In this study, we introduce a distribution network design problem to determine optimal resources required (vehicles, field executives) to operate a city-logistics network with a given distribution of sellers and customers. Since the last decade, the e-commerce market is characterized by enormous growth and changes. This growth has caused and still causes major inefficiencies within the delivery to customer part of the supply chain known as ‘last mile’. On the other hand, pickup from the sellers, to bring goods into the supply chain also known as ‘first mile’ creates opportunities of marginal utilization in the city-logistics supply chain. Urban areas represent great challenges for logistic organizations for better economic efficiency in order to meet seller and customer requirements.

We initially investigated standard clustering techniques such as partitioning based clustering (k-means, k-mediods), hierarchical clustering (agglomerative, divisive) and density based clustering (dbscan, optics). Being a multivariate optimization problem, we develop a hybrid clustering model to generate time-constrained and resource-constrained clusters. These clusters also help to identify optimal hub locations to act as starting points for seller pick-up and customer delivery activities. This dynamic model will improve the productivity of resources considering the supply load variability and unpredictable demand nodes.  The vehicular routing within a cluster to calculate the travel time is done using the meta-heuristic method of ant colony optimization (ACO). This clustering model will help to reorganize logistics more efficiently and deal with urban logistics challenges.