The main focus of this work is to investigate experimentally the transition to turbulence of a yield stress shear thinning fluid in Hagen-Poiseuille flow. By combining direct high speed imaging of the flow structures with Laser Doppler Velocimetry (LDV), we provide a systematic description of the different flow regimes from laminar to fully turbulent. Each flow regime is characterized by measurements of the radial velocity, velocity fluctuations, and turbulence intensity profiles. In addition we estimate the autocorrelation, the probability distribution, and the structure functions in an attempt to further characterize transition. For all cases tested, our results indicate that transition occurs only when the Reynolds stresses of the flow equals or exceeds the yield stress of the fluid, i.e. the plug is broken before transition commences. Once in transition and when turbulent, the behavior of the yield stress fluid is somewhat similar to a (simpler) shear thinning fluid. We have also observed the shape of slugs during transition and find that their leading edges to be highly elongated and located off the central axis of the pipe, for the non-Newtonian fluids examined. Finally we present a new phenomenological approach for quantifying laminar-turbulent transition in pipe flow. This criterion is based on averaging a local Reynolds number to give ReG. Our localised parameter shows strong radial variations that are maximal at approximately the radial positions where puffs first appear during the first stages of turbulent transition.

The main focus of this work is to investigate experimentally the transition to turbulence of a yield stress shear thinning fluid in Hagen-Poiseuille flow. By combining direct high speed imaging of the flow structures with Laser Doppler Velocimetry (LDV), we provide a systematic description of the different flow regimes from laminar to fully turbulent. Each flow regime is characterized by measurements of the radial velocity, velocity fluctuations, and turbulence intensity profiles. In addition we estimate the autocorrelation, the probability distribution, and the structure functions in an attempt to further characterize transition. For all cases tested, our results indicate that transition occurs only when the Reynolds stresses of the flow equals or exceeds the yield stress of the fluid, i.e. the plug is broken before transition commences. Once in transition and when turbulent, the behavior of the yield stress fluid is somewhat similar to a (simpler) shear thinning fluid. We have also observed the shape of slugs during transition and find that their leading edges to be highly elongated and located off the central axis of the pipe, for the non-Newtonian fluids examined. Finally we present a new phenomenological approach for quantifying laminar-turbulent transition in pipe flow. This criterion is based on averaging a local Reynolds number to give ReG. Our localised parameter shows strong radial variations that are maximal at approximately the radial positions where puffs first appear during the first stages of turbulent transition. / Applied Science, Faculty of / Mechanical Engineering, Department of / Graduate

Many flow regimes in Nuclear Reactor Safety Research are characterized by multiphase flows, with one phase being a continuous liquid and the other phase consisting of gas or vapour of the liquid phase. In dependence on the void fraction of the gaseous phase the flow regimes e.g. in vertical pipes are varying from bubbly flows with low and higher volume fraction of bubbles to slug flow, churn turbulent flow, annular flow and finally to droplet flow. In the regime of bubbly and slug flow the multiphase flow shows a spectrum of different bubble sizes. While disperse bubbly flows with low gas volume fraction are mostly mono-disperse, an increase of the gas volume fraction leads to a broader bubble size distribution due to breakup and coalescence of bubbles. Bubbles of different sizes are subject to lateral migration due to forces acting in lateral direction different from the main drag force direction. The bubble lift force was found to change the sign dependent on the bubble size. Consequently this lateral migration leads to a de-mixing of small and large bubbles and to further coalescence of large bubbles migrating towards the pipe center into even larger Taylor bubbles or slugs. An adequate modeling has to consider all these phenomena. A Multi Bubble Size Class Test Solver has been developed to investigate these effects and test the influence of different model approaches. Basing on the results of these investigations a generalized inhomogeneous Multiple Size Group (MUSIG) Model based on the Eulerian modeling framework has been proposed and was finally implemented into the CFD code CFX. Within this model the dispersed gaseous phase is divided into N inhomogeneous velocity groups (phases) and each of these groups is subdivided into Mj bubble size classes. Bubble breakup and coalescence processes between all bubble size classes Mj are taken into account by appropriate models. The inhomogeneous MUSIG model has been validated against experimental data from the TOPFLOW test facility.

