Coupled partitioned fluid-motion solver in modelling the response of marine structures

• Main Author: Chow, Jeng Hei
• Other Authors: School of Mechanical and Aerospace Engineering
• Format: Theses and Dissertations
• Language: English
• Published: 2017
• Subjects: DRNTU::Engineering::Mathematics and analysis::Simulations
• Online Access: http://hdl.handle.net/10356/71810
• Institution: Nanyang Technological University

This thesis presents an improvement to traditional methods in resolving the six degree of freedom rigid body motion of floating structures under realistic waves. The popular approach to the problem involves coupling depth-integrated Boussinesq waves equations with the panel method for the solution. The method compromises on the details close to the structure and phenomenon such as water-on-deck and slamming are out of the scope for those solvers. In the current study, a strongly coupled partitioned six degree of freedom rigid body motion solver was developed by coupling the Navier-Stokes Volume of Fluid solver with a motion solver within its corrector loops. The implicit method to solve the rigid body motion is based on Adams-Bashforth-Moulten predictor-corrector steps, an improvement over the Euler and the Leapfrog schemes for fast convergence. To ensure a balance between stability and fast convergence, Aitkens dynamic under-relaxation has been implemented on the accelerations. Results showed great stability and around 70% – 80% reduction in simulation times in a direct comparison. The final version of the solver can be set to run the fluid-motion iterative loops until convergence has been achieved for the pressure and velocities of the fluid and accelerations of the motion. The solver was eventually validated with an experimental design floating wind turbine setup on a tension leg platform. After tuning of the unknown variables, decay tests performed on the system with the coupled solver resulted in accurate estimations of the natural frequencies and damping ratios. Together with a modified restrain system to model the tendons, the regular and focused waves simulations of the floating system were found to be well-predicted. The current study also aims to match the functionality of the coupled Boussinesq-panel method, thus an improved wave generation boundary condition based on previous work was implemented in OpenFOAM® to receive wave data from a pre-ran Boussinesq waves solver. The additional cubic interpolation of wave data onto the numerical wave generation paddles helped preserve the accuracy of the transfer, especially for directional waves. The transfer was validated with both a uni-directional wave flume and also diffraction cases.