Experiments in Fluids 12, 407-412 (1992)

Experiments in Fluids �9 Springer-Verlag 1992

Visualization study of the flow in and around

N. Fujisawa and F. Gotoh

Dept. of Mechanical Engineering, Gunma University, Kiryu, 376 Japan

a Savonius rotor

Abstract. Flow in and around a Savonius rotor has been studied by flow visualization experiments, and the rotation effect is discussed in comparison with the measured pressure distributions on the blade surfaces. It is observed that the flow separating regions on the blade surfaces are fairly reduced by the rotation effect and the flow through the overlap is weakened by the appearance of resisting flow. The former contributes to the torque production of the rotating rotor while the latter acts as a resistance. These phenomena together with the stagnation effect on the front side of the rotor contribute to the power producing mechanism of the Savonius rotor.

1 Introduction

The Savonius rotor has been used widely in small-scale ap- plications of wind energy conversion systems. The flow field of the rotor, which is closely related to the torque and the power performance, has attracted special attention because of the importance of clariflng the power-producing mecha- nism. Hence some work has been done on flow visualization experiments (Clayton 1978; Jones et al. 1978; Bergeles and Athanassiadis 1982; Fujisawa et al. 1987; Fujisawa and Shi- rai 1987; Ushiyama and Nagai 1988; Modi and Fernando 1989). Most of them are studies on the time-averaged flow field independent of the rotor angles. To investigate the pow- er-producing mechanism, it is important to discuss the in- stantaneous flow field in and around the rotor. Concerning this subject, Fujisawa et al. (1987) and Fujisawa and Shirai (1987) have studied the instantaneous flow around a Savo- nius rotor with the smoke-wire technique. They found the appearance of a Coanda-like flow pattern on the advancing blade, which suggests a lift force contribution to the power mechanism of the rotor.

In the present paper, further flow visualization studies are described on the instantaneous flow field in and around a Savonius rotor with and without rotation, and the rotation effect is discussed in comparison with the measured pressure distributions on the blade surfaces Special attention is given to the separation characteristics of the external flow at vari- ous tip speed ratios, since it is expected to have a strong influence on the power-producing mechanism of the rotor.

2 Experiments

2.1 Experimental apparatus

A schematic diagram of the experimental apparatus is shown in Fig. 1. Uniform flow is produced in an open circuit type wind tunnel by a squared nozzle with an outlet width of 900 mm. The test Savonius rotor is placed at the tunnel center and 700 mm downstream of the nozzle exit. The mea- sured velocity distribution at this position with the rotor removed is uniform to within +0 .5% in the central area of 750 m m x 750 mm (Fujisawa and Shirai 1987). As the Savonius rotor is situated within a projected area of 300 m m x 300 mm, the present experiment is performed in a uniform flow.

The test Savonius rotor has two semi-circular blades with an overlap between them. Opt imum design parameters are adopted from previous experiments by Sheldahl et al. (1978), Sivasegaram (1978), and Ushiyama and Nagai (1988): di- mensionless overlap distance G (= a/(2 R)) = 0.15, end plate diameter D o = 1.1 D, and aspect ratio A s ( = H/D) = 1, where D is the rotor diameter and H the rotor height. The rotor shafts between the blades are removed. The physical size of

~z~/,U .~_~__~Advancing blade

Fig. 1. Rotor configuration and geometrical parameters

408

CCD camera Smoke wire r Savonius

/ ~ ; ; ~ rotor

Wind) / I 1 ~ 1 - ~ , IArLaser I / _ ~ Cylindrical

lens Sm~ U D.C motor n~ ~ -

Control '$et t l ing~'---~S mok e ~ B l o w e J valve chamber_~ tank valve

Exhaust valve

Fig. 2. Experimental arrangement for visualizing the external flow of a Savonius rotor

Experiments in Fluids 12 (1992)

Smoke ~ , \ wire,q, I1: I \

Savonius ~ rotor

D.C. motor

Outlet of wind tunnel /

C.C.D camera

Fig. 3. Experimental arrangement for visualizing the internal flow of a Savonius rotor

the rotor is D = H = 300 mm, and the thickness of the blade is 2 ram. It is made of transparent material so that the inside flow field can be observed.

2.2 Visualization of external flow

Figure 2 shows the experimental arrangement for visualizing the external flow field around the Savonius rotor. The main flow is visualized by the smoke-wire technique (Fujisawa et al. 1987; Fujisawa and Shirai 1987), while the wake flow is observed simultaneously by injecting smoke through the nozzle located at the upstream end of the rotor. The observa- tions are made at the mid-plane of the rotor using a light sheet from a 3W argon-ion laser and a CCD camera with a shutter speed of 0.002 s. Experiments are carried out at a flow velocity U = 1.5 m/s.

2.3 Visualization of internal flow

Figure 3 shows the experimental arrangement for visualizing the internal flow field. The smoke-wire is set in the mid-plane of the Savonius rotor. Heating of the wire is provided by a direct current of 40 V to 60 V which is supplied through the rotating contacts at the end plate. The flow is observed by a CCD camera, and the illumination is provided by three lamps of 300 W. The experiment is performed at a flow veloc- ity U = 0.7 m/s.

