totm nov 2023 SpaceX Discussion Thread

[darthgently]

4 hours ago, MashAndBangers said:

Sonic booms in the air!
And the rocket's red glare!
Gave proof through the night!
That our flap was still there!

*bows*

And it waved.  Like a pilot wagging the wings.  Still here!

[mikegarrison]

That was kind of a spectacular failure and spectacular success all wrapped up in one. A spectacle either way.

Looked like the fairing around the wing root (I guess I'll call it that) burned through first. Hard to tell, though. I'm sure they have experts who will be able to reconstruct what happened better than I can.

I didn't like seeing that "smoke" around the flap earlier in the flight, so I wasn't terribly surprised to see it eventually burn through. If that smoke wasn't just something in the ambient atmosphere, then it likely indicated something volatile was cooking off. (If it was something in the ambient atmosphere, I would have expected to see it continuously, rather than coming and going.)

Edited June 6 by mikegarrison

[CatastrophicFailure]

On 6/5/2024 at 12:37 PM, Terwin said:

As of yet, no starship has been lost due to loss of a tile.

Welp… seems this adage stubbornly clings to truth… kinda like that fin…

holey— 

9 hours ago, JoeSchmuckatelli said:

Oh yeah - watching the booster telemetry go from over 4,000 km/h at 20 km altitude to less than a thousand at 1km before lighting the rocket and then the hoover was amazing.  The physicality of all that - wow!

Our brains are all still locked to meters per second, not kilometers per hour. The mental translation takes a minute. 

[sevenperforce]

[snip]

37 minutes ago, mikegarrison said:

That was kind of a spectacular failure and spectacular success all wrapped up in one. A spectacle either way.

Looked like the fairing around the wing root (I guess I'll call it that) burned through first. Hard to tell, though. I'm sure they have experts who will be able to reconstruct what happened better than I can.

I found it really interesting that the burn-through happened LONG after expected peak heating/temp. This suggests that aerodynamic forces, rather than heat alone, were a dominant force in the burn-through.

The strakes are a temperature hot point. I wonder if the heat transfer past the tiles into the skin was enough to damage the skin structurally, which led to later weakness and peeling away once it got down in the denser, less-searing-but-still-very-spicy atmosphere. It could also be a glue issue. Those tiles around the strakes are glued on, so if the peak heating weakened the glue on a material integrity level, the glue still could have held until it got down into the thicker atmosphere, at which point aero forces ripped off the tiles and allowed the hot air (not hot enough to threaten the tiles but hot enough to melt steel) to enter. 

Edited June 6 by Vanamonde

[PakledHostage]

So I did a rough (very rough) integration of the speed vs. time plot on the previous page by counting squares. From first plasma to splash down is about 5300 km. That seems long? My knowledge of Space Shuttle re-entry comes from playing Orbiter, but the Shuttle re-entered over a shorter distance than that by on the order of 1000+ km. Could that have been a factor in the various burn throughs? (As EveryDayAstronaut pointed out, it looked from the glow like something outside the camera's field of view was on fire on Starship before the landing burn started.) The longer exposure to that re-entry plasma would have given more time for heat to permeate the structure. As @Minmus Taster pointed out in a post a few pages back, it spent a long time at 68 km altitude and the deceleration wasn't as aggressive as I'd have expected? Was the re-entry trajectory nominal?

[Pthigrivi]

Damn yall that was awesome. Really psyched for Tim Dodd's starbase walkthrough vid. 

[cubinator]

1 hour ago, sevenperforce said:

I found it really interesting that the burn-through happened LONG after expected peak heating/temp. This suggests that aerodynamic forces, rather than heat alone, were a dominant force in the burn-through.

I think it's more of an endurance thing and there is still more than enough heat to go around even if it's technically hit a peak. Like turning your oven off but leaving food inside, it might still burn. The steel was pretty obviously going soft, which isn't caused by drag.

