Estate Autunno 2000

<div class=Section1>ILS Glide Path could be a serial killer - beware of it, always! <o:p></o:p>

Analysis of a Controlled Flight Toward Terrain [CFTT]<o:p></o:p>

english version by Cpt. Orlando Giacich

Let us recall the definition of Controlled Flight Into Terrain from which the definition of Controlled Flight Toward Terrain derives.<o:p></o:p>

<div style='border:solid windowtext .5pt;padding:1.0pt 4.0pt 1.0pt 4.0pt; background:#D9D9D9'>CFIT: An event where a mechanically normally functioning airplane is inadvertently flown into ground, water, or an obstacle.<o:p></o:p>

With the word "inadvertently" <o:p></o:p> this definition highlights the fact that impact is unexpected because of the pilots' lack of awareness of their position in respect of water, terrain or obstacles.

CFIT generally occur in instrument flight conditions<o:p></o:p> or at night with good visibility but with false perception of one's own position because of lights or other distorting images that prevail upon the correct instrument indications that the pilot refers to.

An impact with the runway as a consequence of a high rate <o:p></o:p>of descent is certainly not a CFIT, even if certain "experts" erroneously state the contrary.

<div style='border:solid windowtext .5pt;padding:1.0pt 4.0pt 1.0pt 4.0pt; background:#D9D9D9'>CFTT: An event where an airplane is inadvertently flown toward ground, water or an obstacle, and a prescribed safety margin is exceeded.<o:p></o:p>

(kindly translated by cpt.Orlando Giacich)</div><![if !supportEmptyParas]> <![endif]><o:p></o:p>

Premiss

It was in July of a few years ago that the AIDS (flight parameters printouts) of a B747 aircraft flying from New York (JFK) to Milan MALPENSA were examined. The reading highlighted an exceedance (flight safety jargoon for: excess) listed as FLAT APP. The relevant exceedance, in accordance with the procedures of the operational parameters surveillance program, indicated the following:

Radioaltimeter height below 500 ft<o:p></o:p> within a time interval of 6 minutes and 2 minutes prior to touchdown.

At that time I was in charge of the B747 sector of the Alitalia Flight Safety Service. I started an investigation on this incident two months later (September). The reason for this delay was due to different factors, among these the lack of knowledge of the criteria to be used to explain such an event by individuals responsible for the Operations Bodies.

In spite of the delay the following items were still available:<o:p></o:p>

· <![if !supportLists]><![endif]>AIDS printouts [the printouts of the flight tape recordings<o:p></o:p>] of the last 5 minutes of the flight

· <![if !supportLists]><![endif]>relevant meteo folder, Computerized Flight Plan (CFP), <o:p></o:p>Flight LOG

· <![if !supportLists]><![endif]>Vertical and horizontal paths graph reduced in scale and last 5 minutes flight path printout with geographic co-ordinates and graphs.<o:p></o:p>

Towards the end of September I also had at my disposal the Malpensa airport METAR at the time of the event.<o:p></o:p>

In the first phases of the investigation I deemed necessary to obtain further elements such as:<o:p></o:p>

· <![if !supportLists]><![endif]>the part of the printout relevant to the flight phase immediately preceding the event<o:p></o:p>

· <![if !supportLists]><![endif]>the recordings of the Ground Air Ground communications<o:p></o:p>

· <![if !supportLists]><![endif]>the recordings of the radar tracings <o:p></o:p>

None of these documents were available because of the time elapsed since the event.<o:p></o:p>

I considered the case of analyzing the recorded phase of flight through a computerized visual system but unfortunately the unavailability of a tape recording wouldn't allow it. These tapes are limitedly available within a period of 30/40 days<o:p></o:p>. In spite of all these negatives, thanks to the cooperation of the Flight Safety Service technical personnel, we were able to insert the data available from the printout into a program [named FAIR, many of you should remember it) and reproduce those four minutes on a VHS cassette. This cassette was subsequently used to illustrate that episode's sequence of events during the pilots' recurrent training meetings.

