Optimized time increments for stress time series in the substructure of a floating offshore wind turbine using a hybrid frequency–time domain approach

This study proposes an efficient hybrid frequency–time domain approach to reconstruct stress time series in the substructure of a floating offshore wind turbine (FOWT). The proposed framework combines hydrodynamic pressure response amplitude operators (RAOs) obtained from Wamit with concentrated load histories—accelerations, tower-base forces, and fairlead tensions—derived from OpenFAST. Unit stress RAOs were computed using linear static finite element analyses (FEAs) in Abaqus and linearly superposed to generate total stress histories. The hybrid approach was validated against transient dynamic FEAs through a simplified cantilever-beam model, demonstrating excellent agreement except for slight discrepancies attributed to dynamic amplification effects that are inherently captured only in transient dynamic FEAs. The proposed framework was subsequently applied to the VolturnUS-S 15 MW reference FOWT. Hydrodynamic analyses were performed in the frequency domain using Wamit, while fully coupled aero-hydro-servo-elastic simulations were conducted in the time domain using OpenFAST under two design load cases (DLC6.1 for ultimate limit state and DLC1.2 for fatigue limit state). The hybrid method successfully reconstructed stress time series in the substructure by combining the frequency-domain pressure RAOs with the time-domain load histories. A parametric study was conducted to determine an optimal time increment for stress reconstruction. Increasing the time increment for stress reconstruction significantly reduced computational time and storage requirements while moderately influencing stress accuracy. When the time increment for stress reconstruction is 0.1 s, the maximum stress deviation in the ultimate limit state was <5 %, and the cumulative fatigue damage deviation in the fatigue limit state was <2.3 %, compared with reference results obtained at the time increment for stress reconstruction of 0.0125 s. These findings demonstrate that the proposed hybrid approach can maintain sufficient accuracy while reducing computational cost and data volume by over 90 %. Future work should extend this methodology to a broader range of environmental conditions and additional DLCs to validate its general applicability.