Recommended Key Performance Indicators For Operational Management of Wind Turbines

16th Deep Sea Offshore Wind R&D conference IOP Publishing
IOP Conf. Series: Journal of Physics: Conf. Series 1356 (2019) 012040 doi:10.1088/1742-6596/1356/1/012040
Recommended key performance indicators for

operational management of wind turbines
S Pfaffel1 ID
, S Faulstich1 ID
and S Sheng2 ID
1
Fraunhofer Institute for Energy Economics and Energy System Technology - IEE,
Königstor 59, 34119 Kassel, Germany
2
National Renewable Energy Laboratory (NREL),
15013 Denver West Parkway, Golden, CO 80401, USA
E-mail: sebastian.pfaffel@iee.fraunhofer.de
Abstract. Operational managers of wind turbines usually monitor a big fleet of turbines and
thus need highly condensed information to identify underperforming turbines and to prioritize
their work. Key performance indicators (KPIs) are a solid and frequently used tool for this
purpose. However, the KPIs used in the wind industry are not unified to date, which makes
comparison in the industry difficult. Further, comprehensive standards on a set of KPIs for the
wind industry are missing. This article identifies and recommends KPIs and provides detailed
definitions to make KPIs comparable and to enable benchmarking. The starting point of this
work is an industry survey with 28 participants intended to identify commonly used KPIs, collect
various possible definitions, and prioritize them. Out of a total of 50 KPIs, we discuss in a next
step 33 selected KPIs on performance, maintenance, and reliability in detail and recommend
definitions, most of which are based on international standards. As a result, operators can
easily use these recommendations to base their system of KPIs. By using this unified set of
KPIs, operators can be well-prepared to conduct industrywide comparisons and benchmarks.
The survey and this article will also serve as a basis for committee work of the FGW e.V. to
develop a corresponding technical guideline.
1. Introduction
After the successful construction of a wind turbine, its operating phase begins. Although some
wind turbines (WTs) in Germany are replaced by more modern and, above all, larger types before
the standard design lifetime of 20 years [1] is reached [2], the utilization of the remaining useful
life and continued operation after 20 years is important for many WTs [3, 4, 2]. Furthermore,
a first manufacturer has announced new turbine types having a design life of up to 30 years [5].
This and the fact that operation and maintenance (O&M) accounts for a share between 20%
and almost 40% of the levelized cost of energy (LCOE) [6, 7, 8, 9, 10, 11] demonstrates the high
importance of sound operational management.
WTs are usually monitored 24/7 from a central operator’s control room to react to unexpected
technical problems and quickly identify and trigger appropriate measures (e.g., maintenance
operations). However, the long-term performance of WTs and the achievement of targets
and forecasts in various categories (e.g., technical, financial, safety) must also be monitored
for successful operational management. Key performance indicators (KPIs) are metrics to
objectively describe the performance of an observed unit (e.g., company, technical system,
Content from this work may be used under the terms of the Creative Commons Attribution 3.0 licence. Any further distribution
of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.
Published under licence by IOP Publishing Ltd 1
employee) in a well-defined and traceable way [12, 13]. In order to assess individual results,
KPIs must be compared to a target value that could be a forecast, historic result, or comparable
unit. As in other industries, KPIs are widely used in the wind industry and are part of
contracts, management information, and decision support systems. According to [13, 14],
KPIs have to fulfill the SMART (Specific, Measurable, Achievable, Relevant and Time-bound)
criteria to provide valuable information. To be specific and measurable, KPIs have to be well
defined and unified. Even though guidelines and standards (e.g., International Electrotechnical
Commission (IEC) [15, 16] and VGB [17, 18, 19]) on KPIs for the wind industry exist, the
KPIs actually used throughout the wind industry, as well as their definitions, are still very
heterogeneous and sometimes even contradictory. Books by Tavner [20] or Burton et al. [21]
provide valuable information and definitions but are not classified as a guideline or standard.
Publications by Gonzalez et al. [22] or the European Commission [23] evaluate KPIs for the
wind industry. Specifically, Gonzalez et al. [22] provide a comprehensive list of valuable KPIs
for O&M of WTs. Further research papers [24, 25, 26, 27, 28, 29] make use of KPIs and define
single KPIs to their needs, but do not originally intend to provide a set of KPI definitions.
Based on the publications described here as well as standards and guidelines, the objective
of this article is to recommend the most important KPIs along with their specific definitions
for streamlining day-to-day work of the operational management of WTs and industrywide
benchmarking. While recent publications use SMART criteria [22, 13, 14] or even data mining
techniques [30] to select a set of KPIs, we decided instead to reach out to the industry and survey
experts. For this purpose, we conducted a literature review to gather KPIs that were assumed to
be of relevance to O&M of WTs in previous research and international and national standards
and guidelines. Main starting point were the publications by Gonzalez et al. [22] and VGB
[17, 18, 19], complemented by the expert knowledge of the authors from running the WInD-Pool
[31] initiative as well as the gearbox reliability database [32]. KPIs and their definitions were
collected, summarized, and categorized to form the basis for a survey among experts involved
in O&M of WTs.
The survey (see Section 2) aims to determine the relevance of the individual KPIs and their
various definitions, obtain missing KPIs, and prioritize future work. Thus the prioritization is
based on the actual use of KPIs in the wind industry and not on a purely theoretical evaluation
and literature review as done in previous publications. Next (see Section 3), the most relevant
KPIs, their purpose, and application—as well as their definitions—are discussed in detail. While
the prioritization of KPIs is based on the survey results, the single recommended KPI definitions
are mainly based on a detailed review of literature and standards as well as the experience of
the authors. This results in a comprehensive list of unified KPIs including detailed definitions
to achieve a common understanding of the most important KPIs and enable industrywide
comparisons (see Section 4). The detailed KPI evaluation of this article focuses on performance,
maintenance, and reliability metrics. For Health, Safety and the Environment (HSE)- and
Finance-KPIs, we evaluated only the results of the survey. The results are expected to serve as
a basis for the development of a corresponding guideline by the Fördergesellschaft Windenergie
und andere Dezentrale Energien (FGW e.V.) (see Section 5).
2. Survey on KPIs
To focus on KPIs relevant for the day-to-day work performed as part of wind turbine O&M and
to prioritize standardization efforts, we conducted a survey to determine the most important
KPIs used in the wind industry. The survey was part of the work to add a section regarding the
analyses of WT data to the technical guideline part 7 [33] of the FGW e.V., which deals with
operation and maintenance of WT. Prepared by the Fraunhofer Institute for Energy Economics
and Energy System Technology (Fraunhofer IEE) and coordinated with members of the working
group, the request to participate in the survey was circulated by the working group as well as
2
the FGW e.V. to members of a selected target group. The surveyed companies are located in
German-speaking areas but are active internationally in many cases.
2.1. Survey design

The survey is based on a list of 34 potentially relevant KPIs gathered during a literature
search and complemented by the knowledge of experts within the FGW e.V. working group.
In accordance with the different categories of KPIs, the survey is divided into the following five
sections:
• Performance KPIs
• Maintenance KPIs
• Reliability KPIs
• HSE KPIs
• Finance KPIs
Following this categorization, the web-based survey also consists of five sections, which are
responded to according to the previously indicated order. Depending on the survey participant’s
expertise and interest, he/she has the option of skipping individual questions, entire sections,
or exiting the survey early. Answering questions about a KPI is therefore also considered as an
expression of its relevance. Background information about the survey participant is requested
but not mandatory. Participants can remain anonymous or provide contact details for follow-up
questions. For each KPI, the following questions are asked:
• Is the KPI used in your company?
• Which definition is used?
• Which data serves as a basis?
• How important is the KPI?
The survey comprises a total of 144 questions. Therefore, a decrease in participant motivation
to answer later questions has to be considered in the survey design and analyses of the answers.
As a result, KPI categories with the greatest expected differences to other industries are queried
at the beginning. We expect that finance and HSE KPIs are not wind-specific.
2.2. Survey results

