Professional Documents
Culture Documents
The PDF electronic version of this document available at the DNV website dnv.com is the official version. If there
are any inconsistencies between the PDF version and any other available version, the PDF version shall prevail.
DNV AS
FOREWORD
This service document has been prepared based on available knowledge, technology and/or information at the time of issuance of this
document. The use of this document by other parties than DNV is at the user's sole risk. DNV does not accept any liability or responsibility
for loss or damages resulting from any use of this document.
CHANGES – CURRENT
Changes - current
This is a new document.
DNV AS
Acknowledgements
Changes - current
This recommended practice has been developed based on the results of a joint industry project (JIP). The
following companies, listed in alphabetical order, are acknowledged for their contributions to the JIP:
— EDF
— EnBW
— Equinor
— Goldwind
— IFPEN
— RWE
— Shell
— Southern
— Company
— Vaisala
— Vattenfall
— ZX Lidars
DNV AS
CONTENTS
Contents
Changes – current.................................................................................................. 3
Acknowledgements................................................................................. 4
Section 1 General.................................................................................................... 7
1.1 Introduction......................................................................................7
1.2 Objective...........................................................................................7
1.3 Scope................................................................................................ 7
1.4 Application........................................................................................ 7
1.5 References........................................................................................ 8
1.6 Definitions and abbreviations........................................................... 8
DNV AS
Contents
Section 8 Bibliography.......................................................................................... 25
8.1 Bibliography.................................................................................... 25
Changes – historic................................................................................................ 26
DNV AS
SECTION 1 GENERAL
1.1 Introduction
Measurements using remote sensing devices have become increasingly important for the wind industry.
Lidars (light detection and ranging) are now commonly used in many wind energy applications as
they provide a cost-efficient way of measuring horizontal wind speeds and directions in comparison to
conventional anemometers and wind vanes on met masts. The first order quantities (wind speed and wind
direction) from lidars correlate very well with met mast data and are already being deployed for turbine
testing, site assessment, and energy assessment modelling, etc. The second order quantity turbulence
intensity (TI) from lidars is not yet widely accepted, even when a met mast is available for verification.
This recommended practice (RP) provides recommendations for the use of TI measured by ground-based
vertical profiling lidars (GBL) in wind energy applications. The RP’s main purpose to define acceptance
criteria (AC) that provide clear limits when GBL TI is used as an alternative to a co-located cup or sonic
anemometer TI. The AC employ error metrics to represent the deviation between lidar and met mast (cup or
sonic anemometry) TI in a meaningful way.
Applicable standards are currently based on the use of TI measurements from cup or sonic anemometry. This
impedes the application of lidar measured TI. This RP provides a method for accepting TI measurements from
GBL when validated against a co-located met mast.
This recommended practice supplements DNV-ST-0437.
1.2 Objective
The objectives of this RP are to:
— provide a method for using GBL TI measurements in combination with co-located met mast data
— serve as a supporting document when considering the use of GBL TI measurements for the application
of wind resource and wind turbine (WT) certification in combination with other applicable standards,
recommended practices, and guidelines.
1.3 Scope
This RP provides recommended acceptance criteria (AC) for GBL TI measurements for:
— loads - site suitability and load validation type testing
— energy assessment modelling.
The scope is limited to measurement data from:
— onshore sites with a 'flat' terrain as defined by IEC 61400-1 or IEC 61400-12-1 as it applies to the use
case
— offshore sites as defined by IEC 61400-3-1 or DNV-ST-0437.
This RP does not address the use of other kinds of lidars, such as floating, nacelle-mounted, or scanning
lidars.
1.4 Application
The RP is applicable for the use of GBL TI in:
— determining the site suitability and validating loads for type testing the WT design and certifyingf WTs
— preconstruction energy assessment wind farm modelling and reporting.
The methods proposed are technology agnostic and intended to be applicable for all GBL models and types.
DNV AS
1.5 References
Table 1-1 lists DNV references used in this document.
