L.A. Benitez et alInt. Journal of Engineering Research and Applications www.ijera.

com
ISSN : 2248-9622, Vol. 4, Issue 4( Version 1), April 2014, pp.114-119

RESEARCH ARTICLE

OPEN ACCESS

Two Noveland Simple Correlations to Estimate Equilibrium

Water Dew Point for Natural Gas Dehydration Process
Leonel Alberto Benitez1*, Juan Pablo Gutirrez1, Eleonora Erdmann2, Liliana
Ale Ruiz3, Enrique Tarifa4
1

Instituto de Investigaciones para la Industria Qumica (INIQUI-UNSa-CONICET). Consejo de Investigaciones

(CIUNSa). Facultad de Ingeniera, Universidad Nacional de Salta (UNSa).Avda. Bolivia N 5150 Salta
(4400), Argentina.
2
Instituto Tecnolgico de Buenos Aires (ITBA). Instituto de Investigaciones para la Industria Qumica (INIQUIUNSa-CONICET). Av. Madero 399 Ciudad de Buenos Aires (C1106ACD), Argentina.
3
Consejo de Investigaciones (CIUNSa). Facultad de Ingeniera, Universidad Nacional de Salta (UNSa). Avda.
Bolivia N 5150 Salta (4400), Argentina.
4
Facultad de Ingeniera, Universidad Nacional de Jujuy (UNJu) - CONICET. taloPalanca N10 San Salvador
de Jujuy (4600), Argentina.

ABSTRACT
Water is probably the most undesirable component found in crude natural gas because its presence can produce
hydrate formation, and it can also lead to corrosion or erosion problems in pipes and equipment. Natural gas
must be dehydrated before being transported through a long distance to ensure an efficient and trouble-free
operation. Thermodynamic modelling of triethyleneglycol (TEG)-water system is still rather inaccurate,
especially with regard to systems at high temperature and high TEG concentration. As a consequence, design
and operation of absorber towers are affected by the lack of accurate data. Two novel correlations have been
developed to estimate the equilibrium water dew point of a natural gas stream by evaluating experimental data
and literature. These data were collected and analyzed by means of images scanned with MATLAB software
R2012B version. An average percentage error is of 1-2% for linear correlation and it is of 2-3% for non-linear
correlation. Results are quite accurate and they are consistent with literature data. Due to the simplicity and
precision of the correlations developed in this work, the equations obtained have a great practical value.
Consequently, they allow process engineers to perform a quick check of the water dew point at different
conditions without using complex expressions or graphics.
Keywords-Correlation, Dew Point, Natural Gas, TEG-water system, Triethylene glycol, Water Dehydration

I. INTRODUCTION
Natural gas is a vital component of the
world's supply of energy. It is a fossil fuel which is
one of the cleanest, safest, and most useful energy
sources. Natural gas burns more cleanly than other
hydrocarbon fuels, such as oil and coal, and produces
less carbon dioxide per unit of energy released.
Natural gas is a hydrocarbon gas mixture which is
formed primarily of methane, ethane, propane and
butane; but commonly includes some impurities such
as water, carbon dioxide, nitrogen, hydrogen sulfide
and traces of heavier condensable hydrocarbons.
Water is probably the most undesirable
component found in raw natural gas. Its presence
could produce important clogging and plugging in
pipelines, valves and other devices. Normal gas flow
is stopped and obstructed in transport lines due to
hydrate formation, Fig.1 (a) and (b) [1].
Furthermore, it could lead to corrosion or
erosion problems in pipes and equipment, caused by
the dissolution of CO2 and H2S in water.
www.ijera.com

