A metamodeling with CFD method for hydrodynamic optimisations of deflectors on a multi-wing trawl door

摘要

In the present work, a metamodeling with Computational Fluid Dynamics (CFD) method for the hydrodynamic optimisations of deflectors on the multi-wing trawl door is introduced. Then a comparative analysis of Kriging and Artificial Neural Networks metamodeling methods is carried out, using the effects of relative camber, thickness and the installed angles of deflectors on the hydrodynamics as a case study for optimisations. Based on CFD simulations (verified by the wind tunnel experiment), a strict procedure including metamodeling and Multi-Objective Genetic Algorithm is established. Efficiency and accuracy between the two methods are discussed. Finally, this study studied the flow patterns of the optimal otter board in the higher working efficiency state. The results showed that Kriging performs better than Artificial Neural Networks methods in 2-factor metamodeling. However, in the 3-factor investigation, Artificial Neural Networks has more advantages in the prediction of drag, while Kriging is better in the lift component of hydrodynamics. The relative camber has the considerable impacts on the hydrodynamics of the trawl door, whereas the thickness of deflectors shows the prominent influence on the resistances, followed by lift forces. The effect of installed angles on the hydrodynamic loadings is more remarkable than that of thickness and relative camber. In flow visualisation, the disappearance of the vortex on the deflector suction side indicates the rationality of the newly-proposed procedure used for optimisations.