Blower Poc_dgfinal

Final Year Undergraduate Student Project, 2008

Active Separation Control in a Radial Blower: Proof of Concept Guy Arzuan Faculty of Mechanical Engineering Technion – Israel Institute of Technology, Haifa, Israel Advisor: Dr. David Greenblatt

Abstract This project was undertaken as a proof-of-concept study for active separation control in a radial blower. Experiments were performed on a modified 0.5kW radial blower, where acoustic perturbations were introduced into the impeller housing of fully stalled blades. Increases in plenum pressure of up to 40% were achieved and, based on tuft-based flow visualization, it was concluded that the pressure increases were brought about due to excitation and deflection of the leading-edge separated shear layer. Optimum dimensionless control frequencies were found to be O(0.5), irrespective of the blade orientation or number of blades. Equally important, the maximum control effect (pressure rise) was achieved at only 2% of the fan input power. Backward bladed impeller blades exhibited slightly larger increases in pressure coefficients due to their being fully stalled at zero flowrate conditions. The dependence of blower performance on reduced frequency was remarkably similar to that seen on airfoils at similar Reynolds numbers under periodic excitation. To the best of our knowledge, these data are the first and only demonstration of shear layer or separation control on the blades of a rotating machine.

Final Year Undergraduate Student Project, 2008

Contents 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 2. Objective & Scope. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 3. Experimental Setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 4. Governing parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 5. Experimental Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 6. Discussion of Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 6.1 Effect of Actuator Power. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 6.2 Effect of Reduced Frequencies. . . . . . . . . . . . . . . . . . . . . . . . . . . . 12 6.3 Flow Visualization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .16 6.4 Error estimation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26 7. Main Conclusions . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27 Appendix A - Drawings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28

2

Final Year Undergraduate Student Project, 2008

1. Introduction It has been known for several decades that flow separation from solid surfaces can be controlled by the introduction of periodic perturbations. Initial demonstrations were made on airfoils by introducing acoustic signals into the test sections of wind tunnels. This was typically achieved using conventional loud speakers or specialized acoustic drivers. Typically, significant post-stall lift coefficient increases are observed on airfoils; as much as 50% of the post-stall lift coefficient value. Two basic mechanisms were identified for lift enhancement: forcing of laminar-turbulent transition; and direct control of the separated shear layer (see Greenblatt & Wygnanski, 2000). In recent years, investigators have dispensed with acoustic drivers and prefer to use actuators that are mounted on, or within, the airfoil or wing itself. This approach is far more effective and efficient because the perturbations can be applied only where they are needed and much larger hydrodynamic perturbations are possible with a much smaller power input. Common methods of actuation include surface mounted actuators and zero mass-flux blowing. An extensive review of modern techniques can be found in Greenblatt & Wygnanski (2000). In recent years perturbations have been introduced by means of surface-mounted plasma actuators (e.g. Corke et al, 2008). They are light and easy to affix to aerodynamic surfaces, but their momentum production seems somewhat limited. Despite the rapid increase in active separation control studies, coupled with advances in actuation, control of separation on rotating machinery has received very little attention. This is surprising because blade stall is a major factor leading to reduced efficiency, vibrations, damage and noise. Moreover, fans and blowers consume approximately 20% of the total energy in the European Union and in the US (e.g. Cory, 2004). It is therefore clear that effective control of blade stall would impact dramatically on performance, leading to potential large energy savings. In the context of rotating machinery, all studies appear to be confined to stationary simulated turbomachinery flows (e.g. Hultgren & Ashpis, 2003; Ramakumar & Jacob, 2007). In fact, active separation control on turbine blades invariably involves a simulated static pressure gradient or studies performed in linear cascades. This is because considerable technical difficulties are encountered when attempting to apply active control on rotating fan blades. Firstly, it is difficult to establish, a priori, where to place

3

Final Year Undergraduate Student Project, 2008 actuators on the blades when very little is known about the nature of stall. Secondly, the implementation of actuators would require either electrical or pneumatic modification to the blading system. Thirdly, the actuators themselves would be subjected to significant centrifugal forces.