[sevenperforce]

17 minutes ago, PakledHostage said:

So I did a rough (very rough) integration of the speed vs. time plot on the previous page by counting squares. From first plasma to splash down is about 5300 km. That seems long? My knowledge of Space Shuttle re-entry comes from playing Orbiter, but the Shuttle re-entered over a shorter distance than that by on the order of 1000+ km. Could that have been a factor in the various burn throughs? (As EveryDayAstronaut pointed out, it looked from the glow like something outside the camera's field of view was on fire on Starship before the landing burn started.) The longer exposure to that re-entry plasma would have given more time for heat to permeate the structure. As @Minmus Taster pointed out in a post a few pages back, it spent a long time at 68 km altitude and the deceleration wasn't as aggressive as I'd have expected? Was the re-entry trajectory nominal?

Spending a long time at higher altitude is a good thing for Starship.

There are (basically) two different types of heat shields: ablative heat shields and radiative heat shields. Ablative heat shields function by allowing their surface to boil away to shed heat, while radiative heat shields employ a low-conductivity, high-emissivity material which can handle high heat by simply absorbing it and radiating it away into the environment. Ablative heat shields have the advantage of being able to handle higher peak temperatures because their surface can simply boil away faster to adjust, while radiative heat shields are able to handle greater total heating because they don’t get used up.

If you have an ablative heat shield, peak heating is really not an issue: you can get as hot as you like and it’s not a problem because your heat shield will just boil off more quickly. Total heating, however, is a bigger issue: you need to get through the peak as quickly as possible or your entire well heat shield will boil away before you finish. As a result, you want a steep trajectory to slow down quickly. With a radiative heat shield, you need to limit peak heating so you need a more gradual re-entry with as much lift as possible. The more body lift you can achieve, the longer it will take you to slow down and the lower the peak heating will be.

Starship, like the Shuttle, uses a radiative heat shield, so it wants to enter with an attitude that prioritizes body lift and keeps converting speed into altitude for as long as possible to reduce peak heating.

The Shuttle had an additional issue, though: it needed to be able to achieve subsonic glide for landings, so it had WAY too much body lift in re-entry and had to execute complex S-curves to avoid (essentially) skipping off the atmosphere and losing aerodynamic control. Starship has a much more flexible control envelope since it doesn’t have to carry its entire launch structure to orbit and doesn’t have to worry about subsonic glide, so it can glide for as long and high as it wants. 

Edited June 7 by sevenperforce

[cubinator]

1 minute ago, sevenperforce said:

Starship, like the Shuttle, uses a radiative heat shield, so it wants to enter with an attitude that prioritizes body lift and keeps converting speed into altitude for as long as possible to reduce peak heating.

The end result seems to be an unexpectedly comfy ride down.

1 hour ago, sevenperforce said:

The strakes are a temperature hot point. I wonder if the heat transfer past the tiles into the skin was enough to damage the skin structurally, which led to later weakness and peeling away once it got down in the denser, less-searing-but-still-very-spicy atmosphere. It could also be a glue issue. Those tiles around the strakes are glued on, so if the peak heating weakened the glue on a material integrity level, the glue still could have held until it got down into the thicker atmosphere, at which point aero forces ripped off the tiles and allowed the hot air (not hot enough to threaten the tiles but hot enough to melt steel) to enter. 

One of the other things I was thinking about as I rewatched it is you can see the tiles popping off more or less one by one at certain points. I had the idea that thermal expansion of the overheated steel might stress the tile connections enough to contribute to breaking them off. Of course, the aerodynamic load is a factor there too, as the whole structure was very worn down at that point.

[PakledHostage]

3 minutes ago, sevenperforce said:

Starship, like the Shuttle, uses a radiative heat shield, so it wants to enter with an attitude that prioritizes body lift and keeps converting speed into altitude for as long as possible to reduce peak heating.

Sure. But that doesn't explain why Starship's re-entry seemed so much longer and with less deceleration than shuttle's, since they both use tiles. The tiles work to insulate as well as radiate. Longer exposure means more heat manages to permeate that insulation. As I wrote above,  I wonder to what degree the re-entry was nominal?

[CatastrophicFailure]

10 hours ago, Minmus Taster said:

Why are we not losing alttitude?

Out of likes : (

Lots of plasma, yikes

Because it’s really and truly flying.  Also keep in mind this descent would otherwise be much steeper, it’s only 3/4 of an orbit with perigee below the surface somewhere around Hawaii. On a “normal” trajectory that pe would still be above the surface, so this crazy reentry may be even more punishing than “usual.” 

