Journal of Intelligent Manufacturing, 12, 521±534, 2001 # 2001 Kluwer Academic Publishers. Manufactured in The Netherlands. An optimization methodology for intermodal terminal management L. M. GAMBARDELLA,* M. MASTROLILLI, A. E. RIZZOLIand M.ZAFFALON IDSIAÐIstituto Dalle Molle di Studi sull'Intelligenza Arti®ciale, Galleria 2, CH-6928 Manno, Switzerland E-mail: luca@idsia.ch; monaldo@idsia.ch; andrea@idsia.ch; zaffalon@idsia.ch A solution to the problems of resource allocation and scheduling of loading and unloading operations in a container terminal is presented. The two problems are formulated and solved hierarchically. First, the solution of the resource allocation problem returns, over a number of work shifts, a set of quay cranes used to load and unload containers from the moored ships and the set of yard cranes to store those containers on the yard. Then, a scheduling problem is formulated to compute the loading and unloading lists of containers for each allocated crane. The feasibility of the solution is veri®ed against a detailed, discrete-event based, simulation model of the terminal. The simulation results show that the optimized resource allocation, which reduces the costs by 1 3 , can be effectively adopted in combination with the optimized loading and unloading list. Moreover, the simulation shows that the optimized lists reduce the number of crane con¯icts on the yard and the average length of the truck queues in the terminal. Keywords: Intermodal terminals, ¯exible job shop, resources allocation, containers optimization 1. Introduction Intercontinental cargo transport has been continuously increasing since the advent of container shipping in the 1950s. Nowadays 95% of world cargo is moved by ship. Ports play the role of exchange hubs where containers are moved from ships to trains and trucks and their ef®ciency is fundamental to sustain the mounting cargo traf®c. To cope with increased traf®c demands and with decreasing pro®t margins the terminal operators need to improve the management of terminal processes such as ship berthing, ship loading and unloading, straddle crane routing, resource allocation, yard space management. All of these problems are strictly interrelated; for instance, the choice of the position of the containers on the yard impacts on the decisions made regarding the allocation of resources, the allocation of resources constrains the scheduling of loading and unloading operations, and so on. The complexity of these processes makes it necessary to use ef®cient methods for the optimization of the overall system: operations research techniques and simulation have been proven particularly apt to solve these kinds of problems (see, for instance, Kozan and Preston, 1999; Kim and Bae, 1999; Kim and Kim, 1999; Magnanti and Wong, 1984). The approach currently adopted by most terminal managers splits the problem by treating each ship as an independent entity: a ship planner is dedicated to plan loading and unloading operations for a single ship and to allocate the needed resources in terms of quay cranes, yard cranes, lifters and manpower. Since there is no cooperation among different ship planners, this method is a source of con¯icts and performance degradation mainly in the case of resources that must be shared among parallel loading and unloading *Corresponding author.