To improve the understanding and analysis of internal combustion engines, this thesis develops a thermodynamic model that simplifies the complex processes within an engine cylinder into a single, uniform system. The core energy conversion phases of compression and expansion were first simulated using a computationally effective ”zero-dimensional” method, which was first applied to a diesel engine using MATLAB. A key investigation involved comparing four established empirical models used to estimate in-cylinder heat loss, a critical factor for engine efficiency. This comparison revealed that there are significant variations in the predicted engine performance depending on the model choice. The diesel simulation was then validated against experimental data and refined using more realistic gas properties and a dual-zone combustion model to better represent the diesel combustion.

A major contribution of this work was adapting the validated framework to a hydrogen-fuelled spark-ignition engine, testing the model’s flexibility for alternative fuels. This involved recalibrating fuel properties and the combustion model to match hydrogen’s unique burning characteristics. When compared against experimental results, the simulation accurately replicated in-cylinder pressure and energy release patterns. The Woschni heat loss model was identified as most representative for the hydrogen experiments; however, the simulation consistently overestimated the engine’s work output and thermal efficiency.

The study concludes that these simplified models are highly effective for analysing internal thermodynamic trends and comparing the behaviour of different fuels. However, their ability to predict absolute performance is fundamentally limited by the accuracy of the empirical sub-models used for heat transfer and, most significantly, combustion. This work underscores that achieving high-accuracy performance predictions from such models requires robust experimental calibration to account for real-world combustion inefficiencies not captured by idealised functions.

Turbulent flows can be found in many industrial applications, and have a profound impact on the drag resistance and heat transfer characteristics of engineering equipment. While the behavior of turbulence is well-documented for canonical flow cases, its behavior under complex conditions is largely unknown. This poses a significant challenge when designing engineering equipment, since it is essential to accurately predict the impact of variable property flows and surface roughness on pressure losses and heat transfer rates. Yet such practical cases are beyond the scope of canonical flow data. During recent years, the combination of advances in machine learning and GPU-accelerated flow solvers has yielded promising results in the fields of turbulence and scalar transport. Using a few modern GPUs and state-of-the-art algorithms, it is now possible to simulate highly complex turbulent flows in short periods of time. This has enabled further progress in the field of machine learning for fluid mechanics, since high-fidelity turbulent flow databases can be easily generated, and new models can be trained to predict increasingly complex flow effects accurately. In this project, multiple challenges are addressed, such as optimizing GPU-accelerated DNS solvers for extreme-scale simulations, creating data-augmented RANS turbulence models for flows with strong variations in their thermophysical properties, and developing machine learning models for turbulent flows past rough surfaces.

Prior to working in GPU-accelerated simulations, the sub-project about machine learning for variable property flows focused on building data-augmented RANS turbulence models using a technique known as FIML (Field Inversion Machine Learning). In this framework, non-linear optimization is first performed to obtain an ideal set of corrections, after which a neural network (or another model) is trained to predict the observed corrections. When working with variable-property flows, several challenges arise, such as not knowing beforehand the local fluid properties due to thermal changes. This makes it difficult, for instance, to calculate the input features for the neural network. The solution created is a feedback loop, where CFD predictions are used to re-calculate the local fluid properties and to update the neural network input features, etc. The results show that the data-augmented RANS model can accurately improve the predictions for unique flow cases, which need unusually high corrections not observed in the training set. Additionally, a weighted relaxation factor methodology is proposed, to ensure convergence of the RANS models after inserting neural network corrections. The final results show that, for the most challenging CFD case identified, our machine learning system is able to reduce the L-infinity error on the velocity profile from 23.4% to 4.0%.

To generate the required amounts of data for machine learning studies regarding complex geometries like roughness, it was necessary to develop a GPU-accelerated DNS solver. This work focuses on the implementation of a parallel tridiagonal solver for extreme-scale simulations, and the creation of a new cross-platform communication library for supercomputers with either AMD or NVIDIA GPUs. In general terms, turbulent flow simulations in GPUs can be highly efficient, since all operations can be mapped to different GPU threads. However, large-scale data transfers are the main performance bottleneck of GPU-based simulations. While halo exchanges between GPU sub-domains have a minimal impact, the large-scale transpose operations needed for 3D arrays inside Poisson/Helmhotz solvers occupy most of the total running time. Therefore, any transpose operation avoided will drastically improve the running times for the entire DNS solver. By using a parallel tridiagonal solver, it is possible to reduce the number of transposes (for full 3D arrays) by 50% for 2D pencil decompositions inside the Poisson/Helmhotz solvers, and to replace these avoided transposes by simplified operations with a computational cost resembling halo exchanges. Additionally, in this work, a new opportunity to speed up simulations was found, by re-deriving the coefficients of parallel tridiagonal solvers to substantially improve the efficiency and the GPU parallelization of the DNS solver for implicit 1-D diffusion equations. Based on these improvements, it is observed in the results that the efficiency of the DNS solver is substantially improved for extreme-scale simulations in the LUMI and Leonardo supercomputers. Moreover, we show that the entire DNS solver can operate using 2D pencil decompositions without performance degradation compared to most optimal 1D decompositions available for smaller systems. Therefore, the parallel tridiagonal solver enables high-performance in extreme-scale simulations, where 1D decompositions are not feasible.

