This study develops a method to inversely estimate wave heights, periods, and directions from hull motion data through an Artificial Neural Network. Hull motion data, measured for approximately 2.5 years from a Floating Production, Storage, and Offloading vessel currently operating at the Liza field, are collected. Corresponding wave, current, and wind data are also collected. Statistical data, i.e., hourly mean and standard deviation of 6-degree-of-freedom motions along with hourly wind, current, and draft information, are selected as potential inputs for the ANN. A correlation matrix, which shows the correlation between parameters, is used to discover the correct combination of inputs. Then, different input combinations are taken into account to identify a correct combination of inputs with the highest estimation accuracy, which provides the best results when using highly correlated inputs rather than using all variables as inputs. An R-squared of 0.72 is achieved for significant wave height, with the estimated peak period and wave direction resulting in the Root Mean Square Errors of 1.45 s and 7.7 deg, respectively. The results demonstrate the practicability of the developed inverse estimation tool in offshore platform operations.

This study utilizes artificial neural network (ANN) models to inversely estimate the near-real-time 3D ocean vertical current profile (magnitudes and directions) using floater-mounted motion sensors with (or without) two bi-axial inclinometers on the top portion of a riser. A coupled time-domain dynamics model that takes into account the interaction between the FPSO (Floating Production Storage and Offloading platform), mooring lines, risers, and cables is established to provide inputs and outputs for ANN models. The dynamics simulations are performed using the 2-year-long measured environmental (wind, wave, current) conditions. Then, three ANN models with different combinations of input variables are evaluated, and the process of hyperparameter selection is conducted to increase the prediction accuracy compared to the actually measured current profiles. The root-mean-square errors (RMSEs) and R-squared of current speed, the contour plots of measured and estimated current speed and direction, representative motions of the FPSO, and inclination and azimuthal angles of a riser are systematically presented. The results demonstrate that the proposed ANN model adequately estimates the 3D current profile from FPSO motions while the accuracy especially at deeper depth is improved by considering bi-axial inclinometers on the riser as additional inputs.

This paper presents a machine learning method to predict the dynamic and structural behaviors of the submerged floating tunnel (SFT) based on an artificial neural network (ANN) and numerical sensors. The training and testing data are generated by the tunnel-mooring coupled time-domain hydro-elastic simulations under various random wave excitations. Then, ANN is constructed with an input layer, hidden layers, and an output layer. The input layer consists of the numerical angle and acceleration signals while the output is the tunnel's estimated displacements, bending moments, and mooring tensions. The number of hidden layers, neurons in each layer, and epochs for stable performance are selected based on the parametric study. For high prediction accuracy, Rectified Linear Unit (ReLU) and Root Mean Square Propagation (RMSProp) are adopted as activation functions and optimizers. In addition, a cost-effective sensor combination at an optimal location is investigated to achieve good performance while using a minimal number of sensors. The optimized sensor combination with one angle sensor and one accelerometer successfully predicts the dynamic and structural responses based on not only R-squared and root-mean-square errors but also representative time histories, spectra, and prediction accuracy plots.

A hydraulic accumulator is a key component that absorbs the shock and vibrations of a hydraulic system of a marine diesel engine and controls the volume change according to the pressure change of the hydraulic oil. During the operation of a hydraulic accumulator, repeated concentrated stress is applied to the lowermost thread of the lower shell. The purpose of this study is to numerically investigate the structural safety and fatigue life for hydraulic accumulators under extreme pressure conditions. Structural and fatigue analysis were performed using finite element modeling under the operating conditions and test conditions of 30MPa and 40MPa pressure, respectively. As a result of the structural analysis, the maximum von Mises stress did not exceed the tensile strength under the operating conditions, whereas the von Mises stress of 1029 MPa under the test conditions exceeded the tensile strength: thus, the structural stability was not secured. The fatigue life was calculated using fatigue analysis with a design life of 106. Further, fatigue failure occurred within a short duration as it represented 160 cycles.

