how tight is too tight? 2

[macmet]

I started a new job recently at a district energy centre. And one of the parts of my job is construction supervisor for all the various energy units. And part of that job is ensuring everything meets requirements based on various testing, notably pressure tests.

At my last job, in a different industry we would have to do pressure tests, but they weren't critical and if the system held pressure for a couple hours we were just gave them a pass. No trending of pressure, etc.

But at this new job, we test to the consultant's required specs. And I find most tests are technically failing, but I feel like the test results show the systems are good enough to a warrant a pass. But regardless, they didn't meet the spec, the test failed and they have to redo it. And of course I'm the one that gets to tell the contractors, with years of experience, to do it all over again because they dropped 1 or 2 psi outside the allowed range.

Does anyone else feel like specifications sometimes are a little too tight? I feel like if the consultants were on the site looking at the systems, they'd give them a passing grade, but they're not, and it's me who is obligated to say that the test failed or take on the responsibility myself.

How do you guys handle that situation? If I was in the contractors shoes I'd think my company is ridiculous for these tight specifications.

Maybe I'm just venting over my frustration with consultants who are sitting in an office 100 kms away (no offense to those who are consultants).

 

[TheTick]

And I find most tests are technically failing, but I feel like the test results show the systems are good enough to a warrant a pass.

Imagine repeating this in front of a jury.

 

[MechEng2005]

The contractors were aware of the specs prior to testing (I would hope!). Maybe "the contractors, with years of experience" should know what it takes to meet the specs.

Standards exist for a reason. In this case, it is not your job to pick them, only to test them and report the results. If they wanted you to have "wiggle room" they would have put words like "professional judgement" in the requirements, not numerical pressure readings.

You wouldn't want a heart surgeon to have gotten a grade of "not technically passing, but close enough because they were being picky" and sent on his way into your heart.

-- MechEng2005

 

[MiketheEngineer]

This is one of those non-gray areas. It either passes or it doesn't. You don't even have to quibble about it.

 

[macmet]

Ya, you guys are right.

Just feel like it's tedious. But you're right, and they're re-doing it on Monday.

Thanks for the input.

 

[SomptingGuy]

Fixed testing methods, with clear pass/fail criteria sound good to me. It takes away any possibility for someone to put you under pressure to pass something that shouldn't be passed.

- Steve

 

[moltenmetal]

Leakage at an unknown location is a bad thing: indeed that's why you're doing the testing in the first place.

Leakage at a known location might give sufficient reason to not require repeating a test. Leakage from a valve packing is such an example. Do you de-pressurize, fix the valve packing and then repeat the test, or do you pass it because you know where the leak was? That's a judgment call, and unfortunately it's your client's judgment rather than yours.

Yes, it's possible for a test criterion to be set too stringently by an ignorant client. Sometimes this requires that the system be broken into smaller pieces to permit testing, leaving the connections BETWEEN these pieces untested or incompletely tested. That is not in the client's interest.

It's also possible to do a hydrotest poorly enough, i.e. by not ensuring that the equipment is 100% full of water without any trapped air, for a test to pass despite significant leakage.

 

[macmet]

I just used the pressure the test as an example. We have other tests where I feel they're too stringent.

But as someone said, the test requirements are in the specs when the contractors bid. They should be aware and do the test in a manner they should pass.

I feel like their work is good enough to pass, maybe they just need to do conduct the test differently...

 

[cvg]

have you done a risk assessment in order to determine that "the systems are good enough to a warrant a pass"? Did the design engineer do it? It is possible that assessment might indicate undue caution regarding possible leakage, but how would you know otherwise?

 

[Twoballcane]

In the test world, the spec is the spec period. You can put in tolerances like 5% below and 20% higher or which way would give you more margin for the design you give higher percentage. However, one can ask for a waiver from the customer if the customer wants to take the risk and accept the data. But, when something fails later down the line, it will bite you in the butt and down grade your reputation. One can rationalize whatever they want to be right, but keeping to the spec will keep one honest (and sleep at night).

