Scrapped an Alum 80 6351 tank today

[Diving Dubai]

The safety problem is that the VE inspection process is not believed to be perfect. Some cracks will be missed.

In my opinion as an ex NDT/NDI specialist (certified to Level III in Eddy Current Flaw detection and 4 other disciplines by the US ASNT and the UK PCN standards), Eddy Current (or VE) should be perfectly acceptable for this inspection

However the equipment used for tank inspection is "primitive" and "low tech" and the training given by PSI etc is laughable. 2 hours to learn VE inspection!! It's not even funny. It's not the operators fault, they know no better. They've been trained by a supposedly reputable agency and are led to believe that the training is sufficient. Right!

The fact they require "false positives" to be visually inspected is a joke. There is no such thing as a false positive. It either a defect or its mechanical artifact that's allowable. VE indications can be open to interpretation, but that's part of the skill - knowing what the machine is telling you, of having the proper equipment where you can filter out errant signals (using a properly developed inspection procedure)

Is VE the best method to inspect? While yes it can detect surface and subsurface defects, depending on the equipment and technique, it has it's limitations.

A crack running parallel to the probe path (circumferentially) will be much much harder to detect than a crack running at 90 degrees to the probe path.

SLC can and does propagate from a material mid point, that is near the centre of the material. The Deeper it is, teh harder to detect especially with the material thickness of Ali cylinders

A better way would be to use ultrasonic inspection (UFD)where you can easily see defects all the way through the material. You can have a simple automatic rig where is spins the cylinder and the probe scans the full length and gives a go/no go report.

We did this in the 90's with aircraft wheels and trained operators over 1 full day to do it

Guess what? UFD could detect pitting corrosion too on both Ali and Steel cylinders (VE on ferrous material is a PITA unless very skilled). The test equipment would cost around the same as a decent compressor

Unfortunately the industry has gone cheap. The training is inadequate, I feel for operators who believe (rightfully) that they've been properly trained, and the test equipment recommended would be better used as a doorstop

 

[2airishuman]

The reality is there has been no catastrophic failures of properly inspected 6351-T6 aluminum tanks since the initial 18 month eddy current protocols were first implemented around 2002, nor the final 5 year eddy current requirement at re-qualification adopted by the DOT in January 2007. Thats 0 out of 25,000,000 tanks.

I found a reference to a rupture in June 2007 but it's not clear if it was in the USA or not.

Most of the originally manufactured 6351 cylinders are out of service at this point. There's a PSI whitepaper on this that does confirm the 25 million number, but that's the estimated number manufactured between 1971 and 1990. The federal government estimated that appropximately 7 million were in service as of 1994. Some were manufactured under special permits that have now expired. Substantially all those in firehouse or industrial SCBA service have been replaced because the SCBA systems they are part of have been long since upgraded. Few if any dive shops have any left -- even in Minnesota, where there is no salt water and where dive shop cylinders get much lighter use than in, say, Florida -- the shops I've spoken with have gradually replaced their cylinders and have none left manufactured prior to 1990.

Aluminum cylinders in any sort of daily service do not last indefinitely and are typically removed from service after 10-15 years because the cost of reconditioning exceeds the cost of scrapping and replacing, with tax consequences also playing a role.

The PHMSA report linked above also indicates that visual eddy tests miss about 3% of flaws in 6391 cylinders and falsely condemn about 4% of cylinders inspected.

How many are left? Well, who knows, I'd guess under a million in the U.S. -- mostly SCBA cylinders in industrial standby or emergency egress service plus some SCUBA ones belonging mostly to former divers who have them in the back of the garage or closet somewhere. Maybe a few CO2 fire extinguishers.

One way or the other, it's not 0 out of 25,000,000, it's more like 0 out of 500,000 with a good deal of that 0 being due to luck.

 

[2airishuman]

In my opinion as an ex NDT/NDI specialist (certified to Level III in Eddy Current Flaw detection and 4 other disciplines by the US ASNT and the UK PCN standards) Eddy Current (or VE) should be perfectly acceptable for this inspection

However the equipment used for tank inspection is "primitive" and "low tech" and the training given by PSI etc is laughable. 2 hours to learn VE inspection.. it's not even funny. It's not the operators fault, they know no better. they've been trained by a supposedly reputable agency. Okay then.

