The Repository Root Cause

Kegworth: The Gauge Was Right and Nobody Looked at It

October 9, 2026 · Runtime 10:54 · Watch on YouTube

In January 1989 a fan blade broke inside the left engine of a new Boeing 737-400, and nineteen seconds later the crew throttled back the right one. For three minutes the left engine's vibration indicator sat at the top of its scale. Neither pilot looked at it -- and the captain's reason had once been correct.

Chapters

Transcript

The pointer at the top of its scale 0:00

For about three minutes, one small pointer in the cockpit of a Boeing 737 sat at the very top of its scale. It was the vibration indicator for the left engine, and it was telling the truth. Neither pilot looked at it. One of them had a reason not to, and the reason had once been correct. Before that night was over, the crew had shut down the healthy engine and flown the damaged one all the way to a motorway embankment.

The two stories everyone tells 0:27

This crash usually gets told in one of two ways. The short version is pilot error: two experienced pilots shut down the wrong engine. The longer version blames the airplane's air conditioning -- the captain smelled smoke, believed the cabin air came from the right engine, and on this new model it supposedly didn't. The official report, from Britain's Air Accidents Investigation Branch, quietly takes both stories apart. What it describes is a piece of knowledge that was true, that the captain had earned honestly, and that belonged to a different generation of instrument.

Flight 92, climbing through 28,000 feet 1:03

January 8th, 1989. British Midland Flight 92, from London Heathrow to Belfast, a Boeing 737-400 that had been in service for about two months. Eight crew, one hundred and eighteen passengers. At five minutes past eight in the evening, climbing through about twenty-eight thousand feet, the outer part of one fan blade broke off inside the left engine. The engine began to surge. The airframe shuddered, the galley walls shook, and a smell of burning reached the flight deck. There was no fire warning, and no other alarm of any kind.

Nineteen seconds 1:40

The captain took control and disengaged the autopilot. He asked which engine it was. The first officer, looking at the instruments, answered: it's the le... it's the right one. The captain said: okay, throttle it back. That order came nineteen seconds after the vibration started. According to the flight data recorder, the right engine had been running with steady, normal readings the entire time. Afterwards, the first officer could not remember what he had seen on the panel that made him say right.

The coincidence that confirmed it 2:13

Then something happened that the report calls a fatal misconception. Within a second or two of the right throttle closing, the shuddering stopped. It stopped for an unrelated reason, most probably the autothrottle: disconnecting it had trimmed the power on the damaged left engine just enough to let it recover from its surging. In the cockpit, cause and effect looked perfect. The crew went on to shut the right engine down completely, and as they did, the smell of smoke cleared from the flight deck. The report put it plainly: it would have required an exceptional crew to question that association.

What about the air conditioning? 2:50

So what about the air conditioning? The captain did say, afterwards, that he suspected the right engine because he thought the smoke was coming forward from the cabin, and the cabin air came mostly from the right side. On this airplane, that wasn't quite how it worked. Each engine fed its own air conditioning pack, and the flight deck drew its air directly from the left pack -- the left engine's side. But the report goes further, and this is the part the popular version drops. It judged it unlikely that this reasoning had much influence on a decision taken in nineteen seconds.

What was actually on the panel 3:25

Here is what was actually in front of them. The 737-400 had a new engine display: instead of mechanical dials with full-length needles, two panels of light-emitting diodes, with a short row of lit segments imitating a pointer. The vibration gauges sat on the smaller, secondary panel, where each scale had just thirty-one segments. Investigators tested the system afterwards. The left engine's vibration indicator would have reached its maximum within about two seconds of the blade breaking, and it stayed there for about three minutes.

A pointer easy to lose 4:01

At full deflection, that pointer was easy to lose. It sat right beside the digital readout for the left engine's oil quantity, which was the same colour and was the most prominent thing in that corner of the display. In a survey the investigators ran afterwards, sixty-four percent of the pilots at this airline said the new display was not effective at drawing attention to rapid changes. The report's own verdict was that a mechanical needle, clear of anything else, would have been easier to see.