10 hours ago, AckSed said:

"Did the primary buffer panel just fall off my gorram ship?"

Of all the posts to run out of likes on…

[mikegarrison]

2 hours ago, sevenperforce said:

[snip]

I found it really interesting that the burn-through happened LONG after expected peak heating/temp. This suggests that aerodynamic forces, rather than heat alone, were a dominant force in the burn-through.

The strakes are a temperature hot point. I wonder if the heat transfer past the tiles into the skin was enough to damage the skin structurally, which led to later weakness and peeling away once it got down in the denser, less-searing-but-still-very-spicy atmosphere. It could also be a glue issue. Those tiles around the strakes are glued on, so if the peak heating weakened the glue on a material integrity level, the glue still could have held until it got down into the thicker atmosphere, at which point aero forces ripped off the tiles and allowed the hot air (not hot enough to threaten the tiles but hot enough to melt steel) to enter. 

Heat and temperature are not the same thing. The structure *can't* have been at plasma temperatures, or else it would have been a plasma. As long as the aerodynamic temperature is still higher than the structural temperature, even if it's not as high as it was at its peak, it is still moving heat into the structure.

And as metal heats up, it loses strength. So if the forces are still building and the heat is still building (even if not as quickly as before), the structure could fail even after passing the peak outside temperature. Possibly even after passing the peak internal temperature, if the forces build quickly enough and the heat is not shed quickly enough.

And heat moving through a structure is complicated. Even if the tile surface temperature has started to cool down, it is still hotter than the structure. So even if the tiles are no longer heating up, the structure still can be.

Or maybe your glue idea is correct. As I said in my post, I feared that something volatile was cooking off. Maybe that was the glue. (Of course, at high enough temperatures damn near anything can become "volatile", even the metal....)

52 minutes ago, PakledHostage said:

So I did a rough (very rough) integration of the speed vs. time plot on the previous page by counting squares. From first plasma to splash down is about 5300 km. That seems long? My knowledge of Space Shuttle re-entry comes from playing Orbiter, but the Shuttle re-entered over a shorter distance than that by on the order of 1000+ km. Could that have been a factor in the various burn throughs? (As EveryDayAstronaut pointed out, it looked from the glow like something outside the camera's field of view was on fire on Starship before the landing burn started.) The longer exposure to that re-entry plasma would have given more time for heat to permeate the structure. As @Minmus Taster pointed out in a post a few pages back, it spent a long time at 68 km altitude and the deceleration wasn't as aggressive as I'd have expected? Was the re-entry trajectory nominal?

Yeah, re-entry is a balance between heat and deceleration. Things that can take large acceleration forces (like nuclear warheads) tend to plunge very, very quickly through the atmosphere in order to minimize the amount of time (and therefore heat pickup). Things like crew capsules that have reasons why they don't want to take 100g decelerations tend to go through the re-entry much shallower, which means a longer time at high temperatures which means more heating issues.

Edited June 7 by mikegarrison

[Minmus Taster]

Had to put on a hazmat suit to pull this out of Elon's twitter but voila, some interesting updates on Starship:

TL;DR: Elon thinks first catch attempt should still be on the next flight, forward flaps are still getting moved at some point as has been the plan for some time. Other improvements on the way.

https://x.com/elonmusk/status/1798848426895200567

https://x.com/elonmusk/status/1798745564898840872

https://x.com/elonmusk/status/1798735261645369806

https://x.com/elonmusk/status/1798732390313218305

[PakledHostage]

Interesting that he mentions "SX300" stainless steel. I assume that's a proprietary SpaceX variant of a 300 series stainless steel? I wonder how that alloy stacks up against something like Inconel 718?

[mikegarrison]

15 minutes ago, PakledHostage said:

Interesting that he mentions "SX300" stainless steel. I assume that's a proprietary SpaceX variant of a 300 series stainless steel? I wonder how that alloy stacks up against something like Inconel 718?

Inconel is a "super-alloy" that is primarily nickel. 300-series steel is a steel (that is, primarily iron, with significant inclusions of carbon and also other metals, including nickel).

According to Wikipedia, SX300 is not a 300-series steel at all. (Or any other kind of steel, for that matter.) It is a nickel super-alloy, so much more like Inconel.

Edited June 7 by mikegarrison

[Nuke]

i wouldn't doubt that the only thing proprietary is a tighter thickness tolerance and surface finish.

[mikegarrison]

Iron: can refer to elemental iron, but also steel with about 2-4% carbon (like "cast iron"). Wrought iron is very nearly 0% carbon, but is not pure iron due to the inclusion of slag elements.

Steel: iron-carbon alloy, can also include other metals but is primarily iron. "Spring steel" has less than 1% carbon (sometimes less than 0.5%).

Stainless steel: steel, but with a significant amount of chromium

"super-alloys": usually means nickel-based, but also iron and cobalt might be in the base mix, often with chromium as the the main alloying material, generally with very little carbon

===========

So there are some areas of overlap, but usually if the primary metal is nickel, it's not called steel.

Edited June 7 by mikegarrison

[PakledHostage]

I speculated about Inconel 718, not because I was confusing it with any stainless steel alloy, but because it is extremely strong at high temperatures.  I didn't realize that SX300 is also a nickel alloy. Odd that they repeatedly refer to Starship being made out of stainless steel, and even Elon refers to the alloy as SX300 stainless steel in his tweet. As @mikegarrison points out, nickel alloys aren't steel (stainless or otherwise).

[mikegarrison]

18 minutes ago, PakledHostage said:

I speculated about Inconel 718, not because I was confusing it with any stainless steel alloy, but because it is extremely strong at high temperatures.  I didn't realize that SX300 is also a nickel alloy. Odd that they repeatedly refer to Starship being made out of stainless steel, and even Elon refers to the alloy as SX300 stainless steel in his tweet. As @mikegarrison points out, nickel alloys aren't steel (stainless or otherwise).

Maybe Musk just doesn't like the term "super-alloy" (or "superalloy"). These nickel-based alloys have been called that for about 100 years now, though.

[CatastrophicFailure]

9 hours ago, Ultimate Steve said:

Starlink is at nearly 3 million users

Fwiw I’ve had Starlink for a couple years now, still waiting on that reliable 3G cell tower, let alone 5G… 

 

Also outta likes so…

6 hours ago, JoeSchmuckatelli said:

The above from Max Fagin on Xitter 

Only 1.5G? That is very gentle for a reentry!

(Looking at you, Hubble.)

[Ultimate Steve]

Doing a reliability tally for Raptor in the integrated flight tests. I will generally assume SpaceX is being honest with us and I will not count ignition failures on aborted launch attempts. I will not be going back to the suborbital hops, because that would be a hassle, those were really old Raptors, and the small number of burns wouldn't make much of a dent in the reliability statistics. Attempted burn is defined as the engine being commanded to start. I think it would be silly to say "Raptor failed the landing burn!" when Super Heavy blew up during boostback and the landing burn was never attempted.

IFT-1:

Three engine ignitions were either not attempted or aborted (likely aborted but wording is kind of unclear). A further 3 engines cut out during the flight. There may have been one engine reignition. It is unclear if this actually happened or if it was a telemetry error. There were discrepancies between the graphic and the camera views, which is consistent with the fact that they lost contact with at least one of the engines fairly early on in flight. The upper stage never got to attempted ignition.

Depending on how you count, either 30, 31, 33, or 34 attempted engine burns, and 27 or 28 successful burns (the engine that might have restarted either had 1 success and 1 failure or 1 failure).

From a steel man point of view, an additional 12 Raptors failed as thrust vector control was lost. Either the fires (unclear if they originated in the fuel feed system or in Raptor) or the possible hydraulic unit explosion was likely the culprit.

Best number: 28/31 successes. Reasonable number: 27/33 successes. Worst number: 15/33 successes. I don't think it is fair to say that the fires, even if Raptor related, equate to each of those engines that the failure took out of commission being unreliable, but there will be those that disagree with me. Numbers equal or near equal for per engine reliability, not per burn reliability, the only ambiguity coming from that one possible engine restart.

If I remember right, this was the only full up flight of Raptor 1. I think IFT-2 used Raptor 1s still on the upper stage, but not on Super Heavy.

