Posts about: "ADSB" [Posts: 116 Pages: 6]

I agree with the post above (edited for brevity) - and fear the thread is getting repetitive in the absence of new information.

The only thing I would add is the limited ADS-B data I have seen shows the a/c decelerating rapidly from the first data point onwards. It is possible the shutdown occurred when the a/c was on the ground (e.g. after V1). This may seem unlikely given the distance flown, but you can do the maths - a fast a/c has a lot of stored kinetic energy.

Absolutely.

If the aft EAFR ceased recording at the point (just short of the runway end) where the ADS-B and likely everything else went dark, then the extra 30 seconds of CVR recording from the forward recorder could well be crucial to the investigation.

625 feet is straight from ADS-B data, which is just pressure altitude using sea-level STP. I don't know if it has been mentioned in this thread yet but in the other thread it was pointed out several times that this would have come out to ~100ft in true altitude AGL. The mathematics of that checks out to me (QNH 1001, temperature 37\xb0C, and field elevation of 189 feet).

The aircraft also visibly never gets much more than roughly its own wingspan above the ground in CCTV footage, at least to my eyes.

News articles tend to be fairly unreliable sources of info as far as parameters like altitude go in my experience.

tdracer explained that earlier: T/O power to sub idle on fuel shutoff only takes 1s, at most 2s.

Slamming the throttles back is a lot slower as the FCU (on a traditional engine)/FADEC spins down the engine slowly - I suppose to make sure that the airflow through the engine remains stable.

Regarding the momentum: As the first few seconds of the climb were normal compared to previous T/Os of the same flight (speed & altitude, confirmed by comparison of the RAW ADS-B data) I don't believe the engine failure happened before or on lift-off.

I agree with this. For all the SLF reading and not posting ( good work!) there is a delay between the end of the takeoff run and the gear being selected up. I fly 737’s so any 78 folk feel free to correct me but it looks like this;
PM: “ V1 ….Rotate”
The PF then begins to rotate the aircraft up to a pre determined attitude which is normally between 13 and 15 degrees. They do this at a rate of between 2 and 3 degrees per second so about 5 or 6 seconds later the aircraft is at its climb out attitude. The PM is then looking at their instruments to confirm that the aircraft has a positive rate of climb, this takes a moment, maybe 1 to 3 seconds then;
PM “ positive rate”
PF: “ gear up”.
So minimum 8 seconds but probably longer between the PM calling “rotate” and the gear being selected up.
The relevance of all that is to say that if you suspect that the gear up cycle has been interrupted by a dual engine failure, then the engines may well have been producing thrust up to an altitude of 50-100ft or so, which ties in nicely with the max height reached, distance travelled etc.
Mods this is clearly not a theory, just info for those who don’t fly airliners to aid understanding.

In the CSV data set that can be downloaded from that link the first point with altitude data is 1630 ft short of the departure threshold. That point is 575. The highest alt recorded in the data set is 625. All the points with altitude data overlay the departure runway. I do not understand how anyone is using this data set to determine the maximum altitude which was way past the departure end.




Edit to add - I have made no attempt to correct the raw ADS-B altitude data. There is no need to make any correction to see altitude gain.

I wonder if, given the ADS-B stopped at the threshold, no data at all was recorded after that (including the crew firewalling the throttles)? Happened in the Jeju 737 prang.

Yes, we know it climbed. I suggested use of the granular data to show how high it was, in the context of the 400ft mode changeover point.

I don't have access to a current 787 manual, but have found a couple of unverified references to the VNAV engagement on the net. It appears the VNAV will engage at 400ft also. Happy to be corrected though.

Level-off point, approx 11sec after liftoff:
​​​​​​​

Could you please elaborate on that?

FR24 did do that raw ADS-B data comparison. Remember the GPS position and barometric altitude are sent by the aircraft itself. The altitude is sent in 25 ft intervals, so a shallow curve that is smooth in reality looks janky in the data, due to the rounding of the numbers. From https://www.flightradar24.com/blog/f...rom-ahmedabad/ :
The red line is the accident flight, and it covers approximately 4.3 seconds.
Obviously the altitudes are all uncorrected for barometric pressure, which would've varied with the weather on that day; you kind of have to mentally shift the lines vertically downward. Now I looked for, but couldn't find, the post in the old thread where the rotation was triangulated\xb9, but I remember that it was near the turnoff to the high-speed taxiway, so a few seconds ahead of this ADS-B capture. We only have the video to show us what occurred then.

