Starship SN8 Header Tanks and Loss of Vehicle

[lawlessl]

Anyone?

The best picture I can find is a tweet by Steve Jurvetson Dec 13 12:39AM. The valve shows a symmetrical sculptured disk and what looks like a linear actuator. The shaft seems to be perpendicular to the actuator and approximately vertical. I assume the actuator pulls on a lever on the hidden shaft to release the cams, to unseat it, and then rotate the disk.

[Haur]

Anyone?

The best picture I can find is a tweet by Steve Jurvetson Dec 13 12:39AM. The valve shows a symmetrical sculptured disk and what looks like a linear actuator. The shaft seems to be perpendicular to the actuator and approximately vertical. I assume the actuator pulls on a lever on the hidden shaft to release the cams, to unseat it, and then rotate the disk.

We've actually got some pictures of similar valves open - the vents on the side of SN9, as seen here: https://forum.nasaspaceflight.com/index.php?topic=51332.msg2154751#msg2154751

They certainly seem quite capable of holding tank pressure reliably over the course of mulitple cryotests.

[lawlessl]

We've actually got some pictures of similar valves open - the vents on the side of SN9, as seen here:

Thank you. I am now positive that is a triple offset Butterfly valve on the side of SN9. I do not believe it can be a cam type as the actuator arm is on the wrong side of the disk. The actuator arm must be on the same side of the disk as the shaft otherwise it can not turn the required cams. Also the disk is the shaft in this valve.
Therefore back to the original problem of the header tank under extreme pressure. Those valves should give way as the offset on the shaft makes the two sides different areas. Thus reverse pressure can open it. The mystery of the missing header tank top continues.

[cdebuhr]

We've actually got some pictures of similar valves open - the vents on the side of SN9, as seen here:

Thank you. I am now positive that is a triple offset Butterfly valve on the side of SN9. I do not believe it can be a cam type as the actuator arm is on the wrong side of the disk. The actuator arm must be on the same side of the disk as the shaft otherwise it can not turn the required cams. Also the disk is the shaft in this valve.
Therefore back to the original problem of the header tank under extreme pressure. Those valves should give way as the offset on the shaft makes the two sides different areas. Thus reverse pressure can open it. The mystery of the missing header tank top continues.

ISTM that if the butterfly is carefully designed, this effect could be minimized (although not completely eliminated).  Further, it would not be difficult to include a stop in the valve to prevent over-rotation on closure (I suspect this would be standard in many butterfly valves, but I do not know this with certainty).  Thus one would only have to orient the valve so as to prevent pressure differential (in normal operation) from forcing the valve open.  I'd think you would naturally want to keep the header tank at positive pressure WRT CH4 main, so orient the butterflys accordingly.  I'd think significant negative relative pressure in the CH4 header is something to be avoided so as to not de-stabilize the header tank.  Reverse pressure on the valve is something should never happen in normal operation.  Also note that the pneumatic linear actuators on these valves are double-acting, and likely capable of exerting considerable closure force all on their own.

I'm not sure there's really a problem of the header under extreme pressure, or perhaps its a difference of interpretation of the word extreme.  Without quibbling about specific numbers, I'd say the fuel systems run at low pressure, the vehicle pneumatic and RCS systems are high pressure, and the extreme pressures are restricted to the innards of the Raptors.  If by "extreme", you just mean high enough to rupture the tank (still pretty low though - on the order of 10 bar), then I'm not surprised in the least that the valves held ... its not much worse that what the main tank vent valves (very similar, at least in gross appearance) deal with on a regular basis during cryo testing and normal operation.

As to the missing top of the header tank ... it seems exceedingly likely that it was just fine up to "landing" and blew off during the RUD.  No mystery there.

[aero]

As to the missing top of the header tank ... it seems exceedingly likely that it was just fine up to "landing" and blew off during the RUD.  No mystery there.

No Mystery ... Well, maybe but the question now becomes, "Why?"

Why did the tank blow the top off instead of just denting, crushing, or ripping, like the rest of the Starship? STM there must have been a pressure impulse inside the tank, transferred through an incompressible liquid, not through a compressible gas. The implication is that the header tank was almost completely full of liquid methane when it was shocked by an impact with concrete and that shock propagated across the tank and blew the top off. So why was the engine starved for methane?

[RotoSequence]

No Mystery ... Well, maybe but the question now becomes, "Why?"

Why did the tank blow the top off instead of just denting, crushing, or ripping, like the rest of the Starship? STM there must have been a pressure impulse inside the tank, transferred through an incompressible liquid, not through a compressible gas. The implication is that the header tank was almost completely full of liquid methane when it was shocked by an impact with concrete and that shock propagated across the tank and blew the top off. So why was the engine starved for methane?

Elon gave us the answer almost immediately; a lack of pressure for sufficient fuel flow. "Why was the pressure low?" is a much more difficult question. My hypothesis is that the pitch-kick maneuver rather dramatically mixed the cold liquid methane and hot ullage methane, condensing the gas and forming a very low pressure region inside the ullage volume.

