What was the general response to these claims? Most in industry other than the fanboys treated their claims with healthy skepticism.

Eight years later, even after that company successfully nails a picture-perfect launch and reentry, people are still skeptical that in the end their prices are going to end up much cheaper than anyone else. Heck, even I’m still wondering if they’ll be able to keep the prices they’ve been claiming once they’re really into routine and reliable operations–and I’m about as close as you can get to a koolaid drinkin fanboy without having spittle in the corner of my mouth.

Then there’s this other rocket group. Like the first one, they haven’t actually demonstrated the ability to successfully design and build new rocket vehicles. At least not within my lifetime. They also start making claims about how by implementing some key industry suggestions (this time those found on “Page 38″ of last month’s HEFT report) they can deliver a new vehicle for far less than past experience dictates. Unlike the first team though, this team does have a track record. But it is a track record of 30 years of consistently overrunning budgets and getting major projects canceled.

“But it will be different this time” they say. “If we use the suggestions on ‘Page 38′, we can dramatically improve on the affordability of developing new rocket vehicles.”

Now, it’s not that the suggestions on Page 38 are bad. They’re not. They’re actually pretty good. Just like “using the best practices from Silicon Valley” sounds good too. I’ll admit that I’m kind of curious how on a $20B project they’re going to “Model, test and fly early and often” or “Use small lean projects with highly competent empowered personnel”, or how a project that is more or less designed by Congressional committee is somehow going to “Push decision authority to the lowest level. Trust them to implement and don’t second guess (over-manage)” [Aside: if Congress really intended to allow NASA to do that last one, they wouldn't be specifying the size of the rocket, what hardware it can use, and which contractors they have to maintain contracts for]. I’m also somewhat curious of how many of the items on that list CxP managers would claim they were already doing…

…but leaving all of those specific details aside, I just don’t get why this second group of people gets all offended when the net result from industry is once again healthy skepticism. Especially given their past track record. When you’re trying to get people to entrust you with a multi-billion dollar project that all past experience and your management claim is unlikely to fit within budget or timeline, is it really that offensive when people have a hard time swallowing that somehow one powerpoint slide is going to change everything?

I mean, it is totally possible that like SpaceX, this new team is going to surprise us, and totally knock this SLS project out of the park. Heck, maybe they’ll even come in far enough under budget that Shelby, out of the kindness of his generous soul, will decide to put the savings into commercial crew or propellant depot development. It’s totally possible.

But is it really rude to be skeptical about this outcome?

Just food for thought.

If it fails, commercial space people will switch back to “it was only a test” mode while to Ares-huggers, it will prove, prove, prove that all commercial vehicles (including those with existing proven track records) are all death traps. After all, imagine the national security risk of flying our astronauts on private launch vehicles! I mean, if we’re going to turn LEO crew transportation over to the private sector, we might as well all start learning Chinese and reading the little Red Book, cause them Commies are going to come and sap and impurify our precious bodily fluids.

To me one of the more interesting points is found at the top of page 11.  There were several misleading statements made by several people today about the relative safety of Ares-I compared to commercial crew vehicles.  As Brett put it (my emphasis added):

Second, some have claimed that NASA’s Exploration Systems Architecture Study (ESAS) shows that the current exploration vehicles are safer than commercial crew vehicles. In actuality, commercial crew vehicles were never even analyzed in the ESAS report – the ESAS report only looked at vehicles large enough to carry Orion, such as Ares I and variants of the triple-core Delta IV Heavy, and did not examine the smaller, simple, single-core vehicles, such as Atlas V Medium and Falcon 9 Medium that are sufficiently sized for commercial crew missions.  Moreover, even if ESAS had compared exploration vehicles to commercial crew-sized vehicles, the comparisons would be “apples vs. oranges,” because of the dramatically different tasks of these two types of vehicles.

