Update on Air France 358 (YYZ overrun)

[Seeker]

TSB update

full report

[Mitch Cronin]

From the report: [emphasis added] "The autopilot and auto-thrust systems were engaged for the approach. Both were disconnected at about 350 feet above ground, from which point the crew continued with the approach visually and landed in accordance with the airline's standard operating procedures (SOPs). The aircraft then went slightly above the glide path and arrived over the runway threshold at an estimated height of 100 feet; the normal height at that point is 50 feet. At that time, the indicated airspeed increased from 139 to 154 knots. During the flare, the aircraft entered a heavy shower area, and the crew's forward visibility was significantly reduced as they entered the downpour. The digital flight data recorder (DFDR) recorded wind veered to 330 degrees true, causing a tailwind component of approximately 5 knots. The runway became contaminated with at least 1/4 inch of standing water."

Mitch's amateur analysis: Mother Nature sneezed and they were hooped.

From the report: "The aircraft touched down approximately 4000 feet down the 9000-foot runway. The spoilers deployed automatically after touchdown and the DFDR recorded that the crew applied maximum pressure to the aircraft's brake pedals. The pressure remained constant until the aircraft departed the end of the runway surface."

Mitch's amateur analysis: The report includes a chart which shows that from 50ft altitude, with 1/4 inch of water, a 5-knot tailwind, and with reversers operative, the aircraft needs 6614 ft to stop. This aircraft touched down (thanks to Mother Nature) at 4000 ft, and was at 100 ft at the threshold... extrapolating 2000 ft distance per 50 ft of hight assumes the aircraft had 7000 ft remaining when at 50 ft., so could have had a maximum of 386 ft. remaining when stopped....

From the report: [emphasis added] "The DFDR data show that the thrust resolver angle on the throttles' angular position began to change at 12.8 seconds after touchdown, and that the thrust reversers were fully deployed by 14 seconds. Maximum reverse thrust was observed on the engines 17 seconds after touchdown. The aircraft departed the end of the runway at a ground speed of 79 knots. It came to rest 1090 feet beyond the departure end of the runway."

Mitch's amateur analysis: ??

Mitch's wild guess: The very tired crew were thrown into doubt and confusion by the extremely difficult circumstances and took a few moments to recognize neither had deployed the reversers....?

For Jeff: http://www.tsb.gc.ca/en/reports/air/2005/a...te_20051116.asp

"The DFDR data show that the aircraft landed with 7500 kg of fuel; 4500 kg of trip fuel was required to fly to Ottawa."

[Kip Powick]

At that time, the indicated airspeed increased from 139 to 154 knots

I'm surprised the report did not include the G/S (Groundspeed) as that would be interesting to see..

MITCH.....Yes, to a degree Mothernature intervened, but 100 feet and approx 15 kts high, (assuming 140kt final approach speed), over the button is a bit much and then the late deployment of reversers really compounded the problem.

No one wants to see it but I think "Pilot Error" is going to be the end result ,or at least be a very big contributing factor.

[Vsplat]

Not bad, Mitch. I fly the airplane, and as Kip said, they were in a very tough spot crossing the threshold.

What may have complicated things was the presence of weather in the missed approach area. I don't know what they were looking at on radar, but based on the heavy CB activity over the crash site during the evac, I think they really did not have a good option.

This is sort of the reverse of the low energy go-around box. A hateful place for the crew. And a lot easier to get into than we armchair QB's might think. I'm already gritting my teeth for the inevitable re-appearance of the 'experts' we saw the day of....

Vs

[Homerun]

Kip, as I'm sure you know, published landing distances do not include any credit for reverse thrust. Your speculation is armchair quarterbacking at it's worst.

[Kip Powick]

Kip, as I'm sure you know, published landing distances do not include any credit for reverse thrust.  Your speculation is armchair quarterbacking at it's worst.

