Optimal staging — equal mass ratios (for same Isp): Visual walkthrough

Step 1 — What is a rocket made of? (mass, drawn as a stack)

WHAT. Before any formula, let us just draw the rocket as a stack of blocks. From the top down:

the payload $m_{p}$ — the satellite, the thing we actually want in orbit;
for each stage, a chunk of fuel (gets burned) and a chunk of structure (the empty tank, engine, plumbing).

WHY. Every symbol in this chapter is a mass — a how-much-stuff number measured in kilograms. If you can point to each block, no equation later can confuse you.

PICTURE. Look at the two coloured towers below. The orange blocks are fuel (they vanish as we fly). The violet blocks are structure (dead weight — we will throw them away). The tiny magenta cap on top is the payload $m_{p}$ .

Figure — Optimal staging — equal mass ratios (for same Isp)

Step 2 — The "squeeze factor" of one stage (the mass ratio $R$ )

WHAT. Pick one stage. Just before it fires it has some starting mass — call it $m_{before}$ . It burns its fuel and ends with a smaller burnout mass $m_{after}$ . The ratio of these two is the star of the show:

$R = how many times lighter we got \frac{m _{before}}{m _{after}}$

WHY this ratio and not the difference? Because the physics of a rocket (the Tsiolkovsky rocket equation) says the speed you gain depends on the ratio of before-to-after mass, not how many kilograms you dumped. Doubling the mass always buys the same speed, whether you go $2 \to 1$ tonnes or $200 \to 100$ tonnes. Ratios, not differences, are what rockets "feel."

PICTURE. The red bracket below shows $m_{before}$ (full tank) shrinking to $m_{after}$ (empty tank). $R$ is literally how many times taller the full tower is than the empty one.

Step 3 — Turning ratio into speed (why $ln$ appears)

WHAT. The speed gained by one stage is

$Δ v_{i} = exhaust speed v_{e} \times the log of the squeeze ln R_{i} .$

Here $v_{e}$ is the exhaust speed — how fast gas shoots out the back (bigger engine = bigger $v_{e}$ ). $Δ v_{i}$ ("delta-vee") is the change in the rocket's speed from that stage.

WHY a logarithm? Ask: "If I want twice the speed, how much heavier must I start?" The answer is not twice as heavy — it is squared. Speed adds up while mass ratio multiplies. The one function that turns multiplying into adding is the natural logarithm $ln$ : $ln (A \times B) = ln A + ln B$ . That is exactly why it shows up — it is the tool that converts "stack ratios" into "stack speeds."

PICTURE. The curve below is $ln R$ . Notice it rises fast at first then flattens: the first bit of fuel buys lots of speed, later fuel buys less. This "diminishing returns" shape is the whole reason a single giant stage is wasteful.

Step 4 — Stacking stages: speeds add, ratios multiply

WHAT. Two stages fire one after another. To stage 1, everything above it — stage 2 and the payload — is just "the payload it must lift." So:

$Δ v_{total} = Δ v_{1} + Δ v_{2} = v_{e} ln R_{1} + v_{e} ln R_{2} = v_{e} ln (ratios multiply R_{1} R_{2}) .$

WHY. Speeds gained in sequence simply add (you were already moving, now you go faster). And because $ln$ turns products into sums, adding the two $ln$ 's is the same as taking one $ln$ of the product $R_{1} R_{2}$ . This is the compounding trick: two stages of squeeze $R$ act like one super-stage of squeeze $R^{2}$ .

PICTURE. Below, stage 1 lifts the whole grey "everything above" block; after burnout it drops its violet structure and stage 2 takes over with a lighter load.

Step 5 — The catch: structure ties the stages together

WHAT. If there were no structure, we could pour all mass into fuel and any split of $R_{1}, R_{2}$ with the same product $R_{1} R_{2}$ would give the same speed. But real stages carry dead weight. We measure it with the structural fraction:

$ϵ = \frac{m _{s i}}{m _{s i} + m _{f i}} (same ϵ for every stage, by assumption) .$

$ϵ$ is "what fraction of a stage's wet mass is empty shell." $ϵ = 0.1$ means 10% dead weight.

