This study focuses on the issue of the location of micro-depots in urban areas in connection with the organization of food distribution. Given the rapid growth in demand for food products in cities, the need to reorganizing urban logistics is imperative. The Two-Echelon Location Routing Problem with Time Windows (2E-LP-VRPTW) addresses the unique challenges of transporting perishable goods. The main objective is to determine optimal locations for micro-distribution centers (MDC), whether the city center or on the outskirts, to enhance the efficiency of urban deliveries. To address the complexity of this problem combining location and routing, a method based on Variable Neighborhood Search (VNS) was adopted. This heuristic approach was compared to an exact method to evaluate its performance in different scenarios. The VNS approach achieved near-optimal solutions with computation time reductions of over 98%, especially on large instances with up to 100 retailers. The hybrid zone configuration (IN/OUT) also enabled cost savings of up to 7.2% compared to other layouts. To test and validate our approach, we generated a set of problem instances that simulate realistic urban food distribution settings. These instances were designed to reflect typical challenges encountered in last-mile delivery operations, such as demand variability, vehicle capacities, and service time windows. The obtained results show that the good positioning of the MDCs not only optimizes logistics operations but also ensures reliable deliveries while controlling costs.