Performance database interpolation and constrained nonlinear optimization applied to propulsor blade design

Highest efficiency satisfying prescribed requirements has always been the objective of marine propeller design. As the loading of propeller is increased due to the growing demands of larger and faster ships, cavitation becomes an important issue requiring special consideration. Furthermore, the complexity of modern propulsion systems and unsteady non-axisymmetric inflow have made the design procedure much more complicated and challenging. In general, a marine propeller is designed by using either series charts based on experimental results like B-series of MARIN (van Lammeren et al. 1969), or numerical tools such as the lifting line method and the lifting surface method, as those developed by Prof. Kerwin at MIT. Although the series charts have been well applied to conventional propeller design, the application to non-traditional propellers like ducted or podded propellers is not well established. The conventional numerical methods design a propeller in two steps. At the first stage, the optimum circulation distribution is determined by using lifting line theory satisfying the loading constraints. In the second step, the blade geometry is determined by using lifting surface method to produce the required loading distribution. The new design method (named CAVOPT-BASE) presented in this thesis couples vortex/source lattice method (MPUF3A), database interpolation (least squares method called LSM or linear interpolation method called LINTP) and constrained nonlinear optimization to determine the optimum blade geometry under given operating conditions and loading constraints. At the first stage, a blade geometry is selected as the base geometry to generate a propeller family by multiplying the base geometric parameters with corresponding factors. The performance database associated with the propeller family is constructed by MPUF3A. The objective and constraints functions are then approximated from the database by LSM or LINTP linking propeller performance with design variables. The approximating functions are incorporated into the constrained nonlinear optimization algorithm to solve for the designed blade geometry with the highest efficiency. In the case of ducted/podded propellers design, the duct/pod geometry is supposed to be given and remains unchanged during the design procedure. When a propeller is designed subject to unsteady non-axisymmetric inflow, the effective wake is approximated by GBFLOW, which is a flow field solver based on finite volume method. The presented design method has been used for open/ducted/podded/contrarotating propellers design. The designed propeller performance is rechecked by MPUF3A, which compares well with the results from presented method