On 15 December 1989, a GE CF6-powered Boeing 747-400 (PH-BFC) being operated by KLM on a scheduled international passenger flight from Amsterdam to Tokyo via Anchorage as KL867 was making its initial descent into Anchorage in day VMC when it entered a volcanic ash cloud. Almost immediately, ash dust entered the cabin air and the engines began to malfunction. An application of maximum thrust in an attempt to climb out of the cloud quickly led to the complete failure of all four engines. Restart attempts were eventually successful in restoring enough thrust to complete the flight with ATC assistance in respect of altitude and speed. No adverse consequences for the 231 passengers and 14 crew members on board were reported to have arisen as result of the event. However, a series of subsequent examinations of both the aircraft and its engines (the latter both on and off wing) subsequently disclosed extensive damage especially but not only to all the engines (which were replaced) and to the airframe, the avionics and the electrical systems.
This Summary article relies heavily on reliable and technically competent secondary sources, primarily presentations made by Boeing and US Geological Survey specialists at post-event meetings for which Agendas and/or proceedings were published and accessible. The reason for this exceptional action in the case of an event classified as an ICAO Annex 13 Accident by the investigating agency, the NTSB is that although a very comprehensive investigation was carried out and a correspondingly detailed 764 page long Final Report of that investigation was completed in 1991, and has been extensively referenced, it was never published. The extensive and directly associated “Docket” (which in the NTSB System supports all investigations) was (exceptionally) “not released” either. The released version of the Final Report which has just one page of text giving little more than a highly abbreviated factual account of the event was formally adopted by the NTSB on 30 June 1992 and is referenced at the end of this article along with the other sources used in its preparation.
An Investigation was carried out by the NTSB which identified the event as an Accident. Whilst no reference was made by the NTSB to the downloading of either the CVR or the FDR, it is evident from reliable secondary sources that FDR data was certainly downloaded and analysed as were other relevant aircraft system data and that recorded ATC data was also accessed.
The 51 year old Captain, who was recorded by the Investigation as occupying a “rear seat” in the flight deck at the time the ash cloud was entered, had a total of 13,000 hours flying experience of which just 100 hours were on the 747-400. Although this suggests that an augmented crew was operating the flight, no information is given on the other member(s) of the flight crew or which crew member was acting as PF during the events, it is nevertheless clear that at some early stage the Captain took over as PF for the remainder of the flight.
Following an initial Mount Redoubt eruption episode the previous day, a new one began whilst the flight was en-route some 90 minutes prior to the subsequent entry of the aircraft into the ash cloud. By time it was subsequently confirmed that the ash cloud had reached 40,000 feet amsl and having rapidly extended 154 nm downwind from the eruption on a true bearing of 037° in the general direction of the inbound 747. In preparation for arrival at Anchorage, the flight was initially cleared to descend from FL 350 (position 1 on the illustration below) to FL 250 and based on sight of cloud ahead altered course to avoid it. As the flight approached FL 250, the crew reported that a thin layer of altostratus cloud had been encountered and when the flight reported “reaching FL 250”, ATC responded by asking if the crew had “good sight of the ash cloud at this time”. The response to this question was “yea, it’s just cloudy, it could be ashes. It’s just a little browner than the normal cloud”.
However, having levelled at FL 250, it then “became very dark outside” with the flight crew subsequently reporting having seen “lighted particles” - St. Elmos Fire - pass over the flight deck windshields whilst becoming simultaneously aware of “brownish dust with a sulphurous smell” in the aircraft. Almost immediately (position 2 on the illustration below) the crew followed this with a transmission to ATC “we have to go left now...it’s smoky in the cockpit at the moment” which was approved immediately. Due to the presence of this dust, the flight crew donned their emergency oxygen masks and almost immediately, the Captain instructed the PF to begin a climb “to attempt to get out of the volcanic ash”. ATC were advised accordingly that the intention was to climb to FL 390 and that the aircraft was now in “black cloud”. After what appears to have been no more than 20 seconds of this maximum thrust climb, all four engines flamed out with the climb having reached FL 279 (position 3 on the illustration below) before an involuntary descent began. Two calls to ATC in quick succession advised of this and assistance in the form of “radar vectors” was requested. This call occurred only a minute after the crew had still been in VMC and had reported having the ash plume “in sight”.
