Search

An immiscible, binary fluid lattice Boltzmann model is described and its associated equations of motion are given. It is seen that the lattice Boltzmann scheme is totally isotropic and that it does not suffer from the problems of noisy results and a lack of Galilean invariance which plagued its predecessor: the lattice gas model. The incorporation of a body force into the lattice Boltzmann technique is considered. A method which introduces the body force directly into the governing equation is proposed and is seen to have the desired effect without destroying the Galilean invariance of the original model and without introducing any dependency on the grid orientation. The immiscible, binary fluid models, with the body force incorporated, are applied to simulate interfacial waves between the two fluids. The model parameters allow the interface thickness, the fluid viscosity, the gravitational strength and the relative density of the two fluids to be varied. The wavelength of the wave can also be set during the wave initialisation. Standing waves are simulated for a wide range of the variable parameters and progressive waves for a subset of the parameters. The results are seen to compare well with linear wave theory. When compared with available experimental results the behaviour is seen to be similar.

Hypotheses stipulating that adverse exothermic reduction-oxidation (redox) reactions between aluminium-alloy sheets, used as facings for aluminium composite material (ACM) rainscreen panels, and water vapour (2Al + 3H₂O → Al₂O₃ + 3H₂), contribute to the severity and rate of fire spread during high-rise façade fire events, including the fatal Grenfell Tower incident on 14 June 2017 in London, United Kingdom (U.K.), have circulated amongst both industry and academia in the recent years since the disaster. With a lack of direct research and quantitative evidence available to either conclusively support or refute these claims, an ad-hoc investigation was undertaken providing insight into the significance of such reactions, and the likelihood of occurrence, in the context of a façade fire scenario comprising ACMs. Two common ACM panel variants (a polyolefin blend-based core product, PE, and a fire-retardant version, FR) were characterised, analysed to derive material-specific kinetics, and subjected to a series of bench-scale fire performance testing, involving the use of a cone calorimeter with an in-situ application of liquid-mist and water droplet sprays. Formulated inverse stoichiometric analyses of the experimental smoke yields, with respect to O₂ consumption measurements, were used to identify anomalies in fuel composition following H₂O application, given that H2 combustion does not produce CO₂, in contrast to hydrocarbon fuel (CxHy) content within combustible ACM core products. This was achieved using controlled mist spray experimentation, specifically designed for the relevant research objectives. Contemporary numerical modeling techniques adopting computational fluid dynamics were also applied, providing additional insight through a constructed inverse modeling framework, and innovative coupled pyrolysis-combustion simulations, with two-way verification-validation of experimental-simulation data. From the work undertaken, the evidence compiled indicates insufficient quantities of H2 are produced by Al-H₂O reactions at the conditions attained by ACMs during façade fire events. Consequently, the findings strongly oppose hypotheses of Al-H₂O reactivity contributing to the combustion of ACM façade components and reaffirms the use of water as an appropriate extinguishing agent on aluminium-based façade components; No significant risk to fire spread or severity is anticipated through aluminium-water reactivity. ACM samples with ~30 wt. % polymer content (FR) demonstrated an average 73% reduction in pHRR, and 65% reduction in THR, in comparison to samples containing ~70 wt. % polymer (PE). ACM cladding panels with a low polymer content (≤10 wt. %) that achieve EN A2-s1,d0 classification are considered safe for use in high-rise façade assemblies.

