Jeju Air737-800 Crash at MWX (Muan International Airport, South Korea)

[bstolle]

I am not saying that these pilots should have gone around, I don'' t even know if they had sufficient thrust to sustain flight let alone climb).

All this to say that this 1500 / 1600m touchdown (that's about 55% of the runway) was really, really long.

If thrust would have been sufficient, there wouldn't have been a reason to land in the opposite direction.

I don't know how other airlines train a double engine failure landings (or if they train that at all), but in all companies I worked for, the aim point was usually ~2000ft past the threshold since you want to avoid undershooting the runway at all cost. That's of course with the gear down.
As mentioned before in this thread, the reason for the exceptionally long flare was most likely due to the retracted gear and that he didn't want to perform a hard landing

[Gabriel]

If thrust would have been sufficient, there wouldn't have been a reason to land in the opposite direction.

Understood and agreed.

Maybe not a technical reason, but (and this is a big but / if...)
- If they first lost #1 do to a first bird strike...
- And then had a second bird strike in #2 which surged a couple of times but then recovered...

They might have freaked out and wanted to put it down ASAP, making many mistakes in the process due to time compression, stress, and no time for checklists.

I am not even proposing this as a theory, just playing "what if".

[B_C]

Understood and agreed.

Maybe not a technical reason, but (and this is a big but / if...)
- If they first lost #1 do to a first bird strike...
- And then had a second bird strike in #2 which surged a couple of times but then recovered...

They might have freaked out and wanted to put it down ASAP, making many mistakes in the process due to time compression, stress, and no time for checklists.

I am not even proposing this as a theory, just p[laying "what if".

But there were many incidents on bird strikes like this since 2019. Probably this was more severe than past cases and experiences.

[bstolle]

Agreed, definitely a possible scenerio. The main problem at very low altitude is, if there's too little thrust available, that you instantly know that there will not be enough time to troubleshoot and read the required checklists, let alone take all the required actions.

[bstolle]

But there were many incidents on bird strikes like this since 2019. Probably this was more severe than past cases and experiences.

I'm only aware of 2 double engine failure cases due to birdstrikes, one in 2008 and one in 2009.

[Gabriel]

I went a step further to calculate the speeds and deceleration...

I used this photo to estimate, using perspective calculation, the location down the runway where they touched down, considering the point where the #2 engine skid mark starts:

https://i.abcnewsfe.com/a/95993e7f-200b-48fa-b709-c9ccc33ee5ca/embed-crashsite_1735567056999_hpEmbed_23x16.jpg

Then I used this video to calculate several distance/time coordinates, between the touchdown and the crash. For the touchdown I used the first puff of white smoke that comes from #2 at almost 2 seconds into the clip.

https://www.youtube.com/watch?v=d7vY55cwnlU

Then I advanced the video frame by frame, taking into account that it is streamed at 30 frames per second, to "measure" the "time after touchdown" of different events.

Then I used google maps to measure the distances from the departure-end threshold (except the touchdown distance that I measured form the first picture).

https://drive.google.com/file/d/1aO055SZ5hFrEtkPN_AzSbOJiFQXI8tTQ/view

With all that I made a raw-data table and graph (shown in blue) with distance past departure-end threshold and times.

I estimated that they touched down with 1223m of runway ahead, and measured the embankment to be 260m past the threshold, for a total of 1483m.
From touchdown to slam we have almost 15 seconds, giving an average speed of 100.4 m/s or 195kts.

But that's average speed. Of course the speed was faster at the beginning and slowed down along the skid.

[Gabriel]

So I used the intermediate distance/time points to calculate a best-fit constant-acceleration model.
That model produced a deceleration of 8.2 knots per second (which is not bad at all: with that a plane touching down at 120 knots would take some 15 seconds to stop and would roll out for 555m so with a normal touchdown 500m down the runway the total runway used would be just 955m)

Reconstructing the distances and speeds of each event time using the constant acceleration model, I obtain the table shown in green. Note that "time" is not in the table because time is the input, and I am using the times from the raw-data (blue table).

