Pitch-up Upsets due to ILS False Glide Slope

[1stspotter]

The Dutch Safety Board launched an investigation into an incident at Eindhoven Airport. During this incident, the flight crew of a Boeing 737 were confronted with a 'stick shaker' indicating an approach to stall condition. The flight crew made a successful go-around, after which the aircraft landed safely. The existence of false glide slope reversal emerged during this investigation, and was further analysed. Two reports have been published on the basis of the investigations: 'Stick Shaker Warning on ILS Final' and 'Pitch-Up Upsets due to ILS False Glide Slope'.


The report in English can be found here
Dutch Safety Board | Investigations & Publication | Pitch-up Upsets due to ILS False Glide Slope (http://onderzoeksraad.nl/en/onderzoek/2047/pitch-up-upsets-due-to-ils-false-glide-slope)

An animation is here
https://www.youtube.com/watch?v=PxBoTMs_iEs

[BOAC]

I cannot understand the comment that this 'phenomenon' was unknown to flight crew in 2013. There have been warnings about it for years and the advice has always been to capture from below if possible, and if not to monitor range and height to identify the false g/s. Also arming 'APProach' is NOT a good idea in this situation until on the correct slope.

It does appear to be a poor piece of controlling from Eindhoven ATC!

Edited for Speeling

[Capn Bloggs]

Even the colonials are onto it! :D

http://www.airservicesaustralia.com/aip/current/sup/a14-h14.pdf

[fokker1000]

BOAC,

I couldn't agree more. There's a pretty good reason to monitor height/distance!
I always hard tune a DME when I'm doing a GNSS/RNAV if at all possible.

[Cows getting bigger]

A quick look at a European ATPL theoretical knowledge question bank (the stuff the teenagers learn) indicates that there is a possibility of false GPS lobes above the real GP.

Going back a few decades, a young CGB was told to beware capturing the GP from above.

[golfyankeesierra]

I don't think it is about the the concept of a false glideslope itself but more the autopilot behavior that can result from it.
I guess everybody will be aware of false glideslopes but not the violent pitch up that is the reason of the investigation..…

Is new to me anyway. I would expect a pitch down or high rate of descent on a false glideslope, not a pitch up.

And sometimes you end up high and have to capture it from above. Just happens..

[BOAC]

gys - in my 'time' we were warned that the land of reflected glideslopes was dragon infested, and it does not take much thinking to work out the effect of a reversed GS signal on an automatic system. Good that more now understand!

'Capturing from above' is part of the daily bread and butter for pilots - it just requires commonsense in use of the automatics - ah - if only....................:{

[Centaurus]

Going into Manila a long time ago and vectored high (4000 ft due noise abatement) to final ILS. Did the usual check of DME v height and it didn't tie in with the indicated glide slope. At 8 DME we should be close to 2500 ft if on slope. However, at that time we were above a low layer of stratus but at 3500ft due vectors. The ILS showed we were very slightly high on GS (quarter dot) when it should have shown full scale fly down.

We were Vref 40, no wind and on localiser with rate of descent well above normal for three degree slope. Became visual at 1500 ft with GS still showing slightly high but visual scene showed very high and obviously the glide slope we were on was around five degrees. All manual flying in the 737-200. As well, the T-VASIS was a dogs dinner all over the place. Landed and reported apparent false GS to ATC and crook T-VASIS indications as well. ATC advised they were aware of T-VASIS problem and that there had been recent earth tremors which had displaced the T-VASIS boxes. Hence incomprehensible light indications. However ATC SOP required the T-VASIS be switched on for all jet landings (regardless of serviceability!). No NOTAMs of course...

Some months earlier, an Air Manila (?) Boeing 707 had crashed short of the same runway with full flap, idle power and speed brakes extended. The 707 was a write off but minimal casualties I think. The captain swore he was on glide slope all the way from 15 miles out with no wind. Well he was but it was a false glide slope. But to maintain that glide slope he was forced into a high rate of descent about 1800 fpm hence full dirty up and idle thrust. They were fully visual too. Instead of going around, he pressed on regardless to see if he could make it OK. Asked later by investigators why he did not go around when well outside stable approach criteria, he said he thought the noise of the engines spooling up would scare the passengers :ugh:

Next day, after we reported our episode at Manila, a KingAir flight calibration aircraft tested the ILS and confirmed a flyable glide slope about 5.5 degrees (in other words a false glide slope). That is why the 707 pilot had to dive to fly the false glide slope and hard to believe the captain pressed on. In another life I flew DC3 radio navaid flight calibration aircraft in Australia. Part of the routine ILS test was to start at about 1500 feet on localiser at approx. 5 miles and maintain that height while observing glide slope indication. The test was to look for false glide slopes. It was common to see a false glide slope indication as we got closer to the threshold but these were at 50-70 degrees and of course quite unflyable. We did a similar test for false localiser indications by flying an arc across the localiser at 10 miles. False courses beyond 35 degrees from the localiser were normal and information published in the appropriate pilot documentation. There is no question in my mind that it is good airmanship to always be aware of DME v height cross check (where possible) during any instrument approach as well as a close eye on ground speed readings on short final.