<![if !supportEmptyParas]> <![endif]><o:p></o:p>

HISTORY OF THE FLIGHT<o:p></o:p>

The flight was planned on Atlantic Track Yankee (TRK Y) which coincided with the minimum time track in respect of the available ones.<o:p></o:p>

Planned takeoff fuel was 85,000 kgs and a Block Fuel of 96,000 kgs was requested which included 4000 kgs extra fuel as heavy delays during taxi were expected. <o:p></o:p>

Actually the taxi delay was 40 minutes, burning<o:p></o:p> 1600 kgs.

Later on, the flight was cleared to TRK Z which added 36 minutes <o:p></o:p>to the planned flight time, equivalent to an extra 8000 kgs of fuel burn.

The requested additional fuel did allow the flight to be carried out on TRK Z and the flight landed with a delay of 34 minutes on the estimated time of arrival (ETA). The Remaining Block Fuel resulted to be 11,200 kgs against the 11,000kgs planned by the CFP.<o:p></o:p>

The Flight Level for the crossing was 3<o:p></o:p>50, the CFP planned one, and the step climb to FL 370 was disregarded because available only 1h30m before arrival, therefore the fuel economy would have resulted insignificant.

Weather conditions forecast on the destination and alternate airports reported no significant phenomena.<o:p></o:p>

Metar 0620: 320/2 5000 10br 18/15 1017 no sig<o:p></o:p>

The report above is presented to understand the picture of the situation that preceded the event and could constitute an element of evaluation of the factors leading to that event.<o:p></o:p>

<![if !supportEmptyParas]> <![endif]><o:p></o:p>

ANALYSIS OF THE EVENT<o:p></o:p>

The examination of the printout, being in a numeric form<o:p></o:p> (digital), didn’t allow an immediate observation and understanding of the cause/effect relation that lead to the maneuvers and corrective actions.

One thing it did allow though, and that was to assert the complete absence of a marker signal when the aircraft flew over the Middle Marker while perfectly stabilized on the LOC and G/P. <o:p></o:p>

It was impossible to determine if the autopilot was engaged or not since the FDR of the I-DEMV a/c would not allow to establish that circumstance with an acceptable degree of certainty. This was confirmed by the analysis of further printouts of that same a/c. <o:p></o:p>

I therefore devised a diagram as a function of the coordinates, in millimeters, available from the second printout that reproduced the ground tracking.<o:p></o:p>

Using such a method I was able to interrelate all the recorded parameters and visualize their evolution within each single feature of the flight being analyzed.<o:p></o:p>

<![if !supportEmptyParas]><![endif]><o:p></o:p>

Examination of the flight parameters printout.<o:p></o:p>

The only available printout, as previously mentioned, related the last 5 minutes of the flight, starting at 4000 ft msl (mean sea level).

Configuration was: Flap 10°, Landing Gear UP, IAS 195 kts<o:p></o:p>

According to the ILS indications the a/c was on the right of the LOC (almost full scale) and about ½ DOT above the G/P, according to the deviation indication.<o:p></o:p>

Initially the a/c turned right and the heading increased to 030°, it then started a marked left bank followed by the extension of the flaps down to 20° and immediately after, so was the landing gear while the left bank was reduced. Altitude at that point about 3000 ft msl

About 10 seconds later there was another steeper bank to the left, with a 28° Roll.

Altitude about 2700 ft msl

In the meantime the vertical attitude became negative with an increase of the descent rate to 3000 ft/m at 1850 ft msl [1150ft above ground].<o:p></o:p>

The left bank was reduced again but the tendency remained and the a/c continued its left turn.<o:p></o:p>

The G/P deviation was closing on “ON THE GLIDE” indications while the LOC kept showing left from the a/c position<o:p></o:p>

The high descent rate on the vertical speed indicator, confirmed by the non-adjustment of the thrust setting after the landing gear extension, caused the a/c to lose considerable altitude, taking it well below the ideal glide slope.

In this phase the G/P deviation indicator started to show a “FLY UP” tendency while a positive vertical attitude<o:p></o:p>, as to stop the descent, was detected. The minimum height at this point was 420 ft radioaltimeter from where a decisive thrust intervention was used and a climb to about 850 ft radioaltimeter was carried out.

