Altitude compensation methods — nozzle extension, aerospike: Visual walkthrough

This page rebuilds the parent result from absolute zero, in pictures: why a rocket nozzle is only "perfect" at one altitude, and why lengthening it (an extendable nozzle) raises the expansion ratio and squeezes out a little more thrust in vacuum. Every symbol is earned before it is used.

Parent topic: Altitude compensation methods.

Step 1 — What a nozzle is, and the two areas that matter

WHAT. A rocket nozzle is a tube that pinches to a narrow waist and then flares open. Hot gas rushes in from the left, squeezes through the waist, and blasts out the wide mouth.

WHY these two spots. Only two cross-sections control everything:

the throat (the waist) — its cross-sectional area is $A_{t}$ ("t" for throat),
the exit (the mouth) — its cross-sectional area is $A_{e}$ ("e" for exit).

A cross-sectional area is just "how big is the circle you'd see if you sliced the tube straight across there." $A_{t}$ is the small circle at the waist; $A_{e}$ is the big circle at the mouth.

PICTURE. Look at the figure: the gas (orange) flows left-to-right, the throat is the pinch, the exit is the flared mouth. The two shaded discs are $A_{t}$ and $A_{e}$ .

Figure — Altitude compensation methods — nozzle extension, aerospike

Step 2 — Why "spreading out" means "speeding up" and "cooling down"

WHAT. As the gas fills the widening cone, it spreads over a larger area. Because the same amount of gas per second must pass every slice, spreading out lets it accelerate and drop in pressure.

WHY. Think of a crowd leaving a stadium: in a narrow corridor they're packed and slow; when the corridor opens up they thin out and can walk fast. Gas does the same — as area grows, the flow speeds up, and its pressure (the sideways push of the gas on the walls, in pascals, Pa) falls.

Two symbols we now need:

$p_{e}$ — the exit pressure, the leftover push of the gas at the very mouth,
$p_{a}$ — the ambient pressure, the push of the surrounding air/atmosphere outside.

PICTURE. The figure shows three slices along the cone. Left slice: narrow, high pressure (deep magenta), slow. Right slice: wide, low pressure (pale), fast. The arrows lengthen as the cone widens — that's the exhaust speeding up.

Step 3 — The throat is "choked": what Mach 1 means and why mass flow is locked

WHAT. Before we go further we need one word: Mach number, written $M$ . It is simply "how fast the gas moves compared to the speed of sound in that same gas": $M = \frac{gas speed}{local speed of sound} .$

$M < 1$ — subsonic (slower than sound),
$M = 1$ — sonic (exactly the speed of sound; this is "Mach 1"),
$M > 1$ — supersonic (faster than sound).

WHY the throat sits exactly at $M = 1$ . In a converging-then-diverging tube, the flow speeds up through the shrinking part and reaches exactly the speed of sound at the narrowest slice — the throat. It cannot go faster there, and once it's at $M = 1$ the throat is said to be choked: the mass of gas squeezing through per second is pinned at its maximum and cannot increase no matter what you do downstream.

WHY this matters for us. Choking means the mass flow rate — a quantity we will call $\overset{m}{˙}$ (the kilograms of exhaust per second; it appears fully in the thrust equation in Step 4) — is set entirely by the throat conditions ( $A_{t}$ , chamber pressure, temperature). Lengthening the nozzle downstream of the throat does not touch the throat, so:

PICTURE. The figure marks $M = 1$ at the throat (the choke point) with a padlock icon, $M < 1$ upstream, $M > 1$ downstream, and shows that extending the mouth leaves the padlocked throat untouched.

Step 4 — The thrust equation, term by term (and its sign convention)

WHAT. Thrust $F$ is the forward force the nozzle produces. It has exactly two pieces.

WHY these two pieces. Momentum (throwing mass backward fast) plus a pressure mismatch at the mouth.

PICTURE. The figure shows the rocket with a big forward arrow ( $\overset{m}{˙} v_{e}$ ) and a small correction arrow at the lip. When $p_{e} > p_{a}$ the small arrow points forward; when $p_{e} < p_{a}$ it points backward.

