Surge cycle at 105% Speed

Data showing the full time history of the surge event from breakdown to recovery, from the Kulites mounted upstream and downstream of the fan and in the bypass duct, are presented in Fig 9. As mentioned earlier, these data have been sampled to remove the blade passing frequency and therefore only the large scale flictuations are seen. The figure shows eleven bursts of activity before the fan is recovered. The data show that the initial fbw breakdown has a different signature to the subsequent cycles (Figs. 10a and 10b), with the first burst of activity showing the flow break down initiated by an upstream ‘spike’ feature at 90 degrees circumferentially (Fig 10a). Figure 11 (Ref. 10) shows a schematic of an axisymmetric surge cycle where the surge is initiated at A, the fan jumps to operation with reverse fbw at B. A large pressure rise is required to sustain the reverse flow, which is not available, forcing positive flow again to C where the ft>w cleans up and returns to the primary characteristic at D with pressure recovery, once peak pressure rise is reached at point A the cycle begins again. These features can be seen on the downstream Kulites as the back pressure dropping until the fan recovers allowing peak pressure rise to be achieved before the cycle starts over. The data show the frequency of the surge cycle to be 7.5 Hz, which corresponds to the multiple collapse, and recovery cycles for the average annulus pressure, but superimposed are higher frequencies of the local flow around the annulus. The breakdown from surge into a deep, full span rotating stall has been observed in other tests reported by Pampreen (Ref. 3). As each surge occurs it is noticeable that the initiation becomes more axisymmetric with a non-axisymmetric behaviour once the back pressure falls, Fig. 12.

The 0.18 seconds of the overtip data during the surge cycle are shown in Fig. 13. Three Kulites are shown, the furthest upstream, mid chord and the furthest downstream. The data cover one surge cycle and correspond to one of the intermediate surge cycles. The cycle can be divided into three phases, as indicated on the figure: Phase A is the main ft>w breakdown when surge occurs. Phase B appears to be a non-axisymmetric behaviour akin to rotating stall that develops between surge and recovery. Phase C is a region during which the downstream pressure steadily increases before the flow breaks down again. The time taken for the complete cycle is 0.134 seconds suggesting a frequency of 7.4Hz.

The flow structure can be seen more easily when these data are built into an area distribution. This is achieved by plotting the axial location on one axis and distance travelled by the rotor on the other. Due to the large quantity of data a number of time ’snapshots’ are presented showing the pressure distribution during the surge cycle. These are presented in Fig. 14 (a, b, c). Additionally these figures include the time history of the leading edge gauge with a line indicating the time of that snap shot. Pressure distributions A and B show a fairly clean ft>w structure, as was seen when taking data to define the tip flow structure over the normal (stable) operating range. Pressure distribution C however shows the ft>w breakdown during the surge event. Analysis of the overtip data shows the recovery point is close to the 9th steady state point from peak pressure rise shown on figure 8.

4. Conclusions

For the first time unsteady pressure measurements have been obtained and reported from an overtip Kulite array and Kulites at different axial locations for a transonic wide chord fan during surge and rotating stall. The data provide a clear indication of the operation of the rotor tip during surge at 105% speed and in rotating stall at 60%.

At 60% speed, the overtip measurements show the variation in casing pres­sure distribution through the stall cell and the extent to which the flow field varies during the ’stable’ portion of the rotating stall. The overtip data show two stall cells present, rotating at 63% of the rotor speed. When the throttle was opened, one stall cell dropped out just before the other. Due to the relatively long period of time that the bypass valves take to open it was possible to define the stall recovery point, this cannot be done using steady state measurements.

The surge event lasted 1.4 seconds with overtip data covering a period of 0.18 seconds, or about 25 rotor revolutions. This period included one com­plete surge event, lasting 0.134 seconds, a frequency of 7.4 Hz. Three distinct phases of ft>w could be distinguished; breakdown, rotating stall and recovery. The initial recovery point is just below the normal working line, but the back pressure then rises until the surge point is reached and the surge cycle starts again.

This highly detailed data set is still being evaluated to increase the under­standing of the flow in a transonic fan during rotating stall and surge and how to use these data to improve the fan design.

Acknowledgement

The authors would like to acknowledge the financial support of this pro­gramme of work by the DTI (UK).

References

[1] C Freeman AL Rowe, 1999, Intake engine interactions of a modern large turbofan engine ASME 99-GT-344

[2] EM Greitzer EM 1976 Surge and rotating stall in axial fbw compressors Part I & II Trans ASME Journal of Engineering for Power 98 190-217

[3] RC Pampreen, Compressor Surge and stall, 1993 ISBN 0-933283-05-9

Figure 1. Overtip Kulite arrangement

[4] Tyler JM & Sofrin TG 1962 Axial compressor noise studies Trans SAE 70:309-322

[5] Smith MJT Aircraft noise ISBN 0-521-33186-2

[6] Oppenheim AV & Willsky AS, Signals & Systems ISBN 0-13-809731-3, 1983

[7] I. J. Day NA Cumpsty, The measurement and interpretation of few within rotating stall cells in axial compressors, Journal Mechanical Engineering Science vol. 20 No2 1978

[8] AB McKenzie, Axial few fans and compressors

[9] I. J. Day EM Greitzer NA Cumpsty, Prediction of Compressor Performance in Rotating stall, Journal of Engineering for Power, Jan 1978 vol. 100 No.1

[10] AG Wilson, Stall and Surge in axial few compressors, PHD thesis 1996 Cranfield

Figure 2. Time histories of upstream Kulite during transition from one to two stall cells

TIME (ms)

(Natural) LOG AMPLITUDE PLOTTED

Figure 3. Variation of frequency components with time for 60% speed rotating stall

angle 0-360 (degrees) VFR anti-clack

Figure 4. 60% speed time history of stall initiation

Figure 5. Power spectral density for both rotating stall and post stall data for leading edge Kulite

Figure 6. 60% speed stall recovery – symbols steady state, solid lines unsteady data

Compressible pressure rise coefficient

Flow/mox(Flow100)

Figure 8. AH/U2 against non-dimensional tbw for steady and dynamic behaviour

time (seconds)

time (seconds) Figure 10. Time trace of first surge cycle – (a) upstream Kulite, (b) downstream of fan Kulites

Figure 11. Schematic of the pressure rise characteristic during surge

Figure 12. 5th and 6th surge cycle upstream Kulite showing axisymmetric initiation

Pressure (PS) Pressure (PS) Pressure (PSI)

Figure 13. Zoom of surge cycle from in-passage Kulites

Figure 14. Snap shots of casing pressure distribution during surge event