Anda di halaman 1dari 43

Copyright © 2008, 1997, 1984, 1973, 1963, 1950, 1941, 1934 by The McGraw-Hill Companies, Inc.

All rights reserved. Manufactured in the United

States of America. Except as permitted under the United States Copyright Act of 1976, no part of this publication may be reproduced or distributed
in any form or by any means, or stored in a database or retrieval system, without the prior written permission of the publisher.


The material in this eBook also appears in the print version of this title: 0-07-151127-X.

All trademarks are trademarks of their respective owners. Rather than put a trademark symbol after every occurrence of a trademarked name, we use
names in an editorial fashion only, and to the benefit of the trademark owner, with no intention of infringement of the trademark. Where such desig-
nations appear in this book, they have been printed with initial caps.

McGraw-Hill eBooks are available at special quantity discounts to use as premiums and sales promotions, or for use in corporate training programs.
For more information, please contact George Hoare, Special Sales, at or (212) 904-4069.


This is a copyrighted work and The McGraw-Hill Companies, Inc. (“McGraw-Hill”) and its licensors reserve all rights in and to the work. Use of this
work is subject to these terms. Except as permitted under the Copyright Act of 1976 and the right to store and retrieve one copy of the work, you may
not decompile, disassemble, reverse engineer, reproduce, modify, create derivative works based upon, transmit, distribute, disseminate, sell, publish
or sublicense the work or any part of it without McGraw-Hill’s prior consent. You may use the work for your own noncommercial and personal use;
any other use of the work is strictly prohibited. Your right to use the work may be terminated if you fail to comply with these terms.

FITNESS FOR A PARTICULAR PURPOSE. McGraw-Hill and its licensors do not warrant or guarantee that the functions contained in the work will
meet your requirements or that its operation will be uninterrupted or error free. Neither McGraw-Hill nor its licensors shall be liable to you or
anyone else for any inaccuracy, error or omission, regardless of cause, in the work or for any damages resulting therefrom. McGraw-Hill has no
responsibility for the content of any information accessed through the work. Under no circumstances shall McGraw-Hill and/or its licensors be liable
for any indirect, incidental, special, punitive, consequential or similar damages that result from the use of or inability to use the work, even if any of
them has been advised of the possibility of such damages. This limitation of liability shall apply to any claim or cause whatsoever whether such claim
or cause arises in contract, tort or otherwise.

DOI: 10.1036/007151127X
This page intentionally left blank
Section 4


Hendrick C. Van Ness, D.Eng. Howard P. Isermann Department of Chemical and Bio-
logical Engineering, Rensselaer Polytechnic Institute; Fellow, American Institute of Chemical
Engineers; Member, American Chemical Society (Section Coeditor)

Michael M. Abbott, Ph.D. Deceased; Professor Emeritus, Howard P. Isermann Depart-

ment of Chemical and Biological Engineering, Rensselaer Polytechnic Institute (Section Coeditor)*

INTRODUCTION Pipe Flow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-15

Postulate 1. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-4 Nozzles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-15
Postulate 2 (First Law of Thermodynamics) . . . . . . . . . . . . . . . . . . . . . . 4-4 Throttling Process. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-16
Postulate 3. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-5 Turbines (Expanders) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-16
Postulate 4 (Second Law of Thermodynamics) . . . . . . . . . . . . . . . . . . . . 4-5 Compression Processes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-16
Postulate 5. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-5 Example 1: LNG Vaporization and Compression . . . . . . . . . . . . . . . . 4-17


Constant-Composition Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-6 Partial Molar Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-17
U, H, and S as Functions of T and P or T and V . . . . . . . . . . . . . . . . . 4-6 Gibbs-Duhem Equation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-18
The Ideal Gas Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-7 Partial Molar Equation-of-State Parameters . . . . . . . . . . . . . . . . . . . . 4-18
Residual Properties. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-7 Partial Molar Gibbs Energy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-19
Solution Thermodynamics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-19
PROPERTY CALCULATIONS FOR GASES AND VAPORS Ideal Gas Mixture Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-19
Fugacity and Fugacity Coefficient . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-19
Evaluation of Enthalpy and Entropy in the Ideal Gas State . . . . . . . . . 4-8 Evaluation of Fugacity Coefficients. . . . . . . . . . . . . . . . . . . . . . . . . . . 4-20
Residual Enthalpy and Entropy from PVT Correlations . . . . . . . . . . . . 4-9 Ideal Solution Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-20
Virial Equations of State. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-9 Excess Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-21
Cubic Equations of State . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-11 Property Changes of Mixing. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-21
Pitzer’s Generalized Correlations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-12 Fundamental Property Relations Based on the Gibbs Energy. . . . . . . . 4-21
Fundamental Residual-Property Relation. . . . . . . . . . . . . . . . . . . . . . 4-21
OTHER PROPERTY FORMULATIONS Fundamental Excess-Property Relation . . . . . . . . . . . . . . . . . . . . . . . 4-22
Liquid Phase . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-13 Models for the Excess Gibbs Energy. . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-23
Liquid/Vapor Phase Transition. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-13 Behavior of Binary Liquid Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . 4-26


Mass, Energy, and Entropy Balances for Open Systems . . . . . . . . . . . . 4-14 Criteria . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-26
Mass Balance for Open Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-14 Phase Rule. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-27
General Energy Balance. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-14 Example 2: Application of the Phase Rule . . . . . . . . . . . . . . . . . . . . . 4-27
Energy Balances for Steady-State Flow Processes . . . . . . . . . . . . . . . 4-14 Duhem’s Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-27
Entropy Balance for Open Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-14 Vapor/Liquid Equilibrium . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-28
Summary of Equations of Balance for Open Systems . . . . . . . . . . . . 4-15 Gamma/Phi Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-28
Applications to Flow Processes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-15 Modified Raoult’s Law . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-28
Duct Flow of Compressible Fluids . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-15 Example 3: Dew and Bubble Point Calculations . . . . . . . . . . . . . . . . 4-29

*Dr. Abbott died on May 31, 2006. This, his final contribution to the literature of chemical engineering, is deeply appreciated, as are his earlier contributions to
the handbook.


Copyright © 2008, 1997, 1984, 1973, 1963, 1950, 1941, 1934 by The McGraw-Hill Companies, Inc. Click here for terms of use.

Data Reduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-30 Equilibrium Constants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-36

Solute/Solvent Systems. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-31 Example 6: Single-Reaction Equilibrium . . . . . . . . . . . . . . . . . . . . . . 4-37
K Values, VLE, and Flash Calculations . . . . . . . . . . . . . . . . . . . . . . . . 4-31 Complex Chemical Reaction Equilibria . . . . . . . . . . . . . . . . . . . . . . . 4-38
Example 4: Flash Calculation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-32
Equation-of-State Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-32 THERMODYNAMIC ANALYSIS OF PROCESSES
Extrapolation of Data with Temperature. . . . . . . . . . . . . . . . . . . . . . . 4-34
Example 5: VLE at Several Temperatures . . . . . . . . . . . . . . . . . . . . . 4-34 Calculation of Ideal Work. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-38
Liquid/Liquid and Vapor/Liquid/Liquid Equilibria . . . . . . . . . . . . . . . . 4-35 Lost Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-39
Chemical Reaction Stoichiometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-35 Analysis of Steady-State Steady-Flow Proceses. . . . . . . . . . . . . . . . . . . . 4-39
Chemical Reaction Equilibria . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-35 Example 7: Lost-Work Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-40
Standard Property Changes of Reaction . . . . . . . . . . . . . . . . . . . . . . . 4-35

Nomenclature and Units

Correlation- and application-specific symbols are not shown.
U.S. Customary U.S. Customary
Symbol Definition SI units System units Symbol Definition SI units System units
A Molar (or unit-mass) J/mol [J/kg] Btu/lb mol Pisat Saturation or vapor pressure kPa psi
Helmholtz energy [Btu/lbm] of species i
A Cross-sectional area in flow m ft2 Q Heat J Btu
âi Activity of species i Dimensionless Dimensionless q Volumetric flow rate m3/s ft3/s
in solution ⋅
⎯a Q Rate of heat transfer J/s Btu/s
i Partial parameter, cubic R Universal gas constant J/(mol·K) Btu/(lb·mol·R)
equation of state S Molar (or unit-mass) entropy J/(mol·K) Btu/(lb·mol·R)
B 2d virial coefficient, cm3/mol cm3/mol [J/(kg·K)] [Btu/(lbm·R)]
density expansion ⋅
⎯ 3
SG Rate of entropy generation, J/(K·s) Btu/(R·s)
Bi Partial molar second cm /mol cm3/mol Eq. (4-151)
virial coefficient T Absolute temperature K R
B̂ Reduced second virial Tc Critical temperature K R
coefficient U Molar (or unit-mass) J/mol [J/kg] Btu/(lb·mol)
C 3d virial coefficient, density cm6/mol2 cm6/mol2 internal energy [Btu/lbm]
expansion u Fluid velocity m/s ft/s
Ĉ Reduced third virial coefficient V Molar (or unit-mass) volume m3/mol [m3/kg] ft3/(lb·mol)
D 4th virial coefficient, density cm9/mol3 cm9/mol3 [ft3/lbm]
expansion W Work J Btu
B′ 2d virial coefficient, pressure kPa−1 kPa−1 Ws Shaft work for flow process J Btu
expansion ⋅
Ws Shaft power for flow process J/s Btu/s
C′ 3d virial coefficient, pressure kPa−2 kPa−2 xi Mole fraction in general
expansion xi Mole fraction of species i in
D′ 4th virial coefficient, kPa−3 kPa−3 liquid phase
pressure expansion yi Mole fraction of species i in
Bij Interaction 2d virial cm3/mol cm3/mol vapor phase
coefficient Z Compressibility factor Dimensionless Dimensionless
Cijk Interaction 3d virial cm6/mol2 cm6/mol2 z Elevation above a datum level m ft
CP Heat capacity at constant J/(mol·K) Btu/(lb·mol·R) Superscripts
CV Heat capacity at constant J/(mol·K) Btu/(lb·mol·R) E Denotes excess thermodynamic property
volume id Denotes value for an ideal solution
fi Fugacity of pure species i kPa psi ig Denotes value for an ideal gas
l Denotes liquid phase
fˆi Fugacity of species i in solution kPa psi
G Molar (or unit-mass) J/mol [J/kg] Btu/(lb·mol) lv Denotes phase transition, liquid to vapor
Gibbs energy [Btu/lbm] R Denotes residual thermodynamic property
g Acceleration of gravity m/s2 ft/s2 t Denotes total value of property
g ≡ GE/RT Dimensionless Dimensionless v Denotes vapor phase
H Molar (or unit-mass) enthalpy J/mol [J/kg] Btu/(lb·mol) ∞ Denotes value at infinite dilution
[Btu/lbm] Subscripts
Ki Equilibrium K value, yi /xi Dimensionless Dimensionless
Kj Equilibrium constant for Dimensionless Dimensionless c Denotes value for the critical state
chemical reaction j cv Denotes the control volume
k1 Henry’s constant for kPa psi fs Denotes flowing streams
solute species 1 n Denotes the normal boiling point
M Molar or unit-mass solution r Denotes a reduced value
property (A, G, H, S, U, V) rev Denotes a reversible process
M Mach number Dimensionless Dimensionless
Greek Letters
Mi Molar or unit-mass
pure-species property α, β As superscripts, identify phases
⎯ (Ai, Gi, Hi, Si, Ui, Vi) β Volume expansivity K−1 °R−1
Mi Partial property of species i εj Reaction coordinate for mol lb·mol
in⎯ solution
⎯ ⎯ ⎯ ⎯ ⎯ reaction j
(Ai, Gi, Hi, Si, Ui, Vi) Γi(T) Defined by Eq. (4-196) J/mol Btu/(lb·mol)
MR Residual thermodynamic property γ Heat capacity ratio CP /CV Dimensionless Dimensionless
(AR, GR, HR, SR, UR, VR) γi Activity coefficient of species i Dimensionless Dimensionless
ME Excess thermodynamic property in solution
⎯E (AE, GE, HE, SE, UE, VE) κ Isothermal compressibility kPa−1 psi−1
Mi Partial molar excess thermodynamic µi Chemical potential of species i J/mol Btu/(lb·mol)
property νi,j Stoichiometric number Dimensionless Dimensionless
∆M Property change of mixing of species i in reaction j
(∆ A, ∆G, ∆H, ∆S, ∆U, ∆V) ρ Molar density mol/m3 lb·mol/ft3
∆M°j Standard property change of reaction j σ As subscript, denotes a
(∆Gj°, ∆Hj°, ∆CP°)
j heat reservoir
m Mass kg lbm Φi Defined by Eq. (4-304) Dimensionless Dimensionless
m⋅ Mass flow rate kg/s lbm/s φi Fugacity coefficient of Dimensionless Dimensionless
n Number of moles pure species i
n⋅ Molar flow rate φ̂ i Fugacity coefficient of Dimensionless Dimensionless
ni Number of moles of species i species i in solution
P Absolute pressure kPa psi ω Acentric factor Dimensionless Dimensionless
GENERAL REFERENCES: Abbott, M. M., and H. C. Van Ness, Schaum’s Out- Wiley, New York, 1999. Smith, J. M., H. C. Van Ness, and M. M. Abbott,
line of Theory and Problems of Thermodynamics, 2d ed., McGraw-Hill, New Introduction to Chemical Engineering Thermodynamics, 7th ed., McGraw-
York, 1989. Poling, B. E., J. M. Prausnitz, and J. P. O’Connell, The Properties Hill, New York, 2005. Tester, J. W., and M. Modell, Thermodynamics and Its
of Gases and Liquids, 5th ed., McGraw-Hill, New York, 2001. Prausnitz, J. M., Applications, 3d ed., Prentice-Hall PTR, Upper Saddle River, N.J., 1997. Van
R. N. Lichtenthaler, and E. G. de Azevedo, Molecular Thermodynamics of Ness, H. C., and M. M. Abbott, Classical Thermodynamics of Nonelectrolyte
Fluid-Phase Equilibria, 3d ed., Prentice-Hall PTR, Upper Saddle River, N.J., Solutions: With Applications to Phase Equilibria, McGraw-Hill, New York,
1999. Sandler, S. I., Chemical and Engineering Thermodynamics, 3d ed., 1982.


Thermodynamics is the branch of science that lends substance to the Thermodynamics finds its origin in experience and experiment,
principles of energy transformation in macroscopic systems. The gen- from which are formulated a few postulates that form the foundation
eral restrictions shown by experience to apply to all such transfor- of the subject. The first two deal with energy.
mations are known as the laws of thermodynamics. These laws are
primitive; they cannot be derived from anything more basic. POSTULATE 1
The first law of thermodynamics states that energy is conserved,
that although it can be altered in form and transferred from one place There exists a form of energy, known as internal energy, which for
to another, the total quantity remains constant. Thus the first law of systems at internal equilibrium is an intrinsic property of the system,
thermodynamics depends on the concept of energy, but conversely functionally related to the measurable coordinates that characterize
energy is an essential thermodynamic function because it allows the the system.
first law to be formulated. This coupling is characteristic of the primi-
tive concepts of thermodynamics. POSTULATE 2 (FIRST LAW OF THERMODYNAMICS)
The words system and surroundings are similarly coupled. A system
can be an object, a quantity of matter, or a region of space, selected for The total energy of any system and its surroundings is conserved.
study and set apart (mentally) from everything else, which is called the Internal energy is quite distinct from such external forms as the
surroundings. An envelope, imagined to enclose the system and to kinetic and potential energies of macroscopic bodies. Although it is a
separate it from its surroundings, is called the boundary of the system. macroscopic property, characterized by the macroscopic coordinates
Attributed to this boundary are special properties which may serve T and P, internal energy finds its origin in the kinetic and potential
either to isolate the system from its surroundings or to provide for energies of molecules and submolecular particles. In applications of
interaction in specific ways between the system and surroundings. An the first law of thermodynamics, all forms of energy must be consid-
isolated system exchanges neither matter nor energy with its sur- ered, including the internal energy. It is therefore clear that postulate
roundings. If a system is not isolated, its boundaries may permit 2 depends on postulate 1. For an isolated system the first law requires
exchange of matter or energy or both with its surroundings. If the that its energy be constant. For a closed (but not isolated) system, the
exchange of matter is allowed, the system is said to be open; if only first law requires that energy changes of the system be exactly com-
energy and not matter may be exchanged, the system is closed (but not pensated by energy changes in the surroundings. For such systems
isolated), and its mass is constant. energy is exchanged between a system and its surroundings in two
When a system is isolated, it cannot be affected by its surroundings. forms: heat and work.
Nevertheless, changes may occur within the system that are detectable Heat is energy crossing the system boundary under the influence of
with measuring instruments such as thermometers and pressure gauges. a temperature difference or gradient. A quantity of heat Q represents
However, such changes cannot continue indefinitely, and the system an amount of energy in transit between a system and its surroundings,
must eventually reach a final static condition of internal equilibrium. and is not a property of the system. The convention with respect to
For a closed system which interacts with its surroundings, a final sign makes numerical values of Q positive when heat is added to the
static condition may likewise be reached such that the system is not system and negative when heat leaves the system.
only internally at equilibrium but also in external equilibrium with its Work is again energy in transit between a system and its surround-
surroundings. ings, but resulting from the displacement of an external force acting
The concept of equilibrium is central in thermodynamics, for asso- on the system. Like heat, a quantity of work W represents an amount
ciated with the condition of internal equilibrium is the concept of of energy, and is not a property of the system. The sign convention,
state. A system has an identifiable, reproducible state when all its analogous to that for heat, makes numerical values of W positive when
properties, such as temperature T, pressure P, and molar volume V, work is done on the system by the surroundings and negative when
are fixed. The concepts of state and property are again coupled. One work is done on the surroundings by the system.
can equally well say that the properties of a system are fixed by its When applied to closed (constant-mass) systems in which only
state. Although the properties T, P, and V may be detected with mea- internal-energy changes occur, the first law of thermodynamics is
suring instruments, the existence of the primitive thermodynamic expressed mathematically as
properties (see postulates 1 and 3 following) is recognized much more
dUt = dQ + dW (4-1)
indirectly. The number of properties for which values must be speci-
fied in order to fix the state of a system depends on the nature of the where U is the total internal energy of the system. Note that dQ and
system, and is ultimately determined from experience. dW, differential quantities representing energy exchanges between
When a system is displaced from an equilibrium state, it undergoes the system and its surroundings, serve to account for the energy
a process, a change of state, which continues until its properties attain change of the surroundings. On the other hand, dUt is directly the
new equilibrium values. During such a process, the system may be differential change in internal energy of the system. Integration of Eq.
caused to interact with its surroundings so as to interchange energy in (4-1) gives for a finite process
the forms of heat and work and so to produce in the system changes
∆Ut = Q + W (4-2)
considered desirable for one reason or another. A process that pro-
ceeds so that the system is never displaced more than differentially where ∆U is the finite change given by the difference between the

from an equilibrium state is said to be reversible, because such a final and initial values of Ut. The heat Q and work W are finite quan-
process can be reversed at any point by an infinitesimal change in tities of heat and work; they are not properties of the system or func-
external conditions, causing it to retrace the initial path in the opposite tions of the thermodynamic coordinates that characterize the
direction. system.


POSTULATE 3 For closed systems of this kind the work of a reversible process may
always be calculated from
There exists a property called entropy, which for systems at internal
equilibrium is an intrinsic property of the system, functionally related dWrev = −PdV t (4-4)
to the measurable coordinates that characterize the system. For where P is the absolute pressure and Vt is the total volume of the sys-
reversible processes, changes in this property may be calculated by tem. This equation follows directly from the definition of mechanical
the equation work.
dSt =  (4-3)
where S is the total entropy of the system and T is the absolute tem- The macroscopic properties of homogeneous PVT systems at internal
perature of the system. equilibrium can be expressed as functions of temperature, pressure,
and composition only.
This postulate imposes an idealization, and is the basis for all subse-
POSTULATE 4 (SECOND LAW OF THERMODYNAMICS) quent property relations for PVT systems. The PVT system serves as a
satisfactory model in an enormous number of practical applications.
The entropy change of any system and its surroundings, considered In accepting this model one assumes that the effects of fields (e.g.,
together, resulting from any real process is positive, approaching electric, magnetic, or gravitational) are negligible and that surface and
zero when the process approaches reversibility. viscous shear effects are unimportant.
In the same way that the first law of thermodynamics cannot be Temperature, pressure, and composition are thermodynamic coor-
formulated without the prior recognition of internal energy as a prop- dinates representing conditions imposed upon or exhibited by the sys-
erty, so also the second law can have no complete and quantitative tem, and the functional dependence of the thermodynamic properties
expression without a prior assertion of the existence of entropy as a on these conditions is determined by experiment. This is quite direct
property. for molar or specific volume V, which can be measured, and leads
The second law requires that the entropy of an isolated system immediately to the conclusion that there exists an equation of state
either increase or, in the limit where the system has reached an equi- relating molar volume to temperature, pressure, and composition for
librium state, remain constant. For a closed (but not isolated) system any particular homogeneous PVT system. The equation of state is a
it requires that any entropy decrease in either the system or its sur- primary tool in applications of thermodynamics.
roundings be more than compensated by an entropy increase in the Postulate 5 affirms that the other molar or specific thermodynamic
other part, or that in the limit where the process is reversible, the total properties of PVT systems, such as internal energy U and entropy S,
entropy of the system plus its surroundings be constant. are also functions of temperature, pressure, and composition. These
The fundamental thermodynamic properties that arise in connection molar or unit-mass properties, represented by the plain symbols V, U,
with the first and second laws of thermodynamics are internal energy and S, are independent of system size and are called intensive. Tem-
and entropy. These properties together with the two laws for which they perature, pressure, and the composition variables, such as mole frac-
are essential apply to all types of systems. However, different types of tion, are also intensive. Total-system properties (V t, U t, St) do depend
systems are characterized by different sets of measurable coordinates or on system size and are extensive. For a system containing n mol of
variables. The type of system most commonly encountered in chemical fluid, Mt = nM, where M is a molar property.
technology is one for which the primary characteristic variables are tem- Applications of the thermodynamic postulates necessarily involve
perature T, pressure P, molar volume V, and composition, not all of the abstract quantities of internal energy and entropy. The solution of
which are necessarily independent. Such systems are usually made up any problem in applied thermodynamics is therefore found through
of fluids (liquid or gas) and are called PVT systems. these quantities.


Consider a single-phase closed system in which there are no chemical where subscript n indicates that all mole numbers ni (and hence n)
reactions. Under these restrictions the composition is fixed. If such a are held constant. Comparison with Eq. (4-5) shows that
system undergoes a differential, reversible process, then by Eq. (4-1)
∂(nU) ∂(nU)
dUt = dQrev + dWrev 

= T and 

= −P
Substitution for dQrev and dWrev by Eqs. (4-3) and (4-4) gives
For an open single-phase system, we assume that nU = U (nS, nV,
dUt = T dSt − P dVt
n1, n2, n3, . . .). In consequence,
Although derived for a reversible process, this equation relates prop-
∂(nU) ∂(nU) ∂(nU)
erties only and is valid for any change between equilibrium states in a
closed system. It is equally well written as 
d(nU) = 
∂(nS) nV,n

d(nS) + 
d(nV) +

i ∂ni  

d(nU) = T d(nS) − P d(nV) (4-5) where the summation is over all species present in the system and
where n is the number of moles of fluid in the system and is constant subscript nj indicates that all mole numbers are held constant except
for the special case of a closed, nonreacting system. Note that the ith. Define
n n1 + n2 + n3 + … =
∂ni nS,nV,nj
where i is an index identifying the chemical species present. When U,
S, and V represent specific (unit-mass) properties, n is replaced by m. The expressions for T and −P of the preceding paragraph and the def-
Equation (4-5) shows that for a single-phase, nonreacting, closed inition of µi allow replacement of the partial differential coefficients in
system, nU = u(nS, nV). the preceding equation by T, −P, and µi. The result is Eq. (4-6) of
Table 4-1, where important equations of this section are collected.
∂(nU) ∂(nU)
Then d(nU) =  
∂(nS) nV,n
d(nS) +  
∂(nV) nS,n 
d(nV) Equation (4-6) is the fundamental property relation for single-phase
PVT systems, from which all other equations connecting properties of

TABLE 4-1 Mathematical Structure of Thermodynamic Property Relations

For homogeneous systems of
Primary thermodynamic functions Fundamental property relations constant composition Maxwell equations
∂T ∂P
U = TS − PV +
(4-7) d(nU) = T d(nS) − P d(nV) +
µi dni
(4-6) dU = T dS − P dV (4-14)  = − ∂S

∂V S V

∂T ∂V
H U + PV (4-8) d(nH) = T d(nS) + nV dP +
µi dni
(4-11) dH = T dS + V dP (4-15) ∂P = ∂S

∂P ∂S
A U − TS (4-9) d(nA) = − nS dT − P d(nV) +
µi dni
(4-12) dA = −S dT − P dV (4-16) ∂T = ∂V

∂V ∂S
G H − TS (4-10) d(nG) = − nS dT + nV dP +
µi dni
(4-13) dG = −S dT + V dP (4-17) ∂T = − ∂P

U, H, and S as functions of T and P or T and V Partial derivatives Total derivatives

∂H ∂H ∂H ∂S ∂V

dH = 
∂T P

dT + 
∂P T
dP (4-22) 

∂T P
= T 
∂T  =C P
P (4-28) 
dH = CP dT + V − T 
∂T   dPP

∂S ∂S ∂H ∂S ∂V ∂V
 dT + ∂P ∂P   
dS =  dP (4-23) = T  + V = V − T  (4-29) dS = P dT −  dP (4-33)
∂T P T T ∂P T ∂T P T ∂T P
∂U ∂U ∂U ∂S ∂P

dU = 
∂T V

dT + 
∂V T
dV (4-24) ∂T V
= T 
∂T  V
= CV (4-30)   − P dV
dU = CV dT + T 
∂T V

∂S ∂S ∂U ∂S ∂P ∂P
dS =  dT +  dV (4-25) = T  − P = T  −P (4-31) dS =  dT +  dV (4-35)
∂T V ∂V T T ∂V T ∂T V T ∂T V
U Internal energy; H enthalpy; A Helmoholtz energy; G Gibbs energy.

such systems are derived. The quantity µ i is called the chemical poten- variables, called the canonical variables for the property. The choice
tial of species i, and it plays a vital role in the thermodynamics of of which equation to use in a particular application is dictated by con-
phase and chemical equilibria. venience. However, the Gibbs energy G is special, because of its rela-
Additional property relations follow directly from Eq. (4-6). tion to the canonical variables T, P, and {ni}, the variables of primary
Because ni = xin, where xi is the mole fraction of species i, this equa- interest in chemical processing. Another set of equations results from
tion may be rewritten as the substitutions n = 1 and ni = xi. The resulting equations are of
course less general than their parents. Moreover, because the mole
d(nU) − T d(nS) + P d(nV) −
µi d(xin) = 0 fractions are not independent, mathematical operations requiring
i their independence are invalid.
Expansion of the differentials and collection of like terms yield
dU − T dS + P dV −
µ dx n + U − TS + PV −
x µ dn = 0
i i
i i
For 1 mol of a homogeneous fluid of constant composition, Eqs. (4-6)
Because n and dn are independent and arbitrary, the terms in brackets and (4-11) through (4-13) simplify to Eqs. (4-14) through (4-17) of
must separately be zero. This provides two useful equations: Table 4-1. Because these equations are exact differential expressions,
application of the reciprocity relation for such expressions produces
dU = T dS − P dV +
µi dxi U = TS − PV +
xiµi the common Maxwell relations as described in the subsection “Multi-
i i variable Calculus Applied to Thermodynamics” in Sec. 3. These are
The first is similar to Eq. (4-6). However, Eq. (4-6) applies to a sys- Eqs. (4-18) through (4-21) of Table 4-1, in which the partial deriva-
tem of n mol where n may vary. Here, however, n is unity and invari- tives are taken with composition held constant.
ant. It is therefore subject to the constraints
i xi = 1 and
i dxi = 0. U, H, and S as Functions of T and P or T and V At constant
composition, molar thermodynamic properties can be considered
Mole fractions are not independent of one another, whereas the mole
functions of T and P (postulate 5). Alternatively, because V is related
numbers in Eq. (4-6) are.
to T and P through an equation of state, V can serve rather than P as
The second of the preceding equations dictates the possible com-
the second independent variable. The useful equations for the total
binations of terms that may be defined as additional primary func-
differentials of U, H, and S that result are given in Table 4-1 by Eqs.
tions. Those in common use are shown in Table 4-1 as Eqs. (4-7)
(4-22) through (4-25). The obvious next step is substitution for the
through (4-10). Additional thermodynamic properties are related to
partial differential coefficients in favor of measurable quantities. This
these and arise by arbitrary definition.
purpose is served by definition of two heat capacities, one at constant
Multiplication of Eq. (4-8) of Table 4-1 by n and differentiation
pressure and the other at constant volume:
yield the general expression
d(nH) = d(nU) + P d(nV) + nV dP C P 
∂T  P
Substitution for d(nU) by Eq. (4-6) reduces this result to Eq. (4-11).

