An in-depth comparison of experimental and computational turbulence parameters for in -cylinder engine flows.

Abstract

The in-cylinder fluid motion of an internal combustion engine is a complex, turbulent flow that has a significant impact on the resulting air/fuel mixing, combustion efficiency, and pollutant formation. Therefore, it is necessary to comprehend the physics governing turbulence generation during the engine cycle in order to make future improvements to current engine designs and CFD models. An optical engine operating at 2000 rpm is used in this work to characterize the turbulence levels for high and low swirl flows at various locations. An in-depth comparison of turbulence parameters from PIV measurements and predictions by KIVA-3V using the RNG and standard k-epsilon models is presented. Vanzieleghem (2004) observed that the RNG k-epsilon and standard k-epsilon turbulence models in KIVA-3V predicted different trends and levels of turbulent kinetic energy during the engine cycle. Three different experiments are performed in this work to assess these predictions. Various turbulence parameters are discussed and compared for each case. The issue of cycle-to-cycle variability is addressed by comparing the Reynolds decomposition, Gaussian filter, and proper orthogonal decomposition (POD) results. First, PIV experiments are presented to analyze the flow in the swirl plane of the compression stroke for three crank angles. The experimental high swirl results at every location show agreement with both the RNG and standard k-epsilon model. A Gaussian filter of 16mm compares well with the POD cutoff point. Second, high-resolution PIV data is presented for the same plane. In agreement with Kolmogorov theory, the data shows a roll-off from the inviscid inertial region to the dissipative region of turbulence. This is important information for future subgrid model developments (e.g. LES). Third, the influence of fuel injection on turbulence in the surrounding air is addressed along the central vertical plane in the cylinder to validate the turbulence and spray sub-models. The turbulence models and experimental data show no significant increase in turbulence levels of the surrounding air with fuel injection. Qualitative agreement of the predictions from the turbulence models with experimental high swirl results are found, while the low swirl flow results are completely dissimilar.