11 miles out @ 530 feet AGL
 

Incident: Eastok A320 at Irkutsk on Feb 28th 2012, ILS malfunction leads to premature descent

Chilling....

Good thing the crew caught the error. I wonder if they were in IMC?

Scarey.Hope they survive their career.

Had an FO who would have landed in an orchard if I hadn't forced her to level out. Stupid is hard to fix sometimes.

not an isolated incident
 

Reminds me of the 2011 AAIB safety review report of an A320 [G-MEDA] doing an approach to Addis in IMC....

A British Mediterranean Airbus A-320 aircraft, registration G-MEDA operating as flight number LAJ 6711 on a flight from Alexandria (Bourg-el-Arab), Egypt, to Addis Abeba, Ethiopia, carried out two approaches using the Addis Abeba VHF Omni-Directional Radio Range beacon (ADS VOR) and associated Distance Measuring Equipment (DME). On the second approach the aircraft crossed over a ridge of high ground in Instrument Meteorological Conditions (IMC) and came within 56 ft of terrain at a location 5 nm to the northeast of the airport. As the aircraft crossed the ridge the crew, alerted a few seconds earlier by a radio altimeter (RA) height callout, carried out a go-around; at the same time the Enhanced Ground Proximity Warning System (EGPWS) generated a ‘TOO LOW TERRAIN’ aural alert
The investigation determined that the antenna of the ADS VOR had suffered water ingress and was not functioning correctly. The correct maintenance procedures for the ADS VOR/DME and its associated monitoring equipment were not followed.
The aircraft received erroneous information from the ADS VOR which was fed to the flight deck VOR display, the Flight Management System (FMS), the navigation displays and the EGPWS computer with its associated Terrain Awareness Display (TAD). A single common position 1
source error thus adversely affected all these apparently independent navigation/situational awareness systems.
The existing certification standards for the aircraft navigation systems were met but were not sufficient to protect against this problem.

Nothing to do with Addis Abebba or similar incident in Alexandria many years ago.

I would really like to know what was the airplane configuration by the time they realised it was a wrong signal. The report says 48 seconds, but that height is too low for such short time from platform height...

The ILS glideslope signal was obviously working bad, because you can never have a false glideslope below nominal glideslope, as far as I know. I have had such event myself at home base and we disconnected immediately. It was VMC, however.

GPWS will not warn you in such a case if you configure, until EGPWS TFC function is triggered at about 500 agl or less.

In IMC and with some fatigue, and at night, I would like to think that I would be aware of the wrong capture, but I cannot be sure until I have such a situation...

48 seconds is a couple of miles or so, which makes the descent point roughly 13 to 14 miles out, 3 deg glide path makes platform height broadly 4000 feet agl 'ish'. Descent from 4000 to 530 feet in 48 secs! I think even I would have noticed a rather high (!) ROD and overspeed conditions.

Would somebody else like to try juggling the numbers and come up with a rational explanation?

So assuming they were at platform of 3-4000' AGL at >11.5d it would take a large ROD to achieve 530'AGL 48seconds after GS* - possibly 2500-3000fpm.

This is similar to the AirInter A320 that hit the mountain on approach to Strasbourg. They were performing a VOR NPA but selected -3300fpm not -3.3°.

It just stresses the importance of checking that the aircraft is descending at a sensible ROD.

As for the comments regarding ILS signal validity (on AvHerald), there is (was) an AIC detailing this. Typically LOC is valid inside 25nm and GP either 10nm or 15nm depending on offset from centreline. Some installations it's more than this due to operational requirements. If you look on a Jepp plate, the vertical profile shows the range the GS signal is valid from - take a look at the ILS for Naples runway 24. You take the LOC from 7000' but the glide is only valid approx 10d - so DON'T arm the approach at BENTO!

A4

You beat me to it Beardy :)

A difference here is that NPAs are normally stabilised (constant speed) and ILSs are decelerated. Go down and slow down ain't pretty when you're hot and high.

Shows the importance for height checks, distance verses height. Not just for the NPA's but as a cross check to the trusty ILS.

I have seen a few strange captures in my time, luckily I was always awake enough to catch it and takeover manually.

One springs to mind in Canada when the signal was being affected by a snow bank (I think) anyway after capture it caused the GS signal to fluctuate up and down... And of course the autopilot followed it, so it was click click and back to manual!

[quote+M2002]because you can never have a false glideslope below nominal glideslope, as far as I know. I have had such event myself at home base [/quote] - err, say again?

Anyone check the Notams?

which is why you should check your position on any approach...and why the outermarker and its published crossing altitude is important (or other similiar fix)

and why RADAR usually says: you are five miles from the outer marker/8 miles from the field, maintain XXXX till established, cleared for ILSXX apch.

So much for autoland :)

Radar?????????????!!! Are you sure it was working? - not much use on this approach! Probably never crossed the OM..

I have a sneaking feeling the ILS was u/s........................not knowing the electric jet I don't know what pitfalls await there.