[Nevyn72]

As to the missing top of the header tank ... it seems exceedingly likely that it was just fine up to "landing" and blew off during the RUD.  No mystery there.

No Mystery ... Well, maybe but the question now becomes, "Why?"

Why did the tank blow the top off instead of just denting, crushing, or ripping, like the rest of the Starship? STM there must have been a pressure impulse inside the tank, transferred through an incompressible liquid, not through a compressible gas. The implication is that the header tank was almost completely full of liquid methane when it was shocked by an impact with concrete and that shock propagated across the tank and blew the top off. So why was the engine starved for methane?

Maybe a large downcomer shaped battering ram being forced up from beneath could have caused that result? 

EDIT: Speiling

[john smith 19]

Elon gave us the answer almost immediately; a lack of pressure for sufficient fuel flow. "Why was the pressure low?" is a much more difficult question. My hypothesis is that the pitch-kick maneuver rather dramatically mixed the cold liquid methane and hot ullage methane, condensing the gas and forming a very low pressure region inside the ullage volume.

This has been viewed as an issue in various NASA studies on zero boiloff propellant management. It's usually called "pressure collapse" and is associated with where the fill inlets are located within the tank Vs the ullage pressure inlets.

[aero]

No Mystery ... Well, maybe but the question now becomes, "Why?"

Why did the tank blow the top off instead of just denting, crushing, or ripping, like the rest of the Starship? STM there must have been a pressure impulse inside the tank, transferred through an incompressible liquid, not through a compressible gas. The implication is that the header tank was almost completely full of liquid methane when it was shocked by an impact with concrete and that shock propagated across the tank and blew the top off. So why was the engine starved for methane?

Elon gave us the answer almost immediately; a lack of pressure for sufficient fuel flow. "Why was the pressure low?" is a much more difficult question. My hypothesis is that the pitch-kick maneuver rather dramatically mixed the cold liquid methane and hot ullage methane, condensing the gas and forming a very low pressure region inside the ullage volume.

I agree fundamentally. "A lack of pressure for sufficient fuel flow." But I think the tank was almost if not completely full of liquid methane. That is, "... " mixed the cold liquid methane and hot ullage methane, condensing all of the gas leaving the tank filled with liquid. Then, when the tank was shocked at the bottom, the shock wave propagated through the incompressible liquid and blew the top of the tank off.

Or maybe the shock propagated up through the downcomer into the header tank and then blew the top off.

We are saying the same thing, differing only in a matter of degree. How much additional heat is required in the header tank to maintain gaseous pressurization? And that is an issue for Elon's crew.

[cdebuhr]

No Mystery ... Well, maybe but the question now becomes, "Why?"

Why did the tank blow the top off instead of just denting, crushing, or ripping, like the rest of the Starship? STM there must have been a pressure impulse inside the tank, transferred through an incompressible liquid, not through a compressible gas. The implication is that the header tank was almost completely full of liquid methane when it was shocked by an impact with concrete and that shock propagated across the tank and blew the top off. So why was the engine starved for methane?

Elon gave us the answer almost immediately; a lack of pressure for sufficient fuel flow. "Why was the pressure low?" is a much more difficult question. My hypothesis is that the pitch-kick maneuver rather dramatically mixed the cold liquid methane and hot ullage methane, condensing the gas and forming a very low pressure region inside the ullage volume.

I agree fundamentally. "A lack of pressure for sufficient fuel flow." But I think the tank was almost if not completely full of liquid methane. That is, "... " mixed the cold liquid methane and hot ullage methane, condensing all of the gas leaving the tank filled with liquid. Then, when the tank was shocked at the bottom, the shock wave propagated through the incompressible liquid and blew the top of the tank off.

Or maybe the shock propagated up through the downcomer into the header tank and then blew the top off.

We are saying the same thing, differing only in a matter of degree. How much additional heat is required in the header tank to maintain gaseous pressurization? And that is an issue for Elon's crew.

You can't take a partially full cryo tank, initiate ullage collapse, and somehow end up with a tank of liquid.  If you condensed all the gas, you'd have a vacuum and the tank itself would collapse.  But that would never happen - some of the gas would condense, and you'd be left with less gaseous mass occupying only slightly reduced volume at a reduced temperature and pressure.  That's just how it works.

[lawlessl]

There is the obvious reason. The header tank had a problem. Therefore the first thing they did was to cut off the top to get to inspect the device that was on the end of that pipe that enters the tank on the above photo. Or the sensors or devices that were connected to all the control piping left around the tank. The job sheet said cut off the top at the weld and they did just that.

[daavery]

the hole in top of the header was there immediately after the RUD. well before any clean up activities

[lawlessl]

well before any clean up activities

The whole area was sealed off. The first photograph was more than 24 hours later.

