This thesis investigates the application of data-driven surrogate models within a multi-fidelity framework to aid and accelerate the structural design of motion-compensated offshore gangways. The motivation was found in the literature review, where gaps were found regarding suitability for higher-dimensional problems and multi-type input, sampling techniques, suitability for data-scarcity and suitability for application on complex mechanical systems.
The literature review starts by establishing the foundations of surrogate modelling, including its evolution from response surface methods and Kriging, to more advanced data-driven techniques such as support vector regression and neural networks. A distinction is made between physics-based, data-driven, and hybrid models, highlighting trade-offs in terms of complexity, interpretability, and data requirements. Challenges such as generalizability, extrapolation, and the curse of dimensionality are addressed, alongside the importance of design of experiments (DoE) in constructing effective models. Classical and modern DoE methods are discussed, including Latin hypercube sampling and space-filling strategies, as well as adaptive sampling techniques. Recent trends in the field include hybrid models that integrate physical laws into machine learning algorithms, and multi-fidelity frameworks that fuse low- and high-fidelity data to reduce computational cost. These advances are framed as promising solutions to limitations in data efficiency and model applicability to real-world engineering problems. The case study centers on an offshore gangway structure designed by Ampelmann Operations B.V., a company specializing in motion-compensated systems for transferring personnel and cargo to offshore platforms. The gangway operates under diverse loading conditions such as live loads, environmental forces, and cargo transfers, resulting in thousands of unique load combinations. These are categorized according to certification standards and reduced to 4016 discrete cases for model input. Structural data from an existing, fully designed system forms the basis for both model training and evaluation. A detailed description of the gangway, including its main components and loading scenarios, is provided
to contextualize the design space. The methodology involves several key stages. First, the data of the case study is described in detail. This involves defining a numerical representation of structural severity to act as the model output, condensing global responses such as stress, deflection, and buckling into a single ranking metric for each load case. A low-fidelity analytical model is implemented in Python to provide fast approximations of structural response, extended to include stress and buckling calculations. A high-fidelity finite element
model is provided by the partner company, which is simplified to ensure numerical stability. As an alternative, a simple truss model is used for better suitability for beam model analysis in FEM. As a preliminary sampling, random sampling is proposed for the design of experiment. Then, an intelligent sampling technique based on Euclidean distances is proposed to improve data efficiency by selecting high-fidelity samples in underrepresented regions. After data sampling, a machine learning framework is created to support data-driven surrogate modelling. Three model types are implemented: Ridge regression, KRR and XGBoost. These were implemented after pre-processing the data, and a random grid search is implemented for hyperparameter tuning, combined with cross-validation. The selected performance metrics are the REP, MAPE, PPMCC and the R-squared score. A multi-fidelity framework is introduced to combine predictions from low- and high-fidelity sources using the comprehensive approach.
The results show that modelling issues in the high-fidelity model lead to data that is not physically plausible. Among the surrogate types tested, Ridge regression is inadequate for complex, mixed-type input features. KRR performs moderately, and XGBoost demonstrates strong performance particularly in scenarios with limited high-fidelity data. The intelligent sampling strategy does not improve the performance compared to random sampling. The multi-fidelity model yields marginal improvements for KRR on the complex model, but it does not achieve better performance then a single-fidelity XGBoost model. The simple truss model showed an improvement of XGBoost performance when data is scarce, which confirms the hypothesis. As the simple truss model can be extended to cover the full range of load cases, this could justify the time required for implementing the multi-fidelity framework. However, in
the current state the benefit of multi-fidelity is marginal. Regarding the complex dataset with modelling issues in the source, it is still useful that this research found a well-predicting surrogate in XGBoost, even though multi-fidelity might not be the answer for that dataset due to badly correlated data. The predicting surrogate, proven to have well-predictive capabilities even though the data represents modelling issues, can point engineers to all load cases that result in exceeding members. Therefore, the surrogate could aid engineers by acting as a pointer to load cases that require engineering judgment. Also, the predictive capabilities of the surrogate enable surrogate-based optimization. Both findings can increase efficiency of the structural design of the gangway. The study concludes that surrogate modelling, when carefully implemented, can improve the efficiency of structural design evaluations for offshore gangways. XGBoost appeared to be best suited for this case involving complex, moderate-dimensional input spaces. Multi-fidelity modelling is viable but sensitive to data quality. Future work should focus on extending the framework to surrogate-based optimization and exploring adaptability to different geometries

