Design and Calibration of a Low Speed Wind Tunnel

A low speed, open circuit, laboratory wind tunnel was designed and built to facilitate the testing of airfoils designed for use with straight bladed vertical axis wind turbines. The wind tunnel built was the open-circuit type and was chosen due to the ease of fabrication and low construction costs associated with the build. Wind speed through the tunnel was controlled by a variable speed WEG 3.00 hp motor drive unit with a WEG CFW 08 vector inverter plus motor speed control unit. Calibration tests were performed using a hot-wire anemometer to determine the wind velocity as a function of differential pressure measured using a pressure transducer. The characteristic curve of wind velocity vs inches of water was plotted. The velocity profile for this wind tunnel indicated a turbulent flow regime and there was a good second order polynomial relationship between the measured velocity and differential pressure drop.


INTRODUCTION
The wind tunnel is one of the most common experimental testing facilities for the testing of fluid flow [1,2]. There are many types of wind-tunnels and they can be classified by the flow speed which can divide them into four groups.
Wind-tunnel design is a complex field involving fluid mechanics and engineering aspects [3,4]. The main advantage of the open circuit wind tunnel is in the savings of space and cost. Also, the effect of temperture changes is small and the performance of a fan fitted at the upstream end is not affected by disturbed flow from the working section [5].
The quality of results from experimental measurements obtained around a model in a wind tunnel is dependent on the quality of the free-stream flow. Assuring a high quality free-stream flow is of particular interest for investigations of external flows involving separated shear layers, e.g., separation on a wing, or wake of a bluff body [5]. The quality of the flow in a wind tunnel is mainly characterized by two features, namely, flow uniformity and turbulence intensity. Wind-tunnels represent a useful tool for investigating various flow phenomena. An advantage of using wind-tunnels is that experiments can be performed under controlled flow circumstances compared to experiments in the open environment.
The design of wind tunnel contractions has been based on a pair of cubic polynomials, and the parameter used to optimize the design for a fixed length and contraction ratio, has been the location of the joining point [6,7].

DESCRIPTION OF WIND TUNNEL
In this study the proposed design was an open-circuit suction wind tunnel in which the airflow was completely contained within the wind tunnel laboratory. The air leaving the wind tunnel via the fan flowed through the laboratory room and back through a filter and screen into the wind tunnel. The wind tunnel was the conventional non-return type with one closed working section and was constructed using 9.490 mm thick plywood. The dimensions of the wind tunnel was 3.912 m in overall length, 1.626 m wide and 1.626 m high at the mouth as shown in the schematic, Fig. 1.

Fig. 1. Schematic diagram of the wind tunnel
Wind speed through the tunnel was controlled by a variable speed WEG 3.00 hp motor drive unit with a WEG CFW 08 vector inverter plus motor speed control unit. Wind speeds of 3 m/s 16 m/s were obtained in the wind tunnel over the test section. The hot-wire anemometer position was 2.184 m upwind from the wind-tunnel exit. This area had a stable, uniform flow condition. At this point the working section was 1.219 m wide and 0.6096 m high.
The purpose of building this wind tunnel was to facilitate testing of airfoils at low speeds at the University of the West Indies, St Augustine campus. The primary aim was to accommodate experiments that require using airfoils prototypes and placement of endplates in the test section geometry. The main design criteria were: 1. Open circuit wind-tunnel. 2. Good flow quality. 3. Contraction ratio, CR, of 6-9. 4. Test section aspect ratio of 2 and the maximum test section length possible in the available space. 5. Flow speed in the test section between 3 m/s to 15 m/s. 6. Low noise level. 7. Low cost.
Along with wind tunnel geometry, such as contraction ratio, these flow features were controlled by turbulence manipulating devices, located upstream of the test section. The two most effective turbulence manipulators are honeycomb and mesh screens, each of which serves a specific purpose [5]. The reduction in turbulence intensity for flow through a screen is a function of the screen's pressure drop coefficient (K), which can be calculated using the following formula (equation 1) developed by Wieghardt [8], as recommended by Bradshaw & Mehta [9]. = 6.5 (1) In this equation, β is the open area ratio, u is steamwise fluctuating velocity, v is vertical fluctuating velocity and d is the cylinder diameter.
Turbulence reduction increases with the pressure drop coefficient. From equation 1, K increases with decreasing β; however, Bradshaw & Mehta [9] stated that for β < 0.58 flow non-uniformities starts to develop. Therefore, optimal turbulence reduction is achieved when multiple screens of relatively high open area ratio (β > 0.58) are utilized in series. In this case β is given by equation 2.
where d is the wire diameter and L is the length of the screen

Test Section Assembly
The test section was built from 6.35 mm thick Plexiglas and attached together with silicone and L-Brackets. The dimensions of the test section assembly were 60.96 cm x 60.96 cm and 121.92 cm in length. Fig. 2 shows a picture of the completed test section.

