Control Engineering Practice 13 (2005) 1223–1241 An approach to control automated warehouse systems Francesco Amato a , Francesco Basile b,Ã , Ciro Carbone b , Pasquale Chiacchio b a School of Computer Science and Biomedical Engineering, Universita` degli Studi Magna Græcia di Catanzaro, Via Tommaso Campanella 115, 88100 Catanzaro, Italy b Dipartimento di Ingegneria dell’Informazione ed Ingegneria Elettrica, Universita` degli Studi di Salerno, Via Ponte Don Melillo 1, 84084, Fisciano (SA), Italy Received 18 March 2003; accepted 28 October 2004 Available online 29 January 2005 Abstract Thegoalofthispaperisthedevelopmentofcontrolalgorithmsforthemanagementofanautomatedwarehousesystem.Asusual, the implementation of a control algorithm requires three preliminary steps: development of a reliable model; design of control procedures according to some optimality criteria; validation of these control procedures. As for modelling, a new level in the control architecture, namely an optimizer system, is introduced which performs real-time optimization thus simplifying the low-level control and improving the overall performance. In this context, the role of a detailed model of the whole warehouse is discussed and such a modelisbuiltupbyusingthecoloredtimedPetrinetsframework.Asforcontrol,weproposetwocontrolalgorithms,derivedunder the simplifying continuity assumption of the rack locations, to optimize the operations of the cranes (moving within the aisles of the warehouse) and the operations of the shuttle (moving on a straight line placed between the aisles and the picking/refilling area), respectively. To evaluate the performance of the proposed control algorithms we define three different cost indices. As for the validation, extensive simulations are performed on the model in order to prove the effectiveness of the proposed control algorithms. A further validation of the algorithms has been performed on a test bed in order to take into account communication delays and computation times. Finally, the proposed architecture and control algorithms have been applied to a real plant. r 2004 Elsevier Ltd. All rights reserved. Keywords: Discrete event systems modelling; Colored timed Petri nets; Warehouse automation; Factory automation; Optimization 1. Introduction In the last fifteen years, a big effort has been made to find optimal strategies for planning and control of warehouse systems. These issues have become more and more challenging with the growth and diffusion of modern computer technology which allows the imple- mentation of complex operations in a completely automatic way. Planning involves high level decisions, like assignment of goods to the storage locations (random, class-based, correlated product clustering policies) (VandenBerg, 1999) or the designing of the warehouse system itself. Control problems involve optimal sequencing and scheduling of storage and retrieval requests resulting in the so-called dispatching control. We refer to a general warehouse architecture consist- ing of a number of aisles, each one served by a crane, shuttles, picking/refilling positions and in/out buffers (see Fig. 1 for an example of a real layout). On both sides of each aisle there is a storage rack composed of n r rows and n c columns; moreover, as said, each aisle is served by a crane, capable of moving both vertically and horizontally at the same time, which performs the following operations: (i) picking of the Stock Unit (SU) to be stored at the buffer input of the aisle, referred as the input/output (I/O) point of the aisle; (ii) storage of the SU into the assigned location S ARTICLE IN PRESS www.elsevier.com/locate/conengprac 0967-0661/$-see front matter r 2004 Elsevier Ltd. All rights reserved. doi:10.1016/j.conengprac.2004.10.017 Ã Corresponding author. Tel.: +39089964400; fax: +3906233227957. E-mail addresses: fbasile@unisa.it, balraj@macmill.com (F. Basile).