STATISTICAL CONTROL OF A MOULDING PROCESS

The paper deals with the problem of statistical process control (SPC) applied on the process of moulding tin coated cold rolled steel. The process level fluctuates around the historical mean, subgroup averages in the X-bar control chart show an extensive variation and as a result, frequent signals occur. Considering the high process capability, it would be pointless to seek assignable causes and take measures to remove this variation at the current state of the art; it is regarded an inherent part of the process. From this reason the classical X-bar control chart cannot be used. The solution consists in relaxing control limits so that only major shifts of the process mean are signalled. Several approaches based on a chosen time-dependent distribution model of the process are considered in the paper. Methods of the model identification and formulas for the construction of extended control limits are applied on two controlled quality characteristics, the planarity and length of moulded frames to show calculations for both one-sided and two-sided specifications. Moreover, calculation of capability and performance indices for the time-dependent models is discussed.