Air Safety

The final report on this highly complex accident, published by the Air Accidents Investigation Branch of the Department of Transport in the United Kingdom, contains many significant lessons for aircraft operators and pilots alike. A copy recently became available to ‘Australian Aviation’ and this synopsis of the operational aspects of its findings has been prepared by our air safety specialist, Macarthur Job.

* * *

The Flight

Boeing 737-400, G-OBME, operated by British Midland Airways Ltd, was engaged in a flight from London Heathrow Airport to Aldergrove Airport, Belfast, carrying a crew of eight and 118 passengers. The weather over south eastern England at the time was influenced by a moist south westerly airstream, with scattered stratus and stratocumulus cloud between 1,000 and 3,500ft.

With the first officer flying the aircraft from the right hand seat, G-OBME took off from Heathrow in daylight at 1952 hours U TC. It was subsequently cleared to climb to FL350 on a direct track to the Trent VOA.

Seven minutes later, while still climbing some 20 nautical miles south east of East Midlands Airport at Castle Donnington, moderate to severe vibration suddenly developed and the flight crew became aware of a strong smell of burning. There was no fire or other warning, but the captain took control and disengaged the autopilot. The engine instrument display gave him no immediate indication of the problem, but as smoke and fumes seemed to be coming from the cabin, his understanding of the air conditioning system led him to suspect the trouble lay with the No 2 (right) engine.

The first officer was monitoring the engine indications and when asked which engine was giving trouble, he replied, after an initial hesitatib/1, that it was the right one. Told to “throttle it back”, the first officer disengaged the autothrottle and reduced power on the No 2engine. This appeared to reduce the smell and smoke and the captain thought the vibration had also ceased.

After the first officer had called London Air Traffic Control to report they had an emergency which “looked like an engine fire”, the captain ordered him to shut it down. At this point London ATC advised the crew of the aircraft’s position and asked to which airfield they would like to divert. The first officer replied that it looked as if they “would take it to Castle Donnington” (East Midlands Airport).

While the captain called the company at East Midlands Airport, the first officer began the Engine Failure and Shutdown checklist. After the No 2 engine had been shut down, the captain, sensing that the smoke and smell hadcleared, was satisfied the action they had taken was correct. Meanwhile in the cabin, passengers and cabin attendants had heard an unusual noise, accompanied by vibration “like a car backfiring”. One attendant felt the walls of the forward galley “shuddering”. Some passengers were also aware oflight smoke in the cabin and a smell of burning which they variously described as like “rubber”, “oil” or “hot metal”. Many passengers, as well as the three flight attendants in the rear of the cabin, saw signs of fire- “torching” or “sparks” – coming from the No 1 engine.

After the No 2 engine was shut down, the captain called the cabin service manager to the flight deck to check if there had been smoke in the cabin. As a meal was in the process of being served when the emergency developed, the captain instructed him to clear up the trays and pack everything away.

A minute later the cabin service manager returned to the flight deck to say that the passengers were becoming agitated. The captain then announced over the cabin address system that there was trouble with the right engine which had produced some smoke in the cabin, that it was now shut down, and that they could expect to land at East Midlands Airport in 10 minutes.

Although a number of passengers who had seen the fire in the left engine were puzzled by the reference to the right engine, none brought the discrepancy to the attention of the cabin crew. By this time however, the smoke and smell had dissipated, though several passengers were still aware of vibration.

At this stage the aircraft was only a few nautical miles (nm) south of East Midlands Airport, and London ATC, after clearing it to descend to FL1OO, passed its control to Manchester ATC. Manchester then gave the aircraft radar vectors to descend to the north of the airport and intercept the localiser of the ILS for Runway 27. The airport elevation is 280 feet amsl.

Throughout the descent the captain flew the aircraft manually while the first officer handled the numerous radio communications. The workload on the flight deck remained high as he obtained details of the East Midlands weather and attempted to programme the flight management system to display the landing pattern at the airport, but without success. The actual weather passed to the aircraft at 2011 hours was: wind 250 degrees/10 knots: visibility 10km; cloud 7 oktas, base 1,700 feet; temperature 9 degrees C: QNH 1018.

