Detailed Simulation Approach of a Multicomponent Distillation Column using

Inside Out Method: A Review

Shashwat Bansal and Sampatrao D Manjare*

Department of Chemical Engineering, BITS-Pilani, K.K. Birla, Goa Campus, Goa 403726, India

Abstract:

The paper reviews and presents the inside-out method, an equilibrium based method to simulate the
performance of a multistage distillation column for separating a multicomponent mixture. Different aspects of
the method have gradually been reported over the years but a unified approach to the algorithm is not yet
presented. This paper puts forward a detailed methodological approach to the method highlighting its various
aspects along with a flow chart giving a step by step guidance. The method is compared with another rigorous
method namely the bubble point method to understand the qualitative advantages it provides over this
algorithm.

Keywords: Inside-out method, modeling and simulation, rigorous distillation techniques, two-tier algorithm,
MESH equations, non-ideal distillation

1
 *
Corresponding Author e-mail: manjare@goa.bits-pilani.ac.in, Telephone no +91 8322580112, Fax No +91
8322557030

Introduction:

Distillation is by far the most widely used traditional separation technique in the petroleum, natural gas and
chemical industries. Hence the applications of multicomponent distillation are numerous. Detailed computer
modeling of distillation columns has become an integral part of the design and development process [1]. In
rigorous methods, no assumptions are made to simplify the model equations as in case of shortcut methods [2].
Hence the results obtained from these methods are detailed and help us gain an insight into the system. The
importance of mathematical modeling and simulation is supported by the fact that the final design of multistage,
multicomponent separation equipment requires rigorous determination of temperatures, pressures, stream flow
rates, stream compositions, and heat-transfer rates at each stage [3].
The determination of the above mentioned variables is done by solving material-balance, energy-balance,
summation relations and equilibrium relations for each stage. These equations are collectively known as the
MESH equations. MESH equations are usually a set of non-linear coupled algebraic equations that are written
for each stage of the distillation column and require advanced algorithms for its solution. The solutions for these
equations are tedious but once programmed on a high speed computer the results are obtained in a robust and
rapid manner.

Among the available advanced algorithms for the solution of the MESH equations, the Bubble Point method
suggested by Wang and Henke [4] is commonly used. However this method is recommended for feed(s) within
a narrow range of boiling points and nearly ideal mixtures within limited allowable specifications. Bubble point
method uses equation partitioning in conjunction with equation tearing for solving MESH equations. But this
method fails to converge in case of non-ideal mixtures or when there are large boiling point differences between
different components of feed mixture. The method was found to be limited in terms of allowable specifications.
The only possible specifications are reflux ratio, reboil ratio and product flow rate.

A study by Friday and Smith [5] showed that no one tearing technique could solve all types of problems.
However recent methods like Newton-Raphson [6] and inside-out [7-9] are robust and can solve most of the
problems including non-ideal feed mixtures with great amount of flexibility in terms of the allowable
specifications.

2
In view of above discussion it is clear that a method like Inside-Out has potential for simulating most of the
separation problems with great accuracy. The literature survey [7-21] reveals that the available information on
the detailed methodology of inside out method is very scanty. Therefore in this paper a detailed methodological
approach together with a review for Inside Out method is presented.

Inside Out method:

The inside out method or the two tier algorithm was first presented by Boston and Sullivan [7] and subsequently
developed over the years by Russell [8] and Jelinek [9] among others. The method was initially applicable only
to distillation with an ideal or nearly ideal hydrocarbon feed. Later its application areas were extended to
include any multistage multicomponent steady state liquid vapor separator. Trevino and Kisala [15] extended
the inside out method to absorbers by replacing the inside loop with a sum rates method. Boston and Shah [20]
further extended it to model and simulate three phase distillation and Venkatraman et.al [21] extended the
method to reactive distillation. Different application areas of the inside out method has been described in Seader
and Henley [3]. This method has become very popular because of its ability to solve a wide variety of columns
with large range of specifications to choose from. A detailed list of allowable specification was presented by
Naphtali and Sandholm [6] for Newton Raphson method and is also applicable for inside out method. The
specifications are presented as replacement of condenser and reboiler duties in table 1 and table 2 respectively.
The major advantage of inside out method is that it is insensitive to initial estimates. Most other rigorous
methods are sensitive to initial estimates [17]. The suitable choice of iteration variables in the inside out method
result in a relatively small number of equations and therefore requires less computer memory and computation
time. Convergence behavior of inside out method along with comparison of convergence time with Newton-
Raphson method has been discussed in detail in Simandl [16]. The analysis reveals that inside out has a much
faster convergence rate. Shan et. al [11] presents an efficient method to solve tridiagonal matrix containing off-
band elements making inside out method efficient in solving complex separation problems. Saeger and Bishnoi
[10] have simulated various non-ideal systems using UNIFAC group contribution method.

