Probabilistic and defect tolerant fatigue assessment of wind turbine castings

Abstract

The present thesis deals with probabilistic and defect tolerant fatigue assessment of wind turbine castings. To this end, two types of EN-GJS-400-18-LT ductile cast iron were investigated in this research, clean baseline material in the shape of casting blocks with different thicknesses and also defective material from a rejected wind turbine hub.

To establish the required P–S–N diagrams for safe-life design of wind turbine castings, fatigue specimens with different dimensions machined from baseline casting blocks with different thicknesses. Constant amplitude axial fatigue tests were performed on these specimens at room temperature at R = 0 and R = −1. Geometrical size effect, wall-thickness effect (technological size effect) and mean stress effect on fatigue strength of baseline EN-GJS-400-18-LT material were evaluated and analyzed. Statistical analysis of fatigue data was done by means of the Weibull distribution, and P–S–N diagrams were established. The established P–S–N diagrams showed that the Weibull distribution is well fit to the scatter of the experimentally obtained fatigue life data. Weibull’s weakest-link method was used to evaluate the size effect. It made a satisfactory prediction of the fatigue strength for specimens with different dimensions.

To study damage tolerant design of wind turbine castings, a rejected wind turbine hub was flame cut to several blanks and several defective fatigue specimens were machined from these blanks. Constant amplitude axial fatigue tests were performed on these specimens at room temperature at R = 0 and R = −1. Fatigue strength of baseline EN-GJS-400-18-LT was compared with that of defective material from the rejected wind turbine hub. The effect of graphite nodules and defects type, shape, size and position on fatigue strength of defective material was evaluated. The hypothesis that the endurance observed in an S−N test can be predicted based on the analysis of crack growth from casting defects through defect-free ‘base’ material was tested for the analyzed defective material in this research. It was shown that fatigue life of the analyzed defective cast component is controlled by fatigue crack growth and the slope of S − N curve for baseline EN-GJS-400-18-LT is different than the slope of S − N curve for defective EN-GJS-400-18-LT.

To perform random defect analysis of wind turbine castings, establish the scatter of fatigue life and obtain the probability of failure of these components, 3D X-ray computed tomography was use to detect defects in defective specimens and find the defect size distribution and density of defects (number of defects per unit volume). The obtained defect size distribution and density for the defective material was used in random defect analysis to establish the scatter of fatigue life for defective specimens.

Finally both safe-life and damage-tolerant design philosophies were used to evaluate the fatigue life of an EN-GJS-400-18-LT ductile cast iron block, representative of heavy-section wind turbine castings. The estimated S−N curves for the analyzed component based on these two methods were compared. It was shown that fatigue design of heavy section wind turbine cast iron components based on safe-life design philosophy may result in non-conservative design of these components