Windpower in cold climates – Vindforsk project V-313

732 views

Published on

Presentation by Hans Bergström, Uppsala University at Winterwind 2012, session 3a. Windpower in cold climates – Vindforsk project V-313

Published in: Technology, Business
0 Comments
0 Likes
Statistics
Notes
  • Be the first to comment

  • Be the first to like this

No Downloads
Views
Total views
732
On SlideShare
0
From Embeds
0
Number of Embeds
5
Actions
Shares
0
Downloads
24
Comments
0
Likes
0
Embeds 0
No embeds

No notes for slide

Windpower in cold climates – Vindforsk project V-313

  1. 1. Wind power in cold climates WeatherTech Vindforsk project V-131 Hans Bergström and Petra Thorsson, Uppsala universitet Esbjörn Olsson and Per Undén, SMHI Stefan Söderberg, Weathertech Scandinavia Hans Bergström – Institutionen för geovetenskaper 1Winterwind 2012
  2. 2. V-313 Wind power in cold climates Goals: Increase certainty in estimating production in areas with icing. WeatherTech Especially to determine methods to produce a climatological ice map for Sweden on 1 km2 resolution. The  project  is  a  collabora0on  between:   Uppsala  University  –  Dep.  of  Earth  Sciences  (meteorology)        Hans  Bergström,  Petra  Thorsson   Weathertech  Scandinavia        Stefan  Söderberg   SMHI   Per  Undén,  Esbjörn  Olsson,  Ulf  Andræ,  Björn  Stensen     Hans Bergström – Institutionen för geovetenskaper 2Winterwind 2012
  3. 3. What  do  we  mean  by  an   icing  climatology ?     A  mapping  of  icing  should  give  answers  to  ques0ons  as:   WeatherTech   • How  oHen  icing  occurs   • How  long  the  icing  persists   • How  large  the  amounts  of  ice  are   The  most  important  ques0ons  may  be:     • How  oHen  do  ac0ve  icing  occur?   • How  is  the  produc0on  affected?   Hans Bergström – Institutionen för geovetenskaper 3Winterwind 2012
  4. 4. An  icing  climatology  could  be  based  upon  todays  NWP  models   (Numerical  Weather  Predic0on)  or  upon  meteorological   observa0ons.   Data from weather Data from meteorological models (NWP) or observations WeatherTech Model of ice load dM = E ⋅ w ⋅V ⋅ D − Q (Makkonen) dt dM/dt=ice  growth  (M  =  ismassa,  t  =  Ed)   E  =  accre0on  efficiency   Modelled ice load w  =  liquid  water  content   Comparisons with V  =  wind  speed   observations of ice load D  =  diameter   Q  =  melEng,  sublimaEon   Gives  ice  growth  on  a  0.5  m  long  cylinder   with  30  mm  diameter.   Hans Bergström – Institutionen för geovetenskaper 4Winterwind 2012
  5. 5. Icing measurements – camera and HoloOptics IceMonitor ( rotating cylinder) (O2-Vindkompaniet) WeatherTech Hans Bergström – Institutionen för geovetenskaper 5Winterwind 2012
  6. 6. Measured ice load WeatherTech Hans Bergström – Institutionen för geovetenskaper 6Winterwind 2012
  7. 7. Measured ice load – needs filtering WeatherTech Hans Bergström – Institutionen för geovetenskaper 7Winterwind 2012
  8. 8. Measured ice load – needs filtering – handling of ice drop drop. Thus not at all straightforward to use the icing measurements. WeatherTech Hans Bergström – Institutionen för geovetenskaper 8Winterwind 2012
  9. 9. Models used: ECMWF ERA-interim (~80 km) AROME (1 / 2.5 km) WeatherTech COAMPS (1 / 3 / 9 km) WRF (1 / 3 / 9 km) Hans Bergström – Institutionen för geovetenskaper 9Winterwind 2012
  10. 10. Model domains – AROME: Outer mesh forced by ECMWF Grid resolution: 2.5 x 2.5 km2 Winter season 2009/10 Winter season 2010/11 and 2011/12 WeatherTech Hans Bergström – Institutionen för geovetenskaper 10Winterwind 2012
  11. 11. Model domains – AROME: Outer mesh forced by ECMWF Grid resolution: 1x1 km2 Winter season 2010/11 WeatherTech Hans Bergström – Institutionen för geovetenskaper 11Winterwind 2012
