Failure Rate, Repair Time and Unscheduled O&M Cost
Analysis of Offshore Wind Turbines
James Carroll
1
, Alasdair McDonald
1
, and David McMillan
2
1
Centre for Doctoral Training in Wind Energy Systems, University of Strathclyde, Glasgow, UK
2
Electronic and Electrical Engineering Department, University of Strathclyde, Glasgow, UK
ABSTRACT
Determining and understanding offshore wind turbine failure rates and resource requirement for repair is
vital for modelling and reducing O&M costs and in turn reducing the cost of energy. While few offshore
failure rates have been published in the past even less details on resource requirement for repair exist in the
public domain. Based on ~350 offshore wind turbines throughout Europe this paper provides failure rates for
the overall wind turbine and its sub-assemblies. It also provides failure rates by year of operation, cost
category and failure modes for the components/sub-assemblies that are the highest contributor to the overall
failure rate. Repair times, average repair costs and average number of technicians required for repair are also
detailed in this paper. An onshore to offshore failure rate comparison is carried out for generators and
converters based on this analysis and an analysis carried out in a past publication. The results of this paper
will contribute to offshore wind O&M cost and resource modelling and aid in better decision making for
O&M planners and managers.
KEYWORDS
Failure mode, failure rate, offshore wind turbine, reliability.
Correspondence
James Carroll: j.carroll@strath.ac.uk
1. INTRODUCTION
he reliability of an offshore wind turbine and the resources required to maintain it can make up ~30% of
the overall cost of energy [1]. Typically, a higher failure rate and greater repair resource requirement (i.e.
material cost and labour) leads to a higher cost of energy. Consequently, wind farm developers try to select
wind turbines with low failure rates and those that require the least amount of maintenance resources. Due to
accessibility issues, reliability of turbines becomes even more important as offshore wind energy generation
increases [2,3]. This paper shows the results of an analysis determining the failure rates and resource
requirements for repair of modern multi MW scale offshore wind turbines and their sub-assemblies.
This analysis is based on ~350 offshore wind turbines from a leading manufacturer. All offshore turbines
in this analysis are between 3 and 10 years old and are from between 5 - 10 wind farms throughout Europe.
The full data set consists of over 1768 turbine years of operational data. For confidentiality reasons the exact
number of wind farms/turbines cannot be provided. For the same reasons the exact nominal power, blade size
or drive train configuration of the turbine type used in this analysis is also not provided. However it can be
stated that it is a modern multi MW scale turbine type with an identical blade size and nominal power in all
turbines. It can also be stated that it is a geared turbine with an induction machine. As a guide to the size of
the turbine type, the rotor diameter is between 80m and 120m and the nominal power is between 2 and 4MW.
The novelty of this work lies in the large modern population of offshore wind turbines analysed. The
analysis of the resources required for repair of offshore wind turbines is also novel as little or no past
publications were found with real data in this area during the literature review. Offshore wind farm operation
and maintenance (O&M) cost models need resource requirements for repair as inputs to the models. These
models can be highly sensitive to the accuracy of this data and that data is not currently in the public domain
[4,5]. In some cases onshore input data is used to estimate offshore outputs in these models [2,6]. Inputs such
as failure rates, repair times, number of technicians required for repair and average cost of repair are required.
This paper is unique in providing each of these inputs based on analysis of this large and modern population
of offshore wind turbines. Out of the four input areas mentioned above, failure rates is the area with the most
T
literature available, however this paper is still novel in this area because the majority of the past literature
available is for the failure rates of populations of older and smaller onshore turbines [7,8] rather than offshore
failure rates based on modern multi MW turbines.
2. Offshore O&M Literature Review
As mentioned in the introduction little or no past literature exists in the area of resource requirement for
repair of on or offshore wind turbines. As this paper also includes a failure rate / reliability section, past
literature on the reliability of offshore wind turbines was reviewed. As the offshore wind industry is young
and turbine manufacturers are generally reluctant to release performance data there is a lack of offshore
reliability analyses available in the public domain.
