Indonesian aircraft missing off Jakarta

[LaissezPasser]

Originally Posted by dmba

Alon Soetanto, a passenger on board the flight, told TVOne: 'About three to eight minutes after it took off, I felt like the plane was losing power and unable to rise.

'That happened several times during the flight. We felt like in a roller coaster. Some passengers began to panic and vomit.'

Thanks, I hadn't seen that report. Interestingly, I just checked the original Detik article to which I was referring, and it has just been updated (two hours ago) with an editor's note to say that the aircraft involved in the original Detik article was not PK-LQP; it was a different Lion aircraft, flying the same Denpasar-Jakarta route at a different time than PK-LQP.

[NOC40]

Originally Posted by Vessbot

As I was coming home and getting ready to check the thread again, I was just thinking it needed more blowhard "it can stall at any airspeed" posts, and when I refreshed it I was not disappointed.
In more serious matters, I saw Wiedhopf's plot of the granular FR24 speed and altitude data, and it gave me the idea to grab the spreadsheet myself and calculate the Total Energy and add it to the plot,

Beat me to it!
There's more you can analyse though, but will need to double check my maths
1) the final descent angle looks to me to be -25%, far from a vertical dive into the sea
2) g force is around -0.1g (relative to 0g in level flight) for the dive, then +0.2g for the recovery. Uncomfortable but not excessive?
3) totally agree on the TE analysis. if you subtract a multiple of the cube of the velocity though you can get a rough approximation of engine power. looks to me like they reduced power a little after the first upset, then it varied quite a bit a while after they leveled off
I'll tidy up the charts and paste them later if anyone's interested.

[Derfred]

In short if the UAS is detectable near terra firma,land asap.You have radio altimeter up to 2500',you have IVSI,you have IRS GS,you know your attitude/thrust settings for S&L and approach.Never attempt to engage automation as it is fed with the same false data.Stay in the circuit,NO CHECKLIST,fly the plane,ignore ALL warnings except EGPWS and land.
If its "single side" anomaly,you cross check the ASIs and hand over control to the pilot with the good data.You still return but now you have more time to play with,checklists may be done etc.

This is wrong and very poor advice on many, many levels.

You started well by saying this can be an extremely confusing scenario. Which is exactly why you would not react according to the quoted part of your post above.

Boeing has a very good Airspeed Unreliable checklist, which will save your aircraft if you follow it properly. It takes time, and with fuel available you should use that time to understand what is wrong with your aircraft, and how to get it safely back to terra firma.

The first (memory) part of the checklist gives you an attitude and thrust that will keep your aircraft safe.

The rest of the checklist troubleshoots the problem, identifies an accurate airspeed if there is one, and if there isn’t one, gives you attitudes and thrust to get down.

Do LionAir pilots have good training in this area? I have no idea. I am fortunate to fly for an airline that does, and is practiced every 6 months in the simulator.

[Vessbot]

Originally Posted by Rananim

Hmm...pitot-static anomalies are tricky.They arent trained much,they confuse quickly and are highly unforgiving.If the UAS is apparent after liftoff and it affects "both sides",then the last thing you want to do is prolong the flight.Ideally,you discover the anomaly prior to liftoff during the airspeed crosscheck at 80/100 and abort.However,some anomalies arent detectable until after liftoff.If the statics are the problem,then the ASI will read normally on the takeoff roll but in a climb it is measuring ram versus artificially high static pressure which leads to an ASI underread.You are flying faster than indicated.Opposite is true in a descent.If altimeters dont register a climb after liftoff and you get a windshear warning in calm air and/or a stick shaker you need to act quickly and RETURN.You dont fly around with unreliable basic instruments.You have 3 basic enemies in this situation:a)Confusion-your brain must disregard what your instruments are telling you b)Nuisance warnings-overspeed/shaker are highly distracting and you must block them out or disable them c)Time-the longer you expose yourself to the confusion and chaos the more stressed and fatigued you become.
In short if the UAS is detectable near terra firma,land asap.You have radio altimeter up to 2500',you have IVSI,you have IRS GS,you know your attitude/thrust settings for S&L and approach.Never attempt to engage automation as it is fed with the same false data.Stay in the circuit,NO CHECKLIST,fly the plane,ignore ALL warnings except EGPWS and land.
If its "single side" anomaly,you cross check the ASIs and hand over control to the pilot with the good data.You still return but now you have more time to play with,checklists may be done etc.

