TSB A23P0020: a compressor blade that fatigued in 909 cycles of a 20 000 cycle life
TSB Canada traced a PW150A compressor blade fracture to high-cycle fatigue at a 26 000 rpm resonance. Lessons on dwell time, tip clearance and cold-day stress.
Key takeaways
- A PW150A 2nd stage compressor blade fractured near its root after 909 cycles, against a rotor life limit of 20 000 cycles.
- The TSB found high-cycle fatigue from a subsurface origin, driven by a mechanical resonance mode at 26 000 rpm (plus or minus 50 rpm).
- Cold air below minus 25 °C raised the resonant stress by 2.5 times, and a fuel-saving bleed setting pushed rotor speed into the band.
- Tighter, in-specification tip clearances after an engine restoration let the rotor run faster and dwell longer at the resonance.
- A life limit in cycles says nothing about time spent at a resonance; dwell must be tracked as its own damage driver.
At a glance
- Investigating body
- Transportation Safety Board of Canada (TSB)
- Report
- A23P0020, released 22 September 2026
- Occurrence
- 2 March 2023, DHC-8-402 C-GVEN, diverted to Kelowna, British Columbia
- Failed part
- Low-pressure compressor 2nd stage (LPC2) rotor blade, PW150A engine
- Failure mode
- High-cycle fatigue from a subsurface origin at resonance mode M1/5E (26 000 rpm)
- Rotor usage at failure
- 909 cycles since new, against a 20 000 cycle limit
- Injuries
- None (4 crew, 74 passengers)
On 22 September 2026 the Transportation Safety Board of Canada (TSB) released report A23P0020. It covers a WestJet Encore DHC-8-402 that had an engine surge and exhaust duct fire in March 2023. The aircraft was at about 7 600 m (25 000 ft), transitioning from climb to cruise after leaving Kamloops, when the crew heard a loud bang. They shut down the No. 2 engine, a Pratt & Whitney Canada PW150A, and made a single-engine landing at Kelowna. Nobody was hurt.
The trigger was one low-pressure compressor 2nd stage (LPC2) rotor blade that fractured near its root. The engine control then raised fuel flow, and unburned fuel ignited in the exhaust duct.
What the investigation found
The rotor had been fitted new during an engine restoration in March 2022. At the time of the event it had 909 cycles against a life limit of 20 000. Fractography showed a fatigue crack that started at a crystallographic facet about 0.12 mm (0.0047 in) below the pressure-side surface. The striations were typical of high-cycle fatigue. A second blade on the same rotor had a crack at a matching location. The blades are peened and had no nicks, scratches or corrosion, so the TSB ruled out a surface stress riser.
The driver was a resonance. The TSB and the engine maker found that the blade’s highest stress at the origin occurs at mode M1/5E. That mode is excited at a low-pressure spool speed of 26 000 rpm (plus or minus 50 rpm). The manufacturer judged about 30 minutes of exposure in that band could be damaging. Three conditions pushed this engine there:
| Factor | What the report says |
|---|---|
| Bleed setting | Selecting BLEED MIN in climb and cruise raised spool speed, making the band easier to reach |
| Cold air | Below minus 25 °C, spool speed rises and M1/5E stress increases by 2.5 times |
| Tip clearance | The restoration left smaller, in-specification tip clearances, so the engine ran faster |
| Buffeting flow | Chaotic flow through the compressor added about 10% to the excitation |
Engine monitoring data showed dwell time near the resonance rose after the restoration, mostly on flights between November 2022 and March 2023. The report lists 11 LPC2 blade failures since 2012 that led to in-flight shutdowns. On-wing inspections later found two more cracked rotors in the same fleet. Safety actions include borescope fluorescent penetrant inspections, dwell-time monitoring and revised bleed procedures. Transport Canada AD CF-2025-44 also mandates a FADEC software change that cuts fuel after a significant engine event.
Why it matters for design engineers
This part failed well inside its life limit, because the damage came from a loading regime the limit did not count. A cycle-based life treats each start-stop as one unit of damage. A resonance adds a very large number of small stress cycles in every minute of exposure, and their effect depends on dwell time, air temperature and clearances. None of those show up in a cycle counter.
The subsurface origin also matters. Peening protects against surface-initiated cracks, but once the surface is compressed the weakest point can move inside the material. The report ruled out a notch, so a larger fillet would not have saved this blade. Lower vibratory stress or less time at resonance would have.
| If you | Check |
|---|---|
| Rely on a cycle life limit for a rotating part | Whether any resonance crossing or dwell band exists in service, and whether dwell is tracked separately |
| Allow a range of build clearances or tolerances | How the tolerance extremes shift operating speed or load; run a tolerance stack-up at both ends |
| Assess fatigue at a notch | The stress concentration factor, but also subsurface origins once the surface is peened or rolled |
| Qualify a part at room temperature | Whether cold-day or hot-day conditions change the excitation or the mean stress |
| Let operators choose an efficiency setting | Whether that setting moves the machine into a band your analysis treated as rare |
Sources
This post summarises the primary sources listed above and adds our own engineering analysis. Read the original report or release before relying on any detail. See the editorial policy for how corrections are handled, or follow by RSS.