Looking at the video with LOX tank buckling in mind, I've noticed a couple more things that seem to fit this scenario. Right before the failure, the visible ring of vapor condensation around the Dragon seems to go asymmetric. A sort of bump appears on one side of the Dragon, as if the upper stage is deflected and air flow is being pushed to one side more than the other. Secondly, there is a small protrusion on the side of the upper stage. As the vehicle performed the roll maneuver in the seconds before failure, this protrusion moved mostly into shadow, with only a small speck still illuminated. And then, quite suddenly, for a split second, this illuminated spot disappeared and then came back into view, as if it had briefly moved into shadow. The only reason I can think of for this momentary disappearance is that the upper stage orientation is shifting its angle to the sun, which could be due to aggressive steering and/or flexure of the vehicle with rapid orientation changes of the upper stage. I think maybe the vehicle had been experiencing pretty extreme flexure and strain in the moments before failure. Shear winds could also be a factor.
Even if vehicle bowing had been detected in telemetry, it may not necessarily be immediately flagged as the cause of failure if similar bowing had been observed in previous flights without failure. And strain gauges may not detect the buckling unless the gauge happened to be right near the point of initial failure.
The good news is that, if wall buckling is the cause, then the fix may be relatively painless (compared to other imaginable scenarios). I suspect this is not a quality issue or a manufacturing issue. I suspect the dynamic loads just happened to be more extreme than what the models predicted. The problem might be addressed by several conceivable approaches: tighten upper wind constraints; adjust flight profile to be more shallow with less aggressive steering to minimize the angle of attack and associated asymmetric loading (although I think I've heard that the profile is already pretty shallow, so this may not be a practical solution); reduce the maximum cargo capacity; reduce loads with more engine throttling, limiting maximum loads to a lower level; for "work in progress" tanks, reinforce the skin in this area with additional stiffening (this fix would be a more major change, but not insurmountable. Longer term fix would be to increase the thickness profile at critical locations); increase operating pressure in the LOX tank to oppose buckling forces (although this would likely have other undesirable avalanche effects to engine operation, plumbing, hoop stress, etc.).
Pretty much all of these fixes will involve a negative impact to mass-to-orbit capability, but probably not be a show stopper. My prediction is that the vehicle will be back in operation in a few months, but with reduced cargo capacity pending possible future design changes.
---------- Post added at 05:57 AM ---------- Previous post was at 01:04 AM ----------
He bottle failure was on my list early on. But I really think this would be obvious in telemetry. If LOX pressure were increasing, He bottle pressure would be decreasing much more rapidly in parallel. I think it would have been pretty obvious in telemetry, almost immediately after the incident, that the He bottle blew prior to or in unison with the LOX tank pressure increase.
This is why I settled on a simple LOX tank buckling as the most straight forward failure that seems to fit the data. It wouldn't be preceded by or associated with any other warning signs, other than vehicle structural stress, which may have initially been perceived as normal until further investigation.
Even if vehicle bowing had been detected in telemetry, it may not necessarily be immediately flagged as the cause of failure if similar bowing had been observed in previous flights without failure. And strain gauges may not detect the buckling unless the gauge happened to be right near the point of initial failure.
The good news is that, if wall buckling is the cause, then the fix may be relatively painless (compared to other imaginable scenarios). I suspect this is not a quality issue or a manufacturing issue. I suspect the dynamic loads just happened to be more extreme than what the models predicted. The problem might be addressed by several conceivable approaches: tighten upper wind constraints; adjust flight profile to be more shallow with less aggressive steering to minimize the angle of attack and associated asymmetric loading (although I think I've heard that the profile is already pretty shallow, so this may not be a practical solution); reduce the maximum cargo capacity; reduce loads with more engine throttling, limiting maximum loads to a lower level; for "work in progress" tanks, reinforce the skin in this area with additional stiffening (this fix would be a more major change, but not insurmountable. Longer term fix would be to increase the thickness profile at critical locations); increase operating pressure in the LOX tank to oppose buckling forces (although this would likely have other undesirable avalanche effects to engine operation, plumbing, hoop stress, etc.).
Pretty much all of these fixes will involve a negative impact to mass-to-orbit capability, but probably not be a show stopper. My prediction is that the vehicle will be back in operation in a few months, but with reduced cargo capacity pending possible future design changes.
---------- Post added at 05:57 AM ---------- Previous post was at 01:04 AM ----------
He bottle failure was on my list early on. But I really think this would be obvious in telemetry. If LOX pressure were increasing, He bottle pressure would be decreasing much more rapidly in parallel. I think it would have been pretty obvious in telemetry, almost immediately after the incident, that the He bottle blew prior to or in unison with the LOX tank pressure increase.
This is why I settled on a simple LOX tank buckling as the most straight forward failure that seems to fit the data. It wouldn't be preceded by or associated with any other warning signs, other than vehicle structural stress, which may have initially been perceived as normal until further investigation.
Last edited:
