An Experimental Comparison of System Diagrams and Textual Use Cases for the Identification of Safety Hazards

Requirement defects are more costly to correct the later in the development process they are discovered. The same applies to safety requirements, and defects that remain in the fielded system are then not only costly, but potentially life-threatening. It is important to discover safety hazards as early in the process as possible, and it is thus interesting to integrate safety analysis with techniques used in the early stage of requirements engineering. This paper describes an experiment comparing how well two system diagrams and textual use cases support non-experts in identifying hazards in a simple control system. Results show that system diagrams were better for finding hazards related to peripheral equipment, while for all other kinds of hazards textual use cases were as good or better. Not only the type of representation matters, but also how the information is brought into focus for the analyst, as this might steer the analyst towards noting some hazards but ignoring others. An Experimental Comparison of System Diagrams and Textual Use Cases for the Identification of Safety Hazards