This work determines a suitable value for a dispersion coefficient to be used in the One-Dimensional Advection-Dispersion equation to model dispersion within sewer collection systems. Dispersion coefficients for sewer systems have only recently begun to be studied, and there is not yet an established value that is commonly accepted. The work described in this paper aimed, through observational study, to find a suitable value to be used. Salt tracers were placed in two separate reaches of sewer line. The first line studied was a straight, linear reach of sewer that included three manholes. The tracer was placed in the first manhole and the conductivity was measured at the two consecutive manholes downstream. These measurements were compared to a model developed using the 1D Advection-Dispersion Equation. The flow information and sewer network geometry was used in the model and the dispersion coefficient was adjusted to find a best fit. It was found that a value of 0.18 m2/s for the dispersion coefficient provided the best statistical match. The next reach of sewer that was studied was a reach with a 90 degree angle. This section was chosen to observe the effect that mixing has on dispersion, because of the change in direction of flow. The same procedure was applied, and an optimal dispersion coefficient of 0.22 m2/s was found. These values represent optimal dispersion coefficients under a specific set of conditions. It should not be assumed that they will provide accurate results in all circumstances, but are rather a base point for average flows under dry, stable conditions. Using these values inferences can begin to be made about dispersion characteristics throughout the entire sewer network. This can lead to specific engineering applications, and well as applications in other fields of study.

An experimental study of turbulent air flow in a pipe is reported in this paper. A determination was made of the mean velocity distribution and longitudinal mean turbulent velocity distribution, both in the turbulent core and boundary layer for four different Reynolds numbers from 7300 to 58300. A traversing mechanism was designed in order to measure the turbulence correlations between two points. The variation of the macro scale length, one of the fundamental quantities in recent statistical turbulence theory across the pipe diameter was calculated for Reynolds number equal to 58300, by integrating the correlation curves. The turbulent momentum diffusivity at the center of a pipe was calculated from the correlation study and the dimension less value was found to be 0.111. Ethylene gas was injected into the center of the pipe, and in order to investigate the turbulent mass diffusivity, the concentration distribution curves of ethylene were measured at different test positions downstream from the injection point, for the same series of Reynolds numbers used in the turbulence measurement. A numerical method for calculating the diffusivity was developed. The values of diffusivity obtained. in these experiments show that the assumptions which were used by most of the authors, that of (turbulent mass diffusivity/turbulent momentum diffusivity) has a value between 1.0 to 1.6 is correct. / Thesis / Master of Engineering (ME)

An examination of various methods for mass flow rate measurements was undertaken to evaluate their applicability for measuring refrigerant R-134a two-phase mass flow in refrigeration and air-conditioning equipment. An experimental apparatus was constructed to generate the required two-phase flow conditions. A turbine and a venturi flowmeter were used together with either a capacitance transducer or a gamma densitometer to determine the two-phase mass flow rate. The time-averaged void fraction was measured using a capacitance transducer and a gamma densitometer. Their measurements were in good agreement. Hence, for mass flow rate measurements, the capacitance transducer was used as the void fraction meter because of its ease of operation. A number of models were used to combine the output of either the turbine flowmeter or the venturi flowmeter, with the void fraction measurement to estimate the mass flow rate. It was found that, within the range of experimental data tested in the present work, the venturi flowmeter, in conjunction with Chisholm's model, provided the best agreement with the experimental results. / Thesis / Master of Engineering (MEngr)

The following paper presents a one-dimensional numerical model for simulating transient thermohydraulic pipe flow based on the Method of Characteristics. In addition to mass and momentum conservation, the proposed scheme also guarantees compliance with the laws of thermodynamics by solving the energy equation. The model covers transient changes in fluid properties due to pressure changes, heat transfer and dissipation. The presented methodology also allows the computation of the transient temperature distribution in the pipe wall through an additional ordinary finite difference scheme. The numerical procedure is implemented in the commercial simulation software DSHplus. The capability of the code is examined by comparing the simulation results with theoretical solutions and experimental data.

