Clean Currents 2022

Predictive Maintenance — New Strategy for Predicting Unit Damage: Tech Demo 2 in the Innovation Power House

Hydro project owner Sira-Kvina in Norway owns 1,760 MW of hydropower. Its portfolio consists of multiple units installed in the 1960s and 1970s. With this aging fleet, the need for maintenance is increasing and so is the need for optimal timing of this maintenance.

To address this need, the asset owner is experimenting with applying artificial intelligence in conjunction with a taxonomy of all turbine and generator components, damage types, and tests. Together, the use of AI and the creation of the damage taxonomy is leading to a “hybrid” strategy of digitization of maintenance.

The strategy involves associating relevant fundamental types of damage (e.g., mechanical wear, corrosion, cavitation, fatigue, etc.) to the components of the turbine or generator and then evaluating the damage based on component criticality. Each type of damage is associated with available methods for testing, where each test is given definite and consistent criteria for assessment. Tests range from simple visual inspections conducted during operation to real-time digital tests based on data from SCADA, utilizing statistics and data science.

Demonstrators will show examples of how this hybrid strategy works and how it is already started to generate value for Sira-Kvina.

Presenter Information

Uros Stevanovic

Uros Stevanovic
Technical Manager, Electric Power
Sira-Kvina Power Company
Demonstration Speaker


Ramon Perez

Ramon Perez
AI Solutions Director
Elder Research
Demonstration Speaker