To enable extreme-scale simulations in AMD GPUs, a new cross-platform communication library was created, named diezDecomp. This library is able to achieve high performance working with both CPUs and GPUs in NVIDIA or AMD-based supercomputer. The underlying algorithm corresponds to an advanced implementation that works by directly intersecting the x/y/z bounds of all MPI tasks and scheduling data transfer operations. This allows the implementation of any-to-any transpose operations between mismatched 2D pencil decompositions, with complex communication patterns beyond the scope of traditional all-to-all operations. In extreme-scale simulations, direct x-to-z transposes can improve efficiency while solving for implicit 1D diffusion, but they are not available in existing libraries. Thanks to the flexibility of the diezDecomp library, x-to-z transposes can be easily implemented, and the running times of implicit 1D diffusion solvers were improved up to 55% for extreme-scale simulations in the LUMI supercomputer with 1024 GCDs.

The benefits of machine learning to predict the thermal and hydrodynamic behavior of turbulent flows past rough surfaces are explored in Chapter 4. Due to the complexity of this task, a convolutional neural network was used to (independently) scan the input height maps of rough surfaces, and to generate detailed 2-D maps with the local skin friction factors and Nusselt numbers. The proposed neural network is optimized to have linear time complexity while creating 2D maps, instead of quadratic complexity as in naive approaches. The validation study using randomized surfaces with the Fourier spectrum of grit-blasted surfaces shows that machine learning can make accurate 2D predictions for both the local skin friction factors and Nusselt numbers of rough surfaces, with median deviations of 28.43% for the skin friction factors and 6.37% for the Nusselt numbers respectively. The averaged errors in the predictions for skin friction factors and Nusselt numbers were reduced from 24.9% and 13.5% using traditional correlations to only 8.1% and 2.9% thanks to machine learning.

Since the neural network predictions for the thermal behavior of rough surfaces were particularly promising, a further optimization study is presented in Chapter 5. Here, the objective is to combine the benefits of machine learning and GPU-accelerated simulations to improve convective heat transfer in dimpled-surfaces. Remarkably, it was found that machine learning can find a highly optimized dimpled-surface with a 53% higher Nusselt number using cross-aligned dimples after being trained with only random surfaces (displaying lower thermal performance). After the second iteration of the reinforcement learning loop, it was confirmed that the surface found by machine learning created a flow pattern with helical structures inside the dimples. All other tested parameters had lower thermal performance.

In summary, it is concluded that GPU-based DNS solvers can be optimized to enable extreme-scale simulations with minimal performance degradation. Creating highly flexible communication frameworks, such as the diezDecomp library, is also possible while keeping identical running times as traditional libraries. This research project also showcases how machine learning can be an effective tool in fluid mechanics, to predict the behavior of turbulent flows past complex geometries or to account for changes in flows with strong variations in their thermophysical properties.

High energy demand of contemporary society increasingly relies on offshore production. Offshore floating energy production infrastructure, such as floating wind turbines, photovoltaics, and production storage and offloading units, is central to sustaining the energy supply. Mooring chains, which are crucial elements of these floating units, are particularly vulnerable to degradation mechanisms, such as corrosion, fatigue, and their combination. Their frequent inspection is critical to prevent structural failure, production shutdowns, and severe environmental consequences.

A Digital Twin (DT) is a virtual representation of a physical object that uses real-time data to simulate and predict its behaviour. Digital twins have emerged as promising tools, enhancing the monitoring and management of physical assets using real-time data and predictive models. The effectiveness and reliability of a DT for quantitative integrity assessment are contingent on the quality of the input data and the models employed. When paired with predictive models, DTs offer valuable insights for managing the integrity of offshore mooring systems.

The main contribution of this work is to lay the groundwork for enrichment of DTs for mooring chains with quantitative damage assessment. Assessing the structural integrity of submerged offshore mooring chains presents significant challenges due to their difficult accessibility and the complexities involved in subsea inspections. Additionally, the presence of marine growth on the chain links often requires cleaning before detailed inspections can be conducted. The cleaning process is often undesirable from technical, economic, and environmental perspectives. Currently, no in-water inspection techniques for mooring chains have been reported that do not require the removal of marine growth beforehand. Enabling this is another contribution area of this research.

This thesis proposes a novel non-contact acoustic emission (AE) monitoring approach in underwater environments, enhancing DTs of mooring chains without suffering from the limitations of conventional inspection methods. As a passive ultrasound method, AE is an established non-destructive testing (NDT) technique to detect and monitor corrosion and fatigue. Piezoelectric sensors, typically mounted in contact with the material surface, are often employed to capture high-frequency waves generated by damage initiation and propagation. Three key areas have been explored: (i) the feasibility of detecting and monitoring corrosion-fatigue in submerged conditions, both with and without marine growth, using non-contact AE, (ii) the construction of a DT representation based on AE data to identify the location of corrosion-fatigue damage in mooring chain links, and (iii) the integration of AE data with fatigue models to enhance the DT predictive capabilities for damage prognosis.

