On the Value of Operational Flexibility in Anheuser Busch InBev’s Trailer Loading and Shipment Problem: Data-Driven Approaches and Reinforcement Learning
Abstract
This paper introduces the Anheuser Busch Inbev (ABI)'s trailer shipment problem and suggests data-analytics methodologies to deal with it. The problem is to determine the proper weight of products loaded on a trailer (a transporting container) owned by ABI, which is delivered by a tractor (a motor vehicle) owned by third-party logistics (3PL) providers. ABI must meet a regulation that restricts the gross weight of the truck on the road. However, the challenge comes from the fact that the tractor weight is uncertain and unknown to ABI when it determines the load size on the trailer. We suggest machine learning-based methodologies that compute the optimal load size (weight) of a trailer to minimize the associated cost. Furthermore, we propose a dynamic trailer assignment methodology using reinforcement learning, which further reduces the cost dramatically. Using the transaction-level shipping data obtained from ABI, the suggested methodologies are evaluated. This work introduces a general context of the trailer shipment problem and suggests efficient data-analytics approaches to it, which can be widely applied in diverse industries associated with 3PL and has significant economic implications.