Computational simulation of thunderstorm downbursts and associated wind turbine loads

Wind turbines operate in a constantly changing wind environment. This requires modeling and simulation of extreme events in which the wind turbine operates and a study of associated turbine loads as part of the design practice and/or site assessment. Thunderstorms are transient atmospheric events that occur frequently in some regions of the world and can influence the design of a wind turbine. Downbursts are extreme surface winds that are produced during a thunderstorm. They are both complex to model and their damaging effect on wind turbines has been noted in recent years. In the last few decades, downbursts have been the subject of studies in various fields--- most notably, in aviation. Despite their complexity, generally only empirical models based on observational data have been developed for practical uses. Based on such field data as well as laboratory tests, it is common to model a downburst as a jet impingement on a flat plate. The actual buoyancy-driven flow has been commonly modeled as an equivalent momentum flux-driven flow resulting from the impinging jet. The use of computational fluid dynamics (CFD) to model a downburst based on the idea of an impinging jet offers an alternative approach to experimental and analytical approaches.

Simulation of "downburst"' wind fields using a computational model and analysis of associated loads on a wind turbine operating during such events is the subject of this study. Although downburst-like events have been simulated using commercial CFD software, the resulting wind fields from such simulations have not been used as inflow fields for wind turbine loads analysis. In this study, the commercial CFD software, ANSYS FLUENT 12.0, is used to simulate downburst events and the output wind fields are used as input to loads analysis for a utility-scale 5-MW wind turbine. The inflow wind fields are represented by both non-turbulent and turbulent components---the former are simulated using FLUENT while the latter are simulated as stochastic processes using Fourier techniques together with standard turbulence power spectral density functions and coherence functions. The CFD-based non-turbulent wind fields are compared with those from empirical/analytical approaches; turbine loads are also compared for the two approaches. The study suggests that a CFD-based approach can capture similar wind field characteristics as are modeled in the alternative approach; associated turbine loads are as well not noticeably different with the two approaches.