Application of nowcasting to reduce the impact of irradiance ramps on PV power plants

Kurzfassung

In this Master thesis project, a simulation model is developed to study the potential of short-term irradiance forecasts (nowcasts) in output power smoothing. Hereby, a photovoltaic power plant with a battery energy storage system is simulated. Emphasis is put on the impact of irradiance ramps and the analysis of the nowcast uncertainty. The ongoing integration of variable renewable energy sources into the electrical grids challenges grid operation, e.g. more significant power fluctuations impact the energy balance and energy peaks. One way to reduce fluctuations in power generation is to demand power smoothing to reduce the fluctuations to a maximum. Ramp rate constraints for photovoltaic power plants can be met by introducing a nowcasting system, but the uncertainty of the nowcast is crucial for successful integration. Previous studies have shown that ideal nowcasts (i.e. without uncertainties) can substitute battery energy storage systems completely for the task of output power smoothing. Studying the ability of the nowcasting system to predict irradiance ramps and the corresponding control strategies’ ability to balance output power smoothing and minimize curtailment losses is of interest. Further, the ability of nowcasts with realistic uncertainties to eliminate the necessity of a battery energy storage system for power smoothing is assessed, since economic benefits are expected. To study the research question, I develop a photovoltaic power plant model of an existing power plant in northern Germany. The nowcasting system with a network of sky cameras provides the irradiance data for a selection of days. The selected 18 days in the year 2020 represent a variety of irradiance conditions but are not representative of the irradiance of a complete year at the study site. I validate the accuracy of the model and simulate a benchmark scenario with the validation data from the photovoltaic power plant. The benchmark includes the production data and a battery storage system for output power smoothing without any nowcasting. To this benchmark, I compare the nowcasting strategies to estimate the potential reduction of required battery power and capacity. The chosen simulation approach is common practice in this research field to estimate the potential of nowcasting. By following the approach, the obtained results show the capabilities and limitations of the described nowcasting system and control strategy. The model can simulate the output power of the photovoltaic power plant up to a certain precision. The control strategy is not able to avoid all ramp rate violations without a battery system due to uncertainties of the nowcasts, but it can reduce the number of ramp rate violations by 81.3%, the maximum ramp depth by 47.4%, and the mean ramp depth by 2.2%. Further, the required battery capacity can be reduced by 79.7% and the maximum battery power by 48.3%. The resulting curtailment losses are estimated to be 12.5% compared to the benchmark scenario. As a consequence, the installation of a state-of-the-art nowcasting system can reduce the number and extent of ramp rate violations and thus reduce the strain on the battery energy storage system, which can be of significantly smaller dimension in capacity and power. This could lead to an increased lifetime of the battery. Whether the displayed technical potential provides an economically viable case was not investigated.