Investigation of the grid and intake-generated tumble on the in-cylinder flow in a compression ignition direct-injection engine

Abstract

Abstract

The in-cylinder flow field was numerically investigated for a four-valve per cylinder Caterpillar diesel CAT3401 engine. The computer program used was the KIVA3V code developed by Los Alamos National Laboratory. To assess the effect of modelling assumptions made during the domain discretization, three grid models were developed. The first grid or the ‘complete’ grid was a full 360° mesh with moving valves, intake ports, exhaust ports, and runners. The second grid was a full 360° mesh without moving valves, intake ports, or exhaust ports. The third grid assumed that the in-cylinder flow field was periodic and a 60° mesh was generated to accommodate one of the six centrally located fuel injectors. It also neglected the valves and the intake and exhaust ports. The simplification of the grids is to reduce the massive computational and memory requirements to model these flows.

For the complete grid, the computation was initiated when the piston was at the top dead centre location on the intake stroke. For the second and third grids, the computation was initiated after the intake valve had closed. Complex three-dimensional vortical flow structures or tumble is developed during the intake stroke. These structures were clearly shown when the complete grid was used. By comparing the solution of the complete grid with the solutions of the second and third grids, it was seen that important in-cylinder flow information is lost. To overcome this deficiency, a new parameter, namely the initial tumble ratio, was embedded into the standard KIVA3V code by the present authors. When the simplified second grid is used in conjunction with the initial tumble ratio option, an in-cylinder flow pattern similar to that which was computed with the complete grid was obtained. This indicates that the new option significantly improves the accuracy of in-cylinder flow prediction when the moving valves and intake or exhaust ports are not modelled.

The effects of the values of the initial turbulence kinetic energy (TKE_i) and the initial turbulence length scale (SCL_i) were also studied. It is found that the suggested value of TKE_i used in KIVA3V is too small to predict the actual turbulence kinetic energy in the cylinder when the second and third grids are used. Therefore, a larger value for TKE_i is recommended. The effects of the different grids, the new initial tumble ratio, and the different values of SCL_i and TKE_i on important flow parameters, such as the swirl ratio, the tumble ratio, and the turbulence kinetic energy are compared in detail in this article.