Searching for a Numerical Model for Prediction of Pressure-Swirl Atomizer Internal Flow

Abstract

Numerical prediction of discharge parameters allows design of a pressure-swirl atomizer in a fast and cheap manner, yet it must provide reliable results for a wide range of geometries and operating regimes. Many authors have used different numerical setups for similar cases and often concluded opposite suggestions on numerical setup. This paper compares 2D (two-dimensional) axisymmetric, 3D (three-dimensional) periodic and full 3D numerical models used for estimation of the internal flow characteristics of a pressure-swirl atomizer. The computed results are compared with experimental data in terms of spray cone angle, discharge coefficient (C-D), internal air-core dimensions, and velocity profiles. The three-component velocity was experimentally measured using a Laser Doppler Anemometry in a scaled transparent model of the atomizer. The internal air-core was visualized by a high-speed camera with backlit illumination. Tested conditions covered a wide range of the Reynolds numbers within the inlet ports, Re = 1000, 2000, 4000. The flow was treated as both steady and transient flow. The numerical solver used laminar and several turbulence models, represented by k-epsilon and k-omega models, Reynolds Stress model (RSM) and Large Eddy Simulation (LES). The laminar solver was capable of closely predicting the C-D, air-core dimensions and velocity profiles compared with the experimental results in both 2D and 3D simulations. The LES performed similarly to the laminar solver for low Re and was slightly superior for Re = 4000. The two-equation models were sensitive to proper solving of the near wall flow and were not accurate for low Re. Surprisingly, the RSM produced the worst results.Numerical prediction of discharge parameters allows design of a pressure-swirl atomizer in a fast and cheap manner, yet it must provide reliable results for a wide range of geometries and operating regimes. Many authors have used different numerical setups for similar cases and often concluded opposite suggestions on numerical setup. This paper compares 2D (two-dimensional) axisymmetric, 3D (three-dimensional) periodic and full 3D numerical models used for estimation of the internal flow characteristics of a pressure-swirl atomizer. The computed results are compared with experimental data in terms of spray cone angle, discharge coefficient (C-D), internal air-core dimensions, and velocity profiles. The three-component velocity was experimentally measured using a Laser Doppler Anemometry in a scaled transparent model of the atomizer. The internal air-core was visualized by a high-speed camera with backlit illumination. Tested conditions covered a wide range of the Reynolds numbers within the inlet ports, Re = 1000, 2000, 4000. The flow was treated as both steady and transient flow. The numerical solver used laminar and several turbulence models, represented by k-epsilon and k-omega models, Reynolds Stress model (RSM) and Large Eddy Simulation (LES). The laminar solver was capable of closely predicting the C-D, air-core dimensions and velocity profiles compared with the experimental results in both 2D and 3D simulations. The LES performed similarly to the laminar solver for low Re and was slightly superior for Re = 4000. The two-equation models were sensitive to proper solving of the near wall flow and were not accurate for low Re. Surprisingly, the RSM produced the worst results.