Modules 7 and 8 told you the velocity changes — the "delta-v" — that getting around the solar system requires. We never said what those delta-v's actually cost. The answer is the rocket equation, and it is by far the harshest constraint in spaceflight engineering. It explains why a rocket on the launch pad is mostly fuel, why we still bother with multiple stages, and why ion engines exist.
In an everyday vehicle — a car, a boat, a plane — you change your motion by pushing against something. The car pushes the road, the boat the water, the plane the air. You can accelerate, brake, steer, and as long as you have fuel and there's something around to push against, you're free to drive wherever the topography allows. None of this is true in space. There's no road, no water, no air. There is, in fact, nothing at all to push against.
The only handle physics gives you is momentum conservation. To make yourself move one way, you have to push something else the other way — and you have to do it with material you brought along, because there's nothing else to push. Every spacecraft maneuver therefore reduces to the same trick: throw mass overboard fast enough, and you accelerate in the opposite direction. The faster you throw it, and the more of it you throw, the more your velocity changes.
Velocity change — delta-v ($\Delta v$) — is the right currency for thinking about maneuvers because it captures what the orbit-changing math from module 7 and module 8 actually requires. A Hohmann transfer to GEO doesn't ask for "fuel" — it asks for two specific velocity changes: one at the start to leave LEO, one at the end to circularize at GEO. Skip either burn and the geometry doesn't close. The rocket equation will translate that velocity-change demand into a propellant cost, and we'll see that the relationship is not gentle.
Imagine a rocket at rest in deep space, mass $m$. It ejects a small chunk of propellant, mass $dm_p$, at velocity $v_e$ relative to the rocket (the exhaust velocity). By momentum conservation, the rocket gains a small velocity $dv$ in the opposite direction. Since the propellant carries momentum $-v_e \, dm_p$, the rocket must carry $+v_e\, dm_p$ — and since the rocket's mass dropped by $dm_p$ (from $m$ to $m - dm_p$), its velocity change is $dv = v_e\, dm_p / m$.
That's the per-step relation. Now take a continuous limit: the rocket throws away propellant gradually, its mass shrinks from $m_0$ at start to $m_f$ at end, and the velocity changes accumulate. Setting up $dv = -v_e\, dm/m$ (the minus sign because $dm$ is the rocket's mass decreasing) and integrating from $m_0$ to $m_f$:
Start with the per-instant relation:
$$ dv = -v_e\, \frac{dm}{m} $$Integrate from initial mass $m_0$ to final mass $m_f$:
$$ \int_{0}^{\Delta v} dv = -v_e \int_{m_0}^{m_f} \frac{dm}{m} $$The integral of $1/m$ is $\ln m$:
$$ \boxed{\, \Delta v = v_e \, \ln\!\left( \frac{m_0}{m_f} \right) \,} $$This is the Tsiolkovsky rocket equation, derived by the Russian schoolteacher and rocketry pioneer Konstantin Tsiolkovsky in 1903 — six years before Robert Goddard built America's first liquid-fuel rocket and twenty-three years before Goddard's first liquid-fuel test flight. Tsiolkovsky never flew anything. He worked the math out from first principles in a small Russian town with no laboratory, deriving in one short paper the entire energy economics of spaceflight.
The ratio $m_0 / m_f$ is the mass ratio, often written just as $R$. The propellant is the difference $m_0 - m_f$, so the propellant fraction $m_p/m_0 = 1 - 1/R = 1 - e^{-\Delta v / v_e}$.
Engine designers don't usually quote $v_e$ directly — they quote a related quantity called specific impulse:
$$ I_{sp} = \frac{v_e}{g_0}, \qquad g_0 = 9.807\ \mathrm{m/s^2} $$$I_{sp}$ has units of seconds. (The $g_0$ factor is a historical curiosity from when thrust was measured in kilograms-force; it's purely a unit conversion and has nothing to do with the local gravitational field.) Higher $I_{sp}$ means a more efficient engine — you produce more delta-v per unit propellant mass. Doubling $I_{sp}$ doubles $v_e$ and halves the propellant required for a given $\Delta v$.
Different engine technologies live in dramatically different $I_{sp}$ regimes:
| Engine type | Example | $I_{sp}$ (s) | $v_e$ (km/s) |
|---|---|---|---|
| Solid propellant | Shuttle SRB | 268 | 2.6 |
| Kerosene + LOX | Merlin 1D vac (SpaceX) | 311 | 3.1 |
| Hypergolic | Storable upper stages | 320 | 3.1 |
| Hydrogen + LOX | RL10 (Centaur) | 451 | 4.4 |
| Nuclear thermal | NERVA (tested 1960s) | 850 | 8.3 |
| Hall thruster | SPT-100 | 1,600 | 15.7 |
| Gridded ion | NSTAR (Dawn, DS1) | 3,100 | 30.4 |
| Gridded ion (next-gen) | NEXT (NASA) | 4,170 | 40.9 |
Chemical rockets — the only engines that have ever launched a payload from Earth's surface — top out around $I_{sp} = 450\ \mathrm{s}$, set by basic thermochemistry: hydrogen-oxygen reaction enthalpy. To get higher $I_{sp}$ you have to leave chemistry behind. Nuclear thermal heats hydrogen with a fission reactor to 2,500 K, doubling the exhaust speed. Electric propulsion (Hall, ion) ionizes propellant and accelerates it through electromagnetic fields to ten or twenty times chemical exhaust speeds, but at extremely low thrust — milliNewtons rather than meganewtons. Different engines for different jobs.
