11 ms·

Historically the thought "we have gotten the hang of this" has ended badly when dealing with spacecraft and rockets. The potential failure profile that we are

Historically the thought "we have gotten the hang of this" has ended badly when dealing with spacecraft and rockets. The potential failure profile that we are in now is that there may be somethings that engineers accepted the risk that can lead to loss of mission, vehicle and life. See Apollo 1, Apollo 13, Challenger (after 25 shuttle missions) and Columbia (after a further 88). Its great to see the "routine-ness" - but SpaceX needs to stay humble and stay focused. A accident now could cost everything they have done to this point. -- edited to clarify potential failure mode.

chinathrow 5y ago
> The failure profile that we are in now is that there may be somethings that engineers accepted the risk that can lead to loss of mission, vehicle and life. How did you get to that conclusion regarding Crew Dragon?
- idealmedtech 5y ago
  Crew Dragon only has a handful of missions. What the previous commenter is talking about is risks that may have been ignored/downplayed/unsuccessfully mitigated but that Crew Dragon is simply lucky enough not to have faced. With time and more flights, more things will have a chance to go wrong, and in spaceflight, resting on your laurels is the best way to make sure that when it does go wrong it will be catastrophic.
  - chinathrow 5y ago
    I agree regarding the handful of missions argument with Crew Dragon, but let's not forget all the successful Cargo Dragon missions here as I'm sure they play a role in (ongoing) safety assessments.
    - InTheArena 5y ago
      Sure. It helps. One of the things NASA considered was Shuttle-C, a cargo varient of the space shuttle system, which would have given NASA a way to tinker on the engine /SRB stack without crew onboard. Of course, that would not have saved columbia (the orbiter itself was the problem). This stuff is called rocket science for a reason.
      - progre 5y ago
        Well, still... It's not exactly brain surgery is it?
  - bryanlarsen 5y ago
    NASA calls it "normalizing of deviance", and it's why we lost both Shuttles. For example, the O-Rings were designed as a backup, the force generated by the solid rocket motors was supposed to seal the SRM sections together. But they knew after the first launch that the O-Rings were what was actually sealing the SRM's. But the O-Rings worked, so they didn't stop the program to fix the design. Later they found evidence of blow through the O-Rings, and again decided that since that didn't cause a crash, they were OK. Until they didn't. Tile strike damage was a similar normalization that killed 7 people. Does SpaceX normalize deviance? They fixed the toilet, so they didn't normalize that, but what about the delayed opening of the 4th parachute in the Crew 2 landing? It seems suspicious that they launched Crew 3 so quickly after an apparent anomaly. They do claim that the delayed opening is nominal behaviour seen and accepted in testing -- that this behaviour only happens when the other 3 parachutes are working fine. If one of the others had failed the 4th would have opened quicker due to increased air pressure. https://arstechnica.com/science/2021/11/double-dragon-nasa-plans-a-spacex-splashdown-on-monday-launch-wednesday/ https://arstechnica.com/science/2021/11/double-dragon-nasa-p...
    - rbanffy 5y ago
      > that this behaviour only happens when the other 3 parachutes are working fine I'd love to see the causal relationship between both that makes it safe to land with 3 parachutes (I remember they added the 4th as an added precaution that reminds me of the O-ring issue)
    - thaeli 5y ago
      NASA seems to normalize deviance when they're under political pressure to do so. I'm not sure they can be entirely trusted when it comes to their own launch systems, because of the politics involved - but so far they haven't shown similar hesitancy with commercial launch systems. It's a good separation of concerns; when NASA is "just the customer" and it's not "their rocket" they are in a better position to hold a firm line on safety issues. That NASA did not see the parachute as an issue speaks volumes. It looked bad visually, but having watched enough space launches and how mission control reacts to off-nominal data - clearly they were comfortable with the performance, because they didn't flinch at all. NASA just doesn't sign off on these commercial launches with NASA astronauts on board if they have any inkling that something won't go right - they aren't under the political pressure to "save the Shuttle program" or whatever. I have more concern about NASA getting politically-imposed "go fever" with SLS.
ourmandave 5y ago
You'll know they've really reached routine when people's main complaint is about the in-flight meal.
- pkdpic 5y ago
  You'll know they've really reached routine when people's main complaint is about not having enough leg room :^) Also wondering what the timeline comparison is here too. Wasn't one of the problems with Columbia the advanced age of the shuttle? I'm assuming this isn't comparable to the situation with spacex?
  - dotancohen 5y ago
