Sign In

WIceAtlas

Methodology

Background

The most common way of icing is in-cloud icing which means that cloud base height is at rotor elevation or below and temperature is below zero. WIceAtlas is based on cloud base height (CBH) measurements. To achieve more accurate temperature at hub height vertically interpolated MERRA reanalysis data was used for temperature. 

 In-cloud_icing.png

​The original idea of investigation icing severities by using icing frequency information was initiated in WECO [1] and NEWICETOOLS [2] EU projects. Airport METAR data was used to indicate icing frequencies in Europe (see Figure 1).

weco_and_newicetools_map.png 

Figure 1. WECO icing map (left) and NEWICETOOLS icing map (right)

Dr Lasse Makkonen (VTT) then further developed the method of using METAR data as in-cloud icing information. As a result, an icing map of Europe was published including even more data points than Figure 1 icing maps [3].

Icing_map_of_EU_metar.png 

Figure 2. Icing map of EU from METAR data

Datasets

Observations

WIceAtlas is based on weather observations around the world [4]. Observation time series from 1979 to 2015 was used to create the map. As every meteorological station has different and sometimes irregular sample rates data was interpolated to 1 h resolution. Following filters was used to select representative stations:

  • At least 20 years of data per station
  • At least 70 % of data availability per station

~4500 meteorological stations fulfilled these requirements. Stations can be seen in Figure 3.

WIce_stations.png 

Figure 3. Overview of available WIceAtlas meteorological stations.

Based on temperature and cloud height measurements, a cumulative vertical icing profile was calculated for every meteorological station and a linear fit was applied [5]. Example profile can be seen in Figure 4. This meteorological station specific profile was used with elevation data [6] to create a map (principle presented in Figure 5). The linearized vertical icing profile was interpolated between closest stations for every point of map by using inverse distance weighting. Original elevation data resolution was ~0.004 x 0.004 degrees, but in public icing map resolution is upscaled to 0.05 x 0.05 degrees (~6 x 3 km in scandinavia).

Example_profile.png 

Figure 4. Typical vertical icing profile and linear fit.

 

Figure 5. Map creation principle.

In open access map icing resolution is intentionally reduced. Icing severity is based on IEA ice classification table. In public map IEA classes from 3 to 5 are grouped to "Moderate to high icing frequency" class.

Table 1. IEA ice classification table [7]

 

MERRA reanalysis data

In addition to CBH, temperature and wind speed at different heights is needed to create the icing map. More accurate results can be obtained using MERRA reanalysis data for temperatures and wind speed for higher than 50 m a.g.l. compared to met station measurements 10 m a.g.l. Values for certain heights are interpolated from raw data.

Icing intensity [8] is dependent on wind speed. WIceAtlas timeseries are on-off signals and icing intensity information is not available. However, in order to make site icing measurements comparable with WIceAtlas data, low wind speed periods were filtered out.

Resolution of MERRA data is 0.5 x 2/3 degrees (~55 x 30 km in scandinavia). 

Low temperature climate map

Low temperature climate areas do not automatically mean icing conditions, but when building wind turbines to low temperature areas, special low temperature adaptations to turbines should be considered. The low temperature map was calculated using MERRA reanalysis data from 1979 to 2013. DNV-GL definition for low temperature climate was used [9]:

  • Average annual air temperature below 0°C
  • OR air temperature below -20°C on more than 9 days per year (hourly value)
  • ≥ 10 years data required

Low temperature climate is available as transparent layer in WIceAtlas map.

 

Figure 6. Low temperature climate.

Validation and calibration

Using CBH and temperature as icing indicator produces an on-off icing time series signal and icing intensity is not known. Icing intensity depends on liquid water content, wind speed and shape of the object. WIceAtlas results were compared to publically available, extensive icing analysis from Sweden and Canada [10] [11] over large geographical areas. CBH+T method seems to overestimate icing hours mainly because in addition to liquid water in clouds, clouds also contain solid ice crystal particles not resulting to ice growth. Therefore the map is calibrated. The goal of the calibration was to match production losses of these studies to WIceAtlas meteorological icing using IEA ice classification table. For more information about WIceAtlas validations, see here [12] [13] [14].

 

Acknowledgements

Special thanks to Technocentre éolien in Quebéc, Canada for providing measurements for initial validation of the WIceAtlas method and for generating together the Quebéc icing map [15]. WIceAtlas work and website was funded by Tekes (the Finnish Funding Agency for Innovation) within ITEA2_CAP (Collaborative Analytics Platform) project.

 icing_map_Gaspe_low_res.jpg

Figure 7. Icing map of Quebéc.

References and more information

[1]B. Tammelin et al., "Wind Energy Production in Cold climates (WECO)," FINNISH METEOROLOGICAL INSTITUTE, JOR3-CT95-0014, 1998.
[2]B. Tammelin et al., "New Icetools – Experimental Wind Energy Data from Cold Climate Sites in Europe," 2005.
[3]Ben Bernstein, Lasse Makkonen, and Erkki Järvinen, "European Icing Frequency Derived From Surface Observations," in IWAIS XIII, Andermatt, 2009.
[4]NOAA National Centers For Environmental information, Land-Based Station Data.
[5]Simo Rissanen, Ville Lehtomäki, Øyvind Byrkjedal and Rolv Bredesen, "Comparing observed and modelled cloud base height for ice assessment," in WindEurope, Hamburg, 2016.
[6]Jonathan de Ferranti, DEM elevation data.
[7]IEA Wind, "Recommended Practices for wind energy projects in cold climates", 2011.
[8]International organisation for standardization, "ISO 12494: Athospheric icing of structures," 2001.
[9]DNV GL. (2011) Technical Note 067 – Certification of Wind Turbines for Extreme Tempratures.
[10]Till Beckford, "Estimating energy losses caused by blade icing from pre-construction wind data" , in Winterwind, Piteå, 2015. 
[11]Antoine Lacroix, "Atmospheric Icing Effects on Wind Energy Production in Canada" , in Winterwind, Östersund, 2013.
[12]Ville Lehtomäki and Simo Rissanen, "Simple methodology to map and forecast icing for wind power", in Winterwind, Sundsvall, 2014.
[13]Simo Rissanen and Ville Lehtomäki, "Wind Power Icing Atlas(WIceAtlas) & icing map ofthe world", in Winterwind, Piteå, 2015.
[14]

Simo Rissanen and Ville Lehtomäki, "Global Wind and Icing Optimization Atlas: Case Finland", inWinterwind ,Åre, 2016.

​[15]

​V. Lehtomäki, S. Rissanen and M. Wadham-Gagnon, "Low temperature & icing map for Québec," Quebéc wind conference, Gaspé, Canada, 2014.