The feasibility of monitoring corrosion-fatigue damage in submerged conditions using non-contact AE monitoring has been explored using small-scale corrosion-fatigue experiments. The results demonstrated the effectiveness of the proposed approach, with corrosion-fatigue-induced ultrasound signals being detected with a satisfactory signal-to-noise ratio within the frequency range of 50–450 kHz. Cumulative and rate values of AE parameters provided a reliable representation of damage progression, successfully identifying four distinct stages of damage evolution. AE energy, in particular, proved to be the most promising indicator, especially for highlighting crack formation and rapid growth phases. Corrosion-fatigue-induced signals exhibited significantly higher energy levels, approximately an order of magnitude greater than those induced by corrosion alone. In addition, corrosion-induced damage resulted in fewer ultrasound signals than corrosion-fatigue damage. Further experiments demonstrated that simulated crack signals could be measured on steel plates both with and without marine growth, indicating that ultrasound waves in the frequency range of interest can penetrate marine growth and be detected by non-contact AE transducers in submerged conditions. While marine growth caused a noticeable drop in signal amplitude, the findings suggest that non-contact AE monitoring remains feasible in the presence of marine growth.

Large-scale experiments have been used to construct a DT representation based on AE data to identify the location of corrosion-fatigue damage in mooring chain links. Large-scale corrosion and fatigue tests were conducted to assess the feasibility of detecting, localising, and monitoring corrosion and fatigue damage in mooring chains. The results demonstrated the effectiveness of the proposed approach in monitoring growing damage over time. AE measurements were parameterised to track acoustic activity associated with the initiation and progression of corrosion-fatigue damage. A 3D source localisation algorithm was successfully implemented to localise damage-induced ultrasound signal sources, with the localisation results being mapped onto the surfaces of the mooring chain segment. In the fatigue test, the AE-based DT identified three distinct zones of acoustic activity, which aligned well with post-failure inspection results. The remote AE technique accurately detected and localised all damage indications found during post-failure mechanical testing, showcasing its potential for real-time damage detection and localisation in mooring chain links.

Large-scale fatigue test data were subsequently used to assess the feasibility of fatigue crack growth prognosis in submerged mooring chains using remote AE monitoring. A prognosis model, based on the Paris relation and AE energy, was proposed, and its predictive capabilities were evaluated. The results demonstrated the potential of remote AE monitoring to predict fatigue crack growth in submerged mooring chains. AE energy analysis at various test stages revealed distinct phases of crack growth, including initiation, stable propagation, and acceleration toward failure. The prognosis model effectively reflected these stages, as indicated by changes in AE energy rate. A sensitivity analysis on the model parameters showed that reducing the scaling coefficient B led to overestimated crack growth and shorter predicted fatigue life, while increasing it resulted in underestimated crack growth and longer predicted fatigue life. The power-law exponent p further amplified these effects. The fatigue life predictions underscored the importance of accurately determining the initial crack size and selecting the appropriate crack growth models to improve prognosis accuracy.

This research provides a foundation for enhancing DTs of offshore mooring chains through real-time monitoring and predictive analysis. The proposed system shows potential for autonomous inspections of subsea structures, with possible benefits including reduced inspection costs, condition-based maintenance, and improved safety. Its non-intrusive nature may also lessen disturbance to marine ecosystems. Demonstration under offshore conditions will be an important next step to assess and strengthen its practical value for supporting the safety, efficiency, and sustainability of floating energy infrastructure.

The problem of global warming is becoming every day more and more pressing, leading to the necessity of reducing harmful emissions especially in the shipping sector. Most ships sail with Internal Combustion Engines, reason why it is beneficial to keep using this system, with the prospect of being able, though, to cut down on the emissions. This can be done implementing new fuels, such as PODE and methanol. These two fuels can largely reduce the emissions without the need for major changes to the engine system. Also, they can be produced in an ”eco-friendly” way, reducing the emissions also during the production process. In this study, the working characteristics of a dual fuel engine, considering methanol and PODE are analysed, as well as the production process of PODE from methanol, process that happens on board of the vessel itself. For the dual fuel engine the analysis showed that different ratios of methanol and PODE can be considered for the dual fuel engine and that this system can deliver the required power output. The plant design resulted in a plant consisting of two reactors and a separation system. At last, the necessary fuel can be produced by a plant that is small enough to fit on the ship and not have to reduce the capacity of the vessel significantly.

The global market for the wastewater treatment industry has been compelled to devise new concepts due to several factors. Currently, approximately 80% of untreated wastewater is discharged worldwide. The conventional technology for municipal solid waste treatment, which has been in use for over a century, is only partially effective in terms of effluent quality and energy balance. The highly flammable CH4 gas emitted from municipal solid waste landfills impacts the chemical composition of the atmosphere, potentially affecting the Earth. Thus, new concepts for wastewater treatment should address multiple social challenges related to energy and sanitation technologies. A well–implemented wastewater treatment system not only contributes to mankind wellbeing, which results in more effective population planning, but also ensures the optimal utilization of environmental resources.....

In order to reach the goals of the Paris Agreement, global emissions must decrease at a high rate. Looking at the industrial sector, the largest share of energy is used for process heating, with fossil fuels as its primary source. A way to improve the energy efficiency and reduce the emissions of process heating is by the integration of heat pump technology.

