Rockets, for November 5th

by Duncan Lunan

(First published in different form as ‘Launch Costs in the New Era’, Concatenation, 17th April, 2023.)

For many years, critics of space exploration have routinely exaggerated its cost by a factor of a thousand.  Every space mission, particularly if it failed, was said to have cost ‘billions’.  But while space exploration isn’t cheap, even the Apollo missions cost ‘only’ $150 million each in 1970 dollars.  We’re now seeing individual missions that cost billions, like the James Webb Space Telescope, but that represents many years of development and is expected to provide years of top-class science for it.

Forthcoming Moon missions

Fig. 1. Rollout of Artemis 1 spacecraft on SLS-1 booster, March 2023

The Space Launch System  (Fig. 1), to resume US Moon landings, is rather a different matter.  At $4.1 billion per launch, its commitments are being scaled back.  The Europa Clipper mission to Jupiter will now be launched in 2024 on Elon Musk’s Falcon Heavy, and the ‘Clipper’ name is now a misnomer – to get there will take two years longer than originally planned.  Even the return landing on the Moon has now been reallocated to Musk’s Starship/Superheavy vehicle  (Fig. 2), which has yet to make its first successful flight.  But although the Starships will be much larger than any previous rockets, capable of carrying 100 people into space, Musk has already demonstrated with his Falcon 9 and Falcon Heavy  (Fig. 3)  that huge savings can be made by centralising production, and reusability.  Falcon 9 development cost approximately $300 million, and it’s estimated that would have been $3.1 – 4.0 billion by NASA’s traditional procurement methods (‘Falcon 9’, Wikipedia, accessed 7th March 2023.).

Under the requirements of the US Constitution  (unique, in this sense, in the developed world), every government project’s finances must be voted on, every year, by both Congress and Senate.  That requires work on space projects to be farmed out to as many States as possible, and at the launch of Artemis 1, on SLS-1, the NASA Director stated with pride that components for the vehicle had come from every State in the Union.  When compared with Elon Musk’s methods, it makes the high cost of the Space Launch System easier to understand.

Boosters determine launch sites

For the last 60 years, western commercial satellites have had to fit as best they could on to boosters originally designed as military ICBMs.  With improved engines, longer tanks, upper stages and clip-on boosters, the lift capabilities of progressively more powerful versions of Thor  (Fig. 4), Atlas  (Fig. 5)  and Titan were far beyond the requirements for strategic weapons, and their heaviest lifts were for spy satellites, in polar or geosynchronous orbits, and the largest communications satellites in geosynchronous orbit.  But they all had to be launched from military facilities, adjacent to Kennedy Space Centre or from Vandenburg Air Force Base in California, so there were occasional issues of availability and access.  For that reason Europe’s series of Ariane boosters and Japan’s H-II  (Fig. 6)  had no military origins and were launched from civilian sites, with large comsats in geosynchronous orbit at the top of their payload range.

From the outset, however, it was obvious that not all satellite payloads would be as large as that, and the Ariane boosters came with the option of a dual carrier called SYLDA, which could launch two satellites on a single vehicle, and later a triple carrier, SPELTRA.  Still later an ASAP carrier was added for up to eight subsidiary payloads, aiming to reduce launch costs and provide a wider range of destination orbits, particularly for weather and Earth Resources satellites, in near-polar Sun-synchronous orbits;  but even that came in at around $50 million for a shared launch on Ariane V  (Fig. 7).  When the Soviet Union entered the commercial launch market with its Soyuz and Proton boosters  (Fig. 8), similar multiple launches were on offer, at similar prices.  

Small satellites

All the time, though, advances in technology were making the satellites smaller.  Because of their military uses the GPS navigational satellites were launched on Atlas and later Delta II boosters  (uprated Thors with upper stages and clip-on boosters), but for the 77-satellite Iridium series in the early 1990s, launches were made in sevens on Proton boosters, in fives on Delta II, and in threes on China’s Long March 2C  (Fig. 9;  John Bloom, Eccentric Orbits;  the Iridium Story, Grove Press, 2016.).   The Iridium-NEXT generation are now being launched by Space-X  (Fig. 10).  Elon Musk’s own Starlink series, in Medium Earth orbits inclined to the Equator like GPS and Iridium, have been launched on his Falcon 9 rockets in batches of up to 60  (Fig. 11), and the new Starlink 2 series are in batches of 51. 

