Diesel engines find application in a large number of sectors of modern industries, such as the automotive and maritime. The wide adoption of Diesel engines, along with the rise of environmental concerns created the needs for optimization of fuel efficiency and power output as well as minimization of exhaust gas emissions. These targets must be met while retaining the reliability of Diesel engines. One technique that is used to significantly enhance the reliability of Diesel engines is piston cooling. The working fluid is engine oil which is injected in the cylinder towards the lower side of the piston. The objective of the current study is to simulate the process of cooling with the aid of CFD. The numerical methodologies that are used in the current study are the Reynolds-Averaged Navier-Stokes (RANS) equations for the continuous phase, while for the disperse phase the modelling is conducted with the Lagrangian modelling. The numerical model which is built focuses on the effect that the mesh resolution has on the results of the simulation, especially on the disperse phase of the flow. It can be inferred that there is dependence between the quality of the results and the mesh resolution. It is shown that the grid has to be fine enough to produce quality results, but if there is excessive refinement, the quality of results drops because assumptions of the Lagrangian modelling are not satisfied. Following this phase of the study, a CFD model is built to simulate the phenomenon of droplet impingement on a solid wall. In this set of simulations, the goal is to compare the impingement model that is used in the software with published experimental results of droplet impingement. The results indicate agreement between the CFD model and the experimental process. In the final stage of the thesis, there is the numerical study of heat transfer involved in the process of spray impingement on a hot wall. A CFD model is built to compare the heat transfer which is predicted by the software model and results from relevant published experimental studies. It can be concluded that there is divergence between CFD and published experimental results, stemming from the modelling assumptions of the software. The results of the present thesis can be used in future studies of the company, either as a continuation of the spray cooling topic, or for novel studies such as gallery cooling.

Wall-resolved large-eddy simulations are performed to study the interaction between a supersonic turbulent boundary layer and an impinging shock over a wall. The freestream conditions used were Mach 2 and a moderate friction Reynolds number of 𝑅𝑒𝜏 = 950. A passive control method was implemented using perforations in the interaction region of influence of the reference case (flat plate), with the addition of two separated cavities beneath the perforations. The cavities were designed to work as Helmholtz-like resonators at a separation length-based Strouhal number of 𝑆𝑡_𝐿𝑠𝑒𝑝 ≈ 0.03. For the reference case results consistent with the literature were obtained and a time analysis, based on the cross-correlations between dynamic properties allowed to suggest a sequence of events driving the unsteadiness of the interaction. The controlled case, resulted in a larger region of influence of the interaction, with the blowing-suction mechanism inside the cavities allowing a reduction of the separation length. Large oscillations were found close to the wall for thermodynamic properties, but also skin friction and wall-normal velocity. The topology of the recirculation bubble changed to become less symmetric resulting in a stronger reattachment compression fan. Regarding the unsteadiness of the controlled interaction, a tonal behaviour was found (at 𝑆𝑡_𝐿𝑠𝑒𝑝 ≈ 0.3) for the reflected shock motion associated with the resonant frequency found inside the cavities and also for the bubble volume variation. In general, the resonance within the cavities seemed to be able to affect the dynamics of the interaction, despite maintaining its global topology.

Nature has been optimizing for millions of years the aerodynamics of dragonflies. The main goal of
the present thesis is to understand these mechanisms so that they can be further applied in future
bio-inspired designs.

To start with, the experimental and numerical methodology is described (chapter 3). Next, a
single flapping wing is studied under different conditions (chapter 4). Reynolds number, angle of
attack, wing shape and corrugation effects are characterized. It is shown that dragonflies leading
edge vortex is responsible for a great amount of lift production. Leading edge vortex circulation
increases with Reynolds number, and so does lift. However, drag is also found to be a crucial
contributor to the force balance that sustains dragonflies hovering. Additionally, corrugation effects
improve aerodynamic efficiency in the studied flow regime.

Finally, wing-wing interaction effects are studied numerically in a whole dragonfly (chapter 5). It
is illustrated that phase changes between hindwings and forewings can maximize force production or
be tuned to achieve a more stable and efficient hovering. However, not all phases are appropriate for
maximum efficiency. Phase has to be tuned to maximize wake-capturing mechanisms and therefore
flying efficiency. Finally, vorticity removal mechanisms are depicted to maintain a clean and uniform
background flow that optimizes hovering efficiency.

Many wind turbines experience leading-edge erosion on their blades due to rain and hail impacting at speeds of up to 100 m/s. The impact speed is driven predominantly by the blade tip-speed, which is expected to grow in future turbine generations as they become larger. Erosion can remove substantial amounts of material from the blades. Eventually, the damage can reach deep into the structural layers of the blade, where it then starts to jeopardize its structural integrity. The associated roughening of the blade is accompanied by losses in the annual energy production (AEP). These are estimated to be up to several percent, depending on the severity of the erosion damage. While some leading-edge protection systems have been developed, no satisfying solution has been found, and the mechanisms that lead to erosion have yet to be fully understood. The aim of this thesis is to enhance the understanding of the physical mechanisms that promote erosion, understand which site conditions contribute to erosion and apply the gained insight in the erosion-safe mode.