2.4 Observation of flow separation

The flow on the blade surfaces is visualized by a surface tuft technique. Observations at the same rotor angles are made for every cycle of rotor rotation, which is realized by a pho- tosensor and a stroboscope. The tuft is made of silk thread for embroidery of 0.1 mm in diameter with length 5 ram,

which is unravelled to increase the frequency response of the tuft motion. The frequency response is estimated to about 50 Hz from the motion analysis, which is acceptable for the present observations since the highest rotor frequency is less than 10 Hz. The separation point is judged from the tuft observation, where the centrifugal force is dominant com- pared with the surface flow without any fluctuations with time. The tufts are located every 10 mm in both the stream- wise and spanwise directions.

The separation point is also evaluated from boundary layer calculations using the measured pressure distributions on the rotor surfaces. The experimental details for measuring the surface pressures are given in Fujisawa and Gotoh (1991). According to the transition criteria for fiat plate boundary layers, laminar turbulent transition is not expect- ed here, and hence Thwaites's method for laminar boundary layers is adopted to obtain the separation point. The point of separation is defined by the parameter K ( = O~/v. dUe/ds ) < - 0.082, where s is the streamwise distance, Ue the external flow velocity outside the boundary layers, 01 the momentum thickness of boundary layers, and v the kinematic viscosity of the fluid.

3 Results and discussion

3.1 Flow in and around a still rotor

Figure4a -d shows the instantaneous flow field in and around the still Savonius rotor at various rotor angles of 0 = 0 ~ 45 ~ 90 ~ 135 ~ It presents the visualized external flow (a), the smoke patterns inside the rotor (b), the flow models showing the main features of the flow field (c), and the mea- sured pressure coefficient Cp on the blade surfaces (d). The points of separation and the stagnation are also indicated in the flow models (c), which are obtained both from the visu- alizations and calculations. It is noted that they are in close agreement with each other, which suggests the validity of the

N. Fujisawa and F. Gotoh: Visualization study of the flow in and around a Savonius rotor 409

Fig. 4a-d. Flow in and around a still Savonius rotor (X=0); a visualized flow field; b flow inside the rotor; e flow model; tl surface pressure distribution

present observations of surface flow by the tuft method. It is seen that the external flow on the Savonius rotor separates around both sides of the rotor (a). Therefore, the width of the wake agrees closely with the rotor projected width which varies with the ro tor angles. On the front side of the rotor,

a stagnation point appears on the visualized pictures (a), which moves from the advancing one at small rotor angles 0 = 00-45 ~ to the returning one at large rotor angles 0 = 90 ~ 135 ~ The stagnation force is expected to have a maximum around the rotor angles 0 =45 ~ to 90 ~ which can be expected

410 Experiments in Fluids 12 (1992)

Fig. 5a -d . Flow in and around a Savonius rotor in rotation (X =0.9); a visualized flow field; b flow inside the rotor; c flow model; d surface pressure distribution

from the measured pressure distr ibut ions (d). The difference of the pressure coefficient on both the concave sides (d) causes flow through the overlap, which proceeds from the advancing blade to the returning one. This is also observed in the smoke pat terns inside the ro tor (b). The flow through

the overlap induces the pressure recovery effect on the con- cave side of the returning blade, and hence the drag force is decreased on the returning blade. The flow is strengthened for the rotor angles 0 = 45 ~ to 90 ~ as is expected from the pressure distr ibutions near the overlap (d).

N. Fujisawa and E Gotoh: Visualization study of the flow in and around a Savonius rotor 411

3.2 Flow in and around a rotating rotor 180 =

Figure 5 a - d shows the corresponding flow patterns in and around the rotating Savonius rotor at the tip-speed ratio X (= 0.5 D co~U)= 0.9, giving a maximum power performance [ 9(~ of the rotor, where co is the angular velocity of the rotor. A 8. remarkable change of the flow field induced by the rotor rotation is identified in this figure in comparison with the still rotor (Fig. 4). Significant features observed here are the downward movement of the separation point (c) and the 18 ~ u relative decrease in the pressure coefficient on the convex side of the advancing blade (d). These phenomena can be caused by the occurrence of a Coanda-like flow pattern (a) on the convex side, which appears clearly at small rotor angles of 0 = 0 ~ to 45 ~ The attached flow on the convex side ~ 9(~ tends to separate at large rotor angles of 0=90 ~ to 135 ~ 8- which is due to the outward flow motion at the tip of the advancing blade. This flow is induced by the pressure gradi- ent distributed over the concave side of the advancing blade. 0* The injected flow grows into a vortex circulating in the 186 rotating direction of the rotor, which increases in size down- stream. It is considered that the attached flow patterns of the rotating rotor contribute to the rotating torque of the rotor,

1

as is expected from the pressure distributions (d). On the [ other hand, a relative decrease in the stagnation torque is ] 90* expected here in comparison with the still rotor (Fig. 4), since ~" the relative velocity is decreased on the advancing blade and is increased on the returning one. In addition, the stagnation point moves to the center of the blade due to the rotation (3* effect. It can be seen that the pressure coefficients are de- 180" creased overall by the effect of circulation, which is produced by the rotor rotation. Such a circulation is a steady phe- nomenon, and hence it has been pointed out in the flow visualization experiments by Jones et al. (1978) and Modi ~ 9(~ and Fernando (1989). In comparison with the still rotor, the 8. flow through the overlap is reduced here by the production of resisting flow (b). This flow is expected to reduce the pressure recovery effect on the back side of the returning blade, which is supported by the measured pressure distribu- t~ tions near the overlap (d).