IEC 61400-3-1 Wind energy generation systems - Part 3-1: Design requirements for fixed offshore wind
turbines
IEC 61400-12-1 Wind energy generation systems - Part 12-1: Power performance measurements of electricity
producing wind turbines
Term Definition
shall verbal form used to indicate requirements strictly to be followed in order to conform to the
document
should verbal form used to indicate that among several possibilities one is recommended as
particularly suitable, without mentioning or excluding others
may verbal form used to indicate a course of action permissible within the limits of the document
Term Definition
acceptance criteria acceptable range of GBL TI measurement error, as defined by key performance indicators
(KPI), for a chosen use case
DNV AS
Term Definition
error metrics KPIs that are an expression of the error between the underlying GBL and cup anemometer TI
measurements
ground based light vertically facing lidar system that probes the lower boundary layer to measure wind speed
detection and ranging and direction
GBLs used in the wind industry commonly employ the Doppler principle to measure the
change in frequency between light emission and backscatter from aerosols at select heights
above ground. These changes in frequency are then used to derive wind speed statistics.
turbulence intensity software that modifies the GBL TI measurements to resemble cup TI measurements
adjustment method
1.6.3 Symbols
n = index for measurement point in wind bin
N = total number of measurement points in wind bin
i = index for wind bin.
1.6.4 Abbreviations
The abbreviations described in Table 1-5 are used in this document.
Abbreviation Description
AC acceptance criteria
LOS line-of-sight
ML machine learning
RP recommended practice
TI turbulence intensity
WT wind turbines
DNV AS
SECTION 2 MEASUREMENT OF TURBULENCE INTENSITY USING A
LIDAR
2.1 General
This section provides background information on the measurement of TI using a lidar.
TI data measured by lidars cannot directly replace TI measurements obtained by using cup or sonic
anemometers (hereafter referred to as 'cup'), as their characteristics are fundamentally different. Methods
for wake and turbine load modelling found in current standards, e.g. IEC 61400-1, have been developed from
cup TI data and therefore it is implicitly given that lidar data cannot be used directly.
Lidars and cup anemometers measure TI fundamentally differently for a number of reasons. A cup
anemometer wind speed is a scalar point measurement, and it can be generally expected that the
average wind flow across the cups will be somewhat less variable than the volume of air measured by a
lidar /1/, /2/, /3/, /4/.
Furthermore, TI differences between point (cup anemometer) and volume (lidar) TI measurements are
dependent on atmospheric stability /3/, /4/, /5/. This is generally because under stable conditions turbulent
length scales are small and under unstable conditions the turbulent length scales are larger thus resulting in
lidars generally measuring lower TI in stable conditions and higher TI in unstable conditions relative to a cup
anemometer.
All lidars are susceptible to instrument noise. For some lidars this can lead to the variance being
overestimated by the lidar in unstable conditions (larger scale turbulence) /2/, /3/. This is inherent in all lidar
data, and filtering techniques to remove data spikes from the spectral data are included in many complex
correction methods /1/, /2/, /3/, /9/.
The probe volume or radial beam area may also be ineffectual in capturing small turbulent motion in the
atmosphere. This is because the turbulent structure is too small to be captured across the probe length and
as a result the turbulence intensity measurement is impacted by a low pass filter effect /1/, /2/, /3/, /4/.
In summary, GBL TI is typically significantly different from cup anemometer TI due to volume averaging and
the response to atmospheric stability. To date, there is no industry-wide consensus on how to incorporate
GBL TI in current models or how to adjust the lidar measurement to a cup equivalent TI. This is partly due
to the fact that each lidar model and type observe turbulence differently and therefore a universal correction
method for lidars is not possible.
In addition to the aforementioned factors driving differences between lidar and cup anemometer TI (i.e.
volume, noise, and stability), there is a further unique difference to be considered.
Line-of-sight (LOS) lidar technology is impacted by variance or cross contamination /1/, /2/, /3/, /4/, /6/. As
the LOS measurement assumes a homogenous flow across the circumference of the beam, TI measurement
errors arise when this is not true.
DNV AS
SECTION 3 USE CASES
3.1 General
Lidars are applied in a wide range of wind-energy applications. This RP focuses on the measurement of TI by
GBL to be applied in the use cases described in [3.2] and [3.3].
3.2 Loads
The umbrella term 'loads' covers the following two use cases that are applicable to this RP:
— site suitability: the calculation of site-specific loads using an aero-elastic code and site-specific wind
conditions, showing that site specific loads are within the design loads
— load validation: as part of type the testing for type certification, load measurements are carried out
according to IEC 61400-13. The load measurements are compared with aero-elastic loads in order to
validate the software, model, and method assumptions.
