Clearly insufficient numbers of 10 minute data sets

Info iconThis preview shows page 1. Sign up to view the full content.

View Full Document Right Arrow Icon
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: data show the typical decrease in turbulence intensity with increasing wind speed. Clearly, insufficient numbers of 10-minute data sets are available to allow statistically significant comparisons, but the trends are evident. The low deviations for the single cycle statistics are much more germane to the current analysis. For literally thousands of cycles, standard deviations for V∞ are less than 5% for mean velocities between 5 m/s and 25 m/s. As one would expect, the sonic anemometers show higher deviations than the cup or propeller instruments due to bandwidth. Standard deviations significantly exceed 5% below 5 m/s mean velocity and above 25 m/s. At low mean velocities, this is due to division by low mean velocities. At high mean velocities, this is caused by the tendency of sonic anemometers to give false readings from dust and debris. 0.40 Turbulence Intensity 0.35 0.30 0.25 0.20 0.15 0.10 0.05 0.00 0 10 20 Wind Speed (m/s) Phase II (Prop Vane) Phase IV (Sonic) Phase IV (Cup or Bi-vane) Phase V (Sonic) 30 Phase II 10-min. (Prop Vane) Phase IV 10-min. (Sonic) Phase IV 10-min (Cup) Phase V (Cup or Bi-vane) Figure 5: Turbulence intensity versus average wind speed. Figure 4: Number of cycles as a function of mean velocity (V∞) and yaw error (ϕ). Fig. 5 contains inflow turbulence intensity plotted for all cycles in two different ways. First, standard deviations were The same format and trends are apparent in ϕ (Fig. 6). Yaw error data collected from mechanical bi-vanes are in excellent agreement with the values for ϕ calculated from the sonic anemometer. For the single cycle statistics, yaw error standard deviations are less than 5% for all mean wind speeds between 7 m/s and 25 m/s. Thus, binning aerodynamic performance on cycle mean averaged V∞ and ϕ provide a 5 reasonable global overview of blade performance across a wide range of parametric changes. 25 20 15 3 10 5 0 0 10 20 30 Wind Speed (m/s) Phase II (Prop Vane) Phase IV (Sonic) Phase IV (Cup or Bi-vane) Phase V (Sonic) Normal Force Coefficient (Cn) Yaw Error Standard Deviation 30 portion of the chord pointing radially toward the tip with little deviation. If the inflow direction produces significant yaw error, or if the inflow magnitude creates a local angle of attack near stall, the resulting flow effect can be extremely dynamic. Cyclic three-dimensional unsteady separation and reattachment from both tower shadow and yawed inflow can occur within a single rotation cycle. This three-dimensional separation in near-stall and post-stall flow conditions is responsible for the large aerodynamic loads that exceed static stall values. 2 1.5 1 0.5 0 -10 Phase II 10-min. (Prop Vane) Phase IV 10-min. (Sonic) Phase IV 10-min (Cup) Phase V (Cup or Bi-vane) 0 10 20 30 40 -0.5 Angle of Attack (deg) Phase II, 30% Span Phase II, 80% Span Phase IV, 47% Span Phase IV, 80% Span Figure 6: Yaw error standard deviation versus average wind speed. COMPARISON WITH TWO-DIMENSIONAL RESULTS Normal force coefficients (CN) for both blades and all span locations are shown as a function of local α in Fig. 7. There is excellent agreement with two-dimensional wind tunnel data at α’s lower than static stall (α=15.2°). Above stall, data corresponding to 30% span show the most radical departure, with CN values reaching values of nearly 3.0. For comparison, Cp on the suction surface of an S809 at a Reynolds number of 500,000 never decreases below a value of -5.0. The peak CN occurs at approximately 0.9 [2]. At a Reynolds number of 2,000,000, the peak CN increases to 1.1, again with Cp minima remaining above -5.0 [9]. Notably, in near and post-stall flow, CN values for all span locations greatly exceed two-dimensional wind tunnel data. Associated Cp minima decrease substantially below static minima, as documented below. Visualization using fine thread tufts fixed to the upper blade surface shows attached and twodimensional flow until the onset of stall. As separation initiates at the trailing edge and propagates forward with increasing α, flow over the upper surface becomes immediately t...
View Full Document

Ask a homework question - tutors are online