Compressor Stall & Surge

Why compressors are inherently unstable

Every axial compressor is a pump operating against the pressure gradient it itself creates. Air enters the front stage, gets a small kick of pressure rise from a rotor and a stator pair, then enters the next stage at slightly higher pressure and slightly lower density. Each stage adds another increment, multiplying the overall pressure ratio. A modern high-bypass turbofan core compressor stacks ten to fifteen stages to reach a pressure ratio above forty.

The catch is that pressure rise across each stage depends on the angle at which air meets the blades. The blades are designed for a specific inflow angle — the design point. Move off the design point in the wrong direction and the air separates from the suction surface of the blade, exactly the way it separates from a wing stalled at high angle of attack. The stage can no longer hold the pressure that the downstream stages demand, and air starts to slosh backward.

This vulnerability is structural. Every compressor has a stability boundary on its pressure-ratio-versus-mass-flow plane, called the surge line. Above it, the compressor cannot sustain steady forward flow. The engineering task is to schedule throttle inputs, bleed valves, and stator vanes so that the engine's operating line stays comfortably below this boundary — across the entire flight envelope, in fair and foul air, with new and worn-out hardware.

Rotating stall — the local instability

The milder of the two instabilities is rotating stall. When inflow angle is wrong somewhere around the annulus, a small region of the blade row stalls and stops pumping. Air piles up upstream of the stalled cell and divides around it. The two diverted streams hit adjacent blades at increasingly bad angles — the leading edge of the cell at a worse angle, the trailing edge at a better one. The result is that the stalled cell propagates circumferentially: the leading blade stalls (extending the cell) while the trailing blade unstalls (shortening it).

This propagation speed was first analysed by Howard Emmons at Harvard in 1955. In the rotor's rotating frame, stall cells march around the annulus at typically 30-70% of rotor speed, with about 50% being canonical. From the ground frame, you see the cells rotating in the same direction as the rotor but slower. There can be one large cell, several smaller ones, or a multi-cell pattern that varies stage by stage.

The crucial property is that mass flow continues. Stalled cells produce little or no pressure rise, but unstalled regions still pump. The compressor's average pressure ratio drops, efficiency collapses, and exhaust gas temperature climbs as the engine controller compensates by burning more fuel. Vibration spikes — particularly at frequencies that are not rotor harmonics, a giveaway for rotating-stall diagnosis. But the engine usually keeps running. Pilots may not even notice mild rotating stall except as a vague rumble and a creeping EGT.

Surge — the global instability

Surge is the catastrophic version. When rotating stall reduces overall pressure rise so much that the compressor can no longer hold off the combustor's back-pressure — or when a transient pushes the operating point above the surge line directly — the flow through the entire compressor briefly reverses. The combustor's high-pressure gas, complete with partially burned fuel, blows out the inlet.

The dynamics were modelled by Edward Greitzer at MIT in 1976. Greitzer showed that the compressor coupled to its plenum (combustor) and inlet duct behaves like a Helmholtz resonator with a non-linear pump. The governing parameter is

B = (U / 2 a_p) × √(V_p / (A_c L_c))

where U is rotor tip speed, a_p the speed of sound in the plenum, V_p the plenum volume, A_c the compressor flow area, and L_c its length. Below a critical B (typically around 0.5-1), perturbations decay or settle into rotating stall. Above it, the system oscillates: pressure rises, flow reverses, plenum empties, flow re-establishes, pressure rises again. Period: 0.5 to 1 second for typical jet engine geometry — exactly the deep BANG-BANG-BANG you can hear from outside a stalled engine.

The full reversal is what makes surge dangerous. Hot combustor gas blowing back through the compressor heats the front stages they were never designed to tolerate. Blade tips can rub against the case as the rotor distorts thermally and as axial thrust loads change sign. Fuel-air ratios swing wildly enough to flame out the combustor entirely. And the noise — a startlingly loud, low-frequency bark — has been mistaken for engine explosion or bomb damage. The 1989 United 232 crew, dealing with a fan-disc failure, were initially uncertain whether what they heard was a compressor surge or a tail-mounted engine explosion.