The primary objective of inside out method is to reduce computation time by solving MESH equations using
simple K-values and enthalpy models in an inside loop. An outside loop uses complex K-values and enthalpy
correlations to update parameters for simple K-values and enthalpy models of inside loop. Since the
computation of K-values and enthalpy is less tedious in the inside loop, the MESH equations converge rapidly
and the convergence is stable. For non-ideal systems a simple activity coefficient model accounts for

3
composition effects of K-values in the inside loop. The activity coefficient model is discarded for ideal or near
ideal feed mixtures.

The iteration variables for outer loop are relative volatility, parameters of simple K-value and enthalpy models
and parameters of activity coefficient model at each stage. Relative volatility changes little over range of stage
temperatures and hence greatly eases composition and temperature calculations in the inside loop [17].The
iteration variables of inner loop are related to stripping factor and liquid & vapor side stream withdrawal factors
[3]. Stripping factor serves as an excellent iteration variable because it combines the effect of temperature,
liquid flow rate and vapor flow rate at each stage and accounts for stable convergence.
To further understand the methodology of inside out method, a mathematical model of a single stage liquid
vapor separator at steady state has been presented here.

The Mathematical Model

The model presents a cascade of N counter current stages with the following assumptions
• All stages are equilibrium stages.
• Kinetic and potential energy changes in the system are ignored.
• There is no entrainment of liquid or trapping of bubble vapor in liquid.
Diagram of a single stage out of N stages in the cascade is given in Figure 1.

MESH Equations:

The MESH equation for inside out method, presented here uses volatility and stripping factors along with
enthalpy to model a steady state continuous liquid vapor separator.

The diagram of a cascade of N stages has been given in Figure 2 for a continuous steady state separator.
The following variables are defined to be used in the MESH equations

i) Relative volatility:

ii) Base component stripping factor:

iii) Withdrawal factor for liquid phase:

4
 iv) Withdrawal factor for vapor phase:

Material balance for each component at each stage (M-equations)

------ (1)

Phase equilibrium relations at each stage (E-equations)

---------------------------------------------------------------------------- (2)

Summations of component liquid and vapor flow (S-equations)

----------------------------------------------------------------------------------- (3)

----------------------------------------------------------------------------------- (4)

Energy/Enthalpy balance at each stage (H-equations)

----- (5)

Calculation of liquid and vapor enthalpies along with stripping factors require simple and complex
thermodynamic models as presented below.

Complex thermodynamic models

Values of for liquid and vapor enthalpy and K-values are found out from primitive variables like temperature,
pressure, liquid mole fraction and vapor mole fraction by substituting them into actual thermo physical models.
The thermo physical models constitute of any P-v-T equation of state like Soave Redlich Kwong (S-R-K) or
Peng Robinson. The thermo physical models also constitute of free energy models for liquid activity
coefficients like NRTL and Margules.
The thermo physical models are represented in the form:
5
Calculation of K-values
K-values are calculated using its activity coefficient form as given below.

--------------------------------------------------------------------------------------------- (6)

where is the liquid phase activity coefficient for that component. A suitable choice for the activity

coefficient model based on the feed components must be made before calculating activity coefficient.

is the pure species fugacity coefficient for liquid and is the partial fugacity coefficient for vapor

[22].
Calculation of enthalpy values

The enthalpy of a liquid or vapor stream is the sum of its ideal enthalpy and the departure that would exist for
the liquid or vapor.

Enthalpy for vapor phase can be calculated as follows:

--------------------------------------------------------------------------------------------- (7)

----------------------------------------------------------------------------------------- (8)

Enthalpy for liquid phase can be calculated as follows:

----------------------------------------------------------------------------------------------- (9)

---------------------------------------------------------------------------------------- (10)

6
Method to calculate and has been provided with the detailed methodology.

Simple K-values and Enthalpy models

Simple K-value model

The simple K-value model in terms of a reference/base component Kb follows an empirical Clausius-Clapeyron
form as given below [19].

