Search Documents

By P. T. Bail, J. E. Bobek

In reservoirs containing soft shales interbedded with oil-producing sands, it is possible to have water movement from the shales into the sands during reservoir depletion. This paper presents the results of scaled flow-model tests performed to investigate the efficiency of oil displacement by this vertical influx of water into oil-producing sands. In these model tests, the effects of rate of influx and rock wettability were considered. Oil recoveries by shale-wafer influx and by water flooding were cornpared after 0.4 hydrocarbon volume of water entered the producing sand. This comparison was made for water-wet and oil-wet systems. In water-wet sand, shale-water influx produces slightly less oil than conventional water flooding; in oil-wet sand, shale-water influx produces about one-half as much oil as conventional water flooding. In another type of comparison, it was found that shale-water influx followed by water flooding recovers less oil than water flooding alone. This is true for both water-wet and oil-wet sands. A comparison of tests run at different rates of shale-water influx shows that oil recovery is insensitive to rate in water-wet sand; in oil-wet sand, oil recovery decreases with decreasing rate of influx. INTRODUCTION Localized subsidence of the earth's surface has accompanied the depletion of some oil reservoirs, particularly in California,1, 2 along the Texas Gulf Coast,3 4 and in the Bolivar Coastal fields of Venezuela. Subsidence occurs when fluid withdrawals reduce the pressure in the reservoir. The reasons for subsidence accompanying a reduction in reservoir pressure are not known with certainty.10° However, there is considerable evidence that compaction of soft shale beds adjacent to producing oil sands can be a major cause of this subsi-dence.3, 6, 10-12 When the shales compact, a limited amount of water moves from the shales into the producing sands. A schematic illustration of this oil-producing mechanism is shown in Fig. 1. As the pressure in the sand declines due to oil production, a predominantly vertical pressure gradient develops at the sand-shale boundary. This causes water to move from the shale into the sand as indicated by the solid arrows. This paper reports the results of scaled flow-model tests to investigate the efficiency of oil displacement by the vertical influx of shale water into producing sands. The model tests were not performed with the aim of duplicating the recovery behavior of any specific reservoir. The primary purpose of the tests was to obtain a better understanding of oil displacement by shale-water influx. A knowledge of the oil- and water-production characteristics of even a simplified reservoir system should be of value to engineers estimating the performance of reservoirs where shale-water influx is occurring in conjunction with other production mechanisms. The following limitations and assumptions apply to the model work. 1. The sand of the prototype reservoir is homogeneous and horizontal. 2. The shale-water influx is uniformly distributed along the top and bottom of the sand member being studied. 3. No effects of edge-water drive or dissolved-gas drive that might accompany shale-water influx are considered. PROTOTYPE RESERVOIRS Scaled-model tests were performed to study oil displacement from three prototype reservoir sections. The reservoir sections modeled consist of single sand layers out of a sequence of -sand and -shale layers, as illustrated in Fig. 1. Each prototype represents a homogeneous, high-permeability, unconsolidated sand layer. In the field, the prototype is oriented in such a manner that one end is at the radius of drainage of a producing well and the other end is pointed toward the producing well (Fig. 2). The sand and fluid properties and the dimensions of the three prototypes are given in Table 1. The capillary pressure relationships for the three prototypes are presented in Figs. 3, 4 and 5. The oil-water per-

By Rex B. McLellan, W. A. Oates, William W. Dunn

Vapor transport experiments have been carried out in order to determine the saturation solubility of carbon in cobalt and nickel with respect to graphite over a large temperature range. Some of the uncertainty existing in the thermodynamics of these systems has been removed and accurate estimates of the heat and entropy of solution of carbon in nickel and cobalt have been deduced. The relatitle partial entropies of carbon are considered to be predominantly vibrational in origin. ALTHOUGH the thermodynamic properties of solid solutions of carbon in cobalt and in nickel have been investigated previously, the results of these investigations show a marked lack of agreement amongst themselves. In this investigation simple vapor transport experiments have been undertaken in order to measure the temperature variation of the saturation solubility of carbon in nickel and cobalt with the object of removing to some degree the uncertainties in the thermodynamics of these systems. In the case of the C-Co system, early measurements of the saturation solubility of graphite in Co show results at the same temperature differing by almost an order of magnitude.1-5 Recently Schenck, Frohberg, and Jaspert6 measured the equilibrium between CH4-H2 mixtures and C-Co solid solutions at five temperatures ranging from 850° to 1050°C. Their values of the activity of carbon in the solid solution, deduced from a knowledge of the chemical potential of carbon atoms in the gas, varies linearly with composition at each temperature and the authors concluded that the solutions were Henryan. The saturation solubilities deduced from Schenck, Frohberg, and Jaspert's measurements are in good agreement with the solubility determined at 1000°C by Rao and Nicholson7 and the value at 1000°C deduced from the measurements of smith8 on the equilibrium between CO/CO2 mixtures and C-Co solutions. The object of this investigation was to check the results of Schenck, Frohberg, and Jaspert and also to extend the range of solubility data to higher temperatures. In the case of the C-Ni system, there is fairly good agreement between the values given by schaefer9 for the saturation solubility of carbon C, in nickel, the value given at 1000°C by Rao and Nicholson,7 and that deduced from Smith's8 measurements of the equilibrium between C-Ni solutions and CO/CO2 mixtures at 1000°C. The values of C, deduced from Schenck, Frohberg, and Jaspert's6 measurements of the carbon activity in nickel are also in fairly good agreement. However, there is a large discrepancy between the above set of measurements and those of Lander, Kern, and Beach," who measured the carbon solubility of nickel in equilibrium with graphite by depositing an excess of graphite on the surface of Ni-foils from a gaseous mixture and subsequently diffusing in the carbon. These results are considerably higher than those mentioned above. The determinations6 of carbon activity a, in nickel from the CH4-H2 equilibrium showed that at all the temperatures investigated, the ac-values varied linearly with concentration up to about a, = 0.6 and then began to increase with concentration much more rapidly. On the basis of auxiliary experiments in which C-Ni solutions were equilibrated with graphite, Schenck, Frohberg, and Jaspert concluded that their CH4-H2 solid equilibrations at high values of a, were spurious and a, behaved linearly with composition up to the saturation limit (a, = 1). The short linear extrapolations of Schenck, Frohberg, and Jaspert's a, data yield values of C, in good agreement with those of Lander, Kern, and Beach." The solubility data for nickel discussed here is presented in Fig. 1 in the form of plots of ln[Cc/(l - 2cc)] vs 1/T. The discrepancies mentioned are clearly seen. In this investigation a vapor transport method is used to measure C, over as wide a temperature range

