Why USAF Is Breathing Its Own Fumes
Boeing And Lockheed Strike Back Against Elon Musk, Business Insider
The head of Lockheed Martin’s space division approached the issue in these stark terms during an interview with the Financial Times today, saying, “The government has a certification process that I think everybody ought to adhere to.” But certification isn’t quite this straightforward. The Air Force has only ever certified one company to launch its military and spy satellites: ULA. As Keith Cowing, a former NASA astrobiologist and blogger for NASA Watch explained to Business Insider, this very limited experience stacks the certification process against potential newcomers. “ULA has been launching rockets the traditional way since forever and that’s the basis on which the Air Force and NASA builds their accreditation,” Cowing said. “If someone comes along with a new or possibly better way of launching rockets you have an immediate conflict because the old way of doing things is how the new way is going to be evaluated.”
Agree with KC on this one. What is more amazing is the lack of self-awareness among many people inside NASA and industry on how processes and results became so connected as to be seen as one and the same. Then, to make matters worse, the “how” came to be seen as the requirement, not the end result -“what”.
So along comes a new player, asking only that we focus on “what”, and see if different ways of doing business (how) can get the same result. Test launches, ground tests, analysis, all showing a new way to get to “what” is wanted. And it doesn’t get into some people’s heads that’s exactly what they did back in the day, just with less efficient “how’s”.
Well, putting aside the one’s who are totally aware of this, and couldn’t care less, as they will simply defend their turf, for no reason other than to survive.
Also…a very misleading graphic there by ULA, that scorecard. Being an analyst who sees poorly organized or misrepresented data all the time, by mistake or on purpose, some observations:
1. The graph only shows Falcon 9 version 1.1. No such distinctions on model are made for Atlas of Delta (showing only Atlas 501’s, or Delta mediums for example). If you are not going to make such distinctions in Atlas or Delta models, then all nine of the Falcon 9 launches to date should be on the right, not just 4.
2. Time has not been normalized qualitatively: Such a graph will always make anyone who has been launching longer look better. This is like a new car company coming along and the old one saying “do you want cars from a company that has sold only X number of cars, or from someone who has sold 10 times as many?” This may be of great interest to risk averse car buyer who does not want to get into trying to understand what’s really different, what makes one car better or worse. People do this all the time. Nonetheless. it’s a measure that does not bear on which car is better if you do want to do more homework. In our industry, I think we can do more homework! An early adopter view, or the view of a smart buyer, is very different from a “always bought brand X, always will” sort of view. In the former, such a graph about “cars sold” means little.
3. Time again is not normalized quantitatively: A better comparison would be number of launches since start of business, since first launch. I believe you will find that Falcon 9’s have launched nine times in less total calendar days than the first nine Atlas launches. This sort of normalizing starts to tell you who is accelerating growth or who may or may not catchup to another. (Too early to tell just yet). Either way, incomplete and misleading without the time since first launch variable.
“I believe you will find that Falcon 9’s have launched nine times in less total calendar days than the first nine Atlas launches.”
Your comparison would be wrong. During those first nine Atlas “V”, there were also 11 Atlas II & III launches.
Numbers guy. The earlier Falcon 9 launches should not be included because not the vehicle that is being certified. V1.1 is the only vehicle that counts. If you insist that the earlier SpaceX launches count, then you should count all the Delta II launches. And maybe even launches prior to ULA. But what really counts is the vehicles aimed at doing EELV missions and V1.1 is the only SpaceX vehicle currently being considered by the Air Force for certification.
On the other points, time should not be normalized. Total experience and success record count. Not some normalized value that says they are learning at some rate. That’s voodoo mission assurance. But, if you insist on launches since start of business, let’s again count Delta IIs and perhaps earlier Atlas and Deltas, Titans, and Thors.
Let’s put all that on a chart and see how things stack up.
Fair enough, somewhat. To carry your logic forward though, how does new product X ever convince the older product Y customer that product X is just as fine? That is, if length of track record is all there is to consider? Carry this to it’s natural conclusion and I think you’ll soon have a rather absurd conclusion.
At some point a customer either wants to think about what’s really important alongside what’s possible, or not. So even though a new product may never have the track record of the old one, that familiarity of “the devil we know”, BY DEFINITION, a decision to go with a new product is rarely be guided only by length of track record. External circumstances, such as budgets, or a conscious decision to improve on some other metric of importance, is what ultimately justifies taking risks and moving away from a past track record.
That said, I can sympathize with DoD/AF people who see this all as distraction, a quibble over a billion here or there a year, when much higher stakes such as the military sat’s, or guaranteed performance are considered. Sympathizing though is not the same as believing it’s a path forward for DoD to indefinitely support only cost plus, very expensive, monopoly launchers.
