A taboo search heuristic for the optimisation of a multistage component placement system

The manufacturing of Printed Circuit Boards (PCBs) is an important issue in the electronics industry. The core of this process is the insertion of electronic parts into predefined places on a board. To this purpose a numerically controlled assembly system, consisting of machines working sequentially, is used. Each machine has a number of bins (feeders) containing parts to be inserted in the boards. A bin cannot contain more than one part type. However, a part type can be allocated to various bins (possibly of different machines). At each station an arm grasps a part from a bin, moves to the part location, inserts it and then returns to a bin. When the last part is inserted, another board is processed. Once the system has been designed, three operational decisions are to be taken: the allocation of part types to machines, the assignment of part types to feeders, the pick and place sequencing for each machine. The objective is the minimisation of the time spent by any PCB in the system, i.e. the minimisation of the processing time of the slowest machine. In this paper the whole operational problem is modelled as a min-max allocation-routing problem and a Taboo Search heuristic is presented. This procedure is evaluated using both real world and randomly generated instances. Computational results show that the heuristic allows achieving remarkable cost savings.