[chopsticks]

There is the obvious reason. The header tank had a problem. Therefore the first thing they did was to cut off the top to get to inspect the device that was on the end of that pipe that enters the tank on the above photo. Or the sensors or devices that were connected to all the control piping left around the tank. The job sheet said cut off the top at the weld and they did just that.

Why would they cut around the hole that was already there made from downcomer pushing up from below? The weld marks are obviously there where the downcomer was welded to the tank.

Sent from my Redmi Note 7 using Tapatalk

[dnalioh]

There is the obvious reason. The header tank had a problem. Therefore the first thing they did was to cut off the top to get to inspect the device that was on the end of that pipe that enters the tank on the above photo. Or the sensors or devices that were connected to all the control piping left around the tank. The job sheet said cut off the top at the weld and they did just that.

Why would they cut around the hole that was already there made from downcomer pushing up from below? The weld marks are obviously there where the downcomer was welded to the tank.

Sent from my Redmi Note 7 using Tapatalk

The welds are from the cap. The downcomer sits at the bottom of the header tank.

[Rocket Science]

Perhaps a re-think of the final landing approach might be in order and the extra engineering challenges it demands...

[exilon]

This is risking wandering off into a rerun of the Falcon 9 landing discussions where we had people suggesting nets and grabbers after CRS-6 toppled on landing.

Assuming SpaceX had no data for their models, the most straightforward test would be to make a header tank stand-in and shake it on some hydraulic rams to see how much additional feed they need to overcome the pressure drop during the landing rotation.

More likely they already have all the telemetry needed to update their models, given they knew it was caused by low pressure immediately, and have already tweaked SN9's regulator valve settings to compensate.

[TheRadicalModerate]

There seem to be three hypotheses:

1) Plumbing failure.  Somehow the autogenous gas didn't make it into the LCH4 header tank.  There's nothing specific to support this hypothesis, other than bad plumbing is always a suspect.

2) Slosh.  When the Starship rotation reversed to move from past-vertical back to vertical, enough of the header tank's contents had been burned in the initial rotation from horizontal to past-vertical that the downcomer was uncovered, driving ullage gas into it and causing loss of pressure in the fuel turbopump outlet.  Best evidence to support this is that the green flame started just after the rotation reversal.

3) Condensation collapse of the ullage gas.  Assuming that ullage is snorkeled into the header tank, the main LCH4 tank had been sitting in equilibrium from the time the Raptors shut off at apogee until they fired just before landing (1min 50s).  If the ullage cooled enough to condense during that period, then the main and header would have been at much lower pressure than planned.  When the landing burn began, the pressure was barely adequate for the first part, but the autogenous system couldn't bring it back up to pressure in time for the end of the burn, causing loss of pressure in the fuel turbopump outlet.  Green flame at the end of the burn but not the beginning is pretty decent supporting evidence for this hypothesis.

[TheRadicalModerate]

There is some discussion of this in the engineering master thread but if this thread ends up being allowed to stand, here we go.

We know that the methane flows through the header tank during the main engine burn, but we don't know how the methane header tank is pressurized after restart. There are really two options: (A) either it is pressurized via valves from the main tank, or (B) it is pressurized directly using a separate press line from Raptor.

My armchair suspicion is that the consumption of propellant by the two Raptors sucked liquid methane out of the header tank very fast, and the ullage gas expanded into the space too quickly to be replaced, causing its pressure to drop. Either (A) gas wasn't flowing from the main into the header fast enough, or (B) the autogen press gas from Raptor didn't reach the header tank in time.

Either way, adding an auxiliary pressurization line from gaseous methane COPVs would seem to be the straightforward fix. Either that or (if the answer is (B) above) set the header tank at a higher ullage pressure at MECO, assuming Raptor doesn't mind a higher inlet pressure.

I've attached a diagram showing what (A) might have looked like.

In (A), if the vents between the main and header are open (and they have to be to receive pressure from the main), then the LCH4 is going to leak out of the header back into the main when the vehicle is horizontal.  You can fix this with a check valve, but since those vents have to be able to handle about 3200kg/s when all six engines are burning, that could turn into an interesting piece of plumbing.

I guess you could close the vent on the ventral side of the tank, leaving just the dorsal vent open to the main.  But it seems like the failure tree on that strategy would be both wide and deep.

My money's on (B) long-term, but I could see how they might have decided to see if they could get away with (A) + closing the ventral vent for purposes of this test.

[Steve D]

There probably was no ullage gas in the header tank. The fuel header tank is in the bottom of the fuel tank, correct? With sub cooled liquid methane any methane gas in the header would condense to liquid during the flight. When they start pulling fuel from the header with no ullage gas the pressure would almost instantly go to zero. How much fuel needs to flow to the engines before any gaseous methane makes it to the tank? With no ullage pressure would drop instantly and by the time the tank starts to get pressure from the engine it was too late.