QBlade: a modern tool for the aeroelastic simulation of wind turbines
David Marten
Abstract
Wind turbines are large and complex machines which operate in a highly unsteady environment. Due to the spatially and temporally stochastic nature of the wind resource, a large amount of simulation data is needed when one wants to assess the lifetime of such a machine. Both extreme events, such as storms, strong gusts or significant wind direction changes and normal operation, representative of the stochastic characteristics of the wind site, must be present in the evaluated data set to obtain meaningful results. This necessitates the evaluation of many simulations, until the statistics of the synthesized wind input fields converge to those of field data, measured over long periods of time. The required time steps to resolve the key loading characteristics of the turbine are typically small, scaled to around 3° to 10° of rotor advancement. This leads to an overall number of time steps in the order of 10^6 to 10^7 that need to be evaluated for a lifetime assessment according to the IEC 61400-1 standard. This large number of time steps is a serious constraint when it comes to the selection of suitable methods for the aeroelastic simulation of wind turbines. An aeroelastic simulation tool requires two models, one for the simulation of the wind turbine aerodynamics and one for the simulation of the structural dynamics of the system. Both models are then coupled to simulate and resolve all machine relevant aeroelastic effects. As a part of this work the wind turbine simulation and design software QBlade has been developed. The main goal during this development process was to facilitate the usage of modern simulation models within the wind turbine design and certification process. Generally, higher order methods lead to more accurate simulation results, which enables wind turbine designers and manufacturers to lower the levelized cost of energy by applying smaller structural safety margins and more efficient aeroelastic designs. However, this improved accuracy comes with the penalty of higher computational costs. The application of these methods in the aforementioned certification and design process, where a large number of computations is required, is made possible through the steady increase of processing power of widely available consumer hardware and the application of massive parallelization, leveraging the enormous computational potential of modern graphics processing units (GPUs). This work presents the methods applied in the QBlade simulation code and gives reasoning for their selection and their classification within the range of different simulation methods that can be applied to model wind turbine aero-elastics. Finally, a range of examples for the applications of the aero-elastic simulation framework in QBlade are given.