DNVGL-RP-0175 Edition December 2017

Transcription

1 RECOMMENDED PRACTICE DNVGL-RP-0175 Edition December 2017 The electronic pdf version of this document, available free of charge from is the officially binding version.

2 FOREWORD DNV GL recommended practices contain sound engineering practice and guidance. December 2017 Any comments may be sent by to 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 others than DNV GL is at the user's sole risk. DNV GL does not accept any liability or responsibility for loss or damages resulting from any use of this document.

3 CHANGES - CURRENT This is a new document. Changes - current Recommended practice DNVGL-RP Edition December 2017 Page 3

4 CONTENTS Changes - current...3 Section 1 General Introduction Objective Scope Application References Definitions and abbreviations Procedural requirements Contents Section 2 External conditions General General effects of icing climate on wind turbines Ice class Wind turbine ice modelling Design documentation Section 3 Design loads Load assumptions Load analysis Design documentation Section 4 Site specific icing conditions Site specific external conditions Methods for determining site specific icing conditions Site specific evaluation Design documentation Section 5 Ice detection, ice protection and influence on control and protection system General Ice detection Control and protection system Ice protection Design documentation Section 6 Manuals General Recommended practice DNVGL-RP Edition December 2017 Page 4

5 Section 7 Component design requirements General Rotor blades Electrical components Exposed structures Met masts Design documentation Contents Changes - historic...42 Recommended practice DNVGL-RP Edition December 2017 Page 5

6 SECTION 1 GENERAL 1.1 Introduction This recommended practice (RP) provides principles, technical requirements and guidance for design and documentation of wind turbines under icing conditions. The RP may be used for the design of on- and offshore wind turbines under icing conditions. Icing conditions are in general not considered in the present codes and standards such as GL Rules and Guidelines, DNVGL-ST-0437, IEC series (see references in [1.5]). 1.2 Objective The objectives of this RP are to: provide an internationally accepted level of safety by defining minimum requirements for wind turbines under icing conditions (in combination with referenced standards, recommended practices, guidelines, etc.) provide guidance for determination of loads and advice the analysis methods as relevant for operational (FLS) and extreme (ULS) conditions, but not accidental conditions (ALS) implicated by icing conditions serve as a guideline for designers, suppliers, purchasers, and regulators serve as basis for the design of wind turbines exposed to icing conditions. 1.3 Scope This RP provides principles and technical requirements for wind turbines under icing conditions both onshore and offshore. The RP covers the following topics: external conditions design loads site assessment to identify risk of ice for a potential site ice detection, including de-icing and anti-icing procedures and influence on control and protection system manuals design requirements. General requirements for wind turbine technical applications in low temperature climate (LT) are outside the scope of this RP. They may be found in DNVGL-RP Application This RP is applicable to all types of horizontal axis wind turbines intended for installation in an icing climate (IC). The RP is also applicable to the design of components for the wind turbines under icing conditions. The RP is intended to be applied in its entirety. Nevertheless, certain parts of it may be omitted if the applied certification scheme allows for such reduction in scope, and provided this is properly documented as a part of the certification process. Deviations from the requirements, or the application of alternative means of complying with these requirements, may be acceptable after consultation and agreement. This RP may be applied as part of the technical basis for carrying out type, component and project certification of wind turbines. For vertical axis wind turbines, the presented principles and technical suggestions should be applied as far as possible in a reasonable sense. Recommended practice DNVGL-RP Edition December 2017 Page 6

7 This RP covers the technical requirements to be applied for the DNV GL certification schemes. It is also intended for application in connection with IEC related certification schemes. 1.5 References This document refers to relevant DNV GL service specifications, standards and recommended practices and to international codes and standards, as well as other international publications. Unless specified otherwise in this RP, the latest valid revision of each referenced document applies. Table 1-1 DNV GL documents Document code Title DNV GL RA Icing map of Sweden, DNV GL Renewable Advisory, 2015 DNV GL research report DNVGL-RP-0363 "Abschlussbericht "IcedBlades", Entwicklung von Methoden zur Erhöhung der Verfügbarkeit von Windkraftanlagen in klimatisch kalten Regionen Deutschlands und Nordeuropa", Final report, March 28 th 2015, DOI /GBV: Extreme temperature conditions for wind turbines DNVGL-SE-0073 Project certification of wind farms according to IEC DNVGL-SE-0074 Type and component certification of wind turbines according to IEC DNVGL-SE-0190 DNVGL-SE-0441 DNVGL-ST-0126 DNVGL-ST-0361 DNVGL-ST-0437 DNVGL-ST-0438 Project certification of wind power plants Type and component certification of wind turbines Support structures for wind turbines Machinery for wind turbines Loads and site conditions for wind turbines Control and protection systems for wind turbines Table 1-2 External documents Document code Title EN Eurocode 1: Basis of structural design, December 2010 EN Eurocode 1: Actions on structures - Part 1-3: General Actions - Snow loads, December 2010 EN Eurocode 1: Actions on structures - Part 1-4: General Actions - Wind loads, December 2010 IEC Wind turbines - Part 1: Design requirements, ed.4 IEC Wind turbines - Part 2: Design requirements for small wind turbines, 2013 IEC IEC Wind turbines - Part 3: Design requirements for offshore wind turbines, ed.1 Wind turbines - Part 12-1: Power performance measurements of electricity producing wind turbines, 2017 IEC Wind turbines - Part 22: Conformity testing and certification, 2010 IEC Wind turbines - Part 24: Lightning protection, 2010 Recommended practice DNVGL-RP Edition December 2017 Page 7

