Search Results

The flow through the valves of an engine cylinder head is very complex in nature due to very high gas velocities and strong flow separation. However, it is also the typical situation in almost every engine related flow. In order to gain better understanding of the flow features after the cylinder head, and to gain knowledge of the performance level that can be expected from CFD analysis, flow field measurements and computations were made in an engine rig. Particle image velocimetry (PIV) and paddle wheel measurements have been conducted in a static heavy-duty diesel engine rig to characterize the flow features with different valve lifts and pressure differences. These measurements were compared with CFD predictions of the same engine. The simulations were done with the standard k-ε turbulence model and with the RNG turbulence model using the Star-CD flow solver.

The study aims at providing more accurate initial conditions for turbulence prior to combustion with the help of a four valve, large bore diesel engine CFD model. Combustion simulations are typically done with a sector mesh and initial turbulence in these simulations is usually taken from relatively inaccurate correlations. This study also aims at developing a more accurate initial turbulence correlation for combustion simulations. A one-dimensional model was first used to provide boundary conditions as well as the initial flow conditions at the beginning of the simulation. Steady state and transient boundary conditions were studied. Also, the standard κ - ε and RNG/κ - ε turbulence models were compared. From the averaged values of turbulence kinetic energy and its dissipation rate over the cylinder volume, a re-tuned correlation for defining the initial turbulent conditions at bottom dead center (BDC) prior to the compression stroke is proposed.

A new variable density / physical property wall function formalism has been developed. The new formalism is designed to extend the validity range of wall functions to cover both the low- and high-Reynolds-number domains so that the restrictions on the non-dimensional near-wall mesh resolution can be avoided. The new formalism also accounts for the temperature gradient induced variations of density, viscosity, heat conductivity and specific heat capacity. The new wall function formalism is constructed in conjunction with a modified low-Reynolds-number turbulence model in order to avoid the conflicting requirements of low- and high-Reynolds-number models on the near wall mesh resolution. The new formulation is validated with test simulations of strongly heated air flows in circular tube against measurements and Direct Numerical Simulation (DNS) results.

Hydrotreated vegetable oil (HVO) is a high-cetane number alternative fuel with the potential of drastic emissions reductions in high-pressure diesel engines. In this study the behavior of HVO sprays is investigated computationally and compared with conventional diesel fuel sprays. The simulations are performed with a modified version of the C++ open source code OpenFOAM using Reynolds-averaged conservation equations for mass, species, momentum and energy. The turbulence has been modeled with a modified version of the RNG k-ε model. In particular, the turbulence interaction between the droplets and the gas has been accounted for by introducing appropriate source terms in the turbulence model equations. The spray simulations reflect the setup of the constant-volume combustion cell from which the experimental data were obtained.

A modified version of the Laminar and Turbulent Characteristic Time combustion model and the Hiroyasu-Magnussen soot model have been implemented in the flow solver Star-CD. Combustion simulations of three DI diesel engines, utilizing the standard k-ε turbulence model and a modified version of the RNG k-ε turbulence model, have been performed and evaluated with respect to combustion performance and emissions. Adjustments of the turbulent characteristic combustion time coefficient, which were necessary to match the experimental cylinder peak pressures of the different engines, have been justified in terms of non-equilibrium turbulence considerations. The results confirm the existence of a correlation between the integral length scale and the turbulent time scale. This correlation can be used to predict the combustion time scale in different engines.

The one-equation subgrid scale model for the Large Eddy Simulation (LES) turbulence model has been compared to the popular k-ε RNG turbulence model in very different sized direct injection diesel engines. The cylinder diameters of these engines range between 111 and 200 mm. This has been an initial attempt to study the effect of LES in diesel engines without any modification to the combustion model being used in its Reynolds-averaged Navier-Stokes (RANS) form. Despite some deficiencies in the current LES model being used, it already gave much more structured flow field with approximately the same kind of accuracy in the cylinder pressure predictions than the k-ε RNG turbulence model.