PGE 381L ADVANCED PETROPHYSICS FALL 2001 By Dr. Ekwere J. Peters Professor & Chairman Department of Petroleum and Geosystems Engineering The University of Texas at Austin Austin, Texas 78712 USAAdvanced Petrophysics (PGE 3811) Guidelines for Preparing Homework These are minimum requirements. 1. Organization. Present your work in sequence starting with Question 1, 2, etc., regardless of the order in which you actually solved the problems. Clearly label each problem including subproblems such a, b, ¢, etc., as appropriate. Attempt all parts of a given question. Pay attention to details. Ignoring a part of the question does not make it go away. 2. Appearance Your homework should be neat. You don't have to type it, but your writing should be legible. You can use computer software such as a spreadsheet program to solve the homework problems if it 1s convenient and appropriate to do so. The homework should be stapled at the top left corner. 3, Graphs and Figures Graphs, plotted manually, should be plotted on graph paper and not on plain paper or engineering paper. Plotting graphs with computer software 1s encouraged. In this case, the graph should be printed on plain paper. ‘Axes of graphs should be clearly labelled. Dependent variable should be on the ordinate (vertical axis) and the independent variable should be on the abscissa (horizontal axis),All figures and graphs should be labelled with figure numbers and figure captions. The figures and graphs should be referred to in your homework presentation. 3. Discussion Answers to discussion type questions should meet the usaual requirements of good technical writing. This means using full sentences, good grammar and correct spelling. If you have difficulty in technical ' writing in English, you should consider taking the undergraduate technical writing course (PGE 333T) even if it is not required in your degree plan. To get a graduate degree, you will have to write a thesis, a dissertation or a report. So you can't avoid technical writing in graduate school. You might as well learn to do it correctly. Avoid one word answers without explanation or justification. For example, does this method of porosity measurement lead to effective or total porosity? An answer that merely says “total porosity” simply because the class notes says so is useless. You must present arguments and justification for why you think the method gives total porosity. You need to convince me that you know what you are talking about. When asked to design an experiment to determine a certain property, you should design your experiment in such a way that your design can be implement by a competent technician to determine the property without any further input from you. 4. Strategy for Doing Homework You will normally have one week to do each homework assignment. Start early. Don't leave it until the day before it is due. If you do, you can look forward to alot of frustration and unhappiness because you will find that the homework problems cannot be done overnight.’TABLE OF CONTENTS 1 PETROLEUM RESERVOIR ROCKS... 1.1 INTRODUCTION..... 1.2. MINERAL CONSTITUENTS OF ROCKS—A REVIEW. 1.3 ROCKS... 13.1 Igneous 132 Metamorphic Rocks, 14 133 Sedimentary Rocks. 1.4 CLASSIFICATION OF SEDIMENTARY ROCKS. 14.11 Clastic Sedimentary Rocks.. 1.42 Chemical Sedimentary Rocks. 143 Organic Sedimentary Rocks. 1.5 DISTRIBUTION OF SEDIMENTARY ROCK TYPES. 1.6 SANDSTONE RESERVOIRS (CLASTIC SEDIMENTARY ROCK), 1.7 CARBONATE RESERVOIRS (LIMESTONES AND DOLOMITES). 1.7.1 Classification, 1.7.2 Pore Space.. 1.8 FRACTURED RESERVOIRS. 19 PETROPHYSICS... 2 SINGLE PHASE FLUID STORAGE AND TRANSPORT 2.1 POROSITY. 2.1.1 Factors Affecting Sandstone Porosit 2.13 Typical Reservoir Porosity Values 214 Direct Laboratory Measurement of Porosity 21.5 Indirect Porosity Measurement from Well Logs. 2.1.6 Porosity Distribution.23 24 3. MULTI-PHASE ROCK/FLUID PROPERTIES... 3.1 32 33 34 PERMEABILITY... 23.1 Dimensions and Unit of Permeability 2-26 232 Laboratory Determination of Permeability 2-27 233 Field Determination of Permeability. 2-30 23.4 Non-Darcy Flow.. 231 235 Factors Affecting 2-32 236 Typical Reservoir Permeability Valu 236 ty 237 Permeability-Porosity Correlations. 238 Capillary Tube Models of Porous Media.. 239 Steady State Flow Through Fractures. 23.10 Averaging Permeability Data 23.11 Permeability Distributio: 23.12 Measures of Permeability Heterogeneity: 23.13 Modeling Permeability Heterogeneity, DISPERSIVITY FLUID SATURATIONS. SURFACE AND INTERFACIAL TENSIONS. 3.2.1 Surface Tension. 3.22 Interfacial Tensio1 3.23 Measurements of Surface and Interfacial Tensions. WETTABILITY. 3.3.1 Determination of Wettability. 3-40 3.3.2 Wettability of Petroleum Reservoir: 3-47 3.33. Effect of Wettability on Rock-Fluid Interactions.........3-48 CAPILLARITY AND CAPILLARY PRESSURE. 34.1 Drainage Capillary Pressure Curve for a Water Wet Mediu: 34.2. Conversions of Laboratory Capillary Pressure Data to Reservoir Conditions 343 Averaging Capillary Pressure Data . 3.44 Determination of Initial Static Reservoir Fluid Saturations by Use of Drainage Capillary Pressure Curve. 345 Capillary Pressure Hysteresis. 346 Capillary Imbibition.. 347 Capillary End Effect in a Laboratory Core 34.8 Capillary Pressure Measurements. 3.49 Pore Size Distribution. 34.10 Estimation of Permeability from Capillary Pressure.3-10535 ee AND RELATIVE PERMEABILITIES.... .5.1 Laboratory Measurement of Two-Phase Relative Permeabilities by the Steady State Method.. 35.2 Theory of One Dimensional Immiscible Displacement in a Porous Medium 353 Laboratory Measurement of Two-Phase Relative Permeabilities by the Unsteady State Method evn 3.5.4 One Dimensional Immiscible Displacement with Capillarity. 3.5.5 Factors Affecting Relative Permeabilities. 3.5.6 Three-Phase Relative Permeabilities..1. PETROLEUM RESERVOIR ROCKS 1.1 INTRODUCTION A petroleum reservoir rock is a porous medium that is sufficiently permeable to permit fluid flow through it. In the presence of interconnected fluid phases of different density and viscosity, such as water and hydrocarbons, the movement of the fluids is influenced by gravity and capillary forces. The fluids separate, therefore, in order of density when flow through a permeable stratum is arrested by a zone of low permeability, and in time, a petroleum reservoir is formed in such a trap. Porosity and permeability are two fundamental petrophysical properties of petroleum reservoir rocks. Geologically, a petroleum reservoir is a complex of porous and permeable rock which contains an accumulation of hydrocarbons under a set of geological conditions that prevents escape by gravitational and capillary forces. The initial fluid distribution in the reservoir rock, which is determined by the balance of gravitational and capillary forces, is of significant interest at time of discovery. A rock capable of producing oil, gas and water is called a reservoir rock. In general, to be of commercial value, a reservoir rock must have sufficient thickness, areal extent and pore space to contain a large volume of hydrocarbons and must yield the contained fluids at a satisfactory rate when the reservoir is penetrated by a well. Any buried rock, be it sedimentary, igneous or metamorphic, that meets these conditions may be used as a reservoir rock by migrating hydrocarbons. However, ‘most reservoir rocks are sedimentary rocks. cee Sandstones and carbonates (limestones and dolomites) are the most common reservoir rocks. They contain most of the world’s1-2 petroleum reserves in about equal proportions even though carbonates make up only about 25% of sedimentary rocks. The reservoir character of a rock may be primary such as the intergranular porosity of a sandstone, or secondary, resulting from chemical or physical changes such as dolomitization, solution and fracturing. Shales frequently form the impermeable cap rocks for petroleum traps. The distribution of reservoirs and the trend.of pore space are the end product of numerous natural processes, some depositional and some post-depositional. Their prediction, and the explanation and prediction of their performance, involves the recognition of the genesis of the ancient sediments, the interpretation of which depends upon an understanding of the sedimentary and diagenetic processes. Diagenesis is the processes of physical and chemical changes in sediments after deposition that converts them to consolidated rock such as compaction, cementation, recrystallization and perhaps replacement as in the development of dolomite. 1.2 MINERAL CONSTITUENTS OF ROCKS - A REVIEW The physical and chemical properties of rocks are the consequence of their mineral composition. A mineral is a naturally occurring crystalline inorganic material that has specific physical and chemical properties which are either constant or vary within certain limits. Rock-forming minerals of interest in petroleum engineering can be classified into the following families: silicates, carbonates, oxides, sulfates (sulphates), sulfides (sulphides) and chorides. These are summarized in Table 1.1.1-3 Table 1.1 Rock - Forming Minerals Name Chemical Formula Specific Gravity Silicates Quartz siO2 2.65 Orthoclase KAISi208 2.57 Plagioclase NaAlSi30g 2.62 - 2.76 CaAl2Si208 Clay Al28i205(OH) 25 and many more Carbonates Calcite CaCO3 2.72 Dolomite CaMg(CO3)2 2.85 Oxides Magnetite Fe304 5.18 Hematite Fe203, 49-53 Sulfates Anhydrite CaSO4 2.89 - 2.98 Gypsum CaSO4.2H20 2.32 Barite BaSO4 45 Sulfide Pyrite FeS2 5.02 Chloride Halite NaCl 2.16 Silicates are the most abundant rock-forming minerals on the earth’s crust.1-4 1.3 ROCKS A rock is an aggregate of one or more minerals. There are three classes of rocks: igneous rocks, metamorphic rocks and sedimentary rocks . 13.1 Igneous Rocks These are rocks formed from molten material (called magma) that solidified upon cooling either: 1. At the earth’s surface to form volcanic or extrusive rocks, e.g., basaltic lava flows, volcanic glass and volcanic ash. or 2. Below the surface, usually at great depths, to form plutonic or intrusive rocks, e.g., granites and gabbros. Igneous rocks are the most abundant rocks on the earth’s crust, making up about 64.7% of the earth’s crust. 1.3.2 Metamorphic Rocks These are rocks formed by transformation, generally in the solid state, of pre-existing rocks beneath the surface by heat, pressure and chemically active fluids, e.g., quartz is transformed to quartzite and limestone plus quartz plus heat gives marble and carbon dioxide. Metamorphic rocks are the second most abundant rocks on the earth’s crust, making up 27.4% of the earth’s crust.1.3.3 Sedimentary Rocks These are rocks formed at the surface of the earth either by 1. Accumulation and consolidation of minerals, rocks and/or organisms and vegetation, e.g., sandstone and limestone. or 2. Precipitation from solution such as sea water or surface water, e.g., Salt and limestone. Sedimentary rocks are the source of petroleum and provide the reservoir rock and trap to hold the petroleum in the earth’s crust. Sedimentary rocks are the least abundant rocks on the earth’s crust, making up about 7.9% of the earth’s crust. Because most reservoir rocks are sedimentary rocks, they are of particular interest to us in the study of petrophysics. Therefore, we will examine sedimentary rocks in more detail than igneous and metamorphic rocks. 1.4 CLASSIFICATION OF SEDIMENTARY ROCKS Sedimentary rocks may be classified by origin and composition as clastic, chemical or organic. Tables 1.2, 1.3 and 1.4 show the various rock types for each class. 1.4.1 Clastic Sedimentary Rocks These rocks are composed of fragments or minerals broken from any type of pre-existing rock. Clastic sedimentary rocks are usually classified by grain size as boulder, cobble, gravel, sand, silt and clay. Figure 1.1 shows such a classification known as the Wentworth scale.Table 1.2 Clastic Sedimentary Rocks ROCK NAME COMPONENTS _ [ORIGINAL UNCONSOLIDATED DEBRIS | GRAIN SIZE | ROUNDESS| SORTING ‘onglomerate| rock & mineral ‘gravel 30%>2mm | wellrounded| poor fragments Breccia tock fragments rubble 30%>2mm | angular none Sandstone: Quortzite] quartz grains ‘quartz sand 2mm 31/16 mm | welounded| —_ good Arkose | quartz, 25% feldspar} feldspar-rich sand eroded from a_| 2mm 1/16 mm | subrounded | moderate to granitic source good) =" aie Graywacke, Tegra nme ock aya eebeienae a a il rm | subangular Poor Sitstone ‘quariz, clay mud, silt S1é mm | Wolvsibsin specimen. Shale’ Clay, quartz mud, clay. = 17256em | Not visible in : specimen. eeTable 1.3 Chemical Sedimentary Rocks (Precipitates) ROCK NAME | COMPONENTS DESCRIPTION [Uimestone: Lithographic CoCO3 Fine grained, dense limestone. used in making lithographic prints. ‘Marl CaCOg Fine grained, may be fresh water sediment. Travertine] _CoCO3+ Clay _ | Stoloctites, stalagmites, flow stone, etc. Tufo| ‘Caco3 Formed around springs. Oolitic CaCo3 Tiny spherical concretions formed in shallow marine waters. Gypsum Cg8042H20 _[Evaporite mineral, formed from sea woter as well as ground water. ‘Anhydhrite ‘C0804 Dehydrated gypsum. Halite NaCl Evoporite mineral. Syivite KCI Evaporite mineral. Dolomite CaMg(CO3)2 ‘Common constituent of carbonate rocks. Formed by the alteration of limestone by Mgt* rich waters. “Amorphous Siiea $i02+H20 | Opal, chert, chalcedony, flint, agate, jasper, etc. Limonite Fe oxide + H20___[Rust. nota tue mineral, : Goethite Fe O(OW) Similor to limmonite. : Hematite Fe2CO3 Anhydrous lion oxide. Siderite Fe CO3 ‘Often forms concretions.Table 1.4 Organic Sedimentary Rocks COMPONENTS Calcite (CaCO3) Calcite (CaCO3) Calcite (CaCO3) Calcite (CaCO3) Peat and Coal ‘Organic Plant Matericl Diatomite, Chert NATURE OF SEDIMENT Large shell fragments. Limestone containing fossis. Fine grained: skeletons of microscopic marine plonts and animals. Peat = Lignite + Bitumin + Anthracite. Grade depends on carbon content resulting from varying degrees of metamorphorsis. eLVERY COARSE COARSE MEDIUM COARSE MEDIUM 256,000 4.2564 -8 128,000-4.128+ -7 100,000-4-100 64,000-} 644-6 32,0004 924-5 + 16,0004 164-4 + e000 oben f 190mg | 4000+ 44-24 # 2,000 26-1 8 1.0004 14 O04 400g 1 -E 2: 5004 124 1 3 250-4 Wat 2+ s 1254 By 3 8 to0--0.1 Ei: 6250 71/t6t 4 8: 3125 713 sy 15.62 frst 6 tog. 7812 ae 3.908 8 1.953 fu512 9 0.9765 -1/1024- 10: 10.001 vet units of inch Figure 1.1: Classification of clastic rocks according to texture.1-10 1.4.2 Chemical Sedimentary Rocks These rocks are formed by chemical precipitation as carbonates, e.g., limestone (CaCO3) and dolomite (CaMg(CO3)2) or as evaporties, e.g., gypsum (CaSO4.2H20), anhydrite (CaSO4) and salt (NaCl). 1.4.3 Organic Sedimentary Rocks These rocks are formed by biologic precipitation and by accumulation of organic (plant and animal) material, e.g., peat, coal, diatomite and limestone. 1.5 DISTRIBUTION OF SEDIMENTARY ROCK TYPES Table 1.5 shows the approximate distribution of sedimentary rocks in the earth’s crust. Shales make up over 50% of total sedimentary rock volume in the earth’s crust. Table 1.5 Distribution of Sedimentary Rocks Type % Earth’s Crust % Sedimentary Rock Shale 42 53 Sandstone L7 22 Limestone and Dolomite 2.0 25 Total 79 1001-11 1.6 SANDSTONE RESERVOIRS (CLASTIC SEDIMENTARY ROCK) Sandstones are composed of fragmented materials which have been transported to the site of deposition by water currents and which have been subjected to varying degrees of wave and current action during transport and during deposition. Consequently, sandstone reservoirs vary from clean, well sorted quartz sandstone with well rounded grains (Figure 1.2a) through more angular feldspathic sandstone containing varying amounts of clay (Figure 1.2b), to argillaceous, very poorly sorted sandstone containing varying amounts of rock fragments (Figure 1-1c) all of which may be affected by varying degrees of compaction, cementation, solution and replacement. 1.6.1 Pore Space The basic framework of a sandstone reservoir is formed by the sand grains between which the pore space may or may not contain interstitial fine material and/or cement (Figure 1.3). The amount of this intergranular pore space or porosity is controlled primarily by sorting of the sediment and to a lesser extent by the packing of the grains. Sorting is a measure of the spread of distribution of grain size on either side of an average in a sediment. Theoretically, grain size has no effect on porosity. This is actually true only for spherical grains all of the same size. However, the arrangement of such spheres has a large effect on the porosity of the pack. Porosity is at its maximum for spherical grains but becomes progressively less as the angularity of the grains increases because such grains pack together more closely. Experimental data from artificial sands confirm that the grain size essentially has no influence on porosity. However, porosities of wet packed sands show a general decrease as sorting becomes poorer. This is because1-12 with a mixture of sizes, the smaller grains partially fill the interstices between the larger grains. Permeability being a measure of the ease with which a material transmits fluids, depends primarily upon the size, shapes and extent of the interconnections between individual pores rather than the size of the pores themselves. However, since the interconnections are directly related to the pore size which in turn is related to grain size, there are relationships between these factors and permeability. Krumbein and Monk (1942) have shown that permeability varies as the square of the mean grain diameter and a complex function of sorting (other factors being equal). Experimental data show a marked decrease in permeability as grain size decreases and as sorting becomes poorer. Experience has also shown that the permeability measured normal to the bedding is usually less than permeability measured parallel to the bedding and that large variations in permeability occur in different directions in the bedding plane. Clay in the pore space of a reservoir may affect the performance of a reservoir very adversely (Figure 1.4). The amount and kind of clay, as well as distribution throughout the reservoir rock, has an important bearing on liquid permeability, whereas a small amount has little effect on porosity. If fresh water, for example drilling fluid filtrate, invades a reservoir, certain clays, such as montmorillonite, will swell and plug some of the pore interconnections, drastically reducing the permeability, whereas saline water may have no effect.1-13 Sandstone of epicontinental reatm of sedinentation St. Pecer sand, Hid Continent (polarized Light) f sedimentation (polarized Light) Sandstone of orogente resin of sedimentation Ventura field, California (polarized light) Figure 1.2: Examples of sandstone reservoir rocks. (A) clean, well sorted sandstone, (B) angular, feldspathic sandstone and (C) argillacious, very poorly sorted sandstone.1-14 LEGEND aa Wi Pore space Qtz - Quartz Nicols crossed Kf Potash Feldspar EB colcite cement Pf —Plagioclose Feldspar Gh ~Ghert Ls Limestone Figure 1.3: Framework of reservoir sand with interstitial clay and cement.1-15 1.6.2 Compaction and Cementation The pore space of the original unconsolidated sediment is reduced in ancient rocks by many factors. Compaction and cementation are the most important of these factors but they in turn are affected by composition of the sediment , and the contained fluids and their pressures.Compaction by the weight of the overburden commences as soon as a sediment is deposited. It produces reduction of pore space as a result of: 1. Closer packing of the grains which causes smaller pores and connected passages. 2. Crushing and fracturing of grains, and dissolution at points of contact sometimes accompanied by precipitation of silica in the pore space nearby. Alkaline interstitial water seems to provide conditions more conducive to dissolution than does saline water. 3. Plastic deformation of the softer grains which tend to mold around the harder grains thus destroying pore space. The softer grains may be composed of limestone, shale, siltstone and other rock fragments, and when the amount of such soft material exceeds 10 to 12%, the permeability may be largely destroyed even though some porosity usually remains. Early cementation of sand may produce a rigid framework which will inhibit compaction until the depth of burial exceeds that at which fracturing of the grains and cement is initiated. Abnormal fluid pressure in sandstone reservoirs also inhibits compaction because the overburden is partly supported by the fluid pressure. Reduction of reservoir pressure by production of fluid can lead to compaction of the producing zone by rearrangement of the sand grains. This can produce serious and expensive subsidence of surfaceapt amdty {p01 slonrasau auonspurs Uf Kejo pasiadsip jo sada [e19U99, te SAND GRAIN * Troe general types of dispersed clay in sandstone reservoir rocks: a} disrote particles of kaoknite; (D) pore bing by chlorite; {cl pote bridging by tite. (Drawn from SEM photomicrographs by J. W. Neasham, Trans. Soc. Petrol, Enars, AIME, 1977). OL1-17 facilities such as occurred in the Wilmington field in California and Bolivar District fields in Venezuela. This form of compaction leads to porosity and permeability reduction which is irreversible and which may affect the producing characteristics of the reservoir very adversely. Cementation is the result of recrystallization from solution of silica, carbonates and other soluble materials in the pores of clastic rocks. The most common cementing materials in sandstone reservoirs are silica and calcite but many others do occur. It is not ‘uncommon to find both silica and calcite present in the cement and in such cases, the silica in the form of quartz has precipitated first and the calcite later. Silica cement usually occurs in the form of quartz and grows in optical continuity with the sand grains until finally the crystals interfere with one another and an interlocking network results. Calcite cement is often patchy but may completely fill the pore space. Silica cement appears to have two origins: (1) early cementation before the sands were compacted appreciably, and (2) deposition of silica predissolved by pressure solution during compaction. In the Eocene Wilcox sandstones of the Gulf Coast, the distribution of silica cement can be related to both primary texture and depth of burial. The amount of silica cement tends to increase in coarser and better sorted sands and, to a lesser extent, with depth. Much of this cement appears to be early and unrelated to the compaction process. 