Modeling and performance analysis of sequential zone picking systems

This paper develops an analytical model of sequential zone picking systems. The systems belong to the most popular internal transport and order picking systems in practice, due to their scalability, flexibility, high-throughput ability, and fit-for-use for a wide range of products and order profiles. The major disadvantage of such systems, though, is congestion and blocking under heavy use, leading to long order lead times. In order to diminish blocking and congestion most systems make use of a dynamic block-and-recirculate protocol. The various elements of the system, like conveyor lanes and the pick zones, are modeled as a network of queues with multiple order classes and with capacity constraints on subnetworks, including the dynamic block-and-recirculate protocol. Due to this protocol, however, the stationary distribution of the queueing network is highly intractable. Therefore, an innovative approximation method, using jump-over blocking is proposed to accurately assess key performance statistics such as throughput and recirculation. Multi-class jump-over networks admit a product-form stationary distribution, and can be efficiently evaluated by Mean Value Analysis (MVA) and use of Norton's theorem. The method is most suitable to support rapid and optimal design of complex zone picking systems, in terms of number of segments, number and length of zones, buffer capacities, and storage allocation of products to zones, in order to meet prespecified performance targets. Comparison of the approximation results to simulation show that for a wide range of parameters the mean relative error in the system throughput is typically less than 1%. The approximation is also applied to evaluate a real-life zone picking system of a large wholesaler supplying non-food items.