Regional photovoltaic power fluctuations within frequency regulation control time frames: A study with high‐resolution data

The growth of renewable energy penetration on power grids is reaching high levels in many countries around the world. In such high‐penetration scenarios, the proper characterisation of the intermittency of renewable power generation becomes important to the operation of power grids. In this study, using the most recent high‐resolution global horizontal irradiance data in Japan, we characterise short‐term fluctuations of photovoltaic (PV) power in regional scale, in a scenario of high penetration of PV. We analysed 1 year of data with a resolution of 10 seconds for the Kanto area in Japan. The results show that the distribution of the short‐term fluctuations is non‐Gaussian, presenting a shape similar to a Laplacian distribution. In addition to statistical aspects, a methodology is presented to evaluate the fluctuations considering aspects relevant to power grid operators. Applying this methodology, we show that the highest fluctuations within the time frame of secondary regulation control of frequency reached 5% of the hourly power demand. Fluctuations of PV power within the primary regulation control time frame were within 3% of hourly power demand, and the 99th percentile of the fluctuations was within 1% of it. With the proposed methodology, which can be generally applied, we go beyond the description of statistical properties of short‐term fluctuations of PV, showing also the importance to evaluate them in a context relevant to power grid operators. The short‐term fluctuation distributions were non‐Gaussian, presenting a Laplacian‐like distribution shape. The highest regional PV short‐term fluctuations were within 5% of the average daily power demand. Fluctuations within the governor free control time frame were smaller and had less dependence on the season than those within the load frequency control time frame. Nevertheless, the governor free control time frame–related fluctuations were stronger than the load frequency control time frame–related ones, with extreme values of 20 times the fluctuations' annual standard deviation.