When Jeff Hanley talks about how the Great Oz and supercomputers at NASA show that Ares-I is 3x safer than commercial launch vehicles, I wonder if he’s ever going to release their analyses for actually commercial crew vehicles, or if he’s being accidentally or intentionally dishonest. Because so far we haven’t been shown any data about the safety of actual commercial crew launchers. So far we have lots of data shown for the risks of using existing or modified commercial launch vehicles for launching a massive spacecraft designed to go to and return from the moon, including significant plane change maneuvers to allow anytime returns (ie Orion). It’s interesting to note that over half of the mass on Orion is the oversized launch escape system needed to get away from an SRB you can’t shutoff, and enough propellant for about 1500m/s of maneuvering to reach orbit and then to do in-space ops. That’s above and beyond the RCS propellant on the CM itself.

Most of the stuff that make Orion so massive are flat-out completely unnecessary for an earth-to-LEO crew capsule. You don’t need those kinds of delta-V capabilities. You don’t need as roomy of facilities, since by definition the flight times should be a lot shorter. Etc. There’s a reason why almost all of the proposed commercial crew systems are able to utilize single-stick launchers like Atlas V or Falcon 9–for an actual earth-to-LEO capsule you really don’t need anything bigger.

This realization that earth to LEO capsules can be much smaller than Orion leads to at least two important corollaries that I can think of:

1. Smaller capsules mean higher structural margins.  One of the existing vehicles most often suggested for commercial crew, Atlas V, was designed for the worst-case loading environment of any of its configurations (in this case I believe that would be the Atlas V 551 or 552).  The Atlas V 552 sees much higher max-Q’s than the 401/402 do, and has a much heavier payload on top, which exerts much larger structural loads on the Centaur stage than are seen in the 401/402 configuration.  While the Centaur structures may not meet the 1.4 magic number NASA likes in some of the bigger configurations, as I understand it, it actually exceeds that number in the 401/402 config most likely used for commercial applications.  The Falcon 9 was designed from the start to meet NASA structural margin specs.
2. No need for strapons.  Only one of the commercial crew ideas I’ve seen so far used a vehicle with strapons for crew launch (Dreamchaser).  This alone should make a huge difference in launcher reliability, since there are less things that can go wrong, less staging events, etc.  Most of the commercial launcher ideas they mentioned in ESAS assumed multi-core configurations.

There’s also the possiblity on the Atlas-V of using a dual-engine Centaur configuration to allow for some upper stage engine-out capability, or running the RL-10 at a derated performance level (not sure if that’s something it can do automatically, or if you’d have to make modifications–if you have to modify it it probably isn’t worth it).  With the much lower max-Q, and the ability to shut off the booster engine in case of an abort, I have a hard time believing that Ares-I is really that much more reliable than an actual commercial crew capsule launched on a commercial launch vehicle that has dozens of flights under its belt.

Food for thought.

The mission uses the following tricks to make things work:

• Dual Engine Centaur for this mission is stretched by 50% and includes an “Extended Mission Kit” to allow for it to function for the ~5 days necessary for the mission (normal DEC dry mass is ~5400lb, and the EMK is ~1750lb and includes stuff like extra hydrazine bottles, more batteries, deep space navigation upgrades to avionics, sunshields, etc)
• Command module does a powered lunar swingby to go to L2, thus cutting down on overall dV requirements (~750m/s total required, 335m/s per leg), thus allowing for a much smaller CSM (possibly with the service module integrated into the command module).
• The Stretched Centaur and the Lander break into lunar orbit and descend to the surface instead of continuing to L2.  I’m not positive if this allows you to land anywhere on the lunar surface or not (this is one of the few big questions for this mission mode).  This avoids the extra dV requirements you normally get for stopping everything at L2 first.
• Upper stage performs part of the landing burn (between LOI and the descent burn it provides about 1950m/s out of the total 3050m/s needed for LOI and landing).
• RS-68A Upgraded Delta-IVH. This upgrade is already in engine testing and is badly needed by the DoD, so there’s a good chance this will work out.  Expected payload capacity I’ve heard is 27mT for the system.
• Instead of carrying a second stretched Centaur as a payload on one of the flights, the Atlas V 552 uses the stretched Centaur as its upper stage.  In order to tank up the LH2, it carries an LH2 drop tank between the lander and the command module.  It gets transfered right after reaching orbit, and gets dumped shortly before TLI.