Natural gas treating should be unavoidable

to prevent such problems. Therefore, impurities such
as H2S, CO2 and H2O should be reduced or removed
from natural gas stream in order to use it as a fuel.
The removal is made to satisfy and meet certain
specifications regulated by governmental or private
entities
Natural gas dehydration represents an
important operation in the gas industry. A natural gas
stream must be dehydrated before being carried to
consumption and distribution centers to ensure an
efficient and trouble-free operation. Basically, the
fundamental objective of the dehydration is to
remove water, in vapor phase mainly, from the
treating gas which leaves sweetening process. By far,
the most common and successful technique for
natural gas dehydration is the contact between the gas
and a hygroscopic liquid. Glycol solutions are the
most common liquid absorbents in gas industry.
Glycols are diols whose hydroxyl groups give them a
high affinity for water. Particularly ethylene glycol
114|P a g e

L.A. Benitez et alInt. Journal of Engineering Research and Applications www.ijera.com

ISSN : 2248-9622, Vol. 4, Issue 4( Version 1), April 2014, pp.114-119
(EG), diethylene glycol (DEG), triethylene glycol
(TEG), and tetraethylene glycol (T4EG) allow
reaching different dehydration levels. TEG
Absorption process is one of the most acknowledged
methods used for natural gas dehydration [2].

Figure 1(a). Adhered Hydrate on internal walls on

gas pipelines (Picture from Repsol YPF)

Figure 1(b). A large gas hydrate plug formed in a

subsea hydrocarbon pipeline (Picture from Petrobras
- Brazil).
A TEG dehydration process could be
divided into two main parts, gas dehydration and
solvent regeneration. In the dehydration stage,
water is removed from the gas; while in the
regeneration stage, water is removed from the
solvent. After this process, the fresh glycol solution is
apt again to be used in the contactor tower [4, 5]. The
regeneration stage is a normal operation in the gas
industries. It allows reducing undesirable solvent
waste which implies mainly important economic
saves.
A typical dehydration process diagram is
shown in Fig. 2 [4, 5]. A lean TEG solution is fed at
the top side of the contactor, while natural gas is
introduced at the bottom. It leads to a countercurrent
contact between the gas stream and the lean TEG
solution. The contactor is an absorption column
which allows mass transfer at high pressure and low
temperature. Consequently, the bottom stream, which
leaves the contactor unit, is a rich TEG solution with
www.ijera.com

high concentration of water and some hydrocarbons.

It is depressurized by means of a pressure reducing
valve. This stream enters into the regeneration stage
where it is forced to pass by a flash separator in order
to strip gaseous hydrocarbon and condensates which
are carried by the glycol.
The rich glycol is heated in the rich-lean
heat exchanger. Then it is filtered before being fed to
the regenerator step. The regenerator sector includes
a reboiler, a distillation column (STILL), and a surge
drum. In the regenerator column, the glycol itself is
stripped from the absorbed water. The process occurs
almost at near atmospheric pressure by the
application of external heat. The low water
concentration stream that leaves the surge drum is the
regenerated lean glycol. Then, it is partly cooled in
the lean-rich exchanger. Finally, the solution is
pumped through the glycol cooler before being
recycled to the contactor.

Figure 2.Diagram of a Typical Process Natural

Gas Dehydration with TEG.
Evaluation of a TEG dehydration system
consists of establishing the minimum TEG
concentration required to reach water dew point
specification in the final gas stream [6]. In order to
have a precise design of a dehydration unit, data from
liquid-vapor equilibrium (LVE) for TEG-water
system must be accurate, particularly in the diluted
water region. For this purpose, sophisticated graphs,
charts and correlations are found in literature. They
show these equilibrium data as water dew point of a
natural gas stream in equilibrium with a TEG
solution at diverse contactor temperatures and TEG
concentrations (Worley, Rosman and Parrish) [7].
So, a required TEG concentration can be estimated
for a particular application, or the dew point
depression for a given theoretical TEG concentration
at a given contactor temperature.
A better alternative for the complex graphs
and correlations is the application of the correlations
presented by Bahadori and Vuthaluru. The new
correlation reported has been developed for the
simple estimation of water dew point in a stream of
115|P a g e

L.A. Benitez et alInt. Journal of Engineering Research and Applications www.ijera.com