That means the ADS-B data doesn't really tell us whether the first few seconds of the climb were normal or not.

When we compare the red line to the blue lines, the data tells us the climb rate had already decayed significantly before the accident aircraft passed over the end of the runway, because the red flight path is much more shallow than the blue flight paths.

Please correct me if I'm wrong: to my eye, the data alone does not show that the engines must have failed after rotation, because the data does not demonstrate a normal climb rate.

But likewise, the engines can't have failed much before rotation:
For completeness' sake: you can look at the CCTV video, consider the 787's wingspan a flying 200 ft yardstick, and hopefully agree that the aircraft did not get much higher than 200 ft AAL, if that.

-----
\xb9 I found one of them, anyway. The reference is the CCTV video:

Visually, it looks normal to me compared to other 787s I've seen takeoff. They don't point the nose to the heavens like most jets.

SLF here. Please excuse some ignorant questions, and please delete if this is simply contributing to the 'noise':

1) Why would the Aft EAFR stop recording if the main battery were available?

2) The Aft EAFR has its own dedicated analogue connection from the CAM. Is it the case that other relevant audio streams are not generated and/or transported over the fibre-optic data network to the Aft EAFR when on (main) battery power alone?

There seems to be an assumption that, if corrected for local altimeter, the lines all move down toward the runway as a set.

Wouldn't that only be true if the altimeter setting was the same on all the days those flights were made? Isn't that improbable?

I am not convinced that the FR24 graphic shows that.

The FR24 data shows that, for the accident flight, the first data point received on takeoff was one that included altitude. We know where the aircraft was and we know the uncorrected baro altitude at that point.

We do not know how the altitude of that first point compares to the altitude of the reference flights unless all flights have their altitude adjusted by the prevailing altimeter correction.

Each trace starts in about the same place over the runway but this may not be useful data. I don't think this is the ADS-B ground/air transition. I think it is the point at which reception of ADS-B data becomes possible because of transmission "line of sight". We don't know if data starts being received as a result of increased altitude or because of passing by whatever was blocking the signal.

I'm sure someone could research the altimeter setting for each of the reference flight and produce a corrected data set. That would be interesting and useful data.

That's just my interpretation of the data I have seen. It is not presented as fact.

No, it's not. You got that value from ADS-B, which is barometric altitude at standard pressure, and when you correct for that, the highest value is ~100 ft. AAL.

Instead, look at the CCTV video, and consider that the wing span of the aircraft is ~200 ft. I hope you'll agree AI171 didn't come close to 400 ft. AAL at any point.

Reported 625' pressure altitude, elevation 189', QNH 1001.

So they only got to around 100' height, half the wing span of a Boeing 787? I think it looks higher than that.

189' elevation, QNH1001, the pressure altitude should be around 550'. First readout from ADS-B is 575', highest is 625'. I did not look up if runway departure end is much different there than aerodrome elevation.

Interesting that the readout of pressure altitude doesn't get lower again, after 625', which could support the total loss of all AC (and most DC power?). Transponder possibly stopped working, but one of the pilots was able to transmit a mayday call. So they must have had some kind of electricity available.

So what I mentioned about THR REF / VNAV SPD is probably not applicable. Yes, I did not account for the QNH initially, which makes closer to 100' than 400' AGL.

I will revert to await the official findings then, with a substantial loss of electricity seeming more and more plausible.

The last report received from ADS-B out is 625 ft but, at that point, the aircraft is still over the runway. The video shows the aircraft continued to climb after passing the departure end.

There is no conflict. Simply a lack of ADS-B data.

Edit - Add the ADS-B data points graphic that I had posted June 17.

Thank you - it doesn't sound like it should be possible for a non-faulty autothrottle to activate below 400 feet then. Even if e.g. the "wrong" starting altitude is measured on the runway (wrong QNH, etc), as long as it is measuring 400 feet from that altitude, then it shouldn't matter.

