937852013011666968969

7/25/2019 937852013011666968969

http://slidepdf.com/reader/full/937852013011666968969 1/7

Journal of Physical Science and Application 2 (10) (2012) 434-440

Simulation of MEG Packed Distillation Column Using an

Equilibrium Stage Model-Case Study on Operating

Parameters of Farsa Petrochemical Company:

Assaluyeh-Iran

Noorallh Kasiri and Yousef Dorj

Computer Aided Process Engineering Lab. School of Chemical Eng, Iran University of Sci. and Tech, Iran

Received: April 20, 2012 / Accepted: May 26, 2012 / Published: October 15, 2012.

Abstract: Two types of equilibrium and non-equilibrium stage models are generally used to simulate the mass transfer of packed

distillation column. Using non-equilibrium model requires the calculation of mass transfer coefficients, thus, usually equilibrium-based

methods are preferred to be used for simulations of distillation columns. In this paper, packed column distillation of production of

Mono Ethylene Glycol in FARSA SHIMI Company (Assaluyeh-Iran)’s Ethylene Glycol portion has been simulated through using the

equilibrium model and solving the related equations. The simulation has been carried out in the MATLAB environment. The column

also has been simulated in the Aspen Hysys and Aspen Plus ver. 2006.5 environments. Then, the output has been compared with

software results, designing and operating data of the underlying columns which demonstrate good consistency with the model. Having

the model validated, the effect of some operating parameters has been analyzed through the model.

Key words: Distillation, packed column, modeling, equilibrium and non-equilibrium models.

1. Introduction

Simulation is a powerful tool for process engineers

to develop and design, and it can be used to produce

different products. Modeling the system helps to

determine whether the system can lead to the desired

product or not. A column model can also predict the

condition of the tower under various operations [1].

The process engineers would be interested in finding

ways to increase efficiency and capacity of the units at

minimum costs. The accuracy of the models strongly

depends on the proper definition of the inter-phase

interactions [2]. Consistent with major improvements

in computational technology, there have been great

improvements in modeling and simulation of

multi-phase separation processes and mathematical

modeling has become more flexible and realistic.

Corresponding author: Yousef Dorj, research field:

modeling and simulation. E-mail: [email protected].

Shortcut and rigorous procedures are two calculation

methods used for the design and simulation of such

operations. There are also two fundamentally different

kinds of rigorous models to describe such operation

(i.e., the equilibrium stage model and non-equilibrium

or rate-based model) [3]. Even though for the

conventional structures, formed by structured packing,

due to the complex nature of intra-phase forces,

accurate focus would be difficult. So, modeling the

separation phenomena is usually based on the conceptof stages in which equilibrium or non-equilibrium

models could be utilized.

With major improvements in computational

technology, there have been great improvements in

modeling and simulation of multi-phase separation

processes and mathematical modeling has become

more flexible and realistic. Shortcut and rigorous

procedures are two calculation methods used for the

DAVID PUBLISHING

D

7/25/2019 937852013011666968969


Simulation of MEG Packed Distillation Column Using an Equilibrium Stage Model-case Study onOperating Parameters of Farsa Petrochemical Company-Assaluyeh-Iran

435

design and simulation of such operations. There are

also two fundamentally different kinds of rigorous

models to describe such operation (i.e., the

equilibrium stage model and non-equilibrium or

rate-based model) [3]. Even though for the

conventional structures, formed by structured packing,

due to the complex nature of intra-phase forces,

accurate focus would be difficult. So, modeling the

separation phenomena is usually based on the concept

of stages in which equilibrium or non-equilibrium

models could be utilized.

2. Equilibrium Model

In the past century, equilibrium models have beenwidely used to describe separation processes. The

history of computation of separation process may date

back to 1893, when Sorel 1893 published an equation

for simple and continuous distillation. This equation

includes mass and energy balance for calculating heat

loss. His equations were not widely used until 1921,

when the graphic techniques were presented by

Panchon & Savarit, 1921 [1].

