Numerical Prediction of Autonomous Underwater Vehicle Hydrodynamic Coefficients Using Large Eddy Simulation and Frequency-Domain Decomposition

• Hydrodynamic coefficients of an AUV are identified using LES-based HPMM simulations combined with frequency-domain decomposition. • The proposed frequency-domain decomposition distinguishes added-mass and damping contributions and provides a physically interpretable model. • LES resolves phase-dependent vortex shedding and wake deformation that are not captured by RANS. • LES improves the prediction of translational added-mass and cross-flow damping, while several other derivatives remain less sensitive to turbulence resolution. This study investigates the unsteady flow dynamics and hydrodynamic responses of an autonomous underwater vehicle (AUV) using large eddy simulation (LES) combined with frequency-domain decomposition. Harmonic motions based on the horizontal planar motion mechanism (HPMM) are imposed to generate controlled oscillatory maneuvers, allowing transient vortex shedding and wake deformation to be directly resolved. In contrast to Reynolds-averaged Navier–Stokes equations (RANS) models that rely on time-averaged turbulence closures, LES captures the instantaneous evolution of large-scale vortical structures responsible for unsteady hydrodynamic loading. Fourier-based spectral decomposition of the force and moment histories allows the hydrodynamic response to be expressed in terms of components associated with acceleration and velocity, commonly interpreted as added-mass and damping effects, respectively. The results indicate that added-mass terms arise from fluid inertia, whereas damping components are governed by phase-dependent vortex shedding and wake interactions. LES provides improved predictions of several key hydrodynamic derivatives, particularly the translational added-mass coefficients and the cross-flow damping derivatives associated with pure sway and pure yaw motions.