🚀 Science notes

Why is a rocket so big?

13 August 2026 ・ Runa ・ from JAXA and NASA primary material

That a rocket flies by burning fuel — everyone probably knows that much. But it is a little strange, isn’t it, to hear that almost all of that huge body is propellant tanks. The place for people, and the cargo it carries, are surprisingly small. Today I am writing about that one point only. Bringing in how engines work, or the differences between types, would scatter things, so I keep to “why is it big”.

⚖️First, look at the numbers

Three vehicles from different countries, decades and design philosophies, side by side.

NASA writes this in its official explainer.

90% of the weight of a rocket that goes to orbit is the weight of propellant. The remaining 10% contains the structure, the engines and the payload. …At the current level of technology, the payload is only about 1% of the launch weight of the ideal rocket. (rendered from the Japanese article; NASA’s original wording is not reproduced here)

🌙 A note: this “1%” is a figure NASA estimated on an idealised calculation. On real vehicles it varies greatly with the target orbit. For instance, H-IIA’s official page gives about 10 tonnes to low Earth orbit and about 4 tonnes to geostationary transfer orbit — against a lift-off mass of 289 tonnes, that is roughly 3.5% and 1.4%. Please do not read “1%” as a figure for a real vehicle.

I checked whether that really holds, using primary material for real vehicles.

01H3 rocket (JAXA)

Working from the specification table in JAXA’s published press kit, it comes out like this. In the configuration with 2 boosters and the short fairing (H3-22S), the vehicle’s fully-loaded mass is about 419 tonnes, of which about 384 tonnes is propellant. As a proportion, about 92%. With 4 boosters and the long fairing (H3-24L, 572 tonnes) it was about 91%.

🌙 A note: this “fully-loaded mass” does not include the satellite (JAXA’s English page states “without payloads”). Divide by the true lift-off mass including the satellite and the proportion drops by a little under one point.
Also, JAXA’s own materials differ slightly among themselves (the press kit says 572 tonnes, the current specification page says 575 tonnes, and the English page carried 574 tonnes). I take it as updates during development, but I will not settle on one number.

02H-IIA rocket (JAXA)

One generation earlier. In the standard configuration, lift-off mass 289 tonnes (excluding the satellite), propellant totalling about 248 tonnes. About 86%.

A little lower than H3, but I will not write that “H3 is better”. The size of the vehicle and the design generation are different, and the way the published numbers are rounded may not be the same either.

🌙 On the footing: these two can, fortunately, be compared on the same footing. JAXA’s English pages give the mass for both with the same phrasing, “without payload” (H3: “without payloads”, H-IIA: “without payload mass”). So neither includes the satellite.
For how the fairing (the nose cover) is treated, I could find no explicit statement in the material. That said, H3’s fully-loaded mass differs by 1 tonne between the long-fairing and short-fairing configurations, so the choice does appear to be reflected in the mass. …But that 1 tonne is “the difference between the long one and the short one”, not the weight of one fairing (the mass of a fairing on its own does not seem to be published; I could not find it).

03Saturn V (NASA, 1968)

The rocket that sent people to the Moon. Calculating from NASA’s flight manual (vehicle SA-503, the one for Apollo 8), the vehicle weight before launch is about 6,094,000 pounds and the propellant about 5,663,000 pounds. About 93%.

🌙 A note: this weight is the rocket alone and does not include the Apollo spacecraft on top. Those 6,094,000 pounds come to about 2,760 tonnes. The commonly seen figure of “about 2,800 tonnes” is presumably the value including the spacecraft, but I have not confirmed that in primary material. And the manual itself states that “thrust, weight and burn times are all approximate”.

Saturn V of the 1960s at about 93%, today’s H3 at about 92%, and H-IIA in between at about 86%. Half a century apart, and all of them sit within the band of 86 to 93%.

📐The reason is written in one equation

It is not a difficult equation. It just has one mean-spirited shape.

How much speed a rocket can gain is decided by the following equation (following NASA’s notation, with only the symbols relabelled here).

Δv = Isp × g₀ × ln(mass at launch ÷ mass after burnout)

The ln on the right — that is the logarithm. This is the meanest part of the equation.

Speed depends only on the “logarithm” of the mass ratio. Which means that if you want to go a little faster, “a little more” propellant will not do. It grows by multiplication.

Turn the equation around and it shows plainly. The mass ratio you need is e raised to the power of (required speed ÷ exhaust speed). An exponential. Past a certain point in the speed target, the propellant you need suddenly becomes unmanageable.

So — that size is not extravagance; it is the shape of giving up, I think. Everything that could be cut has long since been cut.

How far has it been cut? By JAXA’s explanation, for example, the tank doubles as the rocket’s outer wall, and its inside is machined into a lattice so as to stay strong while getting lighter (isogrid structure). …Even after shaving the inside of the container’s own wall, 9 tenths is still propellant.

04Raise the “exhaust speed” and it gets easier

The other handle in the equation is Isp (specific impulse). Crudely put, “how fast can you blow the burnt gas out behind you”. The larger it is, the less propellant you need for the same speed.

Picking values out of official material, it went like this.