Invitations to participate in the survey were sent out by the FGW e.V. on 4th October 2017,
with the last survey completed on 1st November 2017. A total of 30 responses were submitted
within 1 month. The answers of 28 participants were considered to be useful, whereas two
participants were excluded because of implausible answers. The 28 participants spent about 17
minutes on the survey, on average (median). As expected, none of them answered all of the
questions. Figure 1 shows that the primary target group of owners/operators and operational
managers accounts for about 50% of all answers. A differentiation of owners/operators and
operational managers may not be necessary in many countries, but was made because of the
specific situation of the German market. The roles of manufacturers, researchers, grid operators,
and consultants were not provided as response options in the original survey design and added
by the participants as free text.
The number of participants is within the expected range and does not allow for a detailed
statistical evaluation or representative results in a rigorous statistical sampling sense. From
our experience, this is still a very decent number of responses for an expert survey in the wind
industry, especially if leading manufacturers and operators are participating. Due to the amount
of time participants spend on the survey, we assume a high motivation and knowledge of the
3
Operational Manager
Manufacturer
Owner/Operator
Researcher
Assessor/Expert
Figure 1. Share of different people
Grid Operator and their roles in the wind industry
Consultant who participated in the FGW e.V.
survey on relevant key performance
0 2 4 6 8 10 12 indicators for the operation and
Participants
maintenance of wind turbines.
participating experts. Thus, conclusions on the importance of a single KPI and prioritization
can be drawn. For this purpose, the absolute importance (frequency of use x importance) is
determined. Furthermore, the participants pointed out missing KPIs and provided information
on possible KPI definitions. Results on the different KPI categories are discussed in the following
subsections. Summarizing tables (Tables 1–5) are ordered by the absolute importance. New
KPIs are highlighted in grey and KPIs without any answers/importance are highlighted in red.
2.2.1. Performance Performance KPIs show the highest importance of all KPI categories. All
28 participants answered whether they make use of performance KPIs, with 20 indicating that
they do; see Table 1. Except for ”Remote-Resets,” all proposed KPIs are among the top 10, with
the highest absolute importance of the whole survey. Furthermore, participants suggested that
10 additional KPIs be considered in a comprehensive KPI systematic. Four of the suggested
KPIs are intended to provide more concrete descriptions for the term, ”wind conditions.” The
proposed market value factor is closely linked to financial results.
Table 1. Performance KPIs inquired about in the survey included more precise KPI definitions
regarding wind conditions and additional KPIs in general.
KPI Answers Use Importance (1–5) Abs. Importance

Power curve 20/20 19/20 4.5 85.5
Wind conditions 20/20 16/20 4.5 72
Average wind speed
Wind speed distribution
Wind direction distribution
Average wind speed/site assessment
Full-load hours 20/20 18/20 3.5 63
Energy consumption 20/20 16/20 3.1 49.6
Capacity factor 20/20 13/20 3.7 48.1
Data availability 20/20 11/20 4.1 45.1
Remote-resets 20/20 5/20 3.2 16
Site quality
No. of telecommunication interruptions
Forecast fulfillment
Operating hours
Specific yield
Market value factor
4
2.2.2. Maintenance Maintenance KPIs are used by a little more than half (16/28) of the
survey participants; see Table 2. While availability metrics are of high importance and part of
the top 10 KPIs, further and more detailed KPIs show a significantly lower importance. The
survey suggests that respondents consider six additional KPIs, as three of them are expected to
increase the level of detail on planned maintenance tasks.
Table 2. Maintenance KPIs inquired about in the survey included more precise KPI definitions
regarding planned maintenance tasks and additional KPIs in general.
KPI Answers Use Importance (1–5) Abs. Importance

Time-based availability 16/16 16/16 4.7 75.2
Production-based availability 16/16 12/16 4.1 49.2
Production ratio
Yield losses by cause
Monetary-based availability
Maintenance tasks 16/16 7/16 4 28
Preventive maintenance tasks 16/16 7/16 3.3 23.1
Number of routine maintenance tasks
Number of inspections/visual inspections
Number of repairs
Reactive maintenance tasks 16/16 7/16 3.3 23.1
Risk priority number (RPN) 16/16 1/16 5 5
2.2.3. Reliability Although only basic reliability KPIs were proposed in the survey, only 10
out of 28 respondents use reliability KPIs at all in their organization; see Table 3. Participants
using reliability KPIs have in most cases the role of an owner/operator or manufacturer. No
additional KPIs were suggested.
Table 3. Reliability KPIs inquired about in the survey.
KPI Answers Use Importance Abs.

(1–5) Importance
Failure rate 10/10 8/10 3.6 28.8
Mean time between failures (MTBF) 10/10 7/10 3.6 25.2
Mean time to repair / restoration (MTTR) 10/10 7/10 3.3 23.1
Mean down time (MDT) 10/10 6/10 3 18
Mean operating time between failures (MOTBF) 10/10 5/10 3.2 16
Mean operating time to failures (MTTF) 10/10 5/10 2.8 14
Repair rate 10/10 3/10 2.3 6.9
2.2.4. Health, safety, and the environment Only five (most manufacturers) out of 27
respondents make use of HSE KPIs in their organization. Thus, the absolute importance of
all KPIs is low, and two KPIs are not used at all; see Table 4. No additional KPIs were
suggested. The low importance of HSE KPIs might be caused by the fact that the German
wind industry consists mainly of small- and medium-sized companies and that at least data on
health and safety related incidents is gathered and analyzed by employer’s liability insurance
associations, which may provide more accurate information.
5
Table 4. HSE KPIs inquired about in the survey.
KPI Answers Use Importance Abs.

(1–5) Importance
Total accident rate 5/5 4/5 3.7 14.8
Total lost time occupational illness frequency 5/5 3/5 2.7 8.1
Fatal accident rate 5/5 2/5 4 8
Recordable injury rate 5/5 2/5 3.5 7
Energy demand per employee and year 5/5 0/5 0 0
Pollutant release per employee and year 5/5 0/5 0 0
2.2.5. Finance Nine out of 22 respondents make use of finance KPIs in their organization.
Overall, finance KPIs are of medium-to-low importance and one KPI is not considered to be
relevant at all. One reason for the low importance of the finance KPIs in the survey can be the
subdivision of the operational management of WTs in Germany. The commercial and technical
management is usually carried out by different departments or even different companies.
Table 5. Finance KPIs asked in the survey including additional KPIs (highlighted in grey).
KPI Answers Use Importance (1-5) Abs. Importance

EBITDA 9/9 5/9 4 20
Maintenance costs 9/9 5/9 3.6 18
Operational expenditures (OPEX) 9/9 4/9 4.4 17.6
Levelized cost of energy (LCOE) 9/9 3/9 3.7 11.1
Debt-service coverage ratio (DSCR) 9/9 2/9 5 10
Free cash flow to equity (FCFE) 9/9 2/9 4.5 9
Break-even price of energy (BEPE) 9/9 1/9 4 4
Loan life coverage ratio (LLCR) 9/9 0/9 0 0
3. Recommended KPIs and definitions

A list of 33 KPIs out of a total of 50 KPIs were identified through the survey and selected based
on the experience of the authors to be recommended KPIs. These KPIs are subject of a more
detailed discussion and effort to provide detailed definitions. See Appendix A, Appendix B and
Appendix C for the discussion and definition of the single KPIs. Following the expertise of
the authors, the detailed evaluation focuses solely on performance, maintenance, and reliability
metrics. We must also mention that some of the KPIs (e.g., on availability) cannot be assigned
exclusively to a single category. The chosen assignment should therefore only be regarded as
an orientation. For each KPI, its purpose, field of application, possible differing definitions,
and required data sources are discussed. A proposed KPI definition is provided, which in case
of doubt, takes the point of view of an owner/operator or the operational management. The
discussion considers literature as well as information and comments provided by industry experts
in the survey. Wherever possible, the definitions are based on accepted standards and guidelines
for power plants or other industries of similar nature. Still, the selected definitions reflect the
experience of the authors and KPIs had to be set by the authors in some cases.
The temporal resolution of KPIs as well the frequency of calculation should be adjusted
to a company’s internal reporting period. As a general recommendation, performance and
6
maintenance KPIs should be calculated on a monthly basis and then rolled up to quarterly and
yearly values. For reliability, HSE and finance KPIs, a yearly evaluation will suffice.
Performance of a WT can be understood as its ability to meet the expectations placed on it.
Thus, performance KPIs (Appendix A) comprise indicators describing operational results like
the capacity factor as well as indicators of the underlying conditions (e.g., wind conditions) under
which the results were achieved. Furthermore, KPIs designed to identify undesired behavior of
the WT can be considered to be performance KPIs as well.
Maintenance KPIs (Appendix B) are mainly intended to monitor the success of maintenance
strategies. The selected maintenance strategy has to fit WT specific requirements, as KPIs like
the production-based or monetary-based availability show. Although the reliability of the WT
technology is always a key factor, historical comparisons or comparisons with other WTs in
particular make it possible to evaluate the results.
Reliability KPIs (Appendix C) monitor the ”ability of an item to perform a required
function...” [34] by assessing the frequency of failures as well as the duration of the corresponding
faults/downtime and required time of repairs. Thus, performance and reliability KPIs are
strongly influenced by the WTs reliability. Working with reliability KPIs requires a discussion
of some prerequisites.
When assessing the reliability of a WT, different levels of detail can be applied. The term
”item” includes ”systems” as well as ”components” but will be mainly used as a synonym for
”system” in this work. Analyses can be carried out on a WT, system or component level. It is
recommended to use a standardized designation system to obtain comparable results. Although
the Reference Designation System for Power Plants (RDS-PP) [35] is the recommended system
for worldwide application, the North American Electric Reliability Corporation - Generating
Availability Data System (NERC-GADS) [36] should be applied in the United States. The level
of detail also determines whether the evaluated object is repairable or not. Even though a WT
and its systems are always considered to be repairable, subsystems and components have to
be replaced in many cases. Different KPIs have to be used for (partially) repairable and non-
repairable systems as Figure 2 shows. KPIs for systems that can be repaired to a state as good
as new are calculated the same way as for non-repairable system.
Start of Operation Failure Restart Failure
Time
(Partially) Repairable Systems
Mean Operating Time To Final Replacement

