Mechanical Engineering - Theses
Permanent URI for this collection
Search Results
1 - 10 of 80
-
ItemPhysical modelling of riblet performance in yawed and non-yawed flows( 2023-11)We introduce a viscous vortex model to predict the drag and heat transfer of riblets, eliminating the need for expensive direct numerical simulations (DNS) or experiments. The footprint of a typical quasi-streamwise vortex, in terms of the spanwise and wall-normal velocities, is extracted from smooth-wall DNS flow fields in close proximity to the surface. The extracted velocities are then averaged and used as boundary conditions in a Stokes-flow problem, wherein riblets with various cross-sectional shapes are embedded. Here, the same smooth-wall based boundary conditions can be used for riblets, as we observe from the DNSs that the quasi-streamwise vortices remain unmodified apart from an offset. In particular, the position of these vortices remain unpinned above small riblets. This smooth-wall-like behaviour persists up to the optimal size of riblets, and also up to a yaw angle of 15 degrees, enabling us to predict the riblet drag and heat transfer within these size and yaw-angle constraints. The present approach is compared with the protrusion height model of Luchini et al. (J. Fluid Mech., vol. 228, 1991, pp. 87–109), which is also based on a Stokes calculation, but represents the vortex with only a uniform spanwise velocity boundary condition. The key novelty of the viscous vortex model is the introduction of a non-zero wall-normal velocity component into the boundary condition. This approach induces transpiration at the riblet crests, as transpiration becomes relevant at increasing riblet sizes. Consequently, we show that the drag and heat-transfer prediction of the present model agree with our DNS data, as well as published data.
-
ItemExploring the Similarity between Velocity and Temperature Fluctuations in Turbulent Boundary Layers with Forced Convection( 2023-10)Improvements in understanding turbulent boundary layers with forced convection enable enhanced heat transfer control solutions in practical engineering applications. However, systematic investigations on how energy-containing eddies, especially large-scale structures, relate to heat transport and how this corresponds to notions of Reynolds analogy remain to be explored further. This study delves into the role of large-scale turbulence on heat-transfer processes and sheds light on the applicability of Reynolds analogy in smooth wall-bounded turbulence with forced convection. A custom dual-wire sensor has been utilised, permitting simultaneous measurements of streamwise velocity and temperature, facilitating a comprehensive exploration of thermal boundary layer dynamics. To explore scenarios with different streamwise origins for the thermal boundary layer, the ratio of the momentum to the thermal boundary layer thickness ($\delta_\theta/\delta$) varies from 0.26 to 1.1 while maintaining a constant momentum boundary layer thickness. Statistical analysis reveals a strong correlation between velocity and temperature. The mean temperature and velocity profiles exhibit similar patterns when the thermal and momentum boundary layers have matched edges, while discrepancies arise with mismatched edges. Using a two-state TNTI-associated model, the mean temperature profiles are predicted. Near the edge of the thermal boundary layer for the case of $\delta_\theta\approx \delta$, one-sided temperature fluctuations are observed, while $u$-fluctuations remain two-sided, highlighting dissimilarity between $u$ and $\theta$. Furthermore, the one-sided bias phenomenon is evident even closer to the wall, as $\delta_\theta/\delta$ decreases, corroborated by the observed skewed temperature signals even within the logarithmic region. Concerning energy spectra, large scales exhibit a greater degree of (anti) correlation between temperature and velocity fluctuations than smaller scales. As the thermal boundary layer becomes thinner, coherence between $u$ and $\theta$ at larger scales appears to diminish. This observation can be interpreted through a simple model where large-scale motions that span the edge of the thermal boundary layer will produce increasingly one-sided temperature fluctuations while velocity fluctuations will maintain a comparatively Gaussian distribution. Two-point measurements with the application of two customised dual-wire sensors entail the estimation of the streamwise/wall-normal aspect ratio, $\mathcal{AR}$. Here, ${\mathcal{AR} = \lambda^\mathcal{H}_x/\mathcal{H}}$, where $\lambda^\mathcal{H}_x$ is the streamwise length of a wall coherent structure, and $\mathcal{H}$ is the wall-normal extent). The coherence results highlight the prominent role of wall coherent motions in both momentum and heat transport. Diminished coherence when $\delta_\theta/\delta \ll 1$ is attributed to temperature fluctuations produced by large-scale motions that extend vertically beyond the edge of the thermal boundary layer (e.g., $\lambda^+_x\gtrsim \mathcal{AR}\delta^+_\theta$). Temperature fluctuations associated with these structures and impacted by the one-sidedness effect bring about a diminished coherence, reinforcing the proposed model and challenging notions of Reynolds analogy for these scales.
