Search Documents

By M. Kaufman, D. Rosenthal

IT would be of great importance to our understanding of the phenomena of fracture in metals if a unique relationship could be established between stress and some easily measurable parameter of deformation up to the fracture. It is known that no such relationshir, has been found when the parameter was the mechanically measured strain,' the major part of which is plastic. The present investigation was initiated to determine, among other things, if the X-ray or lattice strain" might be a • For the definition of lattice strain see ref. 2. more suitable parameter. To this end, annealed stress free*' specimens of 61s aluminum alloy were ** Freedom of stress in this particular case means absence of those stresses which can be checked by the technique employed. tested at room temperature and liquid nitrogen temperature (—195OC), by first maintaining the same temperature from the beginning until the end of the test, and then with crossovers from one temperature to the other. Three major factors had to be taken into consideration in designing the specimens and testing equipment. The test setup must allow X-ray pictures to be taken while the specimen is under load (for as much as 2 hr at a given load). Provision must be made for maintaining the test specimen at the temperature of liquid nitrogen as well as at room temperature. In addition, the specimen must be able to rotate to allow the X,ray beam to take in as large a sample of grains as possible, to maintain centering, and to correct for effects of large grains. The final design of the specimen and equipment is shown in Figs. 1 and 2. The specimen and grips are hollow, enabling liquid nitrogen to be poured in through a funnel-container arrangement while the specimen is rotating. The diameter of the specimens at the 2 in. gage section is 5/16 in. with a 3/16 in. diameter hole through the center. A small pilot valve in the lower grip allows a very small stream of liquid nitrogen to escape, thus assuring circulation of the liquid at all times. A low temperature silicone grease-tissue paper packing is used to seal all joints. The specimen is held in the grips by a pin. Spherical socket joints at top and bottom are used to allow for proper alignment. The grips are connected to the thrust bearings by means of plastic couplings to decrease the heat flow. Rotation is accomplished by means of a motor which drives a plastic gear (integral with the lower coupling) through a semiuniversal joint to allow for deforma- tion of the specimen. The torque applied to the specimen is very small. The maximum stress caused by this torque at the highest loads reached was found to be less than 200 psi, The whole unit was mounted in a 10,000 Ib testing machine. An insulation cage was placed around the specimen to cut down air circulation, and consequent frosting of the specimen, and also to reduce radiation and convection heat losses. A narrow window, covered on both sides with a thin plastic materiai allows the X-ray beam to reach the specimen. Thermocouples placed just outside the pilot hole at the lower grip, in the bottom of the funnel container and near the top of the funnel container permitted checks of the

Jan 1, 1955

By Charles C. Templeton

Any explanation of the nature of multiphase fluid flow through porous media must ultimately involve a knowledge of the mechanism of such flow through the individual pores of the medium. To provide quantitative information of the latter type, a study was made of gas-liquid and liquid-liquid displacements in single uniform Pyrex capillaries with diameters as small as 4 microns. Quantitative observation of air-liquid displacements with zero contact. angle indicates the adequacy of a model based on Poiseuille's Law and the independence of capillary pressure and interfacial velocity. Consequently, this work discounts previous claims that flow is abnormally retarded in very small capillaries. For the displacement of a very viscous oil by water in a capillary initially filled with water, flow is faster than predicted by the above model. This behavior is probably due to the existence of a thin annular film of water between the oil column and the capi1lary wall as postulated by Yuster. INTRODUCTION The need for experimental observation of the microscopic nature of multiphase fluid flow, to further the understanding of the results of macroscopic flow investigations, has been previously recognized. Chatenever and Calhoun1 in connection with the API Project 47B have extensively studied the flow of fluids through layers of glass or lucite spheres between glass or lucite plates. Lowman2 prepared an interesting film concerning flow in a single capillary, using an apparatus which made a moving interface appear stationary in the field of a microscope. The purpose of the present investigation was to deter- mine what contribution might be made to this field by comprehensive quantitative observation of gas-liquid and liquid-liquid displacements in single capillaries of uniform, very small diameter. The techniques used are similar to those of Lowman but incorporate improvements necessary for quantitative work. Quantitative observations must begin with fluids of simple chemical composition to provide a basis for analyzing the more complex behavior that may be expected with natural reservoir fluids. However, some qualitative observations on systems involving crude oils are presented. Experiments of this sort involve the observation of the velocity of the interface separating two fluids, and have been performed in horizontal or vertical capillaries by several investigators.3,4,5,6,7,8 The most general treatment of their mechanics, by Brittin,9 involves forces due to capillarity, viscous resistance, gravity, an "end-drag" effect, and the rate of change of momentum of the capillary contents. Since the range of pore diameters of interest in petroleum reservoirs is about 0.1 to 100 microns, it was desired to perform these experiments in the smallest capillaries possible, to ascertain whether capillary diameter had any effect on the existing theory based on data from larger capillaries.* It was also important to be able more systematically to vary the direction and rate of displacement than was done by previous investigators. Finally, to concentrate attention on viscous and capillary forces by eliminating gravity as a parameter, this study involves only horizontal capillaries. The work of Eley and Pepper' with a "large" (0.766-mm diameter) horizontal capillary is the previous investigation which is most similar to the present method of quantitative observation. Here capillaries with diameters as small as 4 microns will be studied.

Jan 1, 1955

By C. J. Pritchard, R. D. Jacksha, K. V. Raj, C. Sunderman

"In underground metal/nonmetal mines, repeated short-term exposure to high levels of airborne contaminants can become a serious health issue. Currently there are no common mechanisms to control or mitigate these localized short-term high exposures to contaminants. To improve miners’ health and safety, the National Institute for Occupational Safety and Health (NIOSH) Spokane Mining Research Division (SMRD) is developing a Smart Monitoring and Control (SMAC) system for real-time monitoring of mine air quality, with integrated countermeasures to reduce these high concentrations of airborne contaminants in localized sections of mines. To develop and test a SMAC system that could be implemented in an underground mine, researchers at SMRD built a test apparatus incorporating a fan, louver, ducting, and sensors combined with atmospheric monitoring and control software. This system will institute effective countermeasures to reduce contaminant levels, improving miner safety and health. INTRODUCTION Underground metal/nonmetal mines continue to expand production by employing large equipment in greater numbers and by mining into hotter and more challenging environments which are typically further away from fresh air sources. This challenges the ventilation system to provide adequate quantities of fresh air to the workplace and increases the potential for air contamination that could result in immediate and long-term health and safety issues to miners. Currently, airflow and contaminant concentration in mines are typically based on full-shift sampling which may not identify repeated short-term exposures to high levels of airborne contaminants such as NO2, CO, diesel particulate matter (DPM), and dust with high silica contents. Repeated short-term exposure to these high levels of airborne contaminants over long periods can lead to serious health issues which are documented in several studies [1-3]. Based on data from ventilation surveys and contaminant measurements, mines often attempt to minimize exposures by sending more fresh air to the working areas where contaminant concentrations are found to be high. Real-time monitoring and control allows mines the option to monitor contaminants and implement controls in real-time. Real-time monitoring and control of the underground mine atmosphere has been a subject of research since its beginning in Finland at Outokumpu’s Pyhäsalmi Mine in 1969 [4]. More recently, the real-time monitoring and control problem has been addressed using ventilation on demand (VOD) systems [5, 6]. A VOD system attempts to deal with this issue on a global level by controlling fans, louvers, and doors to increase airflow to active work areas and reduce contamination. However, a VOD system only controls the distribution of airflow and may not be helpful in mines where the airflow is at maximum capacity. Researchers at SMRD are developing a SMAC system for real-time monitoring of mine air quality with an integrated ability to trigger countermeasures to reduce high concentrations of airborne contaminants in localized sections of mines. The SMAC system will be designed, tested, and refined in the laboratory environment before being demonstrated in an underground mine. This system will not only try to minimize the gaseous contaminants by regulating the airflow locally, but will also potentially employ a filtration system to reduce the airborne contaminants in case the maximum airflow capacity is reached. This paper presents development of the test apparatus for the SMAC system."

