Sampling plan for destructive tests on small lots

C

chandr

Hi all

I was wondering if someone could maybe help. We are a medical device manufacturing company that needs to revise our sampling plans on our finished products. At the moment we have a AQL of 1% (which we were advised might be a little high), by reducing the AQL the amount of samples for testing increases.

We are currently using ISO 2859-1 for our sampling plan and according to this if we use Inspection level II and an AQL of 1% we need to send 13 samples for destructive testing, however the problem with this is that we make batches of quantity 20.

Does anybody have any advise?

Thanking you in advance. :)
 

Mike S.

Happy to be Alive
Trusted Information Resource
With small lot sizes like this you are going to have a rough time with most "AQL-type" sampling plans.

If you have a history of lots passing, you could perhaps move to reduced sampling at 1% AQL which would cut your sample quantity back a lot.

You could raise your AQL -- like to 1.5. 2.5, or 4%. If you can do additional non-destructive inspection earlier in the process perhaps you can justify this.

How did you come up with the idea of the current sampling plan? Was it a customer requirement?
 

Michael Malis

Quite Involved in Discussions
We are currently using ISO 2859-1 for our sampling plan and according to this if we use Inspection level II and an AQL of 1% we need to send 13 samples for destructive testing, however the problem with this is that we make batches of quantity 20.


One thing you could use is the C=0 or Z1.4 plan.
This will allow you to "work back" and find a sampling plan that corresponds to what you want to sample.

Good Luck,
Michael
 
Last edited by a moderator:

David-D

Involved In Discussions
I don't have Z1.4 or the ISO readily available at the moment but I'll make a couple of suggestions based upon MIL-STD-105 (which is what they were both copied from). I would readily agree with the above poster that with a demonstrated history of successful acceptance (10 sequential lots/batches), you should consider changing to reduced sampling. This would change your sample size from 13 samples per lot/batch down to 5 samples (accept on 0 defects).

Alternatively, you could consider changing your inspection level from General II to something "lower" and of higher risk (and sample size) such as General I or the Special Inspection Levels. In this manner, you would maintain your 1.0% AQL but accept more risk (usually both on supplier and customer sides) with a smaller (closer to A) code letter; unfortunately, in your situation that won't help as you'll still be pointed down to 13(0-1).

Finally, I'd suggest considering using variable sampling techniques if your testing allows. In this manner, I'd suggest using Z1.9 (MIL-STD-414) where you are basically calculating a process capability from the average and standard deviation of the results to assess acceptability. In that case, you’d probably be testing 3, 4, or 7 samples; and if you went to reduced, you’d do 3 samples. I’d strongly recommend you look at your current process and its process capability before you go ahead with this approach as if you have strange distributions (not normal, etc.) this may cause you issues.

Whatever the approach, I’d strongly encourage you to coordinate with your customer(s) in advance so that they’ll fully understand and be prepared for any change on your side and understand that you’re not just arbitrarily changing your quality/acceptance.

Best of Luck,

David
 
C

chandr

JHi all

Thank you so much for your replies. I truly appreciate this.

The current sampling plan was designed (made up by) our DSO who is no longer with the company. We manufacture the products and at each stage have a inspection step as well as a final inspection and QC step.
After that we send some samples for destructive testing. We have now noticed however that we have been using the sampling plan incorrectly and only sending 3 / 20 samples for destructive tests as what these tests test can not be visually seen.

I will look into all the suggestions.
 

Mike S.

Happy to be Alive
Trusted Information Resource
How many 3/20 tests have you done and what were the results? If you always pass the 3/20 testing that could be a clue.

What does your customer say/require as far as destructive testing? Anything?

What do your engineers say are critical characteristics, and can they be measured non-destructively?

Are the destructive tests looking at special processes like welding? If so, perhaps you can control the special processes well enough that 3/20 or even fewer destructive tests are needed to verify the proper execution of the special processes. Or maybe test coupons can be used in lieu of destroying good product.

The more info you provide the more help the Cove folks can give, and this group is a heck of a diverse and knowledgeable group!
 

Bev D

Heretical Statistician
Leader
Super Moderator
Some really good advice here.

destruct testing isn't usually handled with standard attribute testing plans especially with small lot sizes. it does sound like you have a special process and the medical device standards do cover process validation as an appropriate substitute for statistically based acceptance sampling. (and depending on what kind of device you are producing I would say that 1% AQL is way too high; remember the AQL is the ACCEPTABLE defect that you would ship 95% of the time if it were present.)

The best advice given here so far:
talk to your Customer and or regulatory reviewer.
Look into special process validation and control.
If you have a very successful sampling record (zero defects found out of hundreds of tests) you have a much better position from which to transition to the special process approach.

and of course the more details you can provide the better advice we give.
 
C

chandr

Hi Mike and Bev

Thank you so much!

@ Mike, we have always done 3/20 tests as our final pass/fail criteria. If any one of these 3 samples fail we reject the batch. We have previously failed 92 / 691 batches tested for various reasons. Unfortunately we cannot measure the critical characteristics non destructively. These characteristics include burst pressure, compliance, deflation time, bond strength, coating coverage, manner of burst etc. . The processes were unstable at first, but we have implemented and improved a lot. Our clients do not require destructive testing, but as we need to supply them with a product that passes certain criteria, we need to do these.

@ Bev, we manufacture class III medical devices. we are looking at reducing the AQL just not sure how low to go. Our regulatory reviewer has had no input as to what requirements should be met.

Thank you so much!
 

Bev D

Heretical Statistician
Leader
Super Moderator
These characteristics definitely fall under the so-called special processes clause. As such you can institute controls for the process validate that they work and then perform minimal sampling. This sample size will not have to conform to published sampling plan standards.
 
Top Bottom