A static pressure analysis was carried out for all cases shown in Fig. 4 from 0 to 180 mm downstream the rig nozzle, with increments of 20 mm. Both open confinements, 4.A and 4.D, showed similar trends with small fluctuations close to the walls where the shear flow leaving the swirl burner collides with the wall. Differences in centreline static pressure for the other cases were much higher. The reductions in the decay of tangential velocity could be clearly seen when comparing the conical and pyramidal exhaust, 4.B and 4.E. Pressure levels were 42 Pa as compared to 15 Pa. The most significant differences occurred for the sudden confine-

Fig. 20. Section 0, z = 0.00 D, sudden circular confinement, 4.F. Vortex core precession has almost stopped. Colors defined in Fig. 11. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

ments, 4.C and 4.F, with center line pressures of 245 Pa compared to 100 Pa, again confirming the ‘lossy’ nature of the square confinement and effect on tangential velocity levels.

4. Discussion

4.1. Isothermal effects

The co-existence of the PVC and the CRZ was confirmed using Phase Locked PIV, for all cases tested are summarised in Table 2. Their shape and interaction are dependant on the geometry and flow regime. Close to the point of vortex breakdown, the CRZ seems to wobble with an unstable behaviour, being linked to the formation of an intermittent PVC whose traces were often difficult to locate. Higher swirl numbers allowed better visualisation of the PVC and CRZs. The shapes and positions of the structures were influenced by the flow rate, hence Re, since the enlarged shape of UNC_2 compared to UNC_1 demonstrated that higher Re tended to compress the CRZ. This is in agreement with the theory of the appearance of the CRZ [13] as a phenomenon caused by the decay in radial pressure gradients, which in turn cause reverse centreline axial pressure gradients, thus forming the CRZ.

The existence of a canal on the side of the CRZ formed by the PVC confirms their interaction again with the PVC only spiralling for 70–90 up to 0.6D. Afterwards, it reattaches with another vortex which seems to form inside of the PVC, this structure then passes downstream as an enlarged, slowly decaying, slightly wob-

Fig. 21. FFT Analysis typical of cylindrical confinement, 4.B. Similar spectrums were obtained for open (4.A) and conical (4.E) cases. Three harmonics at 55, 110 and 165 Hz.

Summary of all cases studied.

bling central vortex core. Thus, it is confirmed that this swirl burner creates a semi-helical PVC. Even when the PVC is well defined and strong, it still follows a helical shape defined by the CRZ. Thus both structures are caused by the sudden expansion of the swirl burner flow.

The confined cases showed strong evidence of another CRZ(2) formed as a consequence of the reduced rate of decay of tangential velocity arising from the use of a confinement. A corner recirculation zone (ERZ) appears as a consequence of the expansion of the shear flow from the burner passing into the confinement. This has not been extensively quantified.

Particular characteristics observed included the appearance of an enlarged CRZ2 with a similar relationship to CRZ1 when using the circular confinement with no exhaust nozzle. The conical exhaust results showed the presence of CRZ2 with a toroidal shape. This time the attachment between structures was weaker and only a mild region of interaction occurred with CRZ1. The latter was suppressed considerably and its shape was more homogeneous. 

The PVC is of a similar size and extent to that of the square confinement case in the lowest regions of the flow near the burner exit.

The experiments have shown that the PVC can be considerably distorted especially by changes to the exhaust nozzle of the confinement. These effects have been illustrated by other work [25,31], which showed that with a swirl burner furnace arrangement the crescent shaped region of high velocity in the shear flow leaving the swirl burner was enlarged circumferentially, thus clearly altering the signal received from the PVC.