Re: # 1785 “Why it wasn’t recovered earlier (when it was a recoverable situation) is what we're discussing...your guess is as good as mine”.
Using the scenario and analysis at
(#1735), the first indication that the crew’s perception was incorrect might have been approaching Vapp (Vref+5?). The expectation would be for the AT to pick up and maintain the selected speed, possibly allowing a small under-swing to see if the AT would respond … (Vref? 1.3vs). In the accident scenario, the next indication would be stick shake (1.1vs), a further speed reduction of ~20kts, which at 2-3kts/sec (mid - end of the speed trend scale) is 7-10sec later.
Human reaction time is often assumed to be in the 2-4 sec range. In this instance it might be nearer 4sec for the Capt (PNF) to recognise the situation and get in the control loop. It would not be a smooth transition from observing speed decay and waiting for stick shake as in training, it was a sudden onset of a totally unexpected and very attention getting warning. The Capt probably announcing I have control, disconnecting the AP, and applying thrust (but omitting to disengage the AT). The control forces are unexpected due to the mis trim condition (confusion), and the supposed need for two hands push force enables the AT to retard thrust again. In the time taken to read the above the aircraft has stalled and at low altitude could not be recovered.
A dominant issue is that at a critical stage of flight the crew were distracted; either by the surprising / unexpected gear warning; “we don’t get these and if we do they occur at 500ft” … thinks “what’s wrong, we are not at 500 ft” etc, etc.
There could be further distraction from late checklist use / training task which masked the first opportunity to detect the speed error at Vapp.
Logically, if the checklist / training task were the primary distractions you would not need an inaccurate RA to cause this or a similar accident. Solutions involve reducing workload, avoiding distraction, and ‘making time’ by advancing procedures to an earlier, timely point in the operation – slowing the pace of operation.
---------------------------
For the would-be designers, I suggest that you review the guidance material in
CS 25 Large Transport Aircraft particularly 25.1309 (p125) and the AMC (p520) on design, analysis and concepts of risk, safety, etc.
Also, see the more recent AMC 25.1302 (Page 485) on human interface, crew error, etc (good reading for HF training and CRM).
The requirement for a Rad Alt comes from
CS AWO – All Weather Operations, but with a developing industry the RA is now used in many systems and thus CS 25 1309 etc would apply. Interestingly you only require one RA (two displays) for Cat2/3, but there could be exceptions in order to meet reliability/integrity targets.
I suspect that the 737NG did not have to comply with the new CS 25 AMC 1302 (due to grandfather rights), but many of the HF objectives would have been considered as most manufacturers have been working in this area for many years. Thus the NG is not actually ‘new’; however, IIRC most of the flight guidance system (AP, AT, RA) is new and might not be able to claim extensive proving from previous series of aircraft – instead of Grandfather rights it’s more like Grandfather’s Axe!
Note AMC 25 1302, para 3,
”… the applicant(manufacturer)
may assume a qualified flight crew trained in the use of the installed equipment”; this appears to be at odds with operational requirements which enable line training – another mismatch between aircraft certification and operational certification.
CS–AWO 268 (for Cat 2/3 ops) requires
“that the probability of the provision of false height information leading to a hazardous situation is Extremely Remote”. I wonder how this might be interpreted in the context of this accident.
Perhaps this is the basis of the ‘cryptic’ comment in the Boeing communication suggesting that the RA fault (indication) should not have occurred.