A CFD Study of Wind Turbine Aerodynamics Chris Kaminsky *, Austin Filush *, Paul Kasprzak * and Wael Mokhtar ** Department of Mechanical Engineering Grand Valley State University Grand Rapids, Michigan 49504 Email: kaminscw@mail.gvsu.edu, filusaus@gmail.com, kasprzap@mail.gvsu.edu and mokhtarw@gvsu.edu Abstract With an ever increasing energy crisis occurring in the world it will be important to investigate alternative methods of generating power in ways different than, fossil fuels. In fact, one of the biggest sources of energy is all around us all of the time, the wind. It can be harnessed not only by big corporations but by individuals using Vertical Axis Wind Turbines (VAWT). VAWT s offer similar efficiencies as compared with the horizontal axis wind turbines (HAWT) and in fact have several distinct advantages. One advantage is that unlike their HAWT counterparts, they can be placed independently of wind direction. This makes them perfect for locations where the wind direction can change daily. To analyze the effectiveness of a VAWT, methods of computational fluid dynamics (CFD) were used to simulate various airflows and directions. The analysis began with a literature analysis into the subject, in order to determine the types of airfoils that were most effective. After the research was done, the system was modeled in SolidWorks and imported into Star CCM+ to perform a CFD analysis. The first part of the CFD analysis analyzed the 2D flow over the chosen airfoil(s). Next, the analysis looked at the flow over a 3D representation of the airfoil(s). The 2D and 3D simulations used different angles of attack and speeds (15 & 30 mph) to determine when separation occurred at the various speeds. Finally, a full VAWT assembly was created and analyzed at various wind directions at the same wind speeds. The full assembly included 3 airfoils that were attached into a 5ft high, 3 ft diameter structure. Each step of the analysis included importing the CAD file into Star CCM+ as an IGES file, selecting physical models, generating a numerical mesh, and applying boundary conditions. All three portions of research studied scalar and vector properties, such as pressure and velocity. This paper describes the research of a VAWT using the NAC A0012-34 airfoil. * Mechanical Engineering Student ** PhD, Assistant Professor of Mechanical Engineering, ASEE member.
Introduction The energy crisis the world is going to be in, in the next 100 years when the Earth s supply of fossil is no more, isn t foreign matter to anyone. Since the start of the industrial revolution humans have been using fossil fuels to power their machines that make more machines. Society thinks that the gravy train that is fossil fuels will be around for a long time but analysis on the current supplies and the ever growing population of the planet suggest that we may run out very soon. The Klass model assumes a continuous compound rate and it is a computational approximation of when we will run out of the various fossil fuels. Depletion times for oil, coal and natural gas are approximately 35, 107, and 37 years. 1 These figures do not give much time for alternative methods to be found or perfected. The use of VAWT on residential and commercial properties could help to extend these deadlines out further or even eliminate them if the right technologies come about. Literature Review Wind turbines use the kinetic energy of the wind and convert it to mechanical energy. This is then used to produce electricity, grinding of grain or pumping of water (windmills, wind pumps). There are two types of wind turbines, horizontal and vertical. Vertical axis wind turbines (VAWT) have the rotor shaft vertically. These types of wind turbines are advantageous because they do not need to be pointed into the wind in order to function. Also, the generator and gearbox are able to be placed near the ground, thus allowing for easy maintenance. On the other hand, these types of turbines typically have a lot lower rotational speed, meaning higher torques involved. Combine this effect with the very dynamic loading that is generated on the blades and a more expensive drive train is needed. This dynamic loading can however be reduced by using more than 2 blades. 2 The dynamic loading that is generated on VAWTs requires higher material expenditures for a given construction as compared to HAWTs but when compared to the overall costs of a small VAWT installation, the increased cost is negligible. 3 Also, HAWTs are presently cheaper because they have been produced for a long time and in larger numbers. As research into VAWTs increases, the cost for such wind turbines will decrease as the process becomes more efficient. 4 Symmetrical blades were chosen for this project because they are easier to manufacture. A study by Liu Shuqin of Shandong University of Chin focused on the power generation based on changing the type of blades that were used and this was used to give an insight into choosing the blade shapes for our VAWT design. The experimental setup is given below in Table 1. The results showed that self-pitch streamline symmetrical blades generated a higher power output than the others, as shown in Figure 1. 2