The fact they require "false positives" to be visually inspected is a joke. There is no such thing as a false positive. VE indications can be open to interpretation, but that's part of the skill - knowing what the machine is telling you, of having the proper equipment where you can filter out errant signals (using a properly developed inspection procedure)

The PHMSA report showed that even the most expert operators found false positives with eddy current equipment 7% of the time. The cylinders used for the study were destructively evaluated (half using an etching process, half using EDM slices) to confirm that the false positives really were false.

Is VE the best method to inspect? Well yes it can detect surface and subsurface defects, depending on the equipment and technique, but it has it's limitations. A crack running parallel to the probe path (circumstantially) will be much much harder to detect than a crack running at 90 degrees to the probe path. SLC can and does propagate from a material mid point, that is near the centre of the material.

A better way would be to use ultrasonic inspection (UFD)where you can easily see defects all the way through the material. You can have a simple automatic rig where is spins the cylinder and the probe scans the full length and gives a go/no go report.

We did this in the 90's with aircraft wheels and trained operators over 1 full day to do it

Guess what? UFD could detect pitting corrosion too on both Ali and Steel cylinders (VE on ferrous material is a PITA unless very skilled). The test equipment would cost around the same as a decent compressor

Unfortunately the industry has gone cheap. The training is inadequate, I feel for operators who believe (rightfully) that they've been properly trained, and the test equipment recommended would be better used as a doorstop

The same PHMSA report showed that ultrasound testing was less reliable than VE.

 

[JamesBon92007]

So probably like most of you friends of mine that know I dive occasionally give me gear they are not using any more. Last week I was given several Alum 80 tanks all out of hydro. So one of them was from 1988 and was a 6351 alloy. To save any headaches i just scrapped it. The tank looked horrible and the valve was garbage as well. I got .20 per pound for 37 pounds so i got $7.40 for the tank. I though i would have gotten a little bit more for the tank but that was it. For me that i own more than 10 tanks i had no desire to cut it and make it a bell or anything like that.

I traded one for a Trek bicycle frame with a guy who sells crap metal. I think he got about $12 for it. Now if I can just find a woman who wants a pink Trek bicycle.

 

[Diving Dubai]

The PHMSA report showed that even the most expert operators found false positives with eddy current equipment 7% of the time. The cylinders used for the study were destructively evaluated (half using an etching process, half using EDM slices) to confirm that the false positives really were false.

During the development of an inspection procedure, destructive inspection of a component to evaluate the results i.e. to confirm fully your results is perfectly acceptable. I've done it lots myself. I use to work in Aerospace both civil and military.
My point is that once a procedure is finalised it should not detect false positives, of if it does they are either filtered out, or appear as definite indications as such, meaning the signal is different to a real defect

I note that the report considered and "Expert" as the following:

Expert (designated as “X”): the inspector is either the system vendor, a
representative of the manufacturing company,

or someone with years of experience in the specific technique The latter is open to interpretation. Is their experience with the Specific Inspection technique OR in VE or UT? The latter is better.

I'll read the report in detail with interest (Thanks for sharing) but quick scans have left me shaking my head - red dye penetrant went out with the arc for instance.

The same PHMSA report showed that ultrasound testing was less reliable than VE.

A probe in a fluid filled wheel. Of course it was junk. That contraption would have problems finding the valve hole at the top of the tanks.

Trust me on this. With both VE and UT I've developed techniques to find defects as small as 0.020" Sometimes smaller depending on the criticality. So proceedures do can work, but they require the correct equipment and procedure for that component

As an aside every time you fly, the aircraft you fly on (whether on a Boeing or Airbus) will have been inspected using at least 1 procedure I wrote, so you'd better hope I knew what I was doing

 

[DA Aquamaster]

I found a reference to a rupture in June 2007 but it's not clear if it was in the USA or not..../

/....One way or the other, it's not 0 out of 25,000,000, it's more like 0 out of 500,000 with a good deal of that 0 being due to luck.

