What are the accepted standard rules for SPC analysis ?(7 points rising, WECO, etc.)

[dacksc]

I would like to recommend that our firm use a standard set of SPC rules (7 points rising, etc.) for Continuous Improvement and 6 Sigma in the future. The problem as I see it is that our software, books and training material from various different efforts has caused several sets of 'SPC Rules' to be written into work instructions. I would appreciate any help you could give pointing me towards an accepted standard.

thank-you

 

[Marc]

I see this a lot. I saw it most recently at a Ford manufacturing facility. I wondered if the folks who put all the stuff together knew anything about SPC, actually. Most of the stuff was very obviously reactionary to a specific where they included the info in manufacturing documents. They also had some of the rules in their SPC training program (which was weak, to be nice). At the Ford facility I was at there was extremely ineffective communication (and many walls - manufacturing did not like and communicated poorly with with the training folks. In fact, I would characterize the relationship as hostile.

My suggestion:

1. Remove such rules from all manufacturing documents.

2. Have a standard training document.

3. Adequately train personnel.

Remember that SPC concepts are not 'easy' for many people and take time to learn and correcty recognize and apply - weeks and months, not hours or days. If you expect a 1/2 hour or 1 hour session to 'be all they need' you will fail miserably.

Don't expect to get consistent interpretations from different sources. They're sorta like Demings 14 points - quite widely 'ihnterpreted'. Establish your 'intrpretation' (rules) in your training document.

 

[mclayton2000]

Nelson's Rules work well many times.
But it depends....
Its really an economic issue.
What is the cost of responding to an alarm?
Can you afford many false alarms just to make sure that you catch an occasional shift or drift or shock? If so, use ALL the rules. If not, use only the first one or two. Each case is different. One solution is to standardize on all of Nelson's (or WECO) rules per any SPC textbook...but create different levels of action for each rule. Rule One alarm, stop machine and adjust or find root cause of problem. Rule Two alarm, send email to maintenance to watch machine closely. Etc etc etc.
There are drifting machines, stable machines, cheap ones, expensive ones, so the costs are different and the sampling rate is different and the actions are different. Rules are just gambling aides.