I faintly remember an event somewhere in Australia I think.
Here to carrier for the G/S was working, but the modulation was U/S.
That will cause the G/S indicator to always stay dead center regardless of where the plane is in the vertical profile.
The autopilot/autoland then did a to steep descend, that would have caused a crash if the crew had not been alert with height versus distance.
To autoland is not really intended for following a signal that never changes. My guess it it tries to reduce power/pitch down until it "sees" a movement on the G/P. If the G/P stays in center, the plane will continue down.

No idea if this is the case in Irkutsk, but its one explanation.

Btw, in the australian incident the G/P U/S was in notam, and the G/P iden't was suppressed from the carrier... Anyone remenbers to listen for that ;-)

Air NZ. APIA. 29 July 2000

http://asasi.org/papers/2002/Air%20N...ope%20Apia.pdf

Quick, somebody copy and paste the METAR!

BOAC;
NOTAMs, especially historical ones, are difficult to track down. The following site is one of the best I've found in checking current and (for terminals I've checked), historical NOTAMs. Keep the URL but replace the last four letters with another ICAO designator.

Metars, TAFs and NOTAMs

Another difficult item to track down is historical weather. "OGIMET" has, so far, been of enormous help: Formulario para peticion de mensajes aeronauticos , METAR/TAF Reports section.

"Celebrating TAWS Saves, but lessons still to be learnt" is a good read on EGPWS.

In addition to the CAA report to which nike refers, Air New Zealand has produced a 3-part video series, ( , , ), on this incident.

Absolutely.

So why do these cross-checks sometimes not occur? Rushed? Distracted? Singularly-focussed on a failed glideslope?...

The larger question is, why don't go-arounds occur when things do not seem quite right or when the approach becomes really unstable? There still seems to be a strong desire to continue an approach when all indications are that a go-around is required. Seen it in the data...why is this so?

PJ2

So 48 sec after GS capture he was at 530ft AGL and 11.5nm from the threshold. If we assume a speed of 160kts and a vertical speed of 800fpm then at GS capture he would have been at 1170ft AGL and 13.7nm from the threshold. I don't know their SOPs obviously but 1200ft at 14 miles would make me decidedly uncomfortable.

I'm pretty sure the ILS has been at the doctor's recently.

I have the July/10 ILS 30 with a 3.33 GP and an IF 8.7 from the field (and a spot height around 800' above the field) to the left of c/line......:eek:.). There is an OM at 2nm with a 754' ATE min crossing and a platform of 2800' ATE and descent at around 7nm from t/down..

BOAC

I mean I had an event when the G/S was erroneously too low. And that there are not false glideslopes below the nominal one. One thing is an antenna or whatever that is transmitting incorrectly and another is the sidelobes of the beam that produce false glideslopes. These are a 6, 9, 12 deg, etc, if I'm not wrong

They are all way above the nominal 3 degree slope.

If 11.5 NM out and @530 feet they may have catched a slope well below 3 degrees.

Anyone who is in the mood can figure it out....

BTW, sidelobs are not unusual for GS signals.....

too low!
 

This is all about airmanship. That certain quality that cant be assessed or monitored. Call it what you will, hairs on the back of the neck, a third sense that all is not well. Long may it last.

Somebody tell me again about those pilotless aircraft in our future....

BOAC

which is why the plane shouldn't have been that low...wait till the outer marker confirms crossing altitude before you start down below the OM alt

The problem may not be false glideslopes "below" the 3-deg one but of a failed glideslope, period, in which the indication centers and remains centered regardless of actual position.

Psychologically therefore, any and all effort to "maintain the glideslope" (high rate of descent, zero rate of descent, etc) is "rewarded", if the pilot isn't cross-checking the crossing altitudes and the DME as ssr has now twice pointed out. The A320 has a superb ND display which should show the position of everything, like the LOM, the airport and so on.

It is hard to imagine how such data can be ignored or just not seen in favour of following either a lobe or a failed glideslope, but it has happened and will continue to happen unless cross-checks are done. It occurred at Apia, at Irkutsk, and I've seen it, recently, elsewhere.

Astonishing, given detailed map displays and all, but there it is.

Indeed - yet another 'WHAT??!!! Too many these days.

Landing to the south at DFW in an MD80 one day with 1000 and 3 we were cleared for the ILS and the FO started a steep descent exceeding 1,000 FPM. The glide slope was pegged full down and we were in the clouds. Looking at the DME at the runway end and our altitude knew we were not high so told him to level off, the glide slope was wrong. 20 seconds later the glide slope snapped to showing us low.

I advised the tower of what happened and they said we don't have to protect the glide slope if the weather is better than 700 and 2. I said we were coupled so does that mean in the future we should only do the localizer approach since the glide slope can't be trusted? He said an aircraft crossed the ILS protected area because of the adequate weather. We didn't get below 1500 ft but it made me closely monitor what the ILS is telling you and if it is real or not.

BOAC, re, "Indeed - yet another 'WHAT??!!! Too many these days."

And it's not newbies, magenta-liners, innocents (like Bonin & Shaw, et al) who are doing this.

