Optimizing swimming performance of tapered elastic swimmer using a parameterized genetic algorithm

A key bottleneck in the computational design of elastic swimmers is the computational intensity of the fluid-structure interaction simulations. When trying to optimize a design, it is impractical to use iterative algorithms to search the enormous design space to find an optimal solution in a reasonable time. We develop a more efficient approach to evaluate swimmer designs by using a simplified kinematic swimmer model that allows us to estimate swimmer parameters correlated with swimming performance, without the need to perform three-dimensional fully coupled fluid-structure simulations. Specifically, we optimize the design of an oscillating tapered plate that propels itself forward in viscous fluid. To this end, we vary the tapering shape and oscillation frequency of the plate. The optimization is performed using an evolutionary genetic algorithm using the kinematic swimmer model for about 10,000 generations. The criteria for the optimum swimmer kinematics include the tip displacement and standing wave ratio. These results are verified by comparing them with the full-scale CFD simulations revealing good agreement.