-
ItemAir-Sea Interaction: Influence of Airflow Separation and Small-Scale Waves on the Drag over Wind-Waves( 2023-05)The study aims to deepen the understanding of turbulent flows above waves, particularly investigating the interaction of wave structure to the airflow separation and its influence on the drag above wind-waves. A series of experimental studies were performed in this research to examine the behaviour of airflow separation above wind-waves, evaluate the important wave components and spectrum in relation to the wave boundary layer, and investigate the role of small-scale ripples on wave turbulence. The drag coefficient Cd is an essential parameter in air-sea momentum transfer. However, its relationship with wind speed U10 has remained scattered and uncertain for over 30 years. Multiple factors, such as wave steepness, wave breaking, gustiness of the wind, non-linear wind-wave interactions, and wave directionality, can affect Cd, leading to deviations from the predictions of climate models. Furthermore, recent field observations by Powell et al. (2003) and Jarosz et al. (2007) have documented drag coefficient saturation and reduction behaviour during hurricanes. The literature suggests that deviations from predicted drag coefficients, especially at high wind speeds, may be related to airflow separation. The airflow separation in the leeward side of wave crests is often observed and may be linked to the occurrence of small-scale waves. Surface tension has been found to cause the formation of ripples at the leeward side of crests, and the interaction of these small-scale ripples with larger waves is important due to their impact on radiation stress and viscosity. The small-scale waves in the gravity-capillary regime are the main contributor to ocean roughness and affect air momentum transfer. However, a systematic investigation into the influence of gravity-capillary waves on the atmospheric boundary layer was lacking until the Direct Numerical Study (DNS) was conducted by Druzhinin et al. (2019). Therefore, this thesis aims to fill a gap in the current understanding of the role of gravity-capillary waves in the atmospheric boundary layer, particularly in laboratory settings where relatively limited research has been conducted on this topic. The experimental investigations are divided into three main parts: Particle Image Velocimetry (PIV) measurement of airflow above wind-waves, highly resolved wave surface measurement in both spatial and temporal domains, and systematic PIV measurement of solid wavy-wall in a wind tunnel. The PIV measurements in the wave flume employed a Large Field of View (LFOV) setup designed to capture large-scale turbulent motions associated with surface-wave topography, such as airflow separation, while characterizing mean velocity, surface drag, and Reynolds stresses over wind-waves. Additionally, the High Magnification Field of View (HMFOV) setup was focused on investigating airflow separation closely at the leeward side of the crests, as well as other potential parameters affecting airflow separation and drag. This thesis also evaluated the idea of wall similarity on rough-wall boundary layers from Castro (2007) to the flow characteristics above wind-waves, which can be seen as moving roughness. Next is an experimental study that uses a laser-based technique to measure the temporally resolved surface elevation of wind-generated gravity-capillary waves. The aim is to study the structural and physical properties of the wind-wave field by capturing the fine details of waves, especially gravity-capillary waves, at different wind speeds and wave scales. This knowledge is critical to accurately reconstruct wind-waves as solid roughness. Surface elevations were obtained using a dynamic threshold algorithm, and the spatial and temporal spectra were analyzed and compared to the linear wave theory. The study also investigated decomposed waves of different scales, particularly small-scale waves. The PIV experiment at the wave flume showed some degree of similarity between wind-waves and solid walls in the fully-rough regime, which is in line with the findings of Geva and Shemer (2022) over wind waves. In addition, small-scale waves occurring on the leeward side of the dominant waves were found to be closely associated with airflow separation. The results from the laser surface experiment provided valuable insights into the structure of wind-waves and gravity-capillary waves, leading to final PIV experiments on turbulence flow above a solid wall in wind tunnel studies. The practicality of wind tunnel testing with a solid wall allowed for a systematic investigation of the relationship between selected parameters, in this case, gravity-capillary waves/ripples and airflow separation to drag. The wind-waves roughness was reconstructed based on the wind-waves generated in the wave flume, resulting in two types of solid wavy-wall: one with the dominant wavelength alone and the other with the addition of gravity-capillary waves. Our study shows that the small-scale waves influence the separation characteristics, which leads to contradictory behaviour between the two solid wavy-wall. The small-scale ripples play a crucial role in maintaining a momentum deficit region as the Reynolds number increases. This leads to a decrease in the drag coefficient with increasing wind speed and may explain the drop in sea drag above the ocean during extreme wind conditions. These findings emphasize the significant impact of small-scale waves on drag above wind-waves and highlight the critical role of airflow separation in air-sea interaction at certain wave ages.