8 Document code IEA Wind Task 19 IEA Wind Task 19 Title Available Technologies of Wind Energy in cold climates, July 2016, IEA Expert Group Study on Recommended Practices, January 2017, IEA ISO Atmospheric Icing of structures, First edition IWAIS NASA NORSOK N-003 PO190 EWEA WindEurope Summit Winterwind Conference The Spatial Distribution of Icing in Germany Estimated by the Analysis of Weather Station Data and of Direct Measurements of Icing, Wichura, B., IWAIS th International Workshop on Atmospheric Icing of Structures, St. John s, Newfoundland and Labrador, Canada, Sep Theoretical and Experimental Power from Large Horizontal Axis Wind Turbines, D J Larry Viterna, NASA, 1982 NORSOK STANDARD, N-003:2017, Actions and action effects Site assessment for a type certification icing class, Freudenreich, K. et al., PO190 EWEA WindEurope Summit, September 27 th -29 th 2016 Classification based approach for Icing Detection, Khadiri-Yazami Z. et al., Winterwind Conference, Åre, Sweden, Definitions and abbreviations Definition of verbal forms Table 1-3 Definition of verbal forms Term Definition shall should may verbal form used to indicate requirements strictly to be followed in order to conform to the document verbal form used to indicate that among several possibilities one is recommended as particularly suitable, without mentioning or excluding others, or that a certain course of action is preferred but not necessarily required verbal form used to indicate a course of action permissible within the limits of the document Definition of terms Table 1-4 Definition of terms Term atmospheric icing Definition all processes where drifting or falling water droplets, rain, drizzle or wet snow in the atmosphere freeze or stick to any object exposed to the weather cold climate combination of icing climate (IC) and low temperature climate (LT) (see also Figure 2-1) ice class specifies the considered icing conditions for a wind turbine type class, considering icing conditions Recommended practice DNVGL-RP Edition December 2017 Page 8

9 Term ice free status ice protection system icing climate in-cloud icing incubation time instrumental icing low temperature climate meteorological icing normal temperature climate precipitation icing recovery time rime rotor icing sea spray ice support structure Definition ice-free status is the state of ensured absence of ice, especially rotor icing, on the turbine. The status may be necessary to be reached before a turbine restart after a rotor icing event. used to mitigate ice build-up on wind turbines, mostly focussed on the blades, and its associated risks for the wind turbine climate in which icing may occur; the most important meteorological parameters for icing are the humidity in the air (liquid water content, mean volume droplet size and cloud base height), air temperature and wind speed. The icing climate starts at temperatures below +3 C and overlaps partly with the low temperature climate (see also Figure 2-1) icing due to super-cooled water droplets in a cloud or fog delay between the start of meteorological icing and the start of rotor icing or instrumental icing (dependent on the surface, the temperature of the structure and the wind turbine operating point) (see also Figure 2-2) period, during which the ice is present/visible at a structure and/or a meteorological instrument ambient temperature for operation below -10 C and corresponding extreme temperature below -20 C (see also Figure 2-1) period during which the meteorological conditions for ice accretion are favourable (active ice formation) (see also Figure 2-2) ambient temperatures for operation from -10 C to +40 C and corresponding extreme temperature range of -20 C to +50 C icing due to either freezing rain or drizzle or accumulation of wet snow delay between the end of meteorological icing and the end of rotor icing (period during which the ice remains but is not actively formed) (see also Figure 2-2) white ice with in-trapped air period during which a wind turbine rotor is disturbed by ice. In difference to instrumental icing rotating turbine rotor has typically shorter incubation and recovery times than stationary instruments (see also Figure 2-2) icing due to sea spray structural components that support the rotor-nacelle-assembly for on- and offshore wind turbines as tower and foundation structures Definition of symbols and equations Table 1-5 Definition of symbols Symbol Definition C Lpen penalty factor for the lift coefficient [-] C Dpen penalty factor for the drag coefficient [-] C85% chord length at 85% rotor radius [m] M ice mass in kilogram per meter blade length as a distributed load [kg/m] Recommended practice DNVGL-RP Edition December 2017 Page 9