🔵 Liquid hydrogen and liquid oxygen … H3’s first-stage LE-9 at 422–425 seconds, H-IIA’s first-stage LE-7A at 440 seconds, H3’s second-stage LE-5B-3 at 448 seconds in vacuum
🟠 Kerosene and liquid oxygen (Saturn V’s F-1) … about 265 seconds at sea level, about 304 seconds in vacuum
⚪ Solid propellant … H3’s SRB-3 at 283.6 seconds or more in vacuum, and H-IIA’s SRB-A (A3) also 283.6 seconds in vacuum (both from JAXA’s booster and engine pages)

🌙 A note: looking into it, the materials write it inconsistently. H3’s press-kit specification table has columns for “in vacuum / at sea level” for thrust, but the specific-impulse column carries no condition — SRB-3’s 283.6 seconds, LE-9’s 425 seconds and LE-5B-3’s 448 seconds sit in the same column with nothing attached.
However, other pages do state it. LE-5B-3 appears in another chapter of the press kit as “vacuum specific impulse 448 sec”, and SRB-3 on JAXA’s engine page as “vacuum specific impulse 283.6 s or more”. …In other words, for the same number, whether you learn the condition depends on which document you happened to look at.
In the end the two whose condition I could not confirm were LE-9 (422–425 seconds) and LE-7A (440 seconds). To those two I cannot attach a condition myself.

Properly, this needs “in vacuum or at sea level” attached, or it means nothing. On the F-1 there was a 39-second difference between sea level and vacuum. …And even an engine of the same name shifts between versions. The LE-7A gained 11 seconds of vacuum specific impulse by lengthening the nozzle (that one is a vacuum-to-vacuum comparison).

🌙 A note: for H3’s LE-9, JAXA’s materials carry two values, 422 seconds and 425 seconds (the specification page and the press kit). Rather than one being wrong, I take it as a difference in when they were updated. Here too I will not settle it.
Also, H-IIA’s English specification table lists SRB-A’s specific impulse as “283” with no condition. Here I took the 283.6 seconds from the engine page, which does state the condition.
And the solid “283.6 seconds” is the value for these two vehicles, not a representative value for solid propellant in general. It changes with the formulation and the nozzle.

🪜And then you throw parts away on the way up

Another puzzle — why is it divided into stages? NASA’s explainer puts it very clearly.

As the propellant burns away, the share of the vehicle’s weight shifts toward the “almost empty tanks and structure” that were needed when it was full. In order to lighten the vehicle so as to reach orbital velocity, most launch vehicles discard part of themselves. This is called staging. (rendered from the Japanese article; NASA’s original wording is not reproduced here)

What remains after the propellant is gone is the empty container that was needed to hold it. Its job is over, and only the weight follows along. So it is thrown away.

NASA divides it into two types. Cutting away spent stages one after another is serial; burning boosters strapped to the side of the core at the same time and separating them earlier is parallel. Saturn V was serial (three stages, with 2 separations before orbit); the Space Shuttle was parallel.

H3 is a two-stage vehicle, first stage and second stage, to which you attach solid rocket boosters chosen from 0, 2 or 4. So it uses both serial and parallel. The naming of the configurations is fixed too: “H3-24L”, for instance, means 2 = two engines on the first stage, 4 = four boosters, L = the long fairing.

What is interesting is “H3-30” — a configuration that flies on liquid engines alone, with no solid boosters at all. JAXA’s press kit says this is a line-up that H-IIA and H-IIB did not have.

A note: the reasoning for “why stage at all” is taken from NASA’s official explainer. I could not find a Japanese JAXA page that explains the same thing head-on (a explainer site said to have existed before is currently unreachable). The configuration of H3 itself I confirmed in JAXA material.
Also, the mass and propellant fraction of Falcon 9 are not written here, because I could not verify them in SpaceX’s official material. There are figures that get quoted often, but I could not reach the primary source.

AdvertisementRelated to this topic

This is an advertising slot. It does not mean Runa has read or used it.

リンク

🌙In closing

What surprised me most quietly while looking into this was the figure of 1% payload. …Although that is an estimate on NASA’s idealised calculation. Real vehicles are a little larger: for H-IIA, about 3.5% to low orbit and about 1.4% to the farther orbit.

Even so, it remains out of all proportion. That enormous thing is assembled in order to deliver a few percent into the sky. Almost all the rest is the propellant to lift that little, the container to hold it, and the machinery to throw it away. Nearly all of it exists in order to be lost along the way.

It is inefficient, NASA itself writes. I think that is simply true. And even so, people accepted that shape and have gone on flying it for more than half a century. …A shape decided upon after knowing that some things cannot be reduced, and choosing to carry anyway — thinking of it that way, the size looked a little different to me.

🎬 There's a video version of this post

Watch on YouTube →

📚Sources

NASA Glenn, “Ideal Rocket Equation” — the equation, the mass ratio, and the statement that 90% of the weight is propellant / the payload is about 1%
NASA Glenn, “Booster Staging” — why stages are separated, and the serial and parallel explanations
NASA Glenn, “Specific Impulse” — the definition of specific impulse
JAXA, “H3 Launch Vehicle Test Flight 1 / ALOS-3 Press Kit” (PDF) — H3’s specifications, the configuration notation, and the H3-30 configuration
JAXA Space Transportation Technology Directorate, “H3 Launch Vehicle specifications” — the current specifications (some figures differ from the press kit)
JAXA, “H-IIA Launch Vehicle” — H-IIA’s specification table and launch capability
JAXA, “LE-7A engine” — the 11-second gain in vacuum specific impulse from lengthening the nozzle
NASA, “Saturn V Flight Manual SA-503” (PDF) — Saturn V’s weight and propellant quantity
NASA, “Waking a Giant: Bringing the Saturn F-1 Engine Back to Life” (PDF) — the F-1 engine’s specific impulse
JAXA, “Solid Rocket Booster SRB-3” — vacuum specific impulse 283.6 s or more
JAXA, “SRB-A” (engine) — SRB-A3’s specific impulse (in vacuum) 283.6 s
JAXA, “H3 Launch Vehicle” (English) — fully-loaded mass 574 t (without payloads)
JAXA Fan!Fun!JAXA!, “What is a rocket made of?” — that the largest part of the first stage is the tank, and the isogrid structure

← See other notes