Double-Retest of Nonconforming Material to Accept the Material

[J Allen]

When meeting the requirement for raw material validation, by sending out a sample of raw material to an independent lab for test, I encountered a rejection for tensile strength.
The results were low. The original material test report showed compliance.
I asked the independent lab if they could have made a mistake that led to a incorrect finding they said no but would be willing to test another sample.
Question: If a test report indicates noncompliance, can a double retest be sufficient to accept that material if retesting shows acceptable results. Do primes (Lockheed, Boeing, Northrop) accept double re-test to validate a failed test.

 

[Bev D]

It really depends. Ultimately your SQE will have to weigh on this one...

tensile testing can be affected by several measurement system variables. I suggest looking at the data from a continuous data perspective rather than a categorical (pass/fail) perspective.

Plot the two sample results on a chart with the specification line. Are the two results close to each other and the specification or are they very different form each other and far from the specification? Is there a lot of variation within the first and the second sample set? If the first and second sample are far from each other it is more plausible that the difference is due to some error (in sample prep or measurement) and a 'tie breaker' might provide convincing evidence - either way. If the two results are close to each other and the specification you will have difficulty as this is probably simple measurement error and/or part variation. It might be helpful if you provided your data...

Also consider the science of the test and the material...
How many samples are tested? what are the samples? Are they actual material or coupons? do they require some prep? (are they whole or 'cut'.) can you get multiple tests from one sample. Where did the samples come from within the batch? were they 'close to each other' in manufacturing sequence or randomly selected from across the lot? was the same measurement system and test party used for the two samples or are they different?

 

[Sidney Vianna]

Question: If a test report indicates noncompliance, can a double retest be sufficient to accept that material if retesting shows acceptable results. Do primes (Lockheed, Boeing, Northrop) accept double re-test to validate a failed test.

As Bev mentioned, you should involve the customer(s) about the acceptability of retesting before proceeding with the processing of this material. As it stands right now, the material should be deemed rejected as the (periodic? AS9100 mandated?) independent validation test rejected the material.

From a risk based thinking perspective, I would take action with the supplier and assess how much confidence you can have in their CoC's. I would assess if their material properties testing processes are adequate or not. If you can't assess that for yourself, and they have some kind of QMS certification (ISO 9001, AS9100, NADCAP, etc...) I would engage with them as well.

Assuming your customers are ok with the retesting, your supplier should bear the costs, in my opinion. Or, alternatively, sometimes your customer might have the test lab they could use for this test.

All of that (above) assumes the independent test lab you sent the sample to is reliable. Are they 17025 accredited?

 

[krayworth]

Were tests performed in accordance with a testing standard? If so, standards usually specify how many samples must be tested to gather the accurate data. I would also ask the test lab about the calibration of their equipment if the product failing the test was a surprise.

 

[Marc]

... Assuming your customers are ok with the retesting, your supplier should bear the costs, in my opinion. ...

Just a quick thought - Until it is known why the test failed it wouldn't be proper to assign a responsibility. It may be that the outside test lab is at fault (I'd take their offer for a re-test if free), and if so I would want to know what happened. If the customer tested wrong... Problem.

Food for thought with regard to lab tests. Unless I witness the test I'm a skeptic (must be my DoD work history...).

 

[J Allen]

It turns out the failure was due to incorrect calculations for the wall thickness. The lab confirmed the error and on re-test of the original and an addition (2nd sample) both were acceptable.
Thanks for your input.