MTTFR
Mean Operating Time To First Failure Mean Time Between Failure

MTTFF MTBF
Mean Time To Restoration Mean Operating Time Between Failure

MTTR MOTBF ...
...
Mean Down Time

MDT
Non-Repairable Systems or As Good As New
Mean Operating Time To Failure Mean Time To Restoration Mean Operating Time To Failure
MTTF MTTR MTTF
...
...
Mean Down Time
MDT
Figure 2. Reliability mean time measures for (partially) repairable and non-repairable systems
according to ISO [37, 38] and IEC [39] standards.
7
This article focuses on mean time metrics as basic reliability KPIs to describe and compare
the reliability of WTs. Various definitions are used in the literature. All definitions in this
work are based on the International Electrotechnical Vocabulary (IEC 60050-192) [39] as well
as on the International Organization for Standardization (ISO) standards 12489 [37] and 14224
[38] which aim at the oil and gas industry. Further publications [40, 41, 42] provide valuable
information as well, but also slightly different definitions. The discussed KPIs enable monitoring
and benchmarks but are not sufficient for maintenance optimization, which requires detailed
reliability distributions. Refer to [43] for further details on the requirements for maintenance
optimization.
Time is the commonly used lifetime metric of the discussed reliability KPIs and even part
of the KPI names. Its usage to describe the lifetime of systems and components is only a first
step though. More sophisticated and system specific metrics are needed. Examples could be
the energy yield, revolutions or load cycles. As soon as further lifetime metrics are available,
see Guo and Sheng [44] for gearboxes, all reliability KPIs can be calculated using such metrics
instead of time.
Work reports on repairs and replacements are the required data sources to calculate reliability
KPIs; reports on routine maintenance/service are not necessary. However, they can be
supplemented by operational data to obtain information on the total downtime.
4. Conclusions
This article discusses recommended KPIs for the operational management of WTs and therefore
takes the operator’s perspective. Results of an industry survey are presented to determine the
most relevant KPIs. As expected, the limited number of participants does not allow for detailed
statistical evaluations and the survey can’t be considered to be representative in a rigorous
statistical sampling sense. However, from the author’s point of view, a basic prioritization of
the KPIs is reasonable. As a result, the recommended KPIs in this article are not primarily
based on the authors’ experience, as in other publications, but on real application in the industry.
The recommended definitions for each KPI are based on comments from the survey and on a
review of the corresponding literature and standards. Still, it has to be acknowledged that the
recommended definitions are always biased by the experience and opinion of the authors. Based
on the results (Sections 2 and 3) of this article, the following conclusions can be drawn:
• To date, there are already many KPIs on O&M of WTs available and used by operational
managers.
• The set of KPIs used by operational managers as well as the definitions of single KPIs are
heterogeneous and do not enable cross company comparisons or benchmarking.
• Performance KPIs are most important for operational managers. Maintenance and
reliability KPIs are of importance as well but follow with less relevance.
• HSE and finance KPIs are usually not wind specific and already well defined in the literature.
Still, an overview of the related recommended KPIs and their definitions would be beneficial.
• An international technical guideline or standard providing recommendations on a set of
KPIs for the operational management of WTs would be beneficial. Such a document should
comprise a list of KPIs, their definitions and recommendations for their application.
• The previous paper publication of Gonzalez et al. [22] or the technical guideline published by
the VGB [17] already provide a comprehensive list of valuable KPIs. In contrast, the present
article follows a more holistic approach and also recommends detailed KPI definitions, which
are based on internationally accepted standards wherever possible.
• The survey is based on a list of 34 KPIs, of which 3 KPIs are rejected. 16 further KPIs
are recommended by the participants. Out of a total of 50 KPIs, 33 KPIs are discussed in
detail.
8
5. Outlook
The results of the survey and this article will be the starting point for committee work of the
FGW e.V. to develop a corresponding technical guideline. However, the corresponding work is
not finished with this publication but just getting started. Thus, the list of recommended KPIs
is not complete. Further KPIs will be developed within the scientific community or the industry
and can become part of the technical guideline at a certain point. KPIs on the fulfillment of
power curves or to monitor degradation are just two possible examples.
Further work is also necessary regarding recommended approaches for aggregation and
pooling of KPIs as well as regarding the uncertainties of KPIs in general. Aggregated results
can strongly differ depending on the selected aggregation approaches. Annex F of the Institute
of Electrical and Electronics Engineers (IEEE) Standard 762-2006 [45] is a good example of
the importance of a unified approach to make KPIs comparable and shows how this issue can
be addressed. The importance of an uncertainty assessment has been shown in [46, 47] for
the production-based availability. Further work on uncertainty estimations and especially the
uncertainty of additional KPIs is strongly encouraged.
Acknowledgments
The research work of Fraunhofer IEE for this paper was funded by the German Federal Ministry
for Economic Affairs through the WInD-Pool (grant No. 0324031A) project.
This work was authored [in part] by the National Renewable Energy Laboratory, operated
by Alliance for Sustainable Energy, LLC, for the U.S. Department of Energy (DOE) under
Contract No. DE-AC36-08GO28308. Funding provided by the U.S. Department of Energy
Office of Energy Efficiency and Renewable Energy Wind Energy Technologies Office. The
views expressed in the article do not necessarily represent the views of the DOE or the U.S.
Government. The U.S. Government retains and the publisher, by accepting the article for
publication, acknowledges that the U.S. Government retains a nonexclusive, paid-up, irrevocable,
worldwide license to publish or reproduce the published form of this work, or allow others to do
so, for U.S. Government purposes.
References
[1] IEC 2016 Wind energy generation systems - part 1: Design requirements (IEC 61400-1) - draft
[2] Berkhout V, Bergmann D, Cernusko R, Durstewitz M, Faulstich S, Gerhard N, Gromann J, Hahn B, Hartung
M, Härtel P, Hirsch J, Hofsä M, Kuhl M, Kulla S, Pfaffel S, Rubel K, Seel J, Spriesterbach S, Waila D,
Wiegand C, Wiggert M and Wiser R 2017 Wind Energy Report Germany 2016 (Fraunhofer Verlag) ISBN
978-3-8396-1195-1
[3] Loraux C and Brühwiler E 2016 Journal of Physics: Conference Series 753 072023 ISSN 1742-6588 URL
http://iopscience.iop.org/article/10.1088/1742-6596/753/7/072023/pdf
[4] Ziegler L, Gonzalez E, Rubert T, Smolka U and Melero J J 2018 Renewable and Sustainable Energy Reviews
82 1261–1271 ISSN 13640321
[5] ENERCON GmbH 2018 windblatt: the enercon magazine URL https://www.enercon.de/fileadmin/
Redakteur/Medien-Portal/windblatt/pdf/Windblatt_01_18_GB.pdf
[6] Gilman P, Maurer B, Feinberg L, Duerr A, Peterson L, Musial W, Beiter P, Golladay J, Stromberg J,
Johnson I, Boren D and Moore A 2016 National Offshore Wind Strategy: Facilitating the Development of
the Offshore Wind Industry in the United States (U.S. Department of Energy)
[7] Maples B, Saur G, Hand M, van de Pietermen R and Obdam T 2013 Installation, operation, and maintenance
strategies to reduce the cost of offshore wind energy URL https://www.nrel.gov/docs/fy13osti/57403.
pdf
[8] Wiser R and Bolinger M 2017 2016 Wind Technologies Market Report (U.S. Department of Energy)
[9] Lüers S, Wallasch A K and Dr-Ing Rehfeldt K 2015 Kostensituation der Windenergie an Land in Deutsch-
land: Update URL https://www.wind-energie.de/sites/default/files/download/publication/
kostensituation-der-windenergie-land-deutschland-update/20151214_kostensituation_der_
windenergie_an_land_in_deutschland_update.pdf
[10] Hobohm J, Krampe L, Peter F, Gerken A, Heinrich P and Richter M 2015 Kostensenkungspotenziale der
Offshore-Windenergie in Deutschland: Kurzfassung URL https://www.offshore-stiftung.de/sites/
9
offshorelink.de/files/pictures/SOW_Download-KurzfassungStudie_Kostensenkungspotenziale_
Offshore-Windenergie.pdf
[11] Arwas P, Charlesworth D, Clark D, Clay R, Craft G, Donaldson I, Dunlop A, Fox A, Howard R, Lloyd
C, Lovett C, Marshall A and Wiles F 2012 Offshore wind cost reduction: Pathways study URL https:
//www.thecrownestate.co.uk/media/5493/ei-offshore-wind-cost-reduction-pathways-study.pdf
[12] Fitz-Gibbon C T 1990 Performance Indicators (Multilingual Matters) ISBN 9781853590924
[13] Shahin A and Mahbod M A 2007 International Journal of Productivity and Performance Management 56
226–240 ISSN 1741-0401
[14] Podgrski D 2015 Safety Science 73 146–166 ISSN 09257535
[15] IEC 2017 Power performance measurements of electricity producing wind turbines (IEC 61400-12-1)
[16] IEC 2014 Production based availability for wind turbines (IEC 61400-26-2)
[17] VGB PowerTech eV 2015 Wind turbines: Definitions and indicators: Vgb-s-002-05-2015-10-en URL https:
//www.vgb.org/begriffe_versorgungswirtschaft.html?dfid=77610
[18] VGB PowerTech eV 2015 Basic terms of the electricity utility industry: Vgb-s-002-01-2015-en URL
https://www.vgb.org/begriffe_versorgungswirtschaft.html?dfid=77604
[19] VGB PowerTech eV 2016 Technical and commercial key indicators for power plants: Vgb-s-002-03-2016-08-en
URL https://www.vgb.org/begriffe_versorgungswirtschaft.html?dfid=81831
[20] Tavner P 2012 Offshore wind turbines: Reliability, availability, and maintenance (IET Renewable energy
series vol 13) (London: Institution of Engineering and Technology) ISBN 9781849192309
[21] Burton T, Jenkins N, Sharpe D and Bossanyi E 2011 Wind Energy Handbook 2nd ed (Hoboken: John Wiley
& Sons) ISBN 978-0-470-69975-1
[22] Gonzalez E, Nanos E M, Seyr H, Valldecabres L, Yürüen N Y, Smolka U, Muskulus M and Melero J J 2017
Energy Procedia 137 559–570 ISSN 18766102
[23] 2011 Key performance indicators for the european wind industrial initiative: Setis - tpwind URL https:
//setis.ec.europa.eu/system/files/Key_Performance_Indicators_Wind.pdf
[24] Rinne E, Holttinen H, Kiviluoma J and Rissanen S 2018 Nature Energy 3 494–500 ISSN 2058-7546
[25] Boccard N 2009 Energy Policy 37 2679–2688 ISSN 03014215
[26] Ueckerdt F, Hirth L, Luderer G and Edenhofer O 2013 Energy 63 61–75 ISSN 03605442
[27] Ritter M, Shen Z, Lpez Cabrera B, Odening M and Deckert L 2015 Renewable Energy 83 416–424 ISSN
09601481
[28] Krokoszinski H J 2003 Renewable Energy 28 2165–2178 ISSN 09601481
[29] Faulstich S, Hahn B and Tavner P J 2011 Wind Energy 14 327–337 ISSN 10954244
[30] Peral J, Mat A and Marco M 2017 Computer Standards & Interfaces 54 76–85 ISSN 09205489
[31] Faulstich S, Pfaffel S and Hahn B 2015 Performance and reliability benchmarking using the cross-company
initiative wind-pool URL https://express2.converia.de/frontend/index.php?page_id=1645
[32] Sheng S 2014 Gearbox reliability database: Yesterday, today, and tomorrow URL https://www.nrel.gov/
docs/fy15osti/63106.pdf
[33] FGW 2010 Technical guideline for power plants part 7: Operation and maintenance of power plants for
renewable energy - general part
[34] British Standards Institution 12/2010 Maintenance maintenance terminology (bs en 13306)
[35] VGB PowerTech eV (ed) 2014 VGB-Standard RDS-PP: Application Guideline Part 32: Wind Power Plants:
VGB-S823-32-2014-03-EN-DE (VGB PowerTech e.V.) ISBN 978-3-86875-693-7
[36] 2017 GADS wind turbine generation: Data reporting instructions: Effective january 1, 2018
URL https://www.nerc.com/pa/RAPA/gads/wind_DRI_DL/GADS%20Wind%20DRI-2018%20Reporting%
20Period-master.pdf
[37] ISO 2013 Petroleum, petrochemical and natural gas industries – reliability modelling and calculation of safety
systems (ISO/TR 12489)
[38] ISO 22072016 Petroleum, petrochemical and natural gas industries collection and exchange of reliability and
maintenance data for equipment (ISO 14224:2016) URL http://www.irantpm.ir/wp-content/uploads/
2016/12/ISO-14224-2016.pdf
[39] IEC 2015 International electrotechnical vocabulary IEV 60050-192 URL http://www.electropedia.org
[40] 1998 Military handbook: Electronic reliability design handbook URL http://everyspec.com/MIL-HDBK/
MIL-HDBK-0300-0499/download.php?spec=MIL-HDBK-338B.015041.pdf
[41] IEEE 610-1991 standard computer dictionary. a compilation of IEEE standard computer glossaries URL
http://ieeexplore.ieee.org/servlet/opac?punumber=2267
[42] Quanterion Solutions Incorporated (ed) 2015 System reliability toolkit - V: New approaches and practical
applications (Utica, N.Y.: Quanterion Solutions Incorporated) ISBN 978-1-933904-97-9
[43] Hahn B 2017 17. wind farm data collection and reliability assessment for o&m optimization: Expert group
report on recommended practices: First edition, 2017 URL https://www.ieawind.org/whatsNEW/052917/
10
IEAWindTCPRP17ReliabilityData2017.pdf
[44] Guo Y and Sheng S 21022018 Probability of failure for gearbox high-speed-stage
bearings in wind turbines: Nrel/pr-5000-70972 URL https://www.osti.gov/biblio/
1441175-probability-failure-gearbox-high-speed-stage-bearings-wind-turbines
[45] IEEE 2006 Standard definitions for use in reporting electric generating unit reliability, availability, and
productivity
[46] Wilkinson M, Vanni F and Delouvri T 2016 A comparison of production-based availability methods:
Ewea analysis of operating wind farms 2016 URL http://www.ewea.org/events/workshops/wp-content/
uploads/Tech16a-PO-033.pdf
[47] Craig A, Optis M, Fields M J and Moriarty P 2018 Journal of Physics: Conference Series 1037 052021 ISSN
1742-6588 URL http://iopscience.iop.org/article/10.1088/1742-6596/1037/5/052021/pdf
[48] IEC 2013 Power performance of electricity-producing wind turbines based on nacelle anemometry (IEC
61400-12-2)
[49] FGW 2017 Technical guideline for power plants part 6: Determination of wind potential and energy yields
[50] Heier S 2005 Windkraftanlagen: Systemauslegung, Netzintegration und Regelung 4th ed (Wiesbaden and s.l.:
Vieweg+Teubner Verlag) ISBN 9783322940889 URL http://dx.doi.org/10.1007/978-3-322-94087-2
[51] Bethke J, Kampmeyer J and Mengelkamp H T The benefit of wind atlases in wind energy and their verification
[52] Grange S K Technical note: Averaging wind speeds and directions
[53] Sohoni V, Gupta S C and Nema R K 2016 Journal of Energy 2016 1–18
[54] FGW 2018 Technical guideline for power plants part 2: Determination of power curve and standardized
energy yields
[55] FGW 2017 Technical guideline for power plants part 5: determination and application of the reference yields
[56] FGW 2018 Technical guideline for power plants part 10: Determination of site quality following
commissioning
[57] 2016 Control and protection systems for wind turbines: Dnvgl-st-0438: Standard URL https://rules.
dnvgl.com/docs/pdf/DNVGL/ST/2016-04/DNVGL-ST-0438.pdf
[58] Gungor V C and Lambert F C 2006 Computer Networks 50 877–897 ISSN 13891286
[59] Deutscher Bundestag Gesetz für den Ausbau erneuerbarer Energien (Erneuerbare-Energien-Gesetz - EEG
2017) URL http://www.gesetze-im-internet.de/eeg_2014/EEG_2017.pdf
[60] Hirth L 2013 Energy Economics 38 218–236 ISSN 01409883
[61] Pudlik M, Sensfu F and Winkler J Arbeitspaket 4 Welche Faktoren beeinflussen die Entwicklung des
Marktwerts der Erneuerbaren Energien?: Leitstudie Strommarkt
[62] Fernahl A, Götz P, Dr Henkel J and Lenck T Ermittlung des Marktwertes der deutsch-
landweiten Stromerzeugung aus regenerativen Kraftwerken für das Jahr 2015: Studie für die
vier deutschen übertragungsnetzbetreiber im Auftrag der TenneT TSO GmbH: Endbericht - Kurz-
fassung URL https://www.netztransparenz.de/portals/1/Content/EEG-Umlage/EEG-Umlage%202015/
2014-10-08_Endbericht_Marktwertfaktoren_Energy_Brainpool_Kurzfassung_2015.pdf
[63] IEC 2010-12 Time based availability for wind turbines (IEC 61400-26-1)
[64] DIN 2017 Instandhaltung - Wesentliche Leistungskennzahlen für die Instandhaltung
[65] Verein Deutscher Ingenieure 2011 VDI-Richtlinie 3423: Verfügbarkeit von Maschinen und Anlagen: Begriffe,
Zeiterfassung und Berechnung
[66] Hirsch J and Faulstich S 2016 HERA-VPP - high efficiency, reliability, availability of virtual power plants :
Abschlussbericht
[67] Lutz M A, Görg P, Faulstich S, Cernusko R and Pfaffel S 2019 Wind Energy (under review)
[68] Goble W M 2010 Control systems safety evaluation and reliability 3rd ed ISA resources for measurement and
control series (Research Triangle Park, NC: International Society of Automation) ISBN 9781934394809
[69] IEC 1990 International electrotechnical vocabulary IEV 60050-191 (outdated)
Appendix A. Recommended performance KPIs

