Search Documents

By A. M. Khan, I. Fatt

A new type of steady-state thermal conductivity apparatus is presented and described. Results of measurements on Berea sandstone, Solehofen limestone, Bandera sandstone and Bakelite are presented. The measurements were made in a two-hour period at temperatures up to 80C and axial stress to 125 bars. Results show that the thermoelectric heat pump and tbermistor-controlled electric heater can be used in a steady-state apparatus for measuring thermal conductivity of mck. Steady-state is reached within two hours. With Urethane foam insulation, not more than 3 per cent of heat is lost by radial flaw. The apparatus can be used as a comparator or for absolute thermal conductivity determinations. INTRODUCTION Calculations of heat flow through the earth, both for scientific purposes and engineering design, require thermal-conductivity data on rock. However, the high variability of rock often makes handbook thermal-conductivity data of little value. For engineering purposes in particular, it is desirable to have measured data on samples of the rock mass under study. Laboratory methods for measuring thermal conductivity of rock use either time-dependent1-3 or steady - state temperatures .4,5 The time -dependent (transient) techniques give thermal diffusivity from which thermal conductivity can be calculated if the specific heat and density of the sample is known. Most transient techniques require lengthy and involved sample preparation. Also, the accuracy tends to be low. Steady-state techniques are more exact and are the most commonly used. The steady-state methods, however, also have their difficulties. A long time is often required to reach steady state, troubles have been encountered in establishing good thermal contact between the heating and cooling elements and the rock sample, and unaccounted-for heat losses are difficult to prevent. In this paper we present a description of a new type of steady-state thermal conductivity apparatus. With it, measurements have been made in two hours at temperatures up to 80C and axial stress to 125 bars. Sample preparation is simple and no thermal conductivity standard is needed.6 APPARATUS Our steady-state apparatus is an improvement . over the conduction pile used by Zierfuss and van der vliet4 and Somerton.5 A linear heat flux is established through a conduction pile consisting of the rock sample and a standard material of known thermal conductivity in series. The thermal conductivity of the sample is calculated from the ratio of the temperature drop across the sample

Jan 1, 1966

By H. T. Kennedy, J. E. Little

An empirical equation for the prediction of the viscosity of several pure paraffin hydrocarbons and nitrogen is presented. It involves temperature, pressure and six constants of the material, and it applies reliably to both liquids and gases. The equation is similar in form to van der Waal's equation of state. For the paraffin hydrocarbons methane through r-hexane and nitrogen, an average absolute deviation of 1.9 percent was obtained on 1,006 data points described in the literature by 14 authors. When this equation is extended to complex, liquid hydrocarbon mixtures, a correlation was obtained with an average absolute deviation of 9.9 percent. INTRODUCTION Equations describing the flow of gas and liquid through porous media contain the viscosity coefficient of the fluid. If other pertinent variables remain constant, the volume rate of flow is inversely proportional to this coefficient. In dealing with condensate fluids and volatile oils, however, the compositional effects resulting from changing pressure materially affect the viscosity. The effect of compositional changes also may be significant in certain secondary recovery or pressure maintenance processes, notably miscible displacement or gas injection. Early attempts to describe the performance of reservoirs utilized a volumetric material balance method wherein gas and liquid in the reservoir were identified as produced gas and liquid at the surface. This method of analysis proved adequate for reservoirs at moderate temperature and pressure that contained gas with moderately low amounts of condensable materials. The volumetric material balance procedures for "black oil" reservoirs leave much to be desired when applied to condensate and volatile oil reservoirs because phase behavior and compositional changes are relatively more important in these cases. The alternative is a compositional material balance, which in turn, requires a correlation of properties of the reservoir fluid with composition. This paper supplies this correlation in regard to viscosity, for reservoir crude oils. REVIEW OF LITERATURE The literature contains many empirical equations describing the effects of composition, temperature and pressure on the viscosities of pure liquids and binary liquid mixtures. However, the applicability of a majority of these equations is limited to very low pressures and to a small number of systems. Most of them, when applied to complex hydrocarbon systems, are of little value. The lack of utility of the majority of equations results from the fact that they were developed to show the separate effect of temperature, pressure or composition on viscosity, but not to predict the viscosity as a function of all three variables. And with the few exceptions noted below, they were developed to apply to much simpler systems than hydrocarbon mixtures. Eakin1 developed an equation for the viscosity of pure light hydrocarbons applicable in the gas, liquid and dense fluid regions, and in its most general form is written as µ - µo =A(ebP - e-cpk) The pure component viscosity at atmospheric pressure and temperature T, µo, was calculated using the Sutherland equation: µo = BT3/2/T + S The constants in Eakin's equation permitted the term e-CPk to become quite small at high densities (above 0.215 gm/cc), and the general equation would simplify to