During a break in radio traffic as the descent continued, the captain made an attempt to review their handling of the emergency. At this moment however, he was interrupted by a further ATC transmission passing a new radar heading, a further descent clearance to FL40 and instructions to change frequency to East Midlands Approach Control. As soon as contact was reestablished on the new frequency, the first officer began to read the One Engine Inoperative Descent and then the Approach checklists. These were also interrupted when f\pproach Control requested the crew to make a test call to the 1;1irport fire service. The Approach checklist was finally completed when the aircraft was 15nm from touchdown, descending through 6,500 feet. A minute later the captain accepted a new vector of 220 degrees to take the aircraft south of the extended runway centreline and increase its distance from touchdown.

When 13nm from touchdown, it was vectored back towards the runway centreline and the crew increased power on No 1 engine to level the aircraft at 3,000 feet. It was then cleared down to 2,000 feet and the captain began a slow descent, calling successively for increased increments of flap. After intercepting the runway localiser at 2,000 feet agl, the captain requested the undercarriage to be lowered and, as they passed the outer marker 4. 3nm from touchdown, called for 15 degrees of flap. A minute later, when 2. 4nm from touchdown at a height of 900 feet, the No 1 engine abruptly lost power.

The captain at once instructed the first officer to relight the No 2 engine and raised the aircraft’s nose in an attempt to reach the runway. Both actions were in vain and the aircraft descended to the ground a kilometre from the runway threshold, striking the ground on the edge of a field immediately short of the M1 motorway which skirts the airport’s eastern boundary.

The fuselage tail skid and main undercarriage legs were torn off and the aircraft, continuing at high speed through the motorway embankment’s wooden fence, cut a swathe through trees on the sloping eastern embankment. Passing above the eastern carriageway, it demolished a lamp standard in the motorway’s centre strip, and finally crashed with great force against the western embankment. This major impact broke the fuselage into three sections, fracturing it ahead of and behind the wing, the tail section jacknifing over the centre section before coming to rest upside down.

Airport fire service crews waiting in position by the runway were watching the aircraft as it approached. Seeing it descend below the motorway embankment, they drove at high speed to the accident site via the runway and the airport’s eastern crash gate. Arriving in only five minutes, they immediately extinguished the fire in No 1 engine which had grown in intensity, and laid a blanket of foam as a fire protection against leaking fuel.

They were quickly joined by units from the Leicestershire, Nottinghamshire Derbyshire fire services. Ambulances from the same counties followed the fire vehicles to the scene and, assisted initially by passing motorists, their crews set about recovering the survivors from the wreckage and moving them to hospital. RAF helicopters were subsequently called in to uplift some of the more seriously injured.

The task proved to be enormous. Despite the fact that a total of 7 4 road ambulances attended the scene, assisted by personnel from the Army, the RAF and the Derbyshire Miners’ Rescue Team, The last passenger was not extricated until eight hours after the crash.

Thirty nine passengers had been killed in the impact. Sixty seven passengers and seven crew members, including the two pilots,had been seriously injured. Only four passengers and one cabin attendant escaped with minor InJunes. Eight of the

seriously injured passengers subsequently died in hospital.

Investigation

The first sign of any problem had occurred 13 minutes after takeoff while the aircraft was climbing through FL283. At the time the crew experienced vibration and a smell of burning, the cockpit voice recorder (CVR) picked up a “rattling” sound and the flight data recorder (FDR) showed fluctuations in lateral and longitudinal accelerations. The FDR also showed that severe vibration had occurred in No 1 (left) engine, accompanied by marked fluctuations in fan speed (N1), a rise in exhaust gas temperature (EGT) and low, fluctuating fuel flow.

When asked by the captain which engine was causing the trouble, the first officer had said: “It’s the le . . . it’s the right one. ”

“OK – throttle it back,” the captain had responded. At this time, all the No 2 (right) engine’s indications were steady.

Within two seconds of the closure of the No 2 throttle, the fluctuations in lateral and longitudinal accelerations ceased, the No 1 enginefan speed settled at 3% below its previous stable speed, and its EGT stabilised at 50 degrees C above its previous level. These engine parameters remained fairly stable for a further minute until the engine’s power was reduced for descent, but the indicated vibration remained at maximum and the fuel flow was erratic.

The captain’s order to shut down the No 2 engine came 43 seconds after the onset of the vibration, but its execution was delayed when he added: “Seems to be running all right now. Let’s just see if it comes in. ”

After further radio traffic concerning the aircraft’s diversion to East Midlands Airport and a flight attendant’s instruction to passengers over the cabin address system to fasten their safety belts, the first officer told the captain he was about to start the “Engine Failure and Shutdown” checklist, adding: “Seems we have stabilised. We’ve still got the smoke. ” The fuel cock for the No 2 engine was closed and the auxiliary power unit (APU) started two minutes and seven seconds after the initial onset of the vibration.