The high angle false glide slopes were simply part of normal ILS glide slope characteristics and as you would need to be a Stuka dive-bomber to fly a 70 degree glide slope, they were of no flight safety concern. I was surprised to read on Pprune that the Eindhoven standard ILS calibration tests did not include a procedure for checking for false glide slopes as we did 50 years ago.

[latetonite]

One cannot invent sufficient SOP's to bypass common sense.

[tartare]

Jeez - I'm not instrument rated - but that video is a scary watch...
Was completely unaware of this.

[moosp]

Many glide slopes need to be captured when using basic autopilots with care and airmanship. As soon as capture occurs from below the glideslope the autopilot will receive a pitch up command as you are still about a quarter dot below it. The 747 Classic was famous for it, at airports where the glideslope was captured beyond 9 miles.

Thought it was common knowledge.

[BOAC]

It is important to make sure that the focus here is on the way the A/P responded (as designed!) to the reversed signals and not particularly on the rates of descent.

The other point is that there were a few clues missed by crew and ATC.

[safelife]

Afaik all 3 degree glideslope installations will produce a valid mirror slope at 9 degrees (6 degrees higher). I did fly one, in visual conditions, with a Dash 8. Worked...
Was a European airport with fairly modern equipment.

[10 DME ARC]

:confused::confused:
One of my first lessons on my UK IMC rating was to show a false G/P!! That's why most G/P's are only certified for 10nm and 3000ft!!

[MathFox]

Safelife, the point that the DTSB is trying to make is that some installations may show a false glidespath at 6⁰. Most likely depending on wheather and other traffic in the neighborhood.

Also, very interesting to know how the AP reacts to not properly intercepting the glidescope from above... might not safe your underpants, but could safe the plane.

[BOAC]

Can someone explain the rate of pitch change?It is but a short step from a 'fly down' demand when 'below' the false GS (and going down) to the a/c trimming back to 'hold' the false G/S and then rapidly into a rapidly accelerating 'fly up' demand if at all 'above' the false GS.Unless the rate of change of pitch is enormous I don't understand how this can happen? - I guess you would need to be 'there' to know. I suspect the rate of pitch was a touch high!

[Pace]

This has nothing to do with false glide slopes or whatever but all to to do with monitoring and being a pilot not someone carried around the skies with an autopilot.

If the aircraft is pitching up at slow speed to try and capture anything the pilot needs to notice that and correct the situation! its as simple as that

Again this is all to do with over reliance on systems WHICH ARE NOT 100% reliable and a move away from basic piloting skills which were more prevalent in the days where we were not lost in technological pilot aids

[BOAC]

Agreed, but as per post#12The other point is that there were a few clues missed by crew and ATC. and it surely does no harm to 'refresh' those who have forgotten and educate those who apparently don't know? I do not see anyone excusing the crew or ATC.This has nothing to do with false glide slopes or whatever- this event would not have happened WITHOUT a false glideslope.

[nitpicker330]

Not quite the same but do you remember the Air New Zealand 767 G/S incident in APIA?

Here is the link to YouTube

► 10:30► 10:30
NZ60 Erronious ILS Incident, Apia - YouTube
Dec 22, 2006 - Uploaded by musicalaviator
An Air New Zealand 767 encounters a malfunctioning ILS Glideslope system which forces it ...

[deefer dog]

Excuse me, but what is a Final Approach Fix if it is not a confirmation of correct glide slope interception?

And another thing; how could one continue an approach below or beyond the FAF if never actually passing through it?

[galaxy flyer]

Technically, an ILS doesn't have a FAF, it has a glide slope intercept altitude and a Glideslope on course indication at that altitude. (TERPS)

[DCS99]

"One cannot invent sufficient SOP's to bypass common sense."

I will use this phrase to confuse my co-workers in the back-office
.-)

[BOAC]

...and that is the point I made in post #2? Used to be known as airmanship..........

EDIT: Erm - here we go agan with time travel. This was posted in response to Pace (now post #25???), so unless he/she deleted and reposted the 'clock' is squirly.

[Pace]

this event would not have happened WITHOUT a false glideslope.

BOAC

That is not the point I was trying to make ;) The point I was making is we should never trust anything and always cross reference with things like altitude/distance points on the charts to confirm against the glide then in that sense a false glide is irrelevant in the sense that the pilots would notice the spurious indications and react accordingly.

[Onceapilot]

To be certified for use, an ILS G/S must be satisfactorily suppressed to avoid false G/S below the designated G/S. Therefore, the G/S must be captured from below because false glideslopes above the designated glideslope can, and may, exist.

[MathFox]

Onceapilot, capture from below should be the procedure, because it's safe.

However, in practice, ATC asks you to capture from above:= once in a while... a capture from above is also part of a CDA. (And people sleeping under an approach path appreciate any reduction of jet noise.)
But the automation can do very nasty things to you (fly the plane into stall :uhoh: ) if you fail to capture the glidescope from above. Good to be warned about that! :cool:

[Denti]

As BOAC pointed out it is not exactly rocket science. However this whole thread could serve as a nice ad for GLS approaches, no wrong glideslopes to capture there.

[Dan Winterland]

All the time I was a pilot in the RAF on several types, GS intercepts from above were prohibited.

[AtomKraft]

The Dutch are a bit too fond of being 'slick' IMHO.

Most of my captures from above have been at AMS, and I remember the Turkish accident was another example.

I remember one occasion at Schipol where the instruction was 'follow the glide slope, you'll get the localiser in a minute'.

The Dutch are very keen and very good. Sometimes a bit too much so.....

[BOAC]

Hopefully with the spread of VNAV/GPS approaches we will see less of this. It is, however, a useful reminder for crews on the 'perils' of false glideslopes, the need to think about autopilot useage, and the well-worn principle of 'know where you are' plus a lesson for Eindhoven ATC to take (more?) note of the winds.

[nitpicker330]

The Turkish 738 accident in AMS was caused by a erroneous Rad Alt driving the Auto Thrust to Idle.

[AtomKraft]

Nitpicker.

Yes, as they descended onto the glide slope from above.
I was just behind them.

[underfire]

The approach FAF/FAP location is set by the Approach segment OCS meeting the Intermed segment OCS with a 500' ROC. It is possible to have a FAF, then a turn to the final approach segment.

From the video, there were far more problems indicated. ATC vectored the ac to a short final, which the ac descent rate was not able to accomplish. They turned on final and were 1000' high on GS!

[Skyjob]

AtomKraft: Yes, as they descended onto the glide slope from above.

Sorry, no that was not the reason.

While intercepting from above may be common in AMS, the reason why the A/T was idle in that flight was NOT because of intercepting from above, but because the throttle retarded as its logic told it to do thinking it was on ground as it was indicating -8 (number by recall, sorry if slightly off).

Resultant the crash, all 737 crews were made aware of the existing system limitations and relationship between the A/T and RA1, which until then was only known to the few who had in-depth system knowledge.

Hence when RA1 failed, without a failure flag , but displayed erroneous readings with indications showing the aircraft was on ground (or actually a few feet lower than normal when on ramp), the A/T retarded as programmed to do when RAD ALT <~27ft.

Failure of the operating 3 crew members to monitor the approach and instrumentation led then to a stall situation, the aircraft's autopilot trying to maintain a 3 degree glide path while the engines remained at idle, speed bleeding off... the rest is history.

The accident report is available, please read this before posting your reason as stated (intercepting from above) for failing automatics behaviour which should have been monitored by crew who failed to intervene accordingly.

[AtomKraft]

Sky job

Yes, I quite agree about the reasons for the Turkish accident.

Just pointing out that capturing from above was a factor....and that it happens often in Cloggie land.

[BOAC]

Without wishing to prolong the agony here, 'the intercept from above' DID contribute to the AMS accident since the throttles were closed for some time due to the need to descend more rapidly which to some extent 'masked' the throttle closure due to the RadAlt failure.

[No Fly Zone]

Its all about professional level flying and situational awareness. The Pros know what (and and what NOT) to expect from their instruments and how to deal with it. If you're not aware of the potential for glide slope errors, and that your on board automation may not act properly, they you're just not up to standard. Another way to look at it might be, when on final approach, if ANYTHING does not look quite right, get out of there - and figure it out at a safe altitude. I think it is called a GO AROUND. You certainly can demand a do-over, but its smart to claim it before you're over the numbers. Duh? If one is flying at the commercial level, I would hope that you already know this stuff and that you're thinking is generally along the same lines. Being faced with a surprise when crossing the numbers - whatever the situation - is not professional level situational awareness. :=

[Oakape]

The Turkish 738 accident in AMS was caused by a erroneous Rad Alt driving the Auto Thrust to Idle.

