This thesis presents a combined experimental and theoretical study of the combustion aerodynamics and pollutant formation in confined swirling flames. The fuel used in this study was natural gas. In the experimental part of the work, two fuel injection modes are examined as alternatives to the conventional central axial fuel injection mode. These alternatives are (a) introducing the fuel around the periphery of the swirled air jet, and (b) injection radially outwards from the central axis, across the entering swirled air flow. The measurements were performed in a semi-industrial size furnace with a movable-block swirl generator. Four swirl settings were examined, covering swirl number range of 0 to 2.25. The flow patterns (as defined by three time-averaged velocity components and static pressure), combustion patterns (as defined by temperature and species concentrations) and pollutant formation (CO and NOx) were investigated for these two alternative injection modes as well as for the conventional central axial mode to assess the merits of the three systems. The formation of NOx has been studied in greater detail in these three systems. For the flow and combustion patterns, measured along the furnace, the main input variable was swirl, while for the pollutants, measured in the stack, the main input variables were fuel equivalence ratio and swirl. The investigations showed that with the radial fuel injection system, a stable flame was achieved without swirl, while for the peripheral and central axial injection systems, a minimum swirl number of 0.8 was required to establish a stable flame. By introducing some of the combustion air radially outward through a central gun with the peripheral fuel injection system, a stable flame was achieved without swirl. With the central radial and peripheral fuel injection modes, complete combustion can be guaranteed with 5% excess air, while for the central axial fuel injection at least 10% excess air was required to achieve complete combustion. The results of flow and combustion patterns demonstrate that the highest rates of mixing, combustion efficiency and heat transfer from the flame were achieved with the peripheral fuel injection. Increasing the degree of swirl was found to improve these characteristics by producing a more uniform and intense flame. The measurements of NOx at lean and rich conditions showed that this system offers wider scope for NOx reduction through lean combustion and staged combustion (ie both air and fuel staging). The radial fuel injection has also produced much improved mixing and combustion efficiency compared with the central axial fuel injection. However, with this system and with the central axial injection, only air staging can be used to reduce NOx formation. In all fuel injection modes, the strong dependence of NOx generation on flame temperature confirms that its principal formation is by the thermal mechanism. Locally, the effect of swirl on NOx formation was significant, however, its effect on the overall values was small. In the theoretical part of the work, predictions of the overall NOx formation are made using a well-stirred reactor model based on the extended Zeldovich mechanism. The model takes account of the fluctuations of the concentrations of fuel and oxidant in NOx reaction zone. A stochastic analysis has been introduced by the author to calculate the effect of these fluctuations on the NOx formation rate. The results of these predictions compare satisfactorily with the experimental measurements for the three methods of fuel injection. As part of the validation process of existing computational fluid dynamics (CFD) codes, an assessment is made of the ability of a CFD code to model swirling flames. The peripheral fuel injection mode applied to natural gas swirling flame is a novel test case (Beltagui and Maccallum (1988)). The predictions were made using the PHOENICS code, with turbulence and combustion represented by the k-e and the 'eddy break-up' models respectively. The main changes in the combustion patterns caused by switching from central axial to peripheral fuel injection were qualitatively well predicted. For the peripheral fuel injection system, the predicted flow patterns were in reasonable agreement with those previously measured. The quantitative agreement for combustion patterns, however, was good for the non-swirled flow only. This is attributed to the simplified turbulence and combustion models used.

No description available.

No description available.

No description available.

No description available.

No description available.

No description available.

No description available.

Alcohol fuels are expected to become an economic/strategic alternative to oil over the next decade as oil reserves are depleted and countries seek to become more energy self-sufficient. Methanol, produced from natural gas deposits, and ethanol, produced locally by distillation of biomass, offer easily transportable alternatives. The use of a wider range of fuels in spark-ignition engines and the quest for fuel economy whilst meeting exhaust emissions legislation are important issues in engine design. The performance of current and proposed combustion chamber designs needs to be assessed with lean mixtures of both conventional and alternative fuels. The parameters defining combustion chamber performance, initial flame development and cycle-to-cycle variations in combustion may be readily determined using computer in-cylinder combustion models in a diagnostic manner to reduce experimentally acquired cylinder pressure data. This thesis develops and applies two analysis techniques to the study of the combustion of methanol in the lean burning regime with experimental results from three engines. The pressure increment technique, in which the pressure rise owing to combustion at constant volume is computed, is suitable for use directly on microcomputer systems. The two-zone equilibrium theory model, in which the mass burnt to give the measured pressure rise is evaluated, provides a more comprehensive analysis but is demanding in computer power. Higher burning rates were achieved using highly turbulent combustion chambers with methanol and equivalence ratios could be leaned to about 0.8 before cycle-to-cycle variations in combustion limited stable operation. The results obtained indicated the significant phases of initial flame development, the influence of early flame development on subsequent burning rates, and the influence of differing chamber geometries on performance. The combustion process was modelled for use in parametric studies of engine performance based on empirical data.

A brief introduction to the field of pyrotechnic time delays is given, followed by a survey of the published work in the field up to the present time. A description is provided of the experimental techniques, equipment and materials used. The energetics and mechanism of reaction in the boron/lead monoxide and boron/stannic oxide systems have been investigated by comparison with the more familiar, analogous silicon fuelled systems; the emphasis being placed on the boron/lead monoxide system. The heats of reaction of the four fuel/oxidant systems have been measured by the use of bomb calorimetry, the results being discussed in relation to the probable reactions taking place in each system. Conclusions have been drawn as to the nature of the reactions which provide heat outputs in excess of the theoretical values in the boron fuelled systems. The propagation rates of the four fuel/oxidant systems have been measured and the relationship between heat output and propagation rate in each system is discussed. An explanation is proposed for the apparently anomalous behaviour of the boron fuelled systems. The response of the propagation rates in each system to varying consolidation pressure has been investigated. An explanation is proposed for the difference in behaviour displayed by the boron and silicon fuelled systems. A method of non-contact temperature measurement using a recording infra-red brightness pyrometer has been investigated. The reaction temperatures of the four fuel/oxidant systems have been measured and their relationships to the response of the propagation rates of each system to consolidation pressure variations are discussed. The boron/lead monoxide reaction in loose powder mixes has been investigated using differential scanning calorimetry. Hot stage microscopy has been used to investigate boron/lead monoxide reactions in loose powder mixes and between consolidated pellets of the individual reactants. The thermal analysis results obtained demonstrate the effect of the boron oxidation product, B 2 0 3 , on the reaction rate and the role which it plays in the reaction mechanism in the boron/lead monoxide system. A reaction mechanism is proposed for the non-propagative reaction between boron and lead monoxide. The extraction and kinetic analysis of data from the differential scanning calorimetry investigation of the boron/lead monoxide system is discussed. It is demonstrated that the analysis of the dynamic data is virtually impossible due to the complexity of the peak patterns obtained.