Strategies for gas treatment with

adsorbents
Advances in regenerable, solid-bed adsorption processes for natural gas
dehydration and treating can improve selectivity and cut input energy costs
Scott Northrop ExxonMobil Upstream Research Company
Narasimhan Sundaram ExxonMobil Research & Engineering

dsorbents are widely used for contaminant

removal in natural gas treatment. Special
materials, such as molecular sieves, are
used to remove sulphur-containing impurities
like mercaptans, as well as water. Such operation
may be prone to problems, particularly during
the regeneration of these molecular sieves. For
example, unwanted byproducts like carbonyl
sulphide (COS) can be created by the transient
thermal waves produced during regeneration.
Improper regeneration may ultimately contribute to lowered product quality if the temperature
is too low, or time is too short. On the other
hand, excessive temperatures or heating times
may reduce adsorbent life, or cause adsorbate
decomposition (or coking) in some instances.
It is possible to take advantage of modified
cycling, as well as additional adsorbents (such as
alumina), to reduce and control the exposure of
molecular sieves to degradation. Such additional
adsorbents also allow targeting and removal of
other contaminants, including oxygenates like
methanol, that may promote undesirable side
reactions on valuable molecular sieves. Effective
compound bed configurations and process techniques to address the problems sometimes
encountered with conventional molecular sieve
treatment will be discussed.
The use of molecular sieve (MS) adsorbents
has become widespread for contaminant removal
from natural gas. Over the past three to four
decades, they have proven to be suitable and
economic in reducing contaminant levels during
gas processing. However, with increasing treatment require-ments, including more stringent
product/effluent specifications, greater plant size

www.digitalrefining.com/article/1000625

and increasingly challenged feedstocks, there is

considerable incentive to improve MS performance. Simultaneously, decades-long experience
from operating plants has begun to reveal limitations in the conventional use of these materials.
With such guidance from both the past and the
future, MS end users and manufacturers are
compelled to develop effective strategies for
enhancing the use of these materials and maximising their potential.
Key limitations that are recognised in conventional MS operating practice include deactivation
via exposure to certain contaminants, sub-optimal regeneration (eg, refluxing), unwanted
byproducts during regeneration, and lowered
adsorbent life due to increased regeneration
frequency and severity. Our strategy addresses
these limitations using enhanced adsorbent
combin-ations (compound beds), modified thermal and/or pressure cycles and modified partial
desorption regeneration,1 or combinations
thereof. Patents are being sought on some of
these combinations. We are also exploring more
selective adsorbents to reduce hydrocarbon loss
and regeneration energy requirements.

Delaying molecular sieve degradation

The motivation to reduce adsorbent fill cost and
to increase the lifetime of solid-bed dehydrators
has led to the consideration of alternative
adsorbents, including their use in conjunction
with compound beds. Compound (or mixed)
beds consist of two or more layers of different
types of MS for water and mercaptan sulphurs,
for example. Figure 1 shows a schematic diagram
of such a compound bed of adsorbents, where

