Search Results

Understanding the complex, transient spray phenomena associated with Gasoline Direct Injection (GDI) technologies continues to be key when designing injection systems to meet the stringent performance and emissions standards of modern internal combustion engines. Internal flow phenomena, such as string cavitation and hole-to-hole flow variations, are often highly transient and affected by turbulence. To better understand the current degree to which turbulence modeling influences simulations of GDI sprays, RANS and LES simulations have been performed on the multi-hole Spray G injector through the start of injection phase, with results compared to previously available X-Ray radiography data. Specifically, the k-ω SST RANS model and the k-equation LES model with a WALE sub-grid scale stress model have been tested on grids generated with the Generation 3 Spray G geometry, which includes as-produced injector dimensions based on X-Ray radiography measurements.

The salient features of modern gasoline direct injection include cavitation, flash boiling, and plume/plume interaction, depending on the operating conditions. These complex phenomena make the prediction of the spray behavior particularly difficult. The present investigation combines mass-based experimental diagnostics with an advanced, in-house modeling capability in order to provide a multi-faceted study of the Engine Combustion Network’s Spray G injector. First, x-ray tomography is used to distinguish the actual injector geometry from the nominal geometry used in past works. The actual geometry is used as the basis of multidimensional CFD simulations which are compared to x-ray radiography measurements for validation under cold conditions. The influence of nozzle diameter and corner radius are of particular interest. Next, the model is used to simulate flash-boiling conditions, in order to understand how the cold flow behavior corresponds to flashing performance.

Flash boiling conditions, where the fuel is superheated with respect to cylinder pressure, are often found in gasoline direct injection engines. This phenomenon affects the flowrate of the fuel and can cause choking of the nozzle. In this work we present multi-dimensional simulations of flashing internal injector flow. The modeled fluid quality (mass fraction of vapor) tends towards the equilibrium quality based on the Homogenous Relaxation Model. The relaxation time is dependent on the local pressure, the vapor pressure, and the void fraction. Simulations of the internal flow are presented and, where possible, validated with experimental data. Both two- and three- dimensional computational results show geometrically-induced phase change, similar to cavitation, near the nozzle entrance. Near the nozzle exit plane the phase change occurs at all radial locations and can be quite sensitive to temperature.

Past experiments have shown that numerous micro-hole sprays in close proximity produce drop sizes that are sensitive to the nozzle arrangement. Numerical studies have been performed to identify the interaction mechanisms between closely spaced sprays. It is shown that nozzle configurations can lower the drop-gas relative velocity and droplet Weber number, leading to reduced atomization intensity. However, the collisions involving droplets from neighboring sprays have a much greater effect on droplet size. Thus, neighboring sprays primarily interfere with each other through droplet collision.

Computational fluid dynamic (CFD) analysis of in-cylinder events in the automotive industry is heavily dependent on spray simulations for almost all advanced engine concepts. As the upper bound of efficiency in these engines is pursued, the accurate prediction of sprays is critical, since the mixture preparation and ignition, whether by spark or auto-ignition, needs to be precisely controlled. While most of the spray literature closely examines various drop processes, especially breakup, comparatively less attention has been focused on the momentum coupling between the liquid and the gas phases, in particular the numerical aspects. In fact, adjusting models for the evolution of drop size has been one of the dominant means of controlling predictions of spray shape and liquid penetration.

Improved numerical methods and physical models have been applied to droplet collision modeling. Contrary to the common practice of using the fixed gas phase mesh to calculate droplet collision incidence, a new algorithm generates a collision mesh independent of the gas phase mesh at each time step. By partitioning collision cells according to the number density of parcels, the algorithm is capable of achieving higher spatial resolution than that of the gas phase mesh. At the same time this method maintains an adequate statistical sample of parcels in the collision cells to ensure statistically accurate results. The continual random rotation of the mesh ensures that parcels that are near each other will, on average, be in the same collision cell. This produces the physically reasonable result that parcels in close proximity should have the opportunity to collide. Another important advantage of the algorithm is that it can be applied to any orifice configuration.

Lagrangian particle tracking has proven very useful for IC engine spray simulations. However, traditional techniques have been very susceptible to grid dependency. As a consequence, predicted engine performance can depend on the computational mesh. The two main causes of this problem are the collision algorithm and the inter-phase coupling. Using the No Time Counter droplet collision algorithm with an embedded collision mesh practically eliminates grid dependency as a consequence of the collision algorithm. The remaining grid dependency is due to numerical error of the gas phase near the injector and problems with coupling the gas phase and liquid phase. Calculations were made using a novel interpolation scheme that takes advantage of the polar symmetry in the spray. The scheme transforms the velocities into their polar components before doing interpolation. The results show a dramatic decrease in grid dependency.

To model sprays from pressure-swirl atomizers, the connection between the injector and the downstream spray must be considered. A new model for pressure-swirl atomizers is presented which assumes little knowledge of the internal details of the injector, but instead uses available observations of external spray characteristics. First, a correlation for the exit velocity at the injector exit is used to define the liquid film thickness. Next, the film must be modeled as it becomes a thin, liquid sheet and breaks up, forming ligaments and droplets. A linearized instability analysis of the breakup of a viscous, liquid sheet is used as part of the spray boundary condition. The spray angle is estimated from spray photographs and patternator data. A mass averaged spray angle is calculated from the patternator data and used in some of the calculations.

Diesel fuel spray characteristics are thought to depend on cavitation inside the fuel injector nozzles. These nozzles are very small and the fuel flow is very fast, making experimental observation very difficult. Numerical simulation of the two-phase flow is hindered by the severe density difference between the liquid and vapor as well as the existence of complex free surfaces. Recent experimental and numerical advances are now permitting visual observation of real-scale cavitating flow and two-dimensional simulation of the cavitating flow. In contrast to past work, we have chosen to study asymmetric nozzle flow. Asymmetry is more representative of real injector geometry than symmetric nozzles and may yield more reproducible results. Experimental runs of 1 mm long nozzles were made at upstream pressures up to 120 bar with downstream pressures from 1 to 50 bar. Photographs of planar asymmetric nozzles revealed complex transient structures on the fuel-vapor interface.

The flow through diesel fuel injector nozzles is important because of the effects on the spray and the atomization process. Modeling this nozzle flow is complicated by the presence of cavitation inside the nozzles. This investigation uses a two-dimensional, two-phase, transient model of cavitating nozzle flow to observe the individual effects of several nozzle parameters. The injection pressure is varied, as well as several geometric parameters. Results are presented for a range of rounded inlets, from r/D of 1/40 to 1/4. Similarly, results for a range of L/D from 2 to 8 are presented. Finally, the angle of the corner is varied from 50° to 150°. An axisymmetric injector tip is also simulated in order to observe the effects of upstream geometry on the nozzle flow. The injector tip calculations show that the upstream geometry has a small influence on the nozzle flow. The results demonstrate the model's ability to predict cavitating nozzle flow in several different geometries.