The compressor map and surge line

The standard diagnostic plot is the compressor map: pressure ratio on the y-axis, corrected mass flow on the x-axis, with constant-corrected-speed lines drawn across the field and efficiency contours overlaid. The operating line is the trajectory the engine traces as throttle is varied at a given flight condition. The surge line is a roughly upper-left boundary above which the compressor is unstable.

Region	Position on map	Behaviour
Design point	Centre of operating line, max efficiency	Steady, all blades at design incidence
Operating line	Locus of steady throttle settings	Stable; runs through nominal cruise/climb/descent
Surge margin	Gap between operating line and surge line	Reserve for transients, distortion, deterioration
Surge line	Upper-left boundary	Boundary of stable operation
Above surge line	High PR, low ṁ	Rotating stall or full surge
Choke line	Far right of map	Mach 1 in some stage — mass flow can't rise further

Surge margin is typically specified at 15-25% at the worst point of the operating line, defined as (PR_surge − PR_op) / PR_op at the same corrected speed. The margin has to absorb: rapid throttle slams (the operating line transiently overshoots the steady line), inlet distortion (which moves the effective surge line down), foreign-object damage (which moves it down further), and lifetime deterioration (gradual erosion of blade tips and seals). New engines start life with a comfortable margin and lose 5-10% over decades of service.

What pushes an engine into surge

• Throttle slams. A rapid throttle command asks the engine for more fuel and thus more compressor work, but rotor inertia lags. Fuel-burn schedule has to ramp gently — too fast and the operating line shoots up past the surge line before mass flow can catch up.
• Inlet distortion. Crosswinds on the ground, high angle of attack in flight, sideslip during a missed approach, or external pylons can starve a sector of the inlet. The stalled sector becomes the seed for rotating stall. Military fighters quantify this with a 'DC60' distortion index measured during certification.
• Foreign-object damage and blade erosion. A bird strike, runway debris, or sand ingestion bends or chips blade leading edges. The damaged blades produce less pressure rise, dropping the achievable surge line below the operating line at that condition.
• Bleed valve and VSV malfunction. A bleed valve stuck closed during start, or a VSV actuator stuck off-schedule, leaves the compressor operating with the wrong stage matching. Front stages can go into rotating stall during the start sequence; the engine 'hangs' or surges.
• Supersonic inlet unstart. At Mach 2+ the inlet contains a shock train carefully positioned in the duct. A perturbation can pop the terminal shock out forward of the cowl lip — an 'unstart'. The compressor sees a sudden inlet pressure collapse and surges within milliseconds. The SR-71 was famous for it.
• Combustor instability driving back-pressure swings. Acoustic combustion instabilities ('rumble', 'screech') modulate combustor pressure at hundreds of Hz. If they couple to the compressor at the right phase, they can trip stall.

Variable stator vanes — the first historical fix

The fundamental problem at low corrected speed is that the front stages of a multi-stage compressor see the wrong inflow angle. At the design point, rotor blades and fixed stator vanes line up to receive air at the right incidence. At low speed, mass flow drops more than rotor speed drops, so axial velocity falls and inflow becomes too steep relative to the blade chord. Front stages stall; back stages choke.

Pratt & Whitney solved this on the J57 (1953) by making the front-stage stator vanes rotatable. A schedule, originally hydraulic, opens or closes the vanes as a function of corrected speed. Closed vanes deflect the flow back toward the blade chord, restoring incidence. Open vanes — at high speed — get out of the way and minimise loss.

