Chapter 23

Phase Diagrams
F.M. OTKJr., New Mexico Inst. of Mining and Technology J.J. Taber, New Mexico Inst. of Mining and Technology*

Introduction
Petroleum reservoir fluids are complex mixtures containing many hydrocarbon components that range in size from light gases such as methane, ethane, and so on to very large hydrocarbon molecules containing 40 or more carbon atoms. Nonhydrocarbon components such as nitrogen, hydrogen sulfide (HzS), or CO2 also may be present. Water, of course, is usually present in large quantities in all reservoirs. At a given temperature and pressure, the components distribute between whatever solid, liquid, and vapor phases are present. A phase may be defined as that portion of a system that is homogeneous, bounded by a surface, and physically separable from other phases present. Equilibrium phase diagrams offer convenient representations of the ranges of temperature, pressure, and composition within which various combinations of phases coexist. Phase behavior plays an important role in a variety of reservoir engineering applications ranging from pressure maintenance to separator design to enhanced oil recovery (EOR) processes. This chapter reviews the fundamentals of phase diagrams used in such applications. terminates at the critical point. At temperatures above the critical temperature, T,, a single phase forms over the entire range of pressures. Thus, for a single component, the critical temperature is the maximum temperature at which two phases can exist. Critical temperatures of hydrocarbons vary widely. Small hydrocarbon molecules have low critical temperatures, while those of large ones are much higher. Critical pressures generally decline as the molecular size increases. For instance, the critical temperature and pressure of methane are - 117F and 668 psia. For decane, the values are 652F and 304 psia. (Fig. 23.9 shows additional vapor pressure curves and critical points for several other light hydrocarbon molecules.) For many reservoir engineering applications liquid/vapor equilibrium is of greatest interest, though liquid/liquid equilibria are important in some EOR processes. Solid/liquid phase changes, such as asphaltene or paraffin precipitation, occasionally occur in petroleum production operations. Fig. 23.2 shows typical volumetric behavior of a single component in the range of temperatures and pressures near the vapor-pressure curve in Fig. 23.1. If the substance under consideration is placed in a pressure cell at constant temperature T1 below T, and at a low pressure (Point A, for instance), it forms a vapor phase of high volume (low density). If the volume of the sample is decreased with the temperature held constant, the pressure rises. When the pressure reaches p,, (T1 ), the sample begins to condense. The pressure remains constant at the vapor pressure until the sample volume is reduced from the saturated vapor volume (V,.) to that of the saturated liquid (V,). With further reductions in volume, the pressure rises again as the liquid phase is compressed. Note that small decreases in volume give rise to large pressure increases in the liquid phase because of the low compressibility of liquids. At a temperature Tz above the critical temperature, no phase change is observed. Instead, the sample can be

Single-Component Phase Diagrams

Fig. 23.1 summarizes the phase behavior of a single component. The saturation curves shown in Fig. 23.1 indicate the temperatures and pressures at which phase changes occur. At temperatures below the triple point, the component forms a vapor if the pressure is below that indicated by the sublimation cuwe, and forms a solid at pressures above the curve. At pressures and temperatures lying on the sublimation curve, solid and vapor can coexist, and on the melting curve, solid and liquid are in equilibrium. At higher temperatures, liquid and vapor can coexist along the vaporization or vapor-pressure curve. If the pressure is greater than the vapor pressure, a liquid forms, if lower, a vapor. The vapor-pressure curve

PETROLEUM

ENGINEERING

HANDBOOK

MELTING CURVE

i:
SOLID

CRITICAL POINT
3 z Pb

VAPORIZATION

k a

Pd

SUBLIMATION CJRVE

VAPOR

PV2

TC

A MOLE

B FRACTION

XE COMPONENT I

1.0

T Fig. 2X1-Phase

MPERATURE

behavior of a pure component.

Fig. 23.3-Pressure-composition phase diagram for a binary mixture at a temperature below the critical temperature of both components.

Tc TC Tc
I I L I Vc VOLUME 1 Vv

of freedom (F=O). Thus, the maximum number of phases possible is three. Therefore, if three phases coexist in equilibrium (possible only at the triple point), the pressure and temperature are fixed. If only two phases are present for a pure component, then either the temperature or the pressure can be chosen. Once one is chosen, the other is determined. If the two phases are vapor and liquid, for example, choice of the temperature determines the pressure to be the vapor pressure at that temperature. These permitted pressure-temperature values lie on the vapor pressure curve of Fig. 23.1. In a binary system, two phases can exist over a range of temperatures and pressures. The number of degrees of freedom is
F=2-nc, . . . . . . . . . . . . . . . . . . . . . . . . . . . . ...(2)

Fig. 23.2-Volumetric behavior of a pure component in the liqutd/vapor region.

compressed from high volume (low density) and low pressure to low volume (high density) and high pressure with only one phase present.

and hence both the temperature and pressure can be chosen, though there is no guarantee that two phases will occur at a specific choice of T and p. For multicomponent systems, the phase rule provides little guidance, since the number of phases is always far less than the maximum number that can occur. Hence, as the number of components increases, more component concentrations must be known to determine the system.