Table 1: Experimental Setup 5 Parameters Wind Turbine A Wind Turbine B Wind Turbine C Pitch Type Self-pitch Fixed-pitch Self-pitch Number of Blades 5 5 5 Blade Shape Streamline Streamline Arc blades symmetrical blades symmetrical blades Size of Blade 120 cm (L) x 20 cm (W) x 2.5 cm (T) 120 cm (L) x 20 cm (W) x 2.5 cm (T) 120 cm (L) x 20 cm (W) x 1.75 cm (T) Weight of Each 1.65 kg 1.65 kg 1.15 kg Blade Diameter of Arm 2 m 2 m 2 m Rotation Generator 300 W permanent magnet generator 300 W permanent magnet generator 300 W permanent magnet generator Figure 1: Power generation of various types of blades 5 Another reason it was decided to choose symmetric airfoils was that although these types of airfoils are not the most efficient at producing lift, they do produce a lift-stall effect that allows the system to obtain equilibrium. The lift that is created from the wind passing over the airfoil is eventually lost and the blade goes into its stall condition, but the other blades pick up more lift and keep the cycle going. This in turn regulates the speed of the entire VAWT. 2 An aspect ratio of 13.6 was chosen as a result of literature review. A study of an application of design of a small capactity (<10kW) fixed-pitch straight bladed VAWT, similar to application of this design study, utilized this aspect ratio. The study claimed 13.6 to be an efficient aspect ratio for the application. 6 Utilizing the maximum dimensions of the VAWT, the chord length was calculated to be 4.41 inches using this aspect ratio. 3
Present Work The presented study utilized CFD simulations to analyze the flow field around a vertical axis wind turbine (VAWT) that could be used for residential use. The design parameters for the chosen VAWT consisted of having 3 blades, a height of 5 ft and a diameter of 3 ft. Using the UIUC Airfoil Coordinates Database the NACA 001234 airfoil was chosen. A graphical representation of this airfoil as well as its associated properties can be seen below in Figure 2. The airfoils and the full assembly were then analyzed at various angles of attack and for different wind speeds, as listed in Table 2. The objective of the analysis was to be optimized the attack angle in order to obtain the greatest possible lift force. The drafting software that was used to model the VAWT and the far field geometries that surround it and its components was SolidWorks. Below, Figures 3 and 4 show images of the CAD models that were used for the 2D/3D airfoil configurations and for the 3D full assembly respectively. Table 2: Simulations Studied Case Yaw Angle Wind Speed (mph) 2D Airfoil 0 o, 5 o, 10 o, 15 o 15, 30 3D Airfoil 0 o, 5 o, 10 o, 15 o 15, 30 3D Full Assembly 0 o, 30 o, 60 o, 90 o 15, 30 Figure 2: NACA 001234 7 4
Figure 3: VAWT 2D & 3D Airfoil Configurations Figure 4: VAWT Full Assembly Configuration Computational Method The analysis of the VAWT was done using STAR CCM+, developed by CD Adapco as standard commercial software. The analysis used a computational finite volume method to analyze the 2D, 3D airfoil only and 3D full assembly cases with using a segregated flow solver. Modeling of the turbulence in each of the cases was done using the 2 equation SST k-ω model. This selection for the turbulence model was made because the k-ω model offers great analysis in both fully developed flow and along the boundary layer regions. Three models were used in generating the mesh regions around the VAWT that were used in the analysis of the 3 cases. These were the prism layer and surface remesher, and the trimmer. Approximately 600,000 cells were generated for the 2D analysis cases, around 900,000 cells were generated in the various 3D airfoil analysis cases and for the full complete VAWT analysis about 1.3 million cells were used in the simulations. Using STAR CCM s volumetric controls, regions of importance, such as separation regions, were able to be analyzed more accurately. Figures 5a-5c shows the boundary conditions, surface mesh and volume mesh that were used in the analysis of the 2D airfoil cases. Figure 6a-6c shows the boundary conditions, surface mesh 5
and volume mesh that were used in the analysis of the 3D airfoil cases. Figures 7a-7c shows the boundary conditions, surface mesh and volume mesh that were used in the analysis of the 3D VAWT full assembly cases. Figure 5a: 2D Airfoil- Fluid Domain Boundary Conditions Figure 5b: 2D Airfoil- Surface Mesh Figure 5c: 2D Airfoil- Volume Mesh and Plane Section 6
