Maintenance optimization of railway infrastructure includes several kinds of aspects, such as safety, economic, operational, organization and regulatory issues. Among them the regulatory issues, that are fixed, increase the maintenance costs significantly. This is especially true in so-called capillary networks (local regional railway networks), where only the freight transport exists. Hence, the question is how to minimize maintenance costs with respect to regulatory issues? To solve this problem, we propose a clustering approach. The idea is to cluster tracks, considering elements of railway infrastructure as attributes. Once railway tracks are clustered in groups with similar attributes, then the maintenance can be organized more efficiently. In this paper, Variable Neighborhood Search metaheuristic is developed to solve minimum sum of squares clustering problem. Based on the results of clustering and available real and simulated data we report 22% savings in maintenance schedule for clusters.