1. Uncertainty of the wind power estimation using standard meteorological measurements and forecast dataset 1Weidinger, T., 2Kiss, Á., 1Gyöngyösi A. Z., 1Tarjányi, Zs., 3Dobi, I., 3Wantuch, F. and 4Bánfalvi, K. 1Department of Meteorology, and 2Department of Atomic Physics, Eötvös Loránd University, 1117 Budapest Pázmány P. s. 1/A, 3Hungarian Meteorological Service, 1024 Budapest, Kitaibel Pál u.1, and 4Netpoint Bt, 2011 Budaklász, Damjanich u. 25.
Figure 5. Power curves of the 600 kW (E-40) (left) and the 2000 kW (E-70) (right) wind turbines in January 2006 with respect to 10 minute averages.
Figure 1: Wind turbines near Moson- magyaróvár (M). Nominal power are 600 kW (right) and 2 MW (up), respectively. Total nominal power of current equipments is 12.6 MW. (Partly founded by Hungarian and EU projects.)
Figure 2: Annual mean wind speed at 10 m in Hungary based on wind measurements of 60 stations ( ) and 95 stations between (Dobi et al., 2005).
Introduction The first high power wind turbine in Hungary has been implemented at the end of 2000 (Inota, Balaton Highlands, 250 kW). The current total wind power of Hungary is nearly 20 MW. 3.6% of the domestic energy production (~1600GWh) is covered by renewable energy at Western Transdanubia is the windiest part of the country (~3,5 m/s). Low cut-in wind speed turbines suitable for continental conditions represent an economic way of electrical power production in Hungary too, in case of appropriate government grants (Bartholy et al., 2003) (Figure 1). The windiest part of the country is the Small Hungarian Plain adjacent to Burgenland in Austria, where a wind energy system of nearly 150 MW total power is in a stage close to realization and another 280 MW under authorization. One major obstacle to the development is the Hungarian power supply network. It is suitable to receive only 330 MW on the next few years. Even above 200 MW -- due to the variation in wind speed -- significant network deviation are to be expected in the next couple of years.
Figure 3: Annual mean wind speed at 50 m in Hungary according to wind measurements of 55 automated weather stations for the period (Dobi et al., 2005)
Figure 6. Correlation between 10 m and 65 m wind speed (with respect to 10 m average (left) and peak (right) wind speed) for January 2005
The inter-annual variability of wind speed is significant. For instance, in the period we get even larger deviations (Table I) compared to the long term average (3 decades). At Kékestető the mean wind speed was only 3.42 m/s, at Siófok m/s, but contrarily in Pápa it was significantly larger (2.93 m/s) than the long term average. Note that year 2001 was especially windy, while 2003 was specially calm. Recorded annual mean wind speed at Siófok in 2001 and 2003 were 3.31 and 2.27 m/s, respectively. Note that in the data series of Budapest and Győr no such significant variability is present perhaps due to urban effects. Additional uncertainty arises from local effects (land cover, orography, etc). Relocation of a certain station also causes significant changes in the recorded mean wind speed. For instance at Sopron the new location of the station is a former wind mill location from Here annual mean increased by more than 1 m/s. Another relocation in the 1990’s resulted in significant modification in the recorded mean again (currently ~3.6 m/s). Due to inter-annual variability of wind speed, wind power production of a windy year can be as high as the twice of a calm year production (Weidinger, et al., 2005).
Figure 4 demonstrates the measured and interpolated wind speed and wind energy at 10 m and at m, respectively. An accuracy of 0.1 in the value of the p exponent results in an uncertainty of 21% at 80 m. Profile fitting can be performed with larger accuracy with the use of wind tower measurements (usually m in Hungary).
4. Joint analysis of meteorological measurements, wind energy data and numerical model calculations
Wind speed and power production data of different hub height wind turbines, 10 minute measurements of the meteorological observation network and wind data output of numerical weather prediction models give the opportunity to perform new generation studies. Relation between the wind speed measured on the turbine and the power production can be analyzed. Correlation of 10 minute mean and maximum wind speed can be calculated. Optimal wind profile exponent can be estimated. The relation of modeled wind profile and measured 10 m wind to wind measurements on the wind turbine is crucial. Its variation with respect to stability, wind speed and direction and macro-synoptic situation is to be analyzed. The aim of these studies is the development of a prediction algorithm for the design of wind turbines and wind farms, and for the smooth operation of power supply network. This work is in a preliminary stage.
Figure 8. Correlation between 65 m and 115 m wind speed (left panel). Measured and calculated power production of the E-70 wind turbine at 115 m – assuming power law profile (p =0,39) – January 2006.
2. Wind conditions in Hungary, inter-annual variability of wind speed
Annual mean wind speed in Hungary is 2-4 m/s at 10 m above surface (Figure 2). Maximum values occur in spring, minima are in early fall. Inter-annual variability is the nature of our climate and is not a climate change issue. However, wind maps for different periods or those based on different interpolation techniques show significant differences ( m/s or more) though their structures are similar (Tar et al., 2005, Weidinger et al., 2005). Extreme wind speed events were recorded on the western frontier (Sopron) and at the top of Hungary (Kékestető; elevation 1010 m). The deviation in the annual mean wind speed from the long range average can be around 10%. This value is around 15% in the wind tunnel of the Buda hills at Budaőrs (Near Budapest) and 20% at the shore of the Lake Balaton (Siófok) and at the Kékestető.