Jan 1, 1969

By Hubert E. Risser

Although, for about a half-century the percentage of coal production provided through strip mining has steadily increased, recent trends indicate that a peak in percentage (not tonnage) is being approached, principally because of the decline in economic adrantage enjoyed by strip mining compared to underground mining. Factors contributing to the decline are the decrease in the margin of advantage in manpower requirements held by strip mining; an increase in depth and volume of overhurden, requiring larger equipment and capital investment; increasing lack of reserve land blocks large enough to support extensive strip mining operations; more stringent legislation for reclamation of stripped land, with increasingly higher costs of such reclamation; and the increasing costs of land surface rights for strip mining. Already there has been a peak and suhsequent decline in relative importance of strip mining insome major Eastern coal-producing states; trends indicate that a peaking for the whole Eastern region is on the way. Although, in the West the pressures are less severe, they will, nevertheless, reduce to an increasing degree the advantage that strip mining has possessed. For about half a century, strip mining has been providing an ever-increasing share of total coal output in the United States, and in the past 30 years this share has grown from less than 8% to 33.7%, as shown in Fig. 1. Fig. 1 also shows that strip mining is gaining faster in the West than in the East. From 1961 to 1967, output from strip mines grew 63.6 million toqs, or 52.2%. Certainly there is nothing in such performance to indicate any peaking in the level of strip coal-mine production. But the fact is that, if not in actual tonnage, strip mining appears to be reaching its peak in relative importance, at least east of the Mississippi River. And the area east of the Mississippi provides 90.0% of the strip and 94.9% of the total coal production of the nation. The relative economic advantage which spurred strip mining's growth for a third of a century has been diminishing markedly in recent years. Fig. 2 shows evidence of this decline. From 1947 until 1961, strip tonnage output remained fairly stable, at the end of the period showing only a 12.5% decline. Meanwhile, underground coal production fell 44.5%. As the result of the greater decline in underground coal, strip coal in 1961 furnished 30.3% of the total coal produced, compared to only 22.1% in 1947. In the succeeding six years, however, underground coal output increased by 76.5 million tons, while strip output increased only 63.6 million. Further evidence of the strengthening of underground coal's relative position — or if you will, the weakening of strip coal's position — can be found in the major coal stripping states: Illinois, West Kentucky, Ohio, Pennsylvania, Indiana, and West Virginia, which in 1966 provided 80% of the nation's strip coal. Fig. 3 shows what has been happening in three states lying in the Appalachian field and in the total output of these three states. In the figure, the column height represents the output of coal mined by all methods. The number at the base of each column is the percentage mined by stripping. In Pennsylvania and Ohio, the percentage of strip mining increased between 1956 and 1961, but by 1966 had declined from the 1961 level. In West Virginia, a decline occurred during the first five years but was followed by a rise. In January 1968, Coal Age published an article listing the locations and capacities of new large mines in the planning and construction stages that would be operating by 1968-1971. The plans for West Virginia and Pennsylvania indicated all the new mines would be underground. Except to the extent they might replace existing under-

Jan 1, 1970

By P. K. Leung, S. R. Allada, P. M. Dranchuk, D. Quon

The alternating direction explicit procedure (ADEP) makes use of the boundary conditions to reduce multi-dimensional problems to a series of one-dimensional problems. The method, previously applied to reservoirs containing only an under-saturated oil, has now been extended to cover the case of natural gas reservoirs. l his involves solving a non-linear partial differential equation, application of the procedure is straightforward and no calcuIational problems were encountered. INTRODUCTION There has been a growing interest in formulating mathematical models of petroleum and natural gas reservoirs — models which permit the engineer to examine and evaluate the physical and economic consequences of various alternative production policies. The tremendous reduction in the cost of solving such models in recent years has made possible their use as an almost routine management tool. This reduction is the result not only of improved computer hardware but also of the development of more efficient mathematical techniques. The present paper is concerned with the application of a recently proposed numerical method (ADEP) to two-dimensional gas reservoirs. STATEMENT OF THE PROBLEM Given a two-dimensional Region R, bounded by a closed Curve C (Fig. 1) such that the behavior on Curve C is known, the differential mass balance for each fluid phase in R, neglecting gravitational effects and assuming Darcy's law for fluid transport, can be written as:

