Search

The key objective of the work so far was to do the science needed to design better microburners and microreactors for soldier power and other applications. Microcombustion has gone from "impossible" to routine. Microreactors for ammonia reforming shrank by more than a factor of 100. Computational models exist where none existed before., The original document contains color images.

Experimental studies of polymer drag-reduction were carried out in the laboratory of Thomas J. Hanratty. Companion studies were carried out in the polymer laboratory of Anthony J. McHugh so that the performance of drag-reducing polymer solutions can be related to their rheological and rheo-optical properties in simple non-turbulent flows. The work was influenced by the notion that aggregation of polymer molecules could be an important role. A remarkable finding is that molecular weight and molecular weight distribution need not change as the drag-reducing ability of the polymer solution degraded. Rheo-optical studies in a Couette device showed that effective polymer solutions develop turbidity over a range of shear rates characteristic of those used in the flow loop. This study shows that the drag reduction can be enhanced by using mixing and delivery procedures which enhance the formation aggregates.

This research was aimed at understanding turbulence generated at a wall. The goals were to investigate the structure of turbulence in the viscous wall region, to understand how production of turbulence is related to measured structural parameters, to develop models that will allow the computation of velocity and temperature fields, and to explore how the characteristics of wall turbulence are changed by external influences. The principal tools that were used are electrochemical techniques that measure turbulence characteristics in the immediate vicinity of a wall without interfering with the flow, the laser Doppler velocimeter, computer experiments, and pulsed-laser velocimetry. External influences to be studied are imposed flow oscillations, wavy walls and drag-reducing polymers.

The results of a study investigating the effects of an imposed pressure gradient on the viscous wall region of a turbulent flow are presented. Only imposed pressure gradients homogeneous in the streamwise direction are considered. A computational model that neglects streamwise derivatives is used to study imposed pressure gradients that are constant in time. The shear stress at the edge of the viscous wall region, and not the wall shear stress, is shown to be the proper shear stress to use in scaling the coherent eddy structures responsible for the production of the turbulent kinetic in the viscous wall region. Arguments are presented that show the minimum value corresponds to the point of imminent relaminarization of a turbulent flow. Measurements using electrochemical probes are presented of the variation of the wall shear stress, the turbulent fluctuations of the wall shear stress, and the transverse correlation coefficients of the streamwise velocity gradient at the wall that are caused by the imposition of a sinusoidal pressure gradient on the flow of a turbulent fluid through a two inch pipe. The correlation coefficient measurements are used to provide information about the variation of the streak spacing. Results are presented for pulsations having a constant centerline variation of 10% and dimensionless frequencies, ranging from 0.0012 to 0.0915.

Surface pressure variations for turbulent flow over solid wavy surfaces with a wavelength of 2 inches were studied. Measurements were made over waves of two amplitudes, 0.0123 inches and 0.03125 inches. The water flow rate and viscosity were varied so that a range of dimensionless wave numbers (scaled with the friction velocity and the kinematic viscosity) of alpha (+) = 0.0011 to 0.020 were covered. Pressure profiles are described using a three harmonic fit. A linear response, for which only a single harmonic is needed, was observed for all alpha (+) for the small amplitude wave and for small alpha (+) for the large amplitude wave. Of particular interest is the large phase shift of the pressure profile observed with increasing alpha (+). These results are used to test the quasi-laminar model for flow over wavy surfaces, widely used to predict wave generation and drag on small amplitude waves. Keywords: Surface pressure; Turbulence; Wavy surfaces.