Cylinder pressure measurement of diesel engines can be a step toward enabling condition-based maintenance, moving beyond traditional maintenance strategies based solely on engine running hours. This thesis investigates the application of a continuous cylinder pressure measurement system for condition monitoring of marine diesel engines. It addresses the question how such a system can be used effectively for the analysis of engine performance, combustion, and fault detection and diagnosis of a naval vessel’s main engines. To answer this question, a comprehensive methodology is proposed and tested on an ocean-going patrol vessel of the Royal Netherlands Navy. The main diesel engines of this vessel have been retrofitted with a permanent, multi-cylinder pressure measurement system, enabling simultaneous, high-frequency, crank-angle-resolved pressure acquisition. The setup includes piezoelectric sensors, a crank-angle encoder, and a custom synchronisation mechanism. In order to obtain useful information from the raw cylinder pressure data, these data have to be preprocessed first. Preprocessing steps applied in this thesis include:
1. Synchronising pressure data with crank angle using a method that determines the TDC shift
based on a temperature-entropy diagram of a non-firing engine.
2. Applying the correct pressure offset, preferably using the air intake manifold pressure. Alternatively, if this pressure is too unsteady, a linear least-squares regression method is used.
3. Averaging the cylinder pressures over 25 cycles to reduce random noise.
4. Applying cubic-spline smoothing to reduce non-random noise. A tunable smoothing parameter dictates the trade-off between curve smoothness and fidelity to the original data.
The preprocessed data are subsequently used to derive several performance and combustion-related indicators, using a single-zone heat-release model. Performance indicators include indicated mean effective pressure, compression pressure, and peak pressure. Combustion characteristics are assessed through net heat release, thermodynamic efficiency, and crank angles corresponding to key energy release thresholds. Cylinder-to-cylinder comparison of these parameters enables the identification of anomalies that may indicate underlying faults. Rejection criteria should be defined based on manufacturer guidelines, though these may need to be adapted. Notably, since OEM specifications only cover peak pressure and exhaust gas temperature, these criteria can be supplemented with operational experience. A structured diagnostic approach involves distinguishing whether the root cause lies in compression (e.g., valve leakage, excessive blowby) or combustion (e.g., injector malfunction).
In this study, particular attention was given to a suspected malfunctioning fuel-injection pump, identified through deviating exhaust gas temperature. However, analysis of pressure-derived parameters did not support this diagnosis; instead, another cylinder was found to exhibit signs of underperformance. This study further argues that IMEP is a more reliable parameter than peak pressure, while the angle of peak pressure is of limited diagnostic value. Exhaust gas temperature is also found to be less effective for localising faults.
In conclusion, the results of this study suggest that cylinder pressure measurement holds substantial promise for condition monitoring applications. However, validation of rejection criteria requires actual fault cases, which highlights the need for a database of known faults, including their symptoms. The study also recommends routine system checks, automation of data processing, and clear assignment of system ownership. It also recommends to implement a method for determining the correct TDC shift for the engine under load.

The composite cycle engine (CCE) is a radically new aero engine architecture that shows the potential of high fuel burn reductions by integrating the high efficiency potential of the Seiliger cycle that is used in piston engines into a conventional turbofan engine cycle. To minimise the weight penalty and increased NOx emissions from the integration of such a piston engine, a CCE concept is proposed that incorporates a free-piston engine (FPE) architecture and homogeneous charge compression ignition (HCCI) combustion. Little research has been conducted on the combination of these three cutting-edge technologies. In this study, the design space of a hydrogen HCCI opposed free-piston linear alternator (OFPLA) engine was explored through two sensitivity analyses on a validated model, showing that scavenging efficiency is critical for OFPLA performance. The bore diameter is the most influential design parameter when the scavenging efficiency is fixed. The OFPLA design is also optimised for integration into a CCE. To maximise total CCE efficiency, the inlet temperature and pressure drop over the OFPLA core should be minimised. Compared to a state-of-the-art conventional turbofan engine, the CCE with an optimally integrated OFPLA core did not achieve the efficiency gains that were reported in the literature. However, higher efficiency gains may be achieved when a holistic CCE optimisation approach is used that also includes the turbofan components.