Jan 1, 1969

By R. M. Asimow

Both smith1 and wells2 have shown that the dif-fusivity of carbon in austenite increases rapidly with increasing carbon concentration. Smith has shown, in addition, that part but not all of this increase can be attributed to the concentration dependence of the activity coefficient of carbon in austenite. From the work of Eshelby', it is known that, because of the image displacements, there is a uniform dilatation of the material in a body containing random centers of dilatation. It is thus tempting to explain the enhanced diffusivity as resulting from this apparent negative pressure caused by the interstitial carbon atoms.* That is, the activation free energy for diffusion is not directly a function of concentration if we consider that the alloy is subjected to a concentration-dependent pressure. To proceed with a quantitative solution, it is necessary to determine the activation volume AV for the diffusion process. An exact calculation of AV is quite a difficult problem and does not appear to have been solved as yet. A simpler procedure is to use the semiempirical, macroscopic theory of seyes' which relates the activation free energy AG to AV. Keyes has shown that for many types of diffusion processes the following relationship is valid Where p is the Gruneisen constant and x the compressibility. This relationship is based on the assumption that AG can be identified with strain-energy terms. In general there will also be contributions to AG due to a change in the energy of the valence electrons of the metal; thus the relation must be considered largely empirical. Surprisingly, for self-diffusion where it is well-known that the electronic contribution is quite important,' the Keyes relation works quite well for many systems. For interstitial diffusion where it might be expected that the relationship is on a firmer theoretical basis, the limited experimental data available is not in particularly good agreement as can be seen from Table I. However, it must be remembered that besides the uncertain basis of the Keyes formula there is considerable uncertainty in some of the experimental values of AV. The only real justification for using the Keyes formula in the present analysis is that it gives a result which satisfactorily solves the problem at hand and, further, there are no more reliable procedures for calculating AV available. The coefficient 3 in Eq. [I] is determined such that the room-temperature value for x is to be used. The activation entropy AS is estimated as 8.95 x 10"3 kcal per OK mole using the values in Table I1 and the Zener" relation AS = PAHIT, where p is a dimensionless thermal coefficient of Young's modulus which is to be evaluated at low values of T/T,, aH is the activation enthalpy, and T, is the melting temperature of the alloy. The resulting value of AG at 1000°C is 25.6 kcal per mole. The Gruneisen constant (p = 3cup/XC,), where p is the density, is evaluated from the room-temperature data in Table I1 after calculating C, from the relation. A value of p = 8.32 g per cu cm based on the value of a. in Table I is used. The result is C, = 5.0 x lo6 erg per g, H = 2.32, and AV = 3.41 cu cm per mole at 1000°C and infinite carbon dilution. It is assumed that the activation volume is not a function of carbon concentration. The theory could be refined to consider AV a function of concentration, but due to the uncertainties in the parameters there seems little justification for this. Carbon dissolved in austenite forms a nonideal solid solution, so that in writing an equation for. the diffusivity of carbon it is necessary to take into account the change in activity coefficient with concentration. The resulting equation is16

Jan 1, 1964

By J. D. Sudbury, J. E. Landers, D. A. Shock

The external corrosion of casing is one of the most important problems facing the present day production man. A 1953 NACE report' estimated the annual cost of casing corrosion at $2,200,000 in 22 fields studied. The fields represented only 10 per cent of the wells in the United States. It was significant that 77 per cent of the failures reported were due to external corrosion. A variety of means for combatting the casing corrosion problem have been discussed in papers such as the NACE survey' mentioned previously. No attempt to revicw these will be made in this paper. Thc scope of this paper has been limited to a discussion of the following major items: (I) an instrument for studying external electrolytic corrosion; (2) use of this instrument to determine the quantity of protective current required to prevent corrosion of casing; and (3) problem of accelerated external corrosion of the oil string at the base of the surface string. ELECTROCHEMICAL MECHANISM OF CORROSION It IS gcncrally belleved that one major cause of external corrosion is the existence of electrolytic cells operating between various formation waters. A theoretical discussion of the role of earth potentials in casing corrosion has been presented by L. de Witte and F. J. Radd.2 A brief picture of the electrolytic corrosion action between two differcnt formations (A and B) is presented in Fig. I. This picture is a review to many but is given to insure clarity. In the anode area, oxidation takes place as iron ions go into solution in formation A as a product of corrosion. In the cathode area. reduction takes place and hydrogen atoms are formed at the boundary between the pipe and formation B. The flow of electrons. as indicated. is into the pipe at the anode or corroding area and out of the pipe into the formation at the cathode. In order to conform with regulations of this journal, the flow of a conventional current (I! is also indicated in this sketch. All subsequent references will be to this so-called conventional current flow and will be indicated by the symbol (I). As this action takes place the layer of hydrogen developed on the cathode offers a resistance to the flow of current in the cell. It is said (hat the cathode has been polarized. This polarizing film can be rcmoved by addition of a so-called dcpolarizing agent which reacts with and removes this hydrogen film. Such an agent must be present in ordcr for this corrosion reaction to go at a rate sufficient to cause a severe corrosion problem. The two common depolarizing agents are: (I) oxy-sen. which readily reacts with hydro gen to form water, and (2) certain bacteria which use atomic hydrogen as part of their life process. It is not difficult to explain the presence of these agents. Oxygen may be present in some very shallow formations. Sulfate reducing bacteria are found to exist in nearly any ground where mud pits might be dug. THE CASING POTENTIAl-PROFILE Instrumentation A sketch of the equipment* uscd in determining casing potential profiles is shown in Fig. 2. The tool is run in the hole using a hoist truck with a three-conductor cable. This tool must be run with tubing and rods out of the hole and should be run with the flow line disconnected. By proper choice of electrode size it can be run in any size casing. The electrodes must be run in an open hole or a hole filled with oil. Measurements cannot be made while immersed in water. The tungsten carbide-tipped knives serve as the electrical robes and arc spaced 25 ft apart. The potential difference along a 25-ft section of casing is measured with a potentinmeter to an accuracy of about 5 mv. Current flowing in the pipe can then

Jan 1, 1956

By C. G. Dunn, F. W. Daniels, M. J. Bolton

IN recent investigations1. a data on relative grain boundary energies in silicon iron have been obtained. The present investigation is a continuation of this work along similar lines for the purpose of obtaining additional information on boundary energies. One aim has been to extend the data for a (110) series, which was formerly studied,' to small differences in orientation, since a recent theory3,4 based on a dislocation model of grain boundaries predicts a E. ? (A-ln ? )* form of energy curve for small differences in orientation. Another aim has been to provide data for a (100) series—all grains having (100) planes parallel with the surface of sheet speci-mens—and to compare the results with the predicted form of energy curve. Finally, information on the nature of approach to equilibrium conditions was to be obtained through observations on the movements of grain boundaries. Recently published papers2,3,5 have treated the mathematical problem of expressing equilibrium conditions for grain boundaries for the general case when boundary energy depends upon boundary orientation. The equilibrium equations, which relate equilibrium angles and grain boundary -free energies, contain additional terms that express the variation of energy with boundary rotation. It has been necessary in the present investigation, however, to neglect these additional terms since no data for their evaluation were available. Dropping the additional terms leads to the following approximate equations which were used: E12/sin ?3 = E18/sin?2 =E28/sin?1 or those of the form: E28 + E13 cos ?3 + E12 cos ?2 = 0 Experimental Procedure Except for sample F1, all specimens were made from two lots of silicon iron called C and L respectively. The compositions of these are listed in table I. In the preparation of 12 three-grain specimens for the present investigation, a controlled grain growth technique, which has been described previously,' was used. After preparation, the specimens were notched at the boundaries to shorten them as a means of providing more rapid approach to equilibrium in the anneals. Initial grain boundary angles were determined from micrographs taken at X500. Seven anneals, totaling about two days at 1300°C and two to four days at 1400°C, were run in an atmosphere of pure dry argon as described in a previous paper.' Boundary changes generally could be followed without metallographic surface preparation, because thermal etching occurred during the anneals. These observations indicated angle changes of as much as 35" in some instances, with the major changes occurring during the first anneal. Annealing was discontinued when the angles reached stationary values. Fig. 1 gives a schematic diagram for purposes of defining crystallographic and boundary directions for three grains with common plane parallel with the sheet. Differences in orientations (expressed by A) are the angles between [00l] directions. They can be calculated with the aid of the ?'s. The grain boundary angles can be calculated from the w'S. Results Results are given in table 11. Types of crystal-lographic planes refer to those parallel with the sheet. Orientations are given in terms of [00l] direc-tions as shown in fig. 1. Except for grain 3 of specimen H6, which was unintentionally tilted 7" out of a (100) plane, all grains generally were within 1" or 2" of the plane specified. The directions of boundaries were measured generally to within 1" on both upper and lower surfaces, and the average direc- tions were determined for use in calculating the grain boundary angles. Variations in grain boundary directions refer to deviations from average directions. Boundaries near the junction point for all specimens generally were very close to perpendicu-