DNV AS
SECTION 4 ERROR METRICS
4.1 General
As the basis for all analysis of TI determined by lidars in comparison to TI determined by cup anemometers,
the definition of appropriate error metrics is required.
In the following, the terms 'error' and 'error metrics' are applied because they are customary in this context.
They are used with the intention to describe the deviation between a lidar and a cup anemometer measured
quantity. Both lidars and cup anemometers deliver correct data according to their physical measurement
principles.
For the use cases analysed in this document, the TI error metrics determined as appropriate are the 'mean
relative bias error' (MRBE) and 'relative root mean square error' (RRMSE). The MRBE defines the relative
error or TI bias between the lidar and cup TI while RRMSE compares the scatter of the two quantities. The
error metrics are described further by Equation (4.1) and Equation (4.2), where the wind bin is denoted by ‘i’
and ‘n’ refers to each point within that bin.
(4.1)
(4.2)
DNV AS
SECTION 5 ACCEPTANCE CRITERIA
5.1 Introduction
As described in Sec.2, physical measurements by a lidar are substantially different to those by a
cup anemometer. Thus, it is to be expected that differences in wind statistics between lidar and cup
measurements will exist. This is especially the case for second order quantities like TI. As industry models
and assumptions include TI from cup anemometers, the accepted tolerance of TI measurement deviations
between the lidar measured TI and the cup measured TI needs to be understood.
The error metrics defined in Sec.4 shall be applied to determine this TI error. The error metrics are used
as key performance indicators (KPI). The smaller the (absolute) value of the KPI, the more alike the
measurement result of a lidar and a cup anemometer is assumed to be. This section defines the AC for which
GBL TI measurements can be used, whether they be direct or adjusted.
According to Sec.3, this RP provides AC for a limited scope of use cases:
— loads
— site suitability
— load validation
— energy assessment modelling.
The TI AC for GBLs will be different for these three use cases as each application has a different sensitivity to
TI. Consequently, the AC for each use case are defined independently.
The AC consist of thresholds for one or more error metrics addressing, for example, the accuracy or precision
of the respective TI measurement.
5.2 Use cases 'site suitability' and 'load validation type testing'
5.2.1 General
Due to the similarities in the approaches to analysing WT loads for site suitability and load validation type
testing, this section covers both of these use cases.
DNV AS
5.2.3 Analysis
The definition of the acceptance criteria in this section is based on an analysis which was performed to
determine the relationship between the error metrics and wind turbine fatigue loads. A load data base was
established consisting of a large number of load simulations. A 10 MW wind turbine model was used for the
simulations, applying a range of wind speeds and TIs as input parameters. DELs were processed for selected
Wöhler slopes and load sensors from the blade root, machinery, and tower. The loads were put into the data
base where they could be accessed for chosen input parameters for wind speed and TI. Based on relevant
standards and certification experience, acceptable margins for certified loads were formulated.
In a further step, measurement data sets consisting of GBL TI and co-located cup data were analysed. TI
error metrics were calculated for each valid 10 min time series of the data set. Linking the simulated load
data base with the measurement data base via wind speed and TI, conclusions regarding acceptable margins
on the error metrics were defined. Based on these results, the AC for GBL TI error were defined and are
provided in [5.2.4].
DNV AS
Figure 5-1 AC for 'site suitability' and 'load validation type testing' use cases (left: turbulence
intensity MRBE and right: turbulence intensity RRMSE)
5.3.2 Analysis
The definition of the acceptance criteria in this section is based on an analysis which was performed to
discover the sensitivity of energy assessment modelling to TI. The focus was on the impact of TI on wake
DNV AS
modelling. The intention of this analysis was to determine the influence of a potential TI error on energy
assessment modelling due to the use of a GBL instead of a cup anemometer as sensor a type.
The analysis included datasets from 34 projects in 19 countries, onshore and offshore, featuring wind farm
sizes from four (4) to 426 turbines and ambient TI at 15 m/s ranging from 4.9% to 15.9%. The energy
production was modelled using the project’s wind speed binned TI and then, for each wind speed bin, the TI
was adjusted by ±2%, ±5% and ±10%. A distribution of the change in energy production from the five TI
scenarios and 34 projects was assembled. The distribution of error was then grouped and the change in the
energy production was evaluated.
