Ariane 5

Heights across the family ran from 46 to 52 metres on a 5.4 metre core, with roughly 15,120 kilonewtons at liftoff, almost all of it from the two solid boosters. The 20 tonne low orbit figure belongs to the ES variant, the one built to push the ATV freighter to the space station. The ECA that flew most of the commercial missions was never sold to low orbit at all: it was a 10,865 kilogram geostationary transfer machine and that is the number its customers bought.

Flight 501 destroyed itself in June 1996 because inertial reference software carried over from Ariane 4 overflowed on a value that the new vehicle produced and the old one never had. It is the most-cited software failure in engineering, and the interesting part is what came after: between April 2003 and December 2017 the vehicle flew eighty-two consecutive successful missions.

The defining commercial compromise was dual launch. A SYLDA adapter stacked two large geostationary satellites in one fairing, which halves the cost per satellite only when you can find two customers whose masses and schedules line up. When you cannot, the rocket waits, and Arianespace spent years managing a manifest around that constraint.

The economics were never as good as the reliability. Published launch costs ran to 150 to 200 million euros while the price offered for a mid-size satellite in the lower berth had fallen to around 50 million by 2014. That gap is a subsidy, and it was the price Europe paid to keep an independent route to orbit. Ariane 6 was pitched at roughly half the Ariane 5 price, which tells you what everyone involved thought of the arrangement.

It finished in July 2023 with 117 flights and 112 successes, a 95.7 percent record, having launched the James Webb Space Telescope so accurately that the observatory gained years of propellant margin on the way out.