THE EVALUATION AND DESIGN OF LPG TREATERS
Jenny Tse
and
Jesse Santos
The Dow Chemical Company
GAS/SPEC Technology Group
Freeport, Texas
Presented at
43rd Annual Laurance Reid
Gas Conditioning Conference
March 1-3, 1993
Norman, Oklahoma
INTRODUCTION
Liquid-liquid extraction is a process used in the purification of liquefied petroleum
gas (LPG) or natural gas liquid (NGL) streams. Purification involves the removal
of acid gas contaminates such as hydrogen sulfide (H2S), carbon dioxide (CO2),
and other sulfur species from liquid hydrocarbon streams containing C2 to C4
compounds. One common method of acid gas removal is by extraction with an
aqueous caustic solution that is discarded when H2S or CO2 is removed. However,
this method is expensive because of the associated chemical and disposal costs. An
alternative method is extraction with an alkanolamine that forms a weak regenerable
salt. These alkanolamines include monoethanolamine (MEA), diethanolamine
(DEA), methyldiethanolamine (MDEA), and formulated MDEA solvents.
This paper reviews the criteria for designing LPG treaters which use alkanolamines
as extraction solvents. Plant case studies of typical LPG treater designs will provide
insight into the physical and chemical aspects dominating the overall mass transfer
in these systems. Finally, these criteria will be employed to troubleshoot an LPG
treater.
OVERVIEW OF LIQUID EXTRACTION WITH ALKANOLAMINES
"Liquid-liquid extraction consists of the following steps:
(1) intimate contacting of solvent with the component to be extracted so that
the solute will be transferred from the solution to the solvent and
(2) separation of two immisible phases."[1]
In the case of LPG treating, the acid gases (the solutes) are removed from the LPG
by contacting the solution with an amine (the solvent). LPG treating can be
considered a special case of extraction where the mass transfer is accompanied by
chemical reactions. This phenomenon is composed of the following steps [1]:
(1) Diffusion of the solute from the bulk of the organic phase to the organicaqueous interface.
(2) Diffusion of solute from the organic-aqueous interface to the bulk
aqueous phase.
(3) Chemical reaction within the aqueous phase.
(4) Diffusion of the reaction products due to the concentration gradient.
FLOW CONFIGURATION
In LPG treating, the aqueous amine phase is usually the continuous phase, while
the organic LPG phase is usually the dispersed phase. The aqueous amine solution
is fed to the top of the column, while the LPG feed enters from the bottom. Once
contact is complete and the phases separate, the rich amine containing the acid gases
will exit the bottom of the column as the extract. The raffinate, or the treated LPG,
will exit from the top of the column. Figure 1 illustrates the flow configuration.
Figure 1. Flow configuration for LPG treating with an
amine.
The selection of the LPG as the dispersed phase and the amine as the continuous
phases provide several advantages in terms of mass transfer:
(1) When the phase with the highest flowrate is dispersed, the total surface
area available for mass transfer is maximized. [2,3]
(2) Dispersion of the LPG into the higher viscosity amine phase reduces the
terminal velocity of the droplets which increase the droplet residence
time and hence the mass transfer.
EQUIPMENT SELECTION
Three types of devices are commonly used for liquid extraction with alkanolamines.
The devices include packed columns, sieve tray columns and static mixers. Of
these, the packed and sieve tray columns are the most prevalent, but static mixers
have been used successfully in a number of applications.
The selection of equipment depends on the treating requirement and the specific
needs of the operation. Static mixers have been used successfully in a number of
application when less than one theoretical stage is required and unit size is crucial.
Packed columns and sieve tray columns are used when 2 to 7 theoretical stages are
needed.[3] Each devices offer its own unique advantages and disadvantages.
Laddha and Degaleesan [1] in addition to Treybel [2] have presented reviews on
each of these devices. This paper will focus specifically on the design of packed
and sieve tray columns.
AMINE SELECTION
The selection of the amine depends on the type of acid gas components that need to
be removed and the specification required. A primary amine such as MEA has a
high affinity for acid gases. This property makes MEA the better choice in
applications where both H2S and CO2 must be removed to very low levels or
where the acid gas partial pressures are extremely low. One drawback, however, is
that MEA reacts irreversibly with carbonyl sulfide (COS) to form nonregenerable
degradation products that reduce the acid-gas removal capacity of the amine. In
Gases with high concentrations of COS, a secondary amine such as DEA should be
considered because it can remove the COS without forming the nonregeneable
products.[4] If the outlet specification cannot be met with a secondary amine, a final
caustic scrub of the LPG may be required to remove the last traces of COS.
MDEA should be used when H2S selectivity and little COS removal are required.
Finally, formulated MDEA solvents can be used in varying applications.