10 Symbol Definition r R RTC radial coordinate along rotor radius [m] rotor radius [m] radius from tower centre [m] ρ Sea Spray Ice ice density due to spray icing [kg/m 3 ] ρ Air Ice θ mean,year air density [kg/m 3 ], assumed to match the typical temperature of icing conditions, which is usually in the range of -5 to 0 C annual average ambient temperature (hourly mean value) [ C] Abbreviations Table 1-6 Abbreviations Abbreviation Description ALS accidental limit state AOA angle of attack [ ] CC FLS IC IEA IEC H LAT LDD LT NT RP ULS UPS cold climate fatigue limit state icing climate International Energy Agency International Electrotechnical Commission hub height [m] low astronomical tide load duration distribution low temperature climate normal temperature climate recommended practice ultimate limit state electrical energy storage device 1.7 Procedural requirements Certification Certification principles and procedures related to certification services for wind turbines are specified in relevant DNV GL service specifications listed in [1.5]. Recommended practice DNVGL-RP Edition December 2017 Page 10

11 1.7.2 Testing If testing is done to demonstrate any function or property of an equipment or material, all such tests should be in compliance with the requirements set out in the corresponding certification scheme. The test plan should contain the following information as a minimum: precise description of the tests to be performed including boundary and test conditions measurement parameters to be recorded along with the planned resolution and sampling rate of the measurement data expected test results and success criteria. The test results should be documented in a test report, which should include the following as a minimum: documentation on measurement set-up reference to the standards followed during the testing and clear indication on any deviations from the standards documentation on component types used and serial numbers of the main components (if for example a wind turbine is under testing data of the following components should be recorded: rotor blades, gearbox, generator, tower, ) calibration information of the measuring equipment test results (plots of measurement parameters), its evaluation and a final conclusion. Recommended practice DNVGL-RP Edition December 2017 Page 11

12 SECTION 2 EXTERNAL CONDITIONS 2.1 General In general, the external conditions to be considered for the design are dependent on the intended site or site type for a wind turbine installation. The icing conditions for global wind turbine loads are mainly defined by the average icing duration over the year, affecting the fatigue limit state (FLS). The ultimate limit state (ULS) under icing conditions is mainly affected by the blade icing distribution. For the exposed structural components (appurtenances such as working platforms, ladders, handrails, heli hoist etc.) of the wind turbine, ULS local loads are defined by the maximum amount of accreted ice. The characteristic load associated with ice accretion should be taken as the load caused by accreted ice with a 50-year return period in combination with the effect of the one-year storm. Icing on wind turbine rotating and non-rotating components may originate from various phenomena (for details see IEA Task 19 and ISO 12494): In-cloud icing is icing due to super-cooled droplets from the cloud which freeze immediately on impact with objects in the airflow. It is the most common icing event for wind turbines. Precipitation icing is freezing rain and wet snow. Sea spray icing is freezing sea spray including interaction spray and white cap spray from the waves. In-cloud icing and precipitation icing comprises all processes where drifting or falling water droplets, rain, drizzle or wet snow in the atmosphere freeze or stick to any object exposed to the weather. Therefore, it is defined as atmospheric icing. Sea spray icing is the freezing sea water mainly from spray of the waves. The joint occurence of accumulation of ice caused by sea spray and atmospheric icing should be considered. The physical properties and the shape of the accreted ice will vary widely according to the variation in meteorological conditions during the ice growth. Two types of atmospheric icing can be defined. Rime ice and glaze ice. Both are specified by ice density and shape and may also have different effect on the aerodynamic properties. In practice, accretions formed of layers of different types of ice may also occur, but from an engineering point of view the types of ice do not need to be described in more detail, ISO gives a schematic outline of the major meteorological parameters controlling ice accretion. The ice density of atmospheric icing varies between ρ ice = 300 kg/m 3 for rime ice and 900 kg/m 3 for glaze ice. Ice accumulation caused by sea water depends on parameters as wind speed, air temperature, sea water temperature, wave height and period and geometry of the structure and response to the waves. It occurs when air temperature is below 0 C. The ice mass due to sea spray icing may be taken to decrease linearly to zero from the level corresponding to the highest wave elevation to a level sea spray may reach. The ice thickness should cover the whole circumference of the element. The conditions for atmospheric icing (icing duration and accumulation) may change significantly with height above ground level, see PO190 EWEA WindEurope Summit and ISO Consequently, a representative height above ground level for a wind turbine rotor is required to take these effects in to account. The IC depends on several factors. The most important ones are relative humidity, air temperature and wind speed. IC starts at temperatures below +3 C and overlaps partly with the low temperature climate (LT). Figure 2-1 gives a schematic definition. Recommended practice DNVGL-RP Edition December 2017 Page 12

13 Figure 2-1 Definition of cold climate, low temperature climate and icing climate adapted with modifications from IEA Wind Task 19 While IC conditions correspond to ice formation or accumulation on the wind turbines and their components, icing may occur in the normal temperature (NT) range. LT conditions correspond to temperatures below the normal environmental conditions as stated in the applicable standards or guidelines used for certification, for example DNVGL-ST-0437, DNVGL-RP-0363 and IEC In addition to the meteorological parameters, the dimensions of the structural components, the orientation and exposure to the wind and the height above ground level influence the icing duration and accretion. Different icing durations may be determined. Meteorological icing is the time period during which the meteorological conditions allow ice accretion. Instrumental icing is the time period during which the ice is present/visible at a structure and/or a meteorological instrument. For a wind turbine, the rotor icing duration is of main interest. This is the period during which ice is present on the rotor blade. The definition is according to Figure 2-2: Figure 2-2 Definition of meteorological icing and rotor icing according to IEC Recommended practice DNVGL-RP Edition December 2017 Page 13