Long term it’s in the national security interest to have multiple launchers at low costs, ever improving on reliability, flight rate and response time. Does this responsibility fall to an operational program? Ideally, no. Ideally separate money and development would fund future stuff all the time. As with NASA though, many things in life are not fair or ideal. Zero sum games make it so that current operational programs get the pressure to improve and develop in new directions just as much as anyone.
Re: how to get from X to Y, the Air Force’s approach is a reasonable balance: (a) a block buy to reduce cost now through economies of scale and (b) introduce competition when it is ready.
The fact is SpaceX is not yet certified, and therefore their product may not be “fine”. Length of record is not “all there is to consider”, but a very important factor. Also, cost is not the only thing either. It is, of course, a mix of these and other factors.
I’d agree that having multiple low cost launchers would be good. But aspirations are not always possible. There are structural limitations to the world launch market that inhibit the ability to get much beyond a boutique business, whether its a US company, European, Russian, Chinese, other.
SpaceX doesn’t have all the answers; it’s not even clear they have a viable long-term answer. The jury is still out whether they can survive full contact with the real world of commercial and government spacecraft customers and their demands for mission unique stuff, schedule predictability and flexibility when they want it, and ability to withstand the volatility of cyclical market demands.
NASA and Air Force are giving them a chance, but SpaceX isn’t there yet. I think it will take 5 years for this experiment to play out before conclusions can be drawn. Other companies, investors building launchers should watch closely to see how this plays out.
Interesting that ULA is only interested in allegedly saving the government money now when it has some competition coming up from behind. Also interesting how the exquisite timing of that block buy deal came along just in time to keep said competition from actually sitting down to eat ULA’s lunch for far longer than the time remaining before certification paper-shuffling is completed.
In the meantime, there’s the not-so-little matter of the Atlas V now having clay feet. Or hasn’t news of the loss of the RD-180 reached your neck of the woods yet?
Oh and PS…if I consider Atlas II and Titans and all that, do I get to add in their development costs to compare against Falcons? Their total costs (development plus launches) divided over launches?
Point being, where DoD and NASA both need to go is toward assisting the development of more efficient and effective US launchers. It’s in their interest long term. The overall equation has to shift to less on development, but creating ever more effective launchers (read: flights per year). Over time more private capital needs to be attracted as well, yielding less up-front expense to the government, for cheaper launchers, seeing many multiples of the current yearly launch rates. Eventually, hundreds of launches a year yiedling more revenue than launch companies see now (but lower per launch costs to customers). This has to happen.
So-the question-what will be written one day about how DoD or NASA helped make this vision so? About the EELV program helping or hindering? Or ISS commercial cargo program helping or not? That’s what’s at stake here, not just ULA and SpaceX quibbling over little bits of scraps at the table.
Let’s put all that on a chart and see how things stack up.
That might not be doing ULA any favors. Those earlier vehicles didn’t necessarily share the enviable reliability record of Atlas V and Delta IV.
All lot is wrong there.
A. Companies do not get certified, vehicles do. The USAF did not certify ULA. EELV’s did not need certification since they were developed by the USAF to USAF standards. Certification only applies to vehicles produced by other organizations. NASA certified Altas V because it did not develop it. Delta II did not need NASA certification because it had enough flight history.
b. There is no “traditional” way that ULA launches rockets since because the process is not static, just like the vehicles and company. ULA is made up from Atlas V program from Lockheed Martin, which was “evolved” from the General Dynamic’s Atlas program and the Martin Titan program. Delta IV “evolved” from the McDonnell Douglas Delta II with some shuttle influence. ULA has been taking the best practices from all its legecy vehicles and companies to incorporate into its own. However, at this time, there still are differences in some of the processes between the vehicles.
Companies do not get certified, vehicles do
Actually, processes get certified which results in a certified vehicle. If a new and less costly process is developed. It is benchmarked against not the results, but the old process.
One thing that is driving the USAF and the Aerospace Corp crazy is the continual improvements that SpaceX is making on the vehicle. The current process demands configuration stability because the underlying assumption that underpins the current process is that change in and of itself, reduces reliability. That may have been true 50 years ago, but we live in a different world today where computer aided design systems, coupled with a strong testing program and automation, results in a reliable product whose performance is continually enhanced.
You need look no farther than the current practice on EE parts in procurement for a sterling example of the failure of the process to keep up with the improved processes in the semiconductor industry. When semiconductors as a whole were far less reliable, then that process made sense. However, today commercial parts failure rates are several orders of magnitude more reliable than military class parts.
