In our last post, we presented the service model run-to-failure. Before diving into the next models, it is important to understand their origin. The run-to-failure service model is most probably the oldest and simplest model. It does not require planning and understanding of the failure mode of the asset. As over time the assets became more complex, it also became more important to understand when it requires maintenance, mainly to avoid dangerous operation conditions or to reduce the downtime.

The field which investigates failure is called reliability engineering. It tries to capture and model failure rates and probability of failure for the components or the entire asset. The resulting research showed that most asset failures can be modeled with the so-called “bathtub” curve. The name origins from the bathtub cross section shape of the failure rate function over time. In the first lifetime period the failure rate is high, but decreases steeply, which is called “infant mortality”. Those failures result from manufacturing tolerances or defects resulting in failure of the asset. Only by carefully selecting and quality checking of components of the asset, those failures can be reduced. When the asset is “runned in”, the flat middle part with a low failure rate is reached. Those failures are random and are unexpected. Towards the end of the lifetime, the failure rate begins to increase, so called “wear out” failures begin to appear.

The bathtub curve is modelled for each component or asset individually and requires a high number of experiments to achieve a statistical significance. Although it is expensive and a lot of work in the beginning, the curve helped to develop new service models which will be presented in the next posts.