Experimental investigation of flow behavior for scalloped and lobed mixers

Abstract

Particle image velocimetry and planar laser-induced fluorescence techniques are used to measure the flow of scalloped, lobed, and splitter mixers. Special attention is given to the understanding of the details of the flow for the scalloped mixers and the role of the notches in enhancing the mixing process. The comparisons between the three mixers are explained by providing detailed measurements on several axial and cross-streamwise planes. Scalloped–lobed mixers enhance the mixing of the two streams by generating an additional pair of small-scale streamwise vortices not observed in the lobed mixer shear layer. Under the same test condition, the impingement of the inner flow on the mixing duct occurs further downstream than the lobed mixer. For the scalloped mixer, the decay of the cross-streamwise velocity starts at its exit plane, while for the lobed mixer, there is an initial magnitude rise before the streamwise decay.

Introduction

Mixer and thrust augmenting mixer/ejector systems have been a key topic in the development of gas turbines exhibiting propulsive efficiency gains, diminished noise and reduced heat signature in the past. Such gains are realized when compact, passive mixers are incorporated to enhance the mixing rate between the core and bypass streams prior to expansion through the nozzle. It is widely known that one powerful mechanism to achieve high mixing rates within short axial distances is by introducing strong streamwise vortices between the co-flowing streams. However, it is not easy to determine the most effective method to introduce these strong streamwise vortices. Therefore, much research to date concentrates on obtaining the optimum mixer geometry for a specific application, balancing gains created by enhanced mixing versus its pressure loss by-product.

Modifying the trailing edge geometry is the most common method for introducing strong, large-scale streamwise vortices into the core–bypass stream. Typical trailing edge treatments consist of complex corrugated lobe shapes imposed onto an annular ring. Further modifications studied by researchers include asymmetric lobes, alternating deep–shallow lobe penetrations, cutback lobes, asymmetric cutback lobes, and notches or scallops cut into the trailing edge. The results from the study by Frost ¹ indicate modest gains for lobed mixing devices by increasing efficiency as much as 3% at specific operating conditions. It is important to add that mixing enhancement of co-flowing streams is also of great interest to other devices. Examples include low emission combustors, ejectors for high lift or jet noise reduction, infrared suppressor nozzles, and supersonic combustion ramjets.

Early researches^2–4 concentrated in identifying mixer shapes that increased overall gross thrust, optimizing mixer geometries for specific flight operating conditions. The results of such experiments could be directly applied to the engine tested with good confidence. However, the complex fluid structures responsible for the performance gains reported remained largely undetermined for use in later designs. Most recent lobed mixer research has concentrated on elucidating the flow structures which promote enhanced mixing. These are often conducted using measuring techniques that restrict flow speeds, temperatures, and in some cases fluids. Typically, gross thrust changes are not measured directly, and the actual performance when applied to full-scale engines must be inferred into the design process. Often this is through verification by computational fluid dynamics (CFD).

Paterson ⁵ is credited with the first quantitative look into lobed mixer flow structures at simulated flight conditions. Paterson ⁵ incorporated laser doppler velocimetry (LDV) to measure mean and turbulent velocity components in the wake of the lobed mixer. He suggested that mixing is accomplished by three main transport mechanisms. Convection by mean velocity components in the radial and circumferential directions dominates initial mixing followed by large-scale turbulent motions with length scales proportional to that of the lobe height. The exhaust flow is further mixed by gradient-type turbulent diffusion of small-scale turbulent structures.

Elliott et al. ⁶ sought to gain insight into the mixing mechanisms by separating the effects of the streamwise secondary circulations discussed by Paterson ⁵ and the shear layer spanwise Kelvin–Helmholtz vortices present in any free shear layer. This was accomplished using two different mixer configurations. Both mixers have the same wavelength (lobe spacing) but incorporate different slopes leading to the trailing edge. The sloped mixer produced streamwise circulations similar to those observed by Paterson, ⁵ while the convoluted plate (straight section prior to trailing edge) did not exhibit these circulations. From experimental results, Elliott et al. ⁶ conjectured that mixing enhancements were largely from the spanwise normal vortices (formed by Kelvin–Helmholtz instabilities) followed by the increased lobed mixer interfacial contact (as compared to free splitter) and finally from the streamwise vortices.

