British Midland Flight 92 (Kegworth)

Date

8 January 1989

Phase of flight

Approach

Location

M1 motorway embankment near Kegworth, Leicestershire, on the approach to East Midlands Airport, United Kingdom

Aircraft

Boeing 737-400

Registration

G-OBME

Operator

British Midland Airways

Flight

British Midland 92 (BD092)

Occupants

126

Fatalities

4739 passengers died at the scene and 8 later in hospital; all 8 crew members survived; 74 occupants were seriously injured

Investigating body

Air Accidents Investigation Branch (United Kingdom)

Final report

AAIB Aircraft Accident Report 4/90 (EW/C1095)

Report date

25 August 1990

Report title

Report on the accident to Boeing 737-400 G-OBME near Kegworth, Leicestershire on 8 January 1989

On 8 January 1989, British Midland flight BD092, a Boeing 737-400 registered G-OBME, was climbing out of London Heathrow for Belfast when an outer panel of one fan blade broke away in its left (No 1) engine. The crew believed the trouble lay in the right engine and shut it down. The damaged left engine kept running until the final approach, then lost power 2.4 NM short of runway 27 at East Midlands Airport, where the crew were diverting. At 20:24:43 UTC the aircraft struck the ground just east of the M1 motorway and came to rest on the motorway's western embankment near Kegworth, Leicestershire.

Of the 8 crew and 118 passengers, 39 passengers died at the scene and 8 more later in hospital. All the crew survived; 74 of the survivors were seriously injured.

The AAIB report is a detailed study of how a crew can form a wrong diagnosis under stress, receive what looks like confirmation, and then stop questioning it. It also examined engine instrument design, communication between cabin and flight deck, and occupant protection.

The flight

G-OBME was a new aircraft, with 521 airframe hours; its certificate of airworthiness had been issued on 3 November 1988. It was powered by two CFM56-3C turbofans, rated at 23,500 lb thrust for takeoff. The 737-400 had a new engine instrument system (EIS) that replaced mechanical pointers with light-emitting diode (LED) displays: N1, EGT, N2 and fuel flow on the primary display, and oil pressure, oil temperature, vibration and hydraulic pressure on the smaller secondary display.

The commander had 13,176 hours and had been a captain with the company for 14 years, but only 763 hours on the 737 and 23 hours on the Series 400. The first officer had 3,290 hours, 192 on the 737 and 53 on the Series 400; he had flown jet transports for only 6 months. The aircraft took off from Heathrow at 19:52 with the first officer flying. All times in the report are UTC.

The accident

At 20:05:05, climbing through FL283 about 20 NM south-south-east of East Midlands Airport, the crew felt moderate to severe vibration and smelt burning. There was no fire warning. The flight data recorder showed the No 1 engine's N1 fluctuating between 99 % and 74 % and its EGT rising from 780 °C to 900 °C, with low and fluctuating fuel flow, over about 22 seconds of compressor stalls. The No 2 engine was steady and normal throughout.

Within 1 to 2 seconds of the No 2 throttle being closed, the aircraft rolled level again, the shuddering stopped, and the No 1 engine settled at an N1 3 % below its previous value and an EGT 50 °C higher. Its vibration indicator, however, stayed at maximum. When the No 2 engine was finally shut down, the smell of smoke cleared from the flight deck, which the commander later said convinced him he had acted correctly.

In the cabin, many passengers and the three cabin attendants in the rear saw fire or sparks from the left engine. The commander told the passengers over the public address that the right engine had been shut down; many who had seen the fire were puzzled, but none raised it with the cabin crew, and the three aft cabin attendants said they had not heard the reference to the right engine.

The descent was flown by hand with a high workload: ATC headings, calls to the company, weather, checklists, and two minutes spent by the first officer trying without success to reprogram the flight management system for East Midlands. The commander's attempt to review the symptoms at 20:12 was cut short by ATC and never resumed.

On final approach the No 1 engine lost power. The commander raised the nose to stretch the glide while the first officer tried to restart the No 2 engine, but flight conditions were outside the windmill start envelope, there was probably too little bleed air, and no power was obtained. The aircraft first struck a field in a 13° nose-up attitude at 113 kt, passed through trees and made its second, major impact on the northbound carriageway and western embankment of the M1. It came to rest about 900 m from the runway threshold. The fuselage broke into three main sections.

The investigation

The engine failure

Fan blade No 17 of the No 1 engine had failed in fatigue and its outer panel had detached. The imbalance made the fan and compressor blades rub their abradable seals, which put smoke and a smell of burning into the air conditioning. The released outer panel probably lodged in the acoustic lining of the intake; when power was raised on the approach it was shaken free and ingested, causing extensive secondary fan damage, compressor stalls and a fire outside the fan case, where the prolonged vibration had loosened fuel and oil unions.

The initial conclusion was that the failure was isolated. Then in June 1989 two more CFM56-3C-1 fan blades failed on UK 737-400s climbing at high altitude: a Dan-Air aircraft, G-BNNL, on 9 June, and another British Midland aircraft, G-OBMG, on 11 June. Both crews identified the failed engine correctly, with the help of the vibration indicators. The AAIB found that the -3C-1, rated at higher climb thrust than the earlier -3B-2, had a fan vibration mode excited at high corrected fan speed at altitude, a condition the certification test-bed running had not reproduced. On 12 June 1989 the UK CAA withdrew the type approval of UK-registered 737-400s; an FAA airworthiness directive then required replacement of the fan blades and discs that had run at -3C-1 ratings and restricted engines to -3B-2 ratings.