The deceptive thing about the rocket equation is its log structure. $\Delta v$ scales linearly in fuel only when the propellant fraction is small. Once you start needing $\Delta v$ comparable to $v_e$, the cost climbs exponentially. To see it, rearrange:
$$ \frac{m_0}{m_f} = e^{\Delta v / v_e} $$Each extra factor of $v_e$ in $\Delta v$ multiplies the mass ratio by $e \approx 2.72$. So:
For a kerolox engine ($v_e \approx 3\ \mathrm{km/s}$), getting to LEO takes about 9.4 km/s of $\Delta v$ (the orbital velocity of 7.8 km/s plus about 1.6 km/s of gravity and atmospheric losses during the climb). That's $9.4/3 \approx 3.1$ — comfortably in "almost all of you is propellant" territory: a single-stage kerolox vehicle to LEO is about 95% fuel by liftoff mass. With hydrolox you can shave it to about 88%, which is why upper stages so often use hydrogen.
Here's where the kinematic delta-v's from module 7 become real engineering numbers. The LEO-to-GEO Hohmann transfer needed two burns — that's not optional. The first burn (about 2.46 km/s) raises apoapsis from LEO to GEO altitude, putting you on the transfer ellipse. The second burn (about 1.48 km/s), six hours later at the top of that ellipse, raises periapsis to GEO altitude and circularizes you there. Total: 3.94 km/s, both burns required.
Plug each burn through Tsiolkovsky in turn, with a kerolox engine ($v_e = 3.05\ \mathrm{km/s}$):
After burn 1 — kick to the transfer ellipse, $\Delta v_1 = 2.46\ \mathrm{km/s}$:
$$ R_1 = e^{2.46 / 3.05} = 2.24 \quad \text{(55\% of mass burned)} $$After burn 2 — circularize, $\Delta v_2 = 1.48\ \mathrm{km/s}$ for a total $\Delta v = 3.94\ \mathrm{km/s}$:
$$ R_{\text{total}} = e^{3.94 / 3.05} = 3.64 \quad \text{(72.5\% of original mass spent on propellant)} $$Notice what happens if a designer is tempted to "save fuel" by skipping burn 2. Yes, you'd save about 17 percentage points of propellant mass — but you wouldn't be in GEO. You'd be in a 6,571 × 42,157 km elliptical orbit forever, slowly tracing back through LEO altitude every six hours, never catching the geostationary belt. The two-burn delta-v is non-negotiable; it's what the geometry of changing one circular orbit for another requires. The propellant cost can be optimized — better engines, gravity assists, ion drives — but the delta-v itself is geometry.
If we replaced the kerolox engine with an NSTAR ion thruster ($v_e = 30.4\ \mathrm{km/s}$), the same 3.94 km/s Hohmann would cost a mass ratio of $e^{3.94/30.4} = 1.14$, or just 12% propellant fraction — nearly an order of magnitude better. The catch: ion thrusters produce 0.09 N of thrust, so what was a few minutes of chemical burn would become months of continuous ion firing, and the "Hohmann" approximation no longer applies (we'd need a low-thrust trajectory analysis). Different engines, different missions.
Once you've burned the propellant in a tank, the empty tank is just dead weight. Every subsequent burn now has to lift, accelerate, and decelerate that mass too — it eats into the next $\Delta v$. The trick is obvious in retrospect: throw the empty tank away. Build the rocket as a series of independent vehicles stacked on top of each other, each with its own engines and tanks. When one stage runs dry, drop it.
The Saturn V illustrates the win. Its launch mass was about 2,950 tonnes; the spacecraft after the trans-lunar injection burn was about 47 tonnes. That's a mass ratio of 63. With a $v_e$ around 3 km/s (averaging across stages), a hypothetical single-stage rocket of that ratio could deliver $3 \cdot \ln 63 \approx 12.4\ \mathrm{km/s}$ of $\Delta v$ — but Saturn V actually delivered closer to 17 km/s when you sum the stages. The extra 4-5 km/s is the staging benefit: each stage didn't have to drag the empty tanks of the previous one. Staging effectively turns the log curve into a sum of log curves, and the sum wins.
The widget below lets you set a desired $\Delta v$ and an engine $I_{sp}$ and reads out the resulting mass ratio and propellant fraction. Click an engine preset to jump to a real engine. Click a mission preset to jump to a real $\Delta v$ — including the LEO-to-GEO total that decomposes into two burns.
Slide the $\Delta v$ slider out toward the right end with kerolox selected. By 9-10 km/s the propellant fraction is over 95% — almost all rocket. By 16 km/s (Pluto direct) the mass ratio is over 200; you'd need to launch 200 tonnes to deliver one tonne to Pluto. This is exactly why every outer-solar-system mission (Voyager, Galileo, Cassini, New Horizons) used gravity assists to reduce its true $\Delta v$ requirement, and why Pluto specifically was visited only once, with a tiny payload, on a mission that took ten years. The rocket equation isn't gentle.
Now switch to the ion preset. The same Pluto $\Delta v$ becomes a 41% propellant fraction — the same trip becomes plausible from a fuel standpoint. The catch (which we'll see in module 10) is that ion thrusters take months to apply that delta-v, not minutes. There's no free lunch — you trade fuel cost for time cost. But it's a very different trade than a chemical rocket has to make.
The rocket equation tells you the cost of a given delta-v. The next module — burns and launch windows — answers the rest of the engineering puzzle: when and how to spend that delta-v efficiently. We'll see why every interplanetary mission has tight launch windows separated by months of waiting; why a long burn at low altitude beats a short burn at high altitude (the "Oberth effect" we hinted at in module 8); and why precise timing within a burn matters as much as precise sizing.