The effect of pre-control on small lot size manufacturing
Wayman, Michael J.
MetadataShow full item record
Traditional SPC techniques, such as the Shewhart control charts, require a minimum lot size quantity of 100 pieces in order to obtain sufficient data for their use. Small lot size manufacturers are generally unable to use these techniques because of this requirement. As a result, many small lot size manufacturers still rely upon 100% part inspection by each machine operator or by a separate department. This study examined the use of a process control method known as pre-control as compared to 100% operator inspection in a small lot size manufacturing environment. Although its overall effectiveness as a process control method is uncertain, pre-control has potential in small lot size manufacturing because it only requires 5 pieces to qualify the process. Quality characteristics examined in order to compare the two methods included dimensional variation and departure from the target value. The procedure used was a time series design. An initial sample of 129 parts was machined on a numerically controlled lathe using 100% operator control as the method of process control. After training the same operator in the rules of pre-control, another sample of 129 parts was run on the same machine using pre-control. The 129 piece part samples were made up of similar parts which each contained a single, common feature. Dimensions produced on this common feature were inspected and recorded by the operator. After analyzing the data obtained, the results of the study showed pre-control to be more effective than 100% operator inspection in guiding operators toward a desired target dimension. As quantified by the Taguchi loss function, tangible economic loss occurs as a part dimension departs from its ideal target dimension. The results of the study were inconclusive in regard to the effect of pre-control as compared to 100% inspection on dimensional variation of parts produced. Subtle differences in the overall shape and material removal methods of the different parts used in the 129 piece pre-control sample distorted the overall calculated dispersion. However, analysis of subgroups which were made up of similar shaped parts showed no significant difference in the amount of dispersion present in parts produced by each method.