Sunday 26 May 2019

Starship Lite: from rapid Interplanetary to Interstellar

Elon Musk stated that a stripped-down SpaceX Starship could become an interplanetary boost vehicle able to push probes towards the farthest objects in our Solar System. 

What other missions could the Starship ‘Lite’ do, and how quickly?
Rockets performance scales favourably with size. A larger rocket dedicates less mass to propellant tanks, engines and other equipment relative to the quantity of propellant it can hold. In technical terms, bigger rockets have better mass ratios.

The SpaceX Starship, planned to stand 55m tall, 9m wide and at 1350 tons on the launchpad, increasing to 118m and 4,400 tons once mounted on top of its giant booster stage, makes the most of its size.
Art by Charlie Burgess.
Despite being made of steel, the launcher manages a dry mass of 85 tons. The addition of landing legs, longer propellant tanks and large delta wings likely brings this closer to 90 tons. This gives it a mass ratio of (1350/90): 15. The current versions of the Raptor engines it uses have a sea-level Isp of 330s and a vacuum Isp of 360s. The average Isp over the course of a launch is about 350s.

Tsiolkovsky’ rocket equation gives us the deltaV we can expect from the Starship:
  • DeltaV = ln(Mass ratio) * Isp * 9.81 
The deltaV is in m/s.
The mass ratio is the dimensionless ratio of full to empty weight.
Isp is in seconds, and multiplying it by 9.81 gives the exhaust velocity in m/s.

We find that it can produce 9.3km/s of deltaV. This is enough to reach Low Earth Orbit, and validates claims that it can act as a single-stage-to-orbit vehicle.
Art by Charlie Burgess.
However, these figures are for a Starship with no payload onboard except the vehicle itself, and no reserve propellant to perform a powered landing. Placing 100+ tons in LEO requires the help of the ‘Superheavy’ booster.

Starship Lite

Elon Musk presented two versions of the Starship back in 2017: a crewed version and an uncrewed tanker or cargo-carrier version.

The 85-90 ton figures are for the crewed version. It has to have a large habitable volume, life support systems and other contributors to a larger dry mass.

The uncrewed version can dispense with all that. Its dry mass is reported to be 60-75 tons. The mass ratio increases to 18-22, as good as that of the Falcon 9 booster stage.

This tweet from Elon Musk introduces what we’ll be calling the Starship Lite – a stripped-down version with no features meant for re-entry, recovery or holding a payload. It would be a naked steel tank with an engine at the bottom and used solely in space. 
Starship Lite has a mass ratio of 30, from a wet mass of 1200 tons and a dry mass of 40 tons. It is unknown why the wet mass is lower than previously stated. The engines can be optimized for the vacuum environment – the addition of huge nozzles increases their Isp to 380s.

Going through the deltaV equation again, we find a value of 12.7km/s.
It will likely resemble the vehicle on the right. Art by 'teamonster'.
The vehicle could start out sitting in Low Earth Orbit, fuelled and ready to go. It could be a regular Starship that was converted in space instead of returned to Earth. Filling it up would take about 12 tanker launches.
Alternatively, it could be boosted into an extremely elliptical orbit, reaching out to beyond the Moon in apoapsis (400,000km) and just above the atmosphere in periapsis (200km). Tankers would struggle to match its orbit and deliver more fuel, increasing the number of launches required to fill it up to 70 (!).

For the following sections, we’ll attach various payloads to the Starship Lite and work out which missions can be carried out and how quickly they can get to their destination.

Ultima Thule and beyond

In that same tweet, Elon Musk talks about Starlink satellites converted into probes. They would have a solar-electric propulsion system with an Isp of 1600s, so with the mass ratio of 2, they’d have a deltaV of 10.9km/s. 

Between the elliptical orbit giving some starting velocity, a fully fuelled Starship Lite and the probes with their efficient engines, we can look forwards to some pretty extreme missions.

Adding up the deltaV amounts, we can already tell that the probes can be put into trajectories that escape the Solar System. This is what probes Voyager I and II accomplished.

Let’s look for the time required to reach the original goal: 2014 MU69 ‘Ultima Thule’.

The asteroid orbits at a distance of 44.5 AU from the Sun on average. Because we don’t have a launch date, and we can assume that the launch will be optimally timed and won’t need an inclination adjustment, we can do some simple calculations.

