| dc.description.abstract | The costs of wind energy have to be reduced in order to be competitive with conventional methods to generate electricity. The tower costs contribute significantly to the costs of a wind turbine. This thesis aims to suggest improvements in wind turbine tower design in comparison with industry standards nowadays in order to reduce the costs of wind turbine towers.
Nowadays tower flanges are modeled either not at all or as point masses in the aeroelastic code used for the load simulations. It has been investigated if modeling the mass, geometry and stiffness of the tower flanges influences the simulation results. Modeling the flanges lowers the first two eigenfrequencies (bending) by less than 1%. The mode shapes do not change and the differences in loads (+0.5%) and displacements (+2.5%) are found to be insignificant. The effect of geometry and stiffness is not contributing as much as modeling the mass. Modeling flanges as point masses is sufficient to represent the flanges in the aeroelastic code.
A constraint damping layer between the flange connections is proposed. Such a layer can be used to increase the damping of the tower. In this way, the fatigue loads on the tower can be reduced.
An improved flange design optimization method is suggested. A cost performance function is created, reducing the flange costs of more than 2.5% in comparison with the optimization method used nowadays within Siemens Wind Power.
Standardization of flange connections in wind turbine towers is considered. This is beneficial for the costs of handling equipment and tower internals, as project specific design and certification of these components can be omitted. Other advantages of flange standardization are risk mitigation and supply chain benefits. Standardization of the flanges with a fixed bolt pattern leads to a costs increase of almost 3% for the flanges. Additionally standardizing the flange width makes the flanges 7% more expensive. The cost benefits on tower internals and handling equipment are obvious, however not specifically investigated.
Improved fatigue life prediction methods are suggested. The tower sections can be designed less conservative if sector based fatigue loads will be considered. In combination with sector based SN-curves and stress concentration factors the tower can be orientated in such a way that the fatigue loads are less severe. In practice this means that the door frame should be directed in the least loaded direction. Up to 6% tower mass reduction can be realized for fatigue driven tower designs. However, this is only feasible with accurate load direction predictions.
The Effective Equivalent Stress Hypothesis (EESH), a multiaxial fatigue life prediction model proposed by literature, is implemented to investigate the influence of combined loading on tower welds. The phase angle between the stress components is a variable of the EESH and has significant influence on fatigue life predictions. A method is described that explains how the phase angle between the stress components can be calculated. Analyses showed however that for a tower weld the phase angle is hard to obtain. The original EESH is analyzed and has some drawbacks. The equivalent stress of the EESH is calculated with use of two factors: the effective damage sum ratio and the square root factor. The calculation of the effective damage sum ratio is a complex process for stochastic stress signals and the result is depending on assumptions made. Furthermore, the square root factor is independent of the shear stress magnitude, which does not correspond with reality. A simplified EESH is introduced in which this square root factor is omitted. This model is however not able to incorporate the out-of-phase angle and SCFs anymore. Both the original and simplified EESH are concluded not to be suitable methods for practical fatigue calculations of tower welds because of above mentioned drawbacks. Another method, the Gough-Pollard algorithm, is implemented. This algorithm is simple to apply and difference in damage contribution of the individual stresses is distinguished by using different SN-curves for the normal and shear stresses. For multiaxial fatigue calculations in tower welds, the Gough-Pollard algorithm is the preferred fatigue model out of the considered models. The method however does not incorporate the phase angle, which could be a suggestion for improvement.
According to the above multiaxial fatigue models a load safety factor between 1.16 and 1.40 is required when evaluating fatigue life with the conventional uniaxial method in order to incorporate the additional damage due to combined out-of-phase loading. Usage of correct multiaxial fatigue models instead of safety factors increases the accuracy of fatigue life predictions. More research should be done to generalize and validate the multiaxial fatigue models.
According to the above multiaxial fatigue models, load safety factors between 1.16 and 1.40 are required when evaluating fatigue life in tower welds with the conventional uniaxial method. In this way the additional damage due to combined out-of-phase loading is incorporated in the fatigue life prediction. However, usage of correct multiaxial fatigue models instead of safety factors increases the accuracy of fatigue life predictions. More research should be done to generalize and validate the multiaxial fatigue models. | en |
