A spate of incidents and accidents during 1990s suggested that pilots flying modern ‘glass cockpit’ aircraft “sometimes fail to detect unanticipated and undesirable automation behaviour in time to recover” [2, 8, 9]. Hence it is important to understand the likelihood of human error for the A-320 pilots monitoring the status and behaviour of the automated cockpit.
On one hand a large number (n = 49) of pilots felt that their scan pattern is limited to “one-instrument scan”, thus becoming smaller and quicker; on the other another group (n = 21) were concerned about scan of the primary flight displays (PFD). The latter group felt that there is an increased need for summing up the information from flight mode annunciations from the PFD, engine monitoring displays, multi-function control display unit (MCDU) and navigation display on the horizontal situation indicator (HSI) to keep track of the aircraft, besides that of the automation configuration and behaviour. Thus making the scan wider and more demanding .
Five pilots had problems about fragmentation of feedback, i.e. “information on the status and behaviour of different system components is presented in various locations on the flight deck”. Some pilots also reported that the commanded automation targets are displayed on the flight control unit (FCU), which they look at to assess the status and behaviour of the automation, while the actual automation configuration is to be monitored as flight mode annunciations on the PFD. This is despite emphasis during training that FCU indications are ‘pilot commands’ while flight mode annunciations on PFD are the actual active modes and target settings, leaving scope for human error particularly during high workload situation!
Fourteen pilots stated that basic instrument scan pattern is not viable for the glass cockpit. Information gathering on A-320 was as per the task context, where monitoring is meant to verify “expected changes in the status and behaviour of the automation and to deal with uncertainties about the effects of input on aircraft behaviour” where the pilots ask themselves specific questions . This is in contrast with efficient allocation of attention, as a highly learned and automatic standardised scan pattern, to help pilots assess the aircraft state at a quick glance even if they are busy performing other tasks.
Unlike the conventional check reading practiced as an efficient monitoring strategy using round dials, some of the glass cockpit displays, particularly speed and altitude, require pilots to focus to ‘read’ the displayed information instead of picking it up at a glance. There were also apprehensions raised about compromised efficiency to operate aircraft using traditional basic instrument scan.
Monitoring an automated aircraft, therefore, is a “reflective process, requiring mental effort on part of the pilot, who has to determine where to look next for what kind of information”. Besides being not economical mentally, there is likelihood of missing vital “information in an event-driven domain if the pilot’s expectations or model of the situation is off track”. False or incomplete expectations may lead to monitoring wrong or only part of the requisite information, while parameters not expected to change may be ignored, in turn degrading or delaying the crew’s situational awareness .
- Old Facts, New Insights – Surprises in Glass Cockpit
- Old Facts, New Insights – Lessons from A-320
- Old Facts, New Insights – Lessons from A-320 Part 2
- Old Facts, New Insights – Lessons from A-320 Part 4
Acknowledgement Image courtesy Wikipedia