Dynamics of Fluids in Porous Media

Israel.

Preface

This book is an attempt to present, in an ordered manner, the theory of dynamics (actually, also of statics) of fluids in porous media, as applicable to many disciplines of science and engineering. For some years I have taught courses on flow through porous media, and have treated this subject as a part of other courses, such as ground water hydrology, while at the Technion—Israel Institute of Technology, at M.I.T. where I spent my sabbatical leave (1966–7), and at several other institutions. I have felt the lack of a suitable textbook on this subject. Ideally, such a text should start from first principles of fluid mechanics and mechanics of continua, should show the passage from the microscopic to the macroscopic level of treatment, should emphasize the special features of porous media, establish the macroscopic theory and then show how it is applied to cases of practical interest.

It is rather surprising that in spite of its importance in many fields of practical interest, such as petroleum engineering, ground water hydrology, agricultural engineering and soil mechanics, so small number of treatises is available on fluids in porous media. This circumstance is even more surprising in view of the vast amount of literature published on the subject in a number of scientific and engineering journals. Although dynamics of fluids in porous media could become an interesting interdisciplinary course serving several departments, I believe that the relatively small number of courses offered by universities on the subject is due in part to lack of a suitable textbook. To overcome this lack I prepared notes for my own classes, which I present here in the form of a book, hoping that it will serve others in a similar situation.

The book is designed primarily for advanced undergraduate students and for graduates in fields such as ground water hydrology, soil mechanics, soil physics, drainage and irrigation engineering, sanitary engineering, petroleum engineering and chemical engineering, where flow through porous media plays a fundamental role. The book, I hope, will also serve the needs of scientists and engineers already active in these fields, who require a sound theoretical basis for their work. The emphasis in this book is on understanding the microscopic phenomena occurring in porous media and on their macroscopic description. The reader is led to grasp the meanings of the various parameters and coefficients appearing in the macroscopic descriptions of problems of flow through porous media, and their actual determination, as well as the limitations and approximations inherent in their description. In each case, the objective is to achieve a clear formulation of the flow problem considered and a complete mathematical statement of it in terms of partial differential equations and a set of initial and boundary conditions. Once a flow problem is stated properly in mathematical terms, three methods of solution are possible in principle: analytic solution, numerical solution aided by high speed digital computers and solution by means of laboratory models and analogs. All three tools are described in this book. Typical examples of analytic solutions are scattered throughout the book, but no attempt is made to present a collection of a large number of solved problems. The principles of the numerical method of solution are presented, and a detailed description is given of laboratory models and analogs, their scaling and applications.

Mathematics is employed extensively and the reader is expected to have a good background in advanced engineering mathematics, including such subjects as vector analysis, Cartesian tensor analysis, partial differential equations and elements of the theory of functions.

No attempt is made to give a complete citation of all published literature or to indicate the first author on a particular subject. References selected for citation are those I think represent a more important point of view, are more appropriate from the educational point of view or are more readily available for the average reader.

Obviously a single book, even of this size, cannot include everything related to the subject treated. Although we consider porous media in general, the discussion is limited to media with relatively large pores, thus excluding clays and media with micropores or colloidal-size particles. Similarly, chemical and electrochemical surface phenomena are excluded. The discussion is restricted to Newtonian fluids.

With these objectives and limitations in mind, the book starts with examples of two important porous media: the ground water aquifer and the oil reservoir. An attempt is made to define porous media, and the continuum approach is introduced as a tool for treating phenomena in porous media. This requires the definition of a representative elementary volume based on the definition of porosity. Chapter 2 includes a summary of some important fluid and porous media properties. In chapter 3, the concepts of pressure and piezometric head are introduced. Chapter 4 starts with the definition of velocities and fluxes in a fluid continuum. Then the equations of conservation of mass, momentum and energy in a fluid continuum are presented, and using a porous medium conceptual model these equations are averaged to obtain the basic equations that describe flow through porous media: the equations of volume and mass conservation, including the equation of mass conservation of a species in solution (also called the equation of hydrodynamic dispersion), and the motion equation for the general case of an anisotropic medium and inhomogeneous fluid. Although the basic equations of motion and of mass conservation are developed from first principles in chapter 4, chapters 5 and 6 return to these topics, discussing them from a different point of view, perhaps more suitable for the reader who is less versed in fluid mechanics. Chapter 5 presents the equation of motion, starting from its original one-dimensional form (as suggested by Darcy on the basis of experiments), and extending it to three-dimensional flow, compressible fluids and anisotropic media. This chapter also contains a review of theoretical derivations of Darcy’s law. My objectives in presenting this and similar reviews is to indicate research methods, such as the use of conceptual and statistical models. A section on the motion equation at high Reynolds numbers is also included.

In chapter 6, the control volume approach is introduced as a general tool for developing mass conservation equations, Special attention is devoted to deformable media. Also included in this chapter is the stream function and its relationship to the piezometric head. Once the continuity or mass conservation equations have been established, the next natural step is to consider the initial and boundary conditions. These are discussed in detail in chapter 7. Special attention is given to the phreatic surface boundary condition and to its description in the hodograph plane. The second part of this chapter contains a discussion on various analytic and numerical solution techniques.

Upon reaching this point, the reader should be able to state a problem of flow through porous media in terms of an appropriate partial differential equation and a set of initial and boundary conditions. He should also know the major methods of solution (analog solutions are discussed in chapter 11).

Chapter 8 deals with the problem of flow in unconfined aquifers. This is a problem often encountered in ground water hydrology and in drainage. The Dupuit assumptions are explained and employed to derive the continuity equations for unconfined flow. The hodograph method, as a tool for solving two-dimensional, steady phreatic flow problems, is discussed in detail with many examples. Several linearization techniques and solutions of the nonlinear equation of unconfined flow are also presented in this chapter.