Why the wrong engine

The EIS was serviceable and would have shown the No 1 engine's large parameter swings. The AAIB found that both pilots acted before they had any positive evidence of which engine was at fault, contrary to their training and the Operations Manual, which required them to evaluate all indications before acting. The first officer could not recall what he had seen on the instruments; he half-formed the word "left" before saying "right", perhaps glancing during a 6-second lull between surges. The commander, handling the aircraft after disengaging the autopilot, later explained his choice by the air conditioning: he thought the smoke came from the cabin, whose air came mostly from the right engine. That reasoning suited aircraft he had flown before, but on the 737-400 some cabin air comes from the left engine. The AAIB thought it more likely that he provisionally accepted the first officer's assessment, believing the first officer had seen positive indications.

The decisive moment was a coincidence. The evidence indicated that disengaging the autothrottle, just before the No 2 throttle was closed and at a moment when it was demanding slightly less thrust, let the damaged No 1 engine recover from its stalls. The noise and shuddering ceased as the crew closed the wrong throttle. The AAIB wrote that it "would have required an exceptional crew to question the association between their action and its apparently obvious, and highly desirable, consequence."

The instruments

The airborne vibration monitor (AVM) showed maximum vibration, 5 units, on the No 1 engine within about 2 seconds of the failure, and kept showing it for about 3 minutes. Neither pilot noticed. The commander rarely scanned vibration gauges, believing them unreliable, a view many pilots of earlier jets shared, and neither pilot's training had drawn attention to the newer, more reliable AVM. On the EIS the vibration pointer was a small LED segment, easily lost beside the adjacent oil quantity digits. A Boeing bulletin of March 1988 had introduced a high vibration drill: retard the thrust lever to keep the AVM below 4.0 units, with the note that engine shutdown is not required.

The AAIB found the LED displays less conspicuous than mechanical pointers and noted that the EIS appeared to have been introduced without a thorough evaluation with line pilots. With no EIS-equipped simulator available, a pilot was likely to see abnormal EIS indications first in flight, with a failing engine.

Communication between cabin and flight deck

The AAIB considered it "extremely unfortunate" that the fire evident to many passengers never reached the flight deck. Passengers assume the pilots know; cabin crew avoid interrupting a busy flight deck, and airline training did not then prepare them to pass on what they had seen. The report stated that had some initiative been taken by the cabin crew who saw the left engine's distress, the accident could have been prevented.

Survival

The second impact produced a peak deceleration of about 22 to 28 g, mainly longitudinal, beyond the certification standards for the airframe and furnishings but within human tolerance. Deaths were concentrated where the cabin floor collapsed, forward and just aft of the wing; in the overwing and tail sections the floor and seats stayed largely intact. All but one of the 30 overhead bins fell. A computer simulation for the report gave a much lower head injury value for a braced occupant than an unbraced one.

Probable cause and contributing factors

The AAIB does not use the term "probable cause". Its conclusion was:

It listed five factors that contributed to the incorrect response of the flight crew:

1. The combination of heavy engine vibration, noise, shuddering and a smell of fire was outside their training and experience.
2. They reacted to the initial engine problem prematurely and in a way that was contrary to their training.
3. They did not assimilate the indications on the engine instrument display before they throttled back the No 2 engine.
4. As the No 2 engine was throttled back, the noise and shuddering from the surging No 1 engine ceased, persuading them that they had correctly identified the defective engine.
5. They were not informed of the flames from the No 1 engine seen by many on board, including three cabin attendants in the aft cabin.

The findings added that the decision to land at East Midlands without delay was correct, but created a workload that precluded an effective review, and that no simulator training had been given, or required, on recognising engine failure on the EIS or on decision making for failures outside standard procedures.

Safety recommendations and what changed

The AAIB made 31 recommendations between January 1989 and March 1990. Among them:

• advise pilots that high engine vibration may bring smoke or a smell of burning into the cabin and flight deck through the air conditioning, from blade tips rubbing their seals, and add a procedure for that combination to the 737 Flight Manual;
• train pilots in the value of engine vibration indicators for identifying an engine with a rotating assembly failure, and fit an attention-getting feature when a vibration indicator reaches maximum;
• evaluate new instrument displays with line pilots under normal and abnormal conditions, and give pilots new to electronic displays simulator familiarisation first;
• simulator practice in reprogramming the FMS for unplanned diversions, and ATC guidance on offering a discrete frequency in an emergency;
• joint exercises for pilots and cabin crew, and a review of pilot training that includes the potential of simulator training in flight deck decision making;
• instrumented flight tests of engines for damaging vibration at all altitudes and powers;
• stronger seats, cabin floors and overhead bins, research into upper torso restraint and aft-facing seats, and child seats for infants.

Lessons for pilots

Do not rush the diagnosis. Few engine malfunctions require action within seconds. Fly the aircraft, then read every engine parameter on both engines before touching a thrust lever, and confirm the identification between both pilots. The Boeing training cited in the report told pilots to avoid precipitate action; the order to throttle back came some 19 seconds after the onset.

Know your instruments, especially the ones you rarely look at. A vibration indicator at full scale was the clearest cue to which engine had failed, and both later crews used it. Knowing what normal and abnormal readings look like on your type's display, and how systems such as air conditioning are actually supplied, protects against reasoning from another aircraft. See engine indications.

Review, and keep reviewing. After an engine shutdown, a structured review of what happened, what was done and what is still unexplained is part of engine failure management. Reducing workload, by using the autopilot, asking ATC for a quiet frequency or delaying non-essential calls, makes room for it.

Use the whole crew. Cabin crew and passengers could see what the pilots could not. Modern crew resource management training includes cabin crew and teaches both sides that a short, factual report of what is seen from the cabin, such as fire or damage on a particular side, can be decisive.

Crash Investigations pages summarise official investigation reports for study and exam preparation. The investigating body's report is the authoritative account and prevails wherever it differs from this page. Under ICAO Annex 13, an investigation exists to prevent accidents, not to apportion blame or liability.