As Derfred also said, this attitude is incredibly wrong. In the middle of the sky and the middle of the flight envelope is the best place to sort everything out, turn off the nuisance warnings, figure out which instruments are right, and if there aren't any, relearn how to fly the airplane and set yourself up for success. A hair-on-fire whip back around to the runway before you have a handle on any of these things where now you have to fly a much more accurate speed with a small margin to stall, only has disadvantages and no advantages. The "basic enemies" you listed are true, but the best place to combat them is up high. If the situation is such that it's questionable if you're gonna be able to main control at thousands of feet high, on what basis do you think your chances would be better on a tight turn to short final?

[DaveReidUK]

Originally Posted by Nemrytter

Nor does it exclude the possibility that, as with most other accidents, the Flightradar data is unreliable.

This might be a good opportunity to put this canard to rest.

FR24 data (and that from FlightAware, etc) is indeed unreliable, but only if you fail to understand its limitations. Don't forget that ADS-B in the cornerstone of many nations' (including the USA's) future ATM strategy, so it's got to have something going for it.

Yes, it was never designed for accident investigation purposes, but that's not to say that it's incapable of at least providing a pointer to what happened (though obviously not how or why). Nowadays it's not uncommon to see ADS-B GE plots in AIBs' accident investigation reports.

There are a number of reasons why FR24 needs to be treated with caution, some apparent in the Lion Air data.

Firstly there are still a handful of airliners with inertial-, rather than GPS-derived ADS-B (but that clearly doesn't apply here).

Secondly, because data is typically captured from multiple receivers for a single flight, timing issues arise. Those are readily apparent in the Lion Air data - most, if not all, of the "outliers" in the graphs we have seen in this thread are actually data which is genuine, but which is displayed in the wrong position in the timeseries.

Thirdly, and again it's obvious in the data, is the fact that FR24 typically displays both position and velocity data as if they have been received simultaneously (i.e. with the same timestamp), which isn't how ADS-B works. So, for example, a lat/lon value that hasn't changed between successive records with different timestamps doesn't mean we are looking at a helicopter, the coordinates have simply been copied over from the last genuine position transmission.

So, for example, if you look carefully at this GE plot you can see instances of these issues, which I'm still working on resolving, but to suggest that it therefore bears no relation to what actually happened is, I suggest, being somewhat perverse:

[Gary Brown]

Originally Posted by Derfred

SNIP

Boeing has a very good Airspeed Unreliable checklist, which will save your aircraft if you follow it properly. It takes time, and with fuel available you should use that time to understand what is wrong with your aircraft, and how to get it safely back to terra firma.

The first (memory) part of the checklist gives you an attitude and thrust that will keep your aircraft safe.

The rest of the checklist troubleshoots the problem, identifies an accurate airspeed if there is one, and if there isn’t one, gives you attitudes and thrust to get down.

Do LionAir pilots have good training in this area? I have no idea. I am fortunate to fly for an airline that does, and is practiced every 6 months in the simulator.

Though it's at least worth mentioning that AF447 carried a very good Airbus QRF too, with separate sections for Unreliable Airspeed in the Approach Phase and in the Cruise Phase. And that QRF says more-or-less what yours does. In that case the 3-man AF crew never once looked at the QRF and never once even referred to its existence. What's more, the BEA Final Report speculated (unusually, and only because of the very great puzzle as to why they crew did what they did) that the PF simply chose the Memory Item he had available from training, but that sadly it was for actions for Unreliable Airspeed in the Approach, not the Cruise; and those were very different.

(Should add that in this present case it seems to me that the crew were dealing with somewhat more than a "simple" instrument failure, and may have had some as yet unknown mechanical issue too.)

[.Scott]

So according to this report (based on FlightRadar24 data), there was a sudden upward acceleration before the plunge. This supports some of the speculation that there was an abrupt control maneuver over Va that broke the aircraft.
So perhaps a sluggish elevator with a lot of play was the problem all along - and not a sluggish altimeter.

[Rananim]

A hair-on-fire whip back around to the runway before you have a handle on any of these things

a tight turn to short final?

Well these words are yours not mine.
I reiterate for the benefit of any younger pilots reading....DO NOT PROLONG a flight with UAS having
confirmed the problem just after liftoff.In the cruise over the ocean,fine....follow the advice of derfred/vessbot.
Why climb?You lose your one remaining working altitude reference(radalt)?At low altitude your IRSGS readout
is close to IAS.At altitude it is not.ATC will have the false data from the transponder and cant help you.Keep your takeoff configuration and turn downwind at 2000' radalt,level off, and set thrust to 60% N1.The Boeing checklist of 10deg/80% is not correct for this scenario and will climb you above 2500.Its not a "hair on fire whip back around" at all.No tight turns required just basic attitude thrust flying.

