The Adam Air accident report is certainly interesting, but I was slightly disappointed that it seems to focus primarily on the crew's task management in general and (lack of) actions after the autopilot disconnect, leaving some important questions relating to the IRS/FMS behaviour unanswered. I think by focussing mainly on the inappropriate task divisions and flawed upset recovery, the investigators missed an opportunity to shed some new light on the inherent design limitations (or, if you want: deficiencies) of a dual-IRS setup, system knowledge by the crew, and training regarding IRS/FMS/EFIS/AFDS systems.
The aircraft seemed to be off course early during the incident, tracking roughly to DILAM i.s.o. DIOLA, which necessitated a 24 degree corrective turn to the right 13 minutes later (*). The pilots also reported strange wind-readouts, which suggests an erroneous FMC groundspeed (note that the FMC position and groundspeed are normally based on the left IRS position and radio-navigation updates). The investigators wrote: "given the FMC's default selection of the left IRU while on the ground and the large left IRU velocity errors shortly after takeoff, it is probable that the FMC's velocity divergence test determined that the IRU-R was the erroneous IRU. This would preclude the FMC's position difference test (IRU-Radio position > 4nm) from causing the FMC selected IRU to switch to IRU-R." All these observations seem to suggest that the left IRU data was faulty, yet the crew decided to switch the right IRU to ATT. As far as I can see, the report does not attempt to find an explanation for this choice, and in fact it even often refers to "the IRU" with no further specification. At one time during the incident, the crew commented: "now the left one is good, the right one is different, you are kidding". Was there any particular reason for them to reach this conclusion at that particular moment?
The IRS FAULT checklist didn't provide assistance for a partial IRS failure with no FAULT light, as encountered by this crew. This confused them a lot, as evidenced by the comments from the pilots: "It's not fault", "The IRS is erroneous", "But the fault must be illuminated, capt", "It is, it is not fault", etc. As far as I see, the report does not attempt to analyse the suitability of this QRH procedure, or try to establish if the assistence offered by the QRH is sufficient for such failures.
Some time after the crew had switched the IRS-R to ATT, the PIC said "put it back on nav again, put it back on nav again." Once an IRS is switched to ATT, it is impossible to re-align it during flight, so putting it back in NAV is useless. I know it is all too easy to judge the crew for lack of knowledge after the facts, but this is such fundamental knowledge that I'd have expected the accident report to mention this explicitly. As far as I see, it doesn't, missing an opportunity to identify a training deficiency (at least enhancing the argument the report makes about deficient IRS system training and maybe identify lacking training on crew coordination for unrecoverable actions) and perhaps also to improve the IRS FAULT procedure in the QRH.
Crew-coordination during the last few minutes seems to have been negatively influenced by the fact that the copilot lost most of his EADI and EHSI indications. This would have affected his ability to monitor the flight-path (now having to check the indications of the captain and/or stby horizon). As far as I can see, the report doesn't analyse this difficulty, nor does it (counterfactually) highlight that the copilot's displays could have been restored using the IRS transfer switch.
And finally, if the left (not the right) IRS was causing the navigation troubles, can we really be sure that the Euler angles (pitch, roll and heading) of this remaining left IRS were indeed accurate?
(*) A related question that remains unanswered: why did the air traffic controller who exclaimed: "Where is Adam direct to? My God, he is flying north!" not inform the pilots straightaway? Why did it take 13 minutes before ATC questioned the aircraft's heading?
Last edited by xetroV; 1st Apr 2008 at 13:55.