Phase Rule
The maximum number of phases that can coexist at fixed temperature and pressure is determined by the number of components present. The phase rule states that
F=2+C-P-n,, ...... .... .... . . . . . . (1)

Types of Diagrams
Binary Phase Diagrams Fig. 23.3 shows typical vapor/liquid phase behavior for a binary system at a fixed temperature below the critical temperature of both components. Such a diagram is known as a pressure-composition phase diagram. At pressures below the vapor pressure of Component 2, ~2, any mixture of the two components forms a single vapor phase. At pressures between p 1 and p,,~ , two phases can coexist for some compositions. For instance, at pressure pb two phases will occur if the mole fraction of Component 1 lies between xE and xE. If the mixture

where F is the number of degrees of freedom, C is the number of components, P is the number of phases, and n c is the number of constraints. For a single-component system, the maximum number of phases occurs when there are no constraints (n c =0) and there are no degrees

PHASE DIAGRAMS

23-3

:i ?2!!!%? MERCURY PUhlP

Pep,, PPd Pr,<P<Pb PPb PPb ppdl

--pi pd

--PZPl

--P3P2 PPd2

PUMP

Fig. 23.4-Volumetric behavior of a binary mixture at constant temperature

Fig. 23.6-Volumetric behavior of a binary mixture at a constant temperature showing retrograde vaporization.

Pd2

Pdl

PVZ VAPOR

xI
MOLE FRACTION COMPONENT

K2
I

TEMPERATURE

Fig. 23.5-Pressure-composition phase diagram for a binary mixture at a temperature above the critical temperature of Component 1.

Fig. 23.7--Regions of temperature, pressure, and composition for which two phases occur in a binary liquid/vapor system.

composition is xB, it will be all liquid, if it is xE, all vapor. For 1 mol of mixture of overall composition, z, between -xE and xE, the number of moles of liquid phase is XE --z
XE -xB

L=-.

. . . . . . . . . . . . . . . . . . . . . . . . . . ...(3)

Eq. 3 is known as an inverse lever rule because it is equivalent to a statement concerning the distances along a tie line from the overall composition to the liquid and vapor compositions; L=ZE/BE, where ZE and BE are lengths on the tie line shown in Fig. 23.3. Thus, the amount of liquid is proportional to the distance from the overall composition to the vapor composition divided by the length of the tie line. The line connecting the compositions of phases in equilibrium is known as a tie line. In binary phase diagrams such as Fig. 23.3, the tie lines are always horizontal.

Phase diagrams such as Fig. 23.3 can be determined experimentally by placing a mixture of fixed overall composition in a high-pressure cell and measuring the pressures at which phases appear and disappear. For example, a mixture of composition xa would show the behavior indicated qualitatively in Fig. 23.4. At a pressure less thanpd (Fig. 23.3), the mixture is a vapor. If the mixture is compressed by injecting mercury into the cell, the first liquid, which has composition xA, appears at the dewpoint pressure, pd. As the pressure is further increased, the volume of liquid grows as more and more of the vapor phase condenses. The last vapor, of composition X,R, disappears at the bubblepoint pRssut& pb. If the system temperature is above the critical temperature of one of the components, the phase diagram is similar to that shown in Fig. 23.5. (See Fig. 23.13 for additional examples of this type of phase diagram.) At the higher temperature, the two-phase region no longer extends to the pure Component 1 side of

23-4

PETROLEUM ENGINEERING

HANDBOOK

CRITICAL LOCUS <

TEMPERATURE
Fig. 23.8-Projection of the vapor pressure (p,, and pa) curves and locus of critical points for binary mixtures. Points C, and C2 are the critical points of the pure components. Fig. 23.10-Properties of ternary diagrams

Fig. 23.9-Vapor pressure curves for light hydrocarbons and critical loci for selected hydrocarbon pairs.

the diagram. Instead, there is a critical point, C, at which liquid and vapor phases are identical. The critical point occurs at the maximum pressure of the two-phase region. The volumetric behavior of mixtures containing less Component 1 than the critical mixture is like that shown in Fig. 23.4. Fig. 23.6 shows the volumetric behavior of mixtures containing more Component 1. Compression of mixture of composition x2 (in Fig. 23.5) leads to the appearance of liquid phase of composition xt when pressure pDl is reached. The volume of liquid first grows and then declines with increasing pressure. The liquid phase disappears again when pressure pm is reached. Such behavior is called retrograde vaporization (or retrograde condensation if the pressure is decreasing). If the system temperature is exactly equal to the critical temperature of Component 1, the critical point on the binary pressure-composition phase diagram is located at a Component 1 mole fraction of 1.O. Fig. 23.7 shows the behavior of the two-phase regions as the temperature rises. As the temperature increases, the critical point moves to lower concentrations of Component 1. As the critical temperature of Component 2 is approached, the two-phase region shrinks, disappearing altogether when the critical temperature is reached. A typical locus of critical temperatures and pressures for a pair of hydrocarbons is shown in Fig. 23.8. The critical locus shown in Fig. 23.8 is the projection of the critical curve in Fig. 23.7 onto the p-T plane. Thus, each point on the critical locus represents a critical mixture of different composition, though composition information is not shown on this diagram. For temperatures between the critical temperature of Component 1 and Component 2, the critical pressure of the mixtures can be much

PHASE DIAGRAMS

23-5

Fig. 23.11-Ternary phase diagram at a constant temperature and pressure for a system that forms a liquid and a vapor.