Jan 1, 2017

By E. H. Edwards, J. Washburn, E. R. Parker

The mechanism of strain hardening was discussed in connection with some recent observations on the stress-induced motion of dislocation boundaries and on the simple shear deformation of zinc, cadmium, and copper crystals. THE nature of work hardening accompanying plastic deformation of metals has long been the subject of speculation. As early as 1925, Becker' suggested a thermodynamic process to explain slip based on the statistical probability for the occurrence of local glide steps. The theory that plastic glide was due to the presence of lattice defects known as dislocations was first introduced in 1934 by Taylor: and Orowan, and Polanyi.4 Taylor presented a rather complete analysis of the behavior and interaction of dislocations. He assumed a certain type of stress field to be associated with a single dislocation, and postulated that work hardening would result from plastic flow in two ways. First, dislocations following one another across a slip plane might encounter a barrier which would impede their movement, thereby gradually diminishing and eventually stopping plastic flow on that plane. The second mechanism involved the concept of dislocations traveling in opposite directions on nearby planes attracting one another to form a meta-stable lattice. The interaction of dislocations in such an array was shown to be large enough to require an increase in stress before plastic flow would continue. Kochendorfer5,6 extended the concept of work hardening involving a back stress due to an accumulation of like dislocations on a slip plane. He suggested that the formation of a new dislocation is hindered by those dislocations already present. The hardening resulting from the interaction of newly forming dislocations with bound dislocations was designated as "formative hardening," and indicated to be the only type of hardening necessarily connected with the slip process. In another detailed picture of work hardening, Mott and Nabarro7 assumed that sufficient dislocations are primarily present in the crystal as a consequence of growth irregularities (mosaic structure) to initiate slip. Work hardening resulted from the interaction of migrating dislocations with localized internal stresses produced during deformation. More recently, however, Mott8 as modified his treatment to include the generation of dislocations at Frank-Read sources and he accounts for the hardening by the formation of sessile dislocations. It has been widely accepted that metals of hexagonal structure, usually possessing only one set of glide planes, exhibit mechanical behavior markedly different from those having cubic structures. However, the differences in behavior of the two classes may be less fundamental than suggested. Rohm and Kochendorfer9 have shown that a single crystal of aluminum subjected to approximately simple shear gives the type of stress-strain curve associated with the hexagonal metals. Similar behavior has been demonstrated for gold and silver crystals by Andrade and Henderson.'" Thus when glide in cubic metals is restricted to one plane, the strain hardening characteristics closely resemble those of the hexagonal metals. It would appear, therefore, that an investigation of some of the strain hardening properties of single crystals sheared in simple glide might provide a more complete understanding of the phenomenon of work hardening. This report relates a number of recent observations made on single crystals of copper, zinc, and cadmium tested in simple shear. In addition, pertinent observations of effects accompanying stress-induced movement of dislocation boundaries are reported. Experimental Procedure and Results Details of the production and advantages of the type of single-crystal shear specimen employed for the following tests have been presented in a previous publication.n The method of testing makes

Jan 1, 1954

By Philip R. Strand

Vermiculite is the name used for those micaceous minerals with a ferromagnesian aluminum silicate composition and the unique property of exfoliating to a low density material when heated. Commercially, the exfoliated products are also called vermiculite, or more exactly, expanded vermiculite. In its exfoliated state, vermiculite serves many markets, chief among them construction, agriculture, horticulture, and general industry. Composition and Properties Vermiculite, in a natural state, has the characteristic mica habit, splits readily into thin laminae, flexible but inelastic. In a mineralogical sense, vermiculite is a hydrated silicate with no exact chemical composition. The structural formula for a macroscopic, trioctahedral vermiculite may be written: [ ] The structure of vermiculite is basically that of a talc, and it could be regarded as a trioctahedral member of the smectite group. Macroscopically, vermiculite displays the prominent monoclinic crystal faces that are commonly marked by lines at 60° and 120°. Hardness varies from 1.5 to 2.8 or more; specific gravities are between 2.1 and 2.8; color is amber, bronze, brown, dark green, or black. When heated quickly to an elevated tempera¬ture, vermiculite expands by exfoliating at right angles to the cleavage, into wormlike pieces (the name vermiculite is derived from the Latin `vermiculare,' to breed worms). This characteristic of expansion, the basis for commercial use of the mineral, is the result of the mechanical separation of the layers by the rapid conversion of contained water to steam. The increase in bulk volume of commercial grades is 8 to 12 times, but individual flakes may expand as much as 30 times. All expansion occurs at right angles to the basal cleavage. Vermiculite may also be expanded by a number of chemical processes such as soaking in hydrogen peroxide, weak acids, and other electrolytes. There is a color change during expansion that is dependent upon the composition of the vermiculite and furnace atmospheres. Heating in an oxidizing atmosphere produces dull gray colors. The iron is generally not oxidized during chemical expansion. The exfoliation of the vermiculite crystal results in large pores being formed between groups of platelets. Thus exfoliation can make available a large increase in void volume without significantly changing the surface area of the platelets. This characteristic is important in the chemical carrier applications of vermiculite. Although there has been much research on the chemical and structural composition of vermiculite, an exact formula and composition cannot be described. This is to be expected since vermiculite derives from a number of different mineral sources. As has been proposed, vermiculite is not a single mineral species, but families of related minerals. Additionally, commercial interest in the end product has resulted in the name vermiculite being applied to the very voluminous, bloated products from calcination of baueritized (bleached) phlogopite. This has also resulted in many names for the specific vermiculite found in a particular deposit. All of this difficulty arises from the nature of its crystal chemistry. Consequently, vermiculite is now thought of as several related variable mineral series rather than a group of related species. The vermiculites have a common silica tetrahedral sheet structure with separating ions. It is now generally agreed that vermiculites must also be regarded as true clay minerals. The properties of the mineral, such as cation exchange capacity, organic complexing ability, and vari-

