Why the Air Force is Taking Its Time With New Launch Providers

Atlas V launches OTV3 into orbit from Cape Canaveral. (Credit: Pat Corkery, United Launch Alliance.)
Atlas V launches OTV3 into orbit from Cape Canaveral. (Credit: Pat Corkery, United Launch Alliance.)

Over at InnerSpace.net, Stewart Money is pushing for an end to ULA’s current monopoly on military launches by letting SpaceX fully compete for contracts immediately:

In the meantime, with news of defense cutbacks and the impacts of sequestration, which Administrator Bolden pointed out yesterday is a 10 year program, presented in dire tones almost daily, why exactly is it that United Launch Alliance, utterly uncompetitive on the commercial market, and with no meaningful program of technology improvement remotely on par with that being undertaken by SpaceX, still enjoys a competition-free firewall around 80% of its business, and worse, much worse, is still receiving an annual launch subsidy ranging between $500 million and $1 billion per year?

It’s a good question. The answer lies in understanding how the military performs its duties in keeping the nation safe, and in the different statuses of the two company’s launch vehicles.

If you’re a risk analyst evaluating rockets for the Air Force, there are a few things that you know very well:

  1. ULA’s rockets are extremely reliable.
  2. The launch vehicles have flown enough times for you to be able to quantify just how reliable they are.
  3. The military has a lot of insight into how the launch vehicles are designed, built and operated based upon years of day-to-day interactions with ULA.
  4. The spacecraft you need to launch are extremely expensive and vital to military readiness.
  5. The nature of the payloads means that reliability is far more important than cost.

On the other side of the ledger, you also know the following:

  1. Five launches is not sufficient to know how reliable Falcon 9 will be over the long term.
  2. You have far less insight into the rocket’s design, manufacture and operations than you do with ULA’s launch vehicles.
  3. There have been some serious anomalies during the missions.
  4. SpaceX’s practice of downplaying the seriousness of these problems is alienating the very people who decide on launch vehicle purchases.
  5. The company is about to introduce an upgraded version of the rocket, with changes designed to increase reliability but which also bring uncertainties.

When you add all these things up, it leads to a fairly straightforward calculation:

Yes, you could save money going with Falcon 9s. But, are the saving sufficient to justify the increased risk of going with a less proven rocket? If it works, great. If it doesn’t, the amount you’ve saved is dwarfed by the expense of replacing an extremely expensive satellite. And you’ve now got a hole in the country’s defense posture, the cost of which can’t be easily calculated in dollars and cents.

From that perspective, it makes sense that the military is not rushing to let SpaceX compete for launches at this point. The risks outweigh the benefits at this point. With two reliable — albeit expensive — rockets already flying, there’s no immediate need to add another launch vehicle that is still going through a maturation process. Especially since cost is a secondary issue.

The other positive factor is that SpaceX has other partners that are willing to help it to prove the Falcon 9’s reliability and to retire risks. NASA is paying for a series of cargo resupply missions to the International Space Station. The stakes on these flights are much lower than for national security missions. NASA can easily recover from losing a Dragon full of cargo, which nearly happened on the most recent mission.

SpaceX also has an ever lengthening manifest of commercial launches to fly. Many of those satellites will be going to geosynchronous orbit, a capability the Falcon 9 hasn’t yet demonstrated that is crucial to the military. With the private sector willing to take on those risks, there’s little reason for the Air Force to rush in.

In a broader sense, Money is being a bit too harsh on ULA. The company is hardly standing still in the face of competition. ULA is pursuing a root-and-branch overall of the Centaur upper stage, which includes joint development work with XCOR on a new engine that could dramatically lower costs. None of that is as dramatic as trying to recover the entire rocket for reuse, but these efforts should not be so easily dismissed.

In the meantime, quality control problems with Russian and Ukrainian launches have forced commercial satellite companies to rethink their emphasis on cost savings in favor of reliability.

