Rockets of Today

ATLAS — USA, 1958

Like the Soyuz, this rather large rocket was first developed as a long range ICBM, and was also obsoleted from that role as smaller rockets became capable enough. It went on to far more positive uses: it has not only launched plenty of satellites, but it’s sent up manned missions as well during the Mercury program, and launched early interplanetary probes such as Surveyor and Mariner. Today’s version has a somewhat distinctive look due to its bottom end, which has two combustion chambers and two nozzles on a single engine, plus plumbing that bulges out of the rocket’s side. (This double-nozzle design was something once associated with the Titan booster, which was used for the Gemini program and for many historic interplanetary probes. Titan dwarfed the early Atlases but has been left behind by the V. It was discontinued in 2005.)

The fuel is kerosene, or “RP-1” as it’s officially called when refined for rocketry, combined of course with liquid oxygen, which is called “lox” for short. But embarrassingly, the engines used in the current Atlas versions are the Energomash RD-180, and NPO Energomash is majority owned by the Russian government, and furthermore the import company appears to have been skimming millions of dollars per engine, which may be going directly to Putin. Naturally, there has been a lot of pressure to find an alternative motor. But no American engine builder has anything handy that can work as a replacement, and at this point, the plan is to just replace the Atlas with a new rocket: the Vulcan.

The original Atlas versions had a single nozzle in the middle, but also had two additional nozzles attached on the side, as small protruding bulges. They were not separate boosters, but extra engines which drew fuel from the main tank, and then dropped off once they were no longer needed. They were tuned for high thrust while the center engine was tuned for high specific impulse, consuming fuel more slowly. Without the boosters, it was incapable of liftoff. It also had small vernier thrusters for steering, and unusually, they were mounted up on the sides rather than at the base. The booster’s steel tanks were incapable of supporting their own weight unless pressurized, even when empty. (It’s more common for rockets to require some pressure when full.) The walls were so thin that they couldn’t even be painted without adding excess weight. This unusual construction made the rocket so lightweight that it could take some satellites to orbit with no upper stage. Mercury capsules were launched this way, in fact... and at the time John Glenn rode one into orbit, the rocket’s failure rate was around fifty percent. Over the years these classic Atlases were continually improved, and the tanks were repeatedly lengthened by small increments. Some old stock, including recycled ICBMs, was still being flown to orbit as late as 1995 (and for other purposes maybe as late as 2001), with an assortment of ad-hoc upper stages. Sometimes they’d put three satellites in the fairing, each with its own separate upper stage. Among the versions purpose-built for orbital use, many used Centaur upper stages, which also evolved. The smaller Agena upper stage saw use until 1978.

Atlas V

The Atlas II got rid of the side verniers and added modern roll control thrusters. It also added the option to strap on solid boosters. The III replaced the triple engines with the RD-180. The V replaced the “balloon tank” construction with rigid aluminum in a larger diameter, dropping the last visible remnant of the old missile. (There was no IV.) One legacy of this heritage is that today’s Atlas boosters still go to unusual heights and speeds before the second stage has to do any work. This makes things easier on the Centaur’s low-thrust engine.

Nowadays, lift can be enhanced for larger payloads by strapping up to five solid-fuel boosters onto the sides. But because the original rocket wasn’t designed for this, these strap-ons end up placed in odd and asymmetrical ways. They saw no reason to re-engineer the first stage’s exterior to correct this, even though the Atlas V is 25% bigger in diameter than earlier models. For most of the rocket’s life these boosters were made by Aerojet Rocketdyne, but ULA recently replaced them with Northrop Grumman’s new “GEM 63” model (developed by Orbital ATK before their merger), which is significantly cheaper and a bit more powerful. This will pretty much leave Northrop with a monopoly on large solid rockets in the USA... a monopoly that has little importance as the only major rockets still planning to use them are the SLS (which will fly very rarely) and the Vulcan, which will use a lengthened version of the Atlas booster called GEM 63XL.

(On the other hand, there’s a startup company called Adranos which is entering the market for solid motors. They’ve got a fuel blend with lithium in it, which may not be a good long term choice.)


There’s a choice of “Centaur III” second stages: a classic single-engine model, and a higher thrust one with two engines, which is rarely used... it had actually gone unused for many years until they brought it back for launching the Starliner crew capsule in 2020. The Centaur burns liquid hydrogen rather than kerosene, and ignites it electrically so it can start and stop as many times as needed. (Regardless of whether the igniter is chemical or electrical, it is actually quite challenging to ignite a cryogenic-fuel rocket engine in a vacuum, and to this day a lot of rockets have second stages which cannot re-ignite after their initial burn.) Hydrogen has higher efficiency but lower thrust, so it’s not uncommon to see it used on upper stages while using kerosene on the lower one... but the Centaur was the O.G., the first liquid hydrogen orbital rocket stage. It still uses balloon tanks, with the steel being just half a millimeter thick under the insulation. And these tanks retain the original booster’s diameter of 3.05 meters, or ten feet. The difference of fuels does largely prevent the lower and upper stages from being able to share any parts. Its RL10 engine uses a closed expander cycle. The single-engine Centaur’s thrust is quite low at just 99 kilonewtons, only enough for about a third of a G with full tanks and a large payload. This is one reason why the first stage has to go high and fast.