Jan 1, 1951

By R. E. Collins, Paul B. Crawford

Potentiometric model studies have been made of ertically fractured reservoirs. All fractures originated at he well and extended into the reservoir for various distances. Studies were made to determine the areal sweep efficiencies for line-drive patterns. The areal sweep efficiency for the unfractured system studied war 70.6 per cent. It was found that if the vertical fracture war paralIel with the direction of flood, the areal sweep efficiency may be considerably reduced, approaching zero in some instances. If the fracture was at right angles to the direction of flood, the areal sweep efficiency may be greater than that for the unfractured system. The sweep efficiency depends on the length and orientation of the fracture and direction of the flood. It is concluded that considerable effort should be made to determine the nature and direction of fractures before initiating a flooding program. Such information is required if the maximum sweep efficiency is to be obtained and channeling is to be reduced to a minimum. INTRODUCTION The recent development and application of fracturing techniques to petroleum reservoirs has served to focus considerable attention on the effect such fractures may have on secondary oil recoveries. The opinion has been expressed that fracturing may not alter appreciably the water flooding possibilities. Others have thought that channeling may occur and that the creation of fractures may serve to substantially reduce the possibilities of conducting a successful water flood in the reservoir The purpose of this study was to obtain quantitative estimates of the effect of vertical fractures on the areal sweep efficiencies of line-drive flooding patterns. In presenting this work on vertical fractures, it is not to be inferred that the present fracturing techniques result in producing a vertical fracture. The types of fracture which are created are not known with certainty in every case. Some are believed to be horizontal, others inclined along bedding planes, and some data indicate vertical fracturing. Each fracture will probably exert a different influence on the sweep efficiency of the secondary recovery program; consequently each fractured well should perhaps be considered as a separate problem, requiring a careful analysis. Preliminary work on the effect of fractures on producing capacity of a well indicated that for practical purposes many fractures may be treated as having a nearly infinite permeability This assumption appears to be valid in some instances where the permeability of the fracture is only 10 or 100 times greater than the permeability of the matrix. It is known to depend on the type, orientation and length of the fracture. In realization that fractures of only 0.01 in. width have calculated permeabilities near 5,000 darcys, it is believed that many fractures may be considered as of near infinite permeability without incurring serious error. Primarily for this reason this study was made with fracture analogs of very great permeability compared to the matrix. FRACTURED SYSTEM ANALOG The pattern sweep efficiencies reported here were obtained with the aid of a potentiometric model. The application of this type model for solving reservoir problems has been described in the literature.1,2,3 The electrolyte reservoir was 20 in. wide by 30 in. long by approximately % in. deep. The liquid depth was constant, which corresponded to a constant permeability-feet product from the injection to the production well. The electrolyte reservoir was built to correspond to a unit in the line-drive pattern shown in Fig. 1. This figure shows part of an infinite array of wells having alternate rows of fractured and unfractured wells. The area within the dashed rectangle shows the area generally studied in the model. In using the model to represent a fractured system, a thin strip of copper was soldered to the well or wells and the length and orientation of the strip was made to conform to the vertical fracture under consideration.

Jan 1, 1955

By Robert M. Lewis

Can wet-ground mica be produced from mica schist ores, and if so, would its properties be comparable with products now on the market? This was the problem which prompted the development of a flowsheet for the beneficiation of mica schist. The fine-ground mica industry relies on weathered pegmatites and alaskite ores as sources of raw material. The mica concentrates obtained from these ores are relatively clean and require moderate beneficiation to meet color and bulk-density specifications. Mica obtained from mica schist is iron-stained and decrepitated and requires additional beneficiation to produce a wet-ground mica product. Despite these adversities, there are several advantages to be gained by mining and processing mica schist ores. The mica content of these ores is considerably higher than that of the pegmatites and alaskite ores now being mined, often containing in excess of 40% mica. This higher-grade mica feed should promote a lower mining and milling cost. The mica schist ores with their higher mica content would mean less tailings to handle; however, there would have to be a means for disposing of acid and sulfate wastes. A flowsheet was developed whereby mica could be recovered from mica schist ores and beneficiated in a way that would result in a high-quality end product.' The essential phase of the process involves using reducing reagents to increase the magnetic susceptibility of iron contaminants which then can be removed with a magnetic separator. General Treatment Procedure The mica is floated by either of two methods, an amine float in an acid circuit' or the U.S. Bureau of Mines (USBM) amine-fatty acid float in an alkaline circuit' The flotation mica is ground in a pebble mill to liberate contaminated particles and expose iron-stained surfaces. The concentrate is then conditioned with a reducing reagent and passed through a Frantz Ferrofilter for the removal of magnetic material. The nonmagnetic product is acid-leached to further reduce the iron contamination. The leached material is then ground to product size and checked for specifications. Discussion of Flowsheet and Its Development Flotation: The mica is concentrated by flotation using either of two methods: an amine float in an acid circuit, or the USBM amine-fatty acid float in an alkaline circuit.' The writer does not wish to elaborate on the relative merits of the two flotation procedures; however, after investigating ores from various localities, it was felt that the amine float in an acid circuit was more consistent when using a standard procedure. Iron Reduction and Magnetic Separation: This phase of the flowsheet may be considered the key to the process. It was developed when it became apparent that hot acid leaching had just about reached its maximum effective¬ness for color improvement. S. C. Sun and William Hirsh in "Hydrochloric Acid Leaching of Iron From Pennsylvania Aluminous Clays,"` reported iron extraction for -20-mesh minerals of 82.20% for hematite and only 33.46% for magnetite. The test procedure employs a 19% hydrochloric acid solution, 104°C temperature, and 10% solids. It was felt that magnetic removal of some of the iron contaminants before leaching therefore would be advantageous. Since the iron minerals present as contaminants vary in magnetic susceptibility, it was decided to experiment with reducing reagents as a means of influencing their magnetic properties. Zinc hydrosulfite or sodium hydrosulfite were the reagents used throughout the tests. The procedure developed consists of conditioning the mica concentrate in a closed container at approximately 5% solids with a reducing reagent. Approximately 10.0 lb of zinc hydrosulfite per ton of mica concentrate is used. The ferric iron is reduced to the ferrous state (magnetic) and oxidation is retarded by isolation from the atmosphere in a closed container. After the desired conditioning time has elapsed, the material is passed through a Frantz Ferrofilter where the magnetic particles are retained and discarded as

Jan 1, 1972

By R. A. Morse, C. R. Holmgren

The production of oil by water flooding can be substantially increased by the maintenance of free gas saturation in the reservoir during the flooding operation. This effect is accomplished by the alteration of oil relative permeability characteristics and the occupation by gas of pore space that would otherwise be filled with residual oil. The amount of reduction In residual oil can be calculated from appropriate water-oil relative permeability characteristics. This paper presents experimental data in support of the foregoing conclusions and an example of the calculations. The microscopic pore saturation concepts of the mechanism are discussed. A method of practical application to field floods is presented together with discussion of certain limitations. INTRODUCTION The presence of free gas has been reported by a number of investigators to significantly affect the oil recovery which can be obtained from sandstone flow systems by water flooding.1,2,3,4,5 The effect of gas, noted in every instance, has been to cause lower residual oil saturations than could be obtained by water flooding the same systems in the absence of free gas. The degree of improvement in recovery has been observed to vary widely, depending on the systems used and the conditions of the tests. The increased oil recovery obtained because of the presence of gas during a water flood has been variously attributed to changes in physical characteristics of the oil, selective plugging action of the gas, inclusion of oil mist in the free gas phase, and the additional sweeping or driving action of the free gas. All but the first of these suggestions imply changes in the displacement mechanism. The change in viscosity and inter-facial tension of the oil phase, within the pressure range used for all the experimental work, is certainly not sufficient to account for the differences in residual oil saturation noted dess there is a drastic change in the displacement process. One other effect which logically seems capable of causing differences in residual oil saturation of the magnitude noted in the experimental work is that of simple replacement. In a water-wet system containing oil, water, and gas, it is to be expected that the gas will exist inside the oil. This is the position of minimum free surface energy, since the gas-oil interfacial tension will be less than the gas-water interfacial tension. There is no apparent reason to expect that the existence of free gas within the oil phase should alter the saturation at which the non-wetting phase (now oil and gas) should become discontinuous and hence trapped so as to be unrecoverable by direct displacement by water. If this is the situation, then trapping of a certain percentage of gas saturation during water flood should result, at infinite water-oil ratio, in a like reduction of oil saturation below that attainable by flooding in the absence of free gas. It is visualized that the gas will exist as bubbles inside the discontinuous residual oil as illustrated in Fig. 1, with the size of the oil bubbles being substantially unchanged due to the presence of the gas. As a practical matter, it can be anticipated that the presence of a free gas saturation inside the oil phase will reduce the relative permeability to oil which will exist at any particular water saturation. This reduction will be caused by two factors — -the addition of the gas-oil interface, and the reduction of area available for oil flow in the pores containing gas. This reduction in oil permeability at any particular water saturation will result in water breakthrough at a lower water satu-

