Parametric design and optimization model

Abstract

The offshore wind energy market is highly dynamic and one of the major challenges is to become more cost competitive against other sources of energy. The support structure of an offshore wind turbine shows a high potential for further cost reduction. Jules Dock is investigating in the replacement of a steel tower on a monopile foundation by a Glass Fiber Reinforced Plastic (GFRP) tower, which has a higher specific strength and better corrosion resistance compared to steel. This would enable lower transportation and installation costs and a potentially longer lifetime. Currently Jules Dock assesses the feasibility of a GFRP tower versus a steel tower using an aero-elastic simulation tool. However, such a tool is computationally demanding and requires detailed turbine data, which are often not available in a preliminary design phase. For this reason, a research project has been set-up to develop a parametric design model which can effectively be used without detailed turbine data and aero-elastic simulations, to identify the design driving constraints for preliminary designs and mass optimization purposes. The parametric design model analyses the natural frequency constraint and Ultimate Limit State (ULS). The natural frequency of the support structure is intended to be in the soft-soft region, which will result in the lowest mass possible as has been found in previous research. The natural frequency of the support structure is analyzed with a finite beam element model, in which the Rotor Nacelle Assembly (RNA) is modelled as a point mass. The interaction with soil is also included using a three spring model. For the ULS, the yield and buckling capacity of the support structure are analyzed. For the steel monopile and transition piece, dedicated design standards have been used. For the composite tower, however, these design standards are not existing and therefore analytical solutions from composite tube applications have been used and verified for this purpose. FEM analysis showed that the yield capacity can be determined exactly. The buckling capacity approximation in the parametric design model showed conservative results compared to a dedicated buckling analysis tool. In the parametric design model the maximum loading for the ULS constraints has been analyzed using a quasi-static load analysis method which includes turbulence and dynamic wave loading effects. This method has been compared with the time-domain simulation tool Phatas, which is integrated in Focus6. This comparison showed that also the tower top bending moment should have been included. If this is taken into account, the integrated load analysis method appears to approach the maximum loading for a very stiff tower quite accurately compared with the time-domain simulations. For flexible towers, however, the loads are less accurately approached, since dynamic interactions between wind and wave loading have appeared to be relevant, especially for load cases with a wave excitation frequency close to the natural frequency of the support structure. In the quasi-static load analysis method these dynamic effects are only taken into account for the submerged part of the support structure, but the time-domain simulations have shown that these effects will increase the maximum loading on the support structure above water level as well. For mass optimization and identification of the design driving constraints of this support structure, a genetic algorithm has been developed. Several optimization runs have been made, resulting in multiple feasible solutions with only small differences in support structure mass. While the buckling constraint drives the design of the monopile, transition piece and top of the tower, the lower part of the tower is design driven by the yield constraint. It has been concluded that the loading analysis method in the developed parametric design model should include the tower top bending moment, since this leads to a mass increase of up to 20% of the support structure mass. Next to that, it has been observed that the optimization algorithm tends to converge to solutions with a natural frequency to be as low as possible, such that the dynamic amplification of the wave loading is minimized. For these solutions, the loading analysis method in the parametric design model may have underestimated the maximum loading on the flexible support structure. This needs to be further investigated.