Appendix A.1. Wind conditions
Wind conditions at the site of a WT strongly influence the overall performance. To describe the
wind conditions, various KPIs are used. Based on the results of the survey, the most important
ones will be discussed below.
n
1 X
V̄ = · V10min (A.1)
n
i=1
11
where
V̄ = Average wind speed,

V10min = Measured wind speed averaged over 10 min
Average wind speed (Equation A.1) is the most commonly used yet least informative KPI to
describe wind conditions. It can be calculated as an arithmetic mean based on 5-, 10-, or 15-
minute average values of the nacelle-based wind-speed measurement included in the supervisory
control and data acquisition system (SCADA) data. Without further information on the range
of wind speeds and air density, the informative value on the wind power actually available at the
site is low, and comparisons between different years or WTs can be misleading. As an alternative
KPI, wind power density (Equation A.2) takes the two above-mentioned issues into account and
is more recommended. IEC 61400-12-2 [48] and FGW TR 6 [49] provide basic information
regarding the handling of wind speed measurements but KPI definitions are missing. However,
the definitions of both KPIs can be considered to be general knowledge and are included in
standard literature [21, 50].
n
1 X1 3
P̄Wind = · ρ10min V10min (A.2)
n 2
i=1
where
P̄Wind = Average wind power density,

ρ10min = Air density averaged over 10 min,
V10min = Measured wind speed averaged over 10 min
To compare wind conditions of a certain year to conditions expected in the site assessment
and project planning, the measured results have to be corrected for the long-term variation.
During the site assessment, wind speed and/or wind energy indices [51, 27] are used to determine
representative wind conditions, as defined in FGW TR 6 [49]. It is recommended to apply the
same approach and indices to operational data.
When analyzing and visualizing wind conditions in detail, the wind speed distribution is
usually represented by a histogram and in a next step by a Weibull distribution [50, 21]. Even
though, both tools can be beneficial for detailed analysis, they do not provide easy-to-grasp
information and cannot be counted as KPIs. The same applies for histograms of the wind
direction. Average wind directions are not commonly used in the wind industry but can be
applied as a first step to identify WT misalignment or problems in the wind measurement
system. In any case, it must be ensured that the mean wind direction is calculated vector-based
and not as an arithmetic mean [52].
Appendix A.2. Power curve

The power curve of a WT describes the average power output at different wind speeds normalized
to standardized conditions, especially regarding air density and turbulence. Different phases in
the life cycle of a WT use different power curves. During project development, calculated power
curves for new turbine types are a starting point in many cases and usually replaced by measured
reference power curves later on. In the O&M phase, power curves are used to compare the actual
operational performance of the WT to the intended performance described by the power curve
as well as to calculate losses during downtime or derated operation. It is obvious that the power
curve does not fit the classical characteristics of a KPI but rather is an indispensable tool for
wind turbine O&M, which is necessary for the calculation of further KPIs.
12
Although the survey shows agreement on the basic definition of a power curve, the procedure
for determining it as precisely as possible is complex and many uncertainties have to be
considered [53, 47]. Definitions for power curve calculations are provided by the IEC as
an international standard and are the recommended definitions. While the IEC 61400-12-1
[15] defines the procedure to calculate a reference power curve using additional measurement
equipment like a measuring mast, IEC 61400-12-2 [48] defines the procedure to calculate a power
curve based on SCADA data including wind speed measurements by the nacelle anemometer.
Further definitions/procedures to calculate power curves for specific applications exist on a
national level. Examples are the German technical guidelines TR 2, 5, and 10 [54, 55, 56] of
the FGW e.V. and the technical recommendation TR-1 on wind farm power performance of the
American Wind Energy Association (AWEA) in the United States, which is still unpublished.
The application of different methods for data filtration and power curve modeling can lead to
not negligible differences in the results, as Craig et al. [47] showed.
Although the power curve is used to determine various KPIs, the authors are not aware of a
generally used KPI to evaluate the power curve itself. It would be beneficial to use such a KPI
to compare power curves of different WT or monitor possible changes over time.
Appendix A.3. Full-load hours and capacity factor

Even though full-load hours (Equation A.4) and the capacity factor (Equation A.3) are individual
KPIs, they are discussed together because of their very similar statement and definition. The
capacity factor compares the actual produced energy to the amount of energy that could have
been produced by a WT in perfect wind conditions, thereby enabling continuous operation at
rated power without any interruption/downtime. Thus, the capacity factor is a dimensionless
value that is usually presented as a percentage value. Full-load hours show the amount of
time (hours) that a WT would have needed to operate at rated capacity to produce the actual
produced amount of energy during a specific duration, usually a year.
WActual
CF = · 100% (A.3)
WNominal
where
CF = Capacity factor,
WActual = Generated energy,
WNominal = Nominal possible energy generation
The available definitions for both KPIs from the survey, the literature [24, 25], and the
standards and guidelines show no differences. Recommended sources for the capacity factor are
IEC 61400-26-2 Annex C.2 [16] and the IEEE Std 762 3.17 [45] and for the full-load hours the
VGB-S-002-05 [18].
WActual
tFL = (A.4)
PRated
where
tFL = Full-Load Hours,

WActual = Generated Energy,
PRated = Rated Capacity
Although full-load hours and the capacity factor depend heavily on site- and year-specific
wind conditions as well as the installed turbine type, both KPIs are commonly used to describe
13
the performance of WTs and can easily be converted into each other. Typical applications
are the comparison between different turbines and turbine types at the same site, different sites
having the same turbine type, as well as the comparison to expected values of the site assessment
or previous years. Wind indexes can be used to correct long-term variability [51, 49]. There is
no doubt that there are better KPIs for all of these different applications, but both KPIs are
frequently used in the wind industry, especially because of their universality and simplicity.
Appendix A.4. Data availability

The availability of plausible data (AData , Equation A.5) from the SCADA system is a prerequisite
to determine many of the KPIs discussed in this paper. It is therefore necessary to monitor the
availability of the data using a corresponding KPI. This statement is supported by the results of
the survey. To date, no known standards or guidelines include a description of data availability.
However, there is agreement within the survey/industry on its basic definition. Data availability
is the ratio of time having sufficient and plausible information to the total amount of time. Data
gaps are of course included in the calculation. Data availability can and should be calculated
for different cases. As a first step, the KPI should assess the general availability of data for each
time step. Furthermore, the evaluation should drill down to the individual measurements and
at least analyze power and wind speed measurements individually.