Jan 1, 1951

By G. R. Countryman, G. H. Thomas, I. Fatt

Procedures are described for scaling up turbulent friction pressure drops of inelastic fluids and laminar friction pressure drops of inelastic/elastic fluids in circular ducts. The laminar scale-up calculation method does not require the use of arbitrarily defined flow models, and can be performed directly from laminar capillary viscometry data, or by means of a numerical solution of the integral equation for Poiseuille flow, inputing true shear rate-shearing stress data from other simple shearing experiments. e.g. Couette and plane Poiseuille flows. The turbulent scale up procedure, based on the Dodge-Metzner correlation for inelastic fluids, requires evaluating the characteristic rheological parameters at the existing wall stress under laminar flow conditions. Since the wall stress is the quantity sought in a pipe flow scale-up problem, a trial-and-err or solution is indicated, and therefore a calculation method has been developed and programmed for a medium-size magnetic-drum-memory digital computer. Laminar flow data from other somces may also be adapted to this turbulent scale-upmetbod for inelastic fluids. Comparisons between predicted and experimentally observed friction losses for generalized Reynolds numbers ranging from 3 — 32.000 indicate an error band width of less than 10 per cent in circular tubes ranging in size from 0.5 to 1.89 in. in diameter. INTRODUCTION Considerable effort has been devoted to the problem of establishing an accurate procedure for predicting the laminar and turbulent flow behavior of non-Newtonian materials in pipes from data obtained with small scale laboratory or pilot plant equipment, as evidenced by the variety of reviews on the subject.1-3 Although there does not appear to be a universally accepted scale-up method for laminar flow, the various methods currently employed can be divided into two principal categories depending upon whether a functional relationship between shear rate and shearing stress is assumed, or whether a generalized correlation is adopted. The methods of Hedstrom4 and of Weltmann are typical of the first category, while the generalized correlation of Metzner and Reed6 illustrates the second category. Under appropriate conditions either approach yields a reasonably satisfactory description of laminar flow behavior. In contrast, the prediction of non-Newtonian transitional and turbulent friction losses is much more complex. For example, certain types of materials have been observed to sustain laminar flow at much higher flow rates than Newtonian liquids. When turbulence does develop, the level of agitation is damped to such an extent that the friction factor is lower than that of the original low viscosity Newtonian medium. Extremely broad transition regions and "diameter effects" have also been noted in the turbulent flow of certain chemically complex materials.' Aside from the latter complications, which are attributed in part to viscoelasticity, the existence of a variety of scale-up procedures for predicting transitional and turbulent friction losses is due principally to the chioce of viscosity parameter used in the definition of the Reynolds number. In contrast, Dodge and Metzner have demonstrated that the friction factorReynolds number relationship for inelastic fluids can be satisfactorily generalized provided the fluid property parameters are evaluated at the existing wall shearing stress.* However, the quantity usually sought in a pipe flow scale-up problem is the wall stress, and hence a trial-and-error procedure is required when dealing with non-power law materials. In this report we discuss two calculation methods: (1) for determining the wall stress, and hence the friction pressure drop, for inelastic fluids under conditions of turbulent flow by means of an iterative procedure, and (2) for predicting laminar friction pressure drops without recourse to arbitrarily defined flow models.

By Robert E. Brand, Strathmore R. B. Cooke

For a study of the bonding that occurs in magnetite pellets during oxidation, cubes of magnetite, hematite, and quartz were prepared, each with one surface polished and nearly optically plane. These faces were fired together under pressure and at various temperatures to determine whether bonding would occur. Sections of surfaces that bonded were examined microscopically. An attempt was made to measure the strength effected by this treatment. EARLIER work1,2 has shown that the strength developed by fired magnetite pellets containing silica or silicates is due to three separate effects. Two of these are associated with the formation of slag in the interior of each pellet at temperatures above 1000°C, where the atmosphere is substantially though temporarily neutral. The third is independent of the silica content and is responsible for the development of a hard shell around the pellet during the initial stages of firing, a condition which permits the pellet to assume the increasing load of the overlying burden and prevents extrusion of the slagged and semi-liquid interior. Pellets made from hematite concentrates do not show these effects unless specific additives are incorporated, and it is these differences in behavior which in large part account for the difficulty in producing satisfactory pellets from such concentrates. Elsewhere it has been stated that "the most important factor responsible for the development of strength in fired magnetite pellets containing no additive is the oxidation of magnetite to hematite and the subsequent recrystallization and grain growth of this constituent."' The object of the investigation reported in the present paper was to study in detail the mechanism described above and to evaluate its contribution to the strength of the fired pellet compared with other possible mechanisms, excluding slag bonding. This was effected by placing together polished faces of two cubes prepared from specific minerals and then heating them under constant pressure at different temperatures both in air and in nitrogen. The materials used were magnetite, hematite, and quartz. The magnetite was coarsely crystalline material from western Montana, the hematite was Itabira ore from Brazil, and the quartz was of optical grade but of unknown origin. A high-grade Lebanon (Pennsylvania) magnetite concentrate was used to prepare briquets for the investigation of bonding strength. Analyses of the iron oxide materials are given in Table I. The specimens for furnace treatment were prepared with a carborundum cut-off wheel, which cut the minerals into cubes approximating % in. along each edge. Care was taken to make the sides as parallel as possible. The edges of the faces to be polished were beveled, and the cubes were mounted in brass rings with sealing wax, this being essential to prevent the specimens from rocking during polishing. To prevent excessive friction, the wax was recessed so that only the brass ring and the face of each specimen would be in contact with the polishing lap. Each specimen was then ground with No. 600 carborundum to produce a level surface before polishing. The Graton-Vanderwilt process was not used for polishing because of the long time required by this method, but several other methods were tried. Cloth laps using levigated alumina or chromium oxide did not produce the desired polish. Satisfactory results were obtained by means of a pitch lap: which was used for all the mineral cubes, with levigated alumina as the polishing medium. The lap was operated in a machine similar to those used in the preparation of single optical surfaces. It was run at low speeds and moderate pressures, and polishing was stopped occasionally to prevent overheating of the lap. A good polish with flat surfaces could be obtained in about 5 min. After polishing the sealing wax was removed either in hot water or in a solvent such as turpentine.