WHY it matters — let us chain the masses ourselves. First give a name to the running mass as we climb the stack. Let $M_{0}$ be the whole rocket's launch mass, and let == $M_{i}$ == be the mass still flying after stage $i$ has burned and dropped its empty shell. So $M_{i - 1}$ is exactly "the mass entering stage $i$ " — the payload that stage $i$ must lift. By the very end nothing is left but the payload, so $M_{n} = m_{p}$ .

Now track one stage. Its ratio is $R_{i} = M_{i - 1} / m_{after, i}$ , so the burnout mass is $M_{i - 1} / R_{i}$ and the fuel it burned is $m_{f i} = mass in M_{i - 1} - mass after burn \frac{M _{i - 1}}{R _{i}} = M_{i - 1} (1 - \frac{1}{R _{i}}) .$ Its structure is tied to its fuel by $ϵ$ : rearranging $ϵ = \frac{m _{s i}}{m _{s i} + m _{f i}}$ gives $m_{s i} = \frac{ϵ}{1 - ϵ} m_{f i}$ . After the burn we drop that structure, so the mass handed up is $M_{i} = M_{i - 1} - m_{f i} - m_{s i} = M_{i - 1} \frac{1 - ϵ R _{i}}{( 1 - ϵ ) R _{i}} .$ Each stage multiplies the running mass by that same-shaped factor. Chaining $M_{0} \to M_{1} \to \dots \to M_{n} = m_{p}$ multiplies all the factors together:

$\frac{m _{p}}{M _{0}} = \prod_{i = 1}^{n} shrink factor of stage i \frac{1 - ϵ R _{i}}{( 1 - ϵ ) R _{i}} . (⋆)$

This is the true constraint. Notice it is not " $\prod R_{i} =$ constant" — the $ϵ$ makes each factor curved, and that curvature is what makes one particular split best.

PICTURE. The bar below splits one stage's wet mass into orange fuel and violet structure; the ratio of violet-to-total is $ϵ$ , fixed for all stages.

Step 6 — Why equal ratios win (the balancing picture)

WHAT — name the thing we want. Let == $f$ == be our prize: the total speed measured in units of $v_{e}$ . From Step 4, total speed is $v_{e} \sum_{i} ln R_{i}$ , so dropping the common $v_{e}$ , $f = \sum_{i} ln R_{i} (maximise this),$ subject to constraint $(⋆)$ . Take $ln$ of both sides of $(⋆)$ to turn its product into a sum and call that the constraint function $g$ : $g = \sum_{i} ln \frac{1 - ϵ R _{i}}{( 1 - ϵ ) R _{i}} = ln \frac{m _{p}}{M _{0}} (must stay fixed) .$

WHY the Lagrange condition, term by term. Lagrange multipliers says: at the best split, the "push" you get on $f$ from nudging any $R_{i}$ must be a fixed multiple $λ$ of the "push" on the constraint $g$ — the same $λ$ for every stage. In symbols $\partial f / \partial R_{i} = λ \partial g / \partial R_{i}$ . Let us differentiate each side with respect to $R_{i}$ :

$\frac{\partial f}{\partial R _{i}} = \frac{\partial}{\partial R _{i}} ln R_{i} = \frac{1}{R _{i}}$ — easy.
For $g$ , only the $i$ -th term contains $R_{i}$ . Split its $ln$ : $ln (1 - ϵ R_{i}) - ln R_{i} - ln (1 - ϵ)$ . The last piece is a constant. Differentiating the first two: $\frac{\partial}{\partial R _{i}} ln (1 - ϵ R_{i}) = \frac{- ϵ}{1 - ϵ R _{i}}$ (chain rule brings down $- ϵ$ ), and $\frac{\partial}{\partial R _{i}} (- ln R_{i}) = - \frac{1}{R _{i}}$ .

Putting the pieces together gives the condition, with each term now explained: $\partial f / \partial R_{i} \frac{1}{R _{i}} = λ \partial g / \partial R_{i} (\frac{- ϵ}{1 - ϵ R _{i}} - \frac{1}{R _{i}}) .$

WHY it forces equality. The right side depends only on $R_{i}$ (because $ϵ$ and $λ$ are the same everywhere). So every stage must solve the identical equation — and therefore land on the same answer, which we name $R_{⋆}$ . Unequal ratios cannot all satisfy one identical equation.

