Ähnlich wie Long-term estimates and variability of production losses in icing climates Stefan Söderberg, Magnus Baltscheffsky, WeatherTech Scandinavia (20)
2. Introduction (I)
Icing on structures has for many years been an important
factor to take into account when planning infrastructures such
as power lines in cold climate.
More recently, icing has also become a major issue for the
wind power industry. Today many wind farms are planned in
areas where icing conditions frequently can be expected.
For successful project management and for wind farms in
operation it is vital to learn about the expected production
losses. However, today climate data for icing conditions is
missing.
WeatherTech
3. Introduction (II)
The objective of this study is to investigate the long-
term variation in icing climate and production losses.
To do this we need:
• Weather data with high enough spatial resolution
and a long enough reference period.
• A method to estimate ice load
• A method to estimate production losses due to
icing
WeatherTech
4. Tools (I)
Numerical Weather Prediction Model
Necessary meteorological data was produced by the mesoscale
numerical weather prediction model WRF (www.wrf-model.org).
Initial and lateral boundary conditions were provided by NCEP/NCAR
Reanalysis data. WRF was used to produce hourly data in two
different model set ups:
• A one year time series (20100501-20110431) on a 1x1 km2 model
grid covering wind farms in northern and southern Sweden.
• A 30+ year time series on a 9x9 km2 model grid covering
Scandinavia, Finland, the Baltic countries, northern Poland, and
northern Germany.
WeatherTech
5. Tools (II)
1982
Long-time reference – 9 km resolution
Illustration of the variability in the wind
speed climate. Map showing % of long-
term mean value.
The estimated ice load is not only
dependent on the modelled liquid cloud
water but also on the wind speed. Hence,
one can expect an annual variation in ice
load/production loss similar to or even
larger than what is found in the wind
speed.
Please contact WeatherTech for
access to the animation
WeatherTech
6. Tools (III)
Ice accreation model Assume a rotating cylinder
A modified version of the “Makkonen Growth
model” was used to estimate the ice load: – α1collision efficiency
dM – α2sticking efficiency
= α1α 2α 3 w ∗ A ∗ V − melt
dt – α3accretion efficiency
– wAV water flux
Melting
– energy balance
dM/dt = F(wind, temperature,
pressure, LWC, droplet size
distribution)
WeatherTech
7. Tools (IV)
Production loss estimate - 3D power curve
Modified power curves has been constructed by
using wind farm production and ice load
Power
measurements.
These power curves were then combined in a
3D power curve which give the power production
as a function of wind speed and ice load. Ice load
WeatherTech
8. Results (I)
Example of icing conditions (1 km)
At this particular site in northern Sweden,
icing conditions in 2010/2011 were most
frequent in winds from SSE and S.
A number of events with heavy icing can be
seen in the ice load time series.
WeatherTech
11. Results (IV)
Long-term correction
Nevertheless, over a
winter season we find a
good agreement in
accumulated ice load
and accumulated hours
with active icing (icing
intensity >10g/h/m) when
comparing 1x1 km2 and
the adjusted 9x9 km2
model results.
WeatherTech
12. Results (V)
Distribution of number of active icing hours
Moving from a one year perspective to a 30 year period, the studied site also
display a large number of hours with active icing in the WNW and NW sectors
WeatherTech
13. Results (VI)
Accumulated monthly production and production losses over 30 years
For this site, the production losses are most pronounced in southerly wind regimes. But,
over a 30-year period substantial production losses are also found in other wind sectors.
WeatherTech
14. Results (VII)
Accumulated yearly
production and
production losses over
30 years
Looking at accumulated
yearly values instead of
monthly values reveal a
year to year variability in
the wind direction
dependence.
WeatherTech
15. Results (VIII)
Yearly deviations
from long-term mean
As expected, there is a
substantial yearly
variation in production
and production losses.
Production losses in
individual years can be
twice as large as the
long term mean.
WeatherTech
16. Discussion
The long-term variation in icing climate and production losses have been studied
using model output from WRF. A one year time series from a 1x1 km2 model grid were
combined with a 30 year time series from a 9x9 km2 model grid.
Among the findings are:
• Combining a high resolution model simulation with a coarser reference time series
is a promising method to investigate long-term variations in production losses.
• For the investigated site, production losses are most pronounced in southerly wind
regimes. But, a year to year variability is found in the wind direction dependence.
• A substantial yearly variation in production and production losses were found.
Production losses in individual years can be twice as large as the long term mean.
WeatherTech
17. Contact info
Long-term estimates and variability of production losses in icing climates
Authors:
Stefan Söderberg and Magnus Baltscheffsky
WeatherTech Scandinavia AB
Uppsala Science Park
SE-751 83, Uppsala.
stefan.soderberg@weathertech.se; Phn: +46 (0)70-3932260
magnus@weathertech.se; Phn: +46 (0)70-8631963
WeatherTech