Capability Study - in the beginning of your career what should you have known about the tool

[arnetjen109]

I have been tasked to complete a support work instructions on how to complete a capabiltiy study, how to interpret the study, and all the intricate details. This will possibly be a 15 page report which is supposed to help new quality engineeres or any fresh graduate who may be working on a capability study.

So here is my question:

What did you completely misunderstand when it came to capability studies when you were fresh and new to the quality department?

Example: when i barely had a year under my belt I didn't understand the capability study should have included subgroups that mattered to my process and that it was bound by time so I should have retrieved my samples through a lengthy time period....

 

[optomist1]

how to correctly interpret...the results...as yo will likely see from other posters, the studies have limitations...secondarily knowledge as to proper execution and use of the same...btw whose software are you using?

 

[Miner]

The concept of rational subgroups and how to determine whether your subgroups are rational. Also, not mixing different process streams and how to handle situations with multiple streams.

 

[Emmyd]

How to check your data for normality. This was a big thing when I first started doing capability studies (at least for that set of customers.

 

[AMIT BALLAL]

The misunderstandings I had during my initial days (with regard to SPC):
1. Cp, Cpk or Pp, Ppk calculation is SPC (Control charts to be used first to achieve stability)
2. Pp, Ppk is for short term SPC study
3. Any data can be used for Cp, Cpk / Pp, Ppk calculations (Normality is required for traditional calculation of Cp, Cpk or Pp, Ppk indices. And if data is not normal, either transform data / check underlying distribution and use the calculations accordingly.)

There is still a confusion across the industry on the usage of Cp, Cpk or Pp, Ppk and AIAG SPC manual also recognize this. Hence I use and recommend calculation of both Cp, Cpk and Pp, Ppk indices.

BevD and Miner are experts on this field, they can give you better insights on SPC.

 

[arnetjen109]

how to correctly interpret...the results...as yo will likely see from other posters, the studies have limitations...secondarily knowledge as to proper execution and use of the same...btw whose software are you using?

We use minitab so the main part of the report will be to provide execution instructions for minitab but I will have to include an excel file or maybe the equations so an engineer would understand where the statistic is coming from.

 

[Welshwizard]

Hello Arenetjen,
As well as the study being planned completely at the outset (already mentioned here) that any reliance and action on the reporting of numbers when the process data is not stable is wishful thinking. If a process is unstable in the past how can you extrapolate into the future with any confidence?
That its not a given that a process only contains random variation without intervention, but rather that it's an achievement and a continual challenge.

 

[arnetjen109]

How to check your data for normality. This was a big thing when I first started doing capability studies (at least for that set of customers.

normality first!!! yes. This is a great example of what I am talking about. It is second nature for me. I will prob process map the whole cape study and then write the instruction.

 

[arnetjen109]

The misunderstandings I had during my initial days (with regard to SPC):
1. Cp, Cpk or Pp, Ppk calculation is SPC (Control charts to be used first to achieve stability)
2. Pp, Ppk is for short term SPC study
3. Any data can be used for Cp, Cpk / Pp, Ppk calculations (Normality is required for traditional calculation of Cp, Cpk or Pp, Ppk indices. And if data is not normal, either transform data / check underlying distribution and use the calculations accordingly.)

There is still a confusion across the industry on the usage of Cp, Cpk or Pp, Ppk and AIAG SPC manual also recognize this. Hence I use and recommend calculation of both Cp, Cpk and Pp, Ppk indices.

BevD and Miner are experts on this field, they can give you better insights on SPC.

I appreciate the information. I know for sure that even in my company there is misinterpretation on CP/PP and it is used erroneously to identify a "Good Process"

 

[Miner]

Dr. Donald Wheeler has written many excellent articles on SPC. There are a number of them that address the correct way to calculate the standard deviation and whether the data must be normal.