Here are the major components of the system:

• Command Module: This module is based on the Apollo outer mold line, but only carries two people, and enough life support consumables for the mission.  I budgetted 11,000lb dry and 3250lb of propellant for the capsule (not including RCS propellants).  I assumed hypergols for the stage, with a crappy 314s Isp.  The Apollo CM wet mass was 12.8klb, and the SM weighed 54klb wet, 13.5klb dry.  However, most of the SM mass was due to the CSM performing the LOI burn for the Apollo Stack.  About half of the dry mass of the CSM was the huge main engine, and a good chunk of the remaining mass was electrical equipment and the huge tanks for the 40klb of propellant.  With modern materials, electronics, a smaller crew, solar panels instead of fuel cells, and the much lower propulsive requirements for the Command Module in this architecture, I think 11klb is actually pretty conservative for such a system.  For another comparison the latest CEV numbers I’ve heard (which are pretty far out of date) were ~18klb for a four person capsule.
• Stretched Centaur Lunar Transfer/Crasher Stage:  As mentioned above, this is a dual engine centaur using two RL10A-4-2 engines, but with a 50% barrel stretch to the tanks.  The tanks are actually less than 40% of the dry mass of a centaur stage, but you also need more helium for pressurization of the larger stage…assuming that the 50% greater propellant load requires a 50% higher dry mass should be a conservative estimate.  The idea of a stretched Centaur shouldn’t be too crazy when you realize how many iterations General Dynamics, Martin Marietta, and Lockheed Martin have done on the Centaur just in the past 20 years (including 5m diameter Centaurs for use on Titan IV among other things).  The 1750lb for the extended mission kit is also based on numbers from previous papers LM/ULA has published about converting their stages over for longer-duration missions.  Total dry mass I assumed was 9850lb.  Note that the Atlas V 552 performance numbers also include 5400lb worth of Centaur burnout weight, so you only have to provide ~4450lb worth of “payload” for the Stretched Centaur.  Also note, that if you tank the stretched Centaur up all the way for launch, it should probably increase the payload capacity of the Atlas V 552 a little compared to a normal Centaur, but for purposes of this analysis we’re assuming only the nominal payload of a normal Atlas V 552, to be conservative.
• Single Stage Lunar Lander/Ascender: This stage takes the crew the rest of the way to the lunar surface after the Centaur has provided the first part of the descent burn, and then provides the ascent burn, and the burn to take the crew to the L2 staging point to rendezvous with the Command Module.  I budgetted 1100m/s for its portion of the descent burn, 100m/s to allow for a 90s hover to find the best landing spot, 2650m/s for the lunar surface to L2 burn, and about 50m/s more for contingencies.  This is probably the most aggressive part of the mission.  For this vehicle, I’m assuming a piston-pump-fed LOX/CH4 stage, based off of the piston pump and LOX/Methane engine work XCOR has done  (possibly combined with stuff that we at Masten have done that they haven’t like gimbals, throttling, etc).  The piston pump requires very low net peak suction head, which allows for very low pressure tanks, that can be made of  the LOX/Cryo-compatible Nonburnite composites that XCOR has been devleoping.  XCOR developed the piston pump and Nonburnite composites explicity for making propellant tanks out of shapes that aren’t typical for propellant tanks (in their cases to make the CG numbers work, they wanted to do LOX-filled “wet wings”).  Using this technology, instead of heavy pressure fed tanks and heavy helium tanks, you have lightweight composite tanks that can actually form part of the load-bearing structure of the vehicle.  As I understand it, based on my recollection of their public statements, the piston pumps they’re looking at using scale to about enough flow for a 2500lbf engine in a single pump.  By combining them with the 7500lbf engine XCOR developed (with a nozzle extension of course), you have significantly more thrust than you need for landing.  More importantly, you can possibly make the three pumps operate in a redundant fashion, so the loss of one pump can be tolerated at any point in the mission, and the loss of a second pump can be tolerated through most of the mission.  If done right, the pumps could be “armored” as XCOR calls it, but placed in such a way that they have removable manways between them and the main compartment that would allow for shirtsleeve troubleshooting/repair (the pump compartments would need to be done in a manner that if something went horribly wrong, that any debris/blast would be directed away from the crew cabin…but I can imagine a few ways that could be done).  All told, I’m assuming a 4350lb dry weight, a 9000lb propellant weight, 500lb worth of hardware to be left on the moon, and a 360s Isp.  The LM ascent stage was 4200lb, but held only 65% of the propellant mass, and only about half the propellant volume of this lander, and didn’t have to do landings, and didn’t have to support the crew for as long (about 3 days vs. the target 9 days to give you a week on the surface and 2 days in transity to L2).  But as mentioned above, it used pressure fed tanks, with the mass of a helium blowdown system, had to provide significant RCS capabilities since the stage did not have a gimballed main engine, was using crappy 60s era electronics and electrical systems, and had tanks that were entirely non structural, and also didn’t have access to modern materials like lithium-aluminum or modern composites.  However, the 13,850lb total mass for the lander actually compares pretty well with the 13,510lb currently assumed for the pressure-fed, hypergol-fueld Altair Ascent stage (from this document), which carries 4 crew for the same mission duration.
• Pre-Depot LOX Tank: This ~2.2klb Tank holds ~57.1klb of LOX for the Stretched Centaur.  It includes a docking port (possibly using LIDS technology?), a sunshield, and a Centuar-derived LOX tank.  It gets launched as the sole payload for the Delta-IVH, using up all but about 200lb of its capacity.  But since it is so dense, it might be able to get away with using a shorter (and lighter weight) fairing than is typical for Delta-IVH if that wouldn’t require lots of expensive aero analysis.  This tank, if launched with the LOX pre-chilled can hang out for over a month waiting for the Atlas V 552 launch.
• LH2 Drop Tank: This ~62.5 m^3 tank weighs about 2000lb (with another 2000lb budgetted for connecting structures between the various parts of the launch stack).  It would be housed between the Lander and the Command Module on the Atlas V 552 launch.  It would possibly use 5m tankage derived from the Delta-IV US.  After reaching orbit, the LH2 from this tank would be transfered (using propulsive settling) into the Stretched Centaur.  After the Command Module docks with the Pre-Depot LOX tank, and has transferred all the propellants from that (and discarded the pre-depot LOX tank), the CM and empty LH2 drop tank would separate from the stack, the drop tank would be discarded, and the CM would reattach to the lander much like was done on the Apollo Missions.

