SpaceX’s ‘Starship’ As a Space Tug

This article examines the potential capability of “Starship” to be a space tug for the elliptic orbit departure and arrival processes.  It is merely a first ballpark look,  I would need more precise data for the orbital mechanics,  and for the vehicle weight statement and engine performances,  to obtain more reliable detail.  But these results are “good enough” to show considerable promise!

The notion of space tug assist to reduce the delta-vee (dV) requirements on interplanetary craft for hyperbolic departure from,  and arrival to,  Earth have been discussed in more detail in references 1 and 2.  Some educated guesses for the “Starship” weight statement and engine performances are given in references 3 and 4. 

Presumed Input Data Values

For purposes of this preliminary look,  it is the tug,  not the unspecified interplanetary craft,  that is the focus.  The following data were assumed about the trajectories:  a very extended elliptic departure and arrival orbit,  with perigee altitude same as low circular orbit at about 300 km,  a periapsis speed of about 10.9 km/s,  which is very near Earth escape at that altitude,  and roughly a 9-day period.  Circular orbit is presumed to be 7.8 km/s at 300 km,  with a 90 minute period.  That makes the dV requirement from one orbit to the other just about 3.1 km/s,  and I simply “budgeted” 0.2 km/s dV for any rendezvous and docking requirements.

This is what I assumed about “Starship” for the purposes of this study:  inert (dry) mass 120 metric tons,  maximum-capacity propellant load 1200 metric tons,  and the flaps and heat shield left in place,  pretty much as with the prototypes that are test-flying now.  It flies with 6 Raptor-2 engines,  3 the sea level form,  and 3 the vacuum form.  For the sea level engines in vacuum,  I presumed 230 metric tons-force thrust and 355 sec Isp.  For the vacuum engines,  I presumed 250 metric tons-force thrust and 379 sec Isp.  I presumed a max turndown ratio of 4 for both variants.

Departure and Arrival Missions

The departure mission differs from the arrival mission,  both in dV requirements and in the events sequence.  It is presumed that both vehicle assembly and refueling take place at an unspecified facility located in low Earth orbit (LEO),  to which the tug returns after assisting departures,  and with the interplanetary craft after assisting arrivals.  Reaching such a facility from the Earth’s surface is easiest,  if it is located in LEO at low inclination launching eastward. 

At departure,  the tug is already coupled to the interplanetary craft,  so there is only the dV onto the ellipse,  followed immediately by undocking and the interplanetary craft firing its engines to reach hyperbolic speed.  The unladen tug must then coast once about the extended ellipse,  and then finally supply the dV’s for return to circular at the proper time-of-perigee, and for rendezvous and docking with the unspecified facility there.

For arrivals,  the unladen tug must supply the dV’s for entry onto the ellipse at the proper ellipse perigee time,  plus a rendezvous and docking budget for coupling to the interplanetary craft,  which had already supplied the dV to enter that ellipse from its hyperbolic approach.  The docked pair coast about the ellipse.  Then the laden tug must supply the dV’s for getting from the ellipse back to circular,  plus a rendezvous and docking budget for returning to the unspecified facility.

Methods of Computation

I did this with a spreadsheet,  embodying both rocket equation calculations and a look at vehicle thrust/weight ratio for in-space accelerations.  I simply presumed vehicle accelerations outside the range of 0.2 to 4 gees as “unacceptable”.  These are nothing but by-hand pencil-and-paper-type calculations,  just automated for easy iteration with the spreadsheet. 

The “first cut” single rocket equation calculation provided performance data and a plot showing gross dV capability and accelerations for variable “payload masses” representing the unspecified interplanetary craft,  and various selections of which engines are operating.  Those results are given in Figure 1 below.   This is very unrealistic as a tug estimate,  because the tug’s weight statements are vastly different laden versus unladen,  but it does help to determine how many engines to use for acceptable accelerations,  for various “payload” masses,  including none.

When the weight statement does not change between burns,  one may sum dV’s to a single overall dV requirement (or not,  as desired).  When the weight statement does change,  there must be a separate rocket equation calculation for each weight statement,  and these calculations are linked by those weight statements,  with the linkages consistent with the order in which burns are made.  I chose to represent every tug burn,  3 for the departure mission,  and 4 for the arrival mission. 

The missions portion of the spreadsheet was set up for arbitrary payload mass inputs,  different for each mission,  with the option to reduce propellant mass if excess dV resulted.  I used this to find the max payload masses for each mission,  flown at full propellant load.  Reduce the “payload” mass,  and you may reduce the propellant load.  However,  I did not explore that issue,  having no information at all about the unspecified interplanetary craft that is the “payload” here.

