Ethical question regarding tolerances on test plan 4

[mdgates]

I'm new to engineering, so I'm looking for the opinions of some more experienced engineers regarding the following ethical concern, which I will phrase as a hypothetical.

Suppose there's an aerospace company that's trying to get a product into the market. There's a specification that defines the required performance, and a test plan that's been approved by the would-be customer as a means of demonstrating compliance with the spec. The spec requires the product perform at a temperature of 100°C, so the test plan calls for the unit to be placed in an oven at 100+/-5°C for three hours (+/- 5 minutes), then removed from the oven and operated within five minutes.

The engineering team discovers that the product will not function satisfactorily at 100°C. Engineer A suggests setting the oven to 96°C, and lowering the temperature of the test stand room to improve the product's chances of passing the test. Engineer A further suggests leaving the unit in the oven for exactly 2 hours and 55 minutes. He further suggests building a "quiescent oven" to use instead of the standard convection oven that the company has always used to test similar products.

The question is: Is Engineer A's proposal ethical?

It will be Engineer B's responsibility to execute the test. If Engineer B's boss says that Engineer A's suggestions should be followed, what should Engineer B's response be?

 

[KENAT]

Did the customer approve the test plan, or is it internally defined.

Just FYI if I had specked something to perform at 100°C then I'd expect that the minimum temperature used during testing would be 100°C.

Posting guidelines faq731-376

http://eng-tips.com/market.cfm?

(probably not aimed specifically at you)
What is Engineering anyway: faq1088-1484​

 

[GregLocock]

I don't believe this is a real case. Is it from a textbook or homework? If so the irony meter just went off the scale.


Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[mdgates]

Greg,

Sorry to peg your irony meter, but I did tell you I'm new at this. It's not from a textbook, but it is against company policy to post details of a real case.

Matt

 

[JohnRBaker]

Technically, it would appear that the test was being conducted within the letter of the spec, i.e., within the tolerances given, but only if the specs were understood to be the TOLERANCES for the test environment. However, you already know that your company is on shaky ground since the product is just barely able to pass the test, even by shading the tolerances toward the best possible result, so if later some of the items delivered did fail and someone audited your testing procedure it could be argued that your tests were invalid if what the customer was relating to you were NOT TOLERANCES for the test environment, but rather the TOLERANCES for the functional environment which the product was going to subjected to in production. That is, the expectation was that the testing was done to assure that the product would actually work in an environment where the temp could vary from 95°C to 105°C and could be expected to be in such a working environment for time periods of 2:55 to 3:05. If THAT's the case, then you have failed to verify that your product will perform as the customer had expected it to.

So I guess to put it another way, unless you can get them to put it in writing that the tolerances given are for the testing environment ONLY and are not being given as an actual RANGE across which they're expecting the tests to be conducted, you should be prepared to accept the consequences that may occur if you're assumptions are later questioned as whether you've meet the letter or the intent of the contract.

John R. Baker, P.E.
Product 'Evangelist'
Product Engineering Software
Siemens PLM Software Inc.
Industry Sector
Cypress, CA
Siemens PLM:
UG/NX Museum:

To an Engineer, the glass is twice as big as it needs to be.

 

[woodman88]

Depends on what you report. If the report states that the unit was heated in a "X" oven at exactly 96 degrees for 2 hours and 55 minutes before testing. Then the client should be able to accept it. But are you able to exactly state that the temperature never was below 95 degrees in the oven? Also what information can you supply that comfirms that you were able to. I don't think the oven manufacturer would state that it would.

Garth Dreger PE - AZ Phoenix area
As EOR's we should take the responsibility to design our structures to support the components we allow in our design per that industry standards.

 

[SAITAETGrad]

Interesting question.

Apparently an aluminum manufacturer decided they could save money by adjusting alloy content from nominal values of the aircraft sheet metal specification. If you have ever examined the alloy content ranges on such a specification, you know they are quite wide. With better specification control, they realized a more consistent product and increased profits. Meanwhile an aircraft manufacturer that had developed a massive library of strength, fatigue, and other material allowables for the aircraft sheet metal specification started to notice unexpected issues in the field since all their test data was based on the previous alloy properties. Needless to say this caused quite a row...

In your case, the stakes are likely nowhere similar but there are lessons to be learned. Honestly, I would say that Engineer A's approach is more resourceful than unethical and this is more a training/knowledge issue than anything else.

Arbitrary specifications such as this 100°C requirement come about general because past experience has shown them to be reliable. We can "sharpen the pencil" or otherwise try to squeeze into the requirement but in doing so, you really are shaving away from the reliability of the system.

You can go too far the other way as well, adding conservative requirements. One time I was told by an experienced certification guy that he always likes to have the techs drag flammability samples around on the shop floor to make sure they have a good amount of dirt and oils to simulate degradation of the samples in service before testing. Well, that skews things away from the requirements which again have been established as reasonable.

Stick to the nominal.

 

[JohnRBaker]

A good analogy for any sort of tolerance scheme is to look at it as if the numbers represented a 'budget' which can be 'spent' across the lifecycle of a product's production. If the first guy in the loop 'spends' all of the 'money', that doesn't leave much flexibility for the rest of the downstream contributors.

John R. Baker, P.E.
Product 'Evangelist'
Product Engineering Software
Siemens PLM Software Inc.
Industry Sector
Cypress, CA
Siemens PLM:
UG/NX Museum:

To an Engineer, the glass is twice as big as it needs to be.

