Chapter 4 Single Equilibrium Stages and Flash Cal-

culations

Main Points for Chapter 4

1. The Gibbs phase rule, which applies to intensive vari-

ables at equilibrium, determines the number of indepen-
dent variables that can be speci…ed. This rule can be
extended to the more general determination of the de-
grees of freedom (number of allowable speci…cations) for
a ‡ow system, including consideration of extensive vari-
ables. The intensive and extensive variables are related
by material and energy balance equations together with
phase equilibrium data in the form of equations, tables,
and graphs.

2. Vapor-liquid equilibrium conditions for binary systems

are conveniently represented and determined with T-y-
x, y-x and P-x diagrams. The relative volatility for a
binary system tends to 1.0 as the critical point is ap-
proached.

3. Minimum- or maximum-boiling azeotropes, which are

formed by close-boiling non-ideal liquid mixtures, are
conveniently represented by the same types of diagrams
used for nonazeotropic (Azeotropic) binary mixtures.
Highly non ideal liquid mixtures can form heterogeneous
azeotropes involving two liquid phases.

4. For multicomponent mixtures, vapor-liquid equilibrium

phase compositions and amounts can be determined by
isothermal ‡ash, adiabatic ‡ash, and bubble- and dew-
point calculations. When the mixtures are non ideal,
the computations are best done with process simulation
computer programs.

5. Liquid-liquid equilibrium conditions for ternary mixtures

are best determined graphically from triangular and other
equilibrium diagrams, unless only one of the three com-
ponents (called the solute) is soluble in the two liq-
uid phases and the system is dilute in the solute. In
that case, the conditions can be readily determined al-
gebraically using phase distribution ratios for the solute.

1
 6. Liquid-liquid equilibrium conditions for multicomponent
mixtures of four or more components are best deter-
mined with process-simulation computer programs.

7. Solubility of gases that are only sparingly soluble in a

liquid are well represented by a Henry’s law constant
that depends on temperature.

8. Solid vapor pressure can be used to determine equilib-

rium sublimation and desublimation conditions for gas-
solid systems. Adsorption isotherms and y-x diagrams
are useful in determining adsorption equilibrium condi-
tions for gas mixtures in the presence of a solid adsor-
bent.

The simplest separation process is one which causes sepa-

ration by bringing two different phases into contact until they
reach equilibrium. If the separation factor between the two
phases is large, then this single contacting stage may be suffi-
cient to bring about the desired degree of separation. If not,
then multiple stages can be used, for example as in distil-
lation. The separation factor is determined by the relative
concentrations of a component in the two phases in equilib-
rium. In the case of a vapor-liquid equilibrium between two
mixtures the volatile component, or light key (LK) will have
a higher concentration in the vapor phase in equilibrium with
a liquid phase, which will be rich in the heavy key, (HK).
The ratio of the two volatilities gives the separation factor.
In chapter four we only consider the case of a single stage
separation. Multistage separations rely on the same princi-
ples with the process being accomplished as a series of single
equilibrium stages.

Gibbs Phase
Rule and Equilibrium in a heterogeneous multicomponent system is
Degrees of reached when the temperatures, pressures and chemical po-
tentials for each component in all  phases are equal. This
Freedom gives us ( 1) conditions for temperature, ( 1) condi-
tions for pressure and ( 1) C conditions for chemical po-
tential for a total of

( 1) (C + 2)

total conditions on the system. How many independent vari-

ables do we have? Well, we have temperature and pressure of

2
each phase and the mole fraction for each component but one
in each phase (since the mole fraction of the cth component is
determined by the other C 1 mole fractions) for a total of

 (C 1) + 2 =  (C + 1)

variables. The number of degrees of freedom f for the

system is then the number of variables minus the number of
conditions giving us

f =C +2

which is known as the Gibbs Phase Rule. This tells us

how many independent ways I can change the system vari-
ables and still maintain equilibrium between  phases with
C components. Notice that the description of equilibrium in-
volves only intensive variables, because the size of the system
doesn’t matter in determining phase equilibrium.

Example 1 Consider a three component, two phase system.

