Vertical Axis Wind Turbine Efficiency Improvement Studies


Global climate change is accelerated by energy production using fossil fuels. Wind energy is renewable energy with no impact our environment. For particular cases of high wind, and urban use where noise is a problem, Savonius Vertical Axis Wind Turbine (VAWT) is a suitable design. Since Sovomius design has low efficiency, I wanted to explore ways to improve.

Three concepts are studied: a) varying blade overlap to affect the drag force on the returning blade b) hinged (swinging) blade concept (inspired by bird's wing) where the blades can move when the convex side is facing wind c) a blocker plate positioned to cut off wind on the returning blade to eliminate drag. 

Design of Experiments (DOE) varying wind speed, overlap, and blade position is conducted, and JMP statistical package is used to predict the behavior. Blades were allowed to move such that returning blade swings back and RPM was measured. A Plexiglas board was used as a blocker plate and its angle and location were changed to maximize RPM. VAWT design is characterized by performance coefficient (Cp).

Most publications used simulation studies. Research shows different overlap studies but not using a DOE. The concept of swinging blade is completely novel.

Improvement of VAWT efficiency will help to employ on roof tops in cities to generate energy.

Blocker plate concept gave the best performance by reducing the drag force on returning blade.

Future studies have to be conducted to automatically orient the blocker plate driven by wind flow and its direction.  

Question / Proposal

Is it possible to improve the efficiency of Savonius vertical axis wind turbine by reducing the drag force on the returning blade?

My hypothesis is that if the drag force on the returning blade is reduced, the net torque on the VAWT and the RPM will increase (keeping the wind speed and blade dimensions constant).

The Savonius Vertical Axis Turbine, designed by Finnish Engineer S.J. Savonius (in 1931) consists of semicircular blades arranged in an ‘S’ shape as shown in Fig.1.


Fig.1: Deatils of Savonius VAWT Design

The blades are typically arranged such that there is an overlap (e). If wind blows from left to right as shown in Fig.1, then the particular configuration shown experiences drag force on both blades.

However, the drag force experienced by the blade with concave side facing the wind is higher compared to that experienced by the blade whose convex side is facing the wind, as described in Shape Effects on Drag [4]. The blade rotating in the same direction of wind is called advancing blade, and the blade moving against the wind is called returning blade. It is this difference in drag force on the advancing blade ) and the drag force on the returning blade ) that powers the VAWT.  The Torque of the VAWT is given by


If the drag force on retuning blade is reduced, the Torque, the rotational speed and the efficiency of the VAWT will increase. All three ideas attempt to reduce the drag on the returning blade.



Greenhouse gas emission and resulting climate change and global warming have brought renewed interest to alternative energy sources. Wind power is a renewable energy source and does not pollute our planet.  Two types of wind turbine technologies are in wide use as presented by Kalmikov [1]. Horizontal Axis Wind Turbine (HAWT) is most common type. HAWTs have to be scaled to large dimensions to be commercially viable. For areas with high wind speeds large HAWTs cannot be used since they fail due to high wind forces. Also HAWT depends on wind direction. With rapid urbanization there is a strong need to tap into wind energy in cities. HAWT is large and also generates noise and is not suitable for installation in cities. Wind direction in cities can be random. For these conditions vertical axis wind turbines are more suitable, Marmutova [2]. In addition HAWT poses threat to bird populations, while VAWT is a safer choice.

Two types of VAWTs commonly used are Savonius and Darrieus designs. In this work I have chosen to study Savonius design because it offers many advantages. It does not depend on wind direction and can operate at very low wind speeds. Blade fabrication and maintenance are easy. However, the Savonius design has low efficiency or power coefficient. The purpose of current work is to study design parameters that can improve Savonius VAWT efficiency, Menet [3].  

Birds flap their wings and move in such a way they offer no resistance when flying. I wanted to test the swinging blade (hinged blade) concept based on a bird's wing movement when it flies upwards.

Method / Testing and Redesign

Wooden end plates (diameter: 20 cm) are connected by an axle (wooden rod of height: 25 cm, diameter: 0.635 cm). Metal sheets are cut to form the two blades (height: 19.5 cm and diameter: 10.75 cm). Steel rods are used to secure the blades and also act as hinges. Along the line joining the hinges three holes are made on each side to allow three overlap (x) values (1, 1.5 and 2 cm). The overlap e is given by 2x.



The edge of each blade is lopped to act as a hinge. Holes are also drilled for steel rods in transverse direction so that the blade location (parameter y) can be varied. The VAWT frame made of File Storage Crate. Non-Flanged 8mm Ball Bearings (Actobotics) are inserted into crate ends. Axle is inserted into the inner diameter of the bearings and rotates freely.   