Fig. 23.12-Ternary phase diagram for the methane/ butane/decane system at 160DF 171Cl.

higher than the critical pressure of either component. Thus, two phases can coexist at pressures much greater than the critical pressure of either component. If the difference in molecular weight of the two components is large, the critical locus may reach very high pressures. Fig. 23.9 gives critical loci for some hydrocarbon pairs. The binary phase diagrams reviewed here are those most commonly encountered. More complex phase diagrams involving liquid/liquid and liquid/liquid/vapor equilibria do occur, however, in hydrocarbon systems at very low temperatures (well outside the range of conditions encountered in reservoirs or surface separators) and in COz/crude oil systems at temperatures below about 50C. For reviews of such phase behavior see Refs. 2 and 3. Ternary Phase Diagrams Phase behavior of mixtures containing three components is conveniently represented on a triangular diagram such as that shown in Fig. 23.10a. Such diagrams are based on the property of equilateral triangles that the sum of the perpendicular distances from any point to each side of the diagram is a constant equal to length of any of the sides. Thus, the composition of a point in the interior of the triangle is
x, =-)

(Fig. 23.10b). In addition, mixtures lying on any line connecting a comer with the opposite side contain a constant ratio of the components at the ends of the side (Fig. 23.1Oc). Finally, mixtures of any two compositions, such as A and B in Fig. 23. lOd, lie on a straight line connecting the two points on the ternary diagram. Compositions represented on a ternary diagram can be expressed in volume, mass, or mole fractions. For vapor/liquid equilibrium diagrams, mole fractions are most commonly used. Typical features of a ternary phase diagram for a system that forms a liquid and a vapor at fixed temperature and pressure are shown in Fig. 23.11. Mixtures with overall compositions that lie inside the binodal curve will split into liquid and vapor. Tie lines connect compositions of liquid and vapor phases in equilibrium. Any mixture on one tie line gives the same liquid and vapor compositions. Only the amounts of liquid and vapor change as the overall composition changes from the liquid side of the binodal curve to the vapor side. If the mole fractions of component i in the liquid, vapor, and overall mixture are xi, yi, and zi, the fraction of the total moles in the mixture in the liquid phase is given by
L= Yi -zi yi-xi

........

. . . . . . . ......

. . (5)

Ll
LT

x2 =-)

L2 LT

x3 =-)

L3 LT

where LT=LI +L2 +L3. ...... . . . . . . . . . . . . . (4)

Several other useful propetties of triangular diagrams are also a consequence of this fact. For mixtures along any line parallel to a side of the diagram, the fraction of the component of the comer opposite to that side is constant

Fq. 5 is another lever rule similar to that described for binary diagrams. The liquid and vapor portions of the binodal curve meet at the plait point, a critical point where the liquid and vapor phases are identical. Thus, the plait point mixture has a critical temperature and pressure equal to the conditions for which the diagram is plotted. Depending on the pressure and temperature and components, a plait point may or may not be present. Any one tematy diagram is given for fixed temperature and pressure. As either the temperature or pressure is varied, the location of the binodal curve and

23-6

PETROLEUM ENGINEERING

HANDBOOK

1
P A , 3 z i; 1 A TEMPERATURE

1000

0.2 MOLE

0.4 FRACTION

0.6 METHANE

0.S

: I 0

Fig. 23.14-Pressure-temperature phase diagram for a mixture of fixed composition.

Fig. 23.13-Pressure-composition phase diagrams for methane/butane and methaneldecane binary systems at 160F [71C].

slopes of the tie lines may change. Fig. 23.12 shows the effect of increasing pressure on ternary phase diagrams for mixtures of methane (C 1), butane (C,), and decane (COO) at 160F. 4J The sides of the ternary diagram represent a binary system, so the ternary diagram includes whatever binary tie lines exist at the temperature and pressure of the diagram. Fig. 23.13 shows the corresponding binary phase diagrams for the C l-C4 and C 1-C 1o pairs. The Cd-C 10 pair is not shown because it forms two phases only below the vapor pressure of C4, about 120 psia at 160F (see Fig. 23.9). As shown in Fig. 23.12, at 1,000 psia the two-phase region is a hand that stretches from the C 1-C 10 side of the diagram to the tie line on the C I -Cd side. If the pressure is increased above 1,000 psia, the liquid composition line shifts to higher methane concentrations; methane is more soluble in both C4 and Cl0 at the higher pressure (see Fig. 23.13). The two-phase region detaches from the C ,-C4 side of the diagram at the critical pressure of the C I -C4 pair (about 1,800 psia). As the pressure increases above that critical pressure, the plait point moves into the interior of the diagram (Fig. 23.12, lower diagrams). With further increases in pressure, the two-phase region continues to shrink. It would disappear completely from the diagram if the pressure reached the critical pressure of the Cl-C 10 system at 160F (nearly 5,200 psia). Reservoir Fluid Systems Real reservoir fluids contain many more than the two or three components, so phase composition data can no longer be represented with two or three coordinates. In-

stead, phase diagrams, which give more limited information, are used. Fig. 23.14 shows one such diagram for a multicomponent mixture. Fig. 23.14 gives the region of temperatures and pressures at which the mixture forms two phases. The analog of Fig. 23.14 for a binary system can be obtained by taking a slice at constant mole fraction of Component 1 through the diagram in Fig. 23.7. Also given arc contours of liquid volume fractions, which indicate the fraction of total sample volume occupied by the liquid phase. Fig. 23.14 does not give any compositional information, however. In general, the compositions of coexisting liquid and vapor will be different at each temperature and pressure. At temperatures below the critical temperature (Point C), a sample of the mixture described in Fig. 23.14 splits into two phases at the bubblepoint pressure (Fig. 23.4) when the pressure is reduced from a high level. At temperatures above the critical temperature, dewpoints are observed (Fig. 23.6). In this multicomponent system, the critical temperature is no longer the maximum temperature at which two phases can exist. Instead, the critical point is the temperature and pressure at which the phase compositions and all phase properties are identical. The bubblepoint, dewpoint, and single-phase regions shown in Fig. 23.14 are sometimes used to classify reservoirs. At temperatures above the cricondentherm, the maximum temperature for the formation of two phases, only one phase occurs at any pressure. For instance, if the hydrocarbon mixture of Fig. 23.14 were to occur in a reservoir at temperature TA and pressure PA (Point A), a decline in pressure at approximately constant temperature caused by removal of fluid from the reservoir would not cause the formation of a second phase. While the fluid in the reservoir remains a single phase, the produced gas splits into two phases as it cools and expands to surface temperature and pressure at Point A. Thus, some condensate would be collected at the

