This study investigates the application of generative artificial intelligence visual language models for object detection and obstacle recognition in underwater remotely operated vehicles (ROVs). By combining open-source underwater image datasets with images collected by ROVs, we systematically compare the performance of multiple advanced visual language models. The experimental design encompasses three typical underwater scenarios, aquaculture, marine exploration, and environmental monitoring, to evaluate the models' adaptability under varying underwater environmental conditions. We employ four key indicators for quantitative evaluation: accuracy, which reflects the model's ability to minimize false positives; recall, which measures the completeness of its detection of true targets; F1-score, which comprehensively balances the two; and average precision, which assesses the model's positioning accuracy under an overlap threshold of 50%. The results indicate that model performance is significantly influenced by environmental complexity. For instance, in turbid waters, the recall rate of all models decreases by approximately 15%, underscoring the unique challenges presented by underwater scenes. Additionally, we found that the models' ability to recognize small targets is generally inadequate, necessitating further optimization of the feature extraction architecture or the introduction of domain adaptation training in future work.

Introduction - Climate change is intensifying extreme ocean phenomena, such as increased wave height, posing significant risks to marine infrastructure. This study aims to improve the prediction of maximum wave height (Hmax) and its ratio to significant wave height (Hmax/Hs), using real buoy data from Bilbao-Vizcaya, Cabo de Peñas, Estaca de Bares, and Villano-Sisargas.

Methods -Three predictive models were applied: Linear Regression (LM), Support Vector Regression (SVR), and Random Forest (RF). The study was divided into two phases. In the first, data reduction techniques were analyzed, including instance reduction (through ordered removal of rows) and variable reduction (by eliminating features with low correlation to the target variable). In the second phase, validation techniques were evaluated, specifically Walk-Forward and Rolling Window, testing different window sizes and training set compositions (with observed or predicted values).

Results - Results showed that reducing both the number of instances and variables is feasible without significantly impacting performance metrics (MSE, MAE, RMSE, R²). LM and SVR yielded the best results. While both validation strategies performed similarly, Rolling Window proved faster and more effective with larger windows. However, incorporating predicted values into the training set notably degraded model performance.

Conclusion - This work demonstrates the feasibility of deploying predictive models in real-world settings, enabling early warnings of potentially destructive wave events using real-time buoy data, thereby improving planning and safety in ocean engineering.

In this study, conducted at the Laboratório de Sistemas e Tecnologia Subaquática (LSTS), we tackle the limitations of traditional hydrodynamic modelling approaches by introducing the CFD–HEKF Integrated Modelling and Estimation (CHIME) methodology. The framework derives all hydrodynamic coefficients solely from simulation data, eliminating the need for experimental trials. The approach combines high-fidelity Computational Fluid Dynamics (CFD) simulations with a nonlinear Hybrid Extended Kalman Filter (HEKF) estimator. First, six-Degree-of-Freedom (6 DoF) steady-state CFD simulations of the ISURUS AUV were performed using ANSYS Fluent to extract drag, lift, and fin derivatives. Added-mass coefficients were then calculated through transient simulations using dynamic mesh under free oscillation in three modes. Subsequently, diving and turning manoeuvres were simulated, and the resulting states were input into a MATLAB-based HEKF estimator to estimate unknown damping and added-mass terms. A mesh sensitivity analysis determined that a medium grid (~500,000 cells) provided the optimal trade-off between accuracy and computational cost. Turbulence modelling confirmed that the k–kl–ω model effectively captured laminar-to-transitional regimes. The CFD-derived hydrodynamic coefficients and HEKF-estimated results were benchmarked against analytical and experimental results from the literature. The results reveal that CFD-derived drag, lift, and fin coefficients are within 2% error of experimental values, while added-mass coefficients showed significant improvement over analytical methods. The manoeuvre results, including circular path, tactical diameter, yaw, pitch, and depth, matched field trials within 1-3% error. In contrast, MATLAB simulations using analytical coefficients consistently performed worse. The HEKF-based results demonstrated alignment of states and trajectory within 3% error of CFD results, with a maximum of 10% error of experimental results. The CHIME methodology offers a simulation-only, high-accuracy alternative for full hydrodynamic characterisation of axisymmetric AUVs. The approach provides a streamlined, cost-effective foundation for improving onboard state estimation and integrating it into real-time navigation and control systems.