  12. 12. Model domains – COAMPS/WRF: Outer mesh forced by GFS Grid resolutions: 9x9, 3x3, 1x1 km2 Bliekevare/Tåsjö Kiruna/Luongastunturi/Sjisjka Röberg Glötesvålen/Sveg Aapua Brahehus/Nässjö WeatherTech COAMPS: Winter seasons 2009/2010 (all but Röberg), 2010/2011, and 2011/2012. WRF: Sensitivity tests oct-dec 2009. Winter seasons 2010/2011 and 2011/2012. Hans Bergström – Institutionen för geovetenskaper 12Winterwind 2012
  13. 13. Icing modelling - what do we know today? Experience so far from V-313: WeatherTech • Models catch the icing events • But sometimes large differences between modelled and measured ice amounts • Measuring ice loads very difficult Hans Bergström – Institutionen för geovetenskaper 13Winterwind 2012
  14. 14. Icing climatology: We could have done this today – but the results would depend on which model is used and the model resolution. We want to learn more to get more reliable results WeatherTech ... and how shall the icing climatology be done. Two problems to balance: •  The need for period length to get a representative estimate of the climate •  The need for resolution to get a result accurately reproducing the geographical variations Hans Bergström – Institutionen för geovetenskaper, meteorologi 14Winterwind 2012
  15. 15. Icing climatology: What others have done Finish Wind Atlas – 48 months at 2.5 km Finish icing climate – the same as for wind – WeatherTech Kjeller icing atlas – one year at 1 km Our options Two choices: • A long period (30 years) is dynamically modelled using low resolution (9 km?) – the results are then dowscaled to desired resolution (1 km). • A limited number of periods (totally ~1 year) is modelled with high resolution – the periods are chosen to be representative for the icing cimate. Could possibly be based upon Lamb s weather type . Results weighted together to get the climatology. Hans Bergström – Institutionen för geovetenskaper 15Winterwind 2012
  16. 16. Icing climatology: Dynamical model runs for the whole country Example: WeatherTech • AROME with 2.5 km resolution ~ 1 h / day on 48 processors • 1 km resolution 8 times more (2x2x2) 8-12 h • 30 years takes 10-15 years to run! Unrealistic – although computers will have more power after a few years but still........ If chosen using a coarser resolution – need for downscaling. Hans Bergström – Institutionen för geovetenskaper 16Winterwind 2012
  17. 17. Icing climatology: Do we really need 30 years? To illustrate this compare with observed variations of the energy content of the wind. Varies between 70 % and 150 % on an annual basis (red). WeatherTech Even 30 year averages varies with at least ±10 % around a long time average (black, blue). Probably not smaller variability regarding the icing climate. Southern Sweden Denmark Hans Bergström – Institutionen för geovetenskaper 17Winterwind 2012
  18. 18. Icing climatology: Example of icing variability Variations regarding icing climate could be illustrated by the following example: Icing days in November according to ERA Interim data varies between 33 % and 230 % as compared to the 22 year average 1989-2010 average. WeatherTech How representative is a single year? Or 10 or 30 years? Number of icing days in Novermber Hans Bergström – Institutionen för geovetenskaper 18Winterwind 2012
  19. 19. Icing climatology: Downscaling An example using WRF data from 1 month at 9 and 1 km resolutions. WeatherTech 1 km grid Hans Bergström – Institutionen för geovetenskaper 19Winterwind 2012
  20. 20. Icing climatology: Downscaling WRF topography at 9 km and 1 km resolutions. WeatherTech Hans Bergström – Institutionen för geovetenskaper 20Winterwind 2012
  21. 21. Icing climatology: Downscaling WRF mean wind speed at 9 km and 1 km resolutions. WeatherTech Hans Bergström – Institutionen för geovetenskaper 21Winterwind 2012
  22. 22. Icing climatology: Downscaling WRF temperature at 9 km and 1 km resolutions. WeatherTech Hans Bergström – Institutionen för geovetenskaper 22Winterwind 2012