tNoData
AData = 1 − · 100% (A.5)
tTotal
where
AData = Data availability,

tNoData = Time of no or bad data,
tTotal = Total time
Appendix A.5. Energy consumption factor

WTs always consume energy to supply ancillary systems. Usually those systems are supplied by
the electricity generated by the WT itself if the electricity output of the WT remains positive.
However, during phases of low wind speeds or downtime, the WT turns from a producer to a
consumer of electricity. The described periods are characterized by a negative power output. To
date, a standardized definition of a KPI is not available. In order to make the average energy
consumption comparable, it can be normalized to the rated power, as with the capacity factor.
Because of the rather small values, the result should be presented as a per-mille value and not
as a percentage.
Pn
(P̄i (P < 0) · ti (P < 0))
ECF = i=1 · 1000 h (A.6)
PRated · ∆t
where
ECF = Energy consumption factor,

P̄i (P < 0) = Average power of the ith period having negative power output,
ti (P < 0) = Duration of the ith negative power output,
PRated = Rated capacity,
∆t = Observation period in hours
14
Appendix A.6. Remote resets

Remote resets occur when the protection system of a WT is activated and the WT is forced to
shut down. According to the definitions of DNV GL [57], a manual reset by staff on-site shall
be normally performed in such cases. Instead, remote resets can be handled from the remote
control center to save money and time as long as further measures like a root cause analysis
are taken and the number of remote resets is limited [57]. Remote resets are defined as having
a maximum downtime of 1 [31] or 2 hours [7] in the literature. Still, remote resets should be
identified by the corresponding signal of the SCADA system.
cR
λR = (A.7)
∆t
PcR
i=1 DTi
M DTR = (A.8)
cR
where
λR = Reset rate,
M DTR = Mean downtime of remote resets,
cR = Count of remote resets,
∆t = Observation period in months or years,
DTi = Downtime of the ith reset
It is important to count the number of remote resets per defined period of time (month, year)
to ensure compliance with recommended limitations as well as to detect deviating WTs. The
average downtime per reset should be evaluated for the same reason. Very short downtimes can
imply an insufficient root cause analysis.
Appendix A.7. Telecommunication interruptions

Reliable and fast network connections are essential in many areas of automated and remotely
monitored asset operation [58]. This applies in particular to WTs. WTs are not forced to
stop operation in the event of a telecommunications interruption, but can continue to run
independently and without monitoring by the control centre. However, the loss of a connection
does not only mean a loss of monitoring but also the possibility of human intervention, which is of
high relevance in the event of network bottlenecks, for example. With this in mind, the frequency
and duration of telecommunication interruptions should be monitored so that measures can be
taken if necessary.
cT I
λT I = (A.9)
∆t
PcT I
i=1 DTi
M DTT I = (A.10)
cT I
where
λT I = Telecommunication interruption rate,

M DTT I = Mean downtime of telecommunication interruptions,
cT I = Count of telecommunication interruptions,
∆t = Observation period in months or years,
DTi = Downtime of the ith telecommunication interruption
15
Appendix A.8. Operating time

The operating time of a WT can be counted as defined by IEC [16] or VGB [19, 18]. Operating
time is defined by the information category ”GENERATING” in the IEC standard and as ”
... the period in which a plant converts energy” by the VGB. Both definitions have the same
meaning. Solely times with a positive power output are counted; standby or idling times are not
considered to be operating time. Operating time can be presented either as an absolute value
(e.g., for a year or as a percentage value normalized to the total time).
n
X
tO = ti (P > 0) (A.11)
i=1
where
tO = Operating time,
ti (P > 0) = Time of positive power output for the ith period
Appendix A.9. Site quality

The site quality is a KPI specific to Germany and required by the Renewable Energies Act [59]
for all WT participating in land-based wind tenders. To enable competition among different
regions, remunerations are adjusted to specific wind conditions. Remunerations are lower at
sites with high wind speeds and higher at low wind speed sites. Turbine type and hub-height-
specific reference yields — as defined in [55] — are calculated based on a generic reference site
(100-m hub height). The wind conditions at the site are described by a Rayleigh distribution
using a reference wind speed (vref = 6.45 m/s) and a reference wind gradient described by
the Hellmann exponent (α = 0.25) to define a 100% scenario. The energy yield predicted
during project planning and the potential energy yield during the operational phase of a WT
are compared to the reference yield to determine the site quality. WT specific remunerations
are adjusted accordingly. The potential energy yield is used instead of the actual energy yield so
that technical or operational losses do not affect the level of remuneration. When calculating the
potential energy yield, unavailability of up to 2% are considered to be normal. Further losses
caused by technical reasons or compensated losses due to grid curtailment are added to the
actual energy yield to calculate the potential energy yield. Detailed calculation specifications
are defined in FGW TR 10 [56].
SEy,W T i
SGy,W T i = · 100 (A.12)
R
where
SGy,W T i = Site quality per WT and year,

SEy,W T i = Potential energy yield according to FGW TR10 per WT and year,
R = Turbine type and hub-height-specific reference yield according to FGW TR 5
The operator is legally obliged to determine this KPI. In addition, this KPI can be used to
compare operational results to expected values from the planning phase.
Appendix A.10. Market value factor

The power generation of WTs depends on the fluctuating wind conditions at the respective sites
and does not necessarily meet the electricity demand which also varies. In many countries, wind
power is sold on the spot market, in which the price is determined by supply and demand. The
16
market value factor (M V F , Equation A.13) compares the average electricity price achieved by
an asset (WT) to the average electricity price over a certain period of time. SCADA data and
electricity prices are needed to calculate the market value factor. In Germany, hourly prices of
the day-ahead auctions of the EPEX Spot SE market are used [60, 61, 62].
Pn
(EP EX M CPh · Wh )
M V F = nh=1EP EX M CP P
P (A.13)
h=1
n
h
· nh=1 Wh
where
M V F = Market value factor,

EP EX M CPh = Hourly day-ahead price at EPEX spot,
Wh = Hourly energy yield of the considered asset
Market value factors are not only important during project planning but can also be of
relevance during O&M to update revenue expectations and to determine the value of individual
assets for direct marketing contracts. Additionally, the KPI might be of interest to prioritize
maintenance measures. Prioritizing the maintenance of WTs showing high market value factors
could lead to higher overall financial revenues. The monetary-based availability (see Section
Appendix B.3), which is designed to take differing electricity prices into account, follows a
similar approach.
Appendix B. Recommended maintenance KPIs

Appendix B.1. Time-based availability
The commonly accepted definition of time-based availability (At , Equation B.1) is the ratio
(percentage value) of available time to total (available and unavailable) time. However, a wide
variety of approaches are used to define available and unavailable times and to exclude times from
consideration. The survey names standards, maintenance contracts, manufacturer definitions,
and internal definitions as well as definitions included in software tools as sources in the wind
industry. Further sources are to be expected. Standardized definitions are provided by the
IEC 61400-25-1 [63], which includes definitions from different views. To achieve international
comparability, the recommended definition for O&M follows the System Operational Availability
(B.2.3) of the IEC standard where all downtime except for low wind is considered as not available.
Guidelines like the VGB-S-002-05 [18] or the FGW TR 10 [56] include differing definitions.
Because of the characteristics of wind energy, broad definitions from DIN EN 1534 [64] or VDI
3423 [65] cannot be applied directly.

tUnavailable
At = 1 − · 100% (B.1)
tAvailable + tUnavailable
where
At = Time-based availability,
tAvailable = Time of full and partial performance and low wind,
tUnavailable = Time of other cases except for data gaps
Appendix B.2. Production-based availability and Production Ratio

Production-based availability (AW , Equation B.2) is the ratio of the actual energy yield of a
WT to its potential energy yield. Considering the energy losses during downtimes instead of the
downtime duration as well as in some cases losses caused by derated operation makes production-
based availability a more holistic KPI compared to the time-based availability. Although this
17
basic definition is common sense and information on the actual energy yield is simple to obtain,
various approaches to calculate the potential energy yield are existing. The main differences lie
in the consideration of different types of downtime, of generation losses during operation, and
in methods for determining the potential power of a wind turbine. Answers of the survey show
internal definitions, software-specific definitions, and the standard definitions published within
the IEC 61400-26-2 [16] to be the most commonly applied. Like for time-based availability, the
“System Operational Availability” (B.2.2) of IEC 61400-26-2 is the recommended definition
for WTs operators/owners. In this definition, all differences between potential and actual
production are assumed to be losses, and data gaps are excluded from the calculation; see
Equation (B.2). VGB-S-002-05 [18] provides an alternative definition, which is less detailed
and includes no calculation guidelines. When calculating the production-based availability,
the determination of the potential power is a special challenge where plausible wind speed
measurements and power curves or reference WTs are required to obtain reasonable results.
A comparison of the different methods of the IEC 61400-26-2 [16] to determine the potential
production was published by Wilkinson et al. [46]. A detailed categorization of the operating
modes also enables an evaluation of yield or availability losses by their causes.