Jan 1, 1975

By E. L. Dougherty, J. W. Sheldon

Using the numerical techniques shown in this paper it is possible to compute the simultaneous dynamical behavior of multiple fluid-fluid interfaces in two dimensions. Hence, fluid-fluid inter face models of several oil recovery processes can be constructed which allow prediction of the effect of fluid flow in two dimensions on the behavior of these processes while taking into account various physical effects, such as saturation gradients, phase changes and thermal stimulation. Techniques for constructing fluid-fluid interface approximations for several oil recovery processes are reuiewed. The validity of these techniques is established by comparison to experimental results. The availability of a computer program for computing the behavior of multiple fluid-fluid interface problems makes the use of potentiometric models uneconomic. The results indicate the areal sweep efficiencies obtained from potentiometric model studies at high mobility ratios may be somewhat optimistic. The results also indicate that the areal sweep efficiency curves obtained at high viscosity ratios 2—4 using miscible fluids in porous media are misleading and that considerable care must be taken in defining the mobility ratio for a displacement process, especially for miscible fluids. INTRODUCTION Many of the methods used to estimate areal sweep efficiency assume that the oil recovery process can be represented by one or more fluid-fluid interfaces. When the mobilities of the displaced and displacing phases are assumed equal, this quantity can be computed using analytical mathematical methods.l.2 For non-unit mobility ratio, experimental methods based upon the fluid-fluid interface concept include potentiometric and electrolytic blotter models. Scaled models of porous media have also been used to study this problem. Fay and Prats3 and Aronofsky4 used numerical techniques to compute breakthrough sweep efficiency, but neither of these studies were extended beyond breakthrough. Numerical techniques which have been developed for solving a general fluid-fluid interface problem are presented in a companion paper.5 These techniques have been incorporated into a program for a large-scale digital computer which can treat simultaneously the dynamical behavior of as many as six fluid-fluid interfaces. Thus it is possible to simulate the behavior of several oil recovery processes taking into account such effects as saturation gradients. Techniques for representing several oil recovery processes by fluid-fluid interfaces are presented. Processes considered are conventional water flooding with and without a mobile gas phase, hot water flooding, miscible flooding and enriched gas drive. Other processes, whose effects could be studied by this technique, but are not considered in detail here, are underground combustion and steam injection processes and various slug processes such as gas-water injection and alcohol-solvent processes. Theoretical results obtained with single interfaces for mobility ratios of 0.25, 1, 2, 4, 8, 16, 32 and 64 in a repeated five-spot are compared to experimental results for this geometrical configuration obtained using miscible fluids and conductive analogs. Predicted recovery curves for water flooding in a repeated five-spot are compared to experimental results reported in the literature. Example calculations are shown for a conventional water flood in the presence of a free gas phase and for a hot-water flood. MATHEMATICAL PRELIMINARIES We assume that the rock matrix is homogeneous with constant thickness, absolute permeability and porosity. The fluids are incompressible. We also neglect the effects of gravity.

Jan 1, 1965

By E. S. Machlin, Morris Cohen

GRENINGER and Troiano' were the first to establish the fact that the habit planes of mar-tensitic products are usually planes of high indices. In steels containing 0.55 to 1.4 pct C, the habit plane indices are {225}A.* In steels with more than 1.5 pct C and in Fe-Ni alloys, the indices are {259In. The presence of both habits in certain alloy steels has been reported recently." Several investigators"-" have subsequently attempted to provide an understanding of these complex habit relations. Bowles'6 work is the most recent and the most successful. Like Greninger and Troiano, Bowles concluded that the transformation occurs in two steps, the first, a strain (or displacement) that is homogeneous throughout the volume of the martensitic plate, while the second is a strain that is homogeneous only within relatively small portions of the martensitic plate. The first strain yields the measured tilt of the specimen surface, but does not produce the final martensitic structure. This and the correct lattice orientations are achieved by the second strain. Bowles also reasoned that each strain could be described by the motion of a plane of no rotation and no distortion (hereinafter called the invariant plane) in a direction not necessarily contained in the plane.** In applying these considerations to steels with {225}a habits, Bowles found that the habit plane coincided with the invariant plane of the first strain, a result which signified that the atoms move along certain paths during the transformation. Whether this correlation results from the existence of some sort of a plate nucleus6 lying along the invariant plane of the first strain, or whether the invariant plane primarily defines the preferred directions of propagation of the first strain, will be discussed in a later section. There are two obstacles to the general acceptance of Bowles' treatment: 1—it has not been shown that elements of his double-strain analysis are unique, and 2—it is not known whether his treatment applies to transformations yielding the {2591A habit. Unless these two questions are resolved, the correlation between the habit and invariant planes in the {225}A systems might be regarded as coincidental. Experimental Procedure Using a 70 pct Fe-30 pct Ni alloy as an example of a {2591A habit system, preliminary attempts were made to obtain a double-strain solution based on a Nishiyamax** set of orientation relations and on a Greninger-Troianot set, adopting the stereographic technique described by Bowles" who used the Kurd-jumow-Sachstt set in connection with the {225}A habit. These attempts were unsuccessful because the method is essentially one of trial and error and requires either exceedingly good luck or a prior knowledge of the results being sought. Consequently it was decided to measure independently the strain associated with the formation of a martensitic plate as well as the habit-plane indices of the particular plate. By means of matrix algebra, the measured strain was applied to the austenite lattice, whose orientation with respect to the strain was determined, in order to ascertain whether the resultant structure would have the lattice of martensite. This was not found to be the case, and hence it was assumed that the measured strain was the first of two Bowles-type strains, characterized by the motion of an invariant plane. A solution of this displacement was obtained in terms of the invariant-plane indices and the direction of motion of the atoms. Then the second strain

Jan 1, 1952

By F. I. Stalkup

Vapor-liquid phase equilibrium experiments have been conducted in a static equilibrium cell on mixtures of a light, 45 API stock- tank gravity reservoir fluid and a rich hydrocarbon gas containing approximately 55 mole per cent of intermediate hydrocarbons. Both apressure-vs-composition study of the gas and a simulated reservoir fluid, and a multiple-batch contact simulation of the condensing-gas-drive oil recovery process were performed. In the latter experiments equilibrium gas and liquid compositions were analyzed. Also, conventional, "condensing-gas-drive", long-tube displacement experiments of the reservoir fluid and gases of various richness were performed. The results of these experiments could not be satisfactorily interpreted by the conventional pseudo-ternary-diagram representation of multicom-ponent phase behavior. The results seem to be explained better by considering a bubble-point surface and a dew - point surface joined in a plait-point locus. Portions of the plait-point locus cannot be "seen" directly by the rich hydrocarbon gas because of curvature of the dew-point surface. In such a system, continuous injection of the rich gas over stationary reservoir fluid might form a zone of contiguously miscible compositions from pure rich gas to pure reservoir fluid by: (1) saturating the reservoir fluid with injected gas to the bubble-point surface; (2) creating by mass transfer with fresh injected gas a path of contiguously miscible compositions along the bubble - point surface to the plait-point locus; and (3) creating by mass transfer with additional injected gas a path of gas compositions along the dew-point surface up to the point where direct miscibility results between dew-point fluid and the injected rich gas. INTRODUCTION The use of the pseudo-ternary-phase diagram to illustrate miscible displacement phase behavior has been discussed by several authors.5,7 Such a representation of phase behavior is not rigorous, but the ternary diagram nevertheless gives a qualitative picture of what actually occurs in a miscible displacement process. Fig. 1 is a typical illustration of miscible displacement phase behavior by a ternary diagram. The multicomponent hydrocarbon system is divided into three pseudo-components: a light fraction containing methane and nitrogen, an intermediate fraction containing ethane through hexanes plus carbon dioxide, and a heavy fraction containing heptane and heavier components. A two-phase region is bounded by a dew-point curve and a bubble-point curve, which are joined at the critical point. The concept deduced from such a representation for miscible displacement by a condensing-gas-drive process is as follows: a rich gas G, which lies to the right of the limiting tie line through the critical point C, is injected into the reservoir and contacts reservoir fluid L, saturating the reservoir fluid to give bubble-point fluid L1 and equilibrium dew-point gas GI. Continued injection of rich gas changes the composition of the saturated liquid L1 through a series of liquid compositions lying along the bubble-point curve, until the critical composition C is reached, at which point direct miscibility with the rich gas is achieved. Some equilibrium gas with compositions lying along the dew-point curve from G1 to C is also formed in this process. Thus, continuous injection of rich gas forms a

