| Abstract | The problem addressed was that of developing a modelling tool to deduce reliability indices for general engineering system configurations. A prototyping development philosophy was adopted for the software aspects of the model, within the environment where it is to be used. This was to ensure end-user familiarity, and acceptance. Large complex engineering systems are often composed of a number of smaller sub-system building blocks which can be further broken down into their constituent units. Ideally models of complex engineering systems should have the same type of structure for whatever purpose they are required.
The overall problem was considered from the highest possible level, and an attempt was made to discover the answer to the question "How can the reliability of a general engineering system be modelled so as to include all of its high level functionality and yet miss none of the detail". The answer proposed here to the above question is one of a hierarchical realisation. The system, as viewed from the highest level, can be considered to be a simple network (composed of components, connectors, and directed lines) whose dimensions are constrained by the dimensions of a typical VDU screen. Each component of that network can potentially be another network, and indeed each component of any network can be yet another. This process of creating descending hierarchies has to end, however at a level that Page 1 Abstract has been chosen as the base, where all the components are of a pre-defined type, and selectable from a pallet of 204 such components that can comprise the diagram (system) as a whole.
The resultant system is termed a diagram of hierarchical networks in a general sense, because the constraints that make it a reliability block diagram model are not imposed until the time of analysis. It is at this stage that the database which accompanies each component (and individual network) is examined to determine its nature, and to recover necessary pre-defined reliability data. Each individual network created below the highest level can be considered to be encapsulated within a single super-component at the level of its mother network. Its database is very important because, although it is mostly empty awaiting results derived from lower levels of hierarchy, it contains information relating to how the network is to be processed. Users, have at their disposal an ergonomically efficient hierarchical graphics editor to produce these diagrams which are, in turn, directly interpreted. The resultant is a diagram that can be structurally changed, and re-analysed without there being a need to re-enter information at the keyboard.
Within this implementation, two types of analysis algorithm have been utilised: Minimal Cut Sets, and k-out-of-n:F analysis. Both have been substantially engineered to fit this modelling tool, and when used in conjunction are capable of representing a large variety of the system constructs that appear within reliability block diagrams.
User interface design has also been of prime concern with text based re-locatable windows, and a graphical diagram editor designed for high ergonomic efficiency given a wide range of potential end user. |
|---|
