Dynamic card control strategy in pull manufacturing systems

The dynamic control of cards in pull systems can improve the performance of a manufacturing system in a turbulent demand condition. The traditional fixed card approach can reduce the performance of a system in some cases. This article presents a dynamic card control methodology in order to keep a high level of performance measures. The proposed methodology is based on the observation of customer demand in order to detect the fluctuations and adjust the number of cards. The demand fluctuations is evaluated based on two moving average: the first on a medium horizon and the second on a short horizon. The comparison between the two moving averages is the signal of the controller developed. A simulation environment has been developed to test the proposed methodology in both kanban and Constant Work In Process (CONWIP) systems. The simulations conducted in several scenarios highlight the better performance obtained using the controller. The main performance measures investigated are: the throughput, the queue of orders waiting to be processed, the throughput time of the orders, the number of cards used and the orders delayed. The benefits are more relevant in the case reliability reduction of the manufacturing system