Figure 6a: 3D Airfoil- Fluid Domain Boundary Conditions Figure 6b: 3D Airfoil- Surface Mesh Figure 6c: 3D Airfoil- Volume Mesh and Plane Section 7
Figure 7a: 3D VAWT Full Assembly- Fluid Domain Boundary Conditions Figure 7b: 3D VAWT Full Assembly- Surface Mesh Figure 7c: 3D VAWT Full Assembly- Volume Mesh and Plane Section 8
CFD Results And Discussions The first part of the analysis involved studying the effects that varying attack angles and wind speeds had on the flow regions. As expected, increasing the attack angle of the airfoil created larger regions of separation causing what is known as the stall effect. Increasing the wind speed caused the separation regions to be more exaggerated with a greater amount of turbuluence present. Knowing the stall angle is important to the design of vertical axis wind turbines because of the lift-stall effect that was mentioned earlier in the literature review section. Having a VAWT system with very large separation regions would prove to be very inefficient because of the large amounts of drag that would exist. Figures 8 and 9 below show pressure distribution plots for 2 of the cases, more results can be found in Appendix A. Figure 8 shows the results from the 0 degree attack angle and it shows an almost equal pressure distribution on both the top and bottom of the wing. This makes sense due to the symmetry of the chosen airfoil. Increasing the attack angle to 15 degrees, as shown in Figure 9, shows a dramatic change in the pressure distribution plot. It can be seen that there is a very large low pressure region on top of the wing and the stagnation point has moved to the bottom of the airfoil. Stagnation point Equal pressure distribution Figure 8: 2D Airfoil Pressure Distribution for 0 degrees at 15 mph 9
Low pressure region Stagnation point High pressure Figure 9: 2D Airfoil Pressure Distribution for 15 degrees at 15 mph Figure 10 and 11 below show the results of the velocity distribution for the same 2 cases as shown above, more results can be found in Appendix B. Figure 10 shows the 0 degree attack angle and similar to the pressure distribution, it shows approximately equal velocities both on top and under the airfoil. This is also due to the symmetry of the chosen airfoil. Just as before, increasing the attack angle, as in Figure 11, shows a very large separation region on top of the wing. In this region vortexes and reversal of flow exist in a turbulent manner. Equal velocity distribution Figure 10: 2D Airfoil Velocity Distribution for 0 degrees at 15 mph 10
Turbulent/separation region Higher velocity Lower velocity Figure 11: 2D Airfoil Velocity Distribution for 15 degrees at 15 mph The same analysis that was done above on the 2D airfoil was done to a 3D airfoil model and it produced similar results. Figures 12 and 13 show the pressure distributions on the airfoil along with velocity streamlines and Figures 14 and 15 show the velocity distribution around the airfoil. The streamlines become more turbulent as the angle of attack is increased, this is shown in Figures 12 and 13. Figure 12: 3D Airfoil Pressure Distribution for 0 degrees at 15 mph 11
Figure 13: 3D Airfoil Pressure Distribution for 15 degrees at 15 mph Figure 14: 3D Airfoil Velocity Distribution for 0 degrees at 15 mph 12
Figure 15: 3D Airfoil Velocity Distribution for 15 degrees at 15 mph Further analysis of the VAWT involved studying the net lift force generated at the various angles of attack and wind speeds. The results show that at higher speeds the critical angle of attack happens faster than at lower speeds. Looking at the results from the 2D airfoil analysis of the lift force versus attack angle (Figure 16), the results show that the stall condition occurred at faster wind speed at approximately 8 degrees. The results for the lower wind speed are incomplete and the stall condition could not be seen in the scope of the speeds that were tested for this analysis. The faster wind speed condition produced the most net lift force, which was expected. The net lift force results for the 3D airfoil analysis (Figure 17) show similar results to that of the 2D scenario but this graph does not appear to show clear stall conditions. This could be because the modeling of the 2D airfoil is a simplified analysis and the 3D airfoil. 13
Net Lift Force (lbf) Net Lift Force (lbf) 0.3 0.25 0.2 0.15 0.1 0.05 15 MPH 30 MPH 0-0.05 0 5 10 15 20 Angle of Attack (degrees) Figure 16: 2D Airfoil- Lift Force vs. Attack Angle 2.5 2 1.5 1 0.5 15 mph 30 mph 0 0 5 10 15 20 Angle of Attack (degrees) Figure 16: 2D Airfoil- Lift Force vs. Attack Angle The full assembly 3D results were intended to be run for 4 different angles of attack, 0, 30, 60 and 90 degrees and 2 wind velocities 15mph and 30mph for a total of 8 runs. This would cover a sampling of the angles that turbine will see during operation, and since the pattern repeats after 120 degrees, there is no need to redo those results. Several different mesh models were run until one yielded results that seemed logical. At first from looking at the velocity distribution, Figure 17, the results seemed to make sense. 14
