Physics of Failure
One of the keys to achieving high reliability is the detailed understanding of the failure mechanisms that commonly result in equipment down-time. For priority failure modes the critical operating conditions should be understood, damage driving parameters identified and the relationship between such parameters and damage accumulation should be quantified. Such understanding can be applied to solve a range of problems including test optimisation, understanding and correcting field failures and prognostics-based condition monitoring.
The Physics of Failure approach is based on fundamental physics and engineering principles. Complex systems can often be simplified and represented by analytical models, typically resulting in some loss of accuracy. However, if the assumptions and boundary conditions for such problem reduction are clearly defined and quantified, the models may provide valuable input for the solution of complex problems. For example, Physics of Failure techniques may be applied to identify the load conditions responsible for high-cycle fatigue cracks in the roller bearing of a wind turbine gearbox or combined with statistical methods to quantify the risk of thermal aging occurring in the insulation of generator windings.
Such an approach is particularly effective in cases where the engineer is provided with limited information. Rather than being the exception, it is a general rule in the development of technical systems that all potential loading conditions are not defined a priori, all required measurement data is not available, understanding of material load capacity is incomplete, quality variations are not quantified and so on. The Physics of Failure approach aims to focus on extracting the maximum benefit from available information, whilst making reasonable assumptions to "fill the gaps". The accuracy of modelling (and hence the obtained results) is continuously increased as and when further information becomes available.
This approach to understanding and avoiding system failure has been successfully applied to a wide range of problems and technologies.