Você está na página 1de 5

WIND TURBINE PERFORMANCE GUARANTEE J P Coelingh, Q W M Sluijs, Ecofys bv P.O. Box 8408, NL-3508 RK Utrecht, The Netherlands tel.

++31-30-2808.300, fax ++31-30-2808.301 email: J.Coelingh@Ecofys.nl, website: www.ecofys.com 1 Introduction It is evident that in the wind energy business as an economic activity there is a close relationship between the wind speed and the revenues. In several phases of the lifetime of a wind project this relationship has different application forms. Although there is a distinct common denominator, these forms have different purposes. Here we will give a short overview of several of these forms (most of which are quite familiar). In addition, a new and fairly simple instrument will be presented that can be used in the operational phase of a wind farm: the wind turbine performance guarantee. 2 Wind speed Power output The table below summarises different forms of the relationship between wind speed and power production of a wind turbine. In each phase of the development of a project a different application of this relationship is called for. Phase Preliminary development stage Final development stage Source of wind data Generic meteorological data Application YIELD PREDICTION Calculation with suitable model (WAsP, Windfarmer) YIELD PREDICTION Calculation with suitable model according to Measure-CorrelatePredict PV-CURVE MEASUREMENT POWER PREDICTION Calculation of energy programme using suitable model (Prediktor, WPPT etc.) WIND DERIVATIVE WIND TURBINE PERFORMANCE GUARANTEE Goal Preliminary feasibility study/ comparison of sites (reporting) Investment decision/ project valuation (measurements+reporting)

Short measuring campaign (mast/SODAR)

Operational stage

IEC compatible mast measurements Predicted data from meteorological model

Check on wind turbine performance (product guarantee) Decrease of unbalance penalty/increase grey value (wind power predictions services)

Generic meteorological data Generic meteorological data

reducing fluctuations in cash flow Check on power output performance (performance guarantee)

2.1 Yield prediction (1) In first stage of the development of a wind project the yield of the proposed wind turbines is of interest to decide on further steps to be taken. With little investment and relatively little effort a calculation can be performed, using available information and methods. This type of calculation always has the following elements: generic meteorological data o representative long-term wind speed records technical specifications of wind turbines o Pv-curve o Ct-curve o Position and hub height of wind turbines terrain information o roughness o orography o obstacles meteorological model o based on boundary-layer meteorology o examples: WAsP, WindFarmer It is up to the experience of the user to select and prepare the necessary input, apply the model and interpret and present the output in a suitable way.

2.2 Yield prediction (2) In a more advanced stage of development a further insight in the expected yield may be considered necessary. This depends on whether or not the previous step is judged to be sufficient, e.g. in specific conditions like complex terrain, mountainous areas or offshore sites. Here data maybe scarce, and existing modelling tools insufficient. Therefore, in addition to the previous step a dedicated wind measurement campaign at the proposed site may be undertaken during several months up to 1 or more years. The extra costs are warranted by a reduction in uncertainty in the outcome. On the one hand this may avoid the investment in a project that is not profitable, on the other hand it may lead to more favourable financing conditions since the risks are being reduced. The method is called Measure-Correlate-Predict and is well known. Data measured in a short-term onsite measurement campaign are correlated to a suitable longterm reference data set, in most cases from a meteorological station or an airport. After establishing the justification of the correlation, the short term can be interpreted towards the long term. 2.3 Pv-curve measurement A specific type of measurement campaign in the operational phase of a wind farm is to measure the Pv-curve of a wind turbine. In general, wind turbine suppliers guarantee their wind turbines in two regards: availability Pv-curve A guarantee on availability is fairly easy to apply: it is counting the hours that the wind turbine is actually able to generate. To check the Pv-curve, however, is far more complicated as described in detail in the IEC-norm 61400-12. A specific Pv-curve measurement just to check the specification is fairly expensive, especially with increasing hub heights. Moreover, the Pv-curve of only 1 wind turbine (sometimes 2) can be measured by a single mast, which still does not say anything definite about other wind turbines. For this reason the concept of the Wind Turbine Performance Guarantee has been developed (see further below). 2.4 Wind power prediction Power production by wind turbines varies due to the wind speed changes. However, wind turbines are connected to the grid, which has to accommodate the power variations. The grid operator has to maintain the balance between production and consumption at all times. Therefore in various countries systems have been implemented which make use of power output predictions. These predictions are always based on meteorological input: predicted wind speeds as follow from mesoscale models like HIRLAM. The wind speeds are then translated into power predictions by use of a model. Such models are either physical (representing the physical processes as well as possible) or statistical (making use of neural networks/fuzzy logic), or a combination of both. The actual performance of such models very much depends on the wind data and also on the wind farm portfolio (type and number of wind turbines, geographical distribution, climate characteristics). 2.5 Wind derivative A fairly new concept that is under consideration in the wind energy business is that of the wind derivative. In principle it is an insurance that pays out in years with low average wind speeds when the revenues are below average. Over several years there will be a net cost paid to the insurer, but the cash flow is more steady with less (negative) peaks. This increases the certainty of income, decreases the financial risk and therefore leads to more favourable financing conditions. 3 Wind turbine performance guarantee

