Hydrodynamic shape optimization by high fidelity CFD solver and Gaussian process based response surface method

A computational framework for hydrodynamic shape optimization of complex ship hull form is proposed and applied to improve the calm water performance of the KRISO Container Ship (KCS). The framework relies on three key features: a novel shape morphing method based on a combination of subdivision surfaces and free form deformations, a robust three dimensional viscous computational fluid dynamic solver based on the openFOAM open-source libraries and a Gaussian process-response surface method (GP-RSM) based on ordinary Kriging model which has been created to speed-up the evaluation of the quantity of interest (QoI) of the design process. The accuracy of the hydrodynamic solver is proven by comparing the obtained results against available experimental measurements. A preliminary sensitivity analysis on the mesh size has been carried out aiming at reducing the computational burden required by the CFD predictions. Three GP-RSMs have been trained relying on increasing number of hull designs. Each surrogate model has been cross-validated by both leave-one-out and k-fold techniques. The behaviours of these multi-dimensional surfaces have been analyzed in details by sampling the investigated design space with 107 points according to a Full-Factorial algorithm, highlighting the regions of maximum deviation with respect to the resistance of the reference hull. The three optimum designs provided by the corresponding GP-RSM models have been verified by using high-fidelity CFD simulations with a refined mesh configuration. Calm water resistance, wave patterns and pressure distributions over the selected hull surfaces have been discussed in the light of the generated shape variations.