The total differentials of nA and nG are obtained similarly and are
expressed by Eqs. (4-12) and (4-13). These equations and Eq. (4-6) CV  (4-27)
∂T V
are equivalent forms of the fundamental property relation, and appear
under that heading in Table 4-1. Each expresses a total property—nU, Both are properties of the material and functions of temperature,
nH, nA, and nG—as a function of a particular set of independent pressure, and composition.

Equation (4-15) of Table 4-1 may be divided by dT and restricted 10

to constant P, yielding (∂H/∂T)P as given by the first equality of Eq.
(4-28). Division of Eq. (4-15) by dP and restriction to constant T yield
(∂H/∂P)T as given by the first equality of Eq. (4-29). Equation (4-28) is
completed by Eq. (4-26), and Eq. (4-29) is completed by Eq. (4-21).
Similarly, equations for (∂U/∂T)V and (∂U/∂V)T derive from Eq. (4-14),
and these with Eqs. (4-27) and (4-20) yield Eqs. (4-30) and (4-31) of Z = 1.02
Table 4-1. 1
Equations (4-22), (4-26), and (4-29) combine to yield Eq. (4-32);
Eqs. (4-23), (4-28), and (4-21) to yield Eq. (4-33); Eqs. (4-24), (4-27),
and (4-31) to yield Eq. (4-34); and Eqs. (4-25), (4-30), and (4-20) to
yield Eq. (4-35).
Equations (4-32) and (4-33) are general expressions for the enthalpy
and entropy of homogeneous fluids at constant composition as func-
tions of T and P. Equations (4-34) and (4-35) are general expressions Z = 0.98
Pr 0.1
for the internal energy and entropy of homogeneous fluids at constant
composition as functions of temperature and molar volume. The coef-
ficients of dT, dP, and dV are all composed of measurable quantities.
The Ideal Gas Model An ideal gas is a model gas comprising
imaginary molecules of zero volume that do not interact. Its PVT
behavior is represented by the simplest of equations of state PVig = RT,
where R is a universal constant, values of which are given in Table 1-9.
The following partial derivatives, all taken at constant composition, 0.01
are obtained from this equation:

∂P ∂Vig Vig ∂P

=  =  =  =  = −
V Vig T P PT T Vig

The first two of these relations when substituted appropriately into 0.001
Eqs. (4-29) and (4-31) of Table 4-1 lead to very simple expressions for 0 1 2 3 4
ideal gases:
∂Uig ∂Hig ∂Sig ∂Sig

∂V  ∂P 
=  =0 = − =  FIG. 4-1 Region where Z lies between 0.98 and 1.02, and the ideal-gas equa-
T T T P T Vig tion is a reasonable approximation. [Smith, Van Ness, and Abbott, Introduction
to Chemical Engineering Thermodynamics, 7th ed., p. 104, McGraw-Hill, New
Moreover, Eqs. (4-32) through (4-35) become York (2005).]

CPig R
dHig = CPig dT dSig =  dT −  dP
For the Gibbs energy, Gig = Hig − TSig; whence by Eqs. (4-37) and
CVig R
dU ig = CigV dT dSig =  dT +  dV (4-38):
T Vig
Gig =
yiGigi + RT
yi ln yi (4-39)
In these equations Vig, Uig, CVig, Hig, CPig, and Sig are ideal gas state i i
values—the values that a PVT system would have were the ideal gas
equation the true equation of state. They apply equally to pure species The ideal gas model may serve as a reasonable approximation to real-
and to constant-composition mixtures, and they show that Uig, CVig, Hig, ity under conditions indicated by Fig. 4-1.
and CPig, are functions of temperature only, independent of P and V. Residual Properties The differences between true and ideal gas
The entropy, however, is a function of both T and P or of both T and V. state properties are defined as residual properties MR:
Regardless of composition, the ideal gas volume is given by Vig = RT/P,
and it provides the basis for comparison with true molar volumes MR M − Mig (4-40)
through the compressibility factor Z. By definition, where M is the molar value of an extensive thermodynamic property
of a fluid in its actual state and Mig is its corresponding ideal gas
Z  =  =  (4-36) state value at the same T, P, and composition. Residual properties
Vig RTP RT depend on interactions between molecules and not on characteristics
of individual molecules. Because the ideal gas state presumes the
The ideal gas state properties of mixtures are directly related to the absence of molecular interactions, residual properties reflect devia-
ideal gas state properties of the constituent pure species. For those tions from ideality. The most commonly used residual properties are
ig ig
properties that are independent of P—Uig, Hig, CV , and CP —the mix- as follows:
ture property is the sum of the properties of the pure constituent
species, each weighted by its mole fraction: Residual volume VR V − Vig Residual enthalpy HR H − Hig
M =
ig ig
i (4-37) Residual entropy S S − S Residual Gibbs energy GR G − Gig
R ig

where M can represent any of the properties listed. For the entropy, Useful relations connecting these residual properties derive from
which is a function of both T and P, an additional term is required to Eq. (4-17), an alternative form of which follows from the mathemati-
account for the difference in partial pressure of a species between its cal identity:
pure state and its state in a mixture:
Sig =
yiSigi − R
yi ln yi
i i
(4-38)  RT
G 1 dG − G dT
d   2

Substitution for dG by Eq. (4-17) and for G by Eq. (4-10) gives, after Smith, Van Ness, and Abbott [Introduction to Chemical Engineering
algebraic reduction, Thermodynamics, 7th ed., pp. 210–211, McGraw-Hill, New York
(2005)] show that it is permissible here to set the lower limit of inte-
d  =  dP − 2 dT
(4-41) gration (GR/RT)P=0 equal to zero. Note also that the integrand (Z − 1)/P
remains finite as P → 0. Differentiation of Eq. (4-45) with respect to
T in accord with Eq. (4-44) gives
This equation may be written for the special case of an ideal gas and

subtracted from Eq. (4-41) itself, yielding HR ∂Z

 = −T  (constant T) (4-46)
R R R RT 0 P

d  =  dP − 2 dT (4-42)
RT RT RT Because G = H − TS and Gig = Hig − TSig, then by difference, GR =
VR ∂(GR/RT) HR − TSR, and
As a consequence,  = 
RT ∂P   T
(4-43) SR HR GR
 =  −  (4-47)
and  = −T 
RT ∂T  
Equations (4-45) through (4-47) provide the basis for calculation of
residual properties from PVT correlations. They may be put into gen-
Equation (4-43) provides a direct link to PVT correlations through eralized form by substitution of the relationships
the compressibility factor Z as given by Eq. (4-36). Thus, with V = P = Pc Pr T = TcTr
dP = Pc dPr dT = Tc dTr
VR V − Vig =  −  =  (Z − 1) The resulting equations are
This equation in combination with a rearrangement of Eq. (4-43)  = (Z − 1) r (4-48)
RT 0 Pr

Pr ∂Z

dP  = −Tr2  r (4-49)
d  =  dP = (Z − 1)  (constant T) RTc 0 ∂Tr P Pr

The terms on the right sides of these equations depend only on the
Integration from P = 0 to arbitrary pressure P gives upper limit Pr of the integrals and on the reduced temperature at
which they are evaluated. Thus, values of GR/RT and HR/RTc may be
 (Z − 1) 
GR dP determined once and for all at any reduced temperature and pressure
 = (constant T) (4-45)
RT 0 P from generalized compressibility factor data.


The most satisfactory calculation procedure for the thermodynamic

dT P
properties of gases and vapors is based on ideal gas state heat capaci- S = Sig0 +  − R ln  + SR
P (4-51)
ties and residual properties. Of primary interest are the enthalpy and T T0 P0
entropy; these are given by rearrangement of the residual property The reference state at T0 and P0 is arbitrarily selected, and the values
definitions: assigned to Hig0 and Sig0 are also arbitrary. In practice, only changes in H
and S are of interest, and fixed reference state values ultimately can-
H = Hig + HR and S = Sig + SR cel in their calculation.
These are simple sums of the ideal gas and residual properties, evalu- The ideal gas state heat capacity CPig is a function of T but not of P.
ated separately. For a mixture the heat capacity is simply the molar average
iyiCigP . i
Empirical equations relating CP to T are available for many pure
P = A + BT + CT 2 + DT −2 (4-52)
For the ideal gas state at constant composition: R
where A, B, C, and D are constants characteristic of the particular gas,
dT dP and either C or D is zero. The ratio CPig /R is dimensionless; thus the
dHig = CigP dT and dSig = CigP  − R 
T P units of CPig are those of R. Data for ideal gas state heat capacities are
given for many substances in Table 2-155.
Integration from an initial ideal gas reference state at conditions T0 Evaluation of the integrals ∫ CPig dT and ∫ (CPig /T) dT is accomplished
and P0 to the ideal gas state at T and P gives by substitution for CPig, followed by integration. For temperature
C limits of T0 and T and with τ T/T0, the equations that follow from
Hig = Hig0 + ig
P dT Eq. (4-52) are

 CR dT = AT (τ − 1) + B2 T (τ D τ−1

T ig

C − 1) +  T 03 (τ 3 − 1) +  
dT P 2 2
Sig = Sig0 + ig
P  − R ln  T0
0 0
3 T0 τ
T0 T P0
Substitution into the equations for H and S yields
 RCT dT = A ln τ + BT + CT +  τ+1
T ig

τ T  2 
C  (τ − 1)
P 2
T 0 0 2 2
H=H + ig
P dT + H R
(4-50) T0 0

Equations (4-50) and (4-51) may sometimes be advantageously or three terms, with B′ and C′ depending on temperature and compo-
expressed in alternative form through use of mean heat capacities: sition only. Moreover, the two sets of coefficients are related:
B′ = BRT (4-65)
H = Hig0 + 〈CigP 〉H(T − T0) + HR (4-55)
C′ = (C − B2)(RT)2 (4-66)
S = Sig0 + 〈CigP 〉 S ln  − R ln  + SR (4-56) Values can often be found for B, but not so often for C. Generalized
T0 P0 correlations for both B and C are given by Meng, Duan, and Li [Fluid
Phase Equilibria 226: 109–120 (2004)].
where 〈CigP 〉H and 〈CigP 〉S are mean heat capacities specific, respectively, For pressures up to several bars, the two-term expansion in pres-
for enthalpy and entropy calculations. They are given by the following sure, with B′ given by Eq. (4-65), is usually preferred:
Z = 1 + B′P = 1 + BPRT (4-67)
〈CigP 〉H B C D
 = A +  T0(τ + 1) +  T 20(τ 2 + τ + 1) + 2 (4-57) For supercritical temperatures, it is satisfactory to ever higher pres-
R 2 3 τT 0 sures as the temperature increases. For pressures above the range
where Eq. (4-67) is useful, but below the critical pressure, the virial
〈CigP 〉S τ+1 τ−1
2  ln τ
D expansion in density truncated to three terms is usually suitable:
 = A + BT0 + CT 20 +   (4-58)
R τ 2T 20
Z = 1 + Bρ + Cρ2 (4-68)
Equations for residual enthalpy and entropy may be developed from
each of these expressions. Consider first Eq. (4-67), which is explicit
RESIDUAL ENTHALPY AND ENTROPY in volume. Equations (4-45) and (4-46) are therefore applicable.
FROM PVT CORRELATIONS Direct substitution for Z in Eq. (4-45) gives
The residual properties of gases and vapors depend on their PVT GR BP
behavior. This is often expressed through correlations for the com-  =  (4-69)
pressibility factor Z, defined by Eq. (4-36). Analytical expressions for RT RT
Z as functions of T and P or T and V are known as equations of state. Differentiation of Eq. (4-67) yields
They may also be reformulated to give P as a function of T and V or V
as a function of T and P. ∂Z

∂T  dT
=  −  
dB B P
Virial Equations of State The virial equation in density is an T RT
infinite series expansion of the compressibility factor Z in powers of
molar density ρ (or reciprocal molar volume V−1) about the real gas By Eq. (4-46),
state at zero density (zero pressure): HR

P B dB
Z = 1 + Bρ + Cρ2 + Dρ3 + · · · (4-59)  =   −  (4-70)
The density series virial coefficients B, C, D, . . . depend on tempera- and by Eq. (4-47),
ture and composition only. In practice, truncation is to two or three
terms. The composition dependencies of B and C are given by the SR P dB
 =−   (4-71)
exact mixing rules R R dT
B =

yi yj Bij (4-60) An extensive set of three-parameter corresponding-states correla-

i j
tions has been developed by Pitzer and coworkers [Pitzer, Thermo-
C =

yi yj yk Cijk (4-61) dynamics, 3d ed., App. 3, McGraw-Hill, New York (1995)].
i j k Particularly useful is the one for the second virial coefficient. The
where yi, yj, and yk are mole fractions for a gas mixture and i, j, and k basic equation is
identify species. BPc
The coefficient Bij characterizes a bimolecular interaction between  = B0 + ωB1 (4-72)
molecules i and j, and therefore Bij = Bji. Two kinds of second virial
coefficient arise: Bii and Bjj, wherein the subscripts are the same (i = j), with the acentric factor defined by Eq. (2-17). For pure chemical
and Bij, wherein they are different (i ≠ j ). The first is a virial coefficient species B0 and B1 are functions of reduced temperture only. Substitu-
for a pure species; the second is a mixture property, called a cross coef- tion for B in Eq. (4-67) by this expression gives
ficient. Similarly for the third virial coefficients: Ciii, Cjjj, and Ckkk are P
for the pure species, and Ciij = Ciji = Cjii, . . . are cross coefficients. Z = 1 + (B0 + ωB1)r (4-73)
Although the virial equation itself is easily rationalized on empirical
grounds, the mixing rules of Eqs. (4-60) and (4-61) follow rigorously By differentiation,
∂Z dB0dTr B0 dB1dTr B1
from the methods of statistical mechanics. The temperature deriva-
tives of B and C are given exactly by  = Pr  − 2 + ωPr  − 2
∂Tr P r
Tr Tr Tr Tr
dB dBij

yi yj  (4-62) Upon substitution of these equations into Eqs. (4-48) and (4-49), inte-
dT i j dT gration yields
dC dCijk GR Pr

yi yj yk  (4-63)  = (B0 + ωB1)  (4-74)

dT i j k dT RT Tr
HR dB0 dB1
An alternative form of the virial equation expresses Z as an expan-
sion in powers of pressure about the real gas state at zero pressure  = Pr B0 − Tr  + ω B1 − Tr  (4-75)
RTc dTr dTr
(zero density):
The residual entropy follows from Eq. (4-47):
Z = 1 + B′P + C′P2 + D′P3 + . . . (4-64)
SR dB0 dB1
Equation (4-64) is the virial equation in pressure, and B′, C′, D′, . . .
are the pressure series virial coefficients. Again, truncation is to two
 = − Pr  + ω 
R dTr 
dTr (4-76)

In these equations, B0 and B1 and their derivatives are well repre- B and its temperature derivative. Values of HR and SR are then given
sented by Abbott’s correlations [Smith and Van Ness, Introduction to by Eqs. (4-70) and (4-71).
Chemical Engineering Thermodynamics, 3d ed., p. 87, McGraw-Hill, A primary virtue of Abbott’s correlations for second virial coeffi-
New York (1975)]: cients is simplicity. More complex correlations of somewhat wider
applicability include those by Tsonopoulos [AIChE J. 20: 263–272
0.422 (1974); ibid., 21: 827–829 (1975); ibid., 24: 1112–1115 (1978); Adv. in
B0 = 0.083 −  (4-77) Chemistry Series 182, pp. 143–162 (1979)] and Hayden and O’Con-
nell [Ind. Eng. Chem. Proc. Des. Dev. 14: 209–216 (1975)]. For aque-
0.172 ous systems see Bishop and O’Connell [Ind. Eng. Chem. Res., 44:
B1 = 0.139 −  (4-78) 630–633 (2005)].
Tr4.2 Because Eq. (4-68) is explicit in P, it is incompatible with Eqs. (4-45)
dB0 0.675 and (4-46), and they must be transformed to make V (or molar den-
 =  (4-79) sity ρ) the variable of integration. The resulting equations are given by
dTr T r2.6 Smith, Van Ness, and Abbott [Introduction to Chemical Engineering
dB1 0.722 Thermodynamics, 7th ed., pp. 216–217, McGraw-Hill, New York (2005)]:
 (Z − 1) dρρ
(4-80) GR ρ
dTr T r5.2  = Z − 1 − ln Z + (4-88)
RT 0
Although limited to pressures where the two-term virial equation in
pressure has approximate validity, these correlations are applicable for
most chemical processing conditions. As with all generalized correla-
ρ ∂Z
 = Z − 1 − T 
0 ∂T
  dρρ ρ
tions, they are least accurate for polar and associating molecules.
Although developed for pure materials, these correlations can be By differentiation of Eq. (4-68),
extended to gas or vapor mixtures. Basic to this extension are the mix-

ing rules for the second virial coefficient and its temperature deriva- dB dC
= ρ + ρ 2
tive as given by Eqs. (4-60) and (4-62). Values for the cross coefficients dT
ρ dT
Bij, with i ≠ j, and their derivatives are provided by Eq. (4-72) written
in extended form: Substituting in Eqs. (4-88) and (4-89) for Z by Eq. (4-68) and in Eq.
RTcij 0 (4-89) for the derivative yields, upon integration and reduction,
Bij =  (B + ωij B1) (4-81)
Pcij GR 3
 = 2Bρ +  Cρ2 − ln Z (4-90)
RT 2
where B0, B1, dB0 /dTr, and dB1/dTr are the same functions of Tr as
given by Eqs. (4-77) through (4-80). Differentiation produces HR
dB T dC
 = B − T  ρ + C −   ρ2 (4-91)
RT dT 2 dT
RTcij dB0 dB1

 =   + ωij  The residual entropy is given by Eq. (4-47).
dT Pcij dT dT
In a process calculation, T and P, rather than T and ρ (or T and V),
R dB0 dB1 are usually the favored independent variables. Applications of Eqs.

 =   + ωij  (4-82) (4-90) and (4-91) therefore require prior solution of Eq. (4-68) for Z
dT Pcij dTrij dTrij
or ρ. With Z = P/ρRT, Eq. (4-68) may be written in two equivalent
where Trij = T/Tcij. The following combining rules for ωij, Tcij, and Pcij

are given by Prausnitz, Lichtenthaler, and de Azevedo [Molecular BP
Z 3 − Z 2 −  Z − 2 = 0 (4-92)
Thermodynamics of Fluid-Phase Equilibria, 2d ed., pp. 132 and 162, RT (RT)
Prentice-Hall, Englewood Cliffs, N.J. (1986)]:

B 1 P
ρ3 +  ρ2 +  ρ −  = 0 (4-93)
ωi + ωj C C CRT
ωij =  (4-83)
2 In the event that three real roots obtain for these equations, only the
largest Z (smallest ρ), appropriate for the vapor phase, has physical
Tcij = (TciTcj)12(1 − kij) (4-84) significance, because the virial equations are suitable only for vapors
and gases.
ZcijRTcij Data for third virial coefficients are often lacking, but generalized
Pcij =  (4-85) correlations are available. Equation (4-68) may be rewritten in reduced
form as
Zci + Zcj

Pr Pr 2
with Zcij =  (4-86) Z = 1 + B̂ + Ĉ  (4-94)
2 Tr Z Tr Z

ci + Vcj
V13 13 where B̂ is the reduced second virial coefficient given by Eq. (4-72).

and Vcij =  (4-87) Thus by definition,
In Eq. (4-84), kij is an empirical interaction parameter specific to B̂  = B0 + ωB1 (4-95)
an i − j molecular pair. When i = j and for chemically similar species,
kij = 0. Otherwise, it is a small (usually) positive number evaluated The reduced third virial coefficient Ĉ is defined as
from minimal PVT data or, absence data, set equal to zero.
When i = j, all equations reduce to the appropriate values for a pure CP2c
Ĉ  (4-96)
species. When i ≠ j, these equations define a set of interaction para- R2Tc2
meters without physical significance. For a mixture, values of Bij and
dBij /dT from Eqs. (4-81) and (4-82) are substituted into Eqs. (4-60) A Pitzer-type correlation for Ĉ is then written as
and (4-62) to provide values of the mixture second virial coefficient Ĉ = C0 + ωC1 (4-97)

Correlations for C0 and C1 with reduced temperature are Solution for V is most convenient with the solve routine of a software
package. An initial estimate for V in Eq. (4-103a) is the ideal gas value
0.02432 0.00313 RT/P; for Eq. (4-103b) it is V = b. In either case, iteration is initiated
C0 = 0.01407 +  −  (4-98)
Tr Tr10.5 by substituting the estimate on the right side. The resulting value of V
on the left is returned to the right side, and the process continues until
0.05539 0.00242 the change in V is suitably small.
C1 = − 0.02676 +  −  (4-99) Equations for Z equivalent to Eqs. (4-103) are obtained by substi-
Tr2.7 Tr10.5
tuting V = ZRT/P.
The first is given by, and the second is inspired by, Orbey and Vera
[AIChE J. 29: 107–113 (1983)]. Z−β
Equation (4-94) is cubic in Z; with Tr and Pr specified, solution for Vapor: Z = 1 + β − qβ  (4-104a)
(Z + %β)(Z + σβ)
Z is by iteration. An initial guess of Z = 1 on the right side usually leads
to rapid convergence. 1+β−Z
Another class of equations, known as extended virial equations, was
introduced by Benedict, Webb, and Rubin [J. Chem. Phys. 8: 334–345
Liquid: Z = β + (Z + %b)(Z + σb)  
qβ (4-104b)

(1940); 10: 747–758 (1942)]. This equation contains eight parameters, bP

all functions of composition. It and its modifications, despite their where by definition β  (4-105)
complexity, find application in the petroleum and natural gas indus- RT
tries for light hydrocarbons and a few other commonly encountered
gases [see Lee and Kesler, AIChE J., 21: 510–527 (1975)]. and q  (4-106)
Cubic Equations of State The modern development of cubic bRT
equations of state started in 1949 with publication of the Redlich-
Kwong (RK) equation [Chem. Rev., 44: 233–244 (1949)], and many These dimensionless quantities provide simplification, and when
others have since been proposed. An extensive review is given by combined with Eqs. (4-101) and (4-102), they yield
Valderrama [Ind. Eng. Chem. Res. 42: 1603–1618 (2003)]. Of the
equations published more recently, the two most popular are the Pr
Soave-Redlich-Kwong (SRK) equation, a modification of the RK β=Ω  (4-107)
equation [Chem. Eng. Sci. 27: 1197–1203 (1972)] and the Peng- Tr
Robinson (PR) equation [Ind. Eng. Chem. Fundam. 15: 59–64
(1976)]. All are encompased by a generic cubic equation of state, Ψα(Tr)
written as q=  (4-108)
RT a(T)
P =   −  (4-100)
V − b (V + %b)(V + σb) In Eq. (4-104a) the initial estimate is Z = 1; in Eq. (4-104b) it is Z = β.
Iteration follows the same pattern as for Eqs. (4-103). The final value
For a specific form of this equation, % and σ are pure numbers, the of Z yields the volume root through V = ZRT/P.
same for all substances, whereas parameters a(T) and b are substance- Equations of state, such as the Redlich-Kwong (RK) equation, which
dependent. Suitable estimates of the parameters in cubic equations of expresses Z as a function of Tr and Pr only, yield two-parameter corre-
state are usually found from values for the critical constants Tc and Pc. sponding-states correlations. The SRK equation and the PR equation,
The procedure is discussed by Smith, Van Ness, and Abbott in which the acentric factor ω enters through function α(Tr; ω) as an
[Introduction to Chemical Engineering Thermodynamics, 7th ed., additional parameter, yield three-parameter corresponding-states cor-
pp. 93–94, McGraw-Hill, New York (2005)], and for Eq. (4-100) the relations. The numerical assignments for parameters %, σ, Ω, and Ψ
appropriate equations are given as are given in Table 4-2. Expressions are also given for α(Tr; ω) for the
α(Tr)R2Tc2 SRK and PR equations.
a(T) = ψ  (4-101) As shown by Smith, Van Ness, and Abbott [Introduction to Chemi-
Pc cal Engineering Thermodynamics, 7th ed., pp. 218–219, McGraw-
RT Hill, New York (2005)], Eqs. (4-104) in conjunction with Eqs. (4-88),
b = Ω c (4-102) (4-89), and (4-47) lead to
Function α(Tr) is an empirical expression, specific to a particular form GR
of the equation of state. In these equations ψ and Ω are pure num-  = Z − 1 − ln(Z − β) − qI (4-109)
bers, independent of substance and determined for a particular equa- RT
tion of state from the values assigned to % and σ.
As an equation cubic in V, Eq. (4-100) has three volume roots, of HR d ln α(Tr)
which two may be complex. Physically meaningful values of V are
always real, positive, and greater than parameter b. When T > Tc, solu-
 = Z − 1 +  − 1 qI
d ln Tr  (4-110)

tion for V at any positive value of P yields only one real positive root.
When T = Tc, this is also true, except at the critical pressure, where
three roots exist, all equal to Vc. For T < Tc, only one real positive (liq-
uidlike) root exists at high pressures, but for a range of lower pressures TABLE 4-2 Parameter Assignments for Cubic Equations
there are three. Here, the middle root is of no significance; the small- of State*
est root is a liquid or liquidlike volume, and the largest root is a vapor For use with Eqs. (4-104) through (4-106)
or vaporlike volume.
Eq. of state α(Tr) σ % Ω Ψ
Equation (4-100) may be rearranged to facilitate its solution either
for a vapor or vaporlike volume or for a liquid or liquidlike volume. RK (1949) Tr−1/2 1 0 0.08664 0.42748
SRK (1972) αSRK(Tr; ω)† 1 0 0.08664 0.42748
RT a(T) V−b PR (1976) αPR(Tr; ω)‡ 1 + 2 1 − 2 0.07780 0.45724
Vapor: V =  + b −   (4-103a)
P P (V + %b)(V + σb) *Smith, Van Ness, and Abbott, Introduction to Chemical Engineering Ther-
RT − bP − VP modynamics, 7th ed., p. 98, McGraw-Hill, New York (2005).

