A look into the past: Flight 9 and Ship 36
Starship Flight 10 successfully launched a couple days ago, and it finally broke the streak of failures that Block 2 introduced, with both Booster 16 and Ship 37 making soft splashdowns in the water and reaching their test objectives. However, in this article we’re going to explore the past of the Starship program this year, focusing on what went wrong on Flight 9 and the causes of the ground explosion of Ship 36. That said, let’s begin:
A look into the past
If you’ve followed the Starship program recently, you’ll know that all the past Block 2 launches this year have failed in some way:
Flight 7 featured the debut of the Block 2 Ship, with Ship 33, which lost its engines at T+7m30s due to failures to the propellant feed system caused by higher-than-expected vibrations on ascent due to the new Ship design.
Flight 8 was the second attempt with Block 2 Ship, although the vehicle unexpectedly failed at around T+8m due to a Raptor failure that caused heavy damage in the propulsion system (at least 2 Raptors were literally blown out), causing the Ship to spin out of control and be destroyed.
Flight 9 finally reached SECO after an extensive testing campaign but lost control during the coast phase and was lost on reentry since it was uncontrollable. Due to this, the Ship skipped both the payload deploy and engine relight tests.
And then, Ship 36 unexpectedly exploded ahead of its 6-engine static fire test at Massey’s, pushing back the launch date for Flight 10 by almost 2 months.
So basically, 4 entire Ships were lost this year, although each of them provided critical data and allowed for many improvements on the vehicle.
However, the Booster side has been glowing with success this year: Boosters 14 and 15 were both successfully caught (although with 1 engine out on boostback for B14 and 2 on boostback and 1 on landing for B15) on Flights 7 and 8, respectively, and Booster 14 was even reflown on Flight 9, being expended in the Gulf of Mexico to test a descent with a higher angle of attack, as well as engine-out capability on landing burn. However, Booster 14 exploded a couple of seconds after the start of the landing burn.
What went wrong on Flight 9?
Let’s see what went wrong on Flight 9: we will start with a short but detailed recap of the flight, followed by the failure causes.
Flight 9 lifted off at 18:36 CDT on May 27th, 2025, featuring Booster 14-2 and Ship 35. This flight featured the first reuse of a Super Heavy Booster, which performed flawlessly during ascent, with all 33 engines running correctly.
Then, at about T+2m40s, the hot staging maneuver occurred, where Booster 14 shut down all but 3 of its Raptor engines, while the Ship ignited 6 out of 6 engines and separated away from the booster. Booster 14 then ignited back 13 of its engines for the flip maneuver, conducted in a controlled direction for the first time (thanks to steel plates welding shut some of the holes on the HSR that would’ve let the exhaust of the Ship out). Following the flip, the booster conducted its boostback burn, targeting the Gulf of Mexico, and the HSR was separated shortly afterwards. Booster 14 then began its descent experiments, which included having a higher angle of attack to allow more air to slow the booster down, therefore requiring less propellant during the landing burn.
During the landing burn, 12 out of the 13 planned engines ignited; however, the whole booster exploded a couple of seconds later, 1 km above the ocean surface.
Booster 14 exploding during the landing burn. Credit: SpaceX
Meanwhile, the Ship had ignited all of its engines and was ascending correctly: then, amidst joy and fear, Ship 35 reached SECO, cutting off the 3 RVacs first, followed by the 3 central engines. We saw some small “anomalies,” like hot spots on the Ship skirt and RVac nozzle, but those didn’t cause any problem.
Following SECO, the Ship was scheduled to enter a coast phase, in which it would’ve conducted the deployment of its Starlink simulator satellites and an engine relight test: the simlinks, scheduled to start deployment at T+17 minutes, weren’t deployed due to issues in opening the payload bay door. A few minutes later, the Ship was visually losing control, spinning around at a low rate, but still unacceptable… this had become clear by T+30m, when the commentators on stream confirmed the loss of control of the Ship, announcing they would skip the engine relight and wait for the Ship’s demise: this came at around T+46m, at 59 km, when Ship 35 was engulfed by plasma and broke up.
One of the last views from S35’s uncontrolled reentry. Credit: SpaceX
What were the causes?