The results regarding the error metric TI MRBE were as follows:
— for a TI MRBE of ±2%, the modelled energy production changed by less than ±0.10% for approximately
95% of projects
— for a TI MRBE of ±5%, the modelled energy production changed by less than ±0.25% for approximately
95% of projects
— for a TI MRBE of ±10%, the modelled energy production changed by less than ±0.50% for approximately
95% of projects.
Based on the results of this analysis, the AC for GBL TI error were defined and are provided in [5.3.3].
There are no standards that establish GBL TI error thresholds for energy assessment reports. The decision
on what is acceptable in terms of energy production risks and varying levels of risk tolerance is left to the
judgement of the expert performing the energy assessment. Therefore these results provide guidance
regarding an acceptable TI difference between a cup and GBL in an energy assessment.
DNV AS
Figure 5-2 AC for the 'energy assessment modelling' use case
DNV AS
SECTION 6 TURBULENCE INTENSITY ADJUSTMENT METHODS
6.1 General
6.1.1 Introduction
A comparison of the TI measured by lidar and by a met mast (cup or sonic) shows a difference. This
difference is mostly driven by the higher variability in the volume of air measured by the lidar, atmospheric
stability, instrument noise, and low pass filter effects.
If the difference is small and the error metrics determined for a GBL TI dataset prove to be smaller than the
AC for the specific use case according to Sec.5, the lidar data may replace the met mast data. If any of the
AC are exceeded, direct application is not possible.
The goal of a TI adjustment method is to reduce the error and scatter between GBL and cup TIs
measurements and obtain GBL TIs that resemble cup TI to the best possible extent. If the resulting adjusted
TI meets the respective use case AC defined in Sec.5, then the adjusted lidar measured TI data may be used
for the respective use case.
The adjustment algorithm will be referred to as the TI adjustment method (TI AM) in this section and Sec.7.
6.1.2 Specifications
There are numerous kinds of TI AMs. Some are purely analytical approaches that process the lidar output
data according to experience derived from field tests or empiric studies. Others apply advanced filtering
methods, adjust the wind field reconstruction thereby improving the temporal coherence between the lidar
beams, or use machine learning techniques. A combination of these methods may also be employed. Figure
6-1 is a flowchart example of how GBL TI data can be adjusted using machine learning (ML) techniques. In
the training phase, lidar and met mast data are read from a data base and used to set up the ML model.
In the application phase, the model processes lidar data from a specific site and provides adjusted data as
output.
Figure 6-1 Example flowchart for the training and application of an ML model
In many cases, the TI AMs are tailored to a specific lidar model and type. Suppliers of TI AMs mainly include
lidar OEMs and independent companies and consultants. These GBL TI AMs require statistical 10 min lidar
data and/or 1Hz lidar data that may include spectral raw data.
Usually, a strong relationship exists between a lidar type and the TI AM.
DNV AS
It shall be taken into consideration that the lidar data are sensitive to data length, variance, randomness,
etc. Additionally,
data synchronization and cardinal point alignment (wind vane vs. GBL) are also of importance. These aspects
shall be throughly considered in the context of the performance expected for the TI adjustment of a dataset.
DNV AS
Figure 6-2 Strategy for training and tuning a TI AM
6.4 Examples of TI AM
This subsection provides examples of GBL TI measurements from different sites. Error metrics have been
calculated for the lidar data before and after an exemplary TI AM was applied. Figure 6-3 shows the results
of a TI MRBE between a lidar and reference cup anemometer both with and without a TI AM. The AC band
for the use case 'site suitability' is marked in red. Before the TI adjustment, the TI MRBE data exceed the
AC for wind speed bins below 8 m/s and at 24 m/s. After the TI adjustment, the error metrics are within the
DNV AS
AC band. For TI RRMSE in Figure 6-4, the exceedance appears even for medium wind speeds below 12 m/s.
After the adjustment, compliance with the AC is observed.
