Pogo Oscillation: The Feedback Loop That Shakes a Rocket Apart

The closed loop: how a rocket drives its own vibration

Pogo is a self-excited (autonomous) oscillation: no external forcing is needed, only a feedback loop with net positive gain around a full cycle. Trace the loop in four links:

• Structure → engine: The vehicle is a slender beam. Its lowest longitudinal (axial) modes have the engine and turbopumps at the aft node region. When the structure oscillates axially, the pumps and feedlines physically accelerate with amplitude a(t), producing an inertial pressure fluctuation Δp ≈ ρ·L·a in the propellant column (ρ = density, L = feedline length).
• Engine → feedline: That pressure perturbation reaches the pump inlet and is amplified by pump dynamics and any cavitation compliance at the inducer, modulating the mass flow ṁ delivered to the chamber.
• Feedline → thrust: Chamber pressure p_c, and therefore thrust F ≈ C_F·A_t·p_c, follows the flow perturbation with a transport lag τ (the residence + combustion delay).
• Thrust → structure: The oscillating thrust re-excites the very structural mode that started the chain.

If the round-trip phase brings the thrust perturbation back roughly in phase with structural velocity, the loop pumps energy in on every cycle. The classic stability condition is that the open-loop transfer function G(jω) — the product of structural mobility, feed-system admittance, pump gain, and combustion response — must satisfy |G(jω)| < 1 wherever its phase is near 0°. Cross unity and the amplitude grows exponentially, e^(ζ·ω·t) with an effectively negative damping ratio ζ, until a nonlinearity (cavitation collapse, valve travel, or structural failure) caps it.

The feed-system resonance: an organ pipe full of cryogen

The heart of pogo is that the propellant feedline is not a rigid pipe of incompressible fluid — it is a hydraulic resonator. Model it as a lumped fluid inertance–capacitance system. The fluid inertance is I = ρL/A (units kg·m⁻⁴), the pressure needed to accelerate the column; the compliance C (m³·Pa⁻¹) comes from line elasticity, dissolved/trapped gas, and — dominantly — cavitation bubbles at the pump inducer. Together they form an acoustic oscillator with natural frequency:

• ω_n = 1/√(I·C), i.e. f_n = 1/(2π√(I·C)).

For a LOX line of length L ≈ 15 m and area A ≈ 0.05 m² feeding a partially cavitating inducer, this lands the feed resonance in the single-digit-to-tens-of-Hz range — squarely on top of the vehicle's first and second longitudinal modes. That coincidence is the whole problem. A dry (bubble-free) line resonates high and stiff; but a cavitating inducer adds large, pressure-sensitive compliance that both lowers f_n toward the structural mode and makes the compliance a strong function of net positive suction head (NPSH), so the resonance moves as the tanks drain. A rocket that is stable at liftoff can walk its feed resonance straight into a structural mode 100 s later — exactly the Apollo 6 signature, where pogo appeared only in a specific portion of first-stage burn.

The turbopump itself contributes a complex pump transfer matrix relating inlet/outlet pressure and flow perturbations; the cavitation compliance and the pump's mass-flow gain (∂ṁ/∂p_inlet) are the terms that most often turn the loop unstable. This is why pogo analysis is inseparable from cavitation and NPSH.

Writing down the model: transfer functions and loop gain

Engineers analyze pogo as a linear feedback system and hunt for unstable poles. The building blocks:

• Structural model: a modal representation of the vehicle beam, giving the axial acceleration at the pump per unit thrust — a set of second-order transfer functions H_s(s) = Σ φᵢ²/(s² + 2ζᵢωᵢs + ωᵢ²), where φᵢ is the mode shape at the engine and ζᵢ ≈ 0.005–0.02 (structural damping is small).
• Feed-system model: the inertance/compliance network H_f(s), including the accumulator if fitted, mapping pump-inlet acceleration to inlet-pressure perturbation.
• Pump/engine model: gain and transport lag e^(−sτ) mapping inlet pressure to thrust perturbation; τ is typically a few milliseconds but its phase matters at loop frequency.

The open-loop gain is G(s) = H_s(s)·H_f(s)·H_pump(s). Stability is assessed with a Nyquist criterion (does G(jω) encircle −1?) or, equivalently, by extracting closed-loop poles and requiring all to have negative real parts. The design target, per NASA and military practice, is a gain margin ≥ 6 dB and phase margin ≥ 30° across the full flight envelope and across tank-drain conditions. Because the feed resonance sweeps during burn, the analysis is repeated at many time slices — a stability 'movie', not a single Bode plot. The single most powerful knob is to separate the feed and structural frequencies so the loop gain never peaks where the phase is dangerous.