 

IFT-2:

33 engines lit. 9 of 10 engines relit for boostback. All 13 engines subsequently shut down. Officially, the most likely culprit for this is a clogged LOX filter starving the engines of fuel. This is a problem with the fuel feed system, but I would also accept that one Raptor failed to safely shut down and blew everything up ("SpaceX stated that the most likely root cause was filter blockages where liquid oxygen is supplied to the engines, leading to a loss of inlet pressure in the engines' oxidizer turbopumps that eventually resulted in one engine failing in such a manner that it resulted in the loss of the vehicle") If you want to be that guy, sure, they could be lying to us and all 13 engine shutdowns could be Raptor's fault. Or maybe Raptor should have just coped with no LOX somehow. Of 43 burns, there was only one shutdown that can be definitely said to be Raptor's fault (unless the LOX clog started extremely early). At extreme cope levels, you could say 30/43 successful burns.

Upper stage lit successfully. 

All six engines were shut down by the AFSS after communication was lost with them due to a LOX vent causing a fire. I don't see any justifiable way you can call this a Raptor failure unless you think SpaceX is lying to us, NASA, and the FAA.

Best number: 48/49 successful burns. Reasonable number: 48/49 burns. If you count that one Raptor that wasn't its fault, but did kill the vehicle, I would also accept 47/49. Worst number: 36/49.

Per engine, the numbers are 38/39, 37/39, and 26/39.

 

IFT-3:

33 engines lit and stayed lit. Ten were relit for the boostback, with six shutting down early, officially likely due to LOX filter issues. This triggered a "benign early boostback shutdown". Taken at its worst, the computer told the other 7 engines to turn off even though there was more burn to be done. The problem should not reside with those seven Raptors.

Booster attempted a seven engine landing burn. Two engines ignited, only one was left by the time the booster crashed. No official word on why this failed, I would assume LOX filters again, but it is possible that six Raptors failed for Raptor reasons here. It could also be due to unexpected conditions (reportedly the hot stage ring was ripped off by aero forces, the booster was in the middle of a large shake during engine ignition).

All six engines on the upper stage lit and stayed lit. The test de-orbit burn was cancelled due to non Raptor related issues (lack of attitude control).

Best number: 56/56 burns (all engine failures have a plausible non Raptor explanation). Reasonable number: 50/56 burns (6 landing burn failures). Worst number: 44/56 burns.

Per engine, the numbers are 39/39, 33/39, and 27/39.

 

IFT-4:

32 engines lit and stayed lit. Ten were relit for boostback. 12/13 engines lit for landing. All six engines on the upper stage worked. We don't know how many engines lit for landing because the graphic didn't work. But we have no indication of a failure. I'm gonna call it 2/2, but it could theoretically be 3/3 or 2/3 or even 1/3 or 1/2.

Best number: 63/65 (3 engine landing burn). Reasonable number: 62/64 (2 engine landing burn). Worst number: 61/65 (3 engine landing burn, 2 failures).

Per engine, the numbers are 37/39, 37/39, and 35/39.

This could change if it turns out that an engine failure was due to a clogged filter or something. It can only change in Raptor's favor, though, barring something ridiculous being uncovered.

 

Plugging all of that into a spreadsheet, per burn, Raptor reliability across the Integrated flight tests is 97.0% at its best, 92.1% if we take SpaceX's word, and assume, in the lack of an official statement, that the IFT-3 landing burn failure was Raptor's fault (number is 95.0% if that incident was a filter issue), and 76.8% at its worst. Per engine, the numbers are 95.9%, 89.3%, and 74%. The lower numbers require SpaceX lying to us and attribution of Raptor failures to non Raptor causes and is intended to be an extreme example.

If we just tally flights 2, 3, and 4, mostly just counting Raptor 2s, those numbers change to 98.2%, 94.1%, 82.9%, 97.4%, 91.5%, and 82.1%.

Raptor is not currently producing 230 tons of thrust (official numbers per flight 3 had it at 7130 tons, or 216 tons per engine). We don't know why this is. Maybe the 230 ton number was a one off, a vacuum number, a future version, or de-rating because Raptor would eat itself otherwise. For all we know, this isn't an issue. 7130 is notably 31*230, so maybe at this stage they assume 2 engine out capability and possibly throttle a little low and throttle up if an engine or two goes out. If it is an issue, though, those reliability numbers are starker, as this is with a gentler environment.

