Commercial Crew - Discussion Thread 1

[raketa]

When I develop my program, it is good to know what I am building, but it is better to develop basic program structure, that is flexible to change and then I test and modify and test until I am happy with result.

Then you should stay away from safety- or life-critical systems.  That is a very different world, where opportunities to iterate in the real world are limited, testing is not a sufficient defense, and mistakes can cause death, dismemberment and destruction.

You think if I am talking about service, it is about deliver astronauts to the orbit death or alive, is it fulfilling requirement? I doubt. How long this company will do business for NASA? For example private company will never allow to fly spaceshuttle, even in it is  last iteration. It will be too risky from business reputation point of view. Only NASA a national interest keep this dangerous machine fly and risk every time lost of crew. How many people lost russians in space 4. And on them was during test fly. Their discipline and quality is hundred times worse then any america company. But the design was dumb easy and tested and tested.....

[raketa]

In my world of IT, CDR approach in complicated project took at least 3 times more time and resources(In simple one 10 times and more). It is important that your initial design is flexible. By testing and modifying to you can achieve result in surprisingly short time. Because real  test is best CDR.

This isn't IT.  Not even close.  Don't even *try* to compare the two.

you know that in todays date majority money spend on spacecraft are  on developing programs for spacecraft. You have requirement, than you write your plan how to achieve it, build program/spacecraft and start to test and modify. I know spacecraft takes more hours and more people, but principle is same. You are developing digital or physical product that is able to provide some service.

[TomH]

It is better to build test modify and test again. If NASA prefer paper work , she is guilty by driving development the way that is more in  risk of delay and cost overrun. When I develop my program, it is good to know what I am building, but it is better to develop basic program structure, that is flexible to change and then I test and modify and test until I am happy with result. With rise of 3D printing I think it is moment to use same method in physical product development.

That is an obsolete approach. The power of modern computers makes it possible to identify and eliminate the  vast majority of those mistakes in a virtual realm rather than having them occur in reality after you've bent metal.  The computer modeling follows complex algorithms finds places where problems occur. These problems are eliminated and the design modified long before you manufacture anything. This does not mean every single problem is identified and eliminated, however most are. Risk is reduced by an order of magnitude or more. Physical testing still has to occur, but the long process of physical trial, error, modification, retrial, etc. is highly reduced.

[SWGlassPit]

you know that in todays date majority money spend on spacecraft are  on developing programs for spacecraft. You have requirement, than you write your plan how to achieve it, build program/spacecraft and start to test and modify. I know spacecraft takes more hours and more people, but principle is same. You are developing digital or physical product that is able to provide some service.

It's really not the same.  Physical testing is extremely expensive and often destructive.  Flight hardware is 3 to 5 times as expensive to produce as non-flight hardware, due to rigorous acceptance testing and traceability requirements.  As a result, the process is heavy on the plan and simulate phase, reserving tests for what can't be reliably simulated.  You make it sound like you start building hardware on a half-baked design.  That is lunacy.

[guckyfan]

That is an obsolete approach. The power of modern computers makes it possible to identify and eliminate the  vast majority of those mistakes in a virtual realm rather than having them occur in reality after you've bent metal.  The computer modeling follows complex algorithms finds places where problems occur. These problems are eliminated and the design modified long before you manufacture anything. This does not mean every single problem is identified and eliminated, however most are. Risk is reduced by an order of magnitude or more. Physical testing still has to occur, but the long process of physical trial, error, modification, retrial, etc. is highly reduced.

You are absolutely right. In that brave new world of virtual testing such a dumb thing like a pressure vessel developing cracks on the first pressure test can never happen.

[abaddon]

Not really, because what NASA has required to this point is design, not fabrication. So from the specific technical and legal definitions, Boeing is ahead. Period. Not that Boeing was my choice, but from a technical standpoint, this is what is clear.

Unless NASA never requires hardware to be built, and the completion of CCtCAP is design, my point stands.  Just because NASA hasn't asked for hardware yet doesn't mean they aren't going to need it, obviously.  CCiCAP requirements for Boeing were quite different from SpaceX but the CCtCAP requirements are going to be much more similar, because the end result is the completed vehicle and operations.

SpaceX is clearly ahead.  That isn't to say Boeing can't catch up... now that they have a firm contract in hand including a guaranteed minimum number of flights I expect they will be more serious in spending the money required to complete the project.

To be clear... SpaceX is behind in CCiCAP.  SpaceX is ahead in the race to get a crew vehicle to space first.

[TomH]

From the Sierra Nevada thread, but relevant here, and similar to recent discussion in this thread:

Lack of progress???  Boeing completed CDR, something neither SpaceX nor Sierra Nevada have done!  Their design is further along than either of their competitors -- SpaceX by a few months but Sierra Nevada by a few years.

So, Boeing is ahead of SpaceX in terms of having their design reviewed and approved by NASA (they are done), but behind in building and integrating the hardware.  SpaceX is ahead of Boeing in terms of actually building their design, but behind in terms of having it reviewed and approved by NASA.

