Vehicle routing with stochastic demands: first insights of a real-life case

Abstract

The Vehicle Routing Problem (VRP) has already been widely studied in literature. The main focus is however on deterministic VRPs. Nevertheless, different forms of uncertainty can occur in real life, such as stochastic demands, customers, travel times and service times. Uncertainty in demand is the most common. In this work, a case study is presented in which customer demands are uncertain. The company under study is a logistics service provider who serves different customers in Europe. Customers have to submit transportation requests in which they have to specify their orders that need to be collected. However, when arriving at the customer, often deviations between the actual quantities and the specified quantities are observed. Furthermore, customers are free to specify their orders, for example, in volume, real weight or dimensions. Yet, the planning in the company is made up in load meters, so orders needs to be translated in the correct load meters, which is not an easy task. This makes it difficult to make a good routing plan. This work tries to provide insights by looking at patterns in historic data. Routing plans based on initial and real orders are compared and, based on these insights, a stochastic VRP will be studied.