My this thread is a long one……………………
Lets See,,,,,,,,,,,,,,,,,,Six Sigma…………………Do you think 99.9% is good enough? Well that means 3,500 babies will be dropped on their head this year, or several hundered planes will fall from the sky………
Sigma = square root of variance
variance = sum of squares/ n-1 (sample)
6*Sigma = 99.999999% yield
maybe this isn’t so important if you are making floor tile, but it is if you are making components to incubators for premature newborns.
Statisticly speaking……..The use of six sigma is more than the aforementioned equation, it is using all the tools to keep a tight/normal distribution centered within process specifications. Variation of a process is going to happen. How that variation exists within the process limits is important.
Example, if you plot a process on a chart (X,Y) the process will usually trend or vary from the target or mean. Now calculate sigma and apply control limits to the process say a range of 3sigma. Yes, by looking at a histogram it appears you have variation, however if your data stays within these controls you are well within six sigma. Now break out some more statistical tools. Cp and Cpk are used in a process to tell how a process distribution is shifting. This tool allows one to examine the process using other tools to determine if it is a people, machinery or measurement system issue. Or a combination. Shewhard charts like CUSUM, Xbar-R and XY are common tools for monitoring this process.
In general Six Sigma does not lead to more variation, it accepts normal variation and allows for processes to be kept in complete control……by catching unstable and trending processes before they are a problem……..
.
.