Now, this mission model isn’t perfect.  It uses most of the capabilities of the two launchers without a huge amount of margin (except in the fact that the Atlas V 552 with stretched Centaur probably has some margin built in that isn’t being explicitly called out).  And I’m not a fan of launching the crew on an EELV with 5 solid strapons.  It would be a lot easier if you assumed the development of something like the Common Upper Stage that ULA has been talking about recently.  With that, you would have tons more margin (since a CUS would add nearly 7mT of capacity to the DIVH, and probably at least 5mT to the Atlas V 552–possibly enough to go with less or no strapons on the crew launcher).  But it demonstrates that a 2-launch EELV mission using almost no modifications to existing launch vehicles (beyond the Centaur mods) is within feasibility.

The system also has several good things going for it.  First off, it can deliver lunar crew to the surface without a depot.  It doesn’t need Autonomous Rendezvous and Docking (since the rendezvous and docking can be piloted), or tankers to be developed.  It doesn’t need HLVs or 10m fairings (everything can fit within a stock Atlas V fairing).  It doesn’t need really long term LH2 storage in orbit.  It only requires two launches for the mission, and doesn’t put anywhere near as much launch timing constraints as the ESAS architecture does.  It can provide for cargo missions (~19klb delivered mass to the surface assuming that 2klb of the lander stage is in the form of a removable crew cabin, which just happens to be enough to land a Bigelow Module).

And most importantly, if depots do come into existence, it can immediately take advantage of them.  With just an LEO depot, you can both cut down on the number of EELV launches to just one (and use lower-cost systems like Falcon 9′s, Zenits, Ariane-Vs, Soyuzes, future commercial RLVs, etc to launch the remaining propellant).  Also by getting rid of the huge LH2 drop tank, you simplify the stack, remove about 15klb worth of hardware from the Atlas stack , dropping it to the point where it can possibly be launched by a 502 launch instead of a 552 launch (since the stretched Centaur provides almost as much propellant as a Phase 1 Atlas, which was supposed to boost the LEO capacity of the single-stick Atlas to almost 30klb).  Or you could use that saved mass to beef up the lander and/or command module for more capable missions.

