Human Reliability Assessment
In Deep Blue, we move from the theoretical position that human reliability is strongly affected by the local working conditions and organisational factors. This tenet is in coherence with what advocated by state-of-the-art models of organisational safety. We thus see quantification of human performance as context dependent, as local factors dramatically change frequency and severity of effects of human operator actions.
The first step of our approach is to build realistic activity and incident models, so to predict how humans will be influenced by the system under analysis and the consequent impact on system safety. This entails considering human behaviour in relation to both degraded equipment and organisational performance (failure case) and performance in accordance with design intent (success case).
The activity model is then linked to a task model obtained with techniques that study the role of humans in the new system, such as a Cognitive Task Analysis.
The final aim of this modelling activity is to identify those processes that are more strongly affected by human performance, with even emphasis on positive and negative contributions. In other words, to highlight processes that may present higher variability due to changes in human performance.
In line with the principles of resilience engineering, we focus on the capabilities of all components of a system to respond to safety threats in a robust yet flexible manner, to monitor risk, and to anticipate risk consequences by preparing to absorb them. The responsiveness of such processes to human error gives sufficient reasons to invest resources in making them more resilient, better reasons than those that might be provided by traditional HR frequency estimates.
In our approach, the overall purpose of the HRA is thus to support the identification or elaboration of human-related Safety Objectives, Safety Requirements, and Safety Benefits.