    > Wasn't one of the problems with Columbia the advanced age of the shuttle? No. The problem that led to the loss of Columbia was fundamental and well understood - if underestimated - since STS-1 was inspected after its first landing.
    - Fatnino 5y ago
      STS-1 was known to be missing some tiles while still in space. Like as soon as they opened the payload bay doors they could see holes in the tiles on the OMS pods.
elif 5y ago
The programs you listed with 4 failures launched 164 times including test flights. Falcon 9 has provided 127 successful missions and 1 failure. On visits to the ISS it's 37 vs 25. If you look at the ratio of non-human flights to human flights SpaceX is far more conservative of human life.
- ortusdux 5y ago
  Also, the 1 failure was during a pre-flight static test fire. If it had been a crewed flight, the crew would not have boarded at that point. It's considered a failure because they test fired the rocked with the payload attached, resulting in its destruction. They have since stopped performing this test with the payload attached.
  - croddin 5y ago
    The 1 failure in those numbers is actually CRS-7, not Amos-6. If you count Amos-6 there have been 2 failures but as you pointed out crew wouldn’t have been on board and they don’t static fire with payload anymore. Also for CRS-7 the dragon survived and would have been fine if the parachutes had been programmed to deploy in that situation. Either way there have been a lot of improvements and ~100 successful launches since either of those.
    - ortusdux 5y ago
      Ahh, my mistake. This chart is very informative for those interested: https://en.wikipedia.org/wiki/List_of_Falcon_9_and_Falcon_Heavy_launches#Launch_outcomes https://en.wikipedia.org/wiki/List_of_Falcon_9_and_Falcon_He...
    - Fatnino 5y ago
      Amos6 style explosiw would have been surviveable if there was crew on board because of the launch escape thrusters. See Soyuz T-10a
  - oxplot 5y ago
    Don't forget that assuming a crew onboard Dragon, the launch escape system (LES) is engaged before propellant loading, so even if the tanks blow up, the capsule will escape the explosion and crew will be safe.
rbanffy 5y ago
> and Columbia (after a further 88). I'll never get over the fact that NASA never made an assessment of tile damage after liftoff in over 100 flights before Columbia. All these machines are experimental in the sense we still don't understand completely what happens with them and treating them like 737's is a huge mistake. Even in the case of the 737, the MAX proved we didn't understand a 737 thoroughly enough to ensure safe operation.
- dotancohen 5y ago
  > Even in the case of the 737, the MAX proved we didn't > understand a 737 thoroughly enough to ensure safe operation. Actually, we did. But it was thought economical to ignore proper engine placement and weight distribution, and make up for it with automated elevator trimming. Economical being a conservative word. Designing a new airframe for the newer, larger, more efficient engines would have been very expensive.
- laverya 5y ago
  NASA not running an actual assessment after STS-27R was almost lost to a similar foam strike in 1988 is really the flabbergasting part. If that foam strike had been almost anywhere else on the wing, it would have resulted in the crew and vehicle being lost. But instead it punched a hole right where there was a double-thickness aluminum plate for an antenna mounting, and so didn't burn through. Columbia had a foam strike on the leading edge and wasn't so lucky.
14 5y ago
I think we will lower the bar of what is acceptable. You seem to think they need perfect track record or it’s a no go but we know that is not true. People are willing to take the odds to do something awesome. Look at the first time one of Tesla’s self driving cars caused a fatality. No one stopped using their auto pilot. It will be the same if a rocket full of people blows up, they will investigate the issue and build a safer one the next time. Then with the prospect of heading to space another crew will eagerly line up.
- dotancohen 5y ago
  > they will investigate the issue and build a safer one the next time The parallels between the new Starship and the Comet are many. As excited as I am to see the new Starship flying, and I would fly it, I wouldn't want my children to fly on it. At least, not the landing part. If they could launch and perform all exatmospheric activities in a Starship yet land in a Dragon (or Soyuz) I'd be fine with that.
  - an9n 5y ago
    I think it's going to be a long time and many many launches of Starship before a human is put on it, at least for launch or reentry. Having a reliable separate launch system in Falcon 9/Dragon gives SpaceX a ton of flexibility in terms of launch/reentry on a relatively proven system, with transfer to orbital Starship for Moon/Mars as necessary.
    - dotancohen 5y ago
      I agree that it will take many launches before a human is put on Starship. That said, SpaceX will probably be able to relaunch early Starships with a few months of their initial launches, and might be able to refurbish a Starship for launch within a few weeks in short order. I would not be surprised to see a single Starship perform three launches and landings by the end of 2022. We could see a dozen total Starship launches and landings in that same timeframe. We might see people fly on it by 2023. That's not a long time.