If you have both a LEO and an L1 or L2 depot, the Centaur can top itself up again that depot, and provide a much larger chunk of the descent burn to the lander stack.  With enough propellant left over to return to LLO then to L1/L2 after separating from the lander, allowing the Stretched Centaur to be reused multiple times.  With such a system you could actually soft-land bigger payloads than the Altair cargo lander…and you’d have the capability of making the lander and transfer stage fully reusable.  The transfer stage, since it wouldn’t see atmospheric flight, reentry, lunar dust, or even particularly bad thermal environments should actually be reusable for several flights–the RL10 is after all rated for 200 relights.  The lander may be tougher, but by the time you have an L1/L2 depot, you’ve probably had enough time (and enough surface infrastructure built up) that you can work that out to.

Ok, so maybe it’s not so bad of an idea after all.

The key points I previously touched on were why high flight rates are important for RLVs, and that attaining a high flight-rate would require both technology and market development.  While there are many potential markets for launch vehicles that have been discussed over the years, there are three markets–people, propellants, and “provisions”–that I think are particularly suited to early commercial RLV efforts. 

In this article I want to begin making my case for why I think that flying people is one of the most important RLV markets, by sharing an amusing analogy.

The Black Aluminum Analogy
One of the “fun” classes I took as a grad student was on analysis of composite structures. It was interesting, even though most of the class involved lots of matrix math as we worked our way through from first principles till we could understand how multi-layer composite structures behaved. One of the lectures near the end of the course discussed some more qualitative concepts now that we had a mathematical foundation. One of the ideas Professor Eastman drove home during that lecture was that composites weren’t just “black aluminum”. While it is often possible to make a composite that was roughly isotropic (say using chopped fibers in resin) and then use it as a drop-in replacement for an aluminum part, you’d be wasting a lot of the composite’s potential. To truly unlock the potential of composites, you have to understand and take advantage of their anisotropic nature.

For instance, composites allow you to put strength in the areas and directions you need it, while minimizing the strength in directions it isn’t needed. The upshot being that a properly designed composite part for a given application may, and probably should, look drastically different from an aluminum part for the same job. Another upshot is that there are some applications where what you need really doesn’t line-up well with the advantages of composites, and where an aluminum part might have been a much better choice (X33 LH2 tanks anyone?).

Black Aluminum and RLV Markets
I think this analogy is relevant to the discussion of how to use RLVs. It’s possible to use RLVs as though they were ELVs that just happen to be cheaper, or that just happen to come home from work at the end of the day. Most experienced ELV people I know who look at RLVs treat them that way. They talk about how “there aren’t enough payloads to justify developing an RLV today”, usually followed by a comment that “maybe in 20-30 years there might be”. By payloads, they tend to be thinking of satellites–because that’s one of the only things ELVs are any good for. And for a satellite, especially the way they’re done today, reusability is just a nuisance, unless it happens to make the flight cheaper. It’s just that much less payload available.

Now, this isn’t to say that the only thing RLVs have to offer is the potential of lower launch prices. Eventually, I think you’ll see satellites that take advantage of intact abort capabilities, the ability to do on-orbit checkout before release, etc. I’m just saying that for satellites most of the reusability stuff is of only secondary importance.

People though are different.

People more often then not will be flying round trips. For them, the recovery system isn’t some extraneous feature that is only useful if it makes things cheaper–it’s a fundamental part of the service. Being able to make it back home in one piece even when something goes wrong also tends to be more highly valued by breathing cargo.  The interesting thing is that the needs of the personnel transport market actually turn some of the main “drawbacks” of RLVs into strong benefits.  That recovery system is no longer “parasitic mass” that can’t be used for payload–Now it’s services already provided by the launch vehicle that don’t have to be deducted from the payload.   Of course, it is possible for a manned RLV to carry its crew in a separate capsule just like an ELV, in which case you’d lose a lot of these benefits, but it’s never been obvious to me why that that approach makes any sense.