V = b + (V + %b)(V + σb) 
 α SRK(Tr ; ω) = [1 + (0.480 + 1.574ω − 0.176ω 2) (1 − Tr1/2)]2

Liquid: (4-103b)
a(T) ‡
α PR(Tr; ω) = [1 + (0.37464 + 1.54226ω − 0.26992ω 2) (1 − Tr1/2)]2

SR d ln α(Tr) HR (HR)0 (HR)1

 = ln (Z − β) +  qI (4-111)  =  + ω (4-118)
R d ln Tr RTc RTc RTc

SR (SR)0 (SR)1
Z + σβ  =  + ω

1 (4-119)
where I =  ln  (4-112) R R R
σ−% Z + %β
Pitzer’s original correlations for Z and the derived quantities were
Preliminary to application of these equations Z is found by solution of determined graphically and presented in tabular form. Since then,
either Eq. (4-104a) or (4-104b). analytical refinements to the tables have been developed, with extended
Cubic equations of state may be applied to mixtures through expres- range and accuracy. The most popular Pitzer-type correlation is that of
sions that give the parameters as functions of composition. No estab- Lee and Kesler [AIChE J. 21: 510–527 (1975); see also Smith, Van
lished theory prescribes the form of this dependence, and empirical Ness, and Abbott, Introduction to Chemical Engineering Thermody-
mixing rules are often used to relate mixture parameters to pure- namics, 5th, 6th, and 7th eds., App. E, McGraw-Hill, New York (1996,
species parameters. The simplest realistic expressions are a linear mix- 2001, 2005)]. These tables cover both the liquid and gas phases and
ing rule for parameter b and a quadratic mixing rule for parameter a span the ranges 0.3 ≤ Tr ≤ 4.0 and 0.01 ≤ Pr ≤ 10.0. They list values of
Z0, Z1, (HR)0/RTc, (HR)1/RTc, (SR)0/R, and (SR)1/R.
b =
xi bi (4-113) Lee and Kesler also included a Pitzer-type correlation for vapor
i pressures:
r (Tr) = ln P r (Tr) + ω ln P r (Tr)
ln P sat 0 1
a =

xi xj aij (4-114)
i j
with aij = aji. The aij are of two types: pure-species parameters (like where ln P0r (Tr) = 5.92714 −  − 1.28862 ln Tr + 0.169347T 6r
subscripts) and interaction parameters (unlike subscripts). Parameter
bi is for pure species i. The interaction parameter aij is often evaluated (4-121)
from pure-species parameters by a geometric mean combining rule 15.6875
aij = (aiaj)1/2 (4-115) and ln P1r (Tr) = 15.2518 −  − 13.4721 ln Tr + 0.43577T 6r
These traditional equations yield mixture parameters solely from (4-122)
parameters for the pure constituent species. They are most likely to be The value of ω to be used with Eq. (4-120) is found from the correla-
satisfactory for mixtures comprised of simple and chemically similar tion by requiring that it reproduce the normal boiling point; that is, ω
molecules. for a particular substance is determined from
Pitzer’s Generalized Correlations In addition to the
corresponding-states coorelation for the second virial coefficient, ln Prsat − ln Pr0(Tr )
ω= n
Pitzer and coworkers [Thermodynamics, 3d ed., App. 3, McGraw- ln Plr(Tr )
Hill, New York (1995)] developed a full set of generalized correla-
tions. They have as their basis an equation for the compressibility where Tr is the reduced normal boiling point and Prsat is the reduced
n n

factor, as given by Eq. (2-63): vapor pressure corresponding to 1 standard atmosphere (1.01325 bar).
Although the tables representing the Pitzer correlations are based
Z = Z0 + ωZ1 (2-63) on data for pure materials, they may also be used for the calculation of
where Z0 and Z1 are each functions of reduced temperature Tr and mixture properties. A set of recipes is required relating the parameters
reduced pressure Pr. Acentric factor ω is defined by Eq. (2-17). Cor- Tc, Pc, and ω for a mixture to the pure-species values and to composi-
relations for Z appear in Sec. 2. tion. One such set is given by Eqs. (2-80) through (2-82) in the Seventh
Generalized correlations are developed here for the residual Edition of Perry’s Chemical Engineers’ Handbook (1997). These equa-
enthalpy and residual entropy from Eqs. (4-48) and (4-49). Substitu- tions define pseudoparameters, so called because the defined values of
tion for Z by Eq. (2-63) puts Eq. (4-48) into generalized form: Tpc, Ppc, and ω have no physical significance for the mixture.
The Lee-Kesler correlations provide reliable data for nonpolar and
Pr dP
(Z0 − 1) r + ω
Pr dP
Z1 r
slightly polar gases; errors of less than 3 percent are likely. Larger errors
can be expected in applications to highly polar and associating gases.
The quantum gases (e.g., hydrogen, helium, and neon) do not con-
Differentiation of Eq. (2-63) yields form to the same corresponding-states behavior as do normal fluids.
Prausnitz, Lichtenthaler, and de Azevedo [Molecular Thermodynam-
∂Z ∂Z0 ∂Z1 ics of Fluid-Phase Equilibria, 3d ed., pp. 172–173, Prentice-Hall PTR,

r Pr r Pr
+ ω 
∂Tr  Pr
Upper Saddle River, N.J. (1999)] propose the use of temperature-
dependent effective critical parameters. For hydrogen, the quantum
Substitution for (∂Z∂Tr)P in Eq. (4-49) gives
gas most commonly found in chemical processing, the recommended
equations are
Pr Pr
HR 1

dPr dPr
 = − T 2r   − ωT 2r  (4-117) T 43.6
RTc 0 ∂Tr Pr Pr 0 Pr Pr c =  (for H2) (4-124)
K 1 + 21.8/2.016T
By Eq. (4-47),  =   − 
R Tr RTc RT  
(for H2) (4-125)
Combination of Eqs. (4-116) and (4-117) leads to bar 1 + 44.2/2.016T

Pr Pr
SR 0 1

∂T  ∂T 
dP dP
 = − r + Z 0 − 1 r − ω r + Z1 r Vc 51.5
R 0 Pr Pr 0 Pr Pr  =  (for H2) (4-126)
r r
1 − 9.91/2.016T
If the first terms on the right sides of Eq. (4-117) and of this equation
(including the minus signs) are represented by (HR)0/RTc and (SR)0/R
and if the second terms, excluding ω but including the minus signs, where T is absolute temperature in kelvins. Use of these effective critical
are represented by (HR)1/RTc and (SR)1/R, then parameters for hydrogen requires the further specification that ω = 0.


LIQUID PHASE treatment of this transition is facilitated by definition of property

changes of vaporization ∆Mlv:
Although residual properties have formal reality for liquids as well as
for gases, their advantageous use as small corrections to ideal gas ∆Mlv Mv − Ml (4-138)
state properties is lost. Calculation of property changes for the liquid l v
state are usually based on alternative forms of Eqs. (4-32) through where M and M are molar properties for states of saturated liquid
(4-35), shown in Table 4-1. Useful here are the definitons of two and saturated vapor. Some experimental values of the enthalpy change
liquid-phase properties—the volume expansivity β and the isother- of vaporization ∆Hlv, usually called the latent heat of vaporization, are
mal compressibility κ: listed in Table 2-150.
The enthalpy change and entropy change of vaporization are
1 ∂V directly related:
V ∂T  P
∆Hlv = T ∆Slv (4-139)
1 ∂V This equation follows from Eq. (4-15), because vaporization at the
κ −  
V ∂P  T
(4-128) vapor pressure Psat occurs at constant T.
As shown by Smith, Van Ness, and Abbott [Introduction to Chemi-
cal Engineering Thermodynamics, 7th ed., p. 221, McGraw-Hill, New
∂V ∂V
For V = f (T, P),
 dT +  
dV = 
∂T P ∂P
York (2005)] the heat of vaporization is directly related to the slope of
the vapor-pressure curve.
This equation in combination with Eqs. (4-127) and (4-128) becomes dPsat
∆Hlv = T ∆V lv  (4-140)
dV dT
 = β dT − κ dP (4-129)
V Known as the Clapeyron equation, this exact thermodynamic relation
provides the connection between the properties of the liquid and
∂P β
If V is constant,

∂T V
(4-130) vapor phases.
In application an empirical vapor pressure versus temperature rela-
Because liquid-phase isotherms of P versus V are very steep and tion is required. The simplest such equation is
closely spaced, both β and κ are small. Moreover (outside the critical B
region), they are weak functions of T and P and are often assumed ln P sat = A −  (4-141)
constant at average values. Integration of Eq. (4-129) then gives T
V where A and B are constants for a given chemical species. This equa-
ln 2 = β(T2 − T1) − κ(P2 − P1) (4-131) tion approximates Psat over its entire temperature range from triple
V1 point to critical point. It is also a sound basis for interpolation between
Substitution for the partial derivatives in Eqs. (4-32) through (4-35) reasonably spaced values of T. More satisfactory for general use is the
by Eqs. (4-127) and (4-130) yields Antoine equation
dH = CP dT + (1 − βT)V dP (4-132) B
ln P sat = A −  (4-142)
dS = CP  − βV dP (4-133) The Wagner equation is useful for accurate representation of vapor
pressure data over a wide temperature range. It expresses the reduced
β vapor pressure as a function of reduced temperature
dU = CV dT +  T − P dV (4-134)
Aτ + Bτ1.5 + Cτ 3 + Dτ 6
r = 
ln P sat (4-143)
CV β
dS =  dT +  dV (4-135)
T κ where τ 1 − Tr
Integration of these equations is most common from the saturated- and A, B, C, and D are constants. Values of the constants for either the
liquid state to the state of compressed liquid at constant T. For exam- Wagner equation or the Antoine equation are given for many species
ple, Eqs. (4-132) and (4-133) in integral form become by Poling, Prausnitz, and O’Connell [The Properties of Gases and Liq-
uids, 5th ed., App. A, McGraw-Hill, New York (2001)].


H = Hsat + (1 − βT)V dP
(4-136) Latent heats of vaporization are functions of temperature, and
P experimental values at a particular temperture are often not available.
Recourse is then made to approximate methods. Trouton’s rule of


S = Ssat − β V dP (4-137) 1884 provides a simple check on whether values calculated by other
sat methods are reasonable:
Again, β and V are weak functions of pressure for liquids, and are ∆H lvn
often assumed constant at the values for the saturated liquid at tem-  ∼ 10
perature T. An alternative treatment of V comes from Eq, (4-131),
which for this application can be written Here, Tn is the absolute temperature of the normal boiling point,
and ∆Hnlv is the latent heat at this temperature. The units of ∆Hnlv, R,
V = V exp[−κ(P − P )]
sat sat
and Tn must be chosen so that ∆Hnlv/RTn is dimensionless.
A much more accurate equation is that of Riedel [Chem. Ing. Tech.

The isothermal vaporization of a pure liquid results in a phase change ∆H lvn 1.092(ln Pc − 1.013)
 =  (4-144)
from saturated liquid to saturated vapor at vapor pressure Psat. The RTn 0.930 − Tr n

where Pc is the critical pressure in bars and Tr is the reduced temper-

Watson’s equation [Ind. Eng. Chem. 35: 398–406 (1943)] has found
ature at Tn. This equation provides reasonable approximations; errors wide acceptance:
rarely exceed 5 percent. ∆H lv2 1 − Tr 0.38
Estimates of the latent heat of vaporization of a pure liquid at any
temperature from the known value at a single temperature may be

∆H lv1
1 − Tr


based on an experimental value or on a value estimated by Eq. (4-144). This equation is simple and fairly accurate.


The thermodynamics of flow encompasses mass, energy, and entropy equation u = ṁ ρA; z is elevation above a datum level, and g is the
balances for open systems, i.e., for systems whose boundaries allow local acceleration of gravity.
the inflow and outflow of fluids. The common measures of flow are as Energy Balances for Steady-State Flow Processes Flow
follows: processes for which the first term of Eq. (4-149) is zero are said to
occur at steady state. As discussed with respect to the mass balance,
Mass flow rate m⋅ molar flow rate n⋅ volumetric flow rate q velocity u
this means that the mass of the system within the control volume is
Also ṁ = Mṅ and q = uA constant; it also means that no changes occur with time in the proper-
ties of the fluid within the control volume or at its entrances and exits.
where M is molar mass. Mass flow rate is related to velocity by
No expansion of the control volume is possible under these circum-
ṁ = uAρ (4-146) stances. The only work of the process is shaft work, and the general
energy balance, Eq. (4-149), becomes
where A is the cross-sectional area of a conduit and ρ is mass density. If
ρ is molar density, then this equation yields molar flow rate. Flow rates ⋅ ⋅
∆ H +  u2 + zg m⋅ = Q + Ws (4-150)
m⋅, n⋅, and q measure quantity per unit of time. Although velocity u does 2 fs
not represent quantity of flow, it is an important design parameter. Entropy Balance for Open Systems An entropy balance differs
from an energy balance in a very important way—entropy is not con-
MASS, ENERGY, AND ENTROPY BALANCES served. According to the second law, the entropy changes in the sys-
FOR OPEN SYSTEMS tem and surroundings as the result of any process must be positive,
with a limiting value of zero for a reversible process. Thus, the entropy
Mass and energy balances for an open system are written with respect changes resulting from the process sum not to zero, but to a positive
to a region of space known as a control volume, bounded by an imagi- quantity called the entropy generation term. The statement of bal-
nary control surface that separates it from the surroundings. This sur- ance, expressed as rates, is therefore
face may follow fixed walls or be arbitrarily placed; it may be rigid or

Time rate of

flexible. Net rate of
Mass Balance for Open Systems Because mass is conserved, change in change of
the time rate of change of mass within the control volume equals the entropy of + entropy
net rate of flow of mass into the control volume. The flow is positive flowing streams in control
when directed into the control volume and negative when directed volume
out. The mass balance is expressed mathematically by

Time rate of

Total rate
dmcv change of
 + ∆(ṁ)fs = 0 (4-147) + = of entropy
dt entropy in generation
The operator ∆ signifies the difference between exit and entrance
flows, and the subscript fs indicates that the term encompasses all
The equivalent equation of entropy balance is
flowing streams. When the mass flow rate m⋅ is given by Eq. (4-146),
d(mS)cv dStsurr ⋅
dmcv ∆(Sm⋅ )fs +  +  = SG ≥ 0 (4-151)
 + ∆(ρuA)fs = 0 (4-148) dt dt
dt ⋅
where SG is the entropy generation term. In accord with the second
This form of the mass balance equation is often called the continuity law, it must be positive, with zero as a limiting value. This equation is
equation. For the special case of steady-state flow, the control volume the general rate form of the entropy balance, applicable at any instant.
contains a constant mass of fluid, and the first term of Eq. (4-148) is zero. The three terms on the left are the net rate of gain in entropy of the
General Energy Balance Because energy, like mass, is con- flowing streams, the time rate of change of the entropy of the fluid
served, the time rate of change of energy within the control volume contained within the control volume, and the time rate of change of
equals the net rate of energy transfer into the control volume. Streams the entropy of the surroundings.
flowing into and out of the control volume have associated with them The entropy change of the surroundings results from heat transfer
energy in its internal, potential, and kinetic forms, and all contribute to between system and surroundings. Let Q⋅ j represent the heat-transfer
the energy change of the system. Energy may also flow across the con- rate at a particular location on the control surface associated with
trol surface as heat and work. Smith, Van Ness, and Abbott [Introduc- a surroundings temperature Tσ, j. In accord with Eq. (4-3), the rate
tion to Chemical Engineering Thermodynamics, 7th ed., pp. 47–48, of ⋅entropy change in the surroundings as a result of this transfer is
⋅ , defined with respect to the sys-
McGraw-Hill, New York (2005)] show that the general energy balance − Qj Tσ,j. The minus sign converts Q j
for flow processes is tem, to a heat rate with respect to the surroundings. The third term in
Eq. (4-151) is therefore the sum of all such quantities, and Eq. (4-151)

d(mU)cv 1 can be written
 + ∆ H +  u2 + zg ṁ = Q̇ + Ẇ (4-149)
dt 2 d(mS)cv Q⋅ j ⋅
∆(Sm⋅ )fs +  −
 = SG ≥ 0
⋅ dt j Tσ, j
The work rate⋅ W may be of several forms. Most commonly there is
shaft work Ws. Work may be associated with expansion or contraction For any process, the two kinds of irreversibility are (1) those inter-
of the control volume, and there may be stirring work. The velocity u nal to the control volume and (2) those resulting from heat transfer
in the kinetic energy term is the bulk mean velocity as defined by the across finite temperature differences that may exist between the

TABLE 4-3 Equations of Balance

Balance equations for single-stream
General equations of balance Balance equations for steady-flow processes steady-flow processes
 + ∆(ṁ)fs = 0 (4-147) ∆(ṁ)fs = 0 (4-153) ṁ1 = ṁ2 = ṁ (4-154)

d(mU)cv 1 1
 + ∆ H +  u2 + zg ṁ = Q̇ + Ẇ (4-149) ∆ H +  u2 + zg m⋅ = Q̇ + Ẇs (4-150) ∆H +  + g ∆z = Q + Ws (4-155)
dt 2 fs 2 fs 2

d(mS)cv Q̇j Q̇j Qj

 + ∆(Sṁ)fs −
 = ṠG ≥ 0 (4-152) ∆(Sṁ)fs −
 = ṠG ≥ 0 (4-156) ∆S −
 = SG ≥ 0 (4-157)
dt j Tσ , j j Tσ , j j Tσ, j

system and surroundings. In the limiting case where SG = 0, the the second law, the irreversibilities of fluid friction in adiabatic flow
process is completely reversible, implying that cause an entropy increase in the fluid in the direction of flow. In the
• The process is internally reversible within the control volume. limit as the flow approaches reversibility, this increase approaches
• Heat transfer between the control volume and its surroundings is zero. In general, then, dS/dx ≥ 0.
reversible. Pipe Flow For a pipe of constant cross-sectional area, dA/dx = 0,
Summary of Equations of Balance for Open Systems Only and Eqs. (4-160) and (4-161) reduce to
the most general equations of mass, energy, and entropy balance
appear in the preceding sections. In each case important applications T 1 + βu2/CP ds βu2/CP + M2 ds
dp du
require less general versions. The most common restricted case is for  = −    u = T   
dx V 1 − M2 dx dx 1− M2 dx
steady flow processes, wherein the mass and thermodynamic proper-
ties of the fluid within the control volume are not time-dependent. A
further simplification results when there is but one entrance and one When flow is subsonic, M2 < 1; all terms on the right in these equa-
exit to the control volume. In this event, m⋅ is the same for both tions are then positive, and dP/dx < 0 and du/dx > 0. Pressure there-
streams, and the equations may be divided through by this rate to put fore decreases and velocity increases in the direction of flow. The
them on the basis of a unit amount of fluid flowing through the con- velocity increase is, however, limited, because these inequalities
trol volume. Summarized in Table 4-3 are the basic equations of bal- would reverse if the velocity were to become supersonic. This is not
ance and their important restricted forms. possible in a pipe of constant cross-sectional area, and the maximum
fluid velocity obtainable is the speed of sound, reached at the exit of
the pipe. Here, dS/dx reaches its limiting value of zero. For a dis-
charge pressure low enough, the flow becomes sonic and lengthening
Duct Flow of Compressible Fluids Thermodynamics provides the pipe does not alter this result; the mass rate of flow decreases so
equations interrelating pressure changes, velocity, duct cross-sectional that the sonic velocity is still obtained at the outlet of the lengthened
area, enthalpy, entropy, and specific volume within a flowing stream. pipe.
Considered here is the adiabatic, steady-state, one-dimensional flow According to the equations for supersonic pipe flow, pressure
of a compressible fluid in the absence of shaft work and changes in increases and velocity decreases in the direction of flow. However, this
potential energy. The appropriate energy balance is Eq. (4-155). With flow regime is unstable, and a supersonic stream entering a pipe of
Q, Ws, and ∆z all set equal to zero, constant cross section undergoes a compression shock, the result of
which is an abrupt and finite increase in pressure and decrease in
∆u2 velocity to a subsonic value.
∆H +  = 0
2 Nozzles Nozzle flow is quite different from pipe flow. In a prop-
erly designed nozzle, its cross-sectional area changes with length in
In differential form, dH = − u du (4-158) such a way as to make the flow nearly frictionless. The limit is
reversible flow, for which the rate of entropy increase is zero. In this
The continuity equation given by Eq. (4-148) here becomes d(ρuA) = event dS/dx = 0, and Eqs. (4-160) and (4-161) become
d(uA/V) = 0, whence
dP u 2
1 dA du u 1 dA
dV du dA  =  2   = −  2 
 −  −  =0 (4-159) dx VA 1 − M dx dx A 1 − M dx
V u A
The characteristics of flow depend on whether the flow is subsonic
Smith, Abbott, and Van Ness [Introduction to Chemical Engineer- (M < 1) or supersonic (M > 1). The possibilities are summarized in
ing Thermodynamics, 7th ed., pp. 255–258, McGraw-Hill, New York Table 4-4. Thus, for subsonic flow in a converging nozzle, the velocity
(2005)] show that these basic equations in combination with Eq. (4-15) increases and the pressure decreases as the cross-sectional area
and other property relations yield two very general equations

βu2 dS u2 dA

V(1 − M2)  + T 1 +   −   = 0 (4-160) TABLE 4-4 Nozzle Characteristics
dx CP dx A dx
Subsonic: M < 1 Supersonic: M > 1
βu2/ CP + M2 dS u2 dA Converging Diverging Converging Diverging
du 1
u − T    + 2   = 0 (4-161)
dx 1 − M2 dx 1−M A dx dA
 − + − +
Mach number M is the ratio of the speed of fluid in the duct to the dP
speed of sound in the fluid. The derivatives in these equations are  − + + −
rates of change with length as the fluid passes through a duct. Equa- du
tion (4-160) relates the pressure derivative, and Eq. (4-161), the  + − − +
velocity derivative, to the entropy and area derivatives. According to dx

diminishes. The maximum possible fluid velocity is the speed of

sound, reached at the exit. A converging subsonic nozzle can therefore
deliver a constant flow rate into a region of variable pressure.
Supersonic velocities characterize the diverging section of a prop-
erly designed converging/diverging nozzle. Sonic velocity is reached at 1
the throat, where dA/dx = 0, and a further increase in velocity and
decrease in pressure require a diverging cross-sectional area to
accommodate the increasing volume of flow. The pressure at the H
throat must be low enough for the velocity to become sonic. If this is P1 (H)S
not the case, the diverging section acts as a diffuser—the pressure H
rises and the velocity decreases in the conventional behavior of sub- 2
sonic flow in a diverging section.
An analytical expression relating velocity to pressure in an isen- 2
tropic nozzle is readily derived for an ideal gas with constant heat
capacities. Combination of Eqs. (4-15) and (4-159) for isentropic flow P2 S
u du = − V dP
Integration, with nozzle entrance and exit conditions denoted by 1 S
and 2, yields
FIG. 4-2 Adiabatic expansion process in a turbine or expander. [Smith, Van
 V dP = 
P2 (γ −1)/γ

1 −  
2γP V P Ness, and Abbott, Introduction to Chemical Engineering Thermodynamics, 7th
u22 − u21 = − 2 1 1 2
(4-162) ed., p. 269, McGraw-Hill, New York (2005).]
P1 P 1

where the final term is obtained upon elimination of V by PV γ = const,

an equation valid for ideal gases with constant heat capacities. Here, discharge pressure. Because the actual expansion process is irre-
γ CP/CV. versible, turbine efficiency is defined as
Throttling Process Fluid flowing through a restriction, such as
an orifice, without appreciable change in kinetic or potential energy η 
undergoes a finite pressure drop. This throttling process produces Ws(isentropic)
no shaft work, and in the absence of heat transfer, Eq. (4-155)
reduces to ∆H = 0 or H2 = H1. The process therefore occurs at con- where Ws is the actual shaft work. By Eqs. (4-163) and (4-165),
stant enthalpy. ∆H
The temperature of an ideal gas is not changed by a throttling η=  (4-166)
process, because its enthalpy depends on temperature only. For most (∆H)S
real gases at moderate conditions of T and P, a reduction in pressure Values of η usually range from 0.7 to 0.8.
at constant enthalpy results in a decrease in temperature, although the The HS diagram of Fig. 4-2 compares the path of an actual expan-
effect is usually small. Throttling of a wet vapor to a sufficiently low sion in a turbine with that of an isentropic expansion for the same
pressure causes the liquid to evaporate and the vapor to become intake conditions and the same discharge pressure. The isentropic
superheated. This results in a considerable temperature drop because path is the dashed vertical line from point 1 at intake pressure P1 to
of the evaporation of liquid. point 2′ at P2. The irreversible path (solid line) starts at point 1 and ter-
If a saturated liquid is throttled to a lower pressure, some of the liq- minates at point 2 on the isobar for P2. The process is adiabatic, and
uid vaporizes or flashes, producing a mixture of saturated liquid and irreversibilities cause the path to be directed toward increasing
saturated vapor at the lower pressure. Again, the large temperature entropy. The greater the irreversiblity, the farther point 2 lies to the
drop results from evaporation of liquid. right on the P2 isobar, and the lower the value of η.
Turbines (Expanders) High-velocity streams from nozzles imping- Compression Processes Compressors, pumps, fans, blowers,
ing on blades attached to a rotating shaft form a turbine (or expander) and vacuum pumps are all devices designed to bring about pressure
through which vapor or gas flows in a steady-state expansion process increases. Their energy requirements for steady-state operation are of
which converts internal energy of a high-pressure stream into shaft interest here. Compression of gases may be accomplished in rotating
work. The motive force may be provided by steam (turbine) or by a equipment (high-volume flow) or for high pressures in cylinders with
high-pressure gas (expander). reciprocating pistons. The energy equations are the same; indeed,
In any properly designed turbine, heat transfer and changes in based on the same assumptions, they are the same as for turbines or
potential and kinetic eneregy are negligible. Equation (4-155) there- expanders. Thus, Eqs. (4-159) through (4-161) apply to adiabatic com-
fore reduces to pression.
Ws = ∆H = H2 − H1 (4-163) The isentropic work of compression, as given by Eq. (4-165), is the
minimum shaft work required for compression of a gas from a given
The rate form of this equation is initial state to a given discharge pressure. A compressor efficiency is
Ẇs = ṁ ∆H = ṁ(H2 − H1) (4-164) defined as

When inlet conditions T1 and P1 and discharge pressure P2 are known, Ws(isentropic)
the value of H1 is fixed. In Eq. (4-163) both H2 and Ws are unknown, Ws
and the energy balance alone does not allow their calculation. How-
ever, if the fluid expands reversibly and adiabatically, i.e., isentropi- In view of Eqs. (4-163) and (4-165), this becomes
cally, in the turbine, then S2 = S1. This second equation establishes the (∆H)S
final state of the fluid and allows calculation of H2. Equation (4-164) η  (4-167)
then gives the isentropic work: ∆H

Ws(isentropic) = (∆H)S (4-165) Compressor efficiencies are usually in the range of 0.7 to 0.8.
The compression process is shown on an HS diagram in Fig. 4-3.
The absolute value |Ws |(isentropic) is the maximum work that can be The vertical dashed line rising from point 1 to point 2′ represents the
produced by an adiabatic turbine with given inlet conditions and given reversible adiabatic (isentropic) compression process from P1 to P2.