2. Objectives & Scope The global objective of this study was to assess the viability of separation control on the blades of a radial blower. A radial blower was selected because the flow is known to be partially stalled specific blade configuration throughout their operational envelope. The specific objectives were: 1. To select a small commercial blower on which to conduct the experiments. The main requirement was that the impeller could be replaced with a design suitable for the present investigation. 2. To design and construct a new impeller with a variable set of blades, thereby allowing adjustment for a large number of configurations. 3. To assess the separation control potential of the configuration by introducing periodic acoustic perturbations into the impeller housing. The increase in the blower plenum pressure was used to assess the separation control potential.

4

Final Year Undergraduate Student Project, 2008

3. Experimental Setup In order to fulfill the first objective, a 500-Watt commercial squirrel-cage-type blower was acquired (see fig. 1; courtesy Danciger Laboratories, Technion), where the blower backplate and squirrel-cage impeller could be removed, albeit with some difficulty.

Fig. 1: The O.ERRE 500-Watt commercial squirrel-cage-type used in this study Once the original squirrel cage impeller was removed new back-plate and impeller were designed. The Impeller was designed to be flexible with the option of either 2-bladed or 4bladed configurations. Furthermore, the blades could be deployed in backward bladed configurations and forward bladed configurations form –45° to +45° in steps of 15°. A schematic of the conceptual design is shown in fig. 2a and a photograph of the final 4bladed configuration, installed in the blower, is shown in fig. 2b. Note the holes drilled along the periphery of the back-plate that are were used for changing the orientation of the blading.

5

Final Year Undergraduate Student Project, 2008

Fig. 2a: CAD assembly of the four-bladed configuration and mounting; preliminary design.

Fig. 2b: The final 4 bladed configuration is installed in the blower case. In order to enforce separated flow at the blade leading-edges (innermost part of the impeller), the blades were constructed from thin (2mm thick) plates. In addition, the blades were not curved in any way in order to eliminate the effects of curvature for the purposes of the pilot study. A grating and photovoltaic diode were attached to the motor shaft in order to measure rpm. The output was attached to a frequency counter. The blower was attached to a variable transformer and the supply voltage (Vac) and current (Iac) were monitored.  =Vac× Iac. Power supplied to the blower was calculated according to the relation: W b

A plenum, constructed from wood, was bolted onto the outlet of the blower. The plenum was equipped with four circumferentially distributed static pressure ports. A fine wire mesh was located between the outlet and plenum to equilibrate the plenum pressure. A side view schematic of this assembly is shown in fig. 3 and a photograph of the assembly is shown in fig. 1.

6

Final Year Undergraduate Student Project, 2008

Fig. 3. A schematic of the experimental setup, showing the blower, plemum and speaker. The pressure ports were joined and connected to an inclined alcohol-based U-tube manometer, measuring the plenum pressure pp. The manometer was referenced to the atmospheric pressure p∞ and hence a standard pressure coefficient:

CP ≡

p p − p∞ 1

(1)

2 2 ρ (ω b Ri )

was defined, where ρ is the air density, ω b is the rotational frequency (c.f. rpm) of the impeller blades 2π fb and Ri is the impeller inner radius. A 100mm circular hole was drilled into the downstream end of the plenum and a 60 Wattrated “woofer” acoustic speaker was placed over the hole (see Fig. 4).

7

Final Year Undergraduate Student Project, 2008

Fig. 4: A photograph of the “woofer” acoustic speaker at the end of the plenum. A function generator and amplifier were used to drive the speaker at frequencies ranging between 20Hz to 200Hz. The excitation voltage supplied to the speaker (Ve) was measured using a hand-held digital voltmeter; the speaker impedance Z was measured directly and thus the excitation power was calculated using W e = Ve2 / Z . Based on this calculation, power inputs were limited to 40-watts to avoid damaging the speaker. Thus the blower blades were driven in a fully stalled state, while simultaneously periodic perturbations were introduced in an attempt to control stall. No integer relationship, and hence no phase relationship, was enforced between the blower frequency and speaker excitation frequencies. Initial experiments showed that, for a wide range of speaker frequencies, the pressure within the plenum corresponding to the stalled fan blades increased with increasing speaker power input. Following this a detailed systematic parametric study was conducted, and this is described in section 5. The maximum sound level in the plenum was recorded using a calibrated microphone at 123 dB.