Why he didn't look 4:32

But there was a second reason the pointer went unseen, and it is the root of this case. After the accident, the captain said he rarely included the vibration gauges in his scan, because he believed they were unreliable and prone to spurious readings. That belief was not careless. For most of his career, it was accurate. When the American regulator first required engine vibration indicators in 1974, its own rule admitted that the detectors available were not as reliable as the engines they monitored.

The meters pilots learned to ignore 5:04

The old meters picked up vibration from almost anything, including a bumpy runway on take-off. Airlines lost so much faith in them that they were removed from the Boeing 707, from Concorde, from the DC-9 and from some 747s. Other aircraft were allowed to fly with them switched off at the circuit breaker. The report traces the captain's distrust, most probably, to his years on the DC-9. A generation of pilots had learned, correctly, to look somewhere else.

A different machine with the same name 5:36

The monitor in this 737 was a different machine. It knew the rotation speed of each engine shaft, and it filtered out every signal that wasn't turning at one of those speeds -- which was exactly what freed it from the false alarms that had ruined its predecessors. Nobody told him that. His type training on the 737 gave the vibration monitors no special importance, and no simulator session ever included high engine vibration. The improvement had happened inside the box, and the knowledge that it had happened never reached the person reading it.

First seen in a failing engine 6:11

The new display had a training gap of its own. Both pilots learned it on a one-day course about the differences between the 737-300 and the 400, and the airline had no simulator equipped with it. The report draws the consequence in one sentence: the first time a pilot was likely to see abnormal readings on this display was in flight, in an aircraft with a failing engine. The captain had twenty-three hours on the type. The first officer had fifty-three.

The cabin saw the fire 6:41

There was one more place the right answer existed. Many passengers saw fire coming from the left engine -- they described flames, torching, sparks. So did three cabin attendants at the back of the aircraft. Then the captain announced to the cabin that there was trouble with the right engine, and that it had been shut down. Several passengers heard that and were puzzled. None of them said anything to the crew. The report doesn't blame them: passengers assume the pilot already knows, and nothing in airline training at the time connected the cabin to the flight deck in a moment like that.

The approach 7:18

The crew diverted to East Midlands Airport, and the workload never let up -- radio calls, checklists, a flight management system the first officer could not get to display the new airport. At about seven minutes past the shutdown, the captain started to review what had happened. A radio call interrupted him, and he never picked it back up. On the approach he added power to the left engine, and its vibration indicator went back to maximum. Neither pilot commented.

Nine hundred feet 7:47

At twenty-three minutes past eight, two point four nautical miles from the runway and nine hundred feet above the ground, the damaged engine lost power abruptly. The captain called for the right engine to be restarted. There was not enough time, and at that speed it would not have turned over on its own. The aircraft struck a field, crossed the M1 motorway and came to rest on its western embankment, about nine hundred metres short of the runway. Forty-seven people died. Seventy-four were seriously injured.

Why the blade broke 8:20

And why did the blade break in the first place? It failed by fatigue, from a vibration mode no one knew existed. The 400's engine was a more powerful version of the one on the 737-300, and its fan blades were tested for vibration in a sea-level test cell, with a special nozzle to reach the high fan speeds that happen at altitude. The report says that was not a true test of the fan at high altitude. The dangerous mode only appeared at climb power above ten thousand feet.

A mode no test was looking for 8:52

Afterwards, engineers went back to the old test data, now knowing what to look for. The signs had been there, at such a low amplitude that they were missed -- in the report's words, without the foreknowledge of the exact mode to seek. Two more fan blades broke the same way that June, on two other British 737-400s. Britain withdrew the engine's type approval, and the blades were redesigned. Twice in one airplane, the knowledge that mattered existed only for someone who already knew where to look.

Same pointer, five months later 9:25

And one of those June incidents settles the question about the instrument. On June 9th, 1989, a Dan-Air 737-400 lost a fan blade in the climb, with the same smell of smoke in the cockpit. The captain later said every engine instrument looked alike -- except one. He identified the failed engine solely from the vibration indicator. Same display, same tiny pointer, five months later -- read by a crew that looked at it. In February, the investigators had already asked for pilots' attitude to those gauges to be reviewed.

