Distribution network clustering for reactive power planning and distributed voltage control

According to the inherent characteristics of power systems, the voltage reduction is different in each part of power system. The voltage drop is directly proportional to the demand current and the total impedance between the source and the loads. Therefore, in distribution system voltage/VAr control is more important rather than transmission network, because of higher line R/X ratios rather than the transmission network and customers or end-users receive power through a higher impedance and experience a larger voltage drop in distribution system. This work presents multistage optimal reactive power planning and clustering the distribution network to implement distributed voltage/VAr control approach. At first, reactive power planning to reduce active power losses and save costs has been investigated. According to the result of the first stage, the power system is clustered into some partitions using spectral clustering and reactive power weighted matrix. The optimal number of clusters has been evaluated and determined by four criteria and the clusters are defined by fuzzy c-mean method which gives better vision about cluster boundaries. Finally, a sensitivity-based approach has been used to determine the precise location of VAr resources and the results show that voltage/VAr control in a clustered distribution system has more acceptable outcomes rather than centralized voltage/VAr control in the whole distribution network.