Jan 1, 1967

By J. L. Green

Virginia-Carolina Chemical Corp., in its Florida Operations, uses both roller mills and ball mills in dry grinding of phosphate rock. The roller mills have been in operation since 1954 in the Mining Department. The ball mills were installed and put in operation in 1960 in the Harding Concentrate Superphosphate Plant. The roller mills in the Mining Department are used to grind phosphate rock for outside customers. This involves short notice orders of varying grinds, grades and types of rock. The ball mills grind phosphate rock for use in V-C's own phosphoric acid and concentrated superphosphate plant. These mills grind generally two types of rock for one grind. In 1959, a 10% speed-up was made in the operating speed of the roller mills. This paper deals with the operational gains and resulting cost savings which evolved from this speed-up, and compares the present operations of the roller mills and of the ball mills in grinding phosphate rock. In January 1959 it was decided to determine the feasibility of reducing the grinding cost on the roller mills, per ton of rock ground, by speeding up the mill 10%. At this time, one mill was completely reconditioned so it would be in top condition. Extensive tests were then run both before and after the mill speed-up to determine the resultant cost savings. These tests showed that, due to this 10% speed-up, the mill capacity was increased approximately 20% and grinding power costs per ton were reduced approximately 10%. Table I shows the results of these tests. Each test run actually represents several of the same type rock and grind, averaged together. Each test run lasted approximately three hours. The validity of these runs has been confirmed by subsequent operation in grinding over 1,000,000 tons of phosphate rock. Accurate records have been kept on power consumption and repairs as well as tons produced. Two types of feed rock were used for the tests — pebble phosphate rock and flotation phosphate concentrate rock. The pebble phosphate rock size is approximately 80-90% plus 20 mesh. The flotation concentrate phosphate rock size is approximately 15% plus 20 mesh and 95% plus 100 mesh. The same two types of rock feed were used for all the test runs. Test Nos. 1, 2, and 3 were run at the original mill speed of 85.5 rpm on the vertical shaft. Test No. 1 was run with a feed of high grade flotation concentrates; the desired fineness attempted was 70% passing 200 mesh. Tests No. 2 and No. 3 were run with a feed of low grade pebble. Test No. 2 was run for a desired fineness of 70% passing 200 mesh. Test No. 3 was run for desired fineness of 50% passing 200 mesh.

Jan 1, 1963

By E. Douglas, D. N. Collins, J. R. Stevens

E. Douglas (Dept. of Scientific and lndustrial Re-search, Warren Spring Laboratory, Hertfordshire, England) — The authors are to be congratulated on the considerable improvements they have made in tailoring efficiency and selectivity terms which are generally superior to the alternative forms in current use. However, I feel that their efficiency expression is limited to lower-grade feeds; its accuracy reduces as the feed grade increases. For example, its application to such processes as the beneficiation of glass sands, refractories and clays will invoIve considerable inaccuracies. The reason for this lies in the direct use of recovery, which, with high-grade feeds, biases their empirical expression, R(1 - c/h). With a 90% grade feed, the recovery weighting for the slightest degree of beneficiation, over a zero efficiency sampling operation, will be 90+%+ I disagree with their second basic requirement which states that there can be no fundamental basis for any single quantifying expression. A formula giving a comprehensive representation of concentra- tion efficiency can be derived fundamentally accord-to the procedure outlined in the literature. '' This formula, called the Basic Efficiency formula, is

Jan 1, 1963

By P. J. Closmann

Steam zone growth as a function of time has been calculated for the case of constant rate steam injection into a preheated reservoir. To simplify the calculation a linear temperature profile has been assumed in the cap and base rock at the start of steam injection. The results indicate that at early times augmentation of steam zone growth due to preheating should be greatest. At longer times the steam zone development becomes close to that calculated with no preheating. INTRODUCTION With the increasing application of thermal recovery processes to recover viscous oil, cyclic steam injection has become important in many large-scale projects. In this process, the cycle of steam injection followed by oil production is repeated a number of times. At the beginning of the second and later cycles, steam is injected into a reservoir that has already been heated but that has lost part of its heat both in produced fluids and by conductive heat loss away from the injection zone. In such a case, the temperature level of the injection zone and the temperature distribution of the surrounding rock will affect the growth of the steam zone developed during the subsequent steam injection. Knowledge of the size of steam zone developed is important in determining the amount of oil displaced and the extent of heating in the reservoir. It is also useful to be able to compute the size of the steam zone for cases where thief zones take most of the injected steam. This paper presents a fairly straightforward method of estimating the steam zone developed in a preheated formation, based on certain simplifying assumptions. Some cases of steam injection into a reservoir at its original temperature have already been considered elsewhere.1-3 Because it is difficult or impossible to obtain an accurate representation of the temperature distribution in the reservoir some time after initial heating has taken place, in this work, a linear temperature profile in the cap and base rock is assumed (Figs. 1A and 1B). The linear profile has the advantage of yielding heat flow equations that

