Analysis and performance assessment of the active frequency drift method of islanding prevention

Islanding of photovoltaic (PV) systems can cause a variety of problems and must be prevented. However, if the real and reactive powers of the load and PV system are closely matched, islanding detection by passive methods becomes difficult. Also, most active methods lose effectiveness when there are several PV systems feeding the same island. The active frequency drift method (AFD), also called the frequency bias method, enables islanding detection by forcing the frequency of the voltage in the island to drift up or down. In this paper, AFD is studied analytically, using the describing function analysis technique, and by simulation, using MATLAB. It is shown that AFD has a nondetection zone (NDZ) in which it fails to detect islanding, and that this NDZ includes a range of unity-power-factor loads. Finally, the paper describes a novel method using positive feedback which significantly shrinks the size of the AFD NDZ.