McCormick and colleagues^7,8 performed extensive flow visualizations and quantitative hot-film anemometry measurements on a forced mixer. Through smoke flow visualizations, the presumed existence of horseshoe-shaped vortices found in the lobe troughs at the mixer trailing edge were verified. He concluded that high levels of mixing were accomplished by the interaction between the normal spanwise vortices and the secondary streamwise circulations.

Previous researchers have predicted that the vortical and turbulent structures in lobed mixing flows play important roles in mixing enhancement. Only in recent years, detailed quantitative experimental techniques have been developed that are capable of revealing the evolution and interaction of whole-field unsteady velocity and vorticity in lobed mixing flows. Using modern whole-field flow diagnostic techniques such as planar laser-induced fluorescence (PLIF), particle image velocimetry (PIV), stereoscopic particle image velocimetry (SPIV), and dual-plane stereoscopic PIV techniques, Hu et al. ⁹ conducted a series of studies to address the evolution and interactive characteristics of various vortical and turbulent structures in lobed jet mixing flows instantaneously and quantitatively. Hu et al. ⁹ concluded that enhanced mixing was accomplished through the breakup of streamwise vortices from large to small scales without losses in intensity. These experiments have done much to further our understanding of mixing structures in lobed mixer–ejector flows. Yu et al. ¹⁰ have performed a study of scalloped trailing edge treatments. Velocity measurements were conducted for a two-dimensional (2D) mixer using a two-component LDV system. In it, they suggest that an additional pair of streamwise vortices is formed initially due to the scalloped lobes which provide mechanism for enhance mixing.

Wright et al. ¹¹ conducted detailed experimental and computational investigation of the effects of scalloping on the mixing mechanisms of a scaled 12-lobed turbofan mixer. They found out that lobe scalloping in high swirl conditions resulted in better mixing and improved pressure loss over the mixer without the scallops but at the expense of reduced thrust. In an experimental investigation, Lei et al. ¹² showed that the more rapidly decaying streamwise vorticity presented a more effective mixing process because of the smaller scale but more developed vortices.

Previous researches have all showed the possible benefit by incorporating scallops into the trailing edge through increased mass flow pumping or thrust increases. The objective of this research is to provide additional insight into the flow structures generated by scalloped trailed edge treatments through high-resolution PIV measurements and qualitative mixing measurement of a passive scalar (PLIF). The modification here along with scalloping is that the mixer is not 2D but three-dimensional (3D), and mixing is confined in a circular mixing duct typical to its traditional application.

Experimental configuration and instrumentation

All experiments were performed in a closed circuit water tunnel. The water tunnel was modified from its original configuration to produce the desired core and bypass flow rates to the lobed mixer and test section shown in Figure 1. A free surface supply plenum, clear test section, and return make up the three segments of the water tunnel. Water was supplied to the plenum where it passed through a series of screens and a honeycomb straightener prior to entering the test section. Plenum walls accelerated the fluid to the test section using a 6:1 contraction ratio. The plenum supply pump was powered by a 2-Hp motor coupled to a 20-cm impeller that produced up to 3.4 m³/min of water flow through the test section.

Figure 1.

(a) Water tunnel schematic and (b) bypass flow conditioner.

In addition, the water tunnel was modified to generate a shear flow between two co-flowing streams. The goal was to design and build a piping system that would supply higher velocity core flows, creating velocity ratios on the order of 2:1 relative to the bypass stream. It was determined that an axial-style pump, similar to the original water tunnel pump, was necessary to generate the core flow momentum. Figure 1(a) shows the core flow pump location that generates maximum average velocity of 2.1 m/s and is controlled by an AC to DC motor speed controller. Core flow is plumbed into the water tunnel through 10.16-cm pipe.

