A typical distribution network consists of primary substations, secondary substations, medium-voltage lines, low-voltage lines and many other elements that ensure smooth operation of the system. In the last two decades, new elements and smart grid technologies have occurred and their impact on distribution system’s operation needs to be evaluated. The impact of distributed generation and smart grid technologies on network’s operation must first be tested using simulation models. Often, generic feeder models are used, but in recent years, regional representative feeder test cases have emerged.
To determine representative distribution feeders, a set of distribution feeders described with different parameters needs to be obtained. Using different clustering methods, the obtained feeders can be clustered in a way that distribution feeders in the same cluster share similar properties.
To determine the representative feeders that are typical for the territory of Slovenia, we have performed clustering on a set of feeders from Slovenian distribution network. We have obtained a set of distribution feeders from one of electric utilities in a format supported by QGIS, a desktop geographic information system application. Individual feeders were later presented in MATLAB in a format suitable for further analysis. Using different parameters of medium-voltage lines and transformer stations, we have determined several features that adequately describe feeder properties. The final set consisted of 206 distribution feeders that were clustered using a clustering method, while a suitable number of groups was determined by calculating various internal criteria. After the analysis, a solution with six groups, obtained by k-means++, was chosen. Representative feeders will serve as a basis for a more precise determination of the reference distribution network, which could be used to evaluate the impact of distributed generation and new technologies on distribution network’s operation.