A Location-Routing Model for a Two-Echelon System of Urban

The freight distribution system of many cities suffering from congestion is based on the use of a single echelon of consolidation facilities that are located in the city outskirts. In this thesis, the introduction of a second level of consolidation facilities to be placed within the city and to be supplied from the first consolidation level by night deliveries is proposed as a solution against congestion and other transport externalities. A model to support the implementation planning of this second-level of logistic facilities and hence of a two-echelon system for urban freight distribution is designed. The model solves a specific case of an uncapacitated Location-Routing Problem (LRP) for the second echelon of facilities, by integrating the routing decisions into the location planning process and considering the cost of the first echelon in the optimization problem. As the city distribution system involves large datasets, the LRP has to be solved by the means of a combination of heuristics, in this case, the Tabu Search metaheuristic, which has been widely proved as adequate to solve the Facility Location Problem (FLP) and the Clarke & Wright heuristic, equally proven as useful to solve the Vehicle Routing Problem (VRP). The algorithm of the model is designed using the Matlab programming language. A fictitious study case corresponding to a parcel business that plans to use the two-echelon system to undertake its distribution activities in the city of Barcelona has been created and used to generate problems to study the performance of the model. The results obtained show that the model works properly, obtaining results of an acceptable quality that could not be obtained by other means; results that are improved when incrementing the number of maximum iterations but with a small trade-off in relation to the increase of the computational time. However, the model appears to be non-robust with the parameters and the maximum number of iterations contemplated in this thesis, because its convergence to the best possible solution depends on the configuration chosen as initial solution. In addition, the importance of having integrated the routing decisions into the facility location process has been confirmed and quantified. Finally the simulation by the means of a discount factor affecting the costs associated to the first echelon is introduced and recommended for further investigations. The validation of the model and the calibration of the parameters also remain as proposed next step