Jan 1, 1955

By L. A. Wood, A. G. Winslow

Subsidence has occurred in several areas of the upper Gulf Coast region of Texas, although in most cases this is not evident without precise instrumental leveling. As referred to in this report, the region comprises about 7200 sq miles of the Gulf Coastal Plain and includes all or parts of the following counties: Harris, Fort Bend, Waller, Brazoria, Galveston, Liberty, Chambers, Orange, and Jefferson (Fig. 1). Topographic relief is low, ranging from sea level at the Gulf of Mexico to 300 ft or more in northwestern Harris County. In this upper Gulf Coast region, which embraces the most heavily populated and industrialized parts of Texas, industrial and municipal water supply is obtained chiefly from wells. Largest center of ground-water withdrawal is the Houston-Baytown area, where about 200 million gallons per day was pumped in 1956 for municipal and industrial use. Other large centers of ground-water withdrawal include the Alta Loma sector and the heavily industrialized Texas City area in Galveston County, where about 23 mgd was pumped in 1956. In the Freeport area in Brazoria County about 6 mgd was pumped in 1956 for city and industrial supply. In addition to withdrawals in metropolitan districts, great quantities of ground water are supplied by irrigation wells to thousands of acres of rice. These wells are widely dispersed over many miles, principally in Harris, Waller, Fort Bend, Brazoria, and Liberty counties, and effects of pumping are not so readily noticed as in the industrial and metropolitan sectors. Subsidence: Starting with the 1905-1906 line from Smithville, Tex., to Galveston, the U. S. Coast and Geodetic Survey has extended a network of first and second-order level lines throughout the upper Gulf Coast region. In 1918, when a line was run from Sinton, Tex., to New Orleans, La., a tie was made to the 1905 leveling at Houston. In 1932-1933 a line was run from Palestine, Tex., to Houston, and a good deal of leveling was carried out in the vicinity of Houston. At this time the entire network was adjusted to the Sea Level Datum of 1929. In 1935-1936 and 1942-1943 several new lines were run and considerable releveling was performed. Supplementary adjustments were made after both the 1936 and 1943 levelings in order to reduce the elevations to the Sea Level Datum of 1929. These adjustments probably masked some of the subsidence that had occurred up to that time, although a certain amount of subsidence was noticed in the 1943 leveling. When several lines in the vicinity of Houston and Galveston were releveled in 1951, the regional nature of the subsidence became apparent.' Direct comparison with the 1943 leveling was possible because the datum was not changed. Comparisons with prior leveling were unreliable because of differences in the early and subsequent datums. In 1953-1954 many lines in the region were re-leveled and in 1957 the entire network, including the elevations previously determined, was readjusted to determine the changes in elevations as indicated by various levelings between 1905 and 1953-1954. In this instance the adjusted elevations were held fixed near the boundaries of the region where the bench marks seem to have remained stable. The only sea-level connection held fixed was the connection of the 1905 leveling to the 1903-1906 tidal series at Galveston. Almost the entire region has been at least slightly affected by the subsidence, some areas to a great extent. The profiles of subsidence (Fig. 2) and the map showing the contours (Fig. 3) illustrate both regional and local changes. A large area of subsidence is centered in the Houston-Baytown district, where the maximum recorded is about 2.7 ft between 1905 and 1954. This probably falls short of the actual maximum, since land surface subsided as much as 2.5 ft in the industrial section alone between 1943 and 1954. Sub-

Jan 1, 1960

By T. M. Kegley, J. H. Frye, D. L. McElroy, M. L. Picklesimer, E. E. Stansbury

Measurements of rate of growth, thermodymmic quantities, and partitioning of Mn are reported for high-purity eutectoid Fe-C and Fe-C-Mn steels for the auistenite-pearlite reaction. Evaluztion of the cotnstants of a derived-rate equation with these data indicate that Mn additions cause the decrease observed by increasivg the activation energy and apparent activation entropy or the movenment of complexes at the interface. A diffusion-controlled mechanism is not required. Appreciable partitioning of Mn to cementite occurs entirely behind the austenite-pearlite interface for more than 30°C of subcooling. ThE growth of pearlite from austenite has been extensively studied over a period of years. but the rate-controlling step in the process is not known nor is the mechanisrn of the effect of alloying elements understood. A comprehensive review of the subject has been given by Mehl and Hagel. The present study has been concerned with further considerations of the effects of alloying elements on the rate of growth of pearlite. Analytical relatioriships for the rate of growth of pearlite have been considered by Zener.2 Turnbull,3 Machlin,4 Cahn.5 Frye,6 and Frye et a1.7 The derived expressions in each case involve the free energy of the transformation. the interlamellar spacing. and some rate term of the nature of the diffusion coefficient. mobility or frequency of occurrence of some rate-determining step. The treatments. of course. differ in detail. Further consideration is given in the present paper to the equation of Frye, Stansbury. and McElroy which was derived from modified absolute rate theory without in any way specifying the mechanism of the reaction. This equation is r=c ? T ? F exp -?E*/RT [1] where r = rate of growth in mm/sec c = a constant AT = equilibrium transformation temperature minus actual transformation temperature "C ?F = free energy per mol of austenite minus that of pearlite ?E* = activation energy R= gas constant ?F in Eq. [1] is computed from the following: ?F=?H?T/To - ?T' dT' ?T' ?CdT"/T" [2] To To where AH is the enthalpy per mol of austenite minus that of pearlite at the equilibrium temperature, To, and AC is the difference in specific heat between austenite and pearlite per mol. Subsequently, Darken, Fisher, and Galligan8 have reported that a layer of ferrite separates pearlite from austenite at the growing interface (i,e., the advancing face of Fe3C is in contact with ferrite, not austenite) and the ratio of the thickness of this layer to the spacing of the pearlite has not been found to vary with alloy content or transformation temperature. This observation does not require any alteration in the above equation: however, it may require that a different significance be attached to the factor, AT. The AT term appears in Eq. [I] as a consequence of a factor for the amount of growth per complex transformed at the interface.* It is based *This is not the sole reason for expecting such a relationship. See Ref. 7._______________________________________________________ on a l/2 ?T dependence of the interlamellar spacing. Zener2 has developed a theory which leads to Soa 1/?V where So = the pearlite interlamellar spacing. This relation is as consistent with the data on interlamellar spacing as other relationships, although, as pointed out by Mehl and Hagel,1 the data are not sufficiently precise to consider the relationship as confirmed. The work of Darken, Fisher, and Galligan8 requires that the above be written where f = the thickness of ferrite layer between austenite and pearlite. The present paper is concerned with iron-carbon-manganese eutectoid alloys and reports experimental measurements of the following kinds: 1) the effect of

Jan 1, 1961

By G. W. White

Over the past several years, the steady progress of gas lift from a mysterious art into a developing science has been traceable through the many fine papers that have appeared in Journal of Petroleum Technology. Consequently, this magazine has rightly become a standard for the serious students of gas lift all over the world. The following discussion is offered in a sincere attempt to accurately present certain topics covered in the subject paper in the light of acknowledged state of the art procedures. INSTALLATION DESIGN The first point to be questioned is the author's installation design, beginning with Eq. 6. The use of 1,200 psi in the spacing equation instead of the 1,275 psi noted as available seems, at first glance, to be a simple arithmetic error. Upon further examination, it appears evident that both this 75 psi plus the 100-psi safety factor must be subtracted from available kickoff pressure because of conditions imposed by the characteristics of the ASC valve (see Fig. B-1). Pressure-charged valve design is incorrectly penalized by spacing the first valve using only 1,100 psi. The use of such a high back-pressure makes it appear that a typical ASC installation would require no more valves than a comparable conventional valve installation. Using a value of 100-psi tree pressure, the well could be valved to 11,000 ft very conservatively with seven pressure-operated valves, but would still require 11 ASC valves. The use of 400-B/D spacing gradient for a 140-B/D rate is not an accepted practice using precharged valves. The author would have done well to note the advantages to be gained by reducing the 500-psi tree pressure, which is clearly explained by Beadle, Harlan and Brown.' In Step 1 of Part III of the continuous flow discussion, the author incorrectly applies the Poettmann and Carpenter curves.2 Item 2 is an incorrect statement. An excellent reference on this point is the paper by Baxen-dell and Thomas.' VALVE MECHANICS The next topic relates to the mechanics of precharged valves. Eq. 2 for the ASC valve can be used to describe the condition of equilibrium, wherein the valve is closed with opening pending. When this equation is used to determine relative values of casing, dome and tubing pressure with conditions approaching the transition value, some strange results are obtained. The following uses the author's symbols and values given on page 689. Fig. B-1 shows the condition that exists at the moment an ASC valve has been uncovered during the unloading process (Condition I). The obvious result is a locked condition. The basis for the curve is shown on the figure which includes 1,000 psig casing pressure to uncover the valve. If injection gas is not bled off the casing until the pressure is decreased by an amount exceeding the stabilizer discharging valve pressure, upward adjustment of the casing pressure is mandatory for the valve to open and unloading operations to continue. In Condition I the dome pressure exceeds the casing pressure by 20 psi, and the tubing pressure required to open the valve is 1,212 psig. This tubing pressure would not exist under the assumed conditions. In Condition I1 the casing pressure is increased to the dome pressure, and the tubing pressure required to open the valve is 1,102 psig. It is apparent that the casing pressure must be increased to a value exceeding the dome pressure before unloading operations could continue. In Condition I11 a maximum casing pressure of 100 psi (stabilizer charging valve set pressure) above the dome pressure is reached, and the tubing pressure required to open the valve is 549 psig. The author's statement after Eq. 5 that "the minimum tubing pressure required to open an ASC valve is equal to the casing pressure at depth less the transition pressure of the valve" is incorrect except for Condition I11 only. Otherwise, the equation is invalid.

