A resolution of the turbulence paradox: numerical implementation

Abstract

Sommerfeld paradox (turbulence paradox) roughly says that mathematically the Couette linear shear flow is linearly stable for all values of the Reynolds number, but experimentally transition from the linear shear to turbulence occurs under perturbations of any size when the Reynolds number is large enough. In [Li, Lin 2011], we offered a resolution of this paradox. The aim of this paper is to provide a numerical implementation of the resolution. The main idea of the resolution is that even though the linear shear is linearly stable, slow orbits (also called quasi-steady states) in arbitrarily small neighborhoods of the linear shear can be linearly unstable. The key is that in infinite dimensions, smallness in one norm does not mean smallness in all norms. Our study here focuses upon a sequence of 2D oscillatory shears which are the Couette linear shear plus small amplitude and high frequency sinusoidal shear perturbations. In the fluid velocity variable, the sequence approaches the Couette linear shear (e.g. in $L^2$ norm of velocity), thus it can be viewed as Couette linear shear plus small noises; while in the fluid vorticity variable, the sequence does not approaches the Couette linear shear (e.g. in $H^1$ norm of velocity). Unlike the Couette linear shear, the sequence of oscillatory shears has inviscid linear instability; furthermore, with the sequence of oscillatory shears as potentials, the Orr-Sommerfeld operator has unstable eigenvalues when the Reynolds number is large enough, this should lead to transient nonlinear growth which manifests as transient turbulence as observed in experiments. The main result of this paper verifies this transient growth.