You are on page 1of 26

PROJECT REPORT # 3 (CHE 510)

Instructor: Professor A. Linninger

Calculation of Minimum Reflux for Quaternary Mixtures


using Zero –Volume Method

Submitted by:
Jyothsna Gunamgari

Date: 05-03-02
Table of Contents:

1.Abstract.............................................................................................................................3
2.Introduction.....................................................................................................................4
Mathematical Model for a Column ...............................................................................4
Application for a ternary mixture...................................................................................4
3Derivation of Zero Volume Method.................................................................................8
4.Comparision to Underwood's Method...........................................................................10
5.Procedure to calculate the minimum reflux.................................................................11
Flowsheet for Zero Volume Method.............................................................................12
5.Case Study and Validations...........................................................................................13
6. Limitations.....................................................................................................................20
7. Observations and Conclusions.....................................................................................21
8.Appendix.........................................................................................................................22
Abstract

The project attempts to design distillation systems for ideal multicomponent mixtures
using an iterative procedure. The basic design of the project deals with analysis of the
fixed points of the given system. The procedure is explained using the mathematical
modeling involved and an example of a quaternary mixture distillation column, which
has been designed using this mathematical model.

For a ternary system, it is well known that at minimum reflux, the feed point and the
saddle point of the rectifying profile are in straight line with the stable node of the
stripping profile (incase of a direct split). The geometric alignment has been explored on
the basis that the heaviest component in the distillate is present in traces, say to the order
of 10-5. This paves way for the alignment of the points. The analysis of the quaternary
mixture has no resemblance to the ternary mixture. Many attempts were made to extend
the three-component theory for multicomponent that has yielded no result. However, the
analysis of the four-component mixture [4] has shown that the four-component can be
generalized to n-component system.

In case of quaternary mixture, at minimum reflux the stripping profile pinches onto the
rectifying manifold (which is formed by the fixed points of the rectifying plane, incase of
a direct split). This rectifying manifold varies from being a curved surface in case of non-
ideal mixtures to a planar surface for ideal mixtures. Thus for ideal mixtures, it is called
the rectifying plane. All the rectifying profiles lie in the same rectifying plane for a fixed
reflux ration and for slight variations of the heavy component in the distillate.

A brief look on the degrees of freedom analysis shows that the number of degrees of
freedom for a distillation column is 4. Incase of a ternary mixture, the search for the
minimum reflux, when a boundary-value design method is employed, restricted to
varying the reflux ratio until an intersection between the profiles is found. The four
degrees of freedom in this case would be a reflux ratio ‘r’ and two of the distillate
compositions followed by a bottoms composition. But for a quaternary mixture, the
specification of the reflux ratio ‘r’ and two distillate compositions along with the bottoms
composition leaves another distillate composition (say xD, 3) as an unknown. So in the
case of quaternary mixture, a bivariate search is involved for finding the minimum reflux
ratio and xD, 3 at the same time.

Sensitivity analysis of the rectifying plane [3] shows that the variation of the heaviest
component in the order of 10-8 does not alter the location of the rectifying plane whereas
it alters the profile in the plane. This helps us to make the search for minimum reflux a
univariate problem, in which the intersection of the stripping profile with the rectifying
plane is made a minimum reflux condition followed by a search for the exact intersection
of the stripping and rectifying profiles by varying the heaviest component in trace
amounts. The geometric condition of the intersection of the stripping profile and the
rectifying plane can be explained by an algebraic equation, which is called the zero-
volume method [4]. The zero volume method says that the volume enclosed between the
vectors formed by the fixed points and the feed point is zero at minimum reflux. So
calculating the volume for different reflux ratios and then find the zero crossing makes
the univariate search.

In the present project, we first find the fixed points and then solve for the minimum
reflux by the zero-volume method. Then a search for intersection of the profiles is made
by varying the heaviest component of the distillate in trace amounts to the order of 10 -8.
Case study from Julka and Doherty [4], is taken and verified.