Jan 1, 1966

By Fred H. Poettmann, J. K. Welchon, A. F. Bertuzzi

A series of gas-lift tests was made which verified conclusions reached from previous studies and which showed that gas-lift performance could be calculated if reasonably accurate data on the producing characteristics of the well are available.' The test installation is described. Its essential features consisted of a rate-of-flow controller to regulate the input gas and an auxiliary string of small diameter tubing to conduct the input gas to the point of injection in the eductor tubing. In general, it was shown that for continuous gas-lift operation, injection should be as deep as possible consistent with available gas pressure and rate of production desired. Calculated curves are presented which completely characterize the gas-lift performance of the well tested. INTRODUCTION A correlation based on field data from a large number of flowing and gas-lift wells, covering a wide range of oper- ating conditions, was developed which permitted the logical design and prediction of the performance of continuous gaslift installations.' The procedure made it possible to calculate the depth, pressure, rate at which to inject gas, horsepower requirements, and the effect of production rate and tubing size on these quantities. As a result of these calculations certain significant facts were brought to light or re-emphasized. The presence of efficient ranges of operation and an optimum injection depth for a given rate of production were shown to exist as well as the fact that the lower the tubing pressure the less the horsepower requirements to lift the reservoir fluids. The fact that the gas lift performance can be predicted by calculation does not mean that any given installation will perform over the entire range predicted. Many of the conventional continuous gas-lift installations will not permit operation over the entire range predicted by calculation (particularly in the efficient range of operation) because of the performance characteristics of the installation. Consequently, field tests were initiated for the purpose of investigating and arriving at a simple continuous gas-lift installation which would operate under all conditions predicted by calculation. The first series of tests involved the use of mechanically operated valves. These valves were slip-joint type valves which are opened and closed by lowering or raising the tubing at the surface by means of a hydraulic jack. These valves were chosen because it was hoped that they could be opened or closed at will and did not depend on the gas pressure in the annulus or the pressure of the fluid in the tubing for their operation. Unfortunately, the valves did not operate satisfactorily and it was imossible to open or close each valve at will. Some limited data were obtained, however, which verified the predicted benefits resulting from the use of lower tubing pressures. In the next series of tests the valves were removed and the gas was injected from the annulus through a 14/64-in. orifice equipped with a reverse check valve. A packer was set below the valve to seal off the space between the 2-in. tubing and 5-in. casing. The injection point was located at 3,810 ft. This arrangement provided satisfactory mechanical operation. However, it was impossible to obtain low rates of liquid production at low tubing pressures as predicted by calculation. A minimum gas injection rate existed below which steady flow was impossible. Liquid pro-

Jan 1, 1953

By D. R. Spink

In the preparation of zirconium and hafnium metals, there are many indications that impurity levels control the physical and chemical properties of the metal. The purpose of this work was to develop methods of producing pure raw materials that could be reduced to highly pure metals. Early work by the Carbomcndum Metals Co. and others indicated that certain fused salts were highly efficient in the removal of impurities from gaseous zirconium tetrachloride. We have developed workable fused salts along with a continuous and higly efficient scrubber design which can be used to produce very pure zirconium tetrachloride. Metal made from the purified tetrachloride has been shown to approach crystal bar metal in purity. Scrubber designs are discussed with recommended materials of construction. VARIOUS methods have been used to purify zirconium (and hafnium) tetrachloride. Most of these methods involve thermal degassing initially (the removal of more volatile compounds at temperatures below that where zirconium tetrachloride exhibits a significant vapor pressure) followed by a sublimation step (which conveniently leaves the nonvolatiles behind.) While these procedures effect definite purification, such processing is batchwise and does not result in optimum purification. Early work by The Carborundum Metals Co. and others indicated that fused salts were highly efficient in the removal of impurities from gaseous zirconium tetrachloride. The objective of this work was to develop a continuous fused-salt scrubbing process that would yield high-purity zirconium and hafnium tetrachlorides. The term "fused salt" refers to the molten state of ionic compounds and as used here, specifically refers to alkali metal and alkaline earth halides. Fused salts, because of their many valuable properties, are becoming increasingly im- portant industrially. These valuable properties of the fused salts are: 1) stability at elevated temperatures; 2)low vapor pressure; 3) low viscosity; 4) ability to dissolve many different materials; and 5) good elctctrical conductivity. All but the last property enhances the use of fused salts as scrubbing agents for zirconium and hafnium tetrachlorides. Fused-salt scrubbing may be defined for our purpose as the intimate contact between a gaseous compound and a molten salt with the purpose of removing impurities from the gaseous compound. In this case, the gaseous compound is zirconium tetrachloride which submlimes at 331°C. When this gas is contacted with a molten salt, impurities such as aluminum, iron, and titanium are removed and the resulting product is considered fused-salt scrubbed. Zirconium and hafnium are sold to a purity specification. Generally speaking, normal production without scrubbing is capable of meeting specifications. There are, however, several considerations That make the knowledge of methods of additional purification highly desirable. These are: 1) the control of product analysis; 2) the production of a pure product for special applications; 3) the salvaging of off-specification material; and 4) the use of raw materials with higher impurity levels than 1 he present raw materials. You will note that items 1 and 2 are somewhat complementary to each other and that item 4 assumes that The present process has impurity removal limitations. To better understand the part that fused-salt scrubbing of zirconium tetrachloride plays in the manufacture of zirconium, Fig. 1 contains an illustrated flow sheet of the industrial production of zirconium metal. In referring to Fig. 1, we can point out the specific benefits derived from our ability to purify zirconium tetrachloride. Purification of the crude (hafnium-containing) zirconium

