Divergence is Good!

During the last AWEA Conference on Wind Resource Assessment in Utah, our chairman asked us (the speakers): What’s best, CFD or LES?

We are not going to answer that question since both techniques have their own strengths and weaknesses and have proven to be very useful for their respective applications. The following is just a short reflection on how, within this context, an apparent weakness (“divergence”) may become a strength if secondary effects are taking into account.

Among CFD (Computational Fluid Dynamics) applications on wind energy, RANS (Reynolds Average Navier-Stokes) solvers are by far the most popular ones, to the extent that CFD has become synonymous with RANS in this context.

While it was somewhat of a revolutionary practice a few years ago, the use of RANS techniques to estimate the resource at the wind-farm scale is nowadays almost routine in many Wind & Site departments.

Meanwhile, LES (Large Eddy Simulations) is a version of CFD that requires far more computing power, and in the Vortex LES application, this technique generates results for smaller areas. Its technical challenges explain why LES has mainly been applied for academic purposes until very recently.

But, what is the main difference between RANS and LES? In simple terms, it is “convergence”:

The RANS approach requires the solution of flow equations to converge, which was a challenge or a problem in pioneering CFD days but is not really a technical problem today. Convergence is good. It produces stable and reliable solutions but introduces a concept that is seldom observed in nature: the atmosphere reaching a stationary state.

In contrast, LES solvers are “divergent” (in the sense that they are non-stationary) in nature. If there are nestly driven by different resolution motions (from synoptic to microscale), as in the Vortex application, such a multiscale approach far more closely resembles real-world atmospheric processes, although it is much more complex to handle.

However, being divergent (not reaching the stationary state), this approach gives rise to a very relevant (and useful) secondary output: time-series can be produced.

And, in our opinion, time-series will be the future standard in Wind Resource Assessment, as we will try to explain in another article (keep posted!).