Formation du mélange par injection simple en milieu évaporatif avec une approche RANS

Abstract

The numerical injection of a fuel spray is an essential technique to model the flow of fuel in internal combustion engines, potentially leading to better design. It can contribute to a reduction of the pollutants or to an improvement of the engine performance with a better understanding of the combustion cycle. One of the most important elements of that kind of simulation is the simulation of the liquid phase of the fuel. Unfortunately, most of the commercial CFD software are quite limited on the fragmentation models available, and they usually require a calibration to model the spray accurately. The main objective of this document is to facilitate the calibration process with a better identification of the elements that affect the calibration of the fragmentation models on a fuel spray. The calibration of the fragmentation model requires experimental results with the same environmental conditions that those used for the numerical simulation. This study uses the Spray A results from the Sandia laboratory. This experiment injects dodecane fuel inside a vessel with a pressure and temperature similar to the environmental conditions from an internal combustion engine. The numerical simulations used in this study use a URANS approach the simulation of the turbulence and two fragmentation models: KHRT and SSD. The first part of this study can be summarized by analyzing the individual effects of multiple elements (mesh, turbulence models, constants, etc.) on the numerical simulation. From some numerical results, the turbulence model κ-ε realizable is better suited than the other models tested for this kind of application. Other observations include the fact that fine unstructured mesh (0.25mm) performs similarly to a structured mesh of the same size. The effects of each individual constant on the spray is also studied to determine a possible range of variation for each constant. This interval is useful when calibrating the fragmentation model. The second part of this study can be summarized by the comparison of different calibrated sprays. Six different sprays are calibrated, with three sizes of mesh for each fragmentation model. To calibrate the models, a Box-Benhken design with four or five constants is used for the KHRT model and a factorial design with two constants is used for the SSD model. The objective of the calibration is to match the liquid and vapor penetration of the experimental results. The result of the calibration is mostly successful for the liquid penetration, but the other comparative criterion (vapor penetration, vapor angle and SMD) do not match the experimental data very well. By comparing the results of the two models after calibration, it’s possible to conclude that the KHRT model allows a better match of the experimental. This model is also the one that allows the best control over the physic of the spray. The KHRT model is the recommended model for simulations of this type.

Date	7 Nov 2018
Original language	French
Awarding Institution	École de technologie supérieure
Supervisor	Patrice Seers (Supervisor) & François Garnier (Co-supervisor)