Jan 1, 1969

By D. M. Davidson, P. F. Kerr

The Permian Cutler Formation in Kane Creek, Utah bears uranium ore in fissure-type vertical veins. This fissure vein formation and its associated host rock bleaching, weak argillic alteration, and base-metal sulfide mineralization along with the development of hydrothermal minerals, a 60 m.y. isotopic Pb/U age, and well-crystallized uraninite's structural control, give evidence of ore deposition in a hydrothermal environment consistent with Laramide mineralization. Mica polymorph studies indicate first, that temperature during the hydrothermal alteration or subsequent ore deposition did not exceed 200°; hence Cutler deposits are classified telethermal; and secondly, that rock alteration and ore deposition are closely related both in time and space, as is seen in argillic alteration sequence. Ore deposition in an alkaline to slightly acid reducing environment is postulated. Ore textures and mineral associations and distributions also indicate that carbonate complexes may have been transported with the deposing ores. Bleached rock formations are not primary strati-graphic phenomena. Bleaching, or changes in ferric iron content, results from a chemical process involving reduction and conversion of hematite to a soluble aqueous ion. Nearby Chinle bedded deposits (Moss Rack sandstone), containing shale lenses, fossil wood, organic debris and minor fractures, overly the Cutler formation. Chinle ores have a single reported isotopic age of approximately 200 m.y. Their chemical environment is assumed similar to that of the Cutler ores, though slightly more acidic. Two genetic, paradoxically divergent interpretations of the bedded deposits are derived from available spectrographic, petrographic, X-ray and chemical analyses. Kane Springs canyon is located southwest of Moab, Utah, south of the Colorado River and northwest of the Lisbon Valley mining district (Fig. 1). Uranium deposits occur along the northeast flank of a Kane Creek salt anticline where two geologic formations bear ore, the Permian Cutler and the Triassic Chinle. The Cutler deposits' fissure vein formation is an unusual occurrence in the Colorado Plateau. The Chinle deposits appear to be of the normal sedimentary type. Associated with the vein deposits are a number of features, including argillic alteration of sedimentary strata. The minerals of these deposits and their associated features have been studied in detail in a two year program entailing both field and laboratory techniques. STRATIGRAPHY A stratigraphic section of the lithologic units in the Kane Creek area is given in Table I.

Jan 1, 1967

By E. F. Johnson, V. O. Naumann, D. P. Bossler

A method is presented for calculating individual gas and oil or water and oil relative permeabilities from data obtained during a gas drive or a waterflood experiment performed on a linear porous body. The method has been tested and found both rapid and reliable for normal-sized core samples. INTRODUCTION Individual oil and gas or oil and water relative permeabilities are required for a number of reservoir engineering applications. Chief among these is the evaluation of oil displacement under conditions where gravitational effects are significant, such as a water drive or crestal gas injection in a steeply dipping oil reservoir. Numerous proposed methods of obtaining relative permeability data on reservoir core samples have been too tedious and time consuming for practical use, or have yielded questionable and sometimes inconsistent results. A method has been developed by which the individual relative permeability curves can be calculated from data collected during a displacement test. The method is based on sound theoretical considerations. Using this method, with a properly designed experimental procedure, relative permeability curves can be obtained using core samples of normal size (i.e., 2 to 3 in. in length and 1 to 2 in. in diameter) within a few days after receipt of the core. In a recent publication D. A. Efros1 describes an approach to the calculation of individual relative permeabilities that is based on the same theoretical considerations. We believe the approach described in the present paper is more adaptable to practical application than the method implied by Efros. In addition, comparisons with independently determined relative permeabilities are furnished to substantiate the reliability of the new method. DERIVATION Previously the theory of Buckley and Leverett as extended by Welge2 as been used to calculate the ratio of relative permeabilities. In the derivation which follows, this theory is further extended to permit calculation of the individual relative permeabilities. The theory assumes two conditions which must be achieved before the method is applicable. They are that the flow velocity be high enough to achieve what has been termed stabilized displacement,3 and that the flow velocity is constant at all cross sections of the linear porous body. In stabilized displacement the flowing pressure gradient is high compared with the capillary pressure difference between the flowing phases. The high pressure drop insures that the portion of the core in which capillary effects predominate will be compressed to a negligibly small fraction of the total pore space. The assumption of constant flow velocity at all