In the meantime heading decreased to 300°.<o:p></o:p>

At this moment flaps were selected to 30°and, while the LOC started to leave its full scale position, the G/P indication showed the a/c practically back on the glide after a brief “FLY DOWN” indication. <o:p></o:p>

Thrust setting confirmed the manoeuvre.<o:p></o:p>

Simultaneously the a/c started a counter-turn to the right to stabilize on the LOC and after a light counter-roll to the left it resulted sufficiently stabilized on both the LOC and the G/P, between 500 ft and 400 ft radioaltimeter.<o:p></o:p>

The 140 kts IAS speed and the thrust setting resulted coherent and stayed such till touchdown.<o:p></o:p>

The wind during the approach, for what concerns that part of the printout, presented no values or variations able to particularly influence the flight. <o:p></o:p>

It was not possible to establish if the autothrottle was switched on or not, since that parameter was not reported in the standard printout.<o:p></o:p>

The autopilot operation, or its use, on that a/c presented frequent discrepancies between what was recorded and what the standard printout <o:p></o:p>would show, as detected from previous flights.

The OM signal (which was not flown over) could not be detected but neither the MM signal. This latter circumstance showed the lack of signal reception was probably due to the failure of the receiver on board since no failure of the ground system, concerning the MM, was reported in the NOTAMs.<o:p></o:p>

Finally there was no indication of the GPWS intervention in any of its warning modes.

<![if !supportEmptyParas]> <![endif]><o:p></o:p><![if !supportEmptyParas]> <![endif]><o:p></o:p>

<![if !supportEmptyParas]><![endif]><o:p></o:p><!--[if gte vml 1]><o:wrapblock><v:shapetype

id="_x0000_t75" coordsize="21600,21600" o:spt="75" o:preferrelative="t"

path="m@4@5l@4@11@9@11@9@5xe" filled="f" stroked="f">

<v:stroke joinstyle="miter"/>

<v:formulas>

<v:f eqn="if lineDrawn pixelLineWidth 0"/>

<v:f eqn="sum @0 1 0"/>

<v:f eqn="sum 0 0 @1"/>

<v:f eqn="prod @2 1 2"/>

<v:f eqn="prod @3 21600 pixelWidth"/>

<v:f eqn="prod @3 21600 pixelHeight"/>

<v:f eqn="sum @0 0 1"/>

<v:f eqn="prod @6 1 2"/>

<v:f eqn="prod @7 21600 pixelWidth"/>

<v:f eqn="sum @8 21600 0"/>

<v:f eqn="prod @7 21600 pixelHeight"/>

<v:f eqn="sum @10 21600 0"/>

</v:formulas>

<v:path o:extrusionok="f" gradientshapeok="t" o:connecttype="rect"/>

<o:lock v:ext="edit" aspectratio="t"/>

</v:shapetype><v:shape id="_x0000_s1108" type="#_x0000_t75" style='position:absolute;

left:0;text-align:left;margin-left:-9.95pt;margin-top:0;width:580.45pt;

height:435.25pt;z-index:1'>

<v:imagedata src="./Sp00-MXP%20luglio92_file/image001.wmz" o:title=""/>

<w:wrap type="topAndBottom"/>

</v:shape><o:OLEObject Type="Embed" ProgID="PowerPoint.Show.8"

ShapeID="_x0000_s1108" DrawAspect="Content" ObjectID="_1026042780">

</o:OLEObject>

<![endif]--><![if !vml]><![endif]><![if !supportEmptyParas]><![endif]><o:p></o:p><![if !supportEmptyParas]><![endif]><o:p></o:p>Analysis of the graph<o:p></o:p>

To put some order on the course of the events, as to reconstitute all possible behaviors of the crewmembers, after presenting the synoptic report, I carried out a separate analysis of the horizontal and vertical components of the glide path. I mean a separate visualization of the flight tracking on the horizontal and vertical planes.

This way it became possible to reach conclusions that the crew could more or less confirm, allowing me to formulate some queries, which the crew could clarify.<o:p></o:p>

<![if !supportEmptyParas]> <![endif]><o:p></o:p>

Horizontal plane<o:p></o:p>

The a/c tracking presented a heading diverging from the runway axis since the a/c had <o:p></o:p>crossed the alignment towards the right.