Gas 2009 1

A1, A2 and A3 represent differtop of these compound beds of

Feed
ent solid materials. Adsorbents
MS is known to protect the MS
can be MS such as 4A (often
from unexpected liquid (hydroused for dehydration) or 13X
carbon,
brine,
glycol
or
(used for the removal of larger
condensed water) carry-over
molecules), or other zeolitic
into the bed from upstream
A1
adsorbents. Additional layers
separation facilities sometimes
of specific adsorbents may also
as aerosols. Such fouling probbe used for trace mercury
lems can be reduced by
A2
removal, for example, if the
well-designed separation equiptreated gas is to be fed to an
ment upstream of the MS bed.
LNG plant. Additional layers
The AA can also protect the MS
A3
can also offer significant
from refluxing, when water
protection of sensitive molecucondenses in the upper, cooler
lar
sieves
from
both
part of the bed during the early
contaminants and damaging
stages
of
regeneration
Product
conditions that might result
(discussed below). AA, because
from upsets during the feed
of its higher water capacity, also
cycle, or refluxing during the
helps to ensure that the heavy
regeneration cycle. Compound Figure 1 Schematic of mixed-bed
compound (eg, mercaptan)
beds are well known in the dehydrator
removal specification will be
air
separation
prepurifier
met by reducing the chance of
industry.2
water breakthrough from a compound bed
Typically, contaminated feed gas enters from consisting of 4A MS and 13X MS, for example.
the top of the bed during the adsorption cycle. As Even a relatively small amount of water breakthe bed adsorbs contaminants over the course of through into the 13X MS will reduce the
the service cycle, it ultimately must be taken off- mercaptan removal capacity substantially
line prior to exceeding treated gas specifications. because of the adsorbents greater affinity for the
Usually, another bed is placed in service, and the more polar compound. 13X capacity cannot be
spent bed is regenerated. During the regeneration restored until the water is desorbed.
cycle, dry, heated gas flows in from the bottom of
There are a number of benefits to using an
the bed and progresses upwards through the AA/MS compound bed, including lower overall
packed bed. A thermal wave moves up through adsorbent cost, and higher resistance to liquid
the bed as contaminants are convected out. In upsets and liquid carry-over, as described above.
general, there is a lag between the arrival of the AA can also have a higher equilibrium capacity
thermal front and the arrival of the mass transfer than MS when the feed gas is near saturation, as
front. This can create conditions for unwanted shown by the isotherms in Figure 2.
side reactions.
Regeneration temperatures required for AA
The contaminant-saturated regenerated gas is are, furthermore, lower than those of MS, thus
usually cooled to condense water and heavy reducing the heat requirement to desorb the
hydro-carbons. If other contaminants are water in the AA relative to MS. This lowered
present, the regeneration stream must be treated energy requirement translates to savings in
in some manner. For example, the stream may steam or fuel gas, depending on how the regenbe washed with physical solvent to absorb eration gas is heated. The benefits of compound
sulphur compounds, so that the treated gas may beds for natural gas treatment have been recogbe used for fuel. Due to the unsteady nature of nised by others in the industry, and compound
the regeneration process, a peak of contaminants beds of AA/MS have been used in commercial
is generated. Generally, the solvent system must natural gas dryers at several installations for
be modified to provide a fairly steady concentra- decades.2-7
tion of contaminants to the next process (sulphur
recovery unit (SRU), for example).
Dehydration of sweet gas
The addition of activated alumina (AA) to the A particularly undesirable limitation of

2 Gas 2009

www.digitalrefining.com/article/1000625

.OTOSCALE

Mol sieve

7ATER

Water loading

No to scale

Activated
alumina
Partial pressure H2O

245C
280C
0ARTIALPRESSURE(

Figure 2 Adsorption isotherms for water on MS and water

on AA

Figure 3 Adsorption isotherms for MS at elevated

temperatures

conventional MS usage occurs if there is uncontrolled water condensation during regeneration.

At the start of the regeneration cycle, the MS at
the top of the bed is already near its maximum
water adsorption capacity (for that temperature
and partial pressure of water) because it has
been in drying service cycle for some time. As
the regeneration thermal wave moves up from
the bottom of the vessel, water is driven off the
heated adsorbent and carried in the gas phase.
As the hot gas moves up through the cooler part
of the bed, it cools and may reach the point of
water saturation, particularly if the regeneration
is done at high pressure. Since the remaining
incremental capacity of the MS is small (for fixed
temperature), the water may condense, which
can be damaging to the binder of the sieve particles. The condensed water can then be vapourised
by the advancing thermal wave, potentially causing pressure build-up in the MS particles. If the
rate of temperature rise is too rapid, the particles
may be damaged. Also, if the bed is internally
insulated, it may be possible for liquid water to
get behind the insulation. In sour service, this
could create corrosion issues.
In the worst cases, liquid water may condense
and start flowing down the bed and be revapourised by the advancing regeneration thermal
wave, a phenomenon known as refluxing. The
ultimate result is often a doughnut-shaped
region of powdered MS material at the top of the
adsorbent bed. AA, on the other hand, is generally resistant to damage by liquid water.8
One means that has been suggested to reduce
the effect of refluxing is to use alternate types of
molecular sieves that are mechanically stronger
and more resistant to disintegration.9 Like the

use of AA, however, they do not actually address

the root cause of the problem.
There are alternative methods to address
refluxing. One example is to more slowly ramp
the temperature of the regeneration gas over the
course of the cycle. This will lengthen the regeneration step somewhat, reducing effective
capacity over the course of the full cycle.
However, this is generally preferable to premature ageing and unscheduled change-out of the
MS.
In many cases, the regeneration heater is
immediately activated to maximum temperature
(typically around 315C) and held at that maximum throughout the duration of the regeneration
step. While this minimises the time for regeneration, it also maximises the severity of conditions
for the adsorbent, as well as ancillary equipment
like switching valves, support screens and internal insulation (if present).
In the simplest cases, it may be possible to
reduce the final regeneration temperature. While
a small amount of net adsorption capacity may
be lost (see Figure 3), and the cycle time may be
increased slightly, the decreased rate of water
desorption will tend to decrease the dewpoint of
the gas at the top of the bed, reducing the chance
for reflux. The method has been tested at one of
our many field installations. A reduction of up to
35C in regeneration temperature was possible
(245C vs 280C), with no apparent immediate
effect on the drying cycle. This also has the
benefit of reducing undesirable side reactions, as
well as reducing total energy requirement. One
possible means of countering the increased
regeneration cycle time would be to increase the
hot gas flow rate. However, velocities are limited

www.digitalrefining.com/article/1000625

Gas 2009 3

to those that will not cause lifting of the bed.

Another possible means of handling the simple
reflux case would be to add external heat to the
top end of the vessel when regeneration is begun.
This will not necessarily fully protect the MS in
the middle (in a radial sense) of the upper
portion of the bed, but it will address the area of
greatest thermal inertia, and could prevent the
formation of the ring of disintegrated MS that is
sometimes observed when opening a spent bed.
However, one must be careful that water does
not condense out on carbon steel piping downstream when it is exposed to sour gas.
Even when refluxing does not occur, molecular
sieves are subject to hydrothermal ageing during
normal operation. Since the loss of activity can
be substantial after a three- or five-year lifetime,
the beds are generally designed for end-of-run
(EOR) conditions. Thus, at start-of-run (SOR)
adsorbent conditions, extra capacity in the
design can be used to advantage. One alternative
is to simply run for a longer period in drying
service, rather than for a fixed EOR time.
Another possibility is to utilise lower regeneration temperatures at SOR conditions. Some
residual capacity may be lost during these cycles,
but the severity of regeneration is reduced. This
will increase the total number of cycles achieved
before EOR conditions are reached. Of course,
the regeneration temperatures may have to be
adjusted upwards over the course of time to
compensate for the hydrothermal ageing that
eventually occurs.
Ultimately, there is a trade-off between milder
regeneration conditions, which reduce operating
expenditure (through decreased material changeout frequency and reduced energy), and
increasing capital expenditure. Lower regeneration temperatures reduce net capacity slightly, so
a corresponding increase in bed size is needed to
offset this, increasing capital expenditure. If the
added volume of material requires an additional
dehydration vessel, then the required capital
expenditure clearly exceeds any incremental
operating expenditure savings obtained through
energy savings and increased adsorbent life. For
existing installations at SOR adsorbent conditions, lagniappe in the design can be used to
advantage, since facilities are generally designed
for EOR conditions, when adsorbent activity is
decreased after many cycles. By using lower
temperatures at SOR conditions, one can

4 Gas 2009

increase the total number of cycles achieved

before EOR conditions are reached.
If only dehydration of a sweet gas stream is
required, a common means of avoiding the negative effects of gas-phase impurities (like BTX) is
to use small-pore zeolites (3A or 4A).10
Mercaptans, heavy hydrocarbons and other large
molecules are not able to fit into the small pores.
Thus, they are not present in large concentration
when the thermal wave arrives during the regeneration cycle. Thus, degradation via coking is not
generally an issue (although hydrothermal ageing
may be).

Sour service
It has been long observed4,11 that CO2 and H2S
can react to form COS according to
H2S + CO2 ---> H2O + COS

This reaction is particularly likely to occur at

the elevated temperatures of regeneration.
For dehydration-only service (ie, no H2S
removal), it is possible to use a small-pore zeolite
(3A) MS in order to exclude the reactants from
the pore space.12 Alternatively, it has been shown
that 5A MS catalyses COS formation to a lesser
degree than 4 A MS, since it is less basic than 4A
MS.11

Heavy hydrocarbon removal

In some cases, it is desirable to remove heavy
hydrocarbons from a sour gas stream. We have
successfully used silica gel in this application to
reduce the amount of heavy carbon going to a
solvent-based acid gas removal process.13
Interestingly, regeneration temperatures had to
be lowered to reduce the effects of sulphidation
in the regeneration gas heater coils.

Mercaptan removal
For those services requiring removal of
mercaptans, larger pore-size molecular sieves
(like 13X) are typically used. Heavy hydrocarbons (like BTX) and other impurities (like
methanol) will be co-adsorbed along with the
mercaptans. Consequently, it is possible for these
compounds to crack and form coke if the thermal wave outruns them during regeneration. The
decomposition of mercaptans on 13X has also
been reported.14 Thus, if the impurities are not
desorbed and flushed out before the thermal

www.digitalrefining.com/article/1000625

-ETHANOLINPRODUCTPPMV

wave arrives, they may react in

some fashion, possibly catalysed

by the solid itself. Again,
lowered regeneration temperatures will tend to reduce these
undesirable side reactions.

Another means to reduce
decomposition reactions is to
Mole sieve
ramp the temperature more
only
slowly during the initial phase
of the regeneration. This allows

more time for the molecules to
desorb, diffuse and be convected
Mole sieve
out of the bed before the high
plus alumina
temperature front arrives. A

variation of this theme is to
bring the temperature up to
some intermediate value and
hold it there for some time, then










ramp to the final temperature.
7ETREGENERATIONSTEPTEMPERATUREPPMV WATER#
Ramping, or holding at intermediate temperatures, also
reduces the thermal stress Figure 4 Model results showing methanol in product with and without
placed on the system, but alumina layer
increases the regeneration cycle
time. An alternative to thermal cycling is pres- regeneration stream. The advantages of this
sure-swing cycling, where regeneration is done approach in comparison to using only MS (the
at reduced pressure.
conventional approach) are shown in Figures 4
Of course, any modification to regeneration and 5. These figures were generated with the aid
cycles requires the buy-in of operations, as of an in-house computational simulator of cyclic
adjustment of the regeneration stream clean-up adsorption processes, which derives data from
system may also be needed. It is ideal to have ExxonMobils (XOM) Upstream, Downstream
the material vendor in agreement with the and Chemical companies.
changes, but this may not always occur.
Figure 4 illustrates that the use of AA in the
bottom layer (mole sieve plus alumina points
Compound beds to reduce coking
on the graph) allows for a substantially lower
To minimise the adverse effects of the cracking regeneration temperature than MS alone, while
of heavy components, one strategy is to use still meeting a <1 ppm methanol specification in
compound beds (ie, MS and desiccant). This the product stream after a typical service time.
could include the use of alumina desiccant in the The product specification will not be met for the
product zone; for example, in Zone A3 in Figure mole-sieve-only case for regeneration tempera1. By doing so, one could use a regeneration tures less than 200C. (Decomposition effects
stream with lower temperature, and also a regen- were not considered.)
eration stream that contains a small amount of
Figure 5 shows the potential effect of allowing
water vapour that may suppress coking reac- the wet regeneration to go for too long. (Total
tions.14 The addition of water vapour to the regeneration cycle time is 124 minutes.) If residregeneration stream for a short period may act ual water is left on the MS, product quality will
to desorb heavy molecules (BTX and mercaptans) suffer. This is particularly true for the mole
before the arrival of the thermal wave. In effect, sieve-only case.
it may be a modified desorptive,1 or displacement
type of regeneration.15 The use of alumina Alternative adsorbents
controls the rate at which water is added to the An entirely new class of adsorbents for natural

www.digitalrefining.com/article/1000625

Gas 2009 5

mixed-metal oxide catalysts

used as adsorbents. This
reduces the thermal energy
(steam or direct fire) required
to heat the regeneration gas,
0.75
thereby reducing emissions.
All mole sieve
Reduced severity of regeneration also prolongs the life of the
0.50
adsorbent. In addition, we are
developing modified pressure
swing cycling, including variable
0.25
frequency cycling for a variety
of applications to eliminate
Mole sieve
plus alumina
hydrothermal effects altogether.
0.0
We have also employed the
0
10
20
30
40
50
60
analogy between adsorption
Duration of wet regeneration at 127C, min
from liquids and adsorption
from unsaturated vapours to
Figure 5 Model results showing improved methanol removal with alumina
identify newer, effective adsorblayer
ents from refining operations.
Figure 6, for example, depicts a
gas purification is currently being tested. These typical breakthrough curve in the liquid phase of
adsorbents are based on mixed-metal oxide cata- sulphur-containing compounds obtained when
lysts, which are composed of mixtures of alumina hydrocarbons laden with sulphur are passed
and/or silica substrates with quantities of oxides through a bed of CoMo hydrotreating catalyst.
of metals, including cobalt and molybdenum, The analogy between gas-phase and liquid-phase
among others. Our testing has determined that adsorption can be used to show the enhancethese materials have reasonable capacity for the ment expected in equilibrium capacity. (Note
principal contaminants found in natural gas. that the early breakthrough is due to the presTesting of both fresh and sulphided forms of ence of thiophenes, which are not generally
these catalysts indicates that it may be possible present in appreciable concentration in gas
to recycle catalysts after their use in refining streams.)
operations such as hydro-treating and hydrofinLiquid-phase adsorption is well known to
ing. (The material would likely require coke possess limitations due to lower diffusivities
removal using a standard decoking procedure (lower mass transfer) and higher densities, which
prior to its new life as an adsorbent.) The spent also promote capacity-reducing adsorption noncatalysts also show reduced uptake of non- idealities. The capacities obtained in the
sulphurous
components
(ie,
heavier liquid-phase experiments depicted in Figure 6
hydrocarbons). This feature also makes these indicate good capacity using the analogy of
catalyst materials attractive compared to conven- adsorption from liquids and adsorption from
tional natural gas treatment molecular sieves. gases. The analogy shows that the enhancement
The latter are generally indiscriminate in the of adsorption from the binary gas vs that from
uptake of all heavy components of natural gas. binary liquid is possible when the mixed gas
Selective adsorbents would not remove as many isotherms are compared. The ratio of gas-phase
valuable heavy hydrocarbons from the feed mixed Langmuir loading and liquid-phase
stream. These valuable molecules would then not Langmuir surface excess can be derived as:16
end up in the SRU, where they consume a lot of
c1P
(1+c2P2S)
air (and, if not completely destroyed, soot up the G =
S
S
L
(c
P
c
P
)
(1 + c2P2)
SRU catalyst).
1 1
2 2
Test results have also shown the possibility of
using regeneration techniques of reduced severGas-phase testing of these adsorbents has not
ity (temperature or duration) with the yet been done. It is recognised that these materiMethanol in product, ppmv

1.00

6 Gas 2009

www.digitalrefining.com/article/1000625

Sulphur in effluent, ppmv

als are used in hydrogenating

services, and that they may
become sulphided during
regeneration with sour gas. It
Total sulphur
is also recognised that special
handling of sulphided forms
may be required. However,
Total mercaptan S
given that the materials themselves are essentially free of
charge (spent catalyst that
would
otherwise
require
disposal), and given the
Time
potential for much greater
selectivity towards sulphur
compounds, these materials Figure 6 Breakthrough curves for CoMo adsorbent in liquid hydrocarbon service
warrant further evaluation.
Francis Wu.

Improved control techniques

Optimising mixed-bed technology is naturally
improved with increased instrumentation and
monitoring. Temperature indicators and moisture analysers can reveal residual water loading
capacity as the MS adsorbents degrade. The
adsorption and regeneration cycle times can then
be adjusted to accommodate the remaining overall bed capacity (ie variable cycle operation).6 A
moisture analyser placed near the bottom of the
upper adsorbent layer could gather AA capacity
data over time, and a thermocouple assembly
taking temperature measurements throughout
the height of the bed could be used to check for
regeneration conditions that generate water
refluxing. At a minimum, a thermocouple at the
inlet (top) of the bed could be helpful in ascertaining the arrival of the thermal front during
regeneration. In the case of more sophisticated
regeneration programs, such a thermocouple
could help to optimise the cycles. Other control
improvements are described elsewhere.17

Conclusions
Given the importance of regenerable, solid-bed
adsorption processes for natural gas dehydration
and desulphurisation, work is being done towards
the continuous improvement of these processes.
Compound beds, improved regeneration techniques and new classes of adsorbents can reduce
energy requirements, and potentially improve the
selectivity of the process.
Acknowledgement
This paper is dedicated to the memory of our friend and colleague,

www.digitalrefining.com/article/1000625

References
1 Ruthven D, Principles of Adsorption Processes, John Wiley
and Sons, 1984.
2 Smith D L, Optimize solid bed adsorption systems,
Hydrocarbon Processing, May 1996, 129.
3 Blachman M, McHugh T, Sour gas dehydration technology and
alternatives, Proceedings of the Laurance Reid Gas Conditioning
Conference, 2000.
4 Campbell J M, Gas Conditioning and Processing, 8th ed., 2,
2000, 400, 402, 406, fig. 19.2, 19.5.
5 Kohl A L, Nielsen R B, Gas Purification, 5th ed., 1997, 1049,
1070.
6 Musamma S, Musamma A, Molecular sieve performance
enhancement in Badak LNG, LNG Journal, July/August 2002, 26
28.
7 Axens brochure on Multibed technology, Appendix C.
8 Veldman quoted in Gas Purification, 5th ed., 1997, authors
Kohl A L, Nielsen R B, 10491050.
9 Rastelli H, Shadden J, Extending mole sieve in natural gas
dehydration units, presented at the 86th Annual GPA Convention,
2007.
10 Malina H M, Fundamentals of adsorptive dehydration,
Proceedings of the Laurance Reid Gas Conditioning Conference,
2000.
11 Turnock P H, Gustafson K J, Advances in molecular sieve
technology for natural gas sweetening, Proceedings of the
Laurance Reid Gas Conditioning Conference, 1972.
12 Trent R E, Craig D F, Coleman R L, The practical application of
special molecular sieves to minimize the formation of carbonyl
sulphide during natural gas dehydration, Proceedings of the
Laurance Reid Gas Conditioning Conference, 1993.
13 Iyengar J, Sibal P, Clarke D, Operation and recovery
improvement via heavy hydrocarbon extraction, Proceedings of
the Laurance Reid Gas Conditioning Conference, 1998.
14 Carlsson A F, Last T, Rajani J B, How to avoid excessive mole
sieve deactivation when used for mercaptan removal, presented
at the 84th Annual GPA Convention, 2005.

Gas 2009 7

15 Trent R E, Dehydration with molecular sieves, Proceedings of

the Laurance Reid Gas Conditioning Conference, 2004.
16 Sircar S, Myers A L, J. Phys. Chem., 1971, 76, 23, 3415.
17 Parker D T, Karber H D, Control strategy for molecular sieve
dehydrators, Proceedings of the Laurance Reid Gas Conditioning
Conference, 1998.
Scott Northrop is Engineering Associate at ExxonMobil Upstream
Research Company, Houston, Texas.
Email: scott.northrop@exxonmobil.com
Narasimhan Sundaram is Adsorption Specialist, Downstream and

8 Gas 2009

Upstream, ExxonMobil Research & Engineering, Fairfax, Virginia.

Email: narasimhan.sundaram@exxonmobil.com

Links
More articles from: ExxonMobil Research and
Engineering
More articles from the following category:
Gas Processing/Treatment

www.digitalrefining.com/article/1000625