Soot formation in turbulent vaporised kerosine/air jet flames at elevated pressure
The objective of this thesis is to develop and validate a model of soot formation which is capable of being applied to a computational fluid dynamic (CFD) simulation of gas turbine combustion. The work follows previous research by Moss and Co-workers (Moss et al.1987, Syed 1990, Stewart et al.1991) The concept of the study is to generate a detailed set of experimental data in turbulent flames of kerosine in which the complicating factors of gas turbine combustion - that is 3D geometry and droplet combustion - are removed. This allows more confidence in the computational simulation of the flames and therefore more insight into the soot formation process. There are two components to the work: the experimental and theoretical studies. The first involves the compilation of an experimental dataset of key variables in ethylene and vaporised kerosine jet flames at elevated pressure, the second with the simulation of two of the experimentally studied flames using CFD methods. The main achievement of the study is the generation of a formidable and detailed experimental database for flames at a variety of pressures and conditions. The unexpected finding is the extremely large conversion of carbon to soot found in the flames even at low pressure. This results in high radiant heat losses and measurement difficulties. From the data, it is possible to assess the pressure dependence of soot growth in kerosine flames. Although, at the higher pressures, high soot levels created uncertainties in the measurements, in absolute terms growth rate is shown to be independent of pressure up to 6atm pressure. Above this it increases significantly. The soot model of Moss et al.1988 - originally developed in laminar e~hylene flames - was shown to give excellent agreement in turbulent situations. However, owing to the large radiant heat loss and soot levels, its application to the kerosine flames was more problematic since the assumptions that soot is a perturbation to the gaseous field and that temperature may be accurately described by a single perturbed flamelet were no longer valid. Further models to deal with such situations are proposed and tested. Aside from the obvious relevance of this study to the field of gas turbine combustion, the large radiant heat loss and high soot levels observed in the flames studied here imply a further significance for the study of fire hazards. That a laboratory scale flame maybe made to behave in a similar manner to a much larger pool fire flame is a very useful finding.