This thesis starts in Chapter 2 by analyzing the impact of erosion on the AEP loss by using reduced-order modeling and subsequently compares it with the erosion-safe mode (ESM). The ESM is an alternative operational erosion mitigation strategy that aims to mitigate erosion by reducing the tip-speed of the turbine during precipitation events. It is shown that, depending on the mean wind speed and frequency of damaging rain at the site, the erosion-safe mode can lead to a lower AEP loss in comparison to a mildly eroded blade or a blade that was fitted with a leading-edge protection solution that leads to similar flow disturbance. However, it still needs to be sufficiently understood what rain is damaging and what other site conditions might promote erosion.

A step toward resolving this knowledge gap is taken in Chapter 3 by investigating the behavior of rain droplets before impact with the blade. Contrary to prior state-of-the-art, it is shown that droplets deform and break up near an incoming wind turbine blade. This finding contradicts the current approach in erosion research of modeling rain droplets as circular. It is shown that deformation reduces the impact velocity of rain droplets with the blade. This effect depends on the diameter of the rain droplets and can be in the order of 10 m/s. Small droplets experience significantly more slowdown than larger rain droplets. This reduction highly influences the formation of erosion damage since the main driver for erosion is the impact velocity. As droplet deformation and slowdown depend on the rain droplet diameter, the described effect can be termed drop-size-dependent effect.

Chapter 4 continues the investigation of drop-size-dependent effects in leading-edge erosion. An advanced erosion damage model is built that includes several drop-size-dependent effects. It is shown that the significant drop-size-effects all suggest that the erosiveness of rain droplets increases with increasing droplet diameter. This is found to be true on a per-drop basis but also when normalizing for droplet size. Therefore, selecting an appropriate droplet diameter for experiments and numerical studies is essential since not all droplet diameters contribute equally toward forming erosion damage. Drop-size effects have substantial implications for the ESM, as increasing rain intensities shift the composition of precipitation from primarily small droplets to a composition dominated by larger ones. For an equal rain column, high-intensity precipitation events are, hence, more erosive. It is found that, for a coastal site in the Netherlands, 50 % of the erosion damage is produced by the 10 % highest-rain intensity events. Thus, in ESM operation, it is advantageous to reduce the tip-speed mainly during high-intensity precipitation events to maximize lifetime and minimize AEP loss. However, a precise relation between precipitation intensity and tip-speed that optimizes this objective is not yet known in leading-edge erosion research. A novel semi-analytical approach is devised to bridge this gap, taking into account site conditions, turbine type, and drop-size effects. With this approach, it is possible to extend the erosion lifetime of a contemporary blade by a factor of 13 for a moderate AEP loss of 1 %.

A critical component for the successful utilization of the ESM is the accurate forecasting of precipitation events minutes to hours ahead. However, the best approach for obtaining this information is still debated. For the first time, Chapter 5 benchmarks a state-of-the-art weather-radar-based probabilistic rainfall nowcast product by the Royal Netherlands Meteorological Institute (KNMI). The performance of the nowcast is assessed for various lead times for three sample sites in the Netherlands and for two distinct ESM strategies. The results show that the quality of the nowcast degrades with increasing lead times. The 5- and 15-minute lead times exhibit sufficiently good accuracy and response time for the successful utilization of the ESM. Across the sites, for a large 15 MW turbine, a lifetime extension of factor five can be achieved for an AEP loss of about 1 %.

To summarize, this thesis introduced the highly significant effect of droplet slowdown and deformation occurring in the vicinity of wind turbine blades. It investigated drop-size-dependent effects and established their significance for ESM operation. It provided new theoretical insights into the ESM and used these to devise a method for finding optimal ESM strategies that exploit drop-size effects. Finally, it benchmarked the devised strategies using a state-of-the-art (operational) nowcasting product and showed that the ESM could already be a viable erosion-mitigation strategy.@en

Wall-modeled large eddy simulations (WMLES) are becoming an increasingly viable tool to study complex unsteady turbulent flows. Conventional wall models applied in these simulations are however not applicable to laminar boundary layers. While these encompass only a tiny fraction of the total surface area, erroneous predictions in this region of the flow can greatly impact the downstream flow field. In the present study, a new wall model is proposed by combining the laminar wall model and turbulent wall model with the use of a transition model marking the laminar and turbulent regions. The proposed wall model is applied to the laminar flat plate, wedge and laminar NACA 0012 flow. Results show that errors incurred at the unresolved leading edge, where the similarity solution used by the laminar wall model is invalid, accumulate in the velocity profiles for the flat plate and wedge flow cases. In underresolved regions near the leading of the NACA 0012 or near the tip of the wedge, good approximations to reference data have been found. The proposed wall model is also applied to a high Reynolds number flow involving an airfoil near stall. The proposed wall model shows promising results with good agreement for the skin friction distribution, especially in capturing the laminar skin friction peak if the transition location is known. However, the transition sensor considered for switching between the laminar and turbulent mode of the wall-stress model performs unsatisfactory. Other discrepancies in the results, such as capturing the laminar separation bubble and trailing edge separation are attributed to the relatively coarse meshes used. Last but not least, the computational cost incurred by the new wall model is marginal.