3. Uncertainty of the Power Law Wind Profile Estimation
Two common methods for the estimation of wind profile based on standard (10 m) meteorological measurements include (i) the Monin -- Obukhov similarity theory and (ii) the power law profile approximation. Similarity theory however has large error above the constant flux surface layer (10-20 m in case of nocturnal stable stratification) (Cost Action 710, 1998, Weidinger et. al, 2000). The scope of this section is the analysis of the power law estimation. The independent variables of the power law wind profile formula are the wind speed (Ur) of the reference level (zr) and the stability dependent exponent (p) (Irwin, 1979). The numerical value of the exponent is 1/7 over homogeneous terrain with short vegetation, which is the most common first guess. Dependence on stability (e.g., Pasquill categories) can be assumed through the daily variation of the exponent p. In the daytime convective surface layer its value ranges between 0.07 and 0.1, while by extreme stable stratification p = is suggested. To quantify this large uncertainty, we take the first derivative of the power law formula U(zr) = Ur(z/zr)p with respect to the exponent (p):
4.1 The database
Figure 9. Correlation between measured and calculated wind speed at 65 m and 115 m calculation made with the NCEP ETA model 48 hours lead time (2006. január 15-16).
The continuously developing database consists of 3 different parts: (i) 10 minute meteorological data of the weather station at Mosonmagyaróvár including global radiation from 1. January, (ii) Wind power and wind speed data every 10 minutes of a 65 m high wind turbine 2 km from the station (Enercon E-40, 600 kW) from 1 January 2005, and the same dataset of a 115 m high wind turbine at the same location (Enercon E-70, 2000 kW) from 1 January, (iii) Calculated wind profiles of three different numerical weather prediction models: ALADIN and MM5 model output of the Hungarian Meteorological Service and the ETA output of the Department of Meteorology at Eötvös Loránd University from 1 January, Hourly wind forecast for 48 hours ahead on MM5 and ALADIN models are under development at Hungarian Meteorological Service.
5. Conclusion
Wind energy projects in Hungary can only be economically efficient with government grants currently. A nature of the wind climate in the Carpathian Basin is the large inter-annual variability. Available wind energy of a windy year can be twice of a calm year. Power production and wind data of current wind turbines along with standard meteorological observations and numerical model outputs give an opportunity to predict wind power. This poster reported the first steps of this developement.
Table I. Average wind speed (U) for the period , deviation (%) of wind speed and wind energy of windiest (Umax) and calmest (Umin) years calculated from monthly windroses. (Due to relocation of stations, period begins at *1975 and **1977, respectively).
Station
(See Fig. 2)
U [m/s]
(Umin-U)/U
(Umax-U)/U
(U3min-U3)/U3
(U3max-U3)/U3
Győr (1)
2.70
-11.8%
14.3%
-40.3%
31.5%
Sopron**(2)
4.38
-8.9%
6.7%
-21.6%
13.1%
Szombathely (3)
3.39
-14.1%
14.4%
-33.9%
82.3%
Pápa** (4)
2.56
-11.1%
8.0%
-30.6%
26.3%
Siófok (5)
3.17
-21.0%
23.4%
-42.8%
58.8%
Budaőrs** (6)
2.88
-14.7%
16.5%
-34.8%
58.9%
Budapest (7)
2.87
-15.9%
19.9%
-46.9%
72.7%
Szeged (8)
3.32
-6.5%
9.6%
23.5%
Kékestető (9)
4.40
-23.1%
12.8%
-55.2%
35.5%
In addition, sensitivity of the wind energy (which is proportional to the cube of wind speed) is three times larger.
References Bartholy, J., Radics, K., Bohoczky, F., 2003: Present state of wind energy utilization in Hungary: policy, wind climate, and modelling studies Renew. Sustain. Energy Rev. 7, COST Action 710 – Final report. Harmonization of the pre-processing of meteorological data for atmospheric dispersion models. EUR EN, (Edited by Fisher, B.E.A., Erbrink, J. J., Finardi, S., Jeannet, P., Joffre, S., Morselli, M. G., Pechinger, U., Seibert, P. and Thomson, D. J.). Dobi I, Konkolyné Bihari Z., Szentimrey T., Szépszó G., 2005: Wind maps for Hungary. In: Wind energy in Hungary, ) (In Hungarian) Irwin, J. S., 1979: A theoretical variation of the wind profile power-law exponent as a function of surface roughness and stability. Atmos. Env. 13, Tar, K, Radics, K., Bartholy, J., Dobi, I, 2005: Wind energy in Hungary. Magyar Tudomány, (In Hungarian). Weidinger, T., Pinto, J., Horváth, L., 2000: Effects of uncertainties in universal functions, roughness length, and displacement height on the calculation of surface layer fluxes. Meteorologische Zeitschrift, Vol 9, No Weidinger, T., Kiss, Á., Gyöngyösi, A. Z., Krassován, K. and Papp, B., Uncertainty of available wind energy estimation based on long term meteorological measurements in Hungary. Euromech Colloquium 464b. Wind Energy, October 4-7, 2005 University of Oldenburg, Germany, Book of abstracts, p35.
4.2 Power production in the light of meteorological datasets
Results of 2005 and 2006 January are being analyzed hereby. As the power production (Figure 5) estimated by wind data measured on the turbine has an accuracy less than 5% error, it is sufficient to focus on the comparison of measured and calculated wind speed. The exponent for the extrapolation of 10 minute average is different (larger) from the one for the extrapolation of the peak wind speed. (Figure 6). Exponent in the layer m is significantly different from one in the layer m – regression equation of the wind speed values are different. Standard deviation of power curves based on the estimated wind speed is large (Figure 7). Measured and calculated wind speeds of the ETA model has similar structure (Figure 8).
Figure 4. Relative wind speed (left) and wind energy (right) at different levels compared to the 10 m value as a function of the p exponent.