. ----------------------------------------------------------------------------------------- (11)

The coefficients An and Bn are very weakly dependent on Tn and hence it makes them good iteration variable.
[14]

The values of An and Bn are calculated by calculating values of Kb at two temperatures, one above and one
below the stage temperature keeping the composition constant. To calculate values of Kb the following equation
is used

----------------------------------------------------------------------- (12)

where is a weighing factor for which value is given by the relation

-------------------------------------------------------------------------------- (13)

----------------------------------------------------------------------- (14)

The derivative expression in equation (14) can be found either analytically or using numerical techniques.

Approximate enthalpy model

7
The approximate enthalpy models are used to calculate departure of enthalpy from ideal values. The
expression for approximate enthalpy models is given below.

---------------------------------------------------------------- (15)
---------------------------------------------------------------- (16)

Where T* is a reference temperature. Since all four parameters , , , are in mass units, they are

insensitive to temperature and composition and are well behaved iteration variables [14].
Detailed methodology for inside out method

Section 1: Initialization

This section aims at providing reasonable estimates for liquid flow rate Ln, vapor flow rate Vn, liquid and vapor
mole fractions xm,n, ym,n and stage temperature Tn. To get initial estimates a step by step procedure is given
below.

1) Provide the required specifications of the distillation column viz number of theoretical stages, conditions of
all feeds, feed-stage locations, and column pressure profile.

2) Specify the stage locations for each side stream product withdrawals. Also specify stage locations of
heat exchangers.

3) Specifications for each product including side streams and each intermediate heat exchanger must
be provided.

4) Estimate each product-withdrawal rate if not specified earlier, and estimate value of Vn. Constant molar
overflow can be assumed for Vn determination. Subsequently find out the values of Ln from total material
balance equation as mentioned below.

---------------------------------------------- (17)

5) To find top stage temperature T1 and bottom stage temperature TN combine all feeds and calculate
bubble point and dew point temperature respectively. The intermediate stage temperatures Tn can be
found out by linear interpolation between T1 and TN.
The reference temperature for each stage T* to be used in equations (15-16) is set equal to the initial
temperature profile.

6) The stage composition can be estimated by solving component mass balances using tridiagonal
8
 matrix algorithm with all Kb, n = 1 and [8].

7) To calculate coefficients for Kb model, two Kb values for each stage at temperatures T1= (Tn + ∆T)
and T2= (Tn - ∆T) using equations (12-14) are calculated. The composition is held constant during
these calculations. Then the coefficients for Kb model can be calculated using equations (18) and
(19) given below.

---------------------------------------------------------------------------------- (18)

---------------------------------------------------------------------- (19)

Note: ∆T should be no more than 25 to 50 oF for first trial and should be decreased for successive
trials as the final temperature profile is approached [17].

8) For highly non-ideal feed, an activity coefficient model is added to account for the non-idealities
present. The activity coefficient model is given by

----------------------------------------------------------- (20)

The values of activity coefficients are fitted to liquid component mole fractions. The values of outer
loop variables and for each component are determined from actual activity coefficient
model at current stage temperature and composition.

is replaced by in material balance equation (1). Now the liquid component flow rates
obtained after solving by tridiagonal matrix algorithm will have non-ideality accounted for.

9) At each stage calculate coefficients cn, dn, en, fn for vapor enthalpy (Eq 15) and liquid enthalpy (Eq
16) departure models. To estimate these coefficients calculate enthalpy departure for liquid and
vapor at two temperatures T1 and T2, the procedure for which is given below.

Note: The approximate enthalpy departure functions have units of enthalpy per unit mass rather than
enthalpy per unit mole. Hence before calculating values of cn , dn , en , fn you need to convert the values
of liquid and vapor enthalpy departure obtained from complex thermodynamic models to enthalpy per
unit mass by dividing it with the molecular mass of the respective phase.

9
Molecular mass of vapor (MV) =
Molecular mass of liquid (ML) =

where Mm is the molecular mass of individual components and xm and ym and liquid and vapor mole
fractions respectively [14].

In some systems such as ammonia-water, enthalpy model gives too crude an approximation and is
therefore inappropriate. In such case approximate enthalpy models are not used and enthalpies of
inside loop are calculated directly from enthalpy correlations [17].

The method to calculate enthalpy departures at T1 and T2 is given below.