Homerun

Really ??? Based on this info please show me the error of my ways. Armchair quarterbacking???show me where. These guys had "auto-brake" and the FDR shows reverse being put on late...take into account the speed and height over the button and tyou have...??....all systems worked as advertised in the aircraft so where is my post "armchair quarterbacking" at its worst.??? Over to you.....

(PS..the actual chart is at the link in an above post. Sorry it won't line up when I Cut and Paste)

Aircraft Landing Performance Information

The length of Runway 24L is 9000 feet (2743 metres). Based on the Air France A340-313 Quick Reference Handbook (QRH), page 34G, "Landing Distance Without Autobrake," the following minimum distances would be used to bring the aircraft to a complete stop. It should be noted that, for a dry runway condition, the QRH shows a correction factor of "0" for landing distances with or without reverse thrust; therefore, for a given wind condition, these numbers remain the same.

ACTUAL LANDING DISTANCE

(from 50 feet above ground to complete stop)

Runway Conditions Dry Wet 6.3 mm (1/4 inch)of water

metres feet metres feet metres feet

No wind 1155 3788 1502 4927 1987 6519

5-knot tailwind 1264 4148 1682 5518 2265 7432

No wind, reversers operative 1155 3788 1397 4582 1768 5802

5-knot tailwind, reversers operative 1264 4148 1564 5132 2016 6614

[Guest woxof]

The aircraft then went slightly above the glide path and arrived over the runway threshold at an estimated height of 100 feet; the normal height at that point is 50 feet. At that time, the indicated airspeed increased from 139 to 154 knots. During the flare, the aircraft entered a heavy shower area, and the crew's forward visibility was significantly reduced as they entered the downpour. The digital flight data recorder (DFDR) recorded wind veered to 330 degrees true, causing a tailwind component of approximately 5 knots.

That would appear to corroborate what I posted the day after the accident...

"Within 30 seconds after his passing it was really heavy rain and strong winds.You

know how it is during a really heavy downpour. Air France levelled off it appeared and I

thought perhaps he is overshooting. Then he disappeared from view in the distance. 30

knot winds at least, I thought to myself. And they appear to be crosswind. "

However, how can one assume that they were a very tired crew. The flight from Paris leaves around 1300 local and arrives in Toronto about 7 hours later. Hardly a tiring flight if the crew was well rested for the flight which of course may or may not be the case.

[Homerun]

Kip, your statement about late application of reverse is what I take issue with. You also make it sound like they were stabilized at vref+15. The report explains it as a sudden gust, causing the airspeed to balloon 15 kts, combined with a wind shift to a 5 kt tailwind component. This happened very late in the approach. The flare on the 340 begins just prior to 30ft RA and this is where the throttles are retarded to idle. Compounding this problem was the fact they flew into a heavy rainshower as they entered the landing regime. Let the investigators finish their work before you start throwing stones.

[Kip Powick]

Homerun

Kip, your statement about late application of reverse is what I take issue with.    You also make it sound like they were stabilized at vref+15.  The report explains it as a sudden gust, causing the airspeed to balloon 15 kts, combined with a wind shift to a 5 kt tailwind component.  .  Let the investigators finish their work before you start throwing stones.

Have you actually read the report??

The approach speed was 140 knots, appropriate for the computed aircraft weight of 185 tonnes for the landing. The autopilot and auto-thrust systems were engaged for the approach. Both were disconnected at about 350 feet above ground, from which point the crew continued with the approach visually and landed in accordance with the airline's standard operating procedures (SOPs). The aircraft then went slightly above the glide path and arrived over the runway threshold at an estimated height of 100 feet; the normal height at that point is 50 feet. At that time, the indicated airspeed increased from 139 to 154 knots.

Re Reverse.....

The DFDR data show that the thrust resolver angle on the throttles' angular position began to change at 12.8 seconds after touchdown, and that the thrust reversers were fully deployed by 14 seconds. Maximum reverse thrust was observed on the engines 17 seconds after touchdown

I don't know about you but I was taught that the longer you wait to apply reverse thrust the less effective it is. Perhaps VSplat knows the recommended time frame to apply reverse for an A340 but if memory serves me...the T-rex was about 4-8 seconds after touchdown.

