Measurements with a phase/Doppler interferometric (PDI) single-particle-counting system and the ensemble light-scattering/polarization ratio (ESPR) technique have been carried out in a kerosene spray, introduced vertically upwards into coflowing nonswirling and swirling air flow fields. The droplet mean sizes, determined using the ensemble technique, were found to be smaller than those obtained with the PDI system. Determination of the Sauter mean diameter (D32) for each technique is based on different assumptions which may have an influence on the results. One assumption for the ensemble light-scattering measurements is the form of the droplet size distribution, which in this case incorporates a monomodal log-normal distribution function (LNDF) of prescribed skewness. Measurements taken with the phase/Doppler system have indicated the presence of both positively skewed monomodal profiles with long tails (along the centerline) and bimodal distributions (near the spray boundary). Our results indicate that the LNDF can adequately represent monomodal size distributions found near the spray centerline. Near the spray boundary, the combination of two monomodal log-normal distribution functions is used to represent the measured bimodal size distributions.
To investigate the effect of size distribution on the measurement of droplet size, several different monomodal distribution functions with varying degrees of skewness were examined, as well as a bimodal distribution function. Calculations for the ESPR technique using a log-normal distribution function are evaluated with measured PDI size distributions, and the dependence of polarization ratio on D32 is examined. The results indicate that the geometric mean diameter and the complex refractive index have a significant influence on the polarization ratio, while the geometric standard deviation and the integration limits in droplet size have a small effect. The polarization ratio, however, does not change dramatically if the value of these parameters is kept within the range of interest and the assumed size distribution remains similar to that measured. Therefore, the differences in the measured droplet size cannot be solely attributed to the uncertainties in the size distribution.