This is Guam, Busan, Charlotte NC, Smolensk, Crossair...all CFITs due to total absence of "checking six"..."where are we, what's my required altitude and when can I go down further?" It's just three components - slope, distance and height and they MUST agree, from intercept to touchdown.

And not one crew member in any of these accidents called for a go-around until it was too late. Why? Why especially, are First Officers so damn reluctant to speak up? CRM is thirty years old.

I always thought that beyond 12NM intercepting the GP (not the LOC) was pretty much a no-no - 'tis outside the operating parameters of the installation ... ?

bubbers 44 demonstrated what piloting is all about...always use multiple sources of information to confirm, confirm, confirm.

One time while instructing in a piper turbo arrow (over 30 years ago), a student selected the VOR for an airport instead of the ILS. He didn't identify...but the course indicator was on course and the GS was centered, the flag sticking out of the way. (VMC )

so he started down on glideslope (even though he was not on the ILS). IT was perfect...500 fpm, tracking the VOR like a champ...I took his hood off and said:

you got lucky, otherwise you would be dead...turned up the volume and made sure he knew the difference between the ILS ID and the VOR ID.

A little more than that at 1312 feet and that would be a perfectly normal platform HEIGHT of 400 meters in Russia, but not at 14 miles out.

Glideslopes are often 2.7 degrees there as well. I haven't yet seen the appropriate chart though.

The always on glideslope failure would be much more aparent if flying manually. Any pilot would quickly realise that the G/S is centerede no matter what. With AP, however, it will be more difficult.

- read post 21?

I have a counterexample to that theory in actual flight data. Hand-flown, high rate of descent towards....?, to "keep up with the glideslope"; - Blew the FAF alt and the thousand-foot-to-be-stabilized-by altitude and leveled off at about the same altitude as the Irkutsk A320 - a CFIT but for some visibility at 500ft AGL.

The lesson is, 1) it happens, and, 2) you cannot say to whom it will occur.

Everyone one of us who fly knows how to cross-check altitudes and distances while descending, and why. What we all think (and know) what should have happened, does not explain people's behaviour.

Preparation, a thorough approach briefing with a solid knowledge of SOPs and PF/PM duties and situational awareness through a cultured airmanship are significant preventative, error-trapping measures, as demonstrated by the Air New Zealand B767 crew at Apia. But it can happen to anyone.

There is no such thing as a "bad apple". The "other guy" is us, on any one day.

- wise words. It always seemed a little 'pansy' to check DME against GP intercept when you just KNEW the GP was going to be 'good', didn't you, and then there was always the OM check, wasn't there............................................

Maybe the airport DME was u/s, GP u/s whatever, but as said, with all the 'kit', including a WTFAI map in front of us..........................???? Would this a/c have TAWS with a complete database? That should have started shouting as they penetrated the terrain floor. Perhaps that is why they went round, although I would have thought the TF was well before 530'ATE at 11 miles?

The dangerous one is a GP fail 'on the slope' - there is a big clue if it is actually u/s earlier - the GP needle is 'centered' from or before LOC intercept.

I don't know how they can do this, maybe maintenance, but flying into Burbank one day in VMC conditions the glideslope was perfectly centered. It didn't matter if you were high or low. The ATIS said glide slope was unusable but it didn't help the airliner behind us who got a low altitude alert from tower and said we are right on glide slope. Some days you can't trust anything.

Bubbers44, I'm not sure I understand your confusion. ILS GS on ATIS as unusable and yet you watched it and presumably went low and high deliberately to see how it reacted (with or without a failure indication?), tower said you were on the correct approach path The guy behind you presumably flying in VMC couldn't maintain a constant approach to the runway are you saying that he too was using an unusable GS? Why would either of you want to do what you apparently did? What is it you can't trust the ATIS or the crew?

Like Alitalia 404 in 1990 ?

Like Pan Am 806 in 1974 ? or 812 in the same year ?

EAL 401 in 72 ?


Approach phase accidents
Last updated: 19 July 2012
Statistical information regarding the Approach flight phase. The number of fatal hull-loss accidents and fatalities per year is given. The figures include corporate jet and military transport accidents.

Year Accidents Casualties
2012 4 148
2011 11 230
2010 9 327
2009 8 231
2008 10 208
2007 3 103
2006 4 136
2005 8 114
2004 10 106
2003 9 336
2002 17 589
2001 12 231
2000 12 434
1999 11 127
1998 18 581
1997 16 726
1996 21 540
1995 22 550
1994 17 701
1993 14 215
1992 22 900
1991 13 498
1990 15 417
1989 22 785
1988 30 722
1987 18 402
1986 22 437
1985 11 359
1984 8 134
1983 14 582
1982 9 412
1981 12 187
1980 14 598
1979 22 312
1978 17 546
1977 22 512
1976 16 706
1975 19 625
1974 22 775
1973 28 887
1972 26 1058
1971 18 624
1970 20 305

Given the numbers in 2012 vs. say, 1972, methinks the old eagles had more problems than the newbies....(granted, the equipment has moved on, but...)


:D