No abstract

Flash boiling occurs when a superheated liquid fuel is injected into a sub-atmospheric pressure environment. This phenomenon is often seen in a gasoline direct injection (GDI) engine under part-load conditions. The injector tip is exposed to the combustion environment, increasing the fuel temperature inside the injector up to 80–90° C. Under these flash boiling conditions, the injected liquid fuel will undergo rapid vaporization, affecting the spray structure. The flash boiling spray helps achieve better atomization of the fuel spray. However, in some conditions, due to flashing, a dense vapor cloud will form in the in-cylinder and disturb the air-fuel mixture, and lead to unwanted pollutants. Several numerical models have been developed for the flash boiling sprays. However, most of these models are hard to implement in a multi-cycle GDI engine combustion simulation. The existing flashing-based evaporation models in the CONVERGE code are not dimensionally consistent. In that regard, a phenomenological model of flash boiling spray is developed using the relevant non-dimensional numbers to help GDI engine combustion modeling under such conditions and address the issue of dimensional inconsistency in the source term formulations. A multi-hole GDI injector from the engine combustion network (ECN) called Spray G is considered for this work. The model is developed using the CONVERGE v2.4 user-defined functions (UDFs). The blob injection method is considered for this work, and the simulations are carried out using the Eulerian-Lagrangian framework. An Unsteady Reynolds Averaged Navier-Stokes (URANS) RNG k-ε turbulence model is considered in this study. The KH-RT breakup length model is adopted for modeling spray breakup. The developed model has achieved a reasonable agreement with the experimental spray images. Comparisons of the predictions are made with the data available through ECN, such as spray penetrations and spray patternations.

A large share of passenger vehicles are currently running on gasoline, which is considered the major contributor to carbon emissions from the transportation sector. In gasoline direct injection (GDI), the fuel is introduced directly into the combustion chamber in a spark ignition (SI) engine. GDI has the advantage of operating at a higher compression ratio due to the charge cooling effect, resulting in better fuel utilization and lower pollutant emissions. The spray formed from fuel injection under different engine running conditions governs the charge (air-fuel) characteristics. The present work focuses on experiments using a GDI system with a multi-hole injector for gasoline fuel sprays in a constant volume chamber. The saturation pressure of the fuel present within the nozzle increases with rising temperature (80 to 90°C) from heat transfer due to the placement of the injector directly into the combustion chamber. At part-load running conditions, when the heated fuel injected into an environment has a pressure sufficiently lower (achieved in early injection) than its saturation pressure, it attains the state of superheating and undergoes flash boiling. The abrupt vaporization of injected fuel results from flash boiling and affects spray characteristics. The ratio of chamber pressure to saturation pressure (Pch/Psat) indicates the propensity of flashing for a given super-heated operating condition. Macroscopic analysis of GDI sprays is reported in this study under flashing and non-flashing conditions through different experimental techniques to provide a comprehensive understanding. Experimental studies reported in this kind of literature typically adopt a specific technique to elucidate the details of spray evolution. A Diffused Back Illumination (DBI) imaging technique with a low-speed Nd: YAG laser has been carried out to analyse the liquid phase of the spray, such as penetration, cone angle, and width. Additionally, a novel Structured Laser Illumination Planar Imaging (SLIPI) technique has been adopted to identify the relatively dense spray regions by addressing the difficulty due to multiple scattering of light illuminating from surrounding scatterer media (droplets). SLIPI is used to suppress the effect of the surrounding light artefacts, helping explore the structure in a new dimension. A multi-plume spray is expected to have voids, dense, and dilute portions at different locations, affecting the fuel mixture, and interestingly, it appears after subduing the effects of multi-scattering from a modulated laser sheet using SLIPI. An LED-based spray illumination was also used to compare the DBI and SLIPI results to better understand the utility and effectiveness of different imaging techniques in a densely populated and highly transient spray phenomenon.

A numerical study has been carried out to understand the effects of Unsteady Reynolds Averaged Navier-Stokes (standard k ― ε and RNG k ― ε model) and large eddy simulations (LES) on a multi-hole gasoline direct injection (GDI) system. The fuel injector considered in this study is the Spray G nozzle from the Engine Combustion Network (ECN). A blob injection model, based on empirical rate of injection (ROI) profile, is considered in this study. The latest data on spray penetrations from Engine Combustion Network is used for model validation along with experimental findings on suction velocity and local droplet diameter. The spray breakup is simulated by using the KH-RT breakup length model. The turbulence model constant Cε1, is tuned to match with the experimental data of liquid and vapor penetrations in simulations while using the standard k-ε turbulence model. On the other hand, the Kelvin-Helmholtz breakup model time constant (B1) and Rayleigh Taylor breakup length constant (Cbl) are tuned for the RNG k ― ε turbulence model. From this work it is observed that by increasing the breakup length model constants (Cbl), the radial dispersion of the spray increases, and the extent of breakup is lowered. The set of optimized model parameters used with RNG k - ε is also used for LES modeling studies with different sub-grid models. The spray penetrations with standard k ― ε turbulence (Cε1=1.44) model are reported underpredicting, and the RNG k ― ε and LES sub-grid models predicted well with the latest and recommended data from ECN. In terms of gas axial velocity comparison, the standard k-ε(Cε1=1.44) simulation setup does not perform as well as the simulation setups using RNG k-ε and LES turbulence models (with breakup parameters: Cbl = 16 and B1 = 32). However, the standard k-ε(Cε1=1.44) simulation setup perform better than the simulation setups using RNG k-ε and LES turbulence models (with breakup parameters: Cbl = 16 and B1 = 32) when it comes to predicting local droplet diameter at 15 mm downstream of the injector tip. A parametric study is also performed considering the geometry of the stepped holes in the computational domain. The rate of injection based simulation is initiated at the end of the smaller hole. The case including the stepped holes led to over-prediction compared to the case with the usual computational domain (i.e., without the stepped holes), in terms of spray penetrations, but exhibited higher levels of fluctuations in the spray morphology. Finally, parametric studies were carried out to understand the relative importance of the individual spray sub-models (breakup, evaporation and collision) and the results are conclusive that for a spray simulation the breakup models are the dominant factors.

No abstract