Intuition The big pictureA rocket engine burns propellant at very high chamber pressure (say 100–250 bar). To push propellant into that chamber, the pump outlet pressure must be even higher. Feeding this from tank pressure alone would need dangerously heavy, thick tanks. Instead we use a turbopump : a tiny turbine (spun by hot gas) drives a pump on the same shaft that raises propellant pressure by 100× or more.
WHY it exists: trade a little propellant burned in a gas generator for the ability to run lightweight, low-pressure tanks while still feeding a high-pressure chamber.
The three design pillars are: (1) the centrifugal pump that adds energy to the liquid, (2) the axial turbine stages that extract power from hot gas to spin it, and (3) the NPSH constraint that stops the pump inlet from boiling (cavitating).
Definition Centrifugal pumpA rotating impeller flings liquid outward. The liquid enters near the axis (the eye ) at radius r 1 r_1 r 1 r 2 r_2 r 2 pressure in the surrounding volute/diffuser.
WHAT we want: the specific energy (energy per unit mass) added to the fluid.
HOW — angular momentum bookkeeping. Consider fluid mass flow m ˙ \dot m m ˙ T T T
T = m ˙ ( r 2 c u 2 − r 1 c u 1 ) T = \dot m\,(r_2 c_{u2} - r_1 c_{u1}) T = m ˙ ( r 2 c u 2 − r 1 c u 1 )
where c u c_{u} c u tangential (whirl) component of the absolute fluid velocity. Why tangential only? Only tangential velocity carries angular momentum about the axis.
Power delivered by a shaft spinning at angular speed ω \omega ω P = T ω P = T\omega P = T ω
P = m ˙ ω ( r 2 c u 2 − r 1 c u 1 ) = m ˙ ( u 2 c u 2 − u 1 c u 1 ) P = \dot m\,\omega(r_2 c_{u2} - r_1 c_{u1}) = \dot m\,(u_2 c_{u2} - u_1 c_{u1}) P = m ˙ ω ( r 2 c u 2 − r 1 c u 1 ) = m ˙ ( u 2 c u 2 − u 1 c u 1 )
using blade speed u = ω r u = \omega r u = ω r Why this step? ω r 2 = u 2 \omega r_2 = u_2 ω r 2 = u 2
Energy added per unit mass is P / m ˙ P/\dot m P / m ˙ g g g head H H H
Why c u 1 = 0 c_{u1}=0 c u 1 = 0 Feeding liquid straight in (radial/axial) means no pre-rotation, so all the whirl energy is added by this impeller — simpler and maximizes head for given u 2 u_2 u 2
Intuition Read the formulaHead ↑ \uparrow ↑ u 2 u_2 u 2 c u 2 c_{u2} c u 2 c u 2 ≲ u 2 c_{u2}\lesssim u_2 c u 2 ≲ u 2 H ∼ u 2 2 / g H \sim u_2^2/g H ∼ u 2 2 / g doubling rim speed quadruples pressure rise. That's why turbopump impellers scream at tens of thousands of RPM.
Δ p = ρ g H ⟹ Δ p ≈ ρ u 2 c u 2 \Delta p = \rho g H \implies \Delta p \approx \rho\,u_2 c_{u2} Δ p = ρ g H ⟹ Δ p ≈ ρ u 2 c u 2
Why? Head is energy/weight; multiply by ρ g \rho g ρ g
Worked example Worked: LOX pump headRim radius r 2 = 0.10 m r_2=0.10\text{ m} r 2 = 0.10 m N = 30000 rpm N=30000\text{ rpm} N = 30000 rpm c u 2 = 0.8 u 2 c_{u2}=0.8\,u_2 c u 2 = 0.8 u 2 ρ L O X = 1140 kg/m 3 \rho_{LOX}=1140\text{ kg/m}^3 ρ L O X = 1140 kg/m 3
ω = 2 π N / 60 = 2 π ( 500 ) = 3142 rad/s \omega = 2\pi N/60 = 2\pi(500) = 3142\text{ rad/s} ω = 2 π N /60 = 2 π ( 500 ) = 3142 rad/s Why? rpm→rad/s.u 2 = ω r 2 = 3142 × 0.10 = 314 m/s u_2=\omega r_2 = 3142\times0.10 = 314\text{ m/s} u 2 = ω r 2 = 3142 × 0.10 = 314 m/s Why? blade speed at rim.H = u 2 c u 2 / g = ( 314 ) ( 0.8 × 314 ) / 9.81 = 8040 m H = u_2 c_{u2}/g = (314)(0.8\times314)/9.81 = 8040\text{ m} H = u 2 c u 2 / g = ( 314 ) ( 0.8 × 314 ) /9.81 = 8040 m Why? Euler head, no inlet swirl.Δ p = ρ g H = 1140 × 9.81 × 8040 ≈ 9.0 × 10 7 Pa = 90 bar \Delta p = \rho g H = 1140\times9.81\times8040 \approx 9.0\times10^7\text{ Pa} = 90\text{ bar} Δ p = ρ g H = 1140 × 9.81 × 8040 ≈ 9.0 × 1 0 7 Pa = 90 bar
Definition Axial turbine stageHot gas (from a gas generator or preburner) flows along the axis . Each stage = one row of stationary nozzles/stators (accelerate & turn the gas) + one row of moving rotor blades (extract work). Flow stays roughly at the same radius, hence axial .
