The global energy transition is one of the most urgent technological and societal challenges of the 21st century. To achieve climate goals, it is essential to reduce the use of fossil fuels and replace them with sustainable alternatives. In this context, green hydrogen is receiving increasing attention, particularly as asolution for sectors where direct electrification is difficult or unfeasible, and as a means to balance the intermittency of renewable energy sources. However, hydrogen production via water electrolysis remains energy-intensive and costly, particularly when powered by fluctuating sources like wind. To improve economic viability, system-level optimisation must account for technical phenomena such as dynamic efficiency, degradation, gas crossover, and the performance of power electronics.
This thesis investigates the dynamic behaviour and optimal sizing of a directly coupled wind-powered alkaline electrolyser system, with the aim of minimising the Levelized Cost of Hydrogen (LCOH). Although static operating conditions are often assumed in existing models, this work addresses gaps in the literature by developing a time-resolved simulation model that includes wind variability, dynamic efficiency, degradation effects, realistic operational limits of the electrolyser, and time-dependent efficiency of power electronics.
An electrolyser model was developed and implemented in Python using an Electrical Equivalent Circuit (EEC) approach. An initial analysis compared intra-hour wind power fluctuations with hourly averages, revealing that the impact on hydrogen conversion efficiency was negligible (<0.06% over three 3-hour periods). Given this minimal difference and the widespread availability of hourly wind data across numerous locations, hourly wind data was deemed sufficiently accurate for system-level analysis.
The model was subsequently applied to evaluate system performance across 38 onshore European locations using 2015 wind data, assuming a constant configuration of a 2MW wind turbine coupled to a 1380kW electrolyser (69% ratio). For each location, the wind turbine and electrolyser capacity factors were calculated to assess the geographical variability in system utilisation. In addition, two Dutch sites, one coastal and one inland, were studied in greater detail to analyse annual operational behaviour, power electronics impact, conversion efficiency, hydrogen yield, and degradation patterns. Finally, lifetime simulations over 20 years were performed to evaluate system economics under varying electrolyser sizes, three cost scenarios, and two discount rates. Results showed that optimal electrolyser sizing is highly location-dependent and influenced by design objectives: the size yielding the highest hydrogen production is not necessarily the one that results in the lowest LCOH. In fact, the LCOH-optimal size was consistently smaller. Moreover, cost scenarios affected optimal sizing, with higher capital costs favouring slightly larger systems to offset investment through increased hydrogen output.
Time-resolved modelling further revealed the importance of minimum load constraints (to avoid gas crossover) and degradation effects, which influence system utilisation and stack replacement timing. While lifetime hydrogen production estimates from the dynamic model did not deviate significantly from those based on static assumptions, the dynamic approach enabled more accurate performance forecasting and degradation tracking. This research highlights the necessity of time-resolved modelling for techno-economic assessment of wind-powered hydrogen systems. The developed framework provides a comprehensive foundation for future optimisation studies and supports more accurate design and investment decisions for renewable hydrogen deployment.

During specific operations, offshore marine contractor Heerema Marine Contractors reduces the draft of their Semi-Submersible Crane Vessel (SSCV) such that the floaters are submerged in close proximity to the free surface. This draft is also known as inconvenient draft. At this draft the motions of the vessel do not fully comply anymore with the computed prediction using linear diffraction theory. A proper motion prediction is required to ensure a safe execution of the offshore operation. The key issue is the submerged part of the floater, since discrepancies occur as soon as there is only a small water column on top of the floaters. Motion RAOs are computed via the hydrodynamic coefficients and wave loads. The reason why this leads to discrepancies in the motion RAO is not yet known. In addition, qualitative hydrodynamic data should be gathered for an object submerged in close proximity of the free surface.
Therefore, in this study the cause of the discrepancies in motion RAO is studied on an experimental basis for an object submerged in close proximity to the free surface. The scope is narrowed down to a two-dimensional cross-section of a SSCV-floater. Numerical simulations using linear diffraction theory are performed in WAMIT. The results are compared to experimental data obtained via model tests performed at the towing tank facilities at the faculty 3mE at Delft University of Technology.

