What is a failure ( Quality Assurance matter )

[bernardgreen]

in over 20 thousand unit years of operation we have not had any failure

is a claim by the Quality Assurance Director for his employer's product.

What exactly does that mean in terms of reliability ?

Impossible to be one unit that operated for 20 thousand years without failure

Could be 20 thousand units each of which ran for a year without failing ( or before failure )

Or 240 thousand units that operated for a month before failing

After some pressure the actual figures were provided, about 5000 units in operation for between 2 and 6 years.

And then the defination of "failure" was discussed.

The Q A Director saw "failure" as the unit being unable to operate and needing to be replaced or repaired.

After some pressure it was admitted that a unit that was not able to operate as well as it did when new was not a failure but was "operating with acceptable degredation"

The discussion about how much degredation is acceptable is on going. The manufacturer's Q A director and the potential customer's Q A assessor have very different "opinions" about what level of degredation is acceptable.

 

[Himaginn]

Fair comments.
Does it also suggest that no actual testing was carried out, just the results of field experience. Which as you rightly pointed out could be assessed in multiple ways.

Also "acceptable degradation" could be quantified or qualified.

You don't say how long your particular unit was in use or how the difference in opinions on "acceptable degradation" was quantified or qualified.
Additionally, one could expect, with rigorous testing, that acceptable degradation is qualified after so many hours/years of ongoing use. After which the degradation becomes unacceptable.

 

[Dave54]

If you think about it, "acceptable degradation" applies to virtually everything. Or at least everything I can think of in the physical world. So in some ways the term itself is a bit meaningless. The point it starts to matter is where the degradation becomes unacceptable surely? That may or not be subjective.

It does remind me somewhat though of a conversation had some time ago with a manufacturer.
Them, "We're sorry, but that model is obsolete"
Me, "Not to me it's not"

 

[Himaginn]

Don't or didn't light bulbs have/had an acceptable degradation quality statement on the packaging.

 

[Dave54]

Don't or didn't light bulbs have/had an acceptable degradation quality statement on the packaging.

Never noticed.
I must have a look. I've got some here in boxes.
Do incandescents degrade as far as light output is concerned? I suppose they must, but I don't think I've ever noticed that much difference, but then again, you don't often change working bulbs for the same type
The compact fluo type definitely change colour / brightness over time. Seen that several times when replacing with the exactly the same make / type.

 

[Brigadier]

No apparent factoring in of servicing / programmed replacement of components either.
Is it "fair" to say that a car that does 10k p.a. and is serviced every year "has never let me down", and is therefore more reliable than the unserviced-and-flogged car than develops a fault?

 

[Shutpa]

Around 10 years ago I had an appointment with a consultant regarding a back problem and after extensive investigations including MRI, X-ray, and Ultrasound she came up with the diagnosis that my back problem was the result of acceptable degradation. "Absolute nonsense" I said. "Now, if you really mean fair wear and tear, why the **** don't you say that?" She replied, "I say acceptable degradation, because I like to keep up to date with the latest ideas. And acceptable degradation is today, my best description of your back problem. But, of course, it is fair wear and tear".

Despite all that acceptable degradation/fair wear and tear, for some reason, my back no longer gives me problems.

 

[bernardgreen]

Does it also suggest that no actual testing was carried out, just the results of field experience.

If the manufacturer has carried out testing for degredation then these results have not been made available.

No apparent factoring in of servicing / programmed replacement of components either.

The item does not have any servicable or ( economically ) replacable parts. Failure or un-acceptable reduction in performance would require replacement. As mentioned it is what reduction in performance ( degredation ) can be considered as acceptable before the item has to be replaced.

A similar scenario happened some years ago when the performance of an electro-chemical actuator was under investigation. The manufacturer would not accept a unit had failed if it produced the required movement no matter how long it took from end to end and how much extra power was needed. The concept was good but the chemistry was flawed.

But the classic marketing was from 1998 for an electronic module

of the more than 1500 units produced in 1997 only 6 were returned by customers for repair or replacement

very good. A failure rate of 0.4 % but there was an important bit of the truth missing.

of the more than 1500 units produced in 1997 less than 300 were sold

.

Sales in 1998 were very low and by 2000 the company had ceased trading with massive debts and many unhappy customers

 

[Himaginn]

I suppose it could be argued that in days gone by reliability and reputation were important factors when making cost-conscious decisions.
But it takes a brave adviser to stand up in front of a the Board, the purchasing officer, or whoever, and argue that the CBA, including any NPV, suggest a more expensive purchase based on longevity, reliability or reputation is the better option.
Of course, the IT industry has long been an advocate of built in redundancy with the Redundant Array of Inexpensive Disks, (RAID).
And critical equipment in other spheres has this built in redundancy.