Model-based Identification of Alternative Bidding Zone Configurations from Clustering Algorithms Applied on Locational Marginal Prices

This paper deals with the application of clustering methods to assist the bidding zone review processes in Italy, considering the Locational Marginal Prices (LMPs) as the relevant features. A novel approach based on the definition of the input data for clustering, depending on a number of scenarios defined by the Transmission System Operator, is exploited. The problem under analysis requires additional procedures to solve the challenging issue of incorporating node connection constraints in the clustering algorithm. A dedicated procedure, based on the definition of specific functions, is then applied to develop customised versions of k-means and hierarchical clustering. The customised procedures implemented can identify both wide clusters and outliers, whose location depends on the assessed scenarios.