First of all, the Starship Lite is loaded with a couple of modified Starlink satellites. Let’s suppose 4 of them fit within a 1 ton payload. Mass ratio is reduced to 29.3

To escape Earth, the loaded vehicle burns all of its propellant at periapsis. It is already travelling at 10.9km/s, to which it adds 12.6km/s of deltaV. This gives it an initial velocity relative to Earth of 23.5km/s.

The Oberth effect is significant. Even after gravity slows down the Starship Lite, we expect it to shoot away into interplanetary space at a whopping 20.9km/s.

Earth orbits at 1 AU from the Sun at 29.7km/s. The escape velocity from the Sun at Earth’s orbit is 42km/s. Our Starship Lite leaves Earth and enters interplanetary space 50.7km/s. Another way of putting it is that the Starship is going faster than the Sun’s escape velocity… so it will continue travelling beyond the Solar System and go interstellar. After millions of years, it will meet another star system. A true star ship.

Kerbal Space Program, modified to represent the real Solar System, can give decent approximations of the trajectories possible. If the screenshots taken look too small to read on your screen, right click and open them to full size in a new tab. We position a target in Ultima Thule’s rough orbit and send off a model of the Starship Lite to meet it.
We find that Ultima Thule can be intercepted after about 6 years and 10 months. Our Starship Lite would pass the asteroid by at a blistering 28.6km/s!
Let’s add the deltaV from the probes’ electric engines on top. They can raise the velocity at which they escape Earth by another 10.9km/s, allowing for a total of 31.8km/s relative to the Earth, or an incredible 61.6km/s relative to the Sun.

The increased velocity shortens the travel time to 4 years and 7 months and the modified Starlinks cross the asteroid’s path going even faster. The biggest challenge would be resolving the asteroid in the probe’s cameras before it is out of sight again!

To the planets, quickly

There is plenty left to explore in the Solar System despite decades of probes and dozens of robotic missions. Scientists would love to be able to send a heavy probe loaded with instruments, RTGs, propellant and radiation shielding for long-duration missions to places such as Mercury or Uranus.
The Cassini-Hyugens mission put a lander on Titan and orbited Saturn for 13 years. It represented a 5.7 ton payload. Using the payload capacity of the Starship Lite, we can put together a bigger, heavier and more capable probe. Since we want the probe to spend a long time doing science instead of flying past like at Ultima Thule, we need to have a way to brake and insert the probe into an orbit around its destination. This means that the probe needs propulsion capability.

Now, working out the optimal probe mass ratios, power densities, ion engine endurance and all the other factors that go into proper mission design would take weeks of work and accurate simulation tools. ToughSF does not have access to those resources… so we will cut short the work by fixing the probe mass at 25 tons.

Depending on the mission parameters, those 25 tons could be nearly entirely dedicated to scientific equipment (24 tons dry mass, 1 ton propellant), entirely filled with propellant (1 ton dry mass, 24 tons propellant) and anything in between.
The exact propulsion type is left open. A hypergolic-fuel system with 320s Isp, where a lightweight 2 ton probe carries along 23 tons of propellants would have a deltaV capability of 7.9km/s would be ideal for a rapid gravity assist maneuver deep in Jupiter’s gravity well, where the radiation environment makes solar power tricky at the very least. A Starlink-like electric engine would work best when braking into orbit around Venus or Mercury, where abundant sunlight allows for decent acceleration. Going further, we could even expect a nuclear-electric power system and a HiPEP-derived 6000s Isp engine slowly accumulating velocity in the Outer Solar System; with a mass ratio of just 1.5, it would have a whopping 23.8km/s to perform a braking maneuver at Uranus or Neptune.

Furthermore, we won’t be using the complicated and expensive elliptical orbit as a starting point. While it might be worth it for a once-in-a-lifetime opportunity to visit an interstellar asteroid leaving the Solar System, it would be too expensive of an option for the exploration of our planets. Instead, we will assume a straightforward and cheaper 1,000km starting altitude.

We can go ahead and focus solely on the outbound trajectory from Low Earth Orbit. Where could Starship Lite position this 25 ton probe and how quickly could we get there?

A 25 ton payload increases the Starship Lite’s dry mass to 65 tons. This decreases the mass ratio to 18.46, and its deltaV capability to 10.9km/s.

From Hohmann trajectory data, we know that this is enough to reach every single body in the Solar System. 
However, relying on these trajectories means sometimes waiting for many decades for the probe to reach its destination. Instead, we will look at higher energy trajectories. We rely on Kerbal Space Program again to obtain approximations.

