You are on page 1of 7

EUROPEAN WIND ENERGY CONFERENCE AND EXHIBITION 2009, MARSEILLE, FRANCE

CFD ANALYSIS OF TWO-BUCKET SAVONIUS ROTOR USING FLUENT PACKAGE

RAJAT GUPTA , BIPLAB KUMAR DEBNATH , RANJAN DAS


1

CORRESPONDING AUTHOR: PROFESSOR & DEAN SRC, DEPARTMENT OF MECHANICAL ENGINEERING, NATIONAL INSTITUTE OF TECHNOLOGY, SILCHAR, SILCHAR, CACHAR, ASSAM, INDIA 788010, MOBILE NO: +91 9435171942. EMAIL ID: r_guptanitsil@yahoo.com ASSOCIATE AUTHORS : M.TECH SCHOLAR, THERMAL ENGINEERING, DEPARTMENT OF MECHANICAL ENGINEERING, NATIONAL INSTITUTE OF TECHNOLOGY, SILCHAR, SILCHAR, CACHAR, ASSAM, INDIA 788010 MOBILE NO: +91 9864907742. EMAIL ID: biplab.nitsilchar@gmail.com. : PHD SCHOLAR, DEPARTMENT OF MECHANICAL ENGINEERING, INDIAN INSTITUTE OF TECHNOLOGY, GUWAHATI.

2,3

SUMMERY:

In this paper, a Computational Fluid Dynamics (CFD) analysis using the Fluent package 6.2 was carried out to predict the performance characteristics like, power coefficient (Cp) and tip speed ratio (TSR) of a two-bucket Savonius rotor for without and with 16.2%, 20%, 25%, 30% and 35% overlap conditions. The rotor was 20 cm in height and 8 cm in diameter. A two-dimensional unstructured computational grid was developed for the rotor model. A k- turbulence closure model with enhanced wall treatment function was chosen. A second-order upwind discretisation scheme was adopted for pressure-velocity coupling of the flow. Grid independence test was also conducted to have the best meshing and accuracy of the computational results. The values of Cp obtained computationally were then compared with that of the values of Cp obtained experimentally for all the overlap conditions. The experimental values of Cp for with and without overlap conditions were obtained from the tests conducted previously in an open-circuit subsonic wind tunnel available in the department. The comparison of experimental and computational study is quite encouraging. Keywords: Computational Fluid Dynamics, Savonius rotor, power coefficient, tip speed ratio. .

1. INTRODUCTION: The finite storage of conventional fossil fuels have given us the scope to think about their unavailability in future, and hence, we now have to reduce or replace the usage of fossil fuels by other means. Renewable energy, especially, the wind energy is a very good option for this. Among all the renewable energies present in the world, the wind energy is known to have the highest potential and is environmentally benign too. It has been estimated that roughly 10 million MW of energy are continuously available in the earths wind. According to the Global Wind Energy Council (GWEC), wind energy developing countries (more than 70) have taken the net wind energy installed capacity to a record high of 74,223MW [1]. Worldwide five nations: Germany, USA, Denmark, Spain and India, have 80 percent of the world's installed wind energy capacity. Wind energy continues to be the fastest growing renewable energy source with worldwide wind power installed capacity reaching 14,000 MW [2]. Despite constraints facing supply chains for wind turbines, the annual market for wind continues to increase at a staggering rate of 32% [1]. Greenpeace predicted that about 10% of the required electricity worldwide could be supplied by wind energy by the year 2020, and with the improved technology along with superior economics; experts predict wind power would capture 5% of the world energy market by the year 2020 [3]. The simplest of the modern types of the wind energy conversion system is the Savonius rotor, which was invented by S.J. Savonius in late 1929 [4]. It was very efficient in low wind velocity. It was consisted of two half cylinders facing convex sides to each other so that almost an S-shaped cross section was formed. After testing 30 different models in the wind tunnel as well as in the open air, Savonius found that best of his model had 31% efficiency, and the maximum efficiency of the prototype in the natural wind was 37%. Following Savonius, Bach [5] made some investigation of the Savonius Rotor and the related machines. The highest measured efficiency was 24%. McPherson [6] reported the highest efficiency of 33%, and the maximum power coefficient obtained by Newman [7] was only 20%. Modi et al [8] reported a power coefficient of 0.22. Sivasegaram S. [9] had also done works on Savonius rotor where he studied the efficiency of this rotor under tunnel conditions and reported the maximum power coefficient of 0.25. The objective of the present study was to evaluate the power coefficients of a two-bucket Savonius rotor at without and with 16.2%, 20%, 25%, 30% and 35% overlap conditions using FLUENT CFD