One corollary of this is that a manned RLV doesn’t need to be able to carry anywhere near as much nominal cargo capacity to carry people as an ELV would.   Depending on the details, instead of needing 10-20klb worth of payload capacity for a 4-8 person capsule on an ELV, you might be able to fly a 2-3 person compliment with an RLV that has only 1000-3000lb worth of cargo capability.  In fact, you could consider the Falcon 9/Dragon stack to actually be a ~6klb to orbit 3STO RLV, just as readily as a 20klb to orbit TSTO with a capsule on top. Of the systems you need for a manned spacecraft, most of them already need to exist for an RLV stage–TPS, landing systems, avionics, RCS, power systems and radiators, abort recovery systems, and possibly even some basic life support hardware (if you’re shipping pressurized/biological cargoes like some of the stuff Dragon will be shipping to ISS).

While it’s outside of the scope of this blog post, before I go on, it is worth mentioning that there are two technologies/techniques that accentuate the advantages of RLVs even further–tugs and fast rendezvous techniques. But that’s a discussion for a different day. It’s also worth mentioning that the “black aluminum” treatment of RLVs extends not just to how people think about using them, but also in how people think about developing them. But that is also a discussion for another day.

In my next post in this series, I’m going to discuss a counter-intuitive result that this line of thinking led me to.