A new graphic technique was developed by McCabe

and Theile 1925. They used a simple assumption of

constant tray mole overflow and removed the energy

balance equation. From 1981 to 1989, simulation of

distillation columns has been studied by many

researchers including Seader 1981 and Gani, et al.

1986 [4].

Ruiz and Cameron 1986 have suggested a model to

simulate a continuous distillation [5]. Many

researchers discussed different assumptions used in the

simulation of separation towers and the errors theyhave caused over the years. Ranzi et al 1988 discussed

the effects of energy balance on the simulation

equation [6]. They found that the energy balance must

be taken into account in order to simulate correct phase

behavior. Choe & Luyben 1987 concluded that the

vapor phases should not be ignored [7]. The

equilibrium stage model assumes that the contact time

of the streams is infinite in each stage. The product

distributions and temperature profiles calculated by

using the equilibrium stage model does not correspond

to real stages and therefore, cannot describe the actual

operating conditions in the packed column [3].

Commercial softwares are usually based on

equilibrium models.

3. Non-Equilibrium Model

Non-equilibrium or rate-based models are another

class of models providing a method to model

separation process using direct rates of mass and heat

transfer. Many models have been presented within this

structure. Since the late 1970s, several non-equilibrium

models (Krishnamurthy & Taylor, 1985 [8];Sivasubramanian et al 1987 [9]) have been proposed to

overcome the shortcomings of equilibrium stage

models. All these models have discarded the

assumption that each stage operates at equilibrium.

Instead, they introduced mass and heat transfer

coefficients to describe the varying status at which each

stage operates. The equilibrium stage models are still

widely used in the design and simulation of distillation

processes due to the difficulty of predicting the

required transfer coefficients [3].

In some models, the film and penetration theory are

used for modeling and the required parameters are

achieved by experimental techniques. The film model,

based on comparisons with various experimental

methods is superior to other models. Based on the film

theory, mass transfer resistance is focused on the

two-phase boundary layer of the film. Film thickness

defines the parameters of model that can be estimated

by experimental methods. Obviously, the mass transferoccurring at the two-layer film is only limited to

molecular diffusion. Outside the film, mass transfer

rate disappears completely due to high mixing levels

[10]. In these models, unlike equilibrium models,

thermodynamic equilibrium is assumed only at the

phase interface as demonstrated in Fig. 1. The Mass

balance for each phase has been written separately and

diffusion rate of each component has been shown. In

7/25/2019 937852013011666968969


Sim

436

Fig. 1 A sch

multi-comp

has been c

theory) equ

4. The M

A modifi

model has b

packed disti

and heat t

modeling re

equations t

including eequations,

components

model is ba

output and

practical th

mass trans

methods tha

each stage,

operation

Therefore, t

actual proc

model assu

equilibrium

columns, it

cross sectio

the next. T

model. Wit

effects of e

ulation of MOpera

eme of non-eq

nent proces

lculated by

tions [2].

del Descri

ed method b

een develope

llation colu

ransfer thro

sults are a s

at replace

fficiency eq emperature,

profile of t

ed on calcul

equilibrium

n the non-e

er coefficie

t can determi

this model

ondition all

his method

ess image a

es that the

with the ne

assumes tha

is at equili

is is the m

hin this dis

tertainment

G Packed Dting Paramet

uilibrium sta

es, diffusio

Maxwell-St

tion [3]

sed on the

d for simulat

ns. This mo

ugh the p

et of MESH

he classic

uation. Solvapor and

he column i

ting method

phases. So,

uilibrium

ts. Unlike

ne the operat

has the abil

through t

ould be ab

nd profile.

output of

t one. In th

the phase l

brium with t

in advanta

tillation col

and conduct

istillation Coers of Farsa

e [10].

between ph

fan (gas ki

quilibrium s

ion and desi

el specifies

cked bed.