Bear in mind that this is the first time I set up a tug mission spreadsheet.  It may get revised as it gets used.  But it is general enough to investigate other possible tug designs,  just using different inputs for vehicle characteristics.  An image of this initial spreadsheet is given as Figure 2 below.

Discussion of Differing Mission Results

First,  I used the vehicle characteristics more-or-less representing the “Starship” prototypes being test flown as of this writing.  Those masses include the heat shield and the aerodynamic control flaps.  This allows the vehicle to return for repair and refurbishment,  after “something” wears out from repeated use,  presumably engines. 

Second,  the mass of the unspecified interplanetary craft is much larger at departure than at arrival.  This is not as adverse a result as it might seem.  Any interplanetary craft departing Earth will have a full load of propellants and any other items for its mission,  and this will inherently be heavier.  On arrival back at Earth,  it will be essentially depleted of propellants and presumably unloaded of some mission-related things,  and so will inherently be much lighter.  That more-or-less matches tug capability.

Third,  for departure tug missions,  the docked pair departs at fully-loaded mass,  and the larger mass ratio associated with the dV = 3.1 km/s burn onto the ellipse then reduces that larger mass considerably.  There is no rendezvous and docking to worry about in this phase of the mission.  The tug undocks from the interplanetary craft,  and then coasts around the extended ellipse once,  firing unladen for a dV = 3.3 km/s to cover both getting back to circular and rendezvous and docking with the unspecified orbital facility.  That mass ratio applies to a much smaller unladen mass,  which is why the propellant quantities are so small.  And THAT “laden-then-unladen” burn sequence is also the “magic” of booster flyback recoveries for the Falcon and Starship systems! It does NOT work that advantageously,  for the “unladen-then-laden” burn sequence.

Fourth,  for arrival tug missions,  the tug departs unladen from the unspecified orbital facility,  and must burn for a dV = 3.3 km/s to cover getting onto the ellipse and rendezvous and docking with the unspecified interplanetary craft.  That mass ratio applies to a smaller but still-large mass at ignition,  using a lot of propellant.  Then it gains considerable “payload” mass after docking,  and must still supply dV = 3.3 km/s to cover getting back to circular,  plus rendezvous and docking with the unspecified orbital facility,  while laden.  That is another large mass ratio reducing the burnout mass,  which is now required to be much larger,  being laden.  And THAT unfavorable reverse of the booster flyback “magic” is what limits the size of the arrival payload mass!

Conclusions

First,  “Starship” as currently flown is a very large vehicle,  and so has a very large “payload” capacity if used as a departure scenario tug assisting any unspecified departing interplanetary craft.  These preliminary numbers as reported in the first figure say the size of that “payload” is almost but not quite 500 metric tons!  The arrival “payload” capability is much smaller at 175 metric tons,  but still quite considerable. 

Second,  crudely speaking,  this scales up or down with the tug stage-only mass at its ignition:  that being the inert mass plus the propellant mass.  We are presuming all the “payload” mass is the unspecified interplanetary craft,  and none is inside the “Starship” cargo bay itself.  Other stages or vehicles,  fitted with the right controls and some sort of on-orbit refueling plumbing,  could also serve this tug function,  and still be recovered for refueling and reuse.  Such might include the second stages of the Falcons,  the SLS upper stage,  the Vulcan second stage which is an upgraded Centaur,  and so on.  There many possibilities.

Third,  that being the case,  it should not be very difficult to develop multiple space tug designs out of these various upper stages.  The hardest part will be setting up a space station facility in LEO that functions as both an assembly facility,  and a propellant refill depot facility!   And THAT is the most important result here!  We will need that LEO-based facility to support any space tug-assist for any interplanetary craft.  That is the real gateway to less-expensive interplanetary travel,  not some hard-to-reach space station around the moon!  References 1 and 2 prove this conclusively.

Fourth,  such would also support lunar missions,  as well,  since the extended ellipse could apogee right near the moon’s orbit about the Earth.  It is the 3-body effects when the moon is there at craft apogee,  that then turn this into a figure-8 lunar transfer trajectory.

References

#1. G. W. Johnson,  “Tug-Assisted Arrivals and Departures”,  posted to “exrocketman” 12-1-2024.

#2. G. W. Johnson,  “Elliptic Capture”,  posted to “exrocketman” 10-1-2024.

#3. G. W. Johnson,  “Rocket Engine Calculations”,  posted to “exrocketman” 10-1-2022.

#4. G. W. Johnson,  “Reverse-Engineering Starship/Superheavy 2021”,  posted to “exrocketman” 3-9-2021.

