Demand Model and Vehicle Routing Problem: an Integrated Procedure to Optimize the Urban Goods Movements
A Vitetta, A Polimeni, Mediterranea University of Reggio Calabria, IT
In this paper an integrate procedure for goods movement simulation in an urban area is organized. The procedure consists of two levels: the first is a commodity-based demand model; the second is a vehicle routing problem with time windows.
In this paper an integrate procedure for goods movement simulation in an urban area is organized. The procedure consists of two levels: the first is a commodity-based demand model that simulates the goods movements in terms of quantity; the second is a vehicle routing problem with time windows that simulates the delivery process.
The demand models can be classified referring to the main characteristics on the basis of different criteria.
The urban goods movements are characterized by some vehicles that delivered the goods over to some retailers, leaving from a wholesaler/carrier depot, a logistic platform, an urban distribution center, etc. The problem can be formalized using a Vehicle Routing Problem (VRP).
In this paper a commodity based model to simulate the goods quantity attracted in a zone is proposed. The model structure proposed considers that there is a carrier that supply some retailers, the carrier choice the first zone to serve, then the retailers in the zone, the time of departure from the depot and the path to reach the retailers. The output of the demand model (goods quantity) is an input for the vehicle routing problem.
In literature various models and methods are proposed to formulate and solve the VRP.
The formulation of a VRP is based on the definition of an objective function to consider the benefit/costs that characterize the problem.
Exact or approximate procedures are proposed to solve the VRP.
Then, in this paper, a VRP with Time Windows (VRPTW) is formulated. The problem is formulated using the network theory to define the cost involved in the problem. The objective function considers the path cost between an origin/destination pair (e.g, depot ? client, client ? client, client ? depot) and the fixed cost (fuel consumption, driver pay, etc.).
The path cost is the sum of two elements: the additive costs, that depends on link and flow characteristics, and the non-additive costs.
The additive costs are obtained solving a shortest path problem, the output is a cost matrix. The non ? additive costs are the cost for unloading operations.
The fixed costs consider the costs for fuel consumption, driver/s pay, other costs (maintenance, technical costs, etc.)
The VRPTW is solved using the genetic algorithm.
Association for European Transport