTANCE OF WIND TURBINES

In March 2007, European Union leaders set the 2020 targets, committing to address the continuously
increasing energy production from hydrocarbon sources. The main goal is to become a highly energyefficient and low carbon economy.
In this programme, they mention the 2020, 20-20-20 targets. The programme describes an integrated
approach to climate and energy policy that aims to combat climate change. One of the “20s” refers to
increasing the share of EU energy consumption produced from renewable resources to 20%.
To achieve this target, wind turbines must play an essential role. Good news, one might say - a solution
to the global issue. The bad news however, is how do we ensure continuous performance from a big
“fan” sitting in remote locations and out at sea? We can do this in the same way that we already support
big metal structures out at sea, albeit with some additional challenges!

3

MAINTENANCE SUPPORT

Akin to an oil and gas production platform, the operation is 24/7. The wind turbines are unmanned,
imposing challenges to maintenance campaigns. Space is also limited and only small spare parts can be
stored in the turbine.
Maintenance strategy is one of many topics that have to be explored in detail for construction and
operation projects. The maintenance strategy is supported by a number of resources, some of which will
now be discussed.

3.1 Crew
The maintenance crew is formed of technicians, and sometimes specialised personnel are required to
perform specific maintenance tasks. One of the biggest challenges is getting maintenance crew to and
then on and off the offshore turbines and substations to carry out work. There are two major factors that
influence the approach taken to gaining access:


Travel time – the time needed to shuttle a service crew from the crew base to the place of work.
A number of constraints must be taken into account such as limited shift hours available, the
mobilisation time taken to prepare the crew as well as the travel time to transport crews to
different locations in the wind farm. The goal should be to optimise the utilisation of the crew.



Accessibility – after getting on the wind turbine, there is a window of time where the turbine can
be safely accessed. This will obviously depend on the transportation means and the sea and
weather conditions. For example, significant wave height will prevent a vessel to transfer crew
and equipment.

3.2 Workboats
Workboats are an essential part of any offshore wind maintenance strategy. These boats are essential
for logistical services by transporting technicians and equipment from the shore to the wind farm. The
services are extended depending on how far offshore the sites are. For example, distant offshore sites
may also use workboats to ferry technicians between the offshore base and turbines.

3.3 Helicopter support
Helicopters transport technicians to and from the wind farm. They are particularly important when
environmental conditions, such as sea state conditions, make the wind turbine inaccessible. The

| WHITEPAPER | MAROS LITE | www.dnvgl.com/software

Page 2

helicopter size will depend on the extension of the wind turbines and its distance from the supporting
location as well as the service being delivered. Helicopters are mainly used for crew transportation but
can also be used to transport spare parts.

3.4 Spare parts
Spare parts are typically required to repair shutdowns caused by unscheduled failures or even on an
opportunistic basis. Opportunistic maintenance is commonly practised on a variety of systems, the
objective being to offset unscheduled repairs by planning an inspection, overhaul or renewal of specific
equipment at regular intervals, with the aim of improving long term productivity (reliability) of the
system. As with opportune maintenance there is an optimum strategy to yield maximum cost efficiency.

3.5 Safety equipment
Inspections of safety/performance critical equipment items are essential to maintain safety requirements
and performance standards.
Obviously, inspections must be carried out by qualified personnel and be supported by a number of
safety maintenance resources. Inspection frequency is typically every six-months or annually, depending
on the equipment being addressed.

4

LIFECYCLE COST ANALYSIS (LCC)

Different analyses are used to explore a number of variables that will impact directly not only the uptime
of the system but also operational expenditure and cash flows. Lifecycle cost analysis (LCC) - which is
typically used to evaluate the financial performance of different projects - seem to be even more
important to wind farms when compared to the oil and gas industry.
Wind turbines are a one thousand year old technology, however they have seen a rapid evolution in
design in recent times as their application to utility scale power production has brough investment. An
important aspect of the rapidly changing design and usage philosophies around the technologies is to
consider the potential financial performance. This is fundamental to ensure return on investment and
understand the risk and uncertainties in a venture.
A tool commonly used to the compare the financial aspects of different projects is the Net Present Value
(NPV). Net present value takes account of cash flows from the project and allows us to compare future
projections to their present values by applying a discount factor. After taking this factor into account,
projects become directly comparable. Should the value of the capital inflows exceed those of the
outflows after the selected discount rate has been applied then the project will provide a positive cash
flow, and the greater the value the better. However, if the NPV is negative the returns from the project
are less than the outflows and attempts should be made to minimise the NPV.
To be able to produce an NPV figure, the following information is needed.

