Six Sigma
This is mainly dirrected to those whom have a slight understanding of what six sigma is "supposed" to indicate. I’ve just gone through several days of class and have two (and I think a third is on its way) “Green Belt” project, after completion of one of those, I will be a certified Six Sigma Green Belt. Whatever that means, at least it looks good on a resume.
Now here are my three biggest problems with the system:
a) Six sigma does not mean failures beyond six standard deviations from the mean. It means about 4.5 because of a built-in 1.5 standard deviation fudge factor.
*hand waving* “You see it has been observed that systems tend to drift 1.5 standard deviations after initial gauge R and R measurements”.
b) The only thing people pay attention to is one metric, or variations of it. That metric is CPK or PPK. Which is basically the probability of any failure occurring in your process. It takes some numbers from empirical data and others from a FMEA. That number from the FMEA is how many things can possibly go wrong with the process. That means the more possible failure modes you can make up the better your CPK/PPK becomes without actually improving your process at all.
c) A high degree of hand waving and dependence on pre-made "tools" that ignore important hypothesis testing methods from statistics. This means the likelihood of you creating a statistic with no meaning is very probable.
Now here are my three biggest problems with the system:
a) Six sigma does not mean failures beyond six standard deviations from the mean. It means about 4.5 because of a built-in 1.5 standard deviation fudge factor.
*hand waving* “You see it has been observed that systems tend to drift 1.5 standard deviations after initial gauge R and R measurements”.
b) The only thing people pay attention to is one metric, or variations of it. That metric is CPK or PPK. Which is basically the probability of any failure occurring in your process. It takes some numbers from empirical data and others from a FMEA. That number from the FMEA is how many things can possibly go wrong with the process. That means the more possible failure modes you can make up the better your CPK/PPK becomes without actually improving your process at all.
c) A high degree of hand waving and dependence on pre-made "tools" that ignore important hypothesis testing methods from statistics. This means the likelihood of you creating a statistic with no meaning is very probable.
0 Comments:
Post a Comment
<< Home