In this case, f = 3 2 + 2 = 3 so that we have three de-
grees of freedom. That is, of the variables T (1) , T (2) , P (1) , P (2) ,
(1) (1) (2) (2)
x1 , x2 , x1 , x2 , I can only specify 3 (and still maintain equi-
librium between the two phases) with the rest being determined
by the conditions of equilibrium (the two equations for thermal
and mechanical equilibrium and the 3 equations for chemical
equilibrium). An alternative set of equations which is more
practical is the equality of the temperatures and pressures in
the phases, and the K-value equations ( 1 of them for each
component) and the mole fraction summations for each phase
(two of them in this example).

Note that K-values are not variables, but thermodynamics

functions relating the equilibrium concentrations of a compo-
nent in different phases.
The Gibbs phase rule does not include the variables of the
feed stream nor the extensive variables of the separation oper-
ation and is consequently limited to describing static systems.
It can however, be extended by adding these variables to make
it applicable to process operations. Additional relationships
between the extensive variables and the feed stream variables
must also be introduced to determine the overall number of
degrees of freedom within the separation operation.
Before we extend the Gibbs Phase Rule to ‡ow systems,
let’s look at a different way to determine the degrees of free-
dom of a multicomponent, multiphase, non-‡owing system.

3
For a C component system with  phases in equilibrium we
count the variables and equations as follows:
Variables:
The total number of variables in the system in equilibrium
are the mole fractions of each component in each phase C ,
and the temperature, and pressure of each phase for a total
of:

V = (C + 2) 

variables.
Equations:
To count the number of equations we note that we have
two types of equations: mole fraction summation constraints,
and equilibrium relations. The mole fraction constraints are
of the form
P
C
Xi = 1
i=1

for each phase for a total of  mole fraction summation equa-

tions.
The second type of equation is the equilibrium relation-
ship. The ones we will use most often are the relationships
between the concentrations of each species in two phases in
equilibrium. We know these are just the K-value equations,
for example for a vapor and liquid in equilibrium
YA
KV L = XA

which is just the ratio of the mole fraction of component A

in the gas phase (which is just its partial pressure), to the
mole fraction in liquid phase. We have a K-value relation for
every unique pair of phases in equilibrium for a total of  1
K-value relations for each component. So for C components,
we have C ( 1) K-value equilibrium relations. Adding this
to the  mole fraction summations and 2 ( 1) temperature
and pressure equations we have a total of

ε = C ( 1) + 3 2

equations.
The number of degrees of freedom is just the number of
variables minus the number of equations. This gives us

F =V ε
F = [(C + 2) ] [C ( 1) + 3 2]
F = C +2

4
 which is the same result for the Gibbs Phase rule as I obtained
before using a slightly different method involving equal chem-
ical potentials rather than K-value and mole fraction equa-
tions. This second approach is more straightforward and more
easily extended to ‡ow systems. Let’s try one more example
before we go on.

Example 2 For a ternary (C = 3) system with three phases

in equilibrium we

F =3 3+2= 2

This means I can specify up to 2 degrees of freedom and

still maintain 3 phases in equilibrium. How many variables do
I have? Well, T, and P of each phase and the mole fractions
of all three components in all three phases:

T (1) ,
(1) (2) (2) (3) (3) (1) (1) (1) (2) (2) (2) (3) (3) (3)
P ,T ,P ,T ,P , x1 , x2 , x3 , x1 , x2 , x3 , x1 , x2 , x3

This is a total of 15 variables. From the Gibbs phase rule

we ought to have 13 equations. We know we need 13 equations
because have 13 unknown variables once we specify two of the
…fteen. So in order to determine the 13 remaining variables
we need 13 equations. What equations do we have available
to us? First, we know we have the three mole fraction sum-
mations for each phase. Then we have two K-value relations
for each component which makes 2  3 = 6 equilibrium rela-
tions. So our total number of equations is this 6 plus the 3
from the mole fraction summation plus the four temperature
and pressure equations for a total of 313, which we already
knew.

Extension of
the Gibbs To extend to the ‡ow system we must do exactly the same
process we did for the non-‡ow system except we add some
Phase Rule
new variables: stream variables, and extensive variables; and
to Flow Sys- we add some additional relationships or equations: relations
tems between feed variables, and relations between extensive vari-
ables.
Additional Variables:
For a ‡ow system we have a feed stream, and a possible
heat exchanger. This ‡ow into the separator has the variables:

F, TF , PF , Q, Zi

5
for a total of C +4 additional variables (the C comes from the
C component mole fractions Zi ).
The product streams each add an additional variable which
is the ‡ow rate of that product stream. There are a total of 
product streams since we separate out each equilibrium phase
into a product stream. Of course, in real separation systems
we usually only have two phases and two product streams.
The total number of additional variables of a ‡ow system
over a non-‡owing system is then:

V = C +  + 4

Additional Equations:
In a ‡ow system we have the same equations as we had for
the non-‡ow system with the additional constraints of mass
and energy balances. The mass balances conserve species
so that in steady-state operations the in‡ow a component is
equal to the sum of that component over the product streams.