Throughout current literature, different combinations between fuel cell type, heat engine type and fuel have been considered. Most studies focus on one combination, and operating conditions (fuel cell temperature, fuel cell pressure, cell voltage, fuel utilization factor, etc.) vary throughout these studies. Due to these variations in operating conditions, as well as differences in integration strategy, it is difficult to select a single combination to be the most suitable for marine applications. Additionally, current literature primarily focusses on maximizing system efficiency. Other factors such as system size, weight, IR signature and acoustic emissions are less frequently discussed in current literature. System size and weight are important factors for the feasibility of the combined cycles in current vessel designs, and the IR signature and acoustic emissions are important criteria for naval applications.

In this thesis, a solid oxide fuel cell (SOFC) is combined with an internal combustion engine (ICE) or a gas turbine (GT), and the system is fuelled by diesel or methanol (MeOH). This results in the following four combined cycles:
• Diesel-fuelled SOFC-ICE
• Diesel-fuelled SOFC-GT
• MeOH-fuelled SOFC-ICE
• MeOH-fuelled SOFC-GT
These combinations are compared for naval surface combatants based on system efficiency, annual energy consumption, annual CO2 emissions, exhaust gas temperature, total system weight and total system volume. Additionally, the combined cycles are compared to the current power generation system of a SIGMA surface combatant to determine the relative performance of the combined cycles to a conventional heat engine.

To determine the annual energy consumption, CO2 emissions and exhaust temperature, the combined cycles are modelled in Cycle-Tempo. The SOFC, ICE and GT are modelled as individual models and validated with available reference data. These individual models are then combined into a single model for each combined cycle, which is validated by comparing the results with results from current literature. The size and weight of the system is then obtained by selecting readily available components and summing their dimensions and weights.

For each comparison criteria, a different combined cycle performs the best. To minimize annual energy consumption, total system weight and total system volume, the diesel-fuelled SOFC-ICE combined cycle performs the best. To minimize exhaust gas temperature, the diesel-fuelled SOFC-GT combined cycle is the best choice. To minimize annual CO2 emissions, the MeOH-fuelled SOFC-ICE combined cycle is preferred. The exact performance of the combined cycles for these comparison criteria, depends on the power split between the SOFC and the heat engine. The annual energy consumption for the diesel-fuelled SOFC-ICE is between 4.4% (for a power split of 4.6-95.4) and 18.2% (for a power split of 25-75) lower than the current system of the SIGMA. The weight of this system is between 1.5% lighter (for a power split of 4.6-95.4) and 51.6% heavier (for a power split of 30-70) compared to the current system. For both the diesel-fuelled combined cycles, the exhaust temperature at silent speed can be reduced by 212 ◦C, which is a reduction of 58.3% compared to the current system. At cruising speed, the diesel-fuelled SOFC-GT can reduce the exhaust temperature with up to 207 ◦C at a power split of 30-70, which is a reduction of 65.8% compared to the current system. When using the MeOH-fuelled SOFC-ICE combined cycle, the annual CO2 emissions can be reduced by up to 25.9% compared to the current system for a power split of 25-75.

Depending on the preferred comparison criteria (annual energy consumption, emissions, exhaust temperature or system weight and volume), a different combined cycle and power split is preferred. There is no single solution that performs best for all criteria. It is therefore up to the designer to choose which criteria is most important and select the combined cycle accordingly. For this thesis, annual energy consumption and total system weight and volume are considered most important. For the combined cycles, a trade-off needs to be made between these criteria. A higher power split results generally in a lower annual energy consumption, but this also leads to a higher system weight and volume. To select the best power split for the SIGMA, a utopian solution is defined. This solution consists of the lowest annual energy consumption and the lowest system weight. This solution is impossible to reach, and the combined cycle and power split that lies closest to this utopian solution is deemed the best. This best solution is the diesel-fuelled SOFC-ICE combined cycle with a power split of 8.4-91.6 in favour of the ICE.

