OCR Text |
Show Fi~. 5 sh0v.:s the axial velocity component in the flame. Mean velocity is indicated by the S?lid.hnes whIle the mean +/- the standard deviation is indicated by the dashed lines. By V1eWIn~ .the data. in this way, it is hoped that the reader may get some sense of the width of the probabih.ty denslty function of the data. This is not meant to imply that the distributions are symmetnc-there are many regions in the flow where the distributions are highly skewed-but simply to provide some perspective as to the range over which the distribution might extend. These data are also useful in providing a means to depict the range of uncertainty in the mean val~es. Assuming that the values were sampled from a nonnal population, it can be shown (Miller and Freund, 1977) that the 95% confidence interval for the mean extends to +/- 6.2% of the standard deviation when using an ensemble of 1000 independent samples. That is to say, if !he experiment ~ere repeated indefinitely, 95% of the mean values should lie within a band that IS 6.2% of the WIdth between the dashed lines in Fig. 5. In practice this is not the case due to day-to-day fluctuations in the experiment, asymmetry in the PDFs, etc. Experience has shown that the repeatability band of the mean is closer to +/- 20% of the standard deviation. As such, the proper interpretation of these data is that the mean values are as depicted but with an uncertainty interval of about 20% of the range between the dashed lines, and that the width of the velocity distributions is approximately indicated by the distance between the dashed lines, but that the distribution is not necessarily symmetric about the mean value. The data of Fig. 5 shows six distinct regions in the flowfield. First is the fuel jet or spray region. !he reason why this is referred to as the "fuel jet" instead of just "spray" is that the jet which Issues from the atomizer is, in fact, a two-phase, high momentum jet, initially with equal mass in each ph~se. As such, the effect of this jet on the aerodynamics of the flame is expected to be much different than that of a simple pressure atomized spray with its concomitantly lower momentum. The fuel jet is indicated by the vertical hatching used to denote data bias due to droplets. This region contains the highest velocities and fluctuations in the field. Note that on the centerline at z = 25 mm the standard deviation in the spray velocity is about 10 m/s with a mean value near zero. Since 95% of the samples are contained within +/- three standard deviations of the mean (at this location the PDF is fairly nonnal) the observed velocities at this "stagnation point" range from between about -30 m/s to +30 m/s. This is a feature of the flow which will be evident in much of the data. The mean value often provides a poor (misleading) characterization of the flow, i.e., in many regions the flow is dominated by fluctuations. The fuel jet is approximately centered along the nominal 60- cone of the spray and is discern able from the air jet until about z = 100 mm. After this, the bimodal behavior of velocity vs. radius disappears. This is confmned by the measurement of unseeded data rate used to detennine the biased regions of the flow. Past z = 100 mm, the data rate dropped below the 100Hz used as an indicator of droplet bias. The second region of th~ flow is the main combustion air jet. This jet issues from the annulus around the nozzle and spreads outwards, concentric with the fuel jet. The fluctuations in the main air jet velocity are much smaller than in the fuel jet, typically 20% of the local mean, except in the shear region adjacent to the fuel jet. The axial velocity entrance profile of the main air jet is somewhat different from what might be expected in simple swirling flows, showing an increase at the edge of the nozzle. This increase is due to entrainment from the fuel jet and will be discussed further in the next section. The third region is at the outer edge of the main air jet and is characterized by the absence of a mean axial velocity. This is the external entrainment region and is composed of gases which have been recirculated from the upper regions of the flame by a toroidal vonex ring adjacent to the furnace windows and floor. The innennost edge of this vonex is visible via density gradient · -6- |