Reduction of the uncertainty of wind power predictions using energy storage

AutoresBludszuweit, Hans
Año publicación2009
Categorías
CódigoCP-0583

 

Resumen
The central subject of this thesis is to find out, what is the required size of an energy storage system (ESS) which is able to compensate completely or partially wind power forecast errors and to estimate its cost. Three probabilistic instruments are developed to enable the proposed ESS sizing method. The first one is a persistence model which permits the generation of large time series of forecast data from measured series of wind power. This model is verified with real world wind power forecast data. Secondly, probability density functions (pdf) of forecast errors are estimated with an algorithm based on the Beta distribution. In this context, an existing model is refined, thanks to the large quantity of persistence forecast data available. As a third tool, an online bias correction method based on moving averages is developed, which guarantees that forecast errors have zero mean.

Unserved energy is chosen as sizing objective parameter. It is defined as the cumulative energy of uncompensated forecast errors. This parameter is an indicator of the grade of partial compensation if ESS size is reduced. The reference ESS size is derived from the case where unserved energy becomes zero. Starting from this reference, it is possible to obtain a relationship between the grade of reduction and unserved energy. It is shown that forecast error pdf can be used directly to determine ESS power rating and associated unserved energy. The process for determining ESS energy capacity is more complex. Saturation times are estimated as a function of ESS capacity reduction and unserved energy is calculated using the energy throughput ratio. A case study with real world forecast data shows the importance of bias correction for ESS sizing. A life-cycle cost model is developed which allows the comparison of different ESS technologies. Regions of ESS sizes for minimum costs are identified for five different technologies. Lowest costs are obtained with pumped hydro and highest with lead-acid and flow batteries. Further, large cost reduction potentials are detected for flow batteries and hydrogen storage. Despite its high power-related costs and low efficiency, hydrogen storage appears to be an economically interesting option for the case studied here. Finally, ESS costs are put into a market context. It is assumed that wind energy is sold on a liberalised market and deviations are compensated on the regulation market. In particular, the Spanish electricity market is analysed. As a result of the analysis, daily profiles of regulation costs are identified. Deviation costs are obtained, introducing these profiles in a probabilistic bidding strategy. It is shown the importance of having a good regulation price forecast method with this type of strategies. The results presented in this work lead to the conclusion that the installation of large storage capacities is needed to compensate forecast errors, which cannot be justified by costs derived from forecast deviations. Nevertheless, expected ESS cost reductions combined with added value of storage capacity for the power system may lead to profitable solutions in the near future.