The work presented in this thesis explores the feasibility of the Multiple Mapping Conditioning (MMC) approach and its closures for real (laboratory) flames. Three different configurations with relatively high Reynolds numbers but without considerable degree of extinction and re-ignition are investigated, and results are compared against experimental measurements of mixing and reactive scalar fields and other commonly used models. MMC combines the probability density function (PDF) approach and the conditioning methods via the application of a generalised mapping function to a prescribed reference space. Stochastic and deterministic formulations of MMC exist. Both formulations have been explored here for the case of one dimensional Gaussian reference space that is associated with the evolution of mixture fraction. The chemically reactive species are implicitly conditioned on mixture fraction, and their fluctuations around the conditional means are neglected for the deterministic approach and modelled for the stochastic approach. Regarding the velocity field evolution, the Reynolds Averaged Navier-Stokes equations are solved with a k-[epsilon]turbulence model. In the deterministic context, this work evaluates the ability of MMC to provide accurate and consistent closures for the mixture fraction PDF and the conditional scalar dissipation which do not rely on presumed shape functions for the PDF such as the commonly used [Beta]-PDF. Computed probability distributions agree well with measurements, and a detailed comparison of the modelled conditional and mean scalar dissipation with experimental data and conventional closures demonstrate MMC’s potential. Predictions of reactive species and temperature are in good agreement with experimental data and similar in quality to singly-conditioned, first-order CMC predictions. MMC therefore provides an attractive -since consistent- alternative approach for the modelling of scalar mixing in turbulent reacting flows. In the stochastic context the evolution of the reference space is described by a Markov process that is coupled with a full PDF method for joint scalar evolution. A modified IECM model is applied for the modelling of the mixing operator where the particles mix with their means conditioned on the reference space. The formulation of the closure leads to localness of mixing in the mixture fraction space and consequently localness is expected to be improved in the composition space. Focus is given on the accurate prediction of scattering around the conditional means. Results demonstrate the potential of the method, however some discrepancies are noted in the predictions that can probably be associated with the chemical mechanism and the uncertainties associated with the choice of the minor dissipation time.

No description available.

No description available.

No description available.

No description available.

No description available.

No description available.

No description available.

No description available.

Data centres are developing at a rapid pace with the continued increase in digital demands. Data centre cooling and energy efficiency is a growing topic of interest that requires new engineering solutions. To achieve both better cooling and higher efficiency, liquid-cooled computer systems are being considered as one of the best solutions. Total liquid cooled computers are not new, but with the power densities required for supercomputers have seen resurgence in liquid cooling, in particular solutions that do not require the use of air as a cooling medium. Recently the industry has developed an advanced fully immersed liquid-cooled data centre solution to fulfil this purpose. The core technology of the design is a liquid-cooled computer node (first cooling stage), which relies on density-driven, natural convection that has challenging engineering requirements. This thesis looks at the density-driven, natural convection from a different angle by simplifying the Navier-Stokes equations and Convection-Diffusion equation leading to the development of a Constant Thermal Gradient (CTG) model to solve the natural convection flow analytically. The CTG model yields algebraic solutions for velocity and temperature profiles, thereby it is able to give the flow characteristic length (l*) and indicate the boundary layer thickness directly. The development and usage of the CTG model is the academic achievement in this thesis, and it provides a clearer understanding of natural convection mechanism. This thesis also uses CFD simulation (ANSYS CFX) and laboratory experiment to analyse the heat transfer performance of the liquid-cooled system. A group of CFD simulations of a cavity convection problem has been carried out to find the appropriate approximation factor for the CTG model, hence completing the CTG model and make it ready for further analysis. A full scale CFD simulation has also been carried out to analyse the first cooling stage of the system for a given condition, and a real computer system has also been tested under the same condition. Then a three-step research work-flow has been developed to do heat transfer analysis on a natural convection based liquid-cooled system: CTG model, CFD simulation and experimental test. This thermal analysis work flow provides a knowledge base for further improvement in cooling design of the system, and this is the engineering achievement of this thesis. In order to see the thermal advantages of the fully-immersed liquid-cooled system, other intense real-world tests on the liquid-cooled system have been carried out. One of which is a benchmark test between an advanced back-door water cooled system and a fully-immersed liquid-cooled system; and such benchmark proves the thermal benefit of the fully liquid-cooled solution. The other benchmark is a series of real-world tests on a fully immersed liquid-cooled system which aim to achieve the ASHRAE W5 standard, and it proves the practicality of the liquid-cooled solution. The benchmark test in this thesis was published in the Semi-Therm conference.