Hydrocarbon production from unconventional reservoirs has become more common in the past decade, and there are increasing demands to understand and analyse the geochemical, petrophysical, and geomechanical properties of these resources. The recent developments in intelligent techniques, such as Machine Leaning techniques, Fuzzy Logic, and optimization algorithms, provided the scientists from different disciplines with a robust tool to analyse their data. These methods, similarly, can be used in the upstream industry to enhance the efficiency of field development. In this study, the Nene Marine field, as a high potential candidate for future developments in West Africa, will be analysed to find its best productive zones (sweet spots). Tight sandstones of Djeno formation, which are interbedded with shale layers, are the main productive zones in this field. Initially, the routine methods of reservoir evaluation will be applied to classify the reservoir and detect the sweet spot zones. For this purpose, various reservoir properties, including petrophysical, geochemical, and geomechanical parameters, will be calculated from the available data. To validate the results, the identified sweet spots will be compared with the available mobility data from Djeno formation. Afterward, a 3D model of the reservoir productive zones will be constructed. Development of tight sand reservoirs, in addition to a comprehensive study on reservoir's properties, demands for a precise stimulation plan, in order to make the project economical. Therefore, a comprehensive Hydraulic Fracturing analysis for the Nene field will be proposed. To complement this, advanced artificial intelligent techniques, including supervised and unsupervised ML techniques (Fuzzy C-Means (FCM), Hierarchical clustering, K-Means, Logistic regression, K-Nearest Neighbors (KNN), Random Forest, and Boosting), along with Deep Learning (DL) methods will be employed to identify the sweet spots. In addition, a novel method (Quick Analyser) of reservoir classification, which works based on the cut-off values or optimum values, will be presented. Application of this method will be considered for the Nene field and will be demonstrated for another field from Oman. After examination of the accuracy of the proposed methods for sweet spot identification, all of the methods will be compared and the best method will be determined. Another application of intelligent techniques in this study is the estimation of shear velocity wave (an important parameter for geomechanical analysis), which is performed by using an integration of fuzzy inference systems and optimization algorithms. Finally, the application of ML techniques (unsupervised learning) in reservoir classification, by using seismic SEGY data, for a North American oilfield will be illustrated. The results of this study show the huge potential of advanced AI techniques in petroleum industry.

The world is embarking into a fossil fuel free environment in which the predominant sources will be renewable and sustainable energies. Since renewable energy is usually associated with their intermittent nature, it is important to always have a backup source of energy in times of need and demand. For this, energy storage is an essential component of any renewable energy system. In times of high demand, where the sun is not shining and the wind is not blowing, the existence of a backup energy supply with easy and fast access is necessary. Thus, Thermal Energy Storage (TES) is one of the proposed solutions to aid in this mismatch in consumption and demand caused by the nature of the green technologies (Ali et al., 2020). TES can be an accompaniment of systems such as a Joule-Brayton cycle, Rankine cycle, a heat pump or even solar collectors. Other applications include systems which focus on heat waste recovery, steam generation for turbines, domestic uses and household heating. Energy from TES can be stored daily, monthly or seasonally. This determines the type and size of the system and whether it is a smaller scale domestic or large scale industrial project (Fang et al., 2017). In all cases, however, its main potential to serve as a tool that saves energy and mitigates climate change, keeping CO₂ emissions minimized, and helping countries meet their established environmental goals. Low grade TES is in increasing importance and therefore remains the focus of this thesis.

The phase and volumetric behaviour of reservoir fluid properties, referred to as pressure-volumetemperature (PVT) properties, involve the thermodynamic studies of the fluid with respect to pressure, temperature and its volumetric compositions. PVT properties are usually determined by laboratory experiments performed on the actual samples of the reservoir fluid. Failing that, these fluid properties have been evaluated by some other methods such as equations of state, empirical correlations and recently, machine learning models. Machine learning is basically the prediction of the future with, (supervised learning), or without, (unsupervised learning), prior knowledge of the past. A common problem for the standalone machine learning technique is local minimum. In view of this, ensemble systems and hybrid techniques have been developed successfully for improvement in different fields. This work introduces two different ensemble methods based on support vector regression and regression trees where both ensemble approaches utilise a novel concept tagged "Tying Ranking" in selection of the base models. Also, a hybrid system for reservoir fluid characterisation with a novel way of grouping petroleum fluid properties using intelligent method was developed. The hybrid system uses K-Means clustering for the intuitive grouping along with functional networks for the prediction. The performance and generalisation of the developed models are compared against their standalone and selected empirical models using some statistical measures which are commonly used for performance evaluation in the petroleum industry. In the first category of experimentation, the impact and effect of training the machine learning models with more diverse and bigger data set is shown. Effects of using different functional forms to predict dead oil, saturated and undersaturated viscosity are also explored. In addition, impacts of different statistical measures on the predicted outputs and wrong interpretations of results in the literature are examined. The main statistical measures that are used for comparison are root mean squared errors, average absolute percentage relative error and maximum absolute percentage relative error. For each of the reservoir fluid properties considered in this work, at least one or more of the developed machine learning models have better overall and average performance than all the compared correlations in each category. The superiority of the three developed machine learning models is visible in the trend analysis as they show less deviations in results compared to the empirical correlations and their standalone methods in most cases for all the considered reservoir fluid properties.