Small correction, the 625 ft pressure altitude data point (~100 ft AGL taking temperature, pressure and field elevation into account) is the last received ADS-B data point, which was right around the end of the runway. The aircraft crashed quite a bit further on from that, ~1.5km from the runway end. If you take perspective into account it also appears to continue to climb for a few seconds after passing over the runway end. I think some people in either this thread or the first thread have argued that the ADS-B data cutting out could be indicative of power loss (like with the Jeju Air 2216 accident) and not just spotty coverage, and when I looked at other flights I was inclined to agree - coverage is spotty on the apron and runway, but once planes start climbing the updates become consistent. If that really is when the engines died, ballistics suggests that the aircraft could well attain another 100 - 200 feet off its sheer momentum before the inevitable descent.

Re: mayday call transmission, isn't that easily battery powered? At least on the captain's side.

Sorry, maybe the context was lost - I was responding to a theory that did argue that the aircraft was automatically trying to capture a target altitude that was incorrectly set too low. That has happened before, but the incidents I could find looked very different from this one (one Dash 8 and one A330, both involved the crew activating the autopilot).

slf/ppl here - with a respectable amount of experience in software delivery for real-time/embedded/safety critical systems. Software development in this area really is an engineering discipline and bears no resemblance to common practice in other areas. Couple that with the requirements for function duplication/triplication, harness separation et al then IMHO the chances of FADEC etc software errors are effectively zero.


I'm commenting to make that point but also to link the videos and the FR-24 dataset - (below with my deltas for height/time added)



Extract from FR24 csv dataset


As noted in both threads to date everything was normal until it wasn't - the two values for fpm above are subject to FR24 variance of +/- 25' so even these suggest a normal climb at this stage of flight ca 2,000fpm. FR24 Lat/Longs all follow the centre line.


On this data the climb stops at around 70' AGL and electrical failure around 2s later. Again, as noted in the threads, this aligns with when gear up might have been expected. If the climb stopped because of fuel shutoff then 2s for spool down to electrical failure isn't out of the question.


Looking at the two videos.

The CCTV video indicates a total flight time, from rotation, of about 32s, subjectively levelling off ~14s after rotation.

The rooftop video has a flight time ~14s suggesting the video starts ~18s after rotation.


The rooftop video evidences the RAT as deployed from the beginning - meaning it must have been deployed by at least 16s after rotation - which aligns with the ADS-B indicated electrical failure.


If the forward flight recorder really is being sent to the US for recovery then it's reasonable to assume that the rear recorder contains nothing after the electrical failure and they are hoping the forward recorder captured something from the cockpit in the final 16s.


I don't have any experience of flight deck CRM but I don't see how those timings allow problem identification/misidentification and subsequent action - ie it wasn't down to the crew.


However:

The maximum aircraft height in the CCTV video, as judged by wingspan, appears higher than 71' - though it is certainly less than a wingspan height at the beginning of the rooftop video.


I haven't seen, in the threads, any statement of what happens on the flight deck with a total electrical failure - is it a 4s blackout whilst the RAT deploys and systems restart? - or are there batteries that keep something alive?

Can someone help me with the calculations on how far from the point of our last ADS-B readout we can expect the stricken jet to fly/glide?

I am assuming the take off mass around 190-200 tons with 50 tons of fuel. For the glide phase this is of no importance, however.

Data on the Internet puts the glide ratio of a 789 around 18-21:1. Gear and flaps/slats out should have a significant negative effect. Does anyone have a good take how much? Minus 40%?

From the available data we can infer the plane never was higher than 200\x92 AAL, maybe even 100\x92. If I understood the online sources correctly, the point of impact was only about 20\x92 lower than the average runway level.

If I am not mistaken the distance from the last ADS-B point to the impact site is about 2 km as per Reuters and the Guardian. That would put it at 6,500\x92.

I just can\x92t get these numbers over each other without the aircraft producing thrust. Please help me correcting the numbers.

Well the point of maximum altitude would have been a fair bit closer to the impact zone than the last ADS-B point. The whole incident was about 30 seconds, it was descending for only 20 of those. Given it was travelling at more than 200 mph when it took off, I can well believe it could make well less than a mile in 20 odd seconds