By G. W. Thomas

Jan 1, 1967

By F. W. Bowdish

Analysis of data on the equilibrium exchange of sodium ions from salt solutions for calcium ions, from various fractions of an oxidized calcareous iron ore from Lorraine, France, strongly suggests that the calcium may be attached to the solid by only one of its bonds. Nearly equal values were found for the equilibrium expression from data covering concentration ranges of 1000 fold for the salt and 235 fold for the ion-exchangeable calcium. Differences indicated between the exchangeable calcium contents of the slime and the sandy portions of the ore, as well as the equilibrium relationship itself, are of importance in the flotation of the ore. During an experimental program on the flotation of oxidized calcareous iron ores from Lorraine, France, ion-exchangeable calcium was found in the ore. Because this source of calcium ions, if not controlled, could thwart all efforts at flotation, an extensive study was made of the phenomenon. The part of that study reported here was to obtain data on the equilibrium established between ore solids and solutions of sodium chloride, and to analyze the data to try to understand better the physical chemistry of the exchange. The ore contains oolites of goethite and quartz grains embedded in a matrix consisting of calcite and a brown cement believed to be the residue from oxidation of the silicate minerals found in fresh unoxidized ore. When ground, a substantial fraction of the ore becomes a fine slime in which is concentrated more than half of the alumina content of the ore. It is believed that the alumina occurs as unidentified clay minerals with which most of the ion-exchange capacity of the ore is associated. A preliminary experiment on the exchange of ions from the ore for sodium ions from solutions of NaCl showed that calcium was about ten times as abundant as magnesium in the exchange solutions. Thus calcium was the predominant ion in causing difficulties with flotation, and it was the only one for which data was obtained in this study. It is believed that the omission of magnesium ion exchange from consideration does not invalidate the results. It is possible, however, that inclusion of magnesium in the analyses and calculations or its prior elimination from the samples would have improved the consistency of the results. EXPERIMENTAL Data about the equilibrium exchange of sodium from solutions for calcium from the solids were obtained using two different samples of material. The solids

Jan 1, 1963

By F. J. Chernosky

Most previous studies of coal flotation utilized chemically pure reagents. Since such reagents are not available in quantity, a study of various wmmercially available reagents as frothers was undertaken. Flotation procedures similar to those used in a large number of commercial coal flotation plants were employed. The reagents used were compared on yield-selectivity-wst basis. The selectivity was found to be relatively constant with yields above 70% but the reagent cost varied greatly. The types of reagents available were compared. In the past, a number of papers have been written on the use of various reagents as coal flotation frothers. For the most part, these have been an attempt to correlate the intrinsic chemical and physical proper- ties of the reagents or reagent groups to the floata-bility of various fine coals. Because of the type or nature of evaluations made, chemically pure reagents were normally used. Generally, these reagents are unavailable in commercial quantities. Therefore, the results obtained with these reagents were usually limited in usefulness to the plant operator or metallurgist. This paper presents the results of a study in which commercially available forms of various reagents were used as flotation frothers. Flotation procedures which would conform in general to those employed in a large number of commercial coal flotation plants were utilized. The objective of the test was to compare these reagents on yield-selectivity-cost basis, that is, 1) the yields obtained, 2) the ash levels obtained and 3) the reagent cost relating to these results. REAGENTS The reagents employed have been grouped into the following chemical classes: 1) aliphatic alcohols, 2) terpines, 3) glycols (synthetic water-soluble

Jan 1, 1963

By T. K. Perkins, L. R. Kern, R. E. Wyant

Many factors influence the results of hydraulic fracturing to stimulate well productivity. Most of these factors have been studied and their effects discussed in the literature. However, the movement of sand through the crack during fracturing has received little consideration. Exactly where the sand goes in the fracture, how it gets there, and how this might influence results have not been studied, judging from published reports. To get an answer to some of these questions we have studied sand movement through a simulated crack segment in the laboratory. Although the work is preliminary, we have answered some questions about sand movement in a semiquantitative sense. We are reporting our results at this time in the hope of stimulating interest and further study in this important aspect of fracturing. LABORATORY EXPERIMENTS Fig. 1 is a schematic diagram of equipment used in this study. A segment of a vertical crack 0.25-in. wide was formed between two Plexiglas plates, and movement of sand in fluid through the crack was observed under various conditions. Sand and fluid were injected into the crack at constant rates. The top and bottom of the crack were closed so that the fluid flowed horizontally. Because of gravity most of the sand settled to the bottom before moving very far through the crack. If the fluid velocity exceeded a certain critical value, then all the sand injected was washed on through the crack even though it had settled to the bottom. When the fluid velocity was less than this value when sand injection was started, most of the sand which settled in the crack remained in place until enough of the area open to fluid flow was blocked by the settled sand to increase the fluid velocity to the critical value.. At this point all injected sand moved on through the crack even though it settled to the lower portion of the moving fluid section. All sand previously settled remained in place. When fluid velocity was increased at this paint by increasing fluid injection rate (without also increasing sand injection rate), some of the previously settled sand was washed out until enough area was open to flow that the velocity dropped to the critical value. Conversely, when the fluid velocity was decreased by decreasing the fluid injection rate (without changing the rate of sand injection), additional sand would settle until the critical velocity was again attained. In short, the critical velocity is an equilibrium value which the system will automatically seek, if possible. The equilibrium (or critical) velocity was measured under several conditions. The results of these measurements are shown in Pig. 2 where the equilibrium velocity is shown as a function of the amount of sand injected per minute. The points falling in the shaded band shown in Fig. 2 were obtained with water. The higher viscosities were obtained by adding CMC (carboxymethylcellu-lose). Data indicate that the equilibrium velocity is not very sensitive to sand size or transport fluid viscosity, or to sand injection rate (except at very low rates). Points connected by the solid CUrve in Fig. 2 were obtained with water and 20- to 40-mesh steel shot. The equilibrium velocity is apparently fairly sensitive to difference in density between particle and fluid. Only one run was made with a gelled fluid and that was gelled water at zero sand injection rate as shown in Fig. 2. (Measurements of equilibrium velocity at zero sand injection rate were made by filling the crack