Let’s start with Mercury. It is a difficult planet to get to. The latest attempt, BepiColombo, has to perform multiple flybys of Earth, Venus and Mercury before it can enter Mercury’s orbit 7 years after lift-off. We won’t be so patient! We’ll make for a single transfer to Mercury.
We find that 10.9km/s is enough for a quick 55 day trip to Mercury.

Venus is closer. Earth’s sister planet was last visited by JAXA’s Akatsuki probe in 2010, where it failed to reach the desired orbit due to a malfunctioning engine. It had to wait 5 years before it could try again.
The ‘porkchop’ plots show us a way to get to Venus in a mere 30 days.

Mars has had a permanent robot population since 1971. Insight, the newest inhabitant, took about 7 months to get there.
According to our approximation, we could cut that down to 40 days with the Starship Lite.

Jupiter is far away. Juno had 4 years and 10 months of drifting through space to reach the gas giant.
A more energetic trajectory can carry a 25 ton probe to Jupiter in just under a year.

Saturn can be reached with a Hohmann trajectory lasting 73 months. Cassini-Hyugens took about this long to have a look at the Solar System’s most impressive rings.
If we had a Starship Lite sitting ready, we could have done it in just over 24 months.

Uranus and Neptune will always take a long time to get to, as they are 19.2 and 30.1 AU from the Sun respectively. Voyager 2, the only probe to visit both planets, took 8 years and 5 months to fly past Uranus and a full 12 years to pass Neptune.
Our faster trajectories mean a 4 year trip to Uranus and 7.5 years to Neptune. These long durations simply mean that chemical propulsion, even on the scale enabled by SpaceX vehicles, is not enough to cross over to the Outer Solar System in reasonable durations. It is much more likely that a good portion of the probe’s mass would be dedicated to electric rockets with high Isp that can shorten the trip and brake at the other end.   

Mars Express
Elon Musk’s dream is Mars. Just how quickly could we get to Mars using the Starship Lite?

It will depend of course on the payload we select for the mission. We also want to recover and reuse the Starship vehicle. Previous calculation assumed that once the payload separated from the vehicle, it would carry on into interplanetary or interstellar space, empty and discarded. Regular travel to Mars means that we would have to keep enough fuel in reserve to brake it into an orbit where it can be met by refuelling tankers.

One complication with using the Starship Lite instead of the regular Starship is that it does not have any features that allow it to aerobrake. No heatshield, no wings and no fairings means it must rely solely on its own propellants.

Let’s work out two scenarios: 10 ton fast, 10 ton staged and 10 ton ultrafast.

In the 10 ton fast scenario, we have as the name suggests a payload of 10 tons. The dry mass of the Starship Lite is therefore 40+10: 50 tons. We work out a mass ratio of 24.2 and a deltaV capacity of 11.9km/s.
When using 4.6km/s to leave a 1000km altitude Low Earth Orbit and 6.8km/s to brake into a ~31,000km altitude High Mars Orbit, we get a trip time of 120 days. The total deltaV is 11.5km/s.

In the 10 ton staged scenario, the payload is separated from the Starship Lite before it starts its braking burn. This allows the payload to perform an aerobraking or aerocapture maneuver while the Starship brakes while 10 tons lighter.
It is possible to accelerate by 5.36km/s leaving Earth to reach Mars in 95 days. After detaching from its payload, the Starship Lite can use its remaining 7.2km/s of deltaV to brake into a 170x800km Low Mars Orbit.

The 10 ton ultrafast trajectory consumes the Starship Lite, because we are entering the atmosphere. The payload stages and performs an aerobrake or aerocapture maneuver, just like in the staged scenario, but the Starship Lite burns up alongside it.
By not having to reserve any propellant for braking, we can take the most energetic trajectory possible. We find that using the full 11.9km/s deltaV capacity it has allows for a trip as short as 47 days.

The only caveat is that we must hope the payload’s heatshield can withstand an entry into the Martian atmosphere at over 19km/s!


The Starship Lite would be an amazing booster for sending off probes to all of the Solar System’s planets with much reduced travel times, or carrying significant payloads to destinations rapidly. In the most energetic trajectories, it proves itself to be a true Star ship.


