The aim of the project is to preliminary test a clustering procedure on multivariate road traffic time series in order to separate different paths between the days of the week and between different months. To create the multivariate time series two fundamental variables are considered the flow q(t,x) and the density ρ(t,x) . The clustering technique used is K-means with soft-dynamic time warping to compare the series. To see if the clustering technique is able to generalize well traffic dynamics, it is also applied to unseen data. In order to decide the most suitable number of clusters the silhouette coefficient is computed as well as a similarity measure between nearest clusters based on soft-dynamic time warping. Moreover a lower bound for a minimum number of days inside a cluster is fixed to overcome the trade off between interpretability of the clusters and their number.