4. Governing parameters

8

Final Year Undergraduate Student Project, 2008 The main parameter governing the control of separation from a stationary airfoil or blade in a two-dimensional flow is the dimensionless frequency:

F+ ≡

fe X U∞

(2)

where fe is the excitation (or forcing) frequency, X is the distance from the position of actuation to the trailing-edge of the wing or blade (here blade chord, c) and U∞ is the freestream velocity. A wide range of different investigation indicate that the reduced frequency which produces the largest improvements lift coefficient CL are mainly in the approximate + range 0.3 ≤ Fopt ≤ 2 . For separation control on flat plates at low Reynolds numbers with + leading-edge separation, the optimum reduced frequency is Fopt ~ 0.5 .

In the case of blades within a radial blower, the characteristic velocity is a vector combination of the rotation speed ω bRi and the flowrate through the blades Q/A, where Q is the volumetric flowrate through the blower and A is the blower outlet area. For the purposes of this investigation, as shown in the previous section, we enforced the conditions of zero flow rate (Q=0) and hence equation (2) can be written:

F+ ≡

fe X X ≡ f* ωb Ri 2πRi

(3) where f * = fe / fb

(4)

is the excitation to fan-blade frequency ratio.

5. Experimental Method Following initial observations described in section 3, a systematic parametric study was carried out in the following manner:

9

Final Year Undergraduate Student Project, 2008 1.

Fan blades were set in a 45° backward-bladed configuration.

2.

The fan was driven using a variable transformer at its design speed, namely

fb=2760rpm (ω b=289 rad/s). 3.

Voltage (Vac) and current (Iac) supplied to blower were recorded.

4.

Using equation (3), a speaker excitation frequency fe was selected, corresponding to

a value in the reduced frequency range 0.3 ≤ F + ≤ 2 . 5.

At this frequency, the power supplied to the speaker was gradually increased from

zero until 40Watts (rms). This was ascertained from a direct voltage measurement described above. 6.

Steps 1 to 3 were repeated for a different frequencies in the range described in item

4 above. 7.

Full selected experiments were repeated for the following conditions:

•

Blower rotation speed fb=1680rpm.

•

Two impeller blades instead of four.

•

Fan blades set in a 30° forward-bladed configuration.

7. Tuft-based flow-visualization was carried out using a high-speed camera. This is described in section 6.6 below. 8. Additional experiments were performed for a wide range of perturbation frequencies, and these are discussed in sections 6.2.

6. Discussion of Results 6.1 Effect of Actuator Power Based on the procedure described in items 1-6 above, the plenum pressure was considered as a function of power supplied to the speaker. Fig. 5 shows the controlled pressure coefficient as a function of the speaker excitation power for a range of relevant reduced frequencies, defined in equations 3 and 4. All data show the same basic trend: namely that the plenum pressure increases with increasing excitation power and then saturates. The maximum overall pressure rise is approximately 40% and occurs at reduced frequencies in the range 0.33 ≤ F + ≤ 0.59 . At lower frequencies and higher frequencies, smaller plenum pressure increases are observed.

10

Final Year Undergraduate Student Project, 2008

3

Fig. 5: Plenum pressure coefficient as a function of actuator excitation power input for a backward-bladed impeller; rotational speed = 2760rpm.

In order to gain a clearer perspective on the relative pressure rise and the relative speaker power,

the

same

pressure

data

is

shown

where

pressure

change,

namely

∆ Cp≡ Cp(controlled)–Cp(uncontrolled) is plotted as a function of the relative actuator power, i.e. the W e / W b (see fig. 6). The graph shows that the 40% pressure rise due to

Cp

control is achieved with an increase of only 2% of the fan power input. For the present case, the flowrate Q is zero and hence the air power is also zero. However, for small Q similar results may be expected and hence the air power pQ will also increase by 40% with an increase in actuator power of only 2%. Moreover, because the actuator power is so low, the overall fan efficiency η will also show approximately 40% improvement, where overall fan efficiency is defined as:

11

Final Year Undergraduate Student Project, 2008 η≡

air power measured fan power + actuator power

=

pQ Vac I ac + Ve2 / Z

(5)

At larger Q no clear conclusions can be anticipated regarding the pressure increase because it is not clear how excitation will affect the flowfield at a different operating point. Nevertheless, the significant increase in fan pressure under fully stalled conditions as presented here is very encouraging.

