Despite substantial theoretical progress in recent years, predicting pipeline friction for slurry flows continues to provide challenges for research. The parameter of dominant importance has long been recognized to be the particle diameter and the limiting cases in the spectrum of flow phenomena are understood fairly well, at least at low and moderate solids concentrations.Slurries of very fine particles can be tested in viscometers and are often assumed to behave as continuous media. This assumption is often justified because the particles are flocculated and the interaction between the flocs provides a structure which prevents deposits from forming at low velocities. Although discrepancies between flow parameters observed in viscometers and in pipe flow do arise, these can be attributed to phenomena which are influenced by the finite size of the flocs (Bartosik et al., 1997).Slurries of very large particles are highly stratified when the particle density is different from that of the fluid and pipeline friction for these slurries has been recognized as a combination of fluid friction and Coulomb friction between the particles and the pipe wall. Velocities near the deposition condition are appropriate for pipeline transport and at low velocities, all the immersed weight of the particles will contribute to particle-wall friction. Although a mechanistic description of this type of friction was given by Newitt et al. (1 955), the derivation of Wilson (1 970, 1976) is more rigorous because it considers the stress transmission mechanism and facilitates further mechanistic modelling.The vast majority of "settling" slurries fall between these two limiting cases, displaying deposition velocities and frictional energy losses which may be regarded as combinations of kinetic "fluid-like" friction and Coulombic friction. An indication of the nature of the kinetic friction is the fact that the lower limit to total friction, e.g., when Coulomb friction is unimportant, has long been calculated using a fluid model which employs the density of the slurry and the viscosity of the carrier fluid (e.g., Spells, 1955). This method of calculating kinetic friction was found to be appropriate by Cillies et al. (1 985) for slurries of solids concentration less than 30% to 35% by volume. The model which they employed was based upon Wilson's two-layer model (1 976) and incorporated a number of features which allowed it to be extended to slurries containing particles that do not contribute Coulomb friction.In this model the total solids content is separated into two fractions: a) solids which contribute Coulomb friction; these solids are assumed to be concentrated in a lower layer; and b) solids which are suspended by lift forces derived from the fluid. These solids are assumed to be uniformly distributed within the pipe and combine with the liquid in the two layers to produce pseudofluids whose densities determine kinetic friction and provide buoyancy for the particles contributing Coulomb friction.'Author to whom correspondence may be add...

ecause the deposition condition represents the lower limit to operating velocities for most slurry transport systems, prediction of deposition velocities is an essential step in pipeline design. The increasing scale of mining operations has led to use of larger pipe diameters and higher solids concentrations for tailings transportation. Prudent design of these pipelines has stimulated laboratory investigations of deposition velocities so that the data base available for selecting operating conditions continues to develop.Of the numerous correlations which have been proposed for predicting true that finer particles constitute the majority of mineral slurries for which pipelines must be designed.Many of the correlations which have been presented in the past have considered data obtained in academic investigations. Although wide ranges of particle and fluid properties have been used, the pipe sizes have often been significantly smaller than modern industrial practice finds appropriate. Even when test pipelines of industrial scale are available for use to generate design data, the quantity of solids available for use in the slurry flow tests is often limited because the mine is not in production. For this reason scale-up of deposition velocities to higher solids concentrations and/or larger pipes is often necessary.To improve the correlations and to provide a guide for use in scale-up of laboratory test data, new and existing data obtained in the Saskatchewan Research Council laboratory using a range of pipe sizes have been reexamined and compared with previous work. This data has been obtained under isothermal flow conditions and the viscosity of the carrier fluid (water + fines) has been measured. Deposition Velocity CorrelationsThe form of the correlations was established by Durand (1 953) and may be stated as:In the correlation, FL was presented in graphical form as a function of particle diameter and solids concentration. To generalize the effect of 'Author to whom correspondence may be addressed. E-mail address: shook@ planet.eon.net Keywords: deposition velocity, slurry flow, solid-liquid flow, pipeline flow. particle diameter, Wilson and Judge (1 976) used the available experimental data to obtain an equation which included the particle diameter, pipe diameter and the particle drag coefficient:Equation (2) was considered to be applicable to slurries of particles with median diameters less than about 0.5 mm. A method for predicting velocities for coarser particles was deduced from Wilson's two-layer model and the combined results were presented conveniently in nomographic form (Wilson, 1979) for particles larger than 0.1 5 mm in pipes with diameters greater than 100 mm. The inherent lack of precision of the nomogram is realistic because deposition velocities are difficult to determine with precision and the effects of fluid viscosity and solids concentration are sometimes significant. 704

performance of the bitumen extraction process, i.e., poor conditioning results in poor recovery; and 2. No one, with the possible exception Karl Clark, the inventor of the water-based extraction process, has done more to further the industry's understanding of bitumen conditioning and extraction than Jacob Masliyah. His contributions to the industry are too numerous and too broad in scope to mention: they include important fundamental breakthroughs (e.g. the discovery of the formation of "slime coatings" on bitumen droplets in certain hard-to-process oil sand ores), industry-standard process models (e.g. the steady-state model of the primary gravity separation vessel); supervision of students who have themselves gone on to contribute to oil sands R&D, design, engineering and operations; and industry short-courses that are considered essential training for process engineers and operators alike.Fresh oil sand slurries were prepared and tested in a 100 mm pipeline loop at 37ºC to evaluate the effects of average flow velocity, slurry air content and air injection method (bulk or continuous) on slurry conditioning, i.e., the evolution of the in-pipe processes that promote gravity separation of bitumen-air aggregates from the remainder of the slurry. The potential separability of the bitumen in the slurry was evaluated using a slurry Conditioning Index (CI). When no air was injected into the slurry, the slurry CI was low (≤ 0.1), indicating very poor conditioning. An increase in flow velocity from 2 m/s to 4 m/s and injection of 5% air (by volume) dramatically improved the slurry CI, to ~ 0.6. The improved slurry conditioning observed at the higher velocity is explained by the increased force of fluid turbulence experienced by the particles and the greatly enhanced bitumen-air contact.

