SwRI engineers design, build and test a prototype wind turbine array
By David L. Ransom, P.E. and J. Jeffrey Moore, Ph.D.
An early move toward renewable, wind-derived electric power in the United States was made in the 1930s when rural power distribution was too expensive for energy companies to pursue. Wind turbines of yesteryear were low-power devices that provided just enough power for radios and lights. As part of the movement to develop more renewable energy sources, wind-derived energy is once again a topic of growing interest. Today’s wind turbines are built at the utility scale. Instead of 300 or 400 watts on the ranch, utility wind turbines typically generate up to 3 megawatts, enough to power entire communities. This significant increase in power brings significant technical challenges. Issues such as mechanical reliability, production costs, and energy storage and distribution are of particular interest in today’s push for renewable energy. Southwest Research Institute has been involved in this growing research field, providing third-party engineering design reviews of machinery reliability, modeling torsional dynamics of drivetrain components and testing sub-scale prototypes of alternative concepts.
A team of SwRI engineers recently assisted a client in testing an alternative wind turbine concept that uses an array of several small wind turbines. In this concept, a single turbine of 200-meter diameter can be replaced with an array of 20 turbines, each having a diameter of 45 meters; thus each design has the same swept area, the area swept by the revolving wind turbine blades. The client’s needs included validating the design concept, demonstrating power performance equivalent to that of a single turbine of the same swept area and developing a validated computer model for future design studies. In support of the client’s development program, SwRI was contracted to validate the initial concept through experiment and computer simulation, looking for any possible blade-wake interactions between adjacent rotors, which might serve to either prevent some of the turbines from reaching full power or lead to adverse blade dynamic stresses caused by the flow interaction at the turbine blade tips. The intent of this study was to validate the wind turbine array concept and gain an initial understanding of the anticipated design challenges for a full-scale array. The SwRI team proposed to arrange multiple commercially available wind turbines in an array pattern and perform tests to evaluate the design for performance as well as mechanical integrity. SwRI researchers also used computational fluid dynamics to study the possible blade-wake or tip-vortex interactions between adjacent rotors.
Wind tunnel testing
As a result of the initial evaluation, the SwRI team designed, built and tested a turbine array prototype to verify key performance characteristics and to support future design work on larger turbine arrays. The array consisted of seven 400-watt, commercially available wind turbines constructed on a 25-foot-tall platform. The center turbine of the array was erected at a height of 15 feet to meet the centerline of the wind tunnel at NASA’s Langley Research Center in Hampton, Va. The Langley Full Scale Tunnel (LFST) is a National Historical Landmark significant to the advancement of aeronautical research dating from 1931 and continuing today through model tests of all current front-line U.S. jet fighter aircraft. The tunnel is 50 feet in length, 60 feet wide and 30 feet high.
Performance test results
Researchers added instrumentation to each turbine in SwRI’s seven-rotor array to provide data on center turbine power, blade strain and turbine speed. Several flow visualization features were included in the experimental rig to better understand potential turbine interactions.
The seven-rotor array was tested in a variety of conditions, including varying the number of active turbines, spacing between active turbines, wind speed and array yaw relative to wind direction. The results of all conditions were compared to the baseline performance results of a single turbine. The repeatability of the results was found to be around 4 percent. This demonstrates that even across a wide range of test article configurations, the maximum possible impact to turbine performance is less than 4 percent, suggesting that performance from operation in an array configuration does not significantly affect performance.
The SwRI team conducted a study in which the outer six turbines were initially spaced at 2 percent of the diameter relative to the center turbine. Spacing was then incrementally increased to a maximum of 16 percent. Spacing study results indicated that the influence of the neighboring rotors in the full seven-rotor array was not detectable within the repeatability of the test results. This indicated that any possible effects of rotor interaction on array performance were limited to less than 4 percent when compared to single-rotor performance. Tests run in the yaw condition did show a distinct drop in performance as expected, but the spacing influence in yaw was still negligible.
Using a smoke rake integrated with the LFST traversing mechanism, SwRI engineers injected smoke streams into the flow for overall streamline visualization. Initial results showed periodic fluctuations in the smoke streams corresponding to the vortex shedding frequency of the primary rake structure. This is an unsteady flow phenomenon in which vortices form behind an obstruction in the flow (the rake structure) and periodically detach, creating disturbances in the smoke flow. Because of this, additional smoke testing was performed by rotating the rake so that the outflow ports of the tines were not directly downstream of this wake.