software and then compare them with the power coefficients at the corresponding overlap conditions obtained from a previous experimental work on the same rotor [10]. 2. PHYSICAL MODEL OF THE TURBINE:

Fig2.1: Without overlap.

Fig2.2: With overlap.

The experimental model was nothing but a two bucket Savonius rotor [10]. The buckets of the model had provision for change of overlap ratio using nuts and bolts. The overlap ratios could be varied from without to with 16.2%, 20%, 25%, 30% and 35%. The thickness of each bucket was 5mm. The brief description of the wind tunnel, experimental procedure and results were given in the paper [12]. The two dimensional view of the model without and with overlap ratio was shown in the fig. 2.1 and fig. 2.2 respectively. 3. COMPUTATIONAL ANALYSIS: 3.1 Physical Model: The CFD analysis was performed for Reynolds 6 number of 10 for the model. The computational domain was the top view of four boundaries of the tunnel test section along with the physical model of the rotor. The Fig. 3.1 showed the computational physical domain of the rotor at 20% overlap. The computational domain was discretised using twodimensional unstructured grid. Velocity inlet and

pressure outlet condition were taken on the left boundary and the right boundary respectively.

accordingly, inlet conditions had been updated. These were done according to the experimental observations. 4. CFD SOLVER DESCRIPTION: Since in the VAWT, the buckets or blades rotate in the same plane as the approaching wind it was felt that a two dimensional simulation was sufficient for this typical analysis [11]. FLUENT 6.2.16, Computational Fluid Dynamics (CFD) solver was utilized to solve the governing equations sequentially using the control volume method. The simplest and the most widely used two-equation turbulence model is the k- model that solves two separate transport equations to allow the turbulent kinetic energy and its dissipation rate to be independently determined. The governing equations were integrated over each control volume to construct discretized algebraic equations for the dependent variables. These discretized equations were linearized using an implicit method. The iterations were done to achieve a converged solution. The modeled transport equations for realizable - model are and in the

Fig.3.1: Physical model, boundary conditions and computational domain of the rotor at 20% overlap The top and bottom boundaries for the sidewalls had symmetry boundary conditions. Moving wall boundary conditions were taken for the buckets. For the different overlap conditions, the physical model geometry was changed within the domain, and

In these equations, Gk represents the generation of turbulence kinetic energy due to the mean velocity gradients. Gb is the generation of turbulence kinetic energy due to buoyancy. C1, C2, C1, C2 and C are constants. k and are the turbulent Prandtl numbers for and , respectively. Sk and S are user-defined source terms. All the variables including turbulent kinetic energy k, its dissipation rate were shared by the fluid, and the volume fraction of each fluid in each computational volume was tracked throughout the domain. The default values of the C1, C2, C, k, , defined in the Fluent were 1.44, 1.92, 0.09, 1.0 and 1.3. 5. SOLUTION METHODOLOGY: Appropriate solver, viscous model, material properties, realistic boundary conditions and solution controls provided for this problem were used as follows.

5.1. Solution Specifications: Solver : Steady, turbulent (k- ) model. 3 -5 Material : Air ( = 1.225 Kg/m , = 1.7894 x 10 Kg/m-s). Operating Condition: Atmospheric pressure (1.0132 bar). Boundary conditions: Inlet : Velocity inlet. Sides : Symmetry. Buckets : Moving Wall. Outlet : Pressure Outlet. Solution Controls : Pressure Velocity Coupling: SIMPLE. Under Relaxation Factor: 0.7(Momentum). Discretization: Momentum (Second Order Upwind). Initialization: Inlet condition.