Jan 1, 1962

By H. D. Yoo, M. R. Tek, D. L. Katz

The behavior of gas-storage reservoirs subject to water drive is investigated through analog simulation on an electronic differential analyzer. The simulation technique developed on an LM-10 computer permits the prediction of reservoir volume or pressure resulting from the movement of water in the surrounding aquifer. The method developed on the analog computer consists of setting up an appropriate transfer-function circuit and feeding the arbitrary time-varying boundary conditions as an input signal. The input may be specified as gas reservoir pressure, pore volume or the water flux. Several cases studied include an isolated gas reservoir on a limited aquifer, interference among three reservoirs adjacent to a common aquifer and the growth of gas-storage volume on an aquifer. It is concluded that the method developed on an electronic differential analyzer provides an excellent technique to simulate and investigate the behavior of gas reservoirs subject to water drive. The agreement between the reservoir performance as predicted from the simulation technique and as measured from actual field data is found to be better than the range usually encountered in predicting water-drive behavior. INTRODUCTION It is generally known that some gas-storage reservoirs are located on top of blanket sands of large extent, saturated with brine called aquifers. Because the volume of the body of water associated with aquifers is usually very large and water is compressible, the cyclic pressure variations encountered in normal storage service inevitably cause unsteady, compressible flow conditions in the adjacent aquifers. The solution to radial diffusivity equation for a limited aquifer for constant terminal conditions has been know111 since the early 1930's. The solution for the constant terminal conditions for an infinite aquifer was puhlished in 1949.2 The proldem of handling the unsteady flow of water through a porous medium with time-varying boundary conditions has been studied by many investigators through the use of graphical,3 digital4 computing techniques or on an R-C type of electrical analog called a reservoir analyzer.5 The digital computer performs a series of arithmetic calculations to superimpose the solutions of constant terminal conditions, while the R-C type of analyzer permits the simulation of the continuous medium by means of "lumped-cell" electrical resistance and capacitance elements based on finite-difference approximations. The graphical method is based on the convolution integral solution of the partial differential equalion. The purpose of the present study was to simulate the behavior of an aquifer on a general-purpose, electronic differential analyzer to obtain the influx of water or the change in pressure at the reservoir continuous1r and quickly. The use of the electronic differential analyzer was based on well known solutions for the constant terminal conditions already available in the literature.2,6-8 THEORETICAL DEVELOPMENT MODEL. The gas reservoir (R) is described in relation to an aquifer in Fig. 1. The permeability and porosity of an aquifer sand, the viscosity of the brine, and the compressibility of the brine and the formation are assumed to be constant. It further is assumed that the blanket sand is not thick enough to give appreciable vertical pressure distribution. The model in consideration is circular with radial geometry. The aquifer radius can be of limited or infinite extent. The gas-water interface is assumed to be fully segregated, and the displacement of the gas by

By C. F. Weinaug

In recent years, several methods have been proposed for correlation of physical properties and behavior of hydrocarbon systems. These correlations were found to be unsatis. factory for predicting differential gas liberation and shrinkage data now obtained experimentally. To achieve the desired accuracy, methods were developed to adjust the correlations with two experimental pieces of data. The required data are the saturation pressure at reservoir temperature and the specific volume or density at these conditions. The saturation pressure is used to adjust the vapor-liquid equilibrium-ratio correlations. The density measurement is used to determine the apparent density of liquid methane. With these adjustments, differential gas liberation and shrinkage are predicted within the experimental error. EQUILIBRIUM RATIOS Equilibrium ratios published by the National Gas Assn. of America' were used as a basis for the calculation. The convergence pressure for the sample was determined by using a hypothetical binary system, with methane as one component and the remainder of the system lumped together to form the other component. For the critical temperature of the heavy constituent, the weight average of the critical temperatures of the ethane and heavier components was used. The correlation of pseudo-critical temperature, density and molecular weight of Matthews, Roland and Katzl was used in this calculation for the hexane and heavier fractions. Points for the critical properties of methane and the hypothetical heavy constituent were placed on a chart of the critical loci of hydrocarbon systems. A critical loci for the hypothetical binary system was drawn, and the critical pressure of this system was determined at the reservoir temperature. This critical pressure became the convergence pressure for which equilibrium ratios were read from the NGAA4 charts. The NGAA equilibrium constants for C, through C, were used for the C,; through C, fractions since they were separated on the bases of normal paraffin boiling points. The correlation of Hoffman, Crump and Hocott5 as used to estimate the equilibrium constant for the C fraction. The NGAA eauilibrium-ratio charts are based primarily on paraffin systems and have to be corrected for the presence of naphthenes and aromatics. The correlations in the literature proved too unsatisfactory for determining these corrections. EXperimental data, the saturation pressure, is used with the correlation of Soloman2 to make this adjustment. The correction factors of Soloman were cross-plotted so that the factors for ethane and heavier components were made a function of the factor for methane. a2 = 0.7927 at + 0.1646 .........(1) where a, = correction coefficient for for methane NGAA equilibrium ratio, and a2t = correction coefficient for ethane and heavier NGAA equilibrium ratios. This equation and the equation for the bubble point are solved simultaneously for the correction coefficients. The bubble-point equation is anKnXn = 1.0 . . . . (2) The equilibrium constant for the saturation pressure at reservoir temperature is used. This procedure not only corrects for the relative volatility between components, but also assures that the results will be directly related to the experimental saturation pressure. METHANE DENSITY After determining a set of equilibrium ratios, the apparent density of condensed methane at 60°F and 14.7 psia was estimated. This was found to be necessary because the correlations of Standing3 ere found to be unsatisfactory. The law of additive volumes was employed to make this calculation.

By W. R. Foster, J. M. McMillen, A. S. Odeh

The complete equations of average linear momentum balance for a single-phase fluid in an incompreisible, homogeneous, porous medium are derived. The derivation begins with Euler's equation of motion for a continuum and uses an integral transform recently developed by Slattery. 7 For steady flow of a compressible, Newtonian fluid, the usual equations of motion result. For transient flow, the space-time description of the pressure is determined in the lowest approximation by the telegrapher's equation. From the analysis a new phenomenological coefficient results which connects the viscous traction to the deriv ative of the linear momentum density. The magnitude of this coefficient determines the velocity of sound through the pore structure in this approximation to the pressure field. INTRODUCTION The modification of Darcy's law of momentum balance for steady, single-phase flow through porous media has been discussed for many years. The first modification was suggested by Forcheimerl who added terms of higher order in the velocity. These can be expected to appear because the underlying microscopic equations of momentum balance are themselves nonlinear in the point velocity field. The Reynolds tensor pvv, which represents the convective flux of momentum density, appears in the momentum balance equation. Only in rectilinear flow (parallel stream lines) does the divergence of this tensor vanish. Since the steady flow stream lines in most porous media are not parallel, nonlinear dependence of the pressure gradient on the velocity should naturally appear. This nonlinearity has nothing to do with turbulence in the ordinary sense of random fluctuations in the pressure and velocity fields. It arises simply because the stream lines converge and diverge, even for steady flows. Klinkenberg2 demonstrated that the permeability coefficient in Darcy's law depends on the absolute pressure or, alternatively, on the density field. However, because he neglected inertial terms of the Forcheimer type, his correction coefficient could not be represented by a constant but tended toward a constant as the velocity decreased. Forcheimer's and Klinkenberg's modifications can be combined in a rigorous way to account for both inertia and slip during steady flow. This will be shown in a future paper. The transient change of pressure in porous media has been described by the diffusion equation.3 This form results from eliminating the velocity and density fields from a combination of the equations of motion in the form of Darcy's law, the continuity equation and an equation of state. Fatt4 suggested that the cause of deviations from the prediction of the diffusion equation for pressure transients lies not in the choice of Darcy's law as the equation of motion but on the existence of dead-end pores which might invalidate the averaged equation of continuity. On the other hand, Oroveanu and Pascal5 noted chat the time derivative of the momentum density must be included in the equations of motion since this measures the local rate of change of momentum density. Their differential equation for pressure is the telegrapher's equation (neglecting gravity). However, their form of this equation predicts that the speed of pressure propagation through the pore structure is the same as that through the bulk fluid. M. K. Hubbert6 attempted a derivation of Darcy's law by volume averaging the Navier-Stokes equations. Since these equations represent momentum balance at a point within an open set of points containing the fluid itself, Hubbert's volume averaging cannot lead to tenns involving transfer of momentum between the fluid and the walls of the pores. Once these viscous tractions are lost by choosing a control volume containing only fluid, they cannot