-
ItemNumerical investigations of the flow around a confined flat plate parallel to the flow direction( 2023-09)The two and three-dimensional flows around a confined flat plate that is oriented perpendicular to the flow is simulated with the spectral element code, Nek5000. Direct numerical simulations (DNS) of two-dimensional flows are carried out at low Reynolds numbers \textit{Re} (between 10$\leq$\textit{Re}$\leq$200) and blockage ratios $\beta$ between 0.1$\leq$$\beta$$\leq$0.9. Four distinct flow regimes are observed for the flow over a confined flat plate. The first is a steady symmetric wake where the vortices are evenly distributed along the centreline of the channel. The second is an unsteady symmetric regime which upon time-averaging the instantaneous flow field, the wake profile is symmetric. The third is an asymmetric steady wake where the flow is steady but the vortices in the wake are skewed towards one of the channel walls. A fourth flow regime defined as unsteady asymmetric wake is found for a flat plate at $\beta$ = 0.8 which was not reported in prior studies. %This flow regime has not been reported in prior studies on the confined flow over a confined cylinder. Investigation using linear stability analysis is also documented which confirms the trends reported in this thesis. Effects of the confinement on the wake of a flat plate on the hydrodynamic forces such as the mean drag coefficient (\textit{C\textsubscript{D}}) and Strouhal number (\textit{St}) have also been documented for the same range of $Re$ and $\beta$ and compared with the flow over a confined cylinder. An extension of these two-dimensional simulations is the investigation of the characteristics of the confined flow in the far wake. The far wake behind a confined flat plate is investigated between \textit{Re} = 100 and 150 and blockage ratios $\beta$ between 0.1$\leq$$\beta$$\leq$0.5. One of the parameters calculated is the decay rate of the wake. The data showed that as $\beta$ increases, there was a faster rate of decay of the wake. In addition, an increase in \textit{Re} accelerates the decay in the wake. The cross stream velocity (\textit{v\textsuperscript{$\prime$}/U\textsubscript{Max}}) spectra is also calculated to study the dominant frequencies of the wake and how the frequencies change with $\beta$ and \textit{Re}. It is reported that for $\beta$>0.1, the frequencies generated from the shedding of the wake changed from multiple to single frequency and remains a single frequency even when \textit{Re} is increased unlike in the $\beta$ = 0.1 case where the signal was more sensitive to \textit{Re}. Three-dimensional simulations is performed for a flat plate at \textit{Re} = 750 and compares the present study with results from \cite{narasimhamurthy_numerical_2009}. One parameter of interest is the mean streamwise velocity (\textit{$\overline{U}$/U\textsubscript{Max}}) along $x/D$ and $y/D$. The second parameter is the variation of the drag and lift coefficients (\textit{C$\textsubscript{D}$} and \textit{C$\textsubscript{L}$} ) against time. Interestingly, the mean streamwise velocity profiles along \textit{x/D} and \textit{y/D} show some discrepancies with published data from \cite{narasimhamurthy_numerical_2009}. It is shown that these discrepancies are due to the nature of these high \textit{Re} cases requiring longer averaging times. Data of \textit{C$\textsubscript{D}$} against time are plotted and shows variations in their magnitudes. \cite{najjar1998low} found that even for a modest \textit{Re} of 250, that these variations in the drag exist. They also reported that there were three different regimes of drag. The first is the high drag regime (regime H) where the fluctuations of \textit{C$\textsubscript{D}$} and therefore, the magnitude of drag is large. The second is a low drag regime (regime L) where these fluctuations are low and finally, the third is the transition regime that links regimes H and L. This transition regime shows series of increasing and decreasing drag. Data of the lift coefficient against time also show these fluctuations which correspond to the drag variations. Given these variations, the true averages of all relevant statistics must be calculated over the timescales that are much longer than the lowest frequencies. These three-dimensional simulations were extended to investigate different flow parameters for two \textit{Re} values which are 250 and 750 for a confined flat plate. The spacetime diagram of instantaneous crossflow velocity against non-dimensionalised time is plotted for three different spanwise lengths which were \textit{L\textsubscript{z}/D} = $\pi$, 2$\pi$ and 3$\pi$ for both \textit{Re}. For the span lengths of $\pi$ and 2$\pi$, there is flipping of the wake which meanders along the span. However, when the span is longer such as 3$\pi$, the wake is more stable over long timescales.