SpaceX is therefore ahead of Boeing in terms of the goal of getting their hardware into space on a test flight first, with the notable caveat that if NASA finds something in CDR that they don't like, it could potentially cause rework of already completed hardware that could set SpaceX back as compared to Boeing.

SpaceX is choosing the riskier approach of proceeding further with hardware implementation before CDR is complete.  It will pay off if they come out of CDR relatively clean.

Please correct me (anyone) if I am wrong here.  I think this is why we keep having the "Boeing is ahead" vs. "SpaceX is ahead" debate; because both are right and wrong.

yes I think you clarify it

He is incorrect. No production has been required, only design. Boeing is ahead.

[TomH]

It is better to build test modify and test again.

That is opinion, not fact, and while it is applicable in some cases, it certainly isn't appropriate for all.

In my world of IT, CDR approach in complicated project took at least 3 times more time and resources(In simple one 10 times and more). It is important that your initial design is flexible. By testing and modifying to you can achieve result in surprisingly short time. Because real  test is best CDR.

In the expensive world of aeronautical research and development, the best test is extensive computer modeling before you begin building anything.

[RonM]

That is an obsolete approach. The power of modern computers makes it possible to identify and eliminate the  vast majority of those mistakes in a virtual realm rather than having them occur in reality after you've bent metal.  The computer modeling follows complex algorithms finds places where problems occur. These problems are eliminated and the design modified long before you manufacture anything. This does not mean every single problem is identified and eliminated, however most are. Risk is reduced by an order of magnitude or more. Physical testing still has to occur, but the long process of physical trial, error, modification, retrial, etc. is highly reduced.

You are absolutely right. In that brave new world of virtual testing such a dumb thing like a pressure vessel developing cracks on the first pressure test can never happen.

Assuming the computer model is accurate and there is no guarantee of that. You still will have surprises during physical testing. It won't happen as often as it did in the old days, but it will happen.

Never say never. 

[SWGlassPit]

For those who have played Mario Kart 64, this quibbling over who is ahead reminds me of the Yoshi Valley level. 



There are multiple paths to the same goal, each fraught with its own risks.  It is next to impossible to tell who is ahead unless everyone is on the same track.  We'll know for sure who is ahead when the first crew members fly.

[TomH]

That is an obsolete approach. The power of modern computers makes it possible to identify and eliminate the  vast majority of those mistakes in a virtual realm rather than having them occur in reality after you've bent metal.  The computer modeling follows complex algorithms finds places where problems occur. These problems are eliminated and the design modified long before you manufacture anything. This does not mean every single problem is identified and eliminated, however most are. Risk is reduced by an order of magnitude or more. Physical testing still has to occur, but the long process of physical trial, error, modification, retrial, etc. is highly reduced.

You are absolutely right. In that brave new world of virtual testing such a dumb thing like a pressure vessel developing cracks on the first pressure test can never happen.

As I said, you still have to test, and problems will be found, but the amount of refinement is reduced. There is a distinction.

[zodiacchris]

Yawn! Let's just wait and see...

[TomH]

That is an obsolete approach. The power of modern computers makes it possible to identify and eliminate the  vast majority of those mistakes in a virtual realm rather than having them occur in reality after you've bent metal.  The computer modeling follows complex algorithms finds places where problems occur. These problems are eliminated and the design modified long before you manufacture anything. This does not mean every single problem is identified and eliminated, however most are. Risk is reduced by an order of magnitude or more. Physical testing still has to occur, but the long process of physical trial, error, modification, retrial, etc. is highly reduced.

You are absolutely right. In that brave new world of virtual testing such a dumb thing like a pressure vessel developing cracks on the first pressure test can never happen.

Assuming the computer model is accurate and there is no guarantee of that. You still will have surprises during physical testing. It won't happen as often as it did in the old days, but it will happen.

Never say never. 

guckyfan was being sarcastic (or at best facetious) by making reference to the Orion pressure test vessel. He is saying you can't depend upon computer modeling. You and I are in agreement; modeling reduces the risk, but you still have to test because there will still be some surprises. The salient point here is that NASA had not yet required any physical testing, only design. Based on that criteria, Boeing is ahead in their milestones.

[joek]

SpaceX is on the verge of completing its CDR. I suspect that they will be done by the time CCtCap actually starts (when the protest is over).

We can hope.  Last word appears to be from a  NASA press release 21-Aug:

SpaceX will conduct a critical design review of its ground systems and mission and crew operations plans toward the end of August as it advances Dragon V2 through development. The company also is coming up on the primary structure qualification for the Dragon V2, which is a more advanced version of the cargo-only spacecraft SpaceX uses to transport supplies to the International Space Station.

NB: Only the "ground systems and mission and crew operations plans" are mentioned; nothing about the spacecraft or launch vehicle.  That may simply be an innocent omission in the press release, but I doubt it; per the original CDR scope:

This integrated CDR will cover spacecraft, launch vehicle, and ground and mission operations systems.