PHASE DIAGRAMS

23-7

surface even though only one phase is present in the formation. The amount of condensate collected depends on the operating conditions of the separator (or separators). The lower the temperature at a given pressure, the larger the volume of condensate collected (Fig. 23.14). Dewpoint reservoirs are those for which the reservoir temperature lies between the critical temperature and the cricondentherm for the reservoir fluid. Production of fluid from a reservoir starting at Point B in Fig. 23.14 causes liquid to appear in the reservoir when the dewpoint pressure is reached, and as the pressure declines further, the saturation of liquid increases because of retrograde condensation. Because the saturation of liquid is low, only the vapor phase flows to producing wells. Thus, the overall composition of the fluid remaining in the reservoir changes continuously. However, the phase diagram shown in Fig. 23.14 is for the original composition only. The preferential removal of light hydrocarbon components in the vapor phase generates new hydrocarbon mixtures which have a greater fraction of the heavier hydrocarbons. Differential liberation experiments, in which a sample of the reservoir fluid initially at high pressure is expanded through a sequence of pressures, can be used to investigate the magnitude of the effect of pressure reduction on the vapor composition. At each pressure, a portion of the vapor is removed and analyzed. Such an experiment simulates what happens when condensate is left behind in the reservoir as the pressure declines. As the reservoir fluid becomes heavier, the boundary of the two-phase region in a diagram like Fig. 23.14 shifts to higher temperatures. Thus, the composition change also acts to drive the system toward higher liquid condensation. Such reservoirs are candidates for pressure maintenance by lean gas injection to limit the retrograde loss of condensate or for gas cycling to vaporize and recover some of the liquid hydrocarbons. Bubblepoint reservoirs are those in which the temperature is less than the critical temperature of the reservoir fluid (Point D in Fig. 23.14). These reservoirs are sometimes called undersaturated because there is insufficient gas for a gas phase at that temperature and pressure. Isothermal pressure reduction causes the appearance of a vapor phase at the bubblepoint pressure. Because the compressibility of the liquid phase is much lower than that of a vapor, the pressure in the reservoir declines rapidly during production in the single-phase region. The appearance of the much more compressible vapor phase reduces the rate of pressure decline. The volume of vapor present in the reservoir grows rapidly with reduction of reservoir pressure below the bubblepoint. Because the vapor viscosity is much lower than the liquid viscosity, and the gas relative permeability goes up markedly with increasing gas saturation, the vapor phase flows more easily. Hence, the produced GOR climbs rapidly. Again, pressure maintenance by water drive, water injection, or gas injection can substantially improve oil recovery over the 10 to 20% recovery typical of pressure depletion in these solutiongas-drive reservoirs. As in dewpoint reservoirs, the composition of the reservoir fluid changes continuously once the two-phase region is reached. There is, of course, no reason why initial reservoir temperatures and pressures cannot lie within the twophase region. Oil reservoirs with gas caps and gas reser-

r-----Fig. 23.15~Increase in OAPI gravity with depth: (a) Ordovician Ellenberger reservoirs in Delaware Val Verde basin; (b) Pennsylvanian Tensleep reservoirs in Wyoming.

voirs with some liquids present are common. There also can be considerable variation in the initial composition of the reservoir fluid. The discussion of single-phase, dewpoint, and bubblepoint reservoirs was based on a phase diagram for one fluid composition. Even for one fluid, all the types of behavior occur over a range of temperatures. In actual reservoir settings, the composition of the reservoir fluid correlates with depth and temperature. Deeper reservoirs usually contain lighter oils.j Fig. 23.15 shows the relationships between oil gravity and depth for two basins. The higher temperatures of deeper reservoirs alter the original hydrocarbon mixtures to produce lighter hydrocarbons over geologic time. 6 Low oil gravity, low temperature, and relatively small amounts of dissolved gas all combine to produce bubblepoint reservoirs. High oil gravity, high temperatures, and high GORs produce dewpoint or condensate systems. Phase Diagrams for EOR Processes Phase behavior plays an important role in a variety of EOR processes. Such processes are designed to overcome, in one way or another, the capillary forces that act to trap oil during waterflooding. In surfactant/polymer processes, the effects of capillary forces are reduced by injection of surfactant solutions that contain molecules with oil- and water-soluble portions. Such molecules migrate to the oil/water interface and reduce the interfacial tension, thereby reducing the magnitude of the capillary forces that resist movement of trapped oil. Miscible displacement processes are designed to eliminate interfaces between the oil and the displacing phase, thereby removing the effects of capillary forces between the injected fluid and the oil. Unfortunately, fluids that are strictly miscible with oil are too expensive for general use. Instead, fluids such as methane or methane enriched with intermediate hydrocarbons, CO*, or nitrogen are injected, and the required miscible displacing fluid is generated by mixing of the injected fluid with oil in the reservoir.