1.6.3 Classification Sandstones may be classified by mineral composition (Figure 1.5). The principal types are: (a) quartz sandsones consisting of over 95% detrital quartz, (b) feldspathic sandstone consisting of 5 - 25% feldspar, (c) arkosic sandstone consisting of over 25% feldspar, (d)sublithic sandstone consisting of 5 - 25% rock fragments and (e) lithic sandstone consisting of over 25% rock fragments. Sandstone grain texture consists of five components: (a) size, (b) sorting, (c) shape (sphericity), (d) roundness, and (e) packing. Porosity is independent of grain size but decreases as sorting gets poorer. Close packing reduces porosity and permeability. The effects of shape and roundness on porosity are less definite. Permeability increases with increasing grain size, but decreases with poorer sorting. Permeability generally increases with angularity and decreasing sphericity.QUARTZ nox____.____Chett, Quortrite frags, Vein Quartz it CUNT, ry enn by s Peas IMMATURE FELDSPAR ROCK FRAGMENTS Plutonic rock frogs, Gneiss frogs. Yolcanic, ls, Shale frags; Micas, Chlorite Figure 1.5: Classification of sandstones by composition,1.7 CARBONATE RESERVOIRS (Limestones and Dolomites) Most carbonate rocks, like clastics, are composed of particles of clay to gravel size that were generally deposited in a marine environment. However, they differ from terrigenous clastics in that they are deposited as lime particles which are produced locally, whereas, sandstones are composed of particles transported from an outside source by water currents. They differ even more importantly from sandstones by being subject to more post-depositional diagenesis ranging from simple cementation of the original particles to complete recrystallization or replacement by dolomite or chert. In addition, they are susceptible to solution at any stage in their diagenesis. They are usually more poorly sorted than clastics. Components of carbonate rocks are usually (1) grains of various kinds, (2) lime mud, and (3) carbonate cement precipitated later. There are several kinds of grains, of which four are the most important. These are (1) shell fragments, called “bio”, (2) fragments of previously deposited limestone called “intraclasts”, (3) small round pellets - the excreta of worms, and (4) ooliths - spheres formed. by rolling lime particles along the bottom. Lime mud consists of clay- sized particles of lime. The material between the grains may be primary lime mud deposited at the same time as the grains which would be grain- supported rock, or the grains may be “floating” in lime mud which would be muti-supported rock.1-21 17.1 Classification Carbonates are usually classified according to depositional texture as shown in Figure 1.6. The presence or absence of lime mud and the type and abundance of grains form the basis of the classification. A boundstone consists of original skeletal components bound together during deposition (Figure 1.7). Grainstones consist of packed carbonate grains with the texture being grain-supported and very little lime mud (Figure 1.8). Packstones are grain- supported but contain very substantial amounts of lime mud. Wackestones have a larger amount of lime muds, such that the grains effectively “float” in the mud. Mudstones consist of essentially lime muds only. The presence of lime mud may be most important in the development of porosity in carbonates because under the right conditions, lime mud may be preferentially dolomitized and may also be more readily leached out than’ the grains. Good porosity in carbonate reservoirs is usually due to dolomitization. The largest volume of carbonate petroleum reserves comes from dolomites. Dolomitization occurs from the substitution of magnesium for calcium in half of the sites in a carbonate crystal. A volume loss of 12 to 13% due to dolomitization results in a corresponding increase in porosity. Due to their larger surface area, mud-size grains are more easily dolomitized than sand-sized grains. Thus, the best carbonate reservoirs may have the lowest primary porosity. Dolomitization also creates planar crystal surfaces and harder crystal structures. Thus, dolomites retain more of their porosity during compaction than limestones.1-22 Limestone or Dolomite and Mixtures fu Gra: Skeletal Supported Supported Supported Crystalline ! < 10% Grains > 10% Grains X Fine V Coatse Mudstone al Boundstone Contains Mud Lacks Mud Packstone Grainstone Figure 1. Classification of carbonates by texture.Figure 1.7: Examples of boundstones.Figure 1.8: Examples of grainstones. 1-241-25 1.7.2 Pore Space The porosity, permeability and pore space distribution in carbonate’ reservoir rocks are related to both the depositional environment and the diagenesis of the sediment. They are most commonly of secondary (diagenetic) origin although residual primary pore space does occur. Carbonates have a large range of pore structures due to the complex nature of carbonate constituents and diagenetic features. The pore structures have been classified by Choquette and Pray, 1970) as shown in Figure 1.9. Fabric-selective porosity includes: *Interparticle porosity. ¢Intercrystalline porosity - typical of dolomites. ¢Fenestral porosity - by solution along bedding planes or joint surfaces. Skeletal, framework, molding, or shelter porosity - selective solution of, within, or around fossil material. *Oomoldic porosity - selective solution of ooliths. *Non fabric-selective porosity includes: «Fracture porosity - by stress or shrinkage. *Channel porosity - widening and coalescence of fractures. ¢Vuggy or cavernous porosity.1-26 interparticle Intraparticle Intercrysti wn i ane eulemuece py SELECTIVE ~~ Fenestral NOT FABRIC SELECTIVE ‘Channel vug FABRIC SELECTIVE OR NOT Breccia Burrow Shrinkage Interparticle: porosity between particles Intraparticle: porosity within individual particles or grains Iniercrystal: porosity between crystals Molsic: porosity formed by selective removal of an individual ce of rock Fenestra: pores larger than grain supported interstices (interpari Sheer: Porosity created by the sheltering effect of large sedime Growth Framework: porosity created by inplace growth of a carbonate rack framework | : Not Fabric Selective 4 i Fracture: porosity formed by fracturing Channel; © markedly elongate pores Vue: pores larger than 1/36 mm in diameter and somewhat ccvunt in shane Cavern: very laree channel or vue i Fabric Selective oF Kot i Brec interparticle porosity in breccia } Boring: porosity created by boring organisms - porosity ereated by organism burrowing. porosity produced by sediment shrinkage Figure 1.9: Classification of pore systems in carbonate rocks (Choquette and Pray, 1970).1-27 *Bioturbation porosity - from boring and burrowing. «Breccia porosity - in some cases, really high fracture porosity. In carbonates, porosity and permeability are not well- correlated with grain size or sorting. Porosity and permeability are controlled largely by the amount of fines and by diagenesis. Correlation of petrophysical properties with rock type is thus very difficult. Table 1.6 shows a comparison of the pore space characteristics of clastic and carbonate rocks (Choquette and Pray, 1970). 1.8 FRACTURED RESERVOIRS Natural reservoir fractures are caused by brittle failure, usually due to such factors as (a) folding, (b) faulting, (c) fluid pressure, (d) release of lithostatic pressure, (e) pressure solution, (f) dehydration, (g) weathering, (h) cooling and (i) impact craters. Natural fractures can exist in essentially any type of rock although they are particularly common in carbonates. Naturally fractured reservoirs are usually treated by a dual porosity approach to deal with their properties. The matrix rock (between fractures) usually has reasonable porosity and extremely low permeability. Fractures range in size from hair-size to several millimeters in aperture. Fractures that have not been filled with cement have very high permeabilities, even though they may be fairly widely-spaced. However, the fracture system: generally contains only a small fraction of the reservoir pore space. Thus, the matrix contains the bulk of the reservoir pore volume while the fractures contain the bulk of the reservoir flow capacity.1-28 Table 1.6: Comparison of pore-space properties in clastic and carbonate rocks (Choquette and Pray, 1970). Factor Clastics Carbonates Types of original almost exclusively interparticle commonly pores intergranular predominates, but (interparticle) intraparticle and other types important Size(s) of original pore and pore-throat_ pore and pore-throat pores sizes closely related to _sizes commonly show sedimentary particle size itie relation to and sorting sedimentary particle size or sorting Shape(s) of pores strong dependence on _—_ ranges from: 1) a strong particle shape (always a dependence on particle “negative” of particles) shape (either a “negative” or “positive” of particles) to 2) complete independence of shapes of depositional or diagenetic components Uniformity of pore commonly fairly uniform ranges from: 1) fairly size, shape, and within a single rock type _uniform to 2) extremely distribution heterogeneous, even within a single rock type Influence of often minor, but ranges from: 1) minor diagenesis sometimes major, reduction to 2) complete reduction of original pore destruction or spaces; compaction and modification of original cementation are pore spaces or creation important of new pores; cementation and solution are important Type(s) of ukimate almost exclusively widely varied because of pores intergranular postdepositional modification1-29 Figure 1.10 shows a naturally fractured rock together with its idealized dual porosity approximation. Matrix Fractures Figure 1.10: Idealization of naturally fractured reservoir (Warren and Root, 1963)1.9 PETROPHYSICS Petrophysics is the study of rock properties and their interactions with fluids (gases, liquid hydrocarbons and aqueous solutions). Because petroleum reservoir rocks must have porosity and permeability, we are most interested in the properties of porous and permeable rocks. The purpose of this course is to provide a basic understanding of the physical properties of permeable geologic rocks and the interactions of the various fluids with their interstitial surfaces. Particular emphasis will be placed on the transport properties of the rocks for single phase and multiphase flow.2. SINGLE-PHASE FLUID STORAGE AND TRANSPORT PROPERTIES 2,1 POROSITY Porosity gives an indication of the rock's ability to store fluids. It is defined by the following equation: Vp _ Vb-V. =P tbo Ns (2.1) ow where ¢ = fractional porosity Vp = pore volume Vb = bulk volume Vs = grain volume Porosity may be classified as total or effective porosity. Total porosity accounts for all the pores in the rock (interconnected and isolated pores) whereas effective porosity only accounts for the interconnected pores. Therefore, effective porosity will be less than or equal to total porosity depending on the amount of isolated pores in the rock. From a reservoir engineering standpoint, it is the effective porosity that matters, not the total porosity. Porosity may also be classified as primary or secondary. Primary porosity is that which was formed at the time of deposition whereas secondary porosity was developed after deposition and burial of the formation. Sandstone porosity is practically all primary porosity whereas carbonate porosity tends to be secondary porosity.2-2 2.1.1 Factors Affecting Sandstone Porosity ‘These include packing, sorting and cementation. Packing describes the arrangement of the sand grains relative to one another. Figures 2.1 shows three idealized types of packing for spherical sand grains and their theoretical porosities. The cubic packing has a porosity of 47.6%; the hexagonal packing has a porostiy of 39.5% and the rhombohedral packing has a porosity of 25.9%. The porosity of a pack of of uniform spheres is independent of the grain size. A B c Cuble Packing Hexagonal Packing Rhombohedral Packing 6 =90° 6 =60° 6 =45° Vp =D? Vp =D D sin 6 Vp = D2 «Dsin d= D3 NE Vg =70°/6 =0.8660° Vg <7D7 6 ge Din z/6) Vg =7D9/6 03 WV B- 7/6) o° © ~ D3 (0.866-7/6) O° 88 WE =0,476=47.6% 0.866D = 0.259 = 25.9% =0.395 =39.5% Figure 2.1: Effect of packing on porosity of uniform spheres.2-3 Well sorted sandstone consists of grains having approximately the same size whereas poorly sorted sandstone consists of grains having a wide range of different grain sizes. Poor sorting reduces the porosity of the sandstone as may be seen in Figure 2.2 in which the small grains fit into the pores created by the large grains, reducing the porosity.” ~ Figure 2.2: Effect of sorting on porosity. (A) Irregular grains, (B) Idealized spherical grains.2-4 Table 2.1 shows experimentally measured porosities of various artificial sandpacks. Note the general decrease of porosity with poor sorting for all grain sizes and the approximately constant porosity of the extremely well sorted sands for all grain sizes. Table 2.1 Measured Porosities of Artificial Sandpacks 1.0 1a ‘VERY WELL, 12 Ls 2.0 27 87 MODERATE VERY POOR 1,000 0.710 0.500 0.350 0.250 0.177 0.125 0.088 0.062 MEDIAN DIAMETER (MM)In consolidated rocks, the sand grains are cemented together usually by quartz or carbonates. Cementation reduces the porosity of the sand as shown in Figure 2.3. fa Figure 2.3: Effect of cementation on porosity.2.1.2 Factors Affecting Carbonate Porosity In carbonates, secondary porosity is usually more important than primary porosity. The major sources of secondary porosity are fracturing, solution and chemical replacement. Fractures are cracks in the rock. Although fracture porosity is generally small, the fractures are very useful in allowing fluids to flow more easily through the rock. Therefore, they greatly enhance the flow capacity of the rock. Solution is a chemical reaction in which water with dissolved carbon dioxide reacts with calcium carbonate to form calcium bicarbonate which is soluble. This reaction increases the porosity of the limestone. CO2 + H20 = H2CO3 (2.2) H2CO3 + CaCO3 = Ca(HCO3)2 (2.3) Chemical Replacement is a chemical reaction in which one type of ion replaces another with a resulting shrinkage in the size of the new compound. An example is dolomitization in which some of the calcium ions in calcium carbonate are replaced by magnesium ions to form calcium magnesium carbonate (dolomite). This replacement causes a shrinkage of 12 to 13 %, with a corresponding increase in secondary porosity. 2CaCO3 + MgClz= CaMg(CO3)2 + CaCl2 (2.4)2.1.3 Typical Reservoir Porosity Values Sandstones have porosities that typically range from 8% to 38%, with an average of 18%. About 95% of sandstone porosity is effective porosity. Sandstone porosity is usually mostly intergranular porosity. Carbonates have porosities that typically range from 3% to 15%, with an average of about 8%. About 90% of carbonate porosity is effective porosity. Carbonate porosities are much more difficult to characterize and may consist of (1) intergranular, (2) intercrytalline, (3) fractures and fissures, and (4) vugular porosities. 2.1.4 Direct Laboratory Measurement of Porosity Direct measurement of porosity requires the measurements of two of the three volumes Vp, Vs and Vp. In the laboratory, measurements are usually performed on extracted cores which have been cleaned and dried. Bulk volume can be determined by (1) caliper and (2) fluid displacement. For well machined samples, the dimensions can be measured with a caliper, from which the bulk volume can be calculated. Two types of fluid displacements can be used to determine bulk volume. In the first method, fluid that does not easily penetrate the pores such as mercury is used. The apparatus, which is known as a pycnometer, measures the volume of mercury displaced by the sample (Figure 2.4a). Since mercury does not penetrate the pores at atmospheric pressure, the volume of mercury dispalced is equal to the bulk volume of the sample. In the-second method, fluid which easily saturates the sample is used. The sample is weighed in air, evacuated and then saturated with a liquid (water, kerosene, or toluene). The saturated sample is weighed in air and then weighed immersed in the saturating fluid. The loss in weight of the saturated sample when immersed in the saturating liquid is proportional to the bulk volume of the sample (Archimedes principle).2-8 (a) Bulk volume pycnometer (o) rai volume ( Boyle's tow) ‘Solid reference] E or somple Gos supply “Sample” Reference chamber volume Solid volume v, = vy (Calibration curve vs agoinst p2/p1-pzcan be ‘established using steel blanks ) {c) Bulk and pore volume porosimeter Figure 2.4: Reference mork Isolation volve Dead weight tester Micrometer mercury pump eee Schematics of equipment for measurement of core plug porosity.(a) Bulk volume pycnometer; (b) Boyle’s law porosimeter; (c) Bulk and pore volume porosimeter.Grain volume can be determined by (1) fluid displacement and (2) gas exapansion using Boyle's law porosimeter. The loss in weight of the dry sample and the sample fully immersed in a liquid is proportional to the grain volume. Figure 2.4b shows a schematic diagram of a Boyle’s law porosimeter for grain volume determination by gas expansion. The sample, which is confined in a vessel of known volume Vj, is pressured by gas (air, nitrogen or helium) to a pressure Pj. The vessel of volume Vj is connected to a second vessel of known volume V2 which is initially evacuated. The valve between the two vessels is opened and the pressure in the two vessels is allowed to stabilize at P2. By Boyle’s law (PV = constant), (V1 - Vs)P1 = (V1 - Vs + V2)P2 (2.5) which can be rearranged as Pe _v, (2.6) Vs Hee Pi-P2 The instrument can be calibrated with steel blanks of known volume. Calibration consists of a plot of Vs versus P2/(P -P2).Pore volume can be determined by (1) fluid saturation and (2) mercury injection. The difference in the weight of the saturated sample and the dry sample is proportional to the pore volume. Mercury injection consists of forcing mercury under relatively high pressure into the pores of the sample using a mercury porosimeter (Figure 2.4c). Because the air in the pores is compressed to a negligibly small value, the volume of mercury injected is essentially equal to the connected pore volume of the sample. This is a destructive method because after the test, the sample is no longer suitable for other measurements. Mercury porosimetry is also used to determine capillary pressure and pore size distribution (see Chapter 3). The methods so far described determine the effective porosity of the sample. To determine the total porosity, the sample is ground into a fine powder after bulk volume measurement. The grain volume of the ground sample can be determined either by liquid displacement or by assuming an average grain density. The measurement of porosity on consolidated samples in routine core analysis might generally be expected to yield values of the true fractional porosity plus or minus 0.005, ie., a true value of 27% porosity may be measured between 26.5% and 27.5% porosity.2-11 With the availability of X-ray computed tomography (CT) imaging sytems in research laboratories, it is now possible to measure the porosity distributions in core samples. Peters and Afzal (1992) have made such measurements in an artificial sandpack and a Berea sandstone approximately 60 cm long and 5 cm in diameter. CT imaging gives rise to a very large data set, over 600,000 porosity values in some cases. Therefore, it is convenient to present the results of the measurements as images (Figures 2.5 and 2.6). Notice from Figure 2.5 that a sandpack may not be as uniform as we always assume them to be. The packing technique used in this test introduced significant porosity variation into the pack. The packing history is clearly evident in the image. The dominant feature in the porosity variation of the Berea sandstone is layering which is clearly visible in Figure 2.6. The porosity data also can be presented in histograms (see Figures 2.7 and 2.8). 2.1.5 Indirect Porosity Measurement from Well Logs Conventional logging techniques for measuring porosity are the Density, Neutron and Sonic logs. All of these logs provide an indication of total porosity. Care should always be taken in comparing core versus log-derived porosities, particularly in rocks that have been highly affected by diagenesis. Logs also measure average porosities over much larger volume than conventional laboratory core analysis. Also, laboratory cores have been subjected to decompaction and cleaning which may make them somewhat different than in the native state.cr eee +275 28.285 29.295 30.305 31.315 132.325 Porosity (fraction) Figure 2.5: Porosity image of a sandpack from CT imaging. L = 54.2 cm, d= 4.8 cm. (a) Cross-sectional slice. (b) Longitudinal vertical slice. (Peters and Afzal, 1992).(b) +1450 415.155. +165 0174175 186185 19.195. Porosity (fraction) Figure 2.6: Porosity image of a Berea sandstone from CT imaging. L = 60.2 cm, d = 5.1 cm. (a) Cross-sectional slice. (b) Longitudinal vertical slice. (Peters and Afzal, 1992).2.1.6 Porosity Distribution Sometimes, it is observed that given lithologies in a reservoir with a particular depositional and diagenetic history will show a characteristic frequency distribution. One of the more commonly observed distributions is the truncated normal (“bell-shaped”) distribution. Figures 2.5 and 2.6 show bell-shaped porosity histograms for a sandpack and a Berea sandstone determined by CT scanning. For such distributions, the mean porosity is calculated as the arithmetic average: ao $ =a $i 27 Sometimes, bimodal frequency distributions are abserved, usually reflecting two different geologic facies.5000 ‘50000 45000 ‘35000 25000 15000 10000 6000 FREQUENCY ( # OF OCCURENCES ) POROSITY Figure 2.7: Porosity histogram for a sandpack from CT imaging. L = 54.2 cm, d = 4.8 cm. Mean = 29.7%, Standard deviation = 2.5%.