By L. H. Simons, H. H. Spain

I NT RODUCTION Theory indicates that linear water-floods should exhibit scaling and stabilization properties in both oil-wet and water-wet porous media'. Experimental verification of these proper-ties in oil-wet media was obtained some time ago', and more recently, Perkins', employing a high permeability unconsoli dated sand and a low oil-to-water viscosity ratio, has obtained flooding data in a water-wet medium which follow the pattern of scaling and stabilization. Also, Root and Calhoun3 have found similar trends in the displacement of gas by liquids from unconsolidated sands. In general, however, the experimen-tal evidence appears still somewhat conflicting since scaling and stabilization of water floods in water-wet media has not been observed in most of the studies to date'.".'. The purpose of this paper is to provide a more comprehensive picture of waterflood behavior in water-wet media and to corroborate the validity of the scaling and stabiliza-tion concepts. The need for differentiating between the intrinsic nature of water-oil displacements inside a porous medium and perturbating secondary end effects is discussed, pertinent experimental procedures are outlined and the results of flooding tests conducted at low and at high oil-to-water viscosity ratios on con- solidated, water-wet Alundum cores are analyzed. END EFFECTS IN WATER-WET MEDIA Outlet End Effects The peculiar feature of the outlet end effect in water-wet cores is that it not only results in an excessive wetting phase saturation at the outflow face,'.' but that it also retards the moment of water breakthrough: i.e., causes an undue delay in the appearance of water in the produced effluent. Conditions existing in a water-wet core when water first arrives at the outflow face are schematically shown in Fig. 1. At this flooding stage the saturation distribution, Fig. la, has not yet been distorted by the outlet end effect. A water-invaded pore at the outlet core face, opening into the oil-filled void space between the end of the core and the end plate of the flooding cell, is illustrated in Fig. lb. The curvature of the water-oil interface is concave toward the outlet, and the pressure on the water side of the interface is lower than on the oil side by an amount equal to a certain capillary pressure (Fig. lc). Before the water (of this particular pore) can be produced from the core, it is apparent that the pressure in the water phase must exceed that in the oil outside the core. Accordingly. when water first reaches the outlet face, it will not leave the core but will accumulate inside, and only oil will be produced. Eventually, as the water injection is continued. sufficient pressure is built up in the water to reverse the curvature of the interfaces in the outlet pores, and only then will water be produced from the core. This stage, illustrated in Fig. 2. corresponds to the conventional "water breakthrough" observed in flooding experiments. For purposes of discussion, it is convenient to distinguish between the flooding stage at which water first arrives at the outlet face of a core (Fig. 1) and the stage at which water is first produced from the core ( Fig. 2). The former will be referred to as "water arrival" and the latter water breakthrough. In oil-wet media

