Variable Mass and the Rocket Equation
A rocket on the launch pad is mostly fuel. A Saturn V leaving Earth was about
85\% propellant by mass; a Falcon 9 is over
90\%. The whole machine is a tank that spends the trip throwing itself
away — burning through tonnes of propellant every second and hurling the exhaust out of the
back. That is the puzzle that makes rockets different from every other vehicle you know: a car, a
train, a plane all keep (very nearly) the same mass while they speed up. A rocket's mass is
changing violently the entire time it accelerates.
So how fast can a rocket go? You might reach for F = ma and, seeing the
mass shrink, write F = m(t)\,a with a smaller and smaller
m. That is wrong, and getting it right is the whole point
of this page. The clean way to think about a rocket is not "a lighter and lighter object being
pushed" but "a fixed lump of stuff — rocket plus its fuel — that keeps its total momentum
while it splits itself in two." Follow the momentum and out drops one of the most famous results in
all of engineering: the Tsiolkovsky rocket equation.
A warm-up on frictionless ice
Forget rockets for a moment. Stand on a cart on perfectly frictionless ice, holding a stack of heavy
bricks, everything at rest. Now throw a brick backwards. By
conservation of
momentum, the brick carries momentum one way, so you and the cart must roll off the
other way. Throw a second brick, then a third — each throw nudges your speed up a little more. A
rocket is exactly this, taken to the continuous limit: instead of bricks it flings out a smooth
stream of hot gas, millions of "bricks" a second.
The warm-up already contains the key lesson. Ask "what force acts on me?" and you get
muddled — my mass drops with every brick I release. Ask instead "is the total momentum of
me-plus-all-the-bricks conserved?" and the answer is a clean yes (nothing
external pushes on the ice-and-cart system). That is the bookkeeping that works: draw
the system boundary around everything — rocket and the fuel it is about to expel — and track
the momentum of the whole.
Exactly the same physics. A squid fills its body cavity with water and squirts it out through a
narrow funnel; the water leaves fast in one direction, so the squid jets off in the other. It is a
living rocket, and its "exhaust velocity" (how fast it can squirt the water) sets how much of a kick
it gets — precisely the role the exhaust speed u plays below. Octopuses,
scallops (clapping their shells) and even some jellyfish move the same way. Nature discovered
reaction propulsion a few hundred million years before Tsiolkovsky wrote it down.
Following the momentum through a small instant dt
Here is the careful derivation. Work in an inertial frame (say, deep space, no gravity) and watch the
rocket over a tiny slice of time dt. Let the exhaust leave the engine at a
fixed speed u relative to the rocket — this is the
exhaust speed, a property of the engine and propellant, not of how fast the rocket
happens to be going.
Just before the instant, the whole system is one object:
- mass m, velocity v,
- so total momentum p_i = m\,v.
During the instant the rocket ejects a small puff of gas. Let the rocket's mass
change by dm — note dm is
negative, because the rocket is losing mass, so the mass of gas thrown out is the
positive amount -\,dm. The rocket speeds up by
dv. Just after the instant we have two pieces:
- the rocket: mass m + dm, velocity v + dv;
-
the expelled gas: mass -\,dm, moving backward at speed
u relative to the rocket, so at velocity
v - u in our frame.
The total momentum afterwards is the sum of the two:
p_f = \underbrace{(m + dm)(v + dv)}_{\text{rocket}} \;+\; \underbrace{(-\,dm)(v - u)}_{\text{exhaust}}.
Nothing external acts on the whole system, so momentum is conserved:
p_f = p_i. Multiply everything out:
m v + m\,dv + v\,dm + dm\,dv - v\,dm + u\,dm = m v.
The v\,dm terms cancel, and dm\,dv is a product
of two tiny quantities — negligibly small (second order) — so we drop it. Cancelling the
m v on both sides leaves the clean heart of the matter:
m\,dv + u\,dm = 0 \qquad\Longrightarrow\qquad m\,\frac{dv}{dt} = -\,u\,\frac{dm}{dt}.
Read the right-hand side as a force. The quantity -\,u\,\dfrac{dm}{dt} is
the thrust. Because dm/dt is negative (mass leaving), the
thrust is a positive push forward. Burn fuel faster (a bigger |dm/dt|) or
throw it out faster (a bigger u) and you get more thrust — exactly as your
gut expects.
Integrating to the Tsiolkovsky equation
We have m\,dv = -\,u\,dm. To find the total speed gained over a
whole burn we integrate. Divide by m and add up from the start of the
burn (mass m_0, the "wet" mass) to the end (mass
m_f, the "dry" mass):
\int_{0}^{\Delta v} dv = -\,u \int_{m_0}^{m_f} \frac{dm}{m}.
The integral of 1/m is the natural logarithm. Treating
u as constant (a well-behaved engine), the right side is
-\,u\big[\ln m_f - \ln m_0\big] = u\,\ln\!\dfrac{m_0}{m_f}. That single step
is the whole game:
-
Thrust. A rocket ejecting mass at exhaust speed u
feels a forward thrust
F = -\,u\,\dfrac{dm}{dt} (positive, since
dm/dt < 0).
-
Velocity gained. Burning from initial mass m_0 down to
final mass m_f changes the speed by
\displaystyle \Delta v = u\,\ln\!\frac{m_0}{m_f}.
-
It is the mass ratio that matters. Everything hinges on
R = m_0/m_f (always > 1), and the
dependence is logarithmic, not linear.
Two numbers set a rocket's reach: the exhaust speed u (how good the engine
and propellant are) and the mass ratio R = m_0/m_f (how much of the rocket
is fuel). The achievable speed change \Delta v — the currency of
spaceflight, the "budget" you spend to leave Earth, change orbit, or land — is
u\ln R and nothing else.