23-8

PETROLEUM

ENGINEERING

HANDBOOK

TYPE b1a.b

II(

DEW

POINTS

O-0-b
SURfACTANl b. TYPE III

SURfACTANl

/ & BllHE
Fig. 23.16-Ternary

c. TYPE

II( + )

\ 75% OIL
VOLUME

50%
PERCENT LIQUID

25%
PHASE

5% I

representation of phase diagrams

MOLE

PERCENT

CO2

Phase diagrams typical of those used to explain the behavior of surfactant systems are shown in Fig. 23.16. In those ternary diagrams, the components shown are no longer true thermodynamic components since they are mixtures. A crude oil contains hundreds of components, and the brine and surfactant pseudocomponents may also be complex mixtures. The simplified representation, however, has obvious advantages for describing phase behavior, and it is reasonably accurate as long as each pseudocomponent has approximately the same composition in each phase. In Fig. 23.16a for instance, the oil pseudocomponent can appear in an oil-rich phase or in a phase containing mostly surfactant and brine. If the oil solubilized into the surfactantlbrine phase is nearly the same mixture of hydrocarbons as the original oil, then the representation in terms of pseudocomponents is reasonable. The compositions shown in Fig. 23.16 are in volume fractions. An inverse lever rule similar to Eqs. 3 or 5 gives the relationship between the volumes of the two phases for a given overall composition, as illustrated in Fig. 23.16. Fig. 23.16a is a phase diagram for the liquid/liquid equilibrium behavior typical of mixtures of brines of low salinity with oil. If there is no surfactant present, the oil and brine are immiscible; mixture compositions on the base of the diagram split into essentially pure brine in equilibrium with pure oil. Addition of surfactant causes some oil to be solubilized into a microemulsion rich in brine. That phase is in equilibrium with a phase containing nearly pure oil. Thus in the low-salinity brine, the surfactant partitions into the brine phase,

Fig. 23.17-Typical pressure-composition phase diagram for a binary mixture of CO, with a crude oil at temperatures above 120F [49%].

solubilizing some oil. The plait point in Fig. 23.16a lies close to the oil comer of the diagram. Because only two phases occur and the tie lines all have negative slope, such phase is often called Type II( -). Phase diagrams for high-salinity brines are often similar to Fig. 23.16~. In the high-salinity systems the surfactant partitions into the oil phase and solubilizes water into an oil-external microemulsion. In this case the plait point is close to the brine apex on the ternary diagram. For intermediate salinities, the phase behavior can be more complex, as shown in Fig. 23.16b. According to the phase rule, if the temperature and pressure are set, then up to three phases can coexist for a three component system. If three phases do occur, then the compositions of the phases are fixed at a given temperature and pressure. The three-phase region on a ternary diagram is represented as a triangle (Fig. 23.16b). Any overall composition lying within the three-phase region splits into the same three phases. Only the amounts of each phase change as the overall composition varies in the three-phase region. The edges of the three-phase region are tie lines for the associated two-phase regions. Thus, there is a two-phase region adjacent to each of the sides of the three-phase triangle. In Fig. 23.16b, the two-phase region at low surfac.tant concentrations is too small to show on the diagram. It must be present, however, since oil and brine form only two phases in the absence of surfactant.

PHASE DIAGRAMS

23-9

Fig. 23.18-Pressure composition diagram-Gas 1 system for Rangely oil: 95% CO, and 5% methane gas system at 160aF 171Cl.

Phase behavior of COzlcrude oil systems is often summarized in pressure-composition (p-x) diagrams such as those shown in Fig. 23.17. Fig. 23.18 is an example of a p-x diagram for mixtures of CO1 (containing a small amount of methane contamination) with crude oil from the Rangely field. 8 In such diagrams, the behavior of binary mixtures of COZ with a particular oil is reported for a fixed temperature. Thus, the oil is represented as a single pseudocomponent on such a diagram. Such diagrams indicate bubble- and dewpoint pressures, the regions of pressure and composition for which two or more phases exist, and information about the volume fractions of the phases. However, they provide no information about the compositions of the phases in equilibrium. The reason for the absence of composition data is illustrated in Fig. 23. 19,3 which gives data reported by Metcalfe and Yarborough9 for a ternary system of CO*, Cd, and C 10. Binary phase data for the CO*-Cd (Ref. 10) and COz-C tc (Ref. 11) systems also are included. Fig. 23.19 shows a triangular solid within which all possible compositions (mole fractions) of COz-Cd-C to mixtures for pressures between 400 and 2,000 psia are contained. The two-phase region is bounded by a surface that connects the binary phase envelope for the COz-C ra binary pair to that on the CO+4 side ofthe diagram. That surface is divided into two parts-liquid compositions and vapor compositions. Tie lines (heavy dashed lines in Fig. 23.19) connect the compositions of liquid and vapor phases in equilibrium at a fixed pressure. Thus, the ternary phase diagram for CO2-C4-C tc mixtures at any pressure is just a constant

Fig. 23.19-Phase behavior of CO,-C,-C,, [71 Cl.