Enthalpy departure for liquid

Enthalpy departure for the liquid phase accounts for the enthalpy of vaporization and the effect of
pressure on both liquid and vapor phases. Of these two the enthalpy of vaporization dominates the
departure term and hence is the only one considered for this calculation [3].

The formula for enthalpy departure for liquid is given as [18]

---------------------------- (21)

∆HLn = Enthalpy departure for liquid / Enthalpy of vaporization, KJ/mol

R= Gas constant = 0.008314 KJ/mol K
Tc= Critical temperature, K
Tr= Reduced temperature, T/Tc
T= Temperature, K
ω= acentric factor

This method is applicable in the range 0.6<Tr<1.

After calculating enthalpy departures at T1 and T2 the following expressions are proposed to find
coefficients en and fn.

--------------------------------------------------- (22)

---------------------------------------------------- (23)
10
Enthalpy departure for vapor

The enthalpy departure function for vapor accounts for the effect of pressure. An equation of state
needs to be selected to account for non-idealities. Non-ideal thermodynamic property models include
P-v-T equations of state like Redlich Kwong(R-K), Soave Redlich Kwong(S-R-K) and Peng
Robinson. After a suitable choice of thermodynamic property model has been made, the following
equation is used to find enthalpy departure for vapor

------------------------------------------- (24)

The calculation for vapor phase enthalpy departure using R-K equation of state for example is
presented below.

The Redlich Kwong equation of state [2, 22] is defined as

Where

If this equation of state is to apply to a mixture then mixing rules must be used to average out
constants a and b. The recommended mixing rule for R-K equation of state is

Substituting the generalized equation of state in Redlich Kwong equation and

eliminating molar volume (v) we obtain the following equation

11
 Where

The equation obtained is cubic in Z and can be solved for three roots. Three real roots exist below the
critical temperature with the largest value applying to vapor which is denoted as Zv [3].

The final relationship obtained for enthalpy departure for vapor after substituting R-K equation of state in
the equation (24) is

----------------------------------------------------- (25)

Similarly the procedure can be repeated for other equations of state.

10) Calculate initial values of base stripping factor, withdrawal factor and relative volatilities.

Section 2: Inner loop calculations

This is the section where inner loop calculations begin. The purpose of inner loop calculations is to solve for
MESH equations using an empirical set of thermodynamic property models.

11) Start the inner loop calculation sequence as follows.

i) Solve component mass balance. To ensure a reasonable distribution of flow rates within the
column, scale the stripping factor using the formula given below. This is done only for first
iteration as initial estimates might be poor and may lead to zero or negative flow rate
distribution.

---------------------------------------------------------- (26)

where Sb is a scalar multiplier

The outcome of first step is estimation of each component liquid flow rate and subsequently
each component vapor flow rate is calculated using equation (2).

ii) Set inter stage and product flow rate equal to the sum of component flow using Eq (3-4)
mentioned below

12
 ------------------------------------------------------------ (3)

-------------------------------------------------------------- (4)

The stage liquid and vapor compositions of each component is found out by dividing
component flow rate with total flow rate of liquid and vapor respectively at each stage.

iii) Estimation of stage temperatures is done using Kb method by the formula mentioned below.

----------------------------------------------------------- (27)

------------------------------------------------------------------ (28)

First calculate new values of Kb n in the inner loop using equation (27) and then calculate
new stage temperatures using equation (28).

iv) Calculate enthalpies of all streams using the method given below.

Enthalpy calculation for vapor stream: The vapor stream enthalpy is calculated using the
below mentioned equation.

is the ideal vapor enthalpy which can be calculated using equation (8).

in equation (8) is the ideal vapor enthalpy of pure component. It is calculated by integrating
specific heat of pure components as a function of temperature between two temperature intervals.
Alternatively a polynomial relation of enthalpy with temperature can be used if available.

∆HV is calculated using the following formula. The expression will have units of enthalpy per
unit mole.

13
 ----------------------------------------------- (29)

Enthalpy calculation for liquid stream: The liquid stream enthalpy is calculated using the
below mentioned equation.

is the ideal vapor enthalpy using liquid composition. Enthalpy of vaporization is subtracted
from to obtain liquid enthalpy. The expression for in units of enthalpy per mole is given
by

-------------------------------------------------- (30)

12) Rearrange the energy balance at each stage into the following form as given by Simandl and Svrcek [16]

13) Select inner loop iteration variables for convergence as follows

i) Natural logarithms of base stripping factors (ln(Sb,n))
ii) Natural logarithms of side stream withdrawal (ln(Un/Ln) or ln(Wn/Ln)) for each side stream with a
specification on its purity or some other variable.

