Modelling dynamic blade loads on scaled tidal turbines due to wave-induced unsteady kinematics

Abstract

Accurate prediction of wave-induced unsteady loads on tidal turbines is essential for reliable fatigue assessment and load mitigation design. These loads arise from the turbine rotating through a depth-varying flow field that oscillates at the wave frequency, requiring detailed modelling of the resulting rotor and blade loading. Many engineering models are based on quasi-steady formulations, and limited studies have examined the influence of dynamic effects on blade and rotor loading. Validation against physical datasets with well-characterised flow conditions and simultaneous rotor and blade loads remains critically important. This study investigates the influence of dynamic stall on two hydrofoil sections---NACA63-415 and NACA63-815---used in experimental studies of horizontal-axis turbine loading in waves. Two-dimensional, blade-resolved unsteady Reynolds-averaged Navier--Stokes (uRANS) simulations are compared with two formulations of the Beddoes--Leishman model---the original and the Sheng variant---to evaluate their performance in predicting unsteady loads on tidal turbines with different hydrofoil types and geometries. The uRANS results are used to examine stall onset delay and Leading Edge Vortex (LEV) dynamics across a range of reduced frequencies and mean angles of attack, followed by model calibration. The more cambered NACA63-815 exhibits reduced sensitivity to unsteadiness compared with NACA63-415. The BL model better captures deep-stall hysteresis, while the Sheng model reproduces peak lift with less calibration effort. When applied to a rotor blade subject to wave-induced kinematics represntative of experimental conditions, both models predict similar turbine-level responses, with unsteady loads most significant near the hub. Dynamic effects alter mean thrust and Root Bending Moment (RBM) by about 2-3\%, whereas cyclic fluctuations increase markedly---thrust and RBM standard deviations increase by up to 85\% and 62\%, respectively---highlighting fatigue-relevant effects and identifying conditions where dynamic stall modelling is essential.