During the design stage of the development of a new system, automated fault tree construction would produce results a lot sooner than the manual process and hence be highly beneficial in order to modify the system design based on identified weakest areas. Although much work has been performed in this area, the construction of fault trees is still generally done manually. In this paper, a new methodology of constructing fault trees from a system description is proposed. Multi-state input/output tables are introduced, which have the capability to capture output deviations during the normal operation of a component as well as under the influence of abnormality or failure. Two libraries, namely, a component library and a mark library, are introduced. The former stores component models and the latter stores a range of marks. The main purpose of a mark is to identify a certain feature of the system, such as a feedback loop or multiple redundancies. These two libraries are used to redraw the system in a graphical environment where the designer can witness the system come together and also input the necessary failure data for each component. An algorithm has been developed, that uses input/output tables and marks, to automatically construct fault trees for failure modes of interest. In order to demonstrate this methodology, it is applied to an automotive emission control system, and a fault tree is generated using the algorithm developed in this work.
Read full abstract