Ships are highly advanced marine structures that incorporate state-of-the-art technologies. Nevertheless, they still depend on outdated systems in certain critical areas, such as escape lighting. Escape lighting systems are vital components of shipboard safety infrastructure. However, conventional systems rely heavily on decentralized battery-powered luminaires and manual testing, leading to high maintenance costs and environmental burdens. This study addresses these challenges through an engineering-driven redesign of escape lighting systems. A novel system architecture was developed, integrating photovoltaic energy sources with centralized battery storage and Automatic Testing Systems (ATSs) compliant with the IEC 62034 standard. The system interfaces with both main and emergency power networks, reducing reliance on fossil fuels and minimizing battery usage. Engineering simulations and operational data indicate a 20% reduction in fuel oil consumption per escape light and a threefold decrease in maintenance costs over a vessel’s lifecycle. For a standard vessel equipped with 350 luminaires, the system demonstrates significant operational efficiency and environmental benefits, including reduced emissions and hazardous waste. This work exemplifies how ocean engineering innovations can enhance vessel safety while promoting sustainability. The integration of renewable energy and automated diagnostics into critical shipboard systems represents a forward-looking approach to marine engineering, aligning with global goals for greener maritime operations. Moreover, the proposed system supports compliance with evolving maritime regulations and offers a scalable solution adaptable to various vessel types and operational profiles.

Offshore twin pipelines are increasingly deployed for oil and gas transportation, where wave-induced local scour poses significant threats to structural stability. When waves propagate over submarine pipelines, complex interactions between wave dynamics, seabed response, and sediment transport occur. The seabed generates excess pore pressures and seepage flows that modify sediment stability conditions under wave loading. Previous investigations have primarily focused on single pipeline scour under pure hydrodynamic conditions (Chiew, 1991; Fuhrman et al., 2014). Recent studies examined twin pipeline configurations, with Zhao et al. (2015) and Hu et al. (2019) investigating scour under current conditions, while Li et al. (2020) explored combined wave--current effects. However, these works neglected seabed response mechanisms. Limited research by Guo et al. (2019) demonstrated that upward seepage significantly influences sediment incipient motion around single structures, but comprehensive analysis of wave-induced seabed response effects on twin pipeline scour remains unexplored.

This study applies the PORO-FSSI-SCOUR model (Zhai and Jeng, 2024) to investigate seepage effects on scour around twin tandem pipelines. The computational framework couples hydrodynamic simulation with poro-elastic seabed analysis and incorporates seepage-modified Shields parameters for sediment transport. Parametric studies examine pipeline spacing ratios (G/D = 1-5) and soil properties, including permeability (ks) and the degree of saturation (Sr).

Results indicate that wave-induced seabed seepage substantially enhances scour development compared to conventional models without soil response. The seepage correction factor consistently exceeds unity, and upstream pipelines experience greater scour than downstream structures. Scour interactions intensify as G/D increases from 1 to 4, but twin pipelines exhibit independent behaviour at G/D = 5. ks influences scour magnitude more than Sr. Seabed response mechanisms significantly modify sediment transport around twin pipelines, emphasising the importance of considering soil--structure interactions in offshore pipeline design and stability assessment.

Offshore oil and gas production poses unique problems due to complex reservoir behaviour, limited accessibility, and the high expenditures of subsea infrastructure. Optimizing the performance of offshore production wells is thus critical to ensuring maximum hydrocarbon recovery, economic viability, and long-term field sustainability. Alpha, Beta, Gamma, Delta, and Epsilon are the five anticipated oil-producing wells that are expected to provide a 20-year production span from the Horizon Oilfield, which is situated offshore in south-west Ghana. The purpose of this study is to use PIPESIM to maximize these wells' performance in both natural flow and artificial lift scenarios. Sensitivity experiments on tube size, gas–liquid ratio (GLR), water cut, and reservoir pressure depletion were carried out in addition to Inflow Performance Relationship (IPR) and Vertical Lift Performance (VLP) evaluations. The findings show that by reducing frictional losses, a 7-inch tubing diameter optimizes flow-rate. However, because of hydro-static and back pressure accumulation, increases in GLR and water cut have a negative impact on well performance. The pressures associated with reservoir abandonment varied from 500 to 1000 psia. Electric Submersible Pumps (ESPs) were designed and selected for all wells to sustain production in the late phases of life, resulting in considerable improvements in drawdown and production rates even at 100% water cut. The findings provide strategic insights for optimizing well productivity and recovery in offshore reservoirs with long-term production.

Bucket foundations possess the characteristics of excellent bearing capacity, relatively low economic cost and being suitable for soft clay foundations. These advantages have led to their gradual adoption in various coastal and offshore engineering structures. Recently, bucket foundations with internal compartments have emerged as a preferred choice for the foundations of structures that need to bear large moment loads, such as offshore wind turbines. The inclusion of inner compartments in bucket foundations significantly influences soil flow, soil softening and penetration resistance. In addition, the soil heave and softening generated during their penetration can adversely affect its subsequent bearing capacity, often resulting in a lower bearing capacity than that of the fully embedded foundations. Therefore, it is crucial to investigate the installation process and the bearing capacity of bucket foundations with inner compartments. Based on a series of numerical analyses using the Coupled Eulerian-Lagrangian (CEL) method, this study examined the penetration and installation process and the post-installation bearing capacity of bucket foundations with cruciform inner compartments. Parameters including soil strength heterogeneity, soil sensitivity, relative ductility, soil stiffness coefficient, and the foundation wall thickness ratio have been evaluated. Detailed analyses were conducted on soil flow, soil heave, soil softening and penetration resistance. Furthermore, a formula was proposed to calculate the penetration resistance of bucket foundations with cruciform inner compartments, incorporating the soil strain softening and strain rate effects.

