Initial BFR (Starship) is not much more powerful than Falcon Heavy

In 2016 when Elon Musk unveiled ITS, everyone thought it was ridiculous and huge. It dwarfed the Saturn V. People were scratching their heads as to how it could possibly launch from LC-39A as pictured, since the 42 Raptors were to produce 128 MegaNewtons of thrust, almost 4 times the Saturn V. SpaceX had only ever launched the Falcon 9 at that point and just recently had started landing them. But now they’re proposing something 4 times the thrust of Saturn V with 42 engines?? Outrageous! Surely that will put a bunch of buildings in danger, exceed the limits of the pad, and cause major problems on the Space Coast. And who even NEEDS a rocket that big?? And the upper stage is a reusable SSTO! Impossible!

In the time since then, ITS (the Interplanetary Transport System which was, before 2016, known as the MCT, or Mars Colonial Transporter–a name not likely to go over well with those with a recent history of European overlords) became BFR (Big… um, Falcon Rocket) and then finally Starship. It went from 42 Raptor engines of 300 tons of thrust, down to 31 of 170 tons of thrust and now, in a tweet, Musk said first flights of the booster would only have “around 20” Raptors:

Raptor has achieved its 172 ton thrust, and so absent subcooling of the propellant, that’s likely where it will be for the first operational missions (or perhaps slightly less for margin).

172 tons of thrust times 20 Raptors (could be 19, which packs better) is just 34MN, or just shy of Saturn V. And in the meantime, SpaceX has launched 2 Falcon Heavies, the first one a lower thrust pre-block-5 version and the last one a full thrust block 5 variant. Its 27 engines worked fantastically and produce about 23 MN of thrust. SpaceX has also been building and test-firing flight versions of the Raptor engine, even testing integration into a battleship demo vehicle, the Starhopper. The initial BFR (I’m still calling it that) will have less than 50% more thrust than the rockets SpaceX already is launching. And its possible SpaceX may use, say, 18 engines and throttle them down to 80% for reliability reasons. That’s basically the same as Falcon Heavy, has fewer engines, and addresses all those concerns about flame trench size, infrastructure risk, etc. (Although I suspect that SpaceX will operate with higher thrust.)

BFR is now no longer absurdly over-sized at all. That talking point is over. It’s easily within their demonstrated capability. Fewer staging events also helps. And landing the Super Heavy booster may be easier than landing 3 separate cores simultaneously (no one knows right now). They switched from carbon fiber to stainless steel for fabrication, but that’s probably a step in the right direction if you want the vehicle to fly realsoonnow. Hypothetically (with almost balloon tanks), stainless has the same mass fraction as a carbon fiber (which needs design knock-downs for cryogenics and oxygen, particularly with out-of-autoclave processes) and similar to SpaceX’s current aluminum-lithium alloy. In practice, it seems SpaceX is still literally hammering out the manufacturing process. They have a method that seems to work with Starhopper, but the mass fraction is terrible (built literally by a water tower company). It seems almost like Sea Dragon.

But they don’t HAVE to have extremely good mass ratio. The upper stage doesn’t HAVE to have SSTO-like capability, not at first. It just needs enough to get to orbit with significant payload, say 50 tons. Perhaps it just needs 6.5km/s. That’s also about the delta-v needed to go from the Gateway to LLO then to the lunar surface and back (well, that’s about 6.2km/s total… 5.2km/s if you’re aggressive with your burns).

The difference between 6.5 and 9.5km/s when your exhaust velocity is about 3.7km/s is: e^((9.5-6.5)/3.7) = 2.24. So while SpaceX might be able to theoretically do SSTO with extremely good mass fraction, they can knock that down by a factor of about 2.25 (including long-duration equipment) and still accomplish the mission. That means they may be able to use a fairly crude manufacturing and heatshield system and still accomplish the initial goals of Starship. In fact, Elon has also been discussing an expendable upper stage for certain missions. That gives more options for SpaceX to insert Starship into some operational role much earlier than you might think.

It’s no longer pie in the sky (although the capability is also much lower, but perhaps sufficient). Of course, most of the professional spaceflight community hasn’t grokked that yet. We’ll see how long that takes. With Starlink, it took the splendid orbital launch of 60 satellites for anyone to even raise the question about night time visibility of 12,000 LEO satellites, even though the likely operational visibility was perfectly predictable when it was first announced.

This entry was posted in Uncategorized. Bookmark the permalink.

29 Responses to Initial BFR (Starship) is not much more powerful than Falcon Heavy

  1. George Turner says:

    Yesterday I was doing some calculations on a comment about the 1968 idea of a Saturn V-B, in which the S-IC stage dropped four of the engines in mid-flight (like the early Atlas) to make a stage-and-a-half booster that could deliver 50,000 lbs to LEO. It’s a horribly inefficient way to get there, starting with 7.5 million pounds of thrust to deliver the same payload as that the Saturn IB delivered on only 1.6 million pounds of liftoff thrust. And the dry weight of the stage, after getting to orbit, would be four times heavier than the payload.

    However, as I was looking at all that I punched in some Merlin and Raptor ISP numbers, since the F-1 engines are pretty poor performers. An S-IC using Raptors (and ignoring the lower density fuel) would have almost 50% more mass after first stage separation, if staged at the same flight velocity. If the second and third stages were scaled up proportionately, the LEO payload would likewise be almost 50% more than a regular Saturn V. So in a three-stage design with cyrogenic upper stages, it shouldn’t be hard to get Saturn V’s payload capability with a little over 5 million pounds of liftoff thrust.

    And that circles around to my point, that SpaceX and Blue Origin are focusing on two stage designs, which are great for LEO, but aren’t very optimal for much beyond GTO insertions. The second stage is doing much of the work of the launch, so by the time they get to LEO they have a low fuel fraction. As the subsequent delta V of the mission gets large, as it would for a lunar mission, the dry mass of the second stage is bigger than the payload. The Apollo design, where money wasn’t much of an object and the required performance was pushing the limits, hewed pretty close to using two stages to get to orbit and the third stage for TLI.

