Simulations of spatially evolving flows with local forcing

Abstract

Direct numerical simulation (DNS) and large eddy simulation (LES) have been used as a fundamental tool in the study of turbulence physics and turbulent flow control. Many findings have been observed from numerical simulations for the last two decades. Due to the limitation of available computer resources, however, most direct numerical simulations and large-eddy simulations have been restricted to simple flows, where periodic boundary conditions can be employed in the streamwise direction. The objective of this treatise is to perform spatially-evolving simulations with DNS and LES. In order to simulate spatially-evolving flows correctly, implementation of suitable inflow and outflow conditions is of prime importance. An inflow condition should supply turbulent kinetic energy continuously to maintain turbulence while a proper outflow condition makes the flow pass through the exit boundary with little distortion, so that the interior solution is not polluted by errors from the exit boundary. Implementation of an appropriate inflow condition is more difficult to deal with than the outflow condition, because the influence of the inflow condition persists over large distances downstream. This present study is composed of four parts. In part I, a comparative study is made of inflow conditions for spatially-evolving flows. A new spatio-temporal inflow condition is devised and evaluated with other methods, i.e., temporal, phase jittering and amplitude jittering, and random noise. These methods are validated by testing a large-eddy simulation of turbulent channel flow. Computational results are presented to disclose the ability of inflow conditions and to capture the turbulent statistics with correct phase information and dynamics. The present spatio-temporal inflow condition is found to be generally satisfactory in CPU time and data management. In part II, direct numerical simulations are made of instability in a spatially-evolving channel flow. A local surface suct...