Suppression: the accumulator that breaks the loop

The canonical fix is a pogo suppression accumulator (PSA) — a gas-charged (helium or GOX/GH₂) or spring-loaded chamber teed into the feedline near the pump inlet. It adds a large, tunable compliance C_a in parallel with the line, which does two things: it lowers the feed-system resonance well below the structural modes (detuning the loop), and it acts as a low-pass filter, shunting oscillatory flow so pressure perturbations never reach the pump. Sizing follows the same ω_n = 1/√(I·C) relation:

• Choose a target feed frequency f_target safely below the lowest structural longitudinal mode (often a factor of 2–3 down).
• Solve for required compliance: C_a ≈ 1/((2πf_target)²·I), where I = ρL/A is the effective inertance of the line segment being isolated.
• For a gas-charged unit, compliance is C_a = V_gas/(γ·p) (isentropic gas spring), so pick charge volume V_gas and pressure p to hit C_a — typically several liters of gas at feed pressure.

The Space Shuttle's SSME used a recirculating GOX-charged LOX accumulator on the main oxidizer feed; Saturn V's F-1 engines were retrofitted after Apollo 6 with helium-filled LOX prevalve cavities that dropped the LOX-line resonance out of the danger band and flew clean on Apollo 8. An alternative or complement is the cavitating venturi — a deliberately choked throat where local pressure drops to vapor pressure so downstream flow becomes independent of downstream pressure fluctuations, acoustically decoupling chamber and feed dynamics. Falcon 9 and many modern stages rely on gas-charged accumulators; the design trade is added mass, helium budget, and a pressurization system versus a catastrophic instability.

Numbers and scale: what pogo actually does to a vehicle

Pogo severity is measured as the peak sustained axial acceleration at points of interest — engine gimbal block, guidance package, and crew station. Real cases span two orders of magnitude:

• Gemini/Titan II (1962): first-stage pogo reached about 2.5 g at ~11 Hz, well above the 0.25 g crew limit; oxidizer standpipes and fuel accumulators cut it to ~0.11 g before crewed flight.
• Apollo 6 / Saturn V (1968): ~0.6 g at 5.5 Hz longitudinal, with structural loads that cracked components in the spacecraft adapter and shed a lunar-module panel.
• Ariane 5 and modern LOX/kerosene stages routinely include accumulators as standard design, keeping residual pogo below ~0.2 g.

The energy comes from thrust modulation: even a ±1–3% oscillation in chamber pressure — thrust of order tens of kN peak-to-peak on a MN-class engine — is enough to drive a lightly damped (ζ ≈ 0.01) beam mode to destructive amplitude in a few seconds. Because growth is exponential, the difference between a 5.4 Hz feed resonance and a 5.5 Hz structural mode can be the difference between a benign vehicle and a divergent one. The controlling variables an engineer tunes are, in order of leverage: feed-line compliance (accumulator), pump inlet NPSH/cavitation margin, structural mode frequency and damping, and feedline inertance (line length and diameter).

Design procedure, failure modes, and best practice

A disciplined anti-pogo campaign, per NASA SP-8055 Prevention of Coupled Structure-Propulsion Instability (Pogo), runs like this:

• Model each subsystem: build the modal structural model, the feed-system inertance/compliance network, and the measured pump transfer matrix (including cavitation compliance vs. NPSH).
• Assemble the loop: close G(s) and compute closed-loop poles at many flight times and propellant loadings.
• Check margins: require gain margin ≥ 6 dB, phase margin ≥ 30°, and no pole with positive real part anywhere in the envelope.
• If marginal, detune: add/resize an accumulator to move the feed resonance; adjust line length/diameter; add a cavitating venturi; raise structural damping only as a last resort (it's expensive per dB).
• Verify: hot-fire and flight instrumentation with accelerometers on the thrust structure and pressure transducers on feedlines; confirm the predicted stable margins.

The dominant failure modes and pitfalls: (1) a feed resonance that sweeps into a structural mode mid-burn because compliance changes with tank level and NPSH — the reason single-point analysis is dangerous; (2) an accumulator that is mistuned or loses gas charge, restoring the coupling; (3) higher structural modes (2nd/3rd longitudinal) sneaking into the band on stretched or clustered-engine vehicles; and (4) engine-to-engine phasing on multi-engine stages, where cross-coupling creates modes the single-engine model misses. Best practice today treats pogo suppression as non-negotiable baseline hardware on any crewed liquid stage — an accumulator plus verified ≥6 dB margin — rather than something to be discovered on the first flight, as it historically was.

Pogo vs. the other longitudinal/dynamic instabilities it is often confused with

Property	Pogo oscillation	Combustion instability	Aeroelastic flutter
Frequency band	5–60 Hz (structural)	200 Hz–10 kHz (chamber acoustics)	1–50 Hz (aero-structural)
Energy source	Engine thrust modulation	Chamber pressure/heat release	Aerodynamic freestream
Closing element	Propellant feedline dynamics	Injector–chamber acoustics	Airflow over lifting surface
Direction	Longitudinal (thrust axis)	Transverse/longitudinal acoustic	Torsion + bending
Primary fix	Feedline accumulator / cavitating venturi	Injector baffles, acoustic cavities	Mass balance, stiffening