The test section consists of a clear glass bottom and sidewalls measuring 0.508 m × 0.381 m × 1.524 m (H × W × L). A clear acrylic open-ended cylinder forms the mixing duct and is located in the center of the water tunnel test section. Lobed mixers attach on the upstream side and are counter-bored to pilot onto the mixing duct assuring both the cylinder and nozzle center are concentric with each other.

Three separate mixer nozzles are studied. They include a splitter nozzle (no trailing edge treatment), conventional lobed, and scalloped–lobed trailing edges. The lobed and scalloped–lobed mixers incorporate the same basic dimensions with the scalloped mixer further modified by removing approximately 50% of the lobe sidewalls. The splitter nozzle was machined from 10.16-cm cast acrylic pipe and tapered to thin trailing edge.

A total of eight lobes spaced equally on 45° intervals make up the trailing edge of the nozzles. Detailed lobe dimensions are given in Figure 2. Lobe height and width measure 41 and 15.9 mm, respectively, with corresponding inner and outer penetration angles measuring a conservative 15° each (30° included). The area ratio between the bypass flow and the core flow is 1.97. Inner diameters of the core and mixing duct are 108 and 197 mm, respectively. The mixing duct measures 91-cm long giving room to measure six nozzle diameters downstream comfortably. In addition, Figure 2 shows the pictures of scalloped, lobed, and splitter mixers and locations of lobe peak, midplane, and valley planes.

Figure 2.

(a) 2D versus 3D lobed mixers and (b) lobed and scalloped–lobed mixer dimensions.

The Reynolds number defined by equation (1), where ν is the kinematic viscosity; velocity scale (V_avg) is the average velocity of the mean core ${\overline{\mathrm{V}}}_{\mathrm{core}}$ and bypass ${\overline{\mathrm{V}}}_{\mathrm{bypass}}$ streams; and the characteristic length is the nozzle diameter, D. Also, in the literature, the lobe height, H, is used as the characteristic length scale (2D mixers) and is included here. Reynolds numbers for this study based on D and H are 33,060 and 14,470, respectively

$$\mathrm{Re}=\frac{\left[{\mathrm{V}}_{\mathrm{avg}}=\left(\frac{{\overline{\mathrm{V}}}_{\mathrm{core}}+{\overline{\mathrm{V}}}_{\mathrm{bypass}}}{\mathrm{2}}\right)\right]·\left(\mathrm{D\; or\; H}\right)}{\nu }$$ (1)

Mean core and bypass velocity profiles are obtained through PIV data obtained three nozzle diameters upstream of the mixing interface for lobed and scalloped mixers. The average velocity of the core and bypass streams are given in Table 1 for lobed, scalloped, and splitter mixers.

Table 1.

Average mixer core and bypass velocities.

	${\overline{\mathrm{V}}}_{\mathrm{core}}(\mathrm{m}/\mathrm{s})$	${\overline{\mathrm{V}}}_{\mathrm{bypass}}(\mathrm{m}/\mathrm{s})$	Vavg (m/s)
Lobed mixer	0.44	0.215	0.33
Scalloped mixer	0.42	0.214	0.32
Splitter mixer	0.45	0.215	0.33

Standard PIV is used for the present experiments. The working fluid is water. For particle traces, hollow glass spheres having specific gravity of 1.06 and mean diameter of 10 µm were used. Cross-correlation is performed using the Hart algorithm with erroneous vectors removed and replaced by mean and standard deviation filters. The 16 × 16 pixel interrogation areas (∼2.5 mm × 2.5 mm axially and ∼3.4 mm × 3.4 mm streamwise in physical space) were used. Post-processing of velocity data was accomplished using developed LabVIEW software.

Results

Streamwise mean velocity profiles and flow visualizations, cross-streamwise velocity, turbulence intensity, and vorticity fields are presented in this section.

