Computational and Experimental Study of Two Phase Flow in Pressure Swirl Atomizer

Atomizers are used in many engineering applications including spray combustion in furnaces, diesel engines, direct injection petrol engines and gas turbine engines. They are also commonly used in applying agricultural chemicals to crops, paint spraying, food processing and cooling of nuclear cores. Pressure swirl atomizers occupy a special position amongst other atomizers because they differ in quality of atomization, simplicity of construction, reliability and low pumping power requirements. Turbulent mixing of the liquid and gas in these atomizers is indispensable consideration in the process of atomization. This thesis presents a recent Eulerian modelling of two-phase flow in a pressure swirl atomizer using Computational Fluid Dynamics (CFD) STAR-CD code and assesses its capabilities and validation. In this novel Ʃ-Y_liq atomisation model, an Eulerian description is applied to solve the two-phase flow assuming both liquid and gas phases as a single continuum with high-density variation at large Reynolds and Weber numbers. The transport equations for the liquid mass fraction and interfacial surface density as well as the average density of the liquid and gas phases are modelled, liquid dispersion correctly captured and their numerical results presented. The results also show atomization characteristics such as droplet velocity and predicted droplet Sauter Mean Diameter (SMD) with reasonable order-of-magnitudes. The predictions show good agreement with experimental results obtained from a laser-diffraction-based drop size analyser (Malvern Spraytec). Different RANS turbulence models are evaluated in order to achieve the best configuration in comparison with experimental measurements and the standard k-ε turbulence model has shown the best performance. Parametric studies were conducted to analyse the influence of the liquid viscosity, surface tension, liquid and gas velocities, liquid and gas densities and pressure on the spray droplet SMD at different locations on the spray centre line and radial distances from the symmetry line of the atomizer. A combination of CFD modeling and the statistical Design of Experiments (DoE) technique known as modified Latin Hypercube Designs (LHD) is applied in order to improve SMD predictions from Ʃ-Y_liq atomisation model. With 4-factor DoE, eighty-seven (87) cases were simulated with the variations and combinations in the design variables such as liquid viscosity 0.31 to 200 mPa.s, surface tension 20 to 75 mN/m, nozzle exit orifice diameter 1.5 to 3.5 mm and liquid velocity from 1 to 6 m/s. The results for the SMD at the axial distances along the spray centreline were obtained. Combinatorial optimization was performed to identify and obtain the optimal nozzle exit orifice diameters, operating conditions and fluid properties that give the most minimum droplet SMD at the spray centreline. The results show remarkable improvement on SMD and new SMD correlation for the model