The largest scales of turbulent wall flows
The largest scales of turbulent wall flows
复制标题
最大尺度的湍流壁流
DOI:
--
复制
发表时间:
1999
期刊:
影响因子:
--
通讯作者:
J. Jiménez
中科院分区:
文献类型:
--
作者:
J. Jiménez
The small scales of wall-bounded turbulent flows have received a lot of attention in recent years, especially in the near-wall region, in part because of the availability of direct numerical simulations that made their detailed study possible (Kim, Moin & Moser 1987). Since those simulations had necessarily moderate Reynolds numbers and little or no separation between their largest and smallest scales, the study of the former independently of the latter in them was difficult. The purpose of this paper is to study the flow scales which are of the order of or larger than the channel width or the boundary layer thickness. We will see that their contribution to the integral flow quantities is not negligible. The resolution of experiments and simulations is usually adjusted so that the discretized variables are smooth while the size of the numerical box, or of the experimental record, is chosen so that the correlation functions at distances comparable to the box size decay to a negligible level. The latter is intended to guarantee that there is little energy at scales larger than the box size, but it has to be interpreted with care. The energy in a flow that has been low-passed filtered at scales of order λ is proportional to the integral of the correlation function over separations longer than λ and decays slower than the function itself. Since singular spectra such as those in turbulent flows give rise to algebraically decaying correlation tails, it is possible to have correlations which appear to have decayed but which still have a substantial fraction of the energy in their tails. The peak of the one-dimensional spectrum is moreover typically at k = 0. This becomes important if the filtered signals are the interesting ones such as in acoustics, where sound attenuation decreases with wavelength and only long waves survive at long distances. Large structures are also physically interesting because long wavelengths imply long lifetimes and large volumes, and their integrated coherent effect can be comparable to those of the smaller ones even when their power per unit volume is not. Thus if the one-dimensional power spectrum of a signal tends to a constant E0 as k → 0, the power is contained in wavelengths longer than λ is O(E0/λ), but since the lifetime of each structure is proportional to λ, the total energy per structure is independent of the wavelength. As an example, even a small transverse velocity acting for a long time would lead to substantial modifications of the velocity profile. For a flow to be well represented in this sense implies that its resolved spectrum should decay at the lowest wavenumbers as well as at the highest ones, which may never be true in turbulent flows.