You keep shrinking your numbers, but whether it is o in 25 million, 0 in 1 million, or 0 in 500,000, the fact remains that there has been no documented catastrophic failure of a properly inspected 6351 T-6 tank since the eddy current requirements were put in effect.

Whether you use a denominator of 25 million, 1 million or 500,000, the numerator is still 0 and luck has nothing to do with it.

The 3% or 4% of missed cracks are by definition small and not observable, and the DOT found that it took 7 to 9 years for a crack to propagate to the point of failure, so even if 3% or 4 % of cracks are missed, they will be caught at the next 5 year re-qualification before the tank catastrophically fails, which was the basis for the DOT only requiring a 5 year interval for eddy current testing.

Local shops backstop this DOT required process with an annual VIP. Magnified inspection of the neck and shoulder will again catch a crack before it poses a risk of a catastrophic failure. I think that's also important as I noted one particular tank that passed the eddy current portion of the requalification but then actually leaked at the end of the hydro test. Obviously, a non detected crack propagated to the point of leaking during the test. I am also aware of one other tank just back from re-qualification that was reported by a shop owner to leak on it's first fill. These were both "failures" but not catastrophic failures.

I'd prefer RINs to do the hydro test first, followed by the visual and eddy current inspections increase the potential for a crack to be observable. However, most of them do the visual and eddy inspection first. Some do it as a follow on to the old practice of looking in a tank before you waste time on the hydro test, while some others do it that way because it was listed in that order in the regs.

Whatever.... in any case a visual plus or visual eddy inspection by the local dive shop provides an additional level of safety - over and above that provide by the DOT required VE testing, which again has resulted in 0 failures over the last 10 years, not counting the 5 or so years prior to that when eddy current inspection was required by Luxfer.

Again, luck has nothing to do with it.

 

[2airishuman]

Bear in mind that all this work was done in the early 1990s when the problem with SLC first came to light and while there were still an economically significant number of these cylinders in use. The whole point of the rulemaking was to preserve a reasonable degree of safety without having to prematurely remove some millions (depending on whose number you believe) of 6391 cylinders from service. So they put together a procedure using 1990s technology that an average cylinder technician could follow reasonably accurately.

Their testing showed that they got 7% false positives and something like 3% false negatives using the procedure as a whole the way they expected it to be implemented in the field.

Time goes by, there aren't many of those cylinders left. Ya sure, with today's technology we could probably come up with a better testing methodology, either using 2017 eddy current test gear rather than 1994 eddy current test gear, or by using 2014 ultrasound gear rather than 1994 ultrasound gear.

But we'd be money ahead and just as safe to take the remaining cylinders to the scrapyard. IMO of course.

 

[2airishuman]

The 3% or 4% of missed cracks are by definition small and not observable, and the DOT found that it took 7 to 9 years for a crack to propagate to the point of failure, so even if 3% or 4 % of cracks are missed, they will be caught at the next 5 year re-qualification before the tank catastrophically fails, which was the basis for the DOT only requiring a 5 year interval for eddy current testing.

I know that is the story. I don't believe there is enough evidence to support the claim that the only cracks that are missed are too small to propagate to the point of failure within the 5 year interval.

 

[DA Aquamaster]

It's worth noting that tanks do fail. However, the properly inspected 6351-T6 tanks that have catastrophically failed since 2007 have done so because they were dropped. There was one incident at a shop in north Florida where an aluminum tank being filled with O2 was dropped, and another in south Florida where an individual dropped a tank in his garage. That can happen regardless of the alloy being used.

 

[DA Aquamaster]

I know that is the story. I don't believe there is enough evidence to support the claim that the only cracks that are missed are too small to propagate to the point of failure within the 5 year interval.

The evidence is the total lack of catastrophic failures.

Your argument is much like the argument for trickle down economics. Some unemployed guy on food stamps living in a run down single wide trailer will claim that it could happen, despite it never happening in the last 100 years

It's hard to argue with someone using that kind of logic.