(Peters and Afzal, 1992).FREQUENCY ( # OF OCCURENCES ) Figure 2.8: ‘50000 ‘40000 ‘30000 POROSITY Porosity histogram for a Berea sandstone from CT imaging. L = 60.2 cm, d = 5.1 cm. Mean = 17.3%, Standard deviation = 2.0%.(Peters and Afzal, 1992).2.2 PORE VOLUME COMPRESSIBILITY Rock at reservoir conditions is subjected to overburden stress, while at the surface, recovered core has been relieved of the overburden stresses. It is not usual to perform routine porosity measurements under stress approaching reservoir conditions. Because of this, laboratory- measured porosities are generally expected to be higher than in situ values. Pore volume compressibility can be used to correct laboratory- measured porosity to an in situ value and for other reservoir engineering calculations. Pore volume compressibility is defined as: a2 ali a (2.8) Note that pore volume compressibility is defined without the negative sign because porosity decreases as pore pressure decreases. Hall’s (1953) correlation may be used to obtain rough estimates of pore volume compressibity (Figure 2.9). Compressibilities, however, are highly affected by reservoir type and overburden conditions. For detailed analysis, pore volume compressibility should be measured in the laboratory. Figure 2.10 shows a schematic of a typical device for measuring pore volume compressibilities. These measurements are fairly difficult to make, and must include calibrations for the compressibilities of the fluids and equipment. Pore volume compressibilities are not always constant. For example, in abnormally pressured reservoirs, it is common to find that pore volume compressibility at high reservoir pressures is much larger than the pore volume compressibility at low pressures. This is due to compaction effects as the pore pressure declines.eae = Yt [ { © LIMESTONE | See 10 12 4 16 16 20 22 24 26 POROSITY, percent EFFECTIVE ROCK COMPRESSIBILITY Figure 2.9: Pore volume compressibility versus porosity (Hall, 1953)SMALL BORE CALIBRATED GLASS TUBE PRESSURE CHAMBER 510. 34" WALL THICKNESS HYORAULIC_ PUMP. : = FOR PRESSURING HYORAULIC PUMP ANNULUS. FOR PRESSURING CORE Figure 2.10: Typical apparatus for measuring pore volume compressibility (Hall, 1953).Total system compressibility is a parameter of importance in pressure transient testing. Total system compressibility is defined as the combined compressibility of pore volume and all fluids saturating the medium: C1 = Soco + Sucw + Sycg + cr (2.9) where So, Sw, Sg = oil, water and gas saturations, fraction Co, Cw, Cg = oil, water and gas coefficients of compressibility, psi? c= pore volume coefficient of compressibility, psi* 2.3 PERMEABILITY Permeability gives an indication of the porous mediums ability to transmit fluids (i.e., permit fluid flow). It is defined through Darcy’s law. For a horizontal system, Darcy’s law for single phase flow in differential form is = -kA® 2.10) 9 WN ax arid where q = volumetric flowrate, cm3/s k = _ absolute permeability of the rock, darcy A = cross sectional area in the flow direction, cm2 = _ fluid viscosity, ep 2 = fluid pressure gradient, atm/cm x For steady-state linear flow (Figure 2.9), Darcy’s Law may be integrated to give: =k (pip) aL (2.51)volumetric flowrate, cm3/s absolute permeability of the rock, darcy cross sectional area in the flow direction, cm2 inlet pressure, atm. outlet pressure, atm fluid viscosity, cp length of rock, cmFigure 2.9: Linear flow geometry.For steady-state linear flow in oilfield units, Eq. 2.11 becomes q = 0.001127kA (P1- Po) (2.12) UB L or q=—L kA (p= Pd (2.13) 887.2 pB OL. where q volumetric flowrate, STB/D k absolute permeability of the rock, millidarcy (md) A cross sectional area in the flow direction, ft2 Pi inlet pressure, psig P2 outlet pressure, psig in fluid viscosity, cp L = _length of rock, ft B= __ oil formation volume factor, RB/STB For radial flow of fluids into a wellbore (Figure 2.10), Darcy’s law may be expressed in radial coordinates in field units as 0.00708kh(pe- Pw), (2.14) wes) or__1_Khipe pu) 9" Tha uBix(Z=| (245) production rate, STB/D absolute permeability of the rock, md pay thickness, ft pressure at external radius, psig pressure at the wellbore, psig fluid viscosity, cp formation volume factor, RB/STB external drainage radius, ft wellbore radius, ft natural logarithmFigure 2.10: Radial flow into a wellbore. 2-252.3.1 Dimensions and Unit of Permeability Eq. 2.11 may be rearranged to determine the dimensions of permeability as follows: pq = fol) (2.16) [AlI4p] where the square brackets refer to the dimensions of the enclosed variable and AP is the pressure drop. Using mass (M), length (L) and time (T) as the fundamental dimensions, the dimensions of the variables on the right side of Eq, 2.16 are oL (ql T = M fH it (LJ= L (Als? Ap} = ML {4p} in Thus, [k}= or =? (2.17)2-27 Eq. 2.17 shows that the permeability of a porous medium has the dimensions of length squared. This means that permeability is proportional to some sort of mean square pore diameter of the porous medium. Based on Eq, 2.17, a rational unit of permeability would be the foot squared or centimeter squared. Both units were found to be too large for use with porous media. Therefore, the petroleum industry adopted the darcy as the unit of permeability. It can be shown that 9.869 x 10-9 cm? I darcy = 9,869 x 10°13 m? = 1.062 x 1071! ft? 2.3.2 Laboratory Determination of Permeability Permeability is an empirical parameter which must be determined by measuring all the parameters in Darcy's law. Core permeabilities are usually measured in the laboratory using dry gas (air, nitrogen or helium) as the flowing fluid to minimize rock-fluid reaction. In this case, another form of Darcy’s law for steady-state flow of an ideal gas must be used, recognizing that volumetric flowrate changes with pressure: = kA @i-p) 2.18) Que L j where a qsc = volumetric flowrate measured at base pressure and flowing temperature, cm3/s k = __ absolute permeability of the rock, darcy A =__ cross sectional area in the flow direction, cm?2-28 P; = absolute inlet pressure, atm P2 = absolute outlet pressure, atm Pse = absolute base pressure, atm b= __ fluid viscosity, cp L = = __ length of rock,cm Low-pressure gas permeability measurements give rise to what is termed the Klinkenberg (1941) effect. Klinkenberg effect is due to the fact that at low mean pressures, the mean free path of gas molecules is about the same size as the pores in the rock. This gives rise to gas slippage at the wall of the pores. Asa result, the permeability to gas at low pressure is higher than it should be. Klinkenberg found that at low pressures, measured gas permeabilities are related to the effective liquid (or high- pressure gas) permeabilities by: Weeki + 2) (2.19) kg = — measured gas permeability = effective liquid permeability empirical constant dependent on rock and gas used mean core pressure, (pi + p2)/2 The effective liquid permeability can be determined in the laboratory by measuring gas permeabilities at different average core pressures. A plot of kg versus 1/Pm yields a straight line with an intercept equal to ky, (Figure 2.11). The Klinkenberg effect is not important at normal reservoir pressures.Permeability (md) 50 40 30 20 ° 2 4 6 8 Reciprocal mean pressure (1/atm) Figure 2.11: Klinkenberg permeability correction.2-30 In general, permeability is an anisotropic property of a porous medium, that is, it is directional. Routine core analyses are usually made on core plugs drilled horizontally from a core. It is sometimes possible to specify that plugs be cut along bedding planes. Plugs are sometimes cut vertically if it is desired to obtain vertical as well as horizontal permeabilities. Because of interactions of fluids with reservoir rocks, absolute permeabilities to different fluids may not always be the same. The permeability to brine, for example, is often somewhat less than the Klinkenberg-corrected gas permeability. For this reason, it is usually beneficial to obtain at least a few brine permeability measurements, especially if waterflooding is anticipated. These measurements are time- consuming, and thus are more costly. Another permeability that is sometimes referenced is the oi] permeability at irreducible water saturation. This value is typically around 80% of the absolute permeability. 2.3.3 Field Determination of Permeability Permeabilities can be determined in the field by use of pressure transient tests, essentially measuring the permeability of the in-situ reservoir rock. A transient pressure test consists of changing the flow rate of a well and then recording the bottomhole pressure response as a function of time. The pressure data can be analyzed to obtain formation permeability and other reservoir and well parameters. Oil permeability at irreducible water saturation is normally used for comparison with the permeability from most pressure transient tests that involve only oil flow. For transient flow, a combination of the law of conservation of mass, Darcy’s law and an equation of state for a slightly compressible liquid gives the following pressure equation:2-31 1B OP (2.20) where pressure, atm radial distance from the wellbore, cm permeability, darcy porosity, fraction viscosity, cp total compressibility, vol/vol/atm = time,s Ferny Sete Eq. 2.20 is known as the diffusivity equation. The solutions of this equation for specific initial and boundary conditions are useful for transient pressure analysis and natural water influx calculations. 2.3.4 Non-Darcy Flow At high pressures, particularly in gas wells, an additional pressure drop can result from inertial-turbulence effects. This additional pressure drop is usually expressed as a non-Darcy flow coefficient that shows up in the radial flow equation for gases as: TkhT <(p3 - p2) qe = wp, fi i re Da] (2.21) where gas flowrate at standard conditions, cm3/s absolute permeability, darcy pay thickness, cm reservoir temperature (absolute), K standard temperature (absolute), K2-32 Pe = _ pressure at external radius (absolute), atm Pw = Pressure at wellbore (absolute), atm Psc = Standard pressure (absolute), atm S = __. skin factor, dimensionless D- = __ non-Darcy coefficient, s/cm> fee external drainage radius, cm Tw = wellbore radius, cm H == ___ viscosity, cp Z = gas Z-factor, dimensionless 2.3.5 Factors Affecting Permeability These include compaction, pore size, sorting, cementation, layering and clay swelling. Compaction. Just as compaction reduces porosity, it also reduces permeability. As a result of compaction, the permeability of rocks tends to decrease with depth of burial. Pore size. In general, for a sand, the permeability is proportional to the square of the mean pore size. For a well sorted sand, the pore size is proportional to the grain size. Therefore, for well sorted sands, the permeability is proportional to the grain size. Thus, a well sorted sand with large grain size will have a higher permeability than a well sorted sand with small grain size (Table 2.2).2-33 Table 2.2 Measured Porosities and Permeabilities of Artificial Sandpacks ——— SIZE —* | Coarse Mediun Fine Very Fine sortins | | tower | upper | Lower | upper | tower | Upper | Lover Extrem, well 43% 4% 4% 4am 43% 43% 43% sorted 2400 | 1200 60D 30D 15D 2D 4D V. well 42% 40% 40% 40% ain at aan sorted 2400 | 115d 60D 30D 15D D 4D Well 38% 38% 3% 39% 40% 40% 40% sorted 1500 75D “oD 20D 2D Moder‘ ly 33% 34% 30h 34% 35% sorted 50D 30D isd 2” wD Poorly 30% 31% ah 31% 32% sorted 25D 10D 6D 4 i)2234 To see why grain size affects the permeability of a medium, consider the wetted surface area per unit volume (specific surface area) of a rectangular conduit and a porous medium. Let the porous medium consist of uniform spherical sands of 0.1 mm radius. Figure 2.12 shows that for 1 cubic meter of rectangular conduit, the wetted surface area is4 square meters whereas for the porous medium, the wetted surface area is 24,000 square meters. Thus, the specific surface area of the porous medium is 6,000 times that of the rectangular conduit. During flow, the fluid velocity on the wetted surface area is zero. Therefore, it will require more energy to sustain the same volumetric flow rate through the porous medium than the rectangular conduit. Since it is more difficult to transmit fluid through the porous medium than the conduit, the permeability of the porous medium is less than that of the conduit. If the grain radius of the sand in the porous medium is reduced to 0.01 mm, the specific surface area increases to 240,000 square meters, thereby reducing the permeability even further. Sorting. Poor sorting reduces the pore size and consequently reduces the permeability of a medium (Table 6.2). Cementation reduces the pore size and consequently reduces the permeability of the rock. Layering. Permeability is a directional quantity and can therefore be different in different directions. Because of layering in sedimentary rocks, the horizontal permeabilities in petroleum reservoirs tend to be higher than the vertical permeabilities. Clay swelling, Many consolidated sandstones contain clay and silt, e.g., arkoses and graywackes. Since montmorillonite-type clays absorb fresh water and swell, the permeabilities of such sandstones will be greatly reduced when measured with fresh water. The addition of salts, such as sodium chloride or potassium chloride, will in most cases, eliminate the clay swelling.2-35 nicked. «dm? reed surf = pattick xn lrex number a partcles poke snfrex rember oporicles ke 20%. ee Se 14 bat Ss 3s ope S= 3-02) x abr Ss 24000 m2/m? Figure 2.12: Effect of grain size on the specific surface area of a porous medium.2-36 2.3.6 Typical Reservoir Permeability Values Reservoir permeabilities vary widely, from 0.1 md for a tight gas sand to 3000 md for an unconsolidated sand. Reservoir permeabilities may be loosely described as follows: Very low: k500md 2.3.7 Permeability-Porosity Correlations Since permeability depends on the continuity of pore space, there is not in theory, or in fact, a unique relationship between the porosity of a rock and its permeability. For unconsolidated sands, it is possible to establish relationships between porosity and either some measure of apparent pore diameter or specific surface area and permeability. However, these have very limited application. In the same depositional and diagenetic environment, there is sometimes a reasonable relationship between porosity and permeability, although for a given porosity, permeabilities can vary widely. Figures 2.13 and 2.14 show typical permeability-porosity correlations obtained from core analysis. The correlation of Figure 2.13 consists of about 500 samples from a uniform sandstone reservoir. Apart from showing the general increase of permeability with an increase in porosity, the correlation shows the wide spread, or lack of close relation, between porosity and permeability. The correlation of Figure 2.14 shows the effect of grain size on the porosity-permeability relationships of sandstones.2-37 Attempts have been made to correlate permeability and porosity using an equation of the form: logk=ag+b (2.22) where a and b are constants. This relationship is at best applicable at a local level. Nevertheless, where it exists, it does provide one method of estimating permeability from well logs and even possibly from drill cuttings. 2.3.8 Capillary Tube Models of Porous Media Because rock pore space is made up of interconnected approximately linear flow conduits, attempts have been made to calculate permeability by modeling pore-level flow as if through cylindrical tubes. Such models are sometimes termed “bundle-of- capillary-tubes” models. The starting point for all these models is the Hagen-Poiseuille’s law for steady flow through a single, straight circular capillary of radius r: =m (p1-P2) - PA AP . q sO SL (2.23) where q volumetric flow rate, cm3/s r radius of capillary tube, cm A = __ cross sectional area in the flow direction, cm? PL inlet pressure, dynes/cm? P2 = - outlet pressure, dynes/cm2 Ap pressure drop (pj - p2), dynes/cm2 : L fluid viscosity, cp L = = _ length of capillary tube, cm2-38 ° & Permeobiity - Milidarcys 5 ou 0.05 2 4 16 1 20 22 6 @ 0 Porosity - Per Cent of Total Volume Permeability-porosity correlation for a sandstone Figure 2.13: reservoir (Levorsen, 1967).2-39 Permeability (mD) 19000 8000 6.000) 4000) 2.000] 1.000] 800) 600 400 very coarse to coarse 0 coarse to medium 8 fine 4 silty shaly 2 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 Porosity (%) Figure 2.14: - Effect of grain size on the permeability-porosity correlations for various sandstones. a2-40 ‘A comparison of Eq. 2.23 with Darcy’s law (Eq. 2.11) shows that the permeability of the capillary tube is given by kee (2.24) 8 If the porous medium is conceived to be a bundle of capillary tubes, then it can be shown that the permeability of the medium depends on the pore size distribution and porosity. Consider a system composed of a bundle of n capillary tubes of the same radius and length. For this system, the Hagen-Poiseuille’s law gives sont 4p . stare (2.25) Darcy’s law for the same medium is given by =kA AP a" (2.26) A comparison of Eqs. 2.25 and 2.26 gives the permeability of the system as k ast (2.27) The porosity of the medium is given by ge yee aml ont (2.28) Substituting Eq. 2.28 into Eq, 2.27 gives the permeability of the system as k -= (2.29)Eq. 2.29 is the simplest relationship between porosity and permeability for pores of the same size. Defining a specific surface area (Sp) as the internal area per unit pore volume of the n capillary tubes, we have = n2mL =-2 (2.30) Sr neeL tT oe Substituting Eq. 2.30 into Eq. 2.29 gives =12 ; k=1 3 (231) To account for the fact that in a porous medium, the capillaries are not circular, the constant (1/2) in Eq. 2.31 is replaced by an empirical constant (1/kz). With this modification, Eq. 2.31 becomes k= (2.32) kS which is the Kozeny equation where kz is the Kozeny constant. If the specific surface area (S) is defined in terms of the bulk volume, we have that S = Sp and the Kozeny equation becomes ke Ss 233) Sometimes, the specific surface area (Ss) is defined as the area per unit grain volume, In this case, the Kozeny equation becomes 3 k= —?% _ (2.34) K,S2(1 = 9)?2-42 Note the inverse relationship between permeability and the square of the specific surface area. Since the specific surface area increases as grain size decreases, pemeability should decrease with a decrease in grain size. This is indeed the case. Attempts have been made to improve Kozeny’s equation to make it to better model flow in real porous media. One improvement is to recognize that the bundle of tubes is tortuous, so that the effectvie flow length through any tube (L) is greater than the bulk length of the porous medium. With this improvement, Eq. 2.31 becomes k=? (2.35) KSLSLY where ko is a shape factor (equal to 2 for cylindrical conduits). The ratio (L,/L)? is called the tortuosity of the porous medium. If the tortuosity is designated by the symbol t, Eq. 2.35 becomes k= rd (2.36) By letting the Kozeny constant kz be equal to kot, Eq. 2.36 becomes identical to Eq. 2.32. Carman reported kot to be approximately 5. Others have suggested ko to be in the range 2.5 to 3. To apply Eq. 2.36 to estimate permeability, the porosity, specific surface area and the Kozeny constant must be known. Several techniques have been proposed for the determination of the specific surface area of porous media. These include (1) a statistical method, (2) adsorption methods, (3) the heat of wetting method and (4) a method based on fluid flow developed by Kozeny. oe2.3.9 Steady State Flow Through Fractures To increase the productivities of certain reservoirs, it is often necessary to fracture the reservoirs. It can be shown that Hagen- Poiseuille’s law for steady state flow through a rectangular slit of width w is given by q= WA (Pi- Po) (237) 1 OL where q volumetric flow rate, cm3/s w width of the fracture, cm. A cross sectional area in the flow direction, cm? Pi inlet pressure, dynes/cm? P2 outlet pressure, dynes/cm2 = _ fluid viscosity, cp L = _ length of fracture, cm A comparison of Eq. 2.37 with Darcy's law (Eq. 2.11) shows that the permeability of the fracture is given by k= = (2.38) 2.3.10 Averaging Permeability Data Because of heterogeneity, different portions of the same reservoir may have different permeabilities. It becomes necessary to be able to estimate the apparent or average permeability of various permeability combinations. For linear beds in series (Figure 2.15), the volumetric flow rate is the same for each bed and the total pressure drop is equal to the sum of the2-44 pressure drop across each bed. For n beds in series, the average permeability is given by k=_L_ (2.39) yu isi Xi where k = _ average permeability L = __ total length of the porous medium Li = _lengthofbedi ki = _ permeability of bedi For radial beds in series (Figure 2.16), the volumetric flow rate is the same for each bed and the total pressure drop is equal to the sum of the pressure drop across each bed. For n beds in series, the average permeability is given by In (re/tw) In (ri/ti-1) (2.40) where k = average permeability Te = external drainage radius Tw = wellbore radius i radius of bedi kK permeability of bediFigure 2.15: __ Linear beds in series. Figure 2.16: Radial beds in series.For linear beds in parallel (Figure 2.17), the total volumetric flow rate is the sum of the flow rate of each bed and the pressure drop is the same for each bed. For n beds in parallel, the average permeability is given by > Ai Keiet ya int k average permeability Ai ‘Area of bed i ki = permeability of bedi If the beds have the same width, then x Kihy where Kk = __ average permeability hi = thickness of bed i ki = permeability of bedi (2.41) (2.42) For radial beds in parallel (Figure 2.18), the total volumetric flow rate is the sum of the flow rate of each bed and the pressure drop is the same for each bed. For n beds in parallel, the average permeability is given by Eq. 2.42.Figure 2.17: Linear beds in parallel. Figure 2.18: Radial beds in parallel.2-48 23.11 Permeability Distribution There is evidence from extensive core analyses data