Introduction

The regular practice to design a distillation column is to guess an initial design and
improve it by successive simulations. For ideal mixtures, good methods are available for
estimating the initial guess, such as the well-known Fenske-Underwood-Gilliland
method. For nonideal mixtures, there are no robust methods for estimating minimum
reflux or the number of stages in each column section.

This project describes a minimum reflux calculation method for ideal, multicomponent
distillation for a direct split. The method is based on the geometric alignment of the
saddle pinches in the rectifying profile and the node in the stripping profile along with the
feed point for a direct split and vice versa for an indirect split.

We first start with the explanation of the mathematical model for a column followed by
method to evaluate the fixed points occurring in a distillation column. The alignment of
these points at the minimum reflux condition gives us an algebraic equation. The
minimum reflux calculation using the zero-volume method is then described.

Mathematical Model for a Column

The model explains the separation of a c-component mixture in a single, conventional


distillation column, with one feed stream and two product streams .The assumptions
involved in the model are as follows
1) All stages are theoretical stages: the liquid and vapor leaving a stage are in
equilibrium.
2) The process is isobaric at a known pressure.
3) There is constant molar overflow in each section of the column.
4) Both of the products are saturated liquids.
The overall material balance (lever rule) gives c-2 constraints on the product compositions:
(1)
, i=2,… c-1

And the overall energy balance is:


(2)
The compositions of the liquid and vapor phases in both sections of the column are
determined by the following set of equations. The material balance and equilibrium
equations for the rectifying profile are:

(3)

(4)
j = 0, 1, 2,…,m.

j = 0, 1, 2,…,m. (5)

Similar expressions are obtained for stripping profile:

(6)

(7)
j = 0, 1, 2,…,n.

j = 0, 1, 2,…,n. (8)

The condition for a feasible column is that the profiles match at the feed stage. If stages m and n are one
above and one below the feed stage, respectively, then:
(9)

Application for a Quaternary mixture

The profiles for a quaternary mixture start at the product compositions and end at the
stable nodes; with additional (saddle) fixed points in each profile given by Eqs (11) and
(12). For r<rmin, the profiles do not intersect. At r = rmin, the stripping profile ends
somewhere on the rectifying profile. For constant volatility mixtures and a saturated
liquid feed, the stripping node, the rectifying saddle and feed composition (the points x 1,s,
x2,r, x3,r and xF ) are aligned at minimum reflux. Generally, the curve joining these three
points is nearly linear even for many highly nonideal systems. For other thermodynamic
states of the feed, the point xF should be replaced by a linear combination of xF and x1,s.

The four degrees of freedom chosen here are xD,1, xD,4, xB,1 and r. The feed stage
composition occurs at the intersection of the profiles. The rectifying profile starts at the
distillate composition and approach (c-2) saddles before ending at a stable node.
Therefore for a quaternary mixture, there would be two saddles approached by the
rectifying profile. The stripping profile starts at the bottoms composition and ends at the
stripping node. At minimum reflux, the stripping node is located on the rectifying
manifold. At a reflux ratio higher than the minimum, the stripping profile intersects the
rectifying manifold transversely. For an indirect split, where the heaviest component is
removed in high purity, the situation is symmetrically reversed.
The fixed points in the profiles can be used to find a close approximation to the minimum
reflux (or reboil) ratio without the calculation of the intermediate profiles The fixed
points are found as the roots of the following equations:

(10)
where F(x) for the rectifying section is:
(11)

and for the stripping section:


(12)

At total reflux and reboil, the above two equations respectively become
(13)
and
(14)

Thus, at total reflux (reboil), the fixed points of Eq. 10 are located at all of the pure
components and azeotropes.

Julka and Doherty [4], proposed an algebraic method for the calculation of minimum
reflux ratio. For example, in a direct split one can construct the following set of vectors
originating from x3,r, where superscript 3,r signifies fixed point number 3 in the rectifying
section as shown in Figure 1:

Figure 1 Vectors joining the fixed points.