Residual Monitors: 0.001 (continuity, x-velocity, y-velocity, k, ). Force Monitors: Drag Coefficient and Moment (Torque) Coefficient.

6. RESULTS AND ANALYSIS: After the convergence of the solution, a steady value of the drag coefficient (Cd) and torque coefficient Ct (or Cm) was found. From the equation of the torque coefficient and power coefficient we can get:

5.2: Grid independence test: The correctness of the result is greatly dependent upon the resolution of the grid. But, we can refine the grid density up to a certain limit beyond which, refinement dose not effect significantly on the result obtained. This limit is called the Grid Independent Limit (GIL). In this typical analysis, coefficient of drag (Cd) was taken as the criteria for the test, and the grid refinement was done until the required steady value was not obtained. The various level of refining used to conduct this study are shown in Table 5.1. Figure 5.1 shows the variation of Cd with the number of nodes, taken in the test. Refining Level 1 2 3 4 5 6 7 8 9 10 No. of Nodes 3398 3869 4927 5171 6130 9238 10725 12338 17566 24455 No. of Cells (Triangular) 6534 7456 9544 10020 11938 18082 21008 24210 34574 48256

Ct =

T 1 AV free 2 D 4
(1)

Cp =

Now if we put the value of torque from the equation (1) into the equation (2) we can get,

1 AV free 3 2

(2)

Cp =
=

1 D Ct 2 V free
Cp Ct

(3)

(4)

In the above equations (3, 4) of power coefficient and tip speed ratio, all the variables including torque coefficient were known after fluent analysis. So we could find out the values of power coefficient (Cp). Further the difference in the experimental and computational power coefficients had been found out in the form of the standard deviation from the data set for each overlap condition. The formula used for power coefficient was:

Table 5.1: Overview of all the grids used for the grid independence study
Grid Independence Test

(5) Where, (6)

21.5 21 20.5 20 Cd 19.5 19 18.5 18


C p

Now experimental and computational power coefficients with respect to tip speed ratio and the percentage of deviation between them at different overlap conditions were shown from Figures 6.1 to 6.12.
0.2 0.18 0.16 0.14 0.12 0.1 0.08 0.06 0.04 0.02 0 0 0.1 TSR Experimental Computational) 0.2 0.3

17.5 0 5000 10000 15000 No. of Nodes 20000 25000 30000

Fig 5.1: Grid independence test for 2-bucket Savonius rotor.

Fig.6.1: Variation of Cp with TSR at without overlap.

Fig.6.2: Percentage deviation of computational Cp from the experimental Cp at without overlap


0.2 0.18 0.16 0.14 0.12 0.1 0.08 0.06 0.04 0.02 0 0 0.1 TSR Experimental Computational 0.2 0.3

Fig.6.6: Percentage deviation of computational Cp from the experimental Cp at 20% overlap


0.3 0.25 0.2 C p 0.15 0.1 0.05 0 0 0.1 0.2 TSR Experimental Computational 0.3 0.4

C p

Fig.6.3: Variation of Cp with TSR at 16.2% overlap.

Fig.6.7: Variation of Cp with TSR at 25% overlap.

Fig.6.4: Percentage deviation of computational Cp from the experimental Cp at 16.2% overlap


0.25 0.2 C p 0.15 0.1 0.05 0 0 0.1 0.2 TSR Experimental Computational 0.3 0.4

Fig.6.8: Percentage deviation of computational Cp from the experimental Cp at 25% overlap


0.25 0.2 0.15 0.1 0.05 0 0 0.1 0.2 TSR Experimental Computational 0.3 0.4

Fig.6.5: Variation of Cp with TSR at 20% overlap.

