This paper presents the problem of designing a logistic network for Less Than Truckload (LTL) shipments of freight transport providers. These companies have to collect products from many origins (suppliers) and ship them to many destinations (clients) through multistop routes and consolidate flows into hubs to achieve economies of scale. This problem is known as the Hub Location Routing Problem (HLRP). The main decisions concern the location of hubs, allocation of origin and destination nodes to the hubs, as well as the design of routes serving origin and destination nodes and determination of flows between hubs. We propose a mathematical model for the HLRP with capacity on hubs and single allocation, dedicated to postal service systems, where collections and deliveries may occur within the same routes. A metaheuristic based on a genetic algorithm, improved by a local search is developped to solve the problem. Experiments are conducted on data sets adapted from the Australian Post (AP) data set. Comparison with a standard solver proves the efficiency of the proposed method.