1 Fig. 6: Change in the plenum pressure coefficient as a function of the relative actuator excitation power input W e / W b for a backward-bladed impeller; rotational speed =

0.8

2760rpm.

The possibility that the pressure rise was caused by an acoustic standing wave was ruled out due to the fact that the acoustic wavelength λ =a/fe was always very much larger than the largest blower dimension. Nevertheless, data was acquired for the entire excitation

12

Final Year Undergraduate Student Project, 2008 frequency range with no fan rotation, and no measurable effect was observed on the plenum pressure.

The identical procedure was followed for the impeller blades oriented in a 30° forwardbladed configuration and the data are shown in figs. 7 and 8. The overall trends are similar with some relatively minor differences. The forward bladed configuration achieves a lower maximum plenum pressure without control and when control is applied the plenum pressure increase is smaller. The net result therefore is a similar percentage increase, namely 40%. The differences in pressure rise can be seen by comparing the stall mechanism of backward- and forward-bladed impellers (fig. 9). All “operating points” enforced in this experiment were for (Q=0), corresponding the data points intersecting the pressure axis. We notice that as Q decreases from its maximum, the forward-facing blades stall visibly and then pressure begins to rise again as Q→0. The backward-facing blades exhibit a much gentler stall with pressure continuously decreasing as Q→0. In the former case the flow attaches partially to the blades while in the latter case the flow becomes more separated. Hence when perturbations are introduced, as was performed in this investigation, the more separated flow shows the largest gain in pressure as the effect on the separated shear layer is largest. Conversely, the less separated shear layer shows a smaller increment. It should be expected, however, that at intermediate Q the effect of perturbations will be greatest on the forward-facing blades in the vicinity of the onset of stall. This could be the subject of a further investigation of this work.

13

Final Year Undergraduate Student Project, 2008

3

Fig. 7: Plenum pressure coefficient as a function of actuator excitation power input for a forward-bladed impeller; rotational speed = 2760rpm.

1

∆Cp

Fig. 8: Change in the plenum pressure coefficient as a function of the relative actuator excitation power input W e / W b for a forward-bladed impeller; rotational speed = 2760rpm.

14

2 0.8

Final Year Undergraduate Student Project, 2008

Fig. 9: Measured blower characteristics fir forward and backward bladed impellers – arbitrary linear units (from ????) . 6.2 Effect of Reduced Frequencies Based on the above results, additional data were acquired for a range of perturbation frequencies corresponding to 0.2 ≤ F + ≤ 2 . At each frequency, the saturation pressure was acquired, i.e. the pressure at which no further increases are observed with additional speaker power. Changes in the pressure coefficient as a function of reduced frequency for

15

Final Year Undergraduate Student Project, 2008 the backward bladed impeller at two fan speeds is shown in fig. 10. The data indicate that there is clearly an optimum frequency in the approximate range 0.3 ≤ F + ≤ 0.6 and that this frequency is independent of the fan speed. At higher frequencies, a reduction in pressure is observed for F+>1 at the lower fan speed.

0.8

Fig. 10: Pressure coefficient rise as a function of reduced frequency for the backwardbladed impeller at two different fan speeds.

The identical experiment was performed, but this time with the forward-bladed configuration at the maxim rpm and the data are shown, together with the backward bladed impeller, in fig. 11. The dependence of the pressure coefficient on reduced frequency is similar but seems to exhibit a sharper peak at F+~0.4. Nevertheless, the data for F+>0.7 is almost indistinguishable, within experimental uncertainty, for the backward and forward

∆Cp

bladed configurations. The figure also shows data acquired with only two blades. In general the pressure rise is slightly smaller, but the over trend as a function of reduced frequency is similar. This suggests that the pressure increases are not dependent on the internal geometry of the blower impeller, but rather that the perturbations are interacting with the shear layer over the blades. Further evidence of this is presented in section 6.3 below.