-
ItemThe influence of realistic roughness on turbulent boundary-layers( 2023-10)Up to 90% of the drag that affects engineering systems with wall-bounded turbulent flows results from the turbulence induced skin-friction. In nearly all practical scenarios, this drag is exacerbated due to the presence of surface roughness. Therefore the accurate prediction of this drag penalty is highly desirable. However, current models are limited as they neglect some common aspects of realistic roughness - viz. anisotropy, multi-scaled nature, and temporal variability. This thesis presents three experimental campaigns that address these shortcomings. The effects of anisotropy of roughness are examined by varying ESy while keeping ESx fixed, where ES is the effective slope (in streamwise, x, and spanwise, y, directions). Four cases are investigated – two isotropic cases with ESx = ESy = 0.24 and 0.13, and two anisotropic cases with ESx = 0.24 and ESy = 0.13 and vice-versa. The skin-friction curves, obtained with a drag balance, suggest that anisotropic cases (ESy/ESx < 1) exhibit a higher drag penalty than the isotropic cases (ESy/ESx = 1) at matched ESx. The Hama roughness function obtained from the mean velocity profiles, is seen to increase by 6 to 8% when ESy/ESx is reduced by a factor of nearly 2 across the Reynolds number range of the measurements. For all cases, there is evidence for Townsend’s (1976) outer-layer similarity hypothesis. However, the wall-normal extents of the enhanced dispersive velocities are seen to increase as ESy is decreased and ESx is held constant. This extent is reasonably captured by 0.5Ly, where Ly is the mean spanwise wavelength, as suggested by Chan et al. (2018). Consequently, models for drag prediction should be adapted to also consider anisotropy in the form of ESy. The importance of ESx for drag prediction is evident, yet for practically occurring multi-scale roughness, it can remain unbounded if all the scales of the topography are not resolved. Consequently, two questions are asked: (i) At what flow-defined length scale do the small-scale features affect the drag? (ii) Can ESx reliably predict the drag of two surfaces with matched ESx but different scale composition? To answer these questions, drag balance measurements are conducted with a set of multi-scaled rough surfaces, where only the contribution of the small scales of the topography is varied. The relative increase in the drag penalty is seen to scale with the viscous scaled height of the small-scales, and is appreciable only when their heights exceed 2–3 times the viscous length scale. However, even when the small-scale height is O(10) viscous units, the additional drag penalty seen is much lower (nearly 28%) than another case with matched ESx, but where the ESx results from larger scale features alone. These findings confirm that for multi-scaled surfaces, one should consider which scales contribute to their measure of ESx to avoid incorrect predictions of the drag penalty. Finally, the response of a TBL to a well-defined oscillating rough surface is examined. For the dimensionless frequencies of the roughness studied, the mean flow is seen to oscillate between the static smooth and rough limits. A quasi-steady state is seen at low frequencies, which diminishes as the dimensionless frequency increases. A transition front is identified from the phase-averaged statistics, which suggests that the response of the boundary-layer is due to the growth of internal boundary-layers (IBLs). The response is modelled based on an argument of effective fetch, which relates the phase of the oscillation to a static step change in roughness at an effective upstream position, which in turn permits existing models for IBL growth to be used. This argument is confirmed by comparison with static streamwise heterogeneous data. The modelled front with this argument shows a good match to the front identified from the experiments along with the broad changes in the inclination of the front with increasing dimensionless frequency. Thus, a case can be made that turbulent boundary-layers over oscillating roughness can be modelled as growing internal layers.