The Diffuser Assembly
This assembly housed the fan and electrical wiring for the fan and fan controller. The diffuser assembly was fabricated using 9.525 mm plywood. The first part of the Diffuser Assembly's construction was the Drive Section. This composed of a 77.47 cm diameter fan and the 90.17 cm x 90.17 cm aluminum fan housing (Fig. 3).

Fig. 3. The fan assembly
The area of the Drive Section (the large end of the Diffuser) was 0.79 m 2 which is 2.127 times the area of the end of the Test Section (also the area of the small end of the Diffuser). According to Cockrell and Markland [10] best results are achieved for a ratio is 2:1. The dimension of the diffuser body was 0.790 m 2 tapered to 0.403 m 2 with a length of 1.575 m. The plywood was then attached to one another with nails and silicone was then used to seal the joining. The completed Diffuser assembly is shown in Fig. 4.

Contraction Cone Assembly
This assembly was at the forward end of the tunnel into which the air flowed as it was drawn in by the fan at the back. This assembly consisted of the Contraction Cone and the Settling Chamber. The body of the contraction cone was also made with 9.525 mm plywood. Nails were used to connect the plywood together and sealed with silicone.

Settling chamber
This component was made with two honeycomb mesh and one screen in the large end of the Contraction Cone, to reduce turbulence to a minimum and improves airflow quality. The screen reduced streamwise velocity fluctuations. To find the screen's pressure drop coefficient (K) from equation 1, the open area ratio β from equation 2 was calculated using the wire diameter, d as 0.2 mm and l (length of screen) as 0.9 mm.
This value is greater than 0.58 (for β < 0.58 flow non-uniformities start to develop [9].
The honeycomb with its cell aligned in the flow direction will reduce minor fluctuating variations in transverse velocity with little effect on steamwise velocity because the pressure drop through a honeycomb is small. The honeycomb used was 90 mm long and the hexagonally shaped cells had a diameter of 15 mm i.e. the length to diameter ratio of the cells was about 6. For optimum benefit the cell length should be about 6-8 times its diameter [9].
The primary reason for using the honeycomb is that, with a sufficient length of about 6-8 cell diameters [4], it forms a very effective flow straightening device. The relatively low pressure drop of a honeycomb made it rather ineffective in reducing non-uniformities or fluctuations in the stream wise component but it is very effective in reducing cross-stream components [11].

Contraction cone body
The contraction section was located between the settling chamber and the test sections and served to both increase mean velocities at the test section inlet and moderate inconsistencies in the uniformity of the flow. Large contraction ratios and short contraction lengths are generally more desirable as they reduce the power loss across the screens and the thickness of boundary layers. Small tunnels typically have contraction ratios between 6 and 9 [5]. The area of the entry to the contraction cone was 2.643 square meters. The exit from the contraction cone to the test section was 0.403 square meters. The contraction ratio was therefore 6.55. A description of the main components of the wind tunnel and the associated cost are shown on Table 1.

CALIBRATION OF THE WIND TUNNEL
Experimental measurements were conducted to calibrate the laboratory built wind tunnel. Wind speed through the tunnel was controlled by a variable speed WEG 3.00 hp motor drive unit with a WEG CFW 08 vector inverter plus motor speed control unit. Calibration measurements included the boundary layer velocity profile and the wind velocity verses inches of water expressed in terms of milliamps, in the working section, in terms of the differential pressure values in the contraction cone. Calibration tests were performed using a hot-wire anemometer to determine the wind velocity as a function of differential pressure measured using a pressure transducer

Anemometer Selection
The hotwire anemometer used was the Amprobe TMA20HW. The Amprobe TMA20HW Hot Wire anemometer technology eliminates the use of bearings and rotating parts. The meter is durable and provides good and stable accuracy of the measurements. This is a highly accurate anemometer with 0.1% basic accuracy for precise measurements with an air flow measurements range of 0.3 m/s to 30.0 m/s. The electrical specifications of the Amprobe TMA20HW anemometer are listed below in Table 2.

