Clustering-based definition of the electricity market zones

In the last years, the progressively growing renewables’ production capacity has increased fluctuating infeed. And, as the lines’ transmission capacities have not been upgraded simultaneously, this has resulted in the increase of congestions number in the electricity transmission grids across Europe. Thereby, countries like Germany have lately seen an enormous increase in their redispatch cost. Therefore, the need for an improved congestion management has day by day become more impelling within the continent. In order to cope in an adequate way with the increase of congestions in the short and medium term, before the still necessary power networks upgrade. As a result, having recognized inside the establishment of optimal zonal pricing mechanisms the answer to this necessity, the European Union has emanated in 2011 through the Agency for the Cooperation of Energy Regulators the Framework Guidelines on Capacity Allocation and Congestion Management for Electricity, which actually has tried to clarify the rules for creating optimal zonal configurations. This made sense, since current European electricity pricing schemes are based on uniform configurations or not optimal zonal ones. But unfortunately, the aforementioned European guidelines have only managed to provide general rules on the topic, without being able to successfully carry out the search for an optimal zonal configuration. For this reason, the last years have seen progressively growing the scientific literature regarding the subject. And this thesis fits equally in this scenario, by trying to give a methodology aimed at locating the most suitable technique to deterministically define an optimal zonal configuration. In particular, in fact, this this thesis contains a three level methodology. In the first one, the zonal configurations’ optimality requirements are clearly stated through objective and quantitative parameters, which permit to better decree price zones’ optimality rather than European general rules. In the second one, having uniquely recognized the desired output, the apparently most suitable clustering algorithms to reach it are presented. Since also the geographical clustering could be used to produce price zones, but actually it has to be rejected because only capable of giving the currently existing zonal configurations, which totally differ from optimal ones. Therefore, in the second level of this thesis’ methodology a k-means clustering, a k-medoids one, a hierarchical one, a genetic algorithm and a price differential clustering are deeply described. In both their versions, LMPs-based and PTDFs-based, and together with their changes aimed at better complying with the previously declared zonal configurations’ optimality requirements. After that, the methodology’s third level provides a series of price zones assessment criteria. Intended to evaluate the newly defined zonal configurations’ optimality, and also to allow the comparison among different price zones definition techniques. Eventually, the so created methodology is applied to a real case study represented by a reduced model of the European transmission grid, so as to test its effectiveness. The results highlight that some of the considered clustering algorithms are clearly inappropriate to work out the problem of defining an optimal zonal configuration, like the price differential clustering and the genetic algorithm, while others have comparable performance in terms of defined price zones’ optimality.