Figure 6-3 Turbulence intensity MRBE AC band for the use case 'site suitability', lidar data site A
before and after adjustment
There may be cases for one or more wind speed bins if the error between the adjusted GBL TI and cup TI is
larger than the error between the non-adjusted GBL TI and cup TI. Therefore, the TI AM does not necessarily
always improve the agreement between the two measurement techniques, especially in cases where the
error between the non-adjusted GBL TI and cup TI is comparatively low. As a consequence, TI AMs should
not be applied in these cases.
DNV AS
Figure 6-4 Turbulence intensity RRMSE AC band for the use case 'site suitability', lidar data site A
before and after adjustment
DNV AS
SECTION 7 APPROVAL OF GBL TURBULENCE INTENSITY
MEASUREMENTS
7.1 Introduction
Industry accepted wind energy standards usually require TI measurement data from cup or sonic
anemometers. Alternatively, if another wind speed sensing technique like a GBL is used, the TI shall have
equivalent characteristics to those measured by a cup anemometer, see IEC 61400-13, clause 7.2.1.
The method described in this RP provides a tool to assess if a lidar type is able to measure TI sufficiently
similar to a cup anemometer to meet defined acceptance criteria and give some guidance on how TI AMs can
be applied so that the TI measurements are similar to those from a cup anemometer.
A procedure is required by which GBL TI measurements can formally be approved as providing TI
measurement data to comply with applicable standards.
[7.2] presents the approach that shall be followed when comparing GBL TI measurements against concurrent
co-located met mast measurements. [7.3] outlines further requirements for using GBL TI after a relocation of
the lidar in the vicinity of the initial site.
DNV AS
4) A third party has confirmed that the GBL TI meets the use case AC following the present procedure.
DNV AS
SECTION 8 BIBLIOGRAPHY
8.1 Bibliography
Table 8-1 Bibliography
Dimitrov N, Borraccino A, Peña A, Natarajan A, Mann J. Wind turbine load validation using lidar-based wind
/1/
retrievals. Wind Energy. 2019;22(11):1512-1533.
Newman J F , Clifton A, Churchfield M J , Klein P. Improving lidar turbulence estimates for wind energy. Journal
/2/
of Physics: Conference Series. 2016;753(7): 072010.
Sathe A, Banta R, Pauscher L, Vogstad K, Schlipf D, Wylie S. Estimating turbulence statistics and parameters
/3/
from ground- and nacelle-based lidar measurements, IEA Wind Expert Report. Roskilde: DTU, 2015.
Sathe A, Mann J. A review of turbulence measurements using ground-based wind lidars, Atmospheric
/4/
Measurement Techniques. 2013;6(11):3147-3167.
St Martin CM, Lundquist JK, Clifton A, Poulos GS, Schreck SJ. Wind turbine power production and annual energy
/5/
production depend on atmospheric stability and turbulence. Wind Energy Science. 2016;1(2):221-236.
Kelberlau F, Mann J. Cross-contamination effect on turbulence spectra from Doppler beam swinging wind lidar.
/6/
Wind Energy Science. 2020;5(2):519-541.
DNV GL. WindFarmer: White Paper [Internet]. Høvik: DNV GL; 2016. [cited 24-03-2023].
/7/
Available from: WindFarmer white paper, April 2016 DNV.
Lee JCY, Stuart P, Clifton A, Fields MJ, Perr-Sauer J, Williams L, et al. The Power Curve Working Group's
/8/
assessment of wind turbine power performance prediction methods. Wind Energy Science. 2020;5(1):199-223.
Sathe A, Mann J, Gottschall J, Courtney MS. Can wind lidars measure turbulence? Journal of Atmospheric and
/9/
Oceanic Technology. 2011;28(7):853-868.
DNV AS
CHANGES – HISTORIC
Changes – historic
There are currently no historical changes for this document.
DNV AS
About DNV
DNV is the independent expert in risk management and assurance, operating in more than 100
countries. Through its broad experience and deep expertise DNV advances safety and sustainable
performance, sets industry benchmarks, and inspires and invents solutions.
Whether assessing a new ship design, optimizing the performance of a wind farm, analyzing sensor
data from a gas pipeline or certifying a food company’s supply chain, DNV enables its customers and
their stakeholders to make critical decisions with confidence.
Driven by its purpose, to safeguard life, property, and the environment, DNV helps tackle the
challenges and global transformations facing its customers and the world today and is a trusted
voice for many of the world’s most successful and forward-thinking companies.