Fi = Vi + Li + ... up to the th phase for each

component.

This gives a total of C mass balances which are additional

equations of the ‡ow system. Energy conservation provides
us with one additional equation

FH (f eed) + Q = VH (1) + LH (2) + ... + LH ()

So with these additional C + 1 equations we have a total

change in the number of degrees of freedom from the non-‡ow
system of

F =
V
ε = [C +  + 4] [C + 1]

F =  + 3

If we add these additional degrees of freedom to the Gibbs

phase rule we get the number of degrees of freedom of a ‡ow
system:

Ff low = [C  + 2] + [ + 3]
Ff low = C + 5

So for a for a single-stage vapor-liquid equilibrium process

the additional variables are those introduced by the feed and
product streams. These include zi , TF , PF , F, Q, V, and L.
This adds C + 6 variables to the total independent variables.

6
 If we have  phases in equilibrium, then we have  2 vari-
ables that describe the how the components can be separated
into the 2 product streams. We also have C + 1 additional re-
lationships; C component balances, and the energy balance.
Stream enthalpies are not counted as variables because they
are thermodynamic functions which depend on independent
variables. The degrees of freedom are now

f = (C  + 2) + (C +  + 4) (C + 1)
f =C+5

Binary Vapor-
Liquid Sys- For many binary systems equilibrium data is available.
From this data, which can be graphical, tabular or in the form
tems
of empirical relationships we can obtain the K-value equilib-
rium relationships for the binary we are interested in.
For a binary in two phase equilibrium we have two degrees
of freedom from the Gibbs Phase Rule since C = 2 and  = 2.
For the ‡ow system there are more degrees of freedom: in this
case 7. This is because the additional ‡ow variables increase
our ‡exibility in terms of what we can vary and still maintain
2-phase equilibrium.
Analyzing the non-‡ow system we see that the variables
are the composition of the two phases and the temperature
and pressure of the two phases:

T (1) , T (2) , P (1) , P (2) , XA , YA , XB , YB

From these eight variables we can specify two and pre-

serve two-phase equilibrium. If we choose the pressure and
temperature P and T, then XA ,and YA , and XB ,and YB are
determined from the two K-value relations (which is the com-
position intercepts with the constant temperature tie line) and
XA + XB = 1, and YA + YB = 1.
So specifying the temperature and pressure completely de-
termines the compositions of the two phases in equilibrium.
Specifying the overall composition determines how much of
each phase you have (by the Lever Rule). So, specifying T,
and P …xes the separation factor at a speci…c value of
KA YA /XA
A,B = KB
= (1 YA )/(1 XA )

A chemical engineer designing a separations process thus

has a certain amount of ‡exibility in choosing the operating

7
 conditions of their system. On the other hand, that ‡exibil-
ity is limited to the conditions of thermodynamic equilibrium,
and is often used to optimize some aspect of the process, such
as maximum separation with minimum cost. On the other
hand, choosing the operating conditions incorrectly can lead
to running your system at impractical conditions. For ex-
ample, near a critical point in the phase diagram where the
separation factor goes to 1. At critical points, the physical dif-
ference between two phase disappears and no separation can
take place. Other difficulties that may arise are azeotropes in
the phase diagram.

