Falcon Heavy News Coverage

Bob Zimmerman isn’t impressed.

I’d note that Jackie Wattles is one of CNN’s space reporters (the other is Rachel Crane, who I met at last year’s launch, and who just received a space journalism award in February from the Commercial Space Federation, a few weeks after giving birth).

Speaking of FH coverage, here’s Eric Berger’s take.

[Monday-morning update]


43 thoughts on “Falcon Heavy News Coverage”

  1. Elon said he recovered the $6 million dollar fairings and will fly them again on a Starlink launch. So that leaves the upper stage as the only throw away item.

    1. And if you include Dragon spacecraft in the discussion, the trunk is also thrown away, and only the capsule is reused.

      It seems as if the end points, the booster and the capsule, are the easiest items to make reusable, it is all the middle bits which are hard!

  2. My wife and I have either been involved with or seen at least seven studies of flyback boosters, conducted by both NASA and USAF. At least three contractors independently performed each study, and all of them found the idea feasible. But neither NASA nor the USAF ever attempted to actually develop one. I really admire Elon and his team for taking this on, and doing it so well.

  3. As an aside, I’ve been pondering on cryogenic upper stages from the recent discussions of how to get a Boeing Starship or Dragon 2 into deep space, and I’m thinking such missions would benefit from the development of a much cheaper and much smaller LH2/LOX engine, somewhere in the 5,000 lbf class, for clustering (redundancy) and to drastically shorten the bell to make it easier to fit inside a payload fairing or interstage.

    Existing engines, whether American, Chinese, European, or Russian, all start at about 14,000 lbf thrust and go up from there, with the smallest US option, the RL-10, coming in at over 24,000 lbf. Put that behind any of the new capsules, sans a big hypergolic service module, and they would be pushing around 1 G at burnout.

    Since not of the craft use small LH2/LOX propulsion systems for thrusters, if the main engine fails all the cryogenic fuel becomes useless. But a small upper stage engine would allow clustering, which opens up some design freedom, and adds redundancy and engine-out capability.

    But nobody has a smaller cyrogenic engine, much less a really cheap one.

    1. Typically stages with a single engine have pretty high reliability.
      So clustering becomes less important in that case.

    2. Also in a lot of cases failures in upper stages are due to failed engine starts rather than engine failures in mid flight. So having more engines should actually decrease reliability.

        1. Having more parts means it is more likely you will have a failure. Also upper stages are typically made to be as lightweight as possible because any pound of extra weight you add there is one less pound of payload. The Falcon upper stage is poorly optimized as it is.

          You must think every single upper stage in existence must have been designed by idiots. I can’t remember any last stage with more than two engines.

          ACES was like the only proposal I can remember which could have more engines than that but it was a) never put into production b) they only used so many engines because they did not have the budget to develop a larger engine.

          1. The Saturn S-IV used six RL-10’s, which was changed to a single J-2 for added thrust on the S-IVB. The planned Exploration Upper Stage for the SLS uses four RL-10’s, though producing slightly more thrust than the six on the S-IV.

            Using six should have pretty good engine out performance, and that’s what I’m focused on. Engine failures will happen, and we can’t rule them out in lunar orbit where no abort is possible. The Falcon 9 has had a couple of engine failures and still went on for a successful mission because it’s got plenty of them and the vehicle was designed to prevent a daisy chain failure. If I was whizzing around the moon I think I’d feel better knowing that a dead engine is no worse than having a bad spark plug in a V-8.

            But it also comes down to cost, which depends heavily on whether the engine was developed with the traditional aerospace (Rocketdyne/NASA) approach or more innovative approaches driving the new launch industry.

            Anyway, SpaceX has already looked at a ton of statistics on multi-engine reliability and how it affects mission reliability, with the amount of tolerable engine-outs growing throughout the first stage burn.

            Certainly you are nine times more likely to suffer an engine failure with nine engines, but the question is whether you suffer enough engine failures to cause a mission failure.

            Suppose the criteria for success was having 7 out of 9 engines running. To achieve a 99.95% mission success rate, each engine would only have to be 98% reliable. Now there would be a single engine failure on 16.7% of the missions, and a double engine failure on 1.2% of the missions, but a triple engine failure, and a mission failure, would only occur 0.05% of the time. Engine reliability could drop to 96% before you’d cross into 99.5% mission reliability.

            And if you were still good through TLI, where you’d need most of the engines, you’re certainly not going to still need all that thrust for a lunar return because most of the fuel was expended on TLI and lunar orbit insertion. Due to mass ratios, getting back would probably only require one good engine. Statistically, other failure modes such as a tank rupture or guidance failure are vastly more likely than losing nine of nine engines.

