Lessons Learned: Rockets in the 1980s and 1990s and now

[Jim]

Spacex doesn't do system engineering because they don't believe in it.  They are slowly being forced into it.

[Robotbeat]

Just because Shuttle is the only partial orbital RLV used operationally as an RLV doesn't mean we have to throw our brains away and do the whole cargo cult "Shuttle is the template" thing.

Nor are you being asked to. There was a lot wrong with the design starting with the funding profile, which eliminated more robust full TSTO. Getting a "non-insane" funding profile is the first "lesson learned".

You could build a rocket without any pyros at all. Or filled with them. For an RLV, it probably makes sense to limit the pyros since you, of course, have to replace them every time. Which is just fine (and probably the optimal solution) on an expendable stage or for a spacecraft whose deployment mechanisms are only used one time.

And yet SX went with non pyro shroud separation, which is a one time event...

No it isn't. Remember SpaceX is seriously trying to figure out how to do fairing reuse.

[Kabloona]

If the pyro initiates and you're failing to separate, then the issue is with your mechanism design.  It is not the pyro's fault.  The Taurus fairing separation failures were due to poor material tracibility and not qualifying the design to the proper environments.  They could affect any system, pyro or not.  They were not tested properly, not becuase they were pyro, because Orbital thought they could get away with not delta-qualifying to new environments.

That is correct. But as I said, those issues slipped through the cracks partly because the sep systems in the flight vehicle couldn't be fully tested pre-flight. Thus the use of pyros hinders end-to-end system testing of the integrated vehicle, and makes it more difficult to eliminate potential failures.

As for qual issues, destructive pyro-driven sep systems are also costly and difficult to qual test, leading to the kinds of short-cutting you mentioned. The  Pegasus stage sep failure could be considered due to bad design driven in part by the pyro issue. The linear shaped charge didn't cut the interstage cleanly. It was a design issue partly because it's also extremely costly and difficult to do destructive testing of pyro-driven interstage separation events, so the Pegasus sep system design was not tested well enough to expose that failure mode during qual. If it had been a purely mechanical sep system, it would have been much less costly to test, and more testing could/would gave been done. As it was, Orbital later added a metal belly band around the interstage to help the linear shaped charge shear the composite fibers.

So yes, once again, the pyros themselves are extremely reliable, but once you incorporate them into a one-shot destructive sep system, the article becomes impossible to fully test without destroying it, leading to limited qual testing that may not expose flight failure modes, and no opportunity to fully test the flight article before launch.

[Kabloona]

Spacex doesn't do system engineering because they don't believe in it.  They are slowly being forced into it.

This is how SpaceX said they did systems engineering back in 2012.

https://www.aiaa.org/uploadedFiles/Events/Conferences/2012_Conferences/2012-Complex-Aerospace-Systems-Exchange-Event/Detailed_Program/CASE2012_2-4_Muratore_presentation.pdf

You can debate the pros and cons of their approach, which is different from the traditional NASA/Air Force methods of requirements flowdowns, verification, configuration control, etc. But to say they don't do any systems engineering at all is a bit much. You can't build complex systems like Dragon and F9 without some method of identifying environments, defining and verifying component requirements, and configuration control. If you're doing that, by definition you're doing at least *some* systems engineering.

[Lar]

Spacex doesn't do system engineering because they don't believe in it.  They are slowly being forced into it.

That's quite an extraordinary claim[1]. I don't think you can build complex systems without doing systems engineering. Whether outsiders peering in recognize it as such? Different question.

1 - extraordinary claims require extraordinary evidence

[Katana]

Spacex doesn't do system engineering because they don't believe in it.  They are slowly being forced into it.

That's quite an extraordinary claim[1]. I don't think you can build complex systems without doing systems engineering. Whether outsiders peering in recognize it as such? Different question.

1 - extraordinary claims require extraordinary evidence

In fact SpaceX once advertized on quality control at beginning .

[Robotbeat]

SpaceX takes a "ready fire aim" approach to vehicle development. Sometimes skipping the "ready." It's not proper "systems engineering."

There are some enormous advantages to this. We've also seen some of the disadvantages.

Hopefully SpaceX can mature their operational systems like F9 to a more sustainable model without losing the advantages for their more innovative systems under development like ITS.

[Kabloona]

SpaceX takes a "ready fire aim" approach to vehicle development. Sometimes skipping the "ready." It's not proper "systems engineering."

There are some enormous advantages to this. We've also seen some of the disadvantages.

One advantage of their systems approach that probably most people can agree on is the emphasis on testing. Testing often gets short shrift from "traditional" aerospace companies in favor of analysis, because testing is expensive and analysis often less so. But analysis often fails to account for real-world conditions, so we get components that *should* work in theory, but have poorly-understood characteristics in the real world that may lead to failure.

The discussion of pyro-driven sep systems above is a case in point. Those tend to be heavily analyzed but insufficiently tested because of the cost and difficulty. "Design systems that are testable, then test rigorously at multiple levels of integration" is a SpaceX principle that other aerospace companies would do well to emulate.

Pretty much all the flight failures of systems I've worked on could have been avoided if sufficient ground testing had been done.

[Robotbeat]

BTW, I think much of NASA could benefit from a SpaceX-like approach of "build first, ask questions later." Anything not operational could probably benefit (I wouldn't recommend it for safety critical ISS operations, for instance...). But the lack of an institutional drive to innovate and a risk-averse culture means that only interns or IRAD projects with little hope of broader relevance get to do it.

[Lar]

... and that's as close to SpaceX as we should get. This is not a SpaceX thread.

[john smith 19]

So can we agree that Systems Engineering is quite important and everyone needs to do it?