Also, in the original CCiCap schedule, both the pad abort test and the Dragon primary structure qualification milestones preceded CDR (which are now 9 and 10 months late respectively).  No idea if those are CDR prerequisites, but presumably they preceded CDR by 2-3 months for a reason.

In any case, I think it is premature to say that "SpaceX is on the verge of completing its CDR".

[TomH]

For those who have played Mario Kart 64, this quibbling over who is ahead reminds me of the Yoshi Valley level. 



There are multiple paths to the same goal, each fraught with its own risks.  It is next to impossible to tell who is ahead unless everyone is on the same track.  We'll know for sure who is ahead when the first crew members fly.

The discussion regards who is ahead in terms of reaching NASA established milestones, and in this process there is only one path: the meeting of those milestones. I think members here want to see multiple paths, but NASA only sees the path that it established. That is the sourse of all the dissonance in this and related threads.

[Lars-J]

The discussion regards who is ahead in terms of reaching NASA established milestones, and in this process there is only one path: the meeting of those milestones. I think members here want to see multiple paths, but NASA only sees the path that it established. That is the sourse of all the dissonance in this and related threads.

Only in the imaginary world where SpaceX and Boeing set the same milestones.

[Robotbeat]

When I develop my program, it is good to know what I am building, but it is better to develop basic program structure, that is flexible to change and then I test and modify and test until I am happy with result.

Then you should stay away from safety- or life-critical systems.  That is a very different world, where opportunities to iterate in the real world are limited, testing is not a sufficient defense, and mistakes can cause death, dismemberment and destruction.

I am calling you on this. Test is the ONLY true way you can determine unknown unknowns. A billion CDRs doesn't help you find unknown unknowns, just your known knows or perhaps known unknowns.

I don't buy that NASA's typical approach is the only way. A test-heavy approach is far superior, in my opinion, to a paper-heavy approach when it comes to safety. That's why airplanes don't fly passengers until they've flown dozens of times. Adding another layer of paper only makes it so it's too expensive to test and so have to fly with fewer tests.

NASA/Boeing had 2 fatal Shuttle failures. The traditional management largesse actually got in the way of identifying and fixing the problems beforehand that led to these fatalities. They knew about them well before but were too stiff to respond until it was too late.

So yeah, I don't buy the condescending attitude that NASA and Boeing have a monopoly on safety due to their management structure. I do buy that SpaceX has a huge advantage due to flying the main hardware successfully 6 times already.

Also, Boeing is much further away on a lot of key tests by their own choice. They set the bar quite low so they could hop over it. I'm not going to concede that they're thus somehow ahead in any real terms due to that technicality.

EDIT: If NASA/Boeing were better able to adapt and change post-design, they would've been able to save 14 astronauts' lives. Being adaptable is KEY to safety because, unlike the traditional management assumption, humans are fallible and the design as well, no matter how much preliminary work you do. Your best bet is to do a good job on initial design, prototype, test, then use that to inform final design and keep in place the flexibility to identify and adapt to near-misses as soon as they happen and before anyone dies.

[mkent]

Huh?  I'm not trying to be mean, but do you not know what a CDR is?

A CDR is a powerpoint, nothing more. You don't have to (necessarily) do dev testing, and you certainly don't QTP or ATP anything. QuantumG is totally right here.

No, that is very, very wrong.

The CDR is the Critical Design Review.  It is the second-most important milestone on an aerospace development program -- second only to first flight.

It means that the engineering is essentially done (usually defined as 90% release of the build-to packages).  To reach this point, it means the structural, system, and tooling design is essentially done.  It means the software has largely been written.  It means the manufacturing planning is nearly complete, and it means that purchase orders* for long-lead items have been placed.

Simply put, it is the gate between design and fabrication.  If you haven't passed CDR, you aren't in fabrication, you're still in design.  Boeing is done with design and starting fabrication.  SpaceX is almost done with design and about to start fabrication.  Sierra Nevada is still well within the design phase.

If it weren't for the protest, Boeing would likely be turning dirt at LC-41, installing production tooling in OPF-3, and fabricating parts for the structural test article by now.

And, oh, you're not doing any qualification or acceptance testing until after CDR.  (Qualified to what?  The design isn't done yet.)

* probably contingent in this case, as they hadn't won CCtCap yet.

[ncb1397]

Huh?  I'm not trying to be mean, but do you not know what a CDR is?

A CDR is a powerpoint, nothing more. You don't have to (necessarily) do dev testing, and you certainly don't QTP or ATP anything. QuantumG is totally right here.

No, that is very, very wrong.

The CDR is the Critical Design Review.  It is the second-most important milestone on an aerospace development program -- second only to first flight.
No, that is very, very wrong.

You realize the Falcon 9 + Dragon V2 stack is scheduled to fly early next year right?

[Robotbeat]

...and the Dragon/Falcon9 stack has had half a dozen orbital then reentry and recovery missions, too.