When the captain called the cabin service manager to the cockpit after the No 2 engine had been shut down, he asked: “Did you get smoke back there?” The cabin service manager replied: “We did, yes. ” At this stage the captain instructed him to clear up the passengers’ meal trays. A minute later the cabin service manager returned to say to the captain: “Sorry to trouble you . . . the passengers are very very panicky. ” The captain then made the announcement that they were diverting to East Midlands Airport.

At 2012. 28 hours, almost seven and a half minutes after the onset of the engine trouble, the captain attempted to review their situation during a break in radio traffic. “Now what indications did we actually get,” he began to the first officer. “Just rapid vibrations in the aeroplane . . . smoke . . . ” He was interrupted by further radio transmissions, and the first officer then began the One Engine Descent and Approach check list which was completed at 2017. 33 hours. Shortly afterwards the captain called for the wing flaps to be extended one degree.

At 2020. 03 hours, after the aircraft had been given its final vector to intercept the runway centreline, power was increased on the No 1 engine to level the aircraft momentarily at 3,000 feet. As this occurred, maximum vibration was again recorded on the FDR. The aircraft was then cleared down to 2,000 feet and the captain called successively for two degrees, then five degrees of flap, and on descending through 2,000 feet, for the undercarriage to be extended.

At 2023. 49 hours, with 15 degrees of flap now lowered and the aircraft 2. 4nm from touchdown at a height of 900 feet, there was a crackling noise on the CVR, possibly the result of electrical interference, as the No 1 engine lost power. Immediately afterwards the first officer madea transmission to the tower indicating they were having trouble with the second engine as well, and the captain told him: ” Try lighting the other one up there’s nothing else you can do. ” Seventeen seconds after the loss ofpower, the No 1 engine fire warning bell sounded and the first officer asked if he should shut the engine down. The captain replied in the negative and indicated his intention to try and “stretch the glide”. Seven seconds later again the ground proximity warning sounded and continued with increasing frequency as the aircraft sank below the glidepath. Again the captain called “Try opening the other one up,” but the first officer replied, “She’s not going. ”

A minute after the loss of power and 10 seconds before impact, the captain called over the cabin address system: “Prepare for crash landing, prepare for crash landing!” The stall warning stick shaker then began to operate as the airspeed fell below 125 knots and there were sounds of impact as the CVR ceased recording at 2024. 43 hours.

* * *

The aircraft was almost new, having been granted a British Certificate of Airworthiness only two months before the accident, and had flown only 521 hours. It was equipped with a new technology “glass” cockpit the same as Ansett and Australian 737s. The engine flight instrument system (EFIS) provided a solid state display of engine related parameters, replacing the earlier conventional array of electro mechanical engine instruments. The aircraft was powered by two CFM56-3C high bypass turbofan engines rated at 23,500lbs thrust for takeoff. Bleed high pressure air from the engines’ core compressors provided cabin air conditioning and other aircraft pneumatic services.

As it was clear that the accident had occurred purely as a result of the crew misidentifying the faulty engine and subsequently shutting down the wrong engine, a close investigation was made of all the factors which could have contributed to this situation.

Both engines, complete with their remaining cowlings and pylons were removed from the wings and taken first to the operating company’s base, and then the manufacturers’ factory for detailed examination.

All the damage to the No 2 engine was found consistent with the effects of ground impact, with no evidence of any external fire. Apart from this damage, the internal condition of the engine was consistent with that of a fully serviceable low time engine which had little or no rotational energy at impact.

The No 1 engine on the other hand, showed evidence of severe damage both internally and externally. All fan blades had suffered considerable hard object damage, only nine of the 38 retaining their full length. Although there was only slight evidence of hard object ingestion within the booster section, all stages of the core compressor had some hard object damage. There was also evidence throughout the engine of severe, circumferentially uniform, rubbing of blades on their tip paths and rotating seals. There was considerable sooting of the core, centred around a damaged combustion nozzle attachment. Sooting was also found on. the inside of the high pressure compressor case and inside the low pressure shaft. Most of the external fire damage was towards the rear of the fan case and appeared to have entered the inside through a split in the fan case to frame joint caused by crushing. There was no evidence that this fire had been driven by a fierce airstream.