This study presents the optimization process of a tuned mass damper (TMD) to mitigate the lateral motions and mooring tensions of a submerged floating tunnel (SFT) under seismic excitations. A time-domain hydro-elastic-simulation model is established to solve the coupled dynamics among the tunnel, TMD, and mooring lines. The wet hydro-elastic natural frequencies of the SFT with mooring are estimated. The spring and damping coefficients of TMD are optimized by using metaheuristic optimization algorithms, i.e., harmony search (HS), genetic algorithm (GA), and particle swarm optimization (PSO). HS is coupled with the time-domain dynamic-simulation model to perform the iterative process of updating the coefficients by HS, running the simulation with updated coefficients, returning the results back to HS. The optimized coefficients obtained by HS are also cross-checked by using GA and PSO with the pre-established objective function, which shows consistent results. Subsequently, the effectiveness of TMD with the optimized parameters is tested for a variety of seismic conditions that can cover the most seismic magnitudes. The time histories, spectra, and statistics of SFT dynamic responses and mooring tensions are systematically analyzed and discussed. As intended, the optimized TMD effectively attenuates the resonant hydro-elastic transient motions of the SFT at its lowest lateral natural frequency. The mooring tensions are also significantly reduced by adopting the optimized TMD, especially in large earthquakes.

A floating bridge is an innovative solution for deep-water and long-distance crossing. This paper presents a curved floating bridge's dynamic behaviors under the wind, wave, and current loads. Since the present curved bridge need not have mooring lines, its deep-water application can be more straightforward than conventional straight floating bridges with mooring lines. We solve the coupled interaction among the bridge girders, pontoons, and columns in the time-domain and to consider various load combinations to evaluate each force's contribution to overall dynamic responses. Discrete pontoons are uniformly spaced, and the pontoon's hydrodynamic coefficients and excitation forces are computed in the frequency domain by using the potential-theory-based 3D diffraction/radiation program. In the successive time-domain simulation, the Cummins equation is used for solving the pontoon's dynamics, and the bridge girders and columns are modeled by the beam theory and finite element formulation. Then, all the components are fully coupled to solve the fully-coupled equation of motion. Subsequently, the wet natural frequencies for various bending modes are identified. Then, the time histories and spectra of the girder's dynamic responses are presented and systematically analyzed. The second-order difference-frequency wave force and slowly-varying wind force may significantly affect the girder's lateral responses through resonance if the bridge's lateral bending stiffness is not sufficient. On the other hand, the first-order wave-frequency forces play a crucial role in the vertical responses.

Floating bridges are considered as a solution to the deep-water crossing. In this study, dynamic behaviors of a curved floating bridge, which consists of a girder, columns, and pontoons, are analyzed in the time-domain in waves, winds, and currents. This type does not have mooring lines so that deep-water installation can be much easier than conventional bridges and floating bridges with mooring lines. The key factor is, therefore, to have good global behaviors. In the frequency domain, we computed hydrodynamic added mass, radiation damping, wave excitation forces by using a 3D diffraction/radiation program. Time-domain simulations were further conducted under different loading conditions. Second-order wave-excitation and dynamic wind loads induce significant lateral motions of the girder since a wave-dominant-frequency range is close to its natural frequencies while the first-order wave-excitation force plays an important role in the girder's vertical motion.

This paper presents a machine learning method for detecting the mooring failures of SFT (submerged floating tunnel) based on DNN (deep neural network). The floater-mooring-coupled hydro-elastic time-domain numerical simulations are conducted under various random wave excitations and failure/intact scenarios. Then, the big-data is collected at various locations of numerical motion sensors along the SFT to be used for the present DNN algorithm. In the input layer, tunnel motion-sensor signals and wave conditions are inputted while the output layer provides the probabilities of 21 failure scenarios. In the optimization stage, the numbers of hidden layers, neurons of each layer, and epochs for reliable performance are selected. Several activation functions and optimizers are also tested for the present DNN model, and Sigmoid function and Adamax are respectively adopted to enhance the classification accuracy. Moreover, a systematic sensitivity test with respect to the numbers and arrangements of sensors is performed to find the appropriate sensor combination to achieve target prediction accuracy. The technique of confusion matrix is used to represent the accuracy of the DNN algorithms for various cases, and the classification accuracy as high as 98.1% is obtained with seven sensors. The results of this study demonstrate that the DNN model can effectively monitor the mooring failures of SFTs utilizing real-time sensor signals.

The wave-generation performance of a piston-type wave maker was analyzed using the numerical wave tank technique, and the numerical results were compared with theoretical solutions. A two-dimensional frequency domain analysis was conducted based on the Rankine panel method. Various parameters were used to examine the wave-generation performance, such as the width and gap of the wave board. The effects of the thickness of the wave board and of the gap from the bottom of the tank were evaluated. The difference in the amplitude of the generated wave between the analytical solution and the numerical result was examined, and its causes were addressed due to the gap flow between the bottom of the tank and the wave board. This parametric analysis can be utilized to design an optimum wave make parametric analysis to design an optimum wave maker that can generate waves with amplitudes that can be predicted accurately.