All publications

Journal article · 2026

Uncertainty analysis of oscillating water column experiments under regular and random wave conditions

Renewable Energy

Y.R. Almalki, I. Karmpadakis

DOI
10.1016/j.renene.2026.126002
Publication
published
Volume
271
Article number
126002
Read at the publisher ↗

Commentary

How confident can we be in a laboratory measurement of wave energy performance?

Laboratory testing is essential for developing wave energy technologies, but every measurement carries uncertainty. Understanding that uncertainty is critical before experimental results are used to justify larger prototypes, field deployment or investment decisions.

This study systematically quantified the uncertainty in Oscillating Water Column experiments under both regular and realistic random waves. Repeated testing, high-accuracy optical measurements and Monte Carlo analysis were combined to distinguish measurement error from the natural variability of the waves themselves.

The experiments showed very strong repeatability, with measurement variability generally below 1%. The expanded uncertainty in average power output was below 1% under regular waves and approximately 2.5% under random waves, where part of the additional uncertainty comes from the natural randomness of realistic sea conditions.

Box-and-whisker style charts of relative uncertainty in percent for seven quantities used to assess OWC performance: χ, η, mean η_owc, dη_owc/dt, p, P_owc and mean P_owc. Three columns show Type A uncertainty, Type B uncertainty and expanded uncertainty. The top row (black) is for regular waves and the bottom row (blue) is for irregular waves.
Putting confidence bounds on wave energy experiments. The uncertainty analysis separates experimental variability, measurement uncertainty and their combined effect across the main quantities used to determine OWC performance, comparing controlled regular waves with more realistic random sea conditions.

Source: Almalki, Y.R. & Karmpadakis, I. (2026), Renewable Energy, 271, 126002, Fig. 13. Licensed under CC BY 4.0.

The broader message is simple: a performance number is much more meaningful when we also know how much confidence we can place in it.