The demand for energy will continue to increase in the coming years and offshore wind energy shows great potential to become a key player in Europe’s renewable energy future. The wind flow offshore is more stable and the average wind velocity is higher than onshore. Moreover, no size restriction exists for offshore wind turbines. However, the levelized cost of electricity for offshore wind energy should be decreased in order to ensure that the transition to offshore wind energy is economically feasible. One way to realize this cost reduction is by optimizing the structural design of the offshore wind turbine. As the support structure is one of the main cost items of the offshore wind turbine, structural optimization of this structure should be investigated. In the current support structure design procedure, the turbine designer (TD) is responsible for the design of the tower, whereas the foundation designer (FD) is responsible for the design of the foundation and the transition piece. These designs are driven by the dynamic loads acting on these structures during the lifetime of the offshore wind turbine. Hence, accurate load predictions are a rerequisite to enable design optimization of the support structure. Therefore the TD runs a large number of aero-elastic simulations with the complete offshore wind turbine model to determine the global loads on the offshore wind turbine. From these simulations, loads or displacements at the tower/foundation interface are extracted and provided to the FD. Subsequently, the FD uses these interface responses in a post-processing analysis in order to obtain loads in the foundation structure. However, inaccuracies can arise at two points in the support structure calculation procedure: * In the aero-elastic model if a reduced or simplified foundation model is integrated in order to keep the aero-elastic model compact and to minimize computation costs. * In the post-processing analysis applied by the FD. To retrieve the response of the foundation model the FD can use either interface loads or displacements, applied either in a dynamic or a quasi-static analysis. By combining different model reduction and post-processing methods, various calculation procedures can be defined. In this thesis the accuracy of these different calculation procedures, that eventually determine the design of the offshore support structure, are analyzed. To this end, both a qualitative and a quantitative study are performed. In the first part of this thesis the different calculation procedures are analyzed from a theoretical perspective. Model reduction methods are explained and the impact of the reduction on the accuracy of the results is investigated. Furthermore, the accuracy of the post-processing methods is investigated and the differences between a quasi-static and a dynamic analysis and between a force and a displacement controlled approach will be outlined. The second part of this thesis concerns a case study in which the various calculation procedures will be applied to a representative offshore wind turbine model on both a monopile and jacket type of foundation. Finally, as fatigue is often the main design driver of the support structure, this case study is used to analyze the impact of an error in the response on the fatigue damage result. This study shows that the use of reduced foundation models in the aero-elastic model can decrease the accuracy of the results, as the reduced model is an approximation of the full model. Therefore, in order to obtain accurate results, the offshore wind turbine model with the reduced components should be spectrally and spatially converged within the frequency range of the external load spectrum. With respect to the post-processing methods, it will be shown that a quasi-static analysis provides accurate results only if the first free or fixed interface eigenfrequency of the foundation structure is higher than the highest excitation frequency in the external load for respectively the force or the displacement controlled approach. Moreover, as the first fixed interface eigenfrequency of a structure is higher than its first free interface eigenfrequency, a quasi-static displacement controlled approach will remain accurate up to higher excitation frequencies than a quasi-static force controlled approach. Furthermore, since both a monopile and a jacket based support structure are modeled, it is found that the accuracy of the different calculation procedures strongly depends on the type of foundation structure. This is reflected by the results from the fatigue calculations; as the monopile behaves in a quasi-static manner within the excitation bandwidth the fatigue damage results are relatively accurate for all calculation procedures. However, the jacket shows much more dynamic behavior and subsequently the fatigue damage results of the quasi-static force controlled approach are highly underestimated. Finally, it is shown that when expanding the response of reduced models, the fatigue damage results can be greatly improved through a quasi-static residual load correction. In conclusion, this work gives an overview of the accuracy of different calculation procedures to determine the design of an offshore wind turbine support structure. As the accuracy depends on several aspects (i.e. characteristics of the structure, use of reduced models, post-processing method and external load spectrum), several requirements are formulated for specific calculation procedures in order to make sure the obtained results are accurate. As a result, one can have more confidence in the optimized design of the support structure and over-dimensioning or the application of additional safety factors is unnecessary. In the end, this will lead to a reduction of costs for the support structure which thereby reduces the levelized cost of electricity for offshore wind energy.