Order picking with multiple pickers and due dates

Simultaneous solution of order batching, batch assignment and sequencing, and picker routing problems

Autor/innen

  • André Scholz
  • Daniel Schubert
  • Gerhard Wäscher

DOI:

https://doi.org/10.24352/UB.OVGU-2018-553

Schlagworte:

Order Picking, Order Batching, Batch Sequencing, Picker Routing, Traveling Salesman, Variable Neighborhood Descent

Abstract

In manual picker-to-parts order picking systems of the kind considered in this article, human operators (order pickers) walk or ride through the warehouse, retrieving items from their storage location in order to satisfy a given demand specified by customer orders. Each customer order is characterized by a certain due date until which all requested items included in the order are to be retrieved and brought to the depot. For the actual picking process, customer orders may be grouped (batched) into more substantial picking orders (batches). The items of a picking order are then collected on a picker tour through the warehouse. Thus, the picking process of each customer order in the batch is only completed when the picker returns to the depot after the last item of the batch has been picked. Whether and to which extend due dates are violated (tardiness) depends on how the customer orders are batched, how the batches are assigned to order pickers, how the assigned batches are sequenced and how the pickers are routed through the warehouse. Existing literature has only treated special aspects of this problem (i.e. the batching problem or the routing problem) so far. In this paper, for the first time, an approach is proposed which considers all aspects simultaneously. A mathematical model of the problem is introduced that allows for solving small problem instances in reasonable computing times. For larger instances, a variable neighborhood descent (VND) algorithm is presented which includes various neighborhood structures regarding the batching and sequencing problem. Furthermore, two sophisticated routing algorithms are integrated into the VND algorithm. By means of numerical experiments, it is shown that this algorithm provides solutions of excellent quality.

Veröffentlicht

2018-09-03

Ausgabe

Rubrik

Artikel