For the qualification and validation of two-phase CFD-models for medium and large-scale industrial applications dedicated experiments providing data with high temporal and spatial resolution are required. Fluid dynamic parameter like gas volume fraction, bubble size distribution, velocity or turbulent kinetic energy should be measured locally. Considering the fact, that the used measurement techniques should not affect the flow characteristics, radiation based tomographic methods are the favourite candidate for such measurements. Here the recently developed ultrafast X-ray tomography, is applied to measure the local and temporal gas volume fraction distribution in a vertical pipe. To obtain the required frame rate a rotating X-ray source by a massless electron beam and a static detector ring are used. Experiments on a vertical pipe are well suited for development and validation of closure models for two-phase flows. While vertical pipe flows are axially symmetrically, the boundary conditions are well defined. The evolution of the flow along the pipe can be investigated as well. This report documents the experiments done for co-current upwards and downwards air-water and steam-water flows as well as for counter-current air-water flows. The details of the setup, measuring technique and data evaluation are given. The report also includes a discussion on selected results obtained and on uncertainties.

Syftet med detta arbete är att simulera ett värmesystem i COMSOL Multiphysics, Pipe Flow Module, Non-Isothermal Pipe Flow som innehåller ekvationer och randvillkor för att modellera inkompressibel strömning och värmeöverföring i rör. Data om processen och dess komponenter har samlats in från industrin där arbetet är utfört och i vissa fall modifierats för att bättre beskrivas i programmet. Utifrån insamlad data har en modell byggts upp och två simuleringar har gjorts. En stationär för starten av systemet, den har sedan legat till grund för en dynamisk som simulerar förloppet från start till normaldrift. Tiden det tar för det aktuella fallet att nå drifttemperatur är 16 timmar. En felströmning upptäcktes samt att en av pumparna inte kommer att klara en start från 20 °C.

A comparison of available fluid-to-fluid scaling laws for scaling convective heat transfer at supercritical pressures showed that the ones suggested by Zahlan, Groeneveld and Tavoularis (ZGT) have some advantages. The applicability of the ZGT laws was tested for pairs of fluids including carbon dioxide, water or Refrigerant R134a. The conditions of previous measurements taken in the Supercritical University of Ottawa Loop with CO2 flowing vertically upwards in an electrically heated tube with 8 mm ID were scaled to equivalent conditions in R134a and new measurements of the heat transfer coefficient (HTC) were taken in the same tube using the latter fluid. The inlet pressure was 1.13 times the critical pressure (4.06 MPa), the mass flux was in the range from 212 kg/m^2 s to 1609 kg/m^2 s, the heat flux was in the range from 2 kW/m^2 to 137 kW/m^2, and the inlet temperature was in the range from 62 ℃ to 105 ℃. The HTC at equivalent conditions in water was also determined with the use of transcritical look-up tables. Average and linearly varying corrections to the ZGT scaling laws were derived by statistical analysis for each pair of fluids under NHT or DHT conditions. Such corrections reduced the standard deviation of the scaling error but did not eliminate the presence of large errors under many sets of conditions. As expected, scaling errors were in general larger for DHT than NHT conditions. The present results did not reveal any systematic and correctable dependence of the scaling error upon the mass flux or heat flux but showed that scaling errors became particularly large as the bulk temperature T_b approached the pseudocritical temperature T_pc. In conclusion, the ZGT scaling laws appear to be fairly accurate for the three pairs of fluids considered in the liquid-like region with T_b/T_pc ≤ 0.94 and possibly in the gas-like region with T_b/T_pc ≥ 1.02, whereas outside this range scaling errors could be significant. It was also found that the ZGT scaling laws do not scale accurately the onset of DHT in different fluids.