The diesel-fuelled SOFC-ICE combined cycle is implemented in the general arrangement plan of the SIGMA to assess the feasibility of this system in terms of system volume. The current power generation system (ICEs and diesel gensets) are removed, and the new components for the combined cycles are placed in the newly available space in the vessel. In the current general arrangement of the vessel, a maximum power split of 7.8-92.2 in favour of the ICE can be placed without having to change the design of the vessel. This power split is slightly lower than the ideal power split of 8.4-91.6. This lower power split results in a slightly higher annual
energy consumption and slightly lower system weight and volume compared to the ideal power split. Compared to the current system of the SIGMA, the diesel-fuelled SOFC-ICE combined cycle with a power split of 7.8-91.2 can reduce the annual energy consumption and CO2 emissions by about 6%. The exhaust temperature at silent speed reduces by about 212 ◦C (about 58%) and by about 80 ◦C (about 25%) at cruising speed. This reduction in energy consumption, CO2 emissions and exhaust temperature comes at the cost of a 4% increase in total system weight.

Overall, fuel cell combined cycles can help in reducing the fuel consumption and GHG emissions of the marine sector. Nevertheless, the implementation of fuel cell combined cycles alone is insufficient to reach the net-zero targets set by the IMO as outlined in [2]. For naval surface combatants, the combined cycles only reach a small reduction in fuel consumption and CO2 emissions (6%). The main benefit for these vessels lies in the reduction in exhaust gas temperature. A significant reduction in exhaust gas temperature (58% at silent speed and 25%
at cruising speed), and therefore a reduced IR-signature, is possible with a system small enough to still fit in the current design of the SIGMA.

The assessment of efficiency potential in hybrid electric turbocharged spark-ignition internal combustion engines (SI ICE) presents a significant opportunity for enhancing powertrain performance while reducing fuel consumption and emissions. This study investigates the integration of hybrid electric turbocharging technology for large SI ICE to optimize fuel efficiency and overall engine output. By analysing key performance parameters, including pressure and thermal dynamics, fuel consumption, and emission characteristics, this research provides insights into the feasibility of hybrid turbocharged SI ICE systems as a transitional solution toward sustainable transportation.
A comprehensive GT-POWER modelling and simulation approach of a CAT G3508a SI gas engine is employed to evaluate hybridization strategies for steady state static simulations. Experimental validation through test bench data substantiates the accuracy of the engine model, ensuring practical applicability. The study explores the synergy of electrical recuperation and forced induction of the CAT G3508a to enhance brake thermal efficiency across different operating conditions. Results indicate that the introduction of an electric machine in the turbochargers rotating assembly significantly improves engine efficiency, particularly at partial load conditions, where conventional SI-ICE often suffer from reduced efficiency due to pumping losses. Fuel economy improvements, reductions in carbon emissions and system versatility demonstrate the potential of this technology to bridge the gap between conventional fossil fuels and alternative fuels.

This thesis explores the potential of Variable Speed Generators (VSGs) within hybrid DC power systems for next-generation Anti-Submarine Warfare (ASW) frigates. As naval operations increasingly prioritise efficiency, emissions reduction, and adaptability to transient load variations, VSGs present a compelling alternative to conventional fixed-speed AC generators. While VSGs offer notable improvements in fuel efficiency, their transient performance under dynamic operating conditions remains a critical area of investigation. To assess these dynamics, a mean value simulation model was used and validated against Factory Acceptance Test (FAT) data and naval standards. The findings reveal a fundamental trade-off: although VSGs enhance fuel efficiency in steady-state operations, their transient response presents challenges, particularly in thermal management and stability. Slow turbocharger dynamics lead to prolonged inadequate air excess ratios during load transitions, resulting in incomplete combustion, excessive soot formation, and increased thermal stress. Additionally, VSGs exhibit difficulties in managing speed and voltage fluctuations during abrupt load changes. Despite these challenges, the study identifies several potential solutions, including electrically assisted turbochargers, advanced energy storage systems, and hybrid control strategies. These innovations can mitigate transient inefficiencies, enhancing both stability and responsiveness. In conclusion, while VSGs represent a significant advancement in naval power systems by combining fuel efficiency with operational flexibility, their successful integration requires targeted improvements in dynamic performance and system-level optimization. By addressing these limitations, VSGs can evolve into a more resilient and robust power solution for next-generation naval vessels.

