Generation of Turbulent Inflow Data with a Prescribed Shear-Stress Profile
Abstract
We propose a new method for the generation of turbulent inflow data which avoids a long-range temporal correlation in the artificial velocity signal. It turns out that the method is fairly robust with respect to several of the parameters which can be chosen. These include the spectrum of the random turbulence introduced at the domain inlet, the streamwise extent of the control zone, the number of x, z-planes where body-forces are applied, the streamwise spacing between stations, and the vertical extent of the forcing zone. The method is not restricted to equilibrium situations since turbulence is forced to follow a prescribed target profile. The method is capable to both enhance and damp existing fluctuations in order to achieve a desired target, e.g. the shear stress average (u'v'). The method can easily be extended to consider other targets such as average U or rms profiles. Possibly, the occurrence of "adverse" forces at later stations could be avoided entirely by correcting only part of the "error" at the first station. Similar strategies might be applied in the context of hybrid RANS/LES approaches, where zones exist in which the grid is too coarse for resolving a sufficient range of energy containing eddies, e.g. near the wall. There, the generation of "supporting" stresses induced by body forces might be useful in order to achieve a smooth transition between regions which are modeled via RANS and via LES, respectively.
Document Details
- Document Type
- Technical Report
- Publication Date
- Aug 01, 2001
- Accession Number
- ADP013648
Entities
People
- A. Spille-kohoff
- H. J. Kaltenbach