Seagoing vessels operate in harsh environments which make them especially prone to progressive degradation mechanisms such as fatigue and corrosion. Acoustic emission (AE) monitoring is gaining interest from ship operators and inspectors for its potential as an early-warning structural health monitoring technique for these types of damage. A major challenge facing the implementation of AE is dealing with the background noise. This article presents an experimental study of ultrasonic noise levels in representative environments and conditions AE monitoring. The probability of detection (PoD) is proposed as a quantitative metric for the detection of damage in the presence of operational noise. Measurements were carried out in multiple locations on board of a vessel under different operational conditions. Measurements at cruising speed on hull plates inside the engine room suggest that the ultrasonic background noise level exceeded 90 dB under 100 kHz but rapidly reduced in the higher frequencies associated with the failure mode-related AE signals. The PoD was estimated to be 94% for damage signals above 100 kHz. These results suggest that acoustic emission monitoring has the potential to perform reliably under noisy conditions. This perspective is promising to the future of a structural health monitoring system based on AE measurement.

Visual search is a fundamental cognitive ability. This study investigates whether Multimodal Large Language Models (MLLMs) exhibit human-like difficulty signatures in visual search tasks. We compared search performance of humans (n = 1,250) and MLLMs using identical 2D and 3D stimuli across different set sizes. Both groups showed efficient performance in feature searches, most clearly when the target had a unique color, but performance degradation in conjunction searches as set sizes increased. Additionally, we found strong correlations between human and MLLM error rates (ρ = 0.82), which suggests that MLLMs are sensitive to similar objective complexities, such as stimulus heterogeneity. However, differences were found as well: whereas humans invested extra search time to respond accurately on target-absent trials, MLLMs exhibited extreme present/absent response biases in complex searches. We conclude that MLLMs replicate high-level human performance signatures, yet their underlying computations differ significantly.

Wave impact loads on maritime structures can cause casualties, damage, pollution of the sea and operational delays. Consequently, their extreme values should be accounted for in the design of these structures. However, this is challenging, as wave impact events are both rare and highly complex, requiring both high-fidelity simulations and long analysis durations to reliably quantify the associated design loads. Moreover, existing extreme value prediction methods are neither specifically developed nor adequately validated for wave impact phenomena. We therefore introduce the new Probabilistic Adaptive Screening (PAS) method for predicting extreme non-linear loads on maritime structures. The method integrates copula-based statistical dependence modelling with multi-fidelity screening and adaptive sampling. This framework enables efficient extreme value prediction by statistically mapping low-fidelity indicator variables to high-fidelity impact loads. The method allows for efficient linear potential flow indicators to be used in the low-fidelity stage, even for strongly non-linear load cases. The statistical framework of the method is validated against four weakly and strongly non-linear test cases, including non-linear waves, ship vertical bending moments, green water impact loads, and slamming loads. It is concluded that PAS with optimal settings accurately estimates both the short-term distributions and extreme values in these test cases, with most probable maximum (MPM) values within 2–15% of the reference brute-force Monte-Carlo Simulation (MCS) results. In addition, PAS achieves this performance very efficiently, requiring in the order of 1–3% of the high-fidelity simulation time needed for conventional MCS. These results demonstrate that PAS can reliably reproduce the statistics of both weakly and strongly non-linear extreme load problems, while significantly reducing the associated computational cost compared to MCS.

Corrosion is a leading damage mechanisms in the degradation of marine assets. Acoustic emission (AE) monitoring has gained increasing interest as a technique for continuous monitoring of corrosion damage. This study numerically and experimentally investigates the feasibility of wall thickness loss estimation from the AE signals due to localized corrosion. The interaction of the elastic waves emitted due to the evolution of corrosion damage are influenced by the local thickness and material properties of the structure. A steel plate of (500 mm x 500 mm x 10 mm) with a localized wall thickness loss between 0 and 80% in the center of the plate was considered. The numerical investigation was conducted using a higher-order finite element model. Laboratory experiments were performed on a carbon steel specimen instrumented with 7 AE transducers (40 - 250 kHz). Corrosion damage was artificially introduced in the steel plate by progressively milling a pit in the center. At different stages of wall thickness loss, simulated AE sources were generated. The response of the structure was evaluated based on signal characteristics such as amplitude, rise-time, frequency content, and waveform. A correlation between the signal amplitudes and the wall thickness loss was observed in both experimental and numerical results. This perspective is promising for the feasibility of corrosion-induced wall thickness loss estimation based on AE measurements.