WPotential − WActual
AW = 1 − · 100% (B.2)
WPotential
where
AW = Production-based availability,
WActual = Actual energy yield,
WPotential = Potential energy yield, data gaps are excluded
An additional KPI is the production ratio as defined in IEC 61400-26-2 Annex C.3 [16]. It is
calculated the same way as the production-based availability but only considers periods of normal
operation (full performance). The production ratio is supposed to detect underperformance,
which can be caused by icing, degradation, pitch angle deviations, and so on. Depending on the
specific conditions, production ratios of more than 100% can occur temporarily, which can be a
result of changing turbulences or inaccuracies in wind speed measurement.
Appendix B.3. Monetary-based availability

Monetary-based availability (AM , Equation B.3) takes the availability assessment of WTs to
a new level. First mentioned in this context by Hirsch et al. [66] in 2016, monetary-based
availability is not yet used in the wind industry but has potential to be the upcoming most
important KPI to assess the overall performance of WTs. While the time-based availability
solely considers downtime, the production-based availability focuses on lost energy yield. In
fact, the decisive factor is not lost energy but the loss of revenue. Thus, the monetary-based
availability compares the achieved gross operating profit (GOP) (numerator) to the potential
GOP (denominator). The GOP is calculated using electricity spot market prices to determine
revenues. It is assumed that the potential GOP includes only periods of sufficient electricity
prices above the marginal costs to operate the WT and the WT would be ideally shut down to
18
avoid losses.
GOP
z }| {
Contribution margin
j z
X }| {
P̄i · ti · (pi − ci )
i=1
AM = j j
· 100 % (B.3)
X X
P̄i · ti · (pi − ci ) + (P̄P C (v̄i ) − P̄i ) · ti · (pi − ci )
i=1 i=1
pi >ci pi >ci
states,i ∈S
| {z }
Loss term: Losses due to certain states (S)
| {z }
Maximum possible GOP
where
AM = Monetary-based availability for considered period in %,

GOP = Gross operating profit for considered period (e.g. year, month, etc.),
i = ith timestamp (averaging period usually 10 minutes) from j,
j = Number of timestamps within considered period,
P̄i = Mean power for timestamp i,
ti = Period duration in hours,
pi = Electricity market price for timestamp i,
ci = Marginal costs for timestamp i (user-specific),
P̄P C = Mean power from reference power curve at wind speed v̄i ,
v̄i = Mean wind speed for timestamp i in m/s,
states,i = WT state s for timestamp i,
S = WT-states considered in the loss term (user-specific)
Shifting O&M strategies toward an optimized monetary-based availability will become more
important when more WTs have to rely on market prices instead of guaranteed remuneration.
This can lead to situations where maintenance is carried out in medium or high winds, but at
low or negative electricity prices, while avoiding low wind times and high electricity prices. Of
course some work requires low wind speeds and safety issues and rules must be prioritized. The
presented definition is borrowed from an upcoming publication by Lutz et al. [67], which will
also look at the application and potential of this KPI in detail.
Appendix B.4. Maintenance tasks

Although availability KPIs measure the result of a maintenance strategy, the corresponding effort
needs to be monitored as well. Basic statistics on the different maintenance tasks are a high-level
approach to monitor and compare the maintenance effort for various assets. In this context,
maintenance tasks are defined as all maintenance-related on-site work on a WT by service staff.
All activities during a visit at the turbine are counted as one single maintenance task. If there
are different possibilities for categorization, the dominant category will be selected. Corrective
maintenance dominates preventive maintenance and repair dominates routine maintenance and
inspection is this context. The counted maintenance tasks shall be normalized to a defined period
of time—ideally 1 year. In addition, the average downtime and costs shall be gathered to allow
19
for further insights. The KPIs described in this section are not to be confused with reliability
characteristics (Appendix C). Reliability characteristics show a higher level of detail (system or
component) and are based only on a part (preventive and corrective repair/replacement) of all
maintenance tasks.
cM T,CategoryY
λM T,CategoryY = (B.4)
∆t
PcM T,CategoryY
i=1 DTM T,CategoryY,i
M DTM T,CategoryY = (B.5)
cM T,CategoryY
PcM T,CategoryY
i=1 CM T,CategoryY,i
M M CM T,CategoryY = (B.6)
cM T,CategoryY
where
λM T,CategoryY = Frequency of maintenance tasks per maintenance category,

M DTM T,CategoryY = Mean downtime per maintenance category,
M M CM T,CategoryY = Mean maintenance costs per maintenance category,
DTM T,CategoryY,i = Downtime of the ith maintenance tasks,
CM T,CategoryY,i = Cost of the ith maintenance task,
cM T,CategoryY = Count of maintenance tasks per maintenance category,
∆t = Observation period in years
Additional insights are possible if maintenance tasks and the corresponding analyses are
categorized using maintenance types and activities as depicted in Figure B1. According to
BS EN 13306 [34] and BS EN ISO 14224 [38], the first breakdown divides maintenance tasks
into preventive and corrective maintenance. The standard [34] defines corrective maintenance
to be ”... carried out after fault recognition..., ” whereas preventive maintenance is ”... intended
to reduce the probability of failure or the degradation of the functioning of an item.” Especially
preventive maintenance tasks could be subdivided into further general and abstract maintenance
types (predetermined, condition-based and predictive maintenance). Instead, as requested in the
survey, we propose to categorize the maintenance tasks in a more practical way according to
the associated activities. The most relevant and thus recommended activities are inspection,
routine maintenance, and repair. Because of different legal requirements on inspections (e.g.,
elevator inspections), it is necessary to distinguish between inspections and routine maintenance.
Definitions are provided by the BS EN 13306 [34] and BS EN ISO 14224 [38], which also include
further categories. The category names in both standards differ slightly, thus Figure B1 covers
both naming schemata.
Appendix C. Recommended reliability KPIs

Appendix C.1. Mean operating time to failure and mean operating time to first failure
The mean operating time to failure (MTTF) and mean operating time to first failure (MTTFF)
as defined by the International Electrotechnical Vocabulary (IEV) [39] describe the average
failure free operating time of a system as being after the start of operation or its replacement.
Although MTTF is a KPI for non-repairable systems or systems repaired to a state as good as
new, MTTFF considers the operational time to the first failure of a repairable system only. While
every failure of a non-repairable system is always the first failure, the semantic differentiation
20
Maintenance
Preventive Maintenance Corrective Maintenance
Inspection
Inspection
Routine
Routine Maintenance
Maintenance // Service
Service
Repair
Repair // Replacement
Replacement
Figure B1. Structure to categorize maintenance tasks by the maintenance type and activity
according to BS EN 13306 [34] and BS EN ISO 14224 [38].
of both KPIs is supposed to avoid confusion. Further failures of repairable systems will be
considered in the mean time between failure (MTBF) (see Section Appendix C.2).
PcR,ItemY
∆toperational,ItemY,i
M T T FItemY = i=1 (C.1)
cR,ItemY
PcF R,ItemY
∆toperational,F R,ItemY,i
M T T F FItemY = i=1 (C.2)
cF R,ItemY
PcR,ItemY
M T T F RItemY = i=1 (C.3)
cR,ItemY
where
M T T FItemY = Mean operating time to failure per non-repairable item,
M T T F FItemY = Mean operating time to first failure per (partially) repairable item,
M T T F RItemY = Mean operating time to final replacement per (partially) repairable item,
∆toperational,ItemY,i = Operational time to ith failure in years,
∆toperational,F R,ItemY,i = Operational time to ith first repair in years,
cR,ItemY = Count of replacements per item,
cF R,ItemY = Count of first repairs per item
Some systems are neither fish nor fowl and thus called partially repairable systems (PRSs).
Although many failures are repairable, PRSs have either some components that can not be
repaired and need to be replaced or some failure modes that require a replacement of the whole
system [68]. A detailed acquisition of the components and failure modes in the maintenance
data would allow a clear distinction to be made between repairable and non-repairable cases.
In practice, however, this level of detail is usually not given. Thus, it seems reasonable to use a
metric to describe the average time until a PRS needs to be replaced. Because a corresponding
metric could not be found in either the standards or the literature, it is defined here as mean
operating time to final replacement (MTTFR).
21
Appendix C.2. Mean time between failure and mean operating time between failure
MTBF and mean operating time between failure (MOTBF) as defined by the IEV [39] both
describe the mean time between two subsequent failures of repairable systems. While MTBF
considers the whole time between two failures (failure occurrence to failure occurrence) and is
thus equal to the mean time between downing events (MTBDE) as defined by MIL-HDBK-338B
[40], MOTBF considers the time between restart and the next failure and is a synonym to the
mean up time (MUT). Being synonyms of other terms, MTBDE and MUT are not displayed in
Figure 2. MOTBF was only introduced because there are different definitions in the literature for
the use of MTBF. Because downtime is usually only a fraction of the total time, the differences
between MTBF and MOTBF are usually marginal in the real application. However, the more
frequent and longer failures become, the more important it becomes to differentiate between the
two KPIs.
PcF,ItemY
∆ttotal,ItemY,i
M T BFItemY = i=1 (C.4)
cF,ItemY
PcF,ItemY
M OT BFItemY = M U TItemY = i=1 (C.5)
cF,ItemY
where
M T BFItemY = Mean time between failure per item,

