Validation of different turbulence models to analyse flow characteristics within a toroidal crank-rocker engine using particle image velocimetry

This paper examines the influence of CFD turbulence models on the in-cylinder flow characteristics for the intake and compression strokes within a single cylinder 120 cc toroidal crank-rocker engine. The turbulence models include four Reynold’s Averaged Navier Stokes (RANS) models, standard k-ε, Re-Normalization Group (RNG) k-ε, Realizable k-ε and Shear Stress Transport (SST) k-ω models, as well as one-equation viscosity model, Large Eddy Simulations (LES). This crank-rocker engine has a curved piston-cylinder geometry unlike conventional straight cylinder crank-slider engines and has a compression ratio of 8:1. Tests are conducted as cold-flow analysis at constant engine speed of 4000 rpm, at standard operating conditions. The simulations were performed using renowned CFD code CONVERGE®. Adaptive Mesh Refinement (AMR) was utilized with 4 mm base grid size for efficient calculations to save time and computational cost. The simulation results are validated using experimental Particle Image Velocimetry (PIV) analysis on an optical crank-rocker testing rig. It is observed that the RANS models agree well with the PIV experimental results in general, better than the LES model. But RNG k-ε model qualitatively captures turbulent structures and vortex positions more accurately than the other RANS models. An appreciable increase in velocity magnitude and Turbulent Kinetic Energy (TKE) during the intake is observed as compared to the compression stroke. There is a noticeable increase in the cycle-to-cycle variations in the LES model, which has low vortex stabilization and filters out eddies of certain lengthscale, as compared to the RANS models which has prominent circular flow fields. The investigations carried over suggested that the RNG k-ε model is more suitable for the in-cylinder flow analysis within the crank-rocker engine in terms of accuracy and reliability.