Multidimensional scaling and stock location assignment in a warehouse : an application (Q2711708)

Order picking is one of the most costly activities in a typical warehouse and therefore it needs careful attention at the design stage, mainly in the ''man to good'' picking systems, i.e. in a system where an operator picks goods from several shelves. One of the main subject, both during the design and the management of a warehouse, is the stock location assignment, as stock position is related to material handling time. In the whole duration of picking activities the following components can be identified: NEWLINENEWLINENEWLINEA1. the time to reach the first picking point from the starting position (\(I/O\)) and to reach the \(I/O\) point from the last picking position, or from the position where the pallet is full; NEWLINENEWLINENEWLINEA2. the time to reach, sequentially, all the picking positions in the picking list; NEWLINENEWLINENEWLINEB. the time to position in front of picking position; NEWLINENEWLINENEWLINEC. the time to pick the right quantity for each product in the picking list. NEWLINENEWLINENEWLINEIn order to minimize the whole duration, it needs to reduce these components of time; but A1 and A2 present the greatest impact on the above-mentioned time. These two components can be reduced by modifying the system layout, or by changing the travel policy, or finally, by considering a different location assignment policy, as proposed in the present paper. By means of an application the suitability of a multivariate statistical methodology, as multidimensional scaling (MDS), to solve an optimization problem is shown. In particular, considering the stock location assignment problem in the warehouse of a supermarket chain, the solution gained by applying MDS to a set of seven variables is compared with the one obtainable by considering the usual techniques applied in this context. A wide discussion of results is reported.