Based on experimental data is concluded that linear diffraction theory does not predict physical phenomena that satisfy the boundary conditions and that underlie principles of the theory. Despite the fact that the experimental data does satisfy the boundary conditions of the numerical simulation, discrepancies occur at the hydrodynamic coefficients and wave load. As a result, it can be concluded that linear diffraction theory malfunctions at the inconvenient draft region, and therefore is not the correct theory to determine RAOs for an object on inconvenient draft.
Discrepancies in motion RAO are found to be dominated by the discrepancies the wave load, except for the frequency at which the added mass equals negative inertia of the body. Discrepancies for added mass are less significant than for the damping coefficient and wave load. At higher frequencies inertia becomes dominant over damping and damping deviations affect the RAO less.
The pitch motion about the center of the cross-section does not represent a rotational motion about same the degree of freedom for an SSCV. However, it allowed to experimentally investigate the global numerical extremes. In addition, it led to the finding that it seems that the rotational data is more affected than the heave data.

There is a strong suspicion that poles in the complex plane are the cause of the discrepancies in the hydrodynamic data, which result of the used Green's function by Wehausen and Laitone. This suspicion is based on characteristics of the numerical data in combination with findings in literature and the experimental data, accumulate.

Currently, an approach that makes use of free surface damping is applied to predict the motion of an SSCV at inconvenient draft. Based on experimental data can be concluded that the use of free surface damping without further modifications does not result in an accurate motion prediction.

Lastly, the effect of nonlinearities when increasing the oscillation or wave amplitude. The wave load is found to be more prone to nonlinear effects than the hydrodynamic coefficients.The force signal remained dominantly harmonic, but higher harmonic forces did show up. Furthermore, with the exception of most tests at the largest tested submergence, higher harmonic waves were measured and visually observed during the tests. These higher harmonic waves arose at the transition from the shallow to deep water regime and vice versa. These forces were found to have hardly any effect on the force signal.

Non-photochemical laser-induced nucleation (NPLIN) is a process where a crystalline phase is formed out of solution by exposure to a laser beam. In NPLIN, the nucleation probability is strongly dependent on the beam intensity and weakly dependent on the wavelength. NPLIN offers a feasible alternative to energy-intensive industrial crystallisation methods. Although several mechanisms have been proposed, little is known about NPLIN at the molecular level. Some theories suggest that nucleation rates are enhanced through the heating of nanoparticles by absorption of electromagnetic energy. In this work, molecular dynamics simulations are performed on the clustering of ions in the vicinity of a heated nanoparticle in an aqueous supersaturated KCl solution. The spherical symmetry of a spherical nanoparticle in solution is exploited by modelling a laterally periodic water column of an initial length of 500 Angstrom enclosed between a planar iron(III) oxide nanoparticle surface and a graphene piston. A cavitation bubble is formed after nanoparticle heating, leading to an increase of clustered ions according to a bond order criterion. The clustering should correspond to nucleation in experimental systems because the clusters satisfy Delta G < 0 for locally valid NPT ensembles. The results corroborate concentration based NPLIN mechanisms as the clustering is visibly induced by local solute evaporation. Pressure based mechanisms appear ineffective because no effects of pressure waves are observed. Thermostatting the graphene sheet does not yield observable dissipation of the thermal energy generated through the nanoparticle heating and at the ion clustering sites and, obstructing completion of the cavitation cycle at a feasible simulation duration. It is suggested to repeat the simulations using a conically shaped system and a piston of higher transverse thermal conductivity.

Due to rising concerns about climate change, a lot of research is currently underway with respect to the development of new technologies which can contribute to the decline of atmospheric CO₂, and will allow further penetration of renewables into the energy mix. A promising technology which is currently actively researched is the electrochemical reduction of CO₂ (ERC). This technology utilizes otherwise polluting and unwanted CO₂ and converts
it into value-added products under the influence of an electrical current. The process can therefore be designed as an energy storage mechanism since electrical energy is stored as chemical bonds. In this research, ERC towards formic acid has been investigated from two perspectives.