Jan 1, 1965

By C. B. Alcock, R. G. Hudson

Sulfur pressure measurements above FeS in equilibrium with iron have been carried out by the Knudsen orifice method. A comparison is made of the weight loss of the cell per unit time obtained in the above experiments, that calculated from Richardson and Alcock's measurements of the S2 pressure above FeS in equilibrium with iron, and the dissociation constant for diatomic sulfur from spectroscopic and thermal data. EXTENSIVE equilibrium measurements have been made to determine the sulfur potentials of a great many systems of metallurgical interest.' In order to confirm their reliability, to supply data which may be of use in further thermodynamic calculations, and to establish the reliability of the Knudsen orifice in obtaining accurate thermodynamic values at high temperatures, the present investigation on the pressure of the gaseous species above FeS in equilibrium with iron has been carried out. The Knudsen orifice method has been used for these pressure measurements. Experimental The apparatus and general technique for this investigation are the same as that reported for the determination of the vapor pressure of silver.' To reduce the residual gas pressure in the evacuated chamber, a titanium getter, maintained at run temperature, was introducbed. Several types of Knudsen cells were used—platinum, zircon, and fused silica. The orifice in the zircon cell was fashioned with a small diamond wheel. The same precautions were used in this investigation as were employed in the work on Mo2S2.3 The FeS-Fe mixture was prepared by heating a sample of pure FeS,, supplied by K. K. Kelley of the Bureau of Mines, with excess iron powder, cp grade, obtained from the Fisher Scientific Co. These two substances were heated in a closed Armco iron crucible at 960°C, held in a vacuum of less than 10" mm Hg for 24 hr. The mass was then removed from the cell and ground in an agate mortar. In this form it was charged to the various Knudsen cells. Heating the Knudsen cell for an additional two days was sufficient to obtain constant rates of weight loss of the cell. The data obtained in this investigation are recorded in Table I. To establish whether or not a sulfide of iron is volatile at these temperatures, 0.7646 g of the FeS-Fe mixture, described above and containing 69.2 pet Fe, was placed in the zircon Knudsen cell. Run Nos. 103 to 109 were performed on it. After the completion of these runs, the total amount of iron left in the cell was obtained by chemical analysis. It contained 0.537 g Fe, compared to 0.529 g in the original charge. Thus, there was an apparent gain in iron of 1.5 pet. This is attributed to unknown experimental error, and the conclusion is that no sulfide of iron is volatile. The vapor pressure of pure iron4 is about one thousandth that of the sulfur pressure' at these temperatures. The vapor pressure of iron inside the Knudsen cell will be almost exactly that of pure iron, since in the iron-rich phase the percentage of sulfur is of the order of 0.025 pet.At this low percentage of sulfur Henry's law is probably obeyed by the sulfur. Thus, the iron will obey Raoult's law. In any event, the vapor pressure of iron will be lower than that of pure iron and, since the vapor pressure of pure iron is negligible compared to the pressure of sulfur, the weight loss of the cell due to the vaporization of iron should be negligible. Since neither iron nor FeS is volatile to an appreciable extent, the total weight loss of the cell must be due to the vaporization of sulfur. Discussion and Calculations On the assumption that S, is the only gaseous species present in appreciable quantity, its pressure

Jan 1, 1957

By C. R. Masson, M. L. Pearce

The nitrogen contents of seven specimens of iron and steel were determined by the three methods. Good agreement won generally observed between the results of the isotope -dilution and Kjeldahl determinations, tltzts establishing the reliability of the isotope -dilutzon technique. For some specimens, the vacuum-fusion values at 1650 °C were lower than the corresponding Kjeldahl values by up to 0.0015 pct absolute. These differences were eliminated for a specimen of "pure" iron by performing the vacuum-fusion analyses at 2100 ° to 2240 °C. The reason for this relatively high temperature required m the vaczium-fusion analyses is discussed. ThE determination of nitrogen in iron and steel has been the subject of numerous investigations and a concise review of the extensive literature in this field has recently appeared.' Two methods of analysis are most commonly employed, the vacuum -fusion and the chemical or Kjeldahl method. Except in special cases where the presence of certain alloying elements may present difficulties, both methods are generally regarded as applicable, although the vacuum-fusion method is more sensitive and is often preferred for samples of very low-nitrogen content. The method of isotope dilution, first employed for the determination of gases in metals by Kirshenbaum and Grosse,2 has recently been applied to the determination of nitrogen in iron and steel.3,4 The method is attractive in that it does not require the quantitative extraction of gas from the metal specimen. Staley and Svec3 modified the standard Kjeldahlpro-cedure by introducing, along with the steel sample, a measured amount of ammonium sulfate enriched in the isotope 15N. The nitrogen content was obtained by measuring the isotopic ratio 15N/14N in the resultant solution. The technique involved the distillation of nitrogen as ammonia from alkaline solution and subsequent oxidation of NH4+ in the distillate to gaseous nitrogen. The results for iron and steel samples of commercial origin were often in reasonable agreement with values quoted by the suppliers, although occasional marked discrepancies were observed, particularly with the results of vacuum-fusion analyses. Preliminary work in this laboratory4 on the determination of nitrogen by isotope-dilution in the gas phase yielded results for three samples which were significantly higher than the values obtained by vacuum-fusion. The present paper describes an extension of this work to a wider variety of samples. Analytical results by the Kjeldahl method are also reported. EXPERIMENTAL The technique employed for the isotope-dilution determinations was similar to that described elsewhere.4 Several minor modifications were made to the apparatus. These included the use of molybdenum instead of platinum-rhodium suspending wires when crucibles were degassed at temperatures above 1600°C. The circulatory system used previously for converting CO to CO2 was replaced by a single chamber, the walls of which were cooled directly in liquid nitrogen. Temperatures were measured with an optical pyrometer which had been recently calibrated and, as in the previous study, were minimum values obtained by sighting on the center bottom of the crucible. Temperatures obtained by sighting on the interior walls were always higher by 100' to 130°C with the design of crucible employed. Flat-bottomed crucibles exhibited an even larger temperature difference. To check the efficiency with which crucibles were degassed in the isotope-dilution apparatus, several experiments were done in the vacuum-fusion apparatus, suitably modified4 for isotope-dilution studies. Faster pumping speeds and higher degassing temperatures were possible, the latter due to the use of a generator with a higher power output (10 Kva at 398 kc per sec). In these experiments the combustion of CO was omitted and a correction4 was made for the contribution of CO to the ion current at m/e 28, used in determining the isotopic ratio R = l5NP4N. This correction did not seriously influence the re-