(15)

(16)
(17)
similarly
(18)
The minimum reflux condition is equivalent to
(19)

The determinant can be physically interpreted as follows:


 For a binary mixture it is the length of e1
 For a ternary mixture it is the area spanned between e1 and e2.
 For a quaternary mixture it is the volume spanned by e1, e2 and e3.
 For mixtures with five or more components, it is a generalized “fixed point volume”
defined by equation 19.

The area spanned by the quaternary mixture is given by the determinant of the vectors
described in Eq(19) as shown in Figure 2. It can be observed that the all the vectors are
coplanar at minimum reflux and thus results in zero volume spanned by these vectors.

xs e1
e2

e3

e4

Figure 2 At minimum reflux, the vectors are coplanar and give Zero Volume

The method is described in the appendix.

The method was originally developed by Julka and Doherty,[4] for quaternary mixtures
and was called the “zero volume” method. The method is exact for constant volatility
mixtures; in fact, Underwood’s method can be obtained as a limiting case. For nonideal
mixtures, the method is approximate though well within the accuracy of the other
approximations made for design purposes.

Derivation of the Zero Volume Method:


The proof for the solution to fixed points and the volume covered by the vectors e 1, e2 and
e3 is explained as follows:

For a three component mixture with direct splits the smallest reflux ratio, r for which the
stripping profile intersects the rectifying manifold is defined as the condition of minimum
reflux ratio.

For mixtures with four or more components, the dimension of the search over product
compositions (i.e. C - 3) while designing a feasible column, becomes >= 1. Thus the
boundary value design procedure becomes a multivariate search in both the reflux ratio,
and the C - 3 product compositions. Thus, for a four-component system the boundary
value design procedure is a two-variable search in the reflux ratio, r, and one of the
terminal compositions. For a direct split, i.e. a l(2)/(1)234 split, one possible set of target
composition design variables is xD,1, xD,4 and xB,1, with r and xD,3 corresponding to the
search variables. For such a problem formulation, the minimum reflux ratio, rmin, is thus
defined as “the smallest possible value of the reflux ratio for which a feasible set of (r,
xD,3) exists”.

The analysis of the geometry of the rectifying map for a fixed value of x D,1 and a
negligibly small value of xD,4, the location and orientation of the rectifying plane in the
composition simplex is determined by the reflux ratio, while the value of x D,3 determines
the exact shape and location of the rectifying profile within this plane. This geometric
property of the composition profiles can be used to separate the reflux ratio from the
terminal composition in the search algorithm, thereby transforming the minimum-reflux
problem into two separate one-variable searches in r, and xD,3.

From figures (5) to (9), the locus of the rectifying composition profiles (as x D,3 is varied)
is observed to sweep out the entire rectifying manifold (rectifying profiles for different
values of xD,3). A special property of this surface is that, for any point that lies on this
surface, there exists a unique value of x D,3 such that the rectifying profile passes through
this point. This implies that if, for a specified reflux ratio, the stripping profile intersects
the rectifying plane, then it is always possible to find a unique value of x D,3 such that the
column is feasible, i.e. the two composition profiles intersect. In view of this, the general
definition of minimum reflux can be modified for direct splits to “the smallest reflux
ratio, r for which the stripping profile intersects the rectifying manifold”. This is
explained in detail in the validation part of the project.

The secant to the rectifying profile at the stripping node is a straight line through the
point , where
(20)

This is true regardless of whether the mixture is ideal, constant-volatility, nonideal or


azeotropic. The secant criterion, coupled with the linearity of the separating manifold
requires that, in addition to the stripping node, x 1,s , the point must also lie in the
rectifying plane. For the special case of a saturated liquid feed (q=1), the point is the
same as the feed composition, xF. Hence, at r = rmin, the feed composition as well as the
stripping node, xs lie in the rectifying plane as shown in the figure (3).

xf xs

Figure 3 At minimum reflux, the stripping node and feed point lie on rectifying plane

As the stripping node, the saddle in rectifying profile and feed point, all lie in the same
plane, choosing one of these points as a origin (e.g. x3,r), we can define the following vectors
which were explained in Figure 1:
(21)

(22)

(23)

(24)

At minimum reflux, the stripping profile pinches on the rectifying plane. Further, the
secant criterion implies. Thus, the minimum reflux ratio can be uniquely determined by
requiring that any three of these four vectors be coplanar which can be observed in figure
(2).