Jelinek [9] noted that the use of natural logarithms for iteration variables has a favorable effect on
stability and rate of convergence.

14) The condenser and reboiler duties are usually unknown and therefore the functions of the first and last
stage of energy balances (H1 and HN)) are replaced by discrepancy functions D1 and D2. List of possible
discrepancy functions are given in Table (1-2). Also the specification for each product side stream is
treated as an independent specification equation.

15) Normalize the discrepancy functions D1, Hn and D2. The procedure for normalization is given below.

14
 For a normalized energy balance equation at each stage, it should be written in terms of ratio of energy
entering the stage to energy leaving the stage [17]. The same principle is followed for specification
equations.

16) Calculate Jacobian of partial derivatives of D1, D2 and Hn with respect to inner loop iteration
variables of step 13, analytically or by perturbation of iteration variables and recalculation of
discrepancies from step 11 to 15.

A sample Jacobian matrix for a five stage distillation column with a partial condenser and no side
streams is presented in Figure 3.

The following calculations are done when the Jacobian is evaluated analytically

The analytical expression for derivative of D1 and D2 depends on choice of these specification
functions.

In case of side streams and composition dependant K values & enthalpies, the analytical calculation
of derivatives becomes complicated [17]. Hence the alternative way is to evaluate numerical
derivatives by perturbation of iteration variables.

Where h is the perturbed value.

17) Calculate corrections to the inner loop iteration variable using Newton Raphson iteration method
[6].
15
 ------------------------------------------------------------------------ (31)

-------------------------------------------------------------------------- (32)

is a damping factor. It can be optimally calculated at every iteration using Fibonacci search [23]. The
method by Broyden [24] can be used for subsequent cycles so that re-estimation of jacobian of partial
derivatives is avoided and to ensure that less computation time is required.

To solve for in equation (31), an alternative method viz LU decomposition method [25], which
takes less computation time can also be used.

18) Calculate the square root of the sum of square of the values of normalized discrepancies. Divide it by
the number of stages in the system. If the value is less than 0.0001 then proceed to the next step [17]. If
not then repeat step 16 and 17 with the new value of iteration variable until it converges.

Norm =

where f(x) is either the enthalpy function Hn or the discrepancy function D1 and D2.
If Norm<10-4 then proceed to next step. Else repeat step 16 and 17 with improved value of iteration
variable until it converges.

19) Repeat step 11 to produce new values of primitive variable xm,n, ym,n, vm,n, lm,n, Vn, Ln, Tn

Section 3: Outer loop calculations

This sections aims to check for final convergence of the inside out method. If the outer loop does not converge,
the parameters of simple thermodynamic models based on new primitive variables are updated here.

20) Using primitive variables of the previous step, compute relative volatilities αm,n of all component at
every stage.

21) The convergence criteria for outer loop is given by Simandl and Svrcek [16]

16
 Where volcrit = 0.002* Total number of stages (N)

If the criterion above is met then both outer and inner loops are converged and final solution is
obtained. Otherwise proceed to the next step.

22) Calculate stream enthalpies from complex thermodynamic models. Compute parameters An, Bn, cn, dn,
en, fn as done in steps (7 and 9). Also compute coefficients for activity model an,bn for non ideal
systems using step 8.

23) Calculate values of base stripping factor as well as liquid and vapor withdrawal factors.

24) Repeat the inner loop calculation from step 11 to 18.

The proposed flow chart is presented in Figure 4.

Conclusion :
In this paper the authors have presented the simplified procedure for the simulation of multicomponent
distillation system. The explicit procedures for calculation of K-values and enthalpy values in complex
thermodynamic models have been provided. Also the explicit procedures for calculation of a reference/base
component Kb value in simple K-value model and departure of enthalpy from ideal values in approximate
enthalpy model have been provided. New flowchart has been proposed for the simulation of multicomponent
distillation system