14 Rotor icing is different from instrumental icing. The term rotor icing specifies the phenomenon more accurately for ice affecting a rotating turbine rotor, while instrumental icing commonly refers to stationary (non-rotating) objects. For the rotating turbine rotor, high flow velocity and blade vibrations result typically in shorter incubation and recovery times than for stationary instruments. Rotor icing is defined as the period during which the wind turbine rotor accumulates ice. Ice is accreted continuously on a wind turbine rotor until the meteorological conditions for icing are not present anymore (end of the meteorological icing). Ice will remain on the wind turbine for a certain time the recovery time until ice erodes, sublimates, melts or sheds away from the rotor (end of the rotor icing). 2.2 General effects of icing climate on wind turbines An IC may have following effects on a wind turbine and auxiliary equipment (see also IEA Wind Task 19): icing affects wind measurements and may cause data loss and uncertain readings from the measurement equipment ice induced stall effects may lead to controller instabilities, see also DNV GL research report IcedBlades ice on wind turbine blades may cause aerodynamic and mass imbalance and may increase the loading of components, thus reducing the life time ice on the wind turbine blades may increase the noise level of a wind turbine ice on rotor blades may lead to production losses icing conditions may cause larger uncertainty in energy yield calculations compared to standard conditions sea spray icing can occur on offshore foundation, on structural components and appurtenances, and it may lead to significant ice mass accumulation and change in aerodynamic and hydrodynamic coefficients and projected area sea spray icing can influence safety systems as lifeboats, escape routes, ventilation and communication systems precipitation and in-cloud icing may lead to significant ice mass accumulation and change in aerodynamic and hydrodynamic coefficients and projected area ice mass accretion may cause additional gravity loads leading to the collapse or damage of wind turbine components and measurement masts ice throw and ice fall from the wind turbine may be a safety issue ice throw and ice fall from a wind turbine may damage wind turbine components (e.g. nacelle, blades) and auxiliary equipment (e.g. transformer houses, measurement masts) ice on the wind turbine may cause maintenance and repairs to be more difficult iced appurtenances (e.g. ladders) may prevent access to on- and offshore wind turbines. 2.3 Ice class General This RP defines a wind turbine ice class. The ice class is defined by the rotor icing hours per year and the extreme ice accretion (ice layer, ice mass) for the design requirements of exposed components. An ice class type certification is recommended at least if icing events have been observed during long term measurements (preferably ten years or more) on an average of more than 168 hours per year. For further items to be checked see [2.3.2] Global wind turbine loads Wind turbine loads are cross section loads on main wind turbine components including rotor-nacelleassembly, tower, transition piece and foundation, for coordinate systems as e.g. defined in DNVGL-ST The ice class should be applied if one or more of the following conditions are fulfilled: Recommended practice DNVGL-RP Edition December 2017 Page 14

15 in case the expected icing duration at the site exceeds 168 hours per year or detailed investigations on the icing duration have not been carried out in case the wind turbine controller shows unacceptable instabilities under icing conditions and may cause increased turbine oscillations. The ice class should be applied if the expected icing duration at the site exceeds 84 hours per year or detailed investigations on the icing duration have not been carried out and one or more of the following items is fulfilled: the once-per-revolution excitation frequency (1P) differs from the first tower eigenfrequency by less than 10% also under icing conditions (to be shown by means of a Campbell diagram) the fatigue loads for the tower along the main wind direction are not significantly larger (>30%) than transverse to the main wind direction (blade icing increases usually the tower loads lateral to the main wind direction) the tower consists of material represented by inverse S/N curve slopes greater than 5 the turbine concept is different from three-bladed, horizontal axis, pitch controlled with variable speed. In case none of theconditions listed above is fulfilled, icing duration of up to 168 hours per year need not be considered for load assumptions. However, in any case the controller and safety system should be assessed and possibly modified to fit to the icing conditions, as well as the operation and maintenance manuals Loads on appurtenances Appurtenances are exposed components of (mainly offshore) wind turbines as e.g. heli hoists, gratings, hand rails, platforms and J-tubes. These are prone to icing. The ice class should be applied if the 50-year extreme ice load value for non-rotating components G10 or R10 acc. to ISO (i.e. 10 mm ice thickness) is exceeded. G10 or R10 for non-rotating componentss corresponds to 10 mm of icing of single holes of gratings (having about 30 mm opening). This leads very likely to a blockage of the gratings and additional snow adds to the icing and must be considered. Ice and snow mass loads on nacelle cover and spinner can be assumed to be covered by the standard live load value of 3 kn/m 2 according to DNVGL-ST-0361 [11.2.6]. In case site specific ice and snow mass loads on nacelle cover and spinner exceed 3 kn/m 2, then the increased value should be considered in the design Wind turbine ice class definition The wind turbine ice class is defined in Table 2-1. For the wind turbine ice class S the manufacturer should describe the model used and values for the design parameters in the design documentation. Table 2-1 Wind turbine ice class definition Rotor icing duration for global FLS loads Wind turbine ice class I 750 hours per year Wind turbine ice class S To be defined by the manufacturer 50-year extreme ice load value for appurtenances 30 mm (ISO 12494/ice class G30/R30) Sea spray ice thickness for offshore substructures and appurtenances 80 mm (definition of profile according to Table 7-2 ) Recommended practice DNVGL-RP Edition December 2017 Page 15