Streamwise mean velocity profiles

Axial velocity profiles for lobed and scalloped mixers are shown at selected axial locations on two streamwise planes through the lobe peak and middle planes in Figures 3 and 4, respectively. Axial velocity is normalized by V_avg. Profiles are presented at locations (a) Z/D = 0.25, (b) 1.0, (c) 2.0, (d) 3.0, and (e) 4.0. Scalloped velocity profiles are right of the centerline, while lobed profiles are located left of the centerline.

Figure 3.

Normalized axial velocity profiles on streamwise plane through lobe peak: (a) Z/D = 0.25, (b) Z/D = 1.00, (c) Z/D = 2.00, (d) Z/D = 3.00, and (e) Z/D = 4.00.

Figure 4.

Normalized axial velocity profiles on streamwise plane through midplane: (a) Z/D = 0.25, (b) Z/D = 1.00, (c) Z/D = 2.00, (d) Z/D = 3.00, and (e) Z/D = 4.00.

Figure 3 shows that at Z/D = 0.25, scalloped and lobed profiles have similar behavior for 0 < Y/D < 0.25. For 0.075 < Y/D < 0.5, there is a noticeable decrease in the scalloped profile versus lobed profile due to convection azimuthally outward as will be shown in cross-stream velocity vector fields. For Z/D = 1.0, a similar trend exists at Y/D ≈ 0.5. In this location, the scalloped mixer indicates lower velocity relative to the lobed by similar reasoning. Increasing the axial distance to Z/D = 2.0, axial velocity profiles show the scalloped nozzle velocity outside the core (0.35 < Y/D < 1) to be marginally greater than the lobed case. This trend continues for Z/D = 3.0 and 4.0 axial locations. The unmixed core region shrinks drastically for both lobed and scalloped–lobed mixers versus axial distance. For Z/D = 1, the unmixed region stretches from Y/D = 0 to 0.3. By Z/D = 4.0, this unmixed region has diminished to zero.

Figure 4 presents axial velocity profile on the midplane halfway between lobe peak and valley. At Z/D = 0.25, the convected core flow through the scallops is readily apparent by the local maximum in axial velocity at Y/D ≈ 0.55. The lobed profile does not exhibit this peak at Z/D = 0.1. However, at Z/D = 1.0, the lobed profile indicates a secondary peak at Y/D ≈ 0.85. This peak arises from the large-scale cross-streamwise vortices which are more concentrated near the edge of the shear layer (Y/D ≈ 0.85) for the lobed case compared with scalloped mixer. These stronger concentrated vortices take the higher speed flow near the core of the lobed mixer (lower values of Y/D) and mixes it with the low speed flow regions of flow near the edge of the shear layer (larger values of Y/D). Although this mixing exists for the lobed mixer, but the cross-streamwise vortices are more elongated and more spread for the scalloped mixer compared with the lobed mixer. This local maximum in axial velocity at Y/D ≈ 0.85 remains visible in the lobed case at Z/D = 2.0. Axial velocity profiles at Z/D = 3.0 and 4.0 approach the profiles obtained on the streamwise plane passing through the lobe peak as previously described.

In Figure 5, axial velocity profiles along a streamwise plane passing through the lobe valley show that at Z/D = 0.25, the shear layer is clearly evident at Y/D ≈ 0.4 in both lobed and scalloped profiles. Interestingly, the scalloped mixer exhibits a slightly higher axial speed, by a normalized value of about 0.1, versus the lobed mixer in the outer regions (0.4 < Y/D < 0.95). The local peak in the lobed midplane profile at Z/D = 1.0 is now evident in the lobed valley profile at Z/D = 2.0.

Figure 5.

Streamwise PLIF dye visualization on midplane for (a) lobed and (b) scalloped–lobed and on valley plane for (c) lobed and (d) scalloped mixers and (e) region of interest for cross-streamwise measurements.

