Systems, including safety related systems have become more and more complex and integrated, enabled through extensive use of software. Software tends to drive complexity, and (artificial) agents, created through the use of software, form intricate relationships, interactions, and dependencies with each other, and the environment.
This paper suggests a framework to assist in the safety analysis of complex systems, by using two different concepts: 1) the “CESM-model (CESM: Composition, Environment, Structure, Mechanisms), and 2) “The Method of Levels of Abstraction”.
The CESM-model guides the analyst towards different aspects of the system that can cause the system to become unsafe.
The “The Method of Levels of abstraction” helps the analyst to be consistent, and explicit about the abstraction level that the analysis is carried out at. Being explicit about the abstraction level guides what kind of questions that can meaningfully be asked about the system based on the current analysis.
This is not a new method for safety analysis, but a framework which can be used together with any safety analyzes method in order to be more systematic in the search for causes for unsafe system behavior. Moreover, different method can be assessed using this framework to evaluate to what degree they address aspects that are important for safety.