16 Relation to other ice class definitions: According to IEA Wind Task 19 the following IEA IC site classification has been defined. In Table 2-2 percentage values from IEA Wind Task 19 have been transferred into hourly values. Table 2-2 IEA icing climate site classification, taken from IEA Wind Task 19 IEA Ice Class Meteorological icing duration Instrumental icing duration Reduced production hours per year hours per year % of annual production 5 > 876 > 1752 > to to to to to to to to to 5 1 < 44 < 131 < 0.5 The rotor icing duration is set equal to the instrumental icing duration, see [ ]. 2.4 Wind turbine ice modelling For modelling of the wind turbine ice class the following definitions and parameters should be used Ice mass and mass imbalance of the rotor An extra ice mass on the rotor blades according to Equation (2.1) shall be assumed in the load simulations. An average ice intensity is assumed being representative for all occurring ice intensities (lower ice intensities are conservatively covered, higher ice intensities cause normally the wind turbine to de-rate or shut down, see also IEA Wind Task 19). The ice mass distribution along the blade should be assumed according to Equation (2.1), see also DNV GL research report IcedBlades: where: M = ice mass as distributed load along the rotor radius [kg/m] = scaling parameter [kg/m 3 ] C85% = chord length at 85% rotor radius [m] r = radial coordinate along the rotor radius [m] It is assumed that the ice mass is located on the leading 30% of the blade surface. As a simplified approach the ice mass can be assumed conservatively at the blade leading edge. Inhomogeneous ice mass distribution (two blades iced, one not iced) results from ice shedding off due to vibrations and defrosting. For the FLS analysis the full ice mass according to Equation (2.1) is applied to blade 1 and 2 and 50% of the ice mass according to Equation (2.1) is applied to blade 3. For the ULS analysis the full ice mass according to Equation (2.1) and multiplied by a factor of 2 is applied to blade 1 and 2 and no ice mass is applied to blade 3. Table 2-3 illustrates this. (2.1) Recommended practice DNVGL-RP Edition December 2017 Page 16

17 Table 2-3 Ice mass distribution on blades FLS analysis ULS analysis blade 1 100% of Equation (2.1) 200% of Equation (2.1) blade 2 100% of Equation (2.1) 200% of Equation (2.1) blade 3 50% of Equation (2.1) no ice mass For aeroelastic simulations the ice mass inhomogeneity may be modelled by single applied point masses along the blade Ice mass on non-rotating components Icing mass and snow on non-rotating components as tubular tower, nacelle cover, working platforms, heli hoist may be omitted for the FLS wind turbine loads, see [2.3.2], but shall be considered for ULS appurtenance loads, see [7.4]. Ice mass on primary structural components should be considered in the global load calculation for ULS. For lattice and guyed towers extra ice mass should be considered in the global- and local load assumption according to ISO For the ULS analysis the 50-year extreme ice load should be applied. For the FLS analysis 50% of the 50-year extreme ice load should be applied for the duration given in line rotor icing duration for global FLS loads in Table 2-1. For turbines supported by lattice towers, caution should be exercised because natural frequencies may decrease due to ice mass on the tower. Ice mass due to sea spray ice on open offshore support structures (e.g. jackets) may be omitted for the wind turbine global loads Aerodynamic influence of the rotor Icing of rotor blades leads to changed aerodynamic behaviour of the airfoils, compared to the clean blade. This change can be modelled by considering aerodynamic penalty factors for the iced airfoils. Equation (2.2) and Equation (2.3) describe aerodynamic penalty factors for medium ice accretion. They should be applied for all iced blades. The aerodynamic penalty factors are valid for normal operating conditions (approximately in the range of -2 to +14 angle of attack) and are to be multiplied with the aerodynamic coefficients for the clean airfoils. The penalty factors depend on the angle of attack (AOA): (2.2) where: C Lpen = penalty factor for the lift coefficient [-] C Dpen = penalty factor for the drag coefficient [-] AoA = dimensionless angle of attack, counted in [ ] (2.3) Aerodynamic coefficients for the range outside normal operating conditions (approximately -2 to +14 angle of attack) may be derived according to NASA. Recommended practice DNVGL-RP Edition December 2017 Page 17