Flow visualization

Streamwise flow visualization on midplane and valley planes for lobed and scalloped–lobed regions and the region of interest for cross-streamwise measurements are shown in Figure 5. Raw images presented are obtained using PLIF method. Here, flow moves left to right, and the end of the mixer is visible in the left side of each image. On the midplane, scalloped mixer visualization indicate core flow is moved through the scallops and begin mixing on this plane upstream of the lobed mixer. Note the region above the core flow close to the exit plane is filled with dye in the scalloped case but not in the lobed. The impingement point on the mixing duct wall is visible and is in agreement with velocity results which indicate delayed impingement by the scalloped mixer. On valley planes, the visualized flow shows the point at which the core flow extending from the lobes in peak and midplanes “roll” into the valley plane. This plane forms the furthest distance (1/2 lobe period or 22.5°) dye travels from the lobe peak plane to the lobe valley streamwise plane. The scalloped mixer show small pockets of dye in this plane upstream of similar pockets observed in the lobe case. Between Z/D = 4 and 5, regions outside the core flow fill with mixed dye rapidly, indicating flow is completely turned into the valley plane.

Similar trends between lobed and scalloped mixers exist on the peak streamwise plane. Radial spreading of the core flow in the splitter case is drastically lower than either lobed or scalloped mixers observed by the PLIF technique.

Cross-streamwise velocity fields

Axial cross-planes provide an excellent means of describing the mixing characteristics of the lobed nozzles. Through measurement of cross-stream fluid velocity and molecular mixing of dye from the core to bypass streams, insight into the mechanisms by which enhanced mixing occurs is obtained. Streamwise vorticity is obtained by post-processing of the cross-stream fluid velocity components. The decay of streamwise vorticity measurement is arguably the best indicator of the mixing performance and location of enhanced mixing region of the nozzle.

Axial results are presented for one lobe period at two downstream locations. This region is illustrated by the blue wedge-shaped outline on the raw PLIF image in Figure 5(e). Cross-stream velocity components of the splitter mixer are not given as there is no dominant streamwise structure present. Although measurements have been taken and results have been analyzed for Z/D values of 0.1, 0.25, 0.5, 0.75, 1, 1.25, 1.5, 1.75, 2, 3, and 4, only the results at Z/D values of 0.1 and 4 are presented here.

Ensemble averaged cross-stream fluid velocity vectors normalized with respect to V_avg for lobed and lobed–scalloped nozzles are given in Figures 6 and 7. Ensemble average velocities indicate the presence of the large-scale dominant flow structures. Also included in these figures are sample instantaneous velocity fields. Instantaneous velocity fields provide snapshots in time of smaller scale turbulent structures and eddies that appear intermittently in the flow. Figure 6 gives the velocity field obtained closest to the mixer exit plane at Z/D = 0.10. Images obtained closer to the exit plane exhibited the nozzle outline illuminated by side scatter laser light preventing velocity measurements. Readily observed in the ensemble averaged fields are two shear layers formed by the large-scale counter rotating streamwise vortices. The scalloped field indicates slight bulging at the top and bottom of each shear layer, while the lobed field remains vertical. Instantaneous captures for scalloped and lobed reveal additional small scale but intense vortices shed into the wake.

Figure 6.

Normalized ensemble average and instantaneous cross-stream velocity field at Z/D = 0.10: (a) ensemble average vector field, scalloped–lobed; (b) instantaneous vector field, scalloped–lobed; (c) ensemble average vector field, lobed; and (d) instantaneous vector field, lobed.

Figure 7.

Normalized ensemble average and instantaneous cross-stream velocity field at Z/D = 4: (a) ensemble average vector field, scalloped–lobed; (b) instantaneous vector field, scalloped–lobed; (c) ensemble average vector field, lobed; and (d) instantaneous vector field, lobed.