16

0.4

Final Year Undergraduate Student Project, 2008 A representative Reynolds number for the blower is based on the conditions at boundary layer separation, i.e. at the inner radius Ri. Hence the Reynolds number is:

Re b =

ρωb Ri c µ

(6) On the basis of this definition, representative Reynolds numbers at 1680rpm and 2760rpm are 13,600 and 22,400 respectively

1 Fig. 11: Pressure coefficient rise as a function of reduced frequency for the backwardbladed and forward bladed impellers at 2760 rpm. It is instructive to compare the pressure rise of the blower with the pressure difference (lift)

0.8

across a flat plate airfoil (two-dimensional section) of the same geometry as the impeller blades and at similar Reynolds numbers. Airfoil data acquired by Greenblatt et al (2008) indicating lift coefficient as a function of reduced frequency is shown in fig. 12 and selected corresponding smoke-based flow visualization is shown in fig. 13. In this instance perturbations were supplied at the leading-edge of the airfoil by means of a plasma-based

∆C 17

Final Year Undergraduate Student Project, 2008 body-force. Notwithstanding the different methods of perturbation, the similarities between the pressure rise (figs. 10 and 11) with the lift coefficient increase (fig. 12) are remarkably similar, with both showing an optimum around F+~0.5.

0.5

Fig. 12: Flat plate airfoil lift data as a function of reduced excitation frequency. Perturbations are supplied by plasma-based actuators at the airfoil leading-edge (Greenblatt et al, 2008).

0.4

∆Cl 18

Final Year Undergraduate Student Project, 2008

(a) Baseline

(b) F+=0.42

(c) F+=2.1

Fig. 13: Flat plate airfoil flow visualization for the baseline case and two reduced excitation frequencies. Perturbations are supplied by plasma-based actuators at the airfoil leading-edge (Greenblatt et al, 2008).

0.8

Fig. 14: Pressure coefficient rise as a function of frequency ratio for the backwardbladed impeller at two rotational speeds.

19

Final Year Undergraduate Student Project, 2008 It is evident from equation (3) that the reduced excitation frequency F+ and the dimensionless frequency f * = f e / f b are linearly related by the factor X/2π Ri. The data plotted on this basis are shown in fig. 14. It is evident, therefore, that for a given blower geometry an optimum physical excitation frequency can be established that is only a function of the rotation speed. This result is important as it could be used to establish a relatively simple feed-forward control loop in order to maximize blower pressure. A feedback loop could also be introduced to automate and optimize the system fully.

6.3 Flow Visualization As mentioned in the previous section, it was believed that the measured increases in plenum pressure were due to excitation of the separated shear layer over the leading-edges of the blades. In order to investigate this, 1.5cm long cotton tufts were glued to two opposite blade leading-edges (see the locations of tufts 1-3 in fig. 15). The fan was run at 2760 rpm and filmed under two conditions: (1) no excitation frequency introduced (baseline case) and (2) a control frequency of 70Hz corresponding to F+=0.6 introduced (controlled case). The leading-edge region of the blades was then filmed for both cases with a high speed digital camera (4kH), that was triggered by the photovoltaic diode via the grating mounted on the motor shaft (described in section 3). A white triangle was placed on the impeller back-plate to indicate the relative position for comparisons (see fig. 15).