For a (prospective) wind farm developer closing a contract with a wind turbine supplier is an important issue. One of the main topics for discussion is the quality guarantee of the wind turbines (over the life-time). In general, a product guarantee is given by the manufacturer, in the form of a guaranteed Pv-curve (and a guaranteed availability). The Pvcurve can be checked by a measurement according to the IEC norm 61400-12. Here a method is proposed to guarantee the performance of a wind turbine (or wind farm), while avoiding a measurement mast (expensive due to increasing hub heights). 3.1 Principle Starting point of the new method is that each year the following calculation is being made: Qyear = Plong-term Byear Windexyear Here Qyear is the expected annual production of a specific year; Plong-term is the guaranteed annual production (longterm average); Windexyear the (relative) number that represents the climatological variation in the production of a specific year; and Byear the average availability (with respect to external factors) of a specific year.

The basis of the calculation is that the wind turbine supplier guarantees a certain mean annual production (Plong-term). Then for each year corrections are applied for two external factors, for which the wind turbine supplier cannot be held responsible. The first is the non-availability due to reasons not attributed to the supplier; the second is the annual climatological variation. Here the operational details of the method have been formulated as summarised in the following table: Feasibility phase Tendering phase Calculation of expected yield in order to determine stand point for negotiations Call: formulation of contract terms Review: comparison of offers Contracting: closing the contract Annual determination Windexyear, Byear, Qyear; determination difference for bonus system

Exploitation phase

The possible yield of the proposed wind farm is estimated by calculation (in advance) of the different options. On the one hand these calculations are being used in the negotiations to determine a position with respect to the guaranteed annual production Plong-term. On the other hand the same calculation is repeated annually, in order to allow for the climatological variations in Windexyear. Finally the realised production Qyear and the availability excluding factors not under responsibility of the supplier (Byear) are being determined on an annual basis. Using this information the annual performance can be evaluated, for which a bonus system is in place. In this way a method has been formulated that meets the wishes of project developers with respect to the guarantees of the suppliers with relatively little effort (in any case compared to the alternative of installing and operating a high meteorological tower). 3.2 Implementation The suggested method consists of the following steps: select a suitable meteorological station representative for the proposed site; make a (global) WAsP-calculation for the location with a long-term (historical) data set, and with the actual Pvcurve and wind farm configuration; this is the long-term reference yield; Then the following steps are being executed every year: make the same WAsP-calculation for a single year; divide the one year yield by the reference yield and consider this to be Windexyear Note that every additional year may have influence on the reference value (particularly for an extreme year and/or if the reference period is short); furthermore, no major changes should occur in other external conditions (like the erection of another wind farm nearby or another change in terrain conditions). To illustrate the sensitivity and the accuracy of a method based on WAsP a number of calculations were done for a fictitious wind farm. Here the following starting points have been used: the wind farm consists of 9 wind turbines and is situated near the Dutch coast; wind speed data and a roughness map of KNMI (Royal Dutch Meteorological Institute) have been used; standard yield calculations have been executed with WAsP. In the calculations the absolute values of the results are not important, but the relative differences. The results are presented in the next paragraphs. 3.3 Analysis sensitivity/reliability

3.3.1 Accuracy Looking at the results of Figure 1 it can be seen that the values of Windexyear do not deviate more than 20% of the long-term average that has been set to 100% by definition. The accuracy of Windexyear is estimated at 3%. Because of the fact that the numerator and the denominator have been calculated similarly, possible systematic errors in the calculations are assumed to be eliminated for a large part. Therefore the value of Windexyear is fairly accurate. 3.3.2 Annual variations The mean wind speed at a site varies from year to year. These variations have been illustrated in Figure 1, where for two different wind regimes the yields have been calculated and presented relatively. Over a period of 10 years the mean has been set to 100%, the variations range from 15% to +20%. The wind climate itself is also of influence, due to the alinear effects of the Pv-curve. In fact the figure shows the value of Windexyear for the years 19932002.

