skip to Main Content

What is the right predictive maintenance strategy for you?

It has never been more important for the industry to get to grips with the advantages – and limitations – of modern predictive maintenance approaches.

Wind farm operators – and their O&M teams – face challenges in reducing downtime, operational expenditure and the overall cost of generating wind energy. And you doubtless be aware of the role that advanced predictive maintenance has to play in meeting these challenges. You may already be using this technology to take control of your maintenance strategy.

But, with a host of new players entering the market, each approaching the challenge from a subtly different angle – and at the same time making bold claims about the operational and financial impact of their approach across your turbine fleet – how can you be sure that you aren’t buying in to ‘false positives’?

If we distill the predictive maintenance offering down to its basics, there are two distinct approaches to assembling the data needed to make critical decisions. One builds on established engineering principles, using vibration, inspection and lubrication data to analyse performance and predict component failures. The other makes use of large statistical datasets, applying artificial intelligence (AI) and machine learning algorithms to detect anomalies.

Both have genuine advantages, and exciting applications for wind energy owners and operators worldwide. But let’s not forget that they have their limitations too.

Using engineering-based datasets gives you a great in-depth view of the condition of critical components. However, they are also by nature more resource-intensive to gather, which ultimately results in a deeper, but narrower perspective – and potential blind spots.

Statistical datasets are accessible at a lower cost and give a broader overview of asset conditions. But, while they cover all variables, they may lack depth – and in the end can leave asset owners and their maintenance teams responding to anomalies that, on closer inspection, pose no danger from an engineering perspective.

The truth is, in order to achieve full clarity and control in operational decision-making, the two approaches have to be combined. Use of advanced machine learning and data analytics approaches must be underpinned by robust engineering knowledge. This principle informs ONYX InSight’s entire service offering for our clients, and forms the basis for all of the advice and recommendations we provide.

AI and machine learning algorithms will be able to learn in a smarter way if real-world engineering principles are applied to guide them. Asset owners and operators who take the lead in adopting approaches that couple statistical and engineering-led datasets will be the first to truly realize the benefits offered by predictive maintenance.

All the best,

Bruce Hall


Back To Top