mixtures at 16OOF

pressure (horizontal) slice through the triangular prism. Several such slices at different pressures are shown in Fig. 23.19. At pressures below the critical pressure of CO 2-C 4 mixtures ( 1,184 psia) , both CO 2-C , c mixtures and CO2-C4 mixtures form two phases for some range of COz concentrations. At 400 and 800 psia, the twophase region is a band across the diagram. Above the critical pressure of COz-Cd mixtures, CO2 is miscible with Cd. and ternary slices at higher pressures show a continuous binodal curve on which the locus of liquid compositions meets that of vapor compositions at a plait point. The locus of plait points (labeled P in Fig. 23.19) connects the critical points of the two binary pairs. To see the effect of representing the phase behavior of a ternary system on a pseudobinary diagram, consider a p-x diagram for an oil composed of 70 mol% C ,c and 30 mol% CJ. At any fixed pressure, the mixtures of CO2 and oil which would be investigated in an experiment to determine ap-x diagram lie on a straight line (the dilution line), which connects the original oil composition with the CO2 apex. Thus, a P-X diagram for this

23-10

PETROLEUM ENGINEERING

HANDBOOK

LIQUID

LIQUID

J I

-,I / / /

Fig. 23.20-p-x diagrams for mixtures of CO, with Wasson oil, where L, is Liquid Phase 1 (oil-rich phase), L, is Liquid Phase 2 (COP-rich phase), and V is the vapor phase. Dashed lines indicate constant volume fraction of L, phase.

system is a vertical slice through the triangular prism shown in Fig. 23.19. The saturation pressures on a p-x diagram are those at which the dilution plane intersects the surface which bounds the two-phase region. Bubblepoint pressures (B) occur where the dilution plane intersects the liquid composition side of the two-phase surface, while dewpoint pressures (D) occur at the intersection with vapor compositions. Comparison of the phase envelope on the resulting p-x diagram with binary phase diagrams yields the following observations. 1. Tie lines do not, in general, lie in the dilution plane. Instead, they pierce that plane. This means that the composition of vapor in equilibrium with a bubblepoint mixture on the p-x diagram is not the same as that of the dewpoint mixture at the same pressure. 2. The critical point on the p-x diagram occurs where the locus of plait points pierces the dilution plane. It is not, in general, at the maximum saturation pressure on the p-x diagram. The maximum pressure occurs where the binodal curve is tangent to the dilution plane. The critical point on the p-x diagram can lie on either side of the maximum pressure, depending on the position of locus of plait points on the two-phase surface. It is apparent from Fig. 23.19 that the composition of the original oil has a strong influence on the shape of the saturation pressure curve, and on the location of the critical point on the p-x diagram. If the oil had been richer in C4, the critical pressure and maximum pressure both would have been lower. Thus, it should be anticipated that the appearance of p-x diagrams for C02icrude oil systems should depend on the composition of the oil. Figs. 23.18 and 23.20 illustrate the complexity of

phase behavior observed for COz/crude oil systems. Fig. 23.18 gives the behavior of mixtures of CO2 (with about 5 % methane as a contaminant) with Rangely crude oil at 16OF. The oil itself has a bubblepoint pressure of about 350 psia. Mixtures containing up to about 80 mol% CO;? (+C 1) show bubblepoints, while those containing more CO2 show dewpoints. At the relatively high temperature of the Rangely field, only two phases, a liquid and a vapor, form. At lower temperatures, more complex phase behavior can occur. Figs. 23.20a, b and c show the behavior of mixtures of a dead oil from the Wasson field3 with CO2. At 90F and 105F, the mixtures form a liquid and a vapor at low pressures and two liquid phases at high pressures and high CO2 concentmtions. They form three phases, two liquids and a vapor, for a small range of pressures at high CO* concentrations. The liquid/liquid and liquid/liquid/vapor behavior disappears if the temperature is high enough. At 120F (Fig. 23.2Oc), the three-phase region had disappeared. For the systems studied to date, 120F appears to be a reasonable estimate of the maximum temperature for liquid/liquid/vapor separations. For detailed discussions of such phase behavior, see Refs. 2 and 3.

Calculation of Phase Compositions

Calculations of the compositions of phases that occur for multicomponent mixtures are important for the design of surface separators and for the design of EOR processes such as high-pressure and condensing-gas drives and CO1 floods. There are two widely used methods for such calculations-K-value correlations and equations of state (EOSs).

PHASE DIAGRAMS

23-l 1

The use of K-values, also called equilibrium ratios or equilibrium constants, is based on the behavior of mixtures of gases at relatively low pressures and temperatures. According to Raoults law, the partial vapor pressure phi of component i in a liquid mixture is equal to the product of the mole fraction of component i in the liquid and its pure component vapor pressure p~i=Xjpvj. ... . . f.. . ......... . . (6)

In addition, Daltons law states that the partial pressure of component i in the vapor is p;j =yjpr, .... ......... . . . . . . . . . *. . . . (7)

where yi is the mole fraction of component i in the vapor andp, is the total pressure. Rearrangement of Eqs. 6 and 7 gives the definition of K-value for an ideal (low pressure) system:
K.-Y; I -pvi xi Pr . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (8)

PRESSURE,PSIA Fig. 23.21-Typical equilibrium ratios at 22CPF [104%] (dashed lines are the ideal ratios).