    If that philosophy was applied to Starship, they would concentrate on splitting it between a large re-usable second stage, merely meant to achieve orbit and return, and a deep-space capable third stage and vehicle (which might or might not be integrated together for re-use).

    But that of course depends on whether the added staging complexity is worse than simply making the first and second stages really big. That advantage of going with really big stages early is that they set a lot of the support and logistics decisions, and enhanced upper stages can be added later, whereas significantly enhancing an existing, optimized, three stage rocket generally means starting over with a bigger rocket, at least if big side boosters aren’t still an option.

  2. Chris Stelter says:

    If you do refueling, then it’s essentially the same as having a third stage. Except your third stage is enormous and capable.

    Refueling is a much better approach than three stages, IMHO. It maximizes the benefit of reusability while minimizing the number of stages you need to develop.

  3. George Turner says:

    Well, in a way, refueling means you need four stages instead of three, since the fuel truck has to get launched using two more stages. That still might be simpler than three stages, although you have to figure in a rendezvous with a fuel depot.

    A little while ago I rethought my idea of converting a Raptor to run an added RP-1 turbine, while retaining methane for the fuel-rich pre-burner, which can’t have sooty exhaust products clogging things up downstream.

    If instead they switched from methane to alcohol, which also burns clean, they would hardly have to modify the engines at all. The gain in first stage fuel mass and fuel fraction would more than offset the decrease in ISP.

  4. Chris Stelter says:

    George Turner:
    The difference is those two extra stages can be identical, so you don’t add significantly to the development costs. In fact, SpaceX had shown the same booster being used for tanker flights, so only 3 stages are needed, two of which are identical and two of which stay at Earth for reuse (and the third being reused for several mission purposes). But the biggest advantage of refueling is it allows you to multiply the capability of a rocket by an order of magnitude or more. And no depot is technically required (although it may be desirable, particularly with hydrogen) as you can just refuel with a tanker or with an unmodified upper stage.

    For instance, BFR is supposed to have about 150 tons to orbit (although initially 100 tons, maybe less if thrust is reduced), like Saturn V’s 140 tons. However, if refueled in orbit completely, the upper stage and the payload weigh 1200 tons in LEO, meaning it has performance equivalent to a 1200 ton-to-LEO extremely heavy lift three stage rocket. In fact, since you can even refuel it in highly elliptical Earth orbit, it acts similar to a 3000 ton-to-LEO three or four stage extremely heavy lift rocket. That is how SpaceX could realistically do a fully reusable single-stage lunar lander with 100 tons of payload (or 100 passengers) with a rocket of about the same thrust as Saturn V and without refueling on the lunar surface. Or, a 3 month transit to Mars (or potentially faster if refueled in high elliptical Earth orbit and using an advantageous transit opportunity).

    Of course, you could do a similar refueling concept with a much smaller vehicle. At almost an extreme, Jon Goff points out that the actual on-orbit mass of a typical GEO sat is about 2-3 tons after injection into geosynchronous proper. So a 2-3 ton RLV, if refueled several times, could inject a payload to GEO equivalent to about a typical 15 ton ELV like the heavier variants of Atlas V.

    It’s an incredibly powerful concept, and it’s criminal that NASA is not leveraging it for the 2024 lunar architecture to enable a single-stage, fully reusable lander.

  5. George Turner says:

    I’m a huge fan of fuel depots, but what concerns me with the two-stage solution is that half the launch system required to put something into orbit is tied up with being the orbital vehicle.

    It doesn’t seem like a big deal, but one of the Shuttle program’s development directors (I forget which one) said that when the flight rate stayed low long after they’d expected it to ramp up, they took a long look at why. It was payload integration. So they focused on that as being the turn-around bottleneck. They never made any headway, and the flight rate never increased over the 30 year program because it just takes a long time to integrate a payload to a Shuttle and make all the required configuration changes for each mission. So the average flight rate for a Shuttle was 1.3 missions a year, sometimes reaching a bit above two missions a year.

    A re-usable second stage that always gives it all to get the maximum amount of useful mass to LEO is a nice cut point between the launch system and whatever payloads it is putting up. The second stage can then make an immediate return, go through inspection and maintenance, and get back in the flight queue so it can launch whatever gets stacked on it without any real need for customization or integration.
    That should help keep the launcher crews and the payload crews out of each other’s hair.

    Starship is a very big vehicle, and it has the potential to out-Shuttle the Shuttle in turnaround time. It’s potential payloads are much larger and more complex. It’s targeted mission duration is longer, in some cases much longer. The configuration you’d want for LEO deliver or assembly missions is very different from how you’d configure it for a passenger or lunar mission, in terms of installed systems, consumables, stowage, and such.

    Hopefully their design will minimize most of the potential snag points, having focused on turnaround time and ease of access, but they might find that they need a very diverse fleet of Starships, tailored to different types of missions, and to have several vehicles in process at once, just in case it takes months to turn one around for another launch.

  6. Chris Stelter says:

    Vehicles operating as tankers don’t have complex integration at all. Just fuel it up and launch again. Three stages is just unnecessarily complex.

    SpaceX has already demonstrated a far greater operational tempo than Shuttle ever did (21 launches last year compared to 9 for Shuttle at its peak), while being approximately just as reusable (maybe more) and certainly a LOT less expensive. Lots of people take the wrong lessons from Shuttle. Shuttle doesn’t prove that reuse doesn’t work or that you can’t integrate payloads fast with an RLV. It also doesn’t prove that human spaceflight has to be expensive. It proves that a nationalized system with tons of political and design constraints that are not revisited is not likely to be super competitive.

  7. George Turner says:

    Well, I agree that NASA has figured out the slowest way to do anything, but SpaceX’s flight rate for commercial payloads isn’t a completely sequential operation. They could take a year to integrate a payload and it wouldn’t effect them as long as they had the floor space and personnel for each mission to be prepped in parallel with the others.