The three cases of that little arrow:

Case	Meaning	Pressure arrow
$p_{e} > p_{a}$	Under-expanded (gas still pushing hard)	forward (+)
$p_{e} = p_{a}$	Perfectly expanded	zero
$p_{e} < p_{a}$	Over-expanded (outside air pushes back)	backward (−)

Step 5 — How the shape $ϵ$ actually fixes the exit Mach number and $p_{e}$

WHAT. We keep saying " $p_{e}$ is fixed by the shape." Here is where that becomes literally true. The link between the geometry and the flow speed comes from conservation of mass written for a supersonic flow.

WHY (visual derivation, not a dropped formula). The same $\overset{m}{˙}$ crosses every slice of the nozzle, and mass flow across a slice of area $A$ is $\overset{m}{˙} = ρ A v$ (density $\times$ area $\times$ speed). So along the nozzle $ρ A v$ stays constant. Two facts fight each other as the cone widens:

the gas speeds up (bigger $v$ ), which by itself would need a smaller area,
but the gas thins out and cools dramatically (density $ρ$ crashes) once it is supersonic, and that thinning wins — so the area must grow to keep $ρ A v$ constant.

That tug-of-war — "how much must area grow to let speed rise while density crashes?" — is exactly what the area–Mach relation bottles up into one equation. The messy bracket and exponent are just the isentropic bookkeeping for how $ρ$ and $v$ each depend on $M$ ; the shape of the resulting curve is what matters, and you can read it straight off the figure.

PICTURE. The figure shows $ρ$ , $v$ and their product's balancing act, then plots the resulting curve $ϵ$ vs $M_{e}$ : from the throat ( $M = 1$ , $ϵ = 1$ ) it climbs — every extra bit of Mach number demands a lot more area.

WHY a second relation gives $p_{e}$ . Once $M_{e}$ is known, the isentropic pressure relation converts it into the exit pressure:

Step 6 — Why one nozzle can't be perfect everywhere

WHAT. From Step 5, $p_{e}$ is fixed by the shape $ϵ$ . But $p_{a}$ changes as the rocket climbs — about 101 kPa at sea level, falling toward 0 in space.

WHY it's a conflict. "Perfect" means $p_{e} = p_{a}$ . But $p_{e}$ is a fixed number for a given cone, while $p_{a}$ slides from 101 kPa down to 0. A fixed line can only cross a sliding target once — at exactly one altitude.

PICTURE. The figure plots $p_{a}$ falling with altitude (curved orange line) and a flat dashed $p_{e}$ line for a fixed nozzle. They meet at one point (the perfect altitude). Below it: $p_{e} < p_{a}$ , over-expanded (air crushes in). Above it: $p_{e} > p_{a}$ , under-expanded (wasted pressure).

Step 7 — Growing $ϵ$ geometrically: the extension is just more cone

WHAT. A conical nozzle grows in radius as you walk along the centre axis. Let $x$ be the axial distance measured straight along the centre line (not along the slanted wall). If the throat radius is $R_{t}$ and the cone wall tilts away from the axis by a fixed half-angle $α$ , then after an axial distance $L$ the radius is

$R_{e} = R_{t} + L tan α .$

WHY $tan α$ , and which length is $L$ . Build a right triangle whose horizontal leg lies along the axis (this is the axial length $L$ , adjacent to $α$ ) and whose vertical leg is the extra radius the wall has climbed (opposite $α$ ). The slanted wall itself is the hypotenuse — we do not use its length. "Vertical gained per axial distance travelled" is opposite over adjacent $= tan α$ , so the extra radius is $L tan α$ . This is why $tan$ and not $sin$ or $cos$ : we want radius-gain per axial length, and that ratio is the tangent. (If you instead measured along the slanted wall you'd use $sin α$ — we deliberately use axial $L$ .)

PICTURE. The figure draws the throat radius $R_{t}$ , an axial length $L$ along the centre line, and shades the right triangle: horizontal leg $= L$ on the axis, vertical leg $= L tan α$ , hypotenuse $=$ the cone wall. Extending the nozzle means a bigger axial $L$ , which grows $R_{e}$ , which grows $A_{e}$ , which grows $ϵ$ .