Satellite operators also showed renewed interest in two vehicles that are not major players in the commercial launch market: the Atlas V and Delta IV, both manufactured by United Launch Alliance (ULA). While the Atlas V has done a handful of commercial launches, its prices have typically kept it out of the commercial market, with both vehicles filling up their manifests with less cost-sensitive US government payloads instead.

“Not many years ago, Eutelsat, like many other satellite operators, used Delta and Atlas,” said Michel de Rosen, president of Eutelsat. “Well, these are not available any more, because Uncle Sam is keeping them busy and rich.”

The satellite executives hoped that ULA could find ways to lower the costs of their vehicles and make them more competitive. The Atlas V “is an excellent rocket, but historically their prices have been out of reach for our industry,” said McGlade. “If we can see a more competitive ULA, one where they can launch more rockets, do it on time, and maintain the reliability they’ve had, I think that’s certainly another option.”

Robert Cleave, president of Lockheed Martin Commercial Launch Services, which markets the Atlas V to commercial customers, hinted at the Satellite 2013 launch panel that the company might be making progress on that front. “We’ve got some great news coming out soon about that,” he said. Both a block buy of launch vehicles by the US government, plus an additional block buy by ULA’s shareholders, Boeing and Lockheed Martin, will drive down costs. “With that, our recurring price becomes very competitive in the marketplace.”

So,  the military has a good (if not optimal) situation on its hands. It has two reliable rockets supplied by a company that is working to reduce launch costs. And there is a new entrant in SpaceX that is not only offering lower prices but has clients willing to take on the increased risks associated with Falcon 9’s early flights.

If all goes well, the U.S. launch industry will become much more robust and competitive, with everyone benefiting from much lower prices on a diverse mix of highly reliable launch vehicles. That will take a few years to develop. In the meantime, a military whose primary focus is on reliable access to space can afford to wait.

  • Can the Government not get insurance on military launches? What is the cost of a SpaceX launch fully insured vs the cost of a ULA launch? I bet SpaceX would still come out as the lower launch cost option.

  • Given Falcon 9’s brief history and the reliability of ULA’s rockets, I’m not sure the insurance rates would necessarily be lower for the former. The overall launch costs would probably be less due to SpaceX’s pricing and the higher costs of Atlas V and Delta IV.

    But, that discussion misses the main point of this post. Cost is not the main driver for military launches, reliability is. The cost of failure goes a lot deeper than the mere cost of insuring and replacing the satellite.

  • This analysis would be quite reasonable if the price differences between F9/Atlas V and FH/Delta IV were in, say, the 20-30% range. However, the Atlas is 2-3 times more expensive than F9 and the Delta IV is 4-5 times the FH. (Don’t forget the $1B annual subsidy to ULA that’s separate from what is given for individual launches.) Such enormous price differences cannot just be brushed off as a minor consideration for military launch policy.

    Even in the best of times, the USAF has a finite budget for its launch and space operations. And today that finite budget is shrinking. For that budget there will be a list of top priority missions. As the budget falls, missions at the bottom of the priority list will fall off the list because they are unaffordable, not because they are judged as unneeded.

    So spacecraft that would otherwise be “vital to military readiness” will not be launched. In addition, the budget crunch means that satellites with important but unproven new capabilities will not be launched because they will be below that priority cutoff. Those new capabilities in some cases would become vital to military readiness once proven but the AF will not know this because they cannot afford to fly them.

    So by not taking advantage of such enormous launch savings, military readiness is undermined.

    With prices differences so big, it is not true that what is saved is dwarfed by the cost of a satellite. Just a handful of launches with F9/FH will save enough to buy a new billion dollar satellite. (Of course, a major motivation for creating hugely expensive satellites in the first place is because of hugely expensive and infrequent access to space.)

    There is no proof to back up any claim that the F9/FH are lower cost because of quality shortcuts. This reminds me of the Detroit Big 3 in the 70s and 80s claiming the Japanese carmakers were somehow cheating on quality, performance, etc. In fact, the Japanese were making higher quality, higher performing cars at lower prices than Detroit because of superior designs, leading edge manufacturing techniques and better industrial organization.