A belief that outlived its instrument 10:02

The report made thirty-one safety recommendations. Flight-test every new engine at the altitudes and powers it will really see. Make the vibration indicator draw attention to itself when it hits maximum. Test new cockpit displays on line pilots before certifying them. None of it was about a part that couldn't do its job: the blade had a hidden mode, and the gauge was working perfectly. It was about a correct belief that outlived the instrument it was about. Next case is another 737, the year before. Aloha Airlines Flight 243, April 1988: at twenty-four thousand feet, about eighteen feet of the cabin roof tore away. The weakness in that joint was already known, and had a service bulletin. And one passenger saw a crack in the skin while she was boarding.

Description and sources

British Midland Flight 92, Heathrow to Belfast, 8 January 1989, 126 on board. Climbing through 28,300 feet, the outer panel of one fan blade in the left engine detached. The engine surged, the airframe shuddered and a smell of burning reached the flight deck, with no fire warning. The right engine -- normal throughout -- was throttled back, and the shuddering stopped, most probably because disconnecting the autothrottle let the damaged engine recover. The crew shut the right engine down.

The popular explanation is the air conditioning. On this aircraft the flight deck drew its air from the left pack, and the AAIB judged the reasoning unlikely to have weighed much in nineteen seconds. On the new LED engine display, the left vibration indicator reached maximum within two seconds and stayed there about three minutes.

The captain rarely scanned vibration gauges, believing them unreliable -- true of the previous generation, which the FAA itself called less reliable than the engines it monitored. The 737-400 monitor filtered by shaft speed, and no training said so.

Passengers and three cabin attendants saw fire from the left engine; no one told the flight deck. At 900 feet the damaged engine lost power and the aircraft came down on the M1 embankment. Forty-seven people died. Five months later a Dan-Air crew identified the same failure from the same pointer.

PRINT-READY, FROM THIS CHANNEL

The Failure Atlas, Vol. 01 — Tacoma Narrows · Citicorp Center · Millennium Bridge · Apollo 13 · the 2003 blackout · Hyatt Regency

https://therepository.gumroad.com/l/failure-atlas

PRIMARY SOURCES

Air Accidents Investigation Branch, Aircraft Accident Report 4/90: Report on the accident to Boeing 737-400 G-OBME near Kegworth, Leicestershire on 8 January 1989 (E J Trimble), HMSO, 1990 -- read in full (paras 1.1, 1.5, 1.6.3-1.6.6, 1.17.1-1.17.7, 2.1-2.2.5, 3(b), 4): https://commons.wikimedia.org/wiki/File:4-1990_G-OBME.pdf
Appendices of the same report: https://www.fss.aero/accident-reports/dvdfiles/GB/1989-01-08-UK.pdf
Kegworth air disaster: https://en.wikipedia.org/wiki/Kegworth_air_disaster

ARCHIVE FIGURES — drawings and documents from the official report, with title, author, archive and licence. Crown copyright, used under the Open Government Licence v1.0 (contains public sector information), reframed for the screen.

App. 2, Fig. 3 — AAIB | AAIB Report 4/90 | OGL v1.0
App. 2, Fig. 2 — AAIB | AAIB Report 4/90 | OGL v1.0
App. 4, Fig. 2 — AAIB | AAIB Report 4/90 | OGL v1.0
App. 4, Fig. 5 — AAIB | AAIB Report 4/90 | OGL v1.0
App. 2, Fig. 5 — AAIB | AAIB Report 4/90 | OGL v1.0
App. 2, Fig. 1 — AAIB | AAIB Report 4/90 | OGL v1.0
App. 1, Fig. 8 — AAIB | AAIB Report 4/90 | OGL v1.0
App. 1, Fig. 4(a) — AAIB | AAIB Report 4/90 | OGL v1.0

The 7 technical plates in this video are illustrations generated for the channel by a diffusion image model, styled to match its cyanotype identity. They are diagrams of the system — the figures above are from the official report, the drawings are ours, not photographs of the hardware, and no person is depicted in any of them.

Root Cause investigates why engineered systems fail, using official investigation findings and primary technical literature. Sources for this episode are linked above.

#aviation #engineering #rootcause