  1. You know what would be a pretty cool probe mission in my opinion? One to Venus. Send a bunch of smaller probes for an aerobraking trajectory in the hope that at least a few survive, and have them inflate with a lifting gas (nitrogen, probably) to stay aloft in the Venusian atmosphere, cover their tops in solar PV and have the science equipment on the underside of the balloon. With inflatable probes you could probably do a lot of them fairly light and low volume before they inflate. It would give us so much interesting data and also function as a proof of concept for floating equipment in the Venusian atmosphere.

    1. With a decent solar-electric propulsion bus, we could be sending probes to Venus and Mars on every Starship launch as a secondary payload.

      Having many probes in the Venusian atmosphere, dotted around the planet, would tell us SO MUCH about its weather patterns, surface features and more... it could also prove or disprove the recent theory that Venus used to be an ocean planet.

  2. Technically, it is really a starship now.
    No more PR lying that some people don't like it while they are bending the definition of words to fit their needs. :V

    But it is just too slow to be practical, more like "Voyager 1 and 2 are also starships now, because they are outside our Solar System".
    Perhaps we should have a NSWR version of Starship?

    1. Of course, it is interstellar in the same way that Oumuamua is an 'interstellar visitor'. It'll take roughly 40 to 50 thousand years to get to the closest star at these speeds!

      The SpaceX Starship is not designed in any way that is compatible with a Nuclear Salt Water rocket, sorry.

  3. Nice sims here. Do you have a reference for the 60 to 70 ton estimate for the BFR tanker dry mass? Based on the fact it’s half sized to the earlier ITS launcher, I took the dry mass as ca. 45 to 50 tons. See here for the size of the earlier ITS tanker:

    This small dry mass estimate is in concert with that 40 ton dry mass size Elon gave for the stripped down BFR upper stage with only three engines when you add back on 4 engines and a fairing.

    Perhaps the larger dry mass of 60 to 70 tons for the BFR tanker is coming from addition of reusability systems, such as landing gear, etc. But you would think the ITS tanker’s cited dry mass values would include these systems for reusability. So taking half that dry mass should still be accurate for the reusable BFR tanker.

    In any case, a 45 to 50 ton dry mass should be accurate for an *expendable* BFR tanker, i.e., no reusability systems added. But this gives a 40 to 45 ton payload to LEO as an expendable SSTO. But this is in the 4% payload range common for current expendable launchers. In others words the expendable BFR tanker SSTO would be just as efficient as current expendable launchers.

    BTW, those sims for the Real Solar System mod for Kerbal for interplanetary flight look pretty good. I don’t know how to use it though. Perhaps we could collaborate on using it on my proposal for using the Falcon Heavy for the Europa Clipper mission.

    Bob Clark

    1. Thanks Bob,

      Sadly, I found no statement from SpaceX or Elon Musk on the dry mass of the tanker variant. The 65-75 ton figure is a conservative estimate (hence the +/-15 variation!).

      I do agree that an empty tanker variant would get to LEO, but it would not have enough reserve propellant to return.

      Kerbal Space Program does patched conics, which are decent enough for high thrust, high energy trajectories. It fails when you do planetary flybys or lower accelerations. For that, GMAT is the tool to use.

      Europa Clipper is 6 tons, so even if we slap on a 25% increase in mass for holding and deploying it, the Falcon Heavy in expendable mode would be able to give it 5.8km/s kick out of LEO. I'm taking this page's ( figure of 4.5 tons dry mass for the upper stage. With a payload capacity of 63.8 tons, it means 55.8 tons of fuel is left over for an interplanetary boost.

      If the payload mass is 7 tons instead, then we get an even more impressive 6.08km/s!

      However, these are not enough to get Clipper to Jupiter... I think the minimum is 6.6km/s.

    2. I intended the Falcon Heavy to use an additional stage for the injection into the flight towards Jupiter. It turns out the FH has far more payload capacity than needed. For instance, the FH has a payload capacity to GTO 0f 26.7 tons. This requires a delta-v of 3.5 km/s. A Centaur upper stage has a propellant of about 20 tons, dry mass of 2 tons and Isp of 465 s.

      But the 6 tons of the Europa Clipper added onto the 22 tons gross mass of the Centaur would be 28 tons, a little above the mass to GTO of the FH. So reduce the propellant load of the Centaur a little to 18 tons. Then the delta-v provided by the Centaur would be:

      465*9.81*Ln(1 + 18/(2 + 6)) = 5,380 m/s. Add this onto the 2,500 m/s provided by the FH in getting the payload to GTO to get a total of 7,880 m/s, well above the delta-v needed for the injection towards Jupiter of ca. 6.3+ km/s.

