Friday, February 9, 2018

Launch Costs Comparison 2018

Update 6-1-18:  see red text edits embedded below
Update 6-16-18:  see blue text at end of article,  after the figures
This article compares and correlates unit costs for launchers,  mostly those used commercially.  These data are based upon reported payload capacities and launch costs found in the literature.  The format is cost per unit delivered payload mass,  on the very important assumption that the launcher flies fully loaded.  All figures are at the end of this article.  Click on any figure to see any or all of them enlarged.  You can close that view box and be right back to viewing this article.  

Results are reported in millions of dollars per delivered metric ton,  and in dollars per delivered pound.  To estimate the unit cost when flying at less than full load,  simply divide these unit costs by the fraction of fully-loaded that you intend to fly. 

Scope includes the launchers used in the competitive satellite launch business,  plus a few launchers that were used,  but not competitively,  and the US Space Shuttle as representative of a large spaceplane.  Some of these launchers are no longer in service.  However,  the correlation results are used to predict unit costs for the NASA SLS block 1,  just for comparison.

Data,  Sources,  and Results for “Standard Low Earth Orbit”

“Standard Low Earth Orbit” is 23 degree inclination out of Cape Canaveral,  Florida,  to a 200 km orbit altitude.  This is what the reported payload delivery capabilities in the literature refer to.  These data are for one-way delivery of payload using a simple payload shroud,  not a recoverable capsule,  except for the Space Shuttle.  As researched,  those data are:

Of these,  the US Space Shuttle,  the Titan IVB,  and Falcon-1 are no longer in service.  There is a demonstrated history of reliability problems with Proton-M.  The Titan-IVB was retired in 2005.  The Falcon-1 retired no later than 2012.  The Space Shuttle retired in 2011.  The prices shown are for fully-expendable flights in the case of Falcon-9 and Falcon-Heavy.  There should be some small price break for flying reusably at reduced payload with these two launchers,  although how much is but speculation. 

The trends of unit cost per delivered metric ton (flying fully-loaded) are given in Figure 1.  Figure 2 shows unit cost per delivered pound.  Data are grouped and correlated as “competitive”,  “non-competitive”,  and “spaceplane.  The “competitive” launchers correlate flying fully loaded as:

                Unit cost $M/metric ton = 10.557 e^(-0.033 Wp)  where Wp = delivered payload, metric tons

                Unit cost $/delivered lb  =   4787 e^(-0.033 Wp)   where Wp = delivered payload, metric tons

The launchers marked “competitive” all compete in the commercial (and military) satellite launch businesses,  with market share in part depending upon price.  The launchers marked “non-competitive” never competed commercially,  and thus were not subjected to severe pressure on price.  The model for these assumes the same -0.033 Wp factor,  and uses a coefficient that forces the curve through the average of the Titan IVB and Delta IV Heavy data points: 

                Unit cost $M/metric ton  =   39.1 e^(-0.033 Wp)  where Wp = delivered payload,  metric tons

                Unit cost $/delivered lb = 17,720 e^(-0.033 Wp)  where Wp = delivered payload,  metric tons

This model was extended to 70 metric tons payload to estimate what should be expected for NASA’s SLS block 1 as shown in the figures ($3.878M/m.ton and $1759/lb).  That calculation corresponds to an expected launch cost of $271M,  when NASA’s actual launch cost estimate is $500M,  and its critics estimate twice that.  So instead of only 3 times more expensive than Falcon-Heavy (otherwise comparable in payload) as predicted by the correlation,  SLS block 1 is likely to be at least 6 times more expensive,  and it might even be 12 times more expensive.

Update 6-1-18:  the latest NASA estimates for launching SLS block 1 have grown to about ~$1-2B per launch.  Its critics put that closer to $4B.  For a 70 metric ton payload,  that's $14-56M/m.ton,  or $6500-$26,000/lb.  The range depends upon whose estimates you believe more.  That falls somewhere halfway between all the other launchers and the space shuttle.  SLS is well on its way to being the most costly launch vehicle in human history.  

The Space Shuttle (marked “spaceplane”) is quite different,  in that the delivered payload is but a small fraction of the mass of the recovered vehicle.  All the others are one-way trips to space,  with the delivered payload encased in a shroud.  There are no recovered capsules delivered by these launchers. 

