The solution of maintenance problems is often linked to the reliability of machinery. Even if the equipment is well designed, well maintained and carefully operated there is always a small chance that it will fail in service, and it is uncertain when this failure will occur. Therefore the analysis of engineering reliability deals in probabilities (i.e. likelihoods of success or failure) and probabilistic variables (i.e. quantities that vary randomly, such as times-to-failure). In particular, it deals with the application of statistical techniques to the analysis of patterns of component and equipment failure – and acting on the information thus gained (e.g. replacing or repairing a component at timely intervals, or redesigning a complete engineering system, etc) in order to increase system reliability.
A cost-effective remedial action to improve the reliability of plant should first address those areas of it, the 'vital few' that are the dominant causes of its failure (i.e. the 20% of its component units which cause 80% of its unreliability), and leave until later the 'trivial many' (the remaining 80% of the units which cause just 20% of its unreliability), or even ignore some of these altogether. AIE utilises Pareto Analysis to identify these systems or components, in order to target where efforts to improve reliability will be most effective.
The distribution of times-to-failure of plant items can often be quite well represented by:
- Normal – if they are subject to (intrinsic) wear-out failure;
- Exponential – if subject to quite randomly occurring (extrinsic) failure;
- Hyper-exponential – if subject to (intrinsic) running-in failure.
However, there is one particular distribution, Weibull, which represents any of these three types of failure. This has been found to be particularly useful for the analysis of the variation of item lifetimes. In addition, it has two other practical virtues,
- Applicability via simple graphical techniques,
- The use of defining parameters (i.e. the terms in the formula) which have an everyday engineering significance.
AIE can conduct Weibull Analysis in order to determine accurate representations of failure rates for use in statistical analysis when conduction optimisation of a maintenance programme.
Reliability Block Diagrams
With complex engineering plant, AIE can represent the plant in Reliability Block Diagrams in which the various blocks and the lines joining them represent the constituent sub-systems or units and their interdependencies -their inputs and outputs, to each other, of material, power, control etc. This allows for the assessment of the overall system or component level reliability.
Fault Tree Analysis
Fault Tree Analysis is one of the most common reliability assessment techniques. AIE analyses undesirable failure events of systems by considering the interactions of the components contained within it. To use the approach a fault tree representing the causes of a given failure is constructed, from which both a qualitative and quantitative analysis can be performed.
Event Tree Analysis
When the sequence of failure events is important then the technique of Event Tree Analysis is recommended for use. If the model of the system being analysed is very complex, there is dependencies between the components within the system, or there is a duty/standby consideration then the technique of simulation can deal with these complications.