Multi-Objective Bayesian Optimization of Bioinspired Flexible Nozzles for Underwater Propulsion

The biomechanics of squid pulsed-jet propulsion reveal principles of energy-efficient aquatic locomotion. Squids achieve rapid acceleration and maneuverability by expelling pulsed jets through flexible funnels that dynamically deform to optimize thrust and power expenditure. Building on these biological principles, we design passively deforming flexible nozzles and evaluate their performance using three-dimensional, strongly coupled fluid–structure interaction simulations. Since the performance of these nozzles is sensitive to their geometry, we implement a multi-objective Bayesian optimization framework to maximize impulse while minimizing power consumption. The axisymmetric nozzle is parameterized with spline-based control points that vary radially and axially within constrained limits, while the inlet remains fixed. The optimizer is initialized using Latin hypercube sampling of the design space and iteratively refines designs toward improved performance. Initial single-objective optimization shows that maximizing impulse increases power demand, revealing a trade-off between thrust and efficiency. To address this, the multi-objective approach seeks Pareto-optimal geometries that balance these competing objectives, advancing flexible bioinspired propulsion design.