An experimental study of the optimal design parameters of a wind power tower used to improve the performance of vertical axis wind turbines

Abstract

In order to augment the performance of vertical axis wind turbines, wind power towers have been used because they increase the frontal area. Typically, the wind power tower is installed as a circular column around a vertical axis wind turbine because the vertical axis wind turbine should be operated in an omnidirectional wind. As a result, the performance of the vertical axis wind turbine depends on the design parameters of the wind power tower. An experimental study was conducted in a wind tunnel to investigate the optimal design parameters of the wind power tower. Three different sizes of guide walls were applied to test with various wind power tower design parameters. The tested vertical axis wind turbine consisted of three blades of the NACA0018 profile and its solidity was 0.5. In order to simulate the operation in omnidirectional winds, the wind power tower was fabricated to be rotated. The performance of the vertical axis wind turbine was severely varied depending on the azimuthal location of the wind power tower. Comparison of the performance of the vertical axis wind turbine was performed based on the power coefficient obtained by averaging for the one periodic azimuth angle. The optimal design parameters were estimated using the results obtained under equal experimental conditions. When the non-dimensional inner gap was 0.3, the performance of the vertical axis wind turbine was better than any other gaps.

Introduction

The importance of renewable energy is increasing, due to the continuing depletion of fossil fuels and the effects of fossil fuel emissions on climate change. For these reasons, a large number of studies have been devoted to methods of harvesting renewable energy and reducing the use of fossil fuels. The renewable energy sources that are available for harvesting include wind energy, solar energy, geothermal, biomass, waste thermal energy, ocean energy, and so on. For utilizing wind energy, horizontal axis wind turbines (HAWTs) or vertical axis wind turbines (VAWTs), which are operated by the force of wind passing over angled blades, are widely used. The different characteristics and limitations between HAWT and VAWT were explained well. ¹ In addition to these two common types, many other different wind turbine designs have been developed and used.

VAWTs have several advantages. They are typically seen in urban areas because they operate with lower noise. Since they can operate in omnidirectional winds, they do not need a yaw-control device. The generator on the VAWT can be installed at a lower elevation because the shaft of the VAWT is rotating vertically, whereas a HAWT requires the generator to be mounted on the rotating shaft, which is typically high on a tower. These advantages mean that the VAWT has higher structural stability and simplicity. However, the VAWT designs result in relatively lower power than horizontal types, because each blade generates a wake which affects the performance of the following blade. Moreover, the VAWT usually requires a higher wind speed for self-start. Hence, the VAWT exhibits lower performance when the wind speed is low.

Wind speed is a very important factor in generating energy from wind since wind energy is proportional to the cube of the wind’s speed. Wind speed can be strongly affected by geographical location and environmental circumstances. The fluctuation nature of wind energy could be controlled to make a power network system stable. ² However, for a given circumstance, the important criteria are how efficiently the system converts wind energy to mechanical energy. This efficiency can be achieved when the turbine blade is designed appropriately,^3–7 since the blade performs the key role in converting wind energy to mechanical energy.

The VAWT is able to operate in an omnidirectional wind because its blade rotates in a vertical position with respect to the ground. However, the output power of the vertical turbine blade changes along the azimuthal direction, since the rotating blade experiences various angles of attack depending on its azimuthal location. In order to augment the performance of the VAWT, it would be beneficial to form a better angle of attack on the blade. Since guide vanes could be used to change the wind flow direction locally, they were applied so that the blades of the VAWT operated at a better angle of attack.^8,9 The yaw of the guide vane was also automatically adjusted using a tail-fin. ¹⁰ A stator blade array was used to enhance the performance of the VAWT in a low-speed environment. ¹¹

