Binyet Emmanuel e-mail: emmanuelbinyet@hotmail.com Wang Jun e-mail: wangjhust@163.com Fluid Machinery Department, Huazhong University of Science and Technology, 1037 Luoyu Road, Wuhan 430074, Peoples Republic of China The aim of the present paper is to investigate ways of improving the efciency of a six-bladed Savonius rotor. The efciency of Savonius machines is low because of the negative drag exerted on the convex part of the blades and also because the torque of standard Savonius rotors varies substantially during one rotation and therefore affects the self starting of the rotor at certain wind angles. Improvement of the efciency of the Savonius rotor is car- ried out by increasing the number of blades and also by prevent- ing the wind from impinging on the convex parts. The latter can be done by hiding the convex part of the blades behind a shield or a vane. The present paper shows the results of two-dimensional computational uid dynamics (CFD) computations, indicating a promising increase of the power coefcient from 0.3 to 0.5. [DOI: 10.1115/1.4004549] Keywords: Savonius, vertical axis wind turbine, CFD 1 Introduction In developing countries with low industrialization, lack of elec- trical energy is a major barrier to their progress. Among the vari- ous alternatives, wind turbines of the Savonius type t well because they are easy to manufacture and have low cost. More- over, drag rotors start producing energy at very low wind speeds and can still continue producing electricity at very high wind speeds. This is the reason why this type of turbine is still popular despite its low coefcient of power. The aim of the present paper is to improve the efciency of the Savonius rotors by working on its two main drawbacks: (1) unsta- ble torque and (2) drag caused by the pressure on the convex part. 2 Rotor Concept and Computational Fluid Dynamics (CFD) Model These rotors work under the principle of drag difference: the concave part has a high drag coefcient and the convex part has a low one (Fig. 1). The difference in drag creates a torque M, which makes the rotor spinning with an angular velocity x, giving a me- chanical power P m Mx. FLUENT 6 together with GAMBIT were used for the CFD modeling (Fig. 2). The two-dimensional unstructured mesh is formed by two zones of almost 100,000 triangular cells each. The rotor is inside the moving zone (delimited by a circular interface), which is rotating at a constant angular velocity. The static zone sur- rounds the moving zone and represents a constant wind ow. The SIMPLE algorithm (with default under relaxation factors) to- gether with the turbulence model RNG k-e were used to solve the discrete RANS over the nite volume mesh. The aim of the simulations is to estimate the power curve (C P k). For the standard Savonius rotors, due to the presence of the overlap, the tip speed ratio (TSR) reads k x 2d e =2v, where v denotes the wind speed and x denotes the rotational velocity, e describes the overlap; e d
1 6 gives the best performan- ces, and the efciency is maximum when k is a little below 1 (according to the references). For the six-bladed rotor: k x D 2 =v. The power coefcient reads C P Mx= q 2 v 3 S, M being the overall torque at the axis. The reference surface S is a (vertical) rectangular surface of height h (1 m) and the rotors diameter D as width. The heightdi- ameter ratio for the rotor used in the present simulations is two (then D 0:5m). For the six-bladed rotor, S m 2 D m h m . For the Savonius rotor (see Fig. 1), S m 2 2d e m h m . 3 CFD Computations 3.1 Savonius Rotor. The conventional Savonius rotor does not work efciently for the following two reasons: there is a nega- tive torque resulting of the wind exerting pressure on the convex part of the blades and also because the torque produced varies (from high to null) along the rotation. The computation results shown in Fig. 3 conrm this statement. Figure 3 shows clearly the biggest drawback of the standard Savonius rotor: the torque varies and becomes null along the rota- tion. High torque value occurs only once every 180 deg. In addi- tion, the pressure on the convex blade is all long of the rotation, hindering the advancing blade generating an important vortex (negative drag). The overlap ratio that denes the empty space at the center plays a key role because it reduces the effect of the pressure on the convex part and allows wind owing to the other side, thus impinging on the other concave part. For a better efciency of the turbine, the wind blowing in the rotor should produce torque continuously. Then, more blades can be added in order to capture all the stream tubes owing through the rotor. We therefore chose to use the six-bladed rotor concept, which seems to have optimum characteristics (according to Shi- kha et al. [2]). 