Chances in wind energy

Abstract

Wind is becoming an ever more important source of renewable energy: installed wind turbine power now stands at 60,000 MW worldwide, providing 0.6% of world electricity demand. Still it is important that the cost of wind energy is brought down further, which means that wind turbines must be designed to be exactly as strong as necessary, but no stronger. Hence there is a need to investigate whether the conventional design procedure results in the right degree of conservatism, and if not, how it may be improved. The ideal is to make the design just conservative enough, i.e. to exactly attain the target failure probability. Because wind turbines tend to be located in remote areas, the target value is primarily determined by economic considerations, rather than by public safety issues. The aims of this work are: To quantify total uncertainty in the design procedure, and to find the relative importance of stochastic parameters influencing fatigue loads and strength. To conduct a comparative review of calculation models where necessary. To derive partial safety factors giving minimum unit electricity cost. The scope of the present research is limited to fatigue issues, since extreme loads have been investigated previously to some degree. An inventory of stochastic parameters is made; for each of the parameters the distribution is estimated, and the models currently used in wind turbine design (i.e. the procedures used to estimate characteristic parameters and how to use them in calculations) are reviewed. A limit state function is derived using the concept of life fatigue damage equivalent load range. With the First Order Reliability Method (FORM) and Monte Carlo simulation, yearly failure probabilities due to fatigue are estimated for a wind turbine that is designed exactly according to the standard, and installed following common site admission rules. A simple economic model is used to establish optimal partial factors. The partial factor values found for blades are somewhat smaller than in the standard, while values for hub, nacelle and tower are higher. The explanation for the latter is that two things are currently not taken into account in design calculations according to the standard: firstly, variation and bias in fatigue life prediction; secondly, the fact that a combination of many critical locations (for example in the tower) yields a larger failure probability than just one location. The sensitivity of the partial factor optimisation to changes in various assumptions is investigated. The main conclusions of the work are threefold: Given available data, a larger partial (load or material) factor should be used in fatigue design for cast iron and weld seams. However, the effect of this on design might be limited since hidden safety exists in the construction: material quality and hence fatigue strength are better than assumed, wind turbines are placed in climates that are more benign than they were designed for, and finally, dimensions may be determined by stiffness or extreme load considerations rather than by fatigue. The variation of the limit state function is determined mainly by uncertainty on fatigue strength and fatigue life prediction. Therefore the way forward is to accurately establish fatigue properties and calibrate fatigue life predictions for materials exactly as used in wind turbines. In this way variation may be reduced (and bias removed), and failure probability estimates may be refined. If better information is available, hidden safety may be removed and smaller partial factors used in calculations. The number of critical locations and correlation of loads and fatigue strength at different locations must be taken into account in calculations to establish failure probabilities, and must have influence on the partial factors to be used. Variation and bias of fatigue life predictions must be an explicit input to fatigue design calculations.