Comparison of clustering algorithms for the selection of typical demand days for energy system synthesis

The results show that all clustering methods are able to determine energy systems that are close to the optimal system, however their demand related costs are approximated best and most reliably with k-medoids.