Neural Operators for Immersed-Boundary Soft Swimmers Locomotion

2026-08-07

AI Summary

This work addresses the high computational cost of high-fidelity immersed boundary simulations for fluid–structure interaction in soft-bodied swimmers, which hinders repeated evaluations required for design optimization and control. To overcome this limitation, the authors propose a neural operator-based surrogate model trained on regular-grid data generated from adaptive fluid–structure interaction simulations. Conditioned on swimmer geometry and Reynolds number, the model enables temporally resolved prediction of unsteady flow fields around two- and three-dimensional eel-like swimmers. This study presents the first neural surrogate capable of full-field, multi-physics prediction—including velocity, vorticity, and pressure—for moving-boundary swimmers, demonstrating strong generalization across geometries and parameters. The two-dimensional model achieves a mean relative L² error of 3.51% on five extrapolated high-Reynolds-number trajectories, while the three-dimensional model yields errors of 3.44% (velocity), 5.58% (vorticity), and 19.2% (pressure) on interpolated trajectories.

Abstract

High-fidelity immersed-boundary simulation resolves the coupled motion of a deforming swimmer and its surrounding flow, but the resulting cost limits repeated evaluations for engineering design, parameter studies, and control. We develop neural-operator surrogates for temporal prediction of the hydrodynamic fields generated by planar and volumetric eel swimmers. The surrogates are trained on regular-grid fields exported from adaptive fluid--structure simulations and are conditioned on swimmer geometry and Reynolds number. The planar model jointly predicts two velocity components, scalar vorticity, and pressure. On five held-out high-Reynolds-number trajectories, its full-domain global relative L^2 error is 3.51 %. The volumetric formulation uses three target-specific models with a common multichannel input: one model predicts three-dimensional velocity, one predicts vorticity, and one predicts pressure. Their full-domain global relative L^2 errors on five held-out within-range trajectories are 3.44 %, 5.58 %, and 19.2 %. Together, the results demonstrate the feasibility of field-resolved neural surrogates for moving-boundary swimmer flows while identifying pressure accuracy and physical consistency as priorities for further development.

Problem

• immersed-boundary
• soft swimmers
• hydrodynamic fields
• computational cost
• fluid-structure interaction

Innovation

• neural operators
• immersed-boundary method
• soft swimmer locomotion
• flow field surrogate modeling
• Reynolds-number-conditioned prediction