In the context of an iteration driven development program where a single engine out (or several) doesn't make or break the mission, and where many engine failures were in unique environments, and where we've only just gotten to flight number 4, I'd say those numbers are not great but acceptable. 98.2% reliability, the best justifiable number I was able to come up with, is not acceptable for humans by traditional standards. The lowest, ridiculously far out number, 74% with derating, is grounds for program cancellation.

I cannot find another modern, significant flight history rocket engine reliability number below 98.2%, except for specifically just Vacuum Rutherford, and Delphin on Astra's rocket 3 (there are probably more I wasn't able to find). We don't know how many times it failed, Astra was pretty secretive, but it is at most 97.2% (one definite engine failure out of 7 orbital attempts with 5 engines each).

SSME: 99.75%

Merlin: Somewhere at around 99.7-99.8%, numbers use different launches, are from different times, unsure if they counted the landing burn failures if there were any, or the first Falcon 1 failure.

RD-180: 100%

Rutherford (normal and vacuum): 99.38%

Vacuum Rutherford: 93.75%

 

Granted, all of those examples have more flight history than Starship in terms of flight count (Raptor probably beats RD-180 in engines flown). If Starship were to work flawlessly on the next 3 flights, maximum reliability for Raptor 2 would increase to 99.2%. Still low. 

 

I'm beginning to see where Exoscientist is coming from. I would still argrue that in the context of the Starship program, the current reliability is within the realm of what is expected at this stage. But with conventional thinking, for something that is to power a moon lander, even 98.2%, let alone 94.1%, or 74%, is an absurdly low number.

To suddenly see NASA being okay about, and enthusiastic about reliabilities this low under a conventional scenario would be grounds for conspiratorial thinking. Imagine if NASA had wrapped up SSME engine testing with a 90% success rate with ambiguity as to whether or not the engines were derated, and were still touting the shuttle as the future. I'd think there was a conspiracy at that point.

 

The difference is that these are not conventional circumstances. I don't think that Starship should be looked at the same way other rockets are looked at. 98.2% is not acceptable for something carrying humans. 92.1% can be reasoned with for something that is fault tolerant, pushing the envelope, fairly early in development, under significant active development, has to ignite in difficult scenarios nobody had tried before, will not fly an external payload for at least a few more flights, will not fly humans for at least another 2-3 years, and even then, only on the upper stage (which has not had a single engine failure during the full up tests), and is doing hardware rich testing.

[mikegarrison]

I'm not worried (yet) about the raptor reliability. They can (I assume) find and fix the problems as they show up. I CERTAINLY would not volunteer to fly on a Starship (yet), unless I had a death wish, but that's for a variety of reasons.

What has *always* concerned me is the close packing of the engines and the possibility of fratricidal cascades. SpaceX has admitted having to add a lot of weight to place armor between the engines, but it's still a fundamental issue of the design. If an engine failure can lead to other engine failures, then the more engines you have does not increase redundancy, it only increases risk.

If we see them routinely losing one and only one engine, I take that as a better sign than if we see them losing banks of them.

Edited June 7 by mikegarrison

[CBase]

nice recap by Scott Manley with his comments how heat could sneak into flaps:

comments I liked best

Quote

@DeadCat-42

That was the most real life Kerbal thing I have ever seen .

Quote

@w3vjp568 

Elon: “The best part is no part.”

Flap: “Ok boss, understood!”

Elon: “No, wait!”

Quote

@nathanobrien2262

I hadn't noticed this in the live event but at 6:04 in your video, I think you can see the hot staging ring being overtaken by the booster on the way back down. Awesome!

 

[Nuke]

the raptors are being developed in parallel to starship. today's old model is tomorrow's test article. as far as i know were still using the oldest engines first. its also more useful to fly it with a retrofit of some new feature on an old engine than it is to retool the line for the new revision. then you are on better footing with the new feature and can better commit it to your design. the important thing to look at is whether or not reliability is improving from launch to launch. 

[Exoscientist]

3 hours ago, mikegarrison said:

Maybe Musk just doesn't like the term "super-alloy" (or "superalloy"). These nickel-based alloys have been called that for about 100 years now, though.

 Is it possible the high temperature alloys used for the engine combustion are being confused with the metals used for the tanks?

  Bob Clark