Hot Wire Anemometry (HWA) was used to obtain PVC traces and via Fast Fourier Transform (FFT) analyser to determine the strength of the various harmonics. A DISA 55M10 Standard bridge was utilized with 55P16 Hot Wires to obtain an output signal from the PVC in the vicinity of the burner exhaust.

The probe was placed at 1.00 D from the outlet, positioned orthogonally to the tangential axis and 1.00 D for the radial plane. Greater sensitivity was found when orienting the probe in the tangential direction.

A Dantec DISA 55D26 Signal Conditioner was used with a Low Pass Filter configuration, filtering up to 125 Hz to smooth the signal. In order to improve the outgoing signal, a gain of 10 provided a stronger signal for triggering purposes. The system was monitored using a Tektronic DS2024B Oscilloscope at 2 G samples/s, 200 MHz and four channels.

Fig. 2. Schematic plan view of swirl burner geometry and internal inserts.

A DI 2200 FFT Analyzer was used to characterise the flow. The first harmonics was found in all cases and permitted the characterisation of each flow. However, the presence of second and third modes allowed the selection of cases for further analysis. It was recognised that the amplitude of the signal used was arbitrary, but they were consistent between measurements, with the same probe used throughout the experimental studies, enabling a quantitative comparison between cases. The periodic shape of the signal, coupled with correlation measurements confirmed the existence of a precessing movement.

Phase Locked Particle Image Velocimetry (PIV) was used to characterise the field. The hot wire was placed exactly at the outlet where the strongest and most regular PVC signal was found. The HWA trigger output signal was directed to a BNC Model 500 Pulse Generator (PG) which produced a TTL digital pulse every time the conditioned signal reached a certain level. This level was established at 90% of the highest peak found after 5 min of free running. The pulse was then redirected to a Nd:YAG Litron Laser with a wave length of 532 nm, 5 Hz and to a Hi Sense MkII Camera model C8484–52-05CP, with 1.3 MPixel resolution at 8 bits. A 60 mm Nikon lens was used for resolution purposes, with a depth of view of 1.5 mm. The system was seeded by a water nebulizer.

The vortex structures were framed when the precessing flow crossed the same position, allowing a spatial representation of the same phenomenon every cycle. Moving downstream, the flow would be captured at the same location, allowing a 3D representation of the real flow in space for the radial–tangential (‘‘radial” for the rest of the text) plane. A series of sections were analysed every 3.25 mm (spacing of 0.0407 D between them), captured in pairs with a time delay of 20 ls between image. The stability of the CRZ boundary was also examined in the context of time mean measurements  which indicate high levels of turbulence around the CRZ boundaries. This would indicate the importance of the low frequency PVC type oscillations from that due to higher turbulent frequencies. The triggering points were established from the HWA signal at 90% of the highest peaks recorded after 5 min steady running. Triggering values varied from 1.90 to 2.25 V for all cases.

Fig. 4. Square and circular confinements used.

The measurement in the axial–radial (‘‘axial”) plane permitted the acquisition of vortical and circulatory structures, all connected in space. To allow complete visualisation in this plane the hot wire was moved circumferentially around the outlet lip every 11.25, simulating movement of the laser. Each section will be referred according to the position where the HWA was placed during their acquisition. It must be remembered that this is only a spatial correlation between sections, since each of them is time locked. The measurement system is shown in Fig. 3. The objective was to recognise the position of zero velocities where the PVC and CRZ existed so as to define the boundaries of these structures. After acquisition, a frame-to-frame correlation technique was then carried out at 32  32 pixels, with an overlap of 50% between frames to reduce noise.