First, the feasibility of the commercial production of formic acid compared to other products of ERC was investigated. The electrochemical production of the most common reduction products have been compared based on production costs, energy storage capabilities, toxicity and manageability. Due to the relatively low energy consumption for 2-electron products, namely formic acid, carbon monoxide and oxalic acid, it is found that these products have the most promising business case. Additionally, ERC to formic acid is best studied compared to other products, and high selectivities are commonly reported. Formic acid and methanol are liquid at atmospheric conditions, which is beneficial as relatively large amounts of energy per unit volume can be stored without the need of additional compression or cooling. This will also allow for easy transportation. As hydrogen carrier, formic acid has the advantage that it can be decomposed in H2 and CO₂ near room temperature. In the second part of this research, the use of numerical modeling to study the reduction of CO₂ in an electrochemical cell towards formic acid/formate at elevated CO₂ pressures is presented. The model investigates to impact on the cathodic half-cell of a cell designed for the reduction of CO₂ in aqueous electrolyte solutions at a constant temperature of 25℃, simultaneously assuming non-limiting conditions with respect to the anodic half-cell. The modeled part of the cell has been divided in three main regions, namely the bulk, cathode surface region and the electrode surface, which are discussed separately. The bulk is assumed to be the region of equilibrated concentrations which are constant in time, as they are not dynamically influenced by any mass transport phenomena. Reactants are supplied from the bulk to the electrode surface and products are removed vice versa via the cathode surface region, which is a thin region in the vicinity of the electrode. The transfer of species within this region and the chemical reactions between the species, form a system of diffusion-reaction equations. This system is solved numerically using appropriate boundary conditions. The actual reduction of CO₂ occurs on the electrode surface, and the kinetics of the electrochemical reactions towards HCOO^-, CO and H₂ are described using Tafel-type kinetics.

The electrochemical model has been verified and compared with experimental data, and despite various simplifications has proven to be predictive of the electrochemical reduction of CO₂. It is found that the potentially beneficial effects of an elevated CO₂ pressure on both the production rate and selectivity, as experimentally observed, can be reproduced with reasonable accuracy. The CO₂ concentration at the electrode surface is identified as the main limiting factor for achieving both a high selectivity towards formate and a higher production rate on formate producing metals. The model shows that with an increased CO₂ pressure the amount of CO₂ dissolved into the solution is increased significantly, resulting in a higher concentration of CO₂ at the electrode and less mass transfer limitations.

Energy demand is constantly increasing and the use of fossil fuels causes an accumulation of carbon dioxide (CO2) which is an important environmental problem that needs to be solved. A promising solution to this problem would be the electrochemical reduction of CO2 to useful products, using the surplus of electricity from renewable sources. This would be a way of storing this excess of electricity in chemical bonds. However, this has not reached high efficiency and selectivity needed for establishing its use. For this process, the CO2 is usually dissolved in an aqueous electrolyte. This thesis proposes the new approach, using Deep Eutectic Solvents (DES) which would capture a higher concentration of CO2 helping to make the whole process more efficient. However, not all salt mixtures reach the eutectic point and therefore, the solvents formed are a low-transition temperature mixtures (LTTMs) of the selected salts.

Experimental work was performed to find out if these solvents can be used for electrochemical carbon dioxide reduction. Non-reported LTTMs were synthetized; they were formed with mixtures of the hydrogen bond donor citric acid (CA), fructose (F) and diethanolamine (DEA) with the hydrogen bond acceptor tetrabutylammonium chloride (TBA-Cl) and different quantities of water, which was found necessary to carry out electrochemistry since the solvents formed were too viscous, and this water has an important effect in the electrocatalytic reduction of CO2 as proton source and charge carrier.

These solvents were characterized electrochemically, performing studies to find out on which metallic surface they behave better (wider working potential window, low degree of decomposition and adsorption) and how different water quantities affect this process. These studies were performed using cyclic voltammetry. From this, it was seen that the solvents are more stable during electrochemical process in presence of copper. Moreover, the solvent formed by DEA:TBA-Cl:H2O in the molar proportions 1:1:1,375 was found to have interesting features that resemble CO2 reduction. As result, this solvent was further analysed using electrochemical in-situ Fourier-Transformed Infrared (FTIR) spectroscopy in surface-enhanced attenuated total reflectance configuration. With this, it was seen that the carbon dioxide was captured by the solvent and that there are visible changes when applying potential. Remarkably, the data shows the formation of a dimer OCCO on the Cu electrode surface, stabilized by the solvent.

This project shows for the first time the electrochemical reduction of CO2 in LTTMs, evaluating different solvents, metals and water quantities. Determining that the carbon dioxide reduction is possible with the solvent DEA:TBA-Cl:H2O in the molar proportions 1:1:1,375 in a copper polycrystalline electrode.