No description available.

Stirring tanks are very important equipments used for mixing, separating, chemical reaction, etc. A typical stirring tank is a cylindrical vessel with an agitator driving the fluid and generating turbulence to promote mixing. Flotation cells are widely used stirring tanks in phase separation where multiphase flow is involved. Flotation refers to the process in which air bubbles selectively pick up hydrophobic particles and separate them from hydrophilic solids. This technology is used throughout the mining industry as well as the chemical and petroleum industries. In this research, efforts were made to investigate the multi-phase flow hydrodynamic problems of some flotation cells at different geometrical scales. Pitot-static and five-hope probes were employed to lab- pilot- and commercial-scale tanks for velocity measurements. It was found that the tanks with different scales have similar flow patterns over a range of Reynolds numbers. Based on the velocity measurement results, flotation tanks' performance was evaluated by checking the active volume in the bulk. A fast-response five-hole probe was designed and fabricated to study the turbulence characteristics in flotation cells under single- and multi-phase flow conditions. The jet stream in the rotor-stator domain has much higher turbulence intensity compared with other locations. The turbulent dissipation rate (TDR) in the rotor-stator domain is around 20 times higher than that near tank's wall. The TDR could be used to calculate the bubble and particle slip velocities. An isokinetic sampling probe system was developed to obtain true samples inthe multi-phase flow and then measure the local void fraction. It was found that the air bubbles are carried out by the stream and dispersed to the whole bulk. However, some of the bubbles accumulate in the inactive regions, where higher void fractions were detected. The isokinetic sampling probe was then extended to be an isokinetic borescope system, which was used to detect the bubble-particle aggregates in the tank. Aggregates were found in the high-turbulence level zones. The isokinetic sampling probe and the isokinetic borescope provide new methods for flotation tank tests. An experiment was also set up to study the dynamics of bubble particle impact. Four different modes were found for the collision. The criterion is that if the fluid drainage time is less than the residence time, the attachment will occur, otherwise, the particle will bounce back. / Ph. D.

This work presents the experimental study of the flow generated in the wakes of three three-dimensional bumps in the Virginia Polytechnic Institute and State University Boundary Layer Wind Tunnel. The three bumps examined are named bump 1, small bump 3, and large bump 3, and are the same test cases studied by Byun et al. (2004) and Ma and Simpson (2004) with a LDV system and a quad-wire hot-wire probe, respectively. Various experimental methods are used in this work: For measuring the mean velocity component in the planes examined, a seven-hole pressure probe is used with the data reduction algorithm developed by Johansen et al. (2001). A sixteen-hole pressure rake is used for boundary layer data on the sidewalls and ceiling of the test section and a Pitot-static probe is used to obtain mean velocity magnitude in the centerline of the test section. Specific techniques are developed to minimize the uncertainties due to the apparatus used, and an uncertainty analysis is used to confirm the efficiency of these techniques. Measurements in the wake of bump 1 reveal a strong streamwise vorticity creating large amounts of high moment fluid entrained close to the wall. In the wake of small bump 3, the amount of high momentum fluid entrained close to the wall is small as well as the streamwise vorticity. The flow in the wake of large bump 3 incorporate the characteristics of the two previous bumps by having a relatively large entrainment of high momentum fluid close to the wall and a low generation of streamwise vorticity. In the wakes of the three bumps, a pair of counter rotating vortices is created. The influence of large bump 3 on the incoming flow-field is found to be significant and induces an increase of the boundary layer thickness. By comparing LDV data and quad-wire hot-wire data with seven-hole probe data in the wakes of the bumps at the same locations, it is shown that uncertainties defined for a quasi-steady, non-turbulent flow-field without velocity gradient are bad indicators of the magnitude of the uncertainties in a more complex flow-field. A theoretical framework is discussed to understand the effects of the velocity gradient and of turbulence on the pressures measured by the seven-hole probe. In this fashion, a model is proposed and validated to explain these effects. It is observed that the main contribution to the uncertainties in seven-hole probe measurements due to the velocity gradient and to the turbulence comes from the velocity gradient. To correct for the effects of the velocity gradient on seven-hole probe measurements in an unknown flow-field, a technique is proposed. Using an estimation of the velocity gradient calculated from the seven-hole probe, the proposed model could be used to re-evaluate non-dimensional pressure coefficients used in the data reduction algorithm therefore correcting for the effects of the velocity gradient on seven-hole probe measurements. / Master of Science

A drag study was conducted on an oblate ellipsoid body in the Virginia Tech Stability Wind Tunnel. Two-dimensional wake surveys were taken with a seven-hole probe and an integral momentum method was applied to the results to calculate the drag on the body. Several different model configurations were tested; these included the model oriented at a 0° and 10° angle of attack with respect to the oncoming flow. For both angles, the model was tested with and without flow trip strips. At the 0° angle of attack orientation, data were taken at a speed of 44 m/s. Data with the model at a 10° angle of attack were taken at 44 m/s and 16 m/s. The high speed flow corresponded to a length-based Reynolds number of about 4.3 million; the low speed flow gave a Reynolds number of about 1.6 million. The results indicated that the length-squared drag coefficients ranged from around 0.0026 for the 0° angle of attack test cases and 0.0035 for the 10° angle of attack test cases. The 10° angle of attack cases had higher drag due to the increase in the frontal profile area of the model and the addition of induced drag. The flow trip strips appeared to have a tiny effect on the drag; a slight increase in drag coefficient was seen by their application but it was not outside of the uncertainty in the calculation. At the lower speed, uncertainties in the calculation were so high that the drag results could not be considered with much confidence, but the drag coefficient did decrease from the higher Reynolds number cases. Uncertainty in the drag calculations derived primarily from spatial fluctuations of the mean velocity and total pressure in the wake profile; uncertainty was estimated to be about 16% or less for the 44 m/s test cases. / Master of Science