20

Final Year Undergraduate Student Project, 2008

Fig. 15: A CAD rendering of the impeller showing the location of the cotton tufts. Eight representative phases of the rotation are shown in figs. 16-23. Initial observations indicated that the phase-dependent tuft direction was significantly affected by the perturbations. Before describing the effects it is important to note that the blades tested here are essentially low aspect ratio semispan wings and not two dimensional airfoils. In stationary flows over large aspect ratio wings (AR>4), only the inboard flow, not too close to the tunnel wall or vehicle body can be considered quasi-two-dimensional. In such cases control deflects the shear layer closer to the wall. For the present experiments, where the blades have an equivalent aspect ratio of 2.4, the flow can be expected to be highly threedimensional. On stationary low aspect ratio wings, the tip flow is dominated by a strong vortex. The vortex strengthens when control perturbations are introduced (see Greenblatt & Washburn, 2008). It should thus be appreciated that the present flowfield is far more complex than a simple two dimensional approximation due to the low aspect ratio, rotation of the blades and fully separated nature of the flowfield. It is convenient to discuss the first three phases together (figs. 16-18) as the show a similar tuft behavior. Here we initially consider tuft furthest inboard (number 3– yellow highlight) in order to obtain a qualitative sense of the effect of control. A visual comparison of the

21

Final Year Undergraduate Student Project, 2008 baseline and controlled cases shows that the tuft is deflected closer to the wall by the perturbations. The inherent stiffness of the thread does not allow it to bend fully in the direction of the flow, but the observed deflection indicates that, in a mean sense, the flow must be deflected closer to the wall. In contrast, the tuft located close to the tip is deflected further from the surface, while the midspan tuft shows no meaningful change. Clearly, the effect of control is not uniform along the span and in this sense it is very different from our principal experiences in two-dimensional non-rotating flows. The differences between corresponding tufts of the controlled and uncontrolled cases become successively smaller as the tufted blade moves to the top of the blower housing. This is because the acoustic perturbations are weaker in this region, being further from the speaker and not directly impacted by the plane acoustic wave being introduced at the blower outlet. In the fourth phase (fig. 20), the opposite-bladed outboard tuft is highlighted in purple {this should be purple!} In this instance the tuft near the tip is deflected closer to the blade, while the opposite is true for the inboard tufts. This trend continues from phase 4 though phase 8. It is difficult to explain because it is precisely the opposite trend observed on the opposite blade when it was in the same physical location and indicates a lack of absolute symmetry within the blower casing. Simultaneously, on the opposite blade (discussed above, right hand side tufts in figs. 21-23) shows the same trend, namely that the outboard tuft is closer to the blade with control, while the inboard tufts are furher.

22

Final Year Undergraduate Student Project, 2008

(a) Baseline

(b) Controlled Fig. 16: Individual frames of high speed photographs of the impeller leading-edge region. Tuft number 3, the furthest inboard, is highlighted. Phase 1.

23

Final Year Undergraduate Student Project, 2008

(a) Baseline

(b) Controlled Fig. 17: Individual frames of high speed photographs of the impeller leading-edge region. Tuft number 3, the furthest inboard, is highlighted. Phase 2.

24

Final Year Undergraduate Student Project, 2008

(a) Baseline

(b) Controlled Fig. 18: Individual frames of high speed photographs of the impeller leading-edge region. Tuft number 3, the furthest inboard, is highlighted. Phase 3.

25

Final Year Undergraduate Student Project, 2008

(a) Baseline

(b) Controlled Fig. 19: Individual frames of high speed photographs of the impeller leading-edge region. Tuft number 3, the furthest inboard, is highlighted. Phase 4.

26

Final Year Undergraduate Student Project, 2008

(a) Baseline

(b) Controlled Fig. 20: Individual frames of high speed photographs of the impeller leading-edge region. Tuft number 3, the furthest inboard, is highlighted in yellow; Opposite blade: tuft number 1, closest to the tip, highlighted in purple. Phase 5.

27

Final Year Undergraduate Student Project, 2008

(a) Baseline

(b) Controlled Fig. 21: Individual frames of high speed photographs of the impeller leading-edge region. Both tufts number 1, furthest outboard, are highlighted in purple. Phase 6.

28

Final Year Undergraduate Student Project, 2008

(a) Baseline

(b) Controlled Fig. 22: Individual frames of high speed photographs of the impeller leading-edge region. Both tufts number 1, furthest outboard, are highlighted in purple. Phase 7.

29

Final Year Undergraduate Student Project, 2008

(a) Baseline

(b) Controlled Fig. 23: Individual frames of high speed photographs of the impeller leading-edge region. Both tufts number 1, furthest outboard, are highlighted in purple. Phase 8.

