The inclusion of renewable energy into utilities' generation portfolios has become a highly desirable evolution. For example, in the State of Oregon, the renewable portfolio target is to achieve 25% renewables generation by the year 2025 []. As a consequence of renewable energy targets across the world, wind power is growing at a significant rate. For example, the BPA load management region extending mainly through Washington and Oregon has average load of 10GW with approximately 2.5 GW wind installed, and an approximate additional 5 GW likely within the next 5 years []. The addition of such a large variable generation will strain the stability of the grid and require increased spinning reserve and cycling of thermal plants [, , ]. The variable nature of the resource is illustrated by Fig. 1, which shows the variation of the power output of a large wind farm in Northern Oregon over one day. The wind farm power data sample points (in blue) are 10 minutes apart and are calculated as the average over those 10 minutes. The baseline forecasting (red line) algorithm is a simple one hour persistence model.
Fig. 1. Example daily energy from wind farm. Blue dots: actual wind farm output; red line: forecast energy output.
To determine the energy storage requirements for such a system, analysis of year long data sets were carry out to determine the economic sizing of a battery system capable of mitigating differences in forecast power greater than 0.04 per unit for 90%of the events occurring in any given year. This analysis suggests that incorporation of about 12% of rated power capacity and about 0.5 hours at full rated power output is necessary to achieve adequate system performance. To test this model, an in-lab research grid was designed and constructed, supported by a 480V, 750 kVA dedicated utility supply. The in-lab grid features emulation of several high-power grid sources and loads, including a wind farm, energy storage system, hydro resources, and local loads. The energy storage system is a 25 kW, 50 kWh Zinc Bromine flow battery. The wind farm is emulated using an Arbitrary Waveform Generator (AWG), which functions as a 120 kVA externally controlled source. Experimental validation of the wind farm sizing is currently under way as is verification of battery system lifetime performance assumptions.
This presentation focuses on the problems posed by present energy forecasting assumptions, the sizing of an energy storage system the capable of mitigating the resource fluctuation using technical and economic constraints, and the development of a lab scale grid for experimental evaluation of the energy storage model developed.
[] Summary of Oregon's Renewable Portfolio Standard, Oregon Department of Energy 2007. http://www.oregon.gov/PUC/Oregon_RPS_Summary_Oct2007.pdf accessed 3/09.
[] “Northwest Wind Integration Action Plan,” Northwest Power and Conservation Council, http://www.nwcouncil.org/energy/Wind/library/2007-1.htm
[] “Utility Wind Integration State of the Art,” Utility Wind Integration Group, http://www.uwig.org/UWIGWindIntegration052006.pdf
[] M. Milligan, “Wind Integration Cost and Ancillary Service Impacts,” NREL Technical Report, http://apps1.eere.energy.gov/tribalenergy/pdfs/course wind milligan1.pdf