Fig. 11. First 60 Starlink satellites in orbit, stacked pre-release, May 23, 2019.

Even so, satellites were getting smaller, sufficiently so to justify the development of dedicated small boosters.  The UK’s Ariel series of scientific satellites were among those launched in the 1960s by a small US booster called Scout, fired from Italy’s San Marco platform off Africa in order to reach a greater range of orbital inclinations.  Arianespace, ESA and the Italian Space Agency have jointly developed a booster called Vega which can launch up to 2 tons into Low Earth Orbit.  But ‘Cubesats’, shoebox-sized ‘bus’ vehicles for a wide variety of payloads, were pioneered in the UK by Surrey Space Technology Ltd and are now manufactured in quantity by Clydespace in Glasgow.  In 2003, when the first ones were launched, typically it cost $40,000 to launch one piggybacked on a larger vehicle.  A specialised launcher for them has been added to the Japanese module of the International Space Station  (Fig. 12).  But as the onboard technology lends itself increasingly to practical applications, it’s created a demand for smaller, cheaper and more versatile boosters.

Fig. 12. First Japanese cubesat launch from International Space Station, Oct 4th 2012

There are a number of ways to achieve that, and so many companies are rushing to provide it that it’s hard to cover them all in a single article.  Small launchers allow a variety of new approaches to be tried, among them new launch sites, new launch methods, new propellants and new construction techniques.

Land, sea and air launches

Traditionally, big boosters have sought to gain maximum advantage by launching in the direction of the Earth’s rotation, which provides a bonus of 1000 miles per hour at the Equator.  That puts the former Soviet sites at Baikonur and Plesetsk at a marked disadvantage compared to Kennedy Space Centre, and all of them at a really big disadvantage compared with ESA’s launch site at Kourou in Guiana, only 6 degrees from the Equator.  For launches to polar and highly inclined orbits this factor doesn’t apply, and hitherto they’ve mostly been conducted from Vandenburg AFB, as above.  Having clear launch over sea areas empty of shipping lanes counts for a lot:  Japan’s space activities have been restricted for decades by the fishermen’s lobby, and attempts to find a Pacific island launch site are frustrated by memories of occupation over World War 2.  But the UK’s relatively small size makes coastal sites here attractive, because we have sea all round us, and small rockets can be shipped to sites like the Hebrides, Cornwall, Sutherland and Shetland, without needing the specialised ships and aircraft which take Ariane stages to Kourou.

A promising attempt to get round the issues was made with the Sea Launch Project, whose two ships could sail to optimum launch positions at sea, sending up payloads with the Ukraine’s Zenit booster.  The launch control vessel, Sea Launch Commander, was built on the Clyde in 1997.  Zenit launches from the Equator, as envisaged by Sea Launch, would have allowed 17.25 to 25% higher payloads to be carried than in similar launches from the latitude of Kourou.  The floating launch pad was restored to operation after an explosion in 2007, but launches were halted after 2014 after the Russian annexation of the Crimea.  The company was transferred to Russian ownership, but the Zenit factories in the Ukraine have been turned over to weapon production, and a replacement for it seems unlikely any time soon.

Air launches have the same versatility as sea-going ones, and are also a well tried technique going back to the rocket aircraft of 1945 to 1969, not to mention the lifting bodies of the 1970s, one of which features at the beginning of The Six Million Dollar Man.  The US military began launching small satellites with the Pegasus booster from NASA’s same B-52 in 1990, transferring them to a converted Tristar named Stargazer in 1994  (both Pegasus and Stargazer were starships in Star Trek:  The Next Generation  (John Bloom, Eccentric Orbits, op cit).

As with sea launch, air launches can be made from any latitude, and straight into the required trajectory without needing the distinctive roll manoeuvre of the Space Shuttle and other launches from ground pads.  Releasing the booster at altitude confers significantly better performance, first demonstrated in the Rockoon and Farside launches from balloons in the 1950s  (see ‘Balloons in Space’, ON, 12th February 2023).  And the rocket can be jettisoned if it develops prelaunch problems, as had to be done with one of the X-1 series in the 1950s.

