Optimal placement of power quality monitors by accounting for several key power quality disturbances

Many distribution system operators are installing PQ monitoring systems in their medium-voltage networks. Installation of PQ monitors in all the nodes of the network is uneconomical, and power quality monitors are rather expensive. Therefore, the optimal placement of PQ monitors has to be found, while minimising cost and maximising observability. Researches published so far have presented PQ placement algorithms, but only from the aspect of one PQ disturbance (either dips or harmonics). In this paper, an algorithm for PQ monitor placement is presented, and its novelty is that it is the first research work accounting for several key PQ disturbances at once: harmonics, flicker, unbalance and dips. The chosen PQ disturbances are also modelled using novel modelling techniques, with the PQ disturbance behaviour modelled for a longer time period of one week or month. PQ disturbance behaviour is modelled for a real study case, where known real network topology, customer structure and measurements are used to develop a model validated by PQ measurements. Obtained synthetic PQ data, which are validated by PQ measurements, are used to optimise the PQ monitor locations. To choose the optimal PQ monitor locations, multiple linear regression was used, with the coefficient of determination R2 being the selection criterion. The results show that only two PQ monitors are necessary to monitor the analysed large real medium-voltage distribution grid, with high confidence. The methodology is straightforward and relatively easy to apply, and can be implemented for other medium-voltage distribution networks.