This study presents a hydrodynamic shape optimization of the DARPA Suboff submarine hull using a discrete adjoint solver coupled with mesh morphing techniques within the computational fluid dynamics (CFD). The objective is to minimize total resistance (drag) under steady, uniform flow conditions. A baseline hull form is first analyzed using a Reynolds-Averaged Navier–Stokes (RANS) solver to establish reference resistance values. The adjoint solver is then applied to calculate sensitivity fields, identifying regions where geometric modifications yield the most significant drag reductions. Mesh morphing is employed at the mesh level to iteratively update the geometry based on these sensitivities without requiring re-meshing. The optimization process incorporates a Free Form Deformation (FFD) approach to ensure smooth and continuous shape changes. Validation against experimental data demonstrates accurate predictions of resistance for the baseline hull. Through five optimization iterations, the total drag force is reduced by 6.43%. Analysis of the optimized geometry reveals that the most effective shape modifications occur near the aft section of the hull, reducing pressure gradients and improving flow separation characteristics. The results highlight the potential of adjoint-based methods integrated with mesh morphing for fluid-exposed geometry optimization of complex underwater vehicles. This methodology provides a robust, computationally efficient framework for submarine hull optimization and can be extended to other marine vehicles and hydrodynamic objectives in future applications.

Over 300,000 cetaceans die every year due to human activity—primarily ship collisions and fishing entanglements—with roughly 67% of all whale deaths attributed to human interactions. Many whale and dolphin populations are in decline, facing the threat of near‑extinction. The North Atlantic Right Whale, specifically, is critically endangered, numbering only ~350. To combat this issue, a real-time Right Whale detection network is proposed in this paper, to be composed of 3D-printed buoys deployed across the Atlantic Ocean. The goal is to prevent human-caused Right Whale deaths through monitoring whale locations and guiding ships and fishing nets away from collisions through early warnings. The proposed detection network uses a three-step approach where captured audio is recorded on a buoy and then sent to a cloud-based server that processes and classifies whether auditory cues of whales are detected. Finally, client applications are updated using an online API, conveying real-time locations of whales. Each buoy consists of an ESP32, solar panels, and a hydrophone, costing significantly less than current buoys. The system runs a two-branch Ensemble Learning model that employs a 2D Convolutional Neural Network (CNN) with a Custom Convolutional Block Attention Module (CBAM) in one branch, while the other branch utilizes a Bidirectional Long Short-Term Memory (Bi-LSTM) model. Both branches use extracted spectral, temporal, and harmonic features to detect Right Whales based on their vocalizations. This model was trained on datasets from Cornell University and Watkins Marine Mammal Database, achieving a 0.977 AUROC score with a standard deviation of 0.002, surpassing the 0.7214 benchmark set by Cornell while using fewer parameters at 343,877 compared to pre-trained models. In testing, the design demonstrated reliability, potential scalability, and accuracy. Furthermore, promising results were found when extrapolating to the classification of multiple cetacean species, significantly enhancing marine conservation.

Wave motion analysis plays a crucial role in ensuring accurate and reliable fatigue life predictions for deepwater risers. This study investigates how different modeling approaches and key parameters affect wave-induced motion fatigue damage, using representative metocean conditions typically found in ultra-deepwater risers connected to floating units. Both irregular and regular wave methodologies are considered through time-domain dynamic analysis.

Irregular sea states are modeled using the JONSWAP spectrum, while regular waves follow Airy wave theory. This study also explores the influence of specific wave parameters on fatigue assessments, including wave seed numbers for stochastic realizations, individual wave components, and critical environmental and operational factors that affect wave–riser interactions. A range of riser configurations are analyzed to capture variations in hydrodynamic input and their impact on fatigue loading.

Different strategies are applied to identify critical cases. While irregular wave analysis requires longer simulation times, regular wave methods offer faster alternatives. This study compares their accuracy and relevance for different scenarios. Results show that the choice of parameters and approach significantly affects fatigue predictions, with measurable correlations observed under complex ocean conditions.

Based on these findings, this study offers practical recommendations for selecting appropriate analysis methods depending on riser configuration and wave characteristics. These insights aim to support offshore engineers in improving the reliability of riser design and enhancing the resilience of deepwater systems.