An important trend exhibited by the offshore wind market is the increasing size of wind turbines, leading to longer and stiffer monopiles with larger diameter-to-thickness ratios. Current transport analysis is focused on loads resulting from hydrodynamic accelerations, without taking into account the loads resulting from differences in bending deflection between the vessel and cargo. This investigation examines the structural response of a monopile and sea-fastening system subjected to displacement-based loads. The load case follows from a vessel excited using a regular wave leading to bending deflections and rigid body accelerations. The intermittent contact between the saddles and monopile is modeled by representing the saddle with a unilateral spring. This requires the use of a nonlinear solution method to obtain structural responses. The harmonic nature of hydrodynamic-based loads led to the selection of the harmonic balance method (HBM) to model the cargo-sea-fastening system. A novel understanding is gained of how cargo properties, sea-fastening properties, and sea-fastening arrangements influence the structural response of the coupled cargo-sea-fastening system. Various parametric studies are performed to identify behaviors related to the total structural response. Based on this study, the conclusion can be drawn that a large number of saddles in combination with a low stiffness is desired to minimize the structural response of the cargo and sea-fastening system. Furthermore, the influence of lashing stiffness and pretension is limited with respect to the total response. Both these conclusions also hold for an increase in cargo length and diameter.

Predicting extreme values of strongly non-linear hydrodynamic responses (such as wave impact loads) is crucial for ensuring the safety and reliability of marine and coastal structures. However, this task is challenging due to the complexity and rarity of these responses. Existing methods are often limited to weakly non-linear responses or are very computationally expensive. This paper presents a new multi-fidelity method called ‘Adaptive Screening’, designed to efficiently predict extreme values of strongly non-linear wave-induced responses. These values are critical inputs for structural design and reliability analysis. Adaptive Screening combines elements of screening, multi-fidelity Gaussian Process Regression, and adaptive sampling. We validate its effectiveness through three applications: predicting the most probable maxima of second-order wave crests, vertical bending moments on a ferry, and green water impact loads on a containership. Our results demonstrate that Adaptive Screening outperforms conventional brute-force methods, achieving comparable accuracy in predicting extreme values while significantly reducing high-fidelity simulation times (especially for the most non-linear cases). Like many alternative methods, Adaptive Screening relies on a response-dependent low-fidelity indicator variable. We also show that the method performs well with realistic indicators for a range of applications. The test cases indicate that Adaptive Screening is very promising for the strongly non-linear responses it was designed for.

This paper aims at estimating both unidirectional and multi-directional waves from noisy measured ship motion data, with a focus on the inclusion of the vessel's forward speed to reflect real-world operating conditions. The technique is based on an Adaptive Kalman Filter for estimating wave elevation and wave spectrum parameters, including significant wave height, peak period, and wave direction. The proposed method was tested using simulated ship motion data, and its performance was evaluated by comparing the estimated wave spectrum with reference values used in the simulation model and with results from a widely used baseline frequency domain approach. The results demonstrate that the method effectively estimates the wave spectrum in a short measuring window with a reasonable degree of accuracy when accounting for varying forward speed, indicating strong potential for real-time wave estimation to aid in improving navigation, safety, and operational efficiency.

In the assessment of wave-in-deck loads for new and existing maritime structures typically model tests are carried out. To determine the most critical conditions and measure sufficient impact loads, a range of sea states and various seeds (realisations) for each sea state are tested. Based on these measurements, probability distributions can be derived and design loads determined. In air gap model testing usually only few, if any, impact loads occur per 3-hour seed. This can make it challenging to derive reliable probability distributions of the measured loads, especially when only a few seeds are generated. In addition wave impact forces, such as greenwater loading, slamming, or air gap impacts are typically strongly non-linear, resulting in a large variability of the measured loads. This results in the following questions: How many impacts are needed to derive a reliable distribution? How is the repeatability of individual events affecting the overall distribution? To answer these questions wave-in-deck model tests were carried out in 100 x 3-hour realisations of a 10,000 year North Sea sea state. The resulting probability distributions of the undisturbed wave measurements as well as the measured wave-in-deck loads are presented in this paper with focus on deriving the number of seeds and exposure durations required for a reliable estimate of design loads. The presented study is Part 2 of a combined study on guidance for the convergence and variability of wave crests and impact loading extreme values. The data set of Part 1 ([1]) is based on greenwater loads on a sailing ferry and the data set of Part 2 on wave-in-deck loads on a stationary deck box.

Green water and slamming wave impacts can lead to severe damage or operability issues for marine structures. It is therefore essential to consider their probability and loads in design. This is difficult, as impacts are both hydrodynamically complex and relatively rare. The complexity requires high-fidelity modeling (experiments or CFD), whereas a statistically sound analysis of rare events requires long durations. High-fidelity tools are too demanding to run a Monte-Carlo simulation; low-fidelity tools do not include sufficient physical details. The use of extreme value theory and/or multi-fidelity modeling is therefore required. The present paper reviews the state-of-the-art methods to find wave impact design loads, which include response-conditioning methods, screening methods, and adaptive sampling methods. Their benefits and shortcomings are discussed, as well as challenges for the wave impact problem. One challenge is the role of wave non-linearity. Another is the validation of the different methods; it is hard to obtain long-duration high-fidelity wave impact data.