Hydrofoils can under certain circumstances cause a phase change from liquid water to water vapor. This phenomenon is called cavitation and is caused by the low pressure over the hydrofoil when the vessel exceeds a certain velocity though the water. Cavitation can take different forms depending the angle of attack α and the flow conditions which are characterized with the cavitation number σ . The different types of cavitation can have different effects on flow and loads.
Computational Fluid Dynamics (CFD) is widely used in aero and hydrodynamic design, with (U)RANS most commonly used for CFD in the industry due to its relatively low computational cost while providing sufficiently accurate results. Cavitation models in URANS simulations need a multiphase framework in order to model the liquid/vapor interface of the cavitation bubbles. The Schnerr-Sauer cavitation model uses simplified bubble dynamics equations for relatively fast calculation while providing accurate results. In current project, a Volume of Fluid method is used the Schnerr-Sauer cavitation model is used with URANS CFD simulations to improve and enhance the behaviour and performance prediction of hydrofoils sections under cavitating conditions. Given the industrial context of this project, the simulations are conducted using constrained computational resources.
Validation is performed for a non-cavitating test case using a NACA-0012 section, followed by validation for a cavitating test case using a modified NACA-66 section. Mesh convergence studies have been performed, turbulence models have been compared and the turbulent viscosity has been modified. The final set-up uses the SST turbulence model with a modified turbulent viscosity exponent n = 2.3.
To assess the flow behavior and hydrofoil section performance under cavitating conditions, a comparison is made in CFD using cavitation models, relative to the current practice. This study shows that the lift and drag results for a simulation without cavitation model are underestimated compared to the simulation with cavitation model in conditions of stable cavitation. For conditions with unstable cavitation, strong unsteady disturbed flow and loads are found that are not captured by the simulation without cavitation model. The transition from stable to unstable cavitation is studied by investigating cavitation bubble length and its
corresponding fluctuation as a function of the stability parameter
ps = σ/2(α−α0) . The conditions found for the transition from stable to unstable cavitation are consistent with reference value at about ps = 4. The inception of stable cavitation is found at about ps = 7 which is considered to be more optimized to delay the formation of cavitation compared to the NACA-0012 at ps = 8.5.
The lift, drag and performance polars are studied for several values of σ . The lift and drag polars for lowerσ , i.e. higher cavitation rate, show a stronger increase in both lift and drag for stable cavitation cases. The performance (or Lift over Drag) is slightly increased at α = 4◦ for σ = 1.2 and 1. For higher angles of attack, the increase in drag surpasses the increase in lift and the performance decreases. These finding only hold the stable cavitation cases (α < 8◦ for all tested σ , and α = 6◦ for σ = 1) since the unstable cavitation results are
inconclusive.
The main limitation of the set-up developed in the current project is that the predictions show significant discrepancies in capturing the unstable bubble shedding characteristics, with respect to the reference data. As a result, the cloud cavitation shedding frequency is not accurately captured, resulting in an inadequate representation of vibrations and loading due to cloud cavitation.

Maintaining laminar flow on large swept surfaces of subsonic transport aircraft, i.e. the wings and the stabilisers, is currently posing a considerable challenge for aerodynamic design. Improving the efficiency of aircraft by delaying or removing the laminar-to-turbulent transition process over the wing and tail parts can substantially reduce contaminant emissions. The dominant flow instability causing laminar-turbulent transition of swept-wing flow is the so-called crossflow instability (CFI). Ongoing research at TU Delft has shown potential to delay transition by use of passive mechanisms. As such, a framework has been designed to numerically compute crossflow development and transition to turbulence on swept wings. Through the use of experimental data acquired in wind-tunnel measurements at TU Delft, the CFI development and transition process on swept wings has been modelled numerically by means of Direct Numerical Simulation (DNS). Based on a DNS laminar flow field generated from the pressure distribution along the model surface, a numerical primary CFI mode in good agreement with the experiment was obtained through Non-linear Parabolized Stability Equations (NPSE). Following this steady flow field analysis, the simulation was made unsteady by the implementation of numerical free-stream turbulence. This novel method resulted in unprecedented modelling of the receptivity mechanisms of transition in three-dimensional crossflow cases, overcoming ad-hoc treatments. Both experimental and numerical flow fields indicated a Type-I dominant secondary CFI (i.e. K-H type of response in the laterally inclined shear layer of the stationary crossflow vortex), which consequently carries the formation of near-wall hairpins and ultimately turbulence. Crossflow vortex frequency content also agrees well in the low-frequency band (450Hz ≤ 𝑓 ≤ 3000Hz), whilst the numerical high-frequency content (3500Hz ≤ 𝑓 ≤ 9000𝐻𝑧) does show a distinct delay in amplitude growth throughout the majority of the transition region. Contradicting the promising qualitative analysis of the free-stream turbulence methodology, this discrepancy in the frequency spectrum indicates the major shortcoming in the numerical setup, which was shown to be biased towards introducing more low-frequency disturbances at the inflow boundary.

