Experimental study of subgrid-scale stress and models using particle image velocimetry data

S. Liu
PhD Thesis, The Johns Hopkins University
July 1997, Baltimore MD

ABSTRACT: This research aims to study the properties of subgrid-scale (SGS) models using experimental data. It is composed of two projects. In the first project, the Particle Image Velocimetry (PIV) technique is used to measure the instantaneous two-dimensional velocity distribution on the plane along the jet axis in the far field of a round jet. Using the experimental data, three components of the subgrid-scale (SGS) stress are calculated and compared with existing SGS models. It is shown that the eddy-viscosity model displays very little correlation with the real stress. A new similarity SGS model, which can display substantially higher correlation with the real stress, is proposed. It is based on the 'resolved stress' tensor Lij, which is obtained by filtering products of resolved velocities at a scale equal to twice the grid scale. Dynamic versions of the new similarity model and mixed model are proposed and tested. In the second project, the time evolution of velocity distributions of initially nearly isotropic turbulence undergoing rapid axisymmetric expansion is recorded using the PIV technique. Rapid Distortion Theory (RDT) is used to predict the time evolution of the mean turbulent part of the SGS stresses. The experimental results are compared with the RDT predictions at different scales. The time evolution of the total SGS stress is studied and compared with different models. It is shown that the similarity and mixed models maintain high correlation with the real stress during rapid straining. The studies of SGS energy dissipation lead to time-dependent model coefficients for Smagorinsky and similarity models during rapid straining. The mixed model is shown to be characterized by time-independent model coefficients, and should be preferable in LES.