EC multi-va

ESH equat

ing this seliquid flow

s specified.

s of mass tra

it is much

ethods that

other simul

ing conditio

ity to deter

e packed

le to expres

The equilib

ach phase i

e case of pa

aving a pac

he phase lea

e of the pre

mn model,

ion heat tra

lumn UsingPetrochemic

ases

etic

tage

n of

ass

The

iant

ions

ofand

This

sfer

ore

eed

tion

s on

ine

bed.

the

ium

s in

ked

ing

ing

sent

the

sfer

ma

foll

m

st

ac

m

eq

1 j L

1 j L

in

+1

Fur

of

rati

Fig.

n Equilibriual Company-

also be ana

ows:

(1) The pro

del and spe

ge. The en

count. For si

st be solved

uations for e

ass balance

,1,11,

i j ji j ji Z F yV x

quilibrium e

nd summati

nergy equati

111

F

j j

V

j j

L

j H F H V

hich:

his results i

nknowns le

ther proble

he feed rate

o, condenser

2 An equili

Stage Mod Assaluyeh-I

yzed. The c

ess is simul

ifies the te

tertainment

mulating, th

. Mass and

ch stage (Fi

equations:

(,11

j ji j U L x E

quations are:

ji y

,

j ji

T f K ,

,

on equations

1

C

j

1

C

j

ons would b

).(11

j j

L

j j U L H E

L

j H

V

j H

N (2C+3)

ding to N(C

definition

, temperatur

and reboiler

rium stage fo

el-case Studan

lculations pr

ted using th

perature pr

effect is no

classic ME

nergy balan

. 2) are as f

).().,,

ji j j ji

yW V x

ji ji x

,,.

ji ji j

y xP,,

,,

(S) are:

1, ji

1, ji

:

).( j

V

j j j

L

j E H W V

ji j j

xPT ,

,,

ji j j

yPT ,

,,

quations wit

7) +1 degre

ay include

and compo

pressures a

MESH.

y on

ocedure is a

equilibrium

file for eac

t taken into

H equation

es and other

llows:

0., ji j

x E (1)

(2)

(3)

(4)

(5)

0. j

L

j Q H (6)

(7)

(8)

h N (3C+10)

s of freedom

specificatio

sition, reflu

d definition

.

7/25/2019 937852013011666968969



437

Fig. 3 Diagram for an element of model for MESCHEC.

of composition and temperature profiles through thecolumn, as well as the liquid and gas flow rates on each

stage.

(2) Having specified the composition and

temperature profiles and liquid and gas flow rates on

each stage; it is assumed that a stage is equivalent to a

part of one HETP called a sub-bed. The vapor leaving

top of the sub-bed is in equilibrium with the liquid

leaving the bottom. An element of height dH is selected.

It is assumed that the phases are uniformly distributed

throughout this stage.

Mass balance is the same as Eq. (1). Liquid phase is

at its bubble point and the vapor phase is at its dew

point. It should be noted that the temperature of vapor

and liquid phase leaving the element are not the same,

however the temperature of vapor phase at the top and

liquid phase temperature at the bottom are the same.

The temperature profiles of liquid and vapor phase

would be determined by mass transfer rate. The

temperature of liquid and vapor leaving each stage orany height of the bed are identical. Of course, this

assumption is not always true. Hence, in order to

modify the model, the equations are rewritten as below:

Phase-corrected equilibrium equations model is

written as:

Yii ji

T F Y ,

(9)

Xii ji

T f X ,

(10)

Sum equations can be written as:

11

,

n

j

ji X (11)

11

,

n

j

jiY (12)

As well as the energy balance in equilibrium stage

the following term must also be added to Eq. (6):

dH

T T A iPiP

d

1

dH

T T A iPiP

d

1

(13)

In vacuum distillation gas load (gas velocity× square

root of gas density) is low and therefore, total mass

transfer rate is controlled by the gas phase (Van Winkle

1967)[11]. Thus, the efficiency for the vapor phase can

be written as:

HETPdH i

Y Y Y Y

j N j N

j N ji .

,1,

,1,

(14)

And for liquid phase:

HETP

dH i

X X

X X

j N j N

j N ji .

,,1

,,

(15)

The equations above can be solved using an iterative

method like the bubbles and Newton, Raphsoon.