The performance of the VAWT can also be improved by augmenting the wind energy. In order to augment the wind energy, diffusers can be installed at the exit of the wind turbine. These diffusers can help reduce air pressure at the turbine exit. ¹² This lower pressure increases the relative difference in pressure so that more wind can pass through the wind turbine. A better diffuser to augment the performance of wind turbines was designed.^13,14 However, the application of the diffuser to the VAWT required an additional yaw-control device to align to the wind direction. To avoid this problem, a wind power tower (WPT)^15–17 was applied to the VAWT. The WPT was installed around the VAWT to increase the frontal area of the VAWT. With this modification, the VAWT could be not only operated in omnidirectional winds but also exhibited augmented wind energy. In addition to these approaches, various other methods have been developed to augment the performance of the VAWT, such as increasing the height of the installed location of the wind turbine,^18–20 utilizing the wind augmented by the space between buildings, ²¹ installing a wind collector on the roof of buildings^22,23.

The WPT has advantages when the operating location of the VAWT is higher, such as when the WPT is constructed as a building several stories tall, as shown in Figure 1. In addition, multiple VAWTs can be installed in the WPT. In particular, the guide walls, which are used to support the structure of the WPT, can change the wind flow direction inside the WPT. This changed flow direction could improve the angle of attack on the turbine blades to further augment the performance of the VAWT. However, use of the WPT could degrade the performance of the VAWT because the WPT can block the wind flow, and an inappropriately changed flow direction could produce a worse angle of attack on the blade. Therefore, it is very important to know how the design parameters of the WPT affect the performance of the VAWT. Nonetheless, the literature survey could not show enough information to design a better WPT. In particular, none was shown the degradation effect when the VAWT operated in omnidirectional winds. In order to investigate the effectiveness of the design parameters of the WPT, an experimental study was conducted in a wind tunnel. To simulate the omnidirectional wind in the wind tunnel, the WPT was fabricated to be rotated. The performance of a VAWT installed inside a WPT was measured for various design parameters as well as wind direction. Based on the measured results, the optimal design parameters were studied.

Figure 1.

View of wind power tower with VAWTs.

Experimental facility

The wind tunnel used in this experiment was operated at a power of 132 kW, and its test section was 2 m in height and 2.4 m in width. A VAWT model was designed to be compatible with the size of the test section, and its blades were manufactured from aluminum based on the NAC0018 profile. Three blades were assembled to the shaft of the VAWT. The shaft was manufactured from stainless bar with a diameter of 18 mm. The span of the blade was 575 mm and its chord was determined 83 mm so that the solidity of the VAWT became 0.5. The three blades were installed using struts which were manufactured from aluminum bar to have diameters of 8 mm. The rotating radius ( $R_w$ ) of the VAWT was 250 mm. These VAWT dimensions resulted in a 6.5% blockage of the test section.

The WPT was installed around the VAWT, and only a one-story WPT among the eight-stories WPT shown in Figure 1 was installed due to the size restriction of the test section. The number of guide walls was set to 7. This was determined by not only the computational fluid dynamics technique ²⁴ but also the preliminary test in the wind tunnel with other combination. The predicted output power and the measured output power showed the best performance when seven guide walls were installed. In order to reduce the loss on the blade tip, the height ( $h_g$ ) of the guide wall was established to be 600 mm. Three different sizes of guide walls were designed by changing only the length (L) of the guide wall. The aspect ratios ( $h_g/L$ ) of the guide walls were set to 1.6, 1.85, and 2.18. They were classified as large guide wall (LGW), medium-size guide wall (MGW), and small guide wall (SGW) for the aspect ratios of 1.6, 1.85, and 2.18, respectively. They were designed to be moveable along the radial direction so that the inner radius ( $R_i$ ) and outer radius ( $R_o$ ) of the VAWT could be adjusted. The experiment was conducted under equal conditions in order to compare each result relatively. In this manner, the effectiveness of the design parameters of the WPT was investigated through relative comparisons.

