Employing Adaptive Mesh Refinement for Simulating the Exhaust Gas Recirculation Mixing Process

One significant emissions issue of compression ignition engines that directly influences human health is the production of nitrogen oxides (NOx). Once produced, these species are difficult to convert catalytically in the exhaust and often require a complex aftertreatment system to mitigate their release into the environment. The common methodology by the internal combustion engine community to reduce the amount of NOx is to employ Exhaust Gas Recirculation (EGR) in order to dilute the intake mixture with inert species (e.g., water). This lowers the combustion temperature lessening the thermal NO production mechanism. Improper mixing of EGR with the intake (species in-homogeneity, low levels of mixing turbulence, etc.) can lead to significant cylinder-to-cylinder variation in combustion temperatures and NOx emissions, making it more difficult to achieve regulatory standards.In this effort, a three-dimensional (3-D), transient, computational fluid dynamics (CFD) analysis was performed in order to more accurately model the mixing of EGR and intake for a single-cylinder test engine. Mixing is achieved for this engine by using a small rectangular box in which clean air and engine exhaust for controlled recirculation are mixed prior to engine intake. A matrix of computational analyses at different engine loads, and simulation types (large eddy and Reynolds-averaged Navier-Stokes) at 25% EGR were performed to check computational time and agreement with experimental measurements. Moreover, this effort employs the use of adaptive mesh techniques in order to understand their usage and validate correct implementation for later endeavors including more complex geometries, such as the manifold of a multi-cylinder engine. The simulation results indicate that mass flow rate and temperature of the mixture as it leaves the mixing box agree to within 3% of experimental values. Furthermore, pressures at the air and EGR inlet boundaries showed agreement to around 1% and 12%, respectively, with the experimental measuring points indicated as the reason for the difference. In addition, species mixing of carbon monoxide was uniform to within 440 ppm. Finally, the use of the models may also account for a prior discrepancy in the output power of the single-cylinder engine test stand.Copyright © 2014 by ASME