4.1 Annual Discount Rate
An annual discount rate in percentage form must be ascertained. This will relate the worth, in financial
terms, of a future sum to its present value. For example, assuming a discount rate of 10%, this implies
that £100 returned 1 year from now is worth £90.90 (the calculation is £100/[1+10%]) in today's terms.

| WHITEPAPER | MAROS LITE | www.dnvgl.com/software

Page 3

4.2 Capital Expenditure
Information regarding the capital expenditure of the project can be used. This describes the initial capital
expense outlayed at the beginning of the project and any other expenses incurred during the lifetime of
the project.

4.3 Operating Expenditure
Operating expenditure information is required for manpower, spares and extras. This consists of the cost
in day rates of the resources used, as well as any mobilisation/de-mobilisation costs incurred from their
use. Extra costs can be added if required. The output represents the cost of operating and maintaining
an installation over its lifetime.

4.4 Product Price
Other information required prior to producing the NPV figure is the product price. This is the stipulated
initial price per unit, in the given currency (e.g. £0.16/kWh), and any changes which can be expected to
occur to this price through the life of the system.

5

RAM ANALYSIS

Reliability, Availability and Maintainability (RAM) analysis is a methodology used to predict asset
performance based on reliability and maintainability. As with many other branches of modern
engineering, system performance analysis is probabilistic as opposed to deterministic in nature.
This methodology is well established and used in many domains such as the oil and gas and transport
industries. For wind power applications the RAM methodology has a remarkably similar approach. When
applying RAM analysis to renewable energy there are a number of slight modifications that must be
accounted for such as the wind speed profile and the probability of the wind blowing in different
directions. Applying RAM analysis to renewable energy is an excellent opportunity which leverages
decades of human investment into RAM simulation methodologies and applies it to a progressive,
sustainable industry.

6

CASE STUDY

6.1 Introduction
For this case study, a 3.5 MW wind turbine is considered to be operational from 2015 to 20401. Wind
turbines contain more than 8,000 components, many of which are made from different materials
including steel, cast iron, and concrete. From these 8,000 components, 20 items have been selected as
production critical. These items are listed below.

1

http://www.ewea.org/wind-energy-basics/faq/
How long does a wind turbine work for?
Wind turbines can carry on generating electricity for 20-25 years. Over their lifetime they will be running continuously for as much as 120,000
hours. This compares with the design lifetime of a car engine, which is 4,000 to 6,000 hours.

| WHITEPAPER | MAROS LITE | www.dnvgl.com/software

Page 4

Table 1: List of equipment data

System

Equipment item

Bearing

Bearing A
Bearing B
Sensors
Controller
Converter
Electric control
Shaft
Cooling system
Gearbox
High Speed Shaft coupling
High speed stage bearing
Intermediate speed stage bearing

Control system
Control system
Electrical system
Gearbox

Generator

Hydraulics
Main break
Pitch control
Spare break

Generator
Bearing
Rotor
Hydraulic system
Main break
Pitch control
Spare break

These components will be used as the basis for the Reliability Block Diagram (RBD).
One important factor to take into account when predicting the performance of wind turbines is the
capacity factor. Capacity factor is defined as the actual output over a period of time, compared to its
potential output if it were possible for it to operate at full nameplate capacity continuously over the same
period of time.

2

The daily production rate can be calculated as:
π‘ƒπ‘Ÿπ‘œπ‘‘π‘’π‘π‘‘π‘–π‘œπ‘› π‘Ÿπ‘Žπ‘‘π‘’ (π‘˜π‘Š. β„Ž) = πΆπ‘Žπ‘π‘Žπ‘π‘–π‘‘π‘¦ π‘“π‘Žπ‘π‘‘π‘œπ‘Ÿ × π‘π‘Žπ‘šπ‘’π‘π‘™π‘Žπ‘‘π‘’ π‘π‘Žπ‘π‘Žπ‘π‘–π‘‘π‘¦ × 24

β„Žπ‘œπ‘’π‘Ÿπ‘ 
π‘‘π‘Žπ‘¦π‘ 

This capacity factor accounts for periods of low or no wind, transmission line capacity and electricity
demand. Hence, the performance of a wind turbine is not only dependent on its availability but also on
the weather conditions. The average capacity factor for offshore wind turbines is 41%.
The capacity factor used used in our case study is 35% giving a daily production of:
π‘˜π‘Š. β„Ž
β„Žπ‘œπ‘’π‘Ÿπ‘ 
π·π‘Žπ‘–π‘™π‘¦ π‘π‘Ÿπ‘œπ‘‘π‘’π‘π‘‘π‘–π‘œπ‘› π‘Ÿπ‘Žπ‘‘π‘’ (
) = 0.35 × 3500π‘˜π‘Š × 24
= 29400 π‘˜π‘Š. β„Ž/π‘‘π‘Žπ‘¦
π‘‘π‘Žπ‘¦
π‘‘π‘Žπ‘¦π‘ 

