880514_en-US

liliconcom

Process capability

I have added the specs for all my resposnses. They are all within spec, but when I do a process capability my capability index plot is  all under 1 (Ppk value). 

 

I do not understand why. Even the Ccp are below 1. Do I need to increase the specs range. Can someone help me? 

Advanced Statistical ModelingRe: Process capability

Hi @donadristi,

 

It's hard to comment with just those images, can you provide the dataset? I can see from the images that your spread/variability is quite large on some of your process variables, and a lot of them have a mean shifted away.

 

Thanks,

Ben

Re: Process capability

Thanks for replying. Here is the data set 

Re: Process capability

Hi @donadristi, 

 

Can you provide the JMP data table with spec limits or have that provided with the excel? I can't perform the analysis without it.

 

Thanks,

Ben

Re: Process capability

Here you go. It is the JMP table with the specs in it

Re: Process capability

It isn't a good idea to have a target meeting the USL or LSL level exactly, ideally the target sits between the two. Your settings are far away from where you have set your spec limits. The real question to ask now is - are your spec limits appropriate for your conditions? How have you decided your spec limits? Is it something strictly set as a critical quality marker that your results have to fall within - if so then you may want to investigate why your responses aren't achieving those targets.

 

 

I think you had best review your process and get a better understanding of applying the SPC, there are some good resources here and here. I would also suggest looking at some of the other platforms in JMP like the distribution platform to show a quick overview of what your responses are vs. the limits you are setting.

 

 

Hope that helps!

Thanks,

Ben

Re: Process capability

Thanks a lot Ben. I will have a look and re do the analysis. 

Re: Process capability

hey @donadristi ,

the reason you are getting a low Ppk even when not having any out of spec points is because Ppk is based upon the fitted distribution of your data and not the raw data itself:

So here you can see that due to the variation in your data and the low sample sizes the tails of the fitted distribution are much larger than the raw data and cross over your spec limits.  What the fitted distribution is saying is that if you were to collect a lot more samples, you eventually would start to have out of spec values.  This is why fitting a correct distribution to your data is so critical for Process Capability.

 

Steve