The development of a high-fidelity fluid-structure interaction tool for the simulation of aircraft flight dynamics in the subsonic flow regime is presented. The tool combines a high-fidelity large-eddy simulation code with an immersed boundary method, a multibody solver and a loose coupling scheme between fluid and solid. The development of the code is motivated by the need to accurately and efficiently simulate aircraft flight dynamics at off-design conditions, such as in separated flow states. The accurate simulations are necessary for the continuous optimization of current aircraft designs and the development of novel aircraft concepts. The developments presented in this thesis focus on three fields of the fluid-structure interaction problem. The modeling of solid geometries in the fluid solver through an immersed boundary method, the coupling of the fluid and solid domain through a loose coupling scheme, and the development of a multi-body solver for the simulation of aircrafts and their components. An extensive literature review is presented on these three fields of the fluid-structure interaction problem. The literature review is conducted to select appropriate methods for the final solver. Based on this review, a ghost-cell approach is selected for the immersed boundary method of the solver. A loosely coupled serial staggered procedure is selected to couple the fluid and solid domains in the solver. The floating frame of reference approach is selected for the derivation of the multi-body solver and a Newmark time integration method is selected for the integration of the equations of motion. The mathematical formulation of the selected methods is presented. Novel approaches are derived for the immersed boundary method and multi-body solver. A hybrid-cell treatment is derived to reduce spurious numerical oscillations in flow fields with moving geometries. Further, the integration of wall-modeling approaches into the ghost-cell immersed boundary method is presented. A controller approach based on the time integration scheme of the multi-body solver is derived. The controller allows user-prescribed dynamic positions and orientations of constraints and bodies. The developed fluid-structure interaction solver has been rigorously verified and validated for multiple test cases. Simulations of an oscillating cylinder and moving airfoil are presented for the verification of the hybrid-cell treatment. Results with hybrid-cell treatment enabled and disabled demonstrate the effectiveness of the developed approach v vi Summary in the suppression of spurious oscillations in the flow field. Simulations of a single and double pendulum are presented to verify the implementation of the multi-body equations of motion and constraints. The validation of the developed solver is performed with reference numerical and experimental results. Results of the laminar flow around an in-line oscillating cylinder are in excellent agreement with available numerical and experimental reference results. Simulations of the dynamic stall problem of helicopter blade sections are presented. The results show that the solver accurately predicts the flow features of the dynamic stall problem. Discrepancies between the present code and the available numerical and experimental results are attributed to an insufficient modeling and resolution of the near-wall flow field. Last but not least, flutter of a two degree of freedom NACA0012 airfoil is simulated. The developed solver accurately predicts the presence of stable, limit cycle and flutter regions. Discrepancies were found in the response frequencies between the results of the present code and available numerical and experimental results. The discrepancies are caused by insufficient resolution of the near-wall flow field. The verification and validation simulations proof the effectiveness of the derived methods and the correct implementation. The validation results further show that the solver accurately and efficiently predicts the flow field of complex flows and fluid-structure interaction problems.

During a severe accident in a water­-cooled nuclear power plant, large quantities of hydrogen can be generated in the reactor core. The hydrogen mixed with air presents a potential risk of combustion when the hydrogen concentration reaches flammability limits. If combustion occurs, pressure loads can damage safety systems and even compromise the structural integrity of the nuclear reactor walls. Thus, predicting the pressure loads is an important safety issue to ensure the reliability of critical structures in the event of a severe accident.

Computational Fluid Dynamics (CFD) codes can be used as numerical tools to assess the risks of hy­drogen combustion and predict detailed transients of the pressure loads. The scenarios simulated in nuclear safety engineering can be analyzed with low fidelity models based on reduction assumptions of the physical phenomena. Specifically, the flame gradient models are chosen to model the com­bustion phenomena based on a turbulent flame speed to which the combustion propagates. Different approaches to compute the turbulent flame speed are implemented and compared in this study, in­cluding the Turbulent Flame Closure (TFC) and the Extended Turbulent Flame Closure (ETFC). New models are proposed based on new experimental correlations specifically derived for lean mixtures of hydrogen. Other required libraries as flame radius calculation, radiative heat transfer properties determination, and axisymmetric Adaptive Mesh Refinement are implemented and evaluated. A study on the influence of the ignition source term is also carried out.

The implementation of the solver and the models is verified with the analytical solution of a planar one­-dimensional premixed flame moving into frozen turbulence. Moreover, three different experimen­tal cases, relevant to nuclear safety, are selected to perform the validation of the solver. The first one is a spherical combustion chamber which was experimentally tested with lean hydrogen mixtures and controlled measured homogeneous isotropic turbulence. The second one is the Thermal-hydraulics Hydrogen Aerosol and Iodine (THAI) facility, which is a larger vessel where effects as buoyancy be­ come more relevant. Finally, the third one is a flame acceleration enclosure called ENACCEF­2, where obstacles are placed along a cylindrical tube to promote flame acceleration and the transition from de­flagration to detonation.

