Development of confidence intervals for process capability assessment in short-run manufacturing environment using bootstrap methodology

Statistical techniques can be helpful in quantifying process variability, analyzing variability relative to product requirements or specifications, and assisting development and manufacturing in eliminating or greatly reducing this variability. This general activity is known as process capability analysis, a major component of statistical process control (SPC). SPC methods have been accepted as an effective improvement tool in almost every type of business. One of the most interesting applications occurs in job-shop manufacturing systems, or, generally in any type of system characterized by short production runs.; Much of the industrial use of process capability analysis focuses on computing and interpreting a single value of the desired capability index. Practitioners often forget that those statistics are just point estimates, and, as such, are subject to statistical fluctuation. Consequently, it is proposed that reporting confidence intervals for process capability indices becomes the standard practice. This specifically applies to reporting process capability indices for short run manufacturing situations for which the production ends before the required minimum number of observations is collected. The bootstrap method provides a way to compensate for violated statistical assumptions required to determine process capability from small samples. This research describes the development of a bootstrap method for assessing process capability. It also provides a comparative analysis of the different bootstrap methods that can be used for process capability assessment. Sensitivity analysis and design of experiments are performed to estimate effectiveness of bootstrap methods for evaluating process capability for small sample sizes.