    The EELVs were developed in a manner similar to the SAAs used by NASA for COTS and CCP. I.e. the firms contributed lots of their own money, made design decisions, etc. So claims that the AF has more insight into the EELVs seems on shaky ground. I believe they would have far more insight, for example, into the Merlin 1D than they do into the Russian RD-180 engines.

    Having anomalies on flights of a new vehicle is routine. The EELVs have also had anomalies, including one on the Delta IV that led to a satellite going into the wrong orbit. If anything, the fact that the F9/Dragon systems still completed their missions should bolster confidence in the overall robustness of the systems. NASA has issued no statements whatsoever that they have been less than fully informed on these anomalies.

    It’s reasonable, of course, not to jump immediately from EELVs to F9/FH. However, if institutional bias and inertia were not at play, the USAF would be planning a transition that would start as soon as possible and that would bolster and accelerate the development of the FH, as well as the reusable F9.

  • Cost is not the main driver for military launches, reliability is.

    Or put another way, the cost of the payload is the main driver for military launch services.

    $3 billion per SBIRS, $2 billion per AEHF, $1 billion per GPS III [1]

  • Yeah, that’s exactly right. The launch is a relatively small part of the overall costs of these programs. So, why skimp on the launch vehicle and go with something that still maturing and you believe is riskier? Especially given the time and cost of replacing the satellite you’ve lost and the damage that does to military readiness.

    The military probably needs to focus on reducing the costs of the satellites. That’s where the real cost savings could be found. The focus on launch costs — although important — distracts from that much larger issue And the problem of high launch costs is on its way to being mitigated.

  • The guys who crunch the numbers for the Air Force on reliability and risk don’t view things the same way you do. They are intimately familiar with the many ways that new rockets can fail, and the serious consequences that can result to military readiness when they do. I have no doubt that there is inertia and institutional bias involved, but there are cold, hard numbers behind these analyses. These are facts. And facts are very stubborn things.

    The EELV’s were Air Force-funded projects; Falcon 9 is a NASA one. Big difference. NASA knows more about Falcon 9 than the Air Force. And NASA’s cargo missions have far less at stake.

    I’m not saying that Falcon 9’s are inferior because of shortcuts. The issue is having enough flights to get a good sense of reliability. SpaceX isn’t there yet.

    As pointed out below, the cost of the Air Force’s satellites are very high. To deal with the funding crunch, it would make sense to focus attention there. The issue of launch costs are being addressed, perhaps not at the pace some people would like, but I believe they’re on the way to be mitigated.

  • ” crunch the numbers for the Air Force on reliability and risk don’t view things the same way you do.”

    This is about policy, not about cranking through someone’s
    concocted formula for launch risk. There is no formula that will tell
    you what the country has lost in military capabilities or in
    intelligence gathering because highly promising new systems were never
    flown due to lack of funding. High EELV costs eat up funding and in turn eat up DoD’s space capabilities.

    ” are cold, hard numbers behind these analyses”

    Yes, there are cold hard numbers involved right along side lukewarm approximations, estimates,and out right guesses. You are wildly exaggerating the determinism of such analyses.

    With multi-billion dollar DoD space projects failing one after
    another, excuse me if I’m not impressed with claims of authority by
    USAF analysts on all things space.

    “And facts are very stubborn things”
    Yes, and so are people and organizations when it comes to changing the way they do things.

    “the cost of the Air Force’s satellites are very high.”

    Of course they are high. What is the motivation to split up a satellite’s capabilities into smaller satellites when launch is so expensive and infrequent? The impetus will always be for payloads to cost several times that of launch. High launch costs motivate more and more testing, and more and more redundancy. Infrequent access motivates piling on as much capability as possible while there is an opportunity to get the thing into space.

    If you knew you could cheaply and quickly replace a satellite, then these motivations reverse. Until there are significantly lower launch costs, there won’t be lower cost USAF satellites.