WHY axial (not centrifugal) for the turbine? The turbine handles low-density, high-velocity gas ; axial rows stack in series easily to extract lots of energy in a compact, high-power-density machine.
Same angular-momentum argument, but now the fluid gives up whirl to the rotor:
w turb = u ( c u , in − c u , out ) = u Δ c u w_{\text{turb}} = u\,(c_{u,\text{in}} - c_{u,\text{out}}) = u\,\Delta c_u w turb = u ( c u , in − c u , out ) = u Δ c u
Why the sign flip vs. pump? The pump adds whirl (work in); the turbine removes whirl (work out). It is the same u Δ c u u\,\Delta c_u u Δ c u
The shaft is shared: P turb = P pump / η mech P_{\text{turb}} = P_{\text{pump}}/\eta_{\text{mech}} P turb = P pump / η mech couples the two: the turbine must produce exactly the power the pump demands.
P pump = m ˙ prop g H η pump = m ˙ prop Δ p ρ η pump P_{\text{pump}} = \dfrac{\dot m_{\text{prop}}\,g\,H}{\eta_{\text{pump}}} = \dfrac{\dot m_{\text{prop}}\,\Delta p}{\rho\,\eta_{\text{pump}}} P pump = η pump m ˙ prop g H = ρ η pump m ˙ prop Δ p
Worked example Power matchPump must deliver m ˙ = 200 kg/s \dot m = 200\text{ kg/s} m ˙ = 200 kg/s Δ p = 200 bar = 2 × 10 7 Pa \Delta p = 200\text{ bar}=2\times10^7\text{ Pa} Δ p = 200 bar = 2 × 1 0 7 Pa ρ = 1000 \rho=1000 ρ = 1000 η p = 0.7 \eta_p=0.7 η p = 0.7 P pump = 200 × 2 × 10 7 1000 × 0.7 = 5.7 × 10 6 W = 5.7 MW P_{\text{pump}} = \dfrac{200\times2\times10^7}{1000\times0.7} = 5.7\times10^6\text{ W} = 5.7\text{ MW} P pump = 1000 × 0.7 200 × 2 × 1 0 7 = 5.7 × 1 0 6 W = 5.7 MW m ˙ g ≈ 5 kg/s \dot m_g\approx 5\text{ kg/s} m ˙ g ≈ 5 kg/s c p ≈ 2000 c_p\approx2000 c p ≈ 2000 T 0 = 900 K T_0=900\text{ K} T 0 = 900 K ≥ 5.7 \ge 5.7 ≥ 5.7 5 × 2000 × 900 × 0.6 × 0.5 ≈ 2.7 5\times2000\times900\times0.6\times0.5 \approx 2.7 5 × 2000 × 900 × 0.6 × 0.5 ≈ 2.7 Why this matters: the power match sizes the gas generator.
Intuition The cavitation problemAt the impeller eye , fast-moving liquid has low static pressure (Bernoulli). If that local pressure drops below the propellant's vapor pressure p v p_v p v boils into bubbles — cavitation . Bubbles collapse on the blades → erosion, noise, and collapsed head. LOX and LH₂ are near-boiling already, so this is a make-or-break constraint.
