Failure Modes and Effects Analysis

Abstract

Abstract Technological developments occurring at the beginning of this century and particularly from the 1950s onward led to the construction of large, complex systems which were both productive and dangerous. The word dangerous can have many meanings: unsafe (nuclear power plants), or sensitive but essential (computer systems of a company), etc. Observations have shown that minor causes can lead to major effects resulting in undesirable events. Studies have been undertaken over the last forty years to predict and avoid these events. These studies constitute an interdisciplinary concept known as reliability . The 1970s constitute the turning period for the hardware/software dichotomy in industrial systems, and the importance of software is increasing rapidly. The main feature of software, as compared to hardware, is that it is not subject to physical breakdown due to wear and tear. The idea followed by researchers and practitioners since then is to concentrate effort in the requirement, specification, design, coding, and testing phases of development in order to ensure high quality and zero defect software at delivery time. These efforts have resulted in the accumulation of an enormous set of development methodologies and techniques accompanied by appropriated languages and tools. These languages and tools are used to improve the adherence to software production requirements, e.g., costs, schedule, and technical characteristics such as performance and reliability. Quantitative methods used in the latter steps of development such as reliability estimation or testing techniques help to increase confidence in the software. During the design phase, qualitative methods can be used efficiently for certain types of studies (safety, availability, etc.) for it is widely accepted that fixing a bug in early phases of development minimizes cost and efforts. Failure modes and effects analysis (FMEA) is one of these methods: originally developed for use in a hardware context, the technique was applied during the 1970s to software development. First its utilization by the hardware community is recieved. After a survey of FMEA's application in software, its role in the software process and its place in the software life‐cycle development are described. Procedures for its implementation will constitute the last part of the article.