ISSN : 2248-9622, Vol. 4, Issue 4( Version 1), April 2014, pp.114-119
natural gas in equilibrium with a solution of TEG at
different TEG concentrations and temperatures [8].
In spite of this, thermodynamic modeling of
TEG-water system is still rather inaccurate,
especially with regard to systems at high temperature
and high TEG concentration [3]. As a consequence,
the design and operation stages of absorber towers
are affected by the lack of accurate data. A better
alternative to these complex tools is the application
of simpler correlations.Consequently, two novel
correlations have been developed to estimate the
equilibrium water dew point of a natural gas stream
by evaluating experimental data and literature.

3.

Correlate the coefficients ai and bi, which were

found in the previous steps, with the
concentration of TEG Wi. Indeed, functions
fa(W) and fb(W) were determined such that
provides the best fit to the values ai and bi
respectively. To obtain the appropriate structure
of each of the above functions, a bank of
adjustment functions was used, whose
parameters were adjusted by regression. The best
fit functions were selected to be part of the
correlation. Fig. 3(a) and Fig. 3(b) show
adjustment achieved.

II. METHODS
Mathematical models for prediction of water
dew point of a natural gas stream could be estimated
as a linear function and a non-linear function of
contactor
temperature
for
different
TEG
concentration.
Correlations have been developed by
evaluating experimental data and literature. The
required literature data to develop the correlation is
presented in Figure 20.54 in GPSA (Gas Processors
and Suppliers Association Engineering Data Book,
2004). This figure is based on Parrish, et al. (1986)
[7] equilibrium data, which covers VLE data for
TEGwater system for contactor temperatures
between 80 F and 130 F and TEG concentrations
ranging from 95wt% to 99.99wt%. These data were
collected and analyzed by means of an image
scanning technique with MATLAB software R2012B
version. MATLAB and Image Processing Toolbox
provide a wide set of tools that allow users to easily
handle, process and analyze data from various format
image.
2.1 METHODOLOGY FOR THE DEVELOPMENT OF A
LINEAR CORRELATION

Water dew point (Td) of a natural gas stream

in equilibrium with a TEG solution is correlated as a
linear function of contactor temperature (Tc) for
different TEG concentration in weight percent (W).
Coefficients for this equation are correlated as a new
function of their corresponding TEG concentrations.
The methodology that has been applied to
set correlations coefficients is summarized in the
following steps:
1. The linear function Td=ai+biTc correlate the
water dew point Td of a natural gas stream in
equilibrium with a solution of TEG as function
of temperature of the contactor Tc, for a given
concentration of TEG Wi. As a result, values of
ai and bi are obtained which are associated to the
Wi concentration.
2. Repeat step 1 for different concentrations of
TEG.
www.ijera.com

Figure 3(a). Functions fa(W) that provides the best

fit to the coefficients values ai

Figure 3(b). Functions fb(W) that provides the best

fit to the coefficients values bi
Thus, the first proposal correlation is
obtained, which is listed below:

Td a bTc (1)
a A1

B1W 2
W

C12

D1W 2
W
2

E12

W E2
1

b A2 Exp G 2
Ln

C
2
B2

F1W 2
W 2 G12

D2

(2)

F2 (3)

W dependence is shown in equations (2) and

(3) whose coefficients are reported in Table 1.

116|P a g e

L.A. Benitez et alInt. Journal of Engineering Research and Applications www.ijera.com

ISSN : 2248-9622, Vol. 4, Issue 4( Version 1), April 2014, pp.114-119
Table 1. Coefficients used in Equations (2) and (3)
Coeff. 95%<TEG<99.99% Coeff. 95%<TEG<99.99%

A1

-4.368339E+01

A2

2.109645E+00

B1

-1.662446E+01

B2

4.398181E+01

C1

1.115752E+02

C2

-1.291611E+01

D1

-1.001244E-01

D2

-1.057123E-01

E1

-1.101131E+03

E2

5.603903E+01

F1

-1.587084E-01

F2

-2.988238E-02

G1

1.002125E+02

G2

-5.000000E-01

concentration in weight percent (W). This new

correlation is as follows:

Td A3 B3 LnF3W G3 C3 LnH 3TC I 3

D3 Ln F3W G3 2 E3 Ln H 3TC I 3 2 (4)

The methodology that has been applied to

set correlations coefficients is the one followed for
the linear correlation. The required literature data to
develop this correlation is also presented in Figure
20.54 in GPSA [7]. Table 3 shows the coefficients for
the correlation obtained from the non-linear
regression.
Table 3.Coefficients for the new correlation

Table 2 shows the average absolute

percentage error (AAPE) from the literature reported
data is 1.40% [7 9]. There is also an average
absolute error of 0.35% compared with data source
[7] which were used into the adjust step, with a
maximum error of 1.00% and a minimum of 0.01%.
This demonstrates an excellent accuracy and
performance of the proposed correlation.

Coefficients

95% <TEG< 99.99%

A3

6.107109E+01

B3

1.346193E+01

C3

1.101468E+01

D3

3.014112E+00

Table 2. Linear Estimation, as a linear function of the

temperature of the contactor, in comparison with the
literature reported data (Bahadori y Vuthaluru;
Parrish et al., 1986; Herskowitz and Gottlieb, 1984)
[7 9]

E3

7.489034E-01

F3

4.499560E+00

G3

-1.073135E+03

H3

-5.953315E+00

I3

5.952720E+02

TEG Concn.
WeightPerce
nt

95
97
98
99
99.9
99.97
99.99

Tc (K)
303.1
5
318.1
5
333.1
5
283.1
5
303.1
5
283.1
5
303.1
5

Linear
Correlatio
n

Td (K)

Absolut
e
Error
Percent

278.15

274.43

1.34

284.65

282.80

0.65

290.65

290.21

0.15

243.15

241.32

0.75

232.65

228.20

1.91

210.00

204.72

2.52

211.70

206.51

2.45

Reporte
d
Data

Average absolute percentage error (AAPE)1.40

2.2 METHODOLOGY FOR THE DEVELOPMENT OF A

NON-LINEAR CORRELATION
It should be possible to develop a correlation
with less regression coefficients. Water dew point
(Td) of a natural gas stream in equilibrium with a
TEG solution is correlated as a non-linear function of
contactor temperature (Tc) for different TEG
www.ijera.com

Table 4 shows a contrast between the results

reported by the non-linear correlation with the
literature reported data [7 - 9]; the average absolute
percentage error (AAPE) is of 2.41%. There is also
an average absolute error of 0.46% compared to data
source [7] which were used into the adjust step, with
a maximum error of 1.43% and a minimum of 0.01%.
This demonstrates an excellent accuracy and
performance of the proposed correlation.
For a second time, it is demonstrated that the
results reached by using this method have an
excellent accuracy and performance. Although
absolute error of this correlation is 70% greater than
the deflection of the first linear correlation, the
number of implemented coefficients is reduced to
half.

117|P a g e

L.A. Benitez et alInt. Journal of Engineering Research and Applications www.ijera.com

ISSN : 2248-9622, Vol. 4, Issue 4( Version 1), April 2014, pp.114-119
Table 4. Non-Linear Estimation, as a non-linear
function of the temperature of the contactor, in
comparison with the literature reported data (Parrish
et al., 1986; Herskowitz and Gottlieb, 1984) [7 9]
TEG Concn.
WeightPerce
nt

95
97
98
99
99.9
99.97
99.99

NonLinear
Correlatio
n
Td (K )