Numerical results are presented in terms of the flame front location, the flame front velocity, and the pressure rise. The sensitivity of the results to mesh resolution, time discretization, and initial turbu­lence levels is assessed. The implemented combustion models are simulated with multiple turbulence models, analyzing the results and the influence on one each other. In general, the proposed mod­els represent an approach that provides good predictions in both flame development and pressure dynamics, being more robust than other models previously available

Computational fluid dynamics is nowadays one of the pillars of modern aircraft design, just as impor­tant as experimental wind tunnel testing. Very ambitious goals in regards to performance, efficiency and sustainability are being asked of the aviation industry, the kind that warrant a virtual exploration of the edges of the flight envelope. High­ order has come to be regarded as a necessary ingredient to achieve breakthrough advances in this direction. So much so, in fact, that the major aircraft manufac­turers, governmental aerospace research agencies and top universities worldwide have been acting in coordination to increase the technology readiness level of these approaches, for the last ten years. In this work, I study in significant detail the one aspect that makes high-­order methods ideal candi­dates for enabling the use of more advanced turbulence models in industry: the cost­-efficiency of their discretization—they offer minimal amounts of dispersion and dissipation errors, for a given number of degrees of freedom. I consider three research objects: discontinuous Galerkin spectral method (DGSEM), flux reconstruction (FR) and a novel B-­spline-based discontinuous Galerkin formulation stabilized via algebraic flux correc­tion (DGIGA­-AFC), and characterize their order of accuracy, linear and nonlinear stability characteristics, as well as dis­persion and dissipation errors as a function of wavenumber. Afterwards, I experimentally investigate their relative suitability towards scale­-resolving simulation of compressible and turbulent flows, by solv­ing a number of simple test cases of increasing difficulty (linear advection, inviscid Burgers and Euler equations; all in 1D) using a purpose­-made MATLAB implementation. The proposed isogeometric method (DGIGA) has been found to be at least as viable as the other two, a priori, for the resolution of high­-speed turbulent flows. Moreover, I have found that low dispersion and dissipation need not always be associated to high order, but to a high number of degrees of freedom per patch instead; these two coincide in the more conventional schemes, yet not necessarily in DGIGA. As a nonlinear stabilization mechanism, however, the proposed combination of DGIGA with AFC has turned out to be inferior to existing limiters.

Efficient combustion processes are indispensable in limiting the temperature increase to 1.5 deg C as set by Intergovernmental Panel on Climate Change (IPCC) before 2030 to curtail the effects of global warming. As for the emissions, the main task lies in controlling the air-fuel ratio in the lean regime as to control the emissions of NOx since they reach maximum discharge at the stoichiometric ratio whereas improper mixing can lead to an increase in CO emissions. Thus, the air and fuel interactions need to be studied to achieve control over emissions. To that end, good computational models are necessary to supplement the design process to control both the emissions and combustion instabilities. The latter can severely damage the combustion chamber. Therefore, the proper modeling of the fuel-air interfaces in high-pressures is pivotal as they differ significantly from the low-pressure injection. Various numerical droplet evaporation models are studied in trans/supercritical environments, the fluid is called supercritical when it's above the critical states whereas it is called transcritical when it passes the critical state. As for the droplets, two components were selected namely n-heptane and n-dodecane. The former for 0-D and 1-D models and the latter for 3-D models, the cases were dictated by the availability of experimental works. As for the 0-D and 1-D models, various correlations based on the Nusselt and Sherwood numbers were utilized. The Prandtl number and other non-dimensional numbers were computed by thermophysical property models based on the 1/3 rd rule rather than fixed values. In comparison, the developed 0-D and 1-D models conform to the experimental results and other computational studies ranging from perfect-gas to real-gas against the experimental work available in the literature. It is hypothesized that the differences in the computation of the latent heat of vaporization are more pronounced in the accuracy of the lifetime of the droplet rather than the density of the components. In 3-D models, the liquid-vapor interface is modeled by level-set and phase-field methods. Thermodynamic closure is achieved by the Peng- Robinson equation of state. Prandtl number assumption model is invoked for the computation of the liquid thermal conductivity, Chung model for the calculation of viscosity of mixtures, and Firoozabadi model for the Maxwell-Stefan diffusion coefficient. A basic model is used for the computation of surface tension coefficient for the phase-field 3-D models. A qualitative agreement was observed between the 3D model under this study, numerical work, and the experimental campaign of microscopic droplets for all the three vaporization regimes namely classical, translational and diffusive mixing. All the models yielded olive-shaped droplets. Effects of mesh resolution on the phase-field quantities were studied and contrasted with the same mesh size for the level set method in a 2-D configuration. Recommendations include better surface tension models, thermal conductivity for gaseous mixtures, the inclusion of PC-SAFT equation of state, cross-diffusion terms, and high mesh resolution of the O(-7) m in the droplet region coupled with adaptive meshing based on the gradient of the phase-field parameter.

Dimples are shallow, indented surfaces that attempt to reduce drag in turbulent boundary layer flows. However, the underlying effect of the dimples on the drag is not entirely understood. This thesis sets out to expand our understanding; a numerical investigation of turbulent boundary layer flow over dimpled surfaces is conducted. This research aims to verify the drag results from recent experimental studies and investigate the possible drag-reducing mechanism of a dimpled plate. The research considers shallow, rounded-edge dimples with a staggered layout. Implicit large eddy simulation (ILES) is carried out with the cell spacing close to direct numerical simulations (DNS). Simulation outputs conclude that the dimple plate causes a total drag increase of approximately 1% compared to a smooth plate. This value confirms the results of Spalart et al. and recent wind tunnel measurements within the Aerodynamics group. The turbulent coherent structures are further investigated by performing hole-filtering sampling to the quadrant events. Results suggest that the dimple plate induces a more intense turbulent activity in the buffer layer. The increased occurrence contributes to a higher Reynolds shear stress. The development of quadrant events is further analysed using a Variable-Interval Time-Averaging (VITA) technique. It reveals that the averaged quadrant event development between two plates is almost the same. However, a more extended sweep development is found in the wake region. Given such a mild even evolution, the resulting Reynolds shear stress generation remains the same. Lastly, the coherent structure response is linked to the skin friction response through the Fukagata-Iwamoto-Kasagi (FIK) identity. The resulting decomposition using the FIK identity reveals that dimples contribute to higher total drag due to increased Reynolds shear stress. On the other hand, the observed skin friction reduction seems relative to the mean flow convection.