that reservoir permeability tends to be log-normally distributed whereas porosity tends to be normally distributed. This means that permeability is not normally distributed but the log of permeability is normally distributed. Figure 2.19 shows typical histograms of permeability and log of permeability. Note that the permeability distribution is skewed whereas the log of permeability distribution is normal. For a log normal distribution, the mean value is given by: log=1 ¥ togtk) (2.43) i=l or k= Vikqkok3 ...kn (2.44) The mean given by Eq. 2.44 is usually called the geometric mean of the distribution.PERMEABILITY, md2-50 2.3.12 Meaures of Permeability Heterogeneity Petroleum reservoirs are nearly always heterogeneous. Therefore, it is necessary to be able to quantify the degree of heterogeneity of a reservoir. Dykstra and Parsons (1950) have presented a method of quantifying the degree of heterogeneity of a reservoir based on permeability data from core analysis. The method assumes a log normal permeability distribution and gives rise to a coefficient of permeability variation between zero and one. The Dykstra-Parsons method involves plotting the frequency distribution of the permeability on a log-normal probability graph paper. This is done by arranging the permeability values in descending order and then calculating for each permeability, the percent of the samples with permeabilities greater than that value. Table 2.3 shows an example calculation. Note that to avoid values of zero or 100%, the percent greater than or equal to value is normalized by n+1, where n is the number of samples. The data are plotted on a log-normal probability graph paper as shown in Figure 2.20. Normally, such a plot gives a straight line, at least when the central portion is used. The midpoint of the permeability distribution (%k2 = 50) is the log mean permeability. The Dykstra- Parsons coefficient of permeability variation, V, is determined from the graph as: kso — kes.1 Ve (2.45) (2.45) where kso = permeability value at %k2 = 50 (log mean) - kg4.1 = permeability value at %k2 = 84.1 (one standard deviation)Table 23 Calculation of Dykstra-Parsons Coefficient of Variation Permeability (k) Number of (md) Samples 2k zk 4388 1 43 2640 2 8.7 2543 3 13.0 1579 4 17.4 930 5 21,7 662 6 26.1 441 7 30.4 402 8 34.8 401 9 39.1 378 10 43.5 267 i 47.8 250 2 52.2 249 13 56.5 232 4 60.9 200 15 65.2 136 16 69.6 98 17 73.9 47 18 78.3 30 19 82.6 28 20 87.0 16 21 91.3PERMEABILITY VARIATION, V ee 3.07 10 8 o S SAMPLE PERMEABILITY, MD Y2 5 10 0 3 40500 7 0 MW 5 8 995. PORTION OF TOTAL SAMPLE HAVING HIGHER PERMEABILITY Figure 2.20: Log normal permeability distribution.A homogeneous reservoir has a coefficient of permeability variation that approaches zero whereas an extremely heterogeneous reservoir has a coefficient of permeability variation that approaches one. Petroleum reservoirs typically have coefficients of permeability variation between 0.5 and 0.9. ‘The coefficient of permeability variation is also known as the Dyktra-Parsons coefficient. Another measure of heterogeneity is the Lorenz coefficient . The Lorenz coefficient of permeability variation is obtained by plotting a graph of cumulative kh versus cumulative ¢h, sometimes called a flow capacity plot. Table 2.4 shows a typical calculation while Figure 2.21 shows the cumulative plot for determining the Lorenz coefficient. The Lorenz coefficient is defined from Figure 2.21 as: Lorenz coefficient = area ABCA. (2.46) area ADCA The Lorenz coefficient of permeability variation also varies from zero to one. Unfortunately, the Lorenz coefficient is not a unique measure of reservoir heterogeneity. Several different permeability distributions can give the same value of Lorenz coefficient. For log- normal permeability distribtuions, the Lorenz coefficient is very similar to the Dykstra-Parsons coefficient of permeability variation.2-54 Table 24 Calculation of Lorenz Coefficient of Variation md) ekn mi oh aed Ras aa 2640] 5280] 0.44 2543] 400.1] 0231.9), 0.7771 15791 7995] $8126.91 1 330] 3720] 61846.) 38 ee2] 6366.2] Ts24] aa 822.4] 1.26) 402] 1085.4! 3.4321 “404 2245.6) 1.121 378] ‘3024| __77979.5| 1.44] 267| 1068] _78447.5| 0.84! 250] "3775|__92222.5| ‘B17 249] 7466.1] _63601.6! 4.003) 232 648.6] 84341.2) 0.616] 200] 1480! _85821.2 1.256] 136 1251.2] 4] 36) 1.444] 47] 1.939] 39] 1,728) 28] '38677.9) 0.505) 16| 705.6| _08783.5| 7.056] 131 205.4! 98988.01 2.212] 5988.9] 26.5771FRACTION OF TOTAL FLOW CAPACITY (kh) Os 0.6 04 02 %9 0.2 04 0.6 O08 FRACTION OF TOTAL VOLUME (hd) Figure 2.20: Flow capacity distribution. D 1.02-56 While the Dykstra-Parsons and Lorenz coefficients give a quantitative measure of the permeability variation, they provide no information on the spatial relationship between the permeability values. It is well known that permeability and other petrophysical properties of rocks are not randomly distributed but are spatially correlated. Another measure of heterogeneity is the spatial correlation length which is a measure of the distance over which the neighboring permeability values are related to each other. Mathematically, the correlation length can be obtained from the variogram. The variogram is the variance of the differences between the permeabilities at two locations separated by a distance h, and is defined as: y= Varn) hoc Be (247 he ie 2N, where Y variogram or semivariogram h lag distance kx) permeability at location (x) k(x+h) permeability at location (x+h) Nh number of pairs of permeability samples separated by the distance h When the variogram is plotted against the lag distance, it increases with increasing lag distance until a certain distance is reached at which the variogram levels off (Figure 2.21). The lag distance at which the variogram levels off is defined as the correlation length, or the range of influence, and the value of the variogram at this point is called the sill. If the correlation length is zero, the permeability distribution is fully random. With increasing correlation length, the range of influence of one permeability on its neighbors increases up to a distance equal to the2-57 correlation length. The correlation length is a directional quantity, and in general, will be different in different directions. Figures 2.22 and 2.23 show experimental variograms for X-ray attenuation coefficient (CT numbers) of a sandpack in the transverse and longitudinal directions. Figure 2.22 shows that the vertical and horizontal variograms are essentially the same, confirming the radial symmetry observed in the CT images of the sandpack. The variogram in the longitudinal direction (Figure 2.23) has a wavy or sinusoidal characteristic. This is caused by the longitudinal heterogeneity introduced in the sandpack by the packing method. Figures 2.24 and 2.25 show experimental variograms for X-ray attenuation coefficient (CT numbers) of a Berea sandstone in the transverse and longitudinal directions. Figure 2.24 shows a wavy characteristic in the vertical direction (Y-direction). This is caused by the layering in that direction. Figure 2.25 shows that the correlation length in the longitudinal direction (X-direction) is about 40 cm. Clearly, the semivariogram is able to reveal the spatial heterogeneity in the observed quantity. When the permeability variogram is combined with the Dykstra-Parsons coefficient, one obtains a more complete picture of the permeability heterogeneity of the reservoir.‘correlation length or range of influence Variogram Lag Distance Figure 2.21: An ideal semivariogram.SEMIVARIOGRAM 0.0 1.0 2.0 LAG DISTANCE (cm) Figure 2.22: - Experimental semivariograms of CT numbers in the transverse directions of a dry sandpack (Peters and Afzal, 1992).2-60 SEMIVARIOGRAM ° 10 20 30 LAG DISTANCE (cm) Figure 2.23: ~ Experimental semivariogram of T numbers in the longitudinal direction of a dry sandpack (Peters and Afzal, 1992).SEMIVARIOGRAM Figure 2.24: LAG DISTANCE (cm) Experimental semivariograms of CT numbers in the transverse directions of a dry Berea sandstone (Péters and Afzal, 1992).2-62 1000 800 = = oO 600 Q x < 2 400 = iw a 200 ° igre erinicy gassed Ss nicieu ines iste so scuisecEut4oTsiAs LAG DISTANCE (cm) Figure 2.25: Experimental semivariogram of CT numbers in the Jongitudinal direction of a dry Berea sandstone (Peters and Afzal, 1992). os2.3.13 Modeling Permeability Heterogeneity Using techniques of geostatistics, it is now possible to mathematically simulate permeability distributions of a given Dykstra- Parsons coefficient and correlation length. Such a simulation can be used in one of two ways: (1) to generate hypothetical porous media of varied degree and structure of heterogeneity in order to study the effect of permeability heterogeneity on the performance of a process, such as a waterflood, and (2) to generate intervening permeability distributions between wells while at the same time honoring the permeability distributions observed at the wells through core analysis. The second application of permeability modeling is usually referred to as reservoir characterization. Here, we demonstrate the first application of permeability modeling in which the objective was to investigate the effect of permeability heterogeneity on a waterflood performance. Figure 2.26 shows twelve heterogeneous, 2D permeability fields generated to cover a wide range of Dykstra-Parsons coefficient and correlation length using a computer program developed by Yang (1990). Permeability fields were generated at Dykstra-Parsons coefficients (DP) of 0.01, 0.55 and 0.87 and dimensionless correlation lengths of 0, 0.2, 0.7 and 2.0. The correlation length was made dimensionless by dividing the correlation length by the length of the porous medium in that direction. A DP of 0.01 represents a nearly homogeneous medium while a DP of 0.89 represents an extremely heterogeneous medium. Petroleum reservoirs typically have Dykstra- Parsons coefficients that range from 0.5 to 0.9. A dimensionless correlation length of zero represents an uncorrelated or random permeability distribution; a correlation length of 0.2 represents mild correlation; a correlation length of 2.0 represents extremely strong correlation. Depending on the depositional environment, petroleum reservoirs can have widely different correlation lengths. The value of dimensionless correlation length in the y direction (Ly) was constant at 0.2 for the permeability fields shown in Figure 2.26.DP=0.01 DP=0.55 DP =0.87 Lx=0.0 Lx=02 Lx=07 —<<—$=— OC Te RE MI Partai ety tase rman Ta Tarts Gee le ee see we Figure 2.26: Simulated permeability distributions(Gharbi and Peters, 1993). i 99-72-65 Two observations can be made from the permeability maps of Figure 2.26. First, as the correlation length (Lx) increases, the permeability distributions become more and more stratified. The number of layers is inversely proportional to Ly. In fact, as Lx approaches infinity, for Ly = 0.2, the permeability distribution will consist of exactly five (1/0.2) distinct homogeneous layers. Second, with increasing DP at a given Lx, the contrast in the permeability values increases while their spatial arrangements remain similar. Figure 2.27 shows the permeability histograms which indicate that the permeability fields are log-normally distributed in accordance with observations in sedimentary rocks. Figure 2.28 shows the variograms for the twelve permeability fields which give a visual impression of the degree of correlation in each permeability field. Figure 2.29 compares the simulated oil recovery curves for each of the twelve heterogeneous porous media with that of a laboratory waterflood experiment in a homogeneous sandpack. The following observations can be made from these results. If the heterogeneous media are characterized by low variability in the permeability distributions (low DP), the waterflood response will be essentially the same as in the laboratory sandpack regardless of the structure of the heterogeneity. This is indicated by the agreement between the simulated and the experimental recovery curves in the first column of Fig. 2.29 (DP = 0.01). If the heterogeneous media are characterized by low correlation in the permeability distributions (low Lx), the waterflood response will be essentially the same as in the laboratory sandpack regardless of the variability in the permeability distributions. This is indicated by the agreement between the simulated and the experimental recovery curves in the first row of Figure 2.29 (Lx = 0). If the heterogeneous media are characterized by high variability and high correlation in the permeability distributions,- the waterflood response could be significantly different from that of the laboratory sandpack. This is most clearly sown by the response in the last permeability field in Figure 2.29 (Lx = 2 and DP = 0.87). In this case, the waterflood effeciency is significantly less in the heterogeneous medium than in the laboratory sandpack. VSand bao aak 224 206 2a? 6 ia 70 Lag treet Lago reraentuy Lag ot Perms Figure 2.27: Simulated permeability histograms (Gharbi and Peters, 1993). :aor: 109: Seer _amefememeeenenend a] ve S occ, ink _ : ® 10 fae, 4 ; ont ae i. : =e Feo oe » 0 os ° 0 one m0 ue - 3 cae 0 fone Est 7 . | ener uo B une = wo eo i" zee E oom ° we aoe i a0 oom te ws uo ‘ cone is one - Z ome : oo i Bea once: ist awe °. ! 0 01 02 03 o4 as 90 1 02 Dimensientess Lar Distance Dit 2 ¢4 Lag Distance Figure 2.28: Simulated permeability variograms in the x-direction -(Gharbi and Peters, 1993).i Hl i i | I : 7 ==] “ ==)" ==] I. : 7 i+ | Hl I i 5 | | | fj I Al) - ee Yam tat on ate lao Paw tens Itt Figure 2.29: A comparison of the experimental and simulated oil recovery curves (Gharbi and Peters, 1993).2-69 To investigate the reason for the significant disparity in performance between the laboratory waterflood in a relatively homogeneous sandpack and in certain kinds of heterogeneous media, we examine the simulated saturation maps. Figures 2.30 and 2.31 show the simulated water saturation maps for each of the twelve heterogeneous reservoirs at 0.10 and 0.25 pore volumes injected. We see that the displacement in the heterogeneous media with high DP and high Lx are dominated by channeling of the injected water due to the permeability stratification. This results in significant bypassing of the oil in unswept layers and consequent low oil recovery. By contrast, the displacement in the media with low DP are characterized by excellent sweep comparable with that observed in the CT images of the laboratory waterflood experiment. This results in a displacement performance that is comparable to the laboratory waterflood experiment in the sandpack. We conclude from this study that the performance of an enhanced oil recovery (EOR) displacement in a heterogeneous reservoir could be significantly lower than in a laboratory experiment depending on the degree and structure of the heterogeneity of the reservoir. This conclusion points to the need for proper scaling when using the results of laboratory coreflood experiments in relatively homogeneous media to forecast the expected performance of an EOR process in heterogeneous reservoirs. The methodology developed and presented in this study can be used to accomplish this scaling and prevent erroneous performance forecasts.DP=0.01 DP=0.55 DP=087 Lx=00 Lx=20 Sarees Tere was gaat OTe ee TE OEE ace aieaioe Figure 2.30: Simulated water saturation maps at 0.10 pore volume ' . injected (Gharbi and Peters, 1993). OLDP=0.01 DP=0.55 DP =087 Lx=0.2 Lx=0.7 Lx=20 SOT Ter et ne te Figure 2.31:- Simulated water saturation maps at 0.25 pore volume injected (Gharbi and Peters, 1993). 1L-724 Dispersivity When a miscible fluid displaces another in a porous medium, the displacing fluid tends to mix with the displaced fluid. The result is that a mixing or transition zone develops at the front in which the concentration of the injected fluid decreases from one to zero. Experiment shows that the mixing zone grows as the displacement progresses. This mixing and spreading of the injected fluid is known as hydrodynamic dispersion. Bear (1972) describes dispersion as the "macroscopic outcome of actual movement of individual tracer particles through pores...". Essentially, dispersion is the mixing caused by single-phase fluid movement through a porous medium. What is "mixed" is usually called a tracer, but can be thought of as a concentration of any chemical component within a given phase that is transported through the system. Dispersion is the net result of: (a) molecular diffusion, (b) local velocity gradients within given pores, (c) locally heterogeneous streamline lengths and velocities, and (d) mechanical mixing in pore bodies. Dispersion has practical consequences in contaminant transport in aquifers and in improved oil recovery from petroleum reservoirs. If a miscible contaminant is accidentally introduced into an aquifer at a site, dispersion will cause the contaminant to spread to a larger area as it is being transported by groundwater flow. Eventhough the concentration of the contaminant is reduced by dispersion, a much larger area of the aquifer will become contaminated as a result of dispersion than the original spill area. Thus, a much larger area than the original spill will need to be cleaned up by any contaminant remediation measure. Miscible displacement is the most efficient improved oil recovery method. Because there is no capillary force to trap the displaced oil, it is theoretically possible to recover 100% of the oil by miscible displacement. However, because the injected solvent is usually more expensive than the oil that is to be displaced, it is usually injected in small quantities as slugs2-73 and chased by a less expensive fluid such as water or gas. Dispersion will dilute and reduce the effectiveness of the miscible slug as it is propagated through the reservoir. In this case, dispersion is detrimental to the recovery process. Dispersion is characterized quantitatively by the dispersion coefficient which is measured by relating the results of mixing to an equation called the convection-dispersion equation: aC, yw 8 _ DL eC = =-sbs=0 (2.48) ot if OR Ox Re ax? (2.48) where, C = tracer concentration juid flux, q/A longitudinal dispersion coefficient, L2/t R,= a retardation factor (2 1) that accounts for adsorption of the tracer by the porous medium. 9 = average porosity of the porous medium The solution to this equation for the appropriate set of initial and boundary conditions that describe a displacement experiment is a concentration profile that moves in time and space such as shown in Fig. 2.32. This profile is usually analyzed by measuring the effluent concentration versus time at the outlet end of the core (breakthrough curve). The solution to Eq. 2.48 for the case of no adsorption (R¢ = 1) and continuous tracer injection used to estimate the longitudinal dispersion coefficient is: =U te eal ; Cp{xpb) alee Pe| iD 2.49) where, Cp= dimensionless concentration xD = dimensionless lengthDimensionless concentration (Cy) Mixing zone at ty 0.5 0.5 0.25 0.1 0 02 0.4 06 08 Dimensiontess distance (xp) Figure 2.32: Tracer concentration profiles for a dispersive system. 