Point 1: <o:p></o:p>Distance to touchdown=11 NM, lateral deviation=1.2 NM

At point 1 a left turn was carried out with a bank reaching 14°.<o:p></o:p>

The bank rate was coherent with the speed of a corrective manoeuvre execution<o:p></o:p>.

The bank was gradually reduced and the heading, though decreasing, still showed a deviation.<o:p></o:p>

After twenty seconds or so, the turn was increased reaching 28° bank at point 2, from where it started to decrease down to 5°-10° for another 49 sec. <o:p></o:p>

Point 2: D=8.4 NM, lateral deviaiton=3.2 NM

(What happened in the vertical plane during these critical moments will be examined further on)

The a/c continued its left turn until stabilized on 305°-310° heading at point 3a

Point 3a: D=4.6 NM, lateral deviation 1.7 NM

After about one and a half minutes the a/c decisively turned right to stabilize on the runway axis.

There followed an instant counter-bank to the left probably due to the completion of the stabilizing manoeuvre and we find ourselves to point 4.

Point 4: D=1.6 NM, lateral deviation 100ft left

Remarks<o:p></o:p>

The resulting tracking presented four phases:

a) perception of the lateral deviation (point 1)

b) start of a decisive correction (point 2) subsequently gently reduced, again accentuated and once again reduced until phase c) is reached

c) re-entry heading stabilized (point 3)

d) determined right turn to steady the stabilization (point 4)

The query we then proposed for this analysis was formulated as follows:

Which of these elements had determined the handling of the a/c:

· LOC deviation (ADI/HIS)

· NDB bearings (NOV/MAL)

· ATC communications

· Runway or airport in sight

Vertical plane

While on a diverging heading, the a/c was on a slope coherent with the height/touchdown distance ratio.

At point 1, height is 3340 ft (QFE)

When flaps and landing gear were extended to 20° and down, no thrust increase was noticed to maintain speed and glide.

As a matter of fact, speed was maintained at the expense of the glide path.<o:p></o:p>

An increase of the negative attitude and vertical speed was noticed.

“Strangely enough” the G/P indication was close to the centre.

At point 2 the QFE height was 2310 ft, RA=2470ft

After 20-25 seconds of vertical speed down, a tendency to reduce the negative attitude was noticed, revealing the intention to stop the descent.

Point 3b: D=5.2 NM, height 320 QFE, RA=420ft

At point 3b, the climb up attitude with thrust intervention to stop the descent corrective manoeuvre was firm and responsive.

A climb to 780 ft QFE followed which brought the a/c back to the correct glide slope and definitely stabilizing the a/c on the G/P.

This happened at point 4.

Point 4: height 452 ft QFE (RA=422 ft), coinciding with LOC stabilization

Considerations

The handling of the a/c on the vertical Path was analyzed in four phases:

a) intermediate approach configuration completed (pre-final), speed steady and on the correct slope but no thrust intervention

b) vertical attitude being reduced and increase of the descent rate as a consequence of a)

c) perception of both the vertical deviation and loss of the correct slope

d) attitude and thrust corrections with correct slope gain

This time as well, the query was formulated taking into consideration the same elements used for the horizontal plane, and that is which of these elements:

· G/P deviation indication (ADI/HIS),

· Control of the height through fixes (DME or Radials)

· Runway in sight

· Visible cues on the ground

· Altimeter (QFE, QNH) and/or radioaltimeter indications, call outs,

have determined the 4 phases mentioned above.

<![if !supportEmptyParas]> <![endif]><o:p></o:p>

Interviews<o:p></o:p>

The testimonies that were gathered gave an answer to the questions that remained unanswered after the analysis of the printout and helped to collect other elements useful to the investigation.

The Captain of the flight under discussion, who was recently endorsed on the B747 after a long activity on the A300, collaborated fully to help shed light on the event. He was obviously convinced, after viewing the recorded parameters, that the excess values and, above all, the flight track deviation were much too emphasized compared to the deviations he and his crew perceived during that event. Besides, for what concerns the crew, the whole event became significant only when they were actively involved in the investigation.

The Captain recollected that the reasons that caused the deviation from the centre line of the localizer (ILS runway 35) were due to an oversight, that is the 350° value was not selected on the Captain’s HSI course selector.