Example 1: LNG Vaporization and Compression A port facility

for unloading liquefied natural gas (LNG) is under consideration. The LNG
2 arrives by ship, stored as saturated liquid at 115 K and 1.325 bar, and is unloaded
at the rate of 450 kg s-1. It is proposed to vaporize the LNG with heat discarded
from a heat engine operating between 300 K, the temperature of atmospheric
2 air, and 115 K, the temperature of the vaporizing LNG. The saturated-vapor
LNG so produced is compressed adiabatically to 20 bar, using the work pro-
duced by the heat engine to supply part of the compression work. Estimate the
work to be supplied from an external source.
For estimation purposes we need not be concerned with the design of the
P2 (H)S
H heat engine, but assume that a suitable engine can be built to deliver 30 percent
of the work of a Carnot engine operating between the temperatures of 300 and
115 K. The equations that apply to Carnot engines can be found in any thermo-
dynamics text.
1 By the first law: W = QH − QC
P1 S By the second law:  = 

In combination: W = QC  −1
Here, |W| is the work produced by the Carnot engine; |QC| is the heat trans-
FIG. 4-3 Adiabatic compression process. [Smith, Van Ness, and Abbott, Intro- ferred at the cold temperature, i.e., to vaporize the LNG; TH and TC are the hot
duction to Chemical Engineering Thermodynamics, 7th ed., p. 274, McGraw-Hill, and cold temperatures of the heat reservoirs between which the heat engine
New York (2005).] operates, or 300 and 115 K, respectively. LNG is essentially pure methane, and
enthalpy values from Table 2-281 of the Seventh Edition of Perry’s Chemical
Engineers’ Handbook provide its heat of vaporization:
The actual irreversible compression process follows the solid line ∆Hlv = Hv − Hl = 802.5 − 297.7 = 504.8 kJ kg−1
from point 1 upward and to the right in the direction of increasing For a flow rate of 450 kg s−1,
entropy, terminating at point 2. The more irreversible the process, the
QC = (450)(504.8) = 227,160 kJ s−1
farther this point lies to the right on the P2 isobar, and the lower the
efficiency η of the process. The equation for work gives
Liquids are moved by pumps, usually by rotating equipment. The

same equations apply to adiabatic pumps as to adiabatic compressors. W = (227,160)  − 1 = 3.654 × 105 kJ s−1 = 3.654 × 105 kW
Thus, Eqs. (4-163) through (4-165) and Eq. (4-167) are valid. How- This is the reversible work of a Carnot engine. The assumption is that the actual
ever, application of Eq. (4-163) requires values of the enthalpy of power produced is 30 percent of this, or 1.096 × 105 kW.
compressed (subcooled) liquids, and these are seldom available. The The enthalpy and entropy of saturated vapor at 115 K and 1.325 bar are given
fundamental property relation, Eq. (4-15), provides an alternative. in Table 2-281 of the Seventh Edition of Perry’s as
For an isentropic process, Hv = 802.5 kJ kg−1 and Sv = 9.436 kJ kg−1 K−1
dH = V dP (constant S) Isentropic compression of saturated vapor at 1.325 to 20 bar produces super-
heated vapor with an entropy of 9.436 kJ kg−1 K−1. Interpolation in Table 2-282
Combining this with Eq. (4-165) yields at 20 bar yields an enthalpy of H = 1026.2 kJ kg−1 at 234.65 K. The enthalpy
 V dP
change of isentropic compression is then
Ws(isentropic) = (∆H)S =
P1 ∆HS = 1026.2 − 802.5 = 223.7 kJ kg−1
The usual assumption for liquids (at conditions well removed from the For a compressor efficiency of 75 percent, the actual enthalpy change of com-
critical point) is that V is independent of P. Integration then gives pression is
Ws(isentropic) = (∆H)S = V(P2 − P1) (4-168) ∆H 223.7
∆H = S =  = 298.3 kJ kg−1
η 0.75
Also useful are Eqs. (4-132) and (4-133). Because temperature
changes in the pumped fluid are very small and because the properties The actual enthalpy of superheated LNG at 20 bar is then
of liquids are insensitive to pressure (again at conditions not close to H = 802.5 + 298.3 = 1100.8 kJ kg−1
the critical point), these equations are usually integrated on the Interpolation in Table 2-282 of the Seventh Edition of Perry’s indicates an actual
assumption that CP, V, and β are constant, usually at initial values. temperature of 265.9 K for the compressed LNG, which is quite suitable for its
Thus, to a good approximation entry into the distribution system.
The work of compression is
∆H = CP ∆T + V(1 − βT) ∆P (4-169)
W = m ∆H = (450 kg s−1)(298.3 kJ kg−1) = 1.342 × 105 kJ s−1 = 1.342 × 105 kW
T The estimated power which must be supplied from an external source is
∆S = CP ln 2 − βV∆P (4-170)
T1 Ẇ = 1.342 × 105 − 1.096 × 105 = 24,600 kW


The composition of a system may vary because the system is open or extensive thermodynamic property, say, U, H, S, A, or G. A total-system
because of chemical reactions even in a closed system. The equations property is then nM, where n = Σ i ni and i is the index identifying chem-
developed here apply regardless of the cause of composition changes. ical species. One might expect the solution property M to be related
solely to the properties Mi of the pure chemical species which comprise
PARTIAL MOLAR PROPERTIES the solution. However, no such generally valid relation is known, and the
connection must be established experimentally for every specific system.
For a homogeneous PVT system comprised of any number of chemical Although the chemical species which make up a solution do not have
species, let symbol M represent the molar (or unit-mass) value of an their own individual properties, a solution property may be arbitrarily

apportioned among the individual species. Once an apportioning solution properties in the parent equation are related linearly (in the
recipe is adopted, the assigned property values are quite logically algebraic sense).
treated as though they were indeed properties of the individual species Gibbs-Duhem Equation Differentiation of Eq. (4-174) yields
in solution, and reasoning on this basis leads to valid conclusions. ⎯ ⎯
For a homogeneous PVT system, postulate 5 requires that dM =
xi dMi +
Mi dxi
i i
nM = M(T, P, n1, n2, n3, …)
Because this equation and Eq. (4-173) are both valid in general, their
The total differential of nM is therefore right sides can be equated, yielding
∂(nM) ∂(nM) ∂(nM) ∂M ∂M ⎯

d(nM) = 
∂T P,n

dT + 
dP +

i ∂ni  

∂T P,x
dT + 
∂P T,x
dP −
xi dMi = 0

where subscript n indicates that all mole numbers ni are held constant, This general result, the Gibbs-Duhem equation, imposes a constraint
and subscript nj signifies that all mole numbers are held constant on how the partial properties of any phase may vary with temperature,
except the ith. This equation may also be written pressure, and composition. For the special case where T and P are
∂M ∂M ∂(nM)

d(nM) = n 
∂T P,x
dT + n 
∂P T,x
dP +

i ∂ni  

xi dMi = 0 (constant T, P) (4-176)

where subscript x indicates that all mole fractions are held constant. Symbol M may represent the molar value of any extensive thermo-
The derivatives in the summation are⎯ called partial molar properties. dynamic property, say, V, U, H, S, or G. When M H, the derivatives
They are given the generic symbol Mi and are defined by (∂H/∂H)P and (∂H/∂P)T are given by Eqs. (4-28) and (4-29), and Eqs.
(4-173), (4-174), and (4-175) specialize to
⎯ ∂(nM)
Mi    (4-171) ∂V ⎯
∂ni T,P,nj dH = CP dT + V − T 
∂T   dP +
H dx
P,x i
i i (4-177)
The basis for calculation of partial properties from solution properties
is provided by this equation. Moreover, ⎯
H =
xiHi (4-178)
∂M ∂M ⎯
d(nM) = n  dT + n  dP +
Mi dni (4-172)
∂T ∂P ∂V ⎯
   dP −
x dH = 0
P,x T,x i
CP dT + V − T  i i (4-179)
This equation, valid for any equilibrium phase, either closed or ∂T P,x i
open, attributes changes in total property nM to changes in T and P
and to mole-number changes resulting from mass transfer or chem- Similar equations are readily derived when M takes on other identities.
ical reaction. Equation (4-171), which defines a partial molar property, provides
The following are mathematical identities: a general means by which partial-property values may be determined.
However, for a binary solution an alternative method is useful. Equa-
d(nM) = n dM + M dn dni = d(xi n) = xi dn + n dxi tion (4-174) for a binary solution is
⎯ ⎯
Combining these expressions with Eq. (4-172) and collecting like M = x1M1 + x2M2
terms give Moreover, the Gibbs-Duhem equation for a solution at given T and P,
∂M ∂M ⎯ ⎯ Eq. (4-176), becomes
dM − 
∂T P,x

dT − 
∂P T,x i
dP −
Mi dxi n + M −
Mi xi dn = 0
 ⎯ ⎯
x1 dM1 + x2 dM2 = 0
These two equations combine to yield
Because n and dn are independent and arbitrary, the terms in brack-
ets must separately be zero, whence ⎯ dM
M1 = M + x2  (4-180a)
∂M ∂M ⎯ dx1
dM = 
∂T P,x
dT + 
∂P  T,x
dP +
Mi dxi
⎯ dM
M2 = M − x1  (4-180b)
and dx1

M =
xi Mi (4-174) Thus for a binary solution, the partial properties are given directly as
i functions of composition for given T and P. For multicomponent solu-
The first of these equations is merely a special case of Eq. (4-172); tions such calculations are complex, and direct use of Eq. (4-171) is
however, Eq. (4-174) is a vital new relation. Known as the summabil- appropriate.
ity equation, it provides for the calculation of solution properties from Partial Molar Equation-of-State Parameters The parameters
partial properties, a purpose opposite to that of Eq. (4-171). Thus a in equations of state as applied to mixtures are related to composition
solution property apportioned according to the recipe of Eq. (4-171) by mixing rules. For the second virial coefficient
may be recovered simply by adding the properties attributed to the
B =

yiyjBij (4-60)
individual species, each weighted by its mole fraction in solution. The i j
equations for partial molar properties are valid also for partial specific
properties, in which case m replaces n and {xi} are mass fractions. The partial molar second virial coefficient is by definition
Equation (4-171) applied to the definitions of Eqs. (4-8) through (4-10) ⎯ ∂(nB)
yields the partial-property relations
⎯ ⎯ ⎯ ⎯ ⎯ ⎯ ⎯ ⎯ ⎯
∂ni  T,nj

Hi = Ui + PVi Ai = Ui − TSi Gi = Hi − TSi

Because B is independent of P, this is in accord with Eq. (4-171).
These equations illustrate the parallelism that exists between the These
⎯ two equations lead through derivation to useful expressions
equations for a constant-composition solution and those for the cor- for Bi, as shown in detail by Van Ness and Abbott [Classical Thermo-
responding partial properties. This parallelism exists whenever the dynamics of Nonelectrolyte Solutions: With Applications to Phase

Equilibria, pp. 137–140, McGraw-Hill, New York (1982)]. The sim- limit of zero pressure, and provides a conceptual basis upon which to
plest result is build the structure of solution thermodynamics. Smith, Van Ness, and
⎯ Abbott [Introduction to Chemical Engineering Thermodynamics, 7th
Bi = 2
ykBki − B (4-182) ed., pp. 391–394, McGraw-Hill, New York (2005)] develop the fol-
lowing property relations for the ideal gas model.
An analogous expression follows from Eq. (4-114) for parameter a
in the generic cubic equation of state given by Eqs. (4-100), (4-103), ⎯ ig RT
and (4-104): V i = V igi =  (4-192)
a⎯ = 2 y a − a
k k ki (4-183) Because the enthalpy is independent of pressure,
⎯ ig
This expression is independent of the combining rule [e.g., Eq. (4-114)] H i = H igi (4-193)
used for aki. For the linear
⎯ mixing rule of Eq. (4-113) for b, the result
of derivation is simply bi = bi. ig
where S is evaluated at the mixture T and P. The entropy of an ideal
Partial Molar Gibbs Energy Implicit in Eq. (4-13) is the gas does depend on pressure, and here
relation ⎯ig
Si = Sigi − R ln yi (4-194)
µi = ∂ni T,P,n  j
where Sigi is evaluated at the mixture T and P. ⎯ ⎯ ⎯
From the definition of the Gibbs energy, Gigi = H igi − TSiig. In combi-
Comparison with Eq. (4-171) indicates the following identity: nation with Eqs. (4-193) and (4-194), this becomes

µi = Gi (4-184) ⎯
Gigi = H igi − TSigi + RT ln yi
The reciprocity relation for an exact differential applied to Eq. (4-13) ⎯
produces not only the Maxwell relation, Eq. (4-21), but also two other or µigi Gigi = Gigi + RT ln yi (4-195)
useful equations:
∂(nV) Elimination of Gigi from this equation is accomplished through Eq.
∂µi ⎯

∂P T,n
∂ni T,P,n 
= Vi
(4-185) (4-17), written for pure species i as an ideal gas:
∂µi ∂(nS) ⎯ dGigi = V igi dP =  dP = RT d ln P (constant T)

∂T P,n
= −  
= − Si (4-186) P
Integration gives G igi = Γi(T) + RT ln P (4-196)
Because µ = f (T, P),
where integration constant Γi(T) is a function of temperature only.
⎯ ∂µ ∂µ
Equation (4-195) now becomes
dµi dGi = i dT + i dP
∂T P,n ∂P T,n ⎯
⎯ ⎯ ⎯ µ igi = Gigi = Γi(T) + RT ln (yiP) (4-197)
and dGi = − Si dT + Vi dP (4-187)
By Eq. (4-172) Gid =
yiΓi(T) + RT
ln (yiP) (4-198)
Similarly, in view of Eqs. (4-14), (4-15), and (4-16), i i

⎯ ⎯ ⎯ A dimensional ambiguity is apparent with Eqs. (4-196) through (4-198)

dUi = T dSi − P dVi (4-188) in that P has units, whereas ln P must be dimensionless. In practice
⎯ ⎯ ⎯ this is of no consequence, because only differences in Gibbs energy
dHi = T dSi + Vi dP (4-189) appear, along with ratios of the quantities with units of pressure in the
⎯ ⎯ ⎯ arguments of the logarithm. Consistency in the units of pressure is, of
dAi = − Si dT − P dVi (4-190)
course, required.
These equations again illustrate the fact that for every equation pro- Fugacity and Fugacity Coefficient The chemical potential µi
viding a linear relation among the thermodynamic properties of a plays a vital role in both phase and chemical reaction equilibria. How-
constant-composition solution there exists a parallel relationship for ever, the chemical potential exhibits certain unfortunate characteris-
the partial properties of the species in solution. tics that discourage its use in the solution of practical problems. The
The following equation is a mathematical identity: Gibbs energy, and hence µi, is defined in relation to the internal
energy and entropy, both primitive quantities for which absolute val-
ues are unknown. Moreover, µi approaches negative infinity when

nG 1 nG
d   d(nG) − 2 dT either P or yi approaches zero. While these characteristics do not pre-
⎯ clude the use of chemical potentials, the application of equilibrium
Substitution for d(nG) by Eq. (4-13), with µi = Gi , and for G by criteria is facilitated by introduction of the fugacity, a quantity that
Eq. (4-10) gives, after algebraic reduction, takes the place of µi but that does not exhibit its less desirable charac-

Gi The origin of the fugacity concept resides in Eq. (4-196), an equa-

nG nV nH
d  =  dP − 2 dT +
 dni (4-191) tion valid only for pure species i in the ideal gas state. For a real fluid,
RT RT RT i RT an analogous equation is written as
This result is a useful alternative to the fundamental property relation Gi Γi(T) + RT ln fi (4-199)
given by Eq. (4-13). All terms in this equation have units of moles;
moreover, the enthalpy rather than the entropy appears on the right in which a new property fi replaces the pressure P. This equation
side. serves as a partial definition of the fugacity fi.
Subtraction of Eq. (4-196) from Eq. (4-199), both written for the
same temperature and pressure, gives
Ideal Gas Mixture Model The ideal gas mixture model is useful fi
Gi − Gigi = RT ln 
because it is molecularly based, is analytically simple, is realistic in the P

According to the definition of Eq. (4-40), Gi − Gigi is the residual Ideal Solution Model The ideal gas model is useful as a stan-
Gibbs energy GRi. The dimensionless ratio fi /P is another new property dard of comparison for real gas behavior. This is formalized through
called the fugacity coefficient φi. Thus, residual properties. The ideal solution is similarly useful as a standard
to which real solution behavior may be compared.
GRi = RT ln φi (4-200) The partial molar Gibbs energy or chemical potential of species i in
fi an ideal gas mixture is given by Eq. (4-195), written as
where φi  (4-201) ⎯
P µigi = Gigi = Gigi (T, P) + RT ln yi
The definition of fugacity is completed by setting the ideal gas state This equation takes on new meaning when Gigi (T, P) is replaced by Gi
fugacity of pure species i equal to its pressure, f iig = P. Thus for the (T, P), the Gibbs energy of pure species i in its real physical state of
special case of an ideal gas, GRi = 0, φi = 1, and Eq. (4-196) is recovered gas, liquid, or solid at the mixture T and P. The ideal solution is there-
from Eq. (4-199). fore defined as one for which
The definition of the fugacity of a species in solution is parallel to ⎯
µidi = Gidi Gi(T, P) + RT ln xi (4-209)
the definition of the pure-species fugacity. An equation analogous to
the ideal gas expression, Eq. (4-197), is written for species i in a fluid where superscript id denotes an ideal solution property and xi repre-
mixture sents the mole fraction because application is usually to liquids.
This equation is the basis for development of expressions for all other
µi Γi(T) + RT ln fˆi (4-202) thermodynamic properties of an ideal solution. Equations (4-185) ⎯
where the partial pressure yi P is replaced by fˆi, the fugacity of species and (4-186), applied to an ideal solution with µi replaced by Gi, are
i in solution. Because it is not a partial property, it is identified by a cir- written as
cumflex rather than an overbar. ⎯ ⎯
⎯ ∂Gidi ⎯ ∂Gidi
Subtracting Eq. (4-197) from Eq. (4-202), both written for the same
temperature, pressure, and composition, yields
Vidi =  ∂P T, x and Sidi = − 
∂T P,x 
fˆi Differentiation of Eq. (4-209) yields
µi − µigi = RT ln  ⎯ ⎯
∂Gidi ∂Gi ∂Gidi ∂Gi
=  and + R ln x i
The residual Gibbs energy of a mixture is defined by GR G − Gig, ∂P T,x ∂P T ∂T P,x P
⎯R the ⎯ analogous
⎯ definition of a partial molar residual Gibbs energy is Equation (4-17) implies that
Gi Gi − Gigi = µi − µigi . Therefore
⎯R ∂Gi ∂Gi
Gi = RT ln φ̂i (4-203)

∂P T
= Vi and 
= −S

where by definition φ̂i  (4-204) In combination these sets of equations provide
yiP ⎯ id
Vi = Vi (4-210)
The dimensionless ratio φ̂i is called the fugacity coefficient of species i
in solution. and Si = Si − R ln xi (4-211)
⎯ analog of Eq. (4-200), which relates φi to
Equation (4-203) is the
⎯id ⎯id ⎯id
GRi. For an ideal gas, GRi is necessarily zero; therefore φ̂iig = 1 and Because Hi = Gi + TSi , substitutions by Eqs. (4-209) and (4-211)
ˆfiig = yiP. Thus the fugacity of species i in an ideal gas mixture is equal yield
to its partial pressure.
Evaluation of Fugacity Coefficients Combining Eq. (4-200) Hi = Hi (4-212)
with Eq. (4-45) gives
The summability relation, Eq. (4-174), written for the special case
 (Z − 1) 
GR dP of an ideal solution, may be applied to Eqs. (4-209) through (4-212):
ln φ =  = (4-205)
RT 0 P
Gid =
xiGi + RT
xi ln xi (4-213)
Subscript i is omitted, with the understanding that φ here is for a i i
pure species. Clearly, all correlations for GR/RT are also correlations
for ln φ. V id =
xiVi (4-214)
Equation (4-200) with Eqs. (4-48) and (4-73) yields
Sid =
xiSi − R
xi ln xi (4-215)
 (Z − 1) 
dP P i i
ln φ = = (B + ωB ) 
r 0 1 r
0 P r T r Hid =
xiHi (4-216)
This equation, used in conjunction with Eqs. (4-77) and (4-78), pro-
vides a useful generalized correlation for the fugacity coefficients of A simple equation for the fugacity of a species in an ideal solution
pure species. follows from Eq. (4-209). For the special case of species i in an ideal
A more comprehensive generalized correlation results from Eqs. solution, Eq. (4-202) becomes
(4-200) and (4-116): ⎯
µidi = Gidi = Γi(T) + RT ln fˆ idi
 (Z − 1)  +ω Z 
Pr Pr
dP dP When this equation and Eq. (4-199) are combined with Eq. (4-209),
ln φ = 0 r 1 r

0 P P r 0 r Γi (T) is eliminated, and the resulting expression reduces to

An alternative form is ln φ = ln φ0 + ω ln φ1 (4-207) fˆiid = xi fi (4-217)
This equation, known as the Lewis-Randall rule, shows that the fugac-
 (Z Z
Pr Pr
dP dP ity of each species in an ideal solution is proportional to its mole frac-
where ln φ0 − 1) r ln φ1
0 1 r
0 Pr 0 P r
tion; the proportionality constant is the fugacity of pure species i in the
same physical state as the solution and at the same T and P. Division of
By Eq. (4-207), φ = (φ )(φ )
0 1 ω
(4-208) both sides of Eq. (4-217) by xi P and substitution of φ̂iid for fˆidi xiP [Eq.
(4-204)] and of φi for fi/P [Eq. (4-201)] give the alternative form
Correlations may therefore be presented for φ0 and φ1, as was done by
Lee and Kesler [AIChE J. 21: 510–527 (1975)]. φ̂iid = φi (4-218)