Now turn radius into $ϵ$ . Since area of a circle is $π R^{2}$ : $ϵ = \frac{A _{e}}{A _{t}} = \frac{π R _{e}^{2}}{π R _{t}^{2}} = (\frac{R _{e}}{R _{t}})^{2} = (1 + \frac{L}{R _{t}} tan α)^{2} .$

The $π$ 's cancel — only the radius ratio survives.
Squaring it: doubling the radius quadruples the area. So a modest length increase can raise $ϵ$ a lot.

Step 8 — Lower $p_{e}$ , faster gas: reading the cash-out

WHAT. A bigger $ϵ$ drives $p_{e}$ lower (Step 5), which lets more of the chamber's stored pressure become exhaust speed $v_{e}$ .

WHY (energy-conservation picture, built before the formula). Picture the chamber as a hot, still reservoir holding a fixed budget of energy per kilogram of gas: its heat energy, measured by temperature $T_{0}$ . As that kilogram rushes down the nozzle it trades heat for motion — it cools from $T_{0}$ to a lower exit temperature $T_{e}$ , and every joule of heat it loses reappears as kinetic energy $\frac{1}{2} v_{e}^{2}$ . That is the whole story: $motion gained \frac{1}{2} v_{e}^{2} = heat given up c_{p} (T_{0} - T_{e})$

$c_{p}$ — how many joules it takes to warm one kilogram by one degree (a property of the gas).
$T_{0} - T_{e}$ — the temperature drop; the bigger the drop, the more heat is cashed into speed.

The exit temperature $T_{e}$ is itself set by how far the pressure fell (cooler gas ↔ lower pressure), and threading that in gives the boxed result. Follow the flowchart in the figure: heat budget $T_{0}$ → cool to $T_{e}$ (fixed by $p_{e} / p_{0}$ ) → speed $v_{e}$ .

WHY it saturates (the key picture). As you drop $p_{e}$ toward zero, the bracket flattens out near 1. So the first halving of $p_{e}$ buys a lot of speed; each later halving buys less. That's why doubling $ϵ$ gives diminishing thrust gains — there is no clean $v_{e} \propto ϵ$ law. And remember $\overset{m}{˙}$ is frozen (Step 3), so the entire thrust change comes through $v_{e}$ and the shrinking pressure term.

The right-hand panel of the figure plots $v_{e}$ against $ϵ$ : a fast rise that bends over into a nearly flat ceiling. The stowed and deployed points sit on the flattening part — a real gain, but small.

Step 9 — The edge and degenerate cases (never leave the reader stranded)

WHAT / WHY / PICTURE, all four corners of the map "big or small $ϵ$ " × "sea level or vacuum", plus the two true limits:

Small $ϵ$ (near 1) at sea level. Exit ≈ throat, so gas leaves near Mach 1 with high $p_{e}$ . Here $p_{e}$ can actually exceed sea-level $p_{a}$ or roughly match it, so no violent over-expansion — but you've barely accelerated the gas, so thrust and $I_{s p}$ are poor. It works, it just wastes energy. This is a stubby booster-style nozzle.
Small $ϵ$ (near 1) in vacuum. $p_{a} \approx 0$ , so the pressure term $+ p_{e} A_{e}$ is forward, but $ϵ$ is so small that huge pressure energy is left unconverted — the gas exits still hot and high-pressure. Severely under-expanded: you throw away 20–30% of possible thrust. This is exactly why upper stages want big $ϵ$ .
Big $ϵ$ at sea level (over-expanded). $p_{e} ≪ p_{a}$ , the pressure arrow points backward and can separate the flow from the wall — the exhaust peels off, causing side-loads that can tear the nozzle. This is why you can't just bolt on a giant vacuum nozzle at launch.
Big $ϵ$ in vacuum (the sweet spot). $p_{a} = 0$ so the pressure term is always forward, and thrust is maximised by making $p_{e}$ tiny — i.e. large $ϵ$ . This is the extendable nozzle's home turf.

Two limiting behaviours that bracket everything:

$ϵ \to 1$ exactly: the flare vanishes, gas exits at exactly Mach 1, almost no pressure converted — a plain sonic hole, useless as a real nozzle.
$p_{e} \to 0$ exactly (infinite nozzle): the energy bracket hits 1, $v_{e}$ reaches its absolute ceiling, and the pressure term vanishes. Infinitely long, infinitely heavy — impossible, but it's the limit everything approaches.