What report are you reading that says...The report explains it as a sudden gust, causing the airspeed to balloon 15 kts

If I have misread the Report alluded to in the link in this thread let me know. Throwing stones??? Com'on now I merely stated what I "thought" was a contributing factor and if you read the report I read, you will draw the same conclusions...nice to be bias toward pilot expertise but some times you have to let go and look at the facts...as they are presented up to this time.

[Specs]

Can somebody fill in some missing bits for me here? It's about the delay in applying reverse thrust.

Is it at all possible that crews, given an understanding that carbon brakes work and wear best when hot and unmodulated, would, for that very reason, not immediately select reverse upon touchdown?

[Kip Powick]

Normally....reverse is selected before brakes are applied. A normal landing on a dry bare runway that the pilots know is long enough to stop on use a judicial amount of both...at least that is what I have seen but reverse is always more effective at high speed and decreases in effect as the speed decreases..

In the accident case I would have thought reverse would have been selected immediately the n/w was on the ground or if really excited..a bit before the n/w was fully down.....but...I wasn't there and have not flown the A340...so take what I say with a grain of salt.

In this case I really don't think how carbon fibre brakes work crossed their minds and had they had a boat anchor up front they probably would have thrown that out as well.

The reason for the delay in selecting reverse is probably only known to the flight crew.

[Mitch Cronin]

However, how can one assume that they were a very tired crew. The flight from Paris leaves around 1300 local and arrives in Toronto about 7 hours later. Hardly a tiring flight if the crew was well rested for the flight which of course may or may not be the case.

I wasn't assuming Woxof... just reaching for possible contributing answers for that time between touchdown and reverser lever movement... or, a "wild guess", as I called it... note the question mark.

I understand the predeliction toward a defensive stance among pilots... and I wouldn't want to be one who'd jump to early conclusions... nor have I... but at this point there have been a number of facts related that have eliminated many possibilities, and shed a lot of light....

Among the pertinent questions remaining, it looks to me that the reason for that time passage of almost thirteen seconds will be the kicker. Quite possibly there are some good lessons to be learned by finding out.

[Vsplat]

I have not waded through the report yet, but this discussion of when reverse was selected is referering to the resolver angle. A bit of a caution here, that may not be the same as what the flight crew was attempting to select. Things such as waiting for an interlock to release on the outboards, etc may have influenced things. To be clear, this is nothing more that a discussion of possibilities.

Those who have read the report may already know what indications the aircraft had that it was on the ground. Not necessarily a no-brainer on a heavily contaminated runway.

As an earlier poster correctly noted, we don't yet know all we need to. The actual cause, human or otherwise, is not known, or the TSB would have released it. We'll just have to wait and let them get on with it.

Vs

[Request deviations]

There was a xwind componant at the time of touchdown. What if the crew realized once on the runway that they were not tracking straight down the runway. Reason for delay of selecting reverse getting that straightened out? Just toss it out for what it's worth. Maybe just a garbage thought on my part.

[Homerun]

The aircraft then went slightly above the glide path and arrived over the runway threshold at an estimated height of 100 feet; the normal height at that point is 50 feet. At that time, the indicated airspeed increased from 139 to 154 knots.

That is what I'm referring to Kip. The airspeed increased after they crossed the threshold.

Landing the 340 is a bit different as there are 3 distinct touchdown points. First touchdown is of the lower hanging wheels on the main bogey, at this point a derotation is started and the second touch is the other set of wheels on the main bogey, after the bogey tilts, and coincident with the centre gear. You then fly the nose wheel down for the 3rd touch as airspeed decreases. With autobrake medium selected, braking will begin after a short delay ( several parameters must be satisfied ), even with immediate selection of reverse, braking will have begun prior to reverse coming in, especially with the delay for the interlocks on the outer engines. The report says that max manual braking was begun immediately. This gives you much more braking than autobrake medium and indeed more than the RTO setting used only for takeoff.