Thus the fugacity coefficient of species i in an ideal solution equals the ∆S, and ∆H are the Gibbs energy change of mixing, the volume
fugacity coefficient of pure species i in the same physical state as the change of mixing, the entropy change of mixing, and the enthalpy
solution and at the same T and P. change of mixing. For an ideal solution, each excess property is zero,
Ideal solution behavior is often approximated by solutions com- and for this special case the equations reduce to those shown in the
prised of molecules not too different in size and of the same chemical third column of Table 4-5.
nature. Thus, a mixture of isomers conforms very closely to ideal solu- Property changes of mixing and excess properties are easily calcu-
tion behavior. So do mixtures of adjacent members of a homologous lated one from the other. The most common property changes of mix-
series. ing are the volume change of mixing ∆V and the enthalpy change of
Excess Properties An excess property ME is defined as the dif- mixing ∆H, commonly called the heat of mixing. These properties are
ference between the actual property value of a solution and the value identical to the corresponding excess properties. Moreover, they are
it would have as an ideal solution at the same T, P, and composition. directly measurable, providing an experimental entry into the network
Thus, of equations of solution thermodynamics.
ME M − Mid (4-219)
where M represents the molar (or unit-mass) value of any extensive ON THE GIBBS ENERGY
thermodynamic property (say, V, U, H, S, G). This definition is analo-
gous to the definition of a residual property as given by Eq. (4-40). Of the four fundamental property relations shown in the second col-
However, excess properties have no meaning for pure species, umn of Table 4-1, only Eq. (4-13) has as its special or canonical vari-
whereas residual properties exist for pure species as well as for mix- ables T, P, and {ni}. It is therefore the basis for extension to several
⎯E are defined analogously:
tures. Partial molar excess properties M i useful supplementary thermodynamic properties. Indeed an alterna-
⎯E ⎯ ⎯id tive form has been developed as Eq. (4-191). These equations are the
Mi = Mi − Mi (4-220) first two entries in the upper left quadrant of Table 4-6, which is now
Of particular interest is the partial molar excess Gibbs energy. to be filled out with important derived relationships.
Equation (4-202) may be written as Fundamental Residual-Property Relation Equation (4-191)
⎯ is general and may be written for the special case of an ideal gas
Gi = Γi(T) + RT ln fˆi ⎯ig
nGig nVig nHig

d  =  dP − 2 dT +
In accord with Eq. (4-217) for an ideal solution, this becomes RT RT RT i RT
Gi = Γi(T) + RT ln xi fi Subtraction of this equation from Eq. (4-191) gives

⎯ ⎯id fˆi nGR nVR nHR GRi
By difference Gi − Gi = RT ln 
xifi  RT
d  =  dP − 2 dT +

dni (4-236)
⎯ ⎯ ⎯ ⎯
The left side is the partial excess Gibbs energy GEi ; the dimensionless where the definitions GR G − Gig and GRi Gi − Gigi have been
ratio fixi fi on the right is the activity coefficient of species i in solution, imposed. Equation (4-236) is the fundamental residual-property rela-
given the symbol γi, and by definiton, tion. An alternative form follows by introduction of the fugacity coef-
ficient given by Eq. (4-203). The result is listed as Eq. (4-237) in the
fˆi upper left quadrant of Table 4-6.
γi  (4-221)
xifi Limited forms of this equation are particularly useful. Division by
⎯E dP and restriction to constant T and composition lead to
Thus, Gi = RT ln γi (4-222)
Comparison with Eq. (4-203) shows⎯that Eq. (4-222) relates γ⎯
i to Gi
exactly as Eq. (4-203) relates φ̂i to GRi . For an ideal solution, GEi = 0,
RT ∂P  T, x

and therefore γi = 1. Similarly, the result of division by dT and restriction to constant P and
Property Changes of Mixing A property change of mixing is composition is
defined by ∂(GRRT)
∆M M −
xiMi (4-223)
 = −T 
RT ∂T   P,x
where M represents a molar thermodynamic property of a homoge- Also implicit in Eq. (4-237) is the relation
neous solution and Mi is the molar property of pure species i at the T ∂(nGRRT)
and P of the solution and in the same physical state. Applications are
usually to liquids.

ln φ̂i = 
∂ni T,P,nj

Each of Eqs. (4-213) through (4-216) is an expression for an ideal This equation demonstrates that ln φ̂i is a partial property with respect
solution property, and each may be combined with the defining equa- to GR/RT. The summability relation therefore applies, and
tion for an excess property [Eq. (4-219)], yielding the equations of
the first column of Table 4-5. In view of Eq. (4-223) these may be  =
xi ln φ̂i (4-241)
written as shown in the second column of Table 4-5, where ∆G, ∆V, RT i

TABLE 4-5 Relations Connecting Property Changes of Mixing and Excess Properties
ME in relation to M ME in relation to ∆M Expressions for ∆Mid
G = G −
xiGi − RT
xi ln xi
(4-224) G = ∆G − RT
xi ln xi
(4-228) ∆G = RT
xi ln xi
i i i i

V = V −
(4-225) V = ∆V
(4-229) ∆V = 0

SE = S −
xiSi + R
xi ln xi (4-226) SE = ∆S + R
xi ln xi (4-230) ∆Sid = − R
xi ln xi (4-234)
i i i i

HE = H −
xiHi (4-227) HE = ∆H (4-231) ∆H id = 0 (4-235)

TABLE 4-6 Fundamental Property Relations for the Gibbs Energy and Related Properties
General equations for an open system Equations for 1 mol (constant composition)

d(nG) = nV dP − nS dT +
µi dni (4-13) dG = V dP − S dT (4-17)

nG nV nH Gi G V H

d  =  dP − 2 dT +

d  =  dP − 2 dT

d  =  dP − 2 dT +
ln φ̂i dni
RT i

d  =  dP − 2 dT

d  =  dP − 2 dT +
ln γi dni
d  =  dP − 2 dT

Equations for partial molar properties (constant composition) Gibbs-Duhem equations

⎯ ⎯ ⎯
dGi = dµi = Vi dP − Si dT (4-256) V dP − S dT =
xi dµi (4-260)
⎯ ⎯ ⎯ ⎯
G µi V Hi V H Gi
d i = d 
= i dP − 
dT (4-257)  dP − 2 dT =
xi d 
RT RT i RT  (4-261)
⎯ ⎯R ⎯

Vi dP − HiR dT
d  = d ln φ̂i =  2 (4-258)  dP − 2 dT =
xi d ln φ̂i (4-262)
⎯ ⎯E ⎯E

Vi Hi
d  = d ln γi =  dP − 2 dT (4-259)  dP − 2 dT =
xi d ln γi (4-263)

Application of Eq. (4-240) to an expression giving GR as a function Fundamental Excess-Property Relation Equations for excess
of composition yields an equation for ln φ̂i. In the simplest case of a gas properties are developed in much the same way as those for residual
mixture for which the virial equation [Eq. (4-67)] is appropriate, properties. For the special case of an ideal solution, Eq. (4-191)
Eq. (4-69) provides the relation becomes
nGid nVid nHid

nGR P d  =  dP −  dT +
 =  (nB) RT RT RT 2
i RT
Subtraction of this equation from Eq. (4-191) yields
Differentiation in accord with Eqs. (4-240) and (4-181) yields

P ⎯
ln φ̂i =  Bi (4-242)  RT
d  =  dP − 2 dT +
RT ⎯ ⎯ ⎯
⎯ where the definitions GE G − Gid and GEi Gi − Gidi have been
where Bi is given by Eq. (4-182). For a binary system these equations imposed. Equation (4-247) is the fundamental excess-property rela-
reduce to tion. An alternative form follows by introduction of the activity coeffi-
P cient as given by Eq. (4-222). This result is listed as Eq. (4-248) in the
ln φ̂1 =  (B11 + y22δ12) (4-243a) upper left quadrant of Table 4-6.
RT The following equations are in complete analogy to those for resid-
ual properties.
ln φ̂2 =  (B22 + y21δ12) (4-243b) VE ∂(GERT)
RT  = 
RT ∂P  T, x

where δ12 2B12 − B11 − B22

For the special case of pure species i, these equations reduce to  = −T 
RT ∂T   P, x
ln φi =  Bii (4-244) ∂(nGERT)
For the generic cubic equation of state [Eqs. (4-104)], GR/RT is

ln γi = 
T, P,nj

given by Eq. (4-109), which in view of Eq. (4-200) for a pure species This last equation demonstrates that ln γi is a partial property with
becomes respect to GE/RT, implying also the validity of the summability relation
ln φi = Zi − 1 − ln(Zi − βi) − qiIi (4-245) GE
xi ln γi (4-252)
For species i in solution Smith, Van Ness, and Abbott [Introduction to RT i
Chemical Engineering Thermodynamics, 7th ed., pp. 562–563, The equations of the upper left quadrant of Table 4-6 reduce to
McGraw-Hill, New York (2005)] show that those of the upper right quadrant for n = 1 and dni = 0. Each equation
⎯ in the upper left quadrant has a partial-property analog, as shown in
ln φ̂i =  (Z − 1) − ln(Z − β) − ⎯
qiI (4-246) the lower left quadrant. Each equation of the upper left quadrant is a
b special case of Eq. (4-172) and therefore has associated with it a
Gibbs-Duhem equation of the form of Eq. (4-173). These are shown
Symbols without subscripts represent mixture properties, and I is in the lower right quadrant. The equations of Table 4-6 store an enor-
given by Eq. (4-112). mous amount of information, but they are so general that their direct

application is seldom appropriate. However, by inspection one can Application of Eq. (4-251) yields
write a vast array of relations valid for particular applications. For
ln γ1 = x22 [A12 + 2(A21 − A12) x1] (4-270a)
example, one sees immediately from Eqs. (4-258) and (4-259) that
∂ ln φ̂i ln γ2 = x [A21 + 2(A12 − A21) x2]

Vi 1
 =  (4-264)
∂P T,x RT When x1 = 0, ln γ 1∞ = A12; when x2 = 0, ln γ 2∞ = A21.
⎯R An alternative equation is obtained when the reciprocal quantity
∂ ln φ̂i

Hi x1x2RT/GE is expressed as a linear function of x1:
 = − 2 (4-265)
∂T P,x RT x1x2
⎯E  = A′ + B′(x1 − x2) = A′ + B′(2x1 − 1)
∂ ln γi GE/RT

 =  (4-266)
∂P T,x RT This may also be written as
∂ ln γi

Hi x1x2
 = − 2 (4-267)  = A′(x1 + x2) + B′(x1 − x2) = (A′ + B′)x1 + (A′ − B′) x2

The substitutions A′ + B′ = 1/A′21 and A′ − B′ = 1/A′12 ultimately produce

Excess properties find application in the treatment of liquid solutions. GE A′12 A′21
 =  (4-271)
Of primary importance for engineering calculations is the excess x1x2RT A′12x1 + A′21x2
Gibbs energy GE, because its canonical variables are T, P, and compo-
sition, the variables usually specified or sought in design calculations. The activity coefficients implied by this equation are given by
Knowing GE as a function of T, P, and composition, one can in princi-
A′12 x1 −2

ple compute from it all other excess properties.
The excess volume for liquid mixtures is usually small, and in accord ln γ1 = A′12 1 +  (4-272a)
A′21 x2
with Eq. (4-249) the pressure dependence of GE is usually ignored.
Thus, engineering efforts to model GE center on representing its com- A′21x2 −2
position and temperature dependence. For binary systems at constant
T, GE becomes a function of just x1, and the quantity most conve-

ln γ2 = A′21 1 + 
A′12 x1 (4-272b)

niently represented by an equation is GE/x1x2RT. The simplest proce- These are the van Laar equations. When x1 = 0, ln γ 1∞= A′12; when x2 = 0,
dure is to express this quantity as a power series in x1: ln γ 2∞ = A′21.
The Redlich-Kister expansion, the Margules equations, and the van
GE Laar equations are all special cases of a very general treatment based
 = a + bx1 + cx21 + · · · (constant T) on rational functions, i.e., on equations for GE given by ratios of poly-
nomials [Van Ness and Abbott, Classical Thermodynamics of Nonelec-
An equivalent power series with certain advantages is the Redlich- trolyte Solutions: With Applications to Phase Equilibria, Sec. 5-7,
Kister expansion [Redlich, Kister, and Turnquist, Chem. Eng. Progr. McGraw-Hill, New York (1982)]. Although providing great flexibility
Symp. Ser. No. 2, 48: 49–61 (1952)]: in the fitting of VLE data for binary systems, they are without theo-
retical foundation, with no basis in theory for their extension to multi-
GE component systems. Nor do they incorporate an explicit temperature
 = A + B(x1 − x2) + C(x1 − x2)2 + · · · dependence for the parameters.
x1x2RT Theoretical developments in the molecular thermodynamics of liq-
uid solution behavior are often based on the concept of local composi-
In application, different truncations of this expansion are appropri-
tion, presumed to account for the short-range order and nonrandom
ate, and for each truncation specific expressions for ln γ1 and ln γ2
molecular orientations resulting from differences in molecular size
result from application of Eq. (4-251). When all parameters are zero,
and intermolecular forces. Introduced by G. M. Wilson [J. Am. Chem.
GE/RT = 0, and the solution is ideal. If B = C = . . . = 0, then
Soc. 86: 127−130 (1964)] with the publication of a model for GE, this
concept prompted the development of alternative local composition
 =A models, most notably the NRTL (Non-Random Two-Liquid) equation
x1x2RT of Renon and Prausnitz [AIChE J. 14: 135−144 (1968)] and the UNI-
QUAC (UNIversal QUAsi-Chemical) equation of Abrams and Praus-
where A is a constant for a given temperature. The corresponding nitz [AIChE J. 21: 116−128 (1975)].
equations for ln γ1 and ln γ2 are The Wilson equation, like the Margules and van Laar equations,
contains just two parameters for a binary system (Λ12 and Λ21) and is
ln γ1 = Ax22 (4-268a)
written as
ln γ 2 = Ax21 (4-268b) GE
 = − x1 ln(x1 + x2Λ12) − x2 ln(x2 + x1Λ21) (4−273)
The symmetric nature of these relations is evident. The infinite dilu- RT
tion values of the activity coefficients are ln γ 1∞ = ln γ 2∞ = A.
Λ12 Λ21
If C = · · · = 0, then

ln γ1 = − ln (x1 + x2Λ12) + x2  − 
x1 + x2Λ12 x2 + x1Λ21 (4-274a)
 = A + B(x1 − x2) = A + B(2x1 − 1) Λ12 Λ21

ln γ2 = − ln (x2 + x1Λ21) − x1  − 
x1 + x2Λ12 x2 + x1Λ21 (4-274b)
and in this case GE/x1x2RT is linear in x1. The substitutions A + B = A21
and A−B = A12 transform this expression to the Margules equation: At infinite dilution,

G E ln γ 1∞ = −ln Λ12 + 1 − Λ21 ln γ 2∞ = −ln Λ21 + 1 − Λ12

 = A21x1 + A12x2 (4-269)
x1x2RT Both Λ12 and Λ21 must be positive numbers.

The NRTL equation contains three parameters for a binary system xkΛki
and is written as and ln γi = 1 − ln 
x Λ −

j ij

x Λ k j kj
GE G21τ21 G12τ12 j
 =  +  (4-275) where Λij = 1 for i = j, etc. All indices in these equations refer to the
x1x2RT x1 + x2G21 x2 + x1G12
same species, and all summations are over all species. For each ij pair
G12τ12 there are two parameters, because Λij ≠ Λji. For example, in a ternary
(x + x G ) 
G21 2
ln γ1 = x22 τ21  (4-276a) system the three possible ij pairs are associated with the parameters
x1 + x2G21 2
2 1 12
Λ12, Λ21; Λ13, Λ31; and Λ23, Λ32.
G21τ21 The temperature dependence of the parameters is given by
(x + x G ) 
G12 2
ln γ2 = x21 τ12  (4-276b)
x2 + x1G12 2 Vj −aij
1 2 21
Λij =  exp  i≠j (4-279)
Here G12 = exp(−ατ12) G21 = exp(−ατ21) where Vj and Vi are the molar volumes of pure liquids j and i and aij is
b12 b21 a constant independent of composition and temperature. Molar vol-
and τ12 =  τ21 =  umes Vj and Vi, themselves weak functions of temperature, form ratios
RT RT that in practice may be taken as independent of T, and are usually
where α, b12, and b21, parameters specific to a particular pair of evaluated at or near 25°C.
species, are independent of composition and temperature. The infi- The Wilson parameters Λij and NRTL parameters Gij inherit a
nite dilution values of the activity coefficients are Boltzmann-type T dependence from the origins of the expressions for
G , but it is only approximate. Computations of properties sensitive to
ln γ ∞1 = τ21 + τ12 exp (−ατ12) ln γ ∞2 = τ12 + τ21 exp (−ατ21) this dependence (e.g., heats of mixing and liquid/liquid solubility) are
The local composition models have limited flexibility in the fitting in general only qualitatively correct. However, all parameters are
of data, but they are adequate for most engineering purposes. More- found from data for binary (in contrast to multicomponent) systems,
over, they are implicitly generalizable to multicomponent systems and this makes parameter determination for the local composition
without the introduction of any parameters beyond those required to models a task of manageable proportions.
describe the constitutent binary systems. For example, the Wilson The UNIQUAC equation treats g GERT as made up of two addi-
equation for multicomponent systems is written as tive parts, a combinatorial term gC, accounting for molecular size and
shape differences, and a residual term gR (not a residual property),
GE accounting for molecular interactions:
 = −
xi ln
RT i

x Λ
j ij (4-277)
g = gC + gR (4-280)

2000 T S 2000 T S 1000

1000 1000 T S
J mol1

J mol1

J mol1

0 0
1000 1000
G G 2000
0 1 0 1 0 1
x1 x1 x1

(a) (b) (c)

1000 1000 1000
J mol1

J mol1

J mol1


0 0 0
1000 1000 G 1000

0 1 0 1 0 1
x1 x1 x1

(d) (e) (f)

FIG. 4-4 Property changes of mixing at 50!C for six binary liquid systems: (a) chloroform(1)/n-heptane(2); (b) acetone(1)/
methanol(2); (c) acetone(1)/chloroform(2); (d) ethanol(1)/n-heptane(2); (e) ethanol(1)/chloroform(2); (f ) ethanol(1)/water(2).
[Smith, Van Ness, and Abbott, Introduction to Chemical Engineering Thermodynamics, 7th ed., p. 455, McGraw-Hill, New York

Function gC contains pure-species parameters only, whereas function An expression for ln γ i is found by application of Eq. (4-251) to the
gR incorporates two binary parameters for each pair of molecules. For UNIQUAC equation for g [Eqs. (4-280) through (4-282)]. The result
a multicomponent system, is given by the following equations:
Φi θi ln γ i = ln γ iC + ln γ Ri
g C =
xi ln  + 5
qi xi ln  (4-281) (4-286)
i xi i Φi
Ji Ji
g R = −
qi xi ln 
j θ τ
j ji (4-282) Li 
ln γ iC = 1 − Ji + ln Ji − 5qi 1 −  + ln 
Li (4-287)

Φi  (4-283) 
ln γ iR = qi 1 − ln si −
sj (4-288)

xjrj j
where in addition to Eqs. (4-284) and (4-285),
xi qi
and θi  (4-284) ri

xj qj Ji =  (4-289)
rj xj

Subscript i identifies species, and j is a dummy index; all summations qi

are over all species. Note that τji ≠ τij; however, when i = j, then τii = Li =  (4-290)
τjj = 1. In these equations ri (a relative molecular volume) and qi (a rel-
qj xj
ative molecular surface area) are pure-species parameters. The influ- j
ence of temperature on g enters through the interaction parameters τji
of Eq. (4-282), which are temperature-dependent: si =
θ l τli (4-291)
− (uji − uii) Again subscript i identifies species, and j and l are dummy indices.
τji = exp  (4-285)
RT Values for the parameters of the commonly used models for the
excess Gibbs energy are given by Gmehling, Onken, and Arlt [Vapor-
Parameters for the UNIQUAC equation are therefore values of Liquid Equilibrium Data Collection, Chemistry Data Series, vol. 1,
uji − uii. parts 1–8, DECHEMA, Frankfurt/Main (1974–1990)].

J mol1

J mol1

J mol1

500 TS E 500 1000

0 1 0 1 0 1
x1 x1 x1

(a) (b) (c)

1000 500
J mol1

500 0
J mol1

J mol1

0 0

500 TS E
500 TS E 1000
0 1 0 1 0 1
x1 x1 x1

(d ) (e) (f )
FIG. 4-5 Excess properties at 50!C for six binary liquid systems: (a) chloroform(1)/n-heptane(2); (b) acetone(1)/methanol(2);
(c) acetone(1)/chloroform(2); (d) ethanol(1)/n-heptane(2); (e) ethanol(1)/chloroform(2); (f ) ethanol(1)/water(2). [Smith, Van
Ness, and Abbott, Introduction to Chemical Engineering Thermodynamics, 7th ed., p. 420, McGraw-Hill, New York (2005).]

0.6 0.6 0.2

In 1 In 2
0.4 0.4 0.4 In 2
In 2 In 1
In 1 0.6
0.2 0.2

0 1 0 1 0 1
x1 x1 x1

(a) (b) (c)

3 1.5 1.5

2 In 1 1.0 In 1 1.0 In 1
In 2 In 2
In 2
1 0.5 0.5

0 1 0 1 0 1
x1 x1 x1

(d ) (e) (f)
FIG. 4-6 Activity coefficients at 50!C for six binary liquid systems: (a) chloroform(1)/n-heptane(2); (b)
acetone(1)/methanol(2); (c) acetone(1)/chloroform(2); (d) ethanol(1)/n-heptane(2); (e) ethanol(1)/chloro-
form(2); (f) ethanol(1)/water(2). [Smith, Van Ness, and Abbott, Introduction to Chemical Engineering Ther-
modynamics, 7th ed., p. 445, McGraw-Hill, New York (2005).]

Behavior of Binary Liquid Solutions Property changes of GE allows calculation of SE by Eq. (4-10), written for excess proper-
mixing and excess properties find greatest application in the descrip- ties as
tion of liquid mixtures at low reduced temperatures, i.e., at tempera- HE − GE
tures well below the critical temperature of each constituent species. SE =  (4-292)
The properties of interest to the chemical engineer are VE ( ∆V),
HE ( ∆H), SE, ∆S, GE, and ∆G. The activity coefficient is also of with ∆S then given by Eq. (4-230).
special importance because of its application in phase equilibrium Figure 4-4 displays plots of ∆H, ∆S, and ∆G as functions of compo-
calculations. sition for six binary solutions at 50°C. The corresponding excess prop-
The volume change of mixing (VE = ∆V), the heat of mixing erties are shown in Fig. 4-5; the activity coefficients, derived from
(HE = ∆H), and the excess Gibbs energy GE are experimentally acces- Eq. (4-251), appear in Fig. 4-6. The properties shown here are insen-
sible, ∆V and ∆H by direct measurement and GE (or ln γ i ) indirectly sitive to pressure and for practical purposes represent solution prop-
by reduction of vapor/liquid equilibrium data. Knowledge of HE and erties at 50°C and low pressure (P ≈ 1 bar).


CRITERIA Combination gives dUt − dW − TdSt ≤ 0

The equations developed in preceding sections are for PVT systems in Because mechanical equilibrium is assumed, dW = −PdVt, whence
states of internal equilibrium. The criteria for internal thermal and
mechanical equilibrium simply require uniformity of temperature and
dUt + PdVt − TdSt ≤ 0
pressure throughout the system. The criteria for phase and chemical
reaction equilibria are less obvious. The inequality applies to all incremental changes toward the equilib-
If a closed PVT system of uniform T and P, either homogeneous or rium state, whereas the equality holds at the equilibrium state where
heterogeneous, is in thermal and mechanical equilibrium with its sur- change is reversible.
roundings, but is not at internal equilibrium with respect to mass Constraints put on this expression produce alternative criteria for
transfer or chemical reaction, then changes in the system are irre- the directions of irreversible processes and for the condition of equi-
versible and necessarily bring the system closer to an equilibrium librium. For example, dUSt ,V ≤ 0. Particularly important is fixing T and
t t

state. The first and second laws written for the entire system are P; this produces
dUt = dQ + dW dSt ≥ 
T d(Ut + PVt − TSt)T,P ≤ 0 or dGtT,P ≤ 0