For many years, ’het Loodswezen’ has relied on its tenders to transfer pilots to and from cargo vessels, with the primary focus on ensuring safety of both crew and pilots. In response to evolving emissions regulations and industry developments, the organisation launched the ’Tender of the future’ project. This initiative also provided an opportunity to enhance safety and comfort by redesigning the hulls of their tenders to address slamming, a significant issue observed in the latest tenders, the M-class. This challenge shaped the objective of this Master’s Thesis: "Delivering a new hull design for the tenders of ’het Loodswezen’ which will reduce the vertical accelerations caused by waves compared to the currently used L- and M-class tenders."

The first sub-question towards achieving the main objective is: "Which width is the optimal trade-off between a safe and efficient mode of operation and a high L/B-ratio?" This question was explored from two perspectives. From a safety standpoint, it involved examining the expertise of ’het Loodswezen’, relevant regulations, literature and consultations with Damen to identify the narrowest width that still ensures safety. From a practical perspective, simulations were conducted using Fastship (a tool designed for evaluating vertical accelerations of vessels by analysing cross-sections as falling wedges in the early design stages). These simulations were used to assess the impact of varying widths, achieved by scaling the Stan Pilot 2205, on
vertical accelerations. The combined results of these analyses identified the optimal breadth for a tender of ’het Loodswezen’ as 6.46 metres, a design that reduces the vertical accelerations with 10% compared to current tenders while maintaining safe and efficient operations.

In the subsequent phase, the bow and overall hull design were iteratively refined from a standardised planing hull towards the optimal design concept for a pilot tender of ’het Loodswezen’. Many iterations were performed and all simulated in Fastship in order to analyse its peak vertical accelerations. These peak values were used to define comfort, as the crew will base their actions on the highest impacts rather than on average vertical accelerations. The main focus of the iterations was on:
• Maintaining the required initial stability.
• Increasing the deadrise angle to 25 degrees.
• Introducing a twist by increasing the deadrise angle toward the bow.
• Implementing a spray rail to enhance hydrodynamic performance by ensuring that the water is separated from the hull.
• Raising the deck at the bow to increase the reserve buoyancy to prevent bow diving.
• Widening the deck at the bow to ensure that the same operational procedures can be applied.
These iterations culminated in the final bow and hull design, optimised for the tenders of ’het Loodswezen’. This design reduces peak vertical accelerations in the bow while maintaining resistance and operational efficiency, achieving a balance between performance and comfort for safe and effective pilot transfers.

Reducing the bow flare and narrowing the bow have proven to positively affect vertical accelerations but introduce potential challenges maintaining efficient operation with the current working methods. To address this issue, the next sub-question was posed: "Can the deck be shaped in such a way that the drawbacks to operational efficiency are minimised while the flare in the bow is significantly reduced?" Through several iterations, a deck design was developed that enhances operational performance without substantially increasing vertical accelerations. This was achieved by incorporating overhangs in strategic locations, enabling the tenders to operate effectively at an angle against cargo vessels.

A final design was developed that reduces vertical accelerations with 27.8% compared to a standardised planing hull while maintaining operational efficiency and considering resistance. However, the results should be interpreted with caution, as they are based on simulations using a basic version of Fastship with a limited number of input parameters. These simplifications enabled rapid evaluations but inherently affect the precision of the results.
It is important to note that this design represents a conceptual model. Further research and development are required to obtain reliable results for the actual design. For instance, the LCG plays a crucial role in performance but was, for simplicity, aligned with the LCB in this conceptual phase. In subsequent stages, the LCG should be carefully established to positively influence the results. Additionally, the hull fairing remains to be completed, a necessary step for generating a hull shape suitable for resistance calculations in simulations.
With these advancements, a comprehensive simulation can be conducted by Marin, for which budget has already been allocated. This simulation will provide a more accurate comparison of the new design with the existing L- and M-class tenders of ’het Loodswezen’, ensuring a robust evaluation of its performance.