By John R. Eckel

A Iaboratory drilling rig has been devised and placed in operation which permits the application of hydrostatic, terrastatic, and formation pore pressures to a rock sample for drilling under controlled conditions. Calibrated rests of this equipment in-dicate qualitative agreerment with field results on the effect of weight on bit, rotary speed, and rate of circ~~lation on drilling rate. Tests under pressure, using water and air as the drilling fluids, indicate that when drilling saturated limestone samples, a pressure differential between hydrostatic and formation pressure is the only pressure which affects drilling rate. A differential in either direction causes a reduction in rate, but the same diflerential from the wellbore into the formation causes the greater reduction. Also, in lime.stone, air favorably affected drilling rote only when compared to water at high hydrostatic pressurre, all other conditions remaining constant. Large increases in drilling rate with air could be obtained in the laboratory only while drilling a shale, which offered a limited permeability to nitrogen and which could be drilled with a differential from the formation into the wellbore. In general, the magnitude of the changes due to the pressure effect were small compared to changes in drilling rate obtainable by changing either bit weight or rotary speed. INTRODUCTION In the course of Humble's drilling research program, the desirability of investigating several factors which lend themselves to laboratory rather than field study became obvious. Principal reasons for this are the expense involved in field studies and the difficulty of exercising the required degree of control. Of primary importance are investigations of the effect of fluid pressure on rock rate, the effect of pressure on rock drillability, and a study of the fundamental mechanics of rock failure and chip generation. Several years ago some work in the laboratory on the effect of mud properties on drilling rate was re- ported to the API', but the completion of this work and the investigation of the other variables affecting drilling rate required equipment somewhat more elaborate than that used on these earlier tests. This equipment has been assembled to form Humble's Drilling Engineering laboratory. To date only the effect of pressure on rock drillability while using water and air has been investigated, and this work is not yet complete. Facilities are available to investigate the effect of fluid properties on rate of penetration, and information is being assembled on a current basis on the mechanics of drilling. The sequence of presentation in this progress report is somewhat different from that in which the data were obtained since the order of ex-

By P. K. Guerrero, N. Arbiter

The correlation of power consumed by a tumbling mill with the dimensions, speed, and load has been attempted by three principal methods. One of these, the torque formula, has been reviewed critically elsewhere.' This approach, while analytical, disregards the individual motion and speed of the tumbling bodies and assumes that the gross geometry of the entire load is sufficient to establish the power requirement. A more general approach, using dimensional analysis, has been applied in this laboratory by the present authors1'1, and more recently by Rose.3 This disregards the internal conditions within the mill entirely and in effect is an empirical procedure for correlating data. Finally, the complete analysis of tumbling mill dynamics as first made by Davis' attempts to develop the power requirement from individual ball and rod paths and velocities. Although Davis' result is applicable only at a single speed for a given size and load it could be useful, since his optimum speed is at the upper limits of conventional tumbling mill practice. Where power values predicted by the Davis equation were compared with actual mill requirements, they were found too low for small mills and much too high for the largest mills. This suggested that the agreement with actual intermediate range requirements might be fortuitous, and the development of the equation was examined critically. The probable source of error was found to lie in an erroneous assumption made by Davis in converting the energy of ball or rod motion to power. With the same approach, but with different assumptions, a modified equation has been obtained which predicts results that are usually consistently high. This is to be expected, since it is assumed that there is no slip between load and shell and no interference among tumbling bodies. Both these effects should account for the difference. Development of Equation: When a cylindrical shell charged with a substantial weight of rods or balls is slowly rotated about a horizontal axis, the load surface first assumes an inclination that gradually increases to a maximum. Continued rotation results in failure within the load, characterized by a sliding of the top layers down the slope and a reduction in the angle of inclination. Further rotation again increases the angle, and the cycle repeats. If the speed is increased, there is continuous load failure and build-up and a continuous stream of rods or balls descends over the upcoming mass, which moves in circular arcs. At higher speeds the up-

Jan 1, 1961

By G. R. St. Pierre, A. J. Wilhelem

In connection with the reduction of hematite or magnetite to metallic iron, it appeared desirable to study the rate of reduction of each oxide to the next lower oxide under conditions which excluded all other condensed phases. For this purpose, pellets of Fe,O,, Fe,O,, and oFeOo were prepared. A gas train for the preparation of Ar-Hz-H20 mixtures was employed to establish a continuous flow of a controlled atmosphere over a pellet suspended on a balance. In this manner the following individual reduction steps could be studied: For example, pellets of hematite were reduced solely to magnetite by adjusting the gas atmosphere such that it was oxidizing to wustite. In other words, the reaction was complete when the hematite pellet had been totally converted to magnetite. The pellets were prepared from Baker Chemical ferric oxide. The ferric oxide powder was moistened and handrolled. Air sintering at 950°C produced Fe,0, pellets with a density of 3.65 g per cm3. The pellets ranged in size from 0.6 to 1.2 cm diam. The magnetite and

