I’m taking values from Wertz’s Space Mission Engineering: The New SMAD, Table 11-23 “Historical Launch Vehicle Costs for Predicting SME-SMAD WBS 2.0 Cost”. This was published in 2011 so it will be using values from the 2010 timeframe. This may not be the most accurate values but it’s the easiest thing for me to grab that has ostensible direct comparisons and at 1032 pages I’m not going to have pulled it off the shelf not to use it. All values are launch cost per kg placed in low earth orbit (LEO) It should be noted that different vehicles launch from different azimuths, and so it isn’t a fully normalized comparison (for instance, if you want a low inclination retrograde orbit, forget using any of the Russian launch vehicles) but it’s about the best to do without running hypothetical flight sims of each vehicle from a common launch point.
For medium/intermediate class vehicles it has European (Ariane 44L at $15.0k/kg), Russian (Dnepr at $4.6k/kg, Soyuz at $7.4k/kg), Chinese (Long March 2C at $9.5k/kg, Long March 2E at $7.4k/kg), and US-ULA (Delta 2 at $14.6k/kg, Atlas 2As at $15.4k/kg). Falcon 9 compares at $5.4k/kg; however, this would be the CRS value based the original F9 vehicle, not the current F9v1.1, and newly manifested launch cost has gone up from from $56M for a 10.5mt payload to LEO to around $63M for at 13,150 kg which would make the cost for that vehicle ~ $4.8k/kg.
On the heavy side, the Ariane 4G costs $12.5k/kg, Atlas 5 and Delta 4 Heavy cost $8.6k/kg and $9.5k/kg respectively, Long March 3B is $6.0k/kg, the Proton just undercuts it at $5.9k/kg, and the Zenit 2 skunks it at $4.2k/kg, while the Sea Launch version of that boaters rises to $7.2k/kg. The recently retired STS caps them all at a whopping $14.2k/kg but should really be considered in a different category since 2/3s of the ostensible payloadable mass is given over to its reentry and thermal protection systems, which means on a mass basis the STS should actually be ~$4.7k/kg with a ~72mt load. (This should beg the question of why the mechanically simpler Space Launch System, which more capacity and utilizing much of the heritage STS systems in its design, is projected to exceed the cost of a Shuttle launch and put its payload to orbit costs in the same range at the Delta 4H. The answer is that it has such a low launch rate and requires a large army of skilled labor to test, integrate, and launch the thing, which underscores my earlier point that the cost reduction should be addressed in minimizing the processing flow cost rather than shooting for reuse of existing technology or any kind of reusability for its own sake.) Falcon Heavy has an advertised cost of $85M for a 53mt payload, giving a $1.6k/kg, seemingly a revolutionary bargain.
Before getting out the checkbook, however, the costs bear some deeper examination. For one, the advertised launch costs were estimates developed prior to launching either SpaceX vehicle. Launch costs on the ULA vehicles (Delta IV, Atlas V) were dramatically undervalued prior to operation and the costs above reflect adjustments after the real operating costs were known. SpaceX has flown the F9v1.1 six times now but they’ve had numerous and likely very costly delays which they’ve eaten, and are still trying to refine their processes to get a streamlined flow that will allow for the kind of launch rates they desire. The Falcon Heavy has yet to fly (the first one is next year) and the both the cost and payload capability are still somewhat in flux.
Second, the costs reflect expected reusability with minimal refurbishment (which Musk has repeatedly stated as a goal). It remains to be seen how practicable that goal will be with the current design, and SpaceX is charging the Air Force a premium to use virgin vehicles with the recovery features (i.e. landing legs) removed.
Third, the “launch costs” advertised by SpaceX are what is known as the manifest price; that is, for that bill you get a rocket vehicle, a standard payload adapter (SpaceX defaults to the EELV Standard Interface Specification but will provide other standard interfaces at a nominal cost or custom payload adapters for a premium), a two phase coupled loads analysis cycle, a 10K (ISO 7) spacecraft integration/encapsulation environment, required horizontal integration, and no payload health and status monitoring or external power. If you want anything else, like a higher cleanroom rating, additional CLA verification cycles, H&S or power, vertical integration, H&S monitoring, supplied power, an ESPA Ring or other secondary payload adapter, payload isolation system, et cetera, you either provide it yourself or pay a premium. If you want simulation data or build “paper” to verify qualification and acceptance in order to perform independent quality and mission assurance, you pay a huge premium and still listen to Gwynne Shotwell harp on endlessly about it.
All of these kinds of costs are already represented in the launch costs of existing vehicles (where they are provided) and some of the difference between US/Ariane and Russian/Chinese launchers are reflected in these capabilities. Now, the argument can be made (and has been made by many in the responsive space arena, including your humble author) that modern spacecraft should be designed to be capable of horizontal integration and should be designed to either be self-damping or tolerant to a reasonable spectrum of vibration loads without requiring payload isolation (e.g. “rocksats”), thus reducing the costs associated with iterative load cycle analysis and all of the tender loving care required from many satellites, but given that the payloads that the F9 and FH are likely to carry will consist largely of multiple payload deployments with complex load paths it probably isn’t completely practical to build and fly rocksats that don’t require some measure of cleanliness and tender loving care beyond the minimum.
The capabilities that SpaceX offers beyond basic launch to orbit are that as a US-based company with a (presumably) EELV-certified launch vehicle they don’t have the kind of International Trafficking in Arms Regulation issues that foreign launch providers (even ones from friendly nations) have, and therefore can carry critical national security and NPR 8507.4 Class A and Class B payloads. Aside from ULA, they’re the only company that is or will in the foreseeable future be certified at this level, provided they are successful in getting EELV and Commercial Crew Transportation Capability (CCtCap). They’re also certified for operation on the US Eastern and Western Ranges by the 45th and 30th Space Wings, and can therefore launch from the major coastal launch sites where existing payload processing facilities and logistical support exists, putting them on equal footing with ULA. And so far they’ve had a good record of launch successes with the Falcon 9 vehicle, albeit with some significant anomalies on nearly every launch that they’re working to address. (This isn’t a bash at SpaceX; teething problems with a new design system are common, and it is actually pretty remarkable that they haven’t experience a loss of vehicle yet.) They also claim multiple restart capability on the Stage 2 Merlin Vac-D engine and long duration (several hours) of coasting time, which makes it fairly comparable (though less efficiently performing due to the use of RP-1 fuel) to the Centaur Upper Stage and Delta Cryogenic Second Stage.
On the other hand, by committing to such a large vehicle SpaceX is going to have to support multiple payload deployment, which is a highly complex operation requiring considerable mission specific customization and analysis. The only unitary payloads that need that kind of capability are the NRO payloads, interplanetary missions, and the big commercial telecom sats. The system is way oversized for smaller Earth imaging and other LEO payloads, which is where some of the smallsat/nanosat launchers like Generation Orbit, Electron, et cetera may come in to offer smallsats a ride to their desired orbit rather than just whatever inclination that a Falcon launch may drop them at. The payload to LEO cost is higher owing to the smaller capability (but still requiring a non-scaling amount of labor) but the desire to maximize science or commercial mission objectives may justify the cost versus tagging along as a secondary payload with no real say in the mission.
Stranger
