More efficient order picking for the online shopping rush
One of the perks of online shopping is being able to place orders at any time of day. In the warehouse where orders are fulfilled, such customer flexibility can lead to unpredictable peaks in the flow of incoming orders. Designing a picking system that can ship out packages quickly and efficiently to customers, even when order volumes unexpectedly spike, can be a challenge. PhD candidate Jelmer van der Gaast of Rotterdam School of Management, Erasmus University (RSM) developed models that can help designers to create optimal design and control methods to improve the performance of picking systems.
Van der Gaast first studied how order picking can be optimised in warehouses divided into zones. In each zone, order pickers collect items from this part of the warehouse and deliver them to a collection area where they are combined and prepared for shipping. Zone picking presents warehouse designers with several puzzles. How big should each of the zones be and how many order pickers should be assigned to each? And which products should be allocated in what quantity to which zone?
In his research, Van der Gaast developed an analytical model that can predict how a particular set-up in a zone picking system will perform, even when it has to process fluctuating flows of orders. He verified his results using several datasets from a real warehouse. The results show that the model is able to predict the efficiency of a picking system with very high accuracy. This makes it a useful tool for testing out new systems, the researcher says.
Bottlenecks and buffers
Van der Gaast then expanded his model to predict how the performance of a zone picking system is influenced by the confluences of conveyor belts coming from several zones. Such locations typically act as bottlenecks where congestion occurs and the efficiency of the entire system comes under stress. In this study, Van der Gaast calculated where buffer spaces should be located in each zone to avoid congestion. He discovered that the optimal allocation of buffer spaces in the system depends on the arrival rate of incoming orders.
The ‘milk run’ is a relatively new approach to order picking, in which a picker makes continuous rounds of the warehouse, collecting items as they go, returning after each round to the collection area to offload the items. As soon as a new customer order comes in, it is relayed to the picker on the warehouse floor in real time. The picker then collects the item from its storage location the next time they pass it on their round.
Milk run picking has the advantage of reducing the time order pickers spend retrieving an order, but switching from traditional systems to a milk-run system requires additional investments. Calculations by Van der Gaast demonstrate that milk-run order picking is an attractive and considerably quicker method than traditional systems when orders arrive in quick succession, especially when the products are placed efficiently around the warehouse.
Download Jelmer van der Gaast's PhD thesis 'Stochastic Models for Order Picking Systems' here.