PICTURE. Picture two water tanks connected at the bottom: efficiency settles to the same level in both; push it high on one side and it just spills to the other. The plot shows total speed against how you split a fixed budget — a smooth hill whose peak sits dead centre at $R_{1} = R_{2}$ .

WHY the closed form — solve the equal-ratio constraint. Since all $R_{i}$ equal one value $R_{⋆}$ , the $n$ identical factors in $(⋆)$ become a single factor raised to the $n$ -th power: $\frac{m _{p}}{M _{0}} = (\frac{1 - ϵ R _{⋆}}{( 1 - ϵ ) R _{⋆}})^{n} .$ Take the $n$ -th root of both sides and name that number $k$ : $\frac{1 - ϵ R _{⋆}}{( 1 - ϵ ) R _{⋆}} = (\frac{m _{p}}{M _{0}})^{1/ n} \equiv k .$ Now just solve for $R_{⋆}$ like a normal linear equation: multiply out, $1 - ϵ R_{⋆} = k (1 - ϵ) R_{⋆}$ ; gather the $R_{⋆}$ terms, $1 = (ϵ + k (1 - ϵ)) R_{⋆}$ ; divide: $R_{⋆} = \frac{1}{ϵ + ( 1 - ϵ ) k} .$

Step 7 — The degenerate case: no structure ( $ϵ \to 0$ )

WHAT. Set $ϵ = 0$ . The constraint $(⋆)$ becomes simply $\prod R_{i} = M_{0} / m_{p}$ — a fixed product. Now $Δ v = v_{e} ln (\prod R_{i})$ is fixed no matter how you split!

WHY show this. It is the boundary case that reveals the truth: with zero structure, all splits tie (see the parent's numerical comparison — $R_{1} = R_{2} = 4$ and $R_{1} = 8, R_{2} = 2$ both give $8318$ m/s). The advantage of equal ratios is strict only when $ϵ > 0$ , where the curved constraint punishes lopsided splits. Do not mis-remember this as "always ties" — that is the trap.

PICTURE. Two curves: the flat dashed line (ε = 0, every split equal) and the humped solid line (ε > 0, peak strictly in the middle). Same picture, one lesson.

The one-picture summary

Below, the entire argument on one canvas: the block stack (Step 1) → the squeeze $R$ (Step 2) → the $ln$ that turns ratios into speed (Step 3) → stacking (Step 4) → structure coupling them (Step 5) → the balanced peak at equal $R$ (Step 6), with the $ϵ \to 0$ flat case as the shadow line (Step 7).

Recall Feynman retelling — say it back in plain words

A rocket is a stack of blocks: a precious little payload on top, and below it stages made of fuel we burn and shells we throw away. Each stage's power is its squeeze — how many times lighter it gets while burning. The speed you gain isn't the squeeze itself but its logarithm, because speeds add while squeezes multiply, and the log is the machine that turns multiplying into adding. Stack two stages and their speeds add, so their squeezes multiply — two squeezes of $R$ act like one of $R^{2}$ . The twist is dead structure: every stage drags a fixed fraction $ϵ$ of empty shell, and that ties the stages together through a curved constraint. Because all stages share the same $ϵ$ and the same engine, the "best" equation each stage must obey is identical — so they all settle to the same squeeze $R_{⋆}$ , like water finding one level in connected tanks. In the make-believe world with no structure every split ties, but the moment structure is real, the balanced, equal-ratio split sits on top of the hill.

Recall

What quantity does a stage's mass ratio $R$ measure? ::: How many times lighter the stage becomes between ignition and burnout, $m_{before} / m_{after}$ . Why does $ln$ appear when converting mass ratio to speed? ::: Speeds add while mass ratios multiply; $ln$ turns multiplication into addition, so total speed $= v_{e} ln (\prod R_{i})$ . What single fact forces all optimal ratios to be equal? ::: The Lagrange condition for each stage depends only on that stage's $R_{i}$ , and since $ϵ$ is shared, every stage solves the same equation ⇒ same $R_{⋆}$ . In which case do all splits give equal $Δ v$ ? ::: Only the degenerate $ϵ \to 0$ case, where the constraint fixes the product $\prod R_{i}$ .