2

http://www.nrc.gov/reading-rm/basic-ref/glossary/capacity-factor-net.html

| WHITEPAPER | MAROS LITE | www.dnvgl.com/software

Page 5

Operation and maintenance costs are a large share of the lifecycle cost of offshore wind turbines. The
typical investment3 in operation and maintenance is €13/MWh in 2015. Reducing the cost of the energy
produced by offshore wind projects is a major focus for the offshore wind industry and governments4.
Therefore, understanding the impact of these operational expenditures and how different maintenance
strategies impact the financial feasibility of the wind turbine is vital. Hence, the focus of this case study’s
analysis is to model the complex maintenance strategy. This maintenance strategy comprises of multiple
locations and multiple crews and it is described in more details below.

6.2 Reliability Block Diagrams and Reliability data
A Reliability Block Diagram (RBD) is a logical representation of the system connection, taking into
account the path of success of the system mission, in this case power production. If you have items in
series, when one of them is in a failed state there is no way for the system to move forward. However, if
you have items in parallel, it means that there is more than one “success” path in the system.
The RBD for the wind turbine system is shown in Figure 2:

Figure 2: RBD for the wind turbine system

Each one of the blocks in a reliability block diagram represents one “event” that can lead to production
loss. In the specific case of this model, each one of the blocks represents an equipment item. Below the
equipment level, the user must define failure modes – failure modes are different ways in which the
equipment can fail.
Generic reliability data was added to the model as shown below:

3
4

http://www.wind-energy-the-facts.org/development-of-the-cost-of-offshore-wind-power-up-to-2015.html

The Crown Estate: Cost Reduction Pathways .:. DECC: Offshore Wind Cost Reduction Task Force Report

| WHITEPAPER | MAROS LITE | www.dnvgl.com/software

Page 6

Table 2: Reliability data for the wind turbine system
System

Equipment
item
Bearing A

Failure distribution

Parameter 1

Parameter 2

Repair distribution

Constant repair time

Weibull (no delay)

9.3

2.5

Constant Repair Time

148

Bearing B

Weibull (no delay)

9.3

2.5

Constant Repair Time

148

Sensors

Exponential

6.97

-

Constant Repair Time

49.4

Controller

Weibull (no delay)

12.48

2

Constant Repair Time

104

Converter

Weibull (no delay)

12.48

2

Constant Repair Time

108

Electric
control
Shaft

Weibull (no delay)

12.5

2

Constant Repair Time

106.6

Exponential

56.18

-

Constant Repair Time

291.4

Cooling
system
Gearbox

Weibull (no delay)

13.6

1.1

Constant Repair Time

14

Weibull (no delay)

23.415

1.7

Constant Repair Time

336

High Speed
Shaft
coupling
High speed
stage bearing
Intermediate
speed stage
bearing
Generator

Weibull (no delay)

56.18

2.5

Constant Repair Time

18

Weibull (no delay)

13.5

2.5

Constant Repair Time

312

Weibull (no delay)

15.5

2.5

Constant Repair Time

312

Exponential

15.94

-

Constant Repair Time

210.7

Bearing

Weibull (no delay)

15.3

2.5

Constant Repair Time

148

Rotor

Weibull (no delay)

17.32

3

Constant Repair Time

120

Exponential

25

-

Constant Repair Time

43.2

Main break

Hydraulic
system
Main break

Exponential

14.55

-

Constant Repair Time

125.4

Pitch control

Pitch control

Exponential

19.23

-

Constant Repair Time

95

Spare break

Spare break

Exponential

20

-

Constant Repair Time

78

Bearing

Control
system
Control
system
Electrical
system
Gearbox

Generator

Hydraulics

The reliability data listed above comes from different sources available in the industry. Data points have
been selected to produce more conservative results.