With or without reverse, there was not enough runway left to stop on. That is the point I am trying to make.

[J.O.]

Excellent point Homerun. Once they touched down where they did, the rest of the information vis a vis reverser timing or braking effectiveness is just sauce for the goose. The airplane can't be stopped in the length of pavement they had left.

With the exception of the centre gear, the A330 landing is exactly as you have described, and if one were to time how long it takes from the moment of rear bogie touchdown until all wheels are on the ground and the real stopping begins, the number of seconds would probably be a surprise. Reverse takes a while to become effective on high bypass engines, and with the LGCIU logic, alot of things have to be lined up for the interlocks to release. Then the reversers have to deploy and then, and only then, does the thrust come above idle.

We have yet to benefit from the full analysis by the professional investigators, so it's premature to try to come to a final conclusion at this point. The one lesson I will take away from the current information is simply the reaffirmation of my appreciation for the many risks of continuing an approach with thunderstorms in close proximity to the landing area. Most of us have been lucky to get away with it at least once or twice. But, if you sit and watch a nasty summer cell pass by from the comfort of your living room, you get an appreciation for how fast the conditions can change from relative calm to "Holy Crap!". The temptation to continue an approach to the landing is strong, particularly when there are cells in the missed approach track. It gives me pause when I think of the number of times I've continued an approach on the basis that there were no cells between us and the runway, but there was lots of red in the missed approach track. I'll think twice before doing that again!

While the investigation may conclude that this is a result of crew error, the only way the investigators will gain my respect is if they make a deep analysis into the human factors elements which led to continuation of the approach. I would challenge anyone to honestly say they'd have done much differently than the AF crew when faced with all of the elements they faced that day, many of which we have yet to fully appreciate.

[Kip Powick]

HOMERUN

I don't have a problem with your explanation in your last post re braking etc. I have a problem with ...

The report explains it as a sudden gust, causing the airspeed to balloon 15 kts,

as I can not find any reference to that comment in any document. I don't fly the A340, never have, but how about this scenario..they are 100 feet high so they "push" for the deck, would that then not give a slight increase in IAS? Regardless of all our assumptions, I still feel, rightly or wrongly, and based on the info in all the reports that the drivers are going to carry a substantial part of the 'load".

regards

[Canus Chinookus]

"must have been a gust, monsieur"

[Mitch Cronin]

Can someone help me out here?... It sounds like some of you are suggesting the 12.8 seconds it took from "touchdown" (and exactly what criteria is used for that point in the preliminary report is not given) til the DFDR records the resolver movement on the thrust levers towards reverse is a result of waiting for interlock...?

I don't know the 340, but if the thrust lever assembly is similar to the 320's, and I suspect it is, the resolvers would register the first movement of the thrust levers to reverse idle as the sticks come back to that setting and then remain until interlocks are removed after full deployment.

This: " The DFDR data show that the thrust resolver angle on the throttles' angular position began to change at 12.8 seconds after touchdown, and that the thrust reversers were fully deployed by 14 seconds. Maximum reverse thrust was observed on the engines 17 seconds after touchdown."... tells me reverse wasn't selected until 12.8 seconds after, not that interlocks were not released until then.

So, I guess my questions are...: Do the thrust levers on the 340 move aft when reverse idle is selected, as in the 320, and... would the "touchdown" as recorded by the DFDR be from the same tilt/squat switches that enable selection of reverse?... And when can reverse in fact be selected? ...ie. are the thrust levers prevented from being moved into idle reverse until squat switches are made? nose oleo compressed?... main bogies tilted? ....what? ...if not until nose wheel is down, and the point of "touchdown" mentioned is from the first main bogie tilt switch being made, that could certainly account for a chunk of that 12.8 seconds. ...but if reverse can indeed be selected earlier, I'm still puzzled by the delay.

