the Aerodynamic Offshore Wind, Henning Schmitt, Jörg R. Seume The Science of Making Torque from Wind 2012 October 9-11, 2012 Oldenburg, Germany
1. Motivation and 2. 3. 4. 5. Conclusions and Slide 2 / 12
Motivation Traditional wind turbine (WT) designs consider uncertainties by means of characteristic loads and partial safety factors. Main sources and types of uncertainties Inherent (basic) uncertainties Model uncertainties Human and organization errors Statistical distribution uncertainties Physical model uncertainties Slide 3 / 12 Geometric uncertainties Variations of the rotor blade geometry can arise due to the manufacturing process and the large dimensions of the rotor blades. Detailed information about geometric uncertainties and manufacturing tolerances are not available.
PDF PDF PDF PDF PDF Probabilistic methods use full statistical distributions of uncertain design variables. Model Input Uncertainties Physics-based Model Probabilistic Response Wind Input 1 Output 1 Material Input 2 Input 3 Output 2 Output 3 Bending moment Geometry Input n Output n Deflection Slide 4 / 12 Using Latin Hypercube Sampling (LHS) to investigate the effect of airfoil geometry variations on 1) the lift and drag coefficients, 2) the performance and loads of an offshore wind turbine (OWT).
Airfoil The geometry variations are modelled by a limited number of uncertain airfoil parameters: Max. thickness, t Location of the max. thickness, x t Max. camber, w Location of max. camber, x w Trailing edge thickness, t TE The variations are described by a truncated normal distribution. Slide 5 / 12 Parameter Mean Standard Lower and deviation upper limit t, x t, w, x w 0% 10% -5% / 5% t TE 0% 66.67% -100% / 100%
Physics-based Model and Airfoil geometries and the aeroelastic model are based on the NREL 5MW reference wind turbine (Jonkman et al. 2009). Slide 6 / 12
Lift and Drag Coefficients Profile DU 25 CoV Lift coefficient Drag coefficient Slide 7 / 12
Kolmogorov-Smirnov Goodness-of-Fit Hypothesis Test Slide 8 / 12 Significance level of 5% For angles of attack with attached flow: Lift coefficient normal distributed Drag coefficient lognormal distributed
Effect on the Wind Turbine Performance Assumptions and simulation settings: Changes of the airfoil parameters are uniformly distributed along the blade radius. Wind field without turbulence and shear No dynamic stall model used Wind speed is Rayleigh distributed with v mean =10m/s Slide 9 / 12 Geometry variations have only a small effect on the power curve and also on the annual energy production. The largest CoV of the power output is near the rated wind speed (CoV=0.2%). The annual energy production has a very small CoV of 0.1%.
Effect on the Damage Equivalent Blade Root Bending Moments Flapwise bending moment Edgewise bending moment The scatter of the flapwise bending moment has a CoV 4.4%. Geometry variations hardly affects the edgewise bending moment (CoV 0.4%). No clear and consistent distribution function which describe the scatter of the bending moments Slide 10 / 12
Conclusions Lift and drag coefficients: Airfoil geometry variations have a significant effect on the lift and drag coefficients. For angles of attack with attached flow the scatter of the lift coefficient follows a normal distribution. the scatter of the drag coefficient follows a shifted lognormal distribution. Performance and loads: Geometry variations have only a small effect on the power curve and the annual energy production. The edgewise bending moment is hardly affected by geometry variations. The largest variances of the flapwise bending moment occur near the rated wind speed. Slide 11 / 12
The quality of the results of probabilistic simulations highly depends on the assumptions made for the input variables and the physical model used. Using CFD simulations to get a better estimation of the stall behavior of the airfoils If available, consideration of information about more realistic geometry variations The rotor blade geometries can vary in many different ways: Influence of the radial distribution of the airfoil variances Variation of the chord and twist angle distributions along the blade radius Slide 12 / 12
Thank you for your attention! Contact: Leibniz Universität Hannover Institute of Turbomachinery and Fluid Dynamics Appelstr. 9 D-30167 Hannover Tel.: +49 511 762 2734 E-Mail: Ernst@TFD.Uni-Hannover.de Web: www.tfd.uni-hannover.de