Refinements On Non-linear Term Of Turbulence Predictive Model In Atmospheric Surface Layer

Near-wall region of wall turbulence is difficult to measure in experiments directly,and is essential to the precision of large eddy simulations.The predictive models for turbulence proposed recently quantify the superposition effect and amplitude modulation between coherent structures in their linear and nonlinear term respectively,which can predict the time series of near-wall turbulence at moderate Reynolds numbers and has been widely concerned.However,a lack of calibration experiments at extreme Reynolds numbers hampers a reliable application of the model thereat;additionally,turbulence information in the upper and central logarithmic regions is also valuable for engineering applications,but has never been accurately predicted by existing models.More importantly,several defects in the characterization of amplitude modulation still exsit in the model,including the multi-scale effect and time shift in modulation,bringing challenges to the prediction at higher Reynolds number.Based on the high-quality data of near-neutral atmospheric surface layer flow field collected at the Qingtu Lake in western China,this paper complements the calibration and prediction of predictive models at friction Reynolds number ?O(10~6),and solves the problem of flow prediction at different heights in high Reynolds numbers wall turbulence.Discrepancies resulted from two mainstream forms of the linear term are studied in the atmospheric surface layer firstly,revealing that the linear part of model mainly affects the prediction of total energy and even moments;especially we elucidate the feasibility of spectral linear stochastic estimation in predicting turbulence upward and downward given an input in that a strong superposition occurs simultaneously in upper and lower regions.The function of the nonlinear term in model is discussed further:it redistributes the signal energy in time domain to establish the model's universal signal and determines whether the prediction of energy distribution and odd moments can be reproduced properly.The importance of multi-scale effect and time shift in amplitude modulation in quantifying modulation intensity is illustrated,and the improvements of the universal signal brought by their respective inclusion into the predictive model have been shown.Eventually these two factors are considered simultaneously,giving an accurate wall-normal evolution of the truncation wavelength and time shift,which modifies the nonlinear term of existing model and put forward a refined bi-directional predictive model that is capable of delineating the amplitude modulation precisely at every wall-normal position when an input is given.The model is confirmed in advancements of the prediction in energy distribution and odd moments,and what's more,has excellent potential to extend to other flows.