The PF reported having noticed the airspeed decaying, “initially at a normal rate (given the airplane's attitude), but suddenly very fast”. All airspeed indications were then lost as volcanic dust contaminated the pitot system and at the same time, a stall warning occurred and the stick shaker was activated although without any signs of buffeting. The PF “rather firmly put the nose of the aircraft down to avoid a stall” and initiated a turn to the left in a further attempt to exit the ash cloud. A FIRE Warning for the forward underfloor hold was annunciated but aware that this was triggered by optical sensing rather than heat detection and was likely to be a result of ash/dust rather than smoke, no action was taken.
As the engines ran down, the generators tripped offline and all flight instruments were lost except for those powered by the batteries. During the time the engines were inoperative, the cabin pressure remained within limits so there was no automatic deployment of passenger oxygen masks. The crew stated that they had “elected not to deploy the masks manually because the passenger oxygen mask system would have been contaminated by volcanic dust in the cabin air”.
After a total of seven or eight restart attempts, over a period of 4-5 minutes, engines 1 and 2 were finally restarted at approximately 17,200 feet (position 4 on the illustration below) and an emergency was then declared to ATC. After a further 3 minutes with just engines 1 and 2 running, several further attempts resulted in engines 3 and 4 also being successfully restarted (position 5 on the illustration below). Ash contamination of the pitot static systems had resulted in all air data systems becoming unreliable but the left side altimeter was still indicating correct values. The rudder ratio light was on indicating contamination of this system which had made the aircraft difficult to handle, especially during the period that only engines 1 and 2 were operating. An approximate speed reference was obtained from the inertial guidance ground speed once the veracity of this had been confirmed with ATC.
The aircraft track shown against the limits of ashfall from the eruptions on both 14 & 15 December. [Reproduced from US Geological Survey Bulletin 2047]
ATC then provided radar vectors for a wide right-hand pattern runway 06 and approved further descent to 2000 feet. After emerging from the overcast at around 5000 feet, the Captain reported that he had then had the runway continuously in sight. However, during the approach, the forward view of both pilots was impaired by the "sandblasting" caused by the volcanic ash such that they were both “only able to look forward with their heads positioned well to the side”.
During the final stages of the approach, an "Equipment Cooling" overheat message was annunciated which the Captain elected to disregard because landing was imminent. Touchdown occurred 39 minutes after the loss of all engines and half an hour after they had all been successfully restarted. After completing the landing roll, the Captain cleared the landing runway and taxied the aircraft towards the assigned gate. On reaching it, he transferred control to the First Officer because his vision through the left hand windshield was impaired in such a way that he could not see adequately to perform the final part of the nose in parking procedure.
It was found from a review of satellite data after the event that although the ash cloud encountered had been forecast to move north northeast at 60 knots, it had actually moved at a speed of around double that thereby leading to the aircraft entering it in a position much earlier than ATC had been led to expect. It was noted that ATC radar had “only been able to detect ash for 5-10 minutes after the eruption” at its initial maximum density prior to the expansion into a less dense ash cloud which was nevertheless still hazardous.
Prior to the entry into the ash cloud, both PIREPS & ATC radar had indicated “either the presence of one of more ash clouds in the general area (north of Anchorage) and/or a large dispersion of the main ash cloud in a north easterly direction in the presence of a strong (100 knot) south westerly upper wind”. However, by the time ash cloud was encountered, the base of the overcast was around 5000 feet amsl and it was impossible for controllers to reliably detect the extent of the ash cloud with the available air traffic surveillance radar.