          2. Having more parts only means an increased chance of failure if everything else is equal, but it’s not. The chance of failure of a manufactured object diminishes with production volume, usually quite sharply in the early going. It’s entirely possible that a mass-manufactured smaller engine would exhibit a failure rate equal or superior to a larger, less-produced engine at the individual level. As evidence, I offer the Merlin. It is the most-produced large rocket engine in the world and, by no particular coincidence, is also the most reliable. It’s only in-flight failure occurred quite early in its production history.

            What really recommends engine clustering, though, is its fault-tolerance. This is exactly the engineering principle that undergirds, for example, RAID disk drive arrays in computing systems – mass produce the components to get the highest individual component reliability, then cluster the units so that the failure of any one unit does not bring down the whole complex. The SpaceX rockets are flying RAID arrays.

            I don’t blame the rocket engineers of the 50’s, 60’s and 70’s for their concentration on single big engines or very small clusters of big engines. But RAID arrays have been around nearly four decades now and, yes, I do blame modern rocket engineers – those employed at SpaceX notably excepted – for failing to grasp an engineering principle that is fully as legitimate as the long-time obsession with parts minimization and which they’ve had plenty of time to notice if they weren’t so obdurately tunnel-visioned on “best” ways to do things that have been superceded by applicable advances in other fields.

            IMHO the aerospace engineering profession has been a largely innovation-free zone for the last four decades – endless refinement of legacy approaches, but no fundamental innovation since the advent of stealth technology in the 70’s. No wonder NewSpace is kicking both legacy contractor and NASA ass by the truckload.

            And all the old-school types seem able to summon up in reply are outraged expostulations like, “But that’s not how you do it!”


          3. I think engine cost kept the legacy aerospace companies locked into the traditional approaches. An RL-10 is about $25 million and an RS-25 is $59 million (after NASA pays $193 million per engine to restart and upgrade the production line to build six more engines).

            So the RS-25 costs about $100 per lb of thrust (vac), whereas the RL-10 costs about $1000 per lb of thrust. The engine makers don’t have an incentive to rectify the pricing, and the engine buyers can’t go against that kind of price curve and justify a cluster of engines in their budgets.

            That happens when most of the work is done with touch labor by experts, and production rates are so low that the workers might be struggling to remember how they built the previous engine.

            Elon Musk took a vastly different industrial approach and his Merlin 1D vacuum probably comes in at about $4.00 per pound of thrust.

          4. Quite right in all respects, Mr. T. Artisanal production technology and insane prices made clustering largely a non-starter for far too long a time.

            It’s worth remembering that the “I” in the RAID acronym stands for “Inexpensive.” The whole acronym derives from Redundant Array of Inexpensive Disks. The inexpensive disks in question were those 5-1/4″ units first mass produced for PC’s. Merlins and Raptors are the rocket engine equivalents of mass-produced small disk drives.

        2. If you want to increase reliability you can simply use a pressure fed hypergolic engine. That has much less chances of failure to start.

    3. Well, I looked at a NASA paper on cryogenic upper stage reliability that was undertaken under the Constellation program.

      The overall failure rate is 3%. The trouble is, if that’s how you intend to get back from the moon, that’s also the fatality rate, although perhaps most of the failures would occur during TLI, where you could salvage a return trajectory.

      The failure breakdown is:
      24% Propulsion control and gimbal devices
      17% Combustion & energy control devices
      17% Turbopump
      58.00% engines and engine systems

      Other failure modes are:
      4% Structural
      4% Propellant tank
      8% Guidance and navigation
      13% Propellant management devices
      13% Pogo
      42% other things.

      For a manned vehicle, I would assume that structural failures, GNC failures, and running out of fuel would become non-issues, and Pogo is extremely unlikely because the TLI and other boosts don’t require high accelerations. If those are ruled out, that means the engine systems would be 77% of the remaining failures, and dropping the overall failure rate to 2.25%.

      With a redundant cluster of engines, able to withstand an engine’s catastrophic failure like the first stage of the Falcon 9, those 77% engine failures should likewise go away, in terms of causing a loss of vehicle, cutting the 3% failure rate to 0.5%.

      The study does mention that the Centaur upper stage has a failure rate of 1.2%, but it took a whole lot of flight experience to get it to that level, and I don’t think we’d have many dozens of unmanned capsule flights to get that kind of experience on a stage specifically aimed at manned deep space missions.

      But in part I’m thinking that the RL-10 is too expensive, and the missions would be much more affordable with a cheaper engine developed in-house by SpaceX, Blue Origin, or someone else who would focus on cost. If they were developing their own engine, a cluster of small ones could add benefits that a large one would never could, not just in reliability, but in offering a simple upgrade path such as going from four to five engines, or seven to nine engines, as mission requirements and vehicle growth might dictate. At present the jump is from one to two to three, which are big increments.