    As for their cargo Dragon missions, they still haven’t matched NASA 1965 to 1972 average flight rate, if Saturn IB test flights are included. My cause for concern is that the NASA personnel who didn’t foresee the Shuttle’s turnaround and refurbishment times were highly experienced at maintaining a high flight rate with even more diverse set of conventional launchers, some smaller than Falcon and some much larger. They thought the problem was having to build a new rocket and capsule for every mission, and the Shuttle was going to end that. After landing, they just kick the tires, put a new satellite in the cargo bay, gas it up, mount another pair of SRB’s, and off they’d go. Despite their experience, they did not see the coming problems because those problems were of a different kind.

    People solve the problem they’re focused on, leaving what they didn’t anticipate to plague their first solution. In a test of Starship’s TPS tiles back in March, Elon Musk said “Transpiration cooling will be added wherever we see erosion of the shield. Starship needs to be ready to fly again immediately after landing. Zero refurbishment.” That sounds eerily familiar.

    But experience teaches, and the advantage of new space is that they make mistakes faster, and they correct them faster. If Starship v1.0 has a major turnaround bottleneck, within six months there will be Starship v1.1, and then v1.2 and 1.3, in contrast with NASA that spent 30 years flying their first and very problematic attempt at a re-usable vehicle.

    Going back to the thermal protection issues, NASA would check all the tiles, looking for loose or damaged ones and replacing them as necessary, not because they wanted to slow the turnaround, but because each tile is potentially critical. SpaceX is going with better tiles than the Shuttle, as they’re starting from the tougher and cheaper tiles from the X-37B program, and stainless is going to have vastly more underlying heat tolerance than aluminum. But they’re never going to get to a place where they don’t have to inspect their tiles because a tile can always suffer an impact or get loosened. A Starship with a hole burned through the wrong bit of skin has a good likelihood of making a crater instead of a nice powered touch down.

    Once that sinks in, I’m sure SpaceX will come up with a snazzy tile checker that will use robotic feelers, laser scanners, and ultrasonics to check the whole vehicle in an hour, perhaps just by towing the Starship through the scanner assembly.

  8. Chris Stelter says:

    The goal of zero refurb for a TPS is not an impossible goal, and it’s obviously desirable, so I don’t see how it’s “eerily familiar.” It’s just the same goal. Metallic TPS *does* generally work better in this regard due to being intrinsically tougher than the foam-like Shuttle TPS.

    The difference between SpaceX and NASA is that while NASA saw the huge problems with turnaround of Shuttle with the heatshield tiles, it took them decades to really start implementing higher density tiles that need less refurbishment. An iteration cycle so long that the vehicle retired before it finished. If SpaceX has any problem, however, it’s that they iterate TOO fast. I think they’ll figure out what they need to do. It might not even involve tiles per se.

  9. Pingback: Space-for-All at HobbySpace » Space transport roundup – June.3.2019

  10. Jonathan Goff Jonathan Goff says:

    An interesting point that Dan Rasky (the NASA Ames TPS guy who helped teach SpaceX how to use and make PICA) made at Space Access this year is that while ceramics are “mechanically brittle” as he put it, metallic TPS is “thermally brittle”. I think his point was that metallic TPS tends to be not very robust to hot spots and such. He didn’t think it was impossible to solve, but suggested it as a wrinkle that SpaceX might not be fully appreciating yet. I still think it’s something they can iterate their way through, but depending on the severity of issues (ie if it regularly costs them a Starship in the early days), they might end up regretting iterating on so big of an upper stage.


  11. Jonathan Goff Jonathan Goff says:


    I totally agree that refueling is better than having a third stage–I think refueling an earth-to-LEO optimized TSTO is also better than trying to do a fully reusable earth-to-GTO TSTO… Ie I still think that Starship is way too big, but I’m glad they’re toning down the crazy initially. Having their first full RLV be something that’s about as massive as a Saturn V still smacks of hubris, but you never know.


  12. George Turner says:

    … they might end up regretting iterating on so big of an upper stage.

    The best place to iterate is the upper stage of a Falcon 9, fabricated from stainless, covered with their TPS, and built as a scale model of the latest Starship design. I suspect they’ll do exactly that because it makes so much sense for shaking out not only the materials, but the control laws for their proposed re-entry and landing method, plus validating the aerodynamic modeling of the ascent phase (given the pre-deployed landing legs that aren’t fins, gosh darn it!). Plus, SpaceX prefers flight test to years of modeling and paperwork.

    Regarding Starship’s use on things like lunar missions, remember when we all made fun of movies that showed a Space Shuttle making a rendezvous with a comet or falling into a wormhole on the way to Venus? “But it’s a flying brick! It makes no sense to send it anywhere but low orbit!” We were not wrong, no matter how big the imaginary booster stage.

    The Soyuz approach is a different design path, segregating crew re-entry (and abort) as a distinct and separable role from the overall orbital vehicle. The logic of fuel depots also leads to on-orbit consumable replacement and resupply, which removes the need of return the entire vehicle for refurbishment. That leaves the crew as the only item really requiring re-entry, and that path optimizes toward stuffing them in a PICA covered phone booth and shooting them out like a sci-fi escape pod.

    Although it lacks useful delta V, the ISS is a good example of such an optimized flight vehicle, with the capsules as its re-entry escape pods. We were coming at it from a completely different direction, building and maintaining a space station (as a career path), but physical constraints and technical capabilities tend to produce working solutions that reflect elements of good designs for a given environment.

    Following that logic, the number of Starships in space is the number of Starships launched minus the number returned to Earth. If the number return to Earth goes to zero, the number in space will be equal to the number launched, and we get a sky full of Starships, perhaps with each new ship resupplying twenty earlier ones.