 

[jgailla]

I think it's baloney to game the specs like that, but if a contractor did it I wouldn't have a leg to stand on.
Tolerances are for things that can't be controlled.

 

[woodman88]

Then again you may have a good client who needs the unit to work at 80 or 90 degrees maximum. So they added a safety factor to 95 degrees as a mimimum and spec that 100 degrees plus/minus 5 degrees is the testing they wanted done. Someone really needs to talk to the client and I nominate "Engineer A" first and the "Boss" second.

Garth Dreger PE - AZ Phoenix area
As EOR's we should take the responsibility to design our structures to support the components we allow in our design per that industry standards.

 

[stevenal]

100+/-5°C means the oven is held to 95 to 105 as the thermostat cycles. If you run it at 96, what is the tolerance on that? Drop anything over a degree before the stat kicks on and you'll be under the 95. What's the tolerance on the thermostat itself? Will it always kick on at precisely 95.1? Every time during the three hours? And whats the accuracy and tolerance of the thermometer you are monitoring with? Tested and traceable to NIST? Crowd the edges and the nice loose tolerance tightens.

 

[mdgates]

SAITAETGrad, you say to stick to the nominal, which is probably a good idea. But what if my boss says to game the tolerances? Can I convince him that he's wrong? Do I have an ethical obligation to stick to the nominal, or to inform the customer that we intend not to?

I have a feeling that if my boss was intent on hiding this from the customer, he'd write the test report himself. It's not good when your boss feels he has to do your job for you. On the other hand, gaming the spec creates a culture of dishonesty that could ripple through all the testing we do.

Woodman, if the customer were on board with this, my conscience could rest easy. But I think it's unlikely that "A" or "boss" would agree to tell them, and it's not "B"'s role to inform them.

 

[lacajun]

Unless the "would-be customer" agrees to Engineer A's changes, I would stick with the "agreed to test plan" to demonstrate performance compliance. Never change agreements w/out the customer's full knowledge and consent. Further, I would tell the customer the part doesn't function per the "agreed to test plan" at 100C/3 hours and determine a course of action to meet their requirements, unless they agree to change the test. It's a test to determine performance in its intended purpose, which is eliminating your product.

All of the specifications I've seen and used were stated as XXX +/-Y, which is the mean and its tolerance. The customer expected that and not XXX-Y +/-Y. There is a difference because you're stating different means. If your customer expects XXX +/-Y and routinely gets XXX-Y +/-Y, he will eventually tell you your process is out of control and look for another supplier, if he understands manufacturing. For example, some of the parts I've worked on the in past now have a spec of XXX +/-0. They now have to be dead on. XXX-Y +/-Y is far from dead on in some parts. In some other parts, we moved the mean to gain other improvements.

Is it unethical to be ignorant? No. It is unethical to be devious? Yes. Only you know what's going on completely and can provide your ultimate answer.

Engineers should conduct themselves as professionals. You always treat your customer fairly and honestly. That's the right thing to do. If you're looking for guidance here, you know something stinketh and you probably know what it is. Whatever you do take the high road and protect your customer as well as your working relationships. It's nothing personal. It's good business. Through good business and professionalism, you will gain a good reputation and those are hard to get.

Pamela K. Quillin, P.E.
Quillin Engineering, LLC

 

[dik]

I would suggest that if the product tested was the absolute worst example of the product to ever be produced... you should have no problem... If, however, there was a failure and the testing were audited... you're baked... at whatever temperature! Could be something else... totally unrelated that fails and brings down whatever part you are testing... it might then be 'put under a microscope'.

If life safety were involved... talk to your lawyer... you could be criminally responsible.

I don't agree with the methodology, but to the letter of the spec you have achieved a passing grade... this time!

Dik

 

[GregLocock]

Apologies, there was a phrase there that sounded so textbook-like that I couldn't believe it was real.




Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[drawoh]

mdgates,

You may need to read carefully a contract. If you guarantee that you will pass the tests you describe, then you must test the device at 105[°]C for 3[ ]hours and 5[ ]minutes, or whatever the worst case is.

--
JHG

 

[MiketheEngineer]

Call your lawyer now or later!! You pick..

 

[dwallace1971]

Is this a for a civilian or military application? If this is a civilian application, how are you obtaining FAA approval?

"On the human scale, the laws of Newtonian Physics are non-negotiable"

 

[MintJulep]

The 100 C in the spec is a minimum.

The test procedure is not compliant, and should not be approved.

Engineer A is a scum bag.

Your warranty costs will be high.

 

[IRstuff]

I agree with MJ. The product specification is 100°C, so knowingly testing to less than that as an acceptance test is unethical. The bottom line is that the product design is marginal, and needs to be redesigned. Barring that, you can ask for a waiver and see if the customer will accept that. I don't see anything that even suggests that it will work at 96°C; it just only happens to be within measurement tolerance of the lower limit of the test procedure.

The customer should not have accepted those limits as valid. Assuming ±1°C tolerance on the thermometer, the minimum acceptance test temperature should have been 101°C.

I should also warn you that marginal designs may have other flaws. We came across a sensor that was specified for operation up to 100°C, and it operated to spec at that temperature, but failed to operate at 90°C.

TTFN
faq731-376