Predictive maintenance approaches have been proposed for a while, promising to optimize maintenance efforts and costs, but proving these effects in the field is tedious and time consuming and often leads to a disbelief in the technology. This presentation will therefore review 5 years of field experience in predictive decision support solutions to various railway stakeholders. Amongst others, the value creation in exemplified real-life cases show how wheel life can be extended by more than 20%, bearing failures can be reduced by more than 70% and wheel damages leading to disruption of operation can be prevented in >85% of the cases by implementing a predictive maintenance scheme.

Using these proven outcomes from real-life cases, challenges and lessons learned will be used to show how value creation from predictive maintenance can be realized and how organizational and technical pit falls can be avoided. For that end, the presentation will focus on typical problem areas in heavy haul operations like wheel damages, bogie performance issues, bearing failures and wheel profile wear and how those can be effectively and accurately forecasted by fully automated predictive analytics algorithms operating on wayside data to appropriately providing decision support to different roles in an organization or automatically generating work orders. It will be clearly shown how validation and gaining trust by the users is realized.

According to the experience of the authors it is possible to reach ratios beyond 90/10 of predictive to reactive maintenance actions, yielding unprecedented cost savings and benefits for reliable operation.