      In fact, the higher delta-v will allow a faster flight to Jupiter. You get even higher delta-v if you used more than one cryogenic stage to exit from LEO, rather than GTO. We have much more mass to work with then using the 63+ ton LEO payload of the FH. I estimate using a ca. 10 ton cryogenic stage plus a 20 ton one, such as the Centaur, then the delta-v would be above 9 km/s.

      By the way, similar calculations show you can get an actual *lander* mission to Europa using smaller spacecraft, say, in the few hundred kilo range:

      Bob Clark

    3. Yes, another upper stage is necessary here.
      Did you include the Oberth effect for a boost at periapsis? The initial velocity would be about 10.3km/s

      I would be hesitant to accelerate the trajectory to Jupiter. Saving time is great but it might not be possible for the Clipper probe to deliver enough insertion deltaV if it was designed for the slower Hohmann trajectory...

  4. That passage:

    “This requires a delta-v of 3.5 km/s. A Centaur upper stage has a propellant of about 20 tons, dry mass of 2 tons and Isp of 465 s.”

    should give the delta-v to GTO as 2.5 km/s, not 3.5 km/s.

    Bob Clark

  5. Where can I reach by using 1 ton water propellant with 600km/s ejecting velocity starts at earth orbit?

    1. You need to use the rocket equation:
      DeltaV = ln (Wet mass / Empty Mass) * Exhaust Velocity

      In your case, I don't have an empty mass. All I know is that your wet mass is 1 ton more than your empty mass, and that exhaust velocity is 600km/s.

      If your empty mass is 20 tons, then your wet mass is 21 tons.

      Putting these figures in the rocket equation gives a deltaV of 29.2 km/s. That's enough to escape the Solar System or make rather quick trips to the planets.

  6. If we redesign the ww1 zeppelin by using modern materials and making it more stronger and streamlined, so it can go supersonic in startosphere without problems, and may take 200 tons payload (including first stage) to low orbit at once, and once it in orbit and separating those stages, same mass as zeppelin hundreds tons in orbit.
    Size of this ”ship” doesn't matter in space, people's first thought maybe like ”it's too big”.

    1. A super-strong and ultra-lightweight zeppelin would need a volume of 5,000,000m^3 to reach an altitude of 25,000m if it has to carry 200 tons. That's a cylinder of 100m in diameter and 636m in length.

      The size of ship does not matter in space, but it does matter if it has to start on the ground.

  7. LZ129 Hindenburg Zeppelin:

    Length= 243m
    Diameter= 41.2m
    Volume= 2e+5m3
    Empty weight= 130tons
    Useful left= 232tons

    2e+5m^3/232tons= 8m*8m*13.469828m/ton

    8*8*13.5 cuboid can carry 1ton, 8*8 width is for zeppelin go supersonic.
    To building a basic space platform, needs 200tons, so I need to launching at least 200 cuboids to orbit.

    Airship, zeppelin, blimp and balloon, they don't need wings to stay in the air, they already can floating in enough high altitude, and speeding up to the limits speed if the aerodynamic allows them to reach high speed.
    Airbus A380 has 8.4m*7.15m can goes the maximum speed 0.9 mech, 8m*8m cuboid zeppelin can goes much faster than 1 mech,
    if we linked up all 200 cubion zeppelins together like ”airborne cargo train”, using enough power jet engine in head and speeds up to the 2 mechs in higher altitude until jet engines cannot burn, so changes to rocket engine carry them up to low orbit.

    I think my idea still has issues and problems, telling me.

  8. I’m starting a research project for calculating the actual averaged Isp for an altitude compensating engine on a SSTO:

    I see you are well expert at the kerbal sims for orbital mechanics. Perhaps we can do a Kerbal sim for flights to LEO using alt.comp.

    Bob Clark

  9. There hasn’t been much interest in a private mission to Europa like I was advocating for in my comment from May 29th. However, there is interest in a private mission to Enceladus:

    Perhaps we can do a Kerbal sim of such a private mission using the Falcon Heavy or the Starship.

    Bob Clark

  10. What a great article. Really enjoyed to read and get to know About propellant and These space vehicles in general

  11. If you can't find water at first time on Mars,or it's difficult to drill,may be refine carbon dioxide into carbon monoxide and oxygen,it's also a kind of rocket fuel,the isp is low,but launch from Mars surface dont need to much dv,so it's sufficent.