Tobalcane
"If you avoid failure, you also avoid success."
“Luck is where preparation meets opportunity”

 

[MintJulep]

the test failed and they have to redo it.

So no corrective action? Just repeat the test and hope it passes the second time?

And if it doesn't pass the second time just keep testing until it passes?

Most likely you have pass/fail criteria that are based on "design nominal" values, with some arbitrarily assigned range tack on.

Properly chosen pass/fail criteria need to take into account process variability and instrumentation error.

 

[Ron]

Read "TheTick"'s post again. If it doesn't sink in, read it again. It's the world I live in...you don't want to be there as a defendant.

Pass/Fail...really simple. No interpretation required.

 

[IRstuff]

I think the converse is true, i.e., the tests are not "too tight," the contractors are "too loose," and are hoping to slide by because they cheated on the construction, somewhere, to save a buck or two. THAT's what typically happens. Yes, contractors do know what's required, most of the time, and what's required is often too expensive and competitive bidding forces everyone into a gigantic gray lie, i.e., that they can meet all the requirements ata fraction of what it really costs.

TTFN

FAQ731-376
Chinese prisoner wins Nobel Peace Prize

 

[macmet]

I know you guys are right. It is nice to get some justification for making them redo the test.

But, if the test fails again, they'll have to do it a third time, isolating the suspective cause. And if it turns out that after isolating the valve the system passes, we'll probably just give them a pass on the whole system since it is a non critical leak.

I'd say it was pointless, but I think there is something to say about holding them accountable and making sure they know that we will not let just anything pass.

It's still tough sometimes when you feel like you're splitting hairs. But I do not want to be repeating to a jury that it was close to passing....

 

[dvd]

Showing my ingnorance here. Is the temperature monitored during a test? Wouldn't it be reasonable to have pressure change with changing temperature?

 

[macmet]

Temperature is noted but not monitored. It was one of a series of tests, in the same facility, and none of the other ones showed any significant change in pressure. The pressure of the failed test varied noticably more than the others.

 

[macmet]

I should have said temperature is noted but not recorded. It probably should become part of the spec... I'm not going to get into why it hasn't. That is another thread.

 

[IRstuff]

It's difficult to write specs to that level of detail. If we did, we'd probably never get anyone to bid, since there would be zero wiggle room, and there might never be anyone who could claim compliance, since there's rarely anyway to verify that ALL performance can actually be met under ALL conditions, or any combination thereof. We'd spend a year testing some silly little widget, and not really get any value-added.

TTFN

FAQ731-376
Chinese prisoner wins Nobel Peace Prize

 

[tygerdawg]

Posting here after reading the responses and I am in total ignorance of what precisely "the spec" states. However... I come from the manufacturing engineering world where everything involves rigorous testing to prove / disprove quality. At least, it should. One of the most common "failures of process" I have seen over and over again is having unreasonable faith in your test method and test devices to generate useful data. Are you absolutely certain that the pass/fail data you are generating is accurate? Gages and other data-producing devices should be appropriate to the task, appropriate to the data range, calibrated by 3rd parties, and run through some flavor of a gage repeatability & reproducibility test to prove their ability to produce good, useful data under whatever test conditions are required. Just a word of skeptical caution to make sure you are not generating false negatives or false positives, and causing a lot of unnecessary trouble.

TygerDawg
Blue Technik LLC
Virtuoso Robotics Engineering

http://www.bluetechnik.com

 

[PeterStock]

"They should be aware and do the test in a manner they should pass."

No, they should do the work such that the system passes because it is to spec. Not tweek the test so that the system passes.



Peter Stockhausen
Senior Design Analyst (Checker)
Infotech Aerospace Services

http://www.infotechpr.net

http://www.linkedin.com/profile/view?id=30064526&trk=tab_pro