Help with concept of PPM in SPC

[lmoreno12]

First i would like to apologize for my poor english. I have a doubt about the term PPM used in SPC (control charts, capability analysis), for example: I know that a process with Cpk > 1,67 we have a ppm below 0,5. But this PPM is about parts with some kind of defect (under or above Engineer Specification), or just above or under Control Specifications (Talking about Control charts). If my question isn't very clear please let me know.

 

[Jim Wynne]

First i would like to apologize for my poor english. I have a doubt about the term PPM used in SPC (control charts, capability analysis), for example: I know that a process with Cpk > 1,67 we have a ppm below 0,5. But this PPM is about parts with some kind of defect (under or above Engineer Specification), or just above or under Control Specifications (Talking about Control charts). If my question isn't very clear please let me know.

Welcome to the Cove.

Cpk is an indicator of how well the process is centered between the specification limits, not the control limits. The upper and lower specification limits should be inside the upper and lower control limits. It's possible to have a process that's in a state of statistical control (all points inside the control limits, for example) but producing only nonconforming product.

 

[lmoreno12]

I understand that but my question is about what my customer will accept, i had some instructors that told me, the ppm in SPC is related to points or values out of control limits, and others that says that ppm is related only with specifications limits. To make my point of view more easy: If i had a process under statistical control and with a capability above 1,67. My customer will accept the quantity of problems that has relation to the Cpk that is demanded? ( in this question my customer demmands a cpk minimum in my process above 1,67), if i make a relation with Cpk x PPM a cpk of 1,67 allows me to have about 0,5 ppm in this case, what is tormenting me is, this ppm is about pieces with real problem?

And thank you i always read a lot of topics in here but never posted anything, but this doubt is killing me.

 

[Jim Wynne]

I understand that but my question is about what my customer will accept, i had some instructors that told me, the ppm in SPC is related to points or values out of control limits, and others that says that ppm is related only with specifications limits. To make my point of view more easy: If i had a process under statistical control and with a capability above 1,67. My customer will accept the quantity of problems that has relation to the Cpk that is demanded? ( in this question my customer demmands a cpk minimum in my process above 1,67), if i make a relation with Cpk x PPM a cpk of 1,67 allows me to have about 0,5 ppm in this case, what is tormenting me is, this ppm is about pieces with real problem?

And thank you i always read a lot of topics in here but never posted anything, but this doubt is killing me.

Regardless of specified PPM levels, there is no guarantee that a customer will accept defective products. That said, PPM concerns itself with defectives--products that have one or more defects or nonconformities, or "pieces with real problem" as you put it. Cpk, being a measure of how well the process is centered between the specification limits (not control limits) is sometimes used to predict PPM levels, the theory being that if some Cpk level x is maintained, some number y of defectives can be expected (if not actually realized). That number y may then be used to calculate PPM.

 

[lmoreno12]

Regardless of specified PPM levels, there is no guarantee that a customer will accept defective products. That said, PPM concerns itself with defectives--products that have one or more defects or nonconformities, or "pieces with real problem" as you put it. Cpk, being a measure of how well the process is centered between the specification limits (not control limits) is sometimes used to predict PPM levels, the theory being that if some Cpk level x is maintained, some number y of defectives can be expected (if not actually realized). That number y may then be used to calculate PPM.

Thats the word i tried to remember " Expected", so using the cpk to predict the ppm is expected some "pieces with real problem", considering that the histogram or the softwares (Minitab for example), uses USL and LSL for predict the ppm using cpk right?

I really thank you for helping me, and again sorry for my english.

 

[Jim Wynne]

Thats the word i tried to remember " Expected", so using the cpk to predict the ppm is expected some "pieces with real problem", considering that the histogram or the softwares (Minitab for example), uses USL and LSL for predict the ppm using cpk right?

Correct.

I really thank you for helping me, and again sorry for my english.

You're welcome.

 

[Steve Prevette]

There are 2 difficulties with the commonplace "PPM" declarations that may lead to confusion:

1. Conversion of CPK to PPM is only valid if the data are normally distributed.

2. The Six Sigma folks for mysterious tribal reasons assume a "1.5 sigma shift" and remove that when calculating PPM from CPK.