Applied Fluid Imaging Laboratory

home research people courses

 

Investigations of the effect of buoyancy on turbulent jet mixing

Student: Michelle Clarke (mariposa: jhu.edu)

(Figures: sample images  quantitative results)  (Publications)

Buoyancy plays a significant role in many practical fluid mixing systems, including atmospheric and oceanic flows, and combustion systems. Density differences in combustion systems can approach an order of magnitude owing to combustion heat release. It is well known that jet diffusion flames, for example, entrain less ambient air than the corresponding non-burning jets. This is likely due to a modification of the shear layer instability between the jet and ambient fluid streams. In simulations of reacting flows, the accurate computation of the combustion process requires first that the mixing phenomena in the presence of wide density differences and buoyancy effects be properly represented.

Prior work in this area has focused mainly on point measurements of species concentrations, and thus density. Here, we apply planar imaging techniques to the mixing of a helium jet (molecular weight = 4) issuing vertically upward into a slow air (MW = 29) co-flow. The laser diagnostic method is Rayleigh scattering, using the 532 nm output of an Nd:YAG laser as the light source. Both the helium jet and air co-flow are filtered to eliminate light scattering from particulates, which would overwhelm the Rayleigh scattering signal. Planar imaging offers two particular advantages for these measurements. First, we can examine the large-scale, coherent structure organization of the mixing field directly, which provides a means to assess the effect of buoyancy on the fundamental jet instability modes; and second, the images facilitate the determination of the scalar gradient and scalar dissipation rate fields, allowing investigation of the effect of buoyancy on the molecular mixing scales.

Figure 1 shows a sample Rayleigh scattering image from these experiments. The imaging window spans from approximately 4 to 12 jet nozzle diameters, d, downstream of the jet exit, and extends approximately 3d to either side of the jet centerline.


Figure 1. (left) A sample raw Rayleigh scattering image plane from the helium-air jet. The helium jet fluid has a smaller scattering cross-section than the ambient air, thus appearing dark in the image. (right) The same image plane, after post-processing to give the mole fraction of helium.

 

The raw Rayleigh scattering image is darkest where the helium concentration is highest, because helium has a smaller Rayleigh scattering cross-section than the ambient air. In this helium-air flow, the raw images are easily processed to yield the helium mole fraction, X(He), as also shown in the figure. The jet Reynolds number for this particular flow is Re = 940. Measurements have also been performed with Re = 1300 and Re = 1940. For all three Reynolds numbers, the jets are categorized as momentum driven according to scaling results reported in the literature.



Figure 2. (a) Radial profiles of the mean helium mole fraction (for axial locations x=4d, 8d, 11d). The curves are normalized by the profile maxima and the profile widths. The profiles deviate noticeably from the Gaussian form (dashed line) at large r values. (b) Linear growth of the scalar radial profile width with downstream distance. Set A has jet Reynolds number 940, set B has jet Re = 1300, and set C has jet Re = 1940. The virtual origin positions and the growth rates differ among the three sets. (c) Decay of the centerline mean helium mole fraction with downstream distance. The curves appear to approach the expected 1/x dependence, though again the three curves are not conincident.

 

Figure 2 shows results for the similarity and scaling results from these flows. Figure 2a gives mean radial profiles of the helium mole fraction, from the Re = 1940 case, at axial positions x = 4d, 8d, and 11d. The profiles are normalized by their respective maxima and widths. It is clear that the profiles are self similar, although they depart from the Gaussian profile expected for non-buoyant jets. Figures 2b and 2c show, respectively, the mean jet growth rate and the mean centerline helium mole fraction decay, as functions of downstream distance, x. Canonical jets grow linearly with x, while their centerline jet fluid concentration decays as 1/x. The curves for the different Reynolds numbers show differences, attributable to a sensitivity of the flow to the differing initial conditions. However, the results demonstrate the linear growth rate very clearly, and are consistent with an asymptotic approach to the 1/x centerline concentration decay. The results of Fig. 2 show that the presence of buoyancy has altered the large scales of the jet in terms of the similarity profiles, even though the scaling properties are consistent with the classification of the flow as momentum driven. Ongoing work involves further exploration of the implications of these results for the large-scale flow evolution, and also the investigation of the small-scale mixing properties of the flow as manifested in, for example, the scalar gradient and scalar dissipation rate fields.

Publications (inquiries: email Michelle Clarke, mariposa: jhu.edu)

 

(top of page)

All materials © 2005 Applied Fluid Imaging Laboratory. Last page update 1.7.05.

home  research  people  courses

JHU    WSE    Dept. of Mechanical Engineering