That model estimates a touchdown speed of 255 knots (which is way too fast even with flaps zero and no slats) and a crash speed of 134 knots.

The distances estimated by the model can be compared with the raw-data distances to check how well the model fits the raw data, and it fists not perfectly but quite well.

Finally, the pink table is totally based on the raw data, and I used it to calculate the average speeds in each segment between events.
If the model was perfect, the average speed (marked in the graph with a dotted line with rhombic markers) would be the average between the max and min speed of each segment, which in the graph would be like the dotted line diving the green rectangles in 2 equal parts.
We can see a good match for the 1st and 4th (last) segment, with the dotted line dividing the rectangles in 2 equal parts. But the match is not so good for the 2nd and 3rd segments. However, if we join the 2nd and 3rd segment as one, we can see the that dotted line would go pretty much through the center, so it seems that the error in the 2nd segment is compensated by an opposite error in the 3rd segment. That may be due to some inaccuracy in the measurement of the time or the distance of the event "threshold" that divides the 2nd segment from the 3rd, taking some meters or fraction of a second from one segment to give it to the other.

Overall, I am quite happy that the constant-acceleration model quite matches the raw data. Now, how good is the raw data to begin with? I did my best, but I am not super confident on the accuracy of the raw data, especially in the estimation of the location of the touchdown point from the photo and how well (or bad) it syncs with the puff of smoke in the video.

Final thoughts:

Assuming that the plane did touch down at 255 knots and that the deceleration was 8.2 knots per second constant, how much would it have skidded to stop?
The answer is sad: 2100m, which even if it touched down 700m down the runway (a normal touchdown point) would have resulted in the plane stopping by the end of the runway, and touching down 960m down the runway (a very long touchdown, right at the limits of acceptable if not a bit past the limit) would have still resulted in the plane stopping just short of the embankment.

(Note, all the speeds are groundspeed, but given that the wind was almost calm and that the airport is at sea level, the indicated airspeed must have been very similar)

[bstolle]

That model estimates a touchdown speed of 255 knots (which is way too fast even with flaps zero and no slats) and a crash speed of 134 knots.

Nice calculations. 👍 The video shows a rather normal, if not high AoA at the final stages of the approach (flaps 15 approach AoA should be around 7°)
Flaps 1 V2 at MLW is 150kts. I doubt that they were a lot faster than this.
Below 200 in any case.
Deceleration seem to be very slow. Maybe one engine was still delivering more than idle thrust? Similar to TAM 3054?

[Evan]

The S-turns just before landing makes me think they knew they were too fast and high coming in but the fact that they didn’t go-around makes me think they had a dual-engine strike.

What can any pilot do at that point? Reminds me a bit of the A330 that landed in the Azores after fuel exhaustion, but without the glider pilot experience.

[bstolle]

The S-turns just before landing makes me think they knew they were too fast and high coming in

What can any pilot do at that point?

What S-turns? The small turns on short final were just for alignment with the runway. A much higher AoA and they would have stalled. Again, AoA seems to match the configuration nicely.

At this altitude? Nothing. AC 143 sideslipped AFAIR, but at a way higher altitude.

[Evan]

Why? What would be the reason? This doesn't make any sense.

Not maybe. You have to. It's that simple.

No, you discontinue the approach and why should an engine failure destabilize the flightpath?

I've never heard of multiple birdstrikes which occured on the same flight at different times with different flocks.
Furthermore if you start to expect an double engine failure in a birdstrike or icing environment, you should stay at home. There are failures which aren't trained and don't make sense to be trained.