In chapter 9 the discussion, hitherto confined to single-phase flow, is extended to polyphase flow in porous media, a topic of special interest in petroleum engineering. Starting from the fundamental concepts of saturation, capillary pressure and relative permeability, the motion and continuity equations are established. The case of unsaturated flow as treated by soil physicists is presented as a special case of flow of immiscible fluids, where one of the fluids—the air—is stationary and at constant pressure. Special cases of interest, dealing with infiltration into soils, are considered in more detail. A new concept is introduced: that of an abrupt interface as an approximation replacing the actual transition zone that occurs between two fluids, whether miscible or immiscible. A detailed discussion is presented on the coastal interface, of great interest to ground water hydrologists.

Chapter 10 deals with hydrodynamic dispersion. Again, although the fundamental equation is developed from first principles in chapter 4, a review of several other theories leading to this equation is presented. Special attention is given to the coefficient of dispersion and its relationship to matrix and flow characteristics. A section on heat and mass transfer completes the discussion on hydrodynamic dispersion.

Chapter 11 presents the use of models and analogs, both as research tools and as tools for solving boundary value problems. Following the presentation of a general method for deriving analog scales, a detailed description is given of the sand box model, the electric analogs of various types, the Hele–Shaw analogs and the membrane analog. Recommendations for application are indicated in each case.

In brief, this is the subject matter I have chosen to cover in this book. I have made an effort to present the information in such a way as to require a minimum of supplementary material, except for those who wish to dig more deeply into the subject. A large number of problems and exercises is included in this book.

I should like to express my appreciation to the many individuals who, through their comments and criticism, have contributed to the completion of this book. Special thanks are due to Dr. Y. Bachmat, Dr. C. Braester, Mr. E. A. Hefez and E. Goldshlager, for the help they have given me in reading, discussing and constructively criticizing the draft. Thanks are also due to the Department of Civil Engineering at M.I.T., and especially to Professor C. L. Miller, head of the department, Professor A. T. Ippen and Professor D. R. F. Harleman, who made it possible for me to write a large part of this book while spending a most fruitful year as a visiting professor at M.I.T.

The heaviest burden involved in writing this book was borne by my wife, Siona, who had to put up with the many inconveniences that are unavoidable when one is engaged in writing a book. For her constant encouragement to me throughout the various stages of writing, my hearty gratitude.

I realize that an attempt to represent a systematic account of a theory, such as I have made here, is bound to have defects. I will accept with gratitude all readers’ suggestions directed toward the improvement of this book.

Jacob Bear

Haifa, Israel

1972

CHAPTER 1

Introduction

1.1Aquifers, Ground Water and Oil Reservoirs

Flow through porous media is a topic encountered in many branches of engineering and science, e.g., ground water hydrology, reservoir engineering, soil science, soil mechanics and chemical engineering. Although our objective in this book was to present only the fundamental aspects of this topic, common to all these scientific and applied fields, we thought it appropriate to begin by presenting examples of porous media, and fluids in them, as encountered in practice. The aquifer, which is the porous medium domain treated by the ground water hydrologist, and the oil reservoir, which is the porous medium domain treated by the reservoir engineer, will serve as typical examples for this purpose. Following is a brief description of these domains and the fluids present in them.

1.1.1Definitions

An aquifer (a ground water basin) is a geologic formation, or a stratum, that (a) contains water, and (b) permits significant amounts of water to move through it under ordinary field conditions. Todd (1959) traces the term aquifer to its Latin origin: aqua, meaning water, and –fer from ferre, meaning to bear.

In contradistinction, an aquiclude is a formation that may contain water (even in appreciable amounts), but is incapable of transmitting significant quantities under ordinary field conditions. A clay layer is an example. For all practical purposes, an aquiclude is considered an impervious formation.

An aquitard is a semipervious geologic formation transmitting water at a very slow rate as compared to the aquifer. However, over a large (horizontal) area it may permit the passage of large amounts of water between adjacent aquifers, which it separates from each other. It is often referred to as a leaky formation. An aquifuge is an impervious formation that neither contains nor transmits water.

Ground water is a term used to denote all waters found beneath the ground surface. However, the ground water hydrologist, who is primarily concerned with the water contained in the zone of saturation (par. 1.1.2), uses the term ground water to denote water in this zone. In drainage of agricultural lands, or agronomy, the term ground water is used also to denote the water in the partially saturated layers above the water table. In this book we shall use the term ground water mainly in the sense employed by the ground water hydrologist.

That portion of rock not occupied by solid matter is the void space (also pore space, pores, interstices and fissures). This space contains water and/or air. Only connected interstices can act as elementary conduits within the formation. Figure 1.1.1 (after Meinzer 1942) shows several types of rock interstices. Interstices may range in size from huge limestone caverns to minute subcapillary openings in which water is held primarily by adhesive force. The interstices of a rock can be grouped in two classes: original interstices, mainly in sedimentary and igneous rocks, created by geologic processes at the time the rock was formed, and secondary interstices, mainly in the form of fissures, joints and solution passages, developed after the rock was formed.

FIG. 1.1.1.Diagram showing several types of Rock Interstices. A. Well-sorted sedimsntary deposit having high porosity; B. Poorly sorted sedimentary deposit having low porosity; C. Well-sorted sedimentary deposit consisting of pebbles that are themselves porous, so that the deposit as a whole has a very high porosity; D. Well-sorted sedimentary deposit whose porosity has been diminished by the deposition of mineral matter in the interstices; E. Rock rendered porous by solution; F. Rock rendered porous by fracturing. (After Meinzer, 1942.)

1.1.2The Moisture Distribution in a Vertical Profile

Subsurface water may be divided vertically into zones depending on the relative proportion of the pore space occupied by water: a zone of saturation, in which all pores are completely filled with water, and an overlying zone of aeration, in which the pores contain both gases (mainly air and water vapor) and water.

Figure 1.1.2 shows a schematic distribution of subsurface water. Water (e.g., from precipitation and/or irrigation) infiltrates the ground surface, moves downward, primarily under the influence of gravity, and accumulates, filling all the interconnected interstices of the rock formation above some impervious stratum. A zone of saturation is thus formed above this impervious bedrock. The saturated zone (fig. 1.1.2) is bounded above by a water table, or phreatic surface. This is a surface on which the pressure is atmospheric. It is revealed by the water level in a well penetrating the aquifer, in which the flow is essentially horizontal (sec. 6.6). Actually, saturation extends a certain distance above the water table, depending on the type of soil (par. 9.4.2). Wells, springs and streams are fed by water from the zone of saturation.

