Description

On 31 May 2013, a Boeing 737-800 was radar-vectored onto the ILS LOC at Eindhoven whilst still above the GS after the aircraft was not adequately configured for descent. Soon afterwards, with AP and A/T engaged and APP mode selected, it pitched up rapidly after capturing the GS false upper lobe. As pitch increased to 24º with the AP still engaged after a go-around call, the stick shaker activated. Crew intervention began with pitch at 26.5º and a minimum speed of 97.5KIAS was reached soon afterwards. Recovery to controlled flight was achieved and a second approach was uneventful.

Investigation

An Investigation was carried out by the Dutch Safety Board. The event was not correctly reported by ATC or by the crew to their Operator and this delayed commencement of the Investigation, until Ryanair reported it after becoming aware as the result of a routine review of QAR data 4 days later. The relevant FDR and 2 hour CVR data from the incident aircraft were both overwritten but the full QAR data was still available and was provided by the Operator upon request.

It was noted that both pilots were Spanish nationals whose main experience had been gained on the 737. The Captain had 4260 hours total flying experience which included 3700 hours on the 737 and the First Officer had 670 hours total flying experience which included 410 hours on the 737. A trainee pilot undergoing initial type rating training was occupying the supernumerary flight deck seat in order to observe the flight as 'air experience' but he performed no functions in connection with the operation of the aircraft.

Instrument Meteorological Conditions (IMC) prevailed during the incident approach with a reported cloud base of 300 feet aal and the surface wind was a light north westerly. The upper wind between 2000 and 3000 feet agl, which would have been displayed to the crew, was subsequently estimated to have been from 010º at 30 knots. It was established that the First Officer had been PF and that military control which provides ATS at Eindhoven had issued their normal left hand base leg radar heading towards the runway 21 ILS LOC and issued a standard descent clearance to achieve a 6 nm capture at 2000 feet. However they had no information about the upper wind and all the earlier arrivals had self-positioned to the ILS so they had been unable to deduce its effect and make allowance for it. At the beginning of this base leg (annotated [1] on the recreated vertical profile and track shown in the two diagrams below), the aircraft was descending at 500 fpm but was above the vertical profile which would allow a 3º descent. The crew failed to correct this as speed was reduced and, coupled with the wind (010º /30 knots effect), remained high as the likely LOC closure range reduced.

The aircraft was given a closing heading of 250º at point [2] on the diagram before becoming established on the ILS LOC at approximately 4nm DME - point [3] on the diagram - still well above the ILS GS - and about 2nm closer to the runway than the no wind condition assumed by ATC would have produced. Once on final, with the AP and A/T still engaged and APP mode selected and with a tailwind component present, TWR cleared the aircraft to land. At that point, the Captain "informed the First Officer that it was very unlikely (that) a successful landing would be possible and they should prepare to make a go around". It was noted by the Investigation that the Operator's stabilised approach gate in IMC was 1000 feet aal which if not met required that a go around be flown anyway.

As the aircraft reached 1000 feet aal, approximately 1nm from the runway and around 600 feet above the ILS GS, configured for a flap 40 landing, the GS indicator was observed by the crew to 'come alive' and go to full 'fly up'. The Captain advised the First Officer that "it was probably a false glideslope and called for a go around" but the AP remained engaged. Two seconds later, after the aircraft pitch had changed rapidly from 0.5º down to 24º up and the A/T increased thrust from 30% to 90% in order to maintain the selected airspeed - 135 KIAS - the stick shaker activated. At almost the same time, the First Officer pushed the TOGA button once and the AP was disconnected. As the pitch reached 26.5 degrees nose up, the crew intervened. Airspeed had already rapidly reduced and reached a minimum of 97.5 KCAS two seconds later with the aircraft at 1267 feet QNH and 0.65 nm from the runway threshold. The First Officer continued with the stall recovery manoeuvre and after a further two seconds, stick shaker stopped for one second before restarting. At this point "the Captain helped the First Officer to reduce the AoA in order to regain airspeed resulting in the warning ceasing after (a further) three seconds". The speed was 103 KCAS and increasing and "the crew finished the stall recovery procedure and initiated a climb to 2000 feet to make a second attempt for landing". The aircraft was cleaned up and TWR were informed that a go around was being flown. Radar vectors to a second uneventful approach and landing followed.