The course selected in fact remained set on an unidentified value within the 1^st quadrant (030°-050°) after switching the functions from INS to VORLOC.<o:p></o:p>

The Captain remembered switching on the ILS function of the autopilot and expected the A/P to correct the a/c back towards the LOC even after having crossed it from left to right.

He stated that realizing the correction was not materializing, he decided to intervene on the controls overpowering the A/P forcing it to carry out a left turn..<o:p></o:p>

Shortly after, the F/O noticed the wrong selection on the course selector while the Captain was ordering the 20° flaps gear down configuration.<o:p></o:p>

The Captain recalled having checked that the glide path deviation indication over Novara was coherent.<o:p></o:p>

He stated that good visibility allowed him to see the terrain, distinguish and identify the airport area, encouraging him to continue his approach while regaining the LOC centerline.

More or less at that point, noticing the autopilot’s uselessness because of its inertia the Captain remembered switching the A/P off and flying manually, carrying out a short climb to re-stabilize correctly on the LOC and the G/P.

In any case, crew’s attention was drawn exclusively on regaining the correct lateral position rather than on decreasing the excessive descent rate or by the proximity of the terrain.<o:p></o:p>

The Captain recalled having the radioaids selected as follows:

NOV NDB on both ADF receivers, ILS on VHFNAV1 and MAL VOR on VHFNAV2 while approaching NOVARA; after crossing NOVARA the F/O selected MAL NDB on the ADF1 and RMG NDB on ADF2 and shortly after the ILS on VHFNAV2.<o:p></o:p>

The F/O actually confirmed the Captain’s testimony, particularly for what concerns the perception of the deviation from the correct path, less dramatic from what the printout revealed.<o:p></o:p>

He also confirmed that the good visibility encouraged them to continue the approach even in non-stabilized conditions having in sight the terrain and the airport area.<o:p></o:p>

The F/O also recalled that after crossing NOVARA, in trying to find out what was causing the a/c to deviate from the centerline, he realized that it was the wrong selection on the course selector. He also recalled selecting flaps 20° and gear down along with the radioaids selections, MAL and RMG on the ADFs and ILS on his VHFNAV.<o:p></o:p>

In this phase he was also engaged with the radio communications.<o:p></o:p>

The Captain and the F/O recalled the ATC advising them of their lateral deviation and asking them if they were able to continue their approach.<o:p></o:p>

After mutual consultation they replied affirmatively<o:p></o:p>

The F/O recalled that after having selected the ILS on the VHFNAV2 he noticed the LOC bar in full scale position while the G/P was in the less than one dot “fly up” position<o:p></o:p>

Both pilots confirmed that the F/E suggested to start a go-around but by that time they considered themselves in full climb-up phase with the LOC/GP correct position regained, so for that reason it was decided that the go-around manoeuvre was uncalled for.<o:p></o:p>

Both pilots recall that radar vectoring lead the a/c on a heading to intercept NOVARA directly and not to a point far enough from the FAP (Final Approach Point) as to allow them to capture and follow the LOC comfortably.<o:p></o:p>

They stated that the duties carried out immediately before the event were irrelevant, even if they admitted that their physical conditions were not at their best having spent the night flying. <o:p></o:p>

The pilots stated that they never heard any GPWS intervention as expected in its different modes and they recalled that radioaltimeter selections were carried out correctly.

<![if !supportEmptyParas]> <![endif]><o:p></o:p>

FACTS AND CONCLUSIONS<o:p></o:p>

The examination and the analysis of all the gathered elements allowed to establish:

1. The inefficiency of the a/c’s (I/DEMV) marker receiver.

Note: the receiver’s inefficiency was established comparing previous recordings printouts with the reports filed in Technical Log Book. In response to a couple of these reports, different components were substituted. The technical personnel requested that the equipment be tested on the next flight, a fact that never materialized since the a/c was frequently used on the New York leg notable to use runways without Outer Markers.

2. The recorded G/P (VHFNAV1) deviation signal highlighted A G/P indication practically centered close over NOV, and it stayed so while the a/c was deviating from the centerline going below the 3° geometric glide path.

3. There was a hint of a “fly up” indication only when the a/c was already climbing up and was turning towards the centerline. This indication, which reached almost 2 DOT “fly up” (a/c below the G/P), did not trigger the GPWS in its “glide slope” mode, neither soft nor hard.

