Visual walkthrough — Gas generator cycle — performance penalty vs simplicity

Everything here supports the parent result: $I_{sp,\text{eff}}=\frac{(1-f)v_{e,c}+f{v}_{e,gg}}{g_0}$ We will earn every symbol in it.

---

Step 0 — What are we even measuring?

Before symbols, a picture of the thing. A rocket makes thrust by throwing mass backwards fast. Two questions matter:

1. How fast do we throw each kilogram out the back? → call that the exhaust velocity $v_e$ (units: metres per second, m/s). Bigger $v_e$ = more push per kilogram.
2. How much mass do we throw per second? → the mass flow rate $\dot{m}$ (the little dot means "per second"; units: kilograms per second, kg/s).

PICTURE: the engine on the left, one arrow of propellant leaving to the right at speed $v_e$. The faster and heavier that arrow, the harder the rocket is pushed forward (equal-and-opposite). This is the whole game — see Rocket Thrust Equation.

---

Step 1 — Split the propellant into TWO rivers

WHAT. The gas generator engine does not send all propellant to one place. It splits the flow.

WHY. To spin the pumps you need a turbine, and a turbine needs hot gas. The GG cycle makes that hot gas by burning a little propellant in a side-chamber. So the total river forks into two:

${\dot{m}}_{tot}=\underset{\text{main chamber}}{\underbrace{{\dot{m}}_c}}+\underset{\text{gas generator}}{\underbrace{{\dot{m}}_{gg}}}$

• ${\dot{m}}_{tot}$ — every kilogram per second leaving the tanks. You paid for all of it.
• ${\dot{m}}_c$ — the big river, goes to the main chamber, burns hot, expands through the big nozzle.
• ${\dot{m}}_{gg}$ — the small river, goes to the gas generator, spins the turbine, then gets dumped.

PICTURE: one thick lavender river entering, forking into a fat mint stream (to the chamber) and a thin coral stream (to the turbine). Watch the coral stream: it will be the villain.

---

Step 2 — The fraction $f$: naming the thin river

WHAT. Give the thin river a name relative to the whole: what fraction of the total flow goes to the turbine?

WHY. Because the penalty will turn out to depend only on this fraction, not on the raw numbers. Fractions travel better between engines.

Two immediate consequences, straight from algebra: ${\dot{m}}_{gg}=f{\dot{m}}_{tot}\text{and}{\dot{m}}_c=(1-f){\dot{m}}_{tot}$

• $f$ — the thin (dumped) fraction.
• $1-f$ — the fat (useful) fraction; whatever isn't dumped stays in the main chamber.

PICTURE: a single bar of length 1 chopped into a big $(1-f)$ mint block and a small $f$ coral sliver. That coral sliver is exactly the propellant we are about to waste.

---

Step 3 — WHY the dumped gas is nearly useless

WHAT. Show that the thin river leaves the vehicle slowly — its $v_{e,gg}$ is far below the main $v_{e,c}$.

WHY. Exhaust velocity comes from expanding gas across a big pressure drop. The main-chamber gas starts at high chamber pressure and expands to near-vacuum → huge drop → fast. The turbine exhaust has already given its pressure to the turbine, so it exits the turbine at low pressure and is simply dumped → tiny remaining drop → slow.

Recall from Nozzle Expansion and Pressure Ratio the exhaust-velocity law: $v_e=\sqrt{\frac{2\gamma }{\gamma -1}R{T}_0\left[1-{\left(\frac{p_e}{p_0}\right)}^{\frac{\gamma -1}{\gamma }}\right]}$

• $T_0$ — the gas temperature before expanding.
• $p_0$ — pressure before expanding; $p_e$ — pressure after (at the exit).
• $\gamma$ (gamma) — how "springy" the gas is (ratio of specific heats, $\approx 1.2$ for hot exhaust).
• The bracket is the star: if $p_e$ is almost as big as $p_0$, the ratio $p_e/p_0\to 1$, the bracket $\to 0$, and $v_e\to 0$.

For the dumped gas $p_0$ is already low (turbine took the pressure), so $p_e/p_0$ is near 1 → bracket tiny → $v_{e,gg}$ small. For the main chamber $p_0=p_c$ is huge and $p_e\approx 0$ → bracket near 1 → $v_{e,c}$ large.

PICTURE: two "expansion staircases." The mint staircase (main chamber) falls a long way from high $p_0$ to near-vacuum → long arrow = fast. The coral staircase (turbine dump) barely steps down → stubby arrow = slow. Same bracket, wildly different heights.

---

Step 4 — Momentum is additive: build the total thrust

WHAT. The thrust of the whole engine is the sum of the momentum each river throws out per second.

WHY. Momentum simply adds — two exhaust streams push independently, so their pushes stack. (Ignoring pressure-area terms, which we bundle into $v_e$ as an effective exhaust velocity.)

$F_{tot}=\underset{\text{fast river’s push}}{\underbrace{{\dot{m}}_c{v}_{e,c}}}+\underset{\text{slow river’s push}}{\underbrace{{\dot{m}}_{gg}{v}_{e,gg}}}$

• Each term is (kg/s) × (m/s) = kg·m/s² = newtons of thrust.
• The coral term is small on both factors: small flow and small speed.