6.3 Maintenance resources and priority
Repair tasks will require a different set of maintenance resources. These maintenance resources can be
grouped to cover a specific range of equipment failures e.g. all pump failures will require one crew and
one spare pump. Obviously, some resources will be shared amongst different equipment items such as
crews and some resources are dedicated to specific set of failures – spare pumps for pump failures.
Some of these maintenance resources will have constraints such as number in stock and time to restock
for spare parts. Every time a spare is not available for a job, the time to restock will be of 4 days.

| WHITEPAPER | MAROS LITE | www.dnvgl.com/software

Page 7

Figure 3: maintenance resources constraints
Further constraints can also be added to accommodate the challenges related to the maintenance
strategy of wind turbines. For example, the accessibility issue can be defined – so further to the repair
task it takes around 1-2 hours to access a specific area of the wind turbine.
In addition to information regarding potential constraints, cost data can be assigned to each
maintenance resource. This data consists of the cost in day rates of the resources used, or unit cost for
each spare part as well as any mobilisation/de-mobilisation costs incurred from their use. By summing
up all the cost related to maintenance resources, the operating expenditure can be estimated. This is
described in more detail in the next section.

Figure 4: operational expenditure for the maintenance resources
In order to organise all this information, we have to define Maintenance profiles. Maintenance profiles
are used to group maintenance resources for a specific set of failure events and set repair priority.
For this case study, these profiles will also be used to classify the maintenance tasks required for the
group of equipment items. A list of the different profiles is given below:
ο‚·

Class I Requirements: No Spare part + Access Vessel + 2 Crew member

ο‚·

Class II Requirements: Spare part + Access Vessel (boat) + 2 Crew members.

ο‚·

Class III Requirements: Spare part + Access Vessel (boat) + Helicopter + 6 Crew members.

ο‚·

Class IV Requirements: Spare part + Access Vessel (heavy boat) + Helicopter + 6 Crew
members.

| WHITEPAPER | MAROS LITE | www.dnvgl.com/software

Page 8

Figure 5: Maintenance profiles defined for failures to the Main bearing
A summary of all the maintenance resources is defined in the following table:
Table 3: Crew constraints

Crew
Technician

Shift
08:00 - 16:00

Travel time
2-4 hours

Daily rate
£100/hour

Table 4: Workboat constraints

Workboat
Large
workboat

Number available

Mobilisation

Mobilisation cost

Daily rate

1
1

5
5

£1,500

£40,000

£2,000

£60,000

Table 5: Helicopter constraints

Helicopter

Number
available
1

Mobilisation

Mobilisation cost

Daily rate

3

£2,000

£15,000

Table 6: Spare parts:

Spare Name

Spare Price
(£)

Mobilisation
cost (£)

Lead time (day)

Classes for type of
maintenance

Rotor
Gearbox
Generator
Brake
Main Bearings
Bearing
Shaft

£1,200,000
£700,000
£200,000
£70,000
£120,000
£50,000
£100,000

£24,000
£14,000
£4,000
£1,400
£10,000
£1,000
£3,000

15
10
4
2
1
2
8

Class B
Class D
Class B
Class B
Class D
Class C
Class D

| WHITEPAPER | MAROS LITE | www.dnvgl.com/software

Page 9

6.4 Financial aspects
Capital expenditures (CapEx) is normally a known, fixed value. The average capital expenditure for
offshore wind turbines is: £4,500,000.005.
The Operational Expenditure which is normally based on failures and availability of maintenance
resources is variable and hard to estimate up front.
By using dynamic simulation techniques, changes to the asset can be taken into account. This approach
also allows the analyst to account for variations on the value of money over the predicted years –
discount rates, interest, etc.
The financial calculation can be extended to incorporate product pricing which enables estimation of the
revenue produced.
The product price defined for this study is 0.16£/kWh.
For this case study, the following maintenance cost data is assigned to a number of elements.
Table 7: Maintenance resource cost

Maintenance resources

Mobilisation
cost

Daily
rate

Crew

Technician

£1,000

£2,400

Vessel

Workboat

£1,500

£40,000

Vessel

Large Workboat

£2,000

£60,000

Accessory

Helicopter

£2,000

£15,000

Table 8: Spare part cost

Spare Name

Spare Price
(£)

Mobilisation
cost (£)

Rotor

£1,200,000

£24,000

Gearbox

£700,000

£14,000

Generator

£200,000

£4,000

Brake

£70,000

£1,400

Main Bearings

£120,000

£10,000

Bearing

£50,000

£1,000

Shaft

£100,000

£3,000

There are two options when calculating the NPV– negative NPV or the Standard NPV. The negative NPV
accounts for the potential loss of revenue whereas the standard NPV details the profit.
NPV (Khan, 1993) should have the cash flow discounted back to its present value or the current
estimated product pricing (PP). The cash inflow and cash outflow are summed so the NPV is the
summation of the terms:
5

http://www.scottish-enterprise.com/~/media/SE/Resources/Documents/MNO/Offshore-wind-guide-June-2013.pdf . The capital expenditure of
a typical 5MW offshore wind turbine is £6M. Therefore, a 3.5MW offshore wind turbine is around £4.5M