The spaceplane model for the Space Shuttle assumes the same -0.033 Wp factor as the “competitive” launchers,  with a coefficient that puts the curve through the data point for the Shuttle:

                Unit cost $M/metric ton  =   131 e^(-0.033 Wp)  where Wp = delivered payload,  metric tons

                Unit cost $/delivered lb = 62,580 e^(-0.033 Wp)  where Wp = delivered payload,  metric tons

Re-Scaling Results for Delivery at the International Space Station (ISS)

I used the payload reduction fraction seen with the Space Shuttle as a constant applied to all the launchers still in service,  for estimating unit cost performance delivering to the ISS.  The ISS is located at a higher inclination and a higher orbit altitude.  For the same launcher technical performance,  a launcher’s max payload capability must be reduced when reaching for the more demanding destination. 

Flying with a 7 person crew,  the Space Shuttle is listed as 24 metric tons to standard low Earth orbit.  It could deliver as much as 27.5 tons,  but only with a smaller crew and less supplies.  Flying with a 7 man crew,  its capability to ISS is reduced to 16 tons.  That is 2/3 of the standard low Earth orbit capability with the same crew and supplies.  Applying this 2/3 factor “across the board” with the same launch prices produces Figures 3 (per ton) and 4 (per pound) below. 

I correlated unit cost estimates to ISS only for the “competitive” launchers that are still in service.  These are (of course) somewhat higher than for “standard low Earth orbit”,  because payload capability is lower,  while launch price is not.  This for one-way payload delivery using a simple payload shroud,  not a recoverable capsule.  That model is:

                Unit cost $M/metric ton = 15.428 e^(-0.048 Wp)  where Wp = delivered payload,  metric tons

                Unit cost $/delivered lb   =    6996 e^(-0.048 Wp)  where Wp = delivered payload,  metric tons

Estimating the Effects of Reusability

I based this estimate on what Falcon-9/Cargo Dragon has demonstrated to ISS with re-use of first stages,  when loaded to max cargo for ISS,  versus what I estimate the fully-expendable deliverable payload is to ISS.  The fully expendable estimate is 15.2 metric tons to ISS.  A fully-loaded (for ISS) Cargo Dragon is 8.8 metric tons.  That ratio is 0.5789,  and I assume it applies to Falcon-Heavy for its payload delivery to ISS with re-use of first stage cores.  The results are given in Figure 5,  for both full price and for an arbitrary modest price break:  80%-of-full-price,  representing savings from re-use. 

Estimating What SLS Block 1 Might Really Do (Standard Low Earth Orbit)

SLS Block 1 is said to deliver 70 metric tons to standard low Earth orbit.  NASA says it expects each launch to cost roughly $500M.  NASA’s critics say each launch might cost nearer $1000M = $1B.  Those data correspond to $7.14-to-14.28M/delivered metric ton or $3239-6478/delivered pound (flying fully loaded). 

The “non-competitive” launcher correlation predicts for SLS Block 1 a unit cost of $3.878/delivered metric ton or $1759/delivered pound (flying fully loaded).  Falcon-Heavy has an almost comparable payload (63.8 vs 70 metric tons),  with unit costs of $1.411M/delivered metric ton or $640/delivered pound (flying fully loaded and fully expendably).  SLS will never be reusable,  as that was never considered as a design requirement. 

SLS is expected to fly only once a year,  and not until 2019 or 2020.  Falcon-Heavy flew its maiden test flight in February 2018.  It is scheduled to fly at least two more times in 2018. 

Update 6-1-18:  as indicated above,  even NASA's own cost estimates for an SLS launch have grown dramatically,  as have the estimates of its critics.  It will come nowhere close to what the non-competitive launcher trend predicts,  being higher by a factor closer to 10 than 3.  This is only opinion,  but one would be tempted to say that this travesty is exactly what you should expect from corporate welfare mandated upon NASA by a Congress interested only in "pork".  

Other Launchers to Watch For (That Are Not Yet Flying)

There will be an Ariane 6.  Long March 5 may or may not be flying yet.  United Launch Alliance is designing a new heavy lifter to be called Vulcan.  The Jeff Bezos organization Blue Origin is designing a heavy lifter to be called New Glenn.  Spacex is working on a design called BFR which will be a super-heavy-lifter with a fly-back first stage combined with a second stage that is also a reusable spacecraft. 