The zone of aeration extends from the water table to the ground surface. It usually consists of three subzones: the soil water zone (or belt of soil water), the intermediate zone (or vadose water zone) and the capillary zone (or capillary fringe).

FIG. 1.1.2The distribution of subsurface water.

The soil water zone is adjacent to the ground surface and extends downward through the root zone. The moisture distribution in this zone is affected by conditions at the ground surface: seasonal and diurnal fluctuations of precipitation, irrigation, air temperature and air humidity, and by the presence of a shallow water table. Water in this zone moves downward during infiltration (e.g., from precipitation, flooding of the ground surface or irrigation; sec. 9.4), and upward by evaporation and plant transpiration. Temporarily, during a short period of excessive infiltration, the soil in this zone may be almost completely saturated (gravitational water).

After an extended period of drainage without a supply of water at the soil surface, the amount of moisture remaining in the soil is called field capacity (par. 9.4.3). Below field capacity, the soil contains capillary water, in the form of continuous films around the soil particles, held by surface tension. This water is moved by capillary action and is available to plants. Below a moisture content—called the hygroscopic coefficient (maximum moisture that an initially dry soil will adsorb when brought in contact with an atmosphere of 50% relative humidity at 20°C)—the water in the soil is called hygroscopic water. Water is then unavailable to plants as it forms thin films of moisture adhering strongly to the surface of soil particles (fig. 9.4.2).

The intermediate zone extends from the lower edge of the soil water zone to the upper limit of the capillary zone. It does not exist when the water table is too high, in which case the capillary fringe may extend into the soil water zone, or even to the ground surface. Nonmoving, or pellicular, water in the intermediate zone is held in place by hygroscopic and capillary forces. Temporarily, water moves downward through this zone as gravitational water.

The capillary fringe extends upward from the water table. Its thickness depends on the soil properties and on the uniformity of pore sizes. The capillary rise ranges from practically nothing in coarse materials to as much as 2 ÷ 3 m and more in fine materials (e.g., clay). A detailed discussion on the capillary rise is given in paragraph 9.4.2. Within the capillary zone there is usually a gradual decrease in moisture content with height above the water table. Just above the water table, the pores are practically saturated. Moving higher, only the smaller connected pores contain water. Still higher, only the smallest connected pores are still filled with water. Hence, the upper limit of the capillary fringe has an irregular shape. For practical purposes some average, smooth surface is taken as the upper limit of the capillary fringe (par. 9.4.2), such that below it the soil is practically saturated (say, > 75%).

In the capillary fringe, the pressure is less than atmospheric and vertical as well as horizontal flow of water may take place. When the saturated zone below the water table is much thicker than the capillary fringe, the flow in the latter is often neglected. However, in most drainage problems the flow in the unsaturated zones may be of primary importance.

Obviously, numerous complications are introduced into the schematic moisture distribution described here by the great variability in pore sizes, the presence of permeability layers and by the temporary movement of infiltrating water.

1.1.3Classification of Aquifers

The following brief review of some geological formations that serve as aquifers is based on a work by Thomas (1952).

Most aquifers consist of unconsolidated or partly consolidated gravel and sand. They are located in abandoned or buried valleys, in plains and in intermontane valleys. Some are of a limited area; others may extend over large areas. Their thickness may also vary from several meters to several hundred meters.

Sandstone and conglomerate are the consolidated equivalent of sand and gravel. In these rocks the individual particles have been cemented together, thus reducing permeability.

Limestone formations, varying widely in thickness, density, porosity and permeability, serve as important aquifers in many parts of the world, especially when sizeable proportions of the original rock have been dissolved and removed. Openings in limestone may range from microscopic original pores to large fractures and caverns forming subterranean channels. By dissolving the rock along fractures and fissures the water tends to enlarge them, thus increasing permeability with time. Ultimately, a limestone terrane develops into a karst region. Its macroscopic behavior (i.e., on a large scale) is probably similar to that of a sand and gravel aquifer; on a smaller scale, this similarity is questionable (sec. 1.2). Figure 1.1.1 e and f show examples of rocks rendered porous by solution and fracturing.

Volcanic rocks may form permeable aquifers. Basalt flows are very permeable. The pore space of a basalt aquifer may not be as large as that of loose sands and gravels, but the permeability, owing to the cavernous character of the openings, may be many times greater. Most shallow intrusive rocks in the form of sills, dikes or plugs are low in permeability, and many of them are impervious enough to serve as barriers to ground water flow.

Crystalline and metamorphic rocks are relatively impervious and constitute poor aquifers. When such rocks occur near the ground surface, some permeability may develop by weathering and fracturing.

Clay and coarser materials mixed with clay, although in general having a high porosity, are relatively impervious owing to the small size of their pores.

Aquifers may be regarded as underground storage reservoirs that are replenished naturally by precipitation and influent streams, or through wells and other artificial recharge methods. Water leaves the aquifer naturally through springs or effluent streams and artificially through pumping wells.

The thickness and other vertical dimensions of an aquifer are usually much smaller than the horizontal lengths involved. Therefore, throughout this book, all drawings describing flow in aquifers are highly distorted. The reader should not be misled by the distorted scales of such figures.

Aquifers may be classed as unconfined or confined depending upon the presence or absence of a water table.

A confined aquifer (fig. 1.1.3), also known as a pressure aquifer, is one bounded above and below by impervious formations. In a well penetrating such an aquifer, the water level will rise above the base of the confining formation; it may or may not reach the ground surface. A properly constructed observation well (or a piezometer) has a relatively short screened section (yet not too short with respect to the size of the openings; see sec. 1.3) such that it indicates the piezometric head (sec. 3.3) at a specific point (say, the center of the screen). The water levels in a number of observation wells tapping a certain aquifer define an imaginary surface called the piezometric surface (or isopiestic surface). When the flow in the aquifer is essentially horizontal, such that equipotential surfaces are vertical, the depth of the piezometer opening is immaterial; otherwise, a different piezometric surface is obtained for piezometers that have openings at different elevations. Fortunately, except in the neighborhood of outlets such as partially penetrating wells or springs, the flow in aquifers is essentially horizontal.