Booster 14: SpaceX stated that B14 reached a peak angle of 17° in the angle of attack experiment and broke up due to a structural failure of the fuel transfer tube, a tube that runs from the common dome (between the fuel and LOX tank) through the LOX tank and finally to the engines, feeding fuel to the Raptors. This structural failure caused mixing of propellants and ignitions, blowing the booster up. To mitigate this issue, no change will be made on the remaining Block 2 boosters (Boosters 16 and 15, for Flights 10 and 11, respectively), but the angle of attack will be gradually lowered to avoid such failure. On Block 3 boosters, the transfer tube will be more heavily reinforced to allow even higher angles.
In every other aspect, Booster 14 performed flawlessly, and it probably would’ve splashed down softly if it weren’t for the AoA experiment.
Ship 35: the Ship’s failure is a little bit trickier: unlike what was previously thought, it all was caused by 1 single problem, and it happened on ascent: about 3 minutes after hot staging, high levels of methane were detected in the nosecone, suggesting some sort of leak; this continued for about 2 more minutes, causing a decrease in main fuel tank pressure and an increase in nosecone pressure. Starship’s automatic systems were able to compensate for the pressure decrease in the fuel tank, allowing it to reach SECO. After SECO, the leak of gaseous methane stopped, but there was still too much pressure in the nosecone, so the vents started to release this pressure by venting the GCH4 into space; however, these 2 factors caused an attitude error so great that the automatic system shut down the nosecone vents. This pressure was also pressing onto the mechanism for the payload bay door, which is the reason why the door couldn’t be fully opened.
The RCS thrusters worked for a bit and managed to get the attitude error below its limits, allowing the automatic onboard system to enable nosecone venting once again. However, roughly 40 seconds later, liquid methane was detected in the nosecone, suggesting a leak from the main tank. This leak of cold, liquid methane started lowering the temperature of critical components in the payload bay, prompting the automatic system to go into passivation mode, where all the propellant is vented, and the vehicle is rendered basically inert. This is also the reason for the loss of control, as well as for the skipping of the engine relight. Lastly, unable to put itself in the proper orientation, Ship 35 demised upon reentry.
The cause of these leaks was the failure of the main fuel tank pressurization system diffuser, a can with holes mounted on top of the main fuel tank, which receives methane gas from the Raptor engines for pressurization. Due to the failure of the diffuser, the gas was released into the payload bay/nosecone section instead of the fuel tank, which explains the pressure differences. The liquid methane leak is probably due to the failure of the diffuser, which allowed liquid methane from the fuel tank to get into the nosecone section.
Inside of S35’s payload bay section, full of LCH4 particles. Credit: SpaceX
To mitigate this on future flights, SpaceX redesigned the diffuser to better direct pressurized gas into the fuel tank and reduce the strain on the structure. They recreated the failure mode at McGregor and tested the redesigned diffuser to 10 times its expected service life, with incredible results!
So even if this flight didn’t reach all the objectives, it was still a major step forward in Starship development and informed the teams of a problem they didn’t know could happen.
What happened to S36?
Ship 36 exploding at Massey’s. Credit: D Wise for NSF
Following Flight 9, SpaceX was ready to go ahead with Flight 10 at a record-breaking turnaround time, with a NET date of June 29th. Booster 16 was tested in early June, while Ship 36 went later in the month at Massey’s: there, it conducted a 1-engine static fire, followed by an attempt at a 6-engine static fire… but during that attempt, at 23:01 CDT, Ship 36 violently exploded, destroying the vehicle and heavily damaging the test site.
SpaceX posted a small update the same day, stating that preliminary investigations suggested that a nitrogen COPV (Composite Overwrapped Pressure Vessel) failed in the payload bay, causing an overpressure that burst that section and the tanks, causing ignition of the propellants.
A COPV is a composite cylinder, which can store a high amount of gas in a small volume, thanks to its ability to withstand high pressures… but not always.
Along with the explanation of what went wrong on Flight 9, SpaceX also released a detailed explanation of the Ship 36 incident, stating that the nitrogen COPV failed due to unseen damage and that they now plan to operate COPVs at a lower pressure. They also added better inspections and proof tests prior to propellant loading, updated the acceptance criteria, developed a new NDT (Non Destructive Testing) method of looking for damage, and added external covers to increase protection and better detect damage.
Thanks for having read so far, and be sure to check our websites as more articles come out!
References
https://x.com/dwisecinema/status/1935571311771680949
