The objective of the thesis is to present studies addressing the question: How do we identify that a system drifts or experiences sudden changes in the safety space? Hence, the studies in this thesis focus on the identification of safety performance indicators. This identification is influenced by the way safety is understood. If safety is seen as the absence of accidents and failures, then indicators refer to failures, malfunctions or deviations. Safety critical organizations like airlines, air navigation service providers or oil companies have implemented numerous improvements by using such lagging indicators. These lagging indicators by their nature provide information after the fact. Nevertheless, today’s systems and organizations must be able to function in rapidly changing environments in which there is a great deal of uncertainty. The Resilience Engineering perspective does not see safety as the absence of failures but as something that the organization or the sociotechnical system does. Thus successes and failures are related to the capability of the system or organization to adjust and continue operations in the presence of continuous changes and operational constrains. Given such a perspective, finding indicators that allow an organization to act before something happens, i.e. to be leading, rather than reactive and lagging, is a main challenge.
In order to address this challenge, the work starts with a systematic review of existing methods that could be applied for the identification of safety indicators. The thesis explores methods accounting for the monitoring of failures, deviations and everyday performance. In everyday operation, there are many successes i.e. flights arrive on schedule without significant problems. Hence, the empirical part of the work deals with the understanding the deficiencies and strengths of specific incidents and daily operations. Established and relatively new methods are tested. Each method represents a different perspective on safety having an influence on the identification of indicators. Using a multidisciplinary analysis that combines methods with different perspectives provides broad understanding. The studies document the application of the following methods: 1) Triangulation using a list of outcome and activity indicators, 2) Sequentially Timed Events Plotting (STEP method), 3) storytelling and 4) the Functional Resonance Analysis Method (FRAM method). Findings from the studies and their limitations have been presented and discussed with industry and the research community.
It is essential to understand how a system operates under operational and financial constraints in order to identify that the system drifts or experiences sudden changes. Therefore, the main argument in this thesis is the identification of indicators related to failures and everyday successful operations. For that purpose, it is reasonable to distinguish from among three types of performance indicators: 1) lagging indicators, which refer to what has occurred in the past; 2) current indicators, which refer to what is occurring now; and 3) leading indicators, which refer to what may occur in the future. While, there is considerable data available for lagging indicators, a balanced composition of lagging, current and leading indicators is needed. Hence, the thesis represents a step forwards for the identification of current and leading indicators. The test of the methods provides a tool box for industry. The work also contributes to develop the FRAM method. While most of the subjects are related to aviation some ideas and methods presented in this thesis are useful to other safety critical organizations such as industry in the nuclear, oil and gas or railway sectors.