-
ItemLarge-Eddy Simulation of Premixed Flame Acoustics( 2023-08)Gas turbines are an important part of our power systems. In modern gas turbines employing lean premixed combustion to control pollutant emissions, combustion noise is an important consideration because it can trigger dangerous thermoacoustic instabilities leading to combustor damage. However, combustion noise is difficult to predict using numerical simulations. Instead, empirical approaches are applied to eliminate thermoacoustic instabilities in new combustor designs. To accelerate the design process of cleaner gas turbines and to reduce costs, efficient simulation tools are required for combustion noise. Large-eddy simulation (LES) is one such tool that can be used for analysing real combustion systems. However, the ability of LES to predict combustion noise has only been evaluated using experimental data, which does not provide high-resolution information about the sources of sound in the flame. This thesis presents an assessment of LES of premixed flame acoustics using the flame surface density (FSD) approach as the combustion model. Two direct numerical simulation (DNS) cases of turbulent, premixed, methane-air, round jet flames are used for both a priori and a posteriori validation. First, the DNS data is filtered using Gaussian filters having different sizes. The sound radiated to the far-field is calculated using Lighthill's acoustic analogy. Comparing the pressure data obtained using the unfiltered and filtered heat release rate distributions as the source terms shows that the far-field sound is largely unaffected by the filtering. Next, the FSD modelling approach is assessed using a priori analysis. Models for the two terms in the filtered flame front displacement (FFFD) term are considered: models for the surface-averaged, density-weighted flame displacement speed $\overline{(\rho S_d)}_s$ and for the flame wrinkling factor. The modelling of $\overline{(\rho S_d)}_s$ is found to have a major effect on the sound. Furthermore, it is found that the behaviour of the wrinkling factor and the flame displacement speed near flame annihilation events is not correctly reproduced by FSD modelling. As flame annihilation is an important source of sound in premixed flames, the ability to capture it accurately has implications for LES for combustion noise. The final part of the thesis presents an a posteriori analysis of the same flame configuration as the previous a priori study. The DNS code NTMIX-CHEMKIN is modified for LES employing FSD combustion modelling, central differencing schemes of different orders, and artificial damping. This study confirms that FSD modelling does not correctly capture flame annihilation events, resulting in major changes in the flame shape. The generated sound is significantly overestimated at high frequencies. A model for $\overline{(\rho S_d)}_s$ is proposed to improve the treatment of flame annihilation with FSD modelling. The important factors impacting the performance of LES are also identified and recommendations for further improvement are provided.
-
ItemNo Preview AvailableBlood Flow Dynamics in the Aortic Dissection( 2023-08)Aortic dissection is one of the catastrophic cardiovascular diseases that have high mortality. It refers to an intimal tear in the aortic wall that initiates the formation of a false lumen due to blood flow between the layers of the vessel wall. Decisions about medical management or surgical intervention for long-term dissections are complex and still evolving, depending largely on the individual patient’s condition. In addition to conventional clinical images, the incorporation of more comprehensive physiological data would benefit clinicians in the decision-making process. Recent advancements in four-dimensional phase-contrast magnetic resonance imaging and computational fluid dynamics are promising in providing detailed data on haemodynamic parameters in cardiovascular diseases, including those that are challenging to predict or measure safely in clinical settings. In this work, the robustness and precision of a respiratory-controlled k-space reordering four-dimensional phase-contrast magnetic resonance imaging sequence were evaluated. Imaging data and pressure measurements are used to inform the development of numerical models of dissected aortas. The influence of different inlet boundary conditions on the outcomes of our simulations has also been investigated. The present results indicate that phase-contrast magnetic resonance imaging is valuable for providing patient-specific flow data. The evaluated magnetic resonance imaging sequence is reproducible and accurate in in-vivo flow metrics measurement. Computational fluid dynamics simulations based on multiple imaging modalities hold substantial promise for identifying potential risk factors associated with disease development. To accurately represent physiological haemodynamic parameters in aortic dissection, appropriate inlet boundary conditions and MRI data should be chosen.