[Interrogator]

Interesting to see the feel diff press warning light was illuminated on the last sector. The V/S oscillations and oscillations around 5000ft look very suspicious

[Escape Path]

How on earth did someone calculate a descent rate of over 30000 fpm? That's ~550 kph, vertically. Hardly a believable figure...

As other have pointed, the initial descent (with corresponding speed increase) seems legitimate. It could perhaps point to an initial problem/abnormal indication being detected and reasonably corrected given the controlled climb to 5000-ish ft with ~300 kts. The flight at altitudes near 5000 ft could point to the initial problem being troubleshooted (I'll play the UAS scenario); minor deviations from altitude target (which seem to be 5000 ft) which are consistent with speed indications: ie. altitude goes down, speed goes up; altitude goes up, speed goes down. So maybe they really were applying a UAS checklist and flying pitch/thrust figures and trying to identify the culprit(s). Then something goes terribly wrong, given the fact they lost +2500 ft (~5000 to 2800, which is the last recorded value in some of the graphs posted above) in just under 30 seconds...

[NOC40]

wasn't vertical dive...

Originally Posted by .Scott

So according to this report (based on FlightRadar24 data), there was a sudden upward acceleration before the plunge. This supports some of the speculation that there was an abrupt control maneuver over Va that broke the aircraft.
So perhaps a sluggish elevator with a lot of play was the problem all along - and not a sluggish altimeter.

Does the aircraft provide vertical speed to FR24, or is the above data just derived from altitude vs time? I suspect the latter might explain the spikes.
The data posted earlier shows ground speed ACCELERATING towards the very end. That, and the time vs altitude data suggest a 25 degree descent angle. The graphs are very misleading in this respect.
Correction to my earlier post, and agree with other comments: vertical g was -0.25 on the first upset followed by +0.5 in the pull up.

[DaveReidUK]

Originally Posted by Interrogator

Interesting to see the feel diff press warning light was illuminated on the last sector. The V/S oscillations and oscillations around 5000ft look very suspicious

Somebody please correct me if I'm wrong (I belong to the avionics = PFM school), but the rectification action carried out for that entry in the tech log appears to have been to clean the connector plug on the elevator feel computer followed by a ground test, which was successful.

I'd have thought that (a) the EFC might well have been performing correctly anyway (in flight), given that the warning light responds to differences in either hydraulic pressure or pitot dynamic pressure, which were presumably detected by the EFC and (b) a ground test wouldn't necessarily be able to reproduce either of those conditions.

I'd be interested to know what procedure the AMM prescribes for diagnosing and rectifying this issue.

[A Squared]

Originally Posted by Escape Path

How on earth did someone calculate a descent rate of over 30000 fpm? That's ~550 kph, vertically. Hardly a believable figure...

Well, for an airplane whcih cruises at approximately 900 kph, a descent rate of 550 kph is not outside the realm of possibility. Note that I am not saying that this is what happened, or that it was calculated correctly, just that if the question is: Is it physically possible for a 737 to have a descent rate of 30,000 fpm? The answer is yes.

[Ascend Charlie]

vertical g was -0.25 on the first upset followed by +0.5 in the pull up.

Pull-up?? Less than 1g will not hold level flight, so 0.5 isn't a pull-up, it is just a reduction in the descent rate.

[NOC40]

Originally Posted by Ascend Charlie

Pull-up?? Less than 1g will not hold level flight, so 0.5 isn't a pull-up, it is just a reduction in the descent rate.

add 1g static if you want...

[LandASAP]

Originally Posted by Retired DC9 driver

What is CYA?

Cover- Your- A*s

If Maintenance say "you're good to go", well take that with a grain of salt. Put a snag in the logbook, call maintenance so they " know" you have a snag in the log, then they have to answer it, consult the MEL, and sign off the snag with a signature. Then a copy is in the system, plus there is a answered snag in the aircraft logbook.. Or wait for a part/repair before the aircraft is good to go.

Always keep a copy , if you can of any irregular ops , ie datalink messages from crew sked, "you are legal to fly an extra leg" or whatever..

So we (maintenance) aren't doing our job if the crews tells us something and not doing an Logbook Entry? Ridiculous comment!

It is quite often the case that the crews tells us, that something was wrong and after a look in the book there is not Entry. Belive it or not but we nevertheless have a look onto it.

