Three-dimensional simulations of ignition and combustion of a diesel spray were conducted. The primary goal of the work was to compare two different CFD codes: OpenFOAM, an object-oriented C++ based code, and KIVA-3V. The spray is modelled by the Eulerian-Lagrangian approach in both codes, with several common submodels. Some important sub-models implemented include inter alia a Kelvin-Helmholtz/Rayleigh-Taylor (KH/RT) model for spray break-up, an improved spray collision model, and a Partially Stirred Reactor (PaSR) model for turbulence-chemistry interaction. Both CFD codes solve the chemical reaction equations in a fully coupled manner. A cubic-shaped Cartesian mesh was used in the KIVA-3V simulations, while a polyhedral mesh including a combination of hexagonal and prism-shaped cells was constructed for the OpenFOAM computations.
The effects of high EGR and ambient temperature on the ignition and flame lift-off processes of a diesel spray were investigated. Sandia experiments conducted in a high-pressure and high-temperature constant-volume vessel were chosen for the simulations and validations. A single spray was injected into the vessel, and EGR was mimicked by reducing the oxygen concentration. The diesel reference fuel (n-heptane) was considered. For the study, a medium-size mechanism involving 83 species and 338 reactions was employed. The mechanism was validated using the CHEMKIN II package and the reaction rate constants were adjusted on the basis of measurements of auto-ignition delays of n-heptane/air mixtures in shock-tube experiments (with equivalence ratios ranging from 0.2 to 0.4 at 50 bar, and from 0.5 to 2.0 at 13.5 bar and 41.0 bar), laminar flame speeds (1 atm and 3 atm), and flame structure in burner-stabilized premixed flames (1 atm).
The simulations demonstrate that both CFD codes are capable of spray ignition and combustion studies, though both show strong grid-dependence. The numerical results show that the ignition delay, flame lift-off and combustion temperature of the spray are strongly influenced by EGR and ambient gas temperature. These predictions are in agreement with measurements. Nevertheless, differences are observed between the results predicted by OpenFOAM and those from KIVA-3V, for example, the flame predicted by the former is thinner and longer than that by the latter, which requires further investigation.