5. Simulation of Mono Ethylene Glycol

Column

Ethylene glycol is produced under vacuum

distillation conditions. Water contamination in Glycol

mixture must be removed before the separation of

Mono Ethylene Glycol from heavier Glycols like DEG

& TEG. In Industrial practice, MEG is separated from

heavier Glycols using two vacuum columns in a

sequence.

In this study, top and bottom pressures of T-5001 are

kept at 23 and 25 KPa respectively. The temperature is

then increased from 154 to 160 °C. MEG of high purity

(99.99 wt.%) was drawn out from the first bed and sent

to storage. Heavier product was taken from bottom of

distillation column and sent to next unit to be separated.

In this column, structured packing (of

MELLAPACK250Y type) has been used. There are 3

beds in the columns of heights of 1,470, 4,620, 1,890

mm respectively from top to bottom with a total height

of 7,980 mm.

7/25/2019 937852013011666968969



438

To simulate the column, above equations were

solved in the MATLAB environment. The HETP value

is considered 400 mm based on the manufacturer’s data.

Rault’s law and Antoine equations are used to evaluate

vapor component pressure and the equilibrium

constants as the column is at vacuum pressure. For

vapor and liquid enthalpy, ideal gas and Clausius-

Clapeyron equations have been used. The influence of

conduction heat transfer between packing and fluid has

been neglected. Component concentration and

temperature profiles in the columns are as

demonstrated in Figs. 4 and 5.

As shown in Fig. 4, MEG concentration (increasing

from bottom to top) tends to one. Stages are numberedfrom top to bottom with MEG & TEG concentration

increasing from top to bottom which is in accordance

with actual plant conditions. The MEG product is

drawn out from stage 5 with the maximum MEG and

TEG existing in the reboiler stage.

As shown in Fig. 5, packed bed temperature

increases gradually from top to bottom of the column

with the maximum temperature occurring at the

reboiler. This demonstrates a good agreement with

practical conditions of the column in the plant. The

steep of temperature at the last stage shows the

reboiler’s heat production.

As it can be seen in Fig. 6, temperature profile

obtained from this model has good consistency with

actual operating profile and compared to the results

produced from the Aspen Hysys simulation outputs,

demonstrates even a better match. This column has

been simulated in the Aspen Hysys and Aspen plus Ver.

2006. 5 environments and Tables 3 and 4 show thecomparison between the present model, actual, design

and software simulation results for top and bottom

products.

In Tables 1 and 2, the results of simulation by using

this model and Aspen Plus 2006 & Aspen Hysys 2006

softwares with design and operational data for the

tower’s top and bottom products are compared.

As can be seen in the tables, for main product (MEG),

the value calculated from the model is consistent with

design, operational value and ASPEN Hysys which

indicates that the model is adequately accurate for this

system and provides good agreement with actual data.

This table also shows that Aspen Hysys is preferable to

Fig. 4 Composition profile on packing height (dH = 50

mm).

Fig. 5 Temraure profile in stages.

Fig. 6 Temperature profile Comparison.

7/25/2019 937852013011666968969



439

Table 1 Comparison between the model, practical, design and software results for MEG product.

Component Design (kg/hr) Operation data Aspen Hysys Aspen Plus Present ModelError compared to

design

Error compared

to operation

MEG 50,577.9 50,565.2 50,560 49,783.9 50,577 0.00 0.02

DEG 9.8 5.06 0.003 791.1 4.92 49.80 2.77TEG 0 0 0 12.19 0 0.00 0.00

WATER 0 17.7 5.9 0.76 0 0.00 0.00

Table 2 Comparison between the model, practical, design and software results for bottom product.

Component Design (kg/hr) Aspen Hysys Aspen Plus Model Error with Design

MEG 17,734.7 17,755 18,419.4 17,403.5 1.9

DEG 5,536.3 5,546.2 5,333.4 5,892 6.4

TEG 360.4 360.4 410.34 348.7 3.2

WATER 0 2.995E-09 0 0 0.0

Table 3 The effect of reflux rate on the reboiler duty and product concentration.