| WHITEPAPER | MAROS LITE | www.dnvgl.com/software

Page 10

𝑁

𝑁

𝑃𝑃
𝑃𝑃
π‘†π‘‘π‘Žπ‘›π‘‘π‘Žπ‘Ÿπ‘‘ 𝑁𝑃𝑉 = (∑ (
) (π‘¦π‘’π‘Žπ‘Ÿπ‘™π‘¦ π‘π‘Ÿπ‘œπ‘‘π‘’π‘π‘‘π‘–π‘œπ‘›)) − (πΆπ‘Žπ‘π‘–π‘‘π‘Žπ‘™ 𝑒π‘₯π‘π‘’π‘›π‘‘π‘–π‘‘π‘’π‘Ÿπ‘’) − (∑ (
) (π‘‚π‘π‘’π‘Ÿπ‘Žπ‘‘π‘–π‘œπ‘›π‘Žπ‘™ 𝑒π‘₯π‘π‘’π‘›π‘‘π‘–π‘‘π‘’π‘Ÿπ‘’))
(1 + 𝑖)𝑑
(1 + 𝑖)𝑑
𝑑=1

𝑑=1
𝑁

π‘π‘’π‘”π‘Žπ‘‘π‘–π‘£π‘’ 𝑁𝑃𝑉 = (−πΆπ‘Žπ‘π‘–π‘‘π‘Žπ‘™ 𝑒π‘₯π‘π‘’π‘›π‘‘π‘–π‘‘π‘’π‘Ÿπ‘’) − (∑ (
𝑑=1

𝑁

𝑃𝑃
𝑃𝑃
) (π‘‚π‘π‘’π‘Ÿπ‘Žπ‘‘π‘–π‘œπ‘›π‘Žπ‘™ 𝑒π‘₯π‘π‘’π‘›π‘‘π‘–π‘‘π‘’π‘Ÿπ‘’)) − (∑ (
) (π‘¦π‘’π‘Žπ‘Ÿπ‘™π‘¦ π‘π‘Ÿπ‘œπ‘‘π‘’π‘π‘‘π‘–π‘œπ‘› π‘™π‘œπ‘ π‘ ))
(1 + 𝑖)𝑑
(1 + 𝑖)𝑑
𝑑=1

Where:
t = reference year
i = the discount rate
PP = product price

7

RESULTS

The virtual model for the wind turbine simulates 12500 cycles. This means the model is sampling events
for 500 different lives of 25 years. With this information, the analyst can create a graph that shows how
the Monte Carlo method averages to a stable value after running a number of cycles.

Figure 6: Rolling average a production efficiency from many simulations
The calculated production availability for the system is 96.098% with a standard deviation of 0.331%.
This production availability is averaged from all of the simulated lifecycles. A graph showing the
distribution of the different production availability throughout the different lifecycles can be generated:

| WHITEPAPER | MAROS LITE | www.dnvgl.com/software

Page 11

Figure 7: normal distribution
From this graph, the analyst can assess the probability of different levels of production. Traditionally,
engineers are interested on the P10 and P90 probabilities for a system. In this case study, we can see
that 10% of the estimated production availability lifecycle will not exceed 95.675% and 90% of the
estimated production availability lifecycle will not exceed 96.514%.
The criticality graph shows the production loss associated to each event defined at the virtual model.
This is then ranked to show what events are the biggest contributors to the production loss:

Figure 8: criticality at the gas customer node
For this model, the biggest contributors for the production loss are listed below:
-

The “Planned Maintenance” is responsible for 47.027% of the losses

-

The “Gearbox system” is responsible for 21.605% of the losses

-

The “Generator system” is responsible for 15.532% of the losses

The major contributors to losses can be further investigated. For instance, within the “Planned
Maintenance” share of the pie chart aforementioned, we have different contributors.

| WHITEPAPER | MAROS LITE | www.dnvgl.com/software

Page 12

Figure 9: contributors to planned maintenance losses
The planned maintenance is expected and the ability to quantify how much production is lost due to
planned shutdowns is really important.
However, production loss can also be tracked for the unscheduled outages such as the example below of
the second biggest contributor, the Gearbox
.

The simulation process keeps track of production levels throughout the simulation process and we can
generate a graph showing the losses and the overall production:

| WHITEPAPER | MAROS LITE | www.dnvgl.com/software

Page 13

Figure 10: Production level
One interesting trend on this graph is the increasing blue bar, representing the increasing losses, which
should be expected since we have failure distributions that describe an increasing failures rate with time.
A view of the financial aspects of the venture can also be evaluated looking at the cash flow, operational
expenditure and cumulative revenue.
The graph below shows the lost production recovery opportunity. Thus -£12.1M shows how much money
we can recover if we can optimise the system to avoid production loss.