Fig. 13. Virgin Orbit 747 carrier Cosmic girl, dummy test

Richard Branson’s Launcher One was first intended to be launched by the White Knight carrier used for his space tourism flights, but was reallocated to a converted Boeing 747 called Cosmic Girl  (Fig. 13).  On January 17th, 2021, the second launch attempt successfully placed 10 cubesats in orbit from off the coast of California.  Virgin Orbit’s first UK launch attempt from Cornwall in January 2023 failed due to a dislodged filter in the rocket’s fuel supply, but the vehicle was already well tried and a second UK attempt was expected later in the year;  however, commercial confidence had been lost and the company went into receivership instead. 

Fig. 14. Astraius C-17 launcher, provisionally named Spirit of Prestwick

A rival company, Astraius, is expected to begin launching from Glasgow Prestwick Airport in 2024, using a Boeing C-17 Globemaster called Spirit of Prestwick,  paying tribute to the airport’s history (Anon, ‘Astraius Names Rocket Providers’, Spaceflight, October 2022) (Fig. 14).  Prestwick Airport is ideally suited to space operations, with excellent road and rail facilities, nearby seaports, main approaches over water in one direction and open country in the other;  a long main runway whose centre section was hardened in World War 2, in anticipation of attack by winged V2s;  advanced hazardous cargo facilities;  and an excellent weather record because it’s sheltered by Goat Fell on the island of Arran in the Firth of Clyde.  Its proximity to the satellite manufacturing facility in Glasgow is highly relevant, and the communications company Mangata Networks aims to open another on the airport itself.

Rocket fuels

Traditionally the standard propellant for the first stages of vertical launchers has been RP  (Rocket Propulsion)  kerosene, developed by Shell in the 1950s for the US Project Vanguard and used in conjunction with liquid oxygen.  The most energetic combination possible with current technology replaces the kerosene with liquid hydrogen, mostly used for upper stages until the advent of the Space Shuttle, and now the Artemis booster.  LH2 is notoriously hard to handle, mainly due to its very low temperature, and the US military’s Vulcan successor to the Atlas V is going to use liquid methane instead  (Fig. 15), as advocated by Arthur C. Clarke back in 1951 (Arthur C. Clarke, Interplanetary Flight, Temple Press, 1951.).  The United Launch Alliance Certification-1 mission is scheduled for launch on Christmas Eve 2023, to send the first of the Peregrine lunar landers to the Moon.  (Mike Wall, ‘ULA targets Christmas Eve for debut of new Vulcan Centaur rocket’,, 27th October 2023.)

Fig. 15. United Launch Alliance Vulcan booster, first launch scheduled for Christmas Eve 2023

At the end of World War 2, when the German team behind the V2 and its planned successors were annexed by the US military, Britain fell heir to the alternative German programme featuring hydrogen peroxide.  Personnel and equipment were moved bodily to the UK and by the late 1950s big strides had been made, with the all-rocket supersonic SR-177 fighter already in production before it was cancelled by the government, on the grounds that all UK defence would be handed over to unmanned missiles.  Hydrogen peroxide’s last bow was with the Black Arrow launcher, which put up the UK’s first and last wholly independent satellite just after the cancellation order had been issued  (Fig. 16).  The technology has now been resurrected by the Edinburgh-based Skyrora Ltd, established in 2017, who have already conducted successful test firings at the former RAF Macrihanish airfield on the Kintyre peninsula  (Anon, ‘Skyrora Fires Up Engine and Launcher License’, Spaceflight, October 2022), and set up production facilities in Cumbernauld.

Skyrora will have the option of launching either from the Saxavord launch complex on the Shetland island of Unst  (Anon, ‘Saxavord Readies for Launch’, Spaceflight, February 2023).(Fig. 17), or from Sutherland Spaceport on the A’Mhoine peninsula, or from Nova Scotia.  Sweden and Iceland are also possibilities.  Sutherland Spaceport is being run by another UK launch provider, Orbex, whose Prime booster’s test site is in Kinloss.  Both of the northerly launch sites are to be used by Lockheed Martin  (Anon, ‘Orbex Builds Its Spaceport’, Spaceflight, January 2023). and production facilities may well follow.  The company’s existing facilities in Scotland go back to the Second World War, and there’s plenty of scope for expansion.

Rocket construction

Construction techniques for rockets have followed much the same path for many years – so much so that the end caps for the Saturn V’s fuel tanks were the largest cold-hammered castings ever made, and Wernher von Braun had to scour the retirement homes of Europe to find technicians who knew how to do it.  Traditionally the rockets’ frameworks and skins have been made of aluminium, as thin as possible – the Convair Division of General Dynamics pioneered the use of pressure bracing to maintain the shape of the Atlas, and when the method was passed on to Britain for Blue Streak, De Havilland drew on its wartime stressed-skin experience with the Mosquito fighter-bomber to make the Blue Streak’s skin much thinner, yet stronger – techniques that found their way into the development of Ariane.