An equivalent criterion for minimum reflux is that the volume enclosed by the vectors is
zero. The oriented volume of n vectors in n-space is the determinant of the matrix of the n
column vectors [see Appendix]. Using this definition, the minimum reflux ratio is that
value of the reflux ratio, r, which makes det[e1, e2, e3] = 0 for a quaternary mixture. It has
been observed before that the oriented volume of these vectors is a well-behaved,
monotonic function with respect to r. Hence any of the usual single variable root finding
techniques (e.g. Bisection method, Newton-Raphson, secant method etc.) can be used to
solve this equation for rmin. Results using this approach are shown for a case study and is
validated.
Rmin using Underwood Method and comparison to Zero Vol. Method:

The value of minimum can be calculated by using the Underwood’s equations for a
multi-component mixture with constant volatilities.
In case of a ternary mixture, direct split, the equations for the rectifying section in this
method [1] are given by:

(25)

and for the reflux ratio as


(26)

Similarly for the stripping section


(27)

and for the reboil ratio


(28)

For example for a three component mixture the third order equation produces three roots
for the values of  for the rectifying section in a peculiar order as:
(29)
and for the stripping section as
(30)

According to Underwood, one of the root of Eq(23) coincides with a root for the Eq(25)
for the system operating at minimum reflux.

Assuming all the roots are equal and denoting them by  and writing Eq(23) and Eq(25)
interms of , an equation interms of the feed can be obtained as shown:
(31)

For the case of three component saturated liquid feed, solving the above equation results
in two roots which lie between the volatilities of the components. For a feed with quality
not equal to unity, Eq(28) results in three roots for  .

Hence for minimum reflux conditions, substituting each value of  in the equation for the
reflux ratio (Eq 24)
(32)
(33)

The other equation, which relates the compositions, is


(34)

Finally, Eqs (30) to (32) can be solved for the rmin and two unknown distillate
compositions

In Underwood’s approach the equation in terms of feed composition and a root of Eq(24)
can be written as
(35)

The above equation is just one linear equation in two unknowns in the composition space,
the location and orientation of which are a function of only the distillate composition and
the reflux ratio. Underwood showed that this equation has two distinct solutions x 1 and
x2, which are the limiting compositions when 3 and 2 are the heavy keys, respectively.
Comparing to the Zero Volume Method for ternary mixture, these two compositions refer
to the fixed points (saddle point and rectifying node). Hence the plane defined by Eq(33)
is identical to the rectifying profile.

The criteria that, at minimum reflux, Eq(14) be satisfied can be geometrically interpreted
as requiring the stripping node to lie in the rectifying plane. This geometry is identical to
one of the special geometries associated with the zero-volume framework

Procedure to Calculate the Minimum Reflux:

For the quaternary case, the design variables are taken as xD,1, xD,4 and xB,1, with r and xD,3
corresponding to the search variables. As explained in the section describing the
derivation of zero volume method, for such a problem formulation, the minimum reflux
ratio, rmin, is thus defined as “the smallest possible value of the reflux ratio for which a
feasible set of (r, xD,3) exists”.

An important geometric property of the distillation map is that for direct splits, the
rectifying profile is extremely sensitive to the amount of the heavy components in
distillate (xD,3) but the stripping profile is unaffected by this change in composition.
Hence variations in xD,3 of the order of 10-8 mole fraction can cause the rectifying profile
to move from one side of the composition simplex to the other, while having no effect on
the stripping profile. This has been validated in this project and explained using Fig (5) to
(9).

Thus the procedure involves two steps in an iterative manner as follows:


For the calculation of the fixed points occurring in the profiles, a reflux ratio has to be
assumed (See Eq 11). For the calculation of the determinant given by equation 19, the
fixed points have to be known. Hence, the steps involved are:
1. Assume a reflux ratio at a value of xD,3 = 6.0e-9 and calculate the fixed points.
2. Calculate the determinant (volume) spanned by the fixed points, using the obtained
set of fixed points.