    But the validity of that concept depends on cost. There was no way we were going to fill the sky with Space Shuttles because they were just too expensive to build. The ISS is the most expensive object man has ever built, and it has a maximum crew of six. If a simple and repeated maintenance or resupply task is a hundred times more expensive to do in orbit than in a hanger, returning a Starship is vastly cheaper that not returning it.

    We have so little manned space experience, and what we have is so dominated by the market-distortions of giant government programs, that we’re lacking a lot of the capital and labor data that we’d want before our spreadsheets to start spitting out optimal architectures for sustainable space development.

    In that light, even very bad architectures that get built are vastly more useful than Power Point vehicles, in that they teach us things that we didn’t otherwise figure out on the ground, and the new real-world information will get incorporated into subsequent design approaches. I don’t have to bet that Starship will succeed to bet that it will lead to success, just as modern aviation benefited from the bad approaches as much as the good ones. Until someone has built a 707 and someone else has built a Brabazon, we don’t fully realize why one approach is so much better than the other, and neither could be built without a lot of preceding successes and failures. No one can yet say whether Starship is a 707 or a Brabazon, but as long as multiple private efforts keep rolling forward, we’ll definitely find out.

  13. Chris Stelter says:

    George Turner: “Regarding Starship’s use on things like lunar missions, remember when we all made fun of movies that showed a Space Shuttle making a rendezvous with a comet or falling into a wormhole on the way to Venus? “But it’s a flying brick! It makes no sense to send it anywhere but low orbit!” We were not wrong, no matter how big the imaginary booster stage.”

    Missing the big picture again.

    Shuttle, of course, has very little delta-v without its external tank. If Shuttle were refuelable and had large internal tanks, it’d be another matter. With a suitable reentry strategy (skip) and a properly designed heatshield, being a flying brick is a huge advantage as it allows aerocapture (or skip entry). Aerocapture (or direct/skip entry) are enormous advantages for interplanetary missions, and even with the usual NASA battlestar galactica architectures, aerocapture cuts the IMLEO in half. And it requires a “flying brick” heatshield anyway, so reusing it also for upper stage recovery is a clever approach if you can pull it off (as one of the challenges of aerocapture is launching or assembling the large heatshield).

    And the key isn’t the size of the booster. It’s reuse and refueling. You must grok that. A shiny space future is simply not going to happen without reuse and refueling.

  14. George Turner says:

    Well, if you’ve got a Starship in orbit, it’s halfway to anywhere, as they say.

    However, using it as is for direct lunar missions, instead of relying on techniques like staging and lunar orbit rendezvous, requires a really large mass ratio. This causes problems.

    To send a Starship direct to the lunar surface and get it back for atmospheric re-entry should take a delta V of about 8,650 m/sec. But it still has to do a re-entry burn and touchdown, which they can probably accomplish with another 1,500 m/sec delta V, maybe much less depending on how much drag they can get for the sideways re-entry. Or they might not match the entry burn delta V of the Falcon because the human crew is a bit fragile, which gets into some slam-landing debates. But the landing fuel had to go all the way to the moon and back, and it brings the total mission delta V to 10,150 m/sec.

    The Raptor’s vacuum ISP is 380 seconds, so the required mass ratio is 15.24:1, which is a fuel fraction of 93.4%. According to some sources, the Raptor propellant mass is 1,104,545 kg, so the fully fueled orbital wet mass would be 1,182,120 kg, leaving a landing weight of 77,574 kg. Unfortunately, that’s 7,400 kg less than the listed dry weight for Starship, so the payload is -7,400 kg. With some weight reduction efforts, or maybe fueling it in a higher orbit, they should be able to get the useful lunar payload up to zero. But they have to be sure to leave plenty of margin, because if they come up a little bit short of fuel for the final landing on Earth, the result is a big fireball on network TV.

    But they’ll surely get the numbers to work or they won’t even attempt the mission. Regardless, they still have to fuel the Starship in orbit. It’s LEO payload is listed as 100,000 kg, so let that first launch carry the 100,000 kg as fuel. Then they launch another Starship ten times to finish filling up the first one, or use 16 or so Falcon Heavy missions. Depending on which launcher you use for refueling, you’re still looking at launches totaling 12 to 20 times the total liftoff thrust of a Saturn V to fly one lunar mission with possibly no useful payload.

    The cause of the payload problem is trying to get too much delta V out of a single stage rocket. If it was refueled on the moon, or just left on the moon, that problem would go away, as it would if Starship was used to only go to LLO and back, dropping off a dedicated lander while it’s there.

    Another option for Starship is to not land in Florida at all, because the crew has probably already seen Disney World, and just use the atmosphere for aerobraking back into LEO. That saves the mass of landing fuel, which translates into much greater lunar payload, and saves the cost of launching it back into LEO again. This gets back to the age old question: If you go to the enormous trouble of getting a big ship out of Earth’s deep gravity well, why would you turn around and drop it back in if you don’t really have to?

    But that gets back to launch cost versus construction cost versus orbital refurbishment cost, and those are numbers we haven’t completely nailed down yet.

    Or you could draw up a mission, say using Blue Origin’s cryogenic engines, that hewed reasonably close to Apollo for the crew return portion and devoted the rest of the Starship’s LEO mass to a one-way lunar payload delivery. If you allocated 20,000 kg to the capsule (twice as heavy as Orion), and a 10,000 kg (dry) ascent module atop a 10,000 kg (dry) descent module (4.6 times bigger than the LM), the crewed round-trip portion would only require 14% of the LEO fully fueled wet mass of the Starship. That leaves 86% available for the cargo delivery, so you could set about 240,000 kg of cargo on the lunar surface. That’s almost three times the dry weight of Starship.

  15. Chris Stelter says:

    You don’t have to refuel it just in LEO. Musk said (and some of us predicted) you would want to refuel it in an elliptical Earth orbit, which drastically reduces the required delta-v.

    Also, it doesn’t take 1.5km/s to land. Starship is subsonic before it needs to land and doesn’t necessarily need a reentry burn, so it’s more like 500m/s or so. Skip or multi-pass reentry helps address heatshield and deceleration constraints.