1.0tD = dimensionless time, (pore volumes injected) NPe = Peclet number, dimensionless The dimensionless variables are defined as follows: Chet) - Cp2> pe GG xp = ode LAE: where C, = initial tracer concentration injected tracer concentration L= length of the porous medium q = volumetric injection rate A = system cross-sectional area of the porous medium t = time of injection 9 = average porosity of the porous medium v = intersitial velocity (2.50) (2.51) (2.52) (2.53)The complementary error function, erfc, is related to the integral of the normal (bell-shaped) distribution function and is given by: 2 erfc(z) = af edu (2.54) Let a mixing zone length be defined as the distance between two symmetric concentration points, for example, CD = 0.1 and 0.9. For such a zone, it can be shown that the mixing zone length in dimensionless form is given by: Axp = 3.625,/$D- (2.55) ue Npe orin dimensional form by: Ax = 3.625VDit (2.56) Eqs. 2.55 and 2.56 show that the mixing zone grows in proportion to the square root of time and in proportion to the square root of the dispersion coefficient. Thus, by measuring the length of the mixing zone as a function of time, Eq. 2.56 can be used to calculate the longitudinal dispersion coefficient for the porous medium. The length of the mixing zone can easily be measured by imaging the experiment (Peters et al., 1994). The dispersion coefficient can be determined from a breakthrough 1+) ‘D curve lotting CD versus on probability paper. If the system is plotting Yip papel y' homogeneously dispersive, such a plot will give a straight line. The ‘ 1D : 3.625 difference between——= at Cp = 0.1 and Cp = 0.9 is equal to , the vio \NPe dimensionless mixing zone length at tp=1. Thus:2-77 3.825 = (Jo9—Jo1) (2.57) ‘Ne Jos - Jos where Jos = [(1 - tp¥Vtp] for Cp = 0.9 Joa = [= to¥Vtp] for Cp = 0.1 Eq. 2.57 can be rearranged to calculate the longitudinal dispersion coefficient as = vi fJoa=Jos\? DiL=VvI 3,625 ) (2.58) Alternatively, the longitudinal dispersion coefficient can be determined from the probability plot of the breakthrough curve as Di = (YE foss — Jose)” (2.59) where Joss = [(1 - toYVtp] for Cp = 0.84 Jos = [C1 - tpt] for Cp = 0.16‘The dispersion coefficient has been found to be a function of fluid velocity as follows: " Diep+ Ye) (2.60) Do Do where, v = average interstitial velocity, 7 Do = binary diffusion coefficient, L2/t Dp = mean grain diameter, L a, B, y= properties of a given medium Eq. 2.60 relates a dimensionless dispersion coefficient to the Peclet number defined in terms of the mean grain diameter and binary diffusion coefficient without a porous medium. The first term on the right side of Eq. 2.60 is the molecular diffusion term, whereas the second term is the mechanical dispersion term. The molecular diffusion term, B, has been found to be 1/6F, where F is the electrical formation factor of the porous medium. Figure 233 shows the correlation between dimensionless longitudinal dispersion coefficient and Peclet number obtained experimentally. At Peclet numbers less than 0.02, molecular diffusion dominates the dispersion and the dispersion coefficient is equal to the diffusion coefficient in the porous medium. In this region, Figure 2.33 shows that the the dimensionless dispersion coefficient (B) is about 0.67. This means that the diffusion coefficient for a tracer in a porous medium is less than the diffusion coefficient in a.liquid without a porous medium. This difference is caused by tortousity. In fact, 1//0.67 (="1.5) is an estimate of the tortousity of clastic rocks. At Peclet numbers between 0.02 and 6, both molecular diffusion and mechanical or convective dispersion contribute to the dispersion coefficient. At Peclet numbers above 6, the2-79 Diffusion controls Figure 2.33: Correlation between longitudinal dispersion coefficient and Peclet number (Perkins and Johnston, 1963).2-80 dispersion coefficient is dominated by mechanical or convective dispersion and the effect of molecular diffusion can be neglected. In most rocks and at normal reservoir velocities, the Peclet number is normally greater than 6. The convective term dominates the dispersion coefficient. Also, vis usually close to one. Under these conditions, Eq. 2.60 can be written as Dy = av (2.61) where O, = dispersivity of the porous medium, L v = interstitial velocity, L/t Eq, 2.61 can be used to estimate the longitudinal dispersivity from the dispersion coefficient. In terms of the dispersivity, the Peclet number defined by Eq. 2.53 becomes sk Npe = ge (2.62) The longitudinal dispersivity is, in general, not a function of fluid velocity, making it a primary petrophysical property. However, dispersivities are highly scale-dependent. Figure 2.34 shows values of measured dispersivity as a function of measurement length. Note the logarithmic scale. Reservoir heterogeneity is the cause of this scale effect. Transverse dispersion coefficient varies with Peclet number in the same manner as longitudinal dispersion coefficient. | However, transverse dispersion coefficient is generally lower than the longitudinal dispersion coefficient. Figure 2.35 compares transverse and longitudinal dispersion coefficients as functions of Peclet number.isperivity tm) wot 107 : 98 oo xo x o is eee a C wo o 8 L £ Bot 8 2% «PF ow ie" ah : o #8 xo , we, [ oe E ss co © & tattemand- Barter end Pascuscect (1978) erg cierea mnereee E o sf G Gethar et a. (1985) q © can cary, 188 3 Oo moo ° apo0 0 2-81Di Dr Do'Do 5 | 103 10? eV 101 109 103 10 1077 110102108 30% 108108 Figure 2.35: A comparison of transverse and longitudinal dispersion coefficients. ae3. MULTI-PHASE ROCK/FLUID PROPERTIES The pore space in a petroleum reservoir rock is usually occupied by more than one fluid. In an oil reservoir, water and oil occupy the pore space. In a gas reservoir, water and gas occupy the pore space. In some oil reservoirs, at some stage of depletion, water, oil and gas may occupy the pore space. When more than one fluid occupies a porous medium, a new set of problems arise. The amount of each fluid in the porous medium needs to be determined. Interfacial forces (surface forces) between the immiscible fluids and between the fluids and the rock surface come into play. Because the pore space is of capillary dimensions, capillarity plays a role by separating phases into different pores, giving rise to differences in fluid pressure between the phases (capillary pressure) and differences in the flow capacity (relative permeability) of the rock and fluids. Capillarity also ensures that an immiscible displacement can never be complete. There is always a residual saturation of the displaced fluid that is trapped by capillarity. 3.1 FLUID SATURATIONS The fraction of the pore space occupied by a fluid is the fluid saturation for that fluid. Thus, in general, Fluid Saturation = @.1) —Volume of Fluid_ Rock Pore Volume If Sw = water saturation, So = oil saturattion and Sg = gas saturation, then Sw = Vw/Vp, So = Vo/Vp, and Sg = Vg/Vp, where Vw, Vo, Vg and Vp are the volumes of water, oil, gas and pore space, respectively. For an oil reservoir without a free gas saturation, So + Sw = 1.0. For a gas reservoir without a liquid hydrocarbon saturation, Sg +Sw = 1.0. For an oil reservoir with a free gas saturation, Sp + Sw + Sg = 1.0. Fluid saturation may also be expressed in %. There are two methods of determining the in-situ fluid saturations ina petroleum reservoir rock. The direct approach is to measure the fluid saturations from a core cut from the reservoir. The indirect approach is to measure some other physical property of the rock that can be telated to fluid saturation. The direct approach is discussed here. The indirect approach such as using electric logs or capillary pressure Measurements to estimate water saturation will be discussed later. One method of direct measurement of fluid saturations is the retort method. In this method, a core sample is heated so as to vaporize the water and oil, which are condensed and collected in a small, graduated receiving vessel (Figure 3.1). The volumes of oil and water divided by the pore volume of the core sample give the oil and water saturations. The gas saturation is obtained indirectly by the requirement that saturations must sum to one. There are two disadvantages to the retort method of saturation determination. In order to remove all the oil, it is necessary to heat the core to temperatures in the range of 1000 to 1100 F. At these temperatures, the water of crystallization (hydration) of the rock is driven off, resulting in an estimated water saturation that is higher than the true interstitial (connate) water saturation. The second disadvantage is that the oil when heated to high temperatures has a tendency to crack and coke. This cracking and coking tend to reduce the oil volume resulting in an oil saturation that is less than the true value. Corrections can be made to the retort measurements to make them more accurate.3-3 Thecmecouple a 2 = Sara A? Re oe ? PE x : hs sveroo L per iS Car mm] Nr) fe} [ev S = / — i Water inter y tl i" — Receiving tube. eee = J Figure 3.1: Retort distillation apparatus.a4 Another method of direct saturation measurement is by extraction with a solvent. This is accomplished in a Dean-Stark distillation apparatus (Figure 3.2). The core is placed in the apparatus in such a way that the vapor from a solvent (e.g., toluene) rises through the core and is condensed back over the sample. This process leaches out the oil and water from the sample. The water and solvent are condensed and trapped in a graduated receiver. The water settles to the bottom of the receiver while the solvent refluxes back into the main heating vessel. The extraction is continued until no more water is collected in the receiving vessel. The water saturation is calculated directly from the volume of water expelled from the sample. The oil saturation is calculated indirectly from the weight of the saturated sample before distillation, the weight of the dry sample after distillation and the weight of the water expelled from the sample. Again, the gas saturation is calculated indirectly from the requirement that the saturations must sum to one. To ensure that all the oil has been removed from the sample, the sample may be transferred from the Dean-Stark apparatus to a Soxhlet extractor for further extraction (Figure 3.3). The Soxhlet extractor is similar to the Dean-Stark apparatus except that there is no provision for trapping the extracted liquids. The saturations determined by direct measurements on cores should be treated with caution because they may not represent the in-situ fluid saturations for several reasons. If the core was cut with a water-based drilling mud, it would have been flushed by mud filtrate resulting in a higher water saturation than the connate water saturation. The measured oil saturation in this case would be the residual oil saturation after waterflooding which is less than the original in-situ oil saturation. If the core was cut with an oil-based mud, the water saturation obtained by direct measurement will be essentially the correct corindte water saturation.<— Cooling, Liquid Figure 3.2: Dean-Stark apparatus.3-6 Condenser Electric heater Figure 3.3: Soxhlet extractor.3-7 As the core is brought from the high pressure and temperature of the reservoir to the low pressure and temperature of the laboratory, changes occur in the fluid saturations which can make them considerably different from the original in-situ saturations. The free gas will expand, expelling water and oil in the process. Solution gas will evolve from the oil, expand and further reduce the oil and water volumes. These changes cause the saturations determined by core analysis to be different from the in-situ saturations. Although the saturations determined by direct measurements on cores may not reflect the true in-situ saturations, they do provide useful information about the reservoir. The saturation measurements can be used to approximately locate the gas-oil and water-oil contacts in the reservoir if they are present. Fluid saturations, So, Sw and Sg, only tell us the proportion of each fluid type in the pore space. They do not tell us how the fluids are distributed in the rock. To determine the fluid distribution, we need to consider the interfacial forces and phenomena that arise when immiscible fluids are confined in reservoir pores of capillary dimensions. The important interfacial forces and phenomena include surface tension, interfacial tension, wettability, capillarity and capillary pressure . 3.2. SURFACE AND INTERFACIAL TENSIONS 3.2.1 Surface Tension Surface tension is the contractile force that exists at the interface of a liquid and its vapor (or air). Surface tension makes the surface of a liquid drop act like a membrane. The force is caused by unequal molecular attractions of the fluid particles at the surface as shown in Figure 3.4. The force per unit length (6 = Force/Length) tending to contract the surface of a liquid is a measure of the surface tension of the liquid. It is a property of the liquid and is usually expressed in units of dynes/cm.3-8 ‘SURFACE FILM ‘SURFACE MOLECULES MENISCUS a ‘MOLECULES INTERNAL MOLECULES. PULLED IN ALL DIRECTIONS OL, Figure 3.4: Apparent surface film caused by unequal attraction of surface molecules of a liquidIf the forces acting on a molecule at the surface or interface are different from those acting on a molecule in the body of the liquid, a new interface can only be created if work is done. It can be shown (e.g., by soap film experiment) that the reversible work required to create a new interface is given by : w=-odA G2) where w = reversible work required to create a new area = surface (interfacial) tension dA = increase in area The reversible work shown in Eq.(3.2) corresponds to a free energy quantity. It is apparent from Eq.(3.2) that surface (interfacial) tension can be viewed as free surface energy per unit area with units of erg per square centimeter. Because a system at equilibrium minimizes its free energy, liquids at equilibrium tend to minimize their surface area. A liquid breaks up into spherical drops because a sphere has the smallest surface area per volume. The surface tension of pure water at 70 °F is 72.5 dynes/cm, and at 200 °F is 60.1 dynes/cm. The surface tensions of crude oils at 70 °F range from 24 to 38 dynes/cm. High temperatures and dissolved gas both tend to reduce the surface tension of crude oils. Values on the order of i dyne/cm may be expected at temperatures and pressures exceeding 150 °F and 3,000 psig. Table 3.1 shows the surface tensions of some selected liquids.Table 3.1 Surface Tension of Pure Liquids Liquid TCC) 6 (dynes/cm) Water 20 728 Water 25 72.0 Benzene 20 28.88 Benzene 25 28.22 Toluene 20 28.43 Carbon tetrachloride 20 26.9 Ethanol 20 22.39 n-Octane 20 21.8 Ethyl ether 20 17.01 Factors that affect the surface tension of a liquid include pressure, temperature and solute concentration. An increase in pressure leads to a reduction in the surface tension of a liquid. This is because pressure exerts a compressive force on the surface which reduces the tensile or contractile tendency of the surface. An increase in temperature leads toa reduction in the surface tension of a liquid. The increase in temperature causes increased randomness at the surface which leads to an increase in the surface entropy. It can be shown from thermodynamics that the surface entropy is given by s &I, G3) where SS is the surface entropy and T is the temperature. It is clear from Eq.(3.3) that surface tension decreases with temperature. Figure 3.5 shows the variation of surface tensions of hydrocarbons with temperature.3-11 35 8 ‘Suefoce tension, dyes per em 200-300-400. S00 600 200-109 ° 100 Temperature, 609 F Figure 3.6: Variation of surface tensions of hydrocarbons with temperature. sanThe effect of solute concentration on the surface tension of a liquid depends on the liquid and the nature of the solute. Four general cases may be identified. 1. Liquids having fairly close values of surface tension. Generally, the surface tension of the mixture varies approximately linearly with the composition. For example, for a mixture of acetone and chloroform at 18° C, the surface tension increases linearly with composition (mole % chloroform) from 22 dynes/cm for pure acetone to 27 dynes/cm for pure chloroform. Liquids having widely different values of surface tension. In general, the suface tension of a liquid is reduced substantially by addition of a liquid of lower surface tension, but is only slightly increased by addition of a liquid of higher surface tension. For example, addition of ethanol to water causes a rapid reduction in the surface tension of water with ethanol concentration. However, addition of water to benzene raises its surface tension from 28.2 to only 29.3 dynes/cm. Solutions of inorganic electrolytes. In general, the surface tension increases with solute concentration. For example, the surface tension of water at 20° C will be increased by addition of sodium chloride from 72.8 dynes /cm to 80 dynes/cm at a concentration of 5 moles of sodium chloride per liter of solution. Solutions of colloidal (long chain ) electrolytes. In general, the surface tension decreases with solute concentration but is followed by a region over which the surface tension is virtually unchanged by solute concentration. For example, the surface tension of water at 25°C will be reduced by addition of sodium lauryl sulfaté from 72 dynes/cm to 40 dynes/cm at a concentration of 0.01 moles per liter of solution. The surface tension remains constant above this concentraion.A useful empirical relation often used to calculate surface tension is through the concept of the parachor. The parachor for a pure substance is defined as A= Move G4) PL-Pg where A= parachor M= molecular weight of the liquid o = surface tension of the liquid, dynes/cm py = saturated liquid density , gm/cm3 Pg = Saturated vapor density, gm/cm3 Parachor posseses definite values for specific atoms and structures. Parachors are predicted from the structure of the molecules or can be calculated for pure substances and mixtures from surface tension measurements at atmospheric pressure. The parachors for pure substances are given in Table 3.2. Correlations for parachors with molecular weight are shown in Figures 3.7 and 3.8. The saturated liquid and vapor densities for various liquids are given in Figure 3.9.Table 3.2 Parachors for Computing Surface and Interfacial Tensions (Katz et al., 1959) Constituent Parachor Methane 77.0 Ethane 108.0 Propane 150.3 i-Butane 181.5 n-Butane 181.5 i-Pentane 225.0 n-Pentane 231.5 n-Hexane 271.0 n-Heptane 312.5 n-Octane 351.5 Hydrogen 34 (approx) Nitrogen 41 (approx) Carbon dioxide 78600 7 ¢ 500 L 7 Zz 400 a 3 § 300 5 é 200 100 o 50 100 150 200 Molecular weight Figure 3.7: Parachors for computing interfacial tension of normal paraffin hydrocarbons (Katz et al., 1959).ont a 3 1200 * 1000 = 2 © 800 8 s = E 600 400 200 ° 100 200 300 400 500 7 Molecular weight Figure 3.8: Parachors of heavy fractions for computing interfacial tension of reservoir liquids (Firoozabadi et al., 1988).Density, g/ee ° =o (0 100 200=—« 300 400. S00.« «600.700 Temperature, *F Figure 3.9: Saturated liquid and vapor densities of various siibstances.3-18 Equation (3.4) can be rearranged to calculate the surface tension as o=(ae=Paf 65) 3.2.2 Interfacial Tension Interfacial tension is the contractile force per unit length that exists at the interface of two immiscible fluids, e.g., oil and water (Figure 3.10). The forces acting on the surface molecules are similar to those in the liquid-vapor system, but the mutual attraction of unlike molecules across the interface becomes important. The free energy required to create a fresh interface is referred to as the excess interfacial free energy. The specific excess interfacial free energy is dimensionally equivalent and is numerically equal to the interfacial tension. Like surface tension, the unit of measurement of interfacial tension is dynes/cm (or ergs/cm2). Consider butanol (C4HgOH) with a surface tension of 24 dynes/cm in contact with water with a surface tension of 72 dynes/cm. What will be the interfacial tension between the two liquids? For this system, the interfacial tension is 1.8 dynes/cm, a fairly low number considering the surface tensions of water and butanol. The low interfacial tension indicates that the molecules of butanol must concentrate at the interface, decreasing the contractile tendency of the interface. Interfacial orientation of the butanol is favored because the hydrocarbon chain in the butanol is hydrophobic. This example shows that there is no simple relationship between the surface tensions of liquids and their interfacial tensions. Table 3.2 lists accurately known interfacial tensions for various organic liquids against water.‘SURFACE FILM ‘SURFACE MOLECULES PULLED TOWARD NENSCUS LIQUD CAUSES: TENSION IH SURFACE WOLECULES INTERNAL MOLECULES WATE! PULLED IN ALL }DIRECTIONS Figure 3.10: Apparent surface film caused by unequal attraction of molecules at the interface of two liquidsTable 3.2 Interfacial Tensions between Water and Pure Liquids Liquid TCC) @ (dynes/cm) n-Hexane 20 51.0 n-Octane 20 50.8 Carbon disulphide 20 48.0 Carbon tetrachloride 20 45.1 Carbon tetrachloride 23 43.7 Bromobenzene 25 38.1 Benzene 20 35.0 Benzene 2 34.71 Nitrobenzene 20 26.0 Ethyl ether 20 10.7 n-Octanol 20 8.5 n-Hexanol 23 6.8 Aniline 20 5.85 n-Pentanol 23 44 Ethyl acetate 30 29 Isobutanol 20 21 n-Butanol 20 18 n-Butanol 2 1.6 The presence of a third component of the right kind can significantly reduce the interfacial tension between two liquids. For example, the interfacial tension of water and iso-pentanol is 4.4 dynes/cm. If ethanol is added to the system, the ethanol molecules will adsorb at the interface, thereby reducing the interfacial tension between the water and the iso- pentanol. When 25% by weight of ethanol is added, the interfacial tension is reduced to zero. The system then becomes miscible and forms a single phase. Chemicals that adsorp at interfaces are usually referred to as surface active agents or surfactants. Such chemicals form a monolayer at3-21 the interface. The monolayer is in a state of compression which reduces the contractile tendency of the interface, thereby reducing the interfacial tension. The presence of the adsorbed molecules creates a surface or spreading pressure which reduces the interfacial tension. In fact, o=0,-l1 3.6) where o = reduced interfacial intension Oo = original interfacial tension TI = spreading pressure Surfactants are often used to reduce the interfacial tensions between oil and water in order to improve oil recovery. The interfacial tensions between reservoir water and crude oils have been measured for a number of reservoirs and found to range from 15 to 35 dynes/cm at 70 °F, 8 to 25 dynes/cm at 100 °F, and 8 to 19 dynes/cm at 130 °F. Table 3.3 presents the results of interfacial tension measurements for some fluid pairs. Table 3.3 Typical Interfacial Tensions and Contact Angles for Fluid Pairs (Archer and Wall, 1986) Contact Interfacial Wetting Non-Wetting Angle Tension Phase Phase Conditions oO (@ynes/cm) Brine ou Reservoir 30 30 Brine Oil Laboratory 30 48 Brine Gas Laboratory 0 2 Brine Gas Reservoir 0 50 oi Gas Reservoir 0 4 Gas ‘Mercury Laboratory 140 480 The interfacial tension between reservoir oil and gas can be estimated using parachors as3-22 Beet ie where o = interfacial tension, dynes/cm ‘Aj = parachor of component i xj = mole fraction of component i in the liquid Mj = molecular weight of the liquid pi = saturated liquid density , gm/cm3 yi = mole fraction of component i in the gas Mg = molecular weight of the gas N = total number of components in the mixture p, = saturated gas density, g/cm? Many reservoir phenomena depend on the interfacial tensions between the reservoir fluids and between the reservoir fluids and reservoir rock. Interfacial tensions of reservoir water and oil can be reduced significantly by the addition of surface active agents to either the oil or the water. Some of these surface active agents occur naturally in crude oils. 3.2.3 Measurements of Surface and Interfacial Tensions Several techniques are used to measure surface and interfacial tensions. These include (1) capillary rise method, (2) sessile drop method, @) pendant drop method, (4) ring method and (5) spinning drop method. When a capillary tube is dipped into a wetting liquid, the liquid is spontaneously imbibed (sucked) into the capillary tube (Figure 3.11). The equilibrium height is determined by a balance between the ‘capillary suction force and the pull of gravity. For a circular capillary tube, the capillary suction force is given by 2xrocos@ whereas the gravity pull is given by mr2h(p- Pnw)g- Equating the two forces leads to3-23 oa ThOw- Paws 8) 2cos® where o = interfacial tension between the wetting and nonwetting phases, dynes/em 1 = radius of the capillary tube, cm h = equilibrium height of the wetting phase, cm Pw = density of the wetting phase , gm/cm3 Pnw = density of the nonwetting phase, gm/cm3 @ = the contact angle § = gravitational acceleration, 981 cm/s? In the case of a wetting liquid and air, the density of the air (nonwetting phase) can be neglected. Furthermore, if the liquid fully wets the solid, @ is zero and cos@ is one. Under these conditions, Eq. (3.8) simplifies to (3.9) Equation (3.9) can be used to calculate the surface tension of the wetting liquid from a capillary rise experiment.3-24 Nonwetting Phase Figure 3.11: Capillary rise experiment.The sessile drop method of determining the surface tension of a liquid consists of measuring the number of liquid drops falling from the capillary end of the instrument while the surface of the liquid within the bulb is lowered from the upper to the lower mark (Figure 3.12). The principle of the method is based on the fact that the size of the liquid drop is proportional to surface tension of the liquid. The size of the drop is reached when the surface tension can no longer support its weight. To a first approximation, Widea = 2010 (3.10) where Wideal = weight of the drop that should fall r= external radius of the tube o = surface tension Figure 3.13 shows the sequence of shapes for a drop that detaches from a tip. The detached drop leaves behind some liquid residue. Thus, the actual weight of the drop which is what is measured is less than the ideal weight. To account for this, Eq.