The location of laminar to turbulent transition is an important consideration as turbulent flow is associated with higher skin friction drag and, by extension, lower fuel economy. In unswept boundary layers, natural transition proceeds by the amplification of Tollmien–Schlichting waves. Tollmien–Schlichting waves are convective instabilities with spanwise oriented vorticity. The amplification of these instabilities/perturbations is sensitive to roughness elements, such as forward-facing steps. These surface imperfections are inevitable as steps, gaps, and humps are a byproduct of mismatch between panels of a wing. However, their interaction with Tollmien–Schlichting waves is not very well understood. Direct numerical simulation of the flow field around forward facing steps has been performed in this thesis to gain an in-depth understanding of the particular flow features that stabilise or destabilise the incoming Tollmien–Schlichting wave, with respect to a flat plate zero pressure gradient flow. The forward facing step is found to significantly distort the base flow, its effect scaling with the roughness Reynolds number in the upstream regime. This distortion of the base flow is observed to amplify the incoming instability, both upstream and far downstream. At the step location, however, stabilisation or destabilisation can be observed, depending upon the height of the step. The step causes the incoming Tollmien–Schlichting wave to split into two, just upstream of the step, and leads to two counter-rotating structures at the step location. The interaction of these structures influences downstream growth. Localised stabilisation is observed, at the step location, for step heights that are smaller than the boundary layer displacement thickness. Destabilisation is observed for larger step heights. The upstream base flow distortion is due to an adverse pressure gradient imposed by the forward facing step. The magnitude of the pressure gradient is found to scale with the roughness Reynolds number. The upstream amplification is due to the Tollmien–Schlichting wave encountering the distorted base flow. The response of the Tollmien–Schlichting wave to the distorted base flow is observed to scale with its wavelength. The ratio of the roughness Reynolds number to the wavelength ($\gls{Rehh}/\lambda$) is found to be the governing parameter for the upstream interaction of the step with the instability.

Atmospheric entry is a crucial phase in planetary exploration missions. During entry, the vehicle experiences severe heating at hypersonic speeds. To ensure the survival of the payload, this heating needs to be mitigated using thermal protection systems. Ablative shielding materials dissipate the incoming heat largely by surface reactions leading to material decomposition. Accurate simulations of this flow environment are critical for the efficient design of spacecraft. The main difficulties in this analysis are due to strong shock waves and thermochemical nonequilibrium in the flow field, giving rise to chemical reactions and the excitation of the internal energy modes of the fluid particles. Under these conditions, an accurate assessment of the flow field requires detailed models for evaluating the physicochemical properties of the reacting fluid, and its interaction with the heat shield at the vehicle surface. This thesis considered the coupling of a high-fidelity flow solver with an aerothermodynamic library designed specifically for atmospheric entry applications. The flow solver is a Cartesian grid immersed boundary finite volume code capable of performing high-order accurate simulations. The coupling procedure with the external library involved four modules to extend the applicability of the flow solver to atmospheric entry flight regimes. The first one provides an accurate set of thermodynamic properties acquired from a tailored database for relevant species. The second module supplies transport properties through a rigorous calculation respecting kinetic theory as opposed to simplified models. The third module deals with the finite-rate chemical reactions occurring in the flow. The last module enables the consideration of catalytic and ablative surface reactions as boundary conditions for the flow solver. Thermal nonequilibrium is implemented to consider two temperatures for the conservation of translational-rotational and the vibrational-electronic energies. Gas-surface interactions are implemented for the first time in a conservative immersed interface method. Various test cases have been simulated for the verification and validation of each of these implementations. Good agreement with reference results are obtained. The increase in the computational cost is justified by the significant improvements in the results and the wide range of conditions made accessible for investigation. A novel framework is established, with which flow simulations that are state-of-the-art both in terms of numerical accuracy and fluid physicochemistry can be performed.

Although usually neglected in aerodynamics, radiation can present the main heat transfer mechanism in some aerodynamic applications and cause significant changes in the behaviour of the flow. Such applications typically include very high-temperature conditions such as hypersonic flow during reentry and hot gas in combustion systems. In these cases, radiative heat transfer must be properly accounted for and the divergence of the radiative heat flux must be added as a source term to the energy budget of the flow. To solve the radiative transfer equation, in this thesis work, use is made of the emission reciprocity based Monte Carlo formulation. To generate spectra, for combustion problems, a routine utilising the High-Resolution Transmission Molecular Absorption database was written. For hypersonic problems, NASA’s NEQAIR was adjusted and implemented. The radiation solver was further coupled with INCA CFD and validated using 14 different simulation cases. Several techniques of solution acceleration were also proposed and tested to reduce the computational requirements of the developed solver, including smart spectral discretisation methods and the use of a multilayered grid separating the radiative and spectral solutions.

