Hit and miss are closer together than you might think. A story from the reality of preventive maintenance.

Preventive maintenance is the holy grail of data analytics. But in the same way that the holy grail is something that is impossible to find, achieving a 100% fault detection rate for preventive maintenance is also impossible. This is a matter of statistics.

With this in mind, we want to be transparent about what is possible in the preventive maintenance field using 10-minute SCADA data (the hit) and what is not (the miss).

Together with our customer ImWind, we want to share a true story about such a scenario.

The application i4SEE Heat monitors and simulates multiple temperatures within the wind turbine, including the main bearing temperatures.

Even though the ImWind fleet is monitored using i4SEE Heat, a main bearing failed without any prior indication. Why is i4SEE so sure about that? This is due to the continuous improvement process that is embedded in everything we do at i4SEE. In this case a deep investigation was automatically triggered.

We looked at every detail and every SCADA signal, to check for any patterns that were abnormal before the breakdown. Additionally, we checked through all event logs, service reports and other activities on site and confirmed that prior to the failure, there was no indication of any abnormal behavior.

This might show the limitations of using only SCADA-based condition monitoring or it may indicate a rapid breakdown of the bearing without any measurable indicators in advance. Maybe such events could be spotted in future with additional sensors, but the business case needs to be checked carefully. It should be recognized that the closer you want to be to 100% detection rate, the higher the overall cost of your monitoring strategies.

One thing is certain, ImWind and i4SEE aim to make sure that nothing is overlooked using the available data. In the case in question, it was confirmed by both ImWind and i4SEE that this requirement had been satisfied. Therefore, although the fault could not be detected in advance, we are confident that no amount of time-consuming manual analysis work could have foreseen this event. All available data was used to its maximum potential as part of an automated, scalable and cost-effective analysis.

Furthermore, there are also plenty of positive cases and interesting findings to share. By using i4SEE Heat™ to monitor its portfolio, ImWind is constantly supported in the detection of abnormalities in temperature signals and can address them with their maintenance providers in a precise and efficient way. Even with ImWind’s full-service contract approach, the service partner appreciates valuable insights. Issues such as temperature abnormalities on auxiliary transformers, can be repaired by the maintenance teams before a costly downtime and revenue reduction occurs for both the service partner and ImWind.

This is why Christian Felling from ImWind states: “The i4SEE applications provide wind turbine operators with value at many levels, in good times and in bad”.