PICTURE: the two arrows of Step 1, now with their momentum shown as arrow area. The mint arrow is a big rectangle; the coral arrow is a thin short sliver. Total push = both rectangles glued nose-to-tail.

---

Step 5 — Divide by what you paid for → effective $I_{sp}$

WHAT. Turn total thrust into an efficiency number by dividing by the total weight-flow — because you carried and burned all the propellant, dumped or not.

WHY. $I_{sp}$ is "push per unit weight of propellant spent per second." The dumped fraction still counts as propellant spent, so it belongs in the denominator even though it barely helps the numerator. That mismatch is the penalty.

$I_{sp,\text{eff}}=\frac{F_{tot}}{{\dot{m}}_{tot}{g}_0}=\frac{{\dot{m}}_c{v}_{e,c}+{\dot{m}}_{gg}{v}_{e,gg}}{({\dot{m}}_c+{\dot{m}}_{gg})g_0}$

Now substitute Step 2 (${\dot{m}}_c=(1-f){\dot{m}}_{tot}$, ${\dot{m}}_{gg}=f{\dot{m}}_{tot}$) and cancel ${\dot{m}}_{tot}$ top and bottom:

PICTURE: the total-momentum bar from Step 5 divided by the full-length "you-paid-for-it" bar. The slice you lose is the gap between "if the thin river were as fast as the fat one" and "how slow it actually is."

---

Step 6 — The clean penalty: worst case $v_{e,gg}\to 0$

WHAT. Look at the limiting case where the dumped gas leaves with essentially zero useful speed.

WHY. It gives the memorable rule-of-thumb and the upper bound on the penalty.

Set $v_{e,gg}=0$: $I_{sp,\text{eff}}\approx (1-f)\frac{v_{e,c}}{g_0}=(1-f)I_{sp,\text{ideal}}$

So the penalty is directly the fraction $f$. Dump 4% → lose about 4% of $I_{sp}$.

Edge / degenerate cases — every corner covered:

• $f=0$ → no turbine flow → $I_{sp,\text{eff}}=I_{sp,\text{ideal}}$. (Impossible in practice — pumps need some gas — but the math is continuous and correct.)
• $f=1$ → all propellant dumped through the turbine → $I_{sp,\text{eff}}=v_{e,gg}/g_0$, nearly zero thrust. A rocket that only spins its pumps. Useless, but the formula still holds.
• $v_{e,gg}=v_{e,c}$ (thin river as fast as the fat one) → penalty vanishes; this is the closed-cycle limit, where dumped gas is instead fed back — see Staged Combustion Cycle and Expander Cycle.
• Higher chamber pressure → more pump work → larger $f$ (from the Turbopump Fundamentals power balance) → bigger penalty. GG cycles therefore fade at extreme $p_c$.

PICTURE: a straight line of $I_{sp,\text{eff}}$ versus $f$, sliding down from $I_{sp,\text{ideal}}$ at $f=0$ to $\sim 0$ at $f=1$. The real operating dot sits at $f\approx 0.04$, just a small step down — with the "true" curve (slow-but-nonzero dump) sitting slightly above the worst-case line.

---

The one-picture summary

The whole story on one canvas: propellant river forks (Step 1–2), the two rivers expand across very different pressure drops (Step 3), throw very different momenta (Step 4–5), and dividing by all the propellant you paid for reveals the $f$-sized tax (Step 6).

Recall Feynman retelling — say it back in plain words

A rocket pushes by throwing stuff out the back fast. To pump the fuel we burn a tiny bit of it in a side-pot to spin a turbine. That tiny bit does its job spinning the turbine, but by then it has no pressure left, so when we dump it, it barely dribbles out — almost no push. Trouble is, we still carried and burned it, so it counts against our fuel bill. Efficiency = total push ÷ total fuel burned. The dumped fraction $f$ adds to the "fuel burned" bottom but almost nothing to the "push" top, so efficiency drops by roughly $f$. Dump 4%, lose about 4%. In return we skip all the nightmare high-pressure plumbing of a closed cycle. That's the deal: a few percent of performance for a whole lot of simplicity.

Recall

What does the fraction $f$ represent, and why is it in the denominator of $I_{sp,\text{eff}}$? ::: $f={\dot{m}}_{gg}/{\dot{m}}_{tot}$ is the propellant fraction sent to the turbine and then dumped; it's in the denominator because you spent that propellant, even though it contributes almost no thrust to the numerator. Why is the dumped gas so slow ($v_{e,gg}\ll v_{e,c}$)? ::: It exits the turbine at low pressure, so its remaining pressure ratio $p_e/p_0$ is near 1, making the expansion bracket tiny and $v_e$ small. Worst-case penalty rule of thumb? ::: $I_{sp,\text{eff}}\approx (1-f)I_{sp,\text{ideal}}$ — lose about the fraction $f$. Why is the true penalty a bit smaller than $f$? ::: The dumped gas still leaves with some velocity $v_{e,gg}>0$, adding a small $f{v}_{e,gg}$ kick to the numerator.