The fracture surfaces of all the fan blade failures were examined metallurgically and with the exception of blade No 17, were found to be overload failures. The failure of No 17 exhibited several features consistent with fatigue and after cleaning, was examined both optically and with an electron microscope. The characteristics of the fracture were those of high cycle failure which had evidently originated on the pressure face of the blade, 1 to 1. 5mm from the leading edge.

Detailed examination and testing established that both sets of main engine controls, and the primary and secondary EFIS, together with their associated wiring, were all functioning as designed at the time of impact and correctly displaying input information The captain, who had been flying since 1964 and had logged over 13 000hours, had undergone conversion training on Boeing 737-300 aircraft 13 months before the accident. The first officer had done so seven months before the accident. Both had then done a further one day course on the Boeing 737-400 when the type was introduced to the company three months before the accident. This included an introduction to the type’s solid state EFIS. At this time however, no flight simulator equipped with EFIS was available to company pilots. At the time of the accident the captain had flown 23 hours on Series 400 aircraft and the first officer 53 hours.

Although the Boeing Company had issued an Operations Manual Bulletin on the CFM56-3 engine vibration, recommending a reduction of power when the aircraft’s vibration indicators read in excess of four units, this had not been available at the time of the captain’s 737 conversion training. However, pending an amendment to British Midland Airways’ operations manual, the bulletin had subsequently been issued to all 737 pilots employed by the company. It stipulated that engine shutdown was not required as vibration levels would decrease as the thrust lever was retarded.

Questioned after the accident, both pilots said they could not remember seeing indications of high vibration on the EFIS. The captain added he rarely included vibrati _ on the monitor in he believed them to be unreliable. The investigation established that while earlier types of vibration meters fitted to turbojet aircraft had a history of unreliability, later types such as that fitted to the 737-400 were more reliable and displayed only signals related to the rotating assemblies of the engines.

Analysis

As the aircraft was on climb to its cruising level, the flight crew heard an unusual noise, accompanied by vibration. The noise was heard in the cabin as a series of thuds, and the FDR indicated it was associated with the stalling of the fan with attendant surging of the No 1 engine. The accelerations recorded on the FDR were consistent with the reported shuddering of the walls of the forward galley. These symptoms were accompanied by a smell of fire and some visible smoke in the cockpit. Interpreted by the pilots as evidence of a serious malfunction, the combination drove them to act quickly to contain the condition.

Neither pilot apparently assimilated any positive indication of malfunction from the instruments, yet there was no evidence that these did not display the large variations in engine indications that occurred when the compressor was surging. The No 2 engine instruments showed no variations during this time, but it was not possible to determine why the first officer believed the fault lay with the No 2 engine. When asked by the captain whichengine was at fault, he half formed the word “left” before saying “right”. His hesitation might have arisen from genuine difficulty in interpreting the engine instruments, or he might have observed the instruments only during the six second period of relative stability between the second and third surges.

The captain said the instruments gave him no clear indication of where the trouble lay. Having disengaged the autopilot only seconds after the first surge however, his attention would have been on the handling of the aircraft and the flight instruments. He said he had judged the No 2 engine to be at fault from his knowledge of the aircraft’s air conditioning system. He thought the smoke and fumes were coming from the cabin, and as the air for the cabin came mostly from the No 2 engine, the trouble lay with that engine. While this reasoning might have applied to other aircraft the captain had flown, it did not do so in this case because some of cabin air in the Boeing 737-400 comes from the No 1 engine.

As both pilots reacted to the emergency before they had any positive evidence as to which engine was operating abnormally, their incorrect diagnosis must be attributed to their hasty action and not to any failure of the EFIS.

There then occurred an event that led both pilots into the fatal misconception that the action they had taken in such haste was correct. As soon as the No 2 throttle was retarded, the perceived symptoms of engine malfunction appeared to cease. Certainly the No 1 engine compressor surging and associated shuddering ceased at this time, though the FDR showed that the high vibration level did not. The smell of burning seems not to have intensified, and having failed to note the continuing maximum reading on the No 1 vibration indicator or the fluctuating fuel flow, both pilots became convinced that closing No 2 throttle had dealt with the problem.

Their failure to detect disparities in the readings of the engine instruments is perhaps most important in regard to the vibration indicators. Unlike the fluctuations that would have appeared on the primary engine instruments, the reading of the No 1 engine vibration indicator rose to maximum and remained there for three minutes. It is a matter for conjecture whether or not the crew would have noticed such a maximum reading on a conventional electro mechanical instrument, but there can be little doubt that it would have been easier to see.