By I. Fatt, D. N. Saraf

A method is described for measuring two- and three-phase relative permeabilities in sandstones or sand packs using a nuclear magnetic resonance (NMR) technique to determine fluid saturations. Two- and three-phase relative permeabilities have been determined on Boise sandstone using the NMR technique of saturation measurement. Three-phase relative permeability to water was found to depend only on the water saturation, whereas three-phase permeability to oil depended on both the water and oil saturations. Relative permeability to gas in three-phase flow was found to depend only on the total liquid saturation. INTRODUCTION Three-phase relative permeabilities are extremely useful in calculating field performance for reservoirs being produced by simultaneous water and gas drives. Three-phase relative permeability data are also needed for analyzing solution gas-drive reservoirs which are partially depleted and are being produced by water drive. Some thermal recovery processes involve three-phase flow which require three-phase relative permeability data for predicting reservoir behavior. Unfortunately, three-phase relative permeability measurements have rarely been made. Also, because of the scarcity of three-phase data, it has not been possible to date to relate other measured rock characteristics to the relative permeabilities with a great certainty. Leverett and Lewis,l Reid2 and Snell3 have reported three-phase relative permeability data on unconsolidated sands. Leverett and Lewis used ring electrodes spaced along the length of the sand sample to measure the resistivity of the sample which was assumed to be monotonically related to brine saturation. Gas saturation was determined from pressure-volume measurements. Oil saturation was obtained by material balance on the cell containing the sand sample. This method is involved and time consuming. Another difficulty arises from the fact that the resistivity of the sand is a function not only of saturation of brine but also of the distribution and saturation history of the brine in the pore spaces. Reid2 used a gamma ray absorption technique for measuring liquid saturation. This method has the disadvantage that total liquid saturation rather than oil or brine saturation is all that can be measured and still another method is required to determine the saturations of individual components. snel13 used a neutron bombardment method which also required a separate determination of the individual component saturations. Caudle et al.4 measured three-phase relative permeability on consolidated sandstones using vacuum distillation for determining fluid saturations. Distillation after each reading makes this technique very lengthy and time consuming. Corey et a1. 5 and Naar and Wygal 6 measured three-phase relative permeability on sandstones by the capillary-pressure method. Semipermeable diaphragm assemblies were used at each end of the core specimen to keep the water phase in the core. Gravimetric methods were used to determine fluid saturations. Sarem7 recently reported an unsteady-state method for measuring three-phase relative permeability on sandstones. This method is an extension of Welge's method8 for measuring two-phase relative permeability. Although Sarem's method is simple and comparatively fast, the assumptions involved may oversimplify the problem. Sarem's assumption, that in all rocks relative permeability to each fluid will depend only on the saturation of that fluid, seems to be rather unrealistic. Neglecting capillary effects at the end of the core is also a weak assumption. Donaldson and dean9 measured three-phase relative permeability using a method similar to Sarem's.

By H. J. Beach

This paper describes the properties and uses of cementitious mixtures containing 10 to 20 per cent bentonite and 2 per cent or more sodium chloride. Salt additions improve bentonite cement by increasing early strength and stabilizing viscosity without changing thickening time. The volume yield of cement is increased 20 per cent or more without lowered early strength. The unique properties of salt in this type of system make savings of more than 10 per cent possible where light-weight, low-strength cementitious material is adequate to fill behind pipe. The density range of these mixtures is 12 to 13 1b/gal and the filtration rates are medium-low, aiding the placement of long columns in primary cementing. Where available, sea water is an ideal salt source in preparing these slurries. INTRODUCTION Accelerators increase the rate of cement hydration. In speeding reaction with water, they can affect cement performance by reducing thickening time and increasing early strength after set. Accelerators that cause both effects can be considered "total", and those which produce only one effect can be called "partial". Materials that act conversely to accelerators are retarders. and they also can be total or partial. Partial retarders. make high-temperature cementing possible by prolonging thickening time while impairing early strength only slightly. Most known accelerators are total. A useful partial accelerator would increase strength without altering thickening time. This report concerns material combinations in which a normally total accelerator exerts unique partial properties. Utilizing these limited effects improves bentonite cement. SEPARATE EFFECTS OF SODIUM CHLORIDE AND BENTONITE ON CEMENT Sodium chloride and calcium chloride are commonly used in cement, and at low concentrations their effects are similar. Ludwig1 studied sodium chloride in API Class A and Class E cements.' Figs. 1 and 2 sum- marize his data taken at 140°F on slurry pumpability and early strength. Normal water-to-cement ratios are 0.46 and 0.40, respectively, for Class A and Class E cements. At saturation the salt content is 12 to 14 per cent of the weight of cement solids. The illustrations, therefore, cover the range of salt solubility in the mixing water. The data show that sodium chloride is a total accelerator of the Class E cement studied over the range of solubility. It is total in Class A cement to 6 per cent. Above this concentration salt becomes a retarder— partial from 6 to 10 per cent and total above. These results are general for different brands of Class A cement, although the shift from acceleration to retardation may occur at different salt concentrations for different temperatures. The results are specific for the Class E brand studied. In discussing his data, the investigator emphasized that different Class E cements might respond erratically to salt, depending on the brand chosen. Morgan and Dumbauld3 escribe the properties of bentonite cement in a paper that is a standard reference in using these systems. Bentonite functions in cement as an enabling agent; each bentonite addition equal to 1 per cent of the weight of dry cement makes possible the further addition of 4.5 per cent water.' Thus, 10 per cent bentonite in Class A cement, normally slurried neat with 46 per cent water, enables the slurry to contain another 45 per cent before settling. Water is one of the lightest and cheapest cement extenders; it reduces density and cost. Bentonite cements produce uniform slurries having low susceptibility to solids separation. After set, they offer low resistance to perforation.