Reflux rate 20% decreases 10% decreases Design

Product concentration 0.9937 0.9987 0.9998

Reboiler duty (KW) 17,900 19,730 21,260

Reduction rate Percent for reboiler 16 7

Fig. 7 Feed tray effect of the MEG quality.

Aspen Plus in simulating this system. Compared to

design value, maximum error is associated with DEG

concentration at the bottom product but error between

reality and model is much lower (3%). The error is

mainly due to the ignoring components heavier than

TEG in the calculation procedure in the model. These

assumptions can be corrected as in the actual plant

conditions the heavier component are present with a

0.018 wt% composition.

6. Studying the Effect of Reflux on the

Product Quality and Reboiler Duty

Having validated the model in order to study the

effect of reflux rate, it is reduced to 10 and 20 percent

of the design value. Thus, the result obtained from the

model is shown in Table 3

As it can be seen in the table above, reducing reflux

has little effect on product concentration but reboiler

duty reduces by 7% and 16%.

7. Studying the Effect of Changing the Feed

Tray on the Product Quality

The design feed tray location is 16. Varying the feed

tray location from 10 to 20 leads to little change in

MEG quality but if it changes to lower locations,

product quality will slightly improve (Fig. 7) . Under

these conditions, other operation parameters such as

reflux rate and required condenser duty will also

decrease.

8. Summaries

The model for processing Glycol plants has been

built up in two different Hysys environments. A

mathematical model has also been developed to predict

the columns of the plant. The design and plant

operational data have been used for verification. The

investigation shows that the Hysys based simulations

7/25/2019 937852013011666968969



440

provide good predictions but the predictions based

onthe equilibrium model are of better capability in

predicting thermal behavior. Further studies have been

carried out on the reflux ratio and feed tray location

using the same model.

References

[1] V. Zaree, S. Shahhossini, Dynamic simulation of DEA

column in Arak. PC, in: 13th Iranian Chemical

Engineering Congress, Tehran, 2006.

[2] A. Shilkin, E.Y. Kenig, A new approach to fluid

separation modeling in the columns equipped with

structured packing, Chemical Engineering Journal 110

(2005) 87-100.

[3] L. Yang, K.T. Chuang, A new approach to simulation of

distillation in packed columns, Computers and Chemical

Engineering 24 (2000) 1843-1849.

[4] H.A. Kooijman, Dynamic nonequilibrium column

simulation, PH.D., Thesis, Clarkson University, 1995.

[5] M.S. Sivasubramanian, R. Taylor, R. Krishnamurthy, A

nonequilibrium stage model of multicomponent

separation processes: Part 4. A novel approach to packed

column design, Journal of American Institute of Chemical

Enginneers 33 (2) (1987) 325-327.

[6]

E. Ranzi, M. Rovaglio, T. Faravelli, G. Biardi, Role of

energy balances in dynamic simulation of multicomponent

distillation columns, Computers and Chemical

Engineering 12 (8) (1988) 783-786.

[7] Y.S. Choe, W.L. Luyben, Rigorous dynamic models of

distillation columns: Part 1. Model description and method

of solution, Journal of American Institute of Chemical

Enginneers 31 (3) (1985) 449-455.

[8] R. Krishnamurthy, A nonequilibrium stage model of

multicomponent separation processes: Part 1. Model

description and method of solution, Journal of American

Institute of Chemical Enginneers 33 (2) (1987) 325-327.

[9] R Gani, C.A. Ruiz, I.T. Cameron, A generalized model for

distillation column-I model description and application,

Computer and Chemical Eng 10 (3) (1986) 181-198.

[10]

A.M. Katariya, R.S. Kamath, Kannan M. Moudgalya,

S.M. Mahajani, Non-equilibrioum stage modeling and

non-linear dynamic effects in the synthesis of TEME by

reactive distillation, Computers and Chemical Engineering

32(2008)02243-2255.

[11] Process description and document of Farsa Pc.

Iran-Assaluye.

Documents

937852013011666968969