PICTURE. Four mini-panels arranged as the 2×2 map: (1) small $ϵ$ / sea level — stubby, near Mach 1, matched-ish; (2) small $ϵ$ / vacuum — under-expanded, wasted pressure; (3) big $ϵ$ / sea level — flow peeling off the wall, red backward arrow; (4) big $ϵ$ / vacuum — all-forward arrows, the sweet spot.

The one-picture summary

This final figure stitches the whole chain together: more axial length → bigger $R_{e}$ (via $tan α$ ) → bigger $ϵ = (A_{e} / A_{t})$ → bigger $M_{e}$ → lower $p_{e}$ → the energy bracket climbs → $v_{e}$ rises (but saturates) → thrust nudges up in vacuum, with $\overset{m}{˙}$ locked by the choked throat throughout, all while the fixed-nozzle "one perfect altitude" curve reminds us why we bothered.

Recall Feynman retelling in plain words

A nozzle is a tube that pinches then flares. At the pinch the gas hits exactly the speed of sound — "Mach 1" — and gets choked, which locks the kilograms-per-second forever; nothing you do downstream can change it. The flare's job is to let the now-supersonic gas spread out, which speeds it up and drops its leftover pressure. How much? Conservation of mass says density × area × speed is the same at every slice; once supersonic the gas thins out faster than it speeds up, so area must grow — and the area–Mach curve packages exactly that. The nozzle's shape number $ϵ$ (mouth area over throat area) feeds that curve and spits out exactly one exit Mach number, which fixes the exit pressure — so the shape alone decides $p_{e}$ . Thrust is "mass thrown back per second times its speed" (always forward, our positive direction), plus a small pressure correction that is the only term whose sign can flip. Perfect means exit and outside pressures match — but the outside pressure keeps falling as you climb while a fixed nozzle's exit pressure stays put, so it's only perfect at one height. To do better, grow the nozzle: walking further along the axis at wall angle $α$ adds radius $L tan α$ (radius gained per axial length = tangent), and area grows as radius squared, so a little more axial length gives a lot more $ϵ$ . More $ϵ$ means higher exit Mach, lower exit pressure, and faster gas — but the speed comes from an energy budget (heat traded for motion) that saturates, and $\overset{m}{˙}$ is frozen, so each doubling helps less. In vacuum the pressure correction is always forward, so bigger is better there — which is exactly where extendable nozzles fire. You can't just use a giant nozzle at sea level, because there it over-expands, the outside air shoves the exhaust off the walls, and it can rip apart.

Recall Quick self-test

What does "Mach 1" mean, and where in the nozzle does it occur? ::: Gas speed equal to the local speed of sound; it occurs at the throat, which is choked. Why is $\overset{m}{˙}$ constant when you lengthen the nozzle? ::: The throat stays choked at Mach 1, and lengthening only changes the diverging section downstream, so throat conditions (and thus $\overset{m}{˙}$ ) are untouched. How does the shape $ϵ$ fix $p_{e}$ ? ::: $ϵ$ feeds the area–Mach relation → one exit Mach $M_{e}$ ; then the isentropic pressure relation turns $M_{e}$ into $p_{e}$ . What is the sign convention for $F$ and $v_{e}$ ? ::: Positive $F$ is forward (rocket's acceleration direction); $v_{e}$ is a positive speed; only the pressure term $(p_{e} - p_{a})$ can change sign. What does $tan α$ physically give, and is $L$ axial or along the wall? ::: Radius gained per unit axial length; $L$ is measured along the centre axis, not the slanted wall. Why is there exactly one perfect altitude for a fixed nozzle? ::: $p_{e}$ is fixed by shape but $p_{a}$ slides from 101 kPa to 0, so $p_{e} = p_{a}$ happens at only one height. Why doesn't doubling $ϵ$ double thrust? ::: $v_{e}$ depends on an energy bracket that saturates toward 1 as $p_{e} \to 0$ , and $\overset{m}{˙}$ is frozen, so gains diminish.