Spreading of the large-scale streamwise vortices occur in the enhanced mixing region, Z/D = 0 < enhanced mixing < Z/D = 2.0. The ensemble average in the enhanced mixing region shows increasing azimuthal spreading with increased axial distance. In addition, the results reveal that the cross-stream velocity impingement for the lobed mixer on the mixing duct wall occurs at Z/D = 1.0. Cross-stream impingement for the scalloped nozzle is delayed and occurs further downstream at Z/D = 1.25. This impingement point of the hot core gasses in practical gas-turbine applications is important as local hot spots on the liner may lead to failure. Results suggest that early azimuthal spreading across the scallops dissipates some of the outward radial momentum, thus delaying and weakening the impingement.

The results at Z/D = 3 and 4 for the cross-stream vector fields past the enhanced mixing region show rapid velocity magnitude decay in cross-stream in these planes. The large-scale structures continue to expand in size until reaching the outer limits of the periodic wedge-shaped boundary. This boundary interface is actually the boundary created by the neighboring lobe’s pair of streamwise vortices. It is important to note that even in these locations, the instantaneous captures at these regions contain small-scale vortices with relatively undiminished intensity.

Turbulence intensity contours

Normalized cross-stream velocity magnitude and turbulence intensity contours at Z/D = 0.1 and 4 are given in Figures 8 and 9, respectively. Cross-stream velocity magnitude is normalized by the average core velocities for each mixer. Based on lobe penetration angle of 15°, the cross-stream component can be estimated by equation (2). Resulting maximum cross-stream velocity should measure approximately 26% of the average core velocity. These figures indicate maximum cross-stream components between 26% and 27% of average core velocity

$${\mathrm{V}}_{\mathrm{cs}(\max )}\approx \overline{{\mathrm{V}}_{\mathrm{core}}}·\sin \left({\theta }_{\mathrm{out}}\right)$$ (2)

Figure 8.

Normalized ensemble average cross-stream velocity magnitude and turbulence intensity for scalloped and lobed mixer at Z/D = 0.10: (a) normalized cross-stream velocity magnitude, scalloped–lobed; (b) turbulence intensity (%), scalloped–lobed; (c) normalized cross-stream velocity magnitude, lobed; and (d) turbulence intensity (%), lobed.

Figure 9.

Normalized ensemble average cross-stream velocity magnitude and turbulence intensity for scalloped and lobed mixer at Z/D = 4: (a) normalized cross-stream velocity magnitude, scalloped–lobed; (b) turbulence intensity (%), scalloped–lobed; (c) normalized cross-stream velocity magnitude, lobed; and (d) turbulence intensity (%), lobed.

Cross-stream velocity contours at Z/D = 0.1 illustrate initial differences between scalloped and lobed mixers. The contours for the scalloped nozzle indicate that the outward velocity is dissipated in regions closer to the bottom of the lobe, adjacent to scallops. Also noted is the radially inward dip near the peak of the lobe caused by the physical boundary of the lobe restricting outward momentum and thickening of the boundary layer in the lobe valleys or troughs. With increasing axial distance from the exit plane, the wedge-shaped cross-stream contours are shown to move radially outward as described in the previous section. From these contours, the maximum cross-stream velocity is extracted and plotted versus axial distance in Figure 10(a). Interestingly, the lobed mixer indicates an initial increase in cross-stream velocity, maximizing at Z/D = 0.5, before decaying. Contrary, the scalloped mixer does not exhibit this phenomenon and begins its decay from the first measured plane, Z/D = 0.10. The exact cause for the initial increase in cross-stream velocity for the lobed mixer is unclear.

Figure 10.

(a) Normalized cross-stream velocity and (b) area average turbulence intensity.