Final Note:  Falcon-9 Cargo Dragon to ISS

Full price for a Falcon-9 launch is $62M.  This can send to ISS a Cargo Dragon totaling 8.8 metric tons.  Of that,  only 3.310 metric tons is actual deliverable cargo.  Using that 3.31 tons,  the effective unit costs for delivery to the ISS are: 

                $18.73M/delivered metric ton = $8495/delivered pound

Given the same 80% of full price with reusability,  as was used above,  these data reduce to:

                $14.98M/delivered metric ton = $6796/delivered pound

Compare those with what the Space Shuttle costs were,  delivering 16 metric tons to the ISS at $1.5B per launch:

                $93.75M/delivered metric ton = $42,517/delivered pound

These are the best guesses I have for Enhanced Cygnus on Atlas V 551,  and they are not accurate.  The max deliverable mass to ISS is 12.34 metric tons,  which has to be larger than the loaded Cygnus.  Data gleaned from multiple sites on the internet says the max payload to ISS inside the Cygnus is 3.5 metric tons max.  Cygnus cannot return to Earth.  Each launch is $153M.  Those unit costs are thus crudely:

                $46.M/delivered metric ton = $21,000/delivered pound

I have no reliable data on the cargo version of Soyuz,  riding the R-7 rocket.  Best guesses are max 2.4 metric tons of payload in the capsule,  and a launch cost on the order of $65M.  Those unit costs are:

                $27.1M/delivered metric ton = $12,300/delivered pound

Thus,  cargo Dragon on a Falcon-9 appears to be the most cost-effective means to deliver self-maneuvering and self-rendezvousing cargo to the ISS,  of all the vehicles that have done this task.  

Prior Similar Articles

This article replaces earlier postings on this site.  The best of the older postings is “Access to Space:  Commercial vs Government Rockets”,  dated August 7,  2015.  That one compares multiple rockets with the best inflation-corrected costs I could find or devise.  The one prior to that was “Revised Launch Cost Update” dated September 13,  2012.  It refers in turn to “Revised,  Expanded Launch Cost Data” dated May 26,  2012.  That one in turn was a revision to the original “Launch Cost Data” article dated January 9,  2012.  But this current posting is the best,  with the latest versions of the rockets,  and the most current costs I could find.  I did not inflation-correct costs from 2016 to 2018 values.

Figures Follow:

 Figure 1 – Unit Cost Comparison (per ton) to Standard Low Earth Orbit

 Figure 2 – Unit Cost Comparison (per pound) to Standard Low Earth Orbit

Figure 3 – Unit Cost Comparison (per ton) to ISS

 Figure 4 – Unit Cost Comparison (per pound) to ISS

Figure 5 – Unit Costs for Falcon Vehicles as Payload-in-Shroud to ISS with Re-Use

Update 6-16-18:

The competitive launcher curve-fit to LEO is $M/m.ton = 10.557 e^(-.033*Wp) where Wp is the metric tons of payload actually delivered from within a payload shroud.  It was derived for a variety of commercial launch vehicles,  including Spacex Falcon-9’s whose boosters were never re-flown more than twice,  if at all. Most were not reused.

From Spacex’s website,  posted prices and payloads-to-LEO for vehicles not flown reusably are $62M for a Falcon-9 carrying 22.8 metric tons,  and $90M for a Falcon-Heavy carrying 63.8 metric tons.  These correspond to curve-fit unit price predictions of 4.9748M/m.ton and 1.2858M/m.ton,  respectively. 

Using this and the payload,  back-predicted launch prices are $113M and $82M,  respectively,  compared to $62M and $90M as posted.  The discrepancy can be laid to the scatter in the data generating the curve fit.  It isn’t all that large at +26% for Falcon-9 and -9% for Falcon-Heavy.  But it is important to realize that predictions made this way could easily be wrong by a factor of 2. 

It is far too soon to be talking about what the launch price of a BFR/BFS ought to be.  However,  ignoring that,  and using this same correlation at the claimed 150 metric ton payload deliverable to LEO for the 2018 9-meter diameter version,  the predicted unit price is $0.07478M/m.ton,  and at that payload,  the back-estimated launch price should be near $11.2M.  That supposedly is the price Spacex would charge a customer to deliver his 150 metric tons of cargo to LEO.