Damage to the Aircraft
In summary, the following, which cost in total around 80 million USD to rectify, was subsequently found:
- All four engines suffered extensive damage and had to be replaced. A detailed inspection of engine 1 found that, in the first stage of the HP turbine, ash had initially melted and then re-solidified on the leading edges of the nozzle guide vanes. The solid ash deposits extended along 75 percent or more of each leading edge with an average depth of about 1.5mm. A preliminary hypothesis suggested that the extent of the melted and re-solidified volcanic ash deposit on the high-pressure turbine nozzle vanes increased the operating-line pressure ratio of the compressors, resulting in engine surge. The repeated restart attempts then resulted in partial breakup of the deposit through "thermal shock”. This, in combination with the improved engine-surge margin at lower altitudes, facilitated the successful restart of all four engines.
- The entire pneumatic system, the air conditioning system and the equipment cooling system were heavily contaminated by volcanic ash. Most of the pneumatic system had to be subsequently removed, cleaned and reinstalled.
- The ‘sandblasting’ of both main flight deck windshields meant that they had to be replaced. The leading edges of the wings, winglets, and empennage were sufficiently ‘sandblasted’ to require replacement. Other parts protruding from the airframe such as antennae, probes, ice detectors and AoA vanes had also sustained sufficient damage to require replacement.
- The pitot-static system was both damaged and heavily contaminated. The pitot probes and static ports had to be removed and replaced and the system purged of all ash contaminant.
- The entire interior of the fuselage, apart from the area to the rear of the aft pressure bulkhead, had to be cleaned very carefully. This included all flight deck instrument panels, all circuit breaker panels, the whole passenger cabin including overhead lockers and the void above the cabin ceiling panels (including all systems within the latter) and the entire Environmental Control System (ECS).
- The entire electrical and avionics systems were contaminated and had been exposed to possible overheating due to loss of cooling air and so all electrical and avionics units had to be replaced.
- The smoke detection system was contaminated throughout and the entire system, including associated plumbing and ejectors, had to be replaced.
- The fuel, hydraulic and potable water systems had to be cleaned and checked for proper operation.
- All passenger cabin windows forward of the wing had been “sandblasted” and had to be replaced.
It was considered that “the absence of any thermal damage to the engines was the result of the combination of decisive and skilful actions of the crew and the responsiveness and built-in protection in the FADEC system”.
The Investigation noted the existence of a Boeing Operations Manual Bulletin (OMB) 747-B2-4 which advised that if volcanic activity was encountered, retarding engine thrust to Idle would reduce the build up of ash-related deposits in the engines and improve the engine surge margin. However, the information available does not make it clear if this Bulletin was issued prior to or after the event under investigation. In any event, none of the evidence available indicates that the flight crew involved were aware of its existence prior to the event.
A (published) Boeing paper presented following this event discussed the way in which significant ash ingestion into engines affects them and on that basis provided the justification for the guidance in the OMB quoted above. In part, this explanation was as follows:
The nature of volcanic ash particles is such that molten deposits accumulate rapidly upon ingestion, reducing HP turbine inlet guide vane area and covering turbine airfoil cooling holes. Engine power loss occurring shortly after entering volcanic ash clouds has been attributed to compressor operating parameter changes. The nozzle guide vane throat area is radically reduced, causing the burner pressure (static) and the compressor discharge pressure (static) to increase rapidly. This event causes the engine to surge. In addition, considerable erosion of compressor blades occurs based on ash particle hardness and density as moderated by particle impact velocity (proportional to thrust setting).
Volcanic ash contains materials of different compositions. Of importance to jet engine performance is the melting temperature of these materials. Glasses have a characteristic melting point of between 600°C and 800°C and crystalline particles melt between 1,100°C and 1,200°C. Since thrust settings above flight idle are likely to produce temperatures capable of melting these materials, it is extremely important to immediately retard thrust levers to idle (altitude permitting) if volcanic ash is inadvertently encountered. Reducing thrust to idle will also reduce compressor erosion damage.