chinathrow 5y ago

> The failure profile that we are in now is that there may be somethings that engineers accepted the risk that can lead to loss of mission, vehicle and life. How did you get to that conclusion regarding Crew Dragon?

idealmedtech 5y ago
Crew Dragon only has a handful of missions. What the previous commenter is talking about is risks that may have been ignored/downplayed/unsuccessfully mitigated but that Crew Dragon is simply lucky enough not to have faced. With time and more flights, more things will have a chance to go wrong, and in spaceflight, resting on your laurels is the best way to make sure that when it does go wrong it will be catastrophic.
- chinathrow 5y ago
  I agree regarding the handful of missions argument with Crew Dragon, but let's not forget all the successful Cargo Dragon missions here as I'm sure they play a role in (ongoing) safety assessments.
  - InTheArena 5y ago
    Sure. It helps. One of the things NASA considered was Shuttle-C, a cargo varient of the space shuttle system, which would have given NASA a way to tinker on the engine /SRB stack without crew onboard. Of course, that would not have saved columbia (the orbiter itself was the problem). This stuff is called rocket science for a reason.
    - progre 5y ago
      Well, still... It's not exactly brain surgery is it?
- bryanlarsen 5y ago
  NASA calls it "normalizing of deviance", and it's why we lost both Shuttles. For example, the O-Rings were designed as a backup, the force generated by the solid rocket motors was supposed to seal the SRM sections together. But they knew after the first launch that the O-Rings were what was actually sealing the SRM's. But the O-Rings worked, so they didn't stop the program to fix the design. Later they found evidence of blow through the O-Rings, and again decided that since that didn't cause a crash, they were OK. Until they didn't. Tile strike damage was a similar normalization that killed 7 people. Does SpaceX normalize deviance? They fixed the toilet, so they didn't normalize that, but what about the delayed opening of the 4th parachute in the Crew 2 landing? It seems suspicious that they launched Crew 3 so quickly after an apparent anomaly. They do claim that the delayed opening is nominal behaviour seen and accepted in testing -- that this behaviour only happens when the other 3 parachutes are working fine. If one of the others had failed the 4th would have opened quicker due to increased air pressure. https://arstechnica.com/science/2021/11/double-dragon-nasa-plans-a-spacex-splashdown-on-monday-launch-wednesday/ https://arstechnica.com/science/2021/11/double-dragon-nasa-p...
  - rbanffy 5y ago
    > that this behaviour only happens when the other 3 parachutes are working fine I'd love to see the causal relationship between both that makes it safe to land with 3 parachutes (I remember they added the 4th as an added precaution that reminds me of the O-ring issue)
  - thaeli 5y ago
    NASA seems to normalize deviance when they're under political pressure to do so. I'm not sure they can be entirely trusted when it comes to their own launch systems, because of the politics involved - but so far they haven't shown similar hesitancy with commercial launch systems. It's a good separation of concerns; when NASA is "just the customer" and it's not "their rocket" they are in a better position to hold a firm line on safety issues. That NASA did not see the parachute as an issue speaks volumes. It looked bad visually, but having watched enough space launches and how mission control reacts to off-nominal data - clearly they were comfortable with the performance, because they didn't flinch at all. NASA just doesn't sign off on these commercial launches with NASA astronauts on board if they have any inkling that something won't go right - they aren't under the political pressure to "save the Shuttle program" or whatever. I have more concern about NASA getting politically-imposed "go fever" with SLS.

ourmandave 5y ago

You'll know they've really reached routine when people's main complaint is about the in-flight meal.

pkdpic 5y ago
You'll know they've really reached routine when people's main complaint is about not having enough leg room :^) Also wondering what the timeline comparison is here too. Wasn't one of the problems with Columbia the advanced age of the shuttle? I'm assuming this isn't comparable to the situation with spacex?
- dotancohen 5y ago
  > Wasn't one of the problems with Columbia the advanced age of the shuttle? No. The problem that led to the loss of Columbia was fundamental and well understood - if underestimated - since STS-1 was inspected after its first landing.
  - Fatnino 5y ago
    STS-1 was known to be missing some tiles while still in space. Like as soon as they opened the payload bay doors they could see holes in the tiles on the OMS pods.