Nomenclature

αm,n = Relative volatility with respect to a base/reference component

= Ideal enthalpy departure for liquid

= Ideal enthalpy departure for vapor

= Activity coefficient for liquid

= Value of activity coefficient using approximate model

= Pure species fugacity coefficient for liquid

= Partial fugacity coefficient for vapor

ω = Acentric factor
= Damping factor
An = Constant in Kb model equation

17
Bn = Constant in Kb model equation
Bspec = Specified bottoms flow rate
= Specified key component flow rate in the bottom product

am,n = Constant in activity coefficient model equation

bm,n = Constant in activity coefficient model equation
cn = Constant in vapor enthalpy departure model equation
dn = Constant in vapor enthalpy departure model equation
Dspec = Specified distillate flow rate
en = Constant in liquid enthalpy departure model equation
fn = Constant in liquid enthalpy departure model equation
Fn = Feed flow rate at stage n
= Energy balance function/ Specification function to be converged using Newton Raphson
fm,n = Component feed flow rate at stage n
fn = Constant in liquid enthalpy departure model equation
h= perturbed value for derivative calculation
hLo = Ideal enthalpy for liquid/Enthalpy of vapor using liqud composition
hL = Real enthalpy for liquid
hVo = Ideal enthalpy for vapor
hV = Real enthalpy for vapor
hF = Enthalpy of feed
= Ideal enthalpy of pure component
Km,n = Liquid vapor distribution ratio for component m at stage n
Kb,n = Base component liquid vapor distribution ratio at stage n
lm,n = Component liquid flow rate at stage n
Ln = Liquid flow rate at stage n
= Key component flow rate in the bottoms

MV = Molecular mass for vapor

ML = Molecular mass for liquid
N = Total number of equilibrium stages
Qn = Heat transfer rate at stage n
= Liquid side stream withdrawal factor

18
 = Vapor side stream withdrawal factor

Sm,n = Stripping factor for component m at stage n

Sb = Scalar multiplier
Sb, n = Base stripping factor at stage n
Tn = Temperature at stage n
T* = Reference temperature
= Specified temperature of the last stage/reboiler

= Specified temperature of the first stager/condenser

Tc = Critical temperature of a component

Tr = Reduced temperature
Un = Liquid side stream flow rate at stage n
Vn = Vapor flow rate at stage n
vm,n = Vapor component flow rate at stage n
Wn = Vapor side stream flow rate at stage n
wm,n = weighing factor to be used in equation
Xspec = Specified key component mole fraction
xm,n = mole faction of liquid component m at stage n
= Iteration variable (Stripping factor/ Side stream factor) used to converge
= Correction to the iteration variable
ym,n = mole faction of vapor component m at stage n
Z = Compressibility factor
Zv = Compressibility factor for vapor

Abbreviation:

UNIFAC:
SRK
NRTL

19
 Tables

Table 1: Replacement functions for condenser duty [6]

Specification Equation to replace Condenser duty
Reflux ratio

Condenser temperature

Distillate rate

Key component flow

Key component mole fraction

Table 2: Replacement functions for reboiler duty [6]

Specification Equation to replace Reboiler duty
Reboil ratio

Reboiler temperature

Bottoms rate

Key component flow

Key component mole fraction

20
References:

1. J. R. Haas, UOP LLC, Des Plaines, Illinois, USA Copyright ^ 2000 Academic Press

2. Fouad M. Khouray (2014) Multistage Separation Processes, 4th edition, CRC Press, Taylor and Francis group,
Boca Raton, Florida, pp.457

3. Seader J.D and Henley E.J (2006) Separation Process Principles, 2nd Edition, Wiley-India Publication, New
Delhi, India, 756 pp.

4. Wang, J.C., and G.E. Henke, Hydrocarbon Processing, 45(8), 155–163 (1966).

5. Friday, J.R., and B.D. Smith, AIChE J., 10, 698–707 (1964).

6. Napthali L.M. and Sandholm D.P. (1971) Multicomponent separation calculations by linearization, AIChE
Journal, 17, 148-153.

7. Boston J.F. and Sullivan S.L. (1974), A new class of methods for multicomponent, multistage separation
processes, The Canadian Journal of Chemical Engineering, Volume 52, 52-63

8. Russell, R.A., (1983) A flexible and reliable method solves single tower and crude distillation column
problems, Chem. Eng., 90(20), 53–59.

9. Jelinek J. (1988) The calculation of multistage equilibrium separation problems with various specifications,
Computational Chemical Engineering, Voume12, 195-198.

10. Saegar R.B., and Bishnoi P.R., (1986) A modified “Inside Out” algorithm for simulation of multistage
multicomponent separation processes using UNIFAC group contribution methods, Canadian Journal of
Chemical Engineering, Volume 64, 759-767.