Differential Pressure Sensor Selection
In order to determine the right pressure sensor for the wind tunnel, preliminary tests measured the pressure range for its location. This was done using the wind speed at different locations in the wind tunnel. Experiments performed in the Open Wind Tunnel with a hot wire anemometer showed that the velocity ranged between 3 m/s to 16 m/s. This was related to the dynamic pressure by using Bernoulli's Equation: This equation ignored the effects of compressibility, which was a good assumption for low speed flows. To find how this relates to the tunnel measuring system, or inches of a water column, equation 4 was used.

P = ρgh (4)
The unit used to measure the pressure difference was inches of a water column. To find out the largest pressures a combination of equations 3 and 4 was used. The dynamic pressure in Bernoulli's equation can be set equal to the dynamic pressure in equation 4.
= ℎ (5) To determine the height of the water column equation 6 can be used. This gave a value of 0.015 m, which is 0.592 inches.
Using this data, the Setra Model M264 Differential Pressure Transmitter was selected for this wind tunnel. This low-air pressure transmitter capable of sensing differential pressure in both negative and positive ranges. The Model M264 incorporates a tensioned stainless steel diaphragm to form a variable capacitor that will produce variation in the output signal (4mA to 20mA). It's specifications are listed below in Table 3.

RESULTS
The free-stream speed within the test section was set by measuring the pressure drop across the 6.55:1 contraction. The pressure drop was calibrated against the hot wire anemometer positioned in the mid-span of the test section. The contraction pressure drop was monitored by a Setra Model M264 Differential Pressure Transmitter, while the hot wire anemometer used was an Amprobe TMA20HW. The free-stream speed was obtained from a calibration curve of the wind velocity versus contraction pressure drop in inches of water expressed in terms of milliamperes. Figs. 5 and 6 shows a graphical representation of the results obtained.

Inches of Water
Velocity m/s where A is the area and P is the wetted perimeter.
For turbulent flow in a square duct the shear is nearly constant along the sides, dropping off sharply to zero in the corners. This is because of the phenomenon of turbulent secondary flows in which there are non-zero mean velocity, v and w in the plane of the cross-section. The Hydraulic diameter for this test section is 0.6096 m. Velocity (m/s) Distance from wall (mm)

Fig. 8. Wind velocity vs. wall distance
From documented data [14], b/a = 1. The effective diameter was calculated from equation 9.
The effective diameter was 0.6855 m. Velocity (m/s) Distance from wall (mm)

DISCUSSION
The first test was to produce the velocity profile for this wind tunnel. From Figs. 7 and 8 the results indicate a turbulent flow regime. A lease squares second-degree polynomial curve displayed an almost exact fit to the data with R^2 values of 0.998 and 0.995. The plot of mean velocity vs. distance from the tunnel wall was compared with the profile for turbulent flow in a square duct (Fig. 9).
The experimental data matched closely to the turbulent flow curve indicating that the wind speed was under 12 m/sec and the flow was turbulent. This can be possible if the fan was inducing sufficient unsteadiness in the flow.

Fig. 9. Turbulent velocity profiles
The second test was to establish a relationship between the measured wind velocities to the differential pressure drop in inches of water expressed in milliamps. Fig. 6 shows a second order polynomial relationship between the measured velocity and milliamps with a lease squares data fit showing an R value of 0.9995. Again, this is almost an exact fit, indicating that the data fits very well to the line. Also, from the graph (Fig. 6), it appears that electronic and mechanical noise was not a very big problem and the equation of the regression line, y = 0.0881x 2 + 0.146x + 3.7497, can be used to calculate the wind velocity when the wind tunnel is being used for testing airfoils via the differential pressure reading in the contraction cone. It indicates that the pressure sensor calibration was a second order polynomial fit.

CONCLUSION
1. A turbulent flow regime was noticed in the vertical boundary layer of the wind tunnel. 2. There is a second order polynomial relationship between the wind velocity and the differential pressure drop across the contraction cone section. The R squared value was 0.9995. Velocity-m/s Distance from wall (mm) 3. Electronic and mechanical noise was not a affecting the differential pressure transducer output. 4. The equation derived from the plot of wind velocity vs. milliamps (inches of water) can be used to calculate the wind velocity when the wind tunnel is to be used for testing airfoils. (Wind speed through the tunnel was controlled by a variable speed WEG 3.00 hp motor drive unit with a WEG CFW 08 vector inverter plus motor speed control unit). 5. The total cost was $ 775.00 US dollars.