vix.ing · top · new · best · stats · spec

Data-driven RANS closures for wind turbine wakes under neutral\n conditions

2020/09/16 by Julia Steiner, Steiner, Julia, Richard P. Dwight +3 · 1 citation
Engineering · Environmental Science · #Fluid Dynamics and Vibration Analysis #Wind Energy Research and Development #Wind and Air Flow Studies

paper · pdf · doi:10.48550/arxiv.2009.10816

Abstract

The state-of-the-art in wind-farm flow-physics modeling is Large Eddy\nSimulation (LES) which makes accurate predictions of most relevant physics, but\nrequires extensive computational resources. The next-fidelity model types are\nReynolds-Averaged Navier-Stokes (RANS) which are two orders of magnitude\ncheaper, but resolve only mean quantities and model the effect of turbulence.\nThey often fail to accurately predict key effects, such as the wake recovery\nrate. Custom RANS closures designed for wind-farm wakes exist, but so far do\nnot generalize well: there is substantial room for improvement. In this article\nwe present the first steps towards a systematic data-driven approach to\nderiving new RANS models in the wind-energy setting. Time-averaged LES data is\nused as ground-truth, and we first derive optimal corrective fields for the\nturbulence anisotropy tensor and turbulence kinetic energy (t.k.e.) production.\nThese fields, when injected into the RANS equations (with a baseline\nk-\ε model) reproduce the LES mean-quantities. Next we build a custom\nRANS closure from these corrective fields, using a deterministic symbolic\nregression method to infer algebraic correction as a function of the (resolved)\nmean-flow. The result is a new RANS closure, customized to the training data.\nThe potential of the approach is demonstrated under neutral atmospheric\nconditions for multi-turbine constellations at wind-tunnel scale. The results\nshow significantly improved predictions compared to the baseline closure, for\nboth mean velocity and the t.k.e. fields.\n

Cited by

Related