Definition NPSH (Net Positive Suction Head)The margin of available head at the pump inlet above the vapor pressure, expressed as a liquid column height:
NPSH A = p inlet − p v ρ g \text{NPSH}_A = \dfrac{p_{\text{inlet}} - p_v}{\rho g} NPSH A = ρ g p inlet − p v required NPSHR _R R Safe operation needs NPSH A > NPSH R \text{NPSH}_A > \text{NPSH}_R NPSH A > NPSH R
Apply energy conservation (Bernoulli + losses) from tank surface to pump inlet:
p tank ρ g ⏟ tank press head + z ⏟ static height − h loss ⏟ friction = p inlet ρ g + v 2 2 g \underbrace{\dfrac{p_{\text{tank}}}{\rho g}}_{\text{tank press head}} + \underbrace{z}_{\text{static height}} - \underbrace{h_{\text{loss}}}_{\text{friction}} = \dfrac{p_{\text{inlet}}}{\rho g} + \dfrac{v^2}{2g} tank press head ρ g p tank + static height z − friction h loss = ρ g p inlet + 2 g v 2
Rearrange for the total (stagnation) head available at inlet above vapor pressure:
Worked example Worked: LH₂ suction checkTank pressurized to p tank = 3 bar = 3 × 10 5 p_{\text{tank}}=3\text{ bar}=3\times10^5 p tank = 3 bar = 3 × 1 0 5 p v = 1 × 10 5 Pa p_v=1\times10^5\text{ Pa} p v = 1 × 1 0 5 Pa ρ L H 2 = 71 kg/m 3 \rho_{LH_2}=71\text{ kg/m}^3 ρ L H 2 = 71 kg/m 3 z = 2 m z=2\text{ m} z = 2 m h loss = 1.5 m h_{\text{loss}}=1.5\text{ m} h loss = 1.5 m NPSH A = ( 3 − 1 ) × 10 5 71 × 9.81 + 2 − 1.5 = 287 + 0.5 = 287.5 m \text{NPSH}_A = \dfrac{(3-1)\times10^5}{71\times9.81} + 2 - 1.5 = 287 + 0.5 = 287.5\text{ m} NPSH A = 71 × 9.81 ( 3 − 1 ) × 1 0 5 + 2 − 1.5 = 287 + 0.5 = 287.5 m Why so huge? Because ρ \rho ρ R _R R This is why LH₂ pumps often add an inducer.
Intuition The inducer trickAn inducer is a slender axial screw ahead of the main impeller. It gently pre-pressurizes flow, lowering NPSHR _R R , so the pump tolerates low inlet pressure and the tank can stay light. It is the standard fix when NPSHA _A A
Common mistake "Head depends on the liquid's density."Why it feels right: heavier liquid seems harder to pump. The truth: Euler head H = u 2 c u 2 / g H=u_2c_{u2}/g H = u 2 c u 2 / g no ρ \rho ρ energy per unit weight . The pressure rise Δ p = ρ g H \Delta p=\rho g H Δ p = ρ g H does scale with ρ \rho ρ head (in metres) for LOX and water, but very different Δp. Fix: separate "head" (geometry/speed) from "pressure" (add density).
Common mistake "NPSH is about the pump outlet pressure."Why it feels right: we care about delivering high pressure. Truth: NPSH is entirely an inlet/suction quantity — it's about not boiling before the impeller does its job. Outlet pressure is irrelevant to cavitation onset. Fix: NPSH = inlet margin above vapor pressure.
Common mistake "Turbine and pump can be designed independently."Why it feels right: they're different machines. Truth: they share one shaft, so P turb = P pump P_{\text{turb}}=P_{\text{pump}} P turb = P pump ω \omega ω constrains the other. Fix: always close the power balance.
Common mistake "Faster shaft is always better for head."Why it feels right: H ∼ u 2 2 H\sim u_2^2 H ∼ u 2 2 Truth: higher speed raises inlet velocity → lowers inlet static pressure → raises NPSHR _R R → risks cavitation. Speed is capped by suction, not head. Fix: the NPSH constraint sets max safe RPM (via suction specific speed N s s N_{ss} N ss
Recall Feynman: explain to a 12-year-oldImagine you want to squirt water into a bottle that already has a LOT of pressure inside. A normal squirt bottle can't win. So you use a spinning fan (impeller) that flings the water super fast around a circle — spinning it faster makes the water "heavier-hitting," so it slams into the exit with big pressure. To spin that fan you need a motor, so you blow hot gas through a little windmill (turbine) on the same stick — the gas pushes the windmill, the windmill spins the fan. One catch: if you suck the water in too hard, it starts to boil into bubbles (like a shaken soda), which wrecks the fan. NPSH is just the "how much room before it boils" safety number — keep it positive!
