A heuristic procedure for the design of an urban distribution system for e-commerce deliveries

The paper describes a procedure for the design of logistic systems involved in the “last mile” management for business to consumer e-commerce. Optimization of the resources and minimization of negative environmental impacts are the procedure’s objectives. The procedure was applied in Terni city area in Italy. A heuristic for the Location Routing Problem, adopting genetic algorithm (GA) as a solution procedure, is proposed. Introduction In the Internet age, it is expected the growth of the Business-to-Consumer (BtoC) e-commerce will determine more frequent shipments of small loads of goods to satisfy consumer needs, with a negative impact on the urban mobility (increase of congestion) and the environment (noise, air pollution). This study, in accord with Municipality objectives, is based on a problem framework that includes both the distribution costs, viewed from the perspective of private enterprises, and social and environmental costs, viewed from the perspective of the community. Reducing both costs can be achieved by making the whole system more efficient, for example by increasing load factors and delivery productivity as well as by improving routing and scheduling strategies. The procedure proposed in the paper is based on the work by Taniguchi et al. [1], who proposed a mathematical model for determining the optimal size and location of public logistic terminals, by using queuing theory and nonlinear programming techniques (genetic algorithms). Their model explicitly takes into account traffic conditions on the network by assuming that the assignment of pickup/delivery trucks complies the user equilibrium condition with variable demand. (i.e.: the decision-maker can determine the optimal size and location of logistics terminals but cannot control the assignment of truck traffic). In our paper we aim at dealing with a more general case, which includes the routing problem into the optimization procedure. A complete overview of the state of the art of the combined location-routing problem until 1998 can be found in Min et al. [2]; more recent works in location-routing problem are presented in Wu et al. [3] and Lin et al. [4]. The logistic scheme for urban deliveries The logistic scheme proposed in this paper is based on the presence of two fundamental elements: • the city terminal (terminals), where system performances are improved by means of load consolidation, intensive implementation of advanced information systems and cooperative operations of freight transportation; • the drop-points, where the goods are delivered and consumers pickup their parcels. In order to reduce new trips and parking needs as well as to minimize the discomfort for the consumer who is involved in parcels pickup, the study idea is to join the pickup and regular and/or enjoyable activity of the consumer (for example, the breakfast at the bar). Thus, commercial and public sites as bars, postal offices, newspapers shops and bookstores have been selected as possible drop-points of the delivery scheme. However, home deliveries are reserved for disabled people and for heavy goods. The design procedure A heuristic procedure for the design of such a logistic scheme is here proposed by integrating together transportation system theory and operational research techniques. The main procedure steps are: • split of the study area in “demand zones”, each of them is to be served by one or more drop-points • estimation of the demand of BtoC goods in each zone by applying a standard Logit model [5] • application of AHP (Analytic Hierarchy Process) methodology [6] to compare performances of droppoints having similar accessibility and then to choice the best one by 2 criteria describing both consumers and social points of view. • performing a user equilibrium assignment of passenger cars for compute the network links cost • allocation of the “demand zones” to drop-point • planning and locating the city terminals by a heuristic procedure to obtain an approximate optimal solution of the location-routing problem. The model is based on a genetic algorithm that considers possible tours between drop-points and then searches for optimal facilities location and route design simultaneously; in the GA fitness calculation phase the model solves allocation of the drop-point to city terminals and vehicle routing problems. The main inputs of the problem are socioeconomic variables of demand zones, Origin/Destination matrix of car trips, road network graph, as well as the set of candidate areas for locating the city terminal and delivery fleet size. Application to the town of Terni A preliminary application of the methodology proposed has been carried out to design an urban delivery system for Terni, a 100,000 in-habitants town of Umbria, Italy. Simulation results indicate that the 55% saving of total fleet trip length and 71% of total fuel consumption can be achieved, against a totally non-cooperative scenario taken as reference term.