(3.10) is modified as Wactual = Wideatf = 2nrof 3.1) where f is a correction factor which can be expressed as a function of 1/V1/3, where V is the volume of the drop. Table 3.4 shows the correction factors for various r/V1/3.Figure 3.12: Sessile drop method of measuring surface tension. vUBETE Figure 3.13: Sequence of shapes for a drop.ve £ we 1 fey vESSESEELELH LL 3E3UCLEGLELU Cee eee 0.30 0.7256 0.80 0.6000 0.40 0.6828 0.90 0.5998 0.50 0.6515 1.00 0.6098 0.60 0.6250 1.10 0.6280 0.70 0.6093 1.20 0.6535 ena The pendant-drop method of measuring surface or interfacial tension depends only on the density of the fluids and the dimensions of the drop. Figure 3.14 shows a pendant drop and the relevant dimensions. The surface or interfacial tension is given by of scile.—ed G12) where 6 = surface tension de = maximum diameter of droplet Px = density of the liquid Pg = density of vapor H= constant, a function of de/ds aa & = gravitational accelerationGe Figure 3.14: Surface tension by pendant drop method. 3-283-29 The constant H is tabulated as a function of de/ds. The pendant drop method can be used to measure surface tension or interfacial tension. It can also be adapted for measurements at elevated temperature and pressure. The ring method of determining surface or interfacial tension depends on measuring the force required to pull the ring free of the interface (Figure 3.15). Theoretically, the surface or interfacial tension is given by aE o aL 3.13) where 6 = surface or interfacial tension F = force required to pull the ring free of the interface L =is the circumference of the ring The factor of 2 accounts for the fact that there are two surfaces around the ring. In practice, corrections are needed to account for the mass of liquid lifted by the ring in breaking through the interface (Figure 3.16). Such corrections are made available with the instrument. Figure 3.17 shows a typical instrument, known as the du Nouy tensiometer, that employs the ring method for surface or interfacial tension determination.Section 4-4 Figure 3.15: Surface tension by ring method.Figure 3.17: du Nouy tensiometer.3-33 The spinning drop method of determining surface or interfacial tension is based on measuring the shape of a drop of liquid or gas bubble in a more dense liquid contained in a rotating horizontal tube. The drop or bubble is typically rotated at speeds of 1,200 to 24,000 revolutions per minute (RPM). Under rotation, the original spherical drop or bubble becomes elongated into a cylindrical shape (Figure 3.18). A strobe light is used to visualize the deformed drop. A microscope is used to measure the diameter of the drop. The interfacial tension is given by o=Lap uA @.14) where 6 = surface or interfacial tension, dynes/cm Ap = density difference between the two fluids, gm/cm3 = angular velocity, radians/s indrical radius of the drop, cm r= An instrument based on the spinning drop method has been designed and patented at the University of Texas at Austin by Schechter and Wade (Figure 3.19). The instrument is manufactured and sold by the Chemistry Department at the University of Texas. The instrument is particularly suited for measuring low interfacial tensions and is therefore used extensively in surfactant research. Interfacial tensions as low as 10-6 dyne/cm have been successfully measured with the instrument.Figure 3.18: Cylindrical liquid drops in a spinning drop apparatus. (A) benzene-water system at 20,000 RPM, (B) octane- surfactant system at 6,000 RPM (Cayias et al., 1975).3-35 ata ‘MOTOR aed CONTROL AMPLIFIER aoe ROTOR ass'Y Frequency TEMPERATURE counter CONTROLLER. a Cane Stonat stRose Generator surecy Figure 3.19: Schematic of spinning drop tensiometer (Cayias et al., 1975).3.3 WETTABILITY Wettability is a tendency for one fluid to spread on or adhere to a solid surface in the presence of other immiscible fluids. The fluid that spreads or adheres to the surface is known as the wetting fluid. In a petroleum reservoir, the solid surface is the reservoir rock which may be sandstone, limestone, or dolomite, together with cementing material. The fluids are water, oil and gas. Normally, either water or oil is the wetting phase. Gas is always a nonwetting phase. Consider the water-oil-solid system shown in Figure 3.20. Three interfacial tensions (specific free surface energies) arise: Gos is the solid- oil interfacial tension, Gow is the oil-water interfacial tension and ows is the water-solid interfacial tension. The angle @ is known as the contact angle and is measured through the water (the more dense fluid). The contact angle is frequently used as a measure of the wettability of the solid. At equilibrium, the interfacial tensions are related by the Young - Dupre equation obtained by considering horizontal equilibrium of the point of contact of the interfacial tensions. For Figure 3.20, this equation is given by Gos — Gws= Gow COS 3.15) The free surface energies for the oil-solid and water-solid interfaces cannot be measured readily. However, their difference can be determined by measuring the oil-water interfacial tension and the contact angle. This difference controls the movement of the interface before equilibrium is achieved. The following may be deduced from Eq. (3.15):ROCK SURFACE Figure 3.20: Interfacial tensions in a water-oil-solid system. 3-371. If the free surface energies for the oil-solid (60s) and the water- solid (6ws) interfaces are equal, the left side of Eq.(3.15) is zero. Since the oil-water interfacial (Gow) tension is nonzero, cos@ on the right side of Eq.(3.15) must be zero, giving a contact angle of 90°. A contact angle of 90° means that solid has no preferential wettability for the oil or the water. This is a situation of neutral or intermediate wettability. 2. If GwsOos, then @ > 90°. The solid is said to be preferentially oil wet. 4. Complete spreading of the oil on the surface takes place if @ = 180° and complete spreading of water on the surface takes place if @ = 0°. Complete spreading of crude oil or water on a surface has never been observed with reservoir fluids. Figure 3.21 shows the equilibrium contact angles for a preferentially water wet and a preferentially oil wet system.WATER-WET OIL-WET Figure 3.21: Equilibrium contact angles for a preferentially water wet and a preferentially oil wet system.eau 3.3.1 Determination of Wettability Reservoir wettability is usually determined either by contact angle measurement using reservoir fluids and a pure mineral surface or by an imbibition test on a reservoir core sample using refined oil and a synthetic brine. No wettability determination method involves the simultaneous use of reservoir fluids and reservoir rock. Contact angle is one of the earliest and still most widely used method to evaluate reservoir wettability. The contact angle measurement essentially seeks to establish whether or not the reservoir oil contains surface active agents that could make an originally preferentially water wet mineral surface become preferentially oil wet over time. Accordingly, the contact angle test uses reservoir oil and brine and a pure, clean mineral surface which is known to be preferentially water wet at the outset. In the absence of reservoir brine, synthetic brine is used in the test since the surface active fluid is the oil and not the brine. The solids normally used to represent a sandstone reservoir, a limestone reservoir and a dolomite reservoir are pure quartz (silica), pure calcite and pure dolomite crystal. These pure minerals are known to be preferentially water wet initially. If the oil contains surface active agents, then these will adsorp on the mineral surface over time and increase the degree of oil wetness. This change in the wettability of the surface can be observed and quantified by measuring the contact angle over time until an equilibrium contact angle is obtained. It is reasonable to assume that a similar wetting equilibrium will be approached in the reservoir. The contact angle measurement is performed with a contact angle cell using an-instrument known as a goniometer. The mineral surface is immersed in the brine (or oil) and allowed to equilibrate. A drop’of the oil (or brine) is then introduced with a hypodermic syringe (Figure 3.22). The contact angle is then measured over time. The test can last several weeks depending on the time required to achieve adsorption equilibrium. The3-41 equipment can be adapted for contact angle measurements at elevated pressures and temperatures. Two contact angles are normally measured: the advancing and receding contact angles. The advancing contact angle is the contact angle obtained when water comes into equilibrium with a surface previously in contact with an oil (Figure 3.23). The receding contact angle is the contact angle obtained when oil comes into equilibrium with a surface previously in contact with water. The advancing contact angle is always greater than the receding contact angle. Normally, it is the advancing contact angle that is reported as the contact angle in a wettability test. Figure 3.24 shows the results of a contact angle test. Several interesting observations can be made. The early time contact angle measurements showed the solid to be preferentially water wet. However, as time passed, the degree of water wetness diminished. Eventually, after adsorption equilibrium was achieved the solid was found to be preferentially oil wet. Had the contact angle test been terminated prematurely, the wettability assessment would have been wrong. Note that for this test, over 30 days of aging were needed to establish adsorption equilibrium.o's = Tw's + Tom COSO Figure 3.22: Contact angle cell.a. Advancing angle b. Receding angle Figure 3.23: Advancing and receding contact angles. °: EQUILIBRIUM een CONTACT ANGLE WATER-WET CONTACT ANGLE, DEGREES 8 1 ! ' t 1 ' ' ' 1 ' t 1 ' ! 1 1 ! ' 10 100 1000 10,000 OIL-SOLID INTERFACE AGE, HOURS Figure 3.24: Approach to equilibrium contact angle.3-45 The major advantages of contact angle measurements are the reliability of the results and the relative ease of obtaining uncontaminated reservoir fluid samples compared to uncontaminated reservoir rock samples. The following disadvantages should be noted. (1) Contact angle is measured on a flat, clean, homogeneous mineral surface. Such a surface does not exist in the reservoir. (2) Pure minerals are used in the test to simulate a sandstone, a limestone and a dolomite. Pure minerals may not be reprentative of actual reservoir mineralogy. (3) The test can be very long and requires extreme cleanliness and inertness of the test system. (4) There is evidence that the contact angle is affected by which fluid was first in contact with the solid. In addition to the contact angle, other measures of wettability have been proposed. One is the wettability index proposed by Amott (1959). The Amott wettability index is obtained by a combined imbibition- displacement test on a reservoir core sample using a refined oil and a synthetic brine. After the reservoir core sample has been flushed with brine to a residual oil saturation and evacuated to remove gas, it is then subjected to the following tests: 1. The core is immersed in oil (e.g., kerosene) and the volume of brine displaced by the imbibition of oil is measured after 20 hours in an imbibition cell (Figure 3.25). 2. The core is centrifuged under kerosene and the additional brine displaced by centrifuging is measured. 3. The core is immersed in brine and the volume of oil displaced by the imbibition of brine is measured after 20 hours. 4. The core is centrifuged under brine and the additional oil displaced by centrifuging is measured.Ground glass joint To water reservoir Figure 3.25: Lmbibition cell.The wettability indices of water (WIw) and oil (WIp) are calculated as follows: Ww= = Volume of oil displaced by brine imbibition aa 'W™ Volume of oil displaced by brine imbibition and by displacement ~~” Wlo= ‘Volume of brine displaced by oil imbibition oa Volume of brine displaced by oil imbibition and by displacement If the rock is preferentially water-wet, WIo will be zero and the greater the degree of water wetness, the closer will Wlw be to 1. Similarly, if the rock is preferentially oil wet, Wlw will be zero and the greater the degree of oil wetness, the closer will WIo be to 1. For a rock of intermediate or neutral wettability, Wlw and WI will be zero or close to zero. Sometimes, the difference (Wlw - Wlg) is used to characterize the wettability. In this case, +1 indicates a strongly water-wet rock whereas -1 indicates a strongly oil-wet rock. The Amott wettability index is a reliable measure of the wettability of the core sample. However, the wettability of the core sample may not be representative of the wettability of the reservoir rock because of the difficulty of obtaining an unaltered core sample. The wettability of the core sample may easily be altered by the coring operation. 3.3.2 Wettability of Petroleum Reservoirs The wettabilities of petroleum reservoirs span the entire spectrum from preferentially water wet to preferentially oil wet reservoirs. Treibel et al. (1972) measured the wettabilities of 30 sandstone and 25 carbonate reservoirs by measuring contact angles at the reservoir temperatures using the reservoir oils and a synthetic brine. Quartz crystal was used to represent the sandstone reservoirs whereas calcite crystal was used to represent the limestone and dolomite reservoirs in the contact angle measurements. The wettabilities of the reservoirs were evaluated using an arbitrary contact angle scale. Systems with contact angles from 0 to75° were classified as water wet; systems with contact angles from 75 to 105° were classified as having intermediate wettability and systems with contact angles from 105 to 180° were classified as preferentially oil wet. The results showed that 27% of the reservoirs tested were preferentially water wet, 66% were preferentially oil wet and the remaining 7% were of intermediate wettability. It was found that 43% of the sandstones were preferentially water wet, 50% were preferentially oil wet and 7% were of intermediate wettability. On the other hand, 84% of the carbonate reservoirs were preferentially oil wet, 8% were preferentially water wet and 8% were of intermediate wettability. It would appear from the results of this study that carbonates are more likely to be preferentially oil wet than preferentially water wet. However, this assertion cannot be generalized because the 55 reservoirs used in this study were not obtained by random sampling. A random sample of reservoirs would be needed if the results of the wettability tests are to be given statistical significance. A similar contact angle study by Chiligarian and Chen (1983) on 161 carbonate reservoirs showed 80% of the reservoirs to be preferentially oil wet, 8% to be preferentially water wet and 12% to be of intermediate wettability. These results are consistent with those of Treibel et al... 3.3.3 Effect of Wettability on Rock -Fluid Interactions Wettability has a profound effect on multiphase rock-fluid interactions. Wettability affects (a) the fluid distribution in the porous medium, (b) the magnitude of the irreducible water saturation, (c) the effeciency of immiscible displacement in the porous medium, (d) the residual oil saturation, (e) the capillary pressure of the porous medium, (f) the relative permeability curves of the porous medium and (g) the electrical properties of the porous medium. Wettability determines the fluid distribution in a porous medium. The wetting fluid occupies the small pores, coats the surface of the solid3-49 grains and occupies the corners of the grain contacts. The wetting phase occupies the small pores which have high specific surface areas (S=3(1- )/r) in order to minimize the specific surface free energy of the system. The nonwetting phase occupies the large pores and are located at the center of the pores. These fluid distributions are shown schematically in Figure 3.26 for water wet and oil wet porous media. For the water wet medium, the residual oil is held in the center of the large pores whereas for the oil wet medium, the irreducible water is held in the center of the large pores. Qualitatively, it is apparent that the irreducible water saturation in the oil wet medium will be less than in the water wet medium. Craig (1971) gives the following rule-of-thumb for irreducible water saturations (connate water saturations) for water wet and oil wet reservoirs. For water wet reservoirs, the irreducible water saturations are usually greater than 20 to 25% whereas for oil wet reservoirs they are generally less than 15% and frequently less than 10% of the pore volume. Wettability has a significant effect on the efficiency of immiscible displacement in a porous medium. Figure 3.27 shows schematically the displacement of oil from a water wet and an oil wet porous medium at the pore scale. In the water wet medium, the injected water is imbibed into the medium along the pore walls in a manner that enhances the oil displacement efficiency. The residual oil is trapped at the center of the large pores. In the oil wet medium, the injected water channels through the large pores leaving behind considerable residual oil in the small pores, at the solid contacts and as coatings on the solid grains. From this pore level picture, it is easy to see that the waterflood efficiency will be higher in the water wet medium than in the oil wet medium.Woter wet sond Git wet sond (a) (a) s (bo) fe) K = tod a CS Water MBO Figure 3.26: Fluid distributions as a function of wettability (Pirson, 1958).WATER Cowater Coo ZA ROcK GRAINS (a) Coolwater O1L = EZ ROCK GRAIN () Figure 3.27: Water displacing oil from a pore during a waterflood: (a) strongly water wet medium, (b) strongly oil wet medium (Raza et al., 1968).eee The higher waterflood efficiency of the water wet system than the oil wet system seen at the pore scale manifests itself at the macroscopic scale (core scale) as well. Owens and Archer (1971) performed waterflood experiments in core plugs (1.9 cm diameter and 4.4 cm length) at various wettability conditions. The core plugs were rendered progressively oil wet by dissolving a sulfonate in the oil phase. Figure 3.28 shows the oil recovery curves for the waterfloods as a function of wettability. The decline in the recovery efficiency with increasing oil wetness is obvious. Peters and Hardham (1989) have conducted similar waterflood experiments in unconsolidated sandpacks at a larger scale ( 4.8 cm diameter and 54 cm length) than core plugs. One sandpack was first saturated with brine and the brine was displaced by a viscous silicon- based test oil to establish an irreducible water saturation in contact with the sand grains. The viscous oil was then displaced by brine to simulate a waterflood. A second sandpack was first saturated with the test oil and the test oil was then displaced by brine to simulate a waterflood. Although the wettabilities of the sandpacks were not measured directly, it is believed that the first sandpack behaved as a water wet system whereas the second sandpack behaved as an oil wet system over the short time scale of the fast waterflood experiments. Both waterfloods were imaged by X-ray CT to visualize the insitu fluid saturations in time and space. The oil recovery curves for the two waterflood experiments are shown in Figure 3.29. They show the displacement in the water wet sandpack to be more efficient than in the oil wet sandpack . These results are in agreement with those of Archer and Owens. The low water breakthrough recoveries in this study are due to the high oil-water viscosity ratio of 91. i3-53 CONTACT ANGLE O° WATER-WET ar 20" ser —-— !80" ow- wet © wor = 25 OIL RECOVERY, PERCENT O1L-IN-PLACE ‘0 oz 04 06 08 10. WATER INJECTED, PORE VOLUMES Figure 3.28: Effect of wettability on waterflood performance at an oil- water viscosity ratio of 5 (Archer and Owens, 1971).3-54 ww=t—= WATER-WET SANDPACK augue OIL-WET SANDPACK FRACTIONAL OIL RECOVERY 0 1 2 3 PORE VOLUMES INJECTED Figure 3.29: Effect of wettability on waterflood performance at an oil- water viscosity ratio of 91 (Peters and Hardham, 1989).3-55 Figures 3.30 and 3.31 show the water saturation images for the two waterfloods at 0.10 and 0.25 pore volumes injected. The images for the water wet sandpack show a relatively uniform and efficient displacement of the oil by the water, with relatively high water saturations. In contrast, the images for the oil wet sandpack show a chaotic, fragmented. and inefficient displacement, with relatively low water saturations. These images clearly show the important role of wettability in determining the efficiency of waterfloods at the macroscopic scale. 