0f all the parameters governing pipeline friction for slurry flows, particle diameter has long been recognized as being of paramount importance. Very fine particles, with diameters less than about 10 pm, are usually flocculated and the interaction of these flocs provides the structure which inhibits settling and leads to nonNewtonian behaviour. Particles which are fine but too large to flocculate produce slurry friction which is velocity dependent in a manner resembling that of pure fluids (kinetic friction). At low and moderate concentrations (less than about 30% solids by volume) kinetic friction depends upon slurry density and the viscosity of the carrier liquid.Large particles, with settling velocities which are greater than the velocity fluctuations of turbulent flow, display friction which is insensitive to bulk velocity and which resembles the Coulomb friction of solids in sliding contact. The mechanistic model which has been proposed for these coarse-particle flows (Wilson, 1970(Wilson, , 1976 has been shown to be reliable in numerous subsequent investigations.Many industrial slurries contain particles with diameters small enough to be suspended in part by fluid turbulence and Wilson's coarse-particle model has been adapted to deal with these slurries (Shook et al., 1986; Shook, 1991, 2000). The kinetic and Coulomb mechanisms are assumed to be responsible for the total friction and since the dependence of each mechanism on pipe diameter is well understood, scale-up of laboratory test results to large pipelines should be reliable. The model is based upon idealizations of the concentration and velocity distributions within the pipe so that two superimposed regions (layers) are defined. The frictional contributions of these two layers are evaluated using the kinetic and Coulomb mechanisms. The relative importance of the two mechanisms has been deduced from experiments employing a wide range of particle diameters, pipe diameters and fluid viscosities.Because the model is intended for use with particles of intermediate size and because large quantities of solids are required for experimentation when large pipes are used, most of the laboratory data in the literature referring to narrow particle size distributions were obtained with sand slurries.Sand is a convenient material for use in experimental work, being relatively inexpensive and resistant to abrasive degradation. Its density is close to that of many minerals but is considerably greater than that of coal, which is usually handled in slurry form in cleaning plants. Although many experimental tests have been conducted with coal slurries and some of this data has been incorporated in the two-layer model, the friable character of many Canadian coals has made many test results difficult to interpret quantitatively. The present investigation was undertaken to examine a material with a density close to that of run-of-mine coal (Bakelite) and which was resistant to abrasive degradation. To complement the experimental results, tests were also undertaken using a...

T here are numerous economic and environmental benefits associated with a reduction in energy used in the oil sands extraction process. One method of reducing energy usage is to operate at lower process temperatures. The three oil sands operators (Syncrude Canada Ltd., Suncor Energy Inc. and Albian Sands Energy Inc.) currently extract bitumen from mined oil sands, and obtain much of their production while operating at temperatures much lower than those initially prescribed by the Clark Hot Water Process for oil sands extraction (Clark and Pasternack, 1932).Both laboratory-scale (Luthra et al., 2003;Wallwork, 2003) and pilotscale studies (Mankowski et al., 1999) have shown that air injection can improve bitumen recovery at lower (≤ 50°C) process temperatures. Other studies have shown (Malysa et al., 1999) that bitumen droplets tend to attach to air bubbles that are roughly the same diameter as the droplets themselves. Since air injection is utilized to improve bitumen recovery, it would be advantageous to understand how the air injection method, air concentration, and turbulent flow conditions in the oil sand hydrotransport pipeline affect the size distribution of bubbles. Oil sand hydrotransport pipelines operate at high solids concentrations of 30 to 38% solids by volume, and thereby represent an extremely challenging environment in which to determine bubble size distributions.The results reported here were obtained with a 100 mm pipeline that is 52 m in length, located at the Saskatchewan Research Council, Pipe Flow Technology Centre, Saskatoon, SK. Because of the complexity of air-water-solids experimental studies, we first looked at the behavior of two-phase air-water flows. Only these results will be discussed here. Subsequent papers will discuss the effect of solids (concentration and d 50 ) on bubble size distribution.Images of bubbles dispersed in water, flowing through transparent pipe sections located at three positions along the pipe, were obtained with a high-speed digital camera. Bubble size distributions were determined by analyzing the images. The effects of axial pipe position, measured as the distance from the air injection point, along with average water velocity, air concentration and injector nozzle diameter on bubble size distributions were studied. Distributions are characterized in terms of Sauter mean diameter (d 32 ) and maximum bubble diameter (d 99.8 ).In an earlier study (Razzaque et al., 2003), we evaluated the development of bubble size distributions with axial position for air-water flows in a 25 mm horizontal pipeline that was 35 m in length.