Desktop version

Home arrow Management

  • Increase font
  • Decrease font


<<   CONTENTS   >>

Acting in an Under-specified World

In the previous discussion of SA and sense-making, I suggested that views of sensemaking predicated on the need for a ‘truthful’ internal representation of the outside world might be wrong. Hoffman (2019) goes as far as to suggest that striving for accuracy is inefficient in evolutionary terms. When you open your eyes, what you actually see are a collection of icons, emojis and GIFs that represent attributes of the world. If I look at a bowl of fruit what I initially see is an icon for ‘something good to eat’. When I look at a group of people, I see a collection of emojis that allows me to establish an understanding of the mood of the group without investing too much effort. Having surveyed the world in shorthand, I then focus attention on those elements of interest and render them in finer detail. The ATR 42 case study shows how this might work in an aviation context. On shutdown in Shannon, the aircraft had 480kg remaining in the LH tank. Fuel planning and in-flight monitoring are not just routines but fundamental pilot skills. Given the degree of randomness in the refuelling at both Waterford and Luton, on another day the aircraft could have landed safely at its destination by chance alone. It seems that rather than seeing an accurate representation of actual fuel states, the captain saw a Hoffmanesque icon for ‘fuel state sufficient for the remaining journey’ based primarily on the contents of the RH tank. But it was not just the actions of the captain that created the situation he engaged with on that day. To fully understand this event, we need to place the captain’s actions into a system’s context.

Airworthiness requirements mandate criteria for the design and performance criteria of fuel systems. These were translated into a solution by the aircraft manufacturer who, in turn, communicated with the airline through the embodied technology, incorporating controls and indicators, and the associated documentation, which contained system descriptions and operating instructions. As part of the solution, the manufacturer provided redundancy. In this case, the electronic fuel switches were backed up by manual overrides. The design of the system afforded opportunities to create novel solutions to problems.

Company Standard Operating Procedures (SOPs) are designed to facilitate work; they are formal work instructions that guide activities but also represent embodied knowledge about how to do a specific task. The aircraft’s fuel gauge problem had been known for at least 5 weeks before the flight incident. The company operations manual stated that defects are to be written up after flight (author’s italics). Because the problem was intermittent, it might be construed that the crew did not see the need for reporting a defect as the gauge had not failed completely. Furthermore, the landing at Dublin was an unscheduled, intermediate stop and might not have been understood by the first crew as the end of the flight, which was destined for Luton. A flight crew instruction circulated to remind crews of the appropriate action in the event of a defect simply talked about accepting an aircraft with open entries in the tech log but does not refer to aircraft with known defects that are not entered in the log. Where maintenance assets are limited (either in terms of manpower or spares), it was once common for faults to be written on notepaper and only those that could be easily remedied during turn round were, in consultation with maintenance, written up in the tech log.

If an aircraft was at an outstation, the procedure in this company was for the relevant page in the log to be faxed to base maintenance who will make a decision and fax the annotated page back to the aircraft for insertion in the log. The report noted that it was often difficult to find a fax machine at some destinations (Asynchronous drift resulting from developments in technology?), and so crews operated with open maintenance items rather than incur the delays typically associated with compliance with the instruction in the operations manual.

Although the company’s minimum equipment list contained advice on how to estimate fuel tank contents in the event of a gauge failure, there was no formal requirement to do so, and although the company procedures did require fuel checks to be completed, there were no specific additional requirements in the event of unreliable or faulty fuel gauges. Therefore, the captain’s failure to check the tank contents before the final take-off from Luton was not necessarily a breach of procedure. Formal rules, then, can also provide sufficient latitude for local solutions to real-time problems to be implemented. Work very rarely proceeds as planned, and variations and anomalies are a normal part of the task. The fact that individuals are required to exercise discretion in order to get the job done renders the work process fallible. We can see this in the way in which the ATR 42 crew managed the faulty fuel gauge.

In this episode, we saw two examples of aircraft refuelling. Each iteration comprised different component parts as the captain tried to negotiate solutions to problems as they arose. His performance was framed by company procedures and shaped by prior experience but also involved collaboration with, in this case, the refuellers. Although one agent was willing to operate the manual trigger at the captain’s request, the second declined. Presumably, both agents were constrained by similar rules, but each acted in a manner that satisfied their interpretation of their freedom of action.

This section has attempted to illustrate how individuals construct meanings in support of their actions and how that action, in turn, modifies the world. Some would say that the captain ‘lacked SA’, but I have attempted to show that sense-making is a dynamic process. Organisational factors and aircraft design rules all shaped the way the captain behaved. However, effective metacognition requires performance to be matched against some anticipated set of outcomes. The incident report referred to ‘common sense, prudence and good airmanship’ in suggesting that taking a manual reading of tank contents at Luton would have been advisable under the circumstances, a classic example of hindsight bias. In fact, the report’s authors have captured the extent to which processes are generally under-specified, relying on operator expertise to elaborate on the process should circumstances dictate. Dekker (2002, 2005) discusses the difference between the work ‘as-imagined’ and work

‘as-done’. Shorrock (2016) goes further, identifying ‘work-as-disclosed’ and ‘work- as-prescribed’ as additional artefacts of the process of describing an activity. This incident reveals how the process of refuelling comprised a task that was represented in documents (imagined, prescribed and disclosed) but also a task was understood by various actors (done). Importantly, it also existed as a task as enacted (outcome). The outcome was clearly not the captain’s intention, but the result was almost inevitable given the way the activity was undertaken in a system’s context based on a buggy construction of the world. Furthermore, the action was the resultant of a solution generated in response to a version of the world created anew on each encounter. Importantly, the captain’s actions flowed from the decisions he made.

 
<<   CONTENTS   >>

Related topics