Experimental and Modeling Studies of Two-Point Stochastic Structure in Turbulent Pipe Flow

209 p.Thesis (Ph.D.)--University of Illinois at Urbana-Champaign, 1980.Experimental and modeling studies of large scale eddies of turbulent structure far from the wall region have been conducted in a vertical fully developed water pipe flow (D = 10.16 cm). The turbulent structure was investigated using space and space-time correlations. The axial and radial velocity components were obtained simultaneously at two points in the flow field with two hot film anemometers x-probes. Correlation measurements were performed initially with the fixed probe at Y/R = 0.5 from the pipe wall and with axial and radial separations of the movable probe. More detailed measurements were performed with the fixed probe at Y/R = 0.35 from the pipe wall with combined axial and radial separations of the movable probe to give net separations in planes inclined at 0(DEGREES), 10(DEGREES), 20(DEGREES), 40(DEGREES), 65(DEGREES) and 90(DEGREES) with respect to either the downstream or the upstream direction.Two-point two component axial and radial fluctuating velocity components were obtained utilizing standard constant temperature anemometry techniques. Mean flow, turbulent velocity and shear stress measurements were made to insure that these measurements were consistent with information in the literature. The signals were digitized and both space and space-time correlations for R(,11), R(,22), R(,12) and R(,21) were calculated. Isocorrelation contours, spatial scales, convection velocities as well as convected frame time scales were determined.Departure of these two-point correlations from isotropic behavior was modeled in terms of axisymmetric tensor forms, in which the turbulence is characterized by two scalar quantities defined by four free parameters which are related to the correlation length scales. This model provided a reasonably good description for this flow.Conditional averages of the velocity components have been measured for both time delay and spatial separation. Comparisons of these conditional averages with their computed linear and second order non-linear estimates, utilizing the mean square stochastic error approximation, showed good agreement. Linear estimation combined with the time series for u(x,t) has been used to produce a typical sequence of conditional eddy patterns which show strong resemblance with turbulence structures from flow visualization studies.U of I OnlyRestricted to the U of I community idenfinitely during batch ingest of legacy ETD