-
ItemA Molecular Simulation Study Of Lithium Ion Intercalation Between Bilayer Graphene On 6H-SIC (0001) Surface( 2023-09)Moire patterns are universal in multilayer van der Waals (vdW) materials due to weak interlayer interactions. In such materials, one layer can be easily strained, rotated, or twisted relative to other layers, forming the moire patterns. These moire patterns consist of topologically protected domains and domain walls (TDWs) with different stacking orders, exhibiting novel quantum properties such as topological channel, plasmon reflection, electron transport, and Aharonov-Bohm oscillations. Controlling such topological structures is essential for quantum applications. However, previous control methods, such as external electric field, mechanical rotation, or straining of layers, show disadvantages, including small sample size, poor mechanical operations, and limited dynamic controllability. Ion intercalation shows the potential to alter the stacking orders in graphite and bilayer graphene, making it a promising and scalable experimental technique to control the topological structures in multilayer vdW materials. However, there is limited knowledge about controlling topological structures by intercalation, primarily how intercalation is utilized to control stacking and how the moire patterns evolve during the intercalation process. Employing molecular simulations, my PhD research focuses on the topological structures and their spatio-temporal evolution upon Li+ ions intercalation in a representative bilayer vdW material: bilayer graphene on SiC (0001) substrate (G/B/SiC). In this system, one graphene layer (buffer layer) partially bound with SiC substrate is slightly strained, and the other is free-standing with equilibrium lattice constant. The mismatch strain between the two graphene layers leads to two moire patterns: zebraic patterns and triangular patterns. Combined with experimental observations from the state-of-the-art in situ low-energy electron microscope (LEEM), my thesis encompasses three aspects of work. Firstly, I investigate the configuration of two topological moire patterns in the bilayer graphene G/B/SiC system. It confirms that the zebraic pattern forms due to uniaxial strain, while the triangular pattern forms due to biaxial strain or a combination of biaxial strain and interlayer rotation in this system. Additionally, I show two novel phenomena, e.g., symmetry-broken domains in zebraic patterns and curved TDWs in triangular patterns. We demonstrate that the flexible buffer layer is the key to the symmetry-broken phenomenon, where vdW interaction is the driving force for symmetry-broken in zebraic patterns. We show the dependence of the occurrence of curved TDWs on the buffer layer flexibility in triangular patterns. We also illustrate the morphology of TDW networks transitioning from curved to straight triangles modulated by strain engineering. Second, I study the thermodynamics and dynamics of Li+ ion intercalation and transport in the G/B/SiC. It demonstrates that the adsorption of Li+ ion adsorption is stacking dependent. Moreover, the exceptionally high energy barrier in the domain surrounding AA could limit the diffusion of Li+ ions from the AA domain to other domains. Third, I explore the topological structure evolution in two moire patterns upon Li+ ion intercalation. It turns out that a top-layer-sliding mechanism generates the stacking shifting upon Li+ ion intercalation, leading to diverse TDW network topological structures with different stacking orders. Our in-depth study unravels detailed topological structure evolution in the two moire patterns during intercalation. The obtained knowledge lays grounds for understanding the intercalation dynamics in other vdW materials. It enables approaches to manipulate the topological structures in multilayer vdW heterostructure systems for future quantum devices.
