The Transverse Horizontal Axis Water Turbine (THAWT) has been proposed as a tidal device which can be easily scaled and requires fewer foundations, bearings seals and generators than a more conventional axial-flow device. The THAWT device is a horizontally deployed variant of the Darrieus cross-flow turbine, in which the blades can be oriented into a truss configuration to produce long, stiff multi-bay rotors. This thesis establishes and combines a set of numerical models, which predict the hydrodynamic and structural performance of the THAWT device, with sufficient confidence to assess the feasibility of such a device at a full scale installation and to optimise its performance. Tests of 1/20th scale experimental models of the THAWT device have demonstrated that the truss configured device is capable of producing power with an efficiency close to that of the parallel configured turbine. In addition, variations in the configuration of the scale models have indicated how several design parameters affect the hydrodynamic performance of the device. A two-dimensional Navier-Stokes blade element model has been developed, in which the THAWT device is represented using an actuator cylinder. The addition of a hydrostatic free surface deformation correction has resulted in a model which is capable of matching experimental results with sufficient fidelity and accuracy. Blade loads from the numerical hydrodynamic model have been applied to a beam finite element analysis, to predict the stresses induced in the hydrofoils of the THAWT device. The numerical and hydrodynamic models are combined with a Linear Channel Momentum model to predict the performance of the THAWT device at a full scale tidal location. The effect that the device has on the channel flow indicates how much energy is available for extraction and how this energy might be most efficiently obtained. When considering material fatigue the analysis suggests that the structural design considerations dominate over the hydrodynamic considerations.

Cette thèse de doctorat a été réalisée dans le cadre du projet HARVEST, programme de recherche initié en 2001 au LEGI et consacré au développement d'un nouveau concept d'hydrolienne à axe vertical inspiré des turbines Darrieus pour la récupération de l'énergie cinétique des courants marins et fluviaux. Ce travail s'est focalisé sur la mise en place d'un moyen de mesure par Vélocimétrie par Image de Particules deux dimensions – deux composantes (2D-2C) et deux dimensions – trois composantes (2D-3C). L'objectif est d'une part de constituer une base de données expérimentale pour la validation locale des simulations numériques RANS 2D et 3D menées dans le cadre de travaux précédents, et d'autre part d'améliorer la compréhension des phénomènes hydrodynamiques instationnaires rencontrés dans ces machines et en particulier du décrochage dynamique. La confrontation des mesures expérimentales et des simulations a notamment permis de mettre en évidence les points forts et les limites des modèles numériques dans les différents régimes de fonctionnement de la machine. / This PhD thesis has been carried out within the framework of the HARVEST project. This research program, initiated in 2001 by the LEGI laboratory (Grenoble, France), is devoted to the development of a new marine turbine concept inspired from Darrieus turbines in order to convert kinetic energy of marine, tidal or river currents into electric energy. This work has been focused on the development of an experimental apparatus based on two dimensions – two components (2D-2C) and two dimensions – three components (2D-3C) Particle Image Velocimetry. The objective is to provide an experimental database for the local validation of 2D and 3D RANS computations and to improve our understanding of unsteady hydrodynamics phenomena experienced in this type of turbine and especially of dynamic stall. Comparison between measurements and computations enabled to identify strengths and limitations of numerical models for various operating conditions of this type of turbine.