Laser-Doppler velocimetry measurements have been made of nonseparated velocity fields over solid sinusoidal wavy surfaces. Time-averaged velocity and turbulent intensity data are given. The measurements were conducted over waves on the bottom wall of a rectangular channel with a cross section of twenty four by two inches, the two inch dimension being vertical. Two sets of velocity data were obtained at conditions corresponding to flows where linear and nonlinear shear stress responses are observed. The conditions were wave steepnesses 2a lambda, and channel Reynolds numbers of 0.03125, 6400, and 0.05, 38,800 respectively. The wavelength of the waves was two inches. The viscous wall region was of particular interest. Sufficient measurements were taken to give an accurate representation of both the streamwise and normal variation of the viscous wall region. The data was Fourier analyzed to determine the extent of nonlinearities, the wavelength average flowfields, and the amplitudes and phases of the velocity responses. The physical meaning of the data is interpreted in terms of pressure gradient effects along the wave surfaces. Comparisons of the data with prediction from simple eddy viscosity models are also given. Keywords; Turbulent flow; waves; and Computer programs. (Thesis)

Measurements are presented of the time variation of the wall shear stress caused by the imposition of a sinusoidal oscillation on a turbulent pipe flow. The amplitude of the oscillation is small enough that a linear response is obtained and the dimensionless frequency, omega +, is large compared to that studied by most previous investigators. The most striking feature of the results is a relaxation effect, similar to what has been observed for flow over a wavy surface, whereby the phase angle characterizing the temporal variation of the wall shear stress undergoes a sharp change over a rather narrow range of omega +. At omega + larger than the median frequency of the turbulence there appears to be an interaction between the imposed flow oscillation and the turbulence fluctuations in the viscous sublayer, which is not described by present theories of turbulence.

Measurements were made of the variation of wall shear stress and of the root mean square level of the fluctuating shear stress along a solid wavy surface with an amplitude of 0.014 in. and a wavelength of 2 in. The phase angle associated with the spatial variation of the wave-induced shear stress passes through a well defined maximum when measured as a function of the dimensionless wave number, alpha = 2 pi nu/lambda u star. Measurements of the wall shear stress and in particular the location of this maximum provide an extremely sensitive test of turbulence models which pay particular attention to the viscous wall region. The use of a quasi-laminar assumption, whereby the wave induced changes of the Reynolds stresses are ignored, gives good agreement with the measurements of the variation of the wall shear stress for large values of alpha. An equilibrium turbulence assumption in which the wave induced Reynolds stresses are assumed to adjust instantaneously to changes in flow conditions along the wave surface is valid for only small values of alpha. In the range of 0.0005 alpha 0.2 it is necessary to use a model for wave induced Reynolds stresses which takes account of the fact that the turbulence does not respond immediately to changes in the flow conditions. The mixing length model of Loyd et al. provides the best fit of the data. A modification of this model which takes into account the effect of streamline curvature on the mixing length provides an improved fit of the available pressure measurements over a wavy surface.

Conditionally averaged measurements of flow in the viscous wall region showed that strong spanwise flows at y(+) = 40 are related to strong spanwise flows at the wall that are of the opposite sign. At y(+) = 20 spanwise flows are always in phase with flows at the wall. Characteristics of the dominant eddies at the wall were explored by performing numerical computations using a fixed cell in space. Close to the edge of the viscous sublayer (y(+) = 15) use of a single spatial harmonic (lambda(+) = 100) was found to be adequate in predicting all important features of turbulence. At the edge of the viscous wall layer two spatial harmonics were found to be necessary to describe the flow field. The longer scale outer flow eddies produce most of the spanwise mixing in the wall region. The smaller scale lambda(+) = 100 eddies can account for momentum transport, Reynolds stress production and creation of mean u squared and mean v squared energy in this region. The physical picture of the flow resulting mean u squared and mean v squared from interaction of the wall with the outer layer is as follows. The mean flow energy is primarily produced in the outer layer by the mean pressure gradient and is subsequently transferred into viscous wall region. All the turbulent kinetic energy is produced by the v-w flow in the wall layer and is concentrated in the streamwise component of the velocity. This energy is convected out of the wall layer into the core region, where part of it is transferred to motion of the y-z plane. The cycle of events completes with the transport of mean v squared energy into the wall region and set-up of the v-w field.