By holding onto an obsolete process, the cost of military space systems has skyrocketed and schedules have gone completely to hell. I give one example, generically of a power supply for space. If I want this power supply, there is only one approved vendor for it. The lead time is 36 months because no one in the commercial industry uses the process by which the parts in this power supply are qualified, so a contract has to be let, the parts made, and then they are tested to the military standards. This process itself takes 16-24 months. Then the supply is built and tested.
The cost of this one power supply is in the mid 7 figure range. (don’t ask for more details).
I can build or buy essentially the same power supply on the market for about $50,000 dollars and it is just as reliable as the military power supply. The lead time is less than 60 days.
Because I did not follow the process, I cannot sell this power supply to the government. However, if at my own expense I build and fly this power supply several times, eventually it would be accepted, if I then follow the process, which as the circle closes drives up the cost to the same price as the one provided by the incumbent aerospace contractor.
If however, I fly this power supply along with the spacecraft it works with in the commercial realm, after it is proven in flight, no problem, insurance will cover its failure and it becomes an accepted part of the commercial system. The insurance industry cares about results, and only secondarily on process.
This is at the core of the reason that a military spacecraft costs billions of dollars to start with, which is then the justification for the process for the launch vehicle, which also drives up its costs. The contractors don’t give a rats patootie about changing things as their profit as a contractor is guaranteed under the existing process and the more expensive a system is, the more profit they make.
The new head of DARPA has talked about this subject as recently as last week.
so yes, there is a traditional way and none of the contractors want to change it because from their perspective the system works and as a very well known observer of the aerospace industry said to me; The current culture of the American aerospace industry breeds out innovation and entrepreneurship in the higher echelons of management.
This is what Elon is trying to change…..
If I were advising him I would say fig em and focus on dominating the commercial market and work with the capitalists of silicon valley to enable more commercial players in the aerospace sector.
The traditional contractor system is irretrievably broken and as the DARPA head said just last week, this broken system itself is a threat to national security.
I agree. I see SpaceX doing something different on every flight and I love it. We shouldn’t stifle that. Eventually many of their rocket products will settle down more naturally, but I don’t doubt that they’ll keep on pressing in some areas.
Good info about the DARPA head.
As an aside, it would be nice to see DARPA get more money because it seems to be a very effective agency. Hell, I’d like to see another agency like it for civilian applications. We could call it ARPA.
They have an ARPA-E that was created not that long ago. Great idea.
http://arpa-e.energy.gov
“coupled with a strong testing program”
What evidence is there of that? Grasshopper tests don’t count since that doesn’t apply to delivering payloads into orbit. The Mvac was never tested in a vacuum before flight. Lucky that Cassiope did not require a restart.
Have you actually been to SpaceX’s facility? Do you know what their test program is? Do you understand that the software that is in the F9R is also what they will use to land the stage when it is reusable?
If you want answers, you might want to go to the conferences where their test system is described. It is completely state of the art, and extremely efficient. It just does not have the same process as ULA. What counts is do the rockets work and put payloads into orbit. I guarantee you that if the Russians truly cut off our access to ISS through Soyuz that the Dragon II would be the first one ready to go.
I also think that Dream Chaser might not be that far behind.
It was not lucky. They specifically had for 1st F9 v1.1 mission that did not required restart to test that in safe way.
And you seem to not be aware about test stands. Not to mention things that are less visible by nature, like testing Dragon software in virtual Dragon. All you can see is dozen of computers connected with cables.
IIRC Gwynne said that if the AF wanted a static vehicle SpaceX would maintain 2 versions of the F9, a static one for the AF and a version that was on an upgrade path.
More cost for SpaceX but guess they’d build that into their pricing.
So AF, want a static version? Current price + 30% compliance + 30% F9v1.1. Anyone else, F9 v1.x list price.
Cheers
It will soon be more than 30% + 30% = 60%.
The military ones will basically be made to measure.
Hey, Hey, Ho, Ho!
FAR and cost-plus have to go!
So the Falcon will be reliable enough to launch NASA astronauts but not DOD satellites? That doesn’t seem very logical. I-Shih Chang provides an extensive examination of the reliability question in his study “space launch vehicle reliability. “http://www.ewp.rpi.edu/hart… There isn’t much evidence analytical certification procedures are accurate predictors of reliability. Although he finds that track record is an accurate predictor of reliability, most launch vehicles reach a stable level of reliability within about six launches, assuming design improvements are made to eliminate potential failure modes as soon as they become apparent. It appears the Falcon will be well past this milestone by the first manned flight. There isn’t anything to be gained by keeping the design static as long as each changed is tested several times in unmanned launches before lives are entrusted to it.
As to certification, don’t forget the FAA, which has a very expensive process for certifying aircraft designs that has resulted in uncertified aircraft being considerably more advanced and less expensive. if NASA would take a crack at modernizing and simplifying the certification process, that could really benefit US industry.