Session: 08-06-01 non-presentations

Paper Number: 128458

128458 - Optimization of Underwater Glider Hull Shape Using Deep Learning for Reduced Hydrodynamic Resistance 

The world of underwater exploration and operation has witnessed a significant shift with the advent of underwater autonomous systems. Central to the success of these systems is their hydrodynamic efficiency, especially for underwater gliders. The greater their hydrodynamic efficiency, the longer they can operate, and the further they can travel, making them indispensable tools in the vast, uncharted territories of our oceans. Hence, optimizing this efficiency has become a primary focus in the continued development of these devices.

One of the traditional challenges in achieving this optimization is the hydrodynamic resistance encountered by the gliders, particularly those with traditional myring hull forms. This resistance can be likened to the air resistance or drag faced by airborne vehicles but intensified due to the higher density of water. Reducing this resistance is critical not just for efficiency but also for the operational life and energy consumption of the gliders.

The main objective of our research is multifaceted. Firstly, we aim to closely examine the possibility of utilizing deep learning models to address this pressing challenge. Deep learning, with its ability to handle vast amounts of data and discern patterns beyond human capability, presents a promising avenue for innovative solutions. Our ambition is to harness this potential to predict the hydrodynamic resistance based on various design parameters. Understanding these parameters and their influence on resistance is crucial for designers to make informed decisions during the glider's design phase.

To achieve our objective, we employed a synthesized dataset. This dataset was meticulously curated, drawing from various sources and simulations to encompass a wide range of scenarios and parameters. Using this dataset, we conducted comprehensive training on a neural network. The goal of this training was twofold: to generate accurate predictions of resistance and to gain insights that could inform design changes.

Ultimately, the ability to predict and reduce hydrodynamic resistance using deep learning could revolutionize the design process for underwater gliders. It could pave the way for more efficient, durable, and capable autonomous systems, ready to tackle the challenges of the underwater realm.

Presenting Author: Mukesh Guggilla Indian Institute of Technology Madras

Presenting Author Biography: Mukesh Guggilla is currently a PhD Scholar at the Department of Ocean Engineering, IIT Madras. He is a graduate of Naval Architecture and Ocean Engineering from IIT Madras and the recipient of the prestigious Prime Minister’s Research Fellowship for his doctoral studies. His research interests include Ship Design and Hydrodynamics, Biomimetic Propulsion systems and Machine Learning Design Applications for Surface and underwater vessels.

Authors:

Mukesh Guggilla Indian Institute of Technology Madras
Vijayakumar R Indian Institute of Technology Madras