FIG. 1.1.3.Types of aquifers.

An artesian aquifer is a confined aquifer (or a portion of it) where the elevations of the piezometric surface (say, corresponding to the base of the confining layer) are above ground surface. A well in such an aquifer will flow freely without pumping (artesian or flowing well). Sometimes the term artesian is used to denote a confined aquifer.

Water enters a confined aquifer through an area between confining strata that rise to the ground surface, or where an impervious stratum ends underground, rendering the aquifer unconfined. The region supplying water to a confined aquifer is called a recharge area.

A phreatic aquifer (also called unconfined aquifer or water table aquifer) is one with a water table (phreatic surface) serving as its upper boundary. Actually, above the phreatic surface is a capillary fringe, often neglected in ground water studies. A phreatic aquifer is recharged from the ground surface above it, except where impervious layers of limited horizontal area exist between the phreatic surface and the ground surface.

Aquifers, whether confined or unconfined, that can lose or gain water through either or both of the formations bounding them above and below, are called leaky aquifers. Although these bounding formations may have a relatively high resistance to the flow of water through them, over the large (horizontal) areas of contact involved significant quantities of water may leak through them into or out of a particular aquifer. The amount and direction of leakage is governed in each case by the difference in piezometric head that exists across the semipervious formation. Obviously, the decision in each particular case whether a certain stratum overlying an aquifer is an impervious formation, a semipervious one, or simply another pervious formation having a permeability that differs from that of the aquifer considered, is not a clear-cut one. Often, a layer that is considered semipervious (or leaky) is thin relative to the thickness of the main aquifer.

A phreatic aquifer (or part of it) that rests on a semipervious layer is a leaky phreatic aquifer. A confined aquifer (or part of it) that has at least one semipervious confining stratum is called a leaky confined aquifer.

Figure 1.1.3 shows several aquifers and observation wells. The upper phreatic aquifer is underlain by two confined ones. In the recharge area, aquifer B becomes phreatic. Portions of aquifers A, B and C are leaky, with the direction and rate of leakage determined by the elevation of the piezometric surfaces of each of these aquifers. The boundaries between the various confined and unconfined portions may vary with time as a result of changes in water table and piezometric head elevations. A special case of a phreatic aquifer is the perched aquifer (fig. 1.1.3) that occurs wherever an impervious (or relatively impervious) layer of limited horizontal area is located between the water table of a phreatic aquifer and the ground surface. Another ground water body is then built above this impervious layer. Clay or loam lenses in sedimentary deposits have shallow perched aquifers above them. Sometimes these aquifers exist only a relatively short time as they drain to the underlying phreatic aquifer.

1.1.4Properties of Aquifers

The general properties of an aquifer to transmit, store and yield water are further defined numerically through a number of aquifer parameters. A detailed analysis of these parameters is given throughout this book. Here we shall present a brief general description of some of them in order to supplement the definition of aquifers given above.

The hydraulic conductivity indicates the ability of the aquifer material to conduct water through it under hydraulic gradients. It is a combined property of the porous medium and the fluid flowing through it (sec. 5.5). When the flow in the aquifer is essentially horizontal, the aquifer transmissivity indicates the ability of the aquifer to transmit water through its entire thickness. It is the product of the hydraulic conductivity and the thickness of the aquifer (sec. 6.4).

The storativity of an aquifer (sometimes called the coefficient of storage) indicates the relationship between the changes in the quantity of water stored in an aquifer and the corresponding changes in the elevations of the piezometric surface (or the water table in an unconfined aquifer).

The storativity of a confined aquifer is defined as that volume of water released from (or added to) a vertical column of aquifer of unit horizontal cross-section, per unit of decline (or rise) of the piezometric head (secs. 6.3 and 6.4). Figure 1.1.4a illustrates this concept. The storativity of a confined aquifer is caused by the compressibility of the water and the elastic properties of the aquifer (or of the porous matrix) as a whole; the elasticity of the solid grains, particles, etc., is usually neglected. A detailed discussion is given in chapter 6.

FIG. 1.1.4.Illustrative sketches for defining storativity.

In a phreatic aquifer, the defination of storativity given above remains essentially unchanged, except that the decline is in the phreatic surface. However, a different mechanism causes the variation in the quantity of water stored in a column of aquifer. In the case of a phreatic aquifer, water is actually drained out of the pore space, and air is substituted as the water table drops. However, not all water contained in the pore space is removed by gravity drainage (say, toward a depression in the ground water table caused by a pumping well). A certain amount of water is held in place against gravity in the interstices between grains under molecular forces and surface tension. Hence, the storativity of a phreatic aquifer is less than the porosity by a factor called specific retention (the ratio of water retained against gravity to the bulk volume of a soil sample). Reflecting this phenomenon, the storativity of a phreatic aquifer is often referred to as specific yield. The term effective porosity is also often used in this context. However, one should be careful not to confuse this usage of the term with that effective porosity referring to flow through a porous medium (sec. 5.1). A detailed discussion of specific yield and specific retention is presented in section 9.4.3.

The elastic storativity resulting from compressibility of aquifer and water is much smaller than the specific yield. As an illustration, the storativity of most confined aquifers falls in the range between 10− 3 and 10− 5, whereas the specific yield of most alluvial aquifers falls in the range between 10 and 25%. This indicates that for the same volume of withdrawal (or recharge), changes in piezometric surface elevations are much larger in a confined than in a phreatic aquifer.

In defining storativity for a confined aquifer it is assumed that no time lag is involved and that water is released immediately upon change in head. However, especially in fine-grained materials, there may occur an appreciable time lag due to low hydraulic conductivity restricting movement of water out of storage. This is also true for a phreatic aquifer where the dewatering process takes time.