3.3 Effect of tip-speed ratio

a X=O

/ ,

�9 Separation point

0 Stagnation poi~

b X=0.4

/ O

o X=O.9

d X=1.4 )

d 9d

A o'o// / V/UoS/,

,

180 ~ 270* 360" e ,

Fig. 6a-d. Variations of flow separation region with rotor angle 0 (shaded area shows the attached flow region); a X=0; b X=0.4; e X=0.9; d X=1.4

A significant feature of the flow around the Savonius rotor is the appearance of separation control effects on the convex side of the rotor. The variations of the points of separation and stagnation are indicated in Fig. 6 for various tip-speed ratios and plotted against the rotor angle 0. The points of separation and stagnation are obtained from the tuft obser- vations, and they are shown by the blade angle ~o measured from the outer tip of the blade. It is seen that the attached flow region, which is shown by the shaded area in this figure, is increased as the tip-speed ratio increases. The growing rate of the attached flow region is very large at small tip-speed ratios, X = 0 to 0.4, while it is reduced at large tip-speed

ratios, X = 0.9 to 1.4. This result agrees qualitatively with the measured torque performance of the rotor by Fujisawa and Gotoh (1991), which shows an increasing trend up to X = 0.4 and a decreasing one for still larger tip-speed ratios. The deterioration of the torque performance at large tip-speed ratios can be explained by the decreases in the stagnation torque and in the pressure recovery effect by the flow through the overlap. It is noted that the stagnation point comes around the center of the rotor at large tip-speed ra- tios, which shows the dominating effect of the rotating veloc- ity of the rotor in comparison with the free-stream velocity.

412 Experiments in Fluids 12 (1992)

4 Conclusions References

The instantaneous flow in and a round a Savonius rotor has been studied by flow visualization experiments, and the rota- t ion effect is discussed in compar ison with the measured pressure distr ibut ions on the blade surfaces. The results are summarized as follows:

The flow separat ion region on the convex side of the advancing blade is fairly reduced at small ro tor angles by the ro ta t ion effect, which is observed as a Coanda- l ike flow pat- tern. The at tached flow is p rompted to separate at large ro tor angles by the ou tward flow at the tip of the advancing blade, which is induced by the pressure gradient inside the rotor. Near the b lade-over lapping region, the flow through the overlap is resisted by the induced flow due to rotor rotat ion, and hence the pressure recovery effect on the back side of the returning blade is decreased in compar ison with the still rotor.

The a t tached flow region on the convex side of the rotor grows with the t ip-speed rat io and this t rend becomes satu- rated, which contr ibutes to an improved torque performance at small t ip-speed ratios. The deter iorat ion of the torque performance at large t ip-speed rat ios is caused by the de- creases in the s tagnat ion torque and in the flow through the overlap. Such a flow mechanism explains well the torque and the power performance of the Savonius rotor in a wide range of t ip-speed ratios.

Bergeles, G.; Athanassiadis, N. 1982: On the flow field around a Savonius rotor. Wind Eng. 6, 140 148

Clayton, B. R. 1978: Observations of the flow in and around Savo- nius and Darrieus rotors. In: Proc. 1st BWEA Conf., Cranfield, pp. 24-31

Fujisawa, N.; Gotoh, F. 1991: Pressure measurements and flow visualization study of a Savonius rotor. In: Proc. EWEC'91, Amsterdam, pp. 199-203

Fujisawa, N.; Shirai, H. 1987: Experimental investigation on the unsteady flow field around a Savonius rotor at the maximum power performance. Wind Eng. 11,195 206

Fujisawa, N.; Shirai, H.; Saikawa, Y. 1987: Visualization of unsteady flow-pattern around a Savonius rotor. J. Flow Visual. Soc. Japan 7, 107-111

Jones, C. N.; Litter, R. D.; Manser, B. L. 1978: The Savonius rotor- performance and flow. In: Proc. 1st BWEA Conf., Cranfield, pp. 102-108

Modi, V. J.; Fernando, M. S. U. K. 1989: On the performance of the Savonius wind turbine. ASME J. Solar Energy Eng. 11 I, 71- 81

Sheldahl, R. E.; Blackwell, B. E; Feltz, L. V. 1978: Wind tunnel performance data for two- and three-bucket Savonius rotors. J. Energy 2, 160 164

Sivasegaram, S. 1978: Secondary parameters affecting the perfor- mance of resistance-type vertical-axis wind rotors. Wind Eng. 2, 49-58

Ushiyama, I.; Nagai, H. 1988: Optimum design configurations and performance of Savonius rotors. Wind Eng. 12, 59 75

Received July 15, 1991