Figure 2 shows the geometric configuration of the WPT and the VAWT in a horizontal cross-sectional plane. The WPT was installed on a turn table so that it could be rotated. The rotation was precisely controlled by a stepping motor. The rotation of the WPT affects the performance of the VAWT due to the effect of the guide wall, although the VAWT is independent of the wind flow direction. Using this setup, the performance of the VAWT was measured at several locations according to the azimuth angle ( $\theta$ ). At a minimum, the measurement should be conducted for one periodic angle ( ${\theta }_p$ ), which can be determined by the number of guide walls. In this experiment, it was ${\theta }_p=51.4$ since seven guide walls were used. Figure 2 shows the azimuthal direction and the origin of the azimuth angle was set when a guide wall ( $g_1$ ) was aligned to the wind flow direction. The measurement was conducted in a steady state at 13 different angles for one periodic azimuth angle.

Figure 2.

Guide walls of the wind power tower and a wind turbine in the center plane.

The origin of the coordinate on the WPT was selected to be a point where the center line of the shaft of the VAWT crossed the upper plane of the turn table. The X-direction was set to the wind flow direction, and the Y-direction and Z-direction were set to the side direction and vertical direction, respectively. Figure 3 shows the geometric configuration of the WPT, viewed from upstream. The WPT was installed at a height of $h_t/h_g=0.84$ from the turn table. The tip clearance between the blade tip and the upper or lower plate of the WPT was set to 2% of the span of the blade. Figure 4 shows a picture of the WPT and the VAWT installed on the turn table.

Figure 3.

Wind power tower and wind turbine, viewing upstream.

Figure 4.

Picture of wind power tower and a vertical axis wind turbine.

The shaft of the VAWT was directly connected to a torque sensor, and this torque sensor was serially connected with a generator, as shown in Figure 5. The output power produced in the generator was dissipated on a load bank. This load bank was applied to control the output power of the VAWT using a variable resistance. The output power of the generator was measured using a power-meter. However, this output power did not include losses that were generated by the components of the VAWT, or by the experimental devices linked to the generator. Hence, the output power transferred to the generator was measured using a torque sensor. The data, including temperature and rotational speed, were acquired using a data acquisition system.

Figure 5.

Picture of components connected with a vertical axis wind turbine.

By installing the WPT in a wind tunnel, the velocity in the cross-section plane upstream could be slightly varied depending on the measuring location, due to blockage. In this experiment, the inlet velocity ( $V_{\infty }$ ) was determined by averaging the velocities measured at five locations in a cross-sectional plane upstream. When the inlet velocity was set to 9.5 m/s, the turbulence intensity of 0.86% was obtained using the equal method like the inlet velocity. In addition, total pressure ( $P{t}_{\infty }$ ) and static pressure ( $P{s}_{\infty }$ ) were also measured in the same manner as the inlet velocity. At a cross-section plane downstream ( $X/h_g=31$ ), total pressure ( $P{t}_{\mathrm{ex}}$ ) and static pressure ( $P{s}_{\mathrm{ex}}$ ) were measured to investigate the pressure difference according to the outer radii of the WPT. These properties were measured simultaneously using a multi-channel pressure measuring system. The accuracies of the measuring instruments are shown in Table 1 with their models.

Table 1.

Model and accuracy of instruments.

Instruments	Model	Accuracy	Remarks
Micro-manometer	Furness (FCO510)	±0.25%	–
Pressure scanners	PSI (PSI9016)	±0.15%	Multi-channel
Torque sensor	Dacell (TRD-2KC)	±0.2% load/10°C	Max. 2 kgf cm
Power-meter	Yokokawa (WT-1600)	±0.1%	–
RPM gauge	Ono Sokki (HT-5500)	±0.02% ±1 count	Displayed value
DAQ	Omron (ZR-RX45)	±0.1% F.S	Voltage
Generator	SPG (S9D60-90C)	–	Rated torque 2 kgf cm

Results and analysis

VAWT alone

Before measuring the effectiveness of the WPT, the performance of the VAWT without the WPT was measured. Since the VAWT was manufactured as a small-scale model, the performance of the VAWT could be affected by many factors, including the number of bearings, the efficiency of connecting parts, the manufacturing accuracy of the model, the blade profile, the material of the blade, and so on, even though the same experimental conditions were applied. Therefore, the experimental results were compared with results obtained on the same VAWT model as well as the same connecting parts.