Azeotropic
Systems Azeotropes indicate a departure from Raoult’s Law. Raoult’s
Law describes ideal systems where the activities of the com-
ponents are just equal to their mole fractions. In other words,
species contribute to the thermodynamic properties in propor-
tion to their fractional population in the mixture. Mixtures
which deviate from Raoultian behavior indicate that either
the unlike components are attracting each other or repelling
each other.
If unlike components attract each other more than like
components attract each other, then the boiling point of the
mixture will be higher than the pure components because we
have to introduce additional thermal energy to get molecules
to enter the vapor phase to overcome the attraction in the
liquid. This is called positive deviation from Raoult’s law and
often leads to a maximum boiling point homogeneous
azeotrope. Here < 1.
If unlike components attract each other less than like
components attract each other, then the boiling point of the
mixture will be lower than the pure components because we
have to introduce less thermal energy to get molecules to enter
the vapor phase to overcome the attraction in the liquid. This
is called negative deviation from Raoult’s law and often leads
to a minimum boiling point homogeneous azeotrope.
This is the most common type of azeotrope. Here > 1.
Heterogeneous minimum boiling azeotropes indicate cases
where two liquid phases vaporize into a vapor at a single com-
position. In this case >> 1. This is because in order to get
the liquid to phase separate into two liquid phases the com-
ponents must be highly repulsive..

8
 Important points to remember about azeotropes is that
they:

 limit the separation achievable by distillation because

even with in…nite trays, there is no thermodynamic driving
force to separate beyond the azeotrope.

 The azeotropic point can be moved, for example by low-

ering the pressure, so that the desired composition may
be obtained.. In fact, for some systems, the azeotrope
might not even form if the pressure is changed enough
and the azeotrope can be broken.

 Azeotropes can be used to help separations. Adding a

third component to a system which forms a minimum
boiling point azeotrope with one of the original compo-
nents can be used to entrain that component, allowing
for it to be vaporized away and recovered.

Multicomponent
Flash, Bubble- Isothermal Flash
point, and For an isothermal ‡ash we can use our degree of freedom
analysis for ‡ow systems to determine the variables and equa-
Dew-Point tions of the system.
Calculations Variables: F, V, L, Zi , Yi , Xi , TF , PF , T, P, Q
Which is 3C + 8 variables. We know we can specify C + 5
of them and still maintain two-phase equilibrium in the ‡ow
system. The C + 5 might be Zi , F, TF , PF and two others of
our choosing. If we choose T, and P as the two others, then
this leads to an isothermal ‡ash situation. So specifying Zi ,
F, TF , PF , T, and P leaves us 2C + 3 remaining variables to
be determined so we need 2C + 3 equations.
Equations:

1. Yi = Ki Xi (Phase equilibrium)

2. F Zi = V Yi + LXi (component material balance)

3. F = V + L (total materials balance)

4. hF F + Q = hV V + hL L (energy balance)

P
C P
C
5. Yi = 1 and Xi = 1 (mole fraction summations)
i=1 i=1

9
 Note that there are 2C + 4 equations here. We know from
our previous analysis that we should only have 2C + 3 equa-
tions. This apparent discrepancy is solved by noting that the
total materials balance and the two mole fraction summa-
tions are not all independent. That is, only two of these three
equations are independent. We can then reduce the number
of equations to 2C + 3 which we do by combining the mole
fraction summations into one equation
P
C P
C
Xi Yi = 0
i=1 i=1
To solve these 2C + 3 equations Rachford and Rice de-
veloped a simple procedure. This involves manipulating the
2C + 3 equations as follows. First, rewrite the materials bal-
ance as

L=F V

Substitute this into the component materials balance to

get

F Zi = V Yi + (F V ) Xi

Dividing both sides by F and using the de…nition = V/F

gives

Zi = Yi + (1 ) Xi

Eliminating Yi using Yi = Ki Xi gives

Zi = Ki Xi + (1 ) Xi

which we solve for Xi

Zi
Xi = 1+ ( Ki 1)

Using the phase equilibrium relationship we see that

Ki Zi
Yi = 1+ ( Ki 1)

Substituting this into the combined mole fraction summa-

P
C P
C
tion Xi Yi = 0 gives:
i=1 i=1

P
C
Zi P
C
Ki Zi
1+ ( Ki 1) 1+ ( Ki 1)
=0
i=1 i=1

which simpli…es to:

10
 P
C
Zi (1 Ki )
1+ ( Ki 1)
=0
i=1

Seader de…nes this as the function f ( ) and this is where

the Rachford-Rice procedure begins.
Rachford-Rice Procedure:
P
C
Zi (1 Ki )
Step 1: Solve f ( ) = 1+ ( Ki 1)
= 0 for
i=1
Step 2: Find V from = V/F
Step 3: Substitute into Xi = 1+ (ZKi i 1) to …nd Xi
Ki Zi
Step 4: Find Yi using Yi = 1+ ( Ki 1)
Step 5: Get L from F = V + L
Step 6: Get Q from hF F + Q = hV V + hL L
Steps 2 through 6 are simple linear algebraic expressions.
However, step 1 involves a non-linear equation which is best
solved using a recursive method. Otherwise, a brute force
way would be to plot f ( ) for = 0 to 1 and look for the
intercept with f = 0. A faster solution can be obtained using
Newton’s method.
In Newton’s method, the root of a function is found by
…rst guessing the root , and then using a recursive relation
to determine the subsequent trial root. The next trial root is
given by the function

h ( ( i) )i
f ( i)
( i+1) = ( i) df ( )
d ( i)

which is just
P
C
Zi (1 Ki )
1+ ( Ki 1 )
( i+1) = P
C
i=1
2
Zi (1 Ki )

i=1 [1+ ( Ki 1)]2

(here on the right side of the equation is actually ( i) )

Before using Rachford-Rice, it is important to make sure
that you are in a two-phase equilibrium condition, otherwise
you won’t be doing Flash distillation, and Rachford-Rice pro-
cedure will …nd unphysical roots. Check:
a) If f (0)  0, then the solution is above the bubble point.
b) If f (1)  0, the solutions is below the dew point.
So if both conditions are true, then the mixture is between
the bubble and dew points in two phase equilibrium.
Example 3 A feed of (1) 20 mol % ethane, (2) 20 mol %
isobutane, (3) 20 mol % n-pentane and (4) 40 mol % n-hexane
is ‡ash vaporized at T = 100 C and P = 600kP a. What mole
fraction of the feed is vaporized?
Step 0: Check

11
 The K-values taken from the DePriester Chart for the
four components (in increasing molecular weight are Ki =
[13.0, 2.75, 0.95, 0.45] .
The feed composition is Zi = [0.2, 0.2, 0.2, 0.4]
a) f (0) = 2.52 (so we’re above the bubble point).
b) f (1) = 0. 18753 (so we’re below the dew point).
Now with a starting guess of (1) = 0.5 we get from New-
ton’s method

(2) = 0.77924
(3) = 0. 7586
(4) = 0. 75863

So the third and fourth values of are about the same,

which means we’ve converged on a solution. The solution
makes sense, because it is between zero and one. So the frac-
tion of the feed that is vaporized is 75.863% (I don’t think
we read the K-values with that much accuracy, but I want to
give you a lot of digits so you can check your work).

Example 4 Let’s try another example where we are looking

at a hydrocarbon liquid feed at T = 95 C and P = 700kP a with
a composition of 40, 30, 20, and 10 mol percent of propane (1),
n-butane (2), n-pentane (3), and n-hexane (4) respectively.
What percentage of the feed enters as liquid? This problem
is similar to example 4.1 in Seader, but with the composition
changed. So

Zi = [0.4, 0.3, 0.2, 0.1]

and from the DePriester Chart:

Ki = [4.2, 1.75, 0.74, 0.34]

Step 0: Check:
a) f (0) = 1. 387 (so we are above the bubble point)
b) f (1) = 0. 16895 (so we are above the dew point!)
Check fails!
What would have happened if we would have tried Rachford-
Rice?
With (1) = 0.5, (2) = 1.1144 from Newton’s method. But
we know that V/F = is bound by zero and one so this is an
unphysical value of the vapor ‡ow rate. The answer is then
that all the feed is in vapor form.

12
Example 5 Now if we use the composition of the feed given
in the text Zi = [0.1, 0.2, 0.3, 0.4] where the least volatile com-
ponents are more prominent in the feed, then we should get
the same answer as the text. For example,

(1) = 0.5, (2) = 9. 8163  10 2 , (3) = 0. 12109,

(4) = 0. 12188, and (5) = 0. 12188

Which agrees with the answer in example 4.1. Notice that

by just changing the composition we go from something that
was completely vapor, to something that is mostly liquid (12.2
percent vapor) since:

= 0. 12188.

If we look at the checks in the above examples we notice

that there is maybe a better way to choose our initial value
for V/F. In the …rst example we noticed that a) f (0) = 2.52
and f (1) = 0. 18753. This means that a value for the root
close to one would be a good choice. In fact, after doing the
procedure we found that the root was at 0.76. In the third
example, which is example 4.1 we note that the check gives
f (0) = 0.128, and f (1) = 0.720. Since 0.128 is closer to zero,
the root is probably closer to zero than to one. In fact it is
0.122.