Design of Experiments (DOE): Three independent variables are:  Overlap, x:1–2 cm; Blade location, y:0-4cm and Wind-Speed, v:3.5, 4.5 and 5.7 m/s. Wind speed is measured with an anemometer. Dependent Variable: RPM of VAWT is measured using a digital tachometer. A reflective sticker on the axle allows laser light to detect RPM. The fan is kept at 20cm from the VAWT. DOE parameter space consists of 8 corner points of a 3D cube and the body center with mid-point values of x, y and v. RPM values are measured for each case and average RPM values were used as dependent variable. JMP predicts that RPM increases with increasing x and v and decreases as parameter y increases. The statistical model: RPM = 7.05+77.98x-11.68y+56.2v is accepted since p-value is << 0.05 and R2 = 0.96 indicating that the linear fit is good.





Based on these predictions multiple concepts were tested.

Idea-1: If the overlap e (=2x) is increased, the wind flow through the gap will create a drag force on the concave side of the returning blade and reduce the effective drag force (FDRet) on the returning blade.

Idea-2: If the blades can move (swing) when the convex side is facing the wind for the returning blade B, the drag force on returning blade will reduce.  Stopper SA holds the advancing blade A .

Idea-3: If a blocker plate is placed cutting off wind on the returning blade, the drag force on the returning blade (FDRet) will be eliminated. For each of these cases, Torque is also measured.

Torque measurement:  A wooden arm is connected to the axle and an attached string, passes over a pulley. The other side of the string holds a weight (123g) placed on a digital scale. When the blade (axle) pulls the string the pull force exerted on the weight leads to a reduced weight reading on the digital scale. The apparent weight loss δW (g) gives the force of the pull. Force times the arm length gives the Torque of VAWT.


Overlap variation: When the parameter x is varied keeping all others constant (distance between fan the VAWT frame is 70 cm, Fan setting-III,  y= 0 cm), Fig.7a shows RPM.      


The variation the parameter (y) (Fig.8) and the concept of a hinged blade showed a decrease in the RPM. Data has significant amount of noise due to impact of the blades hitting the stoppers, as seen by the large ranges in the box plot.

Fan location relative to the VAWT was studied by varying the angular orientation and the distance. Results (Fig.9 & 10) show that there can be a large uncertainty in RPM reading depending on the fan location. All the subsequent experiments are conducted keeping these constant at 70cm and 90°respectively.

The concept of blocking wind [5] so that the returning blade does not experience any drag, is tested using a Plexiglas board and varying its angle/distance relative to the VAWT.


For the case m=0 (Fig.11) the blocker plate is touching the VAWT frame, while for m=20 cm (Fig.12) the blocker plate is 20 cm towards the fan (away from the VAWT).


While swinging blade concept suffred from momentum loss, stationary torque test was conducted to and as seen from Fig.14, swinging blade concpet is able to generate higher torque.

Discussion and analysis

With an increase in the overlap of the blades, the RPM improved. Blocker Plate at an angle of 60° with l=0 and m=0 gave the highest performance coefficient (Cp). The calculation details for different factors are shown below [6,7].


When the blocker plate is held at 60°, highest performance coefficient (Cp) is observed. Cp is ratio of power generated by VAWT to the Kinetic Energy stored in the wind. Drag force is proportional to v for a laminar flow and v2 for a turbulent flow [8]. Tip Speed Ratio is ratio of blade tip speed to velocity of wind. As tip speed ratio increases beyond 1, the efficiency of VAWT drops. This is due to the fact that the net drag force reduces as velocity of wind increases.

Overlap increase showed increase in RPM at all three wind speed values. The fan orientation of 95° implies that the wind is directed towards the concave blade (advancing blade) and hence results in higher net drag force. This is similar effect as blocker plate concept. The dependence on the fan orientation is very sensitive. An angle change from 80° to 90° results in doubling the RPM. Hence care should be taken to orient the fan towards the VAWT. The distance dependence is a weaker, with 20 cm change leading to 35 RPM change. RPM is inversely proportional to the distance between the fan and the VAWT.  


