Triplet Irradiance Measurements

Cloud speed and direction govern the dynamic nature of irradiance, and hence the electrical output of PV panels and arrays. For example, ramp rates are directly proportional to the cloud speed and related to the coincidence of cloud direction and array orientation. In this work, we demonstrate that cloud dynamics can be determined using a triplet of silicon irradiance sensors logged at high frequency. Using cross-correlation, cloud speed and direction are calculated from the time shifts of cloud edges detected from sensor pairs. Data was initially recorded for 5 months from March 2016, near Leeds, UK. Further data will be recorded in 2018, near Kampala, Uganda. From the 2016 data, minimum cloud speeds of 4 ms-1 and a maximum of 40 ms-1 were obtained while dominant direction was found to be from the west-northwest. Comparison was then made to hourly wind speed and direction data recorded at 10 m above the ground level from the Bingley SAMOS weather station which is located 4 km away from the triplet of irradiance sensors. Cloud speed is invariably higher than the 10 m wind speed, on account friction between air and the earth’s surface. Nevertheless, we report a strong linear relationship between cloud speed and 10 m wind speed, with a correlation coefficient, R, of 0.9. Regarding comparison of direction, a deflection of 22.50 in the cloud direction clockwise of 10 m wind direction was observed as the dominant ground level wind direction was found to be from the west, while the dominant direction of the clouds was found to be from the west-northwest. Ramp rates were determined and compared with 10 m wind speeds. A low to moderate positive correlation were observed with a minimum and maximum correlation coefficient, R, of 0.2095 and 0.4274 in June and April respectively. These correlations are understood to have been diminished because of noise and solar irradiance reflected and focused onto the sensors by various cloud sides. This work demonstrates that both triplets of irradiance sensors and ground level wind data are useful and low-cost methods for predicting the likely frequency and magnitude of ramp rates of PV arrays.