Droplet size distribution estimation for multi-hole type gasoline DI injectors

Abstract

We propose a method to estimate atomized spray droplet size distribution for multi-hole type nozzles for direct injection gasoline (DIG) engines. In the development of injector nozzles, important factors are spray quality such as spray pattern, penetration, and atomization. Previous studies have already presented methods to predict mean droplet size for a DIG injector by using the relationships among atomization, velocity, and geometric characteristics of its nozzle. However, coarse droplets in spray may cause liquid fuel droplets to remain when ignition starts, which may cause much particulate matter or/and hydrocarbon emissions, even if the spray has the same mean droplet size. To describe characteristics of the spray itself, sometimes not only the mean diameter but also the distribution of droplets must be considered. The difference in droplet size distribution has been considered to be caused by the difference in internal flow characteristics. However, the droplet size distribution is difficult to predict, especially for a DIG injector. As DIG injectors have a narrow fuel passage just upstream of the orifice inlet and short orifice length, the internal flow in an orifice tends to become complex, which affects atomization strongly. This means the droplet size distribution needs to be predicted by using detailed internal flow information. We therefore tried to construct a method to predict the droplet size distribution by using the numerically simulated velocity distribution of the orifice internal flow. This paper discusses a method to extend a previously investigated relationship between velocity and droplet size into a velocity distribution calculated by computational fluid dynamics (CFD). We manufactured sample nozzles with different dimensions to produce a curve converted from simulated velocity into an experimental droplet diameter distribution. As a result, a single conversion curve was observed by using the proposed methodology for differently designed nozzles and different experiment conditions. We concluded that the velocity distribution information from CFD can be used to estimate droplet size distribution from the DIG injector nozzle by the conversion curve.