Computing six sigma when the USL & LSL are varying

P

patkim

When we have a situation where the USL & LSL are varying e.g. 40 - 30 for one type of activity and 35 - 25 for another type of activity.
How do we calculate the process sigma for such situation. The top level activity is still the same but two activities have different USL / LSL expectations.
Thanks,
 
T

Ted Schmitt

Re: compute six sigma when the USL LSL are varying

When we have a situation where the USL & LSL are varying e.g. 40 - 30 for one type of activity and 35 - 25 for another type of activity.
How do we calculate the process sigma for such situation. The top level activity is still the same but two activities have different USL / LSL expectations.
Thanks,

Shouldn´t your process be under control before you try calculating sigma? Reduce the variations first....
This may be off base, and a humble opinion since my knowledge in this area is very limited...
 
J

John Nabors - 2009

Re: compute six sigma when the USL LSL are varying

Patkim-

What is the unit of measurement? Millimeters? Grams? Light-years? I'm probably not smart enough to answer your question, but there are others here who are waaaayy smarter than me and can answer your question given enough information.

Good luck -John
 
D

D.Scott

When we have a situation where the USL & LSL are varying e.g. 40 - 30 for one type of activity and 35 - 25 for another type of activity.
How do we calculate the process sigma for such situation. The top level activity is still the same but two activities have different USL / LSL expectations.
Thanks,

I think you are asking which type of control chart you would use for such a situation?

You say the activity is still the same so I am assuming the standard deviation is the same and only the upper and lower specs are different. You should be using a Target chart (IX if single sample and X-Bar/R if you have 2 to 10 samples. Your plot points will represent the subgroup average deviation from the target value.

Hope this helps.

Dave
 
P

patkim

Unit of measurement is %. e.g. % Margin for profit.
One type of task it will be set as say 40 to 50 another say 45 to 55
The variation should be within say + / - 3 %

Having the datapoints, I am wondering how do we arrive at sigma value for such situation.
Thanks,
 

Jim Wynne

Leader
Admin
Unit of measurement is %. e.g. % Margin for profit.
One type of task it will be set as say 40 to 50 another say 45 to 55
The variation should be within say + / - 3 %

Having the datapoints, I am wondering how do we arrive at sigma value for such situation.
Thanks,

By calculating them separately? If you have two different sets of data (or even the same set with different pass/fail criteria for different situations) it's not helpful to lump them together.
 

Kales Veggie

People: The Vital Few
My suggestions:

Look into short run SPC.

Look into normalizing your data by creating a ratio of the actual / (USL-LSL).
 

bobdoering

Stop X-bar/R Madness!!
Trusted Information Resource
Unit of measurement is %. e.g. % Margin for profit.
One type of task it will be set as say 40 to 50 another say 45 to 55
The variation should be within say + / - 3 %

Having the datapoints, I am wondering how do we arrive at sigma value for such situation.
Thanks,

You could also normalize the two data sets by plotting variance from the mean, centered at 0 with limits of +3/-3. But, without knowing specifics I can't verify whether or not it is meaningful. If it is apples to apples, it might work, but apples to golf balls, well....:cool:
 

Bev D

Heretical Statistician
Leader
Super Moderator
my first instinct is that Jim's advice is correct, but then again I'm not really sure what you're asking. more detail as to specifics and perhaps even an example of the data will help us help you.
 
B

brahmaiah

Re: compute six sigma when the USL LSL are varying

Shouldn´t your process be under control before you try calculating sigma? Reduce the variations first....
This may be off base, and a humble opinion since my knowledge in this area is very limited...
Process sigma is independent of work tolerance.It will remain same even if you double the tolerance and hovers around the midlimit of the process setting and not around the work tolerance.
The same applys to the UCL & LCL of control charts.They are also independent of work tolerance unless you purposefully set them in line.Setting them exactly in line is a difficult exercise.
V.J.Brahmaiah
 
Top Bottom