The conical exhaust case, Fig. 4E, and the sudden confined case, Fig. 4F, showed varied results some of which are shown in Fig. 19. In Case 4.E the PVC is first recorded at the swirl burner exit, extending to 0.49D, then it stops precessing and changes to a centrally located compact vortex core, again wobbling with the PVC frequency. The ‘‘axial” plane showed traces of a very strong vortex structure, whilst a strong separation was observed between CRZ1 and CRZ2. The strength of CRZ2 has decreased whilst being deformed into a lifted, toroidal shape. The toroidal and complex nature of the CRZs is clear, requiring a full three-dimentional analysis. The asymmetry of the shear flow leaving the swirl burner exhaust is also noticeable.

As discussed elsewhere [10,12,33] the formation of CRZs is a function of the axial decay of tangential velocity, hence axial decay of the centrifugal pressure gradient. The conical exhaust and final exit diameter (same as swirl burner) reduce the rate of decay of tangential velocity, thus reducing the axial pressure gradients which drive the formation of CRZs. This gives very complex CRZs similar to those described by Syred and Dahmen [33]. For Case 4.F, some similarity with the square Case 4.C, was evident. However, the strength of the anchored vortex passing into the confinement exhaust is considerably higher than previous cases. This is associated with the small diameter exhaust to the confinement.

Vectorial analysis showed that the bottom of the annular CRZ is irregular and intermittent, being broken up by families of eddies down near the burner exhaust; similar comments apply to the central forward flow.

Fig. 18. Different sections showing the evolution of both structures. (A) CRZ1 and CRZ2 0.00; (B) union between both structures, 135.00; (C) CRZ1 and CRZ2, 168.75. Case 4.D.

 Fig. 19. (A) Section 146.25, conical confinement exhaust, 4.E. (B) Section 112.25, sudden confinement exhaust, 4.F. The strength of the anchored vortex has increased

considerably.

Vortex core precession has been suppressed considerably, confirmed by the weakened hot wire signal; vortex core precession seems to virtually disappear just past the burner exit. However this vortex core can easily start to precess past the exhaust of the confinement, providing there is an appropriate stimulation [13]. The length of CRZ1 now extends to 1.04 D axially, being stable and uniform in shape; CRZ2 was not evident.

In the ‘‘radial” analysis, the existence of a strong central vortex was confirmed. The spiralling movement of the PVC was barely seen in the sudden confinement case, Fig. 20. The conical case, Fig. 4E, still showed traces of spiralling motion in the first planes.

An FFT analysis was performed to verify this conclusion. The open and cylindrical/conical cases showed a FFT signal with three strong harmonics, Fig. 21.

This, if compared with the unconfined case, Fig. 7, shows a more defined signal with reduced amplitude, a consequence of the confinement of the flow and the interaction of secondary structures.

However, when using the sudden square confinement, 4.C, the first harmonic was reduced, whilst the sudden circular confinement, 4.F, showed a substantial suppression of the latter and hence vortex core precession. Sr analysis was performed as previously highlighted and gave the following results at high Re:

Case Asymptotic Strouhal number

As Fig. 4D 0.78

As Fig. 4E 0.80

As Fig. 4F Little precession detected

This shows ways in which the PVC may be suppressed. There is a cost of course in terms of higher pressure drop and the annular shape of the CRZ. Comparison with the square section confinement results is interesting as the losses inherent in firing a swirling flow into a square section give a faster rate of decay of swirl velocity, hence centrifugal pressure gradients and more vortex core precession. 燃气涡轮机英文文献和中文翻译(9):http://www.751com.cn/fanyi/lunwen_51784.html