Warehouse optimization through anticipatory order picking

The growing importance of e-commerce puts warehouses under high pressure. First, small order sizes typically result in efficiency loss of the process of retrieving the requested products from storage. Second, the extreme short lead times require warehouses to respond to incoming customer orders within hours or even minutes, while keeping operational costs at their minimum. This high degree of dynamism, however, leads easily to ad hoc decisions and missed opportunities for optimization. In this research, we focus on the development of decision-support systems for anticipatory order picking, i.e., the retrieval of products from storage before the customer actually placed the order. Next to already confirmed customer orders, a dynamic list of expected (uncertain) orders is generated based on developed forecasting techniques that use historical and real-time data. Next, each time a picker becomes idle, a new pick tour is constructed based on known and expected orders. The advantage is twofold. First, increased opportunities for warehouse optimization appear through better batching procedures as the pool of potential orders that can be picked together is larger. Second, once a customer eventually orders such an anticipated product, the required time to prepare the order and ship the products to the customer will belower as the product has already been retrieved from storage