The 3D Image of the CRZ, Fig. 14, shows that a small part extends back into the swirl burner exit. A convoluted region with a contraction is found in the middle of the CRZ. The top of the recirculation zone is also convoluted, creating an intermediate canal between both regions. Darkened to facilitate its location, the canal in the middle is clearly associated with the PVC.

Fig. 15 shows the combined PVC/CRZ flow with the appearance of two very strong, coherent interacting structures. The PVC matches the portion defined by the canal in the CRZ and confirms the relationship and interdependence between the two structures. One acts as a feedback loop for the other.

3.3. Confined conditions. Square case

The next phase of the experimental work was to perform tests using confined cases as shown in Fig. 4. The operational conditions were similar to that of UNC_1 shown in Table 1. The square open confined case was analysed first in the ‘‘axial” plane, Fig. 16. The

 Fig. 13. Vectorial map used for standard deviation analysis of fluctuation of CRZ boundary. Vortex highlighted. UNC_1 Case, 22.50.

behaviour of the CRZ changed considerably. Opposite to the main CRZ, another CRZ developed, being lifted out from the burner exit completely.

At first sight, this new CRZ seemed to be a toroidal structure which arises from the central recirculation zone, increasing in size at other sections. However, this assumption was soon discarded after a closer analysis of the axial vectors was performed. Relevant eddies in the flow are emphasised. A large central reverse zone is formed. However, two sets of eddies are noticeable both external and internal to the shear layer flow. The external eddies are well

 Fig. 11. (A) Section 11, z = 0.448 D. Two vortices observed. Main PVC and second, new internal vortex. (B) Section 12, z = .488 D. Both vortices join. (C) Section 13, z = 0.529 D. New conjoined vortex. The vortex will remain central in all downstream sections. Case UNC_1. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

 Fig. 14. Recirculation zone 3D image. The intermediate canal is darkened. UNC_1 Case.

Fig. 15. The total flow. Case UNC_1, [x/D].

known, torroidal in nature and arise from the sudden expansion. Indeed they can be largely eliminated by using a quarl on the burner exit [27]. The other set of eddies is located in between the shear flow and the CRZ and was found to be indicative of an initially separate 3D coherent structure. Similar structures were also found by Yazabadi [28] in the exhaust of a cyclone separator.

This new 3D coherent structure is a new separate CRZ (termed CRZ2) that forms close to and interacts with the main CRZ (now CRZ1). Fig. 16 shows three different phase angle sections showing this development. There is also evidence that CRZ1 and CRZ2 reach a height were they merge.

From the ‘‘radial” and ‘‘axial” analysis, the interrelation between the PVC, CRZ1 and CRZ2 had to be found. CRZ1 is very much larger than previously envisaged and highly three-dimentional in nature, being located over a fairly narrow range of phase angle. The region where it interacts with the PVC is considerably smaller. The axial analysis showed evidence of a strong PVC next to the burner exit and persisting up to 0.30D. After this point, the situation becomes unstable and the positioning of the PVC problematic. The PVC is much less noticeable, the circumferential flow is more uniform, whilst the central region contains CRZ1 and CRZ2. The same technique was used for the pyramidal exhaust, 4.B, and sudden confinement, 4.C, Fig. 17.

A feature observed in several of the axial sections, Case 4.B, was the splitting of the CRZ into a strong and weak structure, possibly caused by a strong PVC, in contrast to Case 4.A. For Case 4.C some sections show a CRZ located at the swirl burner exit, still toroidal in nature but somewhat asymmetrical in nature. CRZ1 dominates and CRZ2 was suppressed to such an extent that only small traces appeared in three sections.