Reference [9] describes an availability analysis on a number of UK offshore wind farms. Each of the wind
farms in reference [9] are in the early years of operation, all of which are operational for less than three years.
The paper highlights the need for improvements to be made in availability if the economic targets of these
wind farms are to be met. However it does not look at wind turbine failure rate or sub assembly failure rate
as this paper does, making it difficult to determine which areas to focus on to achieve the required availability
improvements.
One other offshore analysis is detailed in paper [10]. This analysis is based on a single wind farm of 36
turbines. The analysis is based on turbine stoppages rather than turbine failures and the paper states that this
type of analysis cannot be compared to a failure rate analysis because the stops are defined differently than
failures. One of the drivers for this difference is that scheduled operations are included in the turbine stoppage
analysis but not in the turbine failure analysis.
There are more onshore reliability analyses in the public domain than there are offshore. These analyses
cover the onshore turbine as a whole as well as its subassemblies. However as stated in [11] these analyses
are repeatedly based on the same wind turbine populations and failure databases due to the small number of
reliability databases in the public domain [12]. Databases like LWK and WMEP in Germany, WindStats in
Germany and Denmark, Reliawind and a population from Sweden [13,14] are the basis for the analysis in
the papers described in the following paragraphs.
References [7,8] analyze a population that reaches 6,000 onshore wind turbines at the end of an 11 year
period. This population of 6,000 turbines is located in Germany and Denmark and failures have been recorded
in the Windstats and LWK database. The Windstats and LWK database is based on the largest population
encountered in the literature review; however, it contains turbines as old as 20 years and as small as 200kW.
As the population contains these older smaller turbines, questions are raised as to whether the population is
representative of modern multi MW turbines.
The WMEP database is used in references [12,15]. The WMEP database contains failure data for up to
1,500 turbines over a 15 year period throughout Germany. A similar onshore failure rate analysis is carried
out in [13] on a population consisting of turbines from Sweden. This Swedish database runs from 1997 and
builds up to ~ 750 turbines. The work carried out by Reliawind [16] is based on 10 minute SCADA data,
work orders, alarm logs and service records from 350 turbines. This is a smaller population than the other
onshore databases discussed above but it consists of more modern larger onshore turbines.
3. POPULATION ANALYSIS
The population analysed in this paper builds up to ~350 turbines over a five year period. These turbines
come from between 5-10 wind farms. The years of installation for the population are shown in Figure 1. It
can be seen that 68% of the population analysed is between three and five years old and 32% is greater than
5 years old. In total this population provides 1768 turbine years or ~15.5 million hours of turbine operation.
Exact population details cannot be provided for confidentiality reasons.
Figure 1. Population Operational Years
4. FAILURE DATA AND DEFINITIONS
4.1 Failure Definition
There is no standardized way for defining a failure in the wind energy industry. This analysis defines a
failure as a visit to a turbine, outside of a scheduled operation, in which material is consumed; this is
consistent with reference [11]. Material is defined as anything that is used or replaced in the turbine; this
includes everything from consumable materials (such as carbon brushes) to replacement parts such as full
IGBT units and full generators.
Faults that are resolved through remote, automatic or manual restarts are not covered by this definition of
a failure. However, if the faults that are resolved through remote, automatic or manual restarts repeatedly
occur and they require a visit to the turbine in which material is used, the failure is then subsequently captured
in this type of failure definition, providing the visit is outside of a scheduled service. This definition is
somewhat different to that in reference [16], in which a failure is defined as a stoppage of a turbine for one
or more hours that requires at least a manual restart to return it to operation.
4.2 Failure rates and failure rate categories
This paper provides failure rates in a per turbine per year format as seen in [7, 8, 11]. The formula used to
determine failure rate per turbine per year can be seen below. It is the same formula used in [7, 8, 11]:
3-5 Years > 5 Years
Years of Operation
68% 32%
0%
10%
20%
30%
40%
50%
60%
70%
80%
% of Population