[J.O.]

All good questions Mitch, but unless the TSB runs some tests and publishes the results in the report, it's tough to quantify how long it "normally" takes to get full reverse on the A340. I suspect it takes more time than we think. Maybe next time one of our A340 friends sits in the jumpseat they can time it to give us a rough idea.

As Homerun has pointed out, the length of time to get full reverse is somewhat, although not entirely, irrevelant. According to the the manufacturer's numbers, touching down at the 4,000 ft point on that runway when covered with 1/4" of water left them with less than the required stopping distance in any case. Faster reversing may have led to less of an overrun, but I doubt that it would have been enough to prevent the trip into the ravine.

[Mitch Cronin]

Hi Jeff,

It's all the armchair style stuff of course, but I figure I might be forgiven for the interest... Hopefully you know I'm not trying to fry anyone... only to understand....

I didn't see that in that chart... what I saw was distance to make a full stop, from 50 ft up... with a quarter inch of water... a 5 knot tail wind... and reversers operating... was 6614 ft. .... The way I figured it, these guys would have had about 7000 from that point. What have I missed? ( due to wild and wacky winds, they inadvertently found themselves at 100 ft at the threshold with 9000 ft remaining)...

Yep, at that 100 ft, they were also 15 kts fast... so would they continue to hold that extra 15 kts while descending that 50 ft, or would it be bleeding off already? In any case, during the flare, the wind is reported to have swung, resulting in that 5 knot tailwind... I'm guessing they'd have lost some airspeed then, no?

Obviously it's incredibly hard to look at that whole dynamic from our "armchairs" and get a clear picture of what, when, and how... but I don't see the harm in having a go?

I found out from elsewhere that reverse on the 340 can be selected immediately after touchdown... in one operators manual, it suggests selection of reverse even prior to touchdown is possible, by saying "don't"... so when we read that the resolvers recorded no movement in the thrust levers for 12.8 seconds from touchdown (and remember also, I'm assuming a similar thrust lever assembly as the 320 here, where the thrust levers do indeed move when reverse idle is first selected), we then know that, for whatever reason, those seconds passed during which time reverse could have been selected, but wasn't. Yes, there's a delay on the outboard engines for interlock removal of both the outboard engines before either of them can be spooled up in reverse, but that doesn't seem to me to be the issue here.... Reversers were, according to the report, fully deployed 1.2 seconds after the resolvers began to record thrust lever movement, so that delay was minimal...

To me, it seems evident that the weather circumstances caught these guys in something of a lose lose situation.... Not being one to ever have to deal with that sort of thing, I can only imagine the difficulty of having to decide to continue with the landing when things are looking wrong so very late in the game... When they found themselves 15 kts fast and 50 ft high going over the threshold, was that the time to go-around?... As some of you have said, what they saw beyond in the missed approach path may well have been what axed that idea?... Those decisions can hardly be second guessed from these armchairs.... So maybe they were indeed going to run out of runway regardless....

But here's something:

The question of adequate runoff area has been discussed previously... surely minimizing the length of the overrun is valuable!... Their excursion beyond the runway began at 79 kts ground speed, and took them 1090 ft... and resulted in a fire... It seems likely to me that chopping any knots off that initial slide in the grass could have reduced that possibility (perhaps even saving the aircraft?) ...which an earlier application of full reverse is likely to have accomplished.

I'm acutely aware of the sensitivity many of you appear to feel toward any kind of blame being assigned before all the facts are in... and I'm trying very hard not to even accidentally hint at blame.... Please don't get me wrong... I'm only trying to zero in on the relevant questions remaining, given the information we already have. Discussion always seems a good thing to me... ...And I still see that 12.8 seconds as a key question that wants an answer.

Cheers,

Mitch

[J.O.]