C p

Fig.6.9: Variation of Cp with TSR at 30% overlap

deviation of 6.54% were shown in the Figs.6.7 and 6.8 respectively. At 30% overlap condition, maximum Cp and deviation were shown in the Fig.6.9 and 6.10, which were 0.23 at a TSR of 0.2378 and 5.43% respectively. But from Figs.6.11 and 6.12 we got maximum Cp of 0.22 at a TSR of 0.4189 and deviation of 4.53% at 35% overlap condition respectively. 7. CONCLUSION: For the present analysis it had been found that the prediction of power coefficient for the two-bucket Savonius rotor using Fluent Software matched well with the experimental results for each (with and without) overlap conditions. Maximum power coefficient was 0.24 at 20% overlap condition. The percentage deviation for power coefficient at without overlap condition was 2.44% and at 16.2% overlap condition was 1.87% whereas, 3.64% for 20% overlap condition. For 25%, 30% and 35% overlap conditions; the deviations were 6.54%, 5.43% and 4.53% respectively. Hence it can be concluded that the error of the computational work were within the range of 1.5% to 6.5% which satisfactorily validates the work as well. 7. NOMENCLATURE: D d e A T Overall rotor diameter. Density. Bucket diameter of the rotor. Viscosity. Overlap. Swept area. Torque. Angular Velocity. Power coefficient. Experimental Tip speed ratio (). Torque coefficient. Free stream velocity. Standard deviation. Average power coefficient. No of overlap / no of data points in each overlap.

Fig.6.10: Percentage deviation of computational Cp from the experimental Cp at 30% overlap


0.25 0.2 0.15 0.1 0.05 0 0 0.1 0.2 TSR Experimental Computational 0.3 0.4 0.5

C p

Fig.6.11: Variation of Cp with TSR at 35% overlap.

Cp TSR Ct Vfree n Fig.6.12: Percentage deviation of computational Cp from the experimental Cp at 35% overlap. From Fig.6.1 it was found that at without overlap condition maximum Cp was 0.179 at a TSR of 0.275. From Fig.6.2 deviation of computational Cp from the experimental one was found as 2.44% at that overlap. From Fig.6.3 for 16.2% overlap condition, maximum Cp was 0.19 at a TSR of 0.22 and Fig.6.4 showed that deviation was 1.87%. From Fig.6.5 and 6.6 for 20% overlap condition, maximum Cp of 0.24 at a TSR of 0.259 and deviation of 3.64% were obtained respectively. At 25% overlap condition maximum Cp of 0.234 at a TSR of 0.3189 and

9. REFERENCES: [1] Herbert GMJ, et al. A review of wind energy technologies, Renewable and sustainable energy reviews. Renew Sustain Energy 2007; 11:111745. [2] Motiwal S, et al. Wind energy is the fastest growing energy source, Project monitor, July, 2007. [3] Global wind energy market as of 2008. Press release of Global Wind Energy Council; (GWEC), October, 2008. [4] Savonius SJ. The S-rotor and its application,

Mechanical Engineering, 1931; 53:333-338. [5] Bach G. Investigation concerning Savonius rotors and related machines. Translated into English by Brace Research Institute; Quebec, Canada, 1931. [6] Macpherson RB. Design, development and testing of low head high efficiency kinetic energy machine; MSc thesis: University of Massachusetts, Amherst, MA, 1972. [7] Newman BG. Measurements on Savonius rotor with variable gap. Proceedings of Sherbrook; Sherbrook, Canada: University Symposium on Wind Energy, 1974; 116. [8] Modi VJ, et al. Optimal configuration studies and prototype design of a wind energy operated irrigation system. J Wind Eng Ind Aerodyn 1984: 16:85-96. [9] Sivasegaram S. Secondary parameters affecting the performance of resistance type vertical axis wind rotors. Wind Engineering, 1978; 2:49-58. [10] Gupta R, Das R, Sharma KK. Experimental study of Savonius-Darrieus wind machine. International Conference on renewable Energy for Developing Countries, Washington D.C, USA. 2006. [11] Cochran BC, Banks D, Taylor SJ. A three-tiered approach for designing and evaluating performance characteristics of novel WECS; AIAA Inc and ASME 2004. [12] Gupta R, Biswas A, Sharma KK. Comparative study of a three-bucket Savonius rotor with a combined three-bucket Savonius three-bladed Darrieus rotor; Journal of Renewable Energy, 2008; 33(9): 1974-1981.

You might also like