In addition to engine damage or flameout, volcanic ash has a significant effect on the engine's ability to start such that in severe cases, a restart may not be possible. Starting ability is also enhanced by decreasing altitude and by allowing the hot section to cool. However, in the event of an all-engine flameout, an immediate restart should be attempted. Flight crews should be made aware that engine acceleration during start at altitude is very slow when compared to ground starts. Also, volcanic ash damage due to eroded compressors and deposits on fuel nozzles will further increase the time taken for an engine to accelerate during an in-flight start. Compressor bleed (engine and wing anti-ice plus all air conditioning packs) should be on during restarts and during subsequent operation to maximize the engine surge margin.
As a consequence of the above, Boeing, after consultation with the three large engine manufacturers, reiterated the importance of the following key generic points when responding to an inadvertent encounter with volcanic ash (an abbreviated rewording follows):
- Immediately reduce thrust to flight idle. This will lower the EGT, which in turn will reduce the debris build up on the turbine blades and hot-section components. EGT above idle will result in ash solids exceeding their melting point which is the main cause of engine malfunction/failure in ash clouds.
- If engaged the A/T should be disengaged to prevent the system from increasing thrust above idle and due to the reduced surge margins, the number of thrust adjustments should be limited and changes made slowly with smooth thrust-lever movements.
- Exit the ash cloud as quickly as possible and note that the shortest distance/time to exit may require an immediate, descending, 180° turn. Avoid any attempt to climb clear of an ash cloud.
- Note that the engine stall margin is increased by maximising engine bleed air use so airframe anti ice and air conditioning should be on.
- Engine continuous ignition should be on and if autostart is available it should be kept in the ‘on’ position.
- EGT should be monitored and exceedance of limits avoided - if necessary, shut down and then restart.
- In the event that an engine fails to start, try again immediately. Successful engine start may not be possible until airspeed and altitude are within the applicable air start envelope. If an engine is slow to accelerate, EGT will increase slowly too and remember that engines are very slow to accelerate to idle at high altitude, especially in volcanic dust and this may be erroneously interpreted as a failure to start (as may have happened in the investigated event when some autostarts were manually interrupted).
This guidance also covered the consequences of loss of airspeed indications if volcanic ash blocks the pitot system and recommended using the OM procedure for "flight with unreliable airspeed" or, if all airspeed indicators are unreliable or loss of airspeed occurs simultaneously with an all-engine thrust loss, shutdown, or flameout, then the attitude indicator should be used to establish a -1° pitch attitude. Inertial ground speed may be used for reference if indicated airspeed is unreliable or lost and ATC may be able to provide ground speed during an approach to land.
The Probable Cause of the event was formally documented as an “inadvertent encounter with volcanic ash cloud, which resulted in damage from foreign material (foreign object) and subsequent compressor stalling of all engines”.
A Contributory Factor was also identified as “the lack of available information about the ash cloud to all personnel involved”.
No related Safety Recommendations are recorded as having been made as a result of the Investigation.
A ‘Final Report’ of the investigation was adopted by the NTSB 30 June 1992 but it contains only a single page of text and some supporting factual data. The real 764 page Final Report was apparently never published but exists in hard copy and is understood to be amongst a number of old reports due for release online at some future date. Secondary sources which had access to relevant information available during the Investigation and were used in the preparation of this report were primarily the following US Geological Survey (USGS) publications:
- USGS Circular 1065 Program and Abstracts from the First International Symposium on Volcanic Ash and Aviation Safety Seattle (1991)
- USGS Bulletin 2047 Proceedings of the First International Symposium on Volcanic Ash and Aviation Safety (1994)
Information was also obtained from the public record of a Hearing under the auspices of the US Senate Committee on Commerce, Science and Transportation (2006)
Boeing Flight Crew Briefing Video about Volcanic Ash Avoidance
- For further information on this particular incident, see the NTSB Accident Report which contains brief details of the event.
US Geological Survey