The method is explained in detail using the flowsheet in the section. A plot between the
fixed volume and the reflux ratio is made and the zero crossing would give us the
minimum reflux value.

Flowsheet for Zero-Volume Method:

INPUT: Feed Composition (F, xFi,q),column pressure and xD,1, xD,4 and xB,1
Assume a very small value for x D3. From Component balances calculate the
bottoms unknowns D, B, xB,2, xB,4, xD,2 and xD,4

Assume a small value and a very large for the reflux ratio (r 10, r20) in the range

For both the cases, calculate s from Energy Balances and thus find the fixed
points in the rectifying section and stripping nodes using Eqs (11) and (12).
Calculate the area spanned by the vectors formed by these nodes by
calculating their determinant (f(r)).

NO

YES
Is (f(r1i)-f(r2i))2<12 and (r1- r2)2<22 STOP

NO r1i+1=( r1i+ r2i)/2


Is (f(r1 )*f((r1 + r2 )/2)) < 0
i i i
r2i+1= r2i

YES

r1i+1= r1i
r2i+1= =( r1i+ r2i)/2

Case Study and Validation:

A quaternanry mixture of constant volatilities 8.0, 4.0, 2.0 and 1.0 with a feed
composition of 0.25, 0.25, 0.25 and 0.25 respectively is used. The specs mentioned for a
direct split to satisfy the degrees of freedom are:
1. The composition of the most-volatile component in the Distillate, xD, 1 = 0.95
2. The composition of the least-volatile component in the Distillate, xD, 4 = 5.0*10-17
3. The composition of the most-volatile component in the Bottoms, xB, 1 = 1.0*10-3
4. Assume a small value for xD,3 say 6.0*10-9
5. The reflux ratio is varied using bisection method as explained in the flowsheet for
each run.

A part of the results obtained are presented here for few cases to explain the determinant calculation:
r x1,s x2,r x3,r e1 e2 e3 Det

1.9 0.271606 0.166667 0.455727 -0.205727 -0.184121 -0.28906 -3.21135e-3


0.315489 0.026316 0.544278 -0.294278 -0.228789 -0.517962
0.222294 0.807018 0.0 0.25 0.222294 0.807018
0.190611 0.0 0.0 0.25 0.190611 0.0
2.0 0.284147 0.158333 0.436196 -0.186196 -0.152049 -0.277863 -3.94478e-4
0.310050 0.025 0.563807 -0.313807 -0.253757 -0.538807
0.218471 0.816667 0.0 0.25 0.218471 0.816667
0.187332 0.0 0.0 0.25 0.187332 0.0
2.1 0.296064 0.150794 0.418097 -0.168097 -0.122033 -0.267303 2.418444e-3
0.304966 0.023809 0.581904 -0.331904 -0.276938 -0.558095
0.214800 0.825397 0.0 0.25 0.214800 0.825393
0.184171 0.0 0.0 0.25 0.184171 0.0

The plot between the fixed volume, given by the determinant and the reflux ratio is made
to find the zero crossing which gives us the minimum reflux ratio. It is found out to be
2.014.

Volum e vs r

0.003

0.002

0.001

0
2 2.005 2.01 2.015 2.02 2.025 2.03 2.035 2.04 2.045 2.05
Volume

V olum e vs r

-0.001

-0.002

-0.003

-0.004
r
Figure 4 Zero crossing observed at r = 2.014

Validation:
The case study has been validated using tray by tray calculations for the compositions of
the components in rectifying and stripping profile.

As explained before for direct splits, the rectifying profile is extremely sensitive to the
amount of the heavy components in distillate (x D,3) but the stripping profile is unaffected
by this change in composition. Hence, rmin can be calculated by using a small initial guess
for xD,3 and iterativley solving. Once the r min is found, by fixing the r at r min, the value of
xD,3 is iteratively changed and solved for a feasible column.