However, operational costs are still a limiting factor in the widespread uptake of this technology. A potential of industrial heat pumps that will have a positive effect on the operational costs, as well as its general applicability, is the possibility to operate the equipment in a flexible manner. At this point, the potential to respond with high temperature industrial heat pumps to heat demand, the electricity market or to grid congestion, is not yet unlocked.

The focus in this study is on gaining more insight into the dynamic characteristics of industrial heat pump operation. The approach is based on the modeling of a 2 MW high temperature industrial heat pump in the Dymola simulation environment. Based on the system dynamics, a suitable control method is selected and the controller settings are optimized. With this control system in place, the dynamic limitations are studied.

The limiting factor in the considered economizer based heat pump cycle is the ability of the control system to keep the superheat at the screw compressor injection port within acceptable limits during load level changes. The heat pump is found to be able to ramp up and down at a maximum rate of approximately 20%/min. This level of flexibility allows use of the heat pump for both heat demand and electricity price response, as well as for participation in grid load balancing pools. On a higher level, based on these results flexible heat pump operation can be considered possible.

To answer the need for renewable fuels, Zero Emission Fuels is designing a microplant to create methanol from carbon dioxide and water from the ambient air using only renewable energy from solar panels. For this microplant, gaseous carbon dioxide must be compressed from 1 to 50 [bar] with the low mass flow of 336.7 [g/h]. This thesis examines the compression system which uses rolling piston compressors. This thesis also examines a solution to compress the carbon dioxide optimally.

Because of the low mass flows, the compressors are very small and have relatively high friction and leakage losses. Meanwhile, due to the high pressure ratio of the system, the temperature increase due to compression also poses a problem for the lubrication of the system. Therefore, this thesis looks for solutions for these problems while trying to gain a better understanding of what exactly is happening inside the compressors.

For the temperature regulation, different options for active cooling were assessed with a dimensionless analysis. Eventually, the concept of cooling through the compressor housing was chosen. Afterwards, a computational model was made to calculate the inner workings of the compressors and assess whether the concept of active cooling would still work. Then the computational model was verified by doing experimentation.

After that, the results from the model and the experimentation were compared. The model was able to predict the mass flow and the outside heat transfer coefficient accurately. However, the model was still off from the experimental values by quite a margin. After assessing all the data, the system seemed to perform better as a 2-stage system then as a 3-stage system. However, due to the inaccuracies of the model and experimentation, a definite recommendation on the number of stages of the system cannot be formed.

To improve the efficiency of domestic heat energy use, a heat battery was added to a boiler. This heat battery contains a phase change material (PCM) characterized by low thermal conductivity and high specific heat, enabling efficient storage of residual heat energy. Paraffin wax was selected due to its high latent heat storage capacity; however, its low thermal conductivity posed challenges regarding storage time. To address this, a finned tube was introduced, and the feasibility of the heat battery for domestic applications was investigated. The effectiveness of the finned tube was evaluated using both a numerical model and an experiment. In the experiment, a temperature- and flow-adjustable water source was connected to a plain tube and a finned tube, with temperature and heat flux measurements recorded to assess the impact of the fin structure. These conditions were replicated in the numerical model for comparison. The results indicated that heat extraction from the PCM using the finned tube required approximately 60 minutes, compared to nearly five hours with the plain tube. These findings suggest that while the method shows promise, further development is needed before it is suitable for domestic implementation.

As the world population is increasing and cities grow faster than before, food scarcity becomes a bigger issue. To depend less on import and decrease CO2 footprint, food should be produced nearby its consumers. A good solution for food production is a greenhouse which provides a steady high yield throughout the year. Some of the world's fastest growing cities are in Subtropical climates. Subtropical climates are characterized by high temperatures, abundance of sun and high humidity. Due to the high temperatures, cooling in greenhouses in subtropical climates is necessary, but evaporative cooling is not sufficient due to the high humidity. Outside air can be dehumidified by a liquid desiccant to cool the air further with an extra evaporative cooling step. After dehumidifying the air, the absorbed moisture from the air in the desiccant must be removed from the desiccant solution for re-use. This requires a lot of energy, which makes this cooling technique for greenhouses in subtropical climates not economically feasible. The proposed solution is to remove moisture from the desiccant by making use of the hot air that can be collected in the top of the greenhouse. The potential of this solution is assessed in this study. To predict the transfer of mass and heat between air and a calcium chloride solution as desiccant, a mathematical model of the process of mass and heat transfer over a structured cardboard pad is made. An experiment is conducted to validate the mass and heat transfer coefficients of the model. The model is used to evaluate the workings of the proposed system of regeneration. Results show that dehumidification can significantly lower the inlet temperature of the air in the greenhouse, provided that the inlet temperature of the desiccant is low. Regeneration of the desiccant solution with hot air is also possible, provided that the air in the greenhouse is significantly heated by the sun and pre-heating of the desiccant before regeneration is beneficial. The direct coupling of these two systems is not sufficient due to the big difference in temperature of the desiccant that is required at the inlet of the dehumidifier and regenerator. In future work attention should be paid to this temperature difference which is currently limiting the performance of the system. Then it's possible for the concentration of the desiccant to be brought back to the starting conditions for the dehumidifier.

