| This research investigates the effectiveness of the simple linear regression control chart as an alternative approach to both the defects per unit (DPU) and defects per million opportunities (DPMO) control charts, which have numerous limitations when used in a job-shop manufacturing environment. Using the linear regression concept to transform Opportunities and Defects attribute data to variable type, this research also provides a performance comparison of the DPMO control chart to other commonly used variable control charts, specifically the individual data (IR), the moving average (MA), the exponentially weighted moving average (EWMA), the cumulative sum (CUSUM), and the proposed MA and LR control charts.; The guidelines used in this research for selection of a control chart were based on performance factor and simplicity. The results revealed that, although the added step of data transformation of Opportunities and Defects data was required, the method used offered the advantage of flexibility and the ability to expand usage of data to many other control charting methods; moreover, this method demonstrates superior performance over the DPU and DPMO methods. From the performance standpoint, the recommended MA, EWMA and MA/LR control charts outperformed the DPMO control chart in their ability to quickly detect small to moderate upward process shifts as well as downward shifts (i.e., process improvements). Although the proposed linear regression control chart did not outperform its contenders and could not be recommended for use as a sole method of process control, it proved to be particularly useful when paired with the MA control chart. This combined MA/LR scheme was more sensitive to detect small process shifts than the MA chart itself. |