In a small-scale model, unlike an actual scale VAWT, power losses due to the connecting parts have a large effect on the output power obtained from the wind energy. To account for this, the power losses of the VAWT model were measured at various rotating speeds using a motor instead of the generator. During the measurement, wind blew at a speed of 1 m/s, so that the wake generated by the blade would not stay in the rotating region of the blades. The power loss on the generator was also measured in a state of unload, and the devices used to measure it are shown in Figure 6.

Figure 6.

Picture of devices used to measure the power loss on a generator.

In the experiments with the VAWT model, the performance curve could be varied by two major parameters, inlet velocity and load on the generator. In a condition of constant inlet velocity, the tip speed ratio ( $\lambda$ ) of the VAWT decreases when the load is increased. As another method, if the inlet velocity is increased while the load is maintained constant, the tip speed ratio of the VAWT will be increased. Therefore, several performance curves can be obtained depending on the experimental method. The experimental flow chart is shown in Figure 7.

Figure 7.

Experimental flow chart.

Figure 8 shows the performance curves obtained by varying the inlet velocity while the load was kept constant. The load was directly related to the resistance of the load bank, and the load increased when the resistance was decreased. The load in Figure 8 is expressed as the resistance ratio ( $\beta$ ) based on the minimum resistance ( ${\Omega }_{\mathrm{ref}}$ ) used. The error bar is marked only on a power coefficient curve obtained with the lowest load because the uncertainty on all the other curves was similar to this one. The uncertainty analyses for the measured results were conducted with the accuracy of the instruments. The uncertainty of the obtained result was evaluated using a formula ²⁵ derived with the measured variables. This makes it easier to identify data clearly, so, the error bars in the following figures were also marked like Figure 8.

Figure 8.

Variation of power coefficients on a vertical axis wind turbine.

Figure 9 shows the power coefficient curves formed for an equal inlet velocity condition, while the load was changed. The dotted line expresses the performance curve obtained with low inlet velocity. The performance curve for the increased inlet velocity is expressed with the dashed dotted line. In the experiment of VAWT alone, the low power coefficient curves were obtained in a region of low tip speed ratio, as shown in Figure 9. This occurred because the experiment was conducted at a Reynolds number of 4.6 × 10⁴, which was low compared with that of full-scale VAWTs. Other test ²⁶ conducted with a same scaled VAWT showed similar power coefficient curves like these experimental results, although the wind velocity was greatly increased to reach a higher Reynolds number of 1.9 × 10⁵. Even though the results obtained in this experiment showed the lower power coefficient, these can be useful to investigate the optimal design parameters of the WPT through relative comparison when these results are consistent based on the experimental uncertainty.

Figure 9.

Variation of power coefficients obtained based on equal inlet velocity.

VAWT with the WPT

The performance of the VAWT installed inside the WPT varied depending on the relative location of the guide wall against the wind flow direction. As a result, it would require a huge number of performance tests to obtain the performance curve of the VAWT since there are many relative locations in a periodic azimuth angle on the WPT. If the performances are compared relatively, however, it is not necessary to measure the performance curves of the VAWT at every relative location. As shown in the performance curves of the VAWT alone, they did not cross each other when results were obtained by varying the load with a constant inlet velocity, or by varying the inlet velocity at a constant load. In order to compare the performance of the VAWT with the WPT, an experiment was conducted with the same conditions of equal load and equal inlet velocity. The power coefficient was obtained for each relative location. Then, these power coefficients were averaged for one periodic azimuth angle. Since this averaged power coefficient ( $\overline{C}p$ ) was independent of the wind flow direction, this value was used to compare the effectiveness of the WPT for various design parameters.

