Abstract
Smart warehouse automation has emerged as an effective, competitive solution for suppliers and distributors. With the increasing demand for physical storage and distribution services, suppliers and service providers are challenged to respond not only effectively, but with minimal latency. Differentiated service levels for different classes of customer orders have not yet, however, been developed for physical storage and retrieval. In this paper, in the context of smart warehouse automation services, a novel policy, called Differentiated Probabilistic Queuing (DPQ) is developed for servicing customers' orders by Automated Guided Vehicles (AGV). Applying the DPQ policy, the average overall latency of each customer order, the mean overall processing time of this customer's orders in the smart warehouse automation system, is characterised under Poisson customer order arrival patterns. The weighted average latency of all customer orders is optimised over the choice of (1) storage assignment and (2) DPQ policy. Due to the existence of two types of variables, Alternating Minimisation method is applied to solve this joint optimisation problem. Compared with a combination of the classical turn-over rate storage assignment method and FCFS policy, the new approach yields 19.64% lower (better) objective function value with statistical significance. Numerical analysis results also indicate, as expected, that when the smart warehouse system resources become more limited, and the price difference among different classes of customer orders increases, the improvement becomes even more significant.