The reason can be explained as follows: when the value of xD,3 is changed from 1.0 e-10
to 5.0 e-8 the locus of the rectifying profiles sweep out a plane in the composition
simplex.

For xD,3= 1.0 e-10

Figure 5 Rectifying profile for xD,3 = 1.0 e-10

For xD,3= 2.0 e-9


Figure 6 Rectifying profile for xD,3 = 2.0 e-9

For xD,3= 5.0 e-9

Figure 7 Rectifying profile for xD,3 = 5.0 e-9

For xD,3= 1.0 e-8


Figure 8 Rectifying profile for xD,3 = 1.0 e-8

For xD,3= 5.0 e-8

Figure 9 Rectifying profile for xD,3 = 5.0 e-8


Hence, for the values of xD,3 <= 1.0e-8 the rectifying profiles sweep out a plane in the
composition simplex. The location of fixed points is completely insensitive to changes in
the value of xD,3.

The three dimensional diagram for r < rmin i.e., r= 1.75. at a small value of x D,3 (6.0e-9) is
shown next in Fig9. It can be seen that the two profiles do not meet.

Figure 10 Rectifying and stripping profiles for r< rmin

At the same value of xD,3 (6.0e-9) , the rmin is found to be 2.015, when the profiles meet,
i.e., the stripping profile pinches the rectifying plane.

Figure 11 Rectifying and Stripping profiles for r= rmin


xf xs

Figure 12 At r= rmin, Stripping node pinches the Rectifying plane

It can be seen that though the stripping profile pinches the rectifying plane, it does not
intersect it for this value of xD,3 (6.0e-9).i.e, the stripping node is offset from the
rectifying profile. Hence, varying the value of xD,3 gives us the exact value of xD,3
required for the feasible split as 3.0 e-9

Figure 13 Rectifying and Stripping profiles at xD,3 = 3.0 e-9

It can also be seen that the fixed points are now coplanar.
Figure 14 Rectifying profile for xD,3 = 1.0 e-8

Limitations of the Method:

 Finding the fixed points for mixtures exhibiting tangent pinches is not so straight
forward . Methods like bifurcation, arc-length need to be applied for such cases [2].
 It is a method for calculating the approximate value of the minimum reflux ratio.
Hence, it is not completely reliable.
 For nonideal mixtures, the method is approximate though well within the accuracy of
the other approximations made for design purposes. For sloppy splits it gives an
approximate, though not accurate.
 For the case of quaternary mixture the boundary value design procedure is a two-
variable search in the reflux ratio, r, and one of the terminal compositions and for
higher components the number of search variables increases.

Observations and Conclusions:

 Fixed point tracking for the constant volatile zeotropic mixture has been successfully
achieved and the zero volume method has been used to calculate the rmin.
 The rmin obtained from zero volume method has been validated using tray by tray
composition profiles for the quaternary case study and a complete agreement has been
observed between the results from both the methods.
 From the plots of composition vs trays for each component at different x D,3 , it has
been observed that a small change in xD,3, affects the rectifying profile and the
stripping profile remains unaffected. Hence, rmin can be safely calculated by fixing the
value of xD,3 at a small value and at this rmin the value of xD,3 can be varied to find the
feasible column.

References:
1. A. A. Linninger, Lecture Notes on Che 510 “ Advanced Separation Processes”.
Spring, 2002.
2. Z. T. Fidkowski, M. F. Malone, and M. F. Doherty, “ Nonideal Multicomponent
Distillation: Use of Bifurcation Theory for Design”, AIChE Journal, Dec 1991, Vol.
37, No. 12, 1761- 1779.
3. V. Julka and M. F. Doherty, “Geometric Nonlinear Analysis of Multicomponent
Nonideal Distillation: A simple Computer –Aided Design Procedure”, Chemical
Engineering Science, 1993, Vol 48, 1367-1391.
4. V. Julka and M. F. Doherty, “Geometric Behavior And Minimum Flows for Nonideal
Multicomponent Distillation.”, Chemical Engineering Science, 1990, Vol 45, 1801-
1990.
Appendix

