State Estimation in Power Distribution Network Operation

Abstract

The majority of power distribution networks were planned, designed and built as a passive but reliable link between the bulk power transmission point and the in- dividual customer. Enough latent capacity in cables and lines to accommodate anticipated demand growth was allowed and so the system was left unmonitored. Following the signicant development in business regulation, technology evolutions and various government policies towards low carbon renewable generation, it has become necessary to operate the distribution systems efficiently and in a controlled manner. This obviously needs state estimation for network control functions. State estimation is the core function of any energy management system in transmission networks. However little emphasis have been given to the distribution system state estimation, mainly due to the absence of adequate network measurements and also lack of rigorous methodology and tools that could be applied on restricted measure- ments. The scarcity of measured information offers formidable challenge to the state estimator to provide reasonably meaningful estimates of the system states. This introduces bottlenecks in carrying out a range of substation and feeder automa- tion tasks that rely on the quality of the state estimator and opens up many issues like modelling of demand, identification of suitable estimator and placement of new measurements etc. This thesis attempts to address these issues. Thus, the objec- tives of this research are to model the demand as pseudo measurement, identify the state estimation methodology to suite the distribution scenarios and find the effec- tive locations for placing measurements for improving the quality of the estimated quantities. The thesis discusses in detail the criterion for identifying suitable solvers for the distribution system state estimation and stochastic optimisation methods to model the demand. It also discusses a probabilistic technique for identifying effective locations for measurement placement. The robustness of the state estimation algorithm against changes in network topology has been addressed in a statistical framework. All the concepts have been demonstrated on 12-bus radial and 95-bus UKGDS network models.