A parameter characterizing a leaky aquifer is the resistance of the semiconfining layer. It is defined by the ratio of the thickness of this layer to its hydraulic conductivity. As this value becomes larger, the leakage through this layer diminishes. Another parameter, the leakage factor, is the root of the product of the transmissivity of the aquifer and the resistance of the semipervious layer.

The various parameters mentioned above are used as guides in determining whether a certain geological formation is an aquifer, and of which type. These and other parameters, as well as methods for their determination, are discussed in detail in the following chapters.

1.1.5The Oil Reservoir

The oil or gas reservoir is a porous geologic formation that contains in its pore space, in addition to water, at least one hydrocarbon (oil or gas) in a liquid or gaseous phase. Throughout this book it will serve as another example of a porous medium of practical interest to which one may apply the theory of dynamics of fluids in porous media. Although this is by no means a text on reservoir or petroleum engineering, it seems appropriate, as in paragraph 1.1.1 in connection with the aquifer, to present some background comments related to the reservoir and to the production of oil and gas from it. These comments are based mainly on Muskat (1949) and Amyx et al. (1960).

Although several theories exist on the origin of hydrocarbons, scientists currently tend to accept the organic theory, according to which hydrocarbons evolve as the decomposition products of organic material (vegetable and animal) from organisms that lived during early geological ages. The beds, rich in organic material from which hydrocarbons originated, are called source beds. In these beds, the hydrocarbons are created in the form of a large number of tiny bubbles surrounded by the water that fills the pore space. From the source beds, the hydrocarbon droplets migrate to the reservoir rock where they accumulate in quantities of possible commercial interest.

The primary forces causing migration of hydrocarbons are buoyancy and capillarity. As oil and gas are lighter than the water surrounding them in the rock’s pore space, they will in general have an upward flow component (sec. 9.8). The buoyant forces lead to the separation of the gas, oil and water bodies in the pore space of the reservoir rock. To enable accumulation, the upward migration of hydrocarbons to higher beds in the stratigraphic sequence must be restricted by a blanket of impervious (or nearly impervious) material that serves as an upper boundary of the hydrocarbon-bearing rock. Thus, either a natural barrier that causes an abrupt change in the flow direction, or a trap that prevents outflow under prevailing conditions, must exist for hydrocarbon accumulation to take place. Clays, shales and general argillaceous rocks, such as sandy shales and marls, constitute the most common oil-confining strata. In a way, an oil reservoir is similar to a confined aquifer (par. 1.1.3).

Sometimes the sealing blanket is a water-saturated fine grained stratum with a much lower permeability than that of the reservoir rock itself. In this case, such a stratum serves as a barrier to upward seepage through capillary interfacial flow resistance at the point of contact between strata of different pore-size distributions (par. 9.2.5).

Figures 1.1.5 and 1.1.6 show several examples of elementary reservoir traps (after Wilhelm 1945) by a sectional view and by a view of contours of the structural environment. Some reservoirs are complex and result from a combination of two or more of the elementary trap features. One should recognize that a trap is a necessary but not a sufficient condition for hydrocarbon accumulation.

Most producing oil reservoirs are made of sandstone, limestone and dolomite formations. Occasionally other types of rocks also prove productive.

Within the oil reservoir, gravitational forces cause less dense fluids to seek higher positions in the trap. Capillary forces tend to cause the wetting fluid to rise into the interstices containing nonwetting fluid, thus counteracting the effect of gravity in segregating the fluids. In general, water is the wetting fluid with respect to oil and gas, while oil is the wetting fluid with respect to gas.

In fig. 1.1.7 typical hydrocarbon distributions in reservoirs are shown under equilibrium conditions, with transition zones between water and oil and between oil and gas (par. 9.2.3). Under certain conditions of pressure and temperature in the trap, no gas is present. The oil zone contains small amounts of water (connate water; par. 9.2.4). The fraction of pore space occupied by water increases with depth in the oil–water transition zone so that the base of this transition zone is defined by a completely water-saturated pore space.

FIG. 1.1.5.Elementary traps in sectional view (Wilhelm, 1945).

An oil (or gas) well field is comprised of a group of wells penetrating one or more subsurface reservoirs and producing oil (or gas) from these reservoirs. Several sources of energy exist that cause the hydrocarbons (oil or gas) in the reservoir to move toward the wells and to rise to the surface. The major types of energy available for oil and gas production (Muskat 1949) are: (a) energy of compression of oil, gas (in gas caps), and water within the reservoir; (b) gravitational energy of oil in the upper parts of the formation as compared with that at greater depths; (c) energy of compression and solution of the gas dissolved in the oil (and to some extent in the water); and (d) the energy of compression in the water in reservoirs (and aquifers) contiguous to and intercommunicating with the oil-bearing rock. Additional minor forms of energy that may control the reservoir performance should be mentioned for the sake of completeness. These are the differential energy of the internal surfaces of the porous rock for the different fluid phases, and the compression of the rock itself (Muskat 1949, p. 365).

FIG. 1.1.6.Contours of structural environment (Wilhelm, 1945).

As the wells provide outlets for these forms of energy, the energy is expended by the action of forces or pressures exerted in the direction of lower energy levels. These forces serve to overcome the resistance of the rock to the fluid’s flow toward the wells.

Among the four major energy sources, the first is less significant than the others.

Some reservoirs are closed, so that the associated volume of water is quite small and no gas cap or associated active water are present. In this case the energy available to displace the hydrocarbons toward the wells is solely that resulting from liberation of gas from solution in the reservoir oil, with subsequent expansion and expulsion of the oil. This is the third type of energy source listed above. The reservoir in this case is referred to as a solution-gas drive reservoir, or a depletion-drive reservoir. This type of drive is characterized by rapid pressure decline and low efficiency of oil recovery. It usually results in recovery of only a small fraction of the total oil volume present in the formation.

When the source of energy in a reservoir is an expanding free gas cap, but no associated active water body is present, the driving mechanism is referred to as a gas-cap-drive (or sometimes gas-cap-expansion or external gas drive). This is basically a displacement-type drive, the gas displacing the oil ahead as it expands because of pressure reduction.

