FMS altitude input error led to autopilot glideslope drift

[Jeroen D]

Just picked up the 23 July-3 August copy of Flight international.

On page 15 it has an article with the above heading. Inshore a 777 was too low on Melbourne approach after a wrong entry of the runway threshold crossing altitude in the FMS. Crew noticed and went around, so all is well.

Still, it left me with some questions. I am sort of familiar with the displaced threshold function on the 744 FMC for take off purpose. Explained in Bulfer's guide on page 7.11

Not sure how this works? Is it 777 specific. I always thought the 747 and 777 were pretty similar when it came to the FMS functionality.

The article doesn't provide much detail, it does state that captain inadvertently entered the wrong runway threshold crossing and the FO did not validate against the approach charts or relevant manuals. Where is this shown on a approach chart? I'm familiar with the various  DA/DH etc, but this is new to me.

I would have thought that based on the glideslope the threshold crossing altitude is a given for a specific plane.

Anybody any more insights into this particular threshold crossing altitude functionality?

Jeroen

[Hardy Heinlin]

You mean the "glideslope" of a barometric altitude based VNAV approach, not the glideslope of an ILS radio station?


|-|ardy

[Avi]

It sounds like they did an approach with VNAV and not G/S. If you enter wrong (or change) altitude constraints in the FMC, you may screw the approach. In G/S the aircraft is ridding on a radio beam so it is different. They may notice the VNAV path didn't agree with the G/S and aborted.

Cheers,

[Jeroen D]

The article doesnt mention much detail, but from what I gathered they were using VNAV and only when the Co pilot noticed looking at the PAPI did they realize they were too low.

So where would you enter such a thresshold crossing altitude in the FMC? Last leg?
Thanks
Jeroen

[Hardy Heinlin]

Was it very cold in Melbourne?

[Jeroen D]

Not sure when this incident took place. Im traveeling and left the magazine home. But I dont think it ever gets below zero in Melbourne?
Jeroen

[Hardy Heinlin]

So where would you enter such a thresshold crossing altitude in the FMC? Last leg?

In the leg to the runway, i.e. where the waypoint name starts with "RW...". That's not the last leg if you have missed approach legs.

If there is no runway in the legs, use the leg which has the green E/D label on the ND.


|-|ardy

[Hardy Heinlin]

Here's a situ that is prepared for a test.

Right-click on the link below, and Save Link As ... as is in your Aerowinx/Situations folder:

Melbourne RNAV Z approach Rwy 34.situ

[Jeroen D]

Thanks, Ill give it a try later this week when Imget home. Where do you find this thresshold attitude. I looked for it an the approach plates, but either I'm overlooking it, or not recognising it, or its not there?

Jeroen

[Markus Vitzethum]

I'm not sure whether it has been resolved already which flight was affected.

If not, here's my contribution.   It seems like Flight International is discussing this incident.

Here is the final report by the ATSB.

Looks like the flight crew created a RWY34 runway extension placed at 2.8 NM with an altitude crossing restriction of 380' (50' AGL) placed on the extension (and NOT the runway waypoint).
See Fig.7 on p.11 of the ATSB report. I'd say a misunderstanding is quite possible with the wording used in the briefing paper.  :shock:

Markus

[United744]

Threshold elevation is stated on the ground chart. It might appear on the approach plate as well, but would need to check.

[400guy]

I was surprised to see (in the report) the amount of flight time in the past 7 days for one of the pilots.  I guess "30 in 7" doesn't apply?

[Jeroen D]

Thanks for all the responses. The ATSB report makes for interesting reading. Good to see that they did notice the deviation from the G/S.

I wondered if they should not have gone around, rather then level off and continue unitll they intercepted the G/S?

As far as I can see in the approach plate the MDH is 495' and they were flying at 500' AGL. Obviously, they had good visibility, but still. There is a reason for the MDH, things stick up.



Jeroen

[torrence]

Here is the final report by the ATSB.

https://www.dropbox.com/s/gcc6dkd565k10vm/Melbourne%20Visual%20approach%20Rwy%2034.situ?dl=0

Here's a situ file with the basic approach setup as described in the ATSB report (with threshold 0380 entered correctly).  Along with Britjet's excellent tutorials on LNAV/VNAV  and RNAV approaches this case presents a number of interesting aspects of VNAV use.   With the 0380 constraint in the CORRECT place - RW34, reset MCP ALT to 700 ft (listed as DA(H) for the VOR approach) and the descent proceeds from over SHEED (2500A) (reset MCP ALT to missed approach ALT 4000).   Almost immediately aircraft begins a 90 deg turn to the right to intercept the runway centerline.  Rolling out it's as if you are on a Disneyworld ride on rails, right on centerline at 1000 ft.  Disconnect automatics and land normally - should only take ~half the runway.  I think the ATSB report notes the runway length as one of reasons crew selected this approach rather than the ILS on shorter runway.