The choice of solvent may also be influenced by other factors including energy
requirements and solution corrosivity. Several references offer reviews on these
topics. [5,6] In general, MEA is more corrosive and requires more energy for
regeneration than DEA and MDEA.
PACKED COLUMN DESIGN
Column Diameter and Capacity
Column diameter or capacity is determined by the maximum hydraulic flow through
the column or the flood point of the tower. Several investigators, including
Crawford and Wilke [7] have reported correlations to estimate the flooding
velocities in a packed column. In Table 1, the actual and design velocities of three
operating LPG treaters were compared to the predicted flooding velocities. A
modified Crawford-Wilke correlation [8] was used in this evaluation to predict the
column capacity. The recommended design criterion for this correlation is 20% of
predicted flood.[10] Commercial experience with systems removing H2S in LPG
treaters confirmed the predicted flooding point. [3]
The results from Table 1 indicate that the LPG treaters chosen are operating at a
fraction of the predicted flood, ranging from 5 to 22%. Even though Plant 2 and
Plant 3 have exceeded their original design flow rates, the calculations show that
these treaters may still be operating well below flood. The calculation also indicates
that Plant 1 is approaching maximum capacity with the dispersed phase at 22% of
predicted flood velocity. Additional data need to be gathered from Plant 1 to
confirm its capacity.
While the above correlations should be used for design calculations, a rough
estimate of the column diameter or the column capacity can be determined by the
combined amine and hydrocarbon flow rate. Literature references recommend that
for design the combined flow should not exceed 15 8pm per square foot of column
cross sectional area.[9] A survey of several plants in Table 1 indicates that the actual
combined flow ranged from 16.2 to as high as 35.6 gpm/ft 2. Both Plants 2 and 3
have original design flow of less than 15 gpm/ft 2. Plant 1, however, has a design
flow of 55.7 gpm/ft 2. Plant 1 is of particular interest because the treater has never
operated at its designed capacity. Excessive amine
carryover was reported at an amine circulation rate of 200 8pm with an equivalent
flow to diameter ratio of 37.6 gpm/ft 2.
Table 1. Operating and design conditions of three packed LPG treaters.
Type and Size of Packing
The function of packing is to increase the mass transfer rate in the column. The
packing accomplishes this function in three ways. First, it serves to increase the
residence time of the droplets in the column. The total mass transfer is increased if
the droplets must travel the tortuous paths of the packing interstices. Second, the
packing reduces the likelihood of back-mixing within the column. Third, the
packing will distort, coalesce and re-disperse the droplets to enhance internal
circulation and refreshes the film surface for mass transfer.[3]
The types of packing generally selected for LPG treating are either metal or
ceramic. Ceramic packings are preferentially wetted by the aqueous amine solution,
while metal packing will be wetted by either the aqueous or organic phase
depending on initial exposure of the metal surface.[3] Since the dispersed phase is
the LPG and the continuous phase is the amine, the packing should be wetted by
the continuous phase. The packing should not be wetted by the LPG phase because
the droplets will coalesce on the surface of the packing, reducing the efficiency.
Plastic packing will also sacrifice the LPG droplet integrity and is therefore never
specified for LPG treaters.
The packing size is selected on the basis of efficiency and capacity. If the interstitial
void is large, the efficiency will decrease due to channeling at the walls of the
tower. Conversely, if the packing is too small, the capacity of the column is
diminished. For optimum performance, the packing size should be greater than the
critical packing size (dFC). Calculation for the critical packing size can be found in
Perry's Chemical Engineers' Handbook [10]. The critical packing size can be
calculated from the following relation developed by Gaylor and Pratt:
Packing larger than the critical size will allow the droplets to move freely within the
packing interstices, independent of the flowrate.[1,3,10] Otherwise, if the voids
are too small, the packing will facilitate the coalescence of the droplets into larger
drops and reduce the column capacity.
For the cases listed in Table 1, the critical packing size is less than 1/2 inch.
Therefore, the packing should be greater than 1/2 inch. However, for commercial
applications, packing usually range from 1-1/2 to 2 inches. Packing size should be
no greater than 1/8 the tower diameter to prevent channeling. [3,10,11]
TRAYED COLUMN DESIGN
The following section gives some design criteria for sieve tray LPG treaters.
Examples of three operating LPG treaters are given in Table 2. For Plants 1
and 2, not enough data was available to compare the operating conditions with
these criteria. However, adequate data was available for Plant 3. The
troubleshooting section of this paper will give a detailed comparison of the
operating conditions of Plant 3 with these theoretical design criteria.
Table 2. Operating design conditions of three sieve tray LPG treaters.
