- 46 -
2.10
2.10.1
Possibility of INS malfunction
Possible malfunctions of the INS which might have resulted in an aircraft deviating from track to the extent experienced by KE 007 were examined. It was concluded that it was virtually impossible for any technical malfunction or combination of malfunctions in the INS to have resulted in the aircraft maintaining a constant magnetic heading for over five hours.
2.10.2
Malfunctions or failures of one INS with a drift error without a warning on the CDU was not a common event. A similar and concurrent failure of three systems was regarded as a virtual impossibility. The effect of triple mixing involving an undetected error in one or two systems was examined. If one system position differed from the other two by more than 35 NM, the CDU warnings would flash. The actual, unmixed, position indicated by each system could always be checked against the mixed position and against the unmixed positions of the other sets. Because of the 35 NM limit on the acceptable error between sets when triple mixing was used, a supposition that use of triple mixing in the presence of an undetected drift error in one or two sets could not be accepted as a cause of the track deviation experienced. At the least there would have been CDU warnings, except in the virtually impossible event of similar drift errors in all three systems.
2.11
Consideration of the magnetic heading record
2.11.1
The DFDR recorded a constant magnetic heading from three minutes after take-off. The functioning of the INS was based upon true north. The INS, through the autopilot, controlled the aircraft on a true heading calculated from the true great circle track from one waypoint to the next, the true airspeed and the wind experienced. The magnetic heading on any leg reflected the true heading and the magnetic variation and compass deviation. At any waypoint involving a change in the true track, the true heading, and therefore the magnetic heading, would also change. Changes in the true tracks on R20 were small but sufficiently significant to show as alterations in the magnetic heading record. Such alterations with track changes would have shown in the record whether or not there was an error in the controlling INS. No such alterations showed in the DFDR record.
2.11.2
The magnetic variation along the track flown by KE 007 changed from 24°E in the Anchorage area, through 12°E north of NABIE, 4°W on the east coast of Kamchatka and to more than 8°W at Sakhalin Island. Thus the magnetic variation was continuously changing throughout the flight, although the rate of change was reduced with progress to the west. With the maintenance of a set of true headings under INS control the magnetic heading record would have shown changes in concert with the change in magnetic variation, in addition to the alterations resulting from track changes.
2.11.3
It was considered virtually impossible that an INS fault condition could exist that would result in the maintenance of a constant magnetic heading over a long period, particularly with such a large change in magnetic variation and the effect of varying winds. The DFDR record showed that the aircraft was controlled by autopilot A in Command mode for over five hours and maintained a constant magnetic heading. It was concluded that the autopilot could not have been controlled by an INS. It was also concluded from the constant heading record that manual control of the autopilot was not exercised by the crew by use of heading selection.