BIS Big Data and IoT Conference

The Big data and IoT conference in Berlin has now concluded after two days of discussions about data analytics.   There were many interesting points raised with a good mix between academic and industrial views.  For me the clear takeaway points were:

• Data is rather dirty, and needs to be cleaned before it is useful.
• 95% of useful data analytics is really simple stuff (once your data is sufficiently clean).
• You need to know what the data is about and have a clear problem which needs to be solved in mind.
• The people and culture need to be brought together to gain advantages from big data.

There were several talks relating to power curve analysis.  Never believe a power curve until you have filtered out every possible reason for an outlier, and sampled data is better than averages for evaluating power curves.  The use of pairs of turbines to evaluate any changes was suggested, on the grounds that modifications to improve the power curve may affect the nacelle anemometer. 

A couple of presenters highlighted cheap options for extracting more data from the turbines, but the manufacturers showed that there is already quite a lot of data around, and showed how they are using it to reduce the costs of delivering services to their customers, or increasing their output.

It was interesting to see how the views about the use of big data seem to be converging.  The common themes are that CMS systems (vibration) provide very valuable diagnostic information, and that this can be supported by SCADA analysis to identify other turbine problems, and to monitor output.  The next step is to really integrate the recording of work orders and inspection report data into the system, with several speakers suggesting much greater structure is needed in these reports.  At Uniper we have already a good feedback loop created  through our Production Support Portal, which has been successful in encouraging feedback from the site teams, albeit in a rather unstructured form.  My talk seemed to be well received, with several people commenting that it was good to see some concrete examples of what can be achieved.

So this has been an enjoyable and thought provoking couple of days, with the clear message that Big Data is already helping to transform maintenance practices, but there is a lot further to go.