11. Shan L. C., Ping Z.X., Jiang Z. S., Shun T.X. and Guang X.S.(2006) Simulation of multicomponent
multistage separation processes – An improved algorithm and application, The Chinese Journal of Process
Engineering, Volume 6 (2), 247-254.

12. Boston J.F. and Britt H.I. (1978) A radically different formulation and solution of the single stage flash
problem, Computers and Chemical Engineering, Volume 2, 109-122.

13. Fournier R.L. and Boston J.F. (1980) A Quasi Newton algorithm for solving multiphase equilibrium flash
problems, Chem. Eng. Commun, Volume 8, 305-326

14. Boston J.F (1980), Inside Out algorithms for multicomponent separation process calculations, ACS
Symposium Series No. 124, 135-151.

15. Trevino-Lozano R.A. and Kisala T.P. (1984) A simplified absorber model for non linear simultaneous
modular flowsheet calculations, Computers and Chemical Engineering, Volume 8(2), 105-115.

21
16. Simandl J. and Svrcek W.Y. (1991) Extension of the simultaneous solution and Inside Out algorithm to
distillation with chemical reactions, Computers and Chemical Engineering, Volume 15(5), 337-348.

17. Kister, H. Z., (1992) Distillation Design, McGraw-Hill, Inc., New York.

18. Perry’s Chemical Engineers’ Handbook, (1999) 7th Edition, R.H. Perry and D.W. Green, Eds., McGraw-Hill
. pp. 2-350

19. Boston J.F. and Sullivan S.L. (1972), An improved algorithm for solving mass balance equations in
multistage separation processes, The Canadian Journal of Chemical Engineering, Volume 50, 664-668.

20. Boston J.R. and Shah V.B., Paper presented at AIChE National Meeting, Houston, Texas (1979)

21. Venkataraman, S., Chan, W.K., Boston, J.F. (1990) “Reactive distillation using ASPEN PLUS”, Chem. Eng.
Prog., 86(8), 45—54.

22. Smith J.M., Van Ness H.C. and Abbott M.M. (2005) Introduction to Chemical Engineering
Thermodynamics 7th Edition, McGraw-Hill pp. 561.

23. Beveridge, G.S.G., and R.S. Schechter, (1970) Optimization: Theory and Practice, McGraw-Hill, New
York, pp. 180–189.

24. Broyden C.G. (1965) A class of methods for solving non linear simultaneous equations, Math Comp., 19,
577–593.

25. Chapra S.C. and Canale R.P. (2010) Numerical Methods for Engineers, 6th Edition, McGraw Hill
Publications, New York, NY 10020, pp. 274-283

22
 Figures

Vn Ln-1

Vapour side stream Wn ym,n xm,n-1

hVn hLn-1
Tn Tn-1

Valve F
Feed Heat transfer Qn
Fn Stage n
hFn (+) if from stage
xm,n and (-) if to stage
Tn
hLn
ym,n-1 Tn
hVn
Tn Liquid side stream Un
Vn-1 Ln

Figure 1: Equilibrium Stage

23
 V1

Q1
F1 Stage 1
W2 U1
V2 L1
Q2
F2 Stage 2
W3 U2

Un-1
Wn Qn
Fn Stage n
Wn+ Un

UN-2
WN- QN-1
FN-1 Stage N-1
UN-1
WN Vn LN-1
QN
FN Stage N

LN
Figure 2: Cascade of N equilibrium stages

24
 Provide initial estimates and
calculate outer loop parameters
by using initialization procedure

Solve mass balances to obtain liquid component flow rates using Thomas
algorithm. Calculate vapour component flow rate using phase
equilibrium relation

Calculate Total flow rates using summation and stage

temperatures using Kb method
Outer Loop
Inner Loop

Evaluate enthalpies of all streams. Solve for energy balances and specification equations
using Newton Raphson technique. The natural logarithms of stripping factors are used as
iteration variables along with natural logarithms of withdrawal factors

Update iteration Do an inside loop convergence check

variables using No
simultaneous correction. Norm = , Norm<0.0001 for convergence.

Update outside
loop parameters f(x) is an enthalpy function or a specification function.
using the
procedure Yes
mentioned in the
Calculate new values of primitive variables Ln , Vn, Tn,
initialization
xm,n, ym,n and relative volatilities αm,n
section

Perform an outer loop convergence check

No

Where volcrit = 0.002* Total number of stages (N)

Yes

Converged

Figure 4: Proposed Flowchart for inside out method

25