The tyranny of the logarithm
Slide the exhaust speed u and watch how \Delta v
grows as you stuff in more fuel (larger mass ratio R). The curve is a
logarithm: it rises steeply at first, then flattens. Doubling your fuel does
not double your speed — it only adds another u\ln 2 \approx 0.69\,u.
This flattening is the famous tyranny of the rocket equation. Because
\Delta v depends on \ln R, the mass ratio you
need grows exponentially with the speed you want. Invert the equation:
\frac{m_0}{m_f} = e^{\,\Delta v / u}.
Want twice the \Delta v? You don't need twice the mass ratio — you need
the mass ratio squared (since e^{2x} = (e^{x})^2). Chemical
rockets have exhaust speeds of only a few km/s (roughly
u \approx 3{,}000\text{–}4{,}500\ \text{m/s}), while reaching low Earth
orbit needs a \Delta v of about
9{,}400\ \text{m/s}. That forces a mass ratio of roughly
e^{9400/4400} \approx 8.5 — so around
88\% of the rocket must be propellant, before you even add the cost of
gravity and air drag. Push for a higher target and the tank demands become brutal, fast.
Why rockets are built in stages
The exponential is why big rockets are built as stages stacked on top of one
another. The problem with a single giant tank is that near the end of the burn you are still hauling —
and accelerating — a huge, empty metal shell. That dead structural mass sits in the dry mass
m_f, dragging the ratio m_0/m_f down and stealing
\Delta v.
Staging fixes this by throwing the empty tank away. Once a stage's fuel is spent, the
rocket drops the whole spent stage — engines, tanks and all — so the next stage doesn't have to carry
or accelerate that dead weight. The magic is that the \Delta v values of
the stages simply add up, while each stage keeps its own healthy mass ratio:
\Delta v_{\text{total}} = u_1 \ln\!\frac{m_{0,1}}{m_{f,1}} \;+\; u_2 \ln\!\frac{m_{0,2}}{m_{f,2}} \;+\; \cdots
Adding several moderate logarithms beats trying to force one enormous mass ratio out of a single
stage. It is the engineering answer to the tyranny of the exponential — and why almost every rocket
that has reached orbit has shed pieces of itself on the way up.
Worked examples
Example 1 — Δv from a mass ratio. A single-stage rocket has an exhaust speed
u = 3{,}000\ \text{m/s} and a mass ratio
R = m_0/m_f = 5 (so four-fifths of it is fuel). Its speed change is
\Delta v = u\ln R = 3{,}000 \times \ln 5 = 3{,}000 \times 1.609 \approx 4{,}830\ \text{m/s}.
Note how much fuel bought how little speed: 80\% of the rocket was
propellant, and it gained under 5\ \text{km/s} — not even enough on its own
to reach orbit.
Example 2 — the mass ratio for a target Δv. You want a
\Delta v = 9{,}000\ \text{m/s} from an engine with
u = 3{,}500\ \text{m/s}. Invert the rocket equation:
\frac{m_0}{m_f} = e^{\,\Delta v/u} = e^{\,9000/3500} = e^{\,2.571} \approx 13.1.
The launch mass must be about 13 times the empty mass, so roughly
1 - 1/13.1 \approx 92\% of the rocket is propellant. Improving the engine
to u = 4{,}500\ \text{m/s} drops the required ratio to
e^{9000/4500} = e^{2} \approx 7.4 — a much easier machine to build. This
is why a modest gain in exhaust speed is worth so much: it sits outside the exponential.
Example 3 — the cost of going faster. A probe with
u = 3{,}000\ \text{m/s} already has a mass ratio
R = 4, giving \Delta v = 3{,}000\ln 4 \approx 4{,}160\
\text{m/s}. To double that \Delta v we need
R^2 = 16, quadrupling the mass ratio. To double it again we'd need
R = 256. Each doubling of speed costs a squaring of the tank — the
exponential wall that no amount of fuel comfortably climbs.
No — and this is the single most common mistake with rockets. It is tempting to say
"Newton's law is F = ma; the rocket's mass m(t)
just gets smaller as fuel burns, so F = m(t)\,a." That quietly ignores the
momentum that the ejected fuel carries away with it. The correct law is
F = dp/dt applied to a system of fixed contents — and a
rocket that is spraying mass out of the back does not have fixed contents, so you cannot just pull the
changing m outside the derivative.
Watch what goes wrong. Naïvely differentiating p = mv as if the same stuff
stays gives \dfrac{dp}{dt} = m\dfrac{dv}{dt} + v\dfrac{dm}{dt}, and setting
that to zero would predict the rocket slowing down as it loses mass — the opposite of what a
rocket does! The fix is exactly what we did above: enlarge the system to include the expelled gas, so
no mass ever crosses the boundary, and only then apply conservation of momentum. Do that and the
stray v\,dm term cancels against the momentum of the exhaust, leaving the
correct thrust -\,u\,dm/dt.
It doesn't — the dependence is logarithmic, not linear. Because
\Delta v = u\ln(m_0/m_f), adding more fuel raises
m_0 but the extra \Delta v comes only from the
logarithm of the new ratio. If a rocket has mass ratio R and you
add enough fuel to double it to 2R, the gain is
u\ln(2R) - u\ln R = u\ln 2 \approx 0.69\,u — a fixed, modest bonus,
not a doubling of \Delta v. Worse, doubling the fuel means far more
than doubling the tank, because you're now hauling all that extra propellant too. The honest picture
is the flattening curve above: fuel gives diminishing returns, and exhaust speed
u (a linear factor out front) is the lever that really pays.