Numerical Prediction of Experimentally Observed Behavior of a Scale-Model of an Offshore Wind Turbine Supported by a Tension-Leg Platform

Abstract

Abstract Realizing the critical importance the role physical experimental tests play in understanding the dynamics of floating offshore wind turbines, the DeepCwind consortium conducted a one-fiftieth-scale model test program where several floating wind platforms were subjected to a variety of wind and wave loading conditions at the Maritime Research Institute Netherlands (MARIN) wave basin. This paper describes the observed behavior of a tension-leg support platform (TLP), one of three platforms tested, and the systematic effort to predict the measured response with the FAST simulation tool using a model primarily based on consensus geometric and mass properties of the test specimen. The initial model was tuned through adjusting damping and stiffness of the platform and tower using free-response test results. Simulation results showed reasonable agreement with the experimental data for most quantities, including platform surge, tower-base bending moment, and tendon tension across a broad range of tests. Of particular note was the under prediction of platform-pitch acceleration for most scenarios investigated. In light of past findings for TLP pitch behavior, it was hypothesized that second-order sum-frequencies present in the experiment and not considered numerically were at least partially responsible for this error. For experimental conditions that led to slack-line events in the wave basin, FAST simulations sometimes predicted corresponding events. The prediction of the slack-line events and numerical stability of the following dynamic response was encouraging. However, multiple experimental tests showed slack-line events that were not predicted numerically. In particular, the FAST simulations failed to predict slack-line events that occurred while the rotor was stationary. It is believed that under prediction of wind drag on the tower, the exposed portion of the platform, and possibly the rotor was the primary cause for this modeling error. Increasing the accuracy for the pitch response would improve prediction of slack-line events. It is suggested that the following efforts be made to improve FAST's predictive capabilities for turbines supported on floating platforms:Allow specification of platform geometries other than a constant diameter cylinder for the computation of hydrodynamic forces. The need for adjustments to damping would likely decrease if the geometry of the TLP were modeled explicitly (e.g. the larger diameter portion near the bottom of the platform and the platform legs); Implement a dynamic model of the platform tendons so that contributions beyond cable tension are more explicitly considered;Consider second-order hydrodynamics that are believed to be a source of error in the prediction of the platform pitch; andDevelop wave loading routines for FAST that more readily allow for reproduction of the experimentally observed wave time history instead of only matching the wave spectrum, similar to how a variable wind time history can be explicitly defined. Further, future wave tank tests for a TLP supported floating turbine should consider: additional instrumentation to ascertain tower mode shapes to assist in classification of resonant frequencies as specific mode types;the importance of the pitch response of a TLP and how to accurately quantify the pitch response experimentally;higher sampling rates to better capture the response of the tower-bending behavior that, for the tested TLP, appears to be highly coupled with platform dynamics; and better scaling of rotor aerodynamics.