This was a 'factor', not the cause. The cause was the failure by the flight crew to monitor & maintain the correct airspeed.

There is an aviation commandment from way back in the early 1900's that goes like this - "Often shalt thou confirm thine airspeed on final, lest the earth rise up and smite thee."

10 commandments of Flying (http://www.centercomp.com/dc3/1038.html)

https://www.ittc.ku.edu/~evans/stuff/commandments.html

[underfire]

OK,

One of the beauties of GLS is that it is a signal, not multiple beams. The signal can broadcast whatever you tell it too, even curves.

If the OAT gets too high, ATC can clear the ac for a different channel with a GP of 2.8 instead of 3. This is important with CAT III autoland.

Right now, most GLS procedures have 5 channels set up with different GPA's. (when I say most, that means the ones I have designed) (2.5,2.8,3.0,3.1,3.5)

This was to account for aircraft types, and also be used for temp extremes. Heavies are brought in at 2.8, while mediums at 3.0, and lights at 3.1 for wake sep. (guess a bit too much forward thinking there)

So, that is how temperature can be addressed.

No fly zone, yes, exactly, it appears they turned to short final 1000 feet high, theres the GA right there.

[Lord Spandex Masher]

I think it is called a GO AROUND.

Not sure hitting TOGA is the best first move in a pitch up upset!

[underfire]

OK, I see what you are saying, but in reality, it seems to work. I havent heard of any disco's on capture.

If it is hot enough that you are 3.4 on a 3, it is probably NA anyways!

You are basically tricking the box with what it thinks are the altitudes. Due to the heat, it still thinks it is at 3, when it is at 3.4, so the 2.8 should be an issue. We can get a capture pretty far out depending

This is similar to the baro-vnav at extreme cold, coded path is 3.5 to actually have a 3.0.

[400drvr]

Thanks for sharing. That was a very informative video.

:ok:

[McBruce]

Low level, high nose, 97kts...

Interesting read. A low level stall in the NG isn't a great place to be, sounds like they avoided it by the skin on their teeth. One final slice of cheese left there! Thankfully the crew done a good job in recovering the situation.

[nitpicker330]

No worries, at least you watched it!! :ok:

[WeekendFlyer]

Can't help but feel for the crew - they were put in a very sticky position by ATC, whose vectoring did not seem to take account of the wind and the need to maintain enough distance for final approach and GS capture. The animation shows clearly that the vectoring cut a huge segment off the normal path of the STAR, leaving the crew high and fast when they finally turned on to final approach. Why would ATC put an aircraft in that position, particularly in IMC?

Thank goodness for the stall warner and the crew responding properly to it!

[BOAC]

Can't help but feel for the crew - I don't think it requires much to come to a conclusion on the crew input, really. If the slide presented at around 4 minutes is representative, you have to ask, 2nm, over 1500'........no GP 'capture'?

[drfaust]

Everyone's learned about false glideslopes in the benches, but I'm afraid the false G/S nor the automation was the problem here. Sure an Airbus won't give you a stick shaker in that position, but regardless of that, I think the plane reacted the way it's designed to react.

I don't think the crew of that plane views this event as anything different than their own mess, coupled with ATC. Somehow I also doubt that they don't know all these facts we're all regurgitating here. It was a cocked-up approach with some unexpected autopilot/autothrottle behavior. I don't fly the 737 so whether it should be unexpected or not is hard for me to say. However the result was that they binned it and put her down safely the next time around, job well done in my book.

We all make mistakes and missed approaches, they recovered theirs quickly and professionally. I don't understand why the Dutch Safety Board had to conduct a whole investigation to confirm what every pilot worth his license already knows about false glide slopes. Might have been more useful to remind the aviation community that the phrase 'request delaying vectors' is not somehow an admission of utter incompetence, but is to be used where appropriate.

Or am I missing something?

[Onceapilot]

drfaust, you are correct. "request more track miles" or "unable" are good trigger words. Beyond this, busy feed-ins can be reliant upon slick pilot skill and, an incompatible feed-in by ATC might only reflect badly upon the aircrew unless a full investigation is carried out. Therefore, considerable pressure is placed upon aircrew to comply and "rescue" whatever they are given by ATC.:eek: