With the continuous development of offshore floating wind turbines, their positioning ability has gradually attracted the attention of academia and engineering. The freak wave is destructive to the floating structure because of its large energy, strong suddenness and prominent nonlinear characteristics. In this paper, the new Moray-Base floating wind turbine concept designed by Maridea Company is taken as the research object, and the finite element model with a mooring system is established based on the potential flow theory. In order to improve correct the accuracy of the numerical model, the physical model test of the pool was carried out, and five sets of mooring schemes were designed to study the positioning ability of the numerical model improved by the physical model test. This paper focuses on the motion response and mooring line tension characteristics of five mooring schemes under the action of freak waves, which provides a reference for the design of floating wind turbine mooring system.

In order to predict the second-order force and moment of ships sailing in waves, the fluid domain is decomposed into two parts (where the steady flow effects are considered in the internal one), and a multi-domain coupled hydrodynamic model is established based on the continuity of the flow field. Then a higher-order coupled method is developed, in which nine-node quadratic elements are employed to discretize the boundary, Rankine source is adopted to deal with the complex boundary condition near the ship and higher-order translating-pulsating panel source Green’s function (i.e., analytical integral of the Green’s function over the higher-order panel) is derived over the control surface. On this basis, formulas for second-order force and moment in oblique waves are established under the weak scattering assumption. A numerical program is developed to calculate and analyze the second-order wave forces in surge/sway and yaw moment of various ships advancing in different wave directions. The results show that the present method is more stable and accurate than the traditional higher-order translating-pulsating source method based on Gaussian quadrature formula, as well as the lower-order constant panel method. It is generally applicable to different ship types sailing in both head and oblique waves.

In turbulent channel flows, traditional Detached Eddy Simulation (DES) methods suffer from significant deficiencies in capturing detailed turbulent structures due to the improper allocation of modeling scales. DDES models fail to accurately predict the streamwise mean velocity and are prone to the problem of logarithmic layer mismatch. Although IDDES provides some improvements, it still shows considerable discrepancies in the prediction of second-order statistical quantities. As an improved hybrid approach, the Wall-Modelled Hybrid RANS-LES (WM-HRL) framework addresses this limitation by employing a predefined partitioning strategy to rationally allocate modeling scales between RANS and LES regions. This study systematically investigates the performance of the WM-HRL model in turbulent channel flow and compares its results with those obtained from DDES, IDDES, and direct numerical simulation (DNS) data. The findings demonstrate that the WM-HRL model can more accurately predict near-wall turbulent structures and Reynolds stress distributions, effectively overcoming the limitations of traditional DES methods in the near-wall region.

Buoyancy elements are primarily used to provide additional buoyancy to improve the tension distribution along the risers. Risers covered with buoyancy elements exhibit more complex fluid–structure interaction phenomena under ocean currents, and the current understanding of their hydrodynamic characteristics remains limited. This study conducted stationary towing tests on a rigid riser model covered with buoyancy elements. Based on the hydrodynamic unit formed by the riser-buoyancy system, the drag and lift characteristics in uniform flow were investigated for different spacing arrangements and diameters of the buoyancy elements. The time-frequency characteristics of the cross-flow lift were further analyzed using wavelet transforms. The main conclusions are as follows: compared with a bare riser, the drag coefficient of the buoyancy element decreases to approximately 0.9 due to the low aspect ratio; when the diameter ratio of the buoyancy element to the riser reaches 3.52 and both components operate at Reynolds numbers above 10⁵, the RMS lift coefficient of the hydrodynamic unit increases significantly, reaching up to 1.2 when the length ratio equals 1; the dominant frequency of lift on a riser covered with buoyancy elements exhibits time-varying jumps, with a wider fluctuation range than the vortex shedding frequencies of both the riser and the buoyancy element.

In this paper, the fluctuating pressure on the wall of turbulent boundary layer is studied. Based on the governing equations of the fluctuating pressure in the boundary layer, the unified dimensionless expression of the fluctuating pressure in the full wavenumber region is derived in detail, and the influence of Mach number on the self-spectrum and wavenumber-frequency spectrum of the fluctuating pressure is emphatically analyzed. The fluctuating pressure data of cylindrical boundary layer at different Mach numbers (0 < Ma < 0.176) were obtained by acoustic wind tunnel and water tank experiments. Based on the analysis of experimental data, the accuracy of the predicted Mach-number effect on fluctuating pressure is finally verified.

During maritime operations, ships are subject to wave-induced disturbances, resulting in complex multi-degree-of-freedom motions that severely affect the stability of ships during offshore operations. To address this challenge, this study proposes a CNN-BiLSTM hybrid prediction model for real-time ship attitude forecasting and designs a dual-layer Stewart platform compensation mechanism to mitigate ship motion effects. To overcome sensor latency and noise interference, 2011 ship attitude samples collected in the South China Sea were first filtered using a Savitzky-Golay (SG) filter. The processed data were then fed into the CNN-BiLSTM hybrid prediction model, which significantly improved the accuracy of attitude prediction. Furthermore, a platform kinematic model was established using the closed-loop vector method. The predicted data were utilized to compensate the dual-layer Stewart platform, enabling coordinated control between the lower platform (simulating wave disturbances) and the upper platform (performing dynamic compensation). Experimental results demonstrate that the proposed approach achieves effective ship attitude compensation and provides a feasible solution for stable landing operations of unmanned aerial vehicles (UAVs) at sea.

Energy-saving ducts ahead of the propeller are widely applied to merchant ships as highly efficient energy-saving devices. Taking a certain bulk carrier as the research object, this paper adopts CFD technology to establish a numerical analysis method for the hydrodynamic loads of the energy-saving duct ahead of the propeller under ship steering conditions, and verifies the accuracy of the proposed method through convergence analysis and comparison with relevant experimental data. Based on this method, the influence laws of key factors including displacement volume, ship speed, rudder angle and rudder speed on the hydrodynamic loads of the energy-saving duct ahead of the propeller are investigated. The results show that the larger the ship's displacement volume, the greater the hydrodynamic loads acting on the energy-saving duct ahead of the propeller in all directions; the increases in ship speed and rudder angle both lead to an increase in the hydrodynamic loads on the duct in all directions; rudder speed has almost no effect on the hydrodynamic loads of the duct.

Real-time inversion of wave parameters in actual oceans using ship motion sensor data holds significant value for ship performance research. Compared to direct measurement methods, this indirect inversion approach offers advantages in real-time capability, cost efficiency, and ease of installation. Based on potential flow theory, this paper derives the wave-induced hull motion transfer function (response amplitude operator, RAO) using the principle of linear superposition. Combined with Kalman filtering (KF) technology, a time-frequency inversion model for wave parameters is proposed. The objective is to inversely determine the wave height time history and directional spectrum of random multi-directional waves based on the ship's multi-degree-of-freedom motion. Testing and validation were conducted using simulated multi-degree-of-freedom ship motion signals with noise as input. Results demonstrate that the KF model can effectively inverts directional spectra and wave height time histories for each principal wave direction and sea state level, with a maximum significant wave height inversion error of 10.08%. This indicates the KF model's potential for inverting wave directional spectra in real ocean environments.

The dynamic mechanical behavior of ice at high strain rates is crucial for polar engineering safety and ice–structure interaction studies. Based on an improved specimen preparation method and a Split Hopkinson Pressure Bar (SHPB) system, this study conducted high strain rate uniaxial compression tests on pure water ice at −20 ℃ in the range of 1054−1972 s⁻¹. The results show that pure water ice exhibits clear strain rate sensitivity: peak stress increases from 13.2 MPa to 25.1 MPa (≈ 90% increase), while peak strain shows no monotonic trend but remains scattered. Stress-strain curves generally display a sharp-peak profile, followed by rapid decay and the formation of a residual load-bearing platform of 1−7 MPa at strains around 0.05−0.06, with higher strain rates showing a "delayed instability" characterized by slower post-peak decay and higher residual strength. High-speed imaging further reveals a staged failure process of crack initiation, macro-cracking, and granular fragmentation, corresponding to the residual platform. Even under strict specimen screening and temperature control, peak stress still exhibits scatter of about 15%−20% at similar strain rates, suggesting that strain rate alone is insufficient to explain the mechanical response and that future constitutive models should introduce microstructural or statistical parameters. Cross-comparison with existing literature indicates that the present results fall within the established response range but demonstrate higher consistency and reliability due to stricter control of specimen thickness, temperature uniformity, and loading conditions. Overall, this study supplements the high strain rate database of pure water ice at −20 ℃, clarifies its strain rate sensitivity, residual load-bearing mechanism, and staged failure characteristics, and provides experimental evidence for dynamic response analysis and constitutive model development in cold-region engineering.

Compared with the rigid support scheme of traditional ship shafting systems, the bearing load in an elastically supported shafting system is more easily affected by external disturbances. To address the difficulty in accurately obtaining the bearing load under different operating conditions after the ship is launched, this paper proposes a method for predicting the bearing load of elastically supported shafting systems based on transfer learning. This method focuses on the influence of vertical disturbance forces caused by pressure changes in the air spring isolators on the shafting system. It collects strain data from characteristic regions and air spring pressure data during the installation and commissioning phase of the shafting system. These data, together with finite element simulation data, are used to build a bearing load prediction model with a GA-BP neural network algorithm. Validation with measured data shows that this method has good generalization ability and prediction accuracy for untrained operating conditions, with an average error of 8.8%.

Aiming at the difficuties of fatigue evaluation of slamming stress response for high-speed ships, this study analyzes the stress response and fatigue damage based on the measured data of the bow of a small high-speed aluminum alloy ship, by combining with fatigue tests on two typical longitudinal bulkhead-penetrating structures, with the linear cumulative damage theory. The research confirms that the stress response under slamming pressure exhibits obvious impact characteristics, and the stress amplitude no longer follows the Rayleigh distribution. The results show that fatigue cracks may occur at the measuring points after 212 hours of continuous navigation under the test conditions. This achievement deepens the understanding of the dynamic characteristics of slamming stress and provides a reference for the anti-fatigue design of longitudinal bulkhead-penetrating structures of high-speed aluminum alloy ships.

With the widespread application of composite materials in aerospace, shipbuilding, and automotive fields, optimizing the performance of their connecting structures has become a key challenge. This article focuses on three types of composite connections: adhesive bonding, bolted connection, and adhesive-bolted hybrid. The influence of parameters such as adhesive layer thickness, bolt preload, and porosity on the mechanical properties of joints was systematically analyzed through a combination of experiments and numerical simulations. The key parameters are optimized based on Response Surface Methodology (RSM). The research results indicate that increasing the thickness of the adhesive layer leads to intensified deformation but has no significant monotonic effect on the ultimate bearing capacity. The optimal balance between stiffness and strength is achieved when the pre-tightening force of the bolt is 3.0 N·m, while the ultimate load close to the low porosity structure can still be achieved through deformation compensation mechanism when the porosity is 100%. The adhesive-bolted hybrid joint combines the stress dispersion of adhesive bonding with the anti-stripping characteristics of bolts, and its ultimate load is increased by 17% to 40% compared with a single connection form. The quadratic polynomial model established based on the improved response surface method identified the optimal parameter combination was adhesive thickness of 0.58 mm, pre-tension force of 3.1 N·m, and porosity of 93.0%.

Polymethyl Methacrylate (PMMA) has become a key material for pressure-resistant shells of full-view manned submersibles due to its high transparency, light weight and excellent pressure resistance. However, the viscoelastic creep behavior of PMMA material itself has a significant impact on the long-term reliability of the structure. Therefore, the creep buckling mechanism of PMMA pressure spherical shells and their remaining ultimate load carrying capacity were investigated based on the viscoelastic properties of PMMA. Firstly, the non-linear influence law of stress level and time on the creep behavior of PMMA was revealed through creep tensile tests. Based on short-term test data, creep curves applicable to different stress levels were obtained by applying the time-stress equivalence principle while the applicability of the Burgers model and the modified Burgers model was discussed and analyzed. Subsequently, based on the modified Burgers model and numerical simulations, the influence of key factors, like initial geometric defects, the thickness-to-diameter ratio and working depth on the residual ultimate load of spherical shells after different creep durations was analyzed. The research results can provide certain theoretical basis and data support for the design, safety assessment and life prediction of full-view PMMA pressure-resistant structures.

To improve the efficiency and accuracy of ultimate-bending-moment estimation for complex non-symmetric hull sections, a hybrid method integrating genetic-enhanced Bayesian optimization with an improved Smith stepwise failure procedure was proposed. Firstly, for a given section curvature and increment, Latin hypercube sampling was used to generate an initial design and train a Gaussian process (Kriging) surrogate. Based on the surrogate’s posterior mean and variance, an Expected Improvement (EI) acquisition function was constructed, and a genetic algorithm was then employed to maximize EI and select candidate sampling points, thereby enhancing global search capability while keeping the number of high-fidelity evaluations small. Selected candidates were subsequently refined by a constrained derivative-free local optimizer and their high-fidelity evaluations were incorporated into the training set to iteratively update the surrogate, ensuring accurate local search in regions near global optima. With the Dow 1/3-scale frigate model adopted as a test case, the ultimate bending moments under intact and damaged conditions were computed using the traditional Smith method, a PSO-Smith variant and the proposed approach. Results show that the proposed method could precisely resolve the ultimate bending moments for non-symmetric sections, and that, compared with the PSO-Smith approach, it improves computational accuracy by approximately 2% and increases computational efficiency by about 90%, validating its effectiveness and strong potential for engineering applications.

Accurate prediction of radiated sound fields from underwater structures with limited measurement points is crucial for vibration and noise reduction studies. This paper presents an investigation on a rectangular plate and the proposal of a measurement point arrangement strategy based on Gaussian integration point distribution. The surface vibration velocity field of the structure was reconstructed using an L₁ norm sparse solution, and the acoustic radiation sound field was predicted through vibro-acoustic transfer functions. Under the constraint of limited measurement points, the effects of truncated modal order, number of measurement points, and their arrangement on the prediction accuracy of the radiated sound field were analyzed. Results show that, within the mid-to-low frequency range, the Gaussian integration point distribution strategy optimizes sensor placement. Compared with traditional uniformly distributed measurement methods, the proposed strategy could achieve an accurate prediction of the radiated sound field with fewer measurement points, providing theoretical support for engineering applications involving sound field prediction under limited sensor conditions.