[wiedehopf]

Originally Posted by NOC40

Does the aircraft provide vertical speed to FR24, or is the above data just derived from altitude vs time? I suspect the latter might explain the spikes.

Aircraft transmit vertical speed.
Also looking at the data points in question deriving it from altitude and time would not have yielded that rate.

Could very well be bad data. There is error correction and detection employed but i wouldn't rule it out.
In My Armchair Opinion i would have expected at least another positive vertical rate or some values in between.
But looking at the raw data you can see the value is really isolated: (Last column is vertical rate in fpm)

time (UTC) hex callsign latitude longitude Altitude (ft AMSL) squawk Ground speed (kts) track Vertical speed (fpm)
2018-10-28 23:31:34Z.462 0x8a0711 LNI610 -5.84418 107.1058 4850 1742 328 53 -1216
2018-10-28 23:31:35Z.150 0x8a0711 LNI610 -5.84418 107.10581 4850 1742 322 62 -2240
2018-10-28 23:31:36Z.614 0x8a0711 LNI610 -5.84074 107.10828 4775 1742 335 36 -3136
2018-10-28 23:31:39Z.954 0x8a0711 LNI610 -5.83676 107.11111 4575 1742 338 35 -6912
2018-10-28 23:31:40Z.136 0x8a0711 LNI610 -5.83626 107.11149 4625 1742 338 35 -6912
2018-10-28 23:31:40Z.136 0x8a0711 LNI610 -5.83626 107.11149 4625 1742 338 35 -6912
2018-10-28 23:31:42Z.258 0x8a0711 LNI610 -5.83309 107.11376 3900 1742 340 35 -9216
2018-10-28 23:31:45Z.159 0x8a0711 LNI610 -5.8296 107.11618 3650 1742 345 35 7424
2018-10-28 23:31:45Z.232 0x8a0711 LNI610 -5.82884 107.1167 3700 1742 360 34 -8768
2018-10-28 23:31:48Z.180 0x8a0711 LNI610 -5.82683 107.1181 3350 1742 367 34 -10496
2018-10-28 23:31:48Z.300 0x8a0711 LNI610 -5.82522 107.11921 3200 1742 373 34 -11584
2018-10-28 23:31:48Z.396 0x8a0711 LNI610 -5.82431 107.11982 3100 1742 377 34 -13248
2018-10-28 23:31:51Z.024 0x8a0711 LNI610 -5.82061 107.12238 2500 1742 382 34 -16896
2018-10-28 23:31:54Z.392 0x8a0711 LNI610 -5.8161 107.12541 1450 1742 383 34 -26560
2018-10-28 23:31:56Z.030 0x8a0711 LNI610 -5.81346 107.12698 425 1742 360 33 -30976

[flash8]

Is it physically possible for a 737 to have a descent rate of 30,000 fpm?

Adam Air (LOC) over 53,000 feet per minute at one stage and an average of over 20,000 fpm before break up.

[A Squared]

Originally Posted by wiedehopf

Aircraft transmit vertical speed.
Also looking at the data points in question deriving it from altitude and time would not have yielded that rate.


2018-10-28 23:31:51Z.024 0x8a0711 LNI610 -5.82061 107.12238 2500 1742 382 34 -16896
2018-10-28 23:31:54Z.392 0x8a0711 LNI610 -5.8161 107.12541 1450 1742 383 34 -26560
2018-10-28 23:31:56Z.030 0x8a0711 LNI610 -5.81346 107.12698 425 1742 360 33 -30976

Actually, it seems that the descent rate derived from altitude and time matches the reported vertical speed fairly well. Take a look at the two intervals above. In the first interval the airplane descends from 2500 msl to 1450 msl in 3 seconds. 1050 ft/3 seconds equates to 21,000 fpm. the reported vertical speeds are 16,896 fpm at the beginning of the interval and 26,560 fpm at the end. In the second interval the airplane descends from 1450 MSL to 425 MSL in 2 seconds. 1024 ft in 2 seconds equates to a 30,720 fpm. reported vertical speed at the end of the second interval 30,976 fpm. Not a perfect match, but not off by a large amount.

Maybe I'm not following what you mean by: "would not have yielded that rate"

[A Squared]

Originally Posted by flash8

Adam Air (LOC) over 53,000 feet per minute at one stage and an average of over 20,000 fpm before break up.

Exactly, vertical speeds in the range of tens of thousands of feet per minute (down) are not impossible or obviously incorrect, or "hardly believable" as has been claimed.