    Lowering launch costs should have been the top priority of DoD space policy since the 70s. US military space capabilities would be vastly greater now if DoD had done so. Accepting high launch costs was a profound mistake by DoD leadership and it looks like they will continue to fail in this regard for another decade.

  • “This is about policy, not about cranking through someone’s concocted formula for launch risk”

    Policy needs to be based at least in part on data. We’ve got more than 50 years of data on launcher reliability. Professionals crunch these numbers and try to determine risks so that policy can be based on something more than what we want to happen. They don’t go around putting forth a “concocted formula for launch risk” simply to justify an existing policy.

  • I don’t understand why you are going on about the launch risk estimates. Maybe the AF guys are so good they even get 4 decimal precision on their loss of mission estimates just like the MSFC guys do for Ares I/V, SLS etc. So they predict the loss of a satellite in the first X number of launches. My point is that with such huge differences in launch price, a loss of a DoD satellite is more than made up for in launch cost savings, both directly and indirectly.

    We’re talking about factor of 5-6 for FH vs D4, ~$100M vs $500-600M. That’s a huge direct savings that easily pays for a new satellite within two or three launches.

    Indirectly, it is well worth risking a DoD satellite to gain the new space capabilities that will be enabled by such dramatically lower launch costs. There will be a flood of new military space system proposals once such launch prices are available.

    This is about policy judgements not about analysts crunching numbers.

    SpaceX has accomplished three times what only three countries with enormous space programs have accomplished: put a large capsule into space, dock it to a station, and bring it back safely. And they did all of it at a tiny fraction of the standard costs for such systems. It’s absurd for DoD to see all that and still claim SpaceX must spend several years carrying out multiple launches before they will take advantage of such cost savings and new launch capabilities.

  • Clark, you said up the thread a ways: Until there are significantly lower launch costs, there won’t be lower cost USAF satellites.

    I hear this one a lot, but based on our inability to constrain the cost of tactical aircraft, which have none of the ‘limited access’ problems you mention, I am skeptical that lowering launch costs will have much impact on the cost of USAF satellites. You may well be correct for this being the case in the commercial market.

    It’s absurd for DoD to see all that and still claim SpaceX must spend several years carrying out multiple launches before they will take advantage of such cost savings and new launch capabilities.

    I don’t think they are requiring several years; based on my (admitadely crude, but that’s what’s in the standard) number crunching and what I’ve been able to find on the new entrant criteria they will probably qualify if they knock out what’s on their manifest for this year. I say probably because, you’ll note in this set of slides it says,

    Air Force officials indicate commercial launches will count toward certification, provided the launch vehicle configuration used is identical to the vehicle for which the new entrant is pursuing certification.

    This could be unfortunate for SpaceX because they are upgrading so furiously. Some of their previous launches may not “count”. Obviously, in any “real” sense slinging hardware counts for building confidence with the customers whether an analyst armed with a simple binomial failure probability model says it does or not. This position also has real costs for the Air Force in the form of possibly missing out on improved capabilities as SpaceX continues to upgrade to service their commercial customers, which as you say, can negatively impact our military capabilities. Hopefully the real position is more nuanced than the powerpoint grunt in those slides: i.e. have a process for configuration control and upgrade qualification and the new rocket will be ‘in family’ enough for the history to count.

    If you are interested in this topic of innovation in the armed forces I’d encourage you to read Ehrhard’s Dissertation; outsourcing risk buy-down to others (DARPA, NRO, or commercial markets in this case) is not new, and military leaders tend to be more worried about negatively impacting current capabilities than taking risks to get new ones that might not pan out. That said, they did take the risk of funding two EELV providers hoping for uptake in the commercial market which did not pan out…

  • These are insightful comments. SpaceX’s practice of quickly upgrading the launch vehicles has resulted in some uncertainty about when to start counting, as you say. Upgrades can enhance performance and improve safety, but they can create unintended consequences. A Falcon 1 launch failed because the upgraded first-stage engine slammed into the second stage after separation. The first stage had increased thrust they hadn’t accounted for in timing the separation. This issue had happened before in other rocket launches. It was a painful and expensive lesson for SpaceX and one of the three failures that helped put the company on the edge of bankruptcy.