In the report that is before you, a method has been investigated to increase the capacity of domestic hot water tanks. To do so without having to increase the volume of the tank, a phase change material has been included in an annular space around the tank. To test the performance of such a system, three main topics are dealt with in this report. First of all, the most suitable phase change material was selected. There exist many types of phase change materials, all having completely different properties, advantages and disadvantages. Secondly, weak spots were indicated during melting of the phase change material. These spots indicate the locations of the worst heat transfer rate, and to improve the performance of the system, these spots can be treated separately in future experiments or models, for example by placing fins at those locations. Finally, the response time of the phase change material was investigated during a typical discharge cycle to investigate the amount of latent heat that was released to the water. To model the charging- and discharging cycle, the vorticity stream method was used to create a two dimensional axisymmetric model. This model was implemented into MATLAB. The Boussinesque approximation was assumed to take into account buoyancy effects. Furthermore, only diffusive heat transfer was assumed to take place throughout the PCM and the effect of convection was neglected. Results showed that paraffin based PCMs are most suitable for applications in domestic hot water tanks. They are inert and stable, and their thermal properties do not degrade in time. They have suitable melting temperatures and high latent heat. To increase the thermal conductivity, dispersion of high density alumina particles was applied. Although the charging model did not reach the asymptotic range of convergence, it was shown that melting 5.0 cm PCM required long periods of time. After 9 hours, part of the PCM was still in its solid phase. Due to thermal stratification in the tank, the slowest melting rate was seen at the lower outer edges of the tank. During the discharge cycle, it was shown that not all PCM was able to release its latent heat to the water: 16.9% did not solidify. This occurred at an adapted flow rate to match the discharge time for both tanks. For higher flow rates, used in domestic hot water tanks, the amount of liquid PCM would even be greater, resulting in less latent energy release to the water. Based on the fact that 5.0 cm PCM did not increase the heat capacity of the tank, and all the previously mentioned factors, it is concluded that the performance of a hot water tank enhanced with PCM in an annular space does not increase the performance of a domestic hot water tank. Some suggestions were made regarding the possibilities of increasing the performance of the tank in future models and experiments.

Miniaturization of spacecraft has been gaining wide interest in the space industry, given its potential for reducing space missions' costs and providing a novel approach to enhancing and facilitating spaceflight. Recently, a lot of research has been successfully put into this field along with the advancements that make it more feasible, though a major obstacle to achieving the new generation of spacecraft is the technical challenge of fitting a suitable propulsion system. Useful chemical propellants are usually corrosive, flammable, and/or toxic, so alternatives need to be found. The aerospace industry is shifting towards green and nontoxic propulsion systems, so water could be used as an effective propellant, considering its relatively high mass density and low molecular mass. New microelectromechanical systems (MEMS) technologies show promising opportunities for the integration of miniaturized propulsion systems, due to their versatility and robustness. In this thesis, a comparative study of nozzle flow, heat transfer, and thermodynamics in two different thrusters is conducted. One thruster is based on MEMS, with a typically quasi-2D geometry, while the second thruster is based on more conventional technologies and manufacturing techniques, with an axially symmetrical 3D shape. After briefly introducing micropropulsion and discussing the propellant selection and nozzle fabrication along with the background theory related to micropropulsion as well as the analytical and OpenFOAM numerical (DSMC, continuum, and a hybrid approach containing both to accommodate to the variation in Knudsen number throughout the computational domain) modeling methods, the used methodology is based on using OpenFOAM's DSMC solver (dsmcFoam+) following the mesh creation using blockMesh and snappyHexMesh and developed analytical model (using MATLAB and CoolProp) along with an additional VLM ANSYS Fluent CFD model prepared in advance at TU Delft, where their (steady state as well as transient for DSMC) results (including the same conventional and MEMS nozzles) are processed and discussed. The nozzles are simulated for inlet pressures of 5 and 7 bar at inlet temperatures of 550 and 773 K for a total of four cases for each nozzle. To note, many of dsmcFoam+'s functionalities (mass flow rate measurements, inlet pressure boundary condition, axisymmetric capabilities, statistical error measurements, and dynamic load balancing) are implemented and described along with the full methodology, as Blender (with add-ons) and ParaView with a Python script to extract averaged data (along the nozzle and plume region) along with sampleDict are also used in pre and post-processing respectively and the simulations are carried out on a computer cluster. Furthermore, a quite interesting theoretical project on the side has been independently worked on in parallel. It started as a noticed idea that was decided to be explored using equations, which led to extended continuum/kinetic dimensionless numbers for diffusivity and rarefaction intensity relative to the studied object’s timescale. Ultimately, there are tradeoffs to choosing either thruster, where it is impractical to fault one nozzle for not performing better than the other, as it comes back to the desired features and nature of the mission each is undertaking, where compromises have to be made.

