Desktop version

Home arrow Management

  • Increase font
  • Decrease font

<<   CONTENTS   >>

Communication as Information Propagation across a Network

On 10 September 1991, an Embraer 120, N33701, was pulled into the hangar for scheduled maintenance (NTSB, 1992). The time was about 2130 and the second shift was at work. One of the tasks for the night included the removal and replacement of the left and right horizontal stabiliser deice boots. The job required the leading-edge assembly to be removed, the old deice boot to be separated from the stabiliser, the new boot to be bonded to the structure and the leading edge to be reattached to the stabiliser by 47 retaining screws on both the top and bottom surfaces. The job originally was assigned to the third shift, but the second shift supervisor started the job to help with the overall workload.

The hangar supervisor, who I will call SI, was busy with a C check on another aircraft, so he assigned two mechanics (M1 and М2) to assist a second supervisor (S2) in the task of replacing the boots. This second supervisor was normally responsible for the line (routine rectification and replenishment of aircraft) but, on this occasion, had been asked to oversee work on N33701 in the hangar. A quality inspector (QI1), overhearing the mechanics being tasked, and still with an hour to run on his shift, volunteered to lend a hand. Being the smallest of the three, he went up onto the upper surface of the tailplane, 20 ft above ground level, to remove the screws along the upper surface of the leading-edge assemblies. He placed all the screws, apart from a few damaged ones, in a bag, which he then placed on the ‘manlift’, a hydraulically operated platform used for working at heights. He wrote ‘assisted mechanics with removal of deice boots’ in the Inspection Department Written Report to account for his time.

The second supervisor (S2), in the meantime, was observing the progress on the aircraft and saw QI1 on the top of the right-hand stabiliser. The mechanics had come to the end of their shift and the third shift was arriving for work. The oncoming supervisor (S3) checked the supervisors’ shift handover briefing sheet but saw no comments relating to N33701: this is because SI had not yet written up the task. He then inquired directly of SI what work has been done on the aircraft. He asked if work has started on the left stabiliser. SI looked at the aircraft and said ‘No’. S3 then expressed some concerns about the time remaining to finish the job and decided to defer the replacement of the left deicing boot. S3 told SI that he was putting the left boot back into stock. S3 also checked the job card, which gave no indication of any work done on the left stabiliser.

S3 assigned a new mechanic to the task (М3) and told him to check with SI and the two mechanics (Ml and М2) working on the aircraft as to the state of progress.

They reported that they had managed to remove the leading-edge assembly on the right stabiliser and remove most of the boots, but their progress had been hampered by some stripped screws that needed to be drilled out.

As he went off shift at about 2230, Ml remembered briefing SI on progress but could not later remember if he had mentioned the work on the upper left leading edge. Ml also briefed the fourth mechanic (M4) on the job but failed to complete any documentation as the job cards had not been given to him. The job cards were still in the documentation package awaiting the third shift supervisor (S3). Ml remembered telling M4 about the removal of the screws along the left-hand upper leading edge. However, M4 was reassigned before he could start work, but before he left, he then briefed another mechanic (M5) that the screws had been removed and that the boots needed replacing. He saw a bag of screws on the ‘manlift’ and gave them to M5.

The third shift mechanics finish the removal of the right-hand boot and bonded the new one into place. A second quality inspector (QI2) was lending a hand at this stage. QI2 had come on shift early and had checked the department’s written report before QI1 had entered his comment about helping the mechanics. There was no direct verbal communication between QI1 and QI2.

At this point, N33701 was pushed out of the hangar to make room for another aircraft. M5 completed the fitting of the right leading-edge assembly outside, making use of the screws in the bag together with some new ones drawn from stores. When the right leading edge was refitted, M5 noted a few screws left over but paid these no attention as he understood that there had been some replacements used. S3 climbed up to the right-hand stabiliser to check the reconnection of the deicing fluid lines. He did not check the left-hand stabiliser as he was not aware of any work having been done. It was, of course, dark and the top surface of the stabiliser was not, in any case, visible from the ground. There was no direct light on the aircraft. Finally, М3, QI2 and M5 went off to complete the paperwork. М3 returned to clean up around the right leading edge, telling M5 that the replacement of the left assembly had been deferred. The aircraft was then cleared for flight and departed at 0700 from Houston to Laredo. On the return flight, as the aircraft began its descent into the Houston area, the aerodynamic forces on the tail structure tore off the left-hand leading-edge assembly causing the major structural failure of the tail. All 14 passengers and crew were killed.

Three supervisors, two quality inspectors and five mechanics were involved in this scenario with work spread across three shifts and distributed in space between the inside of the hangar and outside on the ramp. Maintenance work is typically controlled through documentation. In this case, the various shifts had access to logs for recording information and job cards for controlling specific actions in relation to the aircraft. These are the primary artefacts associated with communication, but we also have the physical presence of the replacement for the left stabiliser boot and the bag of screws left on the ‘manlift’. These items, too, allowed individuals to construct meaning in the scenario. Although the engineering staff were assigned to specific shifts, the transition between shifts was fluid as individuals reported for duty and began to support the work of others or assume control. As a result, communication was asynchronous. Individuals briefed themselves on the status of work before records had been written up. Verbal updates were incomplete or incorrect.

Finally, personnel did not have equal access to information: mechanics were not issued with work cards as the documents had yet to be issued to the appropriate supervisor.

The bag of screws left over after the leading edge had been refitted shows how meaning is constructed in situ. The screws came, of course, from the left-hand stabiliser. Because the status of the task was not fully understood and, thus, accurately communicated, the screws became ‘leftovers after drawing new screws from stores’ rather than ‘screws not replaced after the job was changed’. The meaning of the ‘bag of screws’ was constructed based on comments made by others. This example neatly illustrates work as the part-processing of tasks that are then handed off to other workers for additional input. Three groups of workers - supervisors, mechanics and quality inspectors - all participated in an asynchronous and distributed sense during the time the aircraft was being worked on. Because of the way people communicated, the actual state of the aircraft was forgotten and, instead, an imagined state was constructed and transmitted between actors. The propositional content of messages was flawed and, as a result, the job was not finished correctly. On 24 October 2002 (TSB, 2005), a DHC-8 lost a three-foot section of the wing leading edge because 14 screws that fasten the section to the bottom of the wing were missing. The aircraft managed to get airborne and land again safely. The events that led up to the wing failing were very similar to those we have just seen.

<<   CONTENTS   >>

Related topics