  • When the Falcon Heavy has flown repeatedly and proven its reliability, then that will give the Air Force another option for heavier satellites. We’re still talking about something that may not fly for another year.

  • “I am skeptical that lowering launch costs will have much impact on the cost of USAF satellites. You may well be correct for this being the case in the commercial market.”

    Like NASA projects, DoD projects start off on top of a high pedestal of infrastructure/fixed costs. So I definitely agree that DoD sat projects will never come close to commercial sat costs regardless of launch prices. However, lower cost access will still have an impact, especially on new systems that currently are not being funded because DoD can’t afford them.

    ” don’t think they are requiring several years;”

    As I understand it, the current plan looks to start using SpaceX launchers in the 2015 time frame but it will be late in the decade before SpaceX could take a substantial share of DoD launches. That will represent billions of dollars in lost savings. Let SpaceX compete for those other 36 launches as well as the 14 or so that DoD currently will allow to be competed.

    “outsourcing risk buy-down to others (DARPA, NRO, or commercial markets in this case) is not new,”

    (Thanks for the reference.) Yes, I’m aware that some of DoD’s projects tried to use commercial systems and these projects did not work out. I think it is important, though, not to generalize over situations that differ significantly. (DoD procurement is hardly a stellar performer for any kind of project, regardless of the level of commercial involvement.) Launching payloads to orbit for the DoD is not a one of a kind type of operation like many of these other cases. There is no significant difference between launching a big com-sat to GEO for Intelsat and USAF. The size of the systems are not so different and they both demand high reliability.

    “they did take the risk of funding two EELV providers hoping for uptake in the commercial market which did not pan out…”

    For DoD the EELV program worked out very well. They got two reliable launchers that back each other up and for a relative bargain price. If Boeing and L-M had not put in their funding, it’s most likely DoD could have funded only one launcher. Boeing/L-M lost some money (or at least did not make as much as they hoped) because their launchers turned out to be too expensive for the sat-com market, which also was a smaller than they had hoped. That’s the sort of thing that happens in the commercial world. They worked out an arrangement that required more funding for DoD but at the end of day this was still to DoD’s advantage compared to a single launcher program solely funded and run from the start by DoD..

    Having a new competitor enter the market is also the sort of thing that happens in the commercial world and it should be taken advantage of by DoD.

  • There is no significant difference between launching a big com-sat to GEO for Intelsat and USAF. The size of the systems are not so different and they both demand high reliability.

    There is a significant difference: the factor of 10+ cost difference between a commercial sat and a national security payload. I’ll grant that dividing the cost of the program by the number of satellites does overstate things because the replacement cost (the cost you’d have to insure) would be lower because of a lot of that cost is non-recurring engineering. I think you are focusing on the similarity in launch operations and missing the point.

    Of course I agree with you that DoD should take advantage of the opportunity presented by an emerging provider with the credible promise of capability to compete in their market.

  • “There is no significant difference between launching a big com-sat to
    GEO for Intelsat and USAF. The size of the systems are not so different
    and they both demand high reliability.”

    SpaceX hasn’t launched a big comsat to GEO for anyone yet. All their flights to date have involved small spacecraft that have not gone further than the space station orbit. They have a really long manifest of commercial comsats to launch, a list that has continually slid to the right as they have worked to master flights to LEO.

    They also have used a version of the Falcon 9 that is being retired now.

    They’re about to introduce a larger version of the rocket that is 30 to 40 percent more powerful and includes redesigned engines that have much greater thrust, longer fuel tanks, and three or four other major changes. Those are significant changes that may well enhance reliability, but they also introduce additional risks that need to be retired.

    Doesn’t it make sense to wait a bit until SpaceX has actually demonstrated the ability to repeatedly and flawlessly deliver comsats to GEO using the upgraded version of its rocket before letting them compete to launch payloads that cost $1-3 billion apiece? Falcon 9 hasn’t done that yet, Falcon Heavy hasn’t even flown yet.

  • Aviation Week has a great story today on how the Air Force is handling new launch vehicles: http://www.aviationweek.com/Article.aspx?id=/article-xml/AW_04_01_2013_p59-562693.xml&p=1

    I’ve summarized the key points here: http://www.parabolicarc.com/2013/04/02/competition-on-the-way-for-u-s-military-launches/

    I thought Scott Correll’s comments were interesting about how most of military’s payloads are in medium or geosynchronous orbit, locations to which SpaceX has never sent a satellite. Not only does he want to know that the upper stage will work, he’s also concerned about whether SpaceX has the bandwidth to perform its commercial missions and to do the military missions and certification. He doesn’t question the company’s engineering ability.

  • The upper stage concern:

    The RL10 is a single-point-of-failure for the Atlas V and Delta IV
    Evolved Expendable Launch Vehicles, underscoring the need to ensure a
    similar problem cannot happen in the future. “We have to find out what
    happened and why, because there is no plan B,” Shelton says. “The cost
    of a launch failure would be staggering,” both in the loss of an
    expensive satellite and in terms of operational impact because forces
    have become so dependent on spaceborne services for their work.

    USAF Still Looking for Cause of RL-10 Low-Thrust Problem

    The SpaceX marketing:

    No Assured Access. EELV initially intended to guarantee assured access to space by having two launch vehicle providers—Boeing with the Delta and Lockheed with the Atlas. The premise that two providers provides assured access breaks down with the Atlas and Delta now since both vehicles rely on the same upper-stage engine and share production facilities. A production failure or disruption at a single location can impact the entire DOD space program.

  • CoastalRon

    Doug said:

    “We’ve got more than 50 years of data on launcher reliability.”

    Well, that’s not all from one rocket, but lots and lots of different designs. The EELV Atlas V is far different in every aspect than the Atlas D that lofted the Mercury capsules. And the Air Force committed to using the Delta IV Heavy as the successor to the Titan IV even though the Delta IV was a completely new vehicle.

    The SpaceX Falcon 9 can trace it’s engine history back to the Apollo LEM motor, so if you wanted to group misc U.S. Government programs together you would have to lump in the Merlin too.

    The Air Force didn’t have any alternatives to relying on the Atlas V and Delta IV EELV’s, so of course they bought rides on them. But now they do have alternatives, so it does make sense to wait until a new launch system has shown what it can do. But according to NASA and the Air Force, that number is about three launches, which for the Falcon 9 v1.1 should be accomplished by the end of this year.

    Combined with the other Falcon 9 v1.0 launches, that should give the Air Force enough company and vehicle history to start saving significant money in the near future.

  • CoastalRon

    Doug said:

    “SpaceX hasn’t launched a big comsat to GEO for anyone yet.”

    No doubt this is a factor that needs to be considered.

    The way the commercial market handles it is by buying backup flights on other launch systems, which apparently they do anyways. The Air Force could implement such a system too – there are known options for dealing with this type of situation.

  • CoastalRon

    Doug said:

    “SpaceX’s practice of quickly upgrading the launch vehicles has resulted in some uncertainty about when to start counting, as you say.”

    I don’t know if I’d say “quickly” – it’s been over 5 years since they had announced the Falcon 9 v1.0. In some ways this might have been slow compared to the rocket evolution that was happening in the 60’s.

    But I think any customer is going to be looking for more than just performance data. They want to see how consistent a company has been, it’s financial performance and backing, and how they deal with adversity.

    Keep in mind that the Air Force didn’t have any alternatives to their EELV program – they were getting Atlas V and Delta IV regardless how their future performance ended up being. SpaceX will have proven their Falcon rocket system far in advance to flying any Air Force payloads.