FIG. 1.1.7.Typical hydrocarbon distribution in reservoirs (Amyx et al., 1960).

The second type of energy source listed above, gravitational, does not, in general, play an important role as a driving mechanism until the reservoir becomes substantially depleted. In some high-relief reservoirs, where the producing wells are located structurally low, oil recovery by gravitational segregation may be rather substantial.

A petroleum reservoir associated with aquifers so active that little or no pressure drop occurs in the petroleum reservoir by the withdrawal of hydrocarbon fluids is referred to as a water-drive reservoir. In the water-drive process, water from surrounding aquifers enters the reservoir almost as fast as the hydrocarbon fluid is withdrawn, thus preventing any substantial decline in pressure. This displacement-type drive is by far the most efficient natural reservoir driving mechanism. The efficiency of water displacement is usually greater than that of gas displacement, regardless of the wetting characteristics of the reservoir rock. When the reservoir formation has a steep dip, the oil–water interface will be of limited area and will provide an edgewater boundary. We then have an edgewater drive (e.g., fig. 9.5.8f). For gently sloping formations, the water–oil interface may underlie an appreciable part of the oil zone. Wells producing above such a water body are subject to a bottom-water-drive (fig. 9.7.10) if the water is mobile and is permitted to invade the oil reservoir at a rate sufficient to replace the withdrawn fluid.

Most petroleum reservoirs are subject to one or more drives, either simultaneously, or at various times during their productive life. Such reservoirs are referred to as combination-drive reservoirs.

In addition to the natural drives described above, artificial supplementation of natural energy by fluid injection (through wells) may be effected to improve the production efficiency of hydrocarbons from reservoirs. The fluid injection may involve the return of gas, water (water flooding) or gas and water. Such operations are often termed pressure maintenance operations. A more general term, which includes such techniques as injection of miscible (with oil) fluids, thermal recovery methods (involving the injection of hot water or steam into the reservoir) and internal combustion methods, is secondary recovery operations. The basic difference between secondary recovery and pressure maintenance operations is that the initial condition for the former is a state of virtually complete depletion of the reservoir pressure or the original natural oil-expulsion energy, whereas in the latter fluid injection is undertaken during the primary oil producing processes before such states are reached.

Details of the various flooding techniques are given in all texts on reservoir engineering. The theory underlying water flooding and gas injection is considered in sections 9.3 and 9.5 and that underlying miscible flooding techniques is presented in chapter 10.

1.2The Porous Medium

Two examples of practical interest—the aquifer encountered in ground water hydrology, and the oil reservoir encountered in petroleum engineering—were introduced in section 1.1. In both cases, flow takes place through a porous medium. Actually, by introducing the term porous medium, and by considering flow of fluids through a porous medium, we have made a basic and important step in our way of thinking. We have introduced the concept of a continuum, which is common to most branches of physics.

The continuum approach is discussed in detail in the following section. Here we shall attempt to define, or at least to describe in relative terms, a porous medium. Examples of porous materials are numerous. Soil, porous or fissured rocks, ceramics, fibrous aggregates, filter paper, sand filters and a loaf of bread are just a few. Somewhat less obvious examples, but still part of this group, are large geologic formations of karstic limestone, where the open passages (such as solution channels or caverns) may be of substantial size and far apart. All of these materials have some characteristics in common that permit them to be grouped and classified as porous media.

Initially, we may attempt to describe a porous medium as a solid with holes. Obviously, a hollow metal cylinder would not normally be classed as a porous medium, nor would a solid block with isolated holes or pores, since we seek to define a porous medium in connection with flow through the medium, and not, for example, in connection with thermal insulation. We might try to improve our definition by stipulating that the pores are interconnected, with at least several continuous paths from one side of the medium to the other, and by somehow specifying a better distribution (in either a regular or random manner) of holes and paths over the entire porous medium domain. Although we are now describing an acceptable porous medium model, the medium described lacks the possibility of exchange of fluid between adjacent paths, and especially those aspects related to three-dimensional flow, which requires a spatial distribution of channels (or paths) within the domain.

Summarizing these preliminary remarks, and looking at figure 1.1.1, we may try to define a porous medium as (Bear, Zaslavsky and Irmay 1968):

(a) a portion of space occupied by heterogeneous or multiphase matter. At least one of the phases comprising this matter is not solid. They may be gaseous and/or liquid phases. The solid phase is called the solid matrix. That space within the porous medium domain that is not part of the solid matrix is referred to as void space (or pore space).

(b) The solid phase should be distributed throughout the porous medium within the domain occupied by a porous medium; solid must be present inside each representative elementary volume (par. 1.3.2). An essential characteristic of a porous medium is that the specific surface (sec. 2.6) of the solid matrix is relatively high. In many respects, this characteristic dictates the behavior of fluids in porous media. Another basic feature of a porous medium is that the various openings comprising the void space are relatively narrow.

(c) At least some of the pores comprising the void space should be interconnected.

The interconnected pore space is sometimes termed the effective pore space. As far as flow through porous media is concerned, unconnected pores may be considered as part of the solid matrix. Certain portions of the interconnected pore space may, in fact, also be ineffective as far as flow through the medium is concerned. For example, pores may be dead-end pores (or blind pores), i.e., pores or channels with only a narrow single connection to the interconnected pore space, so that almost no flow occurs through them. Another way to define this porous medium characteristic is by requiring that any two points within the effective pore space may be connected by a curve that lies completely within it. Moreover, except for special cases, any two such points may be connected by many curves with an arbitrary maximal distance between any two of them. For a finite porous medium domain, this maximal distance is dictated by the domain’s dimensions.

The features described above can scarcely be called a definition as they involve several terms, such as relatively small, more or less evenly distributed, and relatively narrow, that are comparative, rather than absolute. However, in combination, these terms convey to the reader something of the nature of a porous medium. Rideal (1958) also tries to indicate the various characteristics of porous materials, emphasizing the difficulty in arriving at an exact definition still sufficiently general to be applied to the wide variety of porous media. To some of the features described, numerical values may be assigned. To others, mainly those related to the geometry of the solid surfaces, no such values may be assigned. In fact, it is this difficulty in defining the geometry of the solid surfaces, which act as boundaries to the flow in the void space, that forces us to introduce the continuum approach as a tool for handling phenomena in porous media.

