Capability Analysis with various subgroups

Hello guys.



I want to perform a Capability analysis. The product is measured from production and quality control.

In production we usually have more measurements is characteristics like diameter, wall thickness etc in comparison with QC.

For example, see below.

• Order 1 (measurements):
  3 production & 1 in QC.
  
  
• Order 2 (measurements):

5 production & 2 in QC

• Order 3 (measurements):
  6 production & 1 in QC.
  
  
• Order 4 (measurements):

4 production & 1 in QC

• Order 5 (measurements):

8 production & 2 in QC

• Order 6 (measurements):

5 production & 4 in QC

• Order 7 (measurements):

4 production & 3 in QC

• Order 8 (measurements):

4 production & 1 in QC

In production we usually have >= 2 up to 20 measurements.
In Quality Control we usually have >=1 up to 6-8 measurements.



I use Minitab for the calculation:


For production I select the subgroups by order

For QC are the data capable to produce a valid result?

Thank you,

Since you are self proclaimed ‘new to statistics’ the one big caution I give to newbies is that you just never grab an existing pile of data and start torturing it with math. A basic fundamental premise of statistical analysis is that the data are representative of what you want to understand. You will almost always get an answer from a mathematical manipulation of data but that doesn’t mean it is the right answer for the physics of the situation. (A great quote here is that no amount of statistical manipulation can save a flawed experimental design).

Without seeing the data and understanding the process(es) and the subgrouping scheme there is little effective advice to give. I might think that if you use all of the production and QC data together and calculate the average and standard deviation of that data AND you plot the raw data in time sequence of manufacture (not testing) against the specification limits then you might have a reasonable idea of the process capability. But I could be wrong.
Some questions:

• What is the characteristic?
• How is it measured? Did it pass an MSA?
• What is the process that creates the characteristic?
• What is the sample size and frequency of each ‘subgroup’?
• What happens when a production sample fails? Is there 100% screening and removal or rework before QC? (If so then you shouldn’t use the QC measurements)
• What is the QC sample size and frequency? How was it determined? AQL? RQL? Percentage of the lot.

Then I would probably have other questions.

As you no doubt are aware statistical analysis isn’t as simple as finding a some data and selecting a formula to manipulate…

Hello Bev D

Thank you for your comments.

What is the characteristic?
Its the charecteristic from the product (tube) length, diameter, weight etc.

How is it measured? Did it pass an MSA?
with micrometers. Msa grr was marginal

What is the process that creates the characteristic?

The process creates alla the above

What is the sample size and frequency of each ‘subgroup’?

depends from the amount (kg) per order
until 1.000 kg
for example
production will measure each batch (5 batches so 5 measurements)
qc control it will take one sample so 1 measurement

In case of 2.000 kg 10 measurements for production and 2 for qc

this is the typical sampling but sometimes depends the customer and the norm that it's product it falls

the combination varies



What happens when a production sample fails? Is there 100% screening and removal or rework before QC? (If so then you shouldn’t use the QC measurements).

depends how bad is the product..in some cases we re work the product on other cases we scrap it

What is the QC sample size and frequency? How was it determined? AQL? RQL? Percentage of the lot.

see above but percentage is the best approximate

I am waiting for your response.

best regards

These are very low sample sizes - but that is a different discussion.

You can use all of the measurements (failed and passing) from the production data. Do NOT use any QC data as it is from a censored population and will not give you any idea of the capability. You will need around 25-30 batches. Do not censor or cherry pick which batches you use: use the last 25-30 batches produced. Calculate the capability using the Sd of all of the measurements ie the long term capability. There is no statistical manipulation at this point to calculating short term capability - the sample sizes are just too small to rely on any calculation without seeing the data in time series.

This calculation will not have any statistical or practical precision - it will only give you a rough idea of the capability; it’s a staring point.

If you would like you can post your data here (the actual measurements not the statistical summary) and we can provide better advice..

New to statistics said:

What is the process that creates the characteristic?

The process creates alla the above

Click to expand...

Is the tubing seamless (i.e., drawn) or rolled and welded?

Miner said:

Is the tubing seamless (i.e., drawn) or rolled and welded?

Click to expand...

Seamless, Does it affect that much?

Bev d

QC measurements

Bev D

Production measurements

New to statistics said:

Seamless, Does it affect that much?

Click to expand...

It helps us better understand the process and the potential sources of variation. Knowing it is seamless means that many of the dimensions will be autocorrelated and that changes will occur over much longer periods of time. Likely sources will be changes in material and changes to process settings.

So.. that's affecting also and the analysis (calculations) ?
I mean.. you will do something different?
If you have an examples i would love to or more information about similar operations