Mnemonic Remember the three pillars"PUMP TURns, Never Boil"
P ump: H = u 2 c u 2 / g H = u_2 c_{u2}/g H = u 2 c u 2 / g TUR bine: w = u Δ c u w = u\,\Delta c_u w = u Δ c u same equation, opposite sign N PSH: p tank − p v ρ g + z − h loss \dfrac{p_{\text{tank}}-p_v}{\rho g}+z-h_{\text{loss}} ρ g p tank − p v + z − h loss > NPSHR _R R N ever B oils.
Why does a rocket need a turbopump instead of pressure-fed tanks? To feed a very high chamber pressure while keeping tanks light and low-pressure; the turbine trades a little burned gas for huge pump head.
State the Euler pump head with no inlet swirl. H = u 2 c u 2 / g H = u_2 c_{u2}/g H = u 2 c u 2 / g , where
u 2 = ω r 2 u_2=\omega r_2 u 2 = ω r 2 is rim speed and
c u 2 c_{u2} c u 2 is exit whirl velocity.
Why does head not depend on density but pressure rise does? Head is energy per unit
weight (
H = u 2 c u 2 / g H=u_2c_{u2}/g H = u 2 c u 2 / g ); pressure is energy per unit
volume (
Δ p = ρ g H \Delta p=\rho g H Δ p = ρ g H ), so multiplying by
ρ \rho ρ reintroduces density.
Roughly how does head scale with rim speed? H ∼ u 2 2 / g H\sim u_2^2/g H ∼ u 2 2 / g , so doubling shaft speed quadruples head.
Write the turbine stage work per unit mass. w = u Δ c u w = u\,\Delta c_u w = u Δ c u (blade speed × change in whirl), the same Euler relation as the pump but extracting energy.
Why are pump and turbine coupled? Same shaft: common
ω \omega ω and
P turb = P pump / η m e c h P_{\text{turb}}=P_{\text{pump}}/\eta_{mech} P turb = P pump / η m ec h , so sizing one constrains the other.
Define NPSH_A. NPSH A = p t a n k − p v ρ g + z − h l o s s \text{NPSH}_A=\dfrac{p_{tank}-p_v}{\rho g}+z-h_{loss} NPSH A = ρ g p t ank − p v + z − h l oss : inlet head margin above vapor pressure.
What is the safe-operation NPSH condition? NPSH A > NPSH R \text{NPSH}_A > \text{NPSH}_R NPSH A > NPSH R ; otherwise inlet pressure falls below
p v p_v p v and the pump cavitates.
What does an inducer do? A slender axial screw ahead of the impeller that pre-pressurizes flow, lowering NPSH_R so the tank can stay lighter.
What ultimately limits maximum shaft speed? Suction/cavitation (NPSH_R rises with speed), not head production.
Why does LH₂ give huge NPSH values in metres? Its density is tiny (~71 kg/m³), so any pressure margin converts to a very tall equivalent liquid column.
torque T equals m_dot times whirl
Lightweight low-pressure tanks
Intuition Hinglish mein samjhoDekho, rocket ke chamber ke andar pressure bahut zyada hota hai (100-250 bar). Agar hum sirf tank pressure se propellant push karein to tank itne mote aur bhaari banane padenge ki rocket hi udega nahi. Isliye turbopump use karte hain: ek chhota turbine hot gas se ghoomta hai, aur wahi shaft ek centrifugal pump ko spin karta hai jo liquid ka pressure 100 guna tak badha deta hai. Thoda sa propellant gas banane mein "kharch" hota hai, lekin badle mein tank halka rehta hai — yeh ek smart trade-off hai.
Pump ka core formula Euler se aata hai: H = u 2 c u 2 / g H = u_2 c_{u2}/g H = u 2 c u 2 / g u 2 u_2 u 2