Figure 11: Negative Net Present Value

Figure 12: Categorised operating expenditure

| WHITEPAPER | MAROS LITE | www.dnvgl.com/software

Page 14

8

SENSITIVITY ANALYSIS

For this case study, a few sensitivity cases are going to be explored. Sensitivity cases are changes to the
base case used to investigate potential optimisations in system design, maintenance configuration and
operational strategy.

8.1 Planned renewal
Preventive maintenance is commonly practised on a variety of systems, the objective being to offset
unscheduled repairs by planning an inspection, overhaul or renewal of specific equipment at regular
intervals, with the aim of improving long term productivity of the system.
So, in order to optimise system availability, we will implement planned activities that will renew
equipment items with ageing patterns and, therefore, increasing failure rate. The focus of this sensitivity
case will be to the gearbox system which is the second biggest loss contributor.

Figure 13: overall system criticality
Within this subsystem, we can see that the High and Intermediate speed stage bearing are the most
critical items. So we will target these elements when implementing this maintenance strategy.

Figure 14: Criticality contribution for items below the gearbox system

| WHITEPAPER | MAROS LITE | www.dnvgl.com/software

Page 15

The reliability data assigned to each element is shown below:

System
Gearbox

Equipment
item
High speed
stage
bearing
Intermediate
speed stage
bearing

Failure
distribution
Weibull (no
delay)

Characteristic
life
13.5

Shape
factor
2.5

Repair
distribution
Constant
Repair Time

Constant repair
time
312

Weibull (no
delay)

15.5

2.5

Constant
Repair Time

312

The first case will include a Planned Renewal approach only for the High speed stage bearing. Different
time intervals will be tested to assess what is the optimum replacement period of the bearings. The time
frame will be 9, 11 and 13 years.
The production availability of each case is detailed below:

Production efficiency (%)

Standard deviation (%)

96.098
96.181
96.143
96.139

0.331
0.328
0.316
0.328

Base case - No Bearing replacement
Bearing replacement – 9 year
Bearing replacement – 11 year
Bearing replacement – 13 year

This could be graphed to make the comparison easy to understand.

96.2
96.18
96.16
96.14
96.12
96.1
96.08
96.06
96.04

Base case
9 years
11 years
13 years
Average efficiency
Figure 15: Average efficiency for the different cases

Production efficiency is marginally improving but a more detailed analysis involving the cost should be
implemented. Every time the bearing is replaced, a new set of spares is used as well as the crew and
different accessories such as the helicopter.
Therefore, we can see a small step in revenue losses when assessing the negative NPV graph.

| WHITEPAPER | MAROS LITE | www.dnvgl.com/software

Page 16

Step in
revenue loss

Figure 16: Step in revenue loss for the negative NPV
Taking the negative NPV for each scenario, the following table and graph can be produced:

Negative NPV (£ Million)
Base case - No Bearing replacement

-12.1

Bearing replacement – 9 year

-12.2

Bearing replacement – 11 year

-12.2

Bearing replacement – 13 year

-12.1

Implementing this maintenance strategy increases the cost by £0.1M. Since the increase in production
availability is marginal, the decision is not to implement this strategy.
The second case to be investigated is replacing both the High speed stage bearing and Intermediate
speed stage bearing.
The production availability of each case is detailed below:

Base case - No Bearing replacement
Bearing replacement – 9 year
Bearing replacement – 11 year
Bearing replacement – 13 year

Production efficiency (%)

Standard deviation (%)

96.098
96.327
96.253
96.213

0.331
0.336
0.324
0.338

This could be graphed to make the comparison easy to understand.

| WHITEPAPER | MAROS LITE | www.dnvgl.com/software

Page 17

96.4
96.3

Base case

96.2

9 years

96.1

11 years

96

13 years

95.9
Average efficiency
Figure 17: Average efficiency for the different cases
When implementing this to both bearing systems the production availability shows an increase of 0.2%
which is an attractive option considering we are looking at a 25 years life span.
The next step is to assess the financial aspects of this new maintenance strategy. The maintenance cost
estimate can be derived by the model by looking into how much resources have been used on average
throughout the 12500 lifecycles.
A graph showing all of the maintenance resource expenditure can be created, as shown below:

25000000
20000000
15000000
Base case
10000000

9 years
11 years

5000000

13 years
0

Figure 18: Maintenance expenditure for the different cases
There are a few interesting points on this graph:
-

The utilisation of some maintenance resources do not change from one case to another. This
makes sense since the new maintenance strategy impacts only a set of variables. For example,
the workboat and the rotor are not impacted by the new strategy.