-
ItemNumerical investigation of hemodynamics in stenotic vessels( 2023-08)Stenosis refers to any condition in which a vessel becomes abnormally narrow, so the flow through it is restricted. It is associated with an obstruction of a blood vessel lumen resulting in physiologically abnormal hemodynamics in the vessel and further progression of the obstruction. The presence of stenosis results in an acceleration of the flow past it which is accompanied by a noticeable drop in pressure, while the flow upstream and downstream of the stenosis is commonly disturbed due to rapid change in the geometry of the stenosed vessel. Stenotic flows can occur in various physiological contexts, including atherosclerosis, vascular diseases, inflammation, congenital heart defects and etc. Understanding the hemodynamics of stenotic flows is important for diagnosing and monitoring various cardiovascular diseases, as well as for designing medical devices and improving treatment strategies, encompassing pharmaceutical interventions, stent procedures, and surgical interventions. The objective of this thesis is to conduct a series of numerical experiments on stenotic flows under different geometrical and flow conditions using the direct numerical simulation method. In the thesis, the hemodynamics are numerically investigated in surrogate models that mimic vessels and their connections in the presence of stenosis. The thesis explores different origins of stenosis formation, including atherosclerosis, inflammation, subsequent scar tissue formation, and thrombosis. As an example of stenosis arising from the buildup of atherosclerotic deposits on the inner walls of arteries, the thesis considers a surrogate model that mimics the left anterior descending artery (LAD) affected by stenosis. To investigate hemodynamics in the presence of stenosis caused by scar tissue formation, stenoses at anastomoses between the artificial conduit and vessels in the surrogate model of the total cavopulmonary connection (TCPC) are examined. Furthermore, the thesis focuses on platelet activation triggered by the extensional flow in a channel with a stenosis of a hyperbolic shape. The surrogate LAD model yielded results that included various hemodynamic indices. These indices have the potential to predict the advancement of coronary artery disease. The results showed that alterations in the axisymmetric configuration of the stenosis significantly affect both the flow within the stenosis and downstream, emphasizing the importance of considering the asymmetry of the stenosis in diagnostics to enhance accuracy. The findings of the surrogate TCPC model revealed that the hemodynamics within the model are highly responsive to alterations in conduit stenosis geometry and relative lung perfusion. The most physiologically detrimental scenario, characterized by reduced energy efficiency and heightened wall shear stress, is observed in a TCPC with diffuse conduit stenosis and highly uneven lung perfusion. Finally, a noticeable response of platelet activation to the increased extensional and compressional forces in the flow of a channel with a stenosis of a hyperbolic shape was observed at the entrance and exit of the stenosis. In summary, the thesis delves into the hemodynamic environment of the stenotic flows and gains insights into the flow physics associated with physiologically abnormal hemodynamic metrics. Such insights are crucial for assessing the clinical risks and implications faced by cardiovascular disease patients.
-
ItemCoverage Control for Robotic Networks in Dynamic Environments( 2023-06)The coverage control problem involves spatially disseminating a network of mobile agents using distributed control laws, or coverage controllers, over a desired region to locally minimise an associated cost function. Certain nonlinear controllers derived using Lyapunov theory feature desirable properties, such as distributed communication and computation. Previous approaches to the coverage problem have addressed both static and dynamic environments through the choice of density function and agent sensor models. However, stability guarantees under additional dynamics induced by these choices are restricted by significant technical assumptions that simplify the underlying proofs at the expense of limited applicability. In this work, a generalised coverage controller is presented that guarantees practical stability under relaxed technical assumptions. The algorithm, and its convergence, is illustrated via simulation examples. Asymptotic stability of classical Voronoi-based coverage due to Cortes et al.’s coverage controller and its variations have been studied, however rates of convergence are absent. Although convergence guarantees are provided for many of the variations on the coverage control problem, they rarely demonstrate exponential convergence or rely on conjecture to do so. Local exponential convergence opens new avenues of research within coverage control, such as robust controller design or combination with online estimation. This work provides local exponential convergence properties for a multi-agent network using a coverage controller, as well as the existence of a ball around local equilibria for which this holds under the listed assumptions. Mobile robots using coverage controllers will be subject to various disturbances, such as uncertainty, errors and delays. This work shows that a variation of Cortes et al.’s coverage controller also features robust stability properties, specifically input-to-output stability, under the same set of assumptions. Conservative bounds are used to provide theoretical guarantees on stability, and simulations are used to verify the results and highlight practical performance of the controller. Additionally, this work aims to validate this stability property through experimentation on a hardware platform with a variety of different disturbances. These include disturbances that are present in real-world systems, such as estimation errors and time delays, as well as additional dynamics introduced through variations to the coverage control problem of interest.