Jan 1, 1962

By Gordon H. Geiger, John M. Svoboda

One mechanism of metal penetration into mold aggregates by cast steels is vapor state mass transport. In order to further understand and quantify this mechanism, the steady-state diffusion of metallic vapors was studied at 1250" to 1600°C in controlled oxygen pressure atmospheres by measuring the weight loss from a metal bath whose vapors were allowed to diffuse through a porous aggregate compact held at constant temperature. Mass transport rates substantially greater than those predicted from the assumption that the vaporizing species was the elemental metal vapor were observed. Vaporization rates were found to increase with an increase in the oxygen partial pressure in the gas atmosphere with maximum vaporization occurring at oxygen partial pressures just below those in equilibrium with a liquid oxide layer at the given temperature. When the liquid oxide layer was present, no vaporization was observed. Metal carbonyls were ruled out as the vapor species involved because of their thermodynamic properties. The species with the enhanced volatility were deduced to be gaseous metal-oxygen molecules, probably the polynuclear oxides (FeO)2 and (Mn0)2. Diffusion followed the ordinary diffusion mode, and Knudsen diffusion is not important in normal foundry molds. The interdiffusivities of the metallic vapors and the mold atmosphere, through the porous aggregates, were measured. The effective interdif-fusivity of the vapor species and mold atmosphere through the aggregate may be expressed as follows: Deff,12 - — where E is the void fraction of the aggregate and 7 is the "tortuosity factor". Aggregate material had no effect on diffusion rates, and the only relevant parameter was the void fraction. Values for 7 ranged from 3 to 6 for normal foundry aggregates with 4 being considered a good representative value. An area of interest in high-temperature metallurgical processes which has received little attention in the past is the diffusion of metal vapor species in porous aggregates. Research into the metal penetration phenomenon' in foundry molds has shown that a major penetration mechanism is the vapor transport of metallic species into the porous foundry sand molds, followed by condensation of metallic films on the sand grains. Considerable study of the vaporization of metallic species has been reported in the literature, and the vapor transport properties of many metallic species have been determined.2'3 Many instances have been observed where the vaporization behavior could only be explained by the existence of a metallic species with a vapor pressure greater than the elemental metallic species. These species were identified or, in some cases, presumed to be carbonyls,4,5 hydroxides,6-9 or vaporous oxides.10-16 The diffusion of gases through porous aggregates has received considerable attention in the literature pertaining to catalysis, although primarily at low temperature, and more recently in the literature concerned with ore reduction.17-22 The following experiments were devised and carried out in order to determine the nature of the metallic species involved in the metal penetration phenomenon, and the effective diffusivities involved. EXPERIMENTAL APPROACH Steady-state diffusion was selected for experimental convenience. The experimental systems used, shown in Figs. 1 and 2, consisted of a porous aggregate specimen contained in an impermeable alumina tube held above, and just separate from a metal bath. The entire system was held at constant temperature and pressure in a furnace with a controlled oxygen partial pressure atmosphere. After a brief period of nonsteady state behavior, a steady flux of metal vapor through the porous aggregate will develop due to the concentration gradient of metallic vapor in the gas phase. The flux of metallic vapor from pressure may be neglected because there is no pressure gradient. There is no flux from convection because there is no net flow of gas within the metal-aggregate system, and thermal diffusion is negligible compared to the flux from ordinary diffusion. If the material being studied is allowed to vaporize until steady-state behavior is attained, the experimental results may be analyzed using Fick's First Law.

Jan 1, 1970

By J. Chipman, H. Towers, M. Paris

A simple radioactive tracer technique is used to measure the diffusion coefficient of calcium ion in molten slags. In a slag of 40 pct Ca0-40 pct Si0,-20 pct Al203, Dc " at 1450°C is 1.3x10-6 cm' sec-'. THE application of the principles of thermodynamics to steel and iron making problems has proved so fruitful that the kinetic approach has tended to be neglected. It seems probable that in some reactions the rate-determining step is the dif-fusivity of the reactants or of the products. But knowledge of the diffusion coefficients of the constituents of liquid slag and metal is singularly lacking. The present work was undertaken as a first step toward supplying such data for slags. Furthermore, any additional information on the physical properties of liquid slags might be expected to help in elucidating their structure. It is essentially a preliminary investigation, the course of which was determined more by the necessity of establishing a method than by the immediate intrinsic interest of the results. For example, the selection of calcium ion to be studied was dictated by the suitability of the isotope Ca45 in that it is a simple emitter with a long half-life. Of the other ions commonly encountered in slags, the isotopes of iron and manganese have relatively complex radiations as well as the normal disadvantage of requiring control of the oxygen potential of the atmosphere; Mg27 and Al29 have half-lives measured only in minutes; Si3' has a simple decay process but its half-life of 2.8 hr is too short for preliminary studies; the only useful oxygen isotope is the stable O18; and there remain only sulphur and phosphorus both of which have suitable isotopes but were rejected as being of less fundamental interest than calcium. Similarly a lime-alumina- silica slag was used in preference to calcium silicate primarily because of its low melting point and its capacity to form a glass on fairly rapid cooling. Literature Review Very few studies of diffusion in liquid silicate systems have been reported. In order to determine the origin of diopside inclusions in felspathic rocks, Bowen1 investigated the diffusion between diopside and various mixtures of albite and anorthite. Platinum crucibles were used with diopside, the heavier of the two liquids, forming the lower half of the cell and diffusion taking place against gravity. Change in concentration along the diffusion path was found by measuring the refractive index of the quenched glass. Nearly thirty years later McCallum and Barrett2 used essentially the same method to determine the diffusion coefficient of lime in lime-alumina-silica slags. In this case the interest was to find the rate-determining step in the process of slag attack on refractory materials. In an investigation of the kinetics of the sulphur reaction, Derge, Philbrook, and Goldman8 determined the diffusion coefficient of sulphur in a typical blast furnace slag by the simple procedure of melting a slag containing 2 pct S over an identical slag sulphur-free. They make no claim to a high degree of accuracy but suggest that D, is of the order of 10-6 cm2 sec-l. Other high temperature liquid diffusion studies have been concerned with metals. Holbrook, Furnas, and Joseph4 determined the diffusion coefficients of silicon, phosphorus, sulphur, manganese, and carbon in liquid iron by immersing a small carbon crucible containing pure iron-carbon alloy in a large vessel containing iron with a high concentration of the particular element. Their results were confirmed by Paschke and Hauttmann5 using a similar method. Diffusion coefficients in silicates in the glassy state have been studied by Johnson, Bristow, and Blau6 Both Na24 and Ca45 were used as tracers but only the