This expression shows that all irreversible processes occurring at phase. The number of these variables is 2 + (N − 1)π. The masses of
constant T and P proceed in a direction such that the total Gibbs the phases are not phase rule variables, because they have nothing to
energy of the system decreases. Thus the equilibrium state of a closed do with the intensive state of the system.
system is the state with the minimum total Gibbs energy attainable at The equilibrium equations that may be written express chemical
the given T and P. At the equilibrium state, differential variations may potentials or fugacities as functions of T, P, and the phase composi-
occur in the system at constant T and P without producing a change in tions, the phase rule variables:
Gt. This is the meaning of the equilibrium criterion 1. Equation (4-295) for each species, giving (π − 1)N phase equi-
librium equations
dGtT,P = 0 (4-293)
2. Equation (4-296) for each independent chemical reaction, giv-
This equation may be applied to a closed, nonreactive, two-phase ing r equations
system. Each phase taken separately is an open system, capable of The total number of independent equations is therefore (π − 1)N + r.
exchanging mass with the other; Eq. (4-13) is written for each phase: Because the degrees of freedom of the system F is the difference
between the number of variables and the number of equations,
d(nG)′ = −(nS)′ dT + (nV)′ dP +
µ′i dni′ F = 2 + (N − 1)π − (π − 1)N − r
d(nG)″ = −(nS)″ dT + (nV)″ dP +
µ″i dn″i or F=2−π+N−r (4-297)
The number of independent chemical reactions r can be determined
where the primes and double primes denote the two phases; the pre- as follows:
sumption is that T and P are uniform throughout the two phases. The 1. Write formation reactions from the elements for each chemical
change in the Gibbs energy of the two-phase system is the sum of compound present in the system.
these equations. When each total-system property is expressed by an 2. Combine these reaction equations so as to eliminate from the set
equation of the form nM = (nM)′ + (nM)″, this sum is given by all elements not present as elements in the system. A systematic pro-
d(nG) = (nV)dP − (nS)dT +
µ′i dn′i +
µ″i dni″ cedure is to select one equation and combine it with each of the other
i i equations of the set so as to eliminate a particular element. This usu-
If the two-phase system is at equilibrium, then application of Eq. (4-293) ally reduces the set by one equation for each element eliminated,
yields although two or more elements may be simultaneously eliminated.
The resulting set of r equations is a complete set of independent
dGtT,P d(nG)T,P =
µ′i dni′ +
µi″dn″i = 0 reactions. More than one such set is often possible, but all sets num-
i i ber r and are equivalent.
The system is closed and without chemical reaction; material balances
therefore require that dn″i = −dn′i, reducing the preceding equation to Example 2: Application of the Phase Rule
a. For a system of two miscible nonreacting species in vapor/liquid equilibrium,

i (µ′ − µ″)dn′ = 0
i i i F=2−π+N−r=2−2+2−0=2
The 2 degrees of freedom for this system may be satisfied by setting T and P, or
Because the dn′i are independent and arbitrary, it follows that µ′i = µ″i. T and y1, or P and x1, or x1 and y1, etc., at fixed values. Thus for equilibrium at a
This is the criterion of two-phase equilibrium. It is readily generalized particular T and P, this state (if possible at all) exists only at one liquid and one
to multiple phases by successive application to pairs of phases. The vapor composition. Once the 2 degrees of freedom are used up, no further spec-
general result is ification is possible that would restrict the phase rule variables. For example,
one cannot in addition require that the system form an azeotrope (assuming this
µi′ = µi″ = µi″′ = · · · (4-294) is possible), for this requires x1 = y1, an equation not taken into account in the
derivation of the phase rule. Thus the requirement that the system form an
Substitution for each µi by Eq. (4-202) produces the equivalent result: azeotrope imposes a special constraint, making F = 1.
fˆi ′ = fˆi″ = fˆi″′ = · · · (4-295) b. For a gaseous system consisting of CO, CO2, H2, H2O, and CH4 in chemi-
cal reaction equilibrium,
These are the criteria of phase equilibrium applied in the solution of
practical problems.
For the case of equilibrium with respect to chemical reaction within The value of r = 2 is found from the formation reactions:
a single-phase closed system, combination of Eqs. (4-13) and (4-293)
leads immediately to C + 12O2 → CO C + O2 → CO2
H2 + 12O2 → H2O C + 2H2 → CH4

i µ dn = 0
i i (4-296)
Systematic elimination of C and O2 from this set of chemical equations
For a system in which both phase and chemical reaction equilibrium reduces the set to two. Three possible pairs of equations may result, depending
on how the combination of equations is effected. Any pair of the following
prevail, the criteria of Eqs. (4-295) and (4-296) are superimposed. three equations represents a complete set of independent reactions, and all
pairs are equivalent.
CH4 + H2O → CO + 3H2
The intensive state of a PVT system is established when its tempera- CO + H2O → CO2 + H2
ture and pressure and the compositions of all phases are fixed. How-
ever, for equilibrium states not all these variables are independent, CH4 + 2H2O → CO2 + 4H2
and fixing a limited number of them automatically establishes the The result, F = 4, means that one is free to specify, for example, T, P, and two
others. This number of independent variables is given by the phase mole fractions in an equilibrium mixture of these five chemical species, pro-
rule, and it is called the number of degrees of freedom of the system. It vided nothing else is arbitrarily set. Thus it cannot simultaneously be required
is the number of variables that may be arbitrarily specified and that that the system be prepared from specified amounts of particular constituent
must be so specified in order to fix the intensive state of a system at species.
equilibrium. This number is the difference between the number of
variables needed to characterize the system and the number of equa- Duhem’s Theorem Because the phase rule treats only the inten-
tions that may be written connecting these variables. sive state of a system, it applies to both closed and open systems.
For a system containing N chemical species distributed at equilib- Duhem’s theorem, on the other hand, is a rule relating to closed sys-
rium among π phases, the phase rule variables are T and P, presumed tems only: For any closed system formed initially from given masses of
uniform throughout the system, and N − 1 mole fractions in each prescribed chemical species, the equilibrium state is completely

determined by any two properties of the system, provided only that The second step is the evaluation of the change in fugacity of the
the two properties are independently variable at the equilibrium state. liquid with a change in pressure to a value above or below Pisat. For this
The meaning of completely determined is that both the intensive and isothermal change of state from saturated liquid at Pisat to liquid at
extensive states of the system are fixed; not only are T, P, and the pressure P, Eq. (4-17) is integrated to give
phase compositions established, but so also are the masses of the

phases. Gi − Gisat = Vi dP

VAPOR/LIQUID EQUILIBRIUM Equation (4-199) is then written twice: for Gi and for Gsat
i . Subtraction
Vapor/liquid equilibrium (VLE) relationships (as well as other inter- provides another expression for Gi − Gisat:
phase equilibrium relationships) are needed in the solution of many
engineering problems. The required data can be found by experi- fi
Gi − Gisat = RT ln 
ment, but measurements are seldom easy, even for binary systems, fisat
and they become ever more difficult as the number of species
increases. This is the incentive for application of thermodynamics to Equating the two expressions for Gi − Gisat yields

the calculation of phase equilibrium relationships. fi 1 P
The general VLE problem treats a multicomponent system of ln  =  Vi dP
N constituent species for which the independent variables are T, P, fisat RT sat

N − 1 liquid-phase mole fractions, and N − 1 vapor-phase mole frac- Because Vi, the liquid-phase molar volume, is a very weak function of
tions. (Note that Σixi = 1 and Σiyi = 1, where xi and yi represent liquid P at temperatures well below Tc, an excellent approximation is usually
and vapor mole fractions, respectively.) Thus there are 2N indepen- obtained when evaluation of the integral is based on the assumption
dent variables, and application of the phase rule shows that exactly N that Vi is constant at the value for saturated liquid Vil:
of these variables must be fixed to establish the intensive state of the
system. This means that once N variables have been specified, the fi Vil(P − Psat
i )
remaining N variables can be determined by simultaneous solution of ln sat
the N equilibrium relations fi RT
fˆi l = fˆiv i = 1, 2, . . . , N (4-298) i = φ i P i and solving for fi give
Substituting f sat sat sat

where superscripts l and v denote the liquid and vapor phases, Vil(P − P sat
i )
respectively. fi = φ sat
i P i exp 
In practice, either T or P and either the liquid-phase or vapor-phase RT
composition are specified, thus fixing 1 + (N − 1) = N independent The exponential is known as the Poynting factor.
variables. The remaining N variables are then subject to calculation, Equation (4-299) may now be written as
provided that sufficient information is available to allow determina-
tion of all necessary thermodynamic properties. yiPΦi = xi γi Psat
i i = 1, 2, . . . , N (4-303)
Gamma/Phi Approach For many VLE systems of interest, the
pressure is low enough that a relatively simple equation of state, such φ̂i −Vil(P − P sat
i )
where Φi = sa exp  (4-304)
as the two-term virial equation, is satisfactory for the vapor phase. Liquid- φi t
phase behavior, on the other hand, is described by an equation for the
excess Gibbs energy, from which activity coefficients are derived. The If evaluation of φ isat and φ̂i is by Eqs. (4-244) and (4-243), this reduces to
fugacity of species i in the liquid phase is given by Eq. (4-221), and the
vapor-phase fugacity is given by Eq. (4-204). These are here written as ⎯
PBi − Psat
i Bii − Vi (P − Pi )
l sat

fˆil = γi xi fi and fˆiv = φ̂ vi yiP Φi = exp  (4-305)

By Eq. (4-298), γi xi fi = φ̂iyi P i = 1, 2, . . . , N (4-299) ⎯
where Bi is given by Eq. (4-182).
Identifying superscripts l and v are omitted here with the understand- The N equations represented by Eq. (4-303) in conjunction with
ing that γi and fi are liquid-phase properties, whereas φ̂i is a vapor- Eq. (4-305) may be solved for N unknown phase equilibrium vari-
phase property. Applications of Eq. (4-299) represent what is known ables. For a multicomponent system the calculation is formidable, but
as the gamma/phi approach to VLE calculations. well suited to computer solution.
Evaluation of φ̂i is usually by Eq. (4-243), based on the two-term When Eq. (4-303) is applied to VLE for which the vapor phase is an
virial equation of state. The activity coefficient γi is ultimately based on ideal gas and the liquid phase is an ideal solution, it reduces to a very
Eq. (4-251) applied to an expression for GE/RT, as described in the simple expression. For ideal gases, fugacity coefficients φ̂i and φsat
i are

section “Models for the Excess Gibbs Energy.” unity, and the right side of Eq. (4-304) reduces to the Poynting factor.
The fugacity fi of pure compressed liquid i must be evaluated at the For the systems of interest here, this factor is always very close to
T and P of the equilibrium mixture. This is done in two steps. First, unity, and for practical purposes Φi = 1. For ideal solutions, the activ-
one calculates the fugacity coefficient of saturated vapor φvi = φisat by an ity coefficients γi are also unity, and Eq. (4-303) reduces to
integrated form of Eq. (4-205), most commonly by Eq. (4-242) evalu- yiP = xiP sat
i i = 1, 2, . . . , N (4-306)
ated for pure species i at temperature T and the corresponding vapor
pressure P = Pisat. Equation (4-298) written for pure species i becomes an equation which expresses Raoult’s law. It is the simplest possible
equation for VLE and as such fails to provide a realistic representation
fiv = fil = fisat (4-300) of real behavior for most systems. Nevertheless, it is useful as a stan-
dard of comparison.
where fisat indicates the value both for saturated liquid and for saturated Modified Raoult’s Law Of the qualifications that lead to
vapor. Division by Pisat yields corresponding fugacity coefficients: Raoult’s law, the one least often reasonable is the supposition of solu-
tion ideality for the liquid phase. Real solution behavior is reflected by
fisat fiv fil values of activity coefficients that differ from unity. When γi of Eq.
= sat
=  (4-303) is retained in the equilibrium equation, the result is the mod-
Pi Pi P sat
ified Raoult’s law:
or φvi = φ li = φ sat
i (4-301) yiP = xi γi P sat
i i = 1, 2, . . . , N (4-307)

This equation is often adequate when applied to systems at low to with parameters
moderate pressures and is therefore widely used. Bubble point and
i Ai Bi Ci
dew point calculations are only a bit more complex than the same cal-
culations with Raoult’s law. 1 14.3145 2756.22 −45.090
Activity coefficients are functions of temperature and liquid-phase 2 13.8193 2696.04 −48.833
composition and are correlated through equations for the excess
Activity coefficients are given by Eq. (4-274), the Wilson equation:
Gibbs energy. When an appropriate correlating equation for GE is not
available, suitable estimates of activity coefficients may often be ln γ1 = −ln(x1 + x2Λ12) + x2λ (B)
obtained from a group contribution correlation. This is the “solution
of groups” approach, wherein activity coefficients are found as sums of ln γ2 = −ln(x2 + x1Λ21) − x1λ (C)
contributions from the structural groups that make up the molecules
of a solution. The most widely applied such correlations are based on Λ12 Λ21
where λ  − 
the UNIQUAC equation, and they have their origin in the UNIFAC x1 + x2Λ12 x2 + x1Λ21
method (UNIQUAC Functional-group Activity Coefficients), pro-
posed by Fredenslund, Jones, and Prausnitz [AIChE J. 21: 1086–1099 Vj − aij
By Eq. (4-279) Λij =  exp  i≠j
(1975)], and given detailed treatment by Fredenslund, Gmehling, and Vi RT
Rasmussen [Vapor-Liquid Equilibrium Using UNIFAC, Elsevier,
Amsterdam (1977)]. with parameters [Gmehling et al., Vapor-Liquid Data Collection, Chemistry
Subsequent development has led to a variety of applications, Data Series, vol. 1, part 3, DECHEMA, Frankfurt/Main (1983)]
including liquid/liquid equilibria [Magnussen, Rasmussen, and Fre-
a12 a21 V1 V2
denslund, Ind. Eng. Chem. Process Des. Dev. 20: 331–339 (1981)], cal mol−1 cal mol−1 cm3 mol−1 cm3 mol−1
solid/liquid equilibria [Anderson and Prausnitz, Ind. Eng. Chem. Fun-
dam. 17: 269–273 (1978)], solvent activities in polymer solutions 985.05 453.57 74.05 131.61
[Oishi and Prausnitz, Ind. Eng. Chem. Process Des. Dev. 17: 333–339 When T and x1 are given, the calculation is direct, with final values for vapor
(1978)], vapor pressures of pure species [Jensen, Fredenslund, and pressures and activity coefficients given immediately by Eqs. (A), (B), and (C).
Rasmussen, Ind. Eng. Chem. Fundam. 20: 239–246 (1981)], gas sol- In all other cases either T or x1 or both are initially unknown, and calculations
ubilities [Sander, Skjold-Jørgensen, and Rasmussen, Fluid Phase require trial or iteration.
Equilib. 11: 105–126 (1983)], and excess enthalpies [Dang and Tas- a. BUBL P calculation: Find y1 and P, given x1 and T. Calculation here is
sios, Ind. Eng. Chem. Process Des. Dev. 25: 22–31 (1986)]. direct. For x1 = 0.40 and T = 325.15 K (52!C), Eqs. (A), (B), and (C) yield the
values listed in the table on the following page. Equations (4-308) and (4-307)
The range of applicability of the original UNIFAC model has been then become
greatly extended and its reliability enhanced. Its most recent revision
and extension is treated by Wittig, Lohmann, and Gmehling [Ind. P = x1γ1P sat
1 + x2 γ2 P 2 = (0.40)(1.8053)(87.616) + (0.60)(1.2869)(58.105)

Eng. Chem. Res. 42: 183–188 (2003)], wherein are cited earlier perti- = 108.134 kPa
nent papers. Because it is based on temperature-independent para-
meters, its application is largely restricted to 0 to 150°C. x1γ1Psat (0.40)(1.8053)(87.616)
y1 =  =  = 0.5851
Two modified versions of the UNIFAC model, based on temperature- P 108.134
dependent parameters, have come into use. Not only do they provide
a wide temperature range of applicability, but also they allow corre- b. DEW P calculation: Find x1 and P, given y1 and T. With x1 an unknown, the
lation of various kinds of property data, including phase equilibria, activity coefficients cannot be immediately calculated. However, an iteration
infinite dilution activity coefficients, and excess properties. The most scheme based on Eqs. (4-309) and (4-307) is readily devised, and is part of
any solve routine of a software package. Starting values result from setting each
recent revision and extension of the modified UNIFAC (Dortmund) γi = 1. For y1 = 0.4 and T = 325.15 K (52!C), results are listed in the accompany-
model is provided by Gmehling et al. [Ind. Eng. Chem. Res. 41: ing table.
1678–1688 (2002)]. An extended UNIFAC model called KT-UNI- c. BUBL T calculation: Find y1 and T, given x1 and P. With T unknown, nei-
FAC is described in detail by Kang et al. [Ind. Eng. Chem. Res. 41: ther the vapor pressures nor the activity coefficients can be initally calculated.
3260–3273 (2003)]. Both papers contain extensive literature citations. An iteration scheme or a solve routine with starting values for the unknowns is
The UNIFAC model has also been combined with the predictive required. Results for x1 = 0.32 and P = 80 kPa are listed in the accompanying
Soave-Redlich-Kwong (PSRK) equation of state. The procedure is table.
d. DEW T calculation: Find x1 and T, given y1 and P. Again, an iteration
most completely described (with background literature citations) by scheme or a solve routine with starting values for the unknowns is required. For
Horstmann et al. [Fluid Phase Equilibria 227: 157–164 (2005)]. y1 = 0.60 and P = 101.33 kPa, results are listed in the accompanying table.
Because Σ iyi = 1, Eq. (4-307) may be summed over all species to yield e. Azeotrope calculations: As noted in Example 1a, only a single degree of
freedom exists for this special case. The most sensitive quantity for identifying
P =
xi γi P sat
i (4-308) the azeotropic state is the relative volatility, defined as
Alternatively, Eq. (4-307) may be solved for xi, in which case summing y1/x1
over all species yields y2/x2

1 Because yi = xi for the azeotropic state, α12 = 1. Substitution for the two ratios by
P =  (4-309) Eq. (4-307) provides an equation for calculation of α12 from the thermodynamic

yi /γiP sati

γ 1Psat
Example 3: Dew and Bubble Point Calculations As indicated by α12 = 
γ 2P sat
Example 2a, a binary system in vapor/liquid equilibrium has 2 degrees of free-
dom. Thus of the four phase rule variables T, P, x1, and y1, two must be fixed to
allow calculation of the other two, regardless of the formulation of the equilib- Because α12 is a monotonic function of x1, the test of whether an azeotrope
rium equations. Modified Raoult’s law [Eq. (4-307)] may therefore be applied to exists at a given T or P is provided by values of α12 in the limits of x1 = 0 and x1 = 1.
the calculation of any pair of phase rule variables, given the other two. If both values are either > 1 or < 1, no azeotrope exists. But if one value is < 1
The necessary vapor pressures and activity coefficients are supplied by data and the other > 1, an azeotrope necessarily exists at the given T or P. Given T,
correlations. For the system acetone(1)/n-hexane(2), vapor pressures are given the azeotropic composition and pressure is found by seeking the value of P that
by Eq. (4-142), the Antoine equation: makes x1 = y1 or that makes α12 = 1. Similarly, given P, one finds the azeotropic
composition and temperature. Shown in the accompanying table are calculated
Bi azeotropic states for a temperature of 46!C and for a pressure of 101.33 kPa. At
i /kPa = Ai − 
ln P sat i = 1, 2 (A)
TK + Ci 46°C, the limiting values of α12 are 8.289 at x1 = 0 and 0.223 at x1 = 1.

T/K P1sat/ kPa P2sat/ kPa γ1 γ2 x1 y1 P/kPa Subtraction of Eq. (4-316) from Eq. (4-315) gives
a. 325.15 87.616 58.105 1.8053 1.2869 0.4000 0.5851 108.134
dg* γ1 γ *1 d ln γ * d ln γ *

b. 325.15 87.616 58.105 3.5535 1.0237 0.1130 0.4000 87.939  −  = ln  − ln  − x1 1 + x2 2
c. 317.24 65.830 43.591 2.1286 1.1861 0.3200 0.5605 80.000 dx1 dx1 γ2 γ *2 dx1 dx1
d. 322.98 81.125 53.779 1.6473 1.3828 0.4550 0.6000 101.330
e. 319.15 70.634 46.790 1.2700 1.9172 0.6445 = 0.6445 89.707 The differences between like terms represent residuals between
f. 322.58 79.986 53.021 1.2669 1.9111 0.6454 = 0.6454 101.330 derived and experimental values. Defining these residuals as
Given values are italic; calculated results are boldface. γ γ γ*
δg g − g* and δ ln 1 ln 1 − ln 1
Data Reduction Correlations for GE and the activity coefficients γ2 γ2 γ *2
are based on VLE data taken at low to moderate pressures. Group- puts this equation into the form
contribution methods, such as UNIFAC, depend for validity on para-
dδg γ1 d ln γ * d ln γ *

meters evaluated from a large base of such data. The process of
 = δ ln  − x1 1 + x2 2
finding a suitable analytic relation for g ( GERT) as a function of its dx1 γ2 dx1 dx1
independent variables T and x1, thus producing a correlation of VLE
data, is known as data reduction. Although g is in principle also a func- If a data set is reduced so as to yield parameters—α, β, etc.—that
tion of P, the dependence is so weak as to be universally and properly make the δ g residuals scatter about zero, then the derivative on the
neglected. Given here is a brief description of the treatment of data left is effectively zero, and the preceding equation becomes
taken for binary systems under isothermal conditions. A more com-
prehensive development is given by Van Ness [J. Chem. Thermodyn. γ d ln γ * d ln γ *
δ ln 1 = x1 1 + x2 2 (4-317)
27: 113–134 (1995); Pure & Appl. Chem. 67: 859–872 (1995)]. γ2 dx1 dx1
Presumed in all that follows is the existence of an equation inher-
ently capable of correlating values of GE for the liquid phase as a func- The right side of this equation is the quantity required by Eq. (4-313),
tion of x1: the Gibbs-Duhem equation, to be zero for consistent data. The resid-
ual on the left is therefore a direct measure of deviations from the
g GE/RT = G(x1; α, β, . . .) (4-310) Gibbs-Duhem equation. The extent to which values of this residual
where α, β, etc., represent adjustable parameters. fail to scatter about zero measures the departure of the data from con-
The measured variables of binary VLE are x1, y1, T, and P. Experi- sistency with respect to this equation.
mental values of the activity coefficient of species i in the liquid are The data reduction procedure just described provides parameters
related to these variables by Eq. (4-303), written as in the correlating equation for g that make the δg residuals scatter
about zero. This is usually accomplished by finding the parameters
yi*P* that minimize the sum of squares of the residuals. Once these para-
γ i* =  Φi i = 1, 2 (4-311)
i meters are found, they can be used for the calculation of derived val-
ues of both the pressure P and the vapor composition y1. Equation
where Φi is given by Eq. (4-305) and the asterisks denote experimen- (4-303) is solved for yi P and written for species 1 and for species 2.
tal values. A simple summability relation analogous to Eq. (4-252) Adding the two equations gives
defines an experimental value of g*:
g* x1 ln γ *1 + x2 ln γ*2 (4-312) x1γ1Psat x2γ2Psat
P=  + 
1 2
Moreover, Eq. (4-263), the Gibbs-Duhem equation, may be written Φ1 Φ2
for experimental values in a binary system at constant T and P as
y1 =  1
d ln γ* d ln γ* whence by Eq. (4-303), (4-319)
x1 1 + x2 2 = 0 (4-313) Φ1P
dx1 dx1
Because experimental measurements are subject to systematic error, These equations allow calculation of the primary residuals:
sets of values of ln γ *1 and ln γ *2 may not satisfy, i.e., may not be consis- δP P − P* and δy1 y1 − y*1
tent with, the Gibbs-Duhem equation. Thus Eq. (4-313) applied to
sets of experimental values becomes a test of the thermodynamic con- If the experimental values P* and y*1 are closely reproduced by the cor-
sistency of the data, rather than a valid general relationship. relating equation for g, then these residuals, evaluated at the experi-
Values of g provided by the equation used to correlate the data, as mental values of x1, scatter about zero. This is the result obtained when
represented by Eq. (4-310), are called derived values, and produce the data approach thermodynamic consistency. When they do not,
derived values of the activity coefficients by Eqs. (4-180) with M g: these residuals fail to scatter about zero and the correlation for g does
not properly reproduce the experimental values P* and y*1.
dg Such a correlation is unnecessarily divergent. An alternative is to
ln γ1 = g + x2  (4-314a)
dx1 base data reduction on just the P-x1 data subset; this is possible
because the full P-x1-y1 data set includes redundant information.
dg Assuming that the correlating equation is appropriate to the data, one
ln γ2 = g − x1  (4-314b)
dx1 merely searches for values of the parameters α, β, etc., that yield pres-
sures by Eq. (4-318) that are as close as possible to the measured val-
These two equations combine to yield ues. The usual procedure is to minimize the sum of squares of the
residuals δP. Known as Barker’s method [Austral. J. Chem. 6: 207−210
dg γ1 (1953)], it provides the best possible fit of the experimental pressures.
 = ln  (4-315)
dx1 γ2 When experimental y*1 values are not consistent with the P*-x1 data,
Barker’s method cannot lead to calculated y1 values that closely match
This equation is valid for derived property values. The corresponding the experimental y*1 values. With experimental error usually concen-
experimental values are given by differentiation of Eq. (4-312): trated in the y*1 values, the calculated y1 values are likely to be more
nearly correct. Because Barker’s method requires only the P*-x1 data
dg* d ln γ * d ln γ *
 = x1 1 + ln γ *1 + x2 2 − ln γ *2 subset, the measurement of y*1 values is not usually worth the extra
dx1 dx1 dx1 effort, and the correlating parameters α, β, etc., are usually best deter-
mined without them. Hence, many P*-x1 data subsets appear in the
dg* γ* d ln γ * d ln γ *
or  = ln 1 + x1 1 + x2 2 (4-316) literature; they are of course not subject to a test for consistency by the
dx1 γ *2 dx1 dx1 Gibbs-Duhem equation.

For the solvent, species 2, the analog of Eq. (4-319) is

Henry’s law
y2 =  2
x1(γ1γ1∞)k1 x2γ2P sat
f^1 Because y1 + y2 = 1, P =  +  2
φ̂1 Φ2
The same correlation that provides for the evaluation of γ1 also allows
evaluation of γ 1∞.
There remains the problem of finding Henry’s constant from the
available VLE data.
0 1 For equilibrium fˆ1 fˆ1l = fˆ1v = y1Pφ̂1
fˆ1 y1
Division by x1 gives  = Pφ̂1 
FIG. 4-7 Plot of solute fugacity fˆ1 versus solute mole fraction. [Smith, Van x1 x1
Ness, and Abbott, Introduction to Chemical Engineering Thermodynamics, 7th
ed., p. 555, McGraw-Hill, New York (2005).] Henry’s constant is defined as the limit as x1 →0 of the ratio on the left;
∞ y1
k1 = Psat
2 φ̂1 lim 
x1 → 0 x1
The world’s store of VLE data has been compiled by Gmehling et al.
[Vapor-Liquid Equilibrium Data Collection, Chemistry Data Series, The limiting value of y1/x1 can be found by plotting y1/x1 versus x1 and
vol. 1, parts 1–8, DECHEMA, Frankfurt am Main (1979–1990)]. extrapolating to zero.
Solute/Solvent Systems The gamma/phi approach to VLE cal- K Values, VLE, and Flash Calculations A measure of the dis-
culations presumes knowledge of the vapor pressure of each species at tribution of a chemical species between liquid and vapor phases is the
the temperature of interest. For certain binary systems species 1, des- K value, defined as the equilibrium ratio:
ignated the solute, is either unstable at the system temperature or is
supercritical (T > Tc). Its vapor pressure cannot be measured, and its Ki i (4-325)
fugacity as a pure liquid at the system temperature f1 cannot be calcu- xi
lated by Eq. (4-302).
It has no thermodynamic content, but may make for computational
Equations (4-303) and (4-304) are applicable to species 2, desig-
convenience through elimination of one set of mole fractions in favor
nated the solvent, but not to the solute, for which an alternative
of the other. It does characterize “lightness” of a constituent species.
approach is required. Figure 4-7 shows a typical plot of the liquid-
A “light” species, with K > 1, tends to concentrate in the vapor phase
phase fugacity of the solute fˆ1 versus its mole fraction x1 at constant
whereas a “heavy” species, with K < 1, tends to concentrate in the liq-
temperature. Since the curve representing fˆ1 does not extend all the
uid phase.
way to x1 = 1, the location of f1, the liquid-phase fugacity of pure
The rigorous evaluation of a K value follows from Eq. (4-299):
species 1, is not established. The tangent line at the origin, represent-
ing Henry’s law, provides alternative information. The slope of the y γi fi
tangent line is Henry’s constant, defined as Ki i =  (4-326)
xi φ̂iP

k1 lim 1 (4-320) When Raoult’s law applies, Eq. (4-326) reduces to Ki = PisatP. For
x1→0 x1 modified Raoult’s law, Ki = γiPisatP. With Ki = yi xi, these are alterna-
tive expressions of Raoult’s law and modified Raoult’s law. Were
This is the definition of k1 for temperature T and for a pressure equal
Raoult’s valid, K values could be correlated as functions of just T and
to the vapor pressure of the pure solvent P2sat.
P. However, Eq. (4-326) shows that they are in general functions of T,
The activity coefficient of the solute at infinite dilution is
P, {xi}, and {yi}, making convenient and accurate correlation impossi-
fˆ1 1 fˆ ble. Those correlations that do exist are approximate and severely
lim γ1 = lim  =  lim 1 limited in application. The nomographs for K values of light hydrocar-
x1→0 x1→0 x1 f1 f1 1 x1
x →0
bons as functions of T and P, prepared by DePriester [Chem. Eng.
In view of Eq. (4-320), this becomes γ 1∞ = k1 f1, or Progr. Symp. Ser. No. 7, 49: 1–43 (1953)], do allow for an average
effect of composition, but their essential basis is Raoult’s law.
k1 The defining equation for K can be rearranged as yi = Ki xi. The sum
f1 =  (4-321) Σiyi = 1 then yields
γ 1∞
where γ 1∞ represents the infinite dilution value of the activity coeffi-
i K x = 1
i i (4-327)
cient of the solute. Because both k1 and γ 1∞ are evaluated at P2sat, this With the alternative rearrangement xi = yi/Ki, the sum Σi xi = 1 yields
pressure also applies to f1. However, the effect of P on a liquid-phase
fugacity, given by a Poynting factor, is very small and for practical pur- yi
poses may usually be neglected. The activity coefficient of the solute
then becomes
Thus for bubble point calculations, where the xi are known, the prob-
fˆ1 y1Pφ̂1 y1Pφ̂1γ 1∞ lem is to find the set of K values that satisfies Eq. (4-327), whereas for
γ1  =  = 
x1 f1 x1 f1 x1k1 dew point calculations, where the yi are known, the problem is to find
the set of K values that satisfies Eq. (4-328).
For the solute, this equation takes the place of Eqs. (4-303) and (4-304). The flash calculation is a very common application of VLE. Consid-
Solution for y1 gives ered here is the P, T flash, in which are calculated the quantities and
compositions of the vapor and liquid phases in equilibrium at known
x1(γ1γ 1∞)k1
y1 =   (4-322) T, P, and overall composition. This problem is determinate on the
φˆ 1P basis of Duhem’s theorem: For any closed system formed initally from

given masses of prescribed chemical species, the equilibrium state is P

completely determined when any two independent variables are fixed. r
The independent variables are here T and P, and systems are formed
from given masses of nonreacting chemical species.
For 1 mol of a system with overall composition represented by the
set of mole fractions {zi}, let L represent the molar fraction of the sys-
tem that is liquid (mole fractions {xi}) and let V represent the molar t
fraction that is vapor (mole fractions {yi}). The material balance equa-
tions are M W
L +V=1 and zi = xiL + yiV i = 1, 2, . . . , N
Combining these equations to eliminate L gives
zi = xi(1 − V ) + yiV i = 1, 2, . . . , N (4-329) V
Substitute xi = yi /Ki and solve for yi:
yi =  i = 1,2, . . . , N
1 + V(Ki − 1) s
Because Σiyi = 1, this equation, summed over all species, yields FIG. 4-8 A subcritical isotherm on a PV diagram for a pure fluid. [Smith, Van
ziKi Ness, and Abbott, Introduction to Chemical Engineering Thermodynamics, 7th

1 + V (K − 1)
=1 (4-330) ed., p. 557, McGraw-Hill, New York (2005).]

