Calculating Pp for Unilateral Tolerances

U

Ualfliers

I have a similar problem that involves PP & PPk - for unilateral tolerances.
> I have 30 wait times with USL of 30. I have averaged the 30 wait time at
> 8.76. For Pp would it just be 30/by 3.4 and for Ppk for unilateral
> tolerances - Ppu (upper) = (USL-Xbar)/(3*stdev) In my case would
> be(30-8.76)/(3 * Stdev) will this get me to my Ppk for Unilateral tolerance.
> My instructor said follow the model for Cpk for unilateral tolerance. I have
> been struggling with this, I do not do this for a living and work with no one
> that does this. Thanks
 

bobdoering

Stop X-bar/R Madness!!
Trusted Information Resource
Re: How to Calculate Process Capability for Unilateral Tolerances?

Can you post your 30 wait times? It would be handy to see what you are up against. Really, you have a de facto LSL of 0, but as a physical limitation the data will respond differently than a specified limitation (lets say a value of 15). The close you get to zero, the more non-normal it is. You are far enough from 30 that you may start to see some non-normality. I recommend doing a distributional analysis to see what you data's behavior truly is. It will NOT be "half of a normal (Gaussian) curve" - although that seems to be the estimate your instructor is leaning toward.
 

Bev D

Heretical Statistician
Leader
Super Moderator
Re: How to Calculate Process Capability for Unilateral Tolerances?

when you have a unilateral tolerance with a natural bound such as Cycle time there is no rational Cp value or calculation.

This is one of those times - it appears - where your instructor isn't thinking; he or she is applying these indexes by rote. You migh tconsider gentling challenging them to understand How it cold be done and even what value add it might have in understanding the process
 
U

Ualfliers

Here is my problem - I am having a hard time calculating Pp for a Unilateral Tolerance my teacher says not calculate/report the Pp index (not concerned with centering). What should I do, stuck on this.

Data:Wait times21198612998131177101448109915996109Upper spec is 30 minutes7985
 
U

Ualfliers

Re: How to Calculate Process Capability for Unilateral Tolerances?

Here is the info I was given below - I need to calculate Pp for Unilateral Tolerance, not sure how to do it.

Data:​
Wait times​

2​

11​

9​

8​

6​

12​

9​

9​

8​

13​

11​

7​

7​

10​

14​

4​

8​

10​

9​

9​

15​

9​

9​

6​

10​

9​
Upper spec is 30 minutes​
7​

9​

8​

5​
 

Jim Wynne

Leader
Admin
Here is my problem - I am having a hard time calculating Pp for a Unilateral Tolerance my teacher says not calculate/report the Pp index (not concerned with centering). What should I do, stuck on this.

Data:Wait times21198612998131177101448109915996109Upper spec is 30 minutes7985

First, it appears from the wording of your question that you're trying to do something that your teacher told you not to do. In any event, the string of numbers you posted has no delimiters, so we can't tell where one values ends and the next one starts--you should separate them with commas (or dashes, or something).

You might want to scroll down to the bottom of the page where you'll find a list of related threads. There's been a good deal of discussion about calculating capability indices with a unilateral tolerance.
 
U

Ualfliers

Here is my problem - I am having a hard time calculating Pp for a Unilateral Tolerance my teacher says not calculate/report the Pp index (not concerned with centering). What should I do, stuck on this.

Wait times 2,11,9,8,6,12,9,9,8,13,11,7,7,10,14,4,8,10,9,9,15,9,9,6,10,9,7,9,8,5

With a USL 30
 
U

Ualfliers

Re: How to Calculate Process Capability for Unilateral Tolerances?

Hi Bob, reposted it with the numbers being seperated, did not realize they would be clumped together.
 

bobdoering

Stop X-bar/R Madness!!
Trusted Information Resource
Re: How to Calculate Process Capability for Unilateral Tolerances?

I analyzed the data using my preferred approach, which is to find the best fit distribution. If need be, then, calculate (or allow the software to calculate) the capability using transformation. You can see attached that the normal curve p value is only .4822, whereas the Johnson family distribution p value is .988 - making it a much better model for the process. The normal distribution overstates the capability quite a bit. See attached data.
 

Attachments

  • Wait times.docx
    92.7 KB · Views: 155

Jim Wynne

Leader
Admin
Here is my problem - I am having a hard time calculating Pp for a Unilateral Tolerance my teacher says not calculate/report the Pp index (not concerned with centering). What should I do, stuck on this.

Wait times 2,11,9,8,6,12,9,9,8,13,11,7,7,10,14,4,8,10,9,9,15,9,9,6,10,9,7,9,8,5

With a USL 30
You have (as I assume you're aware) responses to your other post with the same question. Best not to do duplicate posts.
 
Top Bottom