In the experiment with the WPT, the inlet velocity was set to 90% of the maximum velocity in the experiment of the VAWT alone (δ = 90%), and the load was set to the resistance ratio of β = 61. The application of low load was necessary so that the VAWT could be studied under many different operating conditions, due to the variation in design parameters on the WPT. Figure 10 shows the power coefficients obtained with three different guide walls. The axis of abscissa was expressed by the non-dimensional azimuth angle ( $\varphi =\theta /{\theta }_p$ ). When the different guide walls were installed, the inner radius and the outer radius of the WPT were simultaneously changed. Thus, these two parameters determined the configuration of the WPT. The inner radius was non-dimensioned to the inner gap ( ${\epsilon }_{\mathrm{in}}=(R_i-R_w)/R_w$ ) based on the rotating radius of the VAWT, and the outer radius was non-dimensioned to the outer gap ( ${\epsilon }_{\mathrm{out}}=(R_o-R_w)/R_w$ ).

Figure 10.

Variation of power coefficient for different inner gaps as (a) SGW, (b) MGW, and (c) LGW.

The power coefficient curves showed very similar trends for the equal inner gap, even though three different guide walls were applied, as shown in Figure 10. When the WPT was rotated from $\varphi =0$ , the power coefficient was gradually decreased and it reached minimum in the region of $\varphi =0.2$ . When the WPT was rotated over $\varphi =0.2$ , the power coefficient was steeply increased and it reached maximum in the region of $\varphi =0.5$ . With increasing rotation of the WPT, the power coefficient gradually decreased, and the rotating angle of the WPT approached the one periodic azimuth angle of $\varphi =1.0$ .

When the inner gap was reduced to ${\epsilon }_{\mathrm{in}}=0.2$ for three different guide walls, the power coefficient was insignificant near the region of $\varphi =0.2$ . This phenomenon was alleviated when the inner gap was increased. Therefore, the application of the WPT was found to deteriorate the performance of the VAWT if the inner radius of the WPT was installed too close to the VAWT. As the inner gap was increased, the variation of the power coefficient for the azimuth angle was also mitigated. It would be flat if the inner gap was increased to infinite. Therefore, the power coefficient curves were found to greatly depend on the inner gap. When the inner gap was ${\epsilon }_{\mathrm{in}}=0.3$ or ${\epsilon }_{\mathrm{in}}=0.4$ , better power coefficients for one periodic azimuth angle were obtained, even when three different guide walls were applied. Thus, the performance of the VAWT can be improved by choosing an appropriate inner radius for the WPT.

To compare the power coefficients based on the same inner gap, instead of the same guide wall, they were redrawn, as shown in Figure 11. For the case of an inner gap of ${\epsilon }_{\mathrm{in}}=0.2$ , the variation of the power coefficient along the azimuth angle was almost the same, even though the outer gaps were different due to the different guide wall, as shown in Figure 11(a). This phenomenon occurred because the effect of the inner gap was so strong that the effect of the outer gap could be neglected. In the WPT, the outer gap did not directly affect the performance of the VAWT by changing the flow direction, like the inner gap, but it contributed to increases in the frontal area of the VAWT. Therefore, the power coefficient curves obtained with the small inner gap were very similar, although different outer gaps were applied.

Figure 11.

Comparison of power coefficients with equal inner gaps as (a) ε_in = 0.2, (b) ε_in = 0.3, and (c) ε_in = 0.4–0.42.

When the inner gap was increased to ${\epsilon }_{\mathrm{in}}=0.3$ from ${\epsilon }_{\mathrm{in}}=0.2$ , the effect of the outer gap was slightly increased, as shown in Figure 11(b). When the inner gap was ${\epsilon }_{\mathrm{in}}=0.4$ , the effect of the outer gap was much increased, as shown in Figure 11(c). Therefore, the effect of the outer gap was related to the effect of the inner gap, and the effect of the outer gap increased with increases in the inner gap. For the same inner gap, the power coefficient was improved by increasing the outer gap because the increased outer gap increased the frontal area. However, the maximum outer gap could be limited by the area ratio between the inlet and exit of the flow passage formed by the guide wall of the WPT. The optimal area ratio can be determined by the flow quantity passing through the passage for a given total pressure upstream.