Appendix A:
Secant Condition at Minimum Reflux

Consider the mapping for the rectifying section:


(36)

Subtracting xmr from both the sides of this equation and rearranging yields

(37)

where,
(38)
is the backward difference operator. Geometrically the operator  can be viewed as the
secant to the invariant rectifying curve that passes through the points xI and xi-1.
Similarly the secant representation for the stripping map is:

(39)

Replacing the reflux ratio and reboil ratio by the absolute flowrates
(40)
and
(41)
As,
(42)
Adding Eq (39) and (40)
(43)
Consider a feasible column with M theoretical trays in the rectifying section and N
theoretical trays in the stripping section (i.e., the feed tray is the Mth tray down from the
condenser and Nth tray up from the reboiler).
For a feasible column, the rectifying and stripping profiles should intersect at the feed tray, and hence,
(44)
and,
(45)
Substituting Eq (42) and (43) into Eq(41) gives,
(46)
An overall energy balance around the feed plate yields
(47)
and
(48)

Thus Eq (44) can be rearranged to give


(49)
where
(50)

Hence for a feasible column operating at or above minimum reflux, the vector
lies in the plane spanned by the secant vectors and . At minimum reflux,
for a direct split, the stripping profile pinches on the rectifying profile, and therefore
and . For this special case, Eq (44) can be simplified as
(51)

Thus at minimum reflux, the secant to the rectifying profile at the feed plate,
(which reduces to the vector , passes through the point . Similarly, for an
indirect split at minimum reboil ratio:
(52)

In this analysis, no assumptions have been made about the vapor-liquid equilibrium.
Therefore, the results in this section are valid for all types of homogeneous mixtures:
ideal, constant relative volatility, nonideal and azeotropic.
Appendix B
Program for finding the rmin from zero volume method:
#include<stdio.h>
#include <stdlib.h>
#include<math.h>
#include "nr.h"
#include "nrutil.h"
#define NRANSI
#define N 4
#define float double

/* Quarternary Zeotropic Mixture with constant relative volatilities of 16.0, 8.0, 4.0 and 1.0*/

/* Program to find the fixed points for a given reflux ratio*/

/* Specs :
Feed : Equimolar Feed
Feed Quality, q = 1.0;
Reflux Ratio : 3.5
Distillate : 0.95, 0.05, 6e-9, 5e-17
Bottoms : 0.001

*/
double xF1 = 0.25, xF2 = 0.25, xF3 = 0.25, xF4 = 0.25, q =1.0;
double r = 1.8;
double xD1 = 0.95, xD2 = 0.05, xD3 = 6e-9, xD4 = 5e-17;
double xB1 = 0.001;
double alpha1 = 8.0, alpha2 = 4.0, alpha3 = 2.0, alpha4 = 1.0;

void recfix(int n,float x[],float f[])


{
double xB2, xB3,xB4, s;

xB2 = xF2 - (xF1-xB1)*(xF2-xD2)/(xF1-xD1);


s = (r + q)* ((xB1-xF1)/(xF1-xD1)) + q -1;
xB3 = xF3 - (xF1-xB1)*(xF3-xD3)/(xF1-xD1);
xB4 = xF4 - (xF1-xB1)*(xF4-xD4)/(xF1-xD1);

// printf("%5s %12.6f %12.6f %12.6f %12.6f \n","xD3,xB2,xB3,s",xD3,xB2,xB3,s);

/*normalizing the variables*/


x[1] = fabs(x[1]); x[2] = fabs(x[2]); x[3] = fabs(x[3]); x[4] = fabs(x[4]);

// x[1] = x[1]/(x[1]+x[2]+x[3]); x[2] = x[2]/(x[1]+x[2]+x[3]); x[3] =


x[3]/(x[1]+x[2]+x[3]);

f[1] = - alpha1*x[1]/(alpha1*x[1] + alpha2*x[2] + alpha3*x[3] + alpha4*x[4]) + r/(r+1.0)* x[1] +


