Wing Shape Optimization of Underwater Floats with Motion Constraints

Abstract

This study investigates the effects of attaching wing shapes to the body of an underwater float, comparing its motion with and without wings. The body is limited to vertical motion. However, with wings attached, it can achieve horizontal and rotational movements, such as sawtooth and spiraling motions. These wings are effective in controlling the motion of the float in response to water flow and reducing drag. To design the wing airfoil shapes for the underwater float, we utilized a Conditional Deep Convolutional Generative Adversarial Network(DCGAN) model to achieve inverse design with high lift-to-drag ratios and lift coefficients. This model extends the basic GAN structure by adding conditions to the inputs of the generator and discriminator, enabling the generation of airfoil shapes with desired aerodynamic performance. Prior to planform optimization, aerodynamic coefficients were compared using a baseline shape for five candidate wing attachment positions. A total of 77 shapes were generated by varying the span and chord parameters within a ±23% range of the baseline values. These shapes were used to train a Gaussian Process Regression (GPR) model, which served as a surrogate model during the optimization process. Dynamic analysis was conducted at an underwater float angle of 35.5° to determine the objective function and constraints. Finally, planform optimization was performed using a Genetic Algorithm (GA). The results demonstrate the potential of wings to enhance the motion control and drag reduction of underwater floats, offering significant improvements in hydrodynamic performance. © 2025, American Institute of Aeronautics and Astronautics Inc, AIAA. All rights reserved.