  • DOE and statistical analysis predicted that increasing parameters x (e/2) and wind speed (v) lead to higher RPM. Based on these results, experiments varying single parameter at a time were conducted.
  • Larger overlap between the blades results in wind passing through the gap pushing the returning blade and reducing drag force on returning blade. As a consequence, the net drag force on VAWT increases leading to higher RPM.
  • Hinged Blade concept suffered from energy loss due to impact from blade hitting the stoppers. Even in the stationary y-parameter variation test, and RPM reduced.
  • Blocker plate concept gave the best performance: 
    • When the blocker plate is touching the VAWT frame, lateral distance of 3 cm effectively blocks the wind from reaching the returning blade. However when the blocker plate is oriented at an angle such that the wind is directed more effectively on the advancing blade, RPM dramatically increases. The angle of 60° is most effective, while 0° and 30° showed no significant difference.
    • When the blocker plate is moved 20 cm away from the VAWT frame there is appreciable difference in RPM for lateral distance of 0 cm. For 3 and 6 cm lateral distance angular orientation of the blocker plate did not show much difference in RPM. In all cases the measurements are highly repeatable with <4% standard deviation (except for < 50 RPM cases).
  • Torque measurements were made by measuring the weight of an object with string attached to an arm fixed on the axle. Weight loss due to string tension over the pulley is an indicator of the pull force.
  • The swinging or hinged blade concept does show 2.5 times higher torque compared to standard Savonius (fixed blade) design.
  • Savonius VAWTs are easy to implement in cities and high wind regions and pose no threat to birds. They also make less noise. 
  • Hinged Blade concept can be tested in future with mechanisam to reduce the impact of the swinging blade on the stoppers.
  • The swinging blade can be shaped similar to a bird's wing and the influence of the blade profile shape can be further explored in future. 
  • Future studies have to be conducted to automatically orient the blocker plate driven by wind flow and its direction. Drag Force can be measured on each blade and the net drag force can be correlated to angular velocity (ω).

About me

Ever since I was a child, I was intruiged by physics and the forces of motion. At a young age I read all about tornadoes, hurricanes, and other natural disasters, and was an expert on them by the age of 6. I was also good with my hands and could build elaborate structures twice my height back then, and spent most of my free time building with "Kappla" blocks  and reading many nonfiction books. Now, I am extremely interested in political problems as well, and like threorizing about science related topics that affect people on a broader scale, such as nuclear energy. I am extremely excited to take chemistry and physics in the following years, as my main interests lie in those subjects. In my free time, I love to play and watch sports such as baseball, basketball, and football. I play all of those sports with the neighborhood kids and I look forward to playing sports because I am extremely competitive. I also play piano and guitar, and I find it relaxing to play the piano and disconnect from the world and submerge myself in music. In the future, I plan to go wherever life takes me, and I find it is too early for me to narrow down the choices of possible fields I will center my life around. If I won Google Scince Fair, I would be incredibly humbled, as I have not won many science fairs and it would really make a difference in my life,

Health & Safety

I was wearing protective goggles, mask and gloves while cutting wooden discs for end plates. Most of the materials I used are safe to handle. Care was taken while cutting the sheet metal (wearing working gloves).

  • Sheet Metal (for blades)
  • Wooden planks (end plates)
  • Wooden rod  (axle)
  • Saw to cut circles
  • Clamps
  • Metal rods (stoppers)
  • Frame to hold VAWT
  • Digital Tachometer
  • Fan
  • Anemometer
  • Pulley and weight
  • Digital Scale

Bibliography, references, and acknowledgements


  1. Kalmikov, A., (2009). Wind Power Fundamentals. MIT Wind Energy Group & Renewable Energy Projects in Action. <>
  2. Marmutova, S, (2016). Performance of a Savonius wind turbine in urban sites using CFD analysis. Ph.D. dissertation. University of Vaasa. Acta Wasaensia, ISBN: 978-952-476-675-3.
  3. Menet, J, (2003).  Increase In the Savonius Rotors Efficiency via a Parametric Investigation.
  4. Shape Effects on Drag, NASA Glenn Research Center,  The Beginner's Guide to Aeronautics. <> (accessed Jan 28, 2018).
  5. Mbaci, M.W, (2006). Renewable Energy Report.
  6. Battisti, L, et al. (2016). Analysis of Different Blade Architectures on small VAWT Performance.    Journal of Physics: Conference Series, 753 062009.
  7. Akhmedov, D. S, (2015). Mathematical Model to Calculate the Performance of Low Power Vertical Axis Wind Turbine. Proceedings of 42nd The IIER International Conference, ISBN: 978-93-85832-23-9.
  8. Mahmoud. N. H, (2010). An experimental study on improvement of Savonius rotor performance. Alexandria Engineering Journal 51, 19–25.
  9. Frikha, S, et al (2016). Incidence Angle Effect on the Turbulent Flow around a Savonius Wind Rotor. American Journal of Energy Research, vol. 4, no. 2: 42-53.




My father Sai Tallavarjula encouraged me to explore topics that involve motion. Upon research I found vertical axis wind turbine designs. My father showed me how to search for past research and what has been published. He also helped me by discussing different designs and how to build the blades, VAWT and the frame to hold it.

Dr. Fred Barez from San Jose State University gave a suggestion to use a pulley and weight to measure the torque, which I further modified to measure weight loss due to string tension.