Numerical studies of reacting and non-reacting underexpanded sonic jets

The objective of this work was to model underexpanded turbulent sonic jets and the lifted diffusion flames which result should ignition occur. This relates to the accidental release of combustible gases from high pressure pipelines. A pressure-based CFD methodology has been employed, incorporating extensions to handle high speed, shock containing flows. The methodology extended previous studies of high speed flow employing pressure based methods to unlimited Mach Numbers. This has been achieved by developing a Mach number dependent differencing scheme for the pressure correction equation. This allowed high pressure ratio jets containing strong normal shocks to be computed. Numerical accuracy was improved by applying a second order accurate total variation diminishing scheme to the convective differencing. A standard two-equation turbulence model was used, with an optional compressibility correction. Predictions are presented over a large range of pressure ratios and extensive comparison to the available experimental data showed the correct shock cell wavelength, but a too rapid decay of the shock cell structure. The compressibility correction had no effect on the shock cell decay, but increased the potential core length, improving the agreement with the experimental data. Far field velocities and mixture fraction profiles also show a good agreement with experimental data when employing the compressibility correction. Following ignition, the resulting underexpanded jet diffusion flames are lifted from the nozzle rim by several diameters due to flame quenching processes. A quenching mechanism based on the strain rate of the largest eddies and applied either in a local manner or using a lift-off threshold was initially calibrated and validated for subsonic lifted diffusion flames, showing excellent agreement to experimental measurements. The mechanism was subsequently applied to underexpanded jet diffusion flames and showed good agreement to experimental data. Predictions of the lift-off height based on cold flow computations showed a reduced accuracy. The eddy dissipation concept, fast chemical reacting system and presumed β-pdf combustion models were applied and compared using the two quenching mechanisms. The various models predicted significantly different temperature fields close to the nozzle, but were similar further downstream. More experimental data is required in order to determine the accuracy of the models. A pseudo nozzle approximation has also been developed which replaced underexpanded jets with perfectly expanded supersonic jets and removed the need to compute the shock containing region. The approximation was successfully validated for both reacting and non-reacting underexpanded jets.