The vehicle industry is increasingly exploring emission-free mobility. Transforming the mobility sector to zero-emission vehicles that consume renewable and low carbon fuels is necessary to reduce the impacts on climate change. Hydrogen powered electric vehicles, or Fuel Cell Electric Vehicles (FCEVs), could make a significant contribution to reduce GHG-emissions and energy use in the transport sector. The current refuelling network for hydrogen vehicles is in its very early stages. The Cogeneration of Hydrogen,
Heat and Power project (CH2P) aims to impact the sector with a more efficient solution for the growing hydrogen infrastructure. It does so by offering very efficient decentralized H2 refuelling stations. The goal of this project is to design a refuelling station that has the ability to refuel hydrogen vehicles and to charge electric vehicles. The onsite CH2P production system is based on SMR and SOFC technology. A thermodynamic model has been designed to size and optimize all the station components. This includes the compression, intercooling, storage, throttling, precooling and dispensing of hydrogen, and the storage and charging of power. Main messages to take away from this analysis are the following: CH2P as a decentralized production method is protable. For higher station utilization the NPV has the potential to increase further by 55 %. Cascade filling of vehicles reduces the total storage and energy demand, and greatly reduces the CAPEX of the station. The optimal configuration consisted only of 950 bar High Pressure Storages. By adding a battery to the station, the system can power five fast-chargers to charge BEVs. However, power production by CH2P is more expensive than grid power and the current electric system is not sufficient for coping with sudden power fluctuations.

Heatpipes are promising devices for geothermal energy extraction owing to their
effectiveness to transport heat. The goal of this research is to validate an analytical heatpipe model with experiments and to investigate the difficulties in designing and constructing geothermal heatpipes. There is a lack of literature and research concerning the operation, performance limits and construction of heatpipes suitable for geothermal heat extraction. A prototype heatpipe is designed based on specifications for geothermal energy extraction and constructed in a laboratory set-up with sensors and data acquisition. The prototype setup collects experimental data and is used to evaluate important parameters, requirements and practical design difficulties. This research shows the difficulties in designing a geothermal heatpipe taking into account fluid choice and physical limitations as well as complications in constructing a properly sealed heatpipe under the influence of repeated heating and cooling. Furthermore it shows the limitations of the analytical model by comparing the model predictions with experimental data.

Heat transfer in multiphase flow plays an important role in nature and in numerous industries such as petrochemical, automotive, food processing, ocean engineering etc. It is becoming increasingly crucial to design more efficient industrial systems to reduce the environmental impact of industrial activities because of global warming and growing awareness about sustainability. To design these efficient systems it is important to thoroughly understand the details of heat transfer in multiphase flow.

This thesis follows the method of Direct Numerical Simulation (DNS) to provide a detailed physical insight of the fluid motion and heat transfer in the model. To accurately model the two phase flow the Coupled Level Set Volume Of Fluid (CLSVOF) method is used. The main advantage of using the CLSVOF method is that it can accurately capture the interface geometry and it has excellent volume conservation capability. The work presented here is an extension to an in-house code developed at TU Delft for Direct Numerical Simulation of two-phase flows using the CLSVOF method. The existing code has been thoroughly validated for the fluid and interface motion but the validation of the heat transfer model remains an unaccomplished task till now. Therefore, the main objective of this thesis is to validate the heat transfer model and then to study the heat transfer in droplets coalescence using this model.

The validation of the heat transfer model was accomplished by calculating the Nusselt number distribution over a bubble surface and comparing it to the available literature. Both the model results and the information from the literature showed very good agreement with each other. After completing the model validation, this model was used to study the heat transfer phenomenon between two coalescing droplets and the surrounding fluid. The results of the droplets coalescence are as per the expectations and are discussed in detail in this thesis. In the process of doing these analyses different ways of calculating the local Nusselt number and the global Nusselt number have been discussed. The validation of the heat transfer model and the analysis done for the case of coalescing droplets paves the way for conducting more complex heat transfer analysis in two-phase flows using this model.

