Capability calculation on unequal bilateral tolerances?

L

lebelg

hello everybody!

I am confused as to how capability calculations should be made on unequal bilateral tolerances (such as : +0.002/-0.005).
I have been checking other threads for it but this is still unclear for me.

Could anyone please help?
:truce:
 

bobdoering

Stop X-bar/R Madness!!
Trusted Information Resource
You use the same calculations as equal bilateral tolerances. The calculation uses the tolerance spread and the process spread, Cp=(USL-LSL)/6sig, and is not concerned with the "target" value. Even Cpk uses the average, not the target: Cpk= the lesser of (USL-Xbar)/3sig , (Xbar-LSL)/3sig. Even though the tolerance appears to tell you to run to a target value below the "middle" of the specification, capability treats the tolerance spread as the entire dimensional opportunity for your process to run to, and your ability to keep your process within that entire dimensional opportunity.

One might argue, though, that if the use of an unequal bilateral tolerances is a desire to skew the location of the data to one side, that reporting Cp=(USL-LSL)/6sig alone would be sufficient to prove you are capable, because Cpk values will appear much worse if you do do not run to the center of the specification. The whole point of the Cpk calculation is to see if you are centered! Good luck! :cool:
 

bobdoering

Stop X-bar/R Madness!!
Trusted Information Resource
If you are curious as to the actual distribution, I use "Distribution Analyzer" from www.variation.com. It also provides Ppks from transformed data. It analyzes probably one of the widest varieties of distributions - and that is its focus. BUT, it is a toll, not an answer. You need to understand the origins of your distributions, and the underlying variations and their distributions - before making any conclusions.
 
Top Bottom