Thanks Henry, for the excellent writeup. The root cause is extremely
disappointing, and their negligence was criminal. Some people choose to
ignore test reports that have bad news - especially if it is 'too late' to
recover schedule or cost. Someone probably saw they all passed proof
testing and thought it was good enough, despite the failed test rings.
Thanks,
Lars Osborne
On Sat, Sep 3, 2016 at 3:51 PM, Henry Spencer <hspencer@xxxxxxxxxxxxx>
wrote:
On Sat, 3 Sep 2016, Michael S. Kelly wrote:
The S-IVB explosion was due to a design flaw of major proportions. The
helium pressure tanks were mounted inside the LH2 tank, to allow more
helium to be stored. The helium tanks were made of titanium, and the
explosion was due to failure of the girth welds from hydrogen embrittlement.
I regret that I have to contradict Mike on some of the details here. :-)
The best account of the S-IVB-503 explosion is Alan Lawrie's article in the
Dec. 2007 issue of Spaceflight (p. 470); in particular, he talked
extensively with Don Brincka, who was the Douglas test director in the
blockhouse when the stage went up, and was heavily involved in the
investigation.
The S-IVB had both "cold" helium tanks inside the LH2 tank, and "ambient"
helium tanks around the engine on the base of the stage (easily visible in
many photos). It was in fact one of the ambient tanks that ruptured to
cause the accident. (This is confirmed by other shorter accounts,
including p. 186 of "Stages to Saturn" [Bilstein, NASA SP-4206].)
The examination of the debris found that three of the helium tanks (all
ambient tanks) had split into their hemispheres -- in fact, only one
hemisphere of each tank was found. (The test area was an old gold-mining
site which had been reworked extensively to recover gold residues from the
tailings, so the ground was uneven, with ponds and lakes everywhere.
Various bits of missing debris, including the missing hemispheres, almost
certainly were underwater; Navy divers, hampered by bad weather and thick
mud, unfortunately didn't find much.) Two of the found hemispheres, from
tanks 65 and 66, were blackened on the exterior, but the hemisphere of tank
69 was bright and shiny. 69's rupture had come first, so its hemispheres
had departed the vicinity *before* the fire and explosion, hence the clean
surface. Its lower hemisphere, the one found, had gone down into the flame
trench; its upper hemisphere had gone clear through the stage and out the
top. 65 and 66 had come apart during the blast.
The problem wasn't that the tanks were titanium; the problem was use of
the wrong welding wire for the girth welds (combining two hemispheres into
a single spherical tank), and general neglect of quality control.
The tanks were titanium 6Al 4V, a fairly standard alloy, and the weld wire
should have been the same alloy. In fact, the shipping container said that
the shipment of three spools of weld wire was all 6Al 4V. However, if you
read the accompanying test certificates carefully, two of the three spools
were Ti 55A -- straight titanium -- not 6Al 4V. The result was welds that
were 30-40% under strength to start with; all the tanks were successfully
proof-tested at 1.6x operating pressure, but some of them were probably
very close to failure then. Later, as the tanks were pressure-cycled --
and those tanks were cycled many times -- needles of titanium hydride
gradually formed at the interface between the two metals, making the
interface layer brittle and further weakening the tanks.
Aggravating the problem, the tank subcontractor's QC paperwork wasn't
being done properly. Worst of all, Douglas wasn't paying attention. Each
batch of new tank welds was accompanied by a weld test ring, which went out
to an independent test lab. Test ring #1 had been fine and had been used
to qualify the subcontractor. Test ring #2 (for tanks 04-60) had *failed*
testing; Douglas claimed ignorance of this until the investigators pointed
out that they had received and paid an invoice for the testing! Test ring
#3, for tanks 61-75, also failed testing, and again Douglas hadn't noticed.
There might have been more of a public fuss about this, had the explosion
not been totally overshadowed by the Apollo fire a week later.
The fix was to move tank manufacture in-house at Douglas, tighten up
quality control in general, and inspect existing tanks to weed out the ones
with bad welds. Plus repairing the test stand and replacing the S-IVB
stage. Plus repairing assorted damaged cars -- the explosion had happened
in late afternoon, just as the day shift was leaving the test site, and
there had been a number of minor car crashes due to distracted drivers!
Henry
