Methods for long-term correction in a post-construction energy yield analysis

Introduction

Post-construction energy yield analyses are useful for re-financing and when a wind farm is going to be sold. The total uncertainty in a post-construction yield analysis is typically half of that in a pre-construction yield analysis. The long-term adjustment of the measured power is one of the major tasks in the post-construction yield analysis. There are different methods available to do the long-term adjustment and the most suitable method depends on the quality of the operational data. The results based on four methods from two wind farms have been compared.

Approach

The used methods can be divided into two different categories, “Historical power curve methods” (HPCM) and “Index methods” (IM). HPCM are using turbine specific power curves derived from measured power. The IM are relating the measured power to either a wind index or a production index.

Main body of abstract

The wind used when deriving power curves in the HPCM can come from the nacelle anemometer or from a numerical weather prediction (NWP) model. When the nacelle wind speed is used, it must be related to a long-term reference wind speed series. The long-term reference is adjusted based on the found relationship and is then applied on the turbine specific power curves to create a long production time series. The annual energy production (and other statistics) can easily be calculated based on the created time series. If a modeled wind speed is used instead of the measured wind speed from the nacelle, no wake effects are included in the wind material and power curves must be derived sector wise. This in turn requires a rather long period with operational data in order for the power curves to be properly defined. The long-term model data is used with the sector wise power curves to create production time series.
In the IM a monthly or weekly index period can be used. Which is more suitable depends on the data availability in the operational data set. The used availability threshold is having a significant effect on the AEP estimations. A monthly or weekly availability requirement of >95% is used to determine if the month/week should be included or not. The selected months are then adjusted to be equivalent to 100% availability. If it is more favorable to use a weekly index compared to a monthly index is determined of how the available data is distributed in the operational data. A month with low overall availability will be discarded, but in some cases it is however possible to extract one or more weeks with enough data from that month.


Conclusion

HPCM based on measured wind speed is suitable if the nacelle anemometer is consistent during the operational period and also for short (six months to one year) operational periods provided that the turbines are operating in full-performance during the majority of the time. Full-performance is defined as “The WTGS is operative and generating according to design specifications with no technical restrictions or limitations which affect generation.” [IEC61400-26-1]. HPCM with modeled wind speed is suitable to use if the nacelle anemometer is inconsistent, making it difficult to create valid turbine specific power curves. But using modeled wind speed require longer operational period to properly define sector wise power curves The index methods are typically good to use if the nacelle anemometer is inconsistent. And by using both weekly and monthly index periods, it is possible to get additional estimates of the annual energy production (AEP). The spread in the results represents the method uncertainty.


Learning objectives
It is important to select the most proper method(s) in a post-construction energy yield analysis. The uncertainty in the long-term correction depends on the length of the operational period and by the used methods. By calculating AEP for different operational periods using several methods it is possible to estimate this uncertainty.