The initial step in solving a P, T flash problem is to find the value of V

which satisfies this equation. Note that V = 1 is always a trivial solu-
tion. P, but of composition. Thus, Eq. (4-331) represents N complex rela-
tionships connecting T, P, {xi}, and {yi}.
Example 4: Flash Calculation The system of Example 3 has the Two widely used cubic equations of state appropriate for VLE calcu-
overall composition z1 = 0.4000 at T = 325.15 K and P = 101.33 kPa. Determine lations, both special cases of Eq. (4-100) [with Eqs. (4-101) and (4-102)],
V, x1, and y1. are the Soave-Redlich-Kwong (SRK) equation and the Peng-Robinson
The BUBL P and DEW P calculations at T = 325.15 K of Example 3a and 3b (PR) equation. The present treatment is applicable to both. The pure
show that for x1 = z1, Pbubl = 108.134 kPa, and for y1 = z1, Pdew = 87.939 kPa. numbers ε, σ, Ψ, and Ω and expressions for α(Tri) specific to these equa-
Because P here lies between these values, the system is in two-phase equilib- tions are listed in Table 4-2. The associated expression for φ̂i is given by
rium, and a flash calculation is appropriate. Eq. (4-246).
The modified Raoult’s law K values are given by
The simplest application of equations of state in vapor/liquid equi-
1 ) (γ2)(Psat
2 ) librium is to the calculation of vapor pressures Pisat of pure liquids.
K1 =  and K2 = 
P P Vapor pressures can of course be measured, but values are also
implicit in cubic equations of state.
z1 − x1 A subcritical PV isotherm, generated by a cubic equation of state, is
Equation (4-329) may be solved for V : V= 
y1 − x1 shown in Fig. 4-8. Three segments are evident. The very steep one on
Equation (4-330) here becomes
the left (rs) is characteristic of liquids. Note that as P → ∞,V → b,
where b is a constant in the cubic equation. The gently sloping seg-
 +  = 1
(z2)(K2) ment on the right (tu) is characteristic of vapors; here P → 0 as
1 + V(K1 − 1) 1 + V(K2 − 1) V → ∞. The middle segment (st), with both a minimum (note P < 0)
A trial calculation illustrates the nature of the solution. Vapor pressures are and a maximum, provides a transition from liquid to vapor, but has no
taken from Example 3a or 3b; a trial value of x1 then allows calculation of γ1 and physical meaning. The actual transition occurs along a horizontal line,
γ2 by Eqs. (B) and (C) of Example 3. The values of K1, K2, and V that result are such as connects points M and W.
substituted into the summation equation. In the unlikely event that the sum is For pure species i, Eq. (4-331) reduces to φvi = φli, which may be
indeed unity, the chosen value of x1 is correct. If not, then successive trials eas- written as
ily lead to this value. Note that the trivial solution giving V = 1 must be avoided.
More elegant solution procedures can of course be employed. The answers are ln φvi = ln φli (4-332)
x1 = 0.2373 y1 = 0.5190 V = 0.5775 For given T, line MW lies at the vapor pressure Pisat if and only if the
with γ1 = 2.5297 γ2 = 1.0997 K1 = 2.1873 K2 = 0.6306 fugacity coefficients for points M and W satisfy Eq. (4-332). These
points then represent saturated liquid and vapor phases in equilib-
rium at temperature T.
Equation-of-State Approach Although the gamma/phi approach The fugacity coefficients in Eq. (4-332) are given by Eq. (4-245):
to VLE is in principle generally applicable to systems comprised of
ln φip = Zi − 1 − ln(Zi − βi) − qi Ii
p p p
subcritical species, in practice it has found use primarily where pres- p = l, v (4-333)
sures are no more than a few bars. Moreover, it is most satisfactory for Expressions for Z iv and Zil come from Eqs. (4-104):
correlation of constant-temperature data. A temperature dependence
for the parameters in expressions for GE is included only for the local Zvi − βi
composition equations, and it is at best approximate. Ziv = 1 + βi − qiβi  (4-334)
A generally applicable alternative to the gamma/phi approach (Zi + %βi)(Ziv + σβi)

results when both the liquid and vapor phases are described by the
1 + βi − Zli

same equation of state. The defining equation for the fugacity coeffi-
Zli = βi + (Z li + %βi)(Zli + σβi)  (4-335)
cient, Eq. (4-204), may be applied to each phase: qiβi
Liquid: fˆi l = φ̂ li xiP Vapor: fˆiv = φ̂ vi yi P
and Iip comes from Eq. (4-112):
By Eq. (4-298), xiφ̂ il = yiφ̂ vi i = 1, 2, . . . , N (4-331)
1 Z pi + σβi
This introduces compositions xi and yi into the equilibrium equations, Iip =  ln  p = l, v (4-336)
but neither is explicit, because the φ̂i are functions, not only of T and σ− % Zip + %βi

The equation-of-state parameters are independent of phase. As defined Each line includes three segments as described for the isotherm of
by Eq. (4-105), βi is a function of P and here becomes Fig. 4-8: the leftmost segment representing a liquid phase and the
rightmost segment, a vapor phase, both with the same composition.
biPsat Each left segment contains a bubble point (saturated liquid), and each
βi  i
RT right segment contains a dew point (saturated vapor). Because these
points for a given line are for the same composition, they do not rep-
The remaining equation-of-state parameters, given by Eqs. (4-101), resent phases in equilibrium and do not lie at the same pressure.
(4-102), and (4-106), are functions of T only and are written here as Shown in Fig. 4-9 is a bubble point B on the solid line and a dew point
D on the dashed line. Because they lie at the same P, they represent
α(Tr )R2Tc2
ai(T) = ψ  i i
(4-338) phases in equilibrium, and the lines are characterized by the liquid
Pc i and vapor compositions.
For a BUBL P calculation, the temperature and the liquid composi-
bi = Ω c i
(4-339) tion are known, and this fixes the location of the PV isotherm for the
Pc i composition of the liquid phase (solid line). The problem then is to
locate a second (dashed) line for a vapor composition such that the
ai(T) line contains a dew point D on its vapor segment that lies at the pres-
qi  (4-340)
biRT sure of the bubble point B on the liquid segment of the solid line. This
pressure is the phase equilibrium pressure, and the composition for
The eight equations (4-332) through (4-337) may be solved for the the dashed line is that of the equilibrium vapor. This equilibrium con-
eight unknowns Pisat, βi, Zil, Ziv, Iil, Iiv, ln φil, and ln φiv. dition is shown by Fig. 4-9.
Perhaps more useful is the reverse calculation whereby an equation- In the absence of a theory to prescribe the composition depen-
of-state parameter is evaluated from a known vapor pressure. dence of parameters for cubic equations of state, empirical mixing
Thus, Eqs. (4-332) and (4-333) may be combined and solved for qi, rules are used to relate mixture parameters to pure-species parame-
yielding ters. The simplest realistic expressions are a linear mixing rule for
Zvi − Zli + ln [(Zli − βi)/(Ziv − βi)] parameter b and a quadratic mixing rule for parameter a, as shown by
qi =  (4-341) Eqs. (4-113) and (4-114). A common combining rule is given by Eq.
Iiv − Iil (4-115). The general mole fraction variable xi is used here because
application is to both liquid and vapor mixtures. These equations,
Expressions for Zil, Ziv, Iil, Iiv, and βi are given by Eqs. (4-334) through
known as van der Waals prescriptions, provide for the evaluation of
(4-337). Because Zil and Ziv depend on qi, an iterative procedure is
mixture parameters solely from parameters for the pure constituent
indicated, with a starting value for qi from a generalized correlation as
species. They find application primarily for mixtures comprised of
given by Eqs. (4-338), (4-339), and (4-340).
simple and chemically similar molecules.
For mixtures the presumption is that the equation of state has
Useful in the application of cubic equations of state to mixtures
exactly the same form as when written for pure species. Equations
are partial equation-of-state parameters. For the parameters of the
(4-104) are therefore applicable, with parameters β and q given by
generic cubic, represented by Eqs. (4-104), (4-105), and (4-106), the
Eqs. (4-105) and (4-106). Here, these parameters, and therefore
definitions are
b and a(T), are functions of composition. Liquid and vapor mixtures in
equilibrium in general have different compositions. The PV isotherms ∂(na)
generated by an equation of state for these different compositions
are represented in Fig. 4-9 by two similar lines: the solid line for the

liquid-phase composition and the dashed line for the vapor-phase

⎯ ∂(nb)
composition. They are displaced from each other because the equation-
of-state parameters are different for the two compositions.
∂ni  T,nj


∂ni  T,nj

These are general equations, valid regardless of the particular mix-
ing or combining rules adopted for the composition dependence of
mixture parameters.
Parameter q is defined in relation to parameters a and b by Eq. (4-106).
nq = 

∂(nq) a⎯i bi
whence ⎯

∂ni T,nj

=q 1+  − 
a b (4-345)

Any two of the three partial parameters form an independent pair, and
any one of them can be found from ⎯ the other two. Because q, a, and b
are not linearly related, q ⎯ ≠ a⎯ b
i i iRT.
Values of φ̂i and φ̂ i as given by Eq. (4-246) are implicit in an equa-
l v

tion of state and with Eq. (4-331) allow calculation of mixture VLE.
Although more complex, the same basic principle applies as for
FIG. 4-9 Two PV isotherms at the same T for mixtures. The solid line is for a pure-species VLE. With φ̂ li a function of T, P, and {xi}, and φ̂Vi a func-
liquid-phase composition; the dashed line is for a vapor-phase composition.
Point B represents a bubble point with the liquid-phase composition; point D
tion of T, P, and {yi}, Eq. (4-331) represents N relations among the
represents a dew point with the vapor-phase composition. When these points lie 2N variables: T, P, (N −1) xi s, and (N−1) yi s. Thus, specification of N
at the same P (as shown), they represent phases in equilibrium. [Smith, Van of these variables, usually either T or P and either the liquid- or vapor-
Ness, and Abbott, Introduction to Chemical Engineering Thermodynamics, 7th phase composition, allows solution for the remaining N variables by
ed., p. 560, McGraw-Hill, New York (2005).] BUBL P, DEW P, BUBL T, and DEW T calculations.

Because of limitations inherent in empirical mixing and combining Combining this equation with Eqs. (4-349) and (4-350) leads to
rules, such as those given by Eqs. (4-113) through (4-115), the equation-
   − 1 
of-state approach has found primary application to systems exhibiting T T1
 =  − 
modest deviations from ideal solution behavior in the liquid phase, RT RT TT0 T RT T1 0
e.g., to systems containing hydrocarbons and cryogenic fluids. How-
ever, since 1990, extensive research has been devoted to developing T T T
−  ln  −  − 1 1 − I
mixing rules that incorporate the excess Gibbs energy or activity coef- R T0 T0 T
ficient data available for many systems. The extensive literature on
this subject is reviewed by Valderrama [Ind. Eng. Chem. Res. 42:

1603–1618 (2003)] and by Twu, Sim, and Tassone [Chem. Eng. 1
where dT dT dT
Progress 98:(11): 58–65 (Nov. 2002)]. RT
T1 T2 P, x
The idea here is to exploit the connection between fugacity coeffi-
cients and activity coefficients provided by their definitions: This general equation employs excess Gibbs energy data at temper-
ature T0, excess enthalpy (heat-of-mixing) data at T1, and excess heat
fˆi fˆi/xiP φˆ capacity data at T2. Integral I depends on the temperature depen-
γi  =  = i dence of CPE. Excess heat capacity data are uncommon, and the T
xifi fi/P φi
dependence is rarely known. Assuming CPE independent of T makes
the integral zero, and the closer T0 and T1 are to T, the less the influ-
Therefore, ln γi = ln φ̂ i − ln φi (4-346) ence of this assumption. When no information is available for CPE and
Because γi is a liquid-phase property, this equation is written for excess enthalpy data are available at only a single temperature, CPE
the liquid phase. Substituting for ln φ̂ i and ln φi by Eqs. (4-245) and must be assumed zero. In this case only the first two terms on the right
Eq. (4-246) gives side of Eq. (4-351) are retained, and it more rapidly becomes impre-
cise as T increases.
b Z−β ⎯I + qI For application of Eq. (4-351) to binary systems at infinite dilution
ln γi = i (Z − 1) − Zi + 1 − ln  − q of one of the constituent species, it is divided by the product x1x2.
b Zi − βi i i i

Symbols without subscripts are mixture properties. Solution for ⎯

x x RT  T
 − 1 
T T1
qi  = 
yields x1x2RT x1x2RT T0 1 2 T T1 0

Z−β −  ln  −  − 1 1
⎯ =  b
qi 1 − Zi + i (Z − 1) − ln  + qiIi − ln γi (4-347) x1x2R T0 T0 T
I b Zi − βi
The assumption here is that C is independent of T, making I = 0. As

Because q⎯ is a partial property, the summability equation provides an P

i shown by Smith, Van Ness and Abbott [Introduction to Chemical
exact mixing rule: Engineering Thermodynamics, 7th ed., p. 437, McGraw-Hill, New
York (2005)],
q =
xi ⎯
qi (4-348)

x x RT
ln γ ∞i
Application of this equation in the solution of VLE problems is illus- 1 2 xi = 0
trated by Smith, Van Ness, and Abbott [Introduction to Chemical
Engineering Thermodynamics, 7th ed., pp. 569–572, McGraw-Hill, The preceding equation may therefore be written as
New York (2005)].
  − 1 
Extrapolation of Data with Temperature Liquid-phase excess- T T1
ln γ ∞i = (ln γ ∞i )T − 
property data for binary systems at near-ambient temperatures appear x1x2RT T T
T1, xi = 0 0
in the literature. They provide for the extrapolation of GE correlations
 ln  −   − 1  
with temperature. The key relations are Eq. (4-250), written as T T T1
−  (4-352)
x1x2R T
xi = 0 T T 0 0

d  = − 2 dT constant P, x
Example 5: VLE at Several Temperatures For the methanol(1)/
acetone(2) system at a base temperature of T0 = 323.15 K (50!C), both VLE data
and the excess-property analog of Eq. (4-26): [Van Ness and Abbott, Int. DATA Ser., Ser. A, Sel. Data Mixtures, 1978: 67
(1978)] and excess enthalpy data [Morris et al., J. Chem. Eng. Data 20: 403–405
dHE = CPE dT constant P, x (1975)] are available. The VLE data are well correlated by the Margules equa-
tions. As noted in connection with Eq. (4-270), parameters A12 and A21 relate
Integration of the first equation from T0 to T gives directly to infinite dilution values of the activity coefficients. Thus, we have from
the VLE data at 323.15 K:

A12 = ln γ ∞1 = 0.6281 and A21 = ln γ ∞2 = 0.6557
 =  dT 2
These values allow calculation of equilibrium pressures through Eqs. (4-270)
and (4-308) for comparison with the measured pressures of the data set. Values
Integration of the second equation from T1 to T yields of Pisat required in Eq. (4-308) are the measured values reported with the data

 C dT
T set. The root-mean-square (rms) value of the pressure differences is given in
HE = H1E + E
(4-350) Table 4-7 as 0.08 kPa, thus confirming the suitability of the Margules equation
T1 for this system. Vapor-phase mole fractions were not reported; hence no value
can be given for rms δy1.
∂CPE Experimental VLE data at 372.8 and 397.7 K are given by Wilsak et al. [Fluid
In addition, 
dCEP = 
∂T P, x
dT Phase Equilib. 28: 13–37 (1986)]. Values of ln γ ∞i and hence of the Margules
parameters for these higher temperatures are found from Eq. (4-352) with CPE = 0.
The required excess enthalpy values at T0 are



x x RT 
x x RT
Integrate from T2 to T: 2
dT = 1.3636 and = 1.0362
T2 P, x 1 2 T0, x1=0 1 2 T0, x2 = 0

TABLE 4-7 VLE Results for Methanol(1)/Acetone(2) the νi are stoichiometric coefficients and the Ai stand for chemical
formulas. The νi themselves are called stoichiometric numbers, and
T, K A12 = ln γ ∞1 A21 = ln γ ∞2 RMS δP/kPa RMS % δP RMS δy1
associated with them is a sign convention such that the value is posi-
323.15 0.6281 0.6557 0.08 0.12 tive for a product and negative for a reactant. More generally, for a
372.8 0.4465 0.5177 0.85 0.22 0.004 system containing N chemical species, any or all of which can partici-
(0.4607) (0.5271) (0.83) (0.006) pate in r chemical reactions, the reactions are represented by the
397.7 0.3725 0.4615 2.46 0.32 0.014
(0.3764) (0.4640) (1.39) (0.013)
0 =
νi,j Ai j = I, II, . . . , r (4-357)
Results of calculations with the Margules equations are displayed as the pri- − for a reactant species
mary entries at each temperature in Table 4-7. The values in parentheses are
from the gamma/phi approach as reported in the papers cited.
where sign (νi, j) = + for a product species
Results for the higher temperatures indicate the quality of predictions based
only on vapor-pressure data for the pure species and on mixture data at 323.15 K. If species i does not participate in reaction j, then vi,j = 0.
Extrapolations based on the same data to still higher temperatures can be The stoichiometric numbers provide relations among the changes
expected to become progressively less accurate. in mole numbers of chemical species which occur as the result of
chemical reaction. Thus, for reaction j
Only the Wilson, NRTL, and UNIQUAC equations are suited to
the treatment of multicomponent systems. For such systems, the ∆n1,j ∆n2,j ∆nN,j
parameters are determined for pairs of species exactly as for a binary  =  =…=  (4-358)
ν1,j ν2,j νN,j
All these terms are equal, and they can be equated to the change in a
LIQUID/LIQUID AND VAPOR/LIQUID/ single quantity ε j, called the reaction coordinate for reaction j, thereby

Equation (4-295) is the basis for both liquid/liquid equilibria (LLE) i = 1, 2, . . . , N

∆ni, j = νi,j ∆ε j
and vapor/liquid/liquid equilibria (VLLE). Thus for LLE with super- j = I, II, . . . , r (4-359)
scripts α and β denoting the two phases, Eq. (4-295) is written as
Because the total change in mole number ∆ni is just the sum of the
fˆαi = fˆβi i = 1, 2, . . . , N (4-353) changes ∆ ni,j resulting from the various reactions,
Eliminating fugacities in favor of activity coefficients gives
∆ni =
∆ni,j =
νi,j ∆ε j i = 1, 2, . . . , N (4-360)
xαi γ αi = xβi γ βi i = 1, 2, . . . , N (4-354) j j

For most LLE applications, the effect of pressure on the γi can be If the initial number of moles of species i is ni and if the convention is

ignored, and Eq. (4-354) then constitutes a set of N equations relating adopted that εj = 0 for each reaction in this initial state, then
equilibrium compositions to one another and to temperature. For a
given temperature, solution of these equations requires a single ni = ni +
νi,j ε j i = 1, 2, . . . , N (4-361)
expression for the composition dependence of GE suitable for both 0
liquid phases. Not all expressions for GE suffice, even in principle,
because some cannot represent liquid/liquid phase splitting. The Equation (4-361) is the basic expression of material balance for a
UNIQUAC equation is suitable, and therefore prediction is possible closed system in which r chemical reactions occur. It shows for a
by UNIFAC models. A special table of parameters for LLE calcula- reacting system that at most r mole-number-related quantities ε j are
tions is given by Magnussen et al. [Ind. Eng. Chem. Process Des. Dev. capable of independent variation. It is not an equilibrium relation, but
20: 331–339 (1981)]. merely an accounting scheme, valid for tracking the progress of the
A comprehensive treatment of LLE is given by Sorensen et al. reactions to arbitrary levels of conversion. The reaction coordinate has
[Fluid Phase Equilib.2: 297–309 (1979); 3: 47–82 (1979); 4: 151–163 units of moles. A change in ε j of 1 mol signifies a mole of reaction,
(1980)]. Data for LLE are collected in a three-part set compiled by meaning that reaction j has proceeded to such an extent that the
Sorensen and Arlt [Liquid-Liquid Equilibrium Data Collection, change in mole number of each reactant and product is equal to its
Chemistry Data Series, vol. 5, parts 1–3, DECHEMA, Frankfurt am stoichiometric number.
Main (1979–1980)].
For vapor/liquid/liquid equilibria, Eq. (4-295) becomes CHEMICAL REACTION EQUILIBRIA
fˆαi = fˆβi = fˆvi i = 1, 2 , . . . , N (4-355) The general criterion of chemical reaction equilibria is given by Eq.
where α and β designate the two liquid phases. With activity coeffi- (4-296). For a system in which just a single reaction occurrs, Eq. (4-361)
cients applied to the liquid phases and fugacity coefficients to the becomes
vapor phase, the 2N equilibrium equations for subcritical VLLE are
ni = ni + νiε 0
whence dni = νi dε
xαi γ αi f αi = yiφ̂ iP
xβi γ βi f βi = yiφ̂ iP
} all i (4-356) Substitution for dni in Eq. (4-296) leads to

i ν µ = 0
i i (4-362)
As for LLE, an expression for GE capable of representing liquid/liquid
phase splitting is required; as for VLE, a vapor-phase equation of state Generalization of this result to multiple reactions produces
for computing the φ̂ i is also needed.

i ν µ =0
i,j i j = I, II, . . . , r (4-363)

CHEMICAL REACTION STOICHIOMETRY Standard Property Changes of Reaction For the reaction
For a phase in which a chemical reaction occurs according to the aA + bB → lL + mM
a standard property change is defined as the property change resulting
ν1A1 + ν2A2 + · · · → ν3A3 + ν4A4 + · · · when a mol of A and b mol of B in their standard states at temperature

T react to form l mol of L and m mol of M in their standard states also The application of Eq. (4-369) requires explicit introduction of
at temperature T. A standard state of species i is its real or hypotheti- composition variables. For gas-phase reactions this is accomplished
cal state as a pure species at temperature T and at a standard state through the fugacity coefficient
pressure P°. The standard property change of reaction j is given the
symbol ∆M°j, and its general mathematical definition is fˆ y φ̂iP
âi i = i 
νi, j M°i (4-364) f i° f i°
For species present as gases in the actual reactive system, the standard However, the standard state for gases is the ideal gas state at the stan-
state is the pure ideal gas at pressure P°. For liquids and solids, it is dard state pressure, for which f i° = P°. Therefore,
usually the state of pure real liquid or solid at P°. The standard state
y φ̂iP
pressure P° is fixed at 100 kPa. Note that the standard states may rep- âi = i 
resent different physical states for different species; any or all the P°
species may be gases, liquids, or solids. and Eq. (4-369) becomes
The most commonly used standard property changes of reaction are

