REDUCING VARIABILITY BY MONITORING AND FEEDBACK ADJUSTMENT

Abstract

ABSTRACT: This paper presents the results obtained using two strategies to eliminate fluctuations in the value of a critical dimension of a car’s braking system critical component. The quality levels required in this product demand great precision in some dimensions and, due to the peculiarities of its manufacturing process, this can only be achieved by making adjustments unit to unit. Although these adjustments are made automatically based on the exact measurements of the components there are fluctuations in the value of one the critical dimension. Since the causes of the fluctuations were unknown, it was decided to approach the problem from a statistical point of view. This was done by introducing changes based on statistical criterions in the algorithm that regulates the adjustments. A first strategy, the most classical, bases the adjustments on the difference between a moving average and the target value; a second one is to make adjustments based on the deviation of the last part with respect to the target value. After defining a set of parameters for each strategy and selecting their optimal values we have simulated the two strategies’ behaviour with the values of three batches representative of the overall production. Both strategies offer good results, but the second is slightly better and easier to implement and therefore it was the one incorporated in the production process. Key Words: Feedback adjustment, statistical process control, exponential weighing, variability reduction.