Thus, for a multicomponent mixture at low pressure, the equilibrium value can be estimated from the vapor pressure, which is a function of temperature only, and the total pressure. The assumption of an ideal gas in Raoults and Daltons laws is reasonable only if the pressure is below about 50 to 100 psia. * At higher pressures, equilibrium ratios are functions of pressure, temperature, and composition. Fig. 23.21 shows a typical set of equilibrium ratios for a hydrocarbon system containing some CO2 at 220F. I3 Also shown (as dashed lines) are the ideal equilibrium ratios. At high pressures, the K-values for ethane and heavier hydrocarbons pass through a minimum and appear to converge to a value of one, in this case at 4,200 psia. This observation is the basis for a widely used empirical correlation for K-values. K-value charts for a variety of convergence pressures and a recommended technique for estimating the convergence pressure are given in GPSAs Engineering Data Book. If K-values ate known or can be estimated, then amounts of liquid and vapor and phase compositions can be calculated easily. Consider 1 mol of a mixture in which the overall mole fraction of the ith component is zi . If the mole fractions of component i in the liquid and vapor are Xi and yi , and the fraction of the mole of mixture that is liquid is L, then a material balance gives Zi=Xil+y~(l-L). ...... . . . . . . . . . . . (9) into Eq. 9

By definition, cXi=Eyi=l, SO cXi-Cyi=O gives the nonlinear function f(L):

which

f(L)=

Ky;;;;)L 1

=o. 1

.. .... .. .

. (12)

Eq. 12 can be solved for L by application of a NewtonRaphson iteration. If Lk is the kth estimate of the solution, an improved estimate is given by Lk+l=Lkpi z where Lk ) .....,,,...,........(13)

The iterative calculation is complete when AL) as given by Eq. 12 and AL=Lk+l -Lk are both smaller than some preset tolerances. Once the liquid mole fraction has been determined, the Xi and yi are obtained from Eqs. 10 and 11. If the mixture is at its bubblepoint pressure, then L= 1 and Czi=l, and Eq. 12 reduces to C(ziKi)=l. ...........,.......... . . . . . ..(15)

Substitution of the definition of Ki=yilXi and rearrangement gives xj = Ki +(l-Ki)L.

Zi

. . . . . . . . . . . . . . . . . . . . . (10)

Similarly, Eq. 9 can be solved for yi, giving

KiZi Yi= Kj +(l-Ki)L.

Thus, if Kis are known as a function of pressure, then the bubblepoint pressure can be obtained as the pressure at which Eq. 15 is satisfied. Bubblepoint pressures are generally most sensitive to the K-values of the lightest components, which axe the largest. If the mixture is at its dewpoint pressure, then L=O, CZi = 1, and c =l. . . . . . . . . . . . . . . . . . . . . . . . . (16)

.. ..,...... . ....

. -(II)

23-12

PETROLEUM ENGINEERING

HANDBOOK

volume is small enough to be close to the constant h, then the pressure increases rapidly as the volume is reduced. Thus, the EOS is qualitatively consistent with liquid behavior when the pressure is high. The calculation of phase compositions is based on the fact that, at thermodynamic equilibrium, the fugacity of each component must be the same in each phase. The fugacity of a component in a phase can be calculated if the volumetric behavior of the phase is known. It can be shown I4 that the fugacity of component i, fi , in a phase is given by

RTlnfi=Sm[ Vt

(2) t

TV -$]dV, 3 l, t

-RTln
Fig. X3.22-Comparison of calculated and measured phase compositions for ternary mixtures of CO,. methane (C,), and decane (C,,), at 160F [71C] and 1,250 psia.

vt

niRT

.... ...

.. .. .

Dewpoint pressures are most sensitive to the smallest Kvalues, those of the heavy components, which often are least accurately known. Thus, there is often more uncertainty in calculated values of dewpoint pressure. The sums of Eqs. 15 and 16 also are useful for determining whether the mixture forms one or two phases. If C(K;z;) < 1, the mixture is all liquid. If C(ziIKi) < 1, the mixture is all vapor. If both C(K,zi)> 1 and C(K;Iz,) > 1, the mixture forms two phases. In recent years, EOSs also have been used extensively for phase equilibrium calculations. Most of the widely used EOSs are refinements of the equation proposed by van der Waals:
RT p= ---b a v2, ,,........................

where T is the temperature, p the pressure, V, the total volume, ni the number of moles of component i, and R the gas constant. If the relationship between pressure, composition, and total volume is known from an EOS, then taPlaniJT,V ,n can be obtained and the integral evaluated. A var!e<y of EOSs have been suggested for hydrocarbon mixtures. For example, the original Redlich-Kwong equation has the form
n,RT V,-0, 2 ntam

P=

T v (v +nb ) t t ,,,

. . . . .

(20)

b,

where n, =Cni is the total number of moles and a,,, and depend on the mixture composition and the critical properties of the components as follows.
(EyiA; 1/2)2R2T52 p )

am=

......... ....

. . (21)

(17) where

where p is the pressure, R the gas constant, T the temperature, and V the molar volume. The constants a and b can be determined for a particular component from thermodynamic constraints at the critical point, which requires that
(ap/w)~, =(ap%v*)T, =o,

A, = Qa(PIpci) , (T,T,i)5,2

.. ............

. . . (22)

and a,= 1 9(2 -1) b

m

EO.4275.

