Cpk vs PpK

Process Capability vs Process Performance
 

On the top right hand corner of this website is a search engine. type in Ppk and boom all the info you could ever want to know is right there. Same thing for Cpk. Google will get you about 3.34 million hits in .26 seconds also

Hi John
Ppk tells u the actual Process performance. It contains both Special Cause as well as Common Cause of variation (also known as Standard Deviation).
Cpk tell u the potential of the process when all the special cause are removed from the current process. It tell the process Potential when all the special cause of variation is removed. It only includes Common Cause of variation for calculation.
Thus Ppk tell the actual performance of the period of which the data is calculated while Cpk tell the potential that one can achive when all Special Cause of variation is removed from the process
the formulas u can refer from any book
Broadle
Total cause of variation= standard Deviation
Common Cause = SMR2 (generally) but could vary with grouping variable
Cheers
Gaurav
  

Hi Gaurav,
One thing is not very clear for me.
I can’t see why, just by using the formula R/D2 (in order to calculate the sigma value), Minitab manages to go from long term to short term and thus “remove all special causes”. I can’t see the logic.  Any idea ?
Thanks
Vincent
 

Hi Vincent
I am not very sure of what Minitab uses. But I will give u the logic
For calculating Short Term standard deviation (SD) with-in batch SD is used. Now if you have 10 Batches and 5 samples each batch. The SD of each batch is considered for calculating Short term SD
SMR2 is generally taken if for every batch there is only 1 sample. Now the logic behind Short Term SD is somewhat similar to 5 point moving average. The special cause is suppressed when the moving average is taken. So if you plot a trend chart (Run Chart) and 5 or 8 point moving average of same data over it you can note the variation in two. The 5 or 8 point moving average will have less variation and special cause is averaged out by other data.   
Though its is pretty difficult to separate out all special cause of variation but this it is suppressed by this method
Hope it helps
for more detail u can contact me on my email
[email protected]
 
Cheers
Gaurav