4. The examination of the recorded parameters, comforted by the pilots testimony, allowed to establish that the type of flying in the phase preceding the interruption of the descent below the glide path was with the autopilot switched on. It also established that the helping cues for flying the a/c were mainly external, although the landing runway had not yet been clearly identified. Reference to the instruments was intermittent on basic elements (speed), occasional on others (height, vertical speed)

5. The Captain checked over NOV that the G/P indication was coherent with the expected reading.

6. The ATC advised the crew of its lateral deviation.

7. 350° figure was not selected on the course selector, an oversight that became a distraction factor.

8. The rushed vectoring by the ATC and the search for the reasons for the non-capture of the LOC, added to the A/P ordering the a/c to descend on an incorrect G/P, caused the delay for the pilots to realize the transition from a condition of secure position (overflying NOV, G/P centered) to a condition of indetermination.

9. There was a corrective action on the A/P from the Captain, probably due to the fact that on the a/c he transitioned from (A/300) such action was accepted as normal operational behavior. Unfortunately on the B/747 such action (overpower), though technically possible, is practically inefficient.

10. Maintaining thrust to idle, even in the 20° flaps and gear down configuration, a situation that should have alerted the Captain on the abnormal behavior of the autopilot on its vertical axis, showed that there still was a residue of operational behavior acquired and consolidated on the a/c he transitioned from.

11. The visibility was good and that contributed to delay more drastic and definitive interventions.

12. There was a huge task load on the F/O who in an interval of 30-40 seconds, the most critical for what concerns the loss of height, detected the wrong selection on the n.1 course selector, selected flaps to 20°, gear down, the two ADFs and the VHFNAV2 and therefore was kept busy waiting for the response to check the result of the frequency change.

13. There was an absence of significant verbal exchanges during the above mentioned time interval from the F/E, probably busy carrying out his checks on the lateral panel. There was nonetheless a request to go around, not peremptory but during the non-stabilized phase, which the pilots did ignore.

<![if !supportEmptyParas]> <![endif]><o:p></o:p>

CONCLUSIVE REMARKS AND RECOMMENDATIONS<o:p></o:p>

The probable factors that contributed for the event to happen are the following:

1. Radar vectoring was inadequate for an early stabilization

2. The missed selection on the course selector of the ILS on the VHFNAV1, which was probably a consequence of the previous factor.

Note: the relation between this factor and the operation of the a/p for tracking the LOC will be explained further on. Right now, this fact is considered mainly an element of distraction.

3. The absence of a radioaid on the field (DME associated to the ILS or the Malpensa VOR) to allow a continuous check of the glide path and give an evaluation of the distance to the runway during the initial approach or the radar vectoring.

4. The indication of the G/P practically on the center during the high descent rate.

5. The autopilot operation that, because of the previous factor, ordered the aircraft towards a negative attitude and consequently a descent rate not immediately detected by the crew.

6. The heavy task load on the F/O due to too many tasks to be carried out simultaneously at a critical moment.

Note: taken into consideration the period of 30-40 seconds of the unnoticed descent rate increase, it coincides with the time it took to carry out the various selections and communications, therefore leaving just a few residual seconds for an instrument cross-check which, at these conditions, can be all but efficient. From this also derives the improper custom of trusting immediate external visual cues, which are not that reliable after all.

7. The non consolidated method of proper use of the a/p, consequence of time restrictions on transition courses and flight familiarization, favored by the unconscious tendency to use practices learned on previous aircraft, the whole accentuated by the conditions of fatigue.

8. The state of reduced alertness attributed to a night spent flying.

Note: this factor is obviously present in any flight operations with similar characteristics and it is the most surreptitious since the pilots themselves often state (as in this case) that it is irrelevant.

9. A non-specified differentiation on the Operations Manual between the G/P check on the FAP (Final Approach Point) and the G/P check on the OM ( in terms of altitude reading and point overflight) that lead to the confusion that both checks have the same value.

Note: this aspect emerged also in the analysis of the DC9-30 disaster that took place in Zurich two years before.

10. The F/E’s insufficient monitoring, probably due to the short time interval during which the deviation from the right trajectory materialized, and other tasks such as checking the landing gear lights and the hydraulic systems on the lateral panel which probably drew his attention from the flight instruments.

