CFD modeling of reacting diesel sprays is an important factor in the advancement of diesel engines. With the future regulations on NOx and soot emissions on diesel engines, it is critical to accurately model the diesel engine spray and combustion process. The soot emissions depend largely on the spray and flame structure in the engines. Thus, it is important that the computational model is able to capture the spray and fuel vapor distribution.
An accurate simulation of spray involves Direct Numerical Simulation of liquid spray atomization and evaporation of atomized drop-lets. Such a detailed simulation will be computationally expensive and not viable for day-to-day engineering applications. As an alter-native to this, most of the commercial CFD programs model liquid sprays using a Lagrangian approach where the spray droplets are tracked along its trajectory and the primary gas phase is modeled as Eulerian fluid. The spray particles are assumed to be point masses that interact with the gas phase via mass, momentum, and energy exchange as they travel through the gas phase. However, simplification of the physics used in this model gives rise to other problems such as low accuracy and grid dependency.
In a study by researchers from Iowa State University and John Deere Power Systems, KIVA-3V is used to model diesel spray simulations. The basic spray model in KIVA-3V is improved by adding KH-RT breakup model and stochastic collision model. However, this spray along with evaporation model underpredicted vapor penetration. It can also be seen that the slope of vapor penetration curve has a sudden decrease when the liquid particles are completely evaporated. The momentum sources from liquid particle vanish as soon as the particles are evaporated. The absence of momentum sources causes the sudden variation of vapor penetration with and without the presence of liquid particles. Thus additional models are required to supply the momentum sources to the gas phase beyond the region where the liquid spray is present.
Vapor penetration was improved by using a vapor particle approach. In this model, the liquid particles evaporate to generate vapor particles and the vapor particles transfer mass to the gas phase by laminar diffusion. However a fine grid was required in the fuel jet region to capture the fuel penetration. The vapor particle approach was further improved by adding gas jet model and also tracking the vapor particle further downstream after the liquid core is completely evaporated. It was shown to improve the grid independency and vapor penetration. The model uses a cut-off distance based on grid size to release the vapor particles. This could pose a problem in practical engine simulations as the grid size could vary in the domain. Other improvements of the vapor particle model have also been proposed in recent studies.
In this study, a new vapor particle model is used in order to improve the spray penetration results. In this spray evaporation model, momentum sources are provided to the gas phase through gaseous particles which does not retain any fuel vapor mass. The gaseous particles are tracked even after the liquid droplet is completely evaporated. This is necessary to provide extra momentum for the gas phase in order to increase the vapor penetration. By providing continuous momentum sources even after the liquid droplet regime, the sudden decrease in the vapor penetration is also avoided. The current model differs from the previous models in releasing the vapor mass to gas phase. Previous models retained the vapor mass to the vapor particle and gradually released it to the gas phase based on certain criteria which depend on cell size. This results in conditions such as very low or no vapor near the nozzle region even though the particle is evaporating, especially in the case of coarse grids. In the new model, the vapor mass is released to the gas phase directly from the liquid droplet and the same mass is added to the vapor particle, which is an imaginary particle used to track the momentum sources. The model is applied to non-reacting diesel sprays to validate the vapor penetration and grid dependency. The model is further applied in reacting spray cases in constant volume chamber and diesel engine simulations.
The researchers’ proposed gas particle model uses imaginary gas particles to provide momentum sources even after the liquid particles are completely evaporated. The evaporated vapor mass is released to the gas phase simultaneous with the expansion of vapor particle. The tracking of momentum sources are terminated when the velocity of gas particle is comparable to that of the local gas phase velocity. The model was applied to reacting and non-reacting sprays and the results are analyzed.
In the case of non-reacting sprays, the model was able to provide improved vapor penetration as well as good grid independency. Gas particle model was applied to model non-reacting sprays at different ambient densities and was able to accurately predict vapor penetrations at all test conditions. Further, the model was used along with an N-heptane reaction mechanism coupled with a two-step soot model for simulating constant volume diesel spray combustion. The lift-off lengths and soot trends from the model were in reasonable agreement with the experimental results. GPM was also used on diesel engine combustion modeling along with the reaction mechanism and soot model. It was able to predict good ignition delay and soot results. Overall the gas particle model was able to obtain considerable improvement from standard KIVA model in predicting the vapor distribution for diesel sprays.
Though the model predicted good vapor penetrations and ignition delays, it over-predicted the cylinder pressure at late-injection cases. Further improvements on the model are required for accurate prediction of cylinder pressure. One area of concern with this current model is the over diffusion of vapor in the radial direction by the momentum sources from the gas particle, which results in a more rigorous combustion. Another recommendation for future work is to improve the tuning factor as a function of other physical quantities so that there is no need of a step change for reacting and non-reacting sprays.
This article is based on SAE technical paper 2013-01-1598 by Sujith Sukumaran and Song-Charng Kong of Iowa State University and Nam Hyo Cho of John Deere Power Systems.