Figure 17: 3D Full Assembly- velocity distribution After analyzing the results further however, several anomalies arose. For one, the residual plot, shown below in Figure 18, had large spikes in turbulent kinetic energy and specific dissipation rate as shown. These errors were far above acceptable results and were cause for concern. Figure 18: Residuals plot for 3D full assembly 15
Also, further post processing tools used did not yield logical results. In the pressure plot below, Figure 19, there were several spots of abnormally large or abnormally negative results. A closeup view is available in Figure 20. These numerical errors resulted in illogical results. Figure 19: Pressure distribution with streamlines Abnormal pressure increase on inside face of airfoil Figure 20: Close-up on pressure distribution Upon further investigation into these results, it was found that the inside of the trailing edge had some large irregularities from the surface mesh, Figure 21. This only happened on a single one of the airfoils each time however. To investigate if this was indeed the cause of the error, several other angles of air were run and it was found that each time, there was a large error on the airfoil where the air hit it nearly perpendicular to the inside trailing edge. 16
Figure 21: Surface mesh irregularities This theory was further confirmed by looking at the velocity distribution at the boundary layer (Figure 22) of the affected airfoil. To eliminate the possibility that it was an issue with the boundary layer itself, several models were run with anywhere between 10 and 20 prism layers. It had no effect; the results were the same every time. Figure 22: Boundary layer analysis of velocity distribution Several attempts were made to fix the issues with the surface mesh such as adding feature curves near the trailing edge and increasing the number of cells all the way to 6 million cells. These fixes were not enough to smooth out the problem areas. A solution could not be found in the time limit of the study so no further results were able to converge. Conclusions The use of the STAR CCM software to analyze the air flow around a vertical axis wind turbine proved to be a very effective tool. A full analysis could be performed before any construction of such a project even began. The modeling of the airfoil in both the 2D and 3D cases allows the user to study the aerodynamics of various geometries at different physical settings to get a true feel for how the specific airfoil might behave in real world applications. Wind speeds of 15 and 30 mph were studied, as well as varying attack angles from 0 to 15 degrees, the results of this research on the NACA 001234 airfoil showed it could be a very viable choice for a residential 17
VAWT. The 2D analysis gave a stall angle of about 8 degrees, however, the 3D analysis, it being more accurate, did not provide us with a stall angle. This suggests that further analysis would provide more information on the critical angle of attack. CFD software is a very strong tool but it can be difficult to master. The results for the 3D full assembly analysis of vertical axis wind turbine were incomplete. Even though the results yielded were less than desirable, analysis of why the simulations failed proved to be very helpful in the mastery of CFD software use. References 1. Shafiee, Shahriar, and Erkan Topal. "When will fossil fuel reserves be diminished?" Energy Policy 37.1 Jan. (2009): 181. Science Direct. Web. 1 Feb. 2012. 2. Ragheb, M. "Vertical Axis Wind Turbines." University of Illinois at Urbana-Champaign. 1 Aug. (2011). Web. 7 Dec. 2011. https://netfiles.uiuc.edu/mragheb/www/npre%20475%20wind%20power%20systems/vertical%20axis %20Wind%20Turbines.pdf 3. Riegler, Hannes. HAWT versus VAWT: Small VAWTs find a clear niche. July/August 2003. www.renewableenergyfocus.com 4. Eriksson, Sandra, Hans Bernhoff, and Mats Leijon. "Evaluation of different turbine concepts for wind power." Renewable and Sustainable Energy Reviews 1224 May (2006). Science Direct. Web. 9 Dec. 2011. 5. Shuqin, Liu. "Magnetic Suspension and Self-pitch for Vertical-axis Wind Turbines." Fundamental and Advanced Topics in Wind Power. Web. 7 Dec. 2011. <http://www.intechopen.com/source/pdfs/16250/intech- Magnetic_suspension_and_self_pitch_for_vertical_axis_wind_turbines.pdf>. 6. Islam, Mazharul, Fixed Pitch Straight-Bladed Vertical Axis Wind Turbine: Design Challenges and Prospective Applications, Virginia Tech University, 2006, http://www.ari.vt.edu/wind-egypt/files 7. UIUC Applied Aerodynamics Group. UIUC Airfoil Coordinates Database. http://www.ae.illinois.edu/mselig/ads/coord_database.html. 1 Dec 2011. 18