1. Overemphasis on Unmanned Cargo: I’ve previously discussed on this blog what I think is one of the key deficiencies of the current COTS approach–the focus on cargo delivery to ISS while ignoring the crew delivery issue.  The problem with a focus on just cargo delivery to ISS is that a cargo delivery capability doesn’t really open up many other markets.  Sure, there may be a few flights here and there for DragonLab, but the reality is that without a passenger delivery capability, there just isn’t much need for such capabilities, outside of NASA’s ISS needs.  Bigelow, for instance isn’t going to be providing much demand for cargo flights if he can’t get people to his station.  The ability to safely fly people to orbit, and to be able to deliver them to/retreive them from space stations is a lot more useful.  Not only would NASA be a potential customer, but also Bigelow, and even free-flights.  And once you start getting more demand for people flying to space, demand for cargo will increase as well.  Basically, by only funding the unmanned part of COTS, NASA is forcing those COTS competitors into markets for which there is little other non-NASA demand.Of course, NASA has lots of reasons for wanting to do things this way. NASA as an institution wants to build and fly its own rockets, and being able to continually point at a US manned spaceflight “gap”, and being able to point at other nation’s anemic manned spaceflight programs as threats, makes it that much easier to continue getting funding for their anachronistic manned spaceflight projects. Think about it. If there were one or two US commercial options for getting people to the space station, do you really think there would be as much urgency for continuing to throw good money after bad on Ares-I? Most congresspeople don’t have much of a vested interest in keeping the Shuttle Workforce humming along sucking up taxpayer dollars. But the idea of there being a “gap” in US manned spaceflight, and US access to the ISS resonates more. Especially when there is no non-NASA alternative. Take that unifying threat away, and all of the sudden convincing Congresspeople in states other than Alabama, Texas, and Florida that it’s a national priority to keep throwing billions of dollars a year keeping people in their district employed, doing something that the market is already providing, is going to be come a lot trickier.
2. The “Skin-In-The-Game” Provision: One of the defining features of NASA’s implementation of COTS under Mike Griffin’s tenure was the requirement that COTS companies match NASA funding, to “put some skin in the game”.  Jorge Frank (whose opinions I normally agree with) really liked this provision, calling it one of the best things about COTS, and stated that “It limits the field to serious providers that are confident that they have a business case for their spacecraft beyond selling rides to NASA.”  The problem is, that speaking from the record, this hasn’t been the case at all.  Look at Orbital Sciences.  They have pretty much stated that they don’t think that there’s any market for Cygnus other than ISS resupply.  SpaceX is trying to do Dragon-Labs, but even then it isn’t a big demand driver.  In both cases, they’re not expecting to make money selling their cargo delivery services to anyone else, but are expecting to make most of their money off of the new launch vehicle that will be developed in order to lift the capsule. Orbital wants to go after the Delta II market, and SpaceX wants to edge-in on the EELV market. But let me come back to that point in a second.Basically, when you combine these first two points, you can see all sorts of perverse incentives created. In order to raise a large amount of cash, you need to have a large and solid market to justify it to investors. But as I mentioned before, there’s not a big non-NASA market for cargo deliveries to LEO stations, and even the NASA market isn’t really that big, all things told. And more importantly, at the time the COTS contracts were handed out, it wasn’t obvious if a COTS competitor would actually get any of that follow-on demand, even if they delivered. Without having a realistic non-NASA source of demand for the capsule part of the equation, the skin-in-the-game requirements pretty much killed the case for anyone trying to propose doing a capsule on an existing launch vehicle. Without developing a passenger delivery capability, the only market that could justify the kind of skin NASA wanted in the game was possibly a launch vehicle market. Which is a good part of why all three of the COTS winners (SpaceX, RpK, and then OSC when RpK couldn’t raise money) were basing their actual market case on developing new launch vehicles.So, not only did the skin-in-the-game requirement make it really hard for entities that didn’t have billionaire backing (or large existing lines of business) to compete, but it also drove the technical and execution risk for the program up by biasing selection towards companies that had to develop both a launch vehicle and a prox-ops spacecraft. Low-technical risk approaches that used existing launch vehicles wouldn’t actually develop hardware that would provide enough non-NASA business to justify enough outside investment to meet NASA’s skin-in-the-game requirements.  Quite frankly, if COTS fails to deliver, there’s a high probability that it will be due to the fact that both COTS competitors need to develop both a launch vehicle and a capsule.
3. Payment for “Soft” Milestones: One of the other distinguishing features of COTS is that it is a firm, fixed-price contract, where payment is only given on achievement of specific milestones.  The idea being that in theory this gives the company a lot more flexibility on how to achieve its goals, while the government only has to pay for actual results, not just for effort.  Unfortunately, this is also a nice theory that got watered down in practice.  If you look at both company’s COTS contracts, you’ll notice that both of them make the vast majority of their money off of meeting “soft” milestones, such as performing design reviews, raising money, etc.  By the time you get to most of the hardware milestones, the government has already paid out most of the value of the contract–which greatly reduces the benefit of this approach.   In fact, if I read SpaceX’s contract information correctly, they get paid for the first two of three COTS demo flights for just getting the flight off the ground–even if it fails.  They don’t have to actually have a succesful COTS mission to collect any but the last payment.One of the key tenets of the original proto-COTS concept Gary Hudson had pitched to O’Keefe’s NASA was that other than an initial pump-priming kickoff payment, all other payments would be for hard technical milestones.  That would’ve reduced the government’s risk a lot, since until technical milestones start being achieved, it only has the kickoff capital at risk.  Second off, it would emphasize rewarding actual successful development of hardware, not just paying for paper analysis like it has always done in the past.There’s actually a fair deal of danger here for COTS and future COTS follow-ons. The worst thing that could happen would be for OSC and SpaceX to collect most of their money, and then have some high profile failures right at the end. The government would see it as having spent lots of money on small space firms, and then losing their shirt. Something like COTS wouldn’t happen again for a long time. Had they stuck to Gary’s suggestion, technical milestones would’ve been earlier in the program, and therefore, if there was a failure, it would’ve been a lot less costly to the government.

Now, I’m not saying that COTS is doomed to failure or anything like that.  I’m hoping and praying that SpaceX and OSC are able to turn this into a success.  I am suggesting though that in the future, for COTS-like programs, that it would be a good idea to make sure that “skin-in-the-game” requirements are better matched with realistic expectations on the size of non-NASA markets, that we be more careful not to bias incentives in a way that encourages larger technical risks than ought to be taken, and that rewards actual hardware success instead of paying a lot for paperwork.