νi,jG°i =
νi,jµ°i (4-365) i (y φ̂ )
i i
= Kj all j (4-371)
i i

νi,jH°i (4-366) where νj Σiνi,j and P° is the standard state pressure of 100 kPa,
i expressed in the same units used for P. The yi’s may be eliminated in
favor of equilibrium values of the reaction coordinates ε j (see Example
j i
(4-367) 6). Then, for fixed temperature Eqs. (4-371) relate the ε j to P. In prin-
i ciple, specification of the pressure allows solution for the ε j. However,
The standard Gibbs energy change of reaction ∆G°j is used in the cal- the problem may be complicated by the dependence of the φ̂i on com-
culation of equilibrium compositions. The standard heat of reaction position, i.e., on the ε j. If the equilibrium mixture is assumed an ideal
∆H°j is used in the calculation of the heat effects of chemical reaction, solution, then [Eq. (4-218)] each φ̂i becomes φi, the fugacity coeffi-
and the standard heat capacity change of reaction is used for extrapo- cient of pure species i at the mixture T and P. This quantity does not
lating ∆H°j and ∆G°j with T. Numerical values for ∆H°j and ∆G°j are com- depend on composition and may be determined from experimental
puted from tabulated formation data, and ∆C°P is determined from data, from a generalized correlation, or from an equation of state.
empirical expressions for the T dependence of the C°P [see, e.g., Eq.
An important special case of Eq. (4-371) results for gas-phase reac-
tions when the phase is assumed an ideal gas. In this event φ̂i = 1, and
Equilibrium Constants For practical application, Eq. (4-363)

must be reformulated. The initial step is elimination of the µi in favor P
of fugacities. Equation (4-199) for species i in its standard state is i (y )
νi, j
= Kj all j (4-372)
subtracted from Eq. (4-202) for species i in the equilibrium mixture,
giving In the general case the evaluation of the φ̂i requires an iterative
process. An initial step is to set each φ̂i equal to unity and to solve the
µi = G°i + RT ln âi (4-368) problem by Eq. (4-372). This provides a set of yi values, allowing
where by definition âi fi/f°i and is called an activity. Substitution of evaluation of the φ̂i by, for example, Eq. (4-243) or (4-246). Equation
this equation into Eq. (4-364) yields, upon rearrangement, (4-371) can then be solved for a new set of yi values, with the process
continued to convergence.

i [ν i, j (G°i + RT ln âi)] = 0 For liquid-phase reactions, Eq. (4-369) is modified by introduction
of the activity coefficient γi = fˆi xifi, where xi is the liquid-phase mole
fraction. The activity is then
i νi, j G°i + RT
ln âνi = 0 i,j

i f
âi i = γ i xi i

νi,jG°i fi° fi°
or ln âνi = 
i i, j

i RT Both fi and fi° represent fugacity of pure liquid i at temperature T, but

at pressures P and P°, respectively. Except in the critical region, pres-
The right side of this equation is a function of temperature only for sure has little effect on the properties of liquids, and the ratio fi fi° is
given reactions and given standard states. Convenience suggests set- often taken as unity. When this is not acceptable, this ratio is evaluated
ting it equal to ln Kj, whence by the equation
i â νi,j
= Kj all j (4-369)
 V dP  
V (P − P°)
i fi 1
ln  = 
f°i RT P° RT
where, by definition, Kj exp 
RT  (4-370)
When the ratio fi fi° is taken as unity, âi = γ i xi, and Eq. (4-369) becomes
Quantity Kj is the chemical reaction equilibrium constant for reaction i (γ x ) i i
= Kj all j (4-373)
j, and ∆G°j is the corresponding standard Gibbs energy change of
reaction [see Eq. (4-365)]. Although called a “constant,” Kj is a func- Here the difficulty is to determine the γ i’s, which depend on the xi’s.
tion of T, but only of T. This problem has not been solved for the general case. Two courses
The activities in Eq. (4-369) provide the connection between the are open: the first is experiment; the second, assumption of solution
equilibrium states of interest and the standard states of the con- ideality. In the latter case, γ i = 1, and Eq. (4-373) reduces to
stituent species, for which data are presumed available. The standard
states are always at the equilibrium temperature. Although the stan- i (x ) i
= Kj all j (4-374)
dard state need not be the same for all species, for a particular species
it must be the state represented by both G°i and the f°i upon which the “law of mass action.” The significant feature of Eqs. (4-372) and
activity âi is based. (4-374), the simplest expressions for gas- and liquid-phase reaction

equilibrium, is that the temperature-, pressure-, and composition- In the more extensive compilations of data, values of ∆G° and ∆H°
dependent terms are distinct and separate. for formation reactions are given for a wide range of temperatures,
The effect of temperature on the equilibrium constant follows from rather than just at the reference temperature T0 = 298.15 K. [See in
Eq. (4-41) written for pure species j in its standard state (wherein the particular TRC Thermodynamic Tables—Hydrocarbons and TRC
pressure Po is fixed): Thermodynamic Tables—Non-hydrocarbons, serial publications of
the Thermodynamics Research Center, Texas A & M Univ. System,
d(Gj°/RT) − Hj° College Station, Tex.; “The NBS Tables of Chemical Thermodynamic
dT RT 2 Properties,” J. Phys. Chem. Ref. Data 11, supp. 2 (1982).] Where data
are lacking, methods of estimation are available; these are reviewed by
With Eqs. (4-365) and (4-366) this equation easily extends to relate Poling, Prausnitz, and O’Connell, The Properties of Gases and Liq-
standard property changes of reaction: uids, 5th ed., chap. 6, McGraw-Hill, New York, 2000. For an estima-
tion procedure based on molecular structure, see Constantinou and
d(∆Gj°RT) − ∆Hj° Gani, Fluid Phase Equilib. 103: 11–22 (1995). See also Sec. 2.
 =  (4-375)
dT RT 2
Example 6: Single-Reaction Equilibrium The hydrogenation of ben-
In view of Eq. (4-370) this may also be written as zene to produce cyclohexane by the reaction

C6H6 + 3H2 → C6H12

d ln Kj ∆Hj°
 =  (4-376)
dT RT 2 is carried out over a catalyst formulated to repress side reactions. Operating con-
ditions cover a pressure range from 10 to 35 bar and a temperature range from
For an endothermic reaction, ∆Hj° is positive and K j increases with 450 to 670 K. Reaction rate increases with increasing T, but because the reac-
increasing T; for an exothermic reaction, it is negative and Kj decreases tion is exothermic the equilibrium conversion decreases with increasing T. A
with increasing T. comprehensive study of the effect of operating variables on the chemical equi-
librium of this reaction has been published by J. Carrero-Mantilla and M. Llano-
Because the standard state pressure is constant, Eq. (4-28) may be Restrepo, Fluid Phase Equilib. 219: 181–193 (2004). Presented here are
extended to relate standard properties of reaction, yielding calculations for a single set of operating conditions, namely, T = 600 K, P = 15
bar, and a molar feed ratio H2/C6H6 = 3, the stoichiometric value. For these
dT conditions we determine the fractional conversion of benzene to cyclohexane.
d∆Hj° = ∆CP°j dT and d∆Sj° = ∆C°Pj  Carrero-Mantilla and Llano-Restrepo express ln K as a function of T by an equa-
tion which for 600 K yields the value K = 0.02874.
Integration of these equations from reference temperature T0 (usually A feed stream containing 3 mol H2 for each 1 mol C6H6 is the basis of calcu-
lation, and for this single reaction, Eq. (4-361) becomes ni = ni + νiε, yielding
298.15 K) to temperature T gives 0

nB = 1 − ε

∆C °

∆H° = ∆H°0 + R P
dT (4-377) nH = 3 − 3ε hydrogen
R T0
nC = ε cyclohexane

∆CP° dT
i n = 4 − 3ε

∆S° = ∆S°0 + R   (4-378) i

T0 R T
Each mole fraction is therefore given by yi = ni (4 − 3ε).
Assume first that the equilibrium mixture is an ideal gas, and apply Eq. (4-372),
where for simplicity subscript j has been supressed. The definition of written for a single reaction, with subscript j omitted and ν = − 3:
G leads directly to ∆G° = ∆H° − T ∆S°. Combining this equation with
Eqs. (4-370), (4-377), and (4-378) yields ε

P ν 4 − 3ε 15 −3

i yi 
= 3  = K = 0.02874 
−∆G° −∆H°0
 ∆RC° dT + ∆RS° +  ∆RC° dTT 1 − ε 3 − 3ε

ln K =  =  − 
T T0
P 0

P P°

4 − 3ε 4 − 3ε  1

ε 4 − 3ε 3
Substituting ∆S°0 = (∆G°0 − ∆H°0)/T0, rearranging, and defining τ TT0
give finally
  (15)−3 = 0.02874
1 − ε 3 − 3ε and ε = 0.815

Thus, the assumption of ideal gases leads to a calculated conversion of 81.5

−∆G°0 ∆H°0 τ − 1
 ∆C°P dT

ln K =  +   −   dT +   (4-379) An alternative assumption is that the equilibrium mixture is an ideal solution.
RT0 RT0 τ T T0 R T0 R T This requires application of Eq. (4-371). However, in the case of an ideal solu-
tion Eq. (4-218) indicates that φ̂idi = φi, in which case Eq. (4-371) for a single
When heat capacity equations have the form of Eq. (4-52), the integrals reaction becomes
are evaluated by equations of exactly the form of Eqs. (4-53) and (4-54),
but with parameters A, B, C, and D replaced by ∆A, ∆B, ∆C, and ∆D, in i (y φ )  

νi P
i i =K
accord with Eq. (4-364). Thus for the ideal gas standard state

 ∆RC° dT = ∆A T (τ − 1) + 
∆B ∆C For purposes of illustration we evaluate the pure-species fugacity coefficients by
0 T (τ 2
− 1) +  T 30(τ 3 − 1) Eq. (4-206), written here as
T0 2 3
(4-380) φi = exp(Bi0 + ωBi1) r i

∆D τ − 1
T0 τ  Tr i

The following table shows values for the various quantities in this equation. Note
that Tc and Pc for hydrogen are effective values as calculated by Eqs. (4-124) and
(4-125) and used with ω = 0.
 ∆RC° dTT = ∆A lnτ + ∆BT + ∆CT + τ∆DT  

0 (τ − 1) (4-381)
2 2
T0 0 Tc Tr Pc Pr ω B0 B1 φ
Equations (4-379) through (4-381) together allow an equation to be C6H6 562.2 1.067 48.98 0.306 0.21 − 0.2972 0.008 0.919
written for lnK as a function of T for any reaction for which appropri- H2 42.8 14.009 19.78 0.758 0.00 0.0768 0.139 1.004
ate data are available. C6H12 553.6 1.084 40.73 0.368 0.21 − 0.2880 0.016 0.908

The equilibrium equation now becomes: The first term on the right is the definition of the chemical potential;
ε therefore,
4 − 3ε
P ν µi +
λkaik = 0

15 i = 1, 2, . . . , N
i (yiφi) 
= 3 
1−ε 3 − 3ε
= K = 0.02874 (4-383)

 0.908 1.004
4 − 3ε 4 − 3ε

However, the chemical potential is given by Eq. (4-368); for gas-phase

Solution yields ε = 0.816 reactions and standard states as the pure ideal gases at Po, this equa-
tion becomes
This result is hardly different from that based on the ideal gas assumption. The fˆi
fugacity coefficients in the equilibrium equation clearly cancel one another. This µi = G°i + RT ln 
is not uncommon in reaction equilibrium calculations, as there are always prod- P°
ucts and reactions, making the ideal gas assumption far more useful than might
be expected.
If G°i is arbitrarily set equal to zero for all elements in their standard
Carrero-Mantilla and Llano-Restrep present results for a wide range of con- states, then for compounds G°i = ∆G°f i, the standard Gibbs energy change
ditions, both for the ideal gas assumption and for calculations wherein φ̂i values of formation of species i. In addition, the fugacity is eliminated—in
are determined from the Soave-Redlich-Kwong equation of state. In no case are favor of the fugacity coefficient by Eq. (4-204), fˆi = yiφ̂iP. With these
these calculated conversions significantly divergent. substitutions, the equation for µi becomes

Complex Chemical Reaction Equilibria When the composi- y φ̂iP

µi = ∆G°f + RT ln i 
tion of an equilibrium mixture is determined by a number of simulta- i

neous reactions, calculations based on equilibrium constants become
complex and tedious. A more direct procedure (and one suitable for Combination with Eq. (4-383) gives
general computer solution) is based on minimization of the total
Gibbs energy Gt in accord with Eq. (4-293). The treatment here is y φ̂iP
∆G°f + RT ln i  +
λkaik = 0 i = 1, 2, . . . , N (4-384)
limited to gas-phase reactions for which the problem is to find the i
P° k
equilibrium composition for given T and P and for a given initial feed.
1. Formulate the constraining material-balance equations, based If species i is an element, ∆G°f is zero. There are N equilibrium equa-

on conservation of the total number of atoms of each element in a sys- tions [Eqs. (4-384)], one for each chemical species, and there are w
tem comprised of w elements. Let subscript k identify a particular material balance equations [Eqs. (4-382)], one for each element—a
atom, and define Ak as the total number of atomic masses of the kth total of N + w equations. The unknowns in these equations are the ni’s
element in the feed. Further, let aik be the number of atoms of the kth (note that yi = ni /Σi ni), of which there are N, and the λk’s, of which
element present in each molecule of chemical species i. The material there are w—a total of N + w unknowns. Thus the number of equa-
balance for element k is then tions is sufficient for the determination of all unknowns.
Equation (4-384) is derived on the presumption that the set {φ̂i}

i n a i ik = Ak k = 1, 2 , . . . , w (4-382) is known. If the phase is an ideal gas, then each φ̂i is unity. If the phase
is an ideal solution, each φ̂i becomes φi and can at least be estimated.
i n a i ik − Ak = 0 k = 1, 2, . . . , w For real gases, each φ̂i is a function of the set {yi}, the quantities being
calculated. Thus an iterative procedure is indicated, initiated with
2. Multiply each element balance by λk, a Lagrange multiplier: each φ̂i set equal to unity. Solution of the equations then provides a
preliminary set {yi}. For low pressures or high temperatures this result

n a
i ik − Ak = 0 k = 1, 2, . . . , w is usually adequate. Where it is not satisfactory, an equation of state
with the preliminary set {yi} gives a new and more nearly correct set
Summed over k, these equations give {φ̂i} for use in Eq. (4-384). Then a new set {yi} is determined. The
process is repeated to convergence. All calculations are well suited to

k λ 
i n a
k i ik

− Ak = 0 computer solution.
In this procedure, the question of what chemical reactions are
3. Form a function F by addition of this sum to Gt: involved never enters directly into any of the equations. However, the
choice of a set of species is entirely equivalent to the choice of a set of
F = Gt +

n a
i ik − Ak independent reactions among the species. In any event, a set of
species or an equivalent set of independent reactions must always be
assumed, and different assumptions produce different results.
Function F is identical with G , because the summation term is zero. A detailed example of a complex gas-phase equilibrium calculation
However, the partial derivatives of F and Gt with respect to ni are dif- is given by Smith, Van Ness, and Abbott [Introduction to Chemical
ferent, because function F incorporates the constraints of the material Engineering Thermodynamics, 5th ed., Example 15.13, pp. 602–604;
balances. 6th ed., Example 13.14, pp. 511–513; 7th ed., Example 13.14,
4. The minimum value of both F and Gt is found when the partial pp. 527–528, McGraw-Hill, New York (1996, 2001, 2005)]. General
derivatives of F with respect to ni are set equal to zero: application of the method to multicomponent, multiphase systems is
treated by Iglesias-Silva et al. [Fluid Phase Equilib. 210: 229–245
∂F ∂Gt

∂n i T,P,n
∂ni  T,P,nj
λkaik = 0
(2003)] and by Sotyan, Ghajar, and Gasem [Ind. Eng. Chem. Res. 42:
3786–3801 (2003)].


Real irreversible processes can be subjected to thermodynamic analy- CALCULATION OF IDEAL WORK
sis. The goal is to calculate the efficiency of energy use or production
and to show how wasted energy is apportioned among the steps of a In any steady-state steady-flow process requiring work, a minimum
process. The treatment here is limited to steady-state steady-flow amount must be expended to bring about a specific change of state in
processes, because of their predominance in chemical technology. the flowing fluid. In a process producing work, a maximum amount is

attainable for a specific change of state in the flowing fluid. In either LOST WORK
case, the limiting value obtains when the specific change of state is
accomplished completely reversibly. The implications of this require- Work that is wasted as the result of irreversibilities in a process is
ment are that called lost work Ẇlost, and it is defined as the difference between the
1. The process is internally reversible within the control volume. actual work of a process and the ideal work for the process. Thus by
2. Heat transfer external to the control volume is reversible. definition,
The second item means that heat exchange between system and sur- Wlost Ws − Wideal (4-392)
roundings must occur at the temperature of the surroundings, presumed
to constitute a heat reservoir at a constant and uniform temperature Tσ . The rate form is Ẇlost Ẇs − Ẇideal (4-393)
This may require Carnot engines or heat pumps internal to the system
that provide for the reversible transfer of heat from the temperatures of The actual work rate comes from Eq. (4-150):
the flowing fluid to that of the surroundings. Because Carnot engines
and heat pumps are cyclic, they undergo no net change of state. Ẇs = ∆ H +  u2 + zg ṁ − Q̇
These conditions are implicit in the entropy balance of Eq. (4-156) 2 fs
when ṠG = 0. If in addition there is but a single surroundings temper-
ature Tσ, this equation becomes Subtracting the ideal work rate as given by Eq. (4-386) yields

Q̇ Ẇlost = Tσ ∆(Sṁ)fs − Q̇ (4-394)

∆(Sṁ)fs −  = 0 (4-385)
Tσ For the special case of a single stream flowing through the control
The energy balance for a steady-state steady-flow process as given
by Eq. (4-150) is Ẇlost = ṁTσ ∆S − Q̇ (4-395)
Division of this equation by ṁ gives
∆ H +  u2 + zg ṁ = Q̇ + Ẇs (4-150)
2 fs Wlost = Tσ ∆S − Q (4-396)
Combining this equation with Eq. (4-385) to eliminate Q̇ yields where the basis is now a unit amount of fluid flowing through the con-
trol volume.
∆ H +  u2 + zg ṁ = Tσ ∆(Sṁ)fs + Ẇs(rev) The total rate of entropy generation (in both system and surround-
2 fs ings) as a result of a process is
where Ẇs(rev) indicates that the shaft work is for a completely

reversible process. This work is called the ideal work Ẇideal. Thus ṠG = ∆(Sṁ)fs −  (4-397)

Ẇideal = ∆ H +  u2 + zg ṁ − Tσ ∆(Sṁ)fs (4-386) Division by ṁ provides an equation based on a unit amount of fluid
2 fs
flowing through the control volume:
In most applications to chemical processes, the kinetic and poten-
tial energy terms are negligible compared with the others; in this Q
event Eq. (4-386) is written as SG = ∆S −  (4-398)

Ẇideal = ∆(Hṁ)fs − Tσ ∆(Sṁ)fs (4-387)
Equations (4-397) and (4-398) are special cases of Eqs. (4-156) and
For the special case of a single stream flowing through the system, Eq. (4-157).
(4-387) becomes Multiplication of Eq. (4-397) by Tσ gives
Ẇideal = ṁ(∆H − Tσ ∆S) (4-388) Tσ ṠG = Tσ ∆(Sṁ)fs − Q̇
Division by ṁ puts this equation on a unit-mass basis: Because the right sides of this equation and of Eq. (4-394) are identi-
cal, it follows that
Wideal = ∆H − Tσ ∆S (4-389)
Ẇlost = Tσ ṠG (4-399)
A completely reversible process is hypothetical, devised solely to For flow on the basis of a unit amount of fluid, this becomes
find the ideal work associated with a given change of state. Its only con-
nection with an actual process is that it brings about the same change Wlost = Tσ SG (4-400)
of state as the actual process, allowing comparison of the actual work of Because the second law of thermodynamics requires
a process with the work of the hypothetical reversible process.
Equations (4-386) through (4-389) give the work of a completely ṠG ≥ 0 and SG ≥ 0
reversible process associated with given property changes in the flow-
ing streams. When the same property changes occur in an actual therefore Ẇlost ≥ 0 and Wlost ≥ 0
process, the actual work Ẇs (or Ws) is given by an energy balance, and
comparison can be made of the actual work with the ideal work. When When a process is completely reversible, the equality holds and the
Ẇideal (or Wideal) is positive, it is the minimum work required to bring lost work is zero. For irreversible processes the inequality holds, and
about a given change in the properties of the flowing streams, and it is the lost work, i.e., the energy that becomes unavailable for work, is
smaller than Ẇs. In this case a thermodynamic efficiency ηt is defined positive. The engineering significance of this result is clear: The
as the ratio of the ideal work to the actual work: greater the irreversibility of a process, the greater the rate of entropy
generation and the greater the amount of energy that becomes
Ẇ eal unavailable for work. Thus every irreversibility carries with it a price.
ηt(work required) = id (4-390)
When Ẇideal (or Wideal) is negative, Ẇideal is the maximum work ANALYSIS OF STEADY-STATE STEADY-FLOW PROCESSES
obtainable from a given change in the properties of the flowing
streams, and it is larger than Ẇs. In this case, the thermodynamic effi- Many processes consist of a number of steps, and lost-work calcula-
ciency is defined as the ratio of the actual work to the ideal work: tions are then made for each step separately. Writing Eq. (4-399) for
each step of the process and summing give
ηt(work produced) = 

Ẇ lost = Tσ

Dividing Eq. (4-399) by this result yields TABLE 4-8 States and Values of Properties for the Process of
Fig. 4-10*
Ẇlost ṠG

ṠG Point P, bar T, K Composition State H, J/mol S, J/(mol·K)
1 55.22 300 Air Superheated 12,046 82.98
Thus an analysis of the lost work, made by calculation of the fraction 2 1.01 295 Pure O2 Superheated 13,460 118.48
that each individual lost-work term represents of the total lost work, is 3 1.01 295 91.48% N2 Superheated 12,074 114.34
the same as an analysis of the rate of entropy generation, made by 4 55.22 147.2 Air Superheated 5,850 52.08
expressing each individual entropy generation term as a fraction of the 5 1.01 79.4 91.48% N2 Saturated vapor 5,773 75.82
sum of all entropy generation terms. 6 1.01 90 Pure O2 Saturated vapor 7,485 83.69
An alternative to the lost-work or entropy generation analysis is a 7 1.01 300 Air Superheated 12,407 117.35
work analysis. This is based on Eq. (4-393), written as *Properties on the basis of Miller and Sullivan, U.S. Bur. Mines Tech. Pap. 424

Ẇ lost = Ẇs − Ẇideal (4-401)

For a work-requiring process, all these work quantities are positive Thus, by Eq. (4-387),
and Ẇs > Ẇideal. The preceding equation is then expressed as
Ẇideal = −144 − (300)(−2.4453) = 589.6 J
Ẇs = Ẇideal +
Ẇlost (4-402)
Calculation of actual work of compression: For simplicity, the work of com-
pression is calculated by the equation for an ideal gas in a three-stage recipro-
A work analysis then gives each of the individual work terms in the cating machine with complete intercooling and with isentropic compression in
summation on the right as a fraction of Ẇs. each stage. The work so calculated is assumed to represent 80 percent of the
For a work-producing process, Ẇs and Ẇideal are negative, and actual work. The following equation may be found in any number of textbooks
Ẇideal > Ẇs. Equation (4-401) in this case is best written as on thermodynamics:

nγ RT1 (γ − 1)nγ
Ẇideal = Ẇs +
(4-403) Ẇs =  −1
0.8(γ − 1) 1

A work analysis here expresses each of the individual work terms on where n = number of stages, here taken as 3
the right as a fraction of Ẇideal. A work analysis cannot be carried out γ = ratio of heat capacities, here taken as 1.4
in the case where a process is so inefficient that Ẇideal is negative, indi- T1 = initial absolute temperature, equal to 300 K
cating that the process should produce work; but Ẇs is positive, indi- P2/P1 = overall pressure ratio, equal to 54.5
cating that the process in fact requires work. A lost-work or entropy R = universal gas constant, equal to 8.314 J/(mol·K)
generation analysis is always possible. The efficiency factor of 0.8 is already included in the equation. Substitution of
the remaining values gives
Example 7: Lost-Work Analysis A work analysis follows for a simple
Linde system for the separation of air into gaseous oxygen and nitrogen, as
depicted in Fig. 4-10. Table 4-8 lists a set of operating conditions for the num- Ẇs =  (54.5)0.4(3)(1.4) − 1 = 15,171 J
bered points of the diagram. Heat leaks into the column of 147 J/mol of enter- (0.8)(0.4)
ing air and into the exchanger of 70 J/mol of entering air have been assumed.
Take Tσ = 300 K. The heat transferred to the surroundings during compression as a result of inter-
The basis for analysis is 1 mol of entering air, assumed to contain 79 mol % N2 cooling and aftercooling to 300 K is found from the first law:
and 21 mol % O2. By a material balance on the nitrogen, 0.79 = 0.9148 x, whence
Q̇ = ṁ(∆H) − Ẇs = (12,046 − 12,407) −15,171 = −15,532 J
x = 0.8636 mol of nitrogen product
Calculation of lost work: Equation (4-394) may be applied to each of the
1 − x = 0.1364 mol of oxygen product major units of the process. For the compressor/cooler,
Calculation of ideal work: If changes in kinetic and potential energies are
neglected, Eq. (4-387) is applicable. From the tabulated data, Ẇlost = (300)[(82.98)(1) − (117.35)(1)] − (−15,532)
= 5221.0 J
∆(Hṁ)fs = (13,460)(0.1364) + (12,074)(0.8636) − (12,407)(1) = −144 J
For the exchanger,
∆(Sṁ)fs = (118.48)(0.1364) + (114.34)(0.8636) − (117.35)(1)= − 2.4453 JK
Ẇlost = (300)[(118.48)(0.1364) + (114.34)(0.8636) + (52.08)(1)
− (75.82)(0.8636) − (83.69)(0.1364) − (82.98)(1)] − 70
= 2063.4 J

Finally, for the rectifier,

Ẇlost = (300)[(75.82)(0.8636) + (83.69)(0.1364) − (52.08)(1)] − 147 = 7297.0 J

Work analysis: Because the process requires work, Eq. (4-402) is appropriate
for a work analysis. The various terms of this equation appear as entries in the
following table and are on the basis of 1 mol of entering air.

% of Ẇs
Ẇideal 589.6 J 3.9
Ẇlost: Compressor/cooler 5,221.0 J 34.4
Ẇlost: Exchanger 2,063.4 J 13.6
Ẇlost: Rectifier 7,297.0 J 48.1
Ẇs 15,171.0 J 100.0

The thermodynamic efficiency of this process as given by Eq. (4-390) is only

3.9 percent. Significant inefficiencies reside with each of the primary units of
FIG. 4-10 Diagram of simple Linde system for air separation. the process.