The J57 powered the Boeing B-52, F-100, F-101, F-102, and the Douglas DC-8. Its VSV mechanism set the architectural template for every modern engine. The F119 (F-22 Raptor) and F135 (F-35) carry VSVs on the first three to five compressor stages, with FADEC-scheduled angles updating continuously to flight condition. Civilian turbofans like the GEnx, GE9X, PW1000G, and Trent XWB all use VSVs as a matter of course.

Inter-stage bleed valves — the second fix

VSVs solve front-stage matching; bleed valves solve back-stage matching. Below about 70% corrected speed — engine start, descent at flight idle — the back stages of the compressor would choke if the front stages were trying to push their full mass flow through. The trick is to vent some of that flow overboard (or into a bypass duct) at one or more intermediate stages, lowering the back-pressure that the front stages have to overcome.

Bleed valves are typically ring-shaped collector slots at chosen stages, opening and closing in concert with the VSV schedule. They're commanded fully open at start, modulated through ground idle and low-power flight, and fully closed above some threshold corrected speed. On modern FADEC-equipped engines the bleed schedule is computed in real time from corrected speed, ambient conditions, and learned engine deterioration trims.

The cost is performance: every kilogram of bleed air is mass flow that doesn't pass through the combustor and turbine. Bleed valves are an explicit trade — efficiency surrendered to gain surge-free starts and low-power operation. Their schedule is one of the most heavily tuned variables in the engine control law, and ground-test campaigns of new engines spend weeks finding it.

FADEC and the surge-margin discipline

Modern aero-engine control is owned by the Full Authority Digital Engine Control (FADEC). It has authority over fuel flow, VSV angle, bleed valve position, and (on geared turbofans) gearbox engagement. Its job during transients is to slew the operating point along a path that respects all the constraints simultaneously — pressure ratio, turbine inlet temperature, surge margin, EGT redline, blade tip clearance.

The classical formulation is a fuel-flow schedule that accelerates the engine along a path on the compressor map that holds a fixed surge margin. During a throttle slam from idle to maximum, the FADEC ramps fuel along a curve that keeps the operating line a fixed delta below the surge line — typically targeting 10-15% margin during the transient. Once the engine reaches the commanded N1, the schedule transitions to a steady-state map with full 20-25% margin.

This is why a modern airliner can go from idle to takeoff thrust on a 4-5 second ramp without surging, while a 1950s engine without FADEC needed pilot finesse to avoid stalling on the same maneuver. The hardware is similar; the control law has improved by orders of magnitude.

The SR-71 inlet unstart

The Lockheed SR-71 Blackbird flew at Mach 3.2 at 80,000+ feet. Its Pratt & Whitney J58 engines ran as conventional turbojets at low speed, then transitioned at about Mach 2 into a hybrid that bypassed compressor air around to the afterburner. By Mach 3, the inlet itself contributed about 80% of the engine's thrust as a giant ramjet, with the J58 acting mainly as a flame holder and starter.

This worked because of a finely tuned spike inlet. A translating centrebody positioned a normal shock just inside the throat. The shock had to sit in a narrow range of axial positions to keep the inlet 'started' — meaning supersonic flow inside the duct decelerating through the normal shock to subsonic at the compressor face. If the shock popped out forward of the cowl lip — an inlet unstart — the inlet went subsonic, mass flow collapsed, and the engine surged within milliseconds.

Unstarts could yaw the aircraft 30+ degrees in the affected direction, slamming the pilot against the canopy. Early SR-71 missions saw multiple unstarts per flight. The fix was the Digital Automatic Flight and Inlet Control System (DAFICS), which monitored inlet conditions hundreds of times per second and trimmed the spike position and bypass doors faster than the pilot could. Unstart rates dropped substantially, but they never reached zero — the Blackbird's unstart remained the signature failure mode of supersonic propulsion until the program ended in 1999.