..................

which gives
a=-

= @Y;BiPT
P

21 R2Tc2
64 PC

, . . . . . . . . . . . . . . . . . . . . . . . (24)

where
B, = nb(P/Pci) (T,T,i) . . . . . . . . . . . . . .

and
b=F, c . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (18)

(25)

where T, is the critical temperature and pc is the critical pressure. van der Waals equation reduces to the ideal gas law if the molar volume is large (low pressure). If the molar

and 2* -1 fib=3 ~0.08664. ... .... . (26)

PHASE DIAGRAMS

23-13

In Eqs. 21 through 26, y, is the mole fraction of component i in the mixture, and pci and T,; are the critical pressure and temperature of component i. The constants Q, and fib arise from the thermodynamic constraints, =0, at the critical point. From cwa v) Tr =(a2piavl)T, Eqs. 19 through 26, an expression for the fugacity of each component in a phase can be obtained. To calculate phase compositions, the following procedure is used. 1. Estimate compositions of liquid and vapor. 2. Calculate fugacities of each component in each phase. 3. Iff,,, =fiL stop. Otherwise, obtain improved phase compositions and return to Step 2. Similar calculations can be performed for liquid/liquid and even liquid/liquid/vapor systems. Because the equations for fugacities are complex and nonlinear, computer implementation of this iterative scheme to find phase compositions is required. The Redlich-Kwong EOS given as an example here is by no means the only equation available. Many modifications to the Redlich-Kwong equation have been proposed to improve the accuracy of the predictions of and equations with different phase compositions, analytical forms are also in use. The Soave modification of the Redlich-Kwong equation and the Peng-Robinson equation are among the most widely used. Ref. 14 gives details of a variety of EOSs, and Ref. 15 is a useful collection of papers relevant to phase equilibrium calculations for hydrocarbon systems. Computer programs for such calculations are available. * EOSs currently in use are quite accurate for mixtures of light hydrocarbons for which critical properties are known and extensive phase behavior data are available. For instance, Fig. 23.22 shows a comparison of phase compositions calculated with the Peng-Robinson EOS with measured values for mixtures of CO*, C 1, and C 10 at 1,250 psia and 160F. For this well-characterized system, the calculated values agreed well with the measured compositions. For crude oil systems, phase behavior predictions are less reliable because the chamcterization of heavy components is less certain. For such systems some experimental data are required to tune the EOS to represent the particular hydrocarbon system. Improvement of the predictive power of EOSs for complex hydrocarbon systems is an area of active current research.

p = PJ,, = pd = pt = PV = pVi =

p:i = PVZ =
pyl

pv2

T,. = T, = T2 = VL = V, = Zi = Q, = Qb = Subscripts

pressure bubblepoint pressure dewpoint pressure total pressure vapor pressure vapor pressure of Component i in liquid mixture partial vapor pressure of Component i in liquid mixture any mixture of two components which form a single vapor phase pressure below vapor pressure of Component z which may form a single vapor phase pressure above vapor pressure of Component z which may form a single vapor phase critical temperature constant temperature below T, constant temperature above T,. saturated liquid volume saturated vapor volume overall mole fraction of the ith component defined in Eq. 23 defined in Eq. 26

C = number of components F = number of degrees of freedom P = number of phases

References
1. Engineering Data Book, Gas Processors Suppliers Assn., ninth edition, Tulsa (1972). 2. Stalkup, F.I. Jr.: Miscible Displacement, Monograph Series, SPE, Dallas (1983) 8. 3. Orr, F.M. Jr. and Jensen, C.M.: Interpretation of PressureComposition Phase Diagrams for CO*-Cmde Oil Systems, Sot.
Pet. Eng. J. (Oct. 1984) 485-97. 4. Reamer, H.H., Fiskin, J.M., and Sage, B.H.: Phase Equilibria in Hydrocarbon Systems, Ind. Eng. Chem. 41 (Dec. 1949) 2871. 5. Sage, B.H. and Lacey, W.N.: Thermodynamic Properties of the Lighter Parafin Hydrocarbons and Nitrogen, Monograph on API

6. 7. 8. 9.

Nomenclature a, = defined by Eq. 21

Ai = defined by Eq. 22 6, = defined by Eq. 24 Bi = defined by Eq. 25 C = critical point when liquid and vapor phases are identical fir. = liquid fugacity of Component i fiV = vapor fugacity of Component i Ki = K-value of Component i L = total moles of liquid-phase in mixture Lk = kth estimate of L by Newton-Raphson iteration no = number of constraints
Gas Processors Suppliers Assn Tulsa, OK

10. 11.

12. 13.

Research Project 37, American Petmleum Inst., New York City (1950). Hunt, J.M.: Petroleum Geochemistry and Geology, W.H. Freeman and Co., San Francisco (1979). Nelson, R.C. and Pope, GA.: Phase Relationships in Chemical Flooding, Sot. Pet. Enr. 1. (Oct. 1978) 325-38. Graue, D.J. and Zana, E.T.: Study of a Possible CO, Flood in the Rangely Field, Colorado, J. Per. Tech. (July 1981) 1312-18. Metcalfe, R.S. and Yarbomugh, L.: The Effect of Phase Equilibria on the CO* Displacement Mechanism, Sot. Pet. Eng. J. (Aug. 1979) 242-52; Trans., AIME, 267. Olds, R.H. el al.: Phase Equilibria in Hydmcarbon Systems, Ind. Eng. Chem. 41 (March 1949) 475-82. Reamer, H.H. and Sage, B.H.: Phase Equilibria in Hydrocarbon Systems. Volumetric and Phase Behavior of the n-Decane-CO2 System, J. Chem. Eng. Data 8, No. 4 (1963) 508-13. Standing, M.B.: Volumetric and Phase Behavior of Oil Field . Hydrocarbon Sysrems, SPE, Dallas (1977). Allen, F.H. and Roe, R.P.: Performance Characteristics of a Volumetric Condensate Reservoir, Trans., AIME (1950) 189,
83-90.

14. Reid, R.C., Prausnitz, J.M., and Sherwood, T.K.: The Properties of Gases and Liquids, third edition, McGraw-Hill Book Co. Inc., New York City (1977). 15. Phase Behavior. Reprint Series, SPE, Dallas (1981) 15.