-

The large workboat expenditure shows a decrease compared to the base case – this means that
this new strategy effectively addresses failures prior its occurrence which optimises the
resources utilisation. The same applies to the helicopter resource.

| WHITEPAPER | MAROS LITE | www.dnvgl.com/software

Page 18

-

The total expenditure is marginally increased by the new strategy when comparing the base case
to the 9 years’ time interval.

-

The expenditure related to the main bearing spare parts increases. However, the replacement is
performed in a controlled environment with all resources available which has a minimum
duration when compared to an unplanned, failure event with complex mobilisation time and
activities.

The negative NPV increases marginally, as shown in the graph below. Therefore there is an increase in
the operating spending which maintain a higher level of availability. Now it is important to understand
what the priority is when operating the wind turbine:
-

Better production availability or highest revenue?

It might sound like a simple answer where the suggested maintenance strategy should be dropped and
the equipment items should run to failure. However, not being able to deliver the production defined in
contractual requirements can actually incur in much bigger fines than the investment in the maintenance
strategy. For this reason, the decision is to prioritise production availability and the decision is to keep
this maintenance strategy for the main bearings in the gearbox.
The renewal period will be 9 years as it shows increased production availability with tolerable revenue
loss.

Figure 19: Negative NPV for the 9 years periodic planned renewal

8.2 Conditional monitoring
If a fault is being condition monitored then it is expected that some prior warning will be given of an
impending failure, however, not in all cases. With advance knowledge of an imminent failure, the
ensuing repair task can be dealt with more effectively. Offsetting the downtime that would otherwise
accrue while mobilizing the required maintenance resources. This can minimize losses in system
performance. Also, it may be possible to plan the repair to take place while there is minimum disruption
to the system, e.g. during periods of low demand.
To model this situation, then it is necessary to generate an alert of impending failure, along with a
known failure time or in some instances generating the failure without any incipient warning.
For selected failures that are considered to be in a continuous condition monitoring environment, then
the following parameters are provided to describe the condition monitoring option:
ο‚·

Probability of successful detection;

| WHITEPAPER | MAROS LITE | www.dnvgl.com/software

Page 19

ο‚·

Incipient failure period.

If an incipient failure is detected then the capacity loss given at failure will prevail for the incipient period
(should be < 100% impact; possibly zero). If the onset of failure is not detected then when the actual
failure occurs its impact will be the capacity loss at repair (most likely 100%).
This approach will not be suitable for situations where the condition monitoring occurs at prescribed
intervals i.e. it must be a continuous process.
After implementing the planned renewal strategy, the new criticality graph shows the Generator as the
second biggest contributor to losses:

Figure 20: Overview of the criticality graph for the 9 years periodic planned renewal
Within the generator, the bearing is the main contributor to losses.

Figure 21: Criticality graph for Generator system
A condition monitoring maintenance strategy is going to be implemented to the bearing in order to
address potential failures. As listed above, the parameters defined for the condition monitoring
maintenance strategy are:
ο‚·

Probability of successful detection = 95%

ο‚·

Incipient failure period = 48 hours

The production availability increases marginally with this new strategy going from

96.327% to 96.391%.

This makes this implementation not worthwhile.

| WHITEPAPER | MAROS LITE | www.dnvgl.com/software

Page 20

9

CONCLUSION

RAM analysis plays a key role when analysing optimal maintenance and operational strategies in the
wind turbines. Informed decisions can be drawn from the model and uncertainty of the production
behaviour can be predicted and therefore avoided or reduced.
By running the base case, the analyst builds an intuitive understanding of how the system may behave
throughout its life.
In addition, a list of possible maintenance strategy options can be created from a base case. Therefore,
the following sensitivities are suggested:
-

Implementing a planned renewal strategy for High speed stage bearing. This change is not
financially effective as the increased average efficiency does not cover the increased
maintenance cost for performing planned renewal of the critical part.

-

Implementing a planned renewal strategy for High speed stage bearing and Intermediate speed
stage bearing. This new strategy is effective as it shows a good return on investment given the
increase in availability.

-

Implementing a condition monitoring strategy for the Bearing in the Generator. This new
strategy shows to not be worthwhile as the increased efficiency is marginal.

This model can be easily extended to incorporate the uncertainty related to the wind profiles, power
curves. New technologies such as Energy Storage Systems (ESSs) can also be incorporated as a buffer
to the stochastic nature of wind. This new area will play an essential role in wind farm applications by
ensuring a higher availability of energy from wind power plants, enabling an increased penetration of
wind power in the energy mix.

