Numerical investigation of a free standing horizontal axis tidal turbine

The thesis describes a set of studies carried out in parallel with a tidal turbine design program which was undertaken by the Turbomachinery Group to which the author belonged to during the duration of this project. Therefore the work presented in this thesis constitutes an exploration of the physics of horizontal axis tidal turbines and of the modelling issues associated with the simulation of these devices. Specifically the investigation centered on the behaviour of the turbine when exposed to the influence of tidal channel waves. The numerical analysis of the free standing, gravity stabilised horizontal axis tidal turbine (HATT) was carried out with the application of Speziale’s Reynolds Stress Model (hereafter know as Speziale, Sarkar and Gatski or SSG model) available in the CFX CFD code. The use of a number of turbulence models was investigated but poor convergence or starting difficulties led to the employment of the SSG Reynolds Stress turbulence model. Simulations for a range of test cases were undertaken ranging from single blade passage cases to transient simulations which included the motion of the turbine and the variation of the flow velocity due to the combined action of waves and the shearing effects of the sea bed boundary layer. A number of CFD results were compared to experimental data acquired from tests conducted on a scale model in the summer of 2009 at the IFREMER test flume in France. An additional source of comparison is provided by data obtained from a BEMT code produced by the CU consultant, Mr Chris Freeman, (Freeman et al., 2009a). A number of numerical models were assembled to analyse the effects of the presence of the pylon and blade-pylon spacing. These were run as steady and unsteady cases. For the steady cases several angular positions were examined to investigate the effects of the transit of the blades through the pylon potential field and across the sea bed boundary layer shear flow. An idealised no-pylon case was analysed to compare with the equivalent model with pylon. On average there was a performance increase of 2% for this configuration when compared to the case with pylon for the datum spacing. The simulations covered four turbine rotational speed cases. These correspond to a startup condition and to rated power, with an intermediate point, and to an overspeed regime. In the overspeed condition the power is essentially unchanged but the thrust reduction is strong. Additional investigations covered the performance of the turbine when yawed and the influence of inflow turbulence. The comparisons between the solutions obtained from steady-state and transient simulations showed that the unsteady approach is preferable to describe these types of flow. Two waves were employed in conjunction with the transient models. The first corresponded to a moderate sea state (1:5m height and 10:0s time period) of the type occurring more frequently. The second case involved a substantial wave (3:0m height and 14:0s time period) which is associated with storm conditions. The datum models which incorporated the most energetic wave showed similar values for the torque observed on the transient simulations without waves for the overspeed and rated power cases respectively. This is a significant finding given that the 15m diameter turbine is immersed in a 35m tidal channel. The peak value for the axial thrust and torque on the large wave simulation is on average 86% and 90% higher than the steady state thrust and torque respectively for the rated power case. The loads imposed on the rotor system for the large wave are approximately 3% higher than those of the regular wave. Similar detailed studies for tidal turbines are non-existent at time of publication. The work is therefore unique in the scope and breath of the simulations it contains. The computational resources required were vast. Transient simulations including wave effects took three weeks of computation time utilizing sixteen processors. Each of these cases required about several terabytes of storage space to record the intermediate transient results.