Some of the above mentioned factors contributed to triggering the event while others, linking together in a complicated logical and chronological interrelation, determined a delay in having the crew recognize the unstable condition on the vertical plane and above all how dangerous it was indeed.

What dominated in that situation was the major worry of the lateral deviation, until a sudden evaluation, and the subsequent awareness-acquisition that there was sufficient margin to recuperate the approach without interrupting it, led them to carry on the approach and landing.

<o:p></o:p>

The whole crew ended up associating the kind of stabilization they were performing to the curved approaches in use on the HKK runway 13 or NYC runway 13.

Following this analysis I proposed the following interventions:

1. To make the Air Traffic Controllers (ENAV) aware of the problems that arise when radar vectors are given in a way to shorten the approach distances.

2. Make the pilots aware in not asking or courteously refuse such radar vectors in normal conditions.

3. Solicit ENAV to install on Malpensa a/p a DME transmitter associated with the ILS frequency and/or to the MAL VOR.

4. Reconsider the training programs and periods for the a/c transition where inadequacy to deliver basic knowledge becomes evident.

5. Plan adequate time and space in Recurrent Training sessions as to discuss how long-haul transmeridian-flights can cause time zone arrhythmia and loss of sleep, their influence on the long-haul pilot’s performance, in particular on their perception, mental process, awareness faculties and/or distribute scientific research, studies publications and articles on this problem.

6. Update the technical/operational documentation as proposed in a Company Circular (sent on 17 January 1992, educed from the Zurich accident analysis) using adequate language as to avoid any misinterpretation and establish the correct priority of those elements apt to assess a position within the airspace.

7. To attune the CRM (Cockpit Resource Management) with the various training occasions where the whole crew is involved.

8. Issue ITO (Informazione Tecnico Operativa) type information to technical crew members (F/Es) on the peculiarity of the emission and broadcast of the various categories (ILS CAT1, CAT2, CAT3) precision-approach-instrument systems’ signals.

9. Extend the exploratory terms of AIDS (surveillance of the operational parameters through the Flight Data Recorder).

Note: the event came to notice because of a 78 ft excess on the 500 ft RA limit where the automatic alerting system was triggered. A different corrective intervention or a terrain profile with a more pronounced depression below the a/c position, would have masked the event, the character of which would be significant enough to classify it as a CFTT.

10. Immediately proceed to the adoption of a more advanced GPWS or, in case the evaluation and the introduction program of such a system had been decided, accelerate as quickly as possible its fleet introduction.

11. Recall the pilots’ attention on the problems connected when flying, by day or by night, is based on outside visual cues. Make them aware of the poor reliability of such cues that could lead to possible if not peculiar ambiguities.

12. Use the event and its conclusive remarks in the recurrent Training sessions to inform and/or publicize its contents, through circulars, as to make aware the technical flying personnel of the problems this event has uncovered.

Conditions that helped this event to materialize<o:p></o:p><![if !supportEmptyParas]> <![endif]><o:p></o:p>

What I have submitted to you is taken from the formal report I presented at that time to the Flight Operations Director [the Flight Safety under him, an organization configuration that was disputed and that has been reinstated, I think] and to Captain Silvano Silenzi, who at that time was responsible for Flight Safety Policies of the Alitalia Group, under the General Manager (Eng. Pavolini).

For example, Captain Silenzi, in this case, totally agreed on the analysis I produced and the recommendations that ensued.

While I got no reply from the Operations Director who, among other things was responsible for not publishing the final report, fortunately the B747 chief pilot made it his duty distributing to the sector’s pilots the results of this analysis.

The most significant conclusions, which were repeatedly highlighted in the original report, were:

a) The fact that the Glide Path needle stayed more or less centered, no warning flag, for most of the descent down to almost 500 ft radioaltimeter.

b) The fact that the autopilot in ILS function failed to capture the Localizer which resulted full scale left on the HSI instrument.

These were the man-machine interface that made, as a matter of saying, the crew’s behaviour unstable.

In reality, factor a) depended on the environment and averaged with the pilots’ flight instruments cognition.