Use the archive tool on the left side of this “exrocketman” page to quickly reach any of these references (or any other article posted here).  All you need is the date of posting and the title.  Click on the year,  then the month,  and then the title if more than one article was posted that month.  This is far easier than scrolling down.

Figures

Update 1-4-2025: added plot axis titles to the Figure 3 data plot made for Centaur-V as a tug.  You may click on any figure to see them all enlarged.  There is an X-out tab top right of that view,  which takes you right back to this article.  

Figure 1 – Spreadsheet Results For Both Gross-Overall And Specific Mission Capabilities

Figure 2 – Image of the Initial Form of the Tug Spreadsheet Set Up for “Starship”

Update (also) 1-2-2025:

I have started looking at other stages for possible tug use,  notably the Centaur,  which uses higher-energy LOX-LH2 propellants.  The Common Centaur stage design uses a common bulkhead between the LOX and LH2 tanks,  and single-wall thin stainless steel tank walls that gain their strength from internal pressure by the “balloon effect”.  The inter-tank bulkhead is two stainless-steel membranes separated by some fiberglass hex as insulation,  to keep the warmer LOX from heating the colder LH2 too rapidly.  Even so,  the useful stage lifetime is only “hours long”,  limited by heating of the hydrogen by the warmer oxygen,  leading to tank overpressure failure. 

Centaur-III is the 3.05 m diameter form used on Atlas-V,  which only uses two RL-10 engines when lofting the Boeing CST-100 “Starliner”.  The other applications are all one-engine,  presumably underlying the data I found.  The tank wall is 300-series stainless steel 0.020 inches (0.51 mm) thick. 

Centaur-V has a larger hydrogen tank diameter,  and comes in only the two-engine form,  for use on the new Vulcan launcher.  Centaur-V initially has only about 40% longer stage lifetime,  but it is said that it will eventually have a few hundred percent longer lifetime,  for much longer missions.

Centaur-III/Common Centaur simply does not have the stage lifetime needed to serve as a tug on ellipses with multi-day periods!  Eventually,  if the expected longer stage lifetimes turn out to be true,  Centaur-V could serve as a tug!  Tug duty could involve stage lifetimes in the range of 10-days to 2 weeks,  before intervention is required to address hydrogen boiloff and overpressure risks. 

I can only estimate,  based on the “balloon pressure” approximation,  that total applied thrust equals the “PA-kick load” associated with the tank pressure.  Thus,  the minimum operating tank pressure for the 1-engine Common Centaur stage would be P = Ftot/Atank = 22,300 lb/11,310 in² = 2 psia min.  For the 2-engine form,  that would be 4 psia min.  The ideal tank failure hoop stress would the ultimate tensile strength of 300-series stainless steel per Mil Handbook 5,  which is about 95 ksi.  The corresponding failure tank pressure is P = 2 s t/D = 31-32 psi.  However,  it is rendered unusable by yielding,  before reaching that value.  This would actually be a bit lower for the larger-diameter Centaur-V,  nearer 20 psia.

Without an accurate empty mass for Centaur-V,  I cannot run reliable tug performance estimates for it!  A “wild educated guess” for the Centaur-V empty mass might be near 3200-3300 kg,  based on a guess for the change in tank mass,  based on propellant masses,  using “97% propellant”.

I ran these numbers as a rough cut for what a “long stage life” Centaur-V might be able to do as a space tug for these elliptic arrival and departure missions,  using an educated guess of 3.25 metric tons for the empty stage mass.  The results are depicted in Figure 3,  using the same spreadsheet as was used to investigate “Starship” as a tug.  I just copied the “Starship” worksheet to another worksheet,  and then changed all the input numbers,  and finally re-iterated for max payload. 

Unladen burns used only one of the two RL-10 engines on the Centaur-V stage to limit vehicle gees. The results look good,  except that the unladen burns for the departure mission were around 6 gees.  It would probably be required to reduce from full thrust for that mission.  That may reduce specific impulse a little,  which would in turn reduce max payload a little. 

Errors in the assumed stage empty mass subtract directly from max payload capability on the tug missions.

Figure 3 – Results for Centaur-V As a Possible Tug (updated with axis titles 1-4-25)

Conclusion to this Update:

It would be fairly easy to install those changes needed to use Centaur-V as a tug to assist interplanetary departures and arrivals,  based from the unspecified facility in LEO.  The hardest part would be achieving the necessary 2-week “stage life” to support extended ellipses to a 10 day period.  The biggest uncertainty is the longer service life necessary for doing these missions routinely.  The RL-10 engines were not intended to be “permanently” re-startable.