For turbulence intensity contours, initially at Z/D = 0.1, the maximum locations of turbulence intensity are located in the shear layer between streams, concentrated in the lobe troughs. It is observed that the scalloped mixer turbulence intensity is spread azimuthally outward, while the lobed mixer remains in the shear layer only. With increasing axial distance, regions of high turbulence intensity concentrate toward the center of the lobe and begin expanding outward (radially). The overall trend of cross-stream turbulence intensity is given in Figure 10(b), by calculating the area averaged turbulence intensity versus axial distance. The scalloped mixer on average exhibits higher turbulence intensity over the lobed mixer (approximately 1% over entire axial range). The rate of increase for the lobed mixer in the enhanced mixing region is greater than that of the scalloped. However, early mixing caused by the scallops allows for higher overall turbulence intensity. Peak values of maximum turbulence intensity for lobed and scalloped mixers are 11.2% (Z/D = 1.50) and 11.8% (Z/D = 1.75).

Streamwise vorticity contours

Normalized streamwise vorticity ${\varpi }_{\mathrm{z}}$ is defined by equation (3). D is the reference diameter of the nozzle, 0.108 m

$${\varpi }_{\mathrm{z}}=\frac{\mathrm{D}}{{\mathrm{V}}_{\mathrm{avg}}}·\left(\frac{\partial \mathrm{v}}{\partial \mathrm{x}}-\frac{\partial \mathrm{u}}{\partial \mathrm{y}}\right)$$ (3)

Figure 11 shows normalized streamwise vorticity contours on axial cross-planes at Z/D = 0.1 and 4. The distinct differences between scalloped and lobed mixers are observed. For scalloped case, four regions of high streamwise vorticity (and low) exist and are located in the lobe peak (two) and lobe valley (two). Conversely, two narrow vertical regions of intense streamwise vorticity exist for the lobed mixer. Note that the contour scale decreases with axial distance to reveal the change in shape of large-scale streamwise vortices. Without decreasing the scale, the change in vorticity is difficult to articulate due to their rapid exponential-like decay. Initial maximum streamwise vorticity at Z/D = 0.10 is 15.2 and 12.4 for lobed and scalloped mixers, respectively.

Figure 11.

Normalized streamwise vorticity at Z/D = 0.1 for (a) scalloped and (b) lobed mixers and at Z/D = 4 for (c) scalloped and (d) lobed mixers.

At Z/D = 0.25, slight spreading in all directions for the lobed and scalloped mixers were observed. Four distinct regions for scalloped and two for lobed remained visible on this plane. A rapid decrease in streamwise vorticity to 12.1 and 10.6 for lobed and scalloped mixers was observed. At Z/D = 0.50, the streamwise vorticity contours began to bulge outward azimuthally at the lobe peaks. This growth pattern continued on successive axial planes.

Figure 12 illustrates the decay rate of streamwise vorticity by plotting the maximum normalized streamwise vorticity versus axial distance. In the enhanced mixing region (0 < Z/D < 2), the maximum streamwise vorticity decays to approximately 26% (both mixers) of its initial strength.

Figure 12.

Normalized maximum streamwise vorticity decay.

Conclusion

The tests conducted here use 3D forced mixers that are confined in a constant area mixing duct which is more representative of its traditional application. Forced lobed and scalloped shapes when compared with simple splitter plate are dominant enhancers of the mixing process. High mixing rates immediately downstream of the lobed and scalloped mixers are clearly shown by the current results. The major mechanism of mixing for lobed and scalloped mixers is the existence of strong cross-streamwise vortices, where the high-speed fluid of the core is mixed with the lower velocity fluid in the bypass stream. It is also suggested from axial velocity results that the scalloped mixer diffuses the outward radial cross-stream velocity by introduction of lower momentum bypass fluid through the scallops.

Cross-stream velocity vector fields and streamwise vorticity distributions indicate the presence of an additional weaker vortex pair near the core of the mixing layer for the scalloped–lobed mixer compared with the forced lobed mixer. This additional weak vortex pair contributes to the bulging of the streamwise vortices observed in the cross-stream velocity vector fields.

The increase in cross-stream velocity for the lobed case between Z/D = 0.1 and 0.5 is a phenomenon quite different from that reported for 2D and 3D mixers exhausting to large, unbounded volumes. Additionally, it was observed that the impingement location for the scalloped mixer is further downstream than that of the lobed mixer.