3.2 Six-Bladed Savonius Rotor. The height, curvature, and arc length (inner and outer diameters) of the six-bladed rotor have a strong inuence on its efciency. The present work has focused on the innerouter diameter ratio. It has been found that a 35% ra- tio achieves higher power output. The wind inside the rotor should be able to ow from blade to blade, so that more blades experience high pressure on their con- cave part, thus raising the overall torque value. The rotor on the left of Fig. 4 has the lowest efciency because the air is trapped inside the blades; the part of the blade near the shaft yet does not produce any signicant torque. It has to be kept in mind that the power output of a nonoptimized multibladed rotor would even be lower than a standard Savonius rotor [4]. The velocity at the center of the rotor is much higher than in the free stream and the pressure is also lower. The air is owing from blade to blade, generating pressure on four concave parts. The inner ow is prominent, giving a more efcient transformation of the kinetic energy into pressure at the concave faces. At B4, the ow is not efciently captured by the blade, leading to a loss in kinetic energy (high exit velocity). The separation bubbles in the wake of the two convex parts (B5 and B6) illustrate well the negative drag. We also notice that there is not much ow in the zone behind B2. There is also a separation bubble in the wake of B2 and B3, result- ing in a lower pressure zone at B3 and B 4, thus lowering the over- all efciency (Fig. 5). Contributed by the Solar Energy Division of ASME for publication in the JOURNAL OF SOLAR ENERGY ENGINEERING. Manuscript received November 18, 2010; nal manuscript received June 28, 2011; published online October 18, 2011. Assoc. Editor: Spyros Voutsinas. Journal of Solar Energy Engineering NOVEMBER 2011, Vol. 133 / 044503-1 Copyright VC 2011 by ASME Downloaded From: http://solarenergyengineering.asmedigitalcollection.asme.org/ on 06/04/2014 Terms of Use: http://asme.org/terms Fig. 2 CFD model, two-dimensional grid, and boundary conditions Fig. 1 Main geometric factors of the Savonius rotor, physical model, and our prototype Fig. 3 Flow over time for a wind speed of 3 m/s 044503-2 / Vol. 133, NOVEMBER 2011 Transactions of the ASME Downloaded From: http://solarenergyengineering.asmedigitalcollection.asme.org/ on 06/04/2014 Terms of Use: http://asme.org/terms 3.3 Improving the Efficiency Using External Devices. The goal is to suppress the pressure exerted on the convex part. Four congurations were investigated: six-bladed rotor, six-bladed rotor with wind full shield and half shield, six-bladed rotor with stator (Fig. 6). Computations, both steady and unsteady, were run for different TSR. The computations results showed that 1. the six-bladed rotor without shielding has a reduced ef- ciency but still higher than a standard Savonius rotor 2. the two shielded rotors and the rotor with stator have a high efciency due to the suppression of the negative drag. 3.3.1 Rotor With Partial Shield. There is no wind inow im- pinging on the convex parts of the blades anymore, suppressing the negative drag. Convex parts are now located in a low pressure zone. Therefore, the power at the rotation axis is higher. Some of the air particles tend to avoid the concave blades due to the high pressure. Because of the little ow in the blade channels, the energy conversion is mostly due to the two rst blades. 3.3.2 Rotor With Full Shield. The main difference with the previous conguration is that the stream lines are directed in the concave parts; the air particles are forced into the blades. There is a high velocity zone at the inlet of the shield due to the lower pres- sure inside the blade channels and the higher pressure in the neck. 3.3.3 Rotor With Stator. Mobile shields have to adjust their position to ensure that the inlet is perpendicular to the wind at any time, thus yielding a complicated design and stability problems. The stator is then a better solution; it brings a very high efciency at optimum angles. However, at some angles, the stator would become an obstacle to the wind ow, meaning a huge drop in the efciency. The stator presented here can improve the efciency at the major wind angles (0, 90, 180, and 270) but would decrease the efciency at minor angles and especially if the wind is parallel to the vanes. The power output therefore depends on the wind direction. Figure 7 (right) shows the computation results for a wind angle of 45 deg; the Cp is very low because the vane is an obstacle to the ow in the concave blade: concave faces cannot efciently capture the wind. Furthermore, an important part of the incoming ow is impinging Fig. 4 CFD ow eld (pressure and streamlines) for different inner diameters under a wind speed of 3 m/s and a TSR of 0.5 Fig. 5 Steady ow for a wind speed of 3 m/s and a TSR of 0.2 and 0.5 (above) and for a wind speed of 6 m/s for a TSR of 0.2 and 0.5 (below) Journal of Solar Energy Engineering NOVEMBER 2011, Vol. 133 / 044503-3 Downloaded From: http://solarenergyengineering.asmedigitalcollection.asme.org/ on 06/04/2014 Terms of Use: http://asme.org/terms on the convex parts and high pressure is applied on the perpendicu- lar vanes. At the optimum angle [Fig. 7 (left)], the drag of this con- guration is substantial; the pressure in front of the rotor is signicantly higher. As a result, streamlines tend to avoid the vanes but most fail to do so and are forced in the rotor yielding a high mass ow rate in the blade channels. The ow slows down in front of the stator, but the total pressure is mostly conserved in the main ow direction. The ow inside the rotor is prominent as four con- cave parts produce torque. However, the stator is not optimum yet since pressure is still exerted on the convex parts (B5, B6). 