Making changes to the technology is not easy: one reason for the failure of the X-33 National Aerospace Plane as a successor to the Space Shuttle was the difficulty of making new fuel tanks of composite materials, which failed during testing in 1999.  The problems have since been solved and the hydrogen tanks of the SLS booster are indeed made of composites, formed in ‘the world’s largest welder’.  But at the other end of the scale, the entire body of the Relativity Space Terran-R booster is 3-D printed metal, and the Electron booster built by Rocket Lab is carbon composite  (Fig. 18).  That company was founded in New Zealand in 2006 and moved its registration to the USA in 2013.  Based at Long Beach, it launches from the Mid-Atlantic Test Facility at Wallops Beach on the east coast  (launch site of the Scout, above, back in the days when it was referred to as ‘the poor man’s missile’).

Fig. 18. Rocket Lab carbon composite Electron booster

Among the Electron’s many technical innovations are the use of battery-powered fuel pumps, rather than the turbines which have been standard since the V2, and the manufacture of the motors by electron beam 3-D printing.  In June 2022 an Electron launched NASA’s CAPSTONE mission to the Moon, arriving in November, and a private mission to Venus is planned for 2025.

In conclusion

Space has allowed me to cite just one example for each of the innovative methods to reduce launch costs, now coming up.  The various companies are unsurprisingly cagey about just how big the savings will be, but the bottom line is that each will have to compete not just with the established launch providers, but also with the other new starts.  It remains to be seen whether only the cheapest will win, or whether the various innovations will prove sufficient in practise to keep them all in business for specialist uses.  One thing that is for sure is that many of the rockets which have been familiar to us for decades will soon be gone.  Titan IV has gone, Atlas V is going because of its dependence on Russian engines  (though the last launch may not be till 2030), Delta IV Heavy will be replaced by Vulcan next year, Ariane V is about to be replaced by the simpler and cheaper Ariane VI  (but with the same payload capability), H-II is near retirement with trial launches of H-III underway, Soyuz launches from Kourou are over, Proton may attract few customers if any from now on.  It will be interesting to come back in a few years and see which of the old guard are left, and which of the new starts have made it.

Elon Musk’s Starship offers a massive drop in launch costs per kilogram.  Due to its size, it could in theory launch all the satellites wanted in any given year, even launching entire huge constellations like Starlink in a single flight, and some recent articles have predicted that Starship could put all the other launch providers out of business.  But Starship is not quite the game-changer it seems.  Orbital plane changes are very expensive to make, using up a great deal of fuel, and they will be very costly for Starship because it’s so big.  However many satellites it carries on each launch, they will all be released at the same inclination to the Equator, though onboard propulsion could move them higher or lower.  Even for the most popular destinations like geosynchronous or Sun-synchronous orbit, it may take time for a Starship launch manifest to fill up, and some customers will prefer to pay for an earlier launch on a smaller rocket. 

Many scientific satellites are in near-unique orbits.  IUE, the International Ultraviolet Explorer, was launched in 1978 into a ‘tundra’ orbit, with a 24-hour period but inclined to the equator, with a triangular ground track over the Atlantic which brought it over the participating nations in turn, eliminating the need for onboard tape recorders, often the first components to fail in satellites of that time.  As a result IUE remained operational till 1996, when it had to be turned off for lack of funds.  By contrast, GOCE, the gravity-mapping satellite, was in an orbit so low that it required continuous low-level thrust, and came down off the Falkland Islands as soon as its fuel was exhausted.  If anyone wished to put a future satellite into an exotic orbit like that, they might have to wait a lifetime for a Starship manifest to fill up for it.  So there will still be a market for dedicated satellite launches, and specialised boosters waiting to provide them, even if Starship dominates the mass market.

When the first version of this article was commissioned from me by Jonathan Cowie of Concatenation early this year, one of his wishes was that I should provide a table comparing the costs of launchers old and new.  For reasons hinted at above, it’s a much harder task than it seems, and the result will be for discussion next week.

(To be continued).

Duncan’s books are available through Amazon and in most cases through bookshops;  details are on his website,

Leave a Reply