Bubble and Dew Points Calculations

If the speci…cations that you are given for your single state
equilibrium separations process are not T and P, but say
V/F = 0, and T or P (which is a bubble point temperature
or pressure, respectively) or V/F = 1, and T or P (which is a
dew point temperature or pressure, respectively) we must use
a procedure different from Rachford-and Rice. Let’s look at
a Bubble Point Temperature Calculation.
Bubble-Point Calculation
At the bubble point V/F = 0 and from f (0) = 0 we can
P
C P
C
…nd that Ki Zi = 1. If the temperature gives Ki Zi > 1
i=1 i=1
then the K-values are too large and we need to reduce the tem-
perature. The following procedure is a simple way of …nding
the bubble-point temperature.
Step 1: Guess the bubble point temperature
Step 2: Determine the K-values from a chart, equation or
plot of K-values versus temperature.

13
 P
C
Step 3: If Ki Zi = 1 then you are at the bubble point
i=1
temperature. Otherwise determine a new T.
Step 4: New Ki (choose a component) is given by
( n)
K
Ki n
( +1)
= P
C
i

Ki Zi
i=1

Step 5: For a DePriester chart draw a line through pres-

sure P and the one new K-value. For other sources of K-Value
data obtain the new K-values at the new temperature. Use
the intercepts of the isotherm with the other K-values to get
new Ki s and with the T axis to get a new T. Go to Step 3
and repeat until it converges.
Let’s …nd the dew point temperature for example 4.1.

Example 6 A hydrocarbon liquid feed with V/F = 1, and

P = 700kP a with a composition of 10, 20, 30, and 40 mol
percent of propane (1), n-butane (2), n-pentane (3), and n-
hexane (4) respectively is in a ‡ash drum.
Step 1: Guess T = 200C (we know that at T = 95 most of
the mixture was liquid, so we must have a smaller T, but let’s
guess way high so we can see how well this works).
Step 2. At this temperature and pressure we read off K =
[9.8, 5.7, 3.1, 1.6] from a DePriester chart.
P
C
Step 3: Ki Zi = 3. 69
i=1
Step 4: New K-value for propane is 9.8/3.69 = 2. 6558
Step 5: K = [2.66, 0.9, 0.35, 0.15]
P
4
Step 3a: Ki Zi = 0. 61
i=1
Step 4a: New K-value for propane is 2.66/0.61 = 4. 3607
Step 5a: K = [4.36, 1.8, 0.8, 0.37]
P
C
Step 3b: Ki Zi = 1. 184
i=1
Step 4b: New K-value for propane is 4.36/1.184 = 3. 6824
Step 5b: K = [3.68, 1.4, 0.59, 0.27]
P
C
Step 3c: Ki Zi = 0.933
i=1
Step 4c: New K-value for propane is 3.68/0.933 = 3. 9443
Step5c: K = [3.94, 1.4, 0.59, 0.27]
P
C
Step 3d: Ki Zi = 0.959
i=1
Step 4d: New K-value for propane is 3.68/0.959 = 3. 8373
At this K-value for propane we get an intercept with the T
axis of about 87C.

14
 If we are given V/F = 0 and T, but not P , then we can
use the same process to …nd the pressure.

In the case of the dew point, the process is slightly differ-

ent.

Dew-Point Calculation
At the dew point V/F = 1 and from f (1) = 0 we can …nd
P
C
Zi
that Ki
= 1. If a certain pressure and temperature gives
i=1
P
C
Zi
Ki
> 1 then the K-values are too small and we need to
i=1
reduce the temperature to reduce the amount of the mixture
that is in the vapor. The following procedure is a simple way
of …nding the dew-point temperature.
Step 1: Guess the dew point temperature.
Step 2: Determine the K-values from the K-value data
source.
PC
Zi
Step 3: If Ki
= 1 then you are at the dew point tem-
i=1
perature. Otherwise determine a new T
Step 4: New Ki (choose a component) is given by
( n)
( n+1) K
Ki = P
C
i
Zi
Ki
i=1

Step 5: Draw a line through pressure P and the one new

K-value. Use the intercepts with the other K-values to get
new Ki s and with the T axis to get a new T. Go to Step 3.
Seader gives an example of determining the dew point pres-
sure using the method of false position, which is similar to
Newton’s method.