It was noted that whilst the crew was aware as they commenced base leg that they were above the vertical profile and aware of "the need to increase the descent rate in order to capture the 3 degree glide slope signal, their predictions (flight path management) about where the 3 degree glide slope signal would be intercepted were incorrect and unrealistic". Crew action to achieve the necessary descent were wholly ineffective. It was concluded that "only a different routing or extensive additional use of speed brakes during the intermediate and base leg of the approach would have significantly improved the descent profile at the start of the final approach".

The Investigation noted that both prior to and following ILS LOC capture, the following indications were available to the crew to support situational awareness:

• The glide scope scale and pointer had been showing full scale fly down deflection throughout the approach until approximately 1.5 nm DME.
• The glide scope scale and pointer was not moving down from full scale fly up as it normally would when approaching the 3 degree glide slope from below.
• The altitude versus DME distance relationship was significantly different to that which would be seen on a normal ILS approach.

It was concluded that once the aircraft was on final approach, the crew should have been immediately able to appreciate that "a safe landing was impossible from their position" based on altitude versus distance to the runway and "a go around should have been initiated". It was specifically observed that the FCTM provides explicit guidance on how to intercept an ILS GS from above if this becomes necessary and concluded that "if the crew had better performed the crosschecks recommended in the Boeing FCTM, it would have been clear to them that a safe landing was impossible from their position".

The Investigation considered that the upset, which had subsequently occurred, could be described as a case of "automation surprise". It was concluded that "the high level of reliable automation in the cockpit can degrade pilot’s basic flying skills for flight path management".

It was noted that the Full Flight Simulators (FFS) available for pilot training were not programmed to pitch up to a false glideslope in the way that the aircraft had (predictably) done. It was also noted that although none of the Operator's aircraft had been fitted with a Vertical Situation Display (VSD) at the time of the investigated event, such a display was available as a retrofit and would have been likely to enhance situational awareness.

In respect of ATC, the Investigation found that TWR had no information or awareness of winds aloft and concluded that, given the surface crosswind affecting the only runway at Eindhoven, "the controller did not take into account the influence of the upper wind when deciding on the runway in use" It also found in respect of the APP radar service provided on the 'Arrival Control' control frequency that:

• the use of radar vectors to guide aircraft to Eindhoven Airport which did not follow the standard published approach was common practice at the time of the occurrence.
• the controller did not inform the flight crew that the impending radar vectoring would take the aircraft off the previously assigned route.
• the controller had no information regarding the winds at altitude and when the aircraft deviated from the planned track, the controller did not make any heading corrections and persisted in conducting the approach as planned.
• the aircraft was handed over to TWR without coordination despite the fact the aircraft was not fully established on the approach.

The formally stated Conclusions of the Investigation were as follows:

1. Eindhoven Tower Control, when deciding on the runway in use, did not have information available regarding the upper winds and thus did not take into consideration these winds in the IFR traffic pattern. The choice of landing runway resulted in the aircraft drifting on base leg and encountering a tailwind on final that influenced the rate of descent.
2. Eindhoven Arrival Control had no information about the upper winds. The controller did not take into account the influence of these winds when giving radar guided approaches. This resulted in a line-up too short for the final approach.
3. Eindhoven Arrival Control did not follow the procedures correctly regarding the following:
   • inform the flight crew beforehand when radar vectors can be expected for the approach;
   • intercepting an ILS glide path should be executed from below in accordance with published procedures;
   • active monitoring of the aircraft flight path during vectoring;
   • transfer of aircraft from Arrival Control to Tower Control without confirmation that the aircraft is established on the ILS or without coordination.
4. The Flight Crew did not take into account the influence of the upper winds. In combination with the aircraft’s high vertical profile and high speed in relation to the runway distance, a landing in accordance with standard operating procedures became impossible. The flight crew did not challenge air traffic control and postponed the decision to make a go-around. It is likely that the crew’s high level of confidence in the very reliable automation in the cockpit contributed to this.
5. The Flight Crew did not have proper guidance procedures to avoid false glide slope capture in relation to the distance to the runway threshold (during an autopilot coupled ILS glide slope approach from above, under instrument meteorological conditions).
6. The Flight Crew initiated the actions for the stall recovery manoeuvre according to the Boeing FCTM. A second stick shaker warning occurred after the control column was relaxed and the crew again correctly initiated the stall recovery manoeuvre.
7. During an autopilot coupled ILS approach the aircraft, flying at an altitude above the normal 3 degree glide slope followed the fly-up signal after crossing the 9 degrees false glide slope. This resulted in a nose high position of the aircraft causing the stick shaker warning to occur.
8. The Boeing 737NG Flight Crew Training Manual did not warn of possible false glide slope capture with a pitch-up upset during an autopilot coupled ILS approach. This resulted in an ‘automation surprise’ for the flight crew.
9. The Eindhoven occurrence was initially reported (to) and assessed by Ryanair as a minor event which did not warrant CVR and FDR retention.

Safety Action taken or notified as intended as a result of the investigated event was noted as including the following:

• The Dutch Safety Board decided to concurrently conduct a more general review of pitch up upsets due to ILS false glideslope and published a separate report on the subject on the same date as the Report on the investigated Serious Incident at Eindhoven.
• Ryanair has introduced a new horizontal landing gate for ILS interception from above which is specified as "the earliest of the following: the FAF, 5 DME, and 4nm in VMC operations / 5nm in IMC". Ryanair has also advised that it intends to implement the VSD modification on its 737 fleet.
• Boeing intends to implement a Flight Control Computer (FCC) modification which will limit aircraft climb rate when in glide slope mode and will "eliminate pitch up" when an aircraft captures a false glideslope. It also intends to add an explicit warning about the risk of pitch up to a false ILS GS to the existing FCTM text on the subject.
• Dutch Air Force ATC Regulations and the Local Operating Procedures affecting Eindhoven ATS have been amended to require that "whenever radar vectors are given for an ILS approach the glide path shall be approached from below".

Four Safety Recommendations were made as a result of the Investigation as follows:

1. that the Netherlands Minister of Defence should ensure that approach control take into account the effect of upper winds during radar vectoring of civil air traffic in military airspace.
2. that the Netherlands Minister of Defence should ensure that when making the choice for the active runway, the influence of the upper winds during the approach should be part of the decision-making in addition to the effect of the surface wind.
3. that Ryanair should ensure that its list of reportable occurrences in the company Operations Manual specifically includes stick shaker and pitch-up upset events.
4. that Ryanair should ensure that when (there is) doubt whether occurrences should be reported by crew at (their) first contact with (the Company) operator, (crew should) assess the occurrence properly (and consider the possibility that) CVR and FDR (data) retention (may be appropriate).

The Final Report of the Investigation was published 26 June 2014.

Related Articles

• Instrument Landing System (ILS)
• Loss of Control
• Situational Awareness
• Operational Use of Flight Path Management Systems
• Recovery from Unusual Aircraft Attitudes
• Unstabilised Approach: Vectoring Resulting in Intercepting the Glidepath from Above

Further Reading

• Pitch-up Upsets Due to ILS False Glide Slope, Dutch Safety Board, June 2014.
• UK CAA Paper 2004/10 Flight Crew Reliance on Automation, December 2014.