18 The aerodynamic influence of the rotor under icing conditions including Equation (2.2) and Equation (2.3) are based on comprehensive literature studies of iced airfoils, see the IcedBlades research project. In case validated aerodynamic coefficients under icing conditions are available for the applied airfoils, this data may be used. Aerodynamic imbalance should be modelled either by applying the penalty factors only to the blade no.1 and 2 and not to the blade no. 3 or by applying the penalty factors to all three blades and an additional pitch error for blade no. 3 of -3 (i.e. lower angle of attack of 3 to compensate for penalty factors on blade no. 3). The aerodynamic penalty factors should be applied identically for all the cases blade iced 50% Equation (2.1), blade iced 100% Equation (2.1) and blade ice 200% Equation (2.1). This means that the aerodynamic penalty factors should be applied to all iced blades, independent on the ice mass distribution along the blade. The aerodynamic imbalance should be applied for all cases where blade no. 1, 2 and 3 do not have the identical ice mass distribution along the blade, i.e. for both FLS and ULS cases Aerodynamic of none-rotating components For lattice and guyed towers additional aerodynamic drag coefficients and influence of the aerodynamics should be considered for the global load assumption, see ISO for details. A correction of the drag coefficient for tubular tower and nacelle due to icing may be omitted Hydrodynamics due to sea spray ice For offshore support structures sea spray ice should be considered for the increased diameter due to ice accretion (according to [2.3.4] or site specific). A correction of the drag and inertia coefficients due to ice accretion may be considered Air density The air density should be corrected in relation to the corresponding air temperature for the time series with simulated icing. The air density should be assumed to match the typical temperature of icing conditions, which is usually in the range of -5 to 0 C. The corrected air density should be considered in the load calculation. Once meteorological icing conditions are over, temperatures may lower and ice may still be present on the blade at temperatures lower than -5 to 0 C. However, as for the representative ice mass in [2.4.1] a representative (average) value for the temperature and air density should be applied. The altered air density and ice accumulation resulting to a shift in turbine operational point may result to: Changes in the power curve unfavourable turbine controller behaviour increase in start-stop cycles for turbines with/without de-icing systems premature and unwanted stall behaviour of turbine blades. See also DNVGL-RP Recommended practice DNVGL-RP Edition December 2017 Page 18

19 2.4.7 Icing duration The icing duration as defined per Table 2-1 should be assumed. This is considered to be representative for most sites prone to icing. Still, a detailed site specific evaluation should be carried out in case significant exceedance of the icing duration according to Table 2-1 may be expected Ice accretion For the design of appurtenances where local loads are essential, ice layer should be applied according to ISO Detailed guidance can be found in Sec Design documentation The following data is relevant for the external conditions of wind turbines under icing conditions and should be documented: defined ice class I with parameters defined according to sections [3.3.4] and [3.4] or: all parameters and definitions for ice class S (e.g. rotor icing duration, ice thickness of the layer for exposed components) sea spray ice thickness for offshore wind turbines. and: operational requirements (extra features for the controller to operate under icing conditions), if applicable ice detection system, if applicable see also [5.5] de-icing and anti-icing strategies (to reduce the rotor icing duration), if applicable see also [5.5]. Recommended practice DNVGL-RP Edition December 2017 Page 19

20 SECTION 3 DESIGN LOADS 3.1 Load assumptions This section provides requirements and information for the global wind turbine loads of the wind turbine and its support structures under icing conditions with ice loads. Loads for the appurtenances to the support structure, nacelle and spinner can be found in [7.4]. The load calculation should consider the modified temperature range, rotor ice mass, aerodynamic and mass imbalance and aerodynamic influence according to sections [2.3] and [2.4]. The load cases under icing conditions are described in Table 3-1. Due to the degradation of blade aerodynamic properties by icing, the impact of possible changed controller behavior under icing conditions on the wind turbine should be investigated. Extra controller features (e.g. ice operation mode see [5.3.4]) to run the turbine under icing condition with more energy yield should be considered in the load assumptions. Effects from de-icing and anti-icing procedures (e.g. heating system, vibrations, etc.) may be considered in the load assumption by reducing the rotor icing duration. The anticipated reduction of icing duration should be verified by prototype testing over at least two winter seasons showing significant icing conditions and durations. Extra mass from de-icing and anti-icing equipment should be considered in the turbine load, if relevant. Effects from ice detection, resulting in stop or derating of the turbine or other ice operation modes may be considered in the load assumption by reducing the rotor icing duration. The distribution of the mean wind speed bins for the icing conditions may be assumed to be identical to the non-iced distribution, unless site specific data show strongly different distributions. For the simulation of time series to be used for fatigue load evaluations the ice mass should be applied identical for the entire icing duration. Table 3-1 Definition of icing design load cases DLC Wind Conditions Icing conditions Load evaluation Partial safety factor for loads 13.1 Power production under icing conditions NTM* vin < vhub < vout Icing modelling according to ice F (fatigue) U (ultimate) N* class *defined according to DNVGL-ST-0437, IEC , IEC and IEC In case the operational strategy of the wind turbine leads to significantly increased start-up and shut-down events compared to non-iced conditions, these increased numbers should be considered in the load assumptions. In case the operational strategy of the wind turbine leads to significantly increased time periods of idling due to icing, this should be considered in the load assumptions. The dynamic characteristics of the wind turbine under icing conditions (possibly by e.g. increased mass, changed eigenfrequencies, reduced rotational speed and power output etc.) might change the controller behaviour or even require changes in the controller strategy and/or parameters. This should be considered in the simulations. Recommended practice DNVGL-RP Edition December 2017 Page 20

