NASA has fingered design failings by SpaceX in a much-delayed report (PDF) on the 2015 explosion of a Falcon 9 on its way to the ISS.
The 28 June 2015 launch followed six successful flights to the orbiting outpost (two on the original Falcon 9 v1.0 and four on the v1.1 incarnation).
Unlucky number seven began normally enough, leaving launchpad LC40 at 14:21 UTC and powering through the first 139 seconds of flight without issue.
At 139 seconds, the second stage of the vehicle experienced what SpaceX called in their July 2015 report an "overpressure event", resulting in a bad day all round.
The Falcon 9 exploded but the Dragon cargo freighter survived and continued transmitting data up until it disappeared over the horizon.
It eventually scattered freshly laundered astronaut pants over the Atlantic as it impacted the ocean.
Almost three years after SpaceX's investigation, NASA has now published its own take on the matter, including some additional detail and recommendations for the commercial space upstart.
It is not clear why the summary report has taken so long to be made public.
NASA's Independent Review Team (IRT) took a look at SpaceX's own investigation and the available evidence and came up with the conclusion that "a helium filled composite overwrapped pressure vessel (COPV) within the Stage 2 LOx tank had become liberated, and had hit the LOx tank dome causing it to rupture", pretty much agreeing with SpaceX.
The IRT investigated two other scenarios, involving the LOx transfer tube that runs to the stage 2 engine, but couldn’t create thermal conditions as spectacularly destructive as punching a hole in the tank itself.
Before SpaceX can pat itself on the back for solving the mystery nearly three years ahead of NASA, the IRT has some stern words.
The use of industrial rather than aerospace-grade steel in rods holding the COPV in place was a design error on the part of SpaceX and a direct cause of the accident.
The IRT also voiced concerns about the telemetry spewed to the ground by the "full thrust" incarnation of the Falcon 9 with data related to the accident lost due to latency caused by buffering in the second stage flight computer.
The report ends by noting that all findings were closed or "mitigated" prior to the last flight of the Falcon 9 v1.1, January 2016's Jason-3 mission.
A Falcon 9 v1.2 exploded on the launchpad later in 2016, destroying LC40 and $200m of Israeli satellite in a face-plantingly foolish test involving fuelling and firing up the engines while the payload was attached. In that incident, the Falcon blew up before the nine Merlin engines got a chance to start.
A slew of successful launches since then means there won't be anything to worry about when Musk starts playing with really big rockets in 2019. ®