Numerical optimization of a propeller behind a ship hull at full scale

A propeller behind a ship hull (the Hamburg test case) at full scale is optimized to minimize the delivered power at a given ship speed by adjusting the geometry and the revolution speed. An uncertainty analysis and a wake validation are made for the CFD (Computational Fluid Dynamics) code SHIPFLOW. The minimum iteration number and grid density were carefully selected to decrease the computation time needed. The optimization is done with a parallel Adaptive Range Genetic Algorithm in the coarse grid and a local optimization starting from the optimized design is performed to check whether it is a local optimal, at least. Then the verification of the optimal propeller for both objective function and constraint in three finer grids is made. To satisfy the constraints in the finer grids, we make a manual correction by adjusting one design variable. At the same time, an automatic optimization starting from the optimum we obtained in the coarse grid is made in the finest grid with a local optimization method, DHC (Dynamic Hill Climbing). The designs by manual correction and automatic optimization are compared. Furthermore, an off-design propeller behind the same ship is also optimized with the same design variables, constraint and objective function. An optimum close to the one obtained in the near-optimum propeller optimization is found. Results indicate that the present optimization procedure can yield lower delivered power for both a near-optimum and an off-design propeller