3.4 CAPILLARITY AND CAPILLARY PRESSURE The combined effects of interfacial tension and wettability will cause a wetting fluid to be spontaneously imbibed or sucked into a capillary tube. This phenomenon is known as capillarity. Capillarity is significant in a porous medium saturated with immiscible fluids because the interconnected pores of the medium are of capillary dimensions. In the capillary rise experiment discussed in Section 3.2.3, Eq. 3.8 gave the equilibrium relationship between the capillary suction force and the gravity force. A force balance for the equilibrium of the meniscus gives (Paw — Pw)nr? = 2nrocos® (3.18) where Ppw = pressure in the nonwetting phase, dynes /cm?2 Py = pressure in the wetting phase, dynes/cm2 r = radius of the capillary tube, cm 6 = interfacial tension between the wetting and nonwetting phases, dynes/cm @ = the contact angle(a) (©) Figure 3.30: Y-DIREETION (eM) X=DIRECTION © = the contact angle g = gravitational acceleration, 981 cm/s? The capillary pressure represents the pressure differential that must be applied in order for a nonwetting fluid to displace a wetting fluid from the capillary tube. This is termed the displacement pressure for the capillary tube. This pressure increases as the radius of the capillary tube decreases. The capillary pressure also is a measure of the tendency of the solid to attract or imbibe the wetting phase and repel the nonwetting phase.3-59 Eq. 3.19 is a special case of a more general equation for the equilibrium pressure difference across the curved interface between two immiscible fluids known as Laplace equation: a 1 Pe i+ 4 @.21) where rz and rz are the two principal radii of curvatures of the interface in two perpendicular planes (Figure 3.32). The pressure on the concave side is greater than that on the convex side of the interface. Several special cases of Laplace equation are of interest. For a spherical liquid drop, ry = r2 = 1, the radius of the drop. The capillary pressure or the excess pressure of the drop is given by 28 (3.22) For the capillary rise experiment, rj = r2=1/cos@. The capillary pressure is given by P, =20c088 (3.23) EE:Figure 3.32: Equilibrium at a curved interface between two immiscible fluids.3-61 For a soap bubble, ry = rp =r, the radius of the bubble. The capillary pressure or excess pressure is given by P= 4g (3.24) where a factor of 2 has been incorporated to account for the two gas- liquid interfaces of a soap bubble. For a flat interface, rj = r2 = infinity. In this case, the capillary pressure is zero. For a pendular ring at the contacts of two spherical sand grains in an idealized porous medium consisting of a cubic pack of uniform spheres, the capillary pressure is given by Eq. 3.21, where rj and rz are as shown in Figure 3.33. If the wetting fluid saturation in the pendular ring is reduced, the radii of curvatures will be reduced and the capillary pressure will increase. Similarly, if the wetting fluid saturation is increased, the radii of curvatures will increase and the capillary pressure will decrease. Therefore, an inverse relationship exists between the capillary pressure and the wetting phase saturation for a porous medium. Of course, for an actual porous medium, the complexity of the pore structure and the fluid interface arrangements therein preclude the use of Eq. 3.21 directly to calculate the capillary pressure. Instead, the capillary pressure is measured experimentally as a function of the wetting fluid saturation. Thus, in general, for a porous medium, P. = oC = fSy, 6, 9) (3.25) where C = (1/r,+1/r,) is the mean curvature of the fluid interfaces at the particular saturation. The mean curvature increases as the wetting phase saturation decreases.Water r Figure 3.33: . Radii of curvatures for pendular rings around spherical sand grain junctions. :3-63 3.4.1 Drainage Capillary Pressure Curve Figure 3.34 shows a typical capillary pressure curve obtained by displacing the wetting phase from a porous medium with a nonwetting phase. A process in which the wetting phase saturation decreases is known as drainage. The converse process in which the wetting phase saturation increases is known as imbibition. The capillary pressure curve has several characteristic features. The curve shows that a minimum positive pressure (Pq) must be applied to the nonwetting phase in order to initiate the drainage. This minimum pressure, which is known as the displacement pressure, the threshold pressure or the entry pressure, is determined by the largest pores connected to the surface of the medium. It can be estimated with Eq. 3.23, where r is the largest pore radius connected to the surface. As the pressure of the nonwetting phase is increased, smaller and smaller pores are invaded by the nonwetting fluid. Eventually, the wetting phase becomes discontinuous and can no longer be displaced from the medium by increasing the capillary pressure. Therefore, an irreducible wetting phase saturation is achieved for the porous medium at a high capillary pressure. At the irreducible wetting phase saturation, the capillary pressure curve becomes nearly vertical. The irreducible wetting phase saturation is a function of the grain size (pore size), the wettability and the interfacial tension. What information does the capillary pressure curve for a reservoir tock provide about the porous medium? If one reservoir has a higher permeability than another, we know that the higher permeability rock will permit faster fluid flow through it, everything else being equal, than the lower permeability rock. Thus, the higher permeability rock is more desirable than the lower permeability rock in a petroleum reservoir. If cone reservoir rock has a higher porosity than another, we know that the higher porosity rock will store more reserves than the lower' porosityP, (psi) 0 { Swi 1.0 Wetting Phase Saturation, (Sw) Figure 3.34: A typical drainage capillary curve.3-65 rock. Therefore, the higher porosity rock is a more desirable reservoir rock than the lower porosity rock. If one rock has a higher capillary pressure at the same wetting phase saturations than another, what can we say about the rocks? Is the rock with the higher capillary pressure curve more desirable or less desirable as a petroleum reservoir than the rock with the lower capillary pressure curve? Figure 3.35 shows the drainage capillary pressure curves for four rocks: A, B, C and D. Rock A has the least displacement pressure. Therefore, it has the largest pores connected to the surface. Its capillary pressure curve remains essentially flat as the wetting phase saturation is decreased from 100% to 60%. This means that many of the pores are invaded by the nonwetting fluid at essentially the same capillary pressure. This indicates the A has uniform pores or is well sorted. Rock A also has the least irreducible wetting phase saturation, indicating that it has relatively larger grains and pores than the other rocks. Rock B has a higher displacement pressure than A. Therefore, it has smaller pores than A. The capillary pressure curve at the high wetting phase saturations is relatively flat, indicating good sorting. Rock B has a higher irreducible water saturation than A which is consistent with its finer grains and pores. Rock C is even more fine grained than B because of its higher displacement pressure. The shape of its capillary pressure curve shows that a higher capillary pressure is required at each wetting phase saturation to desaturate the rock. This means that C has a wider pore size distribution than A and B. Therefore, C is poorly sorted. It has a higher irreducible water saturation than B which is consistent with its finer grains and pores. Rock D is extremely fine grained, extremely poorly sorted and would be a very poor reservoir rock. This observation is based on its very high displacement pressure, very steep capillary pressure curve and very high irreducible water saturation.1S —_ Pe (bar) 100 80 60 40 20 0 <— Sw Sy 0 20 40 60 80 100 °%o of pore volume Figure 3.35: Capillary pressure curves for four different rocks.3-67 Since permeability is proportional to the square of the mean pore size, it is easy to see that Rock A has the highest permeability, followed by B, Cand D in that order. From this discusion, we conclude that the rock with the higher capillary pressure curve is a less desirable reservoir rock than the one with the lower capillary pressure curve. Figure 3.36 shows characteristic shapes of capillary pressure curves. In general, the capillary pressure curve for a porous medium is a function of (1) pore size, (2) pore size distribution, (3) pore geometry, (4) fluid saturation, (5) fluid saturation history, (6) wettability and (7) interfacial tension. 3.4.2 Conversion of Laboratory Capillary Pressure Data to Reservoir Conditions Typically, capillary pressure curves are measured in the laboratory using fluids other than reservoir fluids. It is not uncommon to measure the capillary pressure curves to be used for analyzing an oil-water reservoir using air and water or mercury and air in the laboratory. It becomes necessary to convert the laboratory data to reservoir conditions. This conversion is done as follows. From Eq. 3.23, (Phan = Arcos (3.26) (Peleeree = 1825s 327) Eliminating r from Eqs. 3.26 and 3.27 gives (Fehon = Pehl Peles 28)s s Pe Pe Unsorted Well sorted s s Pe Pe| sesasissiauaaea Well sorted Well sorted Coarse skewness ~ Fine skewness s s Pel Pe| Poorly sorted Poorly sorted Slightly fine skewness Siight coarse skewness Figure 3.36: Characteristic shapes of capillary pressure curves (Archer and Wall, 1986).3-469 This ability to scale the laboratory capillary pressure data to reservoir conditions provides the flexibility of making laboratory capillary pressure measurements with more convenient fluids than reservoir fluids. 3.4.3 Averaging Capillary Pressure Data The capillary pressure curves for rock samples from the same reservoir having different permeabilities will be different. It is often necessary to average the capillary pressure data for cores from the same reservoir to obtain one capillary pressure curve that can be used for reservoir performance analysis. This averaging can be done using the Leverett J function, which is a dimensionless capillary pressure function. Based on Eq. 3.23, a dimensionless capillary pressure curve can be defined as 1G.) = ee 3.29) where r is a characteristic pore dimension of the porous medium. Leverett (1941) suggested a characteritic pore dimension equal to the square root of the ratio of the permeability and porosity of the medium. With this choice, the Leverett J function is given by « Pe, fk Iw) = Seay § (3.30) It is found that the capillary pressure curves for the same rock type ‘but having different permeabilities will give the same J function. This is confirmed by Figure 3.37 which shows the J function for nine unconsolidated sands with widely different permeabilities. The data plot as one curve. Figure 3.38 shows the J function for a carbonate reservoir. The Leverett J functions for different rock types will be different as shown in Figure 3.39.3-70 ual aecanncl T Legend: Permeability Fluids, 12 Oo 214 darcies Woter — Kerosene —} tates i caer O 42-4 " 10% NoCl — Air 1.0 Ee @ 2160 aS Kerosene vi _| : Oo 07 CCK, Air $ Vo 363 Water — Air i eee ae z os a 720" eee icity 5 irene ana! = ls 0.6 | : e oe a 16 5 OE 0.2 o 0 20 40 60 80 1,00 Saturation of Wetting Fluid Figure 3.37: - Leverett J function for unconsolidated sands (Leverett, 1941).3-71 a uid sotertion, ‘gid startin, % to a a 0 % 20 0 aus sotreion, % avid sateratin, te ‘) Gaui soturtion, % te Figure 3.38: Leverett J functions for a carbonate reservoir; (a) all cores; (b) limestone cores; (c) dolomite’ cores; (d) microgranular limestone cores; (e) coarse-grained limestone cores (Brown, 1951).NV oe Koti NLT | ingats shale ‘040 Figure 3.39: Leverett J functions for different rock types(Rose and Bruce, 1949). 2 304030 Lt 70608000 Woter soturation, Sy3-73 3.4.4 Determination of Initial Static Reservoir Fluid Saturations by Use of Drainage Capillary Pressure Curve Initially, the petroleum reservoir was saturated with water before oil migrated into the reservoir and displaced the water. This displacement of a wetting phase by a nonwetting phase is simulated in the laboratory measurement of the drainage capillary pressure curve. The final fluid distribution in the reservoir is determined by the equilibrium between capillary and gravitational forces. Consider the static equilibrium for the water and oil. From hydrostatics, for the water, =~ pug @31) eB where Py = pressure in the water, dynes/cm2 z= height above a datum, cm Pw = density of the water, gm/cm3 = gravitational acceleration, 981 cm/s? Similarly, for the oil, ee 1 & (3.32) where Po = pressure in the oil, dynes/cm2 Po = density of the oil, gm/cm3 Assuming incompressible fluids, Eqs. 3.31 and 3.32 can be integrated to give Py(z) = Pw(0)— page (3.33)‘° 3-74 and Po(z) = Po(0) — pose (3.34) Subtracting Eq. 3.33 from 3.34 gives P.(z) = Pc(0) + Apgz (3.35) where Pc(z) = capillary pressure at a height z above the datum, dynes/cm?2 P,(0) = capillary pressure at the datum, dynes/cm2 Ap = density of water minus density of oil, gm/cm3 We select as datum the free water level at which the capillary pressure is zero. With this choice of datum, Eq. 3.35 becomes Pez) = Apez (3.36) The free water level occurs at a depth do below the oil-water contact given by do = a 37) where Pg is the displacement pressure. Thus, the elevation above the oil- water contact of any particular saturation is given by 638) where3-75 h = elevation above the oil-water contact, cm Pe = capillary pressure corresponding to the desired saturation, dynes /em2 Figure 3.40 shows (1) a typical static fluid distribution in a homogeneous reservoir, (2) the oil-water contact, (3) the free water level and (4) the oil and water pressure profiles. Note the transition zone above the oil-water contact in which the water saturation decreases from 100% to the irreducible water saturation. The height of this transition zone is a function of the wettability, the oil-water density contrast, the oil-water interfacial tension, the pore size and sorting. Ina layered reservoir in which the layers have different capillary pressure curves, the layers are in capillary equilibrium. As a result, saturation discontinuities will occur. However, there will be only one free water level. Figure 3.41 shows a typical water saturation distribution for a well that has penetrated a layered reservoir. 3.4.5 Capillary Pressure Hysteresis Capillary pressure curves show a marked hysteresis depending on whether the curve is determined under a drainage or an imbition process. Figure 3.42 shows typical drainage and spontaneous imbibition capillary pressure curves for the same porous medium. At a given wetting phase saturation, the drainage capillary pressure is higher than the imbibition capillary pressure. At a capillary pressure of zero, the spontaneous imbibition curve terminates at a residual nonwetting phase saturation. Figure 3.43 shows how the imbibition capillary pressure curve can be used to determine the lowest location for clean oil production in a homogeneous reservoir.PRESSURE —> cap ele WELL IMPERMEABLE CAP ROCK REGION OF IRREDACIBLE WATER SATURATION TRANSITION ZONE WATER SATURATION OlL WATER Con TACT Noles To maintain We wetegri. coxpillaey, peau ey. of he cop rock, Ie. cap UK ( Rexp) rocked foe fess than We ehéplacement pressure of Me cep OK: Figure 3.40: Initial static fluid distribution ina homogeneous reservoir.3-77 Profit ot well 0 _ = _n0% vei : OD (6) (c) own | ® 3 ® 4 ® Fw. ‘x Figure 3.41: Fluid distribution for a layered reservoir; (a) well penetrating a layered reservoir; (b) capillary’ pressure curves for the layers; (c) water staturation profile observed at the well (Archer and Wall, 1986).Paw 7 Pw) bition Cure Drainage Curve Pe ( Oo tswi Sw Figure 3.42: Capillary pressure hysteresis.3-79 Welt ‘GolisinIDs s010m sigionpaia| Immobite cil 100% lies in region between icreducible water saturation ‘and 100% water saturation Initial water =cit transition zone 0» “yyy Lillddlea | owarr t { “Cleon cil’ (Pe. 2 2 g g i 5 a 3 3 2 for oll production level (Pc=0) Observed OWS Poy) Free water ———- Lowes! focation Figure 3.43: Static saturation distribution showing the role of the imbibition capillary pressure curve.3-80 Capillary pressure hysteresis is caused by (1) differences in the advancing and receding contact angles, (2) wettability change during drainage and imbibition and (3) the “ink bottle effect” which allows the same capillary pressure to exist at two different wetting phase saturations. Figure 3.44 shows the “ink bottle effect’. Clearly, the wetting phase saturation is less for the imbibition than the drainage process. Capillary pressure hysteresis presents no problem in reservoir engineering analysis as it is usually clear which curve should be used for a particular analysis. The drainage curve should be used for estimating the initial fluid saturation distribution in the reservoir whereas the imbibition curve should be used for analyzing a waterflood performance in a water- wet reservoir. 3.4.6 Capillary Imbibition Consider a reservoir consisting of two layers with different permeabilities and capillary pressure curves as shown in Figure 3.45 (a) and (b). Initially, both layers are in capillary equilibrium at their respective irreducible water saturations. Let this equilibrium be disturbed by waterflooding the two layers. The injected water will advance further into the more permeable layer (Figure 3.45 (©). The oil and water pressures are continuous across the interface between the two layers. Equilibrium across the interface requires Po = Poo 3.39) Pw = Pwo (3.40)Drainage imbibition Figure 3.44 Demonstration of capillary hysteresis (Corey, 1977).3-82 Subtracting Eq. 3.40 from 3.39 gives the condition for equilibrium as Pa = Pe G41) Thus, at equilibrium, the capillary pressures in the two porous media will be equal at their interface. In Figure 3.45 (0), sections Aand D and C and F are in capillary equilibrium, so no fluid exchanges will occur between these sections. Sections B and E are not in capillary equilibrium, so fluid exchanges will occur in an effort to achieve capillary equilibrium. Section E will loose water to section B and gain oil from B while section B will gain water from E and loose oil to E until a new capillary equilibrium is achieved. Thus, water will be imbibed into the less permeable layer from the more permeable layer and oil will be expelled from the less permeable layer into the more permeable layer for subsequent displacement. This fluid exchange is beneficial to the oil recovery process. However, the imbibition process is very slow. Therefore, the water injection rate must be sufficiently slow for imbibition to assist in waterflooding the low permeability layer. Fissured reservoirs (naturally fractured reservoirs) present another example of capillary imbibition. The fractures have zero capillary pressure whereas the matrix blocks have normal capillary pressure curves. When the fractures become 100% saturated with water which comes in contact with the oil saturated matrix blocks, the capillary equilibrium will be disturbed. Water will be imbibed into the matrix blocks, expelling oil from the matrix blocks into the fractures. Ultimately, the oil saturation in the matrix will be reduced to the residual oil saturation.3-83 (ay K2z>kK, (b) Kk. 4, & mine Fe Swi Surg Sui Suse 60 we tet, Figure 3.45 Capillary imbibition; (a) reservoir before waterflooding; (b) capillary pressure curves for the layers; (c) reservoir after waterflooding.3-84 3.4.7 Capillary End Effect in a Laboratory Core Another capillary phenomenon of interest is the capillary end effect often experienced in laboratory coreflooding experiments. The end of the core is in contact with the outside which could be viewed as a second medium with zero capillary pressure. The condition for capillary equilibrium (Eq. 3.41) requires that the capillary pressure inside the core at the outlet end be equal to zero. Consider a porous medium initially saturated with oil and irreducible water saturation. The outlet end of the core is at a higher capillary pressure than the outside. If the medium is waterflooded, initially, only oil will be expelled from the outlet end at a higher capillary pressure than the outside (Figure 3.46 (a)). When the water arrives at the outlet end, however, the system now has a chance to seek capillary equilibrium. This equilibrium will be achieved by the accumulation of the water at the outlet end of the core until the water saturation equals the water saturation at the residual oil saturation at which the imbibition capillary pressure is zero (Figure 3.46 (b)). Thus, water production is delayed until well after water arrival at the outlet end of the core. This phenomenon has several undesirable consequences. The observed water breakthrough oil recovery will be falsely high and the water saturation distribution in the core will be uneven, with the water saturation being higher towards the core outlet than in the rest of the core (Figure 3.46 (c)). These effects are undesirable in laboratory coreflood experiments especially in relative permeability experiments. Capillary end effect can be reduced by increasing the injection rate.