      1. Why LH2 over methane? Better isp? Does that increase overcome the more difficult storage?
        Asking because I have no idea and you do a great job in your explanations

        1. Yeah it’s because of Isp.
          It makes a larger difference on upper stages.

          That said it’s not impossible to use an engine with Raptor technology for it I think. I mean the N1 didn’t use LOX/LH2 either. The engines had even less Isp. The Soviets also didn’t have the luxury of having a base as close to the equator.

      2. Aha. What’s needed is derivative of the XCOR XR-5H25 LH2/LOX piston-pump-fed development engine which was fired in 2013. ULA wants a much bigger (RL-10 size) version for ACES, but here I’m arguing that the small ones might actually be better for a long-term sustainable deep space effort.

        1. You know, I should turn my analysis of a better way to build a stage that’s not being built by using multiple engines that aren’t being developed into a wonderful PowerPoint presentation. I could even include a part that shows how the 2,500 lbf XCOR cyrogenic engine would be perfect for a descent/ascent cluster on the non-existent lunar lander, especially if refueled by ISRU.

      3. This same NASA wants to use Shuttle liquid-fuel engines that shook the daylights out of everything when they were run at partial power. Their excuse for not fixing that particular issue was, I believe, “We will not operate at that power setting, so no reason to fix it.” Now when you want less thrust, you can’t have it. Also, NASA has not left LEO for 46 years or so, so trusting their expertise may be a stretch, since everyone who knows anything about anything above LEO has retired. I say we trust the pot-smoker who actually developed a working rocket that is knocking down the barriers set up by non-performing federal bureaucrats.

    4. Would it even be 1 G? Take a 13 ton Starliner, then slap on a cryogenic stage. That’s at least 2 tons dry mass added. Then use the biggest RL-10, and that only about 0.76 G at burnout. Is that really a problem?

      Supposedly ULA selected the RL-10C-X for their new Vulcan Centaur, in part because Aerojet-Rocketdyne came up with cheaper 3D manufacturing techniques for the RL-10 to lower cost.

      Rather than develop an entirely new rocket engine, it might end up cheaper and more reliable to just stick with the RL-10.

      1. I don’t think Tom Mueller at SpaceX is going to buy any RL-10’s. He developed a vastly more powerful (650,000 lbf) LH2/LOX engine, the TR-106, at TRW, along with the LOX/Kerosene TR-107. Then he went on to apply his ideas to the Merlin.

        1. A Dragon 2 with a 9 tonne metholox stage added, and launched from a Falcon Heavy, would have enough delta-V to go to LLO and return.

          The Kestrel pressure fed engine would be a perfect candidate for conversion to methane.

          I was surprised when I recently read of old NASA experiments with the RS-18 engine. If they can convert a hypergolic engine to use LOX and methane, than SpaceX should be able to convert the Kestrel!


          1. I doubt they’d resurrect the Kestrel, but they could trivially make something that’s a bit better and specifically designed for the application.

            The Kestrel used an ablatively cooled chamber, as did the TR-106 and early Merlins (1A and 1b). Then Tom Mueller switched to a regeneratively cooled combustion chamber for the 1C and beyond. So they have the tooling and experience to make regeneratively cooled throttleable engines with high reliability and at low cost. Regnerative cooling would allow them to keep a vehicle in operation across multiple missions. Combine that with fuel depots and a lot of possibilities open up.

            SpaceX also has recent experience building LOX/Methane engines, and the TR-106 was LOX/LH2, so I see nothing in the path that would be much of a real challenge for them in developing any type of engine they want.

            But what I find interesting is that NASA, ULA, and other old-space companies are stuck debating between using RL-10’s, RS-25s, RS-68’s, or J2X’s, which are all hideously expensive legacy engines. They’ve modernized the RL-10 to take advantage of new manufacturing techniques, such as 3-D printing, but is this really dropping the cost by that much? To be truly competitive, they need to go for things like a 20-fold cost reduction (“Now 95% off!”), and why would an engine manufacturer do that if engine demand is relatively inelastic?

            SpaceX could probably develop an entirely new Kestrel type engine for less than the cost of one or two RL-10s. In contrast, NASA spent $1.5 billion to restart and upgrade the RS-28 production line and build six more engines. I think they spent about the same amount for the J2X program upgrades, and they now say they won’t use one until they start development on a Mars vehicle.