30

Final Year Undergraduate Student Project, 2008 6.4 Error estimation All uncertainties were approximated in the following manner. Consider the function: f = f ( x, y , z , K )

δ f = f ×(

(4)

δx δy δz + + +K ) x y z

(5)

For example: Cp =

∆p 1 × ρ × v2 2

δ Cp = Cp × (

=

ρ × g × h × sin(α ) 1 × ρ × v2 2

δh δv +2 ) h v

(6)

7. Main Conclusions The present investigation established the viability of using periodic perturbations to improve performance in radial blowers whose blades are fully stalled. Increases in plenum pressure of up to 40% were achieved and, based on tuft-based flow visualization, it was concluded that the pressure increases were brought about due to excitation and deflection of the leading-edge separated shear layer. The following specific conclusions were drawn: 1. Optimum reduced control frequencies were found to be F+~0.5 irrespective of the blade orientation, number of blades or fan speed. 2. At optimum control frequencies, the control effect (pressure rise) saturated at relatively low relative power, typically around 2%. 3. Under fully stalled or close to fully stalled conditions, air power and fan efficiency can be expected to increase by up to 40%. 4. Backward bladed impeller blades produced a larger change in pressure coefficients due to their being fully stalled at zero flowrate conditions. 5. The dependence of blower performance on reduced frequency was remarkably similar to that seen on airfoils subject to periodic excitation at similar Reynolds numbers. 31

Final Year Undergraduate Student Project, 2008 6. Tuft-based flow visualization showed a significant effect of control at the blade leading-edges. However this effect did not appear to be symmetric in the absolute frame. To the best knowledge of the authors, these data are the first and only demonstration of shear layer or separation control on the blades of a rotating machine. They show great promise for more sophisticated control techniques. These more advanced techniques can then be exploited to achieve significant energy savings on large industrial machines. This will also have a major impact on the control of separation on axial fans, which also suffer from debilitating blade stall, as well as axial turbines where advanced techniques could be used to reduce compressor stages.

32

Final Year Undergraduate Student Project, 2008

References 1. Corke, T.C, Post, M.L., Orlov, D. M., “SDBD plasma enhanced aerodynamics: concepts, optimization and applications,” Progress in Aerospace Sciences, Vol. 43, 2007, pp. 193–217. 2. Cory, W.T.W., “Fans – just how mature are they,” C631/100/2004 IMechE International Conference on Fans, One Birdcage Walk, London, UK, 9-10 November, 2004, ISBN: 978-1-86058-475-6. 3. Greenblatt, D. and Wygnanski, I., “The control of separation by periodic excitation,” Progress in Aerospace Sciences, Volume 36, Issue 7, pp. 487-545, 2000. 4. Greenblatt, D. and Washburn, A.E., “Influence of Finite Span and Sweep on Active Flow Control Efficacy,” AIAA Journal, Vol. 46, No. 7, 2008, pp. 1675-1694. 5. Greenblatt, D., Göksel, B., Rechenberg, I., Schüle, C., Romann, D., Paschereit, “ Dielectric Barrier Discharge Flow Control at Very Low Flight Reynolds Numbers,” AIAA Journal, Vol. 46, No. 6, 2008, pp. 1528-1541. 6. Ramakumar, K. and Jacob, J.D., “Low pressure turbine blade separation control using plasma actuators,” AIAA Paper 2007-371, 45th AIAA Aerospace Sciences Meeting and Exhibit, January 8–11, 2007, Reno, Nevada 7. Hultgren, L.S. and Ashpis, D.E., “Demonstration of separation delay with glowdischarge plasma actuators,” AIAA Paper 2003-1025, 41st AIAA Aerospace Sciences Meeting and Exhibit, Reno, Nevada, January 6–9, 2003

Acknowledgement The author would like to gratefully acknowledge the staff of the Danciger Laboratories for their assistance at all stages of this project.

33

Final Year Undergraduate Student Project, 2008

Appendix A Engineering Drawings of components manufactured for this project.

34