Diagnosing stall in test cells and in flight

Stall and surge leave distinctive fingerprints:

• Pressure transducer at the compressor exit. A surge looks like a sudden plunge in P3 (compressor discharge pressure) followed by a recovery oscillation at the Helmholtz frequency. The signature is unmistakable.
• Rotor speed transient. The momentary unloading of the compressor causes a brief N2 (high-pressure spool speed) spike, followed by a dip as the rotor decelerates against reversed flow.
• EGT (exhaust gas temperature). Rotating stall raises EGT slowly as the FADEC adds fuel to maintain thrust against falling compressor efficiency. Full surge spikes EGT briefly because mass flow drops while fuel flow lags.
• Vibration spectrum. Rotating stall produces a peak at a non-integer fraction of rotor speed — the cell-propagation frequency. This is a textbook signature in vibration tags.
• Audible cues. Mild stall is a rumble. Full surge is the BANG — sometimes a single bang, sometimes a series at 1-2 Hz until recovery.
• Visual flame. Particularly at night, a surging engine can spit flame out the front of the inlet — gas from the combustor blowing back. Aircraft accident investigators look for soot patterns on the inlet lip as evidence of post-event surge.

Active surge control — where research is going

Since the late 1990s, researchers have pursued active surge control: instead of designing for a passive margin, instrument the compressor with high-bandwidth pressure sensors and use fast bleed valves or air jets to suppress the precursor oscillations before they grow. Greitzer's group at MIT demonstrated factor-of-two reductions in required surge margin on laboratory compressors using injected air at the compressor face. NASA and Rolls-Royce have flown variants on test engines. The promise is to operate closer to the surge line — therefore at higher pressure ratio and better efficiency — without losing reliability.

The catch is system reliability: an active surge controller that fails leaves the engine operating with no margin. Certification of active stability augmentation is a hard regulatory problem; production engines today still rely overwhelmingly on passive margin plus FADEC scheduling. The next decade of geared and open-rotor concepts may finally push active control into production as their compressors get smaller and pressure ratios climb.

Famous incidents involving compressor surge

• SR-71 unstarts (1965-1999). Routine at Mach 3+. The Blackbird's defining quirk and the most-studied supersonic propulsion failure mode in history.
• B-1B birdstrike at Diego Garcia (2001). A B-1B Lancer ingested birds into multiple engines during takeoff, triggering compressor surges. The crew aborted, the aircraft was severely damaged; nobody was hurt.
• QF32 (Qantas Flight 32, 2010). An A380 suffered an uncontained engine failure on a Trent 900, with subsequent compressor disturbances and surge-like events on adjacent damage. Crew landed safely.
• F-22 engine surges in service. Early production F119 engines surged repeatedly during off-design throttle inputs. Software updates to the FADEC schedule and minor hardware modifications reduced the rate to near zero.
• Helicopter engines in hot-and-high conditions. Turboshaft engines at high density altitude routinely operate with reduced surge margin. Pilots in Afghanistan and high-altitude rescue operations learned to manage throttle slowly to avoid compressor stalls during steep climbs.

Common pitfalls and misconceptions

• Treating stall and surge as the same thing. They are related but distinct. Rotating stall is local and survivable; surge is global and damaging. Many incident reports conflate them and confuse maintenance investigations.
• Assuming bigger surge margin is always better. Margin costs efficiency. A 30%-margin engine burns more fuel than a 20%-margin engine at cruise; the economics of long-haul aviation push designers toward thinner margins.
• Ignoring inlet distortion at the test cell. Engines that look healthy on a test stand can stall the moment they're installed in an airframe with imperfect inlet geometry. Inlet distortion certification testing is its own discipline.
• Misreading rotating-stall vibration as imbalance. An imbalance peak sits at exactly rotor frequency; a stall-cell peak sits at a non-integer fraction. Diagnosing one as the other leads to needless balance attempts that don't fix anything.
• Believing pilots can 'feel' the surge line. Modern jet engines hide their state behind the FADEC. The first indication the pilot has of impending surge is often the surge itself — by which point the FADEC has already taken corrective action automatically.