M OT BFItemY = Mean operating time between failure per item,
M U TItemY = Mean up time per item,
∆ttotal,ItemY,i = Time to ith failure in years,
∆toperational,ItemY,i = Operational time to ith failure in years,
cF,ItemY = Count of failures per item
Appendix C.3. Mean time to restoration and mean downtime

Mean downtime (MDT) is the expected or average downtime after a system fails and stops
operation and before it restarts. The definition provided by the IEV [39] is consistent with
definitions in the literature [38, 37, 34]. Downtime is defined as the total time between the stop
and restart of operation of a considered unit while the unit is in a down state [34]. This period of
time includes all subcategories—for example, waiting time, administrative delays, transportation
time, fault detection, and repair time. Whenever possible, failures and the related downtime
should be assigned to the causative system or component to gain detailed results for further use.
PcF,ItemY
DTF,ItemY,i
M T T RItemY = M DTItemY = i=1 (C.6)
cF,ItemY
where
M T T RItemY = Mean time to restoration per item,

M DTItemY = Mean downtime per item,
DTF,ItemY,i = Downtime due to the ith failure,
cF,ItemY = Count of failures per item
MTTR is defined to be the mean time to restoration by the recent version of the IEV (IEC
60050-192) [39] but was formally called mean time to recovery according to the outdated IEC
60050-191 [69], or mean time to repair [20] in the literature. Although MTTR (as defined in the
22
IEV) and MDT are synonyms, other definitions exclude certain times (e.g., for fault detection
or logistics). We recommend adhering to the definition of IEV and using other KPIs, such
as mean fault detection time (MFDT), mean overall repair time (MRT) or mean active repair
time (MART), if required. Figure C1 from the ISO/TR 12489 [37] standard provides a valuable
depiction of interrelationships and includes categories for even deeper assessments. MTTR is
broken down into its subcategories. Depending on the research question, more detailed KPI can
be calculated.
IEV 191 - MTTR Time to restoration (MTTRes) ISO TR 12489IEC 61508 - MTTR
ISO TR 12489
deﬁnitions IEV 191 Overall repairing time (MRT) ISO TR 12489IEC 61508 - MRT
(shaded boxes)
IEV 191 Corrective maintenance time IEV 191 IEV 191
IEV 191 Active corrective maintenance time IEV 191
administrative
fault
administrative
IEV 191 IEV 191 - MRT
delay
Repair time
detection
delay
logistic
delay IEV 191 IEV 191 IEV 191
time technical fault fault
delay function
(MFDT) localization correction
checkout
time time
time
ISO TR 12489
IEC
IEC 61508 Active repair time ISO TR 12489 IEC 61508 61508
IEC 61508 Preparation and or delay
(MART)
a b c d
Figure C1. Taxonomies of MTTR subcategories from ISO/TR 12489 [37].
Appendix C.4. Failure rate and repair rate

The failure rate (λ) as defined by the IEV [39] describes the ”...probability per unit of time that
the item fails...” [38] during a certain period of time. Being a frequency (failures per time),
the failure rate can be provided in different resolutions (year, day, hour...). As for MTTF and
MTTFF, it has to be differentiated between non-repairable and (partially) repairable systems;
see Section Appendix C.1. Accordingly, for repairable systems, it has to be differentiated between
a failure rate for first failures, intermediate failures and final replacements. For non-repairable
systems, the failure rate is also called hazard rate in the literature [40]. For a constant failure
rate, the relationship between the failure rate and other KPIs is described by Equations C.7 -
C.10.
1
λItemY = (C.7)
M T T FItemY
1
λItemY,F F = (C.8)
M T T F FItemY
1
λItemY,IF = (C.9)
M OT BFItemY
1
λItemY,F R = (C.10)
M T T F RItemY
23
where
λItemY = Failure rate per non-repairable item,

λItemY,F F = Failure rate for first failures per (partially) repairable item,
λItemY,IF = Failure rate for intermediate failures per (partially) repairable item,
λItemY,F R = Failure rate for final replacements per partially repairable item,
M T T FItemY = Mean operating time to failure per non-repairable item,
M T T F FItemY = Mean operating time to first failure per (partially) repairable item,
M OT BFItemY = Mean operating time between failure per (partially) repairable item,
M T T F RItemY = Mean operating time to final replacement per partially repairable item
The repair or restoration rate (µ), as defined by the IEV [39], describes the ”...probability
per unit of time that the restoration of a failed item...” [38] is finished during a certain period of
time and is thus the counterpart to the failure rate. For a constant repair rate, the relationship
to the MTTR is defined by Equation C.11.
1
µItemY = (C.11)
M T T RItemY
where
µItemY = Repair rate per item,

M T T RItemY = Mean time to restoration per item
24

Recommended Key Performance Indicators For Operational Management of Wind Turbines

Uploaded by

Document Informationclick to expand document information

Document Informationclick to expand document information

Copyright:

Available Formats

Recommended Key Performance Indicators For Operational Management of Wind Turbines

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Recommended Key Performance Indicators For Operational Management of Wind Turbines

Uploaded by

Copyright:

Available Formats

16th Deep Sea Offshore Wind R&D conference IOP Publishing

Recommended key performance indicators for

2.1. Survey design

2.2. Survey results

KPI Answers Use Importance (1–5) Abs. Importance

KPI Answers Use Importance (1–5) Abs. Importance

Table 3. Reliability KPIs inquired about in the survey.

KPI Answers Use Importance Abs.

Table 4. HSE KPIs inquired about in the survey.

KPI Answers Use Importance Abs.

KPI Answers Use Importance (1-5) Abs. Importance

3. Recommended KPIs and definitions

Start of Operation Failure Restart Failure

Mean Operating Time To Final Replacement

Mean Operating Time To First Failure Mean Time Between Failure

Mean Time To Restoration Mean Operating Time Between Failure

Mean Down Time

Non-Repairable Systems or As Good As New

Appendix A. Recommended performance KPIs

V̄ = Average wind speed,

P̄Wind = Average wind power density,

Appendix A.2. Power curve

Appendix A.3. Full-load hours and capacity factor

tFL = Full-Load Hours,

Appendix A.4. Data availability

AData = Data availability,

Appendix A.5. Energy consumption factor

ECF = Energy consumption factor,

Appendix A.6. Remote resets

Appendix A.7. Telecommunication interruptions

λT I = Telecommunication interruption rate,

Appendix A.8. Operating time

Appendix A.9. Site quality

SGy,W T i = Site quality per WT and year,

Appendix A.10. Market value factor

M V F = Market value factor,

Appendix B. Recommended maintenance KPIs

Appendix B.2. Production-based availability and Production Ratio

Appendix B.3. Monetary-based availability

AM = Monetary-based availability for considered period in %,

Appendix B.4. Maintenance tasks

λM T,CategoryY = Frequency of maintenance tasks per maintenance category,

Appendix C. Recommended reliability KPIs

Preventive Maintenance Corrective Maintenance

M T BFItemY = Mean time between failure per item,

Appendix C.3. Mean time to restoration and mean downtime

M T T RItemY = Mean time to restoration per item,

IEV 191 Corrective maintenance time IEV 191 IEV 191

IEV 191 Active corrective maintenance time IEV 191

IEV 191 IEV 191 - MRT

Figure C1. Taxonomies of MTTR subcategories from ISO/TR 12489 [37].

Appendix C.4. Failure rate and repair rate

λItemY = Failure rate per non-repairable item,

µItemY = Repair rate per item,

You might also like