I have estimated elsewhere on this site that this 9 meter diameter version of the BFS second stage requires 6 tankers to fully refuel its tanks with 1100 metric tons of propellant (80% liquid oxygen,  20% liquid methane),  enough for a one-way trip to a direct aerobraked Mars retropropulsive landing,  with 150 metric tons of payload.  That’s seven launches for each BFS that departs for Mars,  totaling $78.4M in direct launch costs for each BFS embarking on that trip.

Assume for the sake of argument the wild guess that ~$10B are left to spend making this design flight-ready for the Mars trip.  Assume also as another wild guess that another ~10B are needed to fully develop and field test-verify all the equipment needed to sustain a small crew on Mars for 2+ years,  and to emplace there the means to manufacture propellant locally for the return home (1100 metric tons per returning BFS).

Assume also that one ship gets lost and destroyed trying to make its landing.  Could easily be 2 or 3,  but assume it is only one.

Spacex’s posted plans call for 2 uncrewed BFS vehicles each carrying 150 tons of cargo to land during the first Mars opposition,  and 4 more the second opposition,  with two of those carrying small crews and lots of cargo totaling 150 metric tons.   That’s 6 successful BFS flights to Mars,  plus the one we assumed lost,  for a total of 7 attempts. 

At the estimated aggregated launch cost of $78.4M per BFS sent to Mars above,  that’s $549M in launch costs for the 7 attempts;  just under $0.6B.   Add in the $20B to prove out the vehicle and to prove out the hardware,  and you have a temporary base on Mars for under $21B!  Remarkable!  And note how launch costs are miniscule compared to the hardware development and verification costs that we have assumed.  We could easily double or triple the launch prices,  and not really affect this outcome.

Now,  the predicted $11.2M per BFR/BFS launch is for a matured,  well-proven design that flies routinely and reliably.  That’s what the curve-fit really represents.  Right now this thing exists only as a concept in presentations.  A whole lot of resources must be spent developing this vehicle and turning it into that near-perfectly-reliable transport.  That what the wild guess of ~$10B is all about.

The same applies to both crew life support technologies and most especially the on-Mars propellant production hardware.  If that doesn’t work,  the crew dies on Mars.  There is no rescue,  because without refilling on Mars,  the BFS is a one-way trip.  So the propellant production plant has to work!  And it has to produce some 1100 tons of propellant to refill that BFS which returns the crew,  before their supplies run out.  This is not a small piece of equipment!  That’s what the other wild guess of ~$10B is all about.

So,  no matter what,  you have ~$20B in vehicle and hardware development and prove-out costs to amortize across the flights this vehicle will make.   The more it gets used (for any missions,  not just Mars),  the lower that amortized impact on any one vehicle. 

If the only thing ever done with this vehicle is those 7 attempts at Mars,  then there must be a total of 49 BFR/BFS vehicles launched.  $20B spread over 49 vehicles is $408M per vehicle.  That needs to be added to the otherwise-mature launch price of $11.2M,  for about $419M per launch. 

If you do missions to the moon in a similar way to Mars,  that’s another 49 flights under the same assumptions otherwise,  for a total of 98 flights to amortize the development costs.  That would be $204M per vehicle,  for a total price of $215M per flight.

Now,  if the same vehicle gets routinely used to LEO,  that’s more vehicles over which to amortize the development costs.  Say,  100 orbital deliveries,  for a total of 198 flights.  That’s $101M amortized per vehicle,  for a suggested launch price of $112M per vehicle.  

This is the “mass production effect” in action.  Unless somebody else foots some or all of the development bills,  I fear the BFR/BFS vehicle launch price can never be as low as it could be.  But then again,  my curve fit may have been extrapolated too far.   We’ll just have to wait and see what really happens.  One thing is sure:  this thing will be way cheaper to use than NASA’s SLS ever could be!


  1. What's your opinion on laser thermal rockets as a way to reduce launch costs (especially taking the rapid improvements in laser technology into account)?

    1. Nice idea in theory. But we are lightyears away from lasers big enough to do the job, and lightyears away from the pointing/aiming controls precise enough, to make such a thing truly feasible. GW

    2. I get the aiming part (though combat lasers able to intercept mortars have been a thing for several years), but I don't think the size part is valid. People figured out how to do coherent beam combining, meaning you can slap together as many fiber lasers as you need. Fiber lasers max out at 1 kW/fiber AFAIK so you need quite a lot of fibers even for small payloads. But defense contractors are already selling 40 kW combat lasers.

  2. This comment has been removed by a blog administrator.