21 3.2 Load analysis Load analysis should be carried out according to the standards and guidelines such as DNVGL-ST-0437, IEC , IEC and IEC Design documentation The following data for loads of wind turbines under icing conditions should be documented: load case table distribution of the wind speed bins for icing load cases aeroelastic wind turbine model under icing conditions load time series load evaluations in tabled form for FLS and ULS de-icing and Anti-icing strategies (to reduce the rotor icing duration), if applicable sea spray ice thickness for offshore wind turbines for ULS simulation operational requirements (features for the controller to operate under icing conditions), if applicable. Recommended practice DNVGL-RP Edition December 2017 Page 21

22 SECTION 4 SITE SPECIFIC ICING CONDITIONS 4.1 Site specific external conditions The site specific icing conditions should denote all external icing influences acting on the wind turbine. These are influences resulting mainly from meteorological, but possibly also from orographic, topographic and oceanographic conditions as well as from other external sources. The icing conditions prevailing at the installation site should be analysed and the rules and premises to be applied for the design should be documented. This includes the aspects listed in the following sections. Other conditions that are not listed and may influence the design of the wind turbine should also be stated with clear references to background and the reasoning for using the data. The following data is relevant for loads and site conditions of wind turbines under icing conditions and should be documented: general wind farm location data, turbine positions specific site conditions as e.g. vicinity to open waters, vicinity to thermal power plants (emitting large amounts of water vapour) meteorological data oceanographic data, if relevant operation requirements transport and installation requirements maintenance requirements safety issues (roads, people, ice throw etc.). The external conditions may in principle be determined from various sources: measurements existing data bases, if it is possible to calibrate the existing data to adequately fit the local site conditions numerical analysis icing maps a combination of the above. In the analysis report, the following criteria should be documented: data period used and usable data thereof corrections and assumptions applied and the inherent sensitivity for the resulting design values statistical procedures applied error estimate interpretation of results and concluding values for the design. The site specific icing conditions should be taken as basis for analysing the structural integrity of a wind turbine design (e.g. comparison with design conditions or calculation of site specific loads). 4.2 Methods for determining site specific icing conditions General The determination of site specific icing conditions relies on a combination of different methods, commonly including long-term data from data bases, various measurement approaches performed on the actual wind farm site but covering a shorter time, numerical analysis and icing maps in order to determine the design conditions for each installation site. Wherever national or local requirements with respect to icing conditions exist (based on e.g. national or local standards or building codes), these defined conditions should be considered for a site specific evaluation. Recommended practice DNVGL-RP Edition December 2017 Page 22

23 Sensors for measurements In general, it is recommended to follow IEC requirements for meteorological measurements. In icing climate, attention should be paid to the positioning of the sensors as anemometers and wind vanes. In severe icing conditions, the benefit of sensor heating may be lost if neighbouring objects such as supports and masts are allowed to collect ice. Therefore, these surrounding objects should be heated as well to prevent ice accretion. Wind sensors "[...] Wind sensors may be used for determination of the icing duration. Measurement instruments should be used that are suitable for the climate conditions at the site. Ice growth on sensors and/or the meteorological (met) mast may lead to measurement errors and loss of data. Ice on anemometers and wind vanes may cause them to stop or slow down, and ice build-up on booms or lightning rods may also impact the measurements. Measurement errors may give the wrong conclusions regarding the wind speed at the site. The solution for accurate wind measurements in icing climate may be the use of properly heated anemometers and wind vanes. If cup or propeller type anemometers are used, the anemometer's cup or propeller shaft, and post should be heated in order to prevent ice from accumulating and impacting measurement quality. Even when fully heated, these sensors may not always remain ice free in heavy icing conditions. Ultrasonic anemometers may be more robust sensors in icing conditions than cup anemometers but require calibration and heating similarly to cup anemometers. Since ultrasonic anemometers have no moving components, there are no mechanical effects that influence the measurement.[...]" (IEA Wind Task 19, 2016). See also IEA Wind Task 19 (2017). Temperature and atmospheric pressure sensors "[...] Radiation shields around temperature sensors require ventilation to work properly. The ventilation in conventional small shields with lids may become filled with ice or encased in snow, and provide false readings. For atmospheric pressure sensors, it should be ensured that the air intakes are exposed to the surrounding atmospheric pressure and are not obstructed by ice, causing a false reading. The use of high power heating capability and/or large housings such as those used on meteorological stations may be necessary.[...]" (IEA Wind Task 19, 2017). See also IEA Wind Task 19 (2016). Ice detection sensors Various ice detection sensors exist applying a large variety of mechanisms. They may be installed on the turbine nacelle, turbine blade, or on a met mast. It should be noted that icing on rotating components (e.g. turbine blades) may be different from icing on non-rotating components (e.g. nacelle), see also [ ]. In addition, ice detection may be done using only standard sensors applying algorithms that detect icing signals. The level of maturity differs significantly between different methods and sensors. The capability of a sensor may range between detecting ice/no ice, distinguishing between instrumental and meteorological icing, or detecting the severity and intensity of icing. A list of ice detection sensors is given in IEA Wind Task 19 (2016). It is recommended to apply only ice detection sensors with proven technology for the determination of the site specific icing conditions. To increase the availability and reliability and to reduce the uncertainty, it is also recommended to apply at least two independent methods for direct ice measurements or to combine direct ice measurements with instrumental or meteorological icing measurements for validation. Especially webcams are recommended. See also IEA Wind Task 19 (2017). Site power supply A sufficient and reliable power supply for all measurement and support equipment should be ensured also under harsh environmental conditions. "[...] Monitoring systems implemented in arctic and icing climate may require additional power for the use of heated sensors and other equipment [...]" (IEA Wind Task 19, 2017). Depending on the number and type of sensors significant energy consumptions may be required to keep the sensors ice-free and operational. " [...] Power for heated sensors may often be a challenge when grid power is not available [...]" (IEA Wind Task 19, 2017). Independent power supply systems are options. "[...] Remote monitoring should be implemented to allow early warning of power system problems. The design and implementation of remote power systems may be a non-trivial task [...]" (IEA Wind Task 19, 2017). See also IEA Wind Task 19 (2016) Duration of measurements and long-term correlation Significant variations of the icing duration may occur among different winters. Consequently, icing durations should be measured over a longer period, covering at least 10 years. As an alternative, short term site measurements (covering at least one icing season) may be corrected using long term correlations e.g. from nearby weather stations, airports or icing maps with a high spatial resolution. Since local site effects (as for exposed or sheltered sites) can significantly influence the icing duration, the correlation of nearby data to the investigated site should be validated. It should be noted that icing duration may vary significantly with Recommended practice DNVGL-RP Edition December 2017 Page 23