130%

120%

110%

100%

U = 6,4 m/s U = 7,3 m/s

90%

80%

70% 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002

F i gu r e 1 :

Wi n d e x o f t h e fi c t i t i o u s w i n d fa r m fo r t wo me a n w i n d s p e e d s ( 1 0 0 % = a v e r a ge o f 19932002).

3.3.3 Monthly variations The variations in the yield over the months of the year are sizable. In general the wind speed is high in the winter months, and low in summer. For the year 2002 this has been illustrated as an example in Figure 2. The figure shows that in this case the shortest month (February) contributes about 20% to the annual total, while the summer months are ideally suited for planned maintenance without much loss of revenue.
25%

20%

15% U = 6,4 m/s U = 7,3 m/s 10%

5%

0% J F M A M J J A S O N D

F i gu r e 2 :

R e l a t i v e mo n t h l y yi e l d s o f a f i c t i t i o u s wi n d f a r m fo r t wo w i n d c l i ma t e s ( 1 0 0 % = annual total 2002).

3.3.4 Variations within a wind farm Wind farms of large modern wind farms occupy areas of several square kilometres (a typical figure is 8 to 10 MW/km2). This means that the mean wind speed may vary over such a surface as a result of differences in terrain conditions. Specifically in regions with buildings or trees, or situated close to the coast, wind speeds may differ considerably over short distances, leading to differences of yield per wind turbine.

Moreover, within a wind farm the wind flow approaching the wind turbine may be disturbed as a result of the presence of the other wind turbines, the so-called wake or park effect. The position in relation to the other wind turbines and the wind direction determines the extent of the effect for each wind turbine. In general, the wind turbine with the best position towards the prevailing wind direction will experience the least loss due to the wake effect. Depending on the local conditions the differences may be in the order of several percent. This is illustrated in Figure 3. Here the minimum yield divided by the mean is depicted in blue, while the maximum yield divided by the mean is depicted in red.
115%

110%

105%

100%

Minimum/Mean Maximum/Mean

95%

90%

85%
J F M A M J J A S O N D Annual mean

F i gu r e 3 :

R e l a t i v e yi e l d s p e r wi n d t u r b i n e p e r mo n t h fo r 2 0 0 2 .

This figure shows that the difference in annual mean between the best and the worst producing wind turbines is 5% (98% versus 103%). On a monthly basis these differences can be much larger (see e.g. August). 3.4 Guaranteed annual production The determination of the guaranteed annual production is a process of negotiation between the project developer and the wind turbine supplier. The outcome will be dependent on the conditions in the contract with respect to meeting the targets set (bonus/malus). For both parties it is of importance to enter the negotiations well prepared. Therefore it is obvious to define a starting position for the negotiations based on a (independent) yield prediction. The starting position has to be determined in detail with financial projections. 3.5 Annual settlement For the determination of the annual value Windexyear a long-term reference yield prediction is necessary. In the following years it is important to perform the annual calculations in a similar fashion, since Windexyear is a relative number, where numerator and denominator are compatible quantities. In this phase the annual settlement is fairly simple. After the close of the year the next quantities need to be determined: Qyear, the realised annual production (meter reading); Windexyear, to be calculated by (independent) third party on the basis of wind speed data of the year Byear, the non-availability not to be attributed to the wind turbine supplier With these figures the expected annual production is calculated and compared to the guaranteed annual production. The appropriate actions to be taken follow directly from the contract. 4 Conclusions A short overview has been presented of the different forms and applications of the relationship between wind speed and revenues from a wind turbine. The concept of the wind turbine performance guarantee has been introduced and explained. It is a method for a wind farm developer and a wind turbine supplier to enter a contract where there is a mutual benefit for both to generate as much kilowatthours as possible within the conditions set by the wind.

Você também pode gostar