Search Results

After-treatment sensors are used in the ECU feedback control to calibrate the engine operating parameters. Due to their contact with exhaust gases, especially NOx sensors are prone to soot deposition with a consequent decay of their performance. Several phenomena occur at the same time leading to sensor contamination: thermophoresis, unburnt hydrocarbons condensation and eddy diffusion of submicron particles. Conversely, soot combustion and shear forces may act in reducing soot deposition. This study proposes a predictive 3D-CFD model for the analysis of the development of soot deposition layer on the sensor surfaces. Alongside with the implementation of deposit and removal mechanisms, the effects on both thermal properties and shape of the surfaces are taken in account. The latter leads to obtain a more accurate and complete modelling of the phenomenon influencing the sensor overall performance.

The accurate representation of Internal Combustion Engine (ICE) flows via CFD is an extremely complex task: it strongly depends on a combination of highly impacting factors, such as grid resolution (both local and global), choice of the turbulence model, numeric schemes and mesh motion technique. A well-founded choice must be made in order to avoid excessive computational cost and numerical difficulties arising from the combination of fine computational grids, high-order numeric schemes and geometrical complexity typical of ICEs. The paper focuses on the comparison between different mesh motion technologies, namely layer addition and removal, morphing/remapping and overset grids. Different grid strategies for a chosen mesh motion technology are also discussed. The performance of each mesh technology and grid strategy is evaluated in terms of accuracy and computational efficiency (stability, scalability, robustness).

The overall performance of direct injection (DI) engines is strictly correlated to the fuel liquid spray evolution into the cylinder volume. More in detail, spray behavior can drastically affect mixture formation, combustion efficiency, cycle to cycle engine variability, soot amount, and lubricant contamination. For this reason, in DI engine an accurate numerical reproduction of the spray behavior is mandatory. In order to improve the spray simulation accuracy, authors defined a new atomization model based on experimental evidences about ligament and droplet formations from a turbulent liquid jet surface. The proposed atomization approach was based on the assumption that the droplet stripping in a turbulent liquid jet is mainly linked to ligament formations. Reynolds-averaged Navier Stokes (RANS) simulation method was adopted for the continuum phase while the liquid discrete phase is managed by Lagrangian approach.