In the Netherlands most houses and buildings are connected to a gas network since the nineteen seventies to heat our houses and cook. In march of this year the new Dutch government announced that to meet the climate accord of Paris all dutch household should be of the gas net in 2030. To do this multiple alternatives to heat homes are available and one of those which has proven to be feasible over the last years is district heat. In a district heating network hot water is pumped round a city through a network of pipe lines to houses and other buildings. The network is split into a primary and a secondary network. In the primary network the heat is transported from the source, and in the secondary network the heat is pumped from a substation to the customers. The heat usually is rest heat of a STEG- or waste incineration power plant. With the increasing demand for heat due to the disconnection of the gas network, and at the same time the search for a more sustainable heat source for the district heating network like geothermal heat the prices and costs rise rapidly. This makes the heat loss in the system a more important issue also from a financial perspective. The current installed networks are all series 1 single pipe lines of ST/PUR/PE, this means the inner pipe through which the water flows is steal, followed by an insulating PUR layer which is covered by a protective PE layer. To decrease the heat loss from the pipe lines Nuon is looking at two possible options. Series 2 single piping which has a thicker insulating PUR layer and second the twin pipeline where both the steal supply and return lines are
embedded into one insulating PUR layer and a protective PE shell. The heat loss calculations for the buried pipelines in the ground are done using empirical formulas found in literature. These are checked numerical using a pde-solver. Comparing both results it was the concluded the empirical formulas give a good approximation of the heat loss. It also showed the heat loss is for a great deal depended on the temperature in the pipe and the ground temperature. In current heat loss calculations done by Nuon the temperature gradient of the water entering the network and leaving the network in not taken into consideration. By using the ambient temperature and a formula formulated in literature an estimation of the ground temperature at the buried depth of the pipelines can be made. By dividing the network into a number of pieces with length 푑푥 and an individual temperature 푇 the temperature gradient over
the entire network is taken into account. Collected data by Nuon concerning the user demand and the ambient temperature makes it possible to simulate what is the mass flow through the system at any given time and to analyze how the system responds to a demand fluctuation. By analyzing what happens in the networks in term of heat loss during a diurnal heat curve the results show that the heat loss stays nearly constant during a day. There are however big differences between different days in different times of the year. Analyzing what happens over the course of a year especially in the summer during times of very low demand the system under performs. A main reason for this is caused by the
minimum required temperature of 70 degrees Celsius at the customers due to salmonella regulations. This causes a lot of extra mass flow of hot water pumped around the system which heats up the return flow. This is also clearly notable from the efficiency of the network which drops tremendously in the summer. Comparing the difference in heat loss for the series 1, series 2 and twin system the results where as expected an decrease in the amount of heat lost, by 14.6% for series 2 and 39.70% for twin compared to series 1. The overall image of what happens actually stays the same with high return temperature and relative high heat loss and low efficiency in summer. There are a few options to further improve the performance of the system with a few percent by changing the inlet temperature or the location of the bypass valve. Financially speaking the twin system is also the better choice of the three. Compared to series 1 for series 2 the investment costs will increase because the materials
and instalment costs will increase. For the twin system the prices of the materials will increase and the placements of the welds will become more expansive however only half the number of pipe lines and joints is needed, so in total the prices will stay nearly the same compared to series 1. There is a lot of discussion on whether or not the maintenance costs will increase for the twin system, arguments for both cases are given. However the maintenance costs are so small compared to the costs of the heat loss that in every case the twin system is clearly the better choice.

The production process of milk powder consists of multiple stages. This report focuses on the falling film evaporator, which role in the production process is to evaporate the water content from the milk. A falling film evaporator is a large vertically-placed vessel whereby the inside is filled with smaller tubes. Steam enters the vessel and heats up the outside of the smaller tubes. The milk flows in a thin layer only along the inside perimeter of the smaller tubes, so these tubes are not completely filled. The advantage of this flow is that a thin layer of liquid is continuously in contact with the wall so the heating process is equal along the tube. The thin layer is called a falling film because the thickness compared to the length of the flow is very small.

The objective of this research is to determine theoretically the local overall heat transfer coefficient of the falling film and to investigate experimentally the applicability of heat flux sensors by determining the local overall heat transfer coefficient. By investigating the falling film, it is not allowed to disturb the falling film. Once a falling film is disturbed, the falling film will proceed at a different path. Heat flux sensors allow for local non-intrusive measurements. The overall heat transfer coefficient gives information about the thickness of the falling film. Theoretically the overall heat transfer coefficient is calculated by taken the thermal resistances of each component. Experimentally, the overall heat transfer coefficient is measured by the heat flux and the temperature difference between the bulk temperature of the fluid and the sensor at the outside of the tube.

Before the heat flux sensors are used in practice on the falling film evaporator, a setup has been built to experimentally determine the applicability of the heat flux sensor. This setup has been made for a tube filled with water and to create a falling film. The tube filled with water is well-described in theory and used as a reference.

The results of the experimental setup show that the Danfoss 'Koperpasta tube AT' is in good comparison with the theoretical approach. The theoretical overall heat transfer coefficient difference, caused by the mass flow difference of 0.01 kg/s in the falling film evaporator, can be detected by the heat flux sensors taking into account the error margin of the heat flux sensor and temperature sensor.

Environmental regulations are continuously raising the bar towards more advanced diesel engine designs ca- pable of minimizing the emissions of polluting substances. The recent IMO Tier III legislation, entered into force in 2016, is forcing the engine manufacturers to meet a NOx reduction of more than 70% from Tier II for all the ships sailing in designated NOx Emission Control Areas (NECA). Although these zones are, up to now, limited to the North American and U.S. Caribbean Sea NECAs, future regions, such as the North Sea and Baltic NECA in 2021 as well as stricter national and port regulations, are increasing the uncertainty regarding the possible compliance methods.
The Selective Catalytic Reduction (SCR) represents a flexible, proven and commercially available technology capable of reducing more than 80% of the NOx in the exhaust gas. Its adoption started in the ’70s and was aimed at the reduction of stationary source emissions. From early 2000, it has been extended to the automotive industry in order to meet the Euro and EPA legislation for heavy duty and light diesel engines. Despite the evident disadvantages of the high installation cost and the considerable space requirements, the application of the SCR is subjected to a number of operating limits such as the narrow optimal temperature window and the slip of byproducts originated from the chemical reactions. These limitations can be even more severe when the diesel engine undergoes transient and dynamic operations.
The main objective of this research is to gain insights into the behaviour of the SCR technology under steady-state and dynamic operations of the diesel engine. After an accurate literature review of the working principles, reaction kinetics and modeling approach, a first principle model of the SCR has been developed to fulfill the scope. A 1D single channel approach, discretized along the reactor length, has been adopted to sim- ulate the heat exchange and the chemical reactions occurring in the catalyst. To facilitate the integration of the designed model with an existing diesel engine model available in the Maritime and Transport Technology Department of the 3me faculty, the resistance and volume approach has been selected. After the verification of the SCR model, the integrated system "SCR+Engine" has been tested under steady-state conditions, to an- alyze the effect that the added back pressure has on the engine performance, and under dynamic conditions, to simulate "real" load case scenarios that the SCR might experience.
The designed model, although limited from the chemical reaction point of view, is able to predict NOx reduction and ammonia slip under steady-state and transient operations. It can be further used in an early design phase to investigate the feasibility of the SCR inclusion in the drive trains of ships. Future studies are recommended to optimize the model, by taking into account the sulphur influence on the reduction efficiency and the catalyst aging, and to validate it with real experiments on marine SCR systems.

