Clustering Nepalese power system into stable islands considering real power balance

Abstract

With the increasing demands and consecutively the size and complexity of the power system, blackouts have become unavoidable in today's power system. So, the study about plans and strategies for grid restoration after a blackout has also become a major necessity. This paper uses the strategy which first partitions the blackout area into several islands and restores them paralleled to accelerate the restoration task. An agglomerative clustering algorithm has been used to partition the network which solves a constrained optimization to maximize the generation load difference and form the clusters based on it. Two cases have been studied upon in the Integrated Nepalese Power System (INPS). Case I has taken two black start generators as partitioned the system two clusters whose generation load difference was found to be 200.308 MW and 321.778 MW respectively. Similarly, case II has taken three black start generators and partitioned the system into three islands whose generation load difference was found to be 142.632 MW, 313.912 MW, and 118.788 MW. These positive values indicate that the islands can be operated independently before synchronization even if the loads deviate from the forecasted value.