While the introduction of the EFIS represents progress in terms of reliability and maintenance costs, it may be a retrograde step in the presentation of information. The most obvious difference between conventional electro mechanical engine instruments and the EFIS, is that full radius mechanical pointers have been changed to short, light emitting diode pointers moving around the outside of their scales. Much less conspicuous than mechanical pointers, they are less able to give the comparative information provided by the strong cue of parallelism. The EFIS appears to have been introduced without any thorough evaluation of its efficiency in imparting information to line pilots. The circumstances of the accident emphasise how important it is for such evaluations to take place before new developments are introduced to service. The problem was probably exacerbated in this case by the fact that the pilots were introduced to the EFIS before a simulator equipped with this system was available to them. They thus had no opportunity to become familiar within the interpretation of engine problems using the new system. As a result, the first time they saw abnormal engine indications on the EFIS was in flight in an aircraft with a failing engine.

Although the initial misidentification of the damaged engine may be seen as the start of the accident sequence, the captain’s decision to throttle back the No 2 engine did not in itself lead directly to the accident. It is likely that, if the No 1 engine had not ceased surging (probably as a result of the autothrottle being disconnected) at the same time as the No 2 throttle was closed, the accident would not have happened. It is also likely that if the No 2 engine had not been shut down, the accident would not have happened. For this reason some explanation must be sought for the captain’s action in shutting it down.

This engine was operating normally, but because the decision to shut it down was made after its throttle had been closed, the crew could no longer confirm its operation by comparison with the No 1 engine indications. There was no evidence on the CVR however that the crew consulted the engine instruments or attempted any other analysis of their situation before shutting down the No 2 engine.

In the aircraft’s “non-normal” checklists, severe vibration does not necessitate an engine shutdown, provided it is not accompanied by abnormal indications, nor does the presence of fumes or smoke in the cockpit. The crew did not therefore comply with published procedures in shutting down the engine. Nor did they follow the more general instructions in the operations manual or their training, both of which required them to evaluate all available evidence before taking this action.

However, it may be indicative of what was in the crews’ minds that, when the first officer notified the emergency to London ATC, he said: “. . . at the moment it’s looking like an engine fire,”and it is possible that the captain’s hasty action was taken in the belief that the engine was on fire. The operations manual contained no guidance on action to be taken in the event of vibration and smoke or fumes occurring together. Because severe fan shaft vibration can cause damage to the fan and low pressure compressor abradable seals, and fumes can enter the aircraft from these sources, it would have been prudent for the aircraft manufacturer to have included a procedure for such a contingency.

From the start of the emergency descent, the cockpit workload remained high. When, some seven minutes after the No 2 engine was shut down, the captain began to review what had occurred, the reading on the No 1 engine vibration indicator had reduced and there was no other indication of damage to the No 1 engine. When his review was interrupted by a further ATC transmission, he did not resume it, suggesting that he was confident of the aircraft’s safety. There can be little doubt however that the cockpit workload contributed to the crew’s failure to notice the abnormally high reading on the No 1 engine vibration indicator that was evident for nearly four minutes after the onset of the vibration.

Shortly afterwards when power was increased on the No 1 engine, it again registered high vibration which apparently also went unnoticed. At this stage the No 2 engine could still have been restarted, but by the time the No 1 engine lost power four minutes later, the accident had become inevitable. Although the first officer attempted to restart the No 2 engine when this occurred, flight conditions were by this time outside the envelope for a “windmill” start.

It is extremely unfortunate that the observations of the passengers and cabin crew who saw evidence of fire in the left engine did not find their way to the flight deck after the captain announced he had shut down the right engine. Cabin crew are aware that intrusions during busy phases of flight can be distracting for pilots, especially during an emergency. It is notable in this case that the cabin service manager made no attempt to go

to the flight deck until called. Even so, it has to be said that, had some action been taken by the cabin crew who had seen signs

of distress in the left engine, the accident could have been prevented.

The circumstances of this accident were such that the pilots were suddenly confronted with an unseen combination of symptoms that was outside their training or experience. Although it could be argued that such a combination of circumstances could not have been anticipated in terms of crew training, the accident raises questions concerning the level of technical training available to pilots of modern aircraft. It also illustrates the point that such training should include an appreciation of systems response under abnormal conditions, particularly where the symptoms have the potential to mislead under the stress of an inflight emergency.

Cause

The cause of the accident was that the operating crew shut down the No 2 engine after a fan blade fractured in the No 1 engine. The engine subsequently suffered a major thrust loss due to secondary fan damage after power had been increased during the final approach to land