Absolut
e
Error
Percent

278.15

280.41

0.81

284.65

282.44

0.78

290.65

284.86

1.99

243.15

239.11

1.66

232.65

227.86

2.06

210.00

197.17

6.11

211.70

204.37

3.46

Reporte
Tc (K) d Data

303.1
5
318.1
5
333.1
5
283.1
5
303,1
5
283.1
5
303.1
5

Average absolute percentage error (AAPE)2.41

2.3 ASSOCIATED ERRORS

When wet gas is in contact with the rich
TEG solution in the absorption column at a constant
pressure and temperature, equilibrium is attained in
time. Since the gas and TEG are not in contact for a
long enough time to reach the state of equilibrium,
the actual water dew point is always higher than the
equilibrium dew point (Td,eq). Consequently, the
actual dew point of an output gas stream (Td,act)
depends on the TEG circulation rate and the number
of equilibrium stages. A well-designed and properly
operated unit will have an actual water dew point of 6
to 11C higher than the equilibrium dew point [10].

Td ,act Td ,eq 6 11 C

(5)

There is no practical requirement to include

temperatures higher than 130 F in the analysis of this
study. For higher inlet gas temperatures a cooler is
used to limit the TEG loss with dehydrated gas. Both
linear and non-linear correlations should be used up
to 10.300 kPa (abs) with little error. This is because
equilibrium dew points are relatively insensitive to
pressure [7, 8].

III. EXAMPLE
A simple example is presented to illustrate
the associated simplicity with the use of the proposed
correlation for the simple estimation of natural gas
water dew point at different temperatures and
concentrations of TEG.

0.85 Million Sm3/day of a natural gas stream

enters to a TEG contactor at 38 C and 4100 kPa
(abs). The target H2O dew point is 4 C (269.15 K).
www.ijera.com

Calculate the lean TEG concentration in mass percent

at this given temperature (38 C). Assume a 6 C
approach to equilibrium:
To find the solution, value of W must be
iterated until find the one which causes the actual
dew point equal to 269.15 K; Td,act = Td + 6 K. The
following calculations are performed using the two
proposed correlations in this work.
3.1 LINEAR CORRELATION
A- Assume TEG concentration W = 98%
Applying equations (2) and (3), we obtain:
a=15.9124886 y b=0.82334051
Equilibrium water dew point of gas stream was
calculated by equation (1) is Td,eq=272.09 K. Actual
water dew point is Td,act= 272.09 + 6 = 278.09 K.
B- Assume TEG concentration W = 98.9 %
Applying equations (2) and (3), we obtain:
a=23.2088603 y b=0.77423461
Equilibrium water dew point of gas stream was
calculated by equation (1) is Td,eq=264.11 K. Actual
water dew point is Td,act= 264.11 + 6 = 270.11 K.
C- Assume TEG concentration W = 98.97 %
Applying equations (2) and (3), we obtain:
a=23.9550366 y b=0.76887996
Equilibrium water dew point of gas stream
was calculated by equation (1) is Td,eq=263.19 K.
Actual water dew point is Td,act= 263.19 + 6 = 269.19
K.
3.2 NON-LINEAR CORRELATION
a- Assume TEG concentration W = 98 %
Equilibrium water dew point of gas stream
was calculated by equation (4) is Td,eq=271.85 K.
Actual water dew point is Td,act = 271.85 + 6 = 277.85
K.
B- Assume TEG concentration W = 98.9 %
Equilibrium water dew point of gas stream
was calculated by equation (4) is Td,eq=263.26 K.
Actual water dew point is Td,act = 263.26+ 6 = 269.26
K.
C- Assume TEG concentration W = 98.91 %
Equilibrium water dew point of gas stream
was calculated by equation (4) is Td,eq=263.14 K.
Actual water dew point is Td,act = 263.14 + 6 = 269.14
K.