Reynolds-Averaged Navier-Stokes (RANS) turbulence models are in widespread use in the commercial application of Computational Fluid Dynamics (CFD), but lack the capability to accurately compute unsteady flows and aeroacoustics due to its implied averaging of turbulent effects. In order to compute these kinds of flows, Large Eddy Simulation (LES) may be applied. However, this comes with a large computational cost. Hybrid RANS/LES aims to combine the turbulence-resolving aspects of LES, while attempting to approach the computational efficiency of RANS by applying a RANS turbulence model near walls. The X-LES turbulence model developed at NLR has been successfully applied to massively separated flows. A future application of the method is to obtain aeroacoustic predictions of turbulent noise near sharp edges and wakes. In anticipation of this goal, this work aims to assess the accuracy of the X-LES model when it resolves part of the near-wall turbulent spectrum of Turbulent Channel Flow (TCF).

Multiple TCF simulations have been performed at a variety of friction Reynolds numbers. It was found that the presence of resolved turbulent stress in the RANS-domain is not detrimental to the velocity profile in this domain, as it still conforms to the laws governing the viscous sublayer and the log-layer. The presence of so-called superstreaks causes a flow system where streamwise ribbons of high eddy viscosity, originating from within the RANS-domain, are transported away from the wall into the LES-domain. Here they contribute to the lack of development of turbulent stress, leading to an undesirable increase in velocity in the log-layer known as the Log-Layer Mismatch (LLM). Applying stochastic forcing led to a reduction in the LLM and broke up the large streamwise regions of eddy viscosity. It remains unclear in which proportion the disappearance of the high eddy-viscosity regions, and the stochastic forcing itself, contributed to the reduction in LLM.

An attempt has been made in formulating a consistent hybrid RANS/LES framework in which the effects of turbulence on the mean and fluctuating flow are governed by separate turbulence models throughout the entire domain. The initial results are presented in this report, which may be used to further develop the model in the future.

Computational fluid dynamics (CFD) has become an indispensable tool in research and engineering. Even though the performance of computers is increasing at unfathomable speeds, the calculation of many fluid problems remains to be challenging. With the emerging widespread use of Large Eddy Simulation (LES), the computational cost of simulations has further increased. Nowadays, the successful application of CFD is highly dependent on the user. A crucial factor that influences computational performance, as well as the accuracy and reliability of the simulation results, is the underlying spatial discretization of CFD problems. Adaptive mesh refinement has the potential to address these issues, by automatically creating a computationally-efficient and user-independent mesh. The objective of this Master's thesis is to improve the computational efficiency of LES CFD simulations by developing and testing a novel more user-independent AMR error sensor. A series of novel error sensors are proposed that are based on an error estimate, which is obtained during run-time by comparing the results of a fine and coarse-grained simulation. A popular reference error sensor based on the curvature of velocity magnitude is used as a comparison in three different test cases: Mach 3 flow over a forward-facing step, flow over a two-dimensional cylinder at Re=100, and flow over periodic hills at Re=10595. All error sensors performed similarly for the Mach 3 flow over a forward-facing step. Essential flow features such as the bow shock, shock reflections, and slip lines were captured and well resolved. Differences in performance were mainly attributed to the control of the adaption routine. Anisotropic refinement led to a further error reduction in the range of 10 to 15%. Performance in the laminar two-dimensional cylinder case varied substantially. The novel error sensor based on the percentage error in the solution performed the best, leading to up to five times the computational savings in comparison to the reference indicator. Across all tests, anisotropic refinement was not able to lead to any computational savings when considering the run-time of the problem. The performance of all error sensors was underwhelming for the flow over periodic hills. The error sensors failed to refine the upper wall, which, without wall-function, lead to a significant error in the solution. The overall performance of some novel error sensors was shown to be promising. Large computational savings in a robust user-independent AMR routine were presented for the compressible and laminar cases. However, more research is required for the successful applications in turbulent flows. Arguments were presented that highlight why adaption based on a target mesh size in conjunction with mesh coarsening is superior to regular threshold-based adaption. Directions and possible solutions to make adaption for turbulent flows successful are given as well.

For heavy duty diesel engines, piston temperature control is very important in constructing a successful design that meets the demands of increasing power output and stringent emission regulations. Overheating of piston is avoided with engine oil through spray cooling and gallery cooling processes. In this research project, evolution and disintegration of oil jet used for piston cooling by DAF Trucks N.V., is studied numerically using Large Eddy Simulation (LES) and Volume of Fluid (VOF) methodologies. A robust CFD model is built with major focus on the significance of grid resolution in multi-phase LES, and is used to reproduce two test cases. This is followed by characterization of physics involved in oil jet breakup through qualitative inspection. Parameters relevant to the two types of cooling techniques are estimated to see the impact of jet development at different flow rates. Based on the results obtained, the need for grid refinement is assessed and performed for certain cases. Finally, turbulent atmosphere within the crankcase is estimated through a separate simulation and its effect on oil jet is investigated. Results show that for oil in quiescent atmosphere jet turbulence is the dominant force and is the primary cause for disintegration. A clear transition to turbulence is captured with increase in flow rate, as the jet behavior is more and more chaotic with droplet formation and spreading. Importance of grid resolution on droplet capturing is recognized and an isolated analysis shows more droplets being captured with fine meshes. The level of refinement necessary to capture all the droplets still remains an open question. With turbulent atmosphere, no significant change in the jets is obtained until primary breakup and the inertial force of the liquid phase is found to dominate the surrounding flow effects. However, secondary breakup is found to be affected, as aerodynamic interactions increase disintegration and spreading, impacting both the cooling techniques. Results obtained from this research work will be used as primary inputs for further studies in the company on spray cooling and gallery cooling (sloshing flow) leading towards optimization of the piston cooling process.

