Resumen
The rate at which accidents/incidents occur within the Air Transport System (ATS) has remained consistent over the past two decades (European Commission, 2016). If such a rate is maintained and the ATS grows as expected (AirBus, 2013; Boeing, 2013), then the ATS may no longer be able to be considered ultra-safe. The purpose of this work is to develop a framework by a set of defining system complexity parameters that allow mapping of conceptual safety models and risk analysis methods in relation to their complexity domain analysis range. The study asks: What is the complexity domain range for commonly used models/methods? To answer this question the following models and methods are described and analyzed: Domino model, Swiss Cheese model, Bow-Tie model, Functional Resonance Analysis Method (FRAM), Systems-Theoretic Accident Model and Processes (STAMP). The Cynefin framework uses four domains to represent the system complexity spectrum. In combination with the Cynefin domains, the following system complexity parameters are used for the analysis: Coupling - the ability to represent a loosely coupled system, Linearity- the ability to represent non-linear relationships, and Dynamics - the ability to represent a system through time. The obvious/simple domain of the Cynefin framework requires only that a model/method be able to represent tightly coupled, linear, static systems. While the complicated domain requires that a model/method be able to represent a moderately loose level of coupling, a degree of non-linearities among system actors/elements, and at least a moderate ability to dynamically represent the system. The complex domain requires a model/method to be able to represent a very loosely coupled system, with many non-linear relationships, in a highly dynamic manor. The results of this analysis showed that the FRAM method, Bridge model, and RMF are the most capable when assessing complicated highly interconnected systems (such as the ATS). However, it was also concluded that while when the ATS is operating within a normal range these models/methods may be sufficient, they are not able to effectively inform decision makers when on rare occasions the system shifts into the complicated domain. Therefore, since the ATS represents a common system type (having a large number of actors/elements that are very intertwined, and dependent upon the skills of their operators) a new term is coined ? Highly Complicated and Occasionally Complex (HCOC) systems. An HCOC system is one that predominantly follows the same pattern of rules for the majority of its operations, however when occupational novel conditions are encountered which cause the system to shift to the complex domain and thus the underlying analysis methods are no longer sufficient.