COMPUTATION OF MULTIPHASE EQUILIBRIUM PHENOMENA WITH AN EQUATION

OF STATE

ABSTRACT
This paper presents a general approach to multiphase equilibrium
computations using an equation of state. Various types of cakulations are
described: flash calculations, saturation- pressure and saturation-temperature
calculations, and calculations of the feed composition corresponding to a given
saturation pressure and temperature. A stagewise procedure for flash
calculations using the QNSS method allows efficient computation of phase
equilibria for systems with any number of phases. The initial guess of the
composition of an extra phase which is required to start the iteration is
obtained from an analysis of the Gibbs-energy surface. Newtons method is
used to construct multiphase boundaries on various types of diagrams
(pressure-temperature, pressure-composition and temperature-composition).
Multiphase critical points on these boundaries <are also estimated by
interpolation. The algorithms developed are tested for a typical reservoir-oilCO, mixture which exhibits three-phase liquid l-liquid 2-vapor (L,L,V)
separation. The predictions obtained using the Peng-Robinson equation of state
are satisfactory.
INTRODUCTION
Multiphase equilibrium phenomena are common in gas processes. The
presence of three hydrocarbon phases, liquid l-liquid 2-vapor (L1L2V), occurs
frequently during rich-gas or CO, injection into low-temperature (T ( 50C or
122F) reservoirs (Shelton and Yarborough, 1977; Turek et al., 1980; Orr et al.,
1981; Gardner et al., 1981; Orr and Jensen, 1982) and in cryogenic systems
(Luks et al., 1983). This paper presents a general approach to multiphase
equilibrium calcula- tions with particular application to L,L2V systems by using
some recently developed concepts in phase-behavior computations (Michelsen,
1982a, b; Nghiem and Heidemann, 1982; Heidemann, 1983). The techniques
discussed are more general than previous work on L,L,V equilibria (Fussell,
1979; Mehra et al., 1982; R_isnes and Dalen, 1982) and use rigorous Gibbsenergy analysis to generate initial guesses. Most aspects of multiphase
computations are addressed, and efficient schemes for solving the nonlinear
equations are described. The construction of multiphase boundaries on various
types of diagrams (p ressure- temperature, pressure-composition and
temperature-composition) is also considered. Critical points on the multi- phase
boundaries are also estimated during diagram construction.

MULTIPHASE EQUILIBRIA
Phase equilibrium equations

The molar Gibbs energy G of a closed system of np phases and ns components

is given by
1
where I;; is the mole fraction of phase j, fij the fugacity of component i in phase
j, yii the mole fraction of component i in phase j, zi the global mole fraction of
component i, G,? the molar Gibbs energy of component i in the standard state,
R the universal gas constant, and T is absolute temperature. The mole fractions
zi and yji satisfy the condition

2
and the material-balance constraint
3
Since G is minimized at equilibrium, the following necessary but not suffi- cient
conditions can be derived:

where the subscript r denotes a reference phase which is selected arbitrarily. If

the multiphase equilibrium ratios (K values) are defined as by Nghiem and
Heidemann (1982), i.e.,
5
eqn. (4) can be rewritten as

6
7
The fugacity coefficient $I;, can be computed readily from an equation of state
and is a function of p, T and yij. The following relationship can also be derived
from the definition of Kij and the material-balance equation (eqn. (3)):
8
where (K;,=l (i=l,.._, n,)) by definition. A given set of Kij completely determines
the composition and mole fraction of each phase. Indeed, the phase mole
fractions Fj can be obtained by solving the equations
9
And

10
Substitution of eqn. (8) into eqn. (10) yields
11
Usually Fr is taken as a dependent variable by using eqn. (9), and eqn. (11) is
solved for the remaining Fjs by Newtons method. The compositions of the
phases are then evaluated from eqn. (8). The primary unknowns are Kij, T,p, Fj
and (--) (i= l,..., n,;j= l,..., n,; j + r). The variable (--) is used to define the feed
composition as follows:
12
where zi A and ziB are the compositions (mole fractions) of two given mixtures
A and B, respectively. Computation of the feed composition according to eqn.
(12) is useful in petroleum-reservoir engineering, since the variable (--) can be
used to define the fractions of an oil A and a gas B in a mixture. It is easy to
verify that the above variables define the system completely. The total number
of unknowns is (n, + l)(n, - 1) + 3. The equilibrium equations (eqn. (6)) and the
algebraic equations (eqn. (11)) represent a system of (n, + l)( nP - 1) equations.
Hence, three variables have to be fixed before the equations can be solved.

Flash calculations