Using a Software Safety Argument Pattern Catalogue: Two Case Studies

Abstract

Software safety cases encourage developers to carry out only those safety activities that actually reduce risk. In practice this is not always achieved. To help remedy this, the SSEI at the University of York has developed a set of software safety argument patterns. This paper reports on using the patterns in two real-world case studies, evaluating the patterns’ use against criteria that includes flexibility, ability to reveal assurance decits and ability to focus the case on software contributions to hazards. The case studies demonstrated that the safety patterns can be applied to a range of system types regardless of the stage or type of development process, that they help limit safety case activities to those that are significant for achieving safety, and that they help developers nd assurance deficits in their safety case arguments. The case study reports discuss the difficulties of applying the patterns, particularly in the case of users who are unfamiliar with the approach, and the authors recognise in response the need for better instructional material. But the results show that as part of the development of best practice in safety, the patterns promise signicant benets to industrial safety case creators.