By S. Leber

Pole figures and pole distributions were used for the quantitative detevinination of the preferred orientations in swaged tungsten rods and the effect of subsequent wire drawing on the texture. In the swaged rod, the preferred orientation could best be described as a pseudo (111-112) [110] cylindrical texture. A secondary component was assigned the indices (110) [001]. Further reduction by swaging retained these cylindrical textures. Wire drawing caused lattice rotations about the wire axis, converting the cylindrical texture into a normal fiber texture. The conversion was initiated and proceeded lnost rapidly in the center of the wire. The (110) /001] component was converted into a (110) fiber texture by wire drawing. The effective rotations were about axes contained in the lransverse plane. The preferred orientation associated with tungsten wire usually has been assumed to consist of a (110) fiber texture. This implies rotational randomness around the wire axis. Leber1 has shown that rotation about the wire axis may be restricted resulting in a preferred orientation designated as a cylindrical texture. Specific crystal planes tended to remain aligned parallel to the wire surface. This texture seemed to be the result of the swaging operation used in the intermediate stages of fabrication. Bhandary and cullity2 detected similar textures in swaged iron wires. Peck and Thomas3 suggested that the preferential alignment of crystal planes parallel to the wire surface was the result of tangential flow, characteristic of swaging. Wire drawing, following swaging, twisted the crystals about axes parallel to the wire axis, converting the cylindrical texture into a normal fiber texture.4 In this investigation, quantitative techniques were used to more accurately characterize the cylindrical texture and to study its conversion into a normal fiber texture. PROCEDURE The material investigated was lamp-quality tungsten. Three ingots were rod-rolled and swaged to 112 mils diameter, and were then drawn to 10-mil wire. Samples were obtained after each drawing pass. (110) and (200) pole figures were obtained from samples prepared as shown in Fig. 1. To form the sample shown in Fig. l(a), l-in. lengths of wire were glued side by side to form a l-in. square. The composite was ground and electropolished to form a diametral reference surface. For certain specific tests, samples were prepared as shown in Fig. l(b). This composite was made from 10-mil-thick strips prepared by grinding on parallel longitudinal planes. The polished reference surface in this case was, in essence, the wire surface. The two forms of samples [~igs. l(a) and l(b)] were related by a 90-deg rotation around the wire axis. Because of ease in preparation, most data was obtained from the diametral samples. The X-ray data were obtained using a Siemen's pole-figure goniometer. These data were not corrected for absorption or defocusing effects, since only the angular locations of the intensity maxima were of interest. Regions of the pole figure more than 80 deg from the center were not investigated. Partial monochromatization was achieved by the use of unfiltered copper radiation and a krypton-filled proportional counter. No evidence of spurious white radiation effects5 could be detected. Therefore, more elaborate X-ray techniques were not required. Intensity levels were chosen to show all pertinent details of the orientation distributions. The reference plane for the pole figures was the diametral section. The orthogonal reference directions, shown in Fig. 1, were the wire axis, the radial axis per-

Jan 1, 1965

By L. K. Nemeth, H. T. Kennedy

The investigation presented in this paper was performd to develop a relationship between the dewpoint pressure of a hydrocarbon reservoir fluid and its composition, temperature and characteristics of the heptanes-plus fraction. This is part of an over-all program to develop correlations for the essential properties of reservoir fluids with the composition, temperature and pressure. In all, 579 dewpoint pressures, determined experimentally on 480 different condensate systems along with their compositions and temperatures, were obtained from the literature and from previously unpublished sources. Compositions were expressed in terms of mol fractions of methane through heptanes-plus, nitrogen, carbon dioxide and hydrogen sulfide, and the molecular weight and specific gravity of the heptanes-plus fraction. Distribution of the samples assured a good representation of most of the major oil-producing areas in the world. A multiple regression analysis program written for the IBhl 7094 digital computer was used to compare numerous equations and to calculate their coefficients. The final equation containing 11 coefficients resulted in an average absolute deviation of 7.4 percent for all the samples. Segregation of data on the basis of one or more variables did not materially improve the results. The correlation developed is useful in predicting the dewpoint pressure of condensate reservoir fluids from variables normally measured in the laboratory. INTRODUCTION In any discussion of the dewpoint of hydrocarbon mixtures, it should be noted that there are two kinds of dewpoints which must be distinguished. The first type, or normal dewpoint, usually occurs at low pressures as dry gas is compressed to the point where liquid first forms. The second type, sometimes called the retrograde dewpoint, occurs when a gas mixture containing heavy hydrocarbons in solution is depressured until a liquid is formed. The first type, as applied to condensate mixtures, would normally be below atmospheric pressure and would be of no interest in reservoir performance. The second type is of great significance in interpreting reservoir behavior and is the property referred to throughout this paper. Laboratory measurement of dewpoints is laborious, expensive and subject to many errors, and it is desirable to find a reasonably accurate method to predict dewpoint pressures from the composition and other readily available properties normally measured on the fluids. A possible method of calculating dewpoint pressures would be to employ a set of equilibrium ratios (K-values). 1 However, the K-values themselves depend on composition as well as temperature and pressure; presently their use adds nothing to the accuracy of predicting dewpoints. If future correlations provide a significantly better method of estimating K-values for heavy hydrocarbons, they may be of greater use. REVIEW OF LITERATURE In 1944, Olds, Sage and Lacey 2 studied the behavior of fluids from the Paloma field. Properties of seven mixtures were studied at temperatures ranging from 100 to 250 F and at pressures up to 5,000 psia. From the information obtained, the influence of composition and temperature upon the dewpoint was determined. It was found that removal of the intermediate molecular weight components from the mixture resulted in considerable increases in the dewpoint pressures. Results indicated that the effect of changes in temperature was relatively minor when compared with the effect of modifying the mixture by removing the intermediate components. Thus, removal of propane and butanes resulted in an increase in dewpoint pressure more than 10 times greater than a change in temperature from 100 to 250 F. The authors stated that althouah