  23. 23. Icing climatology: Downscaling WRF temperature at 9 km and 1 km resolutions. Lifting of 9 km temperature according to difference between 9 km and 1 km topography will give a new 1 km temperature – in reasonable agreement with 1 km model results. WeatherTech Hans Bergström – Institutionen för geovetenskaper 23Winterwind 2012
  24. 24. Icing climatology: Downscaling WRF icing hours at 9 km resolution. WeatherTech Hans Bergström – Institutionen för geovetenskaper 24Winterwind 2012
  25. 25. Icing climatology: Downscaling WRF icing hours at 9 km and 1 km resolutions. WeatherTech Hans Bergström – Institutionen för geovetenskaper 25Winterwind 2012
  26. 26. Icing climatology: Downscaling WRF icing hours at 9 km and 1 km resolutions. Lifting and adding supersaturated water to the liquid water content results in icing hours in reasonable agreement with the 1 km model results. WeatherTech Hans Bergström – Institutionen för geovetenskaper 26Winterwind 2012
  27. 27. Icing climatology: Downscaling possibel to use – we may loose accuracy as regards local terrain response but we gain accuracy as WeatherTech regards climatological representativity. Alternative: Model representative periods wih high resolution? Using some classification to find typical months ... ........ 27Winterwind 2012
  28. 28. Using the Lamb classification •  Used by Professor H. H. Lamb for the WeatherTech British Isles •  Sorts into 27 classes based on direction of flow and pressure (cyclonic or anti- cyclonic) (8 direction types, 2 pressure types, 16 hybrid classes and 1 undefined) 28Winterwind 2012
  29. 29. Examples of Lamb classes WeatherTech 29Winterwind 2012
  30. 30. Using Lamb classification to create an icing climatology •  It has been found that WeatherTech months with similar patterns in Lamb classification do not share the same distribution of temperature Ten of the 27 Lamb classes (left hand side) Solid lines: monthly mean temperature 1989-2008, dashed lines: monthly mean temperature during specific year. Red lines – Pajala, Black lines – Kvikkjokk, Blue lines - Arjeplog (right hand side) 30Winterwind 2012
  31. 31. Using Lamb classification to create an icing climatology •  Data from both SMHI stations and ERA WeatherTech suggests that to create a climatology of icing one needs to take more into account than just the class Distribution of temperature in the SW class in Southern Norrland in February and March (three years). There is a noticeable spread in the temperature distribution 31Winterwind 2012
  32. 32. Using Lamb classification to create an icing climatology Temperature from ERA WeatherTech data compared with temperature series from two measurement sits in Northern Sweden. Temperature tendencies between air mass (above) and stations (below) show similar patterns 32Winterwind 2012
  33. 33. Could finding the most common wind direction for icing be of help? Finding what classification WeatherTech is most common during an icing event could help create an icing climatology. Figures Above: Icing event from the Bliekevare site – black line: measurement. Red and blue lines: modelled ice load. Below: Same as above but from the Aapua site 33Winterwind 2012
  34. 34. Could finding the most common wind direction for icing be of help? WeatherTech The most common wind directions for icing in the Bliekevare (above) and Aapua (below) sites seems to around south. Figures Above: Directions when there is an ice growth in the month of February 2011 Bliekevare site Below: As above just the last week of February 2011 at Aapua site 34Winterwind 2012
  35. 35. How can we move on with Lamb classification? •  A way to move forward? WeatherTech - Using the Lamb classification to find the classes with most icing - Using the geostrophic wind used to calculate classes in combination with temperature (either) from ERA or measurement sites to model icing - Combining the results into a climatology 35Winterwind 2012
  36. 36. Thank you for your attention! Questions? WeatherTech Hans Bergström hans.bergstrom@met.uu.se Petra Thorsson petra.thorsson@geo.uu.se 36Winterwind 2012

×