Our CEO, Eitan Vesely was featured in the September 2017 edition of Power Magazine. We have summarized the key points in the article. Specifically, Eitan addresses the following:
- How can Machine Learning change the economics of Operations and Maintenance (O&M) in a windfarm?
- What is the financial impact of turbine downtime on a wind farm?
- A comparison of Traditional Predictive Maintenance (PdM) and Machine Learning Predictive Maintenance.
Summary and Key Insights
- O&M spending accounts for up to 30% of the levelized cost per kWh produced over a turbine’s lifetime.
- Turbine downtime is between 7 and 10 days a year.
- Most wind turbines are in remote locations which drives repair costs and repair time up. Crane rental expenses are particularly high and it typically takes more time to transport machinery than it does to repair it.
- Traditional Predictive Maintenance (PdM) is based on manually set control thresholds. There are limitations to the effectiveness of these SCADA systems (only breaches to control thresholds are monitored and most wind turbine failures are not identified).
- With Unsupervised Machine Learning, algorithms detect abnormal turbine sensor behavior and can send alerts of Time to Failure (TTF) and provide Root Cause Failure Analysis (RCFA).
- Presenso Machine Learning for Predictive Asset Maintenance is sensor and asset agnostic and can use the data from any machine and vendor.
- The potential savings are significant: a 20% reduction in the repair and maintenance portion of O&M would yield an average annual cost savings of $11,383 for a 2.5-MW turbine and $34,148 for a 7.5-MW turbine.
For the full article, please click on this link.