Multidisciplinary system design optimisation of oscillating water column power plantsa nonlinear stochastic approach

Resum

This thesis presents the Multidisciplinary System Design Optimisation of Oscillating Water Column (OWC) Power Plants. This work is based on a stochastic nonlinear approach. A novel Vane-Less Contra-Rotating Turbine (VLCRT) has been designed which outperforms any of the existing architectures both in terms of on- and off-design characteristics. A multi-fidelity design framework together with optimisation genetic algorithms was used to develop and improve this OWC component. A comprehensive state of the art review, including both early patents and available studies in the open literature on this technology, has been included in this work. The analysis methods and strategies used at component and system levels studied, for which advantages and limitations have been ascertained. A nonlinear stochastic model for water column oscillations has been developed based on conservation laws and a collocation method. The majority of the available models in the literature are linear which validity is limited to small amplitude oscillations. The results presented in this work have shown that nonlinear terms could have a fundamental impact on the motion of the fluid. The nonlinear stochastic model has been validated and calibrated using unsteady two-dimensional multi-phase CFD RANS simulations in ANSYS™. A matrix of caisson designs was generated using an Optimised Latin Hypercube Design of Experiments such that the likely design space is covered. Correlations for the empirical coefficients in this model were also developed. This strategy ensures the model accuracy whilst greatly reducing the computational burden. A novel axial Vane-Less Contra-Rotating Turbine (VLCRT) for OWC power off-take has been designed using an automated multi-fidelity framework and optimised via stochastic evolutionary algorithms. The optimised VLCRT has several benefits: efficient operation in the aforementioned conditions, low outlet swirl, and acceptable mechanical steady stress. The design framework is integrated by a core tool based upon the Non-Isentropic Simple Radial Equilibrium (NISRE) equation which together with the radial equilibrium through-flow solver Vista™TF are utilised to explore the design space. Unsteady Reynolds-Averaged Navier-Stokes (uRANS) simulations validate and exploit the resulting design from the previous design stage. A Systems Engineering approach has been taken for system design and analysis. The Voice of the Customer has been translated into functional and non-functional performance requirements. Systems architecting based on system functionality and functional interface has been used to group components into sub-systems. A computationally efficient system model has been proposed which included all the components: water column, air system, power turbine, drive train, and electrical generator. The solution strategy allows for efficient simulation of power plant performance under stochastic wave conditions. A multidisciplinary Design System Optimisation problem has been defined based on the former requirements and models. Two optimisation problems have been presented: (1) caisson optimisation, and (2) Multidisciplinary System Design Optimisation (MSDO). The results from the simpler problems have been fed into the more complex problems to enable faster convergence. Optimum designs for each location has been demonstrated to vary when considering component optima or system optima. Therefore, a collection of optimum components do not lead to an optimal system, and hence, full-system optimisation is required. The optimisation studies have been carried out considering the measured sea wave spectrum at four different locations: (1) Bimep (Spain), (2) Wave Hub (UK), (3) Pilot Zone (Portugal), and (4) Galway Bay (Ireland).