Structural design and dynamic analysis of a tension leg platform wind turbine, considering elasticity in the hull NTNU supervisor: Erin Bachynski TUD supervisor: Frank Sliggers Student M.Sc. Kristian Freng Svendsen 1. INTRODUCTION 2. RESEARCH AIM & CONTRIBUTIONS 3. ELASTIC STRUCTURAL MODEL FOR THE HULL Why floating offshore wind turbines? Offshore wind energy can help reach renewable energy targets. Bottom-fixed turbines are not economically feasible in water depths > 40 - 50m. Several deep-water locations have good wind conditions. Why tension leg platform wind turbine (TLPWT)? Promising for intermediate water depths (45-150 m) and deep water ( >150 m) Low production costs (economic advantage) Small motions (dynamic advantage) TLPWT research motivation: No multi-MW prototype exists. Unproven concept – models are needed which describe the physics of the system as accurately as possible. Computational tools used in this study: SIMO-RIFLEX-AeroDyn WAMIT Matlab HydroD ABAQUS This work has three main contributions. The hydrodynamic loads for large-volume structures depend strongly on frequency. Novel methods are needed to include both flexibility and correct hydrodynamic loads in the global analysis. The primary objective of this work is the testing of a new method for implementing frequency-dependent hydrodynamic loads for sections of the hull in an elastic structural model of the hull. Previous research on dynamic analysis of floating offshore wind turbines have been carried out with a rigid hull. In this work, a scantling design was made for the hull of an existing 5-MW TLPWT design by following DNV and ABS guidelines. FE software was used to estimate the hull stiffness, and an elastic structural model was generated for the hull. In this way, the hull elasticity’s effect on the global dynamic behavior of the TLPWT can be investigated. (ref. Bachynski, Design and dynamic analysis of TLP wind turbines, 2014) Previous research on dynamic analysis of floating offshore wind turbines have focused on global dynamic analysis. This work also considers the internal loads in the hull, and investigates dynamic amplification of internal loads due to hull elasticity. Scantling design performed and checked with: Column design made with DNV-RP-C202 Pontoon design made with ABS MODU and DNV-RP-C201 Stiffness estimated with FE software (ABAQUS): Bending and torsional stiffness Cantilever beam tests The hull can now be represented by beam elements in RIFLEX, using the equivalent stiffness estimations.

4. IMPLEMENTING HYDRODYNAMIC LOADS 5. RESULTS Dynamic amplification of motions and internal loads in the hull due to elasticity at 4 selected cross sections. The following simulations were compared: Frequency-domain analysis with a rigid model in WADAM Time-domain analysis with an elastic structure in SIMO-RIFLEX. 3-D panel model of the hull generated in HydroD: Compute frequency-dependent 1st-order radiation and diffraction pressures on each panel in WAMIT. Divide the hull into sections based on its geometry. Integrate panel pressures and compute frequency- dependent hydrodynamic loads for each section. Added mass matrix Radiation damping matrix Excitation force vector Let the loads for each section be represented by one SIMO-body. Insert the SIMO-bodies at selected nodes in the RIFLEX beam element model of the hull. A hydro-elastic coupling is obtained in which: RIFLEX beam elements represent structural deflection characteristics SIMO-bodies represent frequency dependent loads from 1st-order potential theory. NOTE: A Matlab script was written to carry out steps 3 and 4. Assistance in integrating the panel pressures was provided by researchers at MARINTEK. The total hydrodynamic load from all of the SIMO- bodies w.r.t. to WAMIT’s output for the hull as a whole: Total translational load components are exact. Total rotational load components are slightly underestimated. Tnat estimated from simulated decay tests and compared to previously published results for rigid hull model: 43 % increase in heave period 18 % increase in platform pitch period Time-domain simulations in combined wind-wave environments showed dynamic amplification of: Nacelle motions. Tower base bending moment. Tendon tension. Computed dynamic amplification factor 6. CONCLUSIONS 1. Implementation of frequency-dependent hydrodynamic loads was successful. Increasing the number of sections (reducing the section size) makes the solution approach WAMIT’s output for the hull as a whole. 2. The natural periods have increased. 3. Dynamic amplification observed for the nacelle motions, tower base bending moment and tendon tension. 4. Unreasonably high dynamic amplification observed for the vertical internal loads in the hull. This seems to be induced by unreasonably high heave motion amplification. Possible explanation: WADAM and RIFLEX are inherently not comparable in this particular case due to differences in modelling and/or computation methods. Pitch motion decay Visualization of SIMO-RIFLEX model on top of panel model Tower base bending moment comparison