1.3The Continuum Approach to Porous Media

1.3.1The Molecular and Microscopic Levels

The main purpose of section 1.2 is to demonstrate the hopelessness of any attempt to describe in an exact manner the geometry of the internal solid surfaces that bound the flow domain inside a porous medium. Directing our attention to the fluid or fluids contained in the void space, and trying to describe phenomena associated with them, such as motion, mass transport, etc., the same difficulties are encountered.

First, the concept of the fluid itself requires some further elaboration. Actually, fluids are composed of a large number of molecules (overlooking the existence of a submolecular structure) that move about, colliding with each other and with the solid walls of the container in which they are placed. By employing theories of classical mechanics, we could fully describe a given system of molecules: e.g., given their initial positions in space and their momenta, we could predict their future positions. However, despite the apparent simplicity of this approach, it is exceedingly difficult to solve the problem of the motion of even three molecules (assuming that we know all the forces, which is also doubtful). With the advent of high speed digital computers, the many body problem can be attacked, in principle, numerically. It is still impossible, however, to determine the motion of 10²³ molecules in one gram mole of gas. In addition, because the number of molecules is so large, their initial positions and momenta cannot actually be determined, for example, by observation.

It is the embarrassingly large number of equations that ultimately provides a way out, at least under certain conditions. Instead of treating the problems, say of fluid motion, at the molecular level or viewpoint described above, we may adopt a different approach, statistical in nature, to derive information regarding the motion of a system composed of many molecules. By the statistical approach we mean one in which the results of an analysis or an experiment are presented only in statistical form. This means that we can determine the average value of successive measurements, but we cannot predict with certainty the outcome of a single measurement in the future. In this context, an experiment means, for example, the position of a certain molecule at a certain time. Statistical mechanics is an analytical science by which statistical properties of the motion of a very large number of molecules (or of particles in general) may be inferred from laws governing the motion of individual molecules. Many texts are available on this subject, and the reader is referred to them (e.g., Landau and Lifschitz 1958).

When the purpose of abandoning the molecular level of treatment is the description of phenomena as a fluid continuum, the statistical approach is referred to as the macroscopic approach. Our ultimate objective is to handle phenomena in porous media. We shall need for that purpose a still higher, or coarser, level of treatment, reached by averaging phenomena in the fluid continuum filling the void space. We choose, therefore, to refer to the fluid continuum level as the microscopic one. At the microscopic level, we overlook the actual or molecular structure of a fluid and regard it as a continuum.

Essential to the treatment of fluids as continua is the concept of a particle. A particle is an ensemble of many molecules contained in a small volume. Its size is much larger than the mean free path of a single molecule. It should, however, be sufficiently small as compared to the considered fluid domain that by averaging fluid and flow properties over the molecules included in it, meaningful values, i.e., values relevant to the description of bulk fluid properties, will be obtained. These values are then related to some centroid (par. 4.1.1) of the particle. Then, at every point in the domain occupied by a fluid, we have a particle possessing definite dynamic and kinematic properties.

Associated with the question of particle size, or of the elementary volume that should be considered as a point—a physical or material point—within the fluid continuum, is the definition of the fluid’s density or specific mass.

Density is the ratio between the mass Δm of an amount of matter and the volume ΔU occupied by it. If we consider a mathematical point and wish to assign to it a density value, ρ, such that this value will represent the density of a volume of fluid for which this point is a mass centroid, we have to decide on the volume to use.

Following Prandtl and Tietjens (1934), let us consider a point P in the fluid, and let Δmi denote the fluid mass in a sufficiently large volume, ΔUi for which P is a centroid. The average density, ρi of the fluid in ΔUi is ρi = Δmi/ΔUi Obviously, if ΔUi is too large, say of the order of magnitude of the entire field of flow, it is meaningless to assign the value ρi to the point P, i.e., to represent the ratio Δmi/ΔUi for the fluid in the vicinity of P. This is especially true when the fluid is inhomogeneous. To determine how small ΔUi should be in order for ρi to represent the fluid in the neighborhood of P, we gradually reduce ΔUi around P, determining the ratio Δmi/ΔUi for a sequence of volumes δUi : ΔU/1 > ΔU2 > ΔU3 ... The results of these computations are shown in figure 1.3.1. If we start from a sufficiently large ΔUi gradual changes may be observed in ρi if the fluid is inhomogeneous. Fluctuations around ρi = ρi(ΔUi) diminish as ΔUi becomes smaller. Then as ΔUi converges on P, a range of practically no changes in ρi with changes in ΔUi is observed. However, as ΔUi is made smaller and the number of molecules in it becomes smaller too, a range is reached below a certain volume ΔU0 where the number of molecules in ΔUi is so small that any further reduction of ΔUi appreciably affects the ratio Δmi/ΔUi This happens when the characteristic length dimension of ΔUi becomes of the order of magnitude of the average distance λ between the molecules (mean free path of molecules). As ΔUi → 0 very wild fluctuations in the ratio Δmi/ΔUi are observed, and it is meaningless to use ρi as a definition for the density of the fluid at P. Hence the fluid’s density at P is defined as:

The characteristic volume ΔU0 is called the physical point (or material point) of the fluid at the mathematical point P. The volume ΔU0 may now be identified with the volume of a particle at P. By the procedure just described, a material made of a collection of molecules in a vacuum is replaced by a continuum filling the entire space. We obtain a fictitious smooth medium (instead of the molecules), called fluid, for each point of which a continuous function of space ρ is defined. For any two close points P and P′:

In an inhomogeneous fluid, it is interesting to define a characteristic length L:

or lengths, say, in the direction of the three coordinates:

Lx = ρ/(∂ρ / ∂x);Ly = ρ/(∂ρ / ∂y);Lz = ρ/(∂ρ / ∂z)

that characterize the macroscopic changes in ρ in the field of flow; it indicates how rapid these changes in ρ are. The volume L³ (or LxLyLz) may be used as the upper limit for the range of ΔUi for which ρi is independent of ΔUi (fig. 1.3.1). One should note that in (1.3.2):

where λ < Δl0 < L, is a characteristic length of the fluid at P, with (Δl0)³ of the order of ΔU0.

