H — Japan, 1986

The original H-I was built by Mitsubishi as a derivative of their N series rockets, which were licensed copies of Delta. The first H retained the Delta-based first stage but used an original upper stage which burned hydrogen (hence the letter H) and could re-ignite in orbit. The successor H-II was an all-Japanese design. It was replaced with the cheaper and more reliable H-IIA in 2001, and in 2007 it was privatized, so Mitsubishi has full ownership. The H-IIB arrived in 2009, featuring a fatter first stage with two engines. This has mainly been used for space station service. The A remains in use for lighter loads.

Japan had a lot of smaller rockets as they gradually built up their expertise, mostly named after letters of the Greek alphabet. Now those are all gone, except that they’ve gone back to that in naming the new Epsilon. One odd trait of these earlier rockets is that their lower stages were completely innocent of any form of active guidance. They were made this way so as to prove they would be useless as potential ICBMs. They had to be launched from a leaning tower to aim them toward orbit.

The core stage of the H-II burns hydrogen (so the “H” name is more applicable than ever), and is augmented with either two or four stubby solid boosters, sometimes with smaller booster-ettes tucked in between. The H-IIB always uses four, as its whole point is for heavier lifting than the A. The second stage also burns hydrogen. The original H-I upper stage engine used a conventional gas generator cycle, but for the II they switched to an expander bleed cycle — the first engine with that cycle to enter service. (Blue Origin also uses that cycle for their BE-3U upper stage engine on the New Glenn.)

There were a couple of different plans to launch human beings on an H-IIA, with both capsule and spaceplane designs existing on paper, but none were built.

H3

But first they developed an H-III, which keeps the wider tank diameter from the H-IIB. (They also developed the ability to say “H3” instead of using roman numerals.) For this they replaced the dual LE-7A hydrogen engine with either two or three new engines called LE-9. The old engine used staged combustion, but for the new one they are building the world’s biggest expander-cycle engine, apparently on the belief that higher reliability matters more than the higher specific impulse that a closed cycle offers. They’re powerful enough that even with just two engines, this will increase the stage’s thrust considerably. The LE-5B expander on the second stage is getting a small upgrade. The III has similar solid boosters to the ones the IIB used, with a touch more propellant. So far, there’s been no talk of bringing back the mini-boosters that were sometimes used on the IIA.

Otherwise, not much else has changed, so it doesn’t sound like the cost could be reduced all that much, but by adding up a lot of small savings they get a substantial overall improvement. They aimed for a cost of $40 million per launch, which is half of what the old rocket cost, and underprices the Falcon 9. As this neared readiness, the IIB was retired, while the IIA remained active for a few more flughts.

The H3 can have zero, two, or four solid boosters. When no strap-ons are used, they fit three LE-9 engines under it. This is apparently the preferred configuration for low orbit, with the boosters being added on mainly for geosynchronous and higher destinations, though the LEO capacity without them is only about six tons. (They don’t like to give payload specs for LEO at all, focusing on sun-synchronous and geostationary capacities. The first test was sun-sync, which involved a rather severe dogleg in the flight path to stay safely over the ocean.) When the strap-ons are used, two core engines are enough; adding a third turned out to have very little benefit.

Unfortunately the first H3 launch failed. The complicated new first stage was fine; it was the relatively unchanged upper stage that failed to light up. And unfortunately, because of the long delay in getting off the ground, they put quite an expensive satellite onto the test flight. This led to a lot of public criticism and hand-wringing which only further delayed the second attempt. It was almost a year before the second launch got it to orbit... and that flight had a mostly dummy payload. A real payload was finally put up several months after that.

And after that, the H3 started getting more customers than JAXA expected. As the Vulcan and the Ariane 6 struggle to get a good cadence going, orders are coming in for H3 flights. Not bad for a rocket that was just meant to be a national security asset rather than a commercial competitor. The first five launches were all of the H3-22S variant, which uses two solid boosters and a short fairing. The sixth was a 24W (four boosters, wide fairing). Unfortunately the seventh launch (another 22S) failed again: apparently the payload was damaged during ascent, as was its mounting adapter, most likely when the fairing opened. Something violent happened at that moment, leaving the upper stage leaking hydrogen, and the payload no longer attached to it. Once the acceleration paused so the upper stage could separate from the booster, the damaged satellite floated loose and got left behind. Kind of a weird mishap.

For the future they are considering a triple-core H3 Heavy that would be able to lift 28 tons. That would give this system a wide range of capacities comparable to that of the Angara.