    As far as launch costs, the whole point about reusable launch vehicles is that the marginal cost to launch is extremely low. And that’s doubly true for tanking flights where the only payload is propellant.

  16. George Turner says:

    500 m/sec for touchdown sounds good to me.

    My previous calculation was assuming the landing fuel counted against payload, but it probably doesn’t. 500 m/sec with a Raptor’s SL ISP of 330 would be a mass ratio of 1.167, and given the dry weight of 85,000 kg, the minimum re-entry weight would be 99,202, on top of which is the 100,000 kg payload to LEO.

    I figured an elliptical orbit was the way to go, so I punched in the most obvious elliptical orbit, GTO, which would easily convert to a TLI by a nice Oberth burn at perigee. So from a 200 km LEO with a velocity of 7,784 m/sec to a GTO with a 9,880 m/sec perigee velocity takes a delta V of 2,095 m/sec. That would then later be turned into a TLI (normal LEO dV of 3,130 m/sec) by a final 1,034 m/sec burn at perigee, stripping 2,095 m/sec off the fully-fueled mission requirements.

    It sounded great, I thought about an hour ago, and then my numbers turned horrifying.

    You’ve got Starship number one in LEO with it’s minimal weight of 99,202 kg and a fuel payload of 100,000 kg. You do the burn with Raptors at a vacuum ISP of 380, so the required mass ratio is 1.93, which doesn’t sound bad at first. But it means the Starship burns 96,085 kg of fuel to get into a GTO orbit, and it only had 100,000 kg of fuel to play with. So after the burn, its mass is only 103,117 kg, of which only 3,915 is fuel that can be used for something other than the final land burn.

    So if Starship one goes directly from launch to LEO to GTO, it arrives at GTO almost empty. This also applies to subsequent Starships trying to refuel the first one, but those also will need to deorbit, which I think could be done with a 51 m/sec burn at apogee. That burn would require a mass ratio of 1.0139, so their spare fuel at GTO is only 1,715 kg. So it would take about 640 subsequent Starship launches to fill the tank of the first one in GTO. That’s bad. If they launched a refueling mission once a week, a flight rate never even seen before, it would take 12 years to fuel the one in elliptical orbit, which is a lower lunar mission rate than the SLS.

    So going to GTO on an empty fuel tank will definitely not work, and trying to top up a tank that far out by using reusable Starships doesn’t really work either because the amount of fuel each Starship can deliver is virtually nil for a GTO’s delta V. Due to their high on-orbit mass and low payload fraction, they can’t deliver as much mass to GTO as a Falcon 9 v1.0.

    That seems like a really odd result, and maybe I botched a calculation. The Starship is vastly larger and has a higher ISP than a Falcon 9. The required 2,454 m/sec delta V for GTO takes a MR of 1.93 for the Raptors, and the deorbit burn needs a MR of 1.0138, or a total delta V of 1.959. But an unrefueled Starship starts out in LEO with an available MR of only 2.008, so there’s almost nothing left for anything else at that delta V. Now if the 100,000 kg LEO payload weight was an expendable fuel tug, it could deliver almost 50,000 kg to GTO, but that’s adding another stage.

  17. George Turner says:

    Alrighty, I think Starship is fine, but the early Super Heavy isn’t really sized for more than getting Starship up with a pretty big payload (about 4 times more than the Space Shuttle could carry). Starship, the second stage, has to burn almost all of its fuel to get that payload to LEO because the first stage seems to be sized for that particular mission. If the 100,000 kg is payload, Starship’s tanks are essentially empty, with just enough for the de-orbit and touchdown. If the 100,000 kg is propellant, the Starship’s fuel gauge is reading only 9% full. There’s only 2,400 m/sec of delta V in the tanks, assuming there’s no payload in the bay.

    So, taking a page from Falcon Heavy, which can deliver about 2.58 times as much payload to LEO as a Falcon 9 FT, use three Super Heavies to launch Starship. Instead of a minimal LEO weight of 100,000 kg and a payload of 100,000 kg, you could get about 516,000 kg into orbit, which is Starship with 416,000 kg of fuel. The fuel gauge would be reading 38%, the potential mass ratio has gone from 2 to 5.16, and the attainable delta V is 6,100 m/sec.

    Equally important, the second Starship can rendezvous and bump the fuel gauge up to 76%, and the third Starship can fill it up with plenty to spare.

    And if you send the first one into GTO, dropping the fuel gauge to 47.5%, a fourth Starship can fly out to GTO and bump the gauge up to 95%.

    So, that’s four launches of Starship on a 3-core Super Heavy Heavy to get a Starship almost fully fueled in GTO, for a total of 12 re-usable Super Heavy cores. The single core approach took 11 or 12 cores just to fill Starship up in LEO (which this method accomplished with 9 cores) and couldn’t really do anything for a Starship in GTO.

    That seems like a valid way to give Starship true deep-space capability at lower cost.

  18. Chris Stelter says:

    The full Super Heavy (which makes more sense than a 3 core monster) should enable about 150 ton payloads, particularly with down-range landing.

    But a more important thing you’re missing is that refueling in elliptical orbits is most efficiently done with a secondary tanker being refueled in LEO. The tanker fills up, then boosts itself to elliptical orbit. That allows it to act as a third stage for efficient refueling of the Starship lander in elliptical orbit.

    The key to reuse is maximizing the number of flights to minimize the hardware development.

  19. George Turner says:

    Surprisingly, my number tell me that a fully fueled Starship delivering from LEO to GTO is more efficient than using a dedicated tanker. However, the Starship has to be fully fueled, since the barely-fueled mission discussed earlier turned out not to be rather useless due to the mass ratio requirements to GTO and the Starship’s high dry weight.

