Our paper (co-authored with Vincent Karels, Walter Rei and Luuk Veelenturf) is accepted in the European Journal of Operational Research and available here.
The modern logistics service providers face significant pressure due to competition, public expectations, and regulations. The challenge of efficiently organizing the physical distribution process, especially routing and scheduling, has become increasingly difficult. Vehicle Routing Problems (VRPs) lie at the heart of transportation planning, impacting costs and service quality levels. However, VRPs are notoriously hard to solve, especially when faced with uncertainty arising from unexpected events and stochastic travel times.
This research focuses on the effect of different service agreements between customers and logistics service providers. Service agreements determine when customers provide order information and when the logistics service provider announces the visiting time. Providers prefer early order information and a late announcement of visiting time, while customers prefer the opposite to facilitate workforce organization. The problem is inspired by real-world scenarios in the Business to Business (B2B) retail sector, where contracts differ between customers based on their size and needs.
We consider a logistics service provider with customers under two different service agreements. The first agreement involves customers ordering two days in advance and receiving information about their allocated driver and time window on the same day. The second agreement offers the same guarantee but one day in advance. We formulate an intermediate plan for customers under the first agreement while dealing with uncertainty in orders and demands for customers under the second agreement. They introduce an exact algorithm based on branch-and-bound techniques to solve this problem, providing a proven optimal solution.
The contributions of this research include introducing the Vehicle Routing Problem with Multiple Service Agreements, which cannot be solved using existing methods. The novel exact algorithm presented here can handle instances of up to 15 customers, with six of them being stochastic. The paper also includes a comparison between the proposed algorithm and other solution methods, offering numerical results and insights.
In summary, this research addresses the complexities faced by logistics service providers in planning and organizing transportation operations efficiently. By considering different service agreements and introducing a new algorithm, the study provides valuable insights and solutions to optimize the vehicle routing problem with multiple service agreements and stochastic demands.
Tom Van Woensel 