Vehicle Routing with Stochastic Demandedited by Leonora Bianchi, IDSIA ContentsAbout the Vehicle Routing Problem with Stochastic DemandIntroductionThe vehicle routing problem is a classical problem in operation research and it is one of the most challenging combinatorial optimization tasks. Many variations of this problem have been formulated, the main task of the problem being the same: designing the optimal set of routes for a fleet of vehicles in order to serve a given set of customers. The most elementary version of the vehicle routing problem is the capacitated vehicle routing problem (CVRP). The CVRP is described as follows: n customers must be served from a unique depot. Each customer asks for a quantity q i of goods (i = 1,..., n) and a vehicle of capacity Q is available to deliver goods. Since the vehicle capacity is limited, the vehicle has to periodically return to the depot for reloading. In the CVRP, it is not possible to split customer delivery. Therefore, a CVRP solution is a collection of tours where each customer is visited only once and the total tour demand is at most Q. From a graph theoretical point of view the CVRP may be stated as follows: Let G = (C,L) be a complete graph with node set C = (c o , c 1, c2,..., cn) and arc set L = { (c i , cj) : ci, cj Î C, i ¹ j}. In this graph model, c o is the depot and the other nodes are the customers to be served. Each node is associated with a fixed quantity q i of goods to be delivered (a quantity qo = 0 is associated to the depot c o). To each arc (ci, c j) is associated a value dij representing the distance between c i and c j. The goal is to find a set of tours of minimum total distance traveled. Each tour starts from and terminates at the depot co , each node ci (i = 1,..., n) must be visited exactly once, and the quantity of goods to be delivered on a route should never exceed the vehicle capacity Q. The vehicle routing problem with stochastic demandThe vehicle routing problem with stochastic demand (VRPSD) is a variation of the CVRP where each customer demand is uncertain, instead of being exactly known a priori. The VRPSD arises in practice whenever a company faces the problem of deliveries to a set of customers, whose demands are uncertain. In this formulation, it is assumed that the demand q i of customer i is a discrete random variable whose probability distribution is specified by pi k , i.e., the probability that customer i asks for a quantity qi = k of goods, with k = 0,1,..., K, and K<=Q . In the VRPSD it is also assumed that the actual customer demand is known only when the vehicle arrives at the customer's location. In the deterministic VRP routes are planned in such a way that the vehicle always has enough capacity to satisfy all customers' demands on one of its pre-planned routes. When demands are stochastic, the situation is different. In fact, suppose a vehicle visits customers in a certain pre-planned order. It is possible that at some customer location, where the actual demand is revealed, the vehicle capacity becomes attained or exceeded, and a route failure is said to occur. In such a situation, the vehicle must return to the depot to replenish. The action taken by the vehicle after a route failure depends on the routing strategy that the planner decides to adopt. For example, the simplest strategy is to go back to the customer where demand has not been fully satisfied, and than to proceed visiting other customers in the pre-planned order. In general, also the planning of vehicle routes depends on the strategy that the planner wants to adopt. Whatever strategy is adopted for planning the vehicle routes for the VRPSD, the distance traveled by the vehicle is a random quantity, since the number and location of recurse actions is not known in advance. Indeed, the actual route followed by the vehicle might be in general different from the pre-planned one. The goal, in the VRPSD, is to determine a routing policy such that demand at each customer is met, and the expected distance traveled is minimized. Readings
Links
| |
Network Coordinator: Marco Dorigo e-mail: mdorigo@ulb.ac.be Web site responsibles: Christian Blum and Max Manfrin |