Jan 1, 1954

By F. A. Crossley

The additions: o, Sn, Zr, Cb, Mo, and V were studied for their effects on the a/a + Ti3A1 boundary of Ti-A1 base ternary alloys. These additions were chosen because of their importance to commercial titanium alloy practice—oxygen as the major impurity (in terms of influence on mechanical properties) and the others as major alloying additions. The composition and temperature ranges of investigation varied according to the system as shown in Table I. The principal means of investigation was light-microscopy. The basis of the ternary studies was the recently determined titanium-rich end of the Ti-A1 system.' MATERIALS, METHODS, AND PROCEDURES Ingots were prepared in 15 g size for all alloys except those of the Ti-Al-Mo and Ti-A1-V systems which were prepared in 20 g size. These alloys were prepared from high-purity, electrolytically refined titanium obtained from the Bureau of Mines, Boulder City Metallurgical Laboratory, Boulder City, Col., and Chicago Development Corp., Riverdale, Md., for the 15 g and 20 g ingots, respectively. The alloying additions were obtained from commercial sources, and were at least of 99.9 pct purity, except the V-15 wt pct A1 master alloy for vanadium additions. This alloy was of 98.2 pct purity; the major impurities were 0.37 wt pct Fe and 0.38 wt pct Si. The ingots were prepared by multiple melting in a nonconsum-able-electrode arc furnace. The ingots were covered with Markal coating "CR-T22" (Markal Co., Chicago) to protect them from scaling. Markal coating is applied as a slurry of solids in a volatile carrier liquid. The coated article is air dried. Upon heating to the hot working temperature the coating fuses to form a glassy, protective film. The ingots were heated to 1000°C and rolled to 0.16 in. thickness, reheating after each pass. The glassy film was removed by sand blasting. Ten mils were removed from all surfaces by grinding and/or pickling (3 pct HF, 30 pct HN03, balance water). Grinding was always followed by pickling to remove traces of the grindmg medium. Hydrogen removal was accomplished by vacuum annealing at 1000°C for 2 hr to a final pressure of about 0.02 µ Hg. This was followed by annealing in evacuated Vycor bulbs (800°C, 2 hr, ice brine quenched (IBQ)) in order to maximize ductility. The materials were then lightly pickled and cold rolled to 0.12 in. thickness or cracking, whichever occurred first. After rolling, 1/2 by 1/4 in. specimens were cut for the equilibration anneals given in Table 11 and carried out as previously described.1 Final polishing of the specimens for metallographic examination was done by electrolytic means. Specimens were first examined without etching under bright field and polarized light illumination, for this is the only means of detecting the "two-phase" phenomenon.1,2 The term "two-phase syndrome" refers to Ti-A1 al-loys which appear on the bases of their chemical, phys-ical, and mechanical properties, to be two-phase, but are actually single phase and inhomogeneous in composition. Since processing and heat treatment of the alloys of the first group were completed before the two-phase syndrome was discovered, a number of the specimens were found to exhibit this phenomenon—80 pct of those of the Ti-Al-Zr system, 5 pct of those of the Ti-Al-Sn system, 30 pct of those of the Ti-Al-Zr system. The affected specimens were homogenized by heating above the ß transus temperature and quenching. The heat treatment and working schedules to accomplish this are given in Table 111. After rolling, specimens were pickled (3 pctHF, 30 pct HNO3 in water) to remove 1 to 2 mils from the

Jan 1, 1970

By G. S. Cole, G. F. Bolling

Recent experitnents on crystal growth have indicated that thermal convection in the liquid influences the conditions of solidification and subsequent crystal characteristics. However, the important columnar-equiaxed transition in castings has not been examined as a function of this free convection. An experiment is presented here to demonstrate a change in the transition as convection is suppressed. Lead- and tin-based alloys were cast in stainless-steel molds either without special procedures or with thin grids placed horizontally across the tnold diameter in portions of the castings. While the grids allowed some diffusion of heat and mass transport to occur during solidification, massive fluid flow was hindered in the vegiotz of the grids. A larger percentage of columnar grains appeared in the grid region than in other parts of the special castings or than in all regions of normal castings. This structural difference can be explained in terms of thermal convection. THERE is little information about the possible effects of natural convection which may occur during the solidification of a casting. Ruddlel-3 has stated that the movement of liquid metal produced by convection might assist in the transfer of heat, but in the absence of confirmatory evidence he suggested that this idea be treated with caution. chalmers4 considers, but does not assess, the importance of fluid motion in casting, while Tiller5 suggests that, in liquid metals which have extremely high thermal diffusivities, the effects of thermal convection may be negligible. In addition, throughout the literature, there are statements indicating the possible presence of thermal convection but without proofs or evaluations of its influence. Natural convection may be divided into two categories: thermal convection, where liquid flow is due to density differences caused by temperature gradients in the liquid (e.g., liquid near a cold wall falling with respect to hotter liquid further away from the wall); and solute convection, where flow is due to density differences as a result of concentration gradients, see Wagner.6 In most instances thermal convection predominates over solute convection. Nevertheless, even thermal convection has not been studied in any detail and is difficult to handle, since it can vary quantitatively from one experiment to another depending upon the exact positions of heat sources and sinks or the attitude of the container with respect to gravity. In spite of these difficulties it is possible to realize that a higher rate of heat transfer must result with convection than in any situation where conduction acts alone. Thus, the degree of increased heat transfer is the important undecided question in considerations of casting. It will be demonstrated here that a reduction in convective flow leads to a suppression of the columnar-equiaxed transition in a casting and that this suppression is related to a decrease in the degree of heat transfer. Fortunately it is possible to do this by comparative rather than quantitative measures. The insertion of horizontal plates or grids into a simple mold does not appreciably influence heat transfer per se, but in reducing the allowable height through which colder liquid may fall freely it reduces natural thermal convection. This leads to a reduced fluid velocity, a different flow pattern, and an altered heat transfer which can be measured. EXPERIMENTS The alloys were prepared from elements of 99.9 pct nominal purity and cast into an uncoated 2-in.-diameter stainless-steel mold insulated at the bottom with a suitable refractory so as to ensure predominantly radial heat flow during freezing. Casting was performed either in empty molds or in molds containing grids or plates in various configurations . The grids were constructed of aluminum screen (1/32 in. thick) which was cut to the desired shape, cleaned in aqua regia, and coated with a mixture of 2 parts aqua dag graphite -1 part Sauereisen cement mixture, thus supplying a hard, protective, and insulating layer approximately 0.006 in. thick. The

Jan 1, 1965

By Douglas R. Piteau

Douglas R. Piteau (Engineering Geologist, De Beers Corp., Kimberley, South Africa; Presently Consulting Specialist at Nchanga Mine Open Pit, Chingola, Zambia)-Mr. Lutton is to be commeneded on his analyses and presentation of such practical data which can be used for predicting a preliminary optimum slope angle in rock. It appears that he assumes all slopes analyzed to be straight. Although there are overriding factors which affect the general slope configuration, such as structural discontinuities, ground water, and other environmental factors such as weathering conditions, regional stress, etc.,8 the affects of plan geometry under certain conditions also can be of outstanding importance. In investigation of natural slopes, for example, the effects of plan geometry are not entirely appreciated and it is in this respect that discussion is given. Experience in southern Africa has shown that the overall geometry of surface openings in rock, particularly those with dimensions of the order of 1500 ft or less, i.e., where radii of curvature of the slopes are less than, say, 700± ft, must be considered in rational slope-stability analyses. That is, the slope angle is not only a function of slope height, strength of the mass, and geometry of structural discontinuities, but also is dependent upon the plan geometry of the local slopes and the plan shape of opening. Technically speaking, adequate analysis of the slope must be three-dimensional, the third dimension being that of overall plan configuration and the analysis must incorporate its subsequent effects. However, current theories of slope stability all deal with the slope as a two-dimensional problem, i.e., a slice of unit length of an infinitely long slope is considered for a condition of plane strain. In this the tacit assumption is made that the plan radii of the crest and toe of the slope are infinite and no attention is given to the plan geometry of the slope. Clearly this is not the condition encountered in practice, particularly in open pit mining where excavations, relatively speaking, can be small and where these radii of curvature accordingly can have important affects on the slope stability. This is illustrated by the analyses of Piteau and Jennings 9 of the natural slopes which had developed around pipes of four diamond mines in the semiarid Kimberley area of South Africa. Although these slopes consist of dolerite overlying shale, the values of the slope angles of all the pits were corrected for a height of slope of 320 ft assuming that no dolerite occurred. This brought the data onto a comparable basis. These data were analyzed statistically and a clear empirical relationship, as is shown [in Fig. 4], is seen to have formed between angle of slope and curvature of slope, i.e., the angles of slope are greatest where the radii of curvature are least. Also, it can be seen that the slope of each line is different, indicating that the plan shape of the open pits also are significant, the curve being steepest for the open pits which are more circular in shape in contrast to those which are more elongated. The effect of slope curvature on the average angle of slope for the two extreme radii of 200 ft and 1000 ft for all four open pits is 39.5° and 27.3°, respectively. Hence, the concave curvature effect represents an average increase in slope angle of 12.2° from the essentially straight slope. In quantitative terms in Mr. Lutton's Fig. 2, this represents displacing the curve for a 320-ft pit from the second curve from the bottom (i.e., shale in subtropical climate) to that of the third from the bottom (i.e., quartzite in temperate environment).

Jan 1, 1972