Communication in a Systems Model
The discussion, so far, has looked at the nature of communication between individuals. Ultimately, though, collaboration is goal-directed and, so, we need to consider how communication supports goal accomplishment. We have seen that the San Juan crew did not appear to have a shared understanding of the approach to be flown on arrival. The explicit, filed flight plan approach differed from the custom and practice solution, which was a manifestation of local knowledge. The ‘normal’ trajectory was summarised by the captain w'hen he said ‘they give you an ILS to get you inbound, [...] then [...] ask you “you got the airport runway eight in sight?” [...] “clear Lagoon Visual Eight’”. The first goal is to establish an inbound track, and the second goal is to meet the requirements necessary to switch to a different runway. The transition from the ILS Runway 10 to a visual approach to Runway 08 has a critical point at which the tw'o approach paths intersect. The correct approach speed for the approach must also be achieved. The system safety boundary is the point by which surplus energy must be dissipated. The margin associated with safety boundary extends from the point at which the crew are cleared to reduce speed to the target for the approach, which w'ould have been the outer marker on the inbound course to Runway 10, to the point on the runway where sufficient energy has been lost for the aircraft to stop on the paved surface. In a previous chapter, I suggested that the margin was also the zone in which discrepant task-related signals become apparent.
Given the time of day and duration of the flight, fatigue was unlikely to have been a factor. There was a significant experience differential between the captain and the FO. The airline’s induction course included just four supervised sectors before a pilot was cleared to the line (personal communication). The captain w'as not a trainer, and the FO w'as considered competent to undertake line duties. The subsequent investigations did uncover the fact that the FO was receiving counselling and was on medication for anxiety. The flight on that day, then, needs to be seen in the context of the crew dynamic.
The captain’s construction of a w'ake turbulence risk did not take into consideration the actual separation of the two aircraft and the prevailing wind. The risk of turbulence was less that the captain seemed to believe. While the buffering capacity at this stage could cope with his intervention, the margin was reduced. The failure to maintain the speed requested by АТС w'as a signal suggesting that the plan modification w'ould have implications that w'ould need to be evaluated. This w'as subsequently recognised, in retrospect, by the captain: ‘Just ya ya got traffic on a five-mile final behind you anyw'ay. You get too slow because it screw's their sequence up’. The segment of the profile covered by the CVR transcript includes seven points at w'hich the crew are required to complete a checklist. Four checklists are completed, two are partially completed and one is missed out altogether. One of the partially completed checklists includes the step that establishes the target speed for the approach. The failure to complete checklists and to establish key parameters, such as the desired approach speed, were also signals. The relatively high rate of social chatter between the crew suggests that the workload was not high.
The FO’s struggle to reduce speed in the final approach phase while maintaining the correct path to the runway suggests that the system was approaching its boundary. In effect, the FO was fighting the laws of physics. The aircraft’s potential and kinetic energy, coupled with the FO’s skill set, was beginning to exceed the buffering capacity. Ironically, the subsequent intervention of the captain after the initial touchdown was possibly the act that caused the system to transition to unsafe.
Hutchins describes communication as a mechanism by which teams construct a shared memory of the progress of a distributed task. In part, this is reflected in Excerpt 8.7. By constructing a representation of the movements of aircraft in the traffic pattern behind them, what has happened in the past, the captain can reflect on the options remaining for the crew. Unfortunately, at this point, he delegates decisionmaking to the FO.
In more complex, distributed systems of collaboration, communication is often mediated through tools and artefacts. Consider the following example, which involves aircraft ground movement at an airport (TSB. 2004). On 29 October 2004, a BN Islander nearly collided with a DHC Dash 8 on take-off from Vancouver International Airport (see Figure 8.1). The Islander, call sign Navair 612, was holding at Taxiway A at the threshold to Runway 08R on the south side of the runway. The DHC Dash 8, Jazz 8191, was positioned on Taxiway L2, to the north and further along the runaway. It was early morning and still dark at the time of the event. The ground controller issued the taxi instructions for the Dash 8 crew and then completed the flight progress strip. This ‘strip’, a piece of perspex with sections for different types of information, is an example of an artefact used to communicate between controllers. The design of the strip is such that space for information is limited and the information
about the cleared taxiway - in this case, the Dash 8 was cleared to enter the runway at L2 - was obscured. The ground controller passed the flight progress strip to the tower controller w'ho was sitting beside him.
The tower controller cleared the Islander to line up, asking the crew' to move forward to allow' the Dash 8 to line up behind it. The tower controller clearly assumed that the two aircraft were on the same taxi way where there were several other aircraft waiting for take-off at the time. The crew of Jazz 8191 was then cleared to line up. They could not see the Islander nor, given their angle of approach to the runway, could they see the threshold area. Because it was still dark, the Islander could not see the Dash 8.
Controllers are required to name the intersection in the take-off clearance if it differs from the threshold. Unaware of the actual entry point, the tower controller omitted that piece of information. Pilots are not required to repeat the intersection in their read-back of the instruction and so the error was not picked up. The formal rules of communication, designed to validate differences but obviate the need for unnecessary use of a limited medium - the radio frequency - were easily circumvented through the omission of a single item. The design of the flight progress strip represented a risk in that its completion by the ground controller was a catalyst for failure.
Vancouver had area surface detection equipment (ASDE) installed for detecting aircraft moving on the ground. Issues of reliability and dependability meant that it was used as an aid only, and controllers had little faith in it. The Dash 8 was shown as a target on the ASDE at the L2 position, but this was disregarded as a spurious return by the tower controller despite the fact that the return moved after the Dash 8 was cleared onto the runway.
As the Navair 612 started its take-off roll, Jazz 8191 was moving towards the runway, the crew looking ahead and to their left for the Islander they were supposed to be following. At the last moment, they decided to check the up-wind area, to their right, only to see the lights of the Islander coming towards them. They stopped the aircraft and turned on all external lights as the Islander lifted off in front of them.
Communication between two co-located workers, the ground and tower controllers, was mediated through the flight progress strip, but the design of the strip hampered effective communication. The tower controller was nonetheless able to render the communication sensible as he held a view that all aircraft were in a common location. Interestingly, the ground radar showing the location of the Dash 8 was noticed but dismissed. The crews of the two aircraft constructed an understanding of the situation through direct instructions from АТС and also calls to other aircraft. In the case of Jazz 8191, their actions were linked to instructions assigned to another aircraft, Navair 612. While obeying their own instruction, they were also looking out for the other aircraft upon whom their own progress was dependent. If nothing else, they knew that their ow'n take-off could not commence until Navair 612 had departed. Although they had started to move, there was clearly a doubt in their mind. We do not have details of the conversation between the crew of Jazz 8191, but the evidence suggests that they were uncertain of the situation. An instruction from АТС, set in the context of an instruction to another aircraft, triggered a discussion by a crew directed at resolving ambiguity. The indication of Jazz 8191’s position on the ASDE is an example of a signal that lacked sufficient salience to intrude on the controller’s understanding of the situation. It was a busy time of day with many aircraft preparing to depart, so workload might have been a factor. It was also still dark and so any visual cues would have been hard to detect. These events point to how, in collaborative systems, messages propagate across networks. In this example, the network was small, but the next case study explores propagation across both time and space.