The ground ILS equipment has had an evolution that has increased the performance along the geometric axes [runway alignment and 3° glide path] but has deteriorated and made unreliable the signals strictly outside the tolerated and quite small angles on these axes, while older and obsolete equipment used to be more tolerant

In other words, the Localizer signal is generally reliable within an azimuth of a couple of tens degrees and not beyond a twenty, thirty miles distance from the antennas, while the Glide Path is reliable within 1.5° below and 2° above the 3° slope and, more important, not beyond 8°-10° to the left or right of the Localizer.

The fact that these signals, when an ILS selection has been anticipated, are received correctly induce to misinterpretation and the related indications must never be used operatively neither by the pilot nor by the approach controller (even worse by the terminal area controller) to expedite the traffic.

Besides, and this condition is important, not always the flight instruments respond to an incoherent signal with a red flag. The signal may possess characteristics such as to not trigger the needle (which remains almost centered) but strong enough to stop the warning flag intervention.

[once again we are forced to present the haunting events of the Zurich disaster]

This unreliability of the G/P signal has probably caused the non-intervention of the GPWS glide slope warnings. In fact from the printout it resulted that, for the duration of ten seconds, a position of the Glide Path needle should have triggered the (soft) “glide slope” warning.

But if, in the event as told, the pilots did not promptly detected the unreliability of the G/P signal, the autopilot fully “believed it to be true” and that compounded the picture of the situation.

Factor b) was, on the other hand, the first “treachery” feature of the autopilot that worked against the pilots. It was both unexpected and misunderstood.

During the period I was trying to piece the puzzle together, Captain Marcello Ralli, who was B747 deputy chief pilot at that time, had some trials on the simulator and subsequently, on a cargo flight approaching Malpensa airport in more or less similar conditions, autopilot in ILS mode but with the course selector on a selection different from the runway alignment (QFU).

Captain Ralli reported that moving the course selector the autopilot made an erratic tracking to the point of performing manoeuvres diverging from the expected capture sequence.

The technical explanation was that the autopilot interpreted the difference between the selected and the correct value as a wind drift correction and behaved in successive phases as to re-establish the correct tracking. This was a software peculiarity of most autopilots but, since the B747 avionics had passed their sell-by date, on this particular a/c the regaining of the tracking could only materialize if the course selection difference was of a few degrees and definitely after a great number of oscillations.

The more than 30° sent the system in tilt.

A worrying fact was that this technical peculiarity was not part of the sector’s pilots’ cultural knowledge, and in conscience it couldn’t be for the simple reason that in the transition program there was no hint on this subject.

On the other hand, according to us at that time, this technical aspect was not part of most, if not all, pilots’ basic knowledge.

In reality the compression of the transition courses, that passes on the information hastily and sparingly with the danger of ignoring essential aspects, is one of the consequences of the company managing approach founded on maximum economy but with the risk factor not being properly assessed.

In these conditions, scrupulously sticking to the operational standards and to the suggested task performance will certainly help the pilot avoid the minefield in which the machine’s behavior can cause uncertainty.

But, as we have noticed, the variables that could lead to oversights or errors are so frequent in our operational environment, that a conservative philosophy becomes the only defense strategy possible.

Another observation that, though not included in the final report, made me point out risk areas still actual today, was the reliance on outside cues as to determine position and operation of the flight.

If we drastically changed the outside scenario we would have two possibilities.

The first would be an instrument condition in total absence of outside cues.

In this case, I may add, the event would have never materialized. Attention to the intermediate phase would have been greater and, most probably, the oversight of the wrong selection on the course indicator would have never happened.

The second would be the total absence of daylight. The event in the night with good visibility.

The runway lights quite angled, far away from the immediate view, with the black hole ahead and its few scarce lights.

The pilots attention concentrating on regaining the deviation on the horizontal plane rather that the one on the vertical plane and referring to misleading lights outside.

An almost 3000 ft/m rate of descent at 10 seconds from the terrain!

Instead of a CFTT, a dramatic possibility of another CFIT that some body might have called pilot error!

<![if !supportEmptyParas]> CaptCaptCapt Aldo C. Pezzopane

<![endif]><o:p></o:p>

<![if !supportEmptyParas]> <![endif]><o:p></o:p>

</div>