Availability of multistate systems subject to hard and soft failures.

摘要

The purpose of this research is to model the availability and partial availability of components, subsystems, and systems which may fail completely (hard failure) or may continue to operate but at a level below the original design specifications (soft failure). The measure of performance used for this study is availability. Since availability is commonly measured by a ratio of the operating time of a system to the operating time plus the downtime of the system, it includes both reliability and maintainability considerations.; A Markovian approach was used in this research to model systems, subject to both hard and soft failures. If the system completely ceases to operate and is completely unavailable, a hard failure has occurred. If the system operates at some reduced level, but some useful output is produced, then a soft failure has occurred. This study presents a numerically based availability alternative. The study focuses on the concepts and mechanics necessary to model the availability, partial availability, and unavailability of simple systems made up of a few components. A combination of classical Markovian techniques and simulation methods are used in the model.; Extensive examples are developed which illustrate the usefulness of the model. Cases involving non-series, non-parallel configurations and both two- and three-state availability models are developed. In addition, cases are developed with and without repair and preventive maintenance operations. All cases are supported with full input-output documentation.; The model and code are generic in nature and may be applied to a number of applications, as long as the application generally conforms to the assumptions stated. The model has been developed for the microcomputer. Output may be directed to a hard copy device, to a file on a hard disk, or to a diskette. Graphical availability plots may be produced from the output data files by using existing graphical software. The number of components may be increased up to a maximum of four. Parameter information may be input from a file. This feature is especially useful in conducting sensitivity analyses or for studying the effect of component state configurations on system availabilities. The computer code was written in the C-language.