Press Room - Swirling CFD for spray generators

	Swirling CFD for spray generators

	Giovanni Corbinelli, Siemens VDO Automotive S.P.A, Pisa Plant, Italy

	Introduction Injector performance is a key factor in the success of any direct injection spark ignited engine design. The injector design has a direct influence on critical aspects of the engine performance such as fuel consumption, emissions and drivability. For a design to be successful fuel injection has to be carefully controlled in terms of both the static and dynamic flow rate and the characteristics of the spray (cone angle, penetration length, droplet size and droplet velocity). High-pressure swirl injectors are recognized to be one of the best adapted for this application because they generally produce fine spray at a moderate injection pressure. In order for the design of these injectors to be optimized, it is necessary to understand the influence of modifying parameters such as the dimensions of the injector and the conditions under which it operates, upon the fuel spray issued from the nozzle. Such work is very difficult to perform experimentally because the characteristic dimensions of such an injector are typically measured in micrometers rather than millimeters. Under these conditions numerical simulation with a CFD code, such as STAR-CD, can provide insight into the performance of the injector and significantly improve the quality of a design. In this article, STAR-CD is used in order to understand the complex physical phenomena occurring in a HPDI (High Pressure Direct Injection) Swirl injector, and to optimize the geometrical characteristics of the swirl spray generator group in order to increase the injector performance in terms of spray shape and quality. In figure 1 a classic swirl group generator is shown; it is composed by a needle guide, a swirl disc and a seat with the discharge orifice. During the injection the pressurized fuel is forced through tangential passage of the swirl disc into the swirl chamber. It rotates in the chamber before then emerging from the discharge orifice in the form of a thin conical sheet. The swirl motion is clearly shown in figure 2, in which the path of a typical fuel particle is illustrated (colored by the velocity of the particle). Close to the injector axis the strong swirl motion is responsible for the inducing an air core. The film thickness at the exit of the injector is illustrated in figure 3, which shows the liquid film adjacent to the walls and the air core generated at the injector axis. Although it is possible to perform accurate CFD simulations on two-phase systems (in this case fuel and air), such calculations are significantly more expensive to compute than a simple single-phase system and make carrying our parametric studies involving a large number of simulations prohibitively time-consuming. The analysis In a swirl injector the most important parameters to control are the flow rate and the spray cone angle during steady state operations. This study was therefore limited to the simulation of stationary flows: although it is hoped that the conclusion drawn from these analyses can be extended into more expensive transient calculations. In the initial phase of this work, a single geometry was simulated using both single phase and two phases approaches. The purpose of this stage was to estimate how much error is introduced by making a simplifying assumption that full two-phase system can be represented by a single phase. If the level of error could be demonstrated to be sufficiently small, then significant savings can be made both in terms of time and computer resource by using the significantly less expensive single phase technique. The meshes for all calculations were constructed using STAR-CD’s automatic mesh generation module, which utilizes a unique meshing methodology, known as trimmed-cell technology. This technique optimizes the quality and distribution of cells in the flow volume, by constructing mainly regular hexahedral cells in the core of the domain and trimmed cells (hexahedra cut by a plane) near its surfaces. The trimmed cells mesh is surrounded by layers of regular prism cells that allow optimal capture of boundary layers. The mesh used in these analyses is shown in figure 4. Significant differences in the two approaches can be seen in the pressure profiles immediately after the transition cone. The profiles in the early part of the transition cone and at the injector outlet are however nearly identical. When assessed in terms of the key design parameters (such as swirl, outlet axial velocity, outlet pressure and film thickness) the single and two-phase approaches are remarkably similar. For this reason the single phase approach was deemed sufficiently accurate and adopted for all subsequent simulations. Several different geometrical configurations were then simulated in order to investigate the influence of different parameters on the swirl group generator performance. In one example the needle lift was varied parametrically in order to examine its influence on the flow rate and the pressure drop in the sealing band area and to find out a calibration able to guarantee good injector performances. Conclusion These analyses show clearly that CFD codes such as STAR-CD have an important role in the understanding of injector seat behavior. Because the dimensions of injectors are so small, experimental work is extremely limited leaving CFD as the only instrument that allows understanding of the importance of the geometrical characteristics on the injector behavior and detects trends useful for the definition of injector designs. Figures: Fig. 1: Sketch of the swirl group generator Fig. 2: Particle's tracks Fig. 3: Cone air and film thickness Fig. 4: Mesh used Boundary conditions: -Number of cells: 76000 for 1/6th of the whole geometry - Inlet boundary condition: Pressure 120 [bar] - Outlet boundary condition: Average Pressure 1 [bar] - All temperatures are set to 293.15 [K] - Fluid: N-Heptane C7H16 (Vapour and Liquid) - Barotropic Cavitation Model - Turbulence Model: k-e Model (High Reinolds Number) - Fluid incompressible - Differencing schemes: Mars (Second order) Fig. 5: Pressure drop through the seat: Single phase vs Two phases Fig. 6: Isosurface at 3000 Pa, delimiting cone air