(6) 1-0 1 1 Swro 1 { 1 Sw \ 0 Swi Sweo “0 Swi Sw x7> Note: Actual water production occas at ts not at tye. Figure 3.46 Capillary end effect; (a) coreflood; (b) imbibition capillary pressure curve; (c) water saturation profiles.3-86 3.4.8 Capillary Pressure Measurements Three methods are commonly used to measure capillary pressure curves: the restored state method (porous plate method), mercury injection method and centrifuge method. Restored State Method (Porous Plate Method). In this method, capillary pressure is measured by placing the sample, initially saturated with a wetting fluid, in a vessel filled with the nonwetting fluid. The bottom of the vessel consists of a semi-permeable plate which allows the wetting phase displaced from the sample to pass through while blocking the passage of the nonwetting phase. Extending from the porous plate is a graduated tube which allows the volume of the wetting phase displaced to be measured (Figure 3.47). With the sample in place, the pressure of the nonwetting fluid is increased in steps and the system is allowed to achieve equilibrium after each pressure change. The volume of wetting phase displaced at each pressure is measured. The capillary pressure is the nonwetting phase pressure minus the wetting phase pressure at each step. The wetting phase saturation of the sample is determined from the volume of wetting phase displaced at each pressure to obtain the capillary pressure versus saturation relationship. The porous plate is typically made of porcelain or fritted glass. It must have a displacement pressure that is higher than the largest capillary pressure to be measured. This limits the maximum capillary pressure that can be measured with the method to about 200 psi. The porous plate apparatus can be adapted to measure the imbibition capillary pressure curve as well as the drainage curve. The method gives a reliable estimate of the irreducible wetting phase saturation, The major disadvantage of the porous plate method is that it takes too long to obtain the entire capillary pressure curve. It is not unusual for the capillary pressure experiment to take several weeks to complete.Figure 3.47 Nitrogen pressure ‘Soran tube Crude oil Neoprene stopper “Nickel plated ‘spring red olf N— Core ‘Kleenex poper ‘Uitra-tine titted glass disk Porous plate capillary pressure apparatus (Welge and Bruce, 1947).Mercury Injection Method. In this method, capillary pressure is measured by injecting mercury, which is a nonwetting phase, into the sample. The apparatus used in the measurement is shown in Figure 3.48. It consists of a sample cell and a mercury injection pump. A dry sample is placed in the cell and the cell is evacuated. Mercury is injected into the cell until the mercury is level with a graduation on the high-pressure glass capillary above the sample chamber. Nitrogen pressure is then applied in successive increments and at each step, mercury is injected to maintain the mercury level with the graduation on the capillary. From the volume of the cell and the volume of mercury required to fill the cell with the sample before mercury injection into the sample, the bulk volume of the sample can be determined. The mercury-air capillary pressure versus saturation relationship is calculated from the volume of mercury forced into the sample pore space as a function of the applied nitrogen pressure. The mercury injection method is very fast. The capillary pressure curve can be obtained in a matter of hours. The imbibition curve can be obtained very easily by decreasing the nitrogen pressure and withdrawing mercury from the system. Figure 3.49 shows typical capillary pressure curves obtained by mercury injection, mercury withdrawal and mercury reinjection.0-200 psi pressure gauge 0-2,000 psi pressure gouge Regulating valve 1 EE To Lucite window atmosphere Cylinder = U-tube monometer Lucite window Figure 3.48 Capillary pressure cell for mercury injection (Purcell, 1949). aa100 PRESSURE, MPa 1- Injection Curve 2- Withdrawal Curve 3-Reinjection Curve 10 Pa 4.0 Simax 0.5 Swmin 0 MERCURY SATURATION, % PV Figure 3.49 Mercury-air capillary pressure curves.Brown (1951) has shown that the mercury injection method can give essentially the same capillary pressure curve as the restored state method except for a scaling factor. Figures 3.50 and 3.51 compare capillary pressure curves obtained by mercury injection and the restored state method for a sandstone and a limstone core, respectively. The results show good agreement between the two methods. The scaling factors for the sandstone and limestone were 7.5 and 5.5, respectively. These are different from the scaling factor of 5.2 suggested by Purcell (1949) based on the ratio of ocos® of mercury-air and water-air systems. The major disadvantage of the mercury injection method is that the core can no longer be used for other tests after mercury injection. The method also cannot be used to determine the irreducible wetting phase saturation.3-92 225.0 we Ss. LEGEND: Psi © Restored -: state y uO 187.5 jection @ Mercury 150.0 20 _ SANDSTONE COR 15 Porosity—28.1 % ines Permeability — 1.43 darcys Factor — 7.5. : ay 3 37.5 a MERCURY CAPILLARY PRESSURE WATER=NITROGEN CAPILLARY PRESSURE : PSI ° 0 20 40 60 80 100 4,O —> 100 60 60 40. 20 0 gee Hig CiQUID: SATURATION :* PER “GENT Figure 3.50 A comparison of water-nitrogen and mercury-air capillary pressure curves for a sandstone core (Brown, 1951).348 a ° LEGEND: © Restored state @ Mercury injection T eae Psi o ° 0 © ° y o cy b ° a ° es CAPILLARY PRESSURE : y ° CIMESTONE CORE 1 WATER-NITROGEN CAPILLARY PRESSURE: PSI _ MERCURY: 100 come Hg “Ladi SATURATION : PER “cent A comparison of water-nitrogen and mercury-air capillary pressure curves for a limestone core e (Brown, 1951). Figure 3.51Centrifuge Method. In this method, the sample saturated with a wetting fluid is placed in a centrifuge cup containing the nonwetting fluid (Figures 3.52 and 3.53). The sample is rotated at a series of constant angular velocities and the amount of wetting fluid displaced at equilibrium at each velocity is measured with the aid of a stroboscopic light. The only data measured directly in this method are the volume of wetting fluid displaced and the corresponding rotational speed of the centrifuge. These data can be used to derive the capillary pressure versus saturation relationship of the porous medium. The theory underlying the method is that the centrifuge imposes a centrifugal force (typically, over 1000 times the force of gravity) on the sample. This causes the denser wetting fluid to be displaced outward away from the center of rotation and the nonwetting fluid to flow into the sample through the inlet face of the sample. Consider an oil-water system in which water is the wetting phase and oil is the nonwetting phase. At equilibrium, the pressure gradients in the water and oil are given by @Pw = pyr (3.42) dr Po = pow (3.43) dr where Pw = pressure in the water phase, dynes /cm2 Po = pressure in the oil phase, dynes /cm? 1 = arbitrary position in the sample measured from center of rotation, cm @ = angular velocity, radians/s Pw = density of the water, gm/cm3 Po= density of the oil, gm/cm3CENTRIFUGE ARM GRADUATED GLASS TUBE ‘CENTRIFUGE SHIELD Figure 3.52. Positions of core and graduated tube in a centrifuge for measurement of oil-displacing-water capillaty pressure curve (Donaldson et al., 1980).3-96 GRADUATED, GLASS TUBE CENTRIFUGE. RUBBER STOPPER ‘SHIELD RUBBER CUSHION Figure 3.53 Positions of core and graduated tube in a centrifuge for measurement of water-displacing-oil capillary pressure curve (Donaldson et al., 1980). we3-97 Subtracting Eqs. 3.42 from 3.43 gives dP, 2, Be = Ay 3.44) pwr (3.44) Integration of Eq. 3.44 gives - ~Aparte + c (3.45) where C is an integration constant. Application of the Hassler-Brunner boundary condition P, = 0 at r = r2 gives the capillary pressure at any ras Pex Ae g- G46) At any stage of the centrifuge experiment, the highest capillary pressure occurs at the inlet face of the core at r. At the inlet face, the capillary pressure is given by Pas feet (3-73) G7) where Pq = capillary pressure at core inlet, dynes/cm? 1] = radius of core inlet face measured from the center of rotation, om 12 = radius of core outlet facemeasured from the center of rotation, an c= angular velocity (= 2nN/60, where N = centrifuge speed in revolutions per minute), radians/s Ap = density of water minus density of oil, gm/cm3 At any stage of the centrifuge experiment, the water saturation in the core varies from a minimum at the core inlet, ry, to avalue of 1.0 at3-98 the core outlet, r2. Since the capillary pressure at the core inlet can be calculated from Eq. 3.47, it is only necessary to estimate the water saturation at the core inlet in order to obtain the required capillary pressure versus saturation relationship. The water saturation at the core inlet can be estimated from the average core saturation determined from the amount of water displaced at each speed. The average water saturation in the core is given by ” Sydr Sway (3.48) 12-" where Swav = average water saturation Sw = water saturation at distance rin the core Multiplying Eq. 3.48 by 4pw2ry and rearranging gives 2 Ap@rLns wav -| SwApw*ridr 3.49) n where Lis the length of the core. Noting that Apw2Lry is approximately equal to Pcj since L is small relative to ry and r2 and Apw2rjdr is equal to -dP¢1, Eq. 3.49 can be written as Pa PeiSwav = { Sw(PodPe (3.50) 0 aoeDifferentiating Eq. 3.50 gives the water saturation at the core inlet as Sui = 1 ParSwav) G51) Pa or dS, = Sumy + Pei @Sum : Set = Swas + Fol a (3.52) ‘The approximate water saturation equation, Eq. 3.51 or 3.52, is applicable for rj /r2 equal to or greater than 0.8. The centrifuge method is fast and allows the capillary pressure measurement to be completed in a day or less. The method is good for determining the irreducible water saturation. Hassler and Brunner (1945) have presented data that show good agreement between the capillary pressure curve obtained with the centrifuge method and that obtained by the restored state method (Figure 3.54). The disadvantages of the method include (1) inability to measure the displacement pressure since the water saturation at the core inlet is always less than the average water saturation of the core, (2) the Hassler-Brunner boundary condition at the core outlet may be violated at high centrifuge speeds, (3) the calculated water saturation at the core inlet is an approximation, and (4) inability to obtain spontaneous imbibition capillary pressure curve. Melrose (1988) has investigated the Hassler-Brunner boundary condition and concluded that it was unlikely to be violated under normal core analysis conditions. Rajan (1986) has presented a more accurate method for calculating the water saturation at the core inlet.Figure 3.54 3-100 a : al F fl cebarfuge| Mole 4 ae she Copittory Pressure LAtmosphares) —e A comparison of capillary. pressure obtained by centrifuge and by the restored state method (Hassler and Brunner, 1945).os 3-101 The centrifuge can be used to measure the United States Bureau of Mines (USBM) wettability index by carrying out a number of forced water and oil displacement experiments as shown in Figure 3.55. The process labeled 1 is the forced displacement of water by oil. The process labeled 2 is the forced displacement of oil by water. The process labeled 3 is another forced displacement of water by oil. The USBM wettability index is given by 3.53) where the areas Ay and A2 are shown in Figure 3.55. The areas under the capillary pressure curves represent the thermodynamic work required for the respective displacements. Displacement of a nonwetting phase by a wetting phase requires less energy than the displacement of a wetting phase by a nonwetting phase. Therefore, the ratio of the areas under the capillary pressure curves is a measure of the degree of wettability of the porous medium. The USBM wettability index for a water wet medium is positive whereas the index for an oil wet medium is negative. The magnitude of the index is a measure of the degree of wettability preference. A wettability index of zero indicates equal wetting or no preferential wetting by either fluid.3-102 ov Figure 3.55 Determination of USBM wettability index (Donaldson et al,, 1969).3-103 3.4.9 Pore Size Distribution Capillary pressure versus wetting phase saturation curves (or mercury injection data) can be used to estimate the “pore size” distribution of the porous medium based on a bundle of capillary tube model of the porous medium. The pore size distribution function is given by Ritter and Drake (1945) as =P, dVp-V) Dir)= FF a (3.54) where D(z) = pore size distribution function Vp = total pore volume V = volume of pores with radius less than rj P. = capillary pressure ¥j = pore entry radius In Eq. 3.54, D(rj)dr; is the pore volume contributed by pores with entry radius rj and rj+drj. (rj) has units of volume per unit length or length squared. A related distribution function, a(rj), with unit of reciprocal length, can be defined as a pore size probability density function, such that o(rj)drj is the fraction of the total pore volume occupied by pores with radius between rj and rj+drj. In the case of mercury injection, the volume of mercury injected at a given capillary pressure is equal to the volume of pores with radius greater than 1. Based on these considerations, Eq. 3.54 becomes = Pe dSaw oe) =o (3.55) Eq, 3.55 can be used with Eq. 3.23 (Pelt) = 2ocos6/rj) and the capillary pressure curve to calculate the pore size distribution function as shown in Figure 3.56. Eq. 3.55 can also be written in terms of the pore diameter instead of the pore radius.3-104 Pe Incremental Mercury Injection = Pe dS, tT aR in) r Oo TS pore radius Figure 3.56 Calculation of pore size distribution function.3-105 Figure 3.57 shows a typical pore size distribution function for a sandstone as determined by mercury injection. Figure 3.58 shows the distribution functions obtained by Ritter and Drake (1945) for diatomaceous earth and fritted glass. Dullien (1979) has presented data that show that the pore size distribution obained by mercury injection is significantly different from that obtained by photomicrography (Figure 3.59). The results show that the true pore size distribution obtained by photomicrography is” considerably broader than that obtained by mercury injection. The fundamental problem seems to be the bundle of capillary tube model of the porous medium used to interpret the mercury injection data. The porous medium is made up of a system of interconnected pores rather than a bundle of capillary tubes. 3.4.10 Estimation of Permeability from Capillary Pressure Because of the relationship between pore size and capillary pressure, there have been various attempts to estimate permeability from capillary pressure curves. Purcell (1949) used a bundle of capillary tube model of the porous medium to derive the following permeability equation: nei x = (ocos0h a as (3.56) 2} where Fj is a dimensionless lithology factor to account for tortuosity. Purcell suggested a range for the lithology factor between 0.085 and 0.363, with a mean of 0.216. The usefulness of Eq. 3.56 is limited by the bundle of capillary tube model in that saturation is developed in the large pore bodies that are not reflected in the capillary pressures, which are governed by pore throats.3-106 2000 I 1600 Permeability 440 millidorcies Porosity 237 1200 Dlg) {4quore centimeters) g T 000? 004.00 005 -~010«OI2.-—«OA «ONG CONS Equivotent Pore Entry Radius (centimeters) Figure 3.57 Typical pore size distribution function for a sandstone (Burdine et al., 1950).3-107 oO O 2000 4000 6000 8000 {0,000 Pore radius, A ° Diatomaceous earth © UF fritted glass Figure 3.58 Pore size distribution functions for diatomaceous earth and fritted glass (Ritter and Drake, 1945). i3-108 e 8 g fo Meccury Lnjection 4 Thetomiceogrephy PORE VOLUME DISTRIBUTION FUNCTION BY D OR O, ( ni'x10*) a 0 2 «#40 «60 «80 6100 120 PORE OIAMETER 0 OR 0, (um) Figure 3.59 Comparison of mercury injection pore size distribution with photomicrographic pore size distribution (Dullien, 1979).3-109 3.5 Effective and Relative Permeabilities Ina petroleum reservoir, it is possible for two or three fluids to flow simultaneously. Examples are (a) the flow of gas and water in a gas reservoir, (b) the flow of oil and water in an oil reservoir, (c) the flow of oil and gas in an oil reservoir and (d) the flow of oil, water and gas in an oil reservoir. In multiphase flow situations, the absolute permeability of the porous medium is no longer sufficient to calculate the flow rate of each fluid type or to calculate the total flow rate of all the fluids. : In order to make quantitative predictions for multiphase flow, we need to know the permeability to each fluid in the presence of the other fluids in the rock. The permeability of one fluid in the presence of the other fluids is known as the effective permeability to that fluid. To calculate the flow rate of each fluid in multiphase flow, we extend Darcy’s Law to multiphase flow. For example, for simultaneous flow of oil, water and gas in a horizontal linear system, Darcy’s Law is applied to each phase as follows: —keA GP, 3.57) Ho dx a = —KwA dPw qw Tatas (3.58) =A De 3.59) He dx where qo = volumetric oil flow rate, cm3/s qw = volumetric water flow rate, cm3/s 9g = volumetric gas flow rate, cm3/s ko = _ effective permeability to oil, Darcieskw kg A Po Pw Ho. bw ug x wouow 3-110 effective permeability to water, Darcies effective permeability to gas, Darcies cross sectional area normal to flow, cm2 pressure in the oil phase, atm pressure in the water phase, atm pressure in the gas phase, atm oil viscosity, cp ‘water viscosity, cp oil viscosity, cp distance in the direction of flow, cm Eqs. 3.57 to 3.59 show that using the concept of an effective permeability, Darcy’s Law is applied to each phase as if the other phases did not exist. Capillary equilibrium between the phases gives where Pefow Pe/go Po/gw Po- Pw=PevowSw) (3.60) Pg- Po=Pe/go(So) (3.61) Pevow + Pevgo = Pe/gw (3.62) oil-water capillary pressure curve gas-oil capillary pressure curve = gas-water capillary pressure curve It is often more convenient to work with a dimensionless effective permeability known as the relative permeability obtained by dividing the effective permeability by a base permeability such as the absolute permeability of the porous medium. Thus, for the three phase example, using the absolute permeability of the porous medium as the base permeability, the relative permeabilities to oil, water and gas are given by3-111 Ko = e (3.63) Kw = ke (3.64) ky = (3.65) Sometimes, the effective permeability to the nonwetting phase at the irreducible wetting phase saturation is used as the base permeability for defining the relative permeability. In this case, the end point relative permeability to the nonwetting phase will be one. Figure 3.60 shows typical oil-water relative permeability curves for a two-phase, oil-water system. The key features of the relative permeability curves include: (a) the relative permeability curves are nonlinear functions of fluid saturation, (b) the sum of the oil and water relative permeabilities at each saturation is always less than one, () an irreducible water saturation (Syj) at which the relative permeability to water is zero and the relative permeability to oil attains a maximum end point value (Koy), and (d) a residual oil saturation (Sor) at which the relative permeability to oil is zero and the relative permeability to water attains a maximum end point value (kyr). Figure 3.61 shows typical gas-oil relative permeability curves with features that are very similar to the oil-water curves of Figure 3.60.3-112 |-water relative permeability curves. Figure 3.60 - Typical oil-® er ol ertiea gos Setuaton Figure 3.61 3-113 tb) 1.0) ol_# t — 3 ye Sa, Sorex residual gos (eteSm) seturation Typical gas-oil relative permeability curves.A 3-114 3.5.1 Laboratory Measurement of Two-Phase Relative Permeabilities by the Steady State Method The most straight-forward laboratory measurement technique for relative permeabilities is the steady state method. In this method, a mixture of two phases are injected simultaneously into a core at a progression of different relative rates (Figure 3.62). Injection continues ata fixed relative rate until the system attains steady state such that the fluid saturations and the pressure drop across the core no longer change with time. The pressure drop and the injection rates are measured. The relative permeabilities are calculated with the integrated forms of Eqs. 3.57 to 3.59. The saturations are usually calculated by material balance. Atypical sequence of steps for obtaining the imbibition oil-water relative permeabilities for a water-wet system might be as follows: 1. __ Install the clean, dry core sample in the Hassler apparatus as shown in Figure 3.62. Evacuate the core and saturate with brine (wetting phase). Determine the absolute permeability of the core by brine flow. 2. Displace the brine with the test oil until no more water flows from the core. Calculate the irreducible water saturation and the initial cil saturation. Measure the steady state pressure drop and oil injection rate and calculate the relative permeability to oil at irreducible water saturation as kp = EE (3.66) 3. Inject oil and water simultaneously at rates qo and qw such that the ratio, qw/4o, is less than one until steady state is achieved. Steady state is achieved when the injected and produced qw/do ratios are equal and the pressure drop no longer changes with time.3-115 4. | Measure the pressure drop and the water saturation by material balance. Calculate the relative permeabilities to oil and water at the latest water saturation using Eq. 3.66 and Kw = Sete: (3.67) 5. Increase the ratio qw/qo and repeat steps 3 and 4 to calculate the relative permeabilities at higher and higher wetting phase saturations. 6. Inject only water until no more oil flows from the core. Calculate the residual oil saturation. Measure the steady state pressure drop and water injection rate and calculate the relative permeability to water at residual oil saturation. This completes the relative permeability measurements. In the steady state relative permeability experiment, it is necessary to minimize the capillary end effect. This can be accomplished by injecting at a relatively high total rate (qw + qo)- Figures 3.63 and 3.64 show the pressure profiles in the gas and oil phases for a gas-oil steady state relative permeability experiment conducted at two rates by Richardson et al. (1952). At the lower injection rate (Figure 3.63), capillary end effect is apparent whereas at the higher rate (Figure 3.64), there is little or no capillary end effect. Note also, that when the capillary end effect has been eliminated, the pressure drop in each phase is the same. Therefore, the pressure drop measured in either phase is sufficient for calculating both relative permeabilities. The various steady state methods such as the Penn State method, single core dynamic method, dispersed feed method, Hafford method and Hassler method differ primarily in the technique used to minimize or eliminate capillary end effect (Richardson et al., 1952). When capillary end effect has been eliminated, all the steady state methods give the same results as shown in Figures 3.65 and 3.66.