Capacity and Flood Point
The capacity of sieve tray columns is also limited on consideration of the flood
point. Flooding occurs when the flow rate of the continuous phase is so high
that the dispersed phase accumulates under the plates. If the height of the
coalescing layer exceeds the length of the downcomer, entrainment of the
dispersed phase into the downcomer will occur. When the thickness of the
coalescing layer approaches the plate spacing, the treating capacity of the LPG
treater will be diminished.
Tray Design
Unlike a packed column in which dispersion and coalescence occur somewhere
within the packing, dispersion and coalescence occur at every tray in a sieve tray
column. Therefore, for best efficiency, the sieve trays should be designed such that
the droplets will break cleanly from the hole to form the proper sized droplets.
Mayfield and Church recommend that the liquid be allowed to jet and then form
droplets a slight distance away from the holes.[12] The recommended average
velocity throughout the holes should be between 0.5 to 1.0 ft/s. [10]
To ensure that the droplets do not coalesce after they are formed, the pitch, or the
distance between the holes must be kept 3 to 4 hole diameters apart.[1] The holes
are usually 1/8 to 1/4 inch in diameter, set 1/2 to 3/4 inch apart, on square or
triangular pitch.[1,10]
Tray Spacing
The tray spacing is based on the height of the coalescing layer. The accumulation of
the dispersed liquid is calculated by the pressure drop required for the movement of
the liquid through the holes. The total minimum liquid head required may be
expressed by the following equation [1]:
ht = hN + hg+ hc
where, ht = total minimum head required for movement of dispersed
phase, ft
h N = head required to overcome the friction through the
holes, ft
hg = head required to overcome interfacial effects, ft
hc = head required to overcome the effects of the flow of the
continuous phase, ft.
Detailed calculation of each parameter can be found in Reference [1] and [5]. Once
the dispersed phase thickness is determined, the tray spacing can be calculated. For
commercial units, 10 to 12 times the coalescing layer height is adopted to provide
flexibility and to allow for the location of the manhole. In actual practice, the tray
spacing range from 6 inches to 2 feet. [1]
LPG/AMINE DISTIRIBUTORS AND RE-DISTRIBUTORS
In both packed and sieve tray columns, proper distributions of the phases affect
the tower hydraulics and performance, as well as the amine and LPG losses.
Well-designed distributors will evenly disperse the liquids to provide good
contact, while minimizing both losses due to entrainment and the formation of
emulsions. While many types of designs are available, the following discussion
will be limited to the simplest orifice type distributors used in packed columns.
Proper distribution of the LPG is especially critical for packed columns to avoid
mal-distribution and poor efficiency. For packing larger than the critical packing
size, the droplets have a characteristic drop size when the liquids are in
concentration equilibrium.[1,10] The characteristic drop size is given by the
following equation.[10]
If the droplets formed by the distributor are too large, the packing will
breakdown these droplets until they reach their critical size.[1] Extremely large
droplets will reduce the efficiency of the column by decreasing the surface area
available for mass transfer. On the other hand, if the droplets formed at the
distributor orifices are smaller than the characteristic drop size, there may be a
tendency to flood until the drops grow to size.[10] Stable emulsion may also
form if the droplets cannot quickly coalesce.
The droplets' sizes are controlled by the velocity in which the LPG is dispersed
and the size of the distributor orifice. The velocity at the orifice should be limited
to 70 ft/min
(1.25 ft/s). Orifice sizes of 0.19 to 0.25 inches in diameter are commonly used,
although sizes as small as 0.14 to 0.31 inch diameter have also been successful. If
the dispersed phase superficial velocity exceeds 130 ft/min, the number of orifices
should be increased rather than the orifice size.[3]
Flooding of the tower caused by poor dispersed phase hydraulics at the bottom of
the packed bed can be avoided in two ways. The dispersed phase distributor should
be embedded in the packing itself to avoid liquid hold-up at the bottom of the
packed column.[3] However, if the column requires more than one bed, a
redistributor and a second bed disperser plate should be designed to assure
dispersed phase hold-up will not occur between the beds. A different way of
approaching this problem is to couple the primary distributor with a
disperser/packing support plate. While the distributor's function is to introduce the
dispersed phase across the disperser/packing support plate, the disperser plate
should be designed to develop the final droplet size. If this is done, risers should be
placed on the distributor to extend to about 1 inch above the downcomer exit on the
disperser plate for the continuous phase. Redispersion after 6 to 10 feet of packing
is recommended to improve column efficiency.[3]
The amine distributor should be designed to minimize turbulence which may disrupt
phase separation at the top of the column. The continuous phase velocity should be
no greater than 170 ft/min (2.8 ft/s). If the superficial velocities-of the continuous
phase exceed 60 ft/h (1 ft/min), additional downcomers should be installed at the
disperser plate. [3]
TROUBLESHOOTING CASE STUDY
A sieve tray column was employed in treating a C3/C4 liquid hydrocarbon stream
with MDEA in a refinery (See Table 2, Case 3). The new LPG treater had no
apparent problems during startup. However, after severaI weeks of operation, plant
personnel noticed severe emulsion formation in the contactor. The emulsion resulted
in poor phase separation of the LPG from the amine. The consequent entrainment of
liquid hydrocarbons into the regenerator had resulted in unstable operation.