For the design of maritime structures in waves, the extreme values of responses such as motions and wave impact loads are required. Waves and wave-induced responses are stochastic, so such responses should always be related to a probability. This information is not easy to obtain for strongly non-linear responses such as wave impact forces. Usually class rules or direct assessment via experiments or numerical simulations are applied to obtain extreme values for design. This brings up questions related to the convergence of extreme values: how long do we need to test in order to obtain converged statistics for the target duration? Or, vice versa: given testing data, what is the uncertainty of the associated statistics? Often the test or simulation duration is cut up in ‘seeds’ or ‘realisations’, with an exposure duration of one or three hours based on the typical duration of a steady environmental condition at sea, or the time that a ship sails a single course. The required number of seeds for converged results depends on the type of structure and response, the exposure duration, and the desired probability level. The present study provides guidelines for the convergence of most probable maximum (MPM) wave crest heights and MPM green water wave impact forces on a ferry. Long duration experiments were done to gain insight into the required number of seeds, and the effect of fitting. The present paper presents part 1 of this study; part 2 [1] presents similar results for wave-in-deck loads on a stationary deck box.

Catamarans are popular in the offshore sector as they combine good transverse stability and ample deck space with low wave resistance. However, their slender hull shape results in low restoring qualities in heave and pitch motions. The large motions in rough weather can often result in water impacting the underside of the deck connecting the two hulls, a phenomenon called wet deck slamming. The impulse excitation from wet deck slamming can then produce a transient hydroelastic response of the structure called whipping. Whipping excites mode shapes that would not normally be present in the response, as their natural frequencies are significantly higher than the wave encounter frequency. This results in detrimental contributions to fatigue life through high-amplitude cyclical bending moments. Both the calculation of slamming loads and the prediction of resulting structural responses have been a challenge for several decades. The highly nonlinear and three-dimensional character of the phenomenon, combined with the strongly coupled fluid-structure interaction means that it is unpredictable, and even the definition of slamming events has been a matter of disagreement among researchers. Experiments are still a vital part of these investigations, for validating ever-improving numerical techniques. An essential issue with experiments is the extent to which mode shapes and natural frequencies can be emulated in model scale. Traditional hydroelastic models are segmented and use either a flexible backbone or flexible joints to introduce stiffness. This often results in an excellent description of the 2-node bending mode, but an increasing error for higher modes leads to stress inaccuracies. In this investigation, a continuous model of a catamaran is designed and produced for hydroelastic experiments. The advantages and limitations of the concept are identified, the verification against structural models is presented, and the calibration of the measurements is discussed.

Green water and slamming wave impacts can lead to severe damage or operability issues for marine structures. It is therefore essential to consider their probability and loads in design. This is difficult, as impacts are both hydrodynamically complex and relatively rare. The complexity requires high-fidelity modelling (experiments or CFD), whereas a statistically sound analysis of rare events requires long durations. High-fidelity tools are too demanding to run a Monte-Carlo simulation; low-fidelity tools do not include sufficient physical details. The use of extreme value theory and / or multi-fidelity modelling is therefore required. The present paper reviews the state-of-The-Art methods to find wave impact design loads, which include response-conditioning methods, screening methods and adaptive sampling methods. Their benefits and shortcomings are discussed, as well as challenges for the wave impact problem. One challenge is the role of wave non-linearity. Another is the validation of the different methods; it is hard to obtain long-duration high-fidelity wave impact data. A planned case study is introduced, where different techniques will be put to the test and these challenges will be addressed .

Very large waves, such as rogue waves, can be dangerous for ships and offshore structures. There is no consensus on the theoretical occurrence probability of these rogues and ocean measurements containing rogue waves are rare. This paper addresses the long-term occurrence probability of rogue waves using time-extreme (TE) and space–time extreme (STE) statistical wave models, evaluated with a significant amount of historical directional wave data from Monterey Bay, California, USA. A novel understanding of these models is obtained by exploring the area size for which the STE models estimate a higher rogue wave occurrence probability compared to the TE models. For four rogue wave and crest heights the return periods until a 95% rogue wave occurrence probability are estimated. To illustrate an application of this statistical analysis, the risk of rogue waves to a spar-type floating wind turbine is evaluated. Based on both the dynamic response and the long-term occurrence probability of the rogue wave, this research presents a new and advanced approach for a risk analysis of rogue waves to floating offshore structures.

To examine the reliability or performance of marine systems related to dynamic time-varying responses, time domain simulation may be required. But for long durations, brute-force simulation is clearly not feasible from a computational point of view to examine converged extreme value statistics. An additional challenge is defining a set of mutually exclusive and exhaustive wave excitation records which excite the desired extreme response(s). To address these challenges, this work develops a non-linear Design Loads Generator (NL-DLG) process which links wave profiles targeted for extreme responses generated via response-conditioning wave techniques to a probabilistic framework which examines the possible correlation or mutual exclusivity between those wave profiles. The end result is an ensemble of short targeted wave profiles which excite converged extreme value return-period statistics of a defined response that may be excited by multiple correlated processes. The method of the NL-DLG process is explained by applying it towards rare wave groups, where the result is an ensemble of short wave profiles which contain extreme occurrences of different wave groups targeted for the defined exposure duration. These generated wave excitation profiles are compared to physical data from the Pt. Reyes buoy and then numerical simulations are employed to consider rarer events.