The team attempted several variations in the smoke rake arrangements, such as spacing and restricting the number of tubes to obtain clearer images, but inherent turbulence in the flow made it difficult to generate a clear image of the flow field surrounding the turbines. In still images and video, no clear structure of the flow field was evident because of the turbines. Standard and slow-motion video with stroboscopic illumination synchronized to the rotation of the center turbine showed the flow field passing through the turbine planes with no apparent effect. This was evidenced by the fact that the strobe light could not “freeze” images of smoke patterns. Instead, the smoke continually moved in all recorded images.
SwRI researchers used a mixture of titanium dioxide (TiO2), kerosene and oleic acid to record flow patterns on the rotating blades. The mixture was painted onto the front and back sides of one of the center turbine blades, and the tunnel was subsequently brought up to operating condition. This allowed the TiO2 mixture to travel on the surface under the aerodynamic forces while the kerosene evaporated, leaving a trace of the surface flow patterns.
The SwRI team installed a grid of retro-reflective tufts on the array structure downstream of the turbine blade plane to examine flow structure in the wake of the turbines. During operation, the tufts reacted to the local flow field. Photographs of the tuft array showed the flow structure at the tuft grid for the two-rotor case when the center and upper rotors are separated by 2 percent. Capturing images in this configuration was difficult because of the added wake of the photographer and camera when filming from upstream of the turbine array. Nonetheless, the tufts showed some angularity in the wake of the two active turbines and were very well aligned with the mean flow direction everywhere else. A more straight-on view might have shown the dividing line between the swirled flow at the bottom of the top turbine (swirled flow to the left) and the top of the middle turbine (swirled flow to the right).
Strain gage measurements
Researchers installed strain gages on a single blade of the array’s center turbine. The gages were placed at three locations: outboard, midboard and inboard from the tip and data was transmitted through a telemetry system. Measured root mean square (RMS) strain distribution at the 12 m/sec wind speed shows typical cantilever beam shape with peak strain at the inboard (IB) location and minimum strain at the outboard (OB) location. This is an indication of the blade load distribution, which would demonstrate more variation between various spacing configurations if there was a real influence on turbine performance. These results show no clear difference in the RMS strain distribution, even with closely spaced turbines.
Researchers compared data from the frequency spectrum of the measured dynamic strain between the two-rotor data at 2 percent spacing and the full array data at 2 percent spacing, all at 9 meters per second (m/sec) wind speed. The synchronous response was elevated for the two closely spaced rotor tests compared to the single rotor results. This indicates some amount of blade wake interaction, despite the inability to detect interaction with either performance measurement or flow visualization.
Computational fluid dynamics (CFD) simulation
To support future design studies, the research team developed and analyzed a CFD model of the test rotor for various wind speed and turbine spacing conditions. With a CFD simulation that was validated by experimental data, SwRI engineers gained more insight into the structure of the flow through the turbine array. Similar to the wind tunnel tests, a parametric spacing study was performed in CFD simulation with similar results. The CFD-predicted power performance of the turbines was not strongly influenced by the array configuration. However, it was clear that there was some interaction between neighboring turbines, such as a reduction in tangential velocity between two co-rotating turbines.
The freestream wind flowing into the rotor blade usually generates a wake that develops downstream of the blade because of rotation. The wake can be divided into the central vortex generated by the hub and the tip vortex developed from the blade tip. The flow field results in an unusually complex three-dimensional vortex structure that was captured accurately with the CFD simulation for an array of two co-rotating turbines. The interaction between the vortex structure produced by two co-rotating turbines is quantified and power performance for the turbine array is predicted.
Based on the results of this combined experimental and analytical development program, SwRI researchers drew several conclusions regarding the operation of wind turbines in a closely spaced array.
Any possible effects of rotor interaction on array performance were limited to less than 4 percent when compared to single-rotor performance. Based on the single session repeatability, it was more likely that the influence of adjacent rotors was less than one percent for near-optimum tip speed ratios. Operation in a yawed orientation did not impact the array any more than it did a single rotor. Computational fluid dynamics tools can be used to model the multi-rotor concept, which can be very important in evaluating future wind turbine designs.
The results of this activity were encouraging, and the experimental program provided a large amount of data to support future wind turbine array development work. The performance data, flow visualization results and blade strain data can all be used to validate simulation tools as required in future design cycles. The progress made on the CFD modeling provided the knowledge necessary to more accurately model future designs with the ability to validate new ideas and techniques against existing experimental results. The effort expended in the current phase will continue to pay dividends as the wind turbine array concept matures and reaches field operation.
Published in the Summer 2009 issue of Technology Today®, published by Southwest Research Institute. For more information, contact Joe Fohn.