The current diesel technologies involve various operating conditions that lead to a wide spectrum of combustion regimes. The previous combustion models somehow lack the universality across the combustion regimes or suffer from computational cost. Additionally, they are developed in the context of Reynolds Averaged Navier Stokes (RANS). In modeling engine flows, large eddy simulation (LES) is better than RANS at providing temporal and spatial details in the mixing field. To combustion modeling, this implies that an LES-based combustion model would offer higher predictive accuracy than a RANS-based model. Therefore, the objective of this study was to develop a diesel combustion model that is LES-compatible, efficient, and effective over an extensive range of engine conditions including conventional and low-temperature diesels.
The combustion model was developed to cover major regimes in diesel combustion using the kinetically-controlled, quasi-steady homogeneous, quasi-steady flamelet, and partially-premixed combustion modes. The local combustion regime was identified by two combustion indices based on the local thermodynamic and mixing conditions. In the regimes of kinetically-controlled and quasi-steady homogeneous combustion, the combustion model neglected the subgrid-scale (SGS) mixing effects. In the rest of regimes, the model took the SGS mixing effects into account.
The LES mixing models included a dynamic structure model for subgrid stresses, a one-equation viscosity model for SGS scalar fluxes, an algebraic model for SGS scalar dissipation. The established RANS-based spray and wall models were retained with the RANS-derived scales being replaced by the LES-derived scales. All of the models were incorporated into the KIVA3V code to form an LES package for engine simulations.
The LES package was tested over a wide range of engine conditions including the conventional diesel-type operations and the LTC operations. The predictions and measurements of pressure and heat release rates were in excellent agreement in all of the cases. The models predicted the in-cylinder details of spray, large- and small-scale mixing, and combustion processes accurately. Finally, the present model exhibited advantages in both universality and computational efficiency compared to the previous combustion models, such as the Chemkin model, the characteristic time scale model, and the flamelet time scale model.