IV. RESULTS AND DISCUSSIONS

The linear correlation result Td,act=269.19 K
has good concordance with expected water dew point
(Td=269.15 K). So required glycol purity to meets
targeted water dew point is W=98.97 %.The obtained
118|P a g e

L.A. Benitez et alInt. Journal of Engineering Research and Applications www.ijera.com

ISSN : 2248-9622, Vol. 4, Issue 4( Version 1), April 2014, pp.114-119
non-linear correlation result Td,act= 269.14 K has good
concordance with the expected water dew point
(Td=269.15 K). So required glycol concentration to
reach the required water dew point is W =98.91 %.
It is noted that using the correlation
developed by Bahadori and Vuthaluru [8] to solve the
proposed example, they obtained a TEG
concentration equal to W =99.02%. This shows an
excellent performance of proposed correlation.

REFERENCES
[1]

[2]

V. CONCLUSIONS
Two novel correlations have been developed
to estimate the equilibrium water dew point of a
natural gas stream. These equations are much simpler
than currently available models involving a large
number of parameters, requiring more complicated
and longer computations. They tend to replace
cumbersome, complex tables and graphs which are
generally used in the design and operation of gas
dehydration systems. Therefore, results show an
excellent performance of proposed correlations.
Results are quite accurate and they are consistent
with experimental and literature data. Due to the
simplicity and precision of the correlations developed
in this work, the equations obtained have a great
practical value. Due to the simplicity and precision of
the correlations developed in this work, the equations
have a great practical value. Indeed, they are basic
algebraic equations that can be easily implemented in
a spreadsheet. Consequently, they allow process
engineers to perform a quick check of the water dew
point at different conditions without using complex
expressions or graphics. Additionally, these
correlations are also useful in the design of the
absorber in the dehydration process.

www.ijera.com

[4]
[5]

[6]

[7]

[8]
[9]
[10]
[11]

[12]

NOMENCLATURE
Ai
Coefficient
Bi
Coefficient
Ci
Coefficient
Di
Coefficient
Ei
Coefficient
Fi
Coefficient
Gi
Coefficient
Hi
Coefficient
Ii
Coefficient
Td
Water Dew Point, K
Td,eq Equilibrium Water Dew Point, K
Td,act Actual Water Dew Point, K
Tc
Contactor Temperature, K
W
TEG concentration in water
(weight percent)

[3]

Institute of Petroleum Engineering, Centre

for Gas Hydrate Research, Heriot-Watt
University,
Edinburgh.
http://www.pet.hw.ac.uk/research/hydrate/h
ydrates_why.cfm
F. Gironi, M. Maschietti, V. Piemonte,
Modeling triethylene glycol-water systems
for natural gas dehydration, Eighth
International Conference on Chemical and
Process Engineering, Chemical Engineering
Transactions, ISBN 88-95608-00-6.
C.H. Twu, V. Tassoneb, W.D. Simb, S.
Watansiri, Fluid Phase Equilib., 228229,
2005, 213221.
P. Kazemi, R. Hamidi, Petroleum & Coal,
53 (1), 2011, 7177. ISSN 13377027.
K. MohdAtiqueuzzaman, ASM. Maruf,
IOSRJAC, 2 (2), 2012, 01-06. ISSN 2278
5736.
A. Bahadori, Y. Hajizadeh, H.B.Vuthaluru,
M.O.Tade, S. Mokhatab, J. Nat. Gas Chem.,
17, 2008, 298302.
Gas Processors & Suppliers Association
(GPSA), Gas Processors and Suppliers
Association Engineering Data Book
(Twelfth ed., Tulsa, OK, USA, 2004).
A. Bahadori, H. B. Vuthaluru, J. Nat. Gas
Sci. Eng., 1 (3), 2009, 6871.
M. Herskowitz, M. Gottlieb, J. Chem. Eng.
Data, 29, 1984, 173.
P.Gandhidasan, Energy Sources, 25, 2003,
189201.
R.H. Perry, D.W. Green, Perrys chemical
engineers handbook(Eighth ed., McGrawHill, 2007).
E. Erdmann, L. Ale Ruiz, L.A. Benitez, E.
Tarifa, Av. cien. ing., 3 (3), 2012, 119130.
ISSN:
07188706http://www.exeedu.com/publishing.cl/a
v_cienc_ing/

solution,

119|P a g e