Examination of the operating condition such as temperature, pressure, flow rate,
etc., revealed nothing unusual that would cause such severe emulsification of the
phases. However, a closer look at the equipment revealed several inconsistencies in
the designs.
The actual operating range was calculated and compared to the criteria for a typical
sieve tray column (See Table 3).
Table 3. Comparison of actual operating conditions versus a typical design.
The most notable items in this comparison were the high velocities at the orifices of
the distributors. Excessive hydrocarbon velocity was believed to be the primary
cause of the emulsion. At velocities of this magnitude, the LPG exited the distributor
as unsteady jet streams rather than controlled flows. The breakdown of the turbulent
stream resulted in the formation of non-uniform droplets. In addition, the high
velocity stream sheared through the aqueous phase to produce even smaller droplets
that created a very broad drop size distribution. From a geometry standpoint, smaller
aqueous droplets can exist in the interstices of the larger LPG drops to develop more
stable emulsions.
Further examination of the design revealed the direction in which the LPG exited
the distributor to also have an impact on LPG losses. Since the distributor exit
holes were pointed down, the LPG was expected to go some distance downward
before the density difference overrode and the droplets reversed direction. The
distance between the distributor and the tower bottom is approximately 10.5 ft. At
high velocities, some hydrocarbon was apparently carried out with the rich amine.
The high entrance velocity of the amine posed a concern from an amine loss
standpoint. Although 8 to 10 minutes of hydrocarbon retention time above the
amine distributor was allowed in the design, the feed amine velocities disturbed the
amine-hydrocarbon interface and hindered separation.
As a result of these findings, the refiner had additional holes drilled into both
distributors to reduce the LPG and the amine entrance velocities. Since the trays
effectively re-disperse the droplets, the size of the orifice diameter was not critical.
To prevent possible hydrocarbon entrainment into the rich amine, the LPG
distributor was inverted so that the holes were pointed in the upward direction.
The equipment modification appeared to have corrected the problem and eliminated
the formation of emulsion. The second startup of the plant went smoothly with no
carry-over of the LPG into the regenerator.
CONCLUSION
The purpose of this paper is to offer some basic criteria for the design and evaluation
of LPG treaters. The paper also discussed some of the rationale behind the criteria to
give a better understanding of their effects on mass transfer and on the operation of
LPG treaters. Actual operating data from several plants showed that these criteria can
be used with reasonable reliability for evaluation purposes. In almost all the cases,
the LPG treaters were built with high degrees of conservatism in their design.
REFERENCES
[1] Laddha, G.S. and Degaleesan, T.E., Transport Phenomena in Liquid
Extraction, McGraw-Hill, 1978.
[2] Treybal, R.E., Liquid Extraction, 2nd Edition, McGraw-Hill, 1963, 1978.
[3] Stringle, R.F., Random Packings and Packed Towers -Design and
Applications, Gulf Publishing, 1987.
[4] Gas Conditioning Fact Boot The Dow Chemical Company, Midland, Michigan,
1962.
[5] DuPart, M.S., Bacon, T.R., and Edwards, D.J., "Understanding and
Preventing Corrosion in Alkanolamine Gas Treating Plants," Proceedings of
the 1991 Gas Conditioning Conference.
[6] Cross, C. Edwards, D., Santos, J., and Stewart, E., "Gas Treating Through
Accurate Process Modeling of Specialty Amine Plants," Proceedings of the
1990 Gas Conditioning Conference.
[7] Crawford, J.W. and Wilke, C.R., Chem. Eng. Prog., Vol. 47, 1951.
[8] Nemanaitis, R.R., Eckert, J.S., Foote, E.H., and Rollinson, L.R., Chem.
Eng. Prog., Vol. 67, 1971.
[9] Perry, C.R., "Treating System for Ethane Recovery Plants," Proceedings of the
1977 Gas Conditioning Conference.
[1 O] Perry's Chemical Engineers' Handbook, 6th edition, McGraw-Hill, 1984.
[11] DuPart, M.S. and Marchant, B.D.; "Natural Liquid Treating Options and
Experiences," Proceedings of the 1989 Gas Conditioning Conference.
[12] Mayfield, F.D. and Church, W.L., "Liquid-Liquid Extractor Design," Ind.
En=. Chem., Vol. 44, 1952.