Author(s): Martin Wosnik; Peter Bachant
Linked Author(s):
Keywords: Cross-flow turbines; Marine hydrokinetic turbine; Tidal energy; Turbulent wake
Abstract: As designs for cross-flow turbines (CFT) for use in wind and marine hydrokinetic (MHK) energy conversion mature, the research focus is shifting from individual turbines to the optimization of turbine arrays or farms. Optimizing CFT arrays requires detailed understanding of turbine wakes–specifically their interaction with each other and the mechanisms by which energy and momentum are transported from the free stream into the wake. Performance and detailed near-wake characteristics of two three-bladed, vertical axis, “reference”cross-flow turbines were measured in a large cross-section towing tank at the University of New Hampshire (UNH) .The two turbines were the UNH Reference Vertical Axis Turbine (UNH-RVAT) ,with 13. 4 percent rotor solidity (= (8/) ,and the U. S. Department of Energy Reference Model 2 (DOE RM2) ,with 4. 7%solidity. The near-wake was examined using acoustic Doppler velocimetry, where essential features regarding momentum, energy, and vorticity are highlighted. Dominant scales and their relative importance were investigated and compared at various locations in a measurement plane at one diameter downstream. The lower-solidity RM2operates at a sufficiently high tip speed ratio at peak performance to avoid dynamic stall, and generates significantly less turbulent kinetic energy in the near-wake. However, when estimates for the terms in the mean streamwise momentum and mean kinetic energy equation were computed, it was shown that for both turbine wakes the unique mean vertical velocity field of these wakes, characterized by counter-rotating swirling motion, contributes significantly more to mean kinetic energy recovery in the wake than the turbulent transport. This result sheds light on previous CFT studies showing faster downstream wake recovery compared to axial-flow turbines. Implications of these insights on CFTs in array configurations are discussed. Numerical techniques are generally better suited to explore the design parameter space for large arrays than physical models, since adequately scaled experiments–at sufficiently high turbine and blade Reynolds numbers–can be expensive. For accurate prediction of array performance, accurate representations of turbine-wake interaction and wake recovery are important. Shortcomings of commonly used models within Reynolds-Averaged Navier--Stokes (RANS) simulations, e. g. the actuator disk model, to predict the nearwake structure are analyzed, and a path forward towards improved parameterized engineering models to accurately predict the near-wake physics of CFTs is discussed.
Year: 2015