Adiabatic Flash
If Q = 0, and P is speci…ed, you have an adiabatic ‡ash
operation. In this case we can combine the energy balance
and overall mass balance to derive the equation
hv +(1 ) hL hF
g (T ) = 1000
=0

To …nd T for the adiabatic ‡ash operation you …rst guess

a temperature, use Rachford-Rice to determine and then
plot g(T ). This is done for several temperatures until a plot
of g(T ) versus T can be interpolated to …nd the T at which
g(T ) goes through zero.

15
 The problem involves using Newton’s method to …nd V/F
for each T so it is like n Rachford-Rice procedures with the
added difficulty that a) an interpolation must be done to …nd
the root of g(T ), and b) the enthalpies are functions of T, P
and composition which makes plotting g(T ) fairly tedious.

Ternary
Liquid-Liquid In a ternary system you have three components. This
means you have two independent composition variables (since
Systems
the third is determined from the other two). Thus, plotting
the phase diagram is difficult because we can not plot the
phase diagram with T, and Xi in two dimensions as we usu-
ally did for binary phase diagrams. What is usually done is to
plot the phase diagram at a …xed T and P (remember the bi-
nary phase diagrams are usually being plotted at constant P
to make a plot simple in two dimensions) with only the overall
composition varying. The usual plot is done on an equilateral
triangle. An apex represents a pure component. The distance
from a composition point to a side represents the amount of
the component opposite that side in the mixture. That is the
mole or mass fraction of a component is the shortest distance
from the point in this triangle to the side opposite the apex
representing that component.
If there is a phase equilibrium in the system, it will be
represented by a curve on the triangle. Two phases in equi-
librium at a given T and P are connected by tie lines where
the intercepts of the tie lines with the phase boundary gives
the compositions of the two phases in equilibrium. The lever
rule gives the relative amounts of each phase connected by a
tie line.

Example 7 Using …gure 4.15 in Seader consider the mixture

of 40 grams of a mixture of 45 weight percent ethylene glycol
and 55 weight % water is contacted with 60 grams of Furfural
at 25 C and 101kPa.
The feed is then made up of:
a) (0.45) (40g ) = 18g of ethylene glycol, and
b) (0.55) (40g ) = 22g of water.
The solvent S is of course 60g of pure furfural.
Step 1: The solvent point S is at the furfural apex. The
feed point F is at the 55 percent water point on the water-
ethylene glycol binary base. (the same as in example 4.7).
Step 2: The mixing point M is de…ned as M = F + S =
E + R where F and S are the compositions of the two input

16
streams and E and R are the compositions of the two product
streams of the two liquids E and R in equilibrium. Note that
the lines joining E and R and F and S must intersect at point
M and that the line joining E and R is a tie line.
Step 3: The composition of point M is just the mass frac-
tions of the mixture. Since we have 100 grams total, M is just
18% ethylene glycol, 22% water, and 60% furfural.
Applying the lever rule to the point M we can determine
the relative amounts of each input stream.
( S) ( M)
F SM Wc Wc
S
= MF
= ( M) ( C)
Wc Wc

In this case:
F 100 60
S
= 60 0
= 0.66

So the feed stream is 66 percent by weight of the solvent

stream. Check this with the given masses above.
Step 4: The point M lies in the two phase region so there
is a phase separation into the extract phase and the raffinate
phase. The extract (with most of the solvent) is the intersec-
tion of the tie line with the phase boundary nearest the solvent
(furfural) apex. The point E is at a composition of 5 % water,
25 % ethylene glycol and 75% furfural. The tie line intersects
the raffinate side of the boundary at R with a composition of
85 % water, 8 % ethylene glycol and 7% furfural.
Step 5: Using the lever rule we see that the total mass of
R is given by

R = M ME
ER

Often it is convenient to use the masses of one of the compo-

nents (usually the solvent) to do the lever rule calculations.
In this case that is: R = 100g 7575gg 60
7g
g
= 22. 059g
Measuring with a ruler you can obtain 10015/67 = 0. 22388
We can use E = M MR ER
as a check.
60g 7g
E = 100g 75g 7g = 77. 941g which is just 100g 22.059g so
it checks.
Step 6: To obtain the solvent free extract composition we
take the extract composition, remove all of the solvent, and
then calculate the mass fractions. This turns out to be:
Ei =5 % water, 25 % ethylene glycol and 75% furfural. Re-
moving the furfural leaves the 5 percent water and 25 percent
ethylene glycol. These percentages must each be multiplied by
100
EA +EB

17
which in this case is 3.33. The solvent free composition H
is then 16.67 percent water and 83.33 percent ethylene glycol.
This is also given by the intercept of the tangent to the phase
boundary at E with the solvent free side of the composition
triangle.