elif 5y ago

The programs you listed with 4 failures launched 164 times including test flights. Falcon 9 has provided 127 successful missions and 1 failure. On visits to the ISS it's 37 vs 25. If you look at the ratio of non-human flights to human flights SpaceX is far more conservative of human life.

ortusdux 5y ago
Also, the 1 failure was during a pre-flight static test fire. If it had been a crewed flight, the crew would not have boarded at that point. It's considered a failure because they test fired the rocked with the payload attached, resulting in its destruction. They have since stopped performing this test with the payload attached.
- croddin 5y ago
  The 1 failure in those numbers is actually CRS-7, not Amos-6. If you count Amos-6 there have been 2 failures but as you pointed out crew wouldn’t have been on board and they don’t static fire with payload anymore. Also for CRS-7 the dragon survived and would have been fine if the parachutes had been programmed to deploy in that situation. Either way there have been a lot of improvements and ~100 successful launches since either of those.
  - ortusdux 5y ago
    Ahh, my mistake. This chart is very informative for those interested: https://en.wikipedia.org/wiki/List_of_Falcon_9_and_Falcon_Heavy_launches#Launch_outcomes https://en.wikipedia.org/wiki/List_of_Falcon_9_and_Falcon_He...
  - Fatnino 5y ago
    Amos6 style explosiw would have been surviveable if there was crew on board because of the launch escape thrusters. See Soyuz T-10a
- oxplot 5y ago
  Don't forget that assuming a crew onboard Dragon, the launch escape system (LES) is engaged before propellant loading, so even if the tanks blow up, the capsule will escape the explosion and crew will be safe.

rbanffy 5y ago

> and Columbia (after a further 88). I'll never get over the fact that NASA never made an assessment of tile damage after liftoff in over 100 flights before Columbia. All these machines are experimental in the sense we still don't understand completely what happens with them and treating them like 737's is a huge mistake. Even in the case of the 737, the MAX proved we didn't understand a 737 thoroughly enough to ensure safe operation.

dotancohen 5y ago
> Even in the case of the 737, the MAX proved we didn't > understand a 737 thoroughly enough to ensure safe operation. Actually, we did. But it was thought economical to ignore proper engine placement and weight distribution, and make up for it with automated elevator trimming. Economical being a conservative word. Designing a new airframe for the newer, larger, more efficient engines would have been very expensive.
laverya 5y ago
NASA not running an actual assessment after STS-27R was almost lost to a similar foam strike in 1988 is really the flabbergasting part. If that foam strike had been almost anywhere else on the wing, it would have resulted in the crew and vehicle being lost. But instead it punched a hole right where there was a double-thickness aluminum plate for an antenna mounting, and so didn't burn through. Columbia had a foam strike on the leading edge and wasn't so lucky.

14 5y ago

I think we will lower the bar of what is acceptable. You seem to think they need perfect track record or it’s a no go but we know that is not true. People are willing to take the odds to do something awesome. Look at the first time one of Tesla’s self driving cars caused a fatality. No one stopped using their auto pilot. It will be the same if a rocket full of people blows up, they will investigate the issue and build a safer one the next time. Then with the prospect of heading to space another crew will eagerly line up.

dotancohen 5y ago
> they will investigate the issue and build a safer one the next time The parallels between the new Starship and the Comet are many. As excited as I am to see the new Starship flying, and I would fly it, I wouldn't want my children to fly on it. At least, not the landing part. If they could launch and perform all exatmospheric activities in a Starship yet land in a Dragon (or Soyuz) I'd be fine with that.
- an9n 5y ago
  I think it's going to be a long time and many many launches of Starship before a human is put on it, at least for launch or reentry. Having a reliable separate launch system in Falcon 9/Dragon gives SpaceX a ton of flexibility in terms of launch/reentry on a relatively proven system, with transfer to orbital Starship for Moon/Mars as necessary.
  - dotancohen 5y ago
    I agree that it will take many launches before a human is put on Starship. That said, SpaceX will probably be able to relaunch early Starships with a few months of their initial launches, and might be able to refurbish a Starship for launch within a few weeks in short order. I would not be surprised to see a single Starship perform three launches and landings by the end of 2022. We could see a dozen total Starship launches and landings in that same timeframe. We might see people fly on it by 2023. That's not a long time.