In a 2.5-D computational study ²⁴ of full-scale model, the better power coefficient was obtained at ${\epsilon }_{\mathrm{in}}=0.2$ when the outer radius and azimuth angle set to ${\epsilon }_{\mathrm{out}}=2.75$ and $\varphi =0$ , respectively. In the experimental result shown in Figure 11(a), the power coefficient at a fixed inner radius gap of ${\epsilon }_{\mathrm{in}}=0.2$ was slightly increased due to the effect of the increased outer radius when the outer gap was increased. In the comparison of the power coefficients measured at $\varphi =0$ for the equal outer gap, although each outer gap was not exactly same and less than ${\epsilon }_{\mathrm{out}}=2.75$ , they were almost same for the different inner gaps. This happened since the effect of inner gap could be increased for the smaller outer gap.

Since the VAWT inside the WPT was affected by the wind flow direction, the effectiveness of the WPT could be obtained by integrating the power coefficient for at least one periodic azimuth angle, so that it could be independent of the wind direction, as in equation (1). Nonetheless, the measurement of the power coefficient was conducted at 13 locations for one periodic azimuth angle. Thus, an averaged power coefficient ( $\overline{C}p$ ) was obtained by averaging for the one periodic azimuth angle

$$\overline{C}p=\frac{\underset{o}{\overset{2\pi /n}{\int }}Cpd\theta }{\underset{0}{\overset{2\pi /n}{\int }}d\theta }$$

(1)

Figure 12 shows the variation in the averaged power coefficient for the different inner gaps. The averaged power coefficient obtained near the inner gap of ${\epsilon }_{\mathrm{in}}=0.3$ showed better performance than the others. In case of the SGW, the experiment was conducted at an inner gap of ${\epsilon }_{\mathrm{in}}=0.32$ . The averaged power coefficient obtained at the inner gap of ${\epsilon }_{\mathrm{in}}=0.32$ was better than that at the inner gap of ${\epsilon }_{\mathrm{in}}=0.3$ . Therefore, the optimal inner gap would be between ${\epsilon }_{\mathrm{in}}=0.3$ and ${\epsilon }_{\mathrm{in}}=0.4$ .

Figure 12.

Variation of averaged power coefficient with various inner gaps.

Conclusion

In order to investigate the optimal design parameters for a WPT used to augment the performance of a VAWT, an experimental study was conducted in a wind tunnel. Seven guide walls were used for the WPT and three different types of guide walls were applied to test with several design parameters. The inner gap was the major design parameter affecting the performance of the VAWT, and its optimal value was found to be between ${\epsilon }_{\mathrm{in}}=0.3$ and ${\epsilon }_{\mathrm{in}}=0.4$ . When the inner gap was reduced to ${\epsilon }_{\mathrm{in}}=0.2$ , the WPT deteriorated the performance of the VAWT in the azimuthal region of $\varphi =0.2$ . When the inner gap was increased from ${\epsilon }_{\mathrm{in}}=0.2$ to ${\epsilon }_{\mathrm{in}}=0.3$ , the averaged power coefficient was improved by 30%. The outer gap was slightly influenced to the performance of the VAWT, even though an increase in the outer gap did improve the performance of the VAWT for conditions with the same inner gap. For the inner gap of ${\epsilon }_{\mathrm{in}}=0.3$ , the averaged power coefficient was improved by 5% although the outer radius of the WPT was increased by 40% of the rotating radius of the VAWT. Thus, the optimal outer gap could be determined by considering structural stability and economic benefits.