Consider the ternary liquid-liquid system where a feed of

components A and B enter a separator as a mixture. A second
input stream is introduced which is a solvent C. This solvent
is a solvent of B, but does not dissolve A is introduced to re-
move B from the carrier component A. The product streams
are of two immiscible liquids (because A and C do not mix),
one solvent rich called the extract, and one carrier rich called
the raffinate.
The ‡ow rate of solvent C in the solvent stream is S, and
the feed rate of the carrier is FA .
The mass ratio of component B in the feed streams is
(F ) ( R) ( E)
XB = M B
MA
. Similarly, XB = M MA
B
, and XB = M MC
B
for the
raffinate and extract streams, respectively. Now because A
and C do not mix we know that the ‡ow rate of solvent in
the extract stream is equal to the ‡ow rate of solvent in the
solvent stream. Likewise, the ‡ow rate of carrier in the feed
stream is equal to the ‡ow rate of carrier in the raffinate.
As in all separator processes, we analyze the system by
combining mass balances, energy balances and thermodynam-
ics. In this particular case energy doesn’t play a role.
Mass Balance on Solute B
(F ) ( E) ( R)
XB FA = XB S + XB FA

The term on the left is the ‡ow of B into the drum in the feed.
The …rst term on the right is the ‡ow of B out of the drum in
the extract stream and the second term is the ‡ow of B out of
the drum in the raffinate. Notice that the whole process is one
of moving B from the carrier stream to the solvent stream.
Equilibrium Condition

= KB0 XB
( E) ( R)
XB

where KB0 is a distribution ratio (the same thing as a K-value

where instead of partitioning species between vapor and liq-
uid, we partition component between two liquids).
Combining the mass balance and equilibrium condition
gives:
(F )
(F) XB FA
XB = 0 S
FA +KB

18
 and de…ning the extraction factor EB by

KB0 S
EB = FA

gives
( R)
XB 1
(F ) = 1+EB
XB

The larger EB , then the less B is left in the raffinate. So a

larger distribution factor, or solvent ‡ow rate or smaller ‡ow
rate of carrier leads to more extraction.

Multicomponent
Liquid-Liquid Multicomponent liquid-liquid phase equilibria are not eas-
ily represented by a phase diagram. The easiest way of com-
Systems
puting the equilibrium phase compositions is to use the ac-
tivity coefficient relations (as discussed in chapter 2). The
main complication is really that the K-values are typically
composition dependent and so no simple diagrams, such as
DePriester charts or …gure 2.8 are available for liquid-liquid
systems usually. However, if you do have activity coefficient
relations you can use the Rachford-Rice algorithm, with a few
modi…cations. The differences are that:
a) the feed is now a feed plus a solvent,
b) the vapor phase becomes a liquid, say the extract,
c) the liquid becomes the raffinate liquid,
d) the feed composition is now the combined composition
of the feed and solvent streams,
e) the K-value now becomes a distribution coefficient, which
is analogous to the K-value, in that it is a proportionality
constant between the equilibrium compositions of two liquid
phases in equilibrium.
One complication is that the compositions of the two phases
must be estimated to initiate the procedure. Rachford-Rice is
then followed (using the activity relations for the K-values).
New compositions are determined and used to initiate a sec-
ond iteration through the procedure. The procedure is fol-
lowed until the compositions no longer change. This process
involves an inner and outer loop iterative procedure and is
impractical for solution except for by computer.
The following sections will be skipped, although they are
primarily analyzed with methods similar to the ones we have
been discussing above.

19
Solid-Liquid
Systems Leaching
Crystallization
Liquid Adsorption

Gas-Liquid
Systems
Gas-Solid Sys-
tems Sublimation and Desublimation
Gas Adsorption

Multiphase
Systems Approximate Method for Vapor-Liquid-Solid System
Approximate Method for Vapor-Liquid-Liquid System
Rigorous Method for Vapor-Liquid-Liquid System

20