4 Comparison Between the Savonius Rotor and Other Configurations The prototypes from references have similar geometric factors than the present CFD model: overlap ratio of 0.15 [4] and 0.2 [6] as well as semicircular blades. The only different factor is the heightdiameter ratio, which is lower in Refs. [4,6]. There is a quite good overall agreement between the present CFD computa- tions and those of Refs. [4,6]. Discrepancies could be due to Reyn- olds effects (inuence on ow separation) as well as scale effects. Fig. 7 Rotor with stator for a wind speed of 2 m/s and 6 m/s for a TSR of 0.8 and 0.2 Fig. 8 k-Cp curves, comparing the efciency Fig. 6 Flow for a wind speed of 3 m/s, rotor with partial and full shield for a TSR of 1 and 0.5 044503-4 / Vol. 133, NOVEMBER 2011 Transactions of the ASME Downloaded From: http://solarenergyengineering.asmedigitalcollection.asme.org/ on 06/04/2014 Terms of Use: http://asme.org/terms Indeed, models for wind-tunnel tests, because of their reduced scale (height of 1 m [4] and 0.6 m [6]), are more sensitive to 3D effects and vertical ows that both have a negative inuence on the efciency (Fig. 8). The results of the computations showed that the optimized six- bladed rotors aerodynamic efciency is slightly higher than the standard Savonius rotor over a nite range of lambda (00.65) and reaches a Cp max of 0.3. The multibladed rotors (lower operating k) have a lower rotating speed than the standard Savonius rotors. Since the wind is pushing on the blades at each time step, the blades tangential velocity cannot excess the wind speed. The rotor with stator shows a very high efciency (almost reaching 0.5). However, the wind angle was optimum, and the scenario reects an ideal case. In real conditions, wind would always change and efciency would then decrease. This con- cept is the most suitable for the design of an improved drag tur- bine. The rotor with half shield also shows good aerodynamic characteristics (max efciency reaching 0.4). On the other hand, this conguration is more theoretical because of the chal- lenging design due to the adjustment of the shield according to the wind. 5 Conclusion 1. According to the results of the computations, the shielded rotors have a high efciency. Using the six-bladed rotor alone gives a max efciency of 0.3, which is slightly higher than the standard Savonius rotors and should also present bet- ter starting abilities. Numerical simulations indicate that it is possible to improve the efciency of the Savonius concept. 2. The prototype showed that using those external devices (shields, vanes) is complicated for both design and opera- tion. Indeed, the inlet has to remain steadily oriented toward the wind (perpendicular to the wind at any time), thus removing the main advantage of the vertical axis wind tur- bines, which is the independency of the wind direction. In application, using the rotor alone is more convenient. 3. The coefcients of power given by CFD are overestimated because the two-dimensional model reects ideal cases. We expect the coefcients of power on prototypes to be reduced by a 20% factor (0.24 and 0.4 instead of 0.3 and 0.5). References [1] Publication of the World Energy Council Survey of Energy Resources, 2007. [2] Shikha, Bhatti T. S., and Kothari D. P., 2003, Wind Energy Conversion Sys- tems as a Distributed Source of Generation, J. Energy Eng., 129, pp. 6980. [3] Guo, X. S., Wind Energy Technology (Beijing Industrial Chemistry Press, 2007), pp. 8487. [4] Sheldahl, R. E., Blackwell, B. F., and Feltz, L. V., 1978, Wind Tunnel Per- formance Data for Two- and Three-Bucket Savonius Rotors, J. Energy, 2, pp. 160164. [5] Martin, J., 1997, Techniques de lIngenieur, traite Genie energetique Energies eoliennes, Jean Martin Energies eoliennes Techniques de lIngenieur, B8585, p. 11. [6] Yan, L., Hara, Y., and Hayashi, T., 2009, A Wind Tunnel Experimental Study on The Overlap Ratio of Savonius Rotors, Acta Energiae Solaris Sinica, 30, pp. 226231. [7] Fujisawa, N., and Gotoh, F., 1994, Experimental Study on the Aerodynamic Performance of a Savonius Rotor, J. Sol. Energy Eng., 116, pp. 148152. [8] Menet, J.-L., and Bourabaa, N., 2004, Increase in the Savonius Rotors Ef- ciency Via a Parametric Investigation, http://www.mendeley.com/research/ increase-savonius-rotors-efciency-via-parametric-investigation/. [9] Menet, J.-L., Valdes, L.-C., and Menart, B., 2001, A Comparative Calculation of the Wind Turbines Capacities on the Basis of the L-r Criterion, J. Renew- able Energy, 22, pp. 491506. [10] Ogawa, T., 1984, Theoretical Study on the Flow About Savonius Rotor, J. Fluids Eng., 106, pp. 8590. [11] Nakajima, M., Iio, S., and Ikeda, T., 2008, Performance of Double-Step Savo- nius Rotor for Environmentally Friendly Hydraulic Turbine, J. Fluid Sci. Technol., 3, pp. 293300. Journal of Solar Energy Engineering NOVEMBER 2011, Vol. 133 / 044503-5 Downloaded From: http://solarenergyengineering.asmedigitalcollection.asme.org/ on 06/04/2014 Terms of Use: http://asme.org/terms