Jan 1, 1962

By A. W. Ruff

The seismic wave produced by an explosive is very important in blasting. A true understanding of the wave is only important when considering possible structural damage to buildings located near the blast, but it is also of primary importance when considering rock fragmentation. Because of the importance of the seismic wave to fragmentation, a finda-mental study of the seismic wave and its effect was made at the Cananea Consolidated Copper Co.'s mine at Cananea, Sonora, Mexico. Strain gages were used to measure the strain in the rock when nearby explosive charges were detonated. The variance in the circuit voltage caused by changes in the resistance of the strain gage was measured with an oscilloscope and recorded with a Polaroid camera. CONFINED DRILLHOLE CHARGES To record the seismic wave from a confined drillhole explosive charge, a gage was cemented to the bottom of a 4-in diam drillhole. The horizontal gage hole and nine horizontal charge holes were drilled perpendicular to the center line of the 2-39 drift and 4 ft above the floor. The gage hole was 29 in. deep, and the charge holes were drilled so that each bottomed 1 in. deeper than the gage hole. The gage was installed with its axis horizontal and parallel to the 2-39 drift. The charge holes had the following horizontal distances from the gage: hole 3, 3 ft; hole 4, 4 ft; hole 5, 5 ft; hole 9, 9 ft; hole 10, 10 ft; hole 14, 14 ft; and hole 15, 15 ft. Figs. 1 and 2 show the relationships of the gage and the charge holes. Charges of 80 pct special gelatin, which weighed 50, 100, and 150 g were exploded in the charge holes and the seismic waves were recorded. All charges were stemmed with 200 g of mud and detonated with an electric blasting cap. Fig. 3 shows the recorded seismic waves. The reflected components of the waves were too weak to be objectional on these records. Table I shows the data taken from the wave forms. The approximate average velocity of a wave with a 9-ft travel distance was 12,900 fps. Since different charge weights were used, it was necessary to scale the data. The scaled maximum compress ion-strain data and the scaled maximum tens ion-strain data are shown in Figs. 4 and 5, respectively. Figs. 6 and 7 show the strain-distance data from Figs. 4 and 5 converted to a 1-lb charge. The results showed that the wave consisted of an initial compression pulse followed by a tension pulse. The amplitude of the tension pulse approached the amplitude of the compression pulse as the distance between the gage and the shot increased. The decay rate of the seismic wave is very important in the evaluation of the blasting characteristics of a rock. When high-amplitude stresses are produced, as in rock fragmentation, the rate of decay is much more rapid than when small amplitude stresses are produced. In underground mine blasting, the burdens are usually not greater than 6 ft and the decay is very rapid. The rapid seismic-wave decay rate in the fragmentation range means that in practice, if the minimum weight of powder is used to break a specified burden, only a very

Jan 1, 1961

By M. J. Mullikin, J. B. Conway

Equations capable of representing combined first -and second-stage creep behavior include the Andrade, Cottrell-Aytekin, Garofalo, and de Lacombe equations along with a third-degree polynomial in t1/3. Employing some recent high- temperature creep data for arc-cast tungsten the effectiveness of these expressions was evaluated. It was found that, in this instance, the most accurate representations were afforded by the third-degree polynomial in t1/3 and the de Lacombe equation. Also noted was the fact that the Andrade equation gave results just slightly inferior to those obtained with the third-degree polynomial, definitely better than those due to the Cottrell-Aytekin equation, and slightly superior to those furnished by the Garofalo equation. EXPRESSIONS designed to describe combined first -and second-stage creep behavior must possess certain specific characteristics. At time zero the equation must reduce to a strain value corresponding to the instantaneous deformation on loading. Then in the early moments following load application transient-creep behavior must be described to reflect a creep rate which constantly decreases with time. And finally at larger time values linear (or at least nearly linear) strain-time behavior must be accounted for to yield the steady state or minimum creep rate. Some equations, such as the one based on the parabolic creep law, yield fairly satisfactory results in transient analyses but are not effective at all when combined first- and second-stage data are to be analyzed. Other equations give equally accurate results in both instances although in general the complexity of the equation is slightly increased. For example the Andrade1 equation: has found fairly general application in expressing first-stage creep behavior. Furthermore the contribution of the exponential term becomes more and more prominent as the time increases and causes this equation to exhibit a point of inflection at large values of time. With a proper selection of equation constants the curve can be made fairly flat in the region of the inflection point indicating a desirable equation form for combined first- and second-stage creep analysis. Another equation form with the proper characteristic is that of Cottrell and Aytekin:2 e =a +bt1/3 +ct [2] Obviously as time increases the first derivative of this expression approaches a constant value to suggest application to combined first- and second-stage creep analysis. Similarly a third-degree polynomial in t1/3: e -a + bt1/3 + ct2/3 +dt [3] has the same characteristics. Expressions of this type have been shown3 to be very effective in the analysis of transient-creep behavior and the fact that the first derivative has the same characteristic as Eq. [2] suggests use in the combined analysis.

Jan 1, 1967

By R. J. Stokes

S. Floreen (international Nickel Co.)— One fairly simple way to differentiate between em brittle me nt due to surface microcracks or due to a dislocation barrier effect might be to load a brittle rock salt crystal in tension to some stress just below the fracture stress and then immerse the loaded crystal in water. If a surface film is retarding plastic flow then one should observe some plastic deformation of the rock salt as

Jan 1, 1962