737 crash at Schipol

[TSR2]

Thanks for that Rich. I new about the difference between Rad alt and Altimeter, but was thinking in the context of the Netherlands where the its about as flat as it gets, just didn't think it through from a country at the other endd of the scale ;-)

RE the AP, what I was imagining in my head was the aircraft following the GS / ILS so both height and position from the centre of the runway are known. So I was thinking that at some point on approach that it must have dropped below the GS and I was assuming it would know the distance to the runway, so I would have though that if it dropped below GS when it was still some distance from the runway it would have known something was up. The flaw in my thinking though is if it actually ignores GS in reality and sets a rate of decent based on the rad alt rather than the GS "beam" (Am I making any sense?)

[Prop Jockey]

Thanks for that Rich. I new about the difference between Rad alt and Altimeter, but was thinking in the context of the Netherlands where the its about as flat as it gets, just didn't think it through from a country at the other endd of the scale ;-)

RE the AP, what I was imagining in my head was the aircraft following the GS / ILS so both height and position from the centre of the runway are known. So I was thinking that at some point on approach that it must have dropped below the GS and I was assuming it would know the distance to the runway, so I would have though that if it dropped below GS when it was still some distance from the runway it would have known something was up. The flaw in my thinking though is if it actually ignores GS in reality and sets a rate of decent based on the rad alt rather than the GS "beam" (Am I making any sense?)

Yep I'm with you Ben - it uses the beam and deviation from it - the Rad Alt could fluctuate based on changes in terrain on the approach. Now you could use distance coupled with GS deviation ............................ but.......... You'd need to get a reliable distance measurement, which would depend on the type of approach you were flying. If its an RNAV(GNSS) approach then you have good data, but you could be flying a raw data ILS/DME where your distance is coming from the DME, which might not be co-located with the threshold of the runway you're landing on, or another option would be using the ILS without DME because its unservicable but the approach plate says you can still use the ILS anyway. So there could still be some circumstances where you wouldn't have a reliable feed for distance to give the AP.

Cheers

Rich

[TSR2]

Cheers Rich, that makes sense... still it goes to show that the best computers in the world still need the comon sense human brain to be following the proceedures to double check them.

Thanks for taking the time to explain.

[Prop Jockey]

No worries Ben - I think there might be protection further up the chain, in the cross checking of the left Rad Alt against the right Rad Alt - but I don't know that for sure (or if it happens, what logic is used to work out which Rad Alt is discrepant) or what impact a discrepancy would have down the line to the auto thottle. We'll see.

Cheers

Rich

[Chris Trott]

There are multiple checks when in an autoapproach or autoland on Boeing aircraft. The primary prevention of RA failure is pretty simple though if they're flying the plane right - they set the Decision Altitude in the Altimeter Window and the Decision Height on the Radar Altimeter. If the two don't hit at the same time (and there is an aural warning of "Approaching Minimums" and "Minimums" as you approach the DH to remind you to check), then you know something is wrong. If it's wrong - you ABANDON THE APPROACH. The fact that this captain instead advanced the throttle twice instead of going around and or deselecting the autothrottle was inexcusable. The fact that the FO didn't do his job and call for a missed approach was inexcusable as well.

I'm not one to usually throw crews to the wind, but this is basic instrument aviating and basic Crew Resource Management and has caused all too many crashes over the years. I can think of 2 Korean Air crashes and one Air China crash for this reason, but much of those failures are sadly due to the culture in much of Asia where the "Captain" is never questioned and CRM was eschewed for many years after the rest of the world had adopted the program.

[NigelC]

'Evening Ben.

The thing you need to get your head round is that nothing was wrong with the Autopilot. It worked fine.

The Autothrottle uses a signal from the RadAlt at 27' to go into "Retard" mode. When the RA failed at 1950', it went to -8'. This triggered "Retard" as the A/T thought the aircraft was about to land. Therefore the initial descent from platform height was flown with idle thrust. With a cloudbase of 700', the approach should have been stabilised by 1000'. The requirements for a stabilised approach include being on track, profile, configured and on-speed with the required thrust. By 1000' they may have met all these criteria bar the power requirement but as they configured the aeroplane, increasing the drag with each stage, the speed would have been bleeding off faster and faster. As the aircraft decelerated through Vref, the autopilot would have been struggling to maintain the glideslope, thereby increasing the pitch attitude and leading to more speed loss.

The aircraft was set up for a single channel ILS approach, which would have culminated in a disconnect of A/P and A/T at between 150 and 50' for a amnual landing.

On a dual channel approach to an automatic landing, you have two distinct and independent actions as the aircraft approaches the runway. 1, the autopilot initiates the flare manouevre and 2, the autothrottle retards the power.

Once the RA failed, they immediately received a "Too low, Gear" EGPWS warning which was disregarded. The approach was never stabilised (which means at 1000' they should have gone around). They never monitored the airspeed, and once they reached stickshake, they failed to disconnect the A/T before advancing the power levers, meaning that as soon as they let go, the power levers came back to idle. Pressing the TOGA switches would have overridden the Retard mode.

The failure of the RA may be a contributary factor, but there is no way the failure of a Radio Altimeter can make the aircraft stall. There were ample cues and warnings available to make the situation recognisable.

NigelC

[NigelC]

Oh, and by the way, the Captain was PM and the F/O was PF. So it was the F/O who should have been on the throttles and the Capt's shout for the GA. However they were visual through 700' so no reason to call for a go-around. The first time the crew recognised a problem was at 450' when the stickshake went off. It was too late to recover especially as A/T was still engaged and the manual application of power was removed when the Capt took control and the F/O released the power levers. They closed again as the A/T was still in retard!

If you look at the SB Boeing have released, the last paragraph basically says "Crews are reminded to fly the f********g aeroplane!"

[DaveB]

Hmm.. an excellent explanation of events Nigel. Clear, concise and easily understood by the layman.. at least a layman with a modicum of knowledge and understanding of such things ;-)

ATB

DaveB :tab: