Journal of Natural Gas Chemistry 20(2011)471476

Comparison of three methods for natural gas dehydration

Michal Netusil , Pavel Ditl
Department of Process Engineering, Czech Technical University, Prague 6, 166 07, Czech Republic
[ Manuscript received April 6, 2011; revised May 23, 2011 ]

Abstract
This paper compares three methods for natural gas dehydration that are widely applied in industry: (1) absorption by triethylene glycol, (2)
adsorption on solid desiccants and (3) condensation. A comparison is made according to their energy demand and suitability for use. The
energy calculations are performed on a model where 105 Nm3 /h water saturated natural gas is processed at 30 C. The pressure of the gas
varies from 7 to 20 MPa. The required outlet concentration of water in natural gas is equivalent to the dew point temperature of 10 C at gas
pressure of 4 MPa.
Key words
gas reservoir; underground gas storage; natural gas; gas dehydration

1. Introduction UGSs, vapors of higher hydrocarbons also contaminate the

The theme of natural gas (NG) dehydration is closely con-
nected with the storage of NG. There are two basic reasons
why storing NG is an interesting idea. First, it can decrease the
dependency on supply. Second, it can exploit the maximum
capacity of distribution lines. NG is stored in summer periods
when there is lower demand for it, and is withdrawn in winter
periods when significant amounts of NG are used for heating.
Underground Gas Storages (UGSs) are the most advantageous
option for storing large volumes of gas. Nowadays there are
approximately 130 UGSs inside the European Union. Their
total maximum technical storage capacity is around 95 bcm.
According to the latest update, over 70 bcm of additional stor-
age capacity will come on stream in Europe till 2020 [1].
There are three types of UGSs: (1) aquifers, (2) depleted
oil/gas fields and (3) salt cavern reservoirs. Each of these
types possesses its own physical characteristics. Generally,
the allowable pressure of stored gas inside a UGS is up to
20 MPa. The inside pressure increases as the gas is injected
and decreases when there is a withdrawal. The output gas
pressure depends on further pipeline distribution. Distribution
sites normally begin at 7 MPa. The temperature of the gas
usually ranges from 2035 C. The exact temperature varies
with the location of UGS and with the time of year. A disad-
vantage of UGSs is that the gas becomes saturated by water
vapors during the storage. In the case of depleted oil field
stored gas. The distribution specification sets the allowable
water concentration in NG by specifying a dew point temper-
ature (Tdew ) of NG. Tdew is usually taken to be 7 C for NG
at 4 MPa [2]. This value is equivalent to roughly 5 gH2 O /m3
NG at 4 MPa. The water content in NG at saturation is depen-
dent on the temperature and pressure within the UGS. This is
well presented in Figure No.20, Chapter 20, in the GPSA Data
Book (12th Edition). The average value of H2 O in NG is five
times higher than that of required. A dehydration step of NG
from UGS is therefore essential before the gas is distributed.
This paper compares industrially applied dehydration meth-
ods according to their energy demand and suitability for use.
2. Dehydration methods
2.1. Absorption
First dehydration method is absorption of H2 O. Absorp-
tion is usually performed using triethylene glycol (TEG). Ab-
sorption proceeds in a glycol contactor (a tray column or
packet bed) with countercurrent flows of wet NG and TEG.
During the contact, TEG is enriched by H2 O and flows out of
the bottom part of the contactor. Enriched TEG then continues
flowing into the internal heat exchanger, which is incorporated
at the top of the still column. It then flows into the flash drum,

Corresponding author. Tel: +420-2243522714; Fax: +420-224310292; E-mail: netusil.michal@gmail.com

This work was supported by the Inovation and Optimalization of Technologies for Natural Gas Dehydration (No. FR-TI1/173).

Copyright2011, Dalian Institute of Chemical Physics, Chinese Academy of Sciences. All rights reserved.
doi:10.1016/S1003-9953(10)60218-6
472 Michal Netusil et al./ Journal of Natural Gas Chemistry Vol. 20 No. 5 2011

where the flash gases are released and separated from the
stream. TEG then runs to the cold side of the TEG/TEG
heat exchanger. Just afterwards, warmed TEG is filtered and
sprayed into the still column. From there, TEG runs into the
reboiler. In the reboiler, H2 O is boiled out of TEG. The inside
temperature should not exceed 208 C based on the decompo-
sition temperature of TEG. Regenerated TEG is then pumped
back through the hot side of the TEG/TEG and NG/TEG heat
exchanger into the top of the contactor. The entire process is
depicted in Figure 1 [3].

Figure 1. Scheme of absorption dehydration

The purity of the regenerated TEG and the circulation
rate (LTEG /kgH2 O ) limit the obtained output Tdew of NG.
Gas stripping can be implemented to enhance TEG regener-
ation. Proprietary design DRIZO , licensed by Poser-NAT,
COLDFINGER and Gas Conditioners International, have
been patented as an alternative to traditional stripping gas
units. The Drizo regeneration system utilizes a recoverable
solvent as the stripping medium. The patent operates with
isooctane solvent, but the typical composition is about 60%
aromatic hydrocarbons, 30% naphthenes and 10% paraffins.
Water separator of the three-phase solvent is crucial for this
method. The Coldfinger regeneration system employs a cool-
ing coil (the coldfinger) in the vapor space of the surge tank.
The cooling taking place there causes the condensation of a
huge amount of vapors. The condensate is a water rich TEG
mixture, which leads to a further separation process [4]. En-
hanced regeneration systems are depicted in Figure 2.

Figure 2. Scheme of enhanced TEG regeneration systems

 Journal of Natural Gas Chemistry Vol. 20 No. 5 2011 473

2.2. Adsorption specified by time borders A, B, C and D with appropriate bor-

der temperatures TA , TB , TC and TD . Regeneration starts at
The second dehydration method is adsorption of H2 O by point A. The inlet regeneration gas warms the column and the
a solid desiccant. In this method, H2 O is usually adsorbed on adsorbent. Around a temperature of 120 C (TB ), the sorbed
a molecular sieve, silica gel or alumina. A comparison of the humidity starts to evaporate from the pores. The adsorbent
physical properties of each desiccant is shown in Table 1 [5,6]. continues warming more slowly, because a considerable part
of the heat is consumed by water evaporation. From point C,
Table 1. Comparison of the physical properties of desiccants around the temperature of 140 C (TC ), it can be assumed that
used for NG dehydration
all water has been desorbed. Adsorbent is further heated to
Properties Silica gel Alumina Molecular sieve desorb C5+ and other contaminants till point D. The regenera-
Specific area (m2 /g) 750830 210 650800 tion is completed when the outlet temperature of the regener-
Pore volume (cm3 /g) 0.40.45 0.21 0.27 ation gas reaches 180190 C (TD ). Finally, cooling proceeds

Pore diameter (A) 22 26 45 from point D to E. The temperature of the cooling gas should
Design capacity 79 47 912 not decrease below 50 C, in order to prevent any water con-
(kgH2 O /100 kgdesiccant ) densation from the cooling gas [7]. Part of the processed NG
Density (kg/m3 ) 721 800880 690720 is sometimes used as the regeneration gas. Then it is cooled,
Heat capacity (Jkg1 K1 ) 920 240 200 and water condensed when it is separated. After H2 O sepa-
Regeneration temperature ( C) 230 240 290 ration, the regeneration gas is added back into the processed
Heat of desorption (J) 3256 4183 3718 stream.
Source: Tagliabue (2009), Gandhidasan (2001) So-called LBTSA (Layered Bed Temperature-Swing Ad-
sorption) processes are an upgrade of TSA method. Here, the
adsorption column is composed of several layers of different
The amount of adsorbed H2 O molecules increases with adsorbents. Hence, the properties of the separate adsorbents
the gas pressure and decreases with its temperature, which are combined in one column. For example, a combination of
are taken into account when the process parameters are de- silica gel with alumina is used in NG dehydration. Alumina
signed. Adsorption dehydration columns always work period- has better resistance to liquid water, so it is put in the first
ically. Minima of two bed systems are used. Typically, one place to contact the wet NG. This ordering prolongs the life-
bed dries the gas while the other is being regenerated. Regen- time of the silica gel, which is placed below the alumina layer.
eration is performed by preheated gas, as depicted in Figure 3.
2.3. Condensation

The third dehydration method employs gas cooling to turn

Figure 3. Scheme of the temperature swing adsorption dehydration process
The heater for TSA can be realized as an ordinary burner
or as a shell and tube heat exchanger warmed by steam or hot
oil. The regeneration gas flows through the adsorbent into a
cooler (usually using cold air) and then further into the sep-
arator. Most of the desorbed humidity from the adsorbent is
removed there. A downstream flow of wet NG through the
adsorption column is usually applied. In this way, floating
and channeling of an adsorbent is avoided. The regenera-
tion is performed by countercurrent flow in order to provide
complete regeneration from the bottom of the column, where
the last contact of the dried NG with the adsorbent proceeds.
The typical temperature course for regeneration of molecular
sieves is presented by Kumar (1987) [7]. The shape of the
curve representing the course of the outlet regeneration gas
temperature is typically composed of four regions. They are
H2 O molecules into the liquid phase and then removes them
from the stream. Natural gas liquids (NGLs) and condensed
higher hydrocarbons can also be recovered from NG by cool-
ing. The condensation method is therefore usually applied
for simultaneous dehydration and NGL recovery. NG can be
advantageously cooled using the Joule-Thompson effect (JT
effect). The JT effect describes how the temperature of a gas
changes with pressure adjustment. For NG, owing to expan-
sion, the average distance between its molecules increases,
leading to an increase in their potential energy (Van der Waals
forces). During expansion, there is no heat exchange with the
environment or work creation. Therefore, according to the
conservation law, the increase in potential energy leads to a
decrease in kinetic energy and thus a temperature decrease of
NG. However, there is another phenomenon connected with
the cooling of wet NG. Attention should be paid to the forma-
tion of methane hydrate. Methane hydrate is a solid in which a
large amount of methane is trapped within the crystal structure
of water, forming a solid similar to ice. The hydrate produc-
tion from a unit amount of water is higher than the ice forma-
tion. Hydrates formed by cooling may plug the flow. This is
usually prevented by injecting methanol or monoethylengly-
col (MEG) hydrate inhibitors before each cooling. Figure 4
depicts a dehydration method utilizing the JT effect and hy-
drate inhibition.
474 Michal Netusil et al./ Journal of Natural Gas Chemistry Vol. 20 No. 5 2011
The wet NG is throttled in two steps inside the flash tanks.
The lower temperature (due to the JT effect) of the gas stream
in the flash tanks leads to partial condensation of H2 O va-
pors. The droplets created are removed from the gas stream
by a demister inside the flashes. In cases when cooling by
the JT effect is insufficient (the usable pressure difference be-
tween the UGS and the distribution network is insufficient),
the air precooler and the external cooler are turned on. Since
dehydration is normally applied to large volume of NG, the
external coolers need to have good performance, so this type
of cooling is very energy expensive. However, if the usable
pressure difference is large, the JT effect inside the flashes
is so strong that internal heating of the flashes is required to
Figure 4. Scheme of dehydration method utilizing the JT effect and hydrate inhibition
Adsorption dehydration can obtain very low outlet water
concentration of Tdew <50 C, and contaminated gases are
not a problem. Even corrosion of the equipment proceeds at
a slow rate. However, adsorption requires high capital invest-
ment and has high space requirements. The adsorption pro-
cess runs with at least two columns (some lines use three, four
or as many as six). Adsorption columns are taller and heav-
ier than absorption contactors. The allowed flow velocity for
TEG contactors is approximately three times higher than that
for adsorption. This results in an approximately 70% larger
diameter of the adsorption column for the same amount of
processed gas. Industrial experience indicates that the capital
cost for an adsorption line is 23 times higher than that for an
absorption one [8].
Expansion dehydration is the most suitable method in
cases when a high pressure difference is available between
UGS and the distribution connection. However the difference
decreases during the withdrawal period and becomes in-
sufficient, so that an external cooling cycle is needed. A cycle
for hydrate inhibitor regeneration from the condensate sepa-
rated inside the flashes is also required.
3. Experimental
The energy demand for the methods presented here was
compared on the basis of a model, where a volume of
105 Nm3 /h NG from UGS was processed. The NG was H2 O
saturated at 30 C. The pressure of the gas was varied from
defreeze any ice that may form. A condensation method is ap-
plied when suitable conditions for the JT effect are available.
Each of the methods presented here has its advantages
and disadvantages. Absorption by TEG is nowadays the most
widely used method. Outlet Tdew around 10 C is usu-
ally reached. Indeed, with improved reboiler design (Vac-
uum stripping, Drizo, Coldfinger), the outlet Tdew is even 23
times lower. However, TEG has a problem of sulfur or gas
contaminated with higher hydrocarbons. The TEG in the re-
boiler foams, and with time it degrades into a black mud.
BTEX emissions (the acronym for benzene, toluene, ethyl-
benzene and xylenes) in the reboiler vent are a further disad-
vantage.
7 to 20 MPa, but in the case of the condensation method the
pressure range started at 10 MPa. The required outlet concen-
tration of H2 O in NG was equivalent to the dew point temper-
ature of 10 C at gas pressure of 4 MPa.
The calculation of TEG absorption was based on GPSA
(2004) [9]. The results were compared with the paper by
Gandhidasan (2003) [8] and with industrial data provided by
ATEKO a.s. The total energy demand was composed of heat
for TEG regeneration in the reboiler, energy for the pumps,
filtration and after-cooling the lean TEG before entering the
contactor. Enhanced regeneration was not considered. The
basic parameters for the calculation were: regeneration tem-
perature of 200 C, concentration of lean TEG of 98.5% and
circulation ratio of 35 LTEG /kgH2 O .
For calculating adsorption dehydration, molecular sieve
5A was considered to be the most suitable adsorbent. The to-
tal energy demand was directly connected to the regeneration
gas heater, and no other consumption was assumed. The cal-
culations were again based on GPSA (2004). The results were
compared with the papers reported by Gandhidasan (2001)
and Kumar (1987) [7]. The calculation procedure for GPSA
and Gandhidasan arose from the summation of the particular
heats, i.e. the heat for adsorbent warming, the heat for column
warming and the heat for water desorption. Kumars calcu-
lation procedure run differently. The regeneration step was
divided into four regions reproted in Ref. [7]. Afterwards we
determined what individual phenomena proceeded in each re-
gion, what the border and average temperatures were, and how
much energy was required to cover these phenomena. Finally,
 Journal of Natural Gas Chemistry Vol. 20 No. 5 2011
the demands for each region were added. The basic parame-
ters for all procedures were: temperature of the regeneration
gas of 300 C, time of adsorption/regeneration of 12 h and two
column designs.
The condensation method was calculated on the basis
of industrial data provided by TEBODIN s.r.o. and sup-
plementary calculations of the JT effect. The key parame-
ter influencing energy demand was the pressure of NG from
UGS. Because it was not feasible to apply this method under
low pressures, and the provided data start at 10 MPa, the pres-
sure range was adjusted. The total energy demand consisted
of the air pre-cooling unit, the external cooling, the pumps
for MEG injection and condensate off take, the heat for MEG
regeneration, and flash heating.
4. Results
The results obtained by TEG absorption method were the
same for each of the calculation procedures, and good agree-
ment with industrial data was also obtained. However, the cal-
culation procedures by the adsorption method led to different
results. Hence, the average energy demand value was taken
as the reference. The maximum deviation from it was below
20% for all the calculation procedures. The source of the de-
viation lies in the loss factor and the non-steady state factor.
In the case of the condensation method, the calculated values
for the JT effect were in good agreement with the industrial
data, but the amount of data was limited, resulting in limited
representation of the condensation method.
The final energy consumption results for each dehydra-
tion method are summarized and shown in Figure 5.
Figure 5. Final results of energy consumption for each dehydration method
Under low pressures (pressure of NG from UGS
<13 MPa), the condensation method was the most demanding
one. Its demand decreased linearly with pressure to a value of
145 kW for 13 MPa. At this point, the energy demand for the
condensation method was roughly the same as that for the ad-
sorption method. When NG pressure was further increased
from 13 MPa to 16 MPa, the energy demand for the conden-
sation method still decreased, but with a lowering tendency.
475
Under high pressures of NG (>16 MPa), the energy demand
of the condensation method was at its lowest, and it remained
nearly constant with an average value around 36 kW.
The courses of the energy demand for the adsorption and
absorption methods were quite similar: with increasing pres-
sure of dehydrated NG, the energy demand slowly decreased.
The absorption method was less demanding on the whole
pressure scale, and began with consumption of 120 kW at
7 MPa. The adsorption method started with 234 kW at 7 MPa,
but the energy demand decreased slightly more as the pres-
sure of NG in UGS was risen. This led to a gradual decrease
in the difference between these two methods, and the energy
demand at the final pressure values of 20 MPa were equal to
54 kW for absorption and 103 kW for adsorption.
5. Discussion
By far the highest energy demand of the condensation
method at low pressures of NG from UGS is due to the pres-
sure being close to the distribution pressure, so that pressure
cannot be used for the JT effect in flashes. Cooling is then
compensated by the air pre-cooler and the external cooling de-
vice, which are unsuitable for large volume of processed NG.
However, as the pressure difference between UGS and the dis-
tribution site increases, the space for expansion rises and the
JT effect proceeds with increasing impact. This is projected
into a linear decrease in the energy demand of the air pre-
cooler and the external cooling device. From the point where
there is a pressure of NG >14 MPa, flash heating is gradu-
ally turned on to prevent any freezing caused by the strong
JT effect. The energy demand of flash heating is reflected
in the total energy consumption. Finally, at pressures of
NG>16 MPa, total cooling and subsequent condensation are
achieved by the JT effect. The total energy demand, which
consists of flash heating and inhibitor injection and regenera-
tion, remains constant.
For adsorption and absorption dehydration methods, the
similar falling courses of the energy demand with increasing
pressure of NG can be explained by the fact that with increas-
ing pressure within a UDG, the amount of H2 O present in the
NG decreases. Generally, the absorption method consumes
less energy, because the regeneration of TEG is less demand-
ing than adsorbent regeneration. The composition of the total
energy demand of the adsorption method can be divided into
three parts. The heat for H2 O desorption, warming the ad-
sorbent and the column is approximately 55%, 31% and 14%,
respectively. It also has to be assumed that just part of the heat
in the regeneration gas transfers to the adsorbent, the column
and heat loss leave to the atmosphere, and the balance leaves
with the hot gas.
In brief, from the viewpoint of energy demand, the most
appropriate dehydration method in cases of high pressures is
the stored NG condensation method. This holds for NG from
UGS with pressure >15 MPa and distribution pressure re-
quirement of 7 MPa. Under lower pressures, the condensation
method is used if the objective is to recover NGL and remove
476 Michal Netusil et al./ Journal of Natural Gas Chemistry Vol. 20 No. 5 2011

