Autopilot won't engage if the trim is too far out.

FGD135: Almost correct.

On the B737 it's the case that one can only engage an autopilot if there is no control loading on the yoke.

It is for that reason (on the B737 for sure) that it's the PF that selects / presses the autopilot button (as opposed to asking the PNF to do it) as only the PF knows if any pressure is being applied (by himself/herself) to the yoke. Indeed, the engaging of an autopilot is (or should be) the only time that the PF ever makes a selection on the MCP when they are the PF (i.e. during manual flight).

If the PF is applying any kind of control loading to the yoke (e.g. because the aircraft is out-of-trim) then the autopilot will not engage.

The Boeing B737 FCTM also states: "The airplane should be in trim, and the flight director commands should be satisfied before autopilot engagement. This prevents unwanted changes from the desired flight path during autopilot engagement."

For a prime example of f'ing up the activation of an autopilot, look no further than the Ethiopian (ET409) B737 crash out of Beirut on 25th Jan 2010.

The automation management issue in my view started earlier in the approach with the decision to fly the aircraft manually.

I agree with framer that once pointing skywards, it was a handling problem to solve. All the suggested actions are nice zero IAS suggestions but it seems from the moment full thrust was applied, the crew were well behind the aircraft and unable to catch up.

ManaAda...
You obviously do not hail from a Commonwealth nation.

personally I don't think going anywhere near the autopilot on a light weight 737 missed approach is a good idea, as I understand it (I may be mistaken) hitting TO/GA gives you 1-2000ft minute rate of climb, engage the A/P & A/T and now it's going to go for full whack.... in a light weight 737 that is imho recipe for disaster, alt cap, wrong speed set, loads of power off and on etc.... personal preference is to hand fly it all until clean and levelled off and then engage automatics, I put them in sooner though if heavier and not just below the Missed approach altitude,
it would be really interesting to know more about the MCP modes etc e.g. had they pressed the TOGA switch or was FD giving guidance to their approach etc etc.... so much conjecture without all the facts.

The interim report says the go around was most probably initiated because of w/s. So one would expect it was up on the tv. AP and AT off FD on, aggressive max power is what Boeing says and that`s what the N1`s show. In manual flight pressing TOGA switches activates the WEM, A/T will not move the throttles so crew firewalls them. I would assume this to have been the state of affairs until reaching 900 meters when the nose was commanded to pitch down. Why, to stop the climb, there can be no other reason. So why stop the climb, possible reason is weather, worse anticipated at higher altitude. The actual weather at the time of accident says CB`s base at 1080 meters. Looks to me they may have been attempting to avoid entering an active CB cell.

Well, “could have been":
But then we have:
So not a WEM as defined by Boeing. More a normal GA with max thrust, maybe to get out of the turbulence/shear ASAP. It went OK until:
Why? Knowing what was on the FMA would help a lot. Then:
So, firewall power and pitch over into the ground.

I’m starting to think that it was a complete loss of SA, at least from the PF and the control inputs were almost random. You’re going too fast, so you lower the nose and add power... Doesn’t make sense.

Xollob: "personally I don't think going anywhere near the autopilot on a light weight 737 missed approach is a good idea, as I understand it (I may be mistaken) hitting TO/GA gives you 1-2000ft minute rate of climb, engage the A/P & A/T and now it's going to go for full whack.... in a light weight 737 that is imho recipe for disaster, alt cap, wrong speed set, loads of power off and on etc.... personal preference is to hand fly it all until clean and levelled off and then engage automatics, I put them in sooner though if heavier and not just below the Missed approach altitude,"

No, that is wrong. 737 TOGA activation gives 1-2000fpm climb with a single push. Only pushing TOGA a second time at least 1 second after the first push gives full thrust GA. Manual GA and subsequent AP and AT engagement will engage in either LVL CHG or V/S, depending on circumstances, but neither is a hardship.

As for others engaging the AP with such an extreme out of trim situation by releasing the controls, that makes me question the qualifications of the commentator.

I agree. But there is no other way to engage the A/P on the NG. Ie it won't engage if there is any control input. That's why I think it's a dumb idea. If you're flying manually and get out of shape the A/P is not the answer. ( I imagine you already know that and we are in furious agreement)

I don’t think anyone is suggesting throwing your hands in the air and making a stab at the AP engage button as an SOP whenever things get difficult.

However, in this particular accident, the aircraft is in a million pieces and everyone is dead. IF they had engaged some sort of automation, would the outcome have been different? It would be very interesting for someone with access to a sim setting it up with the published parameters (2000’, TOGA, 18degs NU, trimmed for 156kts/F15 with the speed at 180kts) and observing what happens. It might not be pleasant but will it save the aircraft, at least for long enough to give the pilots another chance?

Recurring theme in post is " doesn`t make sense ". If you think within the box well, of course it does not. Most accidents do not make sense, even after all have been explained.
What makes sense here is that the planned landing was not executed, the attempt to level off in the go around could not have been planned. That is when all went wrong, because it was not planned. It could only have been an instant, unpremeditated, spontaneous response to an unexpected event or encounter. Energy management essential in such circumstances must have been precluded by other more pressing priorities.

Could it be possible that the PF was pushing on the cc while temporarily suffering from the false pitch up feeling caused by the somatographic effect, and the PM was pulling on the cc having seen the unusual attitude develop, this could explain the long duration of APND trim input. As long as the PF is pushing harder than the PM is pulling, then the main trim will operate in the direction the PF is commanding.

Also, what use is a HUD during a go round, it's designed to allow you to follow the instruments while looking out. Why would you want to look out?

Whilst I can accept all the fatigue issues regarding perception and cognitive ability, what I cannot understand is the two hour delay before making the decision to try and land again.

I'm a professional involved in healthcare and human factors in such an environment. I would have expected that during that two hours there would have been a point where a check was made on the fatigue factors.

No. It really does not make sense! We need the full CVR, all 2 hours of it, to make any sense of what was going on.

The way I read the report, all the factors point to a classic case of somatogravic illusion. Fatigue provides an additional hole in the cheese, with rational thought being pushed back and gut instinct taking over.

I didn't see the interim report mention it, but I think the reported "do not worry" from the PNF may have referred to the PFs perception of excess pitch up. The subsequent "don't do that!" from the PNF refer to the PFs actions in applying ND trim.

Punitive company culture regarding diverting?
Recordings of communications with company would be interesting.

Framer,

Believe it or not but the company told us that the communication has been "lost" or are not availble due to a technical problem that night...
I must tell you that we all laugh when they told us

I don't know why people think it is impossible that the pilots would not have some lapse of reason. A couple of years ago, driving my car towards a traffic jam, I temporarily forgot which was the accelerator and which the break. It was just a moment of doubt, and happily did not result in an unfortunate result, but I am quite sure this must happen (albeit very rarely) to a lot of people in the wrong situation, especially with fatigue and adrenalin, when automatic responses suddenly get reappraised, or when you find yourself out of the envelope.

the company told us that the communication has been "lost"

Were the communications HF or ACARS or Sat phone? Would they not be on some recorder on board? The duty controller that night MUST know what was said. Put them in front of a judge with the threat of perjury hanging over them.

What was the "technical problem"?