[Additional Thoughts: After getting a good night's sleep, I had a few additional thoughts I wanted to tack on.  First, I wanted to point out what I think is one of the best things about the whole COTS approach--the fact that the government is giving the COTS contractors a lot freer hand in how they go about their development projects.  When you compare this to how NASA's running Constellation, you can see how big of a difference this is we're talking about.  Also, by having fixed-price payments based on technical milestones, it removes the need for anywhere near as much direct oversight, both on company accounting (a big headache for cost-plus contracts), and on the technical side.  If the company doesn't take advantage of specialized NASA resources, and ends up botching a technical milestone, they don't get paid.  The incentives all point in a lot closer to the right direction.

One other comment on the skin-in-the-game question, is to remember the X-33 debacle.  One of the main reasons why LM was given the award (instead of the DC-X team) was that they were willing to put a lot more skin-in-the-game.  The problem is, the willingness to put in money doesn't necessarily correlate with mission success, competence, or even a desire to see the project succeed!  At least one anecdote said that LM put the money in more to prevent the competition from getting something to work than because they really believed on a corporate level that X-33 was going to lead to Venturestar.  On the other hand, I am somewhat wary of giving the whole contract with no skin-in-the-game requirements, and no actual requirements to commercialize things.  While using the other useful features of COTS (firm milestone based payments, less direct overhead/interference) is better than nothing, a good part of the point of COTS was as a pump-priming exercise.  Without incentives clearly placed pushing the COTS winners towards developing these services for commercial applications, a lot of the benefit is wasted.

Lastly, the basic concepts of COTS (fixed-cost milestone-based payments, focusing on areas with a potential for non-NASA customers, etc) could actually be a decent fit for developing other pieces of space infrastructure such as depot, tugs, etc.]

Just a few thoughts from seeing this drawing:

1. It looks like they’re planning on doing a nose-first reentry with a heatshield on the nose taking the brunt of the reentry heating.  A previous rumor (which I may deserve the ignominity of having first spread) was that they were going to try and reenter tail first using the big niobium extention of the engine nozzle as a sort of radiatively cooled TPS.  I asked Elon about this back in January (the only occasion I’ve had to meet him in person), and he just laughed, and then said something about using a normal heat shield and parachutes for landing.
2. With the main propellant tanks empty, the chunks of mass are the engines, the helium pressurant tanks inside the LOX tank, the heat shield, and those mysterious spheres in the front right behind the heat shield.  I’m not sure what those are, but my main guesses would be a) ballast, b) RCS propellants for deorbit and reentry.  There’s a long-shot third possibility as well–the heat shield might not actually be an ablative (or just an ablative) shield.  The tanks could be transpiration coolant.  But I think option “b” is the most likely.  Either way, by locating those tanks up front, it helps drag the CG forward during reentry, which makes the vehicle more aerodynamically stable.
3. Those big black panels at the back look like they could be aerodynamic control surfaces.   By moving them outward a bit, they can possibly drag the aerodynamic center of pressure back enough to make the stage aerodynamically stable.  Alternatively, they might allow a somewhat unstable design to fly by active controls.  There are plenty of examplse of aircraft that would rip themselves to shreds in seconds if it weren’t for their computer controls (due to being very aerodynamically unstable).

All in all, the more I look at it, the more I’ve come to the conclusion that SpaceX could actually pull off recovering this stage.  It may take them a few tries, and as their first stage recovery efforts have shown, there are lots of details that need to be just right for recovery to work.  Not to mention the fact that I still think that splashing turbopump-fed rocket stages in the ocean is a great way to make reusability a pain in the neck.  But I think the fundamental recovery concept is sound, and if they can be succesful enough with their launches, they’ll be able to stick around long enough to make reusability a reality.

I wonder if they can make this work if that will down the road lead others of their competitors to move in that direction for LEO launches.  Centaur is a fairly valuable stage, and I could see a similar configuration possibly making it be reusable as well (though the LM guys would probably use Mid Air Recovery instead of splashdown).  Wouldn’t make sense at current flight rates, but if things like Bigelow’s station come into reality, you might see enough demand to start justifying more creative incremental improvements like this.

I don’t think that these recoverability methods are the end-all, be-all of space launch.  And I don’t even think that many of them are even on the same evolutionary path as the ultimate direction things need to go.  But I do think they’re a useful improvement that could help lower the cost of heavier launches while small RLVs are getting their feet.