The transition in the energy market, from large fossil fuel consumption to the broad diffusion of renewable energies, involves an intermediate phase where more efficient techniques are developed for the existing power generation technologies. Among new combustion techniques, MILD (Moderate or Intense Low-oxygen Dilution) combustion is particularly attractive because of its potential characteristic to enhance thermal efficiency and reduce emissions like nitrogen oxides (NOx). The successful application of MILD combustion requires a significant entrainment of hot combustion products into the fuel and/or oxidizer stream(s), yielding an increase of the reacting mixture temperature over its auto-ignition value. Such intense dilution causes a reduction in peak temperature levels, with a consequent reduction of NOx emissions, and a homogeneous temperature field followed by enhanced flame stability. Also the overall thermal efficiency is improved because of the recuperated heat. The relative ease of obtaining reactant dilution in a full scale burner makes the MILD combustion regime interesting also from a technological point of view. Despite some interesting applications of MILD technique in industrial cases, its broad adoption is prevented by gaps in the knowledge of this combustion regime. Particularly, the development of simple and reliable numerical models is required to allow testing of full scale industrial burners with reasonable computational expense. The present dissertation is focused on how the oxidiser temperature, oxidiser concentration and fuel concentration affect the complex interaction among molecular transport, chemical kinetics and turbulence that leads to self-ignition in MILD combustion. The diffusion-chemistry contribution to ignition is investigated by means of a one dimensional (1D) zero velocity Direct Numerical Simulations (DNS) of two mixing layers representing a cold fuel mixture and a hot diluted oxidiser. Different oxidiser mixtures as well as different fuel blends are considered. Each case studied showed a different ignition behaviour. An in deep investigation of physical and chemical changes observed for each case along the ignition period is provided. A temporal and a spatial scaling methods are proposed to account for ignition behaviour differences and compare cases. The comparison revealed different aspects of the self-ignition process. The differential diffusion effect plays an important role in the early stages of ignition, for cases presenting methane/hydrogen fuel mixture. If high is the level of hydrogen in the fuel blend, major stages of methane (CH4) consumption pathway,from the CH4 dehydrogenation to the carbon dioxide (CO2) release, are significantly affected by hydrogen (H2) chemistry. In the latest stages of ignition, the methane pathway is also affected by the drop in oxygen level. The influence of turbulence on the diffusion-chemistry interaction is studied by means of three-dimensional (3D) Direct Numerical Simulations modelling a methane/hydrogen circular jet mixing with a diluted oxidiser co-flow. The effects of different fuel and oxidiser blends is also considered in the 3D study. In cases where large is the H2 presence in the fuel jet, the presence of turbulent mixing has a minimal effects on early stages of self-ignition, where instead differential diffusion still plays a major role. As turbulence develops, more marked difference between 1D and 3D studies are observed. The role of turbulent mixing dominates over chemistry where the fuel blend includes a low amount hydrogen. For this configuration the temperature increment is strongly limited compared to corresponding 1D study. The outcome of this study is expected to be of use to other researches in MILD combustion, particularly those adopting existing RANS and LES models to MILD combustion cases.

A model for stenosis development in the carotid artery, employing the lattice Boltzmann method, is presented and its suitability is investigated and demonstrated. The development of a stenosis is modelled based on the blood flow; and the subsequent changes in the flow are examined. The model is applied to investigate the role of different haemodynamic markers on stenosis development and also the importance of modelling the non-Newtonian nature of the blood. A stenosis growth model based on the time-averaged velocity magnitude at the wall is considered. The wall position where this marker has a minimum value is determined and the stenosis is allowed to develop at this point. The extent to which the stenosis develops is controlled by two parameters which are introduced. Simulations are then run based on the new geometry and a new position selected for stenosis development. The stenosis developed in this way was seen to be compatible with observations from the literature. Simulations of stenosis development are presented to investigate the effect of the introduced model parameters to determine suitable ranges for their application. A range of parameters are determined over which the stenosis develops in an independent manner. These parameters relate the extent to which the stenosis can grow in each development phase with no physical significance. It is important to find a variable which will describe the evolution in a manner which is independent of the model parameters. As well as the time-averaged velocity magnitude, a selection of alternative markers were applied to the model. The results show that a number of markers involving near-wall velocity, wall shear stress, residential time, stagnation index and second invariant of the strain rate tensor, all resulted in a realistic stenosis. The oscillatory shear index, which gives a measure of the oscillatory nature of the shear, was found to not be a suitable marker, unless it was combined with the wall shear stress in the form of a ratio. The effect of the non-Newtonian nature of blood on the stenosis development model is also considered. Here the non-Newtonian simulation showed noticeable differences compared to Newtonian; however both produced a realistic stenosis.