By E. F. Sumner, C. K. Eilerts

A partial differential equation was developed and programmed to compute the cumulation and flow of the liquid phase of a gas-condensate fluid. It was found that such cumulation near the wellbore may attain saturations permitting liquid flow within periods of the order of 1 day. The radius at which this saturation obtains advances toward the external reservoir boundary with continued recovery, although it may move only 2 to 5 percent of that distance during the productive life of the well. This relatively high saturation can cause a serious dimunition in delivery capacity. The pressure dilferential required for flow of gas through a formation with such saturation may be as much as three times greater than the corresponding differential in a formation where saturation is less than 2 percent. Because of the radial geometry, total reservoir pressure differential required for flow with a zone of relatively high liquid saturation about the well was indicated to be two to three times greater than without the saturation. INTRODUCTION A computing program developed by the authors 1 to integrate the partial differential equation for transient radial flow of gas-condensate fluids required a prevalued function, W(Po) = kD(p0)/M(p0)z0(Po).......(1) related to cumulation of condensate in pores of the reservoir as pressure declines. Methods were available for relating mobility to saturation, saturation to liquid-vapor volume ratio, liquid-vapor volume ratio to pressure and, hence, mobility to pressure to obtain W(pD). However, certain assumptions had to be made in advance of program execution concerning movement of the liquid phase and change in radial distribution of liquid saturation as recovery progressed. Because reservoir fluid mobility and related pressure gradients are markedly dependent on liquid saturation, it was desirable to eliminate the necessity for these assumptions and in this respect improve the reliability of a result obtained by integration. Experimental research was begun to determine the relationship of variables that control the reservoir flow of low liquid-vapor ratio fluids. With the porous medium partly saturated, permeability to a two-phase flowing fluid could not be measured because the viscosity was unknown. Viscosity could not be measured without knowing permeability of the liquid-containing medium; but mobility, the ratio of permeability to viscosity, was measurable. This property, and liquid-vapor volume ratio of the flowing phases, have values determined by nature of the fluid and porous medium and, in addition, were found to be pressure, velocity and saturation dependent. A program developed by Kniazeff and Naville 2 for computing multicom-ponent, two-phase flow used polynomials to represent saturation dependence of permeabilities for gas and for liquid phases, and an effluent viscosity constant under reservoir conditions. The authors developed a program based on a partial differential equation of similar mass balance origin but designed to represent distribution of an unlimited number of components between gas and liquid phases utilizing the experimental information available as to variable dependence. Attention was given to liquid-vapor volume ratio of the flowing phases, a property which is essential as a liquid saturation boundary condition. A program for computing implicitly saturation increase at uniform intervals along the drainage radius was combined with an existing program for calculating pressures, and the two were used alternately in an iterative solution for both saturations and pressures. The function W(pD) needed for calculating pressures was replaced by the ratio of mobility to compressibility factor. BASIC EQUATIONS The partial differential equation for calculating

By D. R. Parrish, W. E. Lamoreaux, W. W. Owens

Several methods are now being used by the industry for determining the gas-oil flow characteristics of reservoir rock samples. Most of the laboratory experirnerltal rlzethods can be classified either as steady-state, unsteady-state, or Hassler-type tests. Although the validity of the results obtained by any specific method is not questiorzed, in general, the complexity of these methods leaves much to be desired. This investigation dealt primarily with the development of a simplified unsteady-state (gas drive) laboratory test method for measuring kg/ko re1ationships of small core samples. A procedure for selecting uniform and representative sandstone rock samples was developetl since it was observed that the reliability of the test resu1ts was dependent upon having a sample which was uniform in porosity and permeability. The importance of factors known to influence the results of this type test, such as oil viscosity, gas expansion, and pressure differential were eva/uated. The method of calculating the test results was found to give valid data over a wide range of gas expansion. The limits of pressure differential and oil viscosity which would insure reliable results were defined. In cornparisorz with other presently used testing methods, the gar drive method requires considerably less apparatus, is less time consuming, and is more readily adaptable to a routine type core test. INTRODUCTION The detcrmination of the gas-oil flow characteristics of reservoir rocks has been given concentrated study by the oil industry for the past decade. Both analytical and experimental methods have been considered. The analytical methods have generally relied on a simplified characterization of porous materials in order that the fundamental laws of fluid flow and capillarity could be utilized. More direct experimental methods have been proposed by numerous production research laboratories with the primary goal being a simple and inexpensive, yet reliable method for measuring the gas-oil flow characteristics of reservoir rock samples. The experimental methods which have been reported in the literature generally can be grouped into one of the following classifications: 1. Steady state1,2,3,4 in which a fixed ratio of flowing fluids is forced through the test sample until equilibrium conditions of saturation and pressure gradient are established. 2. Hassler or stationary liquid method5,6,7 in which a uniform liquid saturation is attained, usually through capillary desaturation, then gas or oil is flowed through the sample at a pressure drop insufficient to disturb the pre-established saturation. :. Non-steady state method1,8,9,10 In which only the gas phase is injected into the sample, causing displacement of the oil phase and resulting in a system of continually changing average saturation and saturation gradient. The steady-state and Hassler methods have been most popular because, no doubt, of the simplicity of the actual relative permeability calculations. The problems involved in these methods were limited to apparatus design and measurement techniques. In the steady-state method, for instance, a constant flowing fluid ratio is established which, at equilibrium, results in an essentially uniform saturation over the entire length of the test sample. Therefore, the basic flow characteristics could be obtained by correlating the average saturation of the test sample with the kg/ko determined at various flowing gas-oil ratios.

Jan 1, 1957

By K. L. Young

The general energy equation, including change in kinetic energy, was solved by numerical integration and used to evaluate simplifying assumptions and application practices over a wide range of conditions. When extreme conditions were encountered, sizable errors were caused by large integration intervals, application of Simpson's rule and neglecting change in kinetic energy. A maximum error of only 1.31 percent was caused by assuming temperature and compressibility constants at their average value. It was discovered that a discontinuity can develop in the integral for the injection case. This discontinuity indicates a point of zero pressure change and is an inflection point in the pressure traverse. INTRODUCTION When a pressure in a gas well is to be calculated, one of the first decisions is to select a method of calculation. In many instances, this selection becomes a problem because the literature, at best, provides an evaluation of any method for only a limited range of conditions. Once a method has been selected, a question often arises as to the size of the calculation interval which should be used. The question regarding calculation interval arises because an analytic solution is not obtainable and approximate solutions must be used. This paper presents an evaluation of major assumptions and application practices of probably the two most widely used methods for calculating steady-state single-phase gas well pressures. The two methods are Cullender and Smith1-" (numerical integration), and average temperature and com-pressibility. The Cullender-Smith method assumes that change in kinetic energy is negligible and is normally applied in two steps with a Simpson's rule correcton. The average temperature and compressibility method, in addition to neglecting kinetic energy change, asumes that temperature and compressibility are constant at their average values. This method is normally applied for wellhead shut-in pressures of less than 2,000 psi, and in one step. Computer programs were written to compute bottom-hole pressure with and without the assumptions, using various approaches. Values of input parameters investigated are shown in Table 1. Flow rate was limited to a maximum of 5,000 and 10,000 Mcf/D for tubing sizes of 1.610 and 1.995 in. ID, respectively. Flow rate was also limited to 10,000 Mcf/D for a tubing size of 2.441 in. ID when wellhead flowing pressure was 100 psia. These limitations were imposed on flow rate so as not to exceed sonic velocity. The z factor routine available necessitated limiting bottom-hole temperature to 240F and wellhead pressure to 3,000 psia. Pressures were compared on the basis of percent deviation from the trapezoidal integration of Eq. l or 2 at 100-ft intervals. A preliminary investigation indicated that a 1,000-ft interval solution would differ from a 50-ft interval solution by less than 0.25 percent; therefore, the 100-ft interval was chosen for a base. For the purpose of comparison, deviations less than 1 percent were considered insignificant. where 111.1 qeC2/d p = kinetic energy term. Eqs. 1 and 2 can be evaluated numerically at specific depths using the trapezoidal rule as shown by Cullender and Smith. If change in kinetic energy is neglected and temperature and compressibility are assumed constant at their average values, Eq. 1 can be integrated to give the average temperature and compressibility equation,