10 ABOUT THE AUTHOR
Victor Borges, RAM Software Product Manager at DNV GL, is a chemical engineer with experience
performing risk and reliability analysis for assets in the oil and gas industry. He is responsible for DNV
GL’s world-leading simulation software packages Maros and Taro.

| WHITEPAPER | MAROS LITE | www.dnvgl.com/software

Page 21

11 REFERENCES
Dewan, A. (2014). Logistic & Service Optimization for O&M of Offshore Wind Farms. Retrieved 05 06,
2015, from Delft University of Technology:
http://www.lr.tudelft.nl/fileadmin/Faculteit/LR/Organisatie/Afdelingen_en_Leerstoelen/Afdeling_
AEWE/Wind_Energy/Education/Masters_Projects/Finished_Master_projects/doc/Ashish_Dewan_r
_UPDATE.pdf
DNV GL, S. u. (2013). Customer Stories Maros and Taro. Retrieved 05 06, 2015, from
https://www.dnvgl.com/cases/shell-global-solutions-4051
DNV GL, S. u. (2015). Maros User-Guide. London: DNV GL, Software.
DNV GL, Software unit. (2013, March 1). Maros and Taro - prime tools for predicting performance.
Retrieved April 15, 2015, from DNV GL Software: https://www.dnvgl.com/cases/shell-globalsolutions-4051
Dowell, J., Zitrou, A., Walls, L., Bedford, T., & Infield, D. (2013). Analysis of Wind and Wave Data to
Assess Maintenance Access to Offshore Wind Farms. Retrieved 2015, from
https://www.strath.ac.uk:
https://www.strath.ac.uk/media/departments/eee/iee/windenergydtc/publications/Dowell2013b.
pdf
Estate, T. C. (2013). A Guide to UK Offshore Wind Operations and Maintenance. Retrieved 05 06, 2015,
from http://www.scottish-enterprise.com/~/media/SE/Resources/Documents/MNO/Offshorewind-guide-June-2013.pdf
EWEA, E. W. (2009). The Economics of Wind Energy. EWEA.
EWEA, E. W. (2012). Factsheets. Retrieved 05 06, 2015, from
http://www.ewea.org/fileadmin/files/library/publications/statistics/Factsheets.pdf
IRENA, I. R. (2012). Renewable Energy Technologies: Cost Analysis Series.
Khan, M. (1993). Theory & Problems in Financial Management. Boston: McGraw Hill Higher Education.
Lange, B., Larsen, S. E., Højstrup, J., & Barthelmie, R. (n.d.). The wind speed profile at offshore wind
farm sites. Retrieved from Research gate:
http://www.researchgate.net/profile/Soren_Larsen3/publication/228416435_The_wind_speed_pr
ofile_at_offshore_wind_farm_sites/links/0c9605251a01a54f5f000000.pdf
Nuclear Regulatory Commission, U. (n.d.). Capacity factor (net). Retrieved 05 06, 2015, from
http://www.nrc.gov/reading-rm/basic-ref/glossary/capacity-factor-net.html
The facts, W. e. (n.d.). Development of the cost of Offshore wind power up to 2015. Retrieved 05 06,
2015, from http://www.wind-energy-the-facts.org/development-of-the-cost-of-offshore-windpower-up-to-2015.html

| WHITEPAPER | MAROS LITE | www.dnvgl.com/software

Page 22

ABOUT DNV GL
Driven by our purpose of safeguarding life, property and the environment, DNV GL enables organizations
to advance the safety and sustainability of their business. We provide classification and technical
assurance along with software and independent expert advisory services to the maritime, oil and gas,
and energy industries. We also provide certification services to customers across a wide range of
industries. Operating in more than 100 countries, our 16,000 professionals are dedicated to helping our
customers make the world safer, smarter and greener.

SOFTWARE
DNV GL is the world-leading provider of software for a safer, smarter and greener future in the energy,
process and maritime industries. Our solutions support a variety of business critical activities including
design and engineering, risk assessment, asset integrity and optimization, QHSE, and ship management.
Our worldwide presence facilitates a strong customer focus and efficient sharing of industry best practice
and standards.

| WHITEPAPER | MAROS LITE | www.dnvgl.com/software

Page 23

Sponsor Documents

Or use your account on DocShare.tips

Hide

Forgot your password?

Or register your new account on DocShare.tips

Hide

Lost your password? Please enter your email address. You will receive a link to create a new password.

Back to log-in

Close