That sounds like quite a valuable lesson to learn sooner rather than later.

[Robotbeat]

So can we agree that Systems Engineering is quite important and everyone needs to do it?

That sounds like quite a valuable lesson to learn sooner rather than later.

Not sure that's the lesson. Following traditional systems engineering from the start is unlikely to significantly improve on the status quo.

[HMXHMX]

So can we agree that Systems Engineering is quite important and everyone needs to do it?

That sounds like quite a valuable lesson to learn sooner rather than later.

Not sure that's the lesson. Following traditional systems engineering from the start is unlikely to significantly improve on the status quo.

The lesson is to have $100M+ in your personal bank account so that you are not answerable to investors, and implacable will to not give up.  Then you have a chance to succeed.

[Lar]

So can we agree that Systems Engineering is quite important and everyone needs to do it?

That sounds like quite a valuable lesson to learn sooner rather than later.

Not sure that's the lesson. Following traditional systems engineering from the start is unlikely to significantly improve on the status quo.

The lesson is to have $100M+ in your personal bank account so that you are not answerable to investors, and implacable will to not give up.  Then you have a chance to succeed.

If that's the only way??? we're hosed.

[HMXHMX]

So can we agree that Systems Engineering is quite important and everyone needs to do it?

That sounds like quite a valuable lesson to learn sooner rather than later.

Not sure that's the lesson. Following traditional systems engineering from the start is unlikely to significantly improve on the status quo.

The lesson is to have $100M+ in your personal bank account so that you are not answerable to investors, and implacable will to not give up.  Then you have a chance to succeed.

If that's the only way??? we're hosed.

I think the lesson is pretty clear. Set all of the ventures of the 1990s and afterwards against the Company-that-shall-not-be-named -- the only one that has been successful to date met those criteria.  One other met the personal funding criterion (Beal) but not the implacable will requirement.

[Katana]

So can we agree that Systems Engineering is quite important and everyone needs to do it?

That sounds like quite a valuable lesson to learn sooner rather than later.

Not sure that's the lesson. Following traditional systems engineering from the start is unlikely to significantly improve on the status quo.

The lesson is to have $100M+ in your personal bank account so that you are not answerable to investors, and implacable will to not give up.  Then you have a chance to succeed.

If that's the only way??? we're hosed.

I think the lesson is pretty clear. Set all of the ventures of the 1990s and afterwards against the Company-that-shall-not-be-named -- the only one that has been successful to date met those criteria.  One other met the personal funding criterion (Beal) but not the implacable will requirement.

OSC does not have such deep pockets at beginning.

[HMXHMX]

So can we agree that Systems Engineering is quite important and everyone needs to do it?

That sounds like quite a valuable lesson to learn sooner rather than later.

Not sure that's the lesson. Following traditional systems engineering from the start is unlikely to significantly improve on the status quo.

The lesson is to have $100M+ in your personal bank account so that you are not answerable to investors, and implacable will to not give up.  Then you have a chance to succeed.

If that's the only way??? we're hosed.

I think the lesson is pretty clear. Set all of the ventures of the 1990s and afterwards against the Company-that-shall-not-be-named -- the only one that has been successful to date met those criteria.  One other met the personal funding criterion (Beal) but not the implacable will requirement.

OSC does not have such deep pockets at beginning.

They had $50M 1983 dollars. That's a pile.

[Lar]

I think the lesson is pretty clear. Set all of the ventures of the 1990s and afterwards against the Company-that-shall-not-be-named -- the only one that has been successful to date met those criteria.  One other met the personal funding criterion (Beal) but not the implacable will requirement.

OSC does not have such deep pockets at beginning.

They had $50M 1983 dollars. That's a pile.

Arguably (at least from this remove) they took a "semi oldSpace" route to where they are now (props to them, for sure)... which maybe cut cash requirements a bit?

[Kabloona]

They had $50M 1983 dollars. That's a pile.

Yes, but in 1982 they had nothing but a little seed money. The interesting part of the story is how they grew the seed.

Three basically unknown guys got together and convinced NASA to give them a contract to develop a new, low-cost upper stage (Transfer Orbit Stage, or TOS) based on "off-the-shelf" IUS components, to be designed and built by their subcontractor Lockheed Martin. Somehow they got NASA interested and won contracts eventually totalling $200+ mil to develop and launch two TOS upper stages. (The $50M was VC money raised to develop the concept and pitch it to NASA before they actually won the contracts.) They then used the profits from the cost-plus TOS contracts to fund Pegasus design and build, and then Taurus, and then Orbcomm, etc, etc. Voila.

So even though they're now more or less a "traditional giant," they really did start as three guys with zero hardware and a little seed money who somehow sold VC's and NASA on the idea that giving them $200+ mil (most of which they passed through to LockMart as their sub) was a good bet. And they won big. Not too many startups are going to be able to follow that model.

So yes, it's always going to be easier just to start with $50M - $100M right off the bat. But in rare cases you can succeed by starting with almost nothing, like Dave Thompson & Co. And it'll take a lot longer.

[john smith 19]

I think the lesson is pretty clear. Set all of the ventures of the 1990s and afterwards against the Company-that-shall-not-be-named -- the only one that has been successful to date met those criteria.  One other met the personal funding criterion (Beal) but not the implacable will requirement.

It's a great simple (but depressing) message.

However I'll note that both XCOR and REL are both still around. No they have not yet achieved their goals but they have survived. AFAIK both were founded by 3 people and both came predominantly from the propulsion end of rocket engineering (as did OSC in a sense) this suggests that a group of three can survive over the long term if they don't have a huge angel investor up front.

I'll also note both (when fully funded) have delivered what they promised when they promised to do so.