Abstract While climate impacts on hydropower output are well-documented, plant efficiency, the critical ratio of electrical energy generated to hydraulic energy input, remains an underexplored metric, particularly in data-limited regions. This study analyzes the efficiency dynamics of the Ruzizi I plant (29.8 MW) from 2000 to 2023 to unravel the interplay between hydrological drivers and operational constraints. Building on the established context of a hydraulic trade-off between water volume and head, we employed machine learning (Multiple Linear Regression, Random Forest, Gradient Boosting) and operational analysis to diagnose efficiency drivers. Results reveal that plant efficiency increased significantly (+ 3.6%-points/decade) and is overwhelmingly governed by discharge ( r = 0.998), with machine learning models confirming the negligible role of head and seasonality. This indicates that efficiency gains are almost entirely flow-dependent, masking the potential negative impact of head loss. The system exhibits strong buffering from Lake Kivu, with efficiency remaining stable during drought but surging by 17–18% during wet years. Crucially, operational analysis identified an optimal load factor range (78–82%) that could improve efficiency by ~ 4% points compared to historical operation. However, a concurrent decline in available capacity factor (− 5.5%/decade) signals emerging non-hydrological constraints. These findings underscore that while water volume currently dominates efficiency gains, long-term sustainability requires managing sediment-induced head loss and optimizing operations within the identified optimal range to mitigate the underlying vulnerabilities in the energy conversion process.
Mugisho et al. (Thu,) studied this question.