1.0/(r+1.0)*xD1;

f[2] = - alpha2*x[2]/(alpha1*x[1] + alpha2*x[2] + alpha3*x[3] + alpha4*x[4]) + r/(r+1.0)* x[2] +


1.0/(r+1.0)*xD2;

f[3] = - alpha3*x[3]/(alpha1*x[1] + alpha2*x[2] + alpha3*x[3] + alpha4*x[4]) + r/(r+1.0)* x[3] +


1.0/(r+1.0)*xD3;

f[4] = - alpha4*x[4]/(alpha1*x[1] + alpha2*x[2] + alpha3*x[3] + alpha4*x[4]) + r/(r+1.0)* x[4] +


1.0/(r+1.0)*xD4;
}

void stripfix(int n,float x[],float f[])


{
double xB2, xB3, xB4, s;

s = (r + q)* ((xB1-xF1)/(xF1-xD1)) + q -1.0;


xB2 = xF2 - (xF1-xB1)*(xF2-xD2)/(xF1-xD1);
xB3 = xF3 - (xF1-xB1)*(xF3-xD3)/(xF1-xD1);
xB4 = xF4 - (xF1-xB1)*(xF4-xD4)/(xF1-xD1);

// printf("%5s %12.6f %12.6f %12.6f %12.6f \n","xD3,xB2,xB3,s",xD3,xB2,xB3,s);

/*normalizing the variables*/


x[1] = fabs(x[1]); x[2] = fabs(x[2]); x[3] = fabs(x[3]); x[4] = fabs(x[4]);

// x[1] = x[1]/(x[1]+x[2]+x[3]); x[2] = x[2]/(x[1]+x[2]+x[3]); x[3] =


x[3]/(x[1]+x[2]+x[3]);

f[1] = alpha1*x[1]/(alpha1*x[1] + alpha2*x[2] + alpha3*x[3] + alpha4*x[4]) - (s + 1.0)/s* x[1] +


1.0/(s)*xB1;

f[2] = alpha2*x[2]/(alpha1*x[1] + alpha2*x[2] + alpha3*x[3] + alpha4*x[4]) - (s + 1.0)/s* x[2] +


1.0/(s)*xB2;
f[3] = alpha3*x[3]/(alpha1*x[1] + alpha2*x[2] + alpha3*x[3] + alpha4*x[4]) - (s + 1.0)/s* x[3] +
1.0/(s)*xB3;

f[4] = alpha4*x[4]/(alpha1*x[1] + alpha2*x[2] + alpha3*x[3] + alpha4*x[4]) - (s + 1.0)/s* x[4] +


1.0/(s)*xB4;
}

int main(void)
{
int i,check, check1;
float *xr,*fr;
float *xs, *fs;

xr=vector(1,N);
fr=vector(1,N);
xs = vector(1,N);
fs = vector(1,N);
while (r <5.0 )
{
xr[1]=0.1e0; xr[2]=0.1e0; xr[3]= 0.7e0; xr[4] = 0.1;

newt(xr,N,&check,recfix);

recfix(N,xr,fr);

if (check) printf("Convergence problems.\n");


printf("\n %7s \t%12.6f \n","r", r);
printf("%7s %3s %12s\n","Index","xr","fr");

for (i=1;i<=N;i++) printf("%5d %12.6f %12.6f\n",i,xr[i],fr[i]);


printf("%f \n", xr[1] + xr[2] + xr[3] + xr[4]);

xs[1] = 0.3; xs[2] = 0.2; xs[3] = 0.2; xs[4] = 0.3;

newt(xs,N,&check1,stripfix);

stripfix(N,xs,fs);

if (check) printf("Convergence problems.\n");


printf("%7s %3s %12s\n","Index","xs","fs");
for (i=1;i<=N;i++) printf("%5d %12.6f %12.6f\n",i,xs[i],fs[i]);
printf("%f", xs[1] + xs[2] + xs[3] + xs[4]);
r = r + 0.05;
}

free_vector(fr,1,N);
free_vector(xr,1,N);
free_vector(xs,1,N);
free_vector(fs,1,N);
return 0;
}
#undef NRANSI

You might also like