Select Page

With the Global Wind Energy Council forecasting moderate growth and a relatively predictable next three years, it is possible to overlook the potential economic impact on the Wind Power Industry from Industry 4.0.

With one exception, there are few P&L items that a wind farm can control. There is little a wind farm can do to impact revenue related metrics since most are based on external factors such as pricing, wind speed etc. On the cost side, the cost of capital and rental payments are not subject to influence. From a wind farm profitability perspective, the major way to generate more bottom line income is to reduce expenditures on asset maintenance and repair.

In the past, there has been limited cross-over between the domains of Operational Technology (OT) and Information Technology (IT).

Using Machine Learning to Reduce O&M Costs

Asset maintenance and repair costs contribute a considerable proportion (ranging from 40 % to 60%) of Operation and Maintenance (O&M) expenditures. For offshore wind farms, best practice O&M cost range from $0.01 per kWh in the US to an average of $0.02 per kWh in Europe.

Predictive Maintenance (PdM) is an OT-driven process that uses SCADA generated data to determine whether pre-defined control thresholds have been breached. For instance, SCADA-based monitoring of drivetrain component temperature of one turbine with multiple reference turbines within the wind farm is used to detect whether the temperature is within the reference range. If the temperature is lower or higher than the reference range, then an alert is triggered for further investigation.

Today, a typical wind farm can generate over 1,000 terabytes data per year. However, most of the sensor data is not analyzed. Using advanced, artificial Intelligence-driven Machine Learning methods, wind farms can unlock the economic potential offered by sensor data. How does it work? Thanks to the significant reduction in computing and data handling costs, algorithms can analyze vast amounts of data in real time. Instead of monitoring the specific conditions of a specific asset, the algorithms are trained to detect abnormal data behavior patterns or even correlations of abnormal behavior patterns.

There are a number of IIoT solutions available for wind turbine operators ranging from GE’s Digital Wind Farm that is built on the Predix platform to Presenso’s Automated Machine Learning for Asset Maintenance.

IT and OT: Convergence or Separation?

The challenge today is that innovations in Machine Learning are occurring rapidly and many wind farms are not yet able to adjust their operations to keep with the change. First, from an organizational perspective, there are still disconnects between IT and OT organization. The cause for these are well known and include the typical turf-war issues (budget, political etc.) inherent in technology disruptions. Second, IT departments will not have the domain expertise in areas such as Machine Learning and Artificial Intelligence.

Given the lack of internal expertise in Machine Learning, it may be tempting to separate IT and OT. From my experience with multiple organization implementing Machine Learning for Asset Maintenance (including the Wind Power Industry), adoption of IIoT practices using big data is dependent on close coordination between the disciplines.

The major challenges that organizations face are how to access the sensor data that is already owned. In many cases, third party vendors act as gatekeepers, thereby giving them de facto ownership of a valuable resource. There are many scenarios where the interests of the vendor and the wind farm operator are not aligned.

It is unrealistic to expect IT to gain deep domain expertise in the areas of Machine Learning. At the same time, for a wind farm to use Machine Learning for Asset Maintenance, IT will need to play a critical role in the following areas:
Data Governance. Putting aside the issue of data ownership, there are numerous processes for data cleansing, storage and access required for IIoT to be implemented. IT groups are often under-funded and/or over-worked. Without access to clean data in real time, the promise of IIoT will not be realized. If IT plays a constructive role, it can help create the data infrastructure needed to support IIoT.
Standards and Processes. IT groups will need to be responsible for setting standards for network access and security protocols. Standards need to be consistent across the organization and should include IIoT related spheres.

Machine Learning: Can IT Develop Domain Expertise?

The areas of data governance and standards outlined above are within the current areas of responsibility of a typical IT group. The challenge for both IT and OT groups in the Wind Power Industry is that there is very little depth of domain expertise in the areas of Artificial Intelligence and Machine Learning. Even if there is a full alignment between both parts of the organization, there is a dearth of big data scientists and engineers.

The existing challenges of remote asset monitoring and the relative high cost of onsite repairs (including the Mean Time to Repair (MTTR) for both onshore and offshore locations) stem from the geographic isolation of wind farms. Given its existing constraints, it is unlikely that the Wind Power Industry is likely to recruit sufficient quantities of highly skilled data scientists into this type of environment.

Can IT develop domain expertise in Machine Learning? Perhaps, but not in the near future.
Recognizing this limitation, wind farms that wish to apply Machine Learning for Asset Maintenance to their turbine assets will need to find alternatives approaches. For instance, Presenso provides a SaaS-based solution and Analytics are provided as a service. This eliminates the need for internal Machine Learning domain expertise.

Conclusion

About two decades ago, internet access became ubiquities and companies were forced to adjust their operations to reflect the online behavior of their customers. As a result, new roles and groups were created within companies. At Industry conferences I attend, I am meeting more Chief Digitalization Officers and VP’s of IIoT. This trend will likely continue and it is incumbent upon wind farm management to create an organization that can support Industry 4.0 transformation.

 

Please share if you like above articleShare on FacebookShare on Google+Tweet about this on TwitterShare on LinkedIn
Eitan Vesely

Eitan Vesely

Formerly a hardware specialist and a support engineer for Applied Materials. Specializing in software-hardware-mechanics interfaces and system overview. Experienced in the field of industrial automation and motion control. Holds a BSc. Mechanical engineering

[:en]

[:]