SIMULATING SPRAY DYNAMICS WITH A FINITE ELEMENT METHOD FOR INTERNAL COMBUSTION ENGINES USING LARGE EDDY SIMULATIONS

Spray dynamics in an internal combustion engine is comprised of complex phenomena while interacting with unsteady turbulence. Physics requires detailed modeling of the dynamics for spray and carrier gases to accurately predict a spray's fate. Large eddy simulation (LES) turbulence modeling approaches are capable predictors of the turbulent processes and of dynamically modeling sub-grid scales, therefore enabling calculation of model coefficients for the smallest resolved scale. Dynamic LES methods are also well suited for unsteady flows associated with spray injection and engine fluid dynamics. In this study, for the first time, a dynamic Verman LES scheme employed in a stabilized a finite element (FE) framework is used to model the spray process with emphasis on injected fuels for simulating internal combustion engines. Spray modeling often comprises a coupled Eulerian-Lagrangian approach to capture the droplet/particle dynamics, where the droplets are modeled in the Lagrangian frame, interacting with the gas. The momentum and heat exchange between the fluid gases and the evaporating and atomizing spray droplets are modeled in a two-way coupling system as described in this paper. Direct injected liquid is modeled in this paper as a spherical ligament of fuel and ligament breakup to atomization use our version of the Kelvin Helmholtz break-up scheme. Discussed are models and methods of the whole system in some detail. The method for simulation of the fluid's momentum, heat transfer, and turbulence are discussed as is the system to evaluate droplet or ligament properties. Validation or results of the modeling are presented on test cases as determined by the Engine Combustion Network, base cases G along with some benchmarking cases. We demonstrate in the paper by using the FE method combined with the Verman LES model, and our version of the Kelvin Helmholtz-Rayleigh Taylor spray break-up model combined with two-way coupling is quite capable of modeling sprays, having distinct and significant advantages over the more traditional finite volume discretization framework using similar spray modeling methods.