Obviously, when the fluid is a gas at very low pressure, λ may be very large and Δl0 and ΔU0 may be as well.

FIG. 1.3.1.Defination of fluid density

Knudsen (1934) defines a dimensionless number Kn = λ/L (called a Knudsen number). When Kn < 0.01, the flowing fluid may be considered as a continuum to which macroscopic considerations (e.g., in the form of partial differential equations) are applicable. This then sets the limits for δl0 or ΔU0. When Kn ∼ 1 we have a slip-flow regime (par. 5.3.3) and when Kn > 1 we have Knudsen flow or "free molecular flow".

A comment seems appropriate regarding time involved in the process described above, because of the continuous rapid motion of the molecules in and out of any volume around P. One should consider figure 1.3.1 as representing what happens at a specified (or mathematical) instant of time around point P, or should regard each point of the curve as an average of several observations taken over an interval of time, Δt0, around the considered instant of time. To determine Δt0 we must go through arguments similar to those involved in the determination of ΔU0. The interval of time should not be too large (as we should then lose information regarding temporal variations of ρ, if such variations take place), i.e., it should be smaller than a characteristic time, T, defined by T = ρ/(∂ρ/∂t). On the other hand, it should not be shorter than a mean free time (Landau and Lifschitz 1958) of a molecule (an interval of time during which each molecule collides once on the average with another molecule). The value of δt0 may be obtained from a curve similar to that shown in figure 1.3.1, but with Δti as abscissa and Δmi as ordinate.

Many other physical phenomena in fluids, observed through their macroscopic manifestations, are the outcome of perpetual molecular motion. Among these we have mass transport by molecular diffusion, heat transfer and momentum transfer, which manifests itself in the form of internal friction or viscosity. In each of these cases, because we are unable to treat transfer phenomena on a molecular level, we average the transfer produced by the individual molecules and pass to a higher level—that of a fluid continuum, referred to in the present text as the microscopic level. In order to describe the various transfer phenomena at this higher level, transfer coefficients are needed. They are molecular diffusivity, thermal diffusivity, kinematic viscosity, etc. Occasionally, to understand phenomena, or the meaning of the microscopic parameters described and their relation to basic molecular properties, it may prove instructive to revert back to the molecular point of view.

To conclude, our discussion has led us from the molecular level to the microscopic level of treating physical phenomena. We now have a fluid continuum enclosed by solid surfaces—the solid surfaces of the porous medium. At each point of this fluid continuum we may define the specific physical, dynamic and kinematic properties of the fluid particle. Can we, however, solve a flow problem in a porous medium at this level? In principle we have at our disposal the theory of fluid mechanics, so that we may derive the details of a fluid’s behavior within the void space. For example, we may use the Navier–Stokes equations for the flow of a viscous fluid to determine the velocity distribution of the fluid in the void space, satisfying specified boundary conditions, say, of vanishing velocity, on all fluid–solid interfaces. However, as we have already shown, it is practically impossible, except in especially simple cases, such as a medium composed of straight capillary tubes, to describe in any exact mathematical manner the complicated geometry of the solid surfaces that bound the flow domain within the porous solid matrix. Moreover, it is often difficult to define the boundary conditions themselves. Consequently, any solution by means of this approach is precluded. In view of the discussion presented above, the obvious way to circumvent these difficulties is to pass to a coarser level of averaging—to the macroscopic level. This is again a continuum approach, but on a higher level.

The following paragraph deals with the macroscopic approach to dynamics of fluids in porous media.

1.3.2Porosity and Representative Elementary Volume

In paragraph 1.3.1 we have seen that essential to the concept of a continuum is the particle, or the physical point, or the representative volume over which an average is performed. This is also true in the passage from the microscopic to the macroscopic level. Our task now is, therefore, to determine the size of the representative porous medium volume around a point P within it. From the discussion in the previous paragraph we know that this volume should be much smaller than the size of the entire flow domain, as otherwise the resulting average cannot represent what happens at P. On the other hand, it must be enough larger than the size of a single pore that it includes a sufficient number of pores to permit the meaningful statistical average required in the continuum concept.

When the medium is inhomogeneous, say, with porosity varying in space, the upper limit of the length dimension of the representative volume should be a characteristic length that indicates, as in (1.3.2), the rate at which changes in porosity take place. The lower limit is related to the size of the pores or grains.

We shall now define volumetric porosity and the representative elementary volume (REV) associated with it, following the same procedure as that described in paragraph 1.3.1 for the definition of density. We choose to define the REV through the concept of porosity, which seems to be the basic porous matrix property. Porosity is, in a sense, the equivalent of density (mass per unit volume) discussed in paragraph 1.3.1. This might be more easily visualized if we defined the REV through the concept of the solid’s bulk density (unit weight of solids = mass of solid per unit volume of medium).

Let P be a mathematical point inside the domain occupied by the porous medium. Consider a volume ΔUi (say, having the shape of a sphere) much larger than a single pore or grain, for which P is the centroid. For this volume we may determine the ratio:

where (ΔUν)i is the volume of void space within ΔUi. Repeating the same procedure, a sequence of values ni(ΔUi), i = 1, 2, 3, ... may be obtained by gradually shrinking the size of ΔUi around P as a centroid: ΔU1 > ΔU2 > ΔU3 ....

For large values of ΔUi the ratio ni may undergo gradual changes as ΔUi is reduced, especially when the considered domain is inhomogeneous (e.g., layers of soil). Below a certain value of ΔUi, depending on the distance of P from boundaries of inhomogeneity, these changes or fluctuations tend to decay, leaving only small-amplitude fluctuations that are due to the random distribution of pore sizes in the neighborhood of P. However, below a certain value ΔU0 we suddenly observe large fluctuations in the ratio ni. This happens as