          2. Kestrel changes all depends on your application. For a Dragon 2/Starliner propulsion stage, a single use, simple and cheap Kestrel is good enough. Such a stage would be single use anyway, just like the trunk of the Dragon 2 and the service module of the Starliner are single use.

            However if you were talking about an orbital maneuvering engine for something like the Starship, then sure a fancier Kestrel makes more sense.

            In fact, now that I think of it, what is the SpaceX plan for the OMS of the Starship? With all the orbital refueling baked into the cake, the Starship will certainly need some kind of OMS engines smaller than the main Raptor engines.

          3. Starship’s dry weight of 187,000 lbs is very comparable to the Space Shuttle orbiter’s 151,000 lbs, which makes estimating the basic requirements pretty easy.

            The Shuttle’s aft RCS thrusters had 870 lbs of thrust, with 24 lb verniers. The two main OMS engines produced 6,000 lbs of thrust each.

            Dracos only have 90 lbs of thrust, so are going to be too small by a factor of 10 for an RCS system on a vehicle that large, so they’ll probably come up with a mid size Draco with about 1000 lbs of thrust.

            If Starship needs an OMS (since it’s designed to do way more than just make orbital adjustments), they might use a vacuum optimized Super Draco as the lowest cost development solution. The under-expanded SL Super Draco has 16,000 lbs thrust, and I’d guess a vacuum optimized Super Draco would have around 22,000 lbs vac. Just one would have almost twice the total OMS thrust on the Shuttle.

  4. I usually agree with Robert but not this time.
    Arianespace had much the same issues when the Ariane 5 came out. None of their clients wanted to use the larger capacity of the Ariane 5 because if there was some issue with the rocket, they couldn’t schedule it into another launcher. Nor could they get competitive pricing for it. It’s a big issue when you are in a time sensitive business which uses leading edge technology like aerospace.

    In fact Arianespace had to design their own dual payload adapter for Arianespace because of this. Eventually satellites got larger as Proton became available as an alternative launch vehicle.

    So I think until a competitor to Falcon Heavy shows up it is likely commercial demand for it will remain low.

    Other than that I think the main deal about Falcon Heavy is that NASA now finally has a heavy launch vehicle at a low cost they can use. This should make a lot of missions which previously weren’t possible doable.

  5. I’m curious about how the media covered the launch, but I cut my cable in 2012 so I don’t have TV any more. The only media I have direct experience with is the Wall Street Journal, which my boss has delivered at work. They had a photo of the launch and a short article on page A3. Last year’s Falcon Heavy test flight was on Page 1 above the fold. That’s pretty decent coverage.

    I commented on another site last night that if the rocket had blown up, it would have been the lead story on every network. Fiery explosions are always popular with news organizations.

    On the one hand, I find that irritating because it emphasizes failure over success. But on the other hand, routine successful trips by car, train, ship, or airplane are not newsworthy. Only crashes, derailings, and sinkings are. So maybe it’s a good thing when successful space launches are not especially noted in the news. It could be a sign that rocket travel is coming of age.

    1. It isn’t good news to cover a launch live, unless its cable news and they can cut away from other programming, because when they hype it up and the launch is delayed or scrubbed, it turns the audience off.

      I didn’t canvas all the networks but the coverage I did see was like, here is some thing that happened today. Then the show a clip of the lift off and landings.

  6. Ed Minchau made an interesting comment:

    Of the NASA centers, only JPL is a Federally-Funded Research and Development Center. One of the recommendations of the Augustine report was to convert all the NASA centers to FFRDCs. Might be time for VP Pence to revisit that report.

    How would that change the activities and focus of those centers?

    1. The primary change would be that people could be fired much more easily, because employees would no longer be civil servants. Which is why it was strenuously opposed.

      BTW, it was the Aldridge Commission that recommended that, not Augustine (AFAIK).

  7. Regarding the prints, they provide just more evidence that SpaceX has an enormous cost advantage over legacy systems.

    You see, for $4.5 million you can buy a 1965 painting of Borman and Lovell on eBay that was used in an Aerojet press release. Their Gemini mission was built by legacy aerospace companies and flown by NASA, which explains why a painting of it done in one evening would cost so much, on par with a middle tier Van Gogh.

    But you can by actual photographic prints of Falcon 9 landings for anywhere from $15 to $100. There’s no way NASA’s painting can compete with that.

    As an aside, the same eBay seller has a used TRW Apollo LM descent engine chamber and bell for $99K. He also has the LM engine housing for $100K. Either one would make a nifty planter because titanium doesn’t rust. They have that added cool factor because they were critical major components for an Apollo lunar module, which isn’t something you find just lying around – except on eBay, apparently.

Leave a Reply

Your email address will not be published. Required fields are marked *