In many ways the Atlas seems archaic and clumsy, but United Launch Alliance (handed down from Lockheed Martin, who took over from General Dynamics, who got it from Convair) managed to continue selling it by keeping the costs reasonable. They’ve even managed to cut prices in response to the competition from SpaceX. As mentioned, they now see the writing on the wall, and are trying to get their replacement ready in the next few years. But don’t underestimate the old-timer: the Atlas V has flown dozens of launches without ever losing a single payload — a record that no other rocket can match. In fact, the preceding Atlas III and Atlas II models also had no failures, so the V was just continuing their streak. We shall see if the Vulcan does likewise. (Other rockets have had longer runs of consecutive successful launches, after initial failures. For instance, Ariane V is now on a longer run that the Atlas V, with each streak having a single blemish where a satellite reached a mildly incorrect orbit which forced it to expend onboard fuel. Delta II had an even better run with 100 consecutive successes, ending with retirement, and of course the Falcon 9 now has a record-shattering launch streak going. The Long March 2/3/4 can claim two separate hundred-launch streaks — the first with one blemish, and the second ongoing now. But only Atlas II/III and Atlas V include their very first launches in such a streak.)

What they lost instead was any excuse to continue buying Russian engines. Congress tried for fifteen years to pass laws and policies which would end our dependence on imported engines for national security access to space, and Lockheed and ULA and Aerojet Rocketdyne have repeatedly evaded or simply ignored this mandate. For instance, when told to develop a US-built copy of the RD-180, Aerojet Rocketdyne accepted fat checks for years to spend on the effort, then announced that they didn’t feel like finishing the job. Getting away with that takes some hard-core cronyism.

Now that there is finally a firm schedule in place to move on from the Atlas V and stop buying RD-180 engines, and ULA has already received all the engines it needs to finish the Atlas’s run (about 25), Russia is hoping that some rocket maker in China will buy them. (They also laughably blustered that they would now sell us no more engines, after we’d already stopped buying.) Since no current Chinese engine is even half as powerful, they may find a buyer. They are aware that this may just lead to Chinese copies of the engine being made, but figure that selling some beats selling none. And meanwhile, Aerojet Rocketdyne eventually did finish their replacement, and now they’ve got no customer for it, except one startup named Firefly which had yet to get things reliably to orbit, but said that when they do, they want to use this engine for the bigger followup rocket they’ll make someday if the first one is successful.

The Atlas V model has never carried a human being, but the plan is that it will do so in its final year or two if the Starliner capsule can qualify soon, and that’s good, because despite the problematic politics of its construction, the V might well be the safest rocket a person could ride on.

Atlas V 401 (no extra boosters): mass 334 t, diam 3.8 m, thrust 3827 kN, imp 3.3 km/s, staged combustion (kerosene), payload 9.8 t (2.9%) [18.5 t (3.2%) with 5 boosters], cost $11M/t, record for “V” (through June 5, 2024) 100/0/0! (1 crewed) — for legacy versions about 268/19/47 (4 crewed)
Records of Atlas orbital versions:
early ad-hoc  ’58–’95  45/10/8
Atlas-Able ’59–’60 0/0/3
Mercury-Atlas  ’59–’63 4/4/2 (4 crewed)
Atlas-Agena ’60–’78 90/0/20
Atlas-Centaur  ’63–’97 60/5/14
Atlas II ’91–’04 63/0/0
Atlas III ’00–’05 6/0/0
— so far: —
Atlas V ’02– 100/0/0 (1 crewed)
Stage name AJ-60A (’03-’20) GEM-63 (2020+) Atlas CCB Centaur III
Role (pos) count booster (S) ×0-5 booster (S) ×0-5 core (1) upper (2)
Diameter (m)   1.58   1.61   3.81   3.05
Liftoff mass (t) 46.7 49.3 305.1  23.1
Empty mass (t)  2.2  5.2 21.1    2.2 *
Fuel mass (t) ~13    ~13.5  ~76.3  ~3.0
Oxidizer mass (t) ~30    ~30.5  ~208     ~17.7 
Fuel type HTPB HTPB+Al kerosene hydrogen
Engine Aerojet-Rocketdyne
RL-10C ×1-2
Power cycle solid solid staged expander
Chamber pres. (bar) 100    267    24  
Ox./fuel ratio   2.3?   2.3?   2.72   5.88
Thrust, vac max (kN) 1690     1650     4152      106.3 *
Thrust, SL initial (kN) ~1110      3827    
Spec. imp, vac (km/s)   2.74   2.74   3.31   4.41
Total imp, vac (t·km/s) 117    120    943    ~91