Yes, there are failures that are so rare and remote that that aren’t trained and the PIC must improvise and make the correct life or death decisions, and this is one of them. It’s not that simple at all. Procedures exist to make anticipated failures more simple and to defend against pilot error. Where procedures do not exist, pilots must use their best judgment. In a situation like this, where there is a high threat of ingesting birds and losing engines, and you are configured and lined up on final, you don’t simply go-around. You must use your best judgment. If it were me, my judgment would tell me to continue the landing because the environment is unsafe to fly in.

[thor]

8:59L: Flight 7C-2216 pilot reports bird strike, declares emergency "Mayday Mayday Mayday" and "Bird strike, bird strike, go-around."
9:00L: Flight 7C-2216 initiates a go-around and requests authorisation to land on runway 19
9:01L: Air traffic control clears the aircraft to land on runway 19.
9:02L: Flight 7C-2216 touches down on the runway about 1,200m down on the 2800m long runway

just 3 minutes between the go-around and the final landing, looks like pilots are in really really desperate situation.
double engine failure? smoke in cockpit?

one of the survivor, a flight crew who was still conscious when retrieved from the wreckage, said it's a normal landing and no brace for impact announcement were made.

[bstolle]

In a situation like this, where there is a high threat of ingesting birds and losing engines, and you are configured and lined up on final, you don’t simply go-around.
If it were me, my judgment would tell me to continue the landing because the environment is unsafe to fly in.

Ah, I see. So if you experience a single engine failure you prefer to endanger the life of all your passengers, your crew and even people on the ground by
not completing the neccessary emergency checklists (e.g. engine shutdown in flight)
not calculating the correct speeds for the single engine landing (or even worse, you land with the originally planned flap setting and speed)
not calculating the new landing distance (risking an overrun due to the increased landing distance)
not performing an emergeny briefing for the passengers.
These are just a few points to consider and chances are that quite a few additional will come to your mind in this situation,
The company will not be happy with your decision and at least there will be a few additional sim sessions and retraining for you IF every thing goes to plan.
If it doesn't, even a small item like a blown tire during the rollout is enough so that things might turn pretty ugly for you fast, as an official accident investigation might start.
You are risking your job and (not only) your life for an event that will not occur with a 99.99% possibility. Your choice and yes, not an easy one to make.

[Gabriel]

Ah, I see. So if you experience a single engine failure you prefer to endanger the life of all your passengers, your crew and even people on the ground by
not completing the neccessary emergency checklists (e.g. engine shutdown in flight)
not calculating the correct speeds for the single engine landing (or even worse, you land with the originally planned flap setting and speed)
not calculating the new landing distance (risking an overrun due to the increased landing distance)
not performing an emergeny briefing for the passengers.
These are just a few points to consider and chances are that quite a few additional will come to your mind in this situation,
The company will not be happy with your decision and at least there will be a few additional sim sessions and retraining for you IF every thing goes to plan.
If it doesn't, even a small item like a blown tire during the rollout is enough so that things might turn pretty ugly for you fast, as an official accident investigation might start.
You are risking your job and (not only) your life for an event that will not occur with a 99.99% possibility. Your choice and yes, not an easy one to make.

Except that...

a) Boeing does encourage the PIC to decide whether to continue with the approach or go around, and
b) You have a frigging 2800m runway ahead.

https://encrypted-tbn0.gstatic.com/images?q=tbn:ANd9GcQuYtGXXBUD8jGjN0c4lAGxXasivPiHJZk27w&s

(sorry for the bad quality)

[bstolle]

a) Boeing does encourage the PIC to decide whether to continue with the approach or go around, and
b) You have a frigging 2800m runway ahead.

a) That's valid for the full flaps, short final case, hence below 500-1000ft. I assume that Boeing included this part after the double engine failure case at 100ft, 1 sec from the threshold.
Furthermore it's the FCTM, the manual with the lowest authority leaving aside all other aspects and disregarding FCOM, OM-A etc.
b) If you experience a runway excursion, e.g. due to a blown tire, runway lenght is irrelevant.