bXJkZXV4QGludGVybm9kZS5vbi5uZXQ=@aviationusenet.com (JB) writes:

I said:

"spoke two"? Sigh; I am regularly surprised by my improving ability to
type major errors more and more frequently. If only I could market the
skill...

Anyhow, I asked about the type of failures [longitudinal vs
circumferential] but she had no ready data. From what I've seen posted,
SC[U]BA tank failures are of two basic types. The most common involves
a longitudinal crack starting at the neck and propagating downward; a
wedge size piece may come off or the tank can split open without ejecting
a chunk.

The other failure is called Sustained Load Cracking and that seems to
be confined to tanks made of 6351 alloy. It appears such can start
anywhere in the tank wall but that's not detailed. Nowhere have I found
a reference to 6351 being used in non-SC[U]BA tanks, such as Boeing
installed in this case.

But it appears the US is a lousy country to base data on, since we have
minimal regulation of the entire field. [Rather like investment bankers,
sigh.]

Not discussed but surely occurring is the user breaking off the
valve/regulator. It's rather hard to indict the tank itself for that
situation.

As JB has noted, there sure does not seem to be a lot of data on
circumferential cracking. The only mentions I find are on big boilers;
along a welded seam.

The 6351/SLC cracking history may be valuable in that as a result,
eddy-current inspection has joined hydrotesting for scuba tanks. (I don't
see a mention of it for DOT tests, but I'm not expert on such regs.)