This is the same situ file except with the 0380 constraint entered INCORRECTLY in the runway 2.8nm extension waypoint (FA34 in the PSX FMS).

https://www.dropbox.com/s/r0gv19pzgqw6vvf/Melbourne%20Visual%20approach%20Rwy%2034%20%202.situ?dl=0

Everything looks deceptively the same as the approach proceeds - note there is no ILS on this runway, and the VNAV PTH indicator looks basically the same since it's tracking a descent path to 380 ft where you told it - FA34 - YIKES!  By the time the turn is initiated you should have noticed the V/S is a lot higher than it should be (~2000 fpm, trying to get down to 380 by FA34) and the ground is coming up fast.  You will get a SINK RATE warning also.  If you immediately disconnect automatics, increase throttle (it will probably be at IDLE HOLD) and level the aircraft you can do what the crew did and fly level at ~500 ft until close to the threshold and then land - er - normally.  There is not much time (about 30-40 sec) for all this to happen after passing SHEED and it was good that the crew had a lot of eyeballs out of the cockpit and the PF noticing the excessive VS in time.

Weird thing - for real world pilots to comment on:  Just for fun I ran the incorrect case and just let it run with no intervention.  I did NOT crash!  I  got the SINK RATE  warning, followed by PULL UP, PULL UP --- and then the engines spooled up, nose came up and sedately climbed on runway centerline to my MCP ALT of 4000.  Does the damn thing really do that in CMD mode with just LNAV and VNAV??  If it does, I'll bet it's only demonstrated in simulators.

The final icing is that when I ran the CORRECT case with no intervention in CMD - I came right in on the 3deg path and hit HARD (probably knocked a few fillings out), bounced ... and then spooled up and - slowly - climbed to about 2500 ft for some reason.   

Cheers,
Torrence

[Hardy Heinlin]

Hi Torrence,

great reproduction! Thank you! :-)

When the E/D is overflown, VNAV changes to ACT CLB mode and climbs with THR REF | VNAV SPD to the new CRZ ALT (i.e. to the missed approach altitude in this case).


Cheers,

|-|ardy

[ScudRunner]

I fly this visual approach to RWY34 at YMML quite frequently in PSX.

To be honest, I fly the first segment (SHEED - ASUKI ) of the descent in V/S ~1100fpm (depending upon wind components), commencing about .2nm before SHEED and then winding the descent rate off to around 800~900fpm as I am about 1/2 way around the turn onto final with the PAPIs in sight. 

LNAV works very well in flying the required ground track during all this.

I find this method works a little better than the VNAV PTH solution as it avoids the higher descent rates required if the aircraft commences descent after passing SHEED using VNAV.

Just my opinion of course...

An excellent review of this incident (by a 777 check captain) can also be found here... http://www.flight.org/sheed-mesen-melbournes-rw34-virgin-australias-boeing-777-flight-path-incident

cheers
S[ud

[Jeroen D]

Hi Torrence,
thanks for those situ's. Very interesting to see how this works.

Jeroen

[United744]

Interesting article!

Questions:

1) did both crew review the modification?
2) why didn't anyone spot TWO waypoints with the same hard altitude constraint?

Anyone got a link to the official report?

[Jeroen D]

Interesting article!

Questions:

1) did both crew review the modification?
2) why didn't anyone spot TWO waypoints with the same hard altitude constraint?

Anyone got a link to the official report?

See the link posted by Markus to the official report
Jeroen

[United744]

Hi,

Thanks - I saw it after I posted.

Interesting issue:

I select "34" in the ARR page, and can either:

* Just set runway extension distance, but unable to set an altitude at the RW34 waypoint ("INVALID INPUT"). It has an FMC computed altitude of 2500 ft at RW34, and I'm unable to change it

or

* Select VFR APP to the runway and alter the glideslope, but not the extended centerline distance (fixed at 8.0 nm).

That makes no sense to me... Why can I not change the extended centerline distance in VFR APP mode (ARR page), or the altitude in the non-VFR arrival selection?