    I’m assuming (all published numbers at this point are speculations) that Starship’s dry weight is 87,000 kg, the fuel allowance for landing brings the minimum orbital weight to 99,020 kg, and the fuel capacity is 1,104,545 kg. The start is 200 km Kennedy LEO and the delta V to GTO is 2,454.58 m/sec, which for a 380 sec ISP requires a mass ratio of 1.932. So with a starting LEO mass of 1,203,565 kg, it arrives at GTO with a mass of 622,871 kg.

    But Starship has to return for aerocapture, and assuming that only takes a 51 m/sec apogee burn (mass ratio of 1.013), it has to maintain a mass of 100,307 kg at GTO for the return burn. Subtracting that from the GTO arrival mass leaves 522,564 kg of fuel to unload. This means that 46.75% of the Kennedy-LEO fuel was delivered to GTO, which was 43.3% of the wet mass in LEO.

    So for the mission’s 1,104,545 kg of fuel, 12,020 kg (1.09%) was used in the landing burn, 1,287 kg (0.12%) was used for de-orbit to aerocapture, 574,895 kg (52.05%) was used for the GTO injection burn, and 516,343 (46.75%) was delivered to GTO.

    To beat that performance, the dedicated tug has a problem because it probably can’t use aerobraking. Aerobraking back into LEO would introduce lots of thermal heat shield requirements and much more complexity in re-establishing a proper low Earth orbit for its next rendezvous and fill up. But if aerobraking is ruled out, is has to do a burn to get back to LEO, and with the same delta V it had used to get to GTO. The second burn doesn’t need very much fuel because most of that gets delivered, but it still creates a more demanding overall mass ratio.

    To match the 46.75% fuel delivery capability of Starship, which was a mass ratio of about 13.7:1 (from LEO to touchdown), the tug needs a mass ratio of 22.7:1, which means a 95.6% fuel fraction. And it has to be pretty well insulated, capable of docking and fuel transfer (with the requisite pumps and valves), and it’s engines have to be capable of an almost unlimited number of restarts.

    It probably can’t match Starship’s delivery efficiency, and it certainly couldn’t significantly improve on it even if its fuel fraction was 99%. It could go with a fuel that had a higher ISP, such as LH2 or even ion thrusters. LH2 would create a great deal of increased complexity and cost, whereas ion thrusters large enough to boost the huge fuel payloads would be something not yet tried.

    A fully fueled Starship in LEO, with no cargo, and just

  20. born01930 says:

    Wouldn’t the starting weight be 1,104,545 + 87,000? I think you are counting the landing burn propellant twice

  21. George Turner says:

    I think my reply got stuck in moderation, probably from linking the online injection orbit calculator, or else I clicked the wrong button (cancel instead of post?)

    So let me try this again, cutting and pasting from a spreadsheet. I think I’m up to my third spreadsheet that’s grown to over 350 rows just for this thread, as I try to suss out the potential performance and limitations of Starship. ^_^

    From an online orbit injection calculator, the change from a 200 km Kennedy orbit to a common GTO using a 380 second ISP was:

    perigee velocity: 9,880 m/sec
    200 km orbit velocity: 7,784.25 m/sec
    injection delta V: 2,454.58 m/sec
    injection mass ratio: 1.9323

    Fully Fueled Starship to GTO
    Starship fuel capacity: 1,104,545 kg
    Starship landing fuel: 12,020 kg
    Starship dry weight: 87,000 kg
    Starship GTO mass with fuel for aerobraking: 100,307 kg
    Starship in LEO, initial mass: 1,191,545 kg
    Starship after GTO injection burn: 616,650 kg
    Starship fuel delivered (the above line minus the 100,307 kg): 516,343 kg
    fuel delivered as % LEO fuel: 46.75%
    fuel delivered as % LEO initial mass: 43.33%

    Of course those numbers depend on the assumption about the required delta V for the landing, and the assumed weights and fuel capacities are from technical speculations on the Internet, so it’s like analyzing Saturn V performance numbers back in the very early days when Huntsville hadn’t even settled on a design. What SpaceX finally builds might not be that close to what we’re discussing.

    Anyway, my Starship numbers depend on it being fully fueled in LEO to get the mass ratio up, since it would otherwise arrive in orbit with either payload and almost no excess fuel, or with excess fuel and almost no payload. So the attainable mass ratio is actually pretty good, at 13.70:1, with a potential delta V of 9,752 m/sec, some of which is taken up by the landing requirements. Those also cut into the delta V a bit because the final burn is with a sea level ISP. I figured that in for the 500 m/sec landing burn.

    The space tug section of the spreadsheet is simply:

    starting fuel fraction: 99.00% (the input)
    starting mass: 100,000 kg (Starship’s published payload, though this doesn’t affect the percentage of fuel delivered)
    starting fuel load: 99,000 kg (fuel fraction times starting mass)
    dry weight: 1,000 kg (starting mass minus fuel load)
    GTO initial mass required for return to LEO: 1,932 kg (from the 1.932 mass ratio)
    mass at GTO: 51,752 kg (from the starting mass and mass ratio for the injection burn)
    fuel delivered: 49,820 kg (subtracting the mass for the start of the return burn from the mass in GTO)
    fuel as % LEO fuel: 50.32%
    fuel as % LEO mass: 49.82%

    The need for the return burn adds a lot of final mass for the injection burn, cutting the attainable mass ratio almost in half, and that’s what puts even a good space tug back on par with a fully fueled Starship that has a very high dry weight.

    So the tug designers would have to look at the difference between the mass rquired to return to LEO with a simple but large perigee burn, versus the mass they would have to add for aerobraking. The aerobraking delta V is much smaller than for re-entry, however the kinetic energy they have to bleed is almost two thirds that of a full re-entry from LEO. (Ignoring potential energy, 7784.25 m/sec is 30.30 megaJoules per kilogram, versus 48.8 MJ/kg for a GTO perigee’s 9980 m/sec velocity. So it still has to bleed 18.5 MJ/kg instead of 30.30 MJ/kg.)

    They could burn the energy with multiple shallow passes, but that could take days or weeks, and that impacts refueling schedules, which means it might take more active space tugs to maintain the same mission schedules. The other major factor is that unless they add plenty of radiation shielding, they’re not going to want astronauts making repeated GTO trips through the Van Allen belt, waiting for fuel to show up.

    And another thing I wonder about is the Starship’s return payload capacity, which Wiki says is 50,000 kg. That’s 57% of the listed dry weight. Starship’s sideways re-entry profile has to keep the center of mass close to the center of pressure or they’ll have to burn a whole lot of fuel the whole way in to maintain the proper orientation. But given the return payload requirements, obviously the payload will have to ride near the aerodynamic center of the vehicle, much like the Space Shuttle’s cargo bay.

    That’s an interesting design challenge, and one you wouldn’t expect from looking at the renderings of the exterior, which look like a simple rocket with engines at the back and a payload up front.

  22. born01930 says:

    Basically what you are saying is SS has to refill twice in LEO and make 2 LEO-GTO trips to fill up the SS in GTO.
    4 Launches:
    -SS #1 to GTO with payload (frozen turkeys bound for Venus) arrives no propellant
    -SS #2 to LEO needs a top off from SS#3 then LEO-GTO and back for propellant transfer to SS #1
    -SS #3 to LEO tops off SS #2 returns, hits 7-11 for a fill-up and a Big Gulp, then back to LEO to top off SS #2 again

  23. George Turner says:

    I think so. Although I picked GTO out of thin air as a good elliptical orbit, if we stick with it we see that the early configuration (that can only manage 100,000 kg of fuel or payload to LEO) can’t make a direct GTO injection of much more than its own dry mass. It has to get refueled in LEO, or arrive in LEO with much more fuel by using a significantly larger booster, or a parallel booster configuration, to get into orbit.

    This investigation came about because even fully fueled in LEO, Starhip’s mass ratio and delta V doesn’t let it deliver useful cargo directly to the moon, so per Chris Stelter’s observation, it needs to be topped up while in an elliptical orbit.

    The GTO orbit shaves about 2,450 m/sec off the lunar mission’s delta V requirement. The TLI burn needs to get the velocity up to something around 10,400 m/sec, which means a delta V from LEO of around 3,050 m/sec. I’m seeing the GTO to TLI burn as taking anywhere from 450 to 600 m/sec, depending on what I use as a baseline, so let’s call it 550 m/sec instead of the direct from LEO requirement.

    Add 914 m/sec for LOI and 2,073 m/sec for lunar descent and landing, and the delta V on the first part of the mission is 3,537 m/sec. At an ISP of 380, that’s a mass ratio of 2.584:1. I fully fueled Starship in GTO with no cargo weighed 1,191,545 kg, so 461,198 kg of that makes it to the lunar surface

    To get back for aerobraking entry to Earth, Starship needs 1,829 m/sec for ascent and 853 m/sec for trans-Earth injection, or 2,682 m/sec total delta V, and thus a mass ratio of 2.054. It has to hit Earth’s atmosphere with a mass of 100,307 kg (so it has enough landing fuel), so it had to launch from the lunar surface with a mass of 205,629 kg. The mass difference is how much cargo was delivered to the lunar surface, and that comes to 255,569 kg, which is significantly more than the -7,000 or so kg in the original LEO direct to moon scenario.

    So the elliptical orbit paid huge dividends in delivered lunar payload capacity, which is this scenario is 2.55 times more than the Starship’s stated LEO payload capacity. That doesn’t worry me because lunar landing has really low G requirements, so it doesn’t cause structural issues.

    So Starship could definitely be the lynch pin of lunar development, if it can be refueled in a high elliptical orbit efficiently, and that takes those bigger booster configurations, which also get rid of the need for a special space tug. That means it can be down with only two required vehicles, Starship and paralleled Super Heavy or a much enlarged Super Heavy.

  24. gbaikie says:

    [[Initial BFR (Starship) is not much more powerful than Falcon Heavy

    But now they’re proposing something 4 times the thrust of Saturn V with 42 engines?? Outrageous! Surely that will put a bunch of buildings in danger, exceed the limits of the pad, and cause major problems on the Space Coast.
    And who even NEEDS a rocket that big?? ]]

    Well Falcon 9 increased it’s payload capability and the idea, likely, is BFR will likely improve in terms it’s size and capability. Anyways, who need a rocket that big?
    Anyone who want to have settlements on Mars.
    But I would say, we are not ready for settlements on Mars- Mars needs to be explored first. We are also not ready for commercial lunar water mining- need some lunar exploration, first.

    I would think that it would a good day, when a person can buy 1 lb of lunar dirt on Earth for about 1 or 2 dollars. And it seems that day, will be when payload is lifted of the lunar surface NOT using chemical rockets, so mag rails or something.
    On that day, we probably have very large chemical rocket leaving Earth, and probably not using space elevators or other means of getting off Earth without chemical rockets.
    So we will have very large chemical rocket and we also have chemical rockets of same size as Falcon-9.
    And it seems these big rockets will be launched from the ocean. Or the vision of Sea Dragons will more or less come true.
    Anyhow, the BFR would launched from the ocean. But near the present time, BFR are probably going start by launching them from land. Or launching them from the ocean would probably be a later modification.
    Not sure why Musk wants the BFR, soon. I suspect it has to do with how to land on Mars.
    One small part of getting lb of lunar dirt for $1, will be cheaply landing stuff on Earth.
    Or it seems currently it’s cheap to land stuff on Earth, but for $1 lunar dirt, even cheaper.
    Question: In distant future [50 year plus] will it always be cheaper to land stuff on Earth as compared to land stuff on Mars?
    And what about Moon, will be more expensive or cheaper landing stuff on the Moon as compared to landing stuff on Earth or Mars.
    For decades I have been interested in this space stuff, and general “rule” is it’s cheaper to land on Mars than it’s to land on the Moon.
    Or other than Mars being Earth like, this has been an apparent “selling point” of the Mars destination vs Lunar destination.
    But around the time of Curiosity rover, more doubt has occurred regarding this issue.
    Or with Moon, we have landed +5 ton on lunar surface. And with Mars, we have not.
    And Blue Moon suppose to land 4 or 5 tons of payload on the Moon.

    But with the Earth price of $1 for lunar dirt, it seems we need non chemical rocket means of doing this, and if you toss something off the Moon, you might be able to catch something on the Moon. Or mass driver of some type which can throw, could possibly, catch.

    Nature, drops rocks on earth which end up hitting surface at few hundred miles per hours. A rock the size of car, hits Earth’s atmosphere, explodes and smaller rocks hit ground at around terminal velocity. Whereas tens of thousands of tons of space rock can punch thru atmosphere and hit the surface like nuclear bomb.
    And roughly with Mars, with things like rocks, have a higher terminal velocity, as compared to Earth [because of thin atmosphere]. But human made things could have slower terminal velocity. Too be foolishly simple, a typical sailboat could land backwards. Crazy, but maybe something vaguely like it. But something big and balloon like is more normal {sane}, or something like the Starship which reduces the speed and uses a fair amount of rocket power to land.

    But as I ramble, it reminds me of how I think Mars should be settled- by having lakes in the tropic zone of Mars. People live under the lakes. Lakes could have frozen surface or liquid surface. And lakes could be flat landing zones.

    But if the moon has mineable water, then the Moon becomes a near term viable destination. And one thing I am not worried about, is that if NASA finds that there is lunar water which is mineable, is idea that the Chinese or Indians or Europeans are going to take advantage of this possible NASA discovery.
    Or idea of passing laws or having wars about who mines the water, is rather foolish.
    And on flip side, US public subsidizing such activity is almost as bad, maybe far worse.
    It’s either mineable or it’s not mineable. And mineable does not mean throwing vast amount tax dollar money at it.
    Mining the moon no matter the circumstances, is not going to be easy. Same goes for Mars settlements.
    And my simple rule is that socialism can not make humans a spacefaring civilization, it require “free markets”. Though we have politically decided that government should explore space to determine if Space could have markets.
    I think it’s good idea, that US govt should pay [US tax payers should pay] for the exploration of space- or beginning stages of exploration of space. BUT in theory space exploration could be done as private enterprise, but not keen on the laws needed and possible conflict it might cause.
    But it might follow the airplane- government fails to make an airplane, and the Wright bros end up doing it.

  25. whitelaughter says:

    gbaikie, the Wright brother’s airplane was simply a German engine in an Australian boxkite: the USA had the advantage of being in close communication with both nations so could combine the two achievements. (To be fair, an impressive thing to do).

    And that need to share information in this fashion is going to be critical for space exploration. And, thanks to the Chinese govt nicking everything they can, and sharing nothing, mindnumbingly difficult.

  26. Stan Witherspoon says:

    The Wright brothers were the only ones at the time who recognized that learning to control the aircraft was the issue, hence their invention of wing warping and coordinated bank turns and lots of practice in gliders at Kitty Hawk. Their history of being bicycle mechanics was not a just a coincidence.
    Sort of like learning how to put a TTW >1 stage back on the ground.

  27. George Turner says:

    The always accurate UK Daily Mail (that’s sarcasm), has a story up about the Raptor engines that cites Elon’s tweets.

    ‘Raptor liberated its oxygen turbine stator (appears to be mechanical, not metal combustion failure), so we need to update the design & replace some parts,’ wrote Musk in the thread.

    ‘Production is ramping exponentially, though. SN6 almost done. Aiming for an engine every 12 hours by end-of-year.’

    ‘Full year production is usually 70 [percent] of peak daily rate, so 500 [per] year. Still, non-trivial at 100,000 tons of thrust [per] year,’

    As production scales up, Musk said that the Raptor’s price will lower dramatically, from about $2 million per unit now to $200,000.

    ‘Since Raptor produces 200 tons of force, cost-per-ton would be $1000.

    ‘Since Raptor produces 200 tons of force, cost-per-ton would be $1000, wrote the CEO in another tweet.

    ‘However, Raptor is designed for 1,000 flights with negligible maintenance, so cost-per-ton over time would actually be $1.’

    That indicates that Superheavy’s 31 engines cost $62 million now, hopefully dropping to $6.2 million in full production. The importance of that shows in a comparison of common rocket engines (although some of the prices I’m basing my table on are quite sketchy).

    $1,000/lbf – RL-10
    $120/lbf – RS-25
    $28/lbf – RS-68
    $27/lbf – RD-180
    $14/lbf – BE-4
    $2/lbf – Merlin 1D
    $4.50/lbf – Raptor now
    $0.45/lbf – Raptor full production
    $0.00045/lbf – Raptor re-used 1000 times

    The final number for the massively re-used raptor is 22 million times cheaper than the RL-10, whereas an expendable, use-it-once and throw-it-away Raptor would only be 2,200 times cheaper than an RL-10.

    Now obviously that lowest number won’t happen because it would mean spending something like $2 per flight per engine, and it would cost more than that for a security guard to log it in as “returned”. But it would mean that engine costs would disappear into general staffing and operations costs, along with janitorial services.

    It also means that nobody who wants to stay in the spaceflight business very long would design anything that used the vastly more expensive expendable legacy engines.

  28. born01930 says:

    These numbers make P2P Starship economically feasible. I don’t think the 1000 people per flight he quoted would work though…seeA380. But it would make international travel so much more bearable.

  29. Zed_WEASEL says:


    The 1000 person P2P Starship is exactly what the USAF Air Mobility Command need for Intra-theatre transfers of entire Infantry battalion to their pre-positioned equipment anywhere within 10000 km within a half hour. Or shipped about 100 tonnes of cargo every couple of hours. Of course you would need a large supply of propellants at the originating and destination spaceports along with the cargo handling infrastructure.

Leave a Reply

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