This thesis investigates the yaw stability of wind-driven vessels using wingsails, addressing the urgent need for decarbonising global shipping. An analytical method is developed to assess directional stability based on aerodynamic and hydrodynamic characteristics, yielding a universal stability criterion validated with SEAMAN data. The study explores practical stability assessment approaches and evaluates the impact of varying hydrodynamic parameters, aerodynamic models, and surge coupling effects.

Results show that while stability diagnostics are invariant under different hydrodynamic models, significant variations occur in destabilised hulls. Including a boundary layer in the wind model has negligible effects on stability but increases computational effort. The closed-loop analysis assesses proportional and derivative feedback control strategies, demonstrating satisfactory stability for both human and autopilot control across all points of sail in the original hull configuration.

Time-domain responses to step rudder inputs indicate small steady errors and oscillatory components under human control, with autopilot control yielding even more robust outcomes. The study highlights the importance of surge coupling in stability analysis, often overlooked in previous models. The thesis concludes that wind-driven vessels can achieve directional stability under active control schemes, providing a foundation for future research on sustainable maritime transportation.

A significant push toward sustainability has emphasized the need to make all vessels as energy efficient as possible. Close attention must be paid to the performance evaluation of the vessel’s main energy consumer, particularly the engine. Moreover, prioritizing the efficient operation of the engine is essential for maintaining the vessel’s operational safety. Engine performance evaluation has been conducted using various methodologies, based on either field testing data or physical and data-driven models. This research focuses on developing a condition monitoring model that is effective even with limited data availability, highlighting the correlation between subsystem degradation and its impact on the overall underperformance of the engine. The literature review examines potential methods for condition monitoring of marine diesel engines, with a focus on their modeling purposes. These methods are classified based on their theoretical foundation: White-box models, which are based on first principles, and Black-box models, which rely on the interrelationships between input and output parameters using available mathematical tools. Additionally, the literature discusses Grey-box models, which combine physics-based principles with data-driven approaches. The modeling purpose of the methods analyzed is identified as either parameter estimation or fault detection and isolation, depending on the available data and the chosen methodologies. To achieve the objective of this thesis, a structured model development process is formulated. This process begins with defining the available parameters and determining which are inputs and which are outputs. The dataset then undergoes pre-processing steps before being divided into training, validation, and testing datasets. The modeling procedure begins by developing a basic model using shop test data (manufacturer data) to benchmark target parameters and determine whether a data-driven or hybrid model is necessary for engine condition monitoring. A neural network is then employed to capture data variance and leverage its ability to identify non-linear relationships. Building on the baseline neural network model, the loss function is customized to incorporate physics-based information into the data-driven model. This integration is achieved using two methods: either by approximating shop test data to represent engine behavior or by incorporating physical equations derived from mean value engine models. These physicsbased elements are introduced as new loss terms within the loss function evaluation. Each of the five developed model variations is then validated for their predictive capacity, with evaluation metrics compared to identify the most promising approach. Additionally, the models are assessed for their ability to generalize to unseen data. After concluding the modeling process, a case study was developed to examine the practical application of the best model in creating a condition monitoring tool. This tool relies on monitoring the residual trends of target parameters and correlating these trends with potential fault types in engine subsystems. By analyzing these residual trends, the model can effectively link subsystem degradation to overall engine underperformance, providing valuable insights for fault detection. Transitioning to the conclusions, it becomes clear that incorporating physics-based information, particularly from mean value models, yields superior results in capturing the variance of actual measurements and accurately predicting different operational conditions. This highlights the critical importance of developing hybrid approaches that combine the strengths of data-driven models while addressing their inconsistencies, thereby producing results closer to the ground truth. Further recommendations include integrating recurrent neural networks and establishing a decision support system to help identify the root causes of degradation and suggest necessary actions. Overall, developing a condition monitoring model proves beneficial for identifying underperformance and ensuring the correct operation of the vessel.