An experimental investigation on shock-shock interactions has been conducted with the aim of studying the transition between Regular (RI) and Mach interactions (MI) under the condition of variable inflow Mach number. The RI contains only oblique shocks while the MI has a quasi-normal shock segment, known as the Mach stem. MI configurations generate larger total pressure losses, which is specially relevant for the design of supersonic inlets and rocket nozzles. The study was conducted in the TST-27 transonic-supersonic wind tunnel at Delft University of Technology, which allows a variation of the Mach number during a run through a flexible nozzle system. The wind tunnel model was built using two full span wedges that deflected the freestream flow to generate a shock-shock interaction. The minimum vertical distance between the wedges was kept constant throughout the runs. The Mach number was determined through continuous pressure readings in both the settling chamber and the test section. Schlieren and Focusing Schlieren systems were used to visualize the shock wave pattern. For all cases, the wind tunnel runs were initialized in the RI domain after which the Mach number was slowly decreased to the MI domain, thereby traversing the whole dual solution domain. The process was then inverted to reach again the RI domain in order to investigate a possible transition hysteresis. The Focusing Schlieren setup was capable of eliminating three-dimensional effects that contaminated the images obtained with the Schlieren visualization system. It also revealed that the main three-dimensional effect in the setup was a boundary layer detachment at the sidewalls. Schlieren visualization was used as the main quantitative flow diagnostics tool, allowing a clear view of the shock pattern and revealing the absence of any hysteresis effect. A custom built image processing software was used to obtain accurate measurements of the Mach stem height (MSH) evolution and detect transition. The MSH measurements confirmed the nonexistence of the hysteresis loop and an increase of the rate of change of the MSH with the Mach number as the deflection angle was decreased. The detected transition points between RI and MI showed a very good experimental repeatability and revealed that transition always occurred at the von Neumann condition regardless of the initial shock configuration. These results suggest that flow perturbations present in the TST-27 wind tunnel may prevent a RI from existing within the theoretical dual solution domain and thus suppressing any hysteresis loop. Further experiments and simulations are required to characterize these perturbations and fully understand the physical mechanism that triggers premature transition.

Prediction of heat loads during hypersonic re-entry is of great interest in space exploration and in the topic of space debris as well. To date, there is no accurate method to reproduce either experimentally or numerically the physics in re-entry conditions. On the numerical side, high-order discontinuous Galerkin methods have potential to improve results achieved with state-of-the-art finite volume methods. However, they suffer from Gibbs-type oscillations around discontinuities such as shocks. This work uses artificial viscosity to tackle this issue.
The artificial viscosity method used in this work contains three user-defined parameters which specify the magnitude, location and width of the artificial viscosity. A parametric study of these parameters on a steady inviscid wedge test case at Mach 2 showed that tuning their values is a non-trivial task and it appeared that in some cases shock smoothness and shock thickness need to be traded-off. Furthermore, it was observed that sufficient refinement in mesh size or polynomial order was needed to obtain satisfying results. The Sod test case was considered to test the method in an unsteady inviscid case. Good results were obtained, although the shock was better resolved than the contact discontinuity. The method was also tested on another unsteady inviscid test case, the Shu-Osher problem. The latter is more challenging due to the shock strength varying in time. Specifying the artificial viscosity parameters was challenging since the set had to account for variation of the shock strength in time. The test case also highlighted the need for a method that keeps variables within their physical bounds.
Prediction of the heat flux on the surface of a half-cylinder in a hypersonic flow was attempted. However, this was unsuccessful and the Mach number was eventually reduced to 3.25. Strong non-physical oscillations appeared in the heat flux profile even with smooth contour plots and pressure profile at the surface of the half-cylinder. The key to obtain a smooth heat flux profile was to ensure no artificial viscosity at all is inserted in the boundary layer. Also, a symmetric mesh was needed to obtain a symmetric profile. Overall, the heat flux results showed that it is possible to simulate this challenging test case with discontinuous Galerkin and a combination of simple methods, including the Lax-Friedrichs flux and a basic smoothing of the artificial viscosity field.
A new strategy to tune the user-defined parameter specifying the location of the artificial viscosity was implemented and tested on the Shu-Osher and half-cylinder test cases. It was found to be greatly beneficial in terms of stability and user-friendliness, but did not fully eliminate the issues in simulating these challenging test cases. This strategy allowed to run the half-cylinder test case beyond the Lax-Friedrichs flux and a polynomial order of 1. The heat flux profile with SLAU came very close to that with Lax-Friedrichs, whereas the difference was significant between first and second orders.