water simultaneously. However, this is usually not the case BTEX benzene, toluene, ethylbenzene and xylenes
when storing NG in a UGS. In cases when insufficient pres-
sure difference is available, the absorption method is therefore References
favored over the adsorption method in terms of energy de-
mand. TEG absorption is nearly twice less demanding. How- [1] Gas infrastructure Europe [Internet]. Brussels: GSE
ever, if a gas contaminated with sulfur or higher hydrocarbons Map Dataset in Excel-format Storage map (version: Au-
is being processed, TEG in the reboiler foams and degrades gust 2010). [cited 2010 Mar 8]. Available from:
with time. This occurs when a depleted oil field is used as http://www.gie.eu/maps data/storage.html
a UGS. Adsorption is preferred in cases when very low T dew [2] NET4GAS [Internet]. Prague: Gas quality pa-
(H2 O concentration lower than 1 ppm) of NG is required, for rameters. [cited 2010 Mar 8]. Available from:
example, when NG is liquefied. http://extranet.transgas.cz/caloricity spec.aspx
[3] Bahadori A, Vuthaluru H B. Energy, 2009, 34(11): 1910
Abbreviations [4] Hubbard R A, Campbell J M. Hydrocarbon Eng, 2000, 5: 71
NG natural gas [5] Tagliabue M, Farrusseng D, Valencia S, Aguado S, Ravon, U,
NGL natural gas liquid Rizzo C, Corma A, Mirodatos C. Chem Eng J, 2009, 155(3):
UGS underground gas storage 553
bcm billion cubic meter [6] Gandhidasan P, Al-Farayedhi A A, Al-Mubarak A A. Energy,
Tdew dew point temperature 2001, 26(9): 855
TEG triethylene glycol [7] Kumar S. Gas Production Engineering. Houston: Gulf Profes-
MEG monoethylenglycol sional Publishing, 1987
TSA temperature swing adsorption [8] Gandhidasan P. Energy Sources, 2003, 25(3): 189
LBTSA layered bed temperature swing adsorption [9] GPSA. Engineering Data Book. 12th ed. Tulsa: GPSA Press,
JT effect Joule-Thompson effect 2004