24 height above sea level, often an increased icing duration is observed with increasing height above sea level. The sensitivity of the obtained site data dependent on the correlation and the applied assumptions should be evaluated Conversion to rotor icing height The frequency of icing conditions may change significantly with height above ground level. Consequently, a representative height above ground level for a wind turbine rotor is required to take these effects in to account. According to IEA Wind Task 19 (2016) this representative height, rotor icing height, is defined as: Rotor icing height = hub height + 1/3 rotor diameter The icing duration should be defined for this rotor icing height. It is recommended to carry out icing duration measurements at rotor icing height. If necessary, data from other heights above ground level should be converted to rotor icing height by applying validated height corrections. These height corrections could be done by using height-dependent icing durations e.g. from nearby weather stations, airports or icing maps with a high spatial resolution. Since local site effects (as for exposed or sheltered sites) can significantly influence the icing duration, the correlation of nearby data to the investigated site should be validated. It should be noted that icing duration may vary significantly with height above sea level, often an increased icing duration is observed with increasing height above sea level. The sensitivity of the obtained site data dependent on the correlation and the applied assumptions should be evaluated Determination of icing duration Determination of icing duration by direct ice measurements Direct ice measurement is done e.g. by observing ice with a camera, measuring the ice mass, etc. A large variety of sensors exists, applying different detection or measurement mechanisms, see also [ ]. It is recommended to apply at least two independent methods for direct ice measurements or to combine direct ice measurements with instrumental or meteorological icing measurements for validation. A validated icing duration based on direct ice measurements may be set equal to the rotor icing duration as a conservative approach, in case long-term correction and hub height conversion are applied Determination of instrumental icing duration Instrumental icing is commonly measured by comparing the effects of icing on one ice-sensitive sensor and one non-ice-sensitive sensor. Often double anemometry technique using one heated and one unheated anemometer is applied. The signals of the unheated anemometer will be disturbed by icing blockage, but not the signals of the heated anemometer. A common setup is the application of one heated ultrasonic anemometer used as undisturbed reference and one unheated cup anemometer. For ice indication, the measured difference between heated and unheated anemometer (at the same mast height) should be at least twice the sum of the two uncertainties ranges of both anemometers. Shadowing effects from the mast should be avoided. Other validated instrumental icing measurement methodology may also be used. Additionally, to the wind speed measurements, contemporaneous meteorological measurements of temperature and humidity may be used to ensure a higher reliability of the determination of instrumental icing duration. If no other measured data available than temperature and relative humidity, then the following boundary conditions for the determination of instrumental icing may be considered according to PO190 EWEA WindEurope: the measured temperature at the reference height is lower than 3 C the measured relative humidity at the reference height is higher than 95%. It should be noted that this approach is very conservative and may lead to strong over-estimation of the icing duration, see also IEA Wind Task 19 (2017) section It should also be noted that measurements or relative humidity below 0 C may be very uncertain. It is recommended to combine this approach with a direct ice measurement system as e.g. camera observations in order to double validate the sensor readings. Recommended practice DNVGL-RP Edition December 2017 Page 24