from this SLF - Probably 1.5 re limit load. But there may be other factors re size ( area) of MAX stabilizer compared to NG and the effects of a longer lever arm re pitch change per degree stabilizer movement. That kind of discussion is well beyond my old wheelhouse/pay grade. MY cliffs note version is that with a longer lever arm AND same size ( area) stabilizer , less stabilizer angle change ( and load ) is needed to give the same nose up/down pitch change.

From the evidence it is clear that this accident was not caused by mechanical leadscrew problems. As soon as the AoA reaches about 20 degrees the PITCHTRIM moves down at a very constant speed, exactly following the described MCAS operation. This is consistently dealth with by the PF through regular TRIMUPMANUAL inputs after which the down trimming halts for a few seconds (the way MCAS has been stated to do). At the end these TRIMUPMANUAL inputs are of shorter duration resulting in a runaway down trim.

The PF was continuously up-trimming as required until those last trim inputs. Those were too short, perhaps because of other distractions. Although the crew was aware of the continuous need to apply up-trim they apparently did not use the auto trim cutout switches.

This is what I think too, note also that the repeated ND automatic trim starts coincident with flaps 0 (as it does on the accident flight, so MCAS) - before that the automatic is sometimes up sometimes down.

Also in the descent there is plenty of trim movement with no auto or manual (electric) signals - my guess is cutouts and manual trim using trim wheel, until after landing.

The weird thing is that the auto trim signal is pretty darn quiet through cruise and descent, suggesting that the data is acquired downstream from the cutouts, but there are two brief sets of down-trim spikes during descent. Seems odd to move the cutouts then, maybe bad data?

Distraction comes to mind.

This may be a wake up call as aviation (and the world) racing towards more automation and artificial intelligence/machine learning. Machine-1, Man-0.

Indeed but it is nothing new: this case reminds me very much of the Turkish crash here at EHAM (deep stall at short final, a 738 was flaring / retarding throttles mid-air after a single radio altimeter failure resulting in a zero altitude reading, this also happened on previous flights of that aircraft but was dealt with promptly, the cause of the RF problem was poorly understood).

Stickshaker
 

On the previous flight the stickshaker was on from flap 0 until landing. For the whole flight!! Also the IAS difference was there for the entire flight. I cannot imagine how mtc would think a simple AOA sensor change would do and release the aircraft for a revenue flight. This is try and error on a life critical fault.

Not wanting to start an A vs. B argument, but the European side of the aircraft manufacturer house went, and to some extent still goes, through automation growing pains. Of course, different philosophies mean that A is using Hal as the master whilst B is more aligned to developing Robocop.

Sorry, tired ramblings but my point is that 21st century accidents are now so often down to the "What's it doing now?" scenario.

You change the whole unit (pictured here) - the ramp is not the place to start taking it apart.


Eight screws, gasket, two electrical connectors, probably less than an hour including functional checks.

I'd be fascinated to know how the unit that was removed subsequently tested on the bench. Since I'm in a betting mood, my money would be on NFF.

Somewhat under the radar and probably not related but on Nov 14th Sunwing 738M near Dulles left ADIRU failure (6 month old aircraft). Per Avherad "when the captain's instruments began to show erroneous indications. The first officer was handed control of the aircraft as his instruments and the standby instruments remained in agreement." Wonder what instruments?

Pilot error ?
 

The previous crew counteracted MCAS for a while before activating the CUTOUT, in accordance with the runaway stabiliser procedure. The accident crew were also counteracting MCAS for six minutes but do not activate the CUTOUT. It seems that the bad AOA data was probably contributing to a fair amount of confusion, but knowing the previous crew had stabiliser trim issues, and then having those same issues themselves, why didn't they activate the CUTOUT ?
The airline would seem to have significant culpability but is it pretty clear case of pilot error ?

Paraphrasing Dr Earl Wiener PhD "Automation reduces workload when the workload is already low. Paradoxically, it increases workload in a cockpit when the workload is already high"

I believe others are asking the same.

'Ony Soerjo Wibowo, an air safety investigator for the Indonesian government, said that on the plane’s third-to-last flight, from the eastern Indonesian city of Manado to Bali, the plane, which was delivered to Lion Air in August, had recorded no airspeed data.' (cited in NY Times)

and Mr. Ony said that part of his investigation centered on why the plane had not been grounded by Lion Air, given that it had experienced multiple airspeed problems.“This is strange,” he said in an interview on Wednesday. “We found several events that we should investigate, but they didn’t report them.”

It's not clear who 'they' is that didn't report them. But, in looking at the leaked maintenance log, sure looks like some issues were not documented by prior crews. Don't know what condition the AoA sensor that replaced the alleged offending sensor was in either - probably whatever they had in store since it's the same PN as the NG uses (see #1418).

I'll be interested to see the factual reports on
- the training records of the flight crew, especially in regard to using the trim cut out procedure
- the records about the maintenance organization and its actions and culture and how it was possible to release the plane fit to fly with IAS disagree, stick shaker on, trim runaway, after a simple vane exchange, or even worse, without any action before the last flight, as the aircraft lost its redundancy
- the log book entries by the crew flying the aircraft on the last 3 sectors, which obviously were not sufficient to address the problem to mtc and the subsequent crew, so this was mentally prepared what could happen
- the certification documents that allow a no majority capable dual sensor system to manipulate essential flight controls and leaves it to the flight crew to identify what might be wrong
​​​It seems much comes down to training, organization, leadership, documentation, culture in the airline, the FAA and Boeing.

A couple of AOA and MCAS details to consider:

1. AOA vanes measure the angle of the airflow at the location of the sensor. The fuselage distorts the flow such that the vane angle does not match the free air AOA seen by the airplane. There is a conversion from degrees of vane to degrees of body AOA. Typically one degree of vane motion corresponds to about half a degree of body AOA change. I'm not sure if the AOA data we have seen here in PPRUNE is in vane degrees or body degrees.

2. When reviewing flight data and trying to determine the status of MCAS and the cutout switches it is important to realize the following:
- MCAS has an authority limit of at most 2.5 degrees of stabilizer (less as speed/Mach increase).
- MCAS will not move the stabilizer an increment more than its authority unless it has been reset by either pilot manual stabilizer trim command or engagement followed by disengagement of the autopilot.
- MCAS would not be reset as a result of mechanical manual stabilizer trim (i.e., pilot physically turning the trim wheel by hand)

Some of the time periods where seeing no automatic nose down trim might lead one to think that the cutout switches have been activated, it could actually be the case that MCAS has already put in its allowed stabilizer motion increment and has not be reset via manual trim and therefore will not move the stabilizer any more regardless of how long the time period is.

Another possibility is that during some of the time periods where there are no automatic trim increments the crew may have been able to engage the autopilot in which case MCAS would be disabled. There would, however, likely be some automatic stabilizer trim with autopilot engaged as fuel burn would require.

With the cutout switches activated, the only means of trimming is via the mechanical trim wheel. It is likely that pilot activated trim via the wheel would not be at the same rate as electric motor trim (and might not be at a continuous, constant rate). Knowing that it might be possible to figure out which stabilizer motions were likely not via electric motor and thus must have been via pilot use of the mechanical trim wheel.