The Beam Dump Facility (BDF) is a new fixed target facility proposed to be installed in the North Area of CERN. Currently in its design phase, the BDF is aimed at the Search for Hidden Particles (SHiP) experiment, which purpose is to investigate the origin of dark matter and other lightly interacting particles. The BDF target/dump sits at the core of the installation, and its aim is to fully absorb the high intensity Super Proton Synchrotron (SPS) beam and produce charmed mesons. The average beam power on target is 305 kw due to which very high thermo-mechanical loads will be generated on the target/dump configuration which can lead to mechanical failure of the target material. This calls for continuous cooling and heat removal from the target material, which requires the design and optimization of a very complex cooling system to ensure the target reliability during operation.
Keeping in view the high velocities that are obtained in the channels of the cooling system of the target assembly, an extensive 3D turbulence modeling of the full scale cooling system is required in order to predict the target water cooling system behavior. A comparative conjugate heat transfer (CHT) study using a simplified 2D geometry of the BDF target for different mesh size was performed for identical y+ values to validate the pressure drop in the cooling channels and the temperature profile in the target blocks with the analytical calculations. Subsequently, a 3D model of the cooling channel was simulated for different Reynolds number and an extensive study was performed to check the behaviour of the flow in the log-law layer. In addition to that the friction factor was validated with the analytical results and with the available literature for various Reynolds number. Thereafter a full scale 3D steady and transient model was simulated using the information obtained from the previous studies. All these simulations were carried out in Ansys Fluent. The energy deposition in space on the target blocks was obtained via FLUKA MonteCarlo simulations. The variation of HTC in different channels and the fluid-solid interface temperature is found out to be in accordance with the analytical calculations. The pressure drop and the temperature rise from inlet to outlet of the cooling system is also in confirmation with the design parameters. Transient simulations were performed subsequently in order to study the impact of time and space varying energy deposition in the target blocks on the overall flow. With the given inlet velocity, the boiling at the surface of the target blocks is not expected to happen. As a final part of this study, in order to make the numerical model more robust, a mesh sensitivity analysis was done in order to optimize the mesh size, especially in the boundary layer region.

This report presents the first simple and comprehensive method to effectively compare different infrastructure configurations for reservoirs that have been found suitable for long-term storage of carbon dioxide. Modelling the different transport phenomena along the chain, this study identified a fixed set of input parameters and used them to evaluate the differences between existing and new infrastructure. The report then explores how these differences affect the conditioning steps the CO2 must undergo before injection.The model was then applied to a a test case, the P18-2 natural gas reservoir 30km north-west of the Hague. The results for this test case show that an auxiliary pumping step is required for supercritical injection at the desired pressure. Subsequently, the results are used to develop a conceptual design for a module that can either be installed on a new platform or retrofitted to existing offshore production facilities, allowing them to be used for conditioning and injection of CO2. Last, the economic feasibility analysis demonstrates that reusing existing infrastructure can reduce the costs of transport and storage of CO2 by over 50%.

Zero Emission Fuels B.V. develops a standalone System that produces renewable methanol from air and sunlight. In this thesis the steady-state model of this System is replaced with a Simulation Tool that can predict its behaviour for every minute of a year. The goal is to verify how technical design parameters influence methanol production and to determine if the business case target of 6.7 mole methanol per day is achievable. This thesis will contribute to the understanding of the System, speeds-up the design process and makes it possible to see the impact of fluctuating weather conditions throughout the year at different locations on earth. The System as it was at January 1, 2018 is implemented in the Simulation Tool and its production determined for Tucson, Arizona in the year 2005. The result is an average daily production of 3.0 mole methanol with fluctuations from 1.0 mole during a cloudy winter day to 4.2 mole during a sunny summer day. The minute by minute production is found to be influenced by all System design parameters. They must thus be designed together to make sure the System performs optimally. The Alkaline Electrolysis uses the most energy of all subsystems. Four options to improve the System are identified. These are 1) improving the power the Solar Panel produces, 2) improving the efficiency of each subsystem, 3) improving how the System is controlled and 4) integrating heat. A selection of these improvements shows that it is possible to produce 4.8 mole methanol per day. The observed seasonal and daily fluctuating methanol production demonstrates the importance of the dynamic model developed in this thesis. Compared to the steady-state model it gives a more realistic view of methanol production and can be used to improve System design. However, to reach the business case target more technical improvements are needed. It should be realised that the design decisions will be based on technical, economical and political arguments.