All good points Mitch and far be it for me to say that any of them are invalid. I don't think it's armchair quarterbacking to raise these points, but I think we'd agree there's a whole lot more to the story. My concern is that if we discuss these elements without a full appreciation of all of the factors involved, we are left with an incomplete answer as to what transpired. Since this is public domain, I'm only trying to keep some balance for the sake of clarity.

To your point as to the distance remaining, your assumption of them having 7,000 ft remaining only covers for the extra 50 ft of altitude over the threshold. I agree with your numbers, given that the nomal threshold crossing height typically results in a touchdown around the 1,000 to 1,500 ft mark. In addition to that though, we must account for the fact that every knot of airspeed above the target adds to the landing distance. While I don't have the A340 tailwind numbers, a good rule of thumb is to add between 100 and 200 feet to the landing distance for each knot above the target approach speed. So, if they were 15 kts fast over the threshold, one can conservatively estimate that the landing distance would increase by some 1,500 ft, and maybe as much as 3,000 ft. The actual touchdown point of approximately 4,000 ft from the threshold bears this out. That extra airspeed would also make it more difficult to get the wheels on the ground due to a tendency to float and/or balloon the flare, particularly if the pilot applied the same amount of control effort into the flare as they would at the normal approach speed.

The tailwind component may cause the airspeed to decrease, but this also serves to increase the landing distance by a significant margin. Add in the cells on the far end of the runway and we start to see how right you are as to them potentially being in a no win scenario once the decision to land was taken.

I believe that the basic operating philosophy of the A340 reverse system is similar to the A320 and the A330, which I have flown. I can't comment further on the length of time it took to get reverse, for the reasons I stated above as to unknown factors. I still maintain that even if it were proven that there was a delay due to human factors, it would only have affected the outcome marginally.

As to the runway overrun area, yes there are things that can be done to reduce the severity of an overrun incident, and in cases like YYZ where there is a significant drop, I would be pleased to see some of the newer technologies applied. That would only serve to reduce the severity of the incident, but does nothing to reduce the probability of it happening again (which is already very low, IMHO).

One could argue that the time to go around may have been well before they crossed the threshold, if one assumes that the go-around area was covered in heavy thunderstorm activity. I only say this with the benefit of hindsight, and I am prepared to admit I have continued my share of approaches in similar conditions. On further reflection, I now believe that doing so increases the risk of the no win situation significantly. Who's to say that we may never have to abandon the approach for any number of reasons, from a blocked runway to a mechanical fault to a call from the back because someone is out of their seat. Every approach should include a suitable path by which it can be abandoned. I would hope that all pilots will take this as a lesson as to what can happen when doing an approach in severe weather, and that they'll look at the weather in the entire approach and go-around areas prior to deciding to continue. That for me is the biggest lesson so far.

Jeff

[Mitch Cronin]

Thanks for the response Jeff...

"On further reflection, I now believe that doing so increases the risk of the no win situation significantly. Who's to say that we may never have to abandon the approach for any number of reasons, from a blocked runway to a mechanical fault to a call from the back because someone is out of their seat. Every approach should include a suitable path by which it can be abandoned. I would hope that all pilots will take this as a lesson as to what can happen when doing an approach in severe weather, and that they'll look at the weather in the entire approach and go-around areas prior to deciding to continue. That for me is the biggest lesson so far."

.... That'll be why I think discussion is good. It promotes reflection... which might even save your hide.

That's interesting to hear that 100 to 200 feet for each extra knot of airspeed...

As for reverse... I guess I wanted learned confirmation that 12.8 seconds was unusually long, and worth exploring. I believe I'm learning that's too difficult to say, so, and in any case, I'll continue waiting for what more facts may be told... with the rest of you.

In the spirit of learning... it strikes me that if mistakes are made which then can be examined to find contributing factors... if people pay attention, that's a good thing, right!?

Sheoot, that's like turnin' mud inta gold, ain't it?

Cheers,