I
I
ABSORPTION AND EXTRACTION
S Y M POSlUM' Held under the auspices of the Division of Industrial and. Engineering Chemistry of the American Uhermcal Society at Columbia University, New York, N. Y.,December 28 and 29, 1936,
Carbon Dioxide Scrubbing
by Amine Solutions L.B. GREGORY Standard Oil Company of Louisiana, Baton Rouge, La.
Plant and experimental pilot plant data are presented on the scrubbing of carbon dioxide from hydrogenation process gas by means of water solutions of monoethanolamine, triethanolamine, and diaminoisopropanol. Monoethanolamine and diaminoisopropanol are similar from the standpoint of carbon dioxide absorption capacity and speed ; triethanolamine is only about half as effective as the other two. Care must be taken in the use of these compounds to keep the system free of oxygen as far as possible: otherwise the solutions become corrosive to carbon steel.
W. G. SCHARMA" Standard Oil Development Company, Bayway, N. J.
bing agent, but during the past year additional equipment has been installed to increase scrubbing capacity by the use of an amine of more recent development-diaminoisopropanol. Plant and experimental pilot plant data have been assembled on the action of monoethanolamine (M. E. A), triethanolamine (T. E. A,), and diaminoisopropanol (Dapol) with gas of the composition just given.
Large-Scale Operation on T. E. A. The carbon dioxide scrubbing system originally installed a t the Baton Rouge hydrogenation plant was designed to operate with triethanolamine (T. E. A.) of 50 per cent (by volume) concentration. The properties of T. E. A. are listed in Table I. It is a viscous, high-boiling liquid and absorbs carbon dioxide to the extent of 0.5 mole per mole of T. E. 8, The plant consists of two complete units which can be operated in parallel. Each unit has a scrubbing tower, an activating tower, and the necessary pumps, exchangers, and coolers. Raw hydrogen gas, containing approximately 77 per cent hydrogen, 18 per cent carbon dioxide, and 5 per cent carbon monoxide, methane, and nitrogen, is scrubbed a t an operating pressure of 200 to 235 pounds per square inch. The scrubbed gas containing 0.0 to 0.5 per cent carbon dioxide is then compressed to 3500 pounds per square inch for use in the hydrogenation units. The scrubbers employed in the above operation are 6 feet in diameter and 50 feet high, and contain twenty-four bellcap trays with twelve Jackson-Mann type bells per plate. Perforated plates containing l/s-inch holes on llrinch centers were installed over the bells to increase the degree of contact between liquid and gas and thereby the scrubbing efficiency. Actified solution enters the top of the scrubber and is released from the bottom by an automatic valve actuated by a level controller. The solution passes through exchangers and a steam preheater to the top of the actifying towers.
A
PROBLEM of increasing industrial importance is the removal of carbon dioxide from gases. I n some cases the carbon dioxide is recovered, either for further chemical processing or for use as a refrigerant; in others, notably in the manufacture of hydrogen from carbonaceous material, it is removed as an undesired impurity. For either purpose a washing process is commonly used, in which the carbon dioxide is removed from a gaseous mixture through its preferential solubility in certain liquid media. Most important of these solutions, and of comparatively recent development, are the organic amines which show a much higher carbon dioxide solubility than the older types of solutions such as water or potassium carbonate. At the hydrogenation plant of the Standard Oil Company of Louisiana a t Baton Rouge, carbon dioxide is removed from raw hydrogen prepared by the catalytic reformation and conversion of natural gas with steam. The raw gas contains approximately 18 per cent carbon dioxide, and this is reduced to essentially 0.0 per cent by scrubbing with solutions of organic amines.* Triethanolamine has been used as the scrub2 Additional papers presented at this symposium appear on pages 270-318 of our March issue, and on pages 4 4 7 4 5 9 of our April issue. * The use of aliphatic aminas for removing aoidic gases from gas mixtures is covered by patents owned by the Oirdler Corporation, Louisville, Ky.
5 14
INDTJSTRIAL AND ENGINEERING CHEMISTRY
MAY. 1937
FIGURE 1 . EXPERIMENTAL CARBON DIOXIDE PILOTPLANT
TABLE
I.
Structure
Boiling Point
T.E. A .
N(CHzCH20H)s
208 (10 mm.)
Dspol
HpCNHi
119 (4 mm.)
M.E. A.
HzCNHz
172
-0.
5
all operating at atmospheric pressure. The carbon dioxide absorbed in the scrubber is partly flashed off in the flash chamber and the remainder is stripped out in the actifier. The stripped carbon dioxide, containing some steam and vaporized solution, is passed through a condenser and then vented, while the condensate is pumped back into the flash chamber. Stripping in the actifier is accomplished by means of closed steam in the actifier kettle. Stripped solution is then cooled and pumped back into the scrubber. Both scrubber and actifier were built to contain the same packed height as the large-scale scrubber and actifier with bubble plates removed. This was done to simplify construction and provide data on the use of packing material to replace the trays used in the large plant. The scrubber ( l l g / l inches inside diameter) was fist packed to a height of 35 feet with 3/Anch stoneware Raschig rings; the actifier (12 inches inside diameter) was packed to a height of 26 feet below the feed inlet and 6 feet above the inlet with 2-inch rings.
AGENTS FOR
CARBON DIOXIDE REMOV.4L Normal Strength -Properties of Normal So1ns.in HzO Sp. heat Viscosity Sp. gr. Approx. Commercial Purity. '..% Soln. at looo F. at 100' F. at 80' F. Cot So1y.S .. 3 Wt. 7 0 Centipoises-Cu. ft./uat. Water, not over 1; monoethanolamine, 60" 0.80-0; 85 6.5 1,085 4 .s not over 2.6. diethanolamine not over 15; trie)thanolarnine, no; less than 80 Diaminoisopropanol, 90; water and 30 5.0 1.040 11.7 0.85-0.90 higher amines, 10
PROPERTIES OF SCRUBBING
Materiai
515
Monoethanolamine, 97-100
30
0.80-0.85
...
1.012
10
HnbOH Volume per cent. b At 1 atmosphere COz pressure and 80' F.
These towers are 6 feet in diameter and 50 feet high, and contain sixteen bell-cap trays with twenty Jackson-Mann type bells per tray. Under test operation which was confined to one scrubber and one actifier, the carbon dioxide content of the gas was lowered to 0.4 per cent while scrubbing 5,600,000 cubic feet of hydrogen per day (calculated from inlet gas on the basis of pure hydrogen a t 60" F. and 1 atmosphere pressure). Table I1 shows the effect of installing the perforated plates over the bell caps in the scrubbing towers. In spite of a somewhat high solution outlet temperature, design conditions of 6,000,000 cubic feet of hydrogen per day, containing 0.5 per cent carbon dioxide, were essentially met after the perforated plates were installed on the trays. Without. the plates, the gas was reduced to only 2.3 per cent carbon dioxide.
Pilot Plant Results During the past two years consideration has been given to enlarging the capacity of the Baton Rouge plant. In order to carry this out, an experimental pilot unit was built to investigate the scrubbing characteristics of M. E. A. and Dapol in relation to those of T. E. A. A flow sheet of the pilot unit is shown in Figure 1. It consisted of a scrubber built to withstand pressures up to 500 pounds per square inch, connected by means of a pressurereducing valve to steam preheater, flash chamber, and actifier,
TABLE IT. OPERATION OF HYDROGENATION PLANT WITH T. E. A. Gas rate ( 1 0 0 7 Hz), C I L ft./day COP inlet gas s/, cot exit gas '% Circulation, 'gal./min. S o h concn., vol. yo T. E. A. Soln. temp., F . : To scrubber From scrubber Actifier steam 1bJday COX in actified soln., cu. ft./gal. ~~
With Perforated Without Perforated Plates on Trays Plates on Trays 5,600,000 5,600,000 16.5 17.0 0.4 2.3 275 288 47 49 146 170
240,000 0.32
148 166
... ..,
~
The gas used in all tests was a raw hydrogen gas of composition previously indicated for the large-scale operation. It was supplied to the scrubber a t a pressure varying from 200 to 235 pounds per square inch except where otherwise stated. All quantities of gas and steam are expressed as cubic feet a t 60" F. and 1 atmosphere pressure.
Triethanolamine Solution During the first experimental runs on T. E. A. (Table 111, runs A and B), the scrubbing tower was packed, as stated in the previous section, with 3/d-inch Raschig rings. I n these runs the solution concentration was maintained close to 50 per cent T. E. A. by volume. The raw gas was passed through the tower in increasing amounts, using the minimum quantity of solution to remove the carbon dioxide. At a throughput of 6000 cubic feet of raw gas per hour and a solu-
INDUSTRIAL AND ENGINEERING CHEMISTRY
516
VOL. 29, NO. 5
PLANT OPERATION USING T. E. A.0
Run A
B
c
D E
F
G
H I
-Rates on Exptl. SorubberRaw gas T. E. A. Cu. f t . / h r . Gal./hr. 70concn. 6,000 400 46.1
--TABLE111. PILOTTemperature Scrubber Inlet* Outlet Preheat F. F. F.
COz/Gal. Soln. After Actifier flash soln. Cu. ft. Cu. ft.
Actifier kettle F.
..
SteamC
Cot in Exit Gas
Cu. ft./hr.
%
11.000
0.0
....
0.0
144
148
199
221
6 000 7:800 9,100 9,400
370 491 500 500
35.0 37.4 41.5 43.4
142 147 140 152
152 153 142 159
202 207 202 201
227 224 222 235
1:i7 2.50 1.18
1:23 0.72 0.63
10,000 12,700 10,400
9 400 11'000 111200 12,000
560 596 595 660
47.7 46.4 47.0 44.0
147 149 165 155
150 157 172 157
192 196 200 202
226 225 229 227
1:38
0:39
13,500 13,200 13,200 16,000
a Inlet Con, 18 per cent.
b
Measured a t cooler outlet.
..
0.535
.. ..
..
Remarks 3/a-in. Raschig rings in scrubber tower Flooded 2-in. Raschig rings in scrubber
0.0
.......
0.0 2.6
High scrubber temp. and low circulation
0.0
0.0 1.2 0,s
Includes steam used in preheater.
.......
.
. . ... . , . . . . . , Tower flooded a t higher liquor rates
~
TABLEIv.
P I L O T P L A N T O P E R A T I O N USINQ
DAPOLAT
A SYSTEM P R E S S U R E O F
--Temperat jure PreRates onI Exptl. Scrubber -Scrubberheater --Aotifier-Run Raw D apolInlet Outlet outlet Top Kettle No. gas F. CY. f. t ../ h r Gal./hr. % concn. a F. O F . a F. F. 7 -
.
2 3 3
3A
4 5 6 7 8 9 10 11 12 a
12 000 12'000 121000 12000 19'000 181400 20000 19:550 19300 18:650 19 850 191800 24.000
340 425 425 340 625 525 525 522 625 525 525 625 525
31.1 29.4 24.7 24.9 29.5 21.2 32.0 30.3 30.3 31.3 2617 27.0
... .. . ...
ioi
203 200 203 200 202 197 207 205
220 220 221 229 225 231 227 226 226 226 232 226 226
--
SteamPreheat Actifier Cu. It. p e r hr.
....
... ....
15,000" 19,000a 26,000a
....
7,280 7000 7:240 7,100 6600 10:192 10,192
,...
15,200 16,850 15,960 15,960 17,000 16,100 16,100
con
235
in Exit Gas
% 0.2 0.0 0.0 0.0 0.5 0.4 0.6 0.3 0.1 0.0 0.0 0.0 3.0
POUNDS PER SQUARE
INCHG A G E
--
COa/Gal. Soln.-----. Dapol Scrubber After Stripped Flashed exit flash liquor liquor cu. ft. cu. j t . cu. ft. % 10.62
...
...
9:9 7.2 10.1
... ... ...
...
8.18 10.4
8.10
..
7 :05 9.0 8.7 9.90
.. .. ..
0:85
..
Conon. Condensate
%
4.36 4.03 2.90 2.91 5.78 4.0 5.04 4.74 5.04 5.77
..
1.54
26:4 27.7 24.8
0,066 9.8
4: 89 4.17
24:5
.. .. .. ..
. .
.... ,...
0.144 0.204 0.167 19.8 0.0376 0.132
Total steam.
tion rate of 370 gallons per hour (45 per cent T. E. A.), the tower flooded and solution appeared in the exit gas trap. In both runs the carbon dioxide was reduced to 0.0 per cent (analysis made with Orsat apparatus) while decreasing the solution rate until carbon dioxide first showed up in the exit gas. The test period was then taken a t a slightly higher solution rate. Extrapolating these data to plant-scale operation indicated that a/4-inch Raschig rings would limit the operation using one scrubber and one actifier to about 4,000,000 cubic feet of hydrogen per day, assuming equivalent efficiency of operation. This limitation was apparently imposed by flooding tendency rather than absorption capacity. A short run was next made circulating pure water through the system to determine whether the carry-over was due to liquid viscosity or entrainment. At 18,000 cubic feet per hour of raw gas and as high as 700 gallons per hour of water, there was no carry-over. This gas rate is equivalent to 16,300,000 cubic feet per day of raw gas on the plant scrubber. It was then concluded that the T. E. A. solution viscosity was too great and that 2-inch Raschig rings a t least would be necessary to obtain any great amount of scrubber capacity. After changing over to 2-inch rings in the scrubber, the runs were continued with increasing gas rates to reach maximum capacity. At 12,000 cubic feet per hour of raw gas (run I, Table 111), it was possible to obtain gas with 0.8 per cent carbon dioxide content, using 650 gallons per hour of 44 per cent T. E. A.; higher solution rates were impossible because the scrubber was again a t the flood point. From these data it was estimated that the use of 2-inch rings would permit operation at 8,000,000 cubic feet of pure hydrogen per day in the plant scrubber, using 350 gallons per minute of 50 per cent T. E. A. This solution rate is equivalent to 44 gallons per minute of 50 per cent solution per million cubic feet of hydrogen per day; this unit solution rate was also obtained on the plant scrubber using trays, when scrubbing 5,600,000 cubic feet of hydrogen pes day to 0.9 per cent carbon dioxide. Runs G and H (Table 111) show the effect of scrubbing
temperature. In the first of these runs the carbon dioxide was completely removed a t an exit scrubber temperature of 157" F. With all other conditions the same, the carbon dioxide content increased to 1.2 per cent with an increase to 172" F. of the scrubber outlet solution. The experimental actifies operating a t about 225" F. kettle temperature stripped the solution to approximately 0.5 cubic foot of carbon dioxide per gallon of 50 per cent solution. The steam quantities shown in Table I11 include steam used for preheating since no exchangers were used. These quantities are consequently higher than those normally incurred in plant operation (Table 11).
Diaminoisopropanol Solution Dapol has a far greater affinity for carbon dioxide than T.
E. A. Some of its physical properties are listed in Table I. Normally it is used in a 30 per cent by weight solution in water and absorbs carbon dioxide mole for mole. Because of its high heat of absorption (Girdler Company reports approximately 750 B. t. u. per pound of carbon dioxide absorbed) , it is inadvisable to use solutions stronger than 30 per cent by weight. Greater strengths would result in too high a solution temperature upon absorption of carbon dioxide, thereby giving lowered capacity unless cooling is carried out in the scrubber. The same experimental scrubbing system used for the T. E. A. investigations was used for Dapol, and the same gas (raw hydrogen containing 18 per cent carbon dioxide) was scrubbed a t a system pressure of 235 pounds per square inch. The first runs (Table IV) were made a t an inlet gas rate of 12,000 cubic feet per hour, and in all these the carbon dioxide was almost completely removed. The required liquor rate varied considerably, partly because of changes in concentration and partly because of variation in stripping efficiency. Further runs were made at inlet gas rates from 18,400 to 24,000 cubic feet per hour. At these rates the scrubber o p erated satisfactorily with no entrainment or flooding difficulties. The exit carbon dioxide a t gas rates as high as
1NDUSTRIAL AND ENGINEERING CHEMISTRY
MAY, 1937
are always lower than for the stripped liquor. The analytical procedure does not allow for the increased weight of the solution due to carbon dioxide absorbed. If this is corrected, the inlet and exit concentrations check very well. The runs at 50 pounds per square inch scrubbing pressure are included more from the standpoint of completeness. Discrepancies found in the limited amount of data obtained prevent any rigid generalization on the effect of lowered operating pressure. Allowing for heat losses, a 30 per cent Dapol solution cannot absorb much over 5 cubic feet of carbon dioxide per gallon without having the temperature rise to over 165" F. This can be offset by the use of cooling coils in the tower or by circulating an excess of solution. With lower operating pressures and the same degree of actification, this becomes less necessary because the carbon dioxide capacity of the solution is accordingly lower.
19,850 cubic feet per hour was 0.0 per cent and amounted a t times to as much as 0.6 per cent, depending on the stripping efficiency. At these higher rates it became apparent that the actifier was operating very close to the flood point so that slight variations in kettle steam resulted in boiling over. This is shown by runs 9 and 10 in which a steam increase of only 1000 cubic feet per hour resulted in Dapol carry-over into the condensate. At the 24,000 cubic feet of hydrogen .gas rate, it was impossible to scrub the gas completely because of the limitation mentioned above. The scrubbing tower, however, was apparently far from overloaded, and it is certain that if the stripper could have handled more liquor, the carbon dioxide would have been completely removed. In most of the runs the liquor was stripped down to 4.65.5 cubic feet of carbon dioxide per gallon of 30 per cent solution. Solution with this carbon dioxide content is low enough in carbon dioxide vapor pressure still to permit reduction to 0.0 per cent carbon dioxide in the gas, provided sufficient solution is circulated. For the runs in which carbon dioxide appeared in the exit gas, a 30 per cent Dapol solution held approximately 11.5 cubic feet of carbon dioxide per gallon a t 150" F. for a carbon dioxide partial pressure of 3 atmospheres. Assuming normal stripping to 5 cubic feet of carbon dioxide per gallon, the net capacity of the material is 6.5 cubic feet of carbon dioxide per gallon. Extrapolating these data to large plant operation, it appears that 13,500,000 cubic feet of hydrogen per day could be handled by one packed scrubber and actifier. Actually the limitation was imposed by the actifier, and consequently the scrubbing capacity may well exceed this figure. Another set of runs was made a t a system pressure of 50 pounds per square inch (Table V). The carbon dioxide analyses of the scrubber exit liquor were corrected for concentration (amine content) to 30 per cent; in this correction the small carbon dioxide solubility in the water was neglected. The corrected carbon dioxide contents indicate that a 30 per cent solution held approximately 8.75 cubic feet of carbon dioxide per gallon a t 150' F. for a carbon dioxide partial pressure of 0.8 atmosphere. The amine concentrations obtained at the scrubber exit TABLEV. PILOTPLANT OPERATIONUSING DAPOLAT Rates on Exptl. Scrubber -ScrubberInlet Outlet Raw gas ---DapolF. a F. Cu. ft./hr. Gal./hr. %conen. 152 163 10,000 330 27.4 100 148 10000 275 24.6 110 148 l0:OOO 265 25.3
TABLE VI. Cu. It./ hr.
18 000 17'700 18:050 17,700 17 225 17:500 13,075 12,862 13,425 19000 18'440 19:250 20,550 21,800 18,050
PILOT PLANT OPERATION USINGM. E. A.
.
Temperature Steam Exptl. Scrubber -ScrubberPre- YAotifierto -M, E. A.Inlet Outlet heater Top Kettle Kettle Gal./ % hr. concn. ' F. ' F. F. F . ' F . Lb./hr. 108 135 192 182 228 24.8 186 230 ... 104 136 196 26.0 ... 98 140 195 198 229 30.2 202 232 152 198 103 35.4 199 204 231 102 160 36.2 825 158 195 200 235 31.6 100 204 221 690 97 147 208 30.5 222 149 206 690 95 207 29.6
...
... ...
180 300 300 300 300 300 300
32.1 31.6 30.5 30.6 24.3 26.8 30.2
94 97 98 98 97 97 100
148 147 148 150 143 145 144
Monoethanolamine Solution M. E. A. has the same general chemical structure as T. E. A. but is considerably lower boiling and less viscous. Its physical properties are given in Table I. This material, like Dapol, has a high heat of carbon dioxide absorption, approximately 720 B. t. u. per pound of carbon dioxide absorbed. The experimental scrubbing system was used in testing this material in a 30 per cent (by weight) solution in water, scrubbing a raw hydrogen gas containing 11 to 16 per cent carbon dioxide a t a total pressure of 200 pounds per square inch gage. Operation during these tests was rather erratic. The greater portion of the hydrogenation plant had been shut down for new construction, and consequently the compressor on the pilot unit was working over capacity, necessitating the by-passing of scrubbed gas to the pilot unit. This accounts for the varying amounts of inlet carbon dioxide noted in Table VI. The scrubber data showed that a 30 per cent M. E. A. solution could hold about 10.5 cubic feet of carbon dioxide per gallon with a carbon dioxide partial pressure of 1.7 to 2.2 atmospheres. Assuming that the fresh solution has been stripped to 2.5 cubic feet per gallon, the net carbon dioxide absorption will be 8 cubic feet per gallon of 30 per cent M. E. A. Because of limitations in actifier capacity (as was the case SYBTIMPRESSURE OF 50 POUNDS PER SQUARE INCH GAQE
CO, -COn/Gal. S o l n . 7 -Dapol Temperature in Actifier Exit Scrubber Flashed Stripped Flashed Pre7ActifierGas exit liquor liquor liquor Steam heat T o p Kettle C u . f t . C z ~ . f t . Cu.ft. % F. O F. F . Cu.ft./hr. % 2.89 22.7 7.68 6.35 200 218 17,100 0.2 197 2.71 19.9 5.94 16,050 0.5 7.1 200 218 198 2.61 20.8 7.62 5.64 224 16,050 0.4 196 202
7 -
Rates on Raw gas
A
517
206 203 204 203 200 201 204
206 198 200 198 193 197 198
221 225 227 224 225 224 223
690 690 690 690 795 740 740
AT A
-COPInlet gas
-
Concn. Gage SerubPressure ber Stripped Conof exit liquor denaate Scrubber
%
%
%
L b . / s q . in
25.9 22.6 23.8
27.4 24.6 25.3
0.268 0.55 0.493
60 51 51
SCRUBBER PRESSURE O F 200 P O U N D 5
PER SQUARE
-COa/Gal. S o l n . 7 -M. E. A. Conon.-Exit Scrubber Flash Stripped Flash Conden- Scrubber gas outlet drum liquor drum sate outlet
% '
%
11.6 12.5 11.2 11.8 12.2 14.2 14.5 14.5
0.5 0.4 2.5 0.0
14.6 16.0 14.8 15.1 15.2 15.4 15.4
1.4 1.6 1.3 2.0 3.6 4.0 3.8
0.0
0.8 0.0
0.2
C u . ft. 6.1
Cu. ft. Cu. f t .
%
%
7.3 8.8 8.2 7.9 7.0 7.4
5.0 5.3 5.5 7.8 7.1 7.3 6.2 6.3
3.4 3.1 3.0 3.2 3.6 2.1 2.2 1.6
22.6 25.1 26.3 33.4 32.8 33.7 32.4 32.1
13.9 0.38 0.24 0.54 0.5 0.38 0.61 2.0
7.8
6.5
2.0 3.04 2.78 1.89 0.91 2.18
26.2
.. .. .. .. .. ..
0.51
...
... ... ...
... ... ...
... ... ... ...
... *.
...
0.49 0.39 16.9 0.44 0.52
INCH Remarks
% 24.0 Actifier flooding 2i:2
....
t
32.8 33.9 ..... 29.3 ..... 30.7 29.1 Actifier' siightly flooding 29.6
...... .. .. ..
..... .....
..... Actifier'5ooding
....,
5 18
INDUSTRIAL AND ENGINEERING CHEMISTRY
VOL. 29, NO. 5
with the Dapol investigations), it was impossible to determine the ultimate scrubber capacity. The highest throughput reached was 21,800 cubic feet of raw gas per hour, but a t this rate the exit carbon dioxide was 4.0 per cent. However, the foul liquor from the scrubber contained 11.0 cubic feet per gallon (26.8 per cent solution), indicating that the absorption rate was not the limiting factor but that an insufficient solution rate was used.
tinuously circulated over the 3-inch spiral tile contained in the condenser and cooled externally in shell and tube coolers, A constant level is automatically held in the condenser by releasing excess condensate to the actifier. In this way the amine is scrubbed from the carbon dioxide-steam mixture leaving the actifier. Solution concentrations are maintained constant by returning water which would otherwise be lost to the atmosphere.
From the data obtained, i t is believed that 30 per cent M. E. A. solution is just about the equivalent of 30 per cent
The new equipment has been operating quite satisfactorily to date, but gas rates have not as yet been high enough to warrant a direct comparison either with former operation on T. E. A. or with the experimental data obtained on Dapol.
Dapol solution for the removal of carbon dioxide under operating conditions normal to the hydrogenation plant operation. Because of its rather high vapor pressure, there was some question as to whether M. E. A. losses in the scrubbed gas would be serious. Two determinations were made of the amine content of the scrubbed gas (temperature 100" F., pressure 200 pounds per square inch gage) and these showed an M. E. A, content of 0.262 and 0.257 gram per 1000 cubic feet (60' F., 1 atmosphere) of scrubbed gas. These values represent the loss by vaporization and possibly entrainment.
Large-Scale Operation on Dapol From the results of the pilot-plant scrubbing experiments, Dapol was selected as the material to be used in the enlarged hydrogenation plant unit. Figure 2 is a photograph of the enlarged unit. One new actifying tower was added, similar in construction to the existing towers. These towers are shown a t the left with three small flash drums located near the top of each actifier. Xo change was made in the two scrubbers a t the right of the actifiers. Coolers, exchangers, preheaters, and pumps (not shown in Figure 2) were arranged in three parallel passes so that either scrubber could be connected to any of the actifiers by suitable valve adjustments. The vertical drum installed between the two actifiers on the left is a direct contact condenser where amine solution is recovered from the carbon dioxide and steam coming from the actifier. Condensate containing a trace of amine is con-
Corrosion In constructing plant equipment for operation with the organic amines, it is necessary to exclude the use of any copper-, tin-, or zinc-bearing alloys because of corrosive action on these materials. During the initial operation of the T. E. A. scrubbing system a t the Baton Rouge hydrogenation plant, serious corrosion difficulties were encountered on carbon steel equipment. This corrosion first appeared a t the automatic release valve which releases solution from the scrubber. The corrosion a t this point destroyed the operability of the valve after approximately one day of service. Substitution of KA2 for carbon steel eliminated attack a t this point. During this period corrosion was also obtained in the lines and exchangers handling the carbon-dioxide-bearing solution. The attack was mainly localized to places on which the solution impinged or changed direction. For example, serious corrosion was obtained a t bends and exchanger tube sheets. The cause of this corrosive condition was not immediately appreciated since no attack had been experienced on small equipment used in experimentally investigating the scrubbing properties of this solution. An experimental program was initiated using this experimental equipment under conditions which duplicated the full-scale operation as nearly as possible. In order to duplicate the velocity and temperature conditions of the plant equipment, the carbon-dioxide-saturated solution leaving the experimental scrubber was passed through a hot
MAY, 1937
INDUSTRIAL AND ENGINEERING CHEMISTRY
water bath, the temperature of which was adjusted to that existing at the exit of the plant scrubber. A number of orifice plates were installed between flanges in the submerged exit line. These orifices were threaded to hold a small bolt which was to serve as the test specimen. Holes were drilled through the shank and head of the bolt so that the stream had to make a right-angle bend in passing through. The effect of-velocity w a s studied by varying 50 t h e s i z e of t h e holes and also by placing the release 4.0 valve ahead or behind the specimens t o permit gas-flashing from 3.0 the solution during i t s p a s s a g e through the speci2.0 mens. This method of studying corrosion I.o was somewhat hampered by the discovery t h a t 0 essentially no corr o s i o n w a s obLENGTH OF RUN - HRS. tained when using F I a r m E 3. CORROSION STUDY BEFORE T. E. A. from fresh AND AFTER RELEASEVALVE the unused makeD = diameter of speoimen, inches. a t zero up supply for the time, D = 0.0825 inch; ?ant. soiution = 10 gallons per our large plant. Howe v e r , w h e n solution which had been in circulation in the large plant was charged to the experimental unit, intense corrosion resulted. Figure 3 gives the data obtained during this investigation. The general significance of the results may be more fully understood by assuming that two identical specimens (initial diameter, 0.0625 inch) were located before and after the release valve a t the start of the run. The specimen after the valve was removed a t various intervals during the run and the loss in weight and increase in diameter were noted. Curve A indicates that the gradual increase in diameter of the specimens finally reduced the velocity to a point where corrosion practically stopped. Curve C shows the negligible weight loss obtained with the specimen located before the release valve during the operation a t 200" F.; after 45 hours the loss in weight of the specimen after the release valve was fifty-eight fold greater than that before the valve. Curve B is a continuation of the study a t 145" F. instead of 200" F. The corrosion on the specimen before the release valve was quite similar to that a t 200" F. This may indicate either that a small temperature effect occurred, or that the order of magnitude of the loss in weight obtained was too small to establish differences in corrosion tendency satisfactori€y In continuously using the charge of solution obtained from the circulating system of the plant, it was found that it gradually lost its high corrosive activity and approached the low activity exhibited by fresh material. This effect was eliminated in the previous study by frequent change of solution. Loss of corrosiveness indicated that probably the raw hydrogen gas was responsible for the effects noted since this was the only variable which had not been duplicated in the experimental set-up. The hydrogen a t both the large plant and the experimental unit is produced by the reformation and conversion of natural gas, the only difference existing in the method of cooling. In the small unit the hot gas from the
.
519
converter was cooled by bubbling through a well water seal, whereas the large plant employed a direct-contact -packed tower with river water as the cooling medium. Consequently, considerably more water was used in the case of the plant cooler, and therefore more dissolved oxygen was available for the gas stream. ddditional small-scale laboratory experiments were then performed using fresh 50 per cent T. E. A. solution which was continuously circulated over steel wool by a gas lift employing either air or carbon dioxide as the lifting agent. The effect of temperature was obtained by electrically heating the reservoir from which the gas lift removed the solution. Carbon dioxide was continuously passed into the reservoir through an alundum thimble to maintain a saturated solution, regardless of the types of gas used in the lift. Samples of the solution were periodically analyzed for iron. The results showed that after 48 hours a t 145" F. a 50 per cent T. E. A. solution containing 2.7 to 3.0 cubic feet of carbon dioxide per gallon dissolved 0.1 gram of iron per liter when using carbon dioxide as the lift medium, and 1.65 grams of iron per liter when using air as the lift medium. These experiments tended to indicate that the basic cause of the severe corrosion noted a t the large plant probably could be attributed to traces of oxygen which entered the scrubbing tower with the raw hydrogen; oxidation of the T. E. A. resulted, and this material was then capable of adding iron in the form of a complex organic compound. Among the sources of oxygen which may cause the above reaction, the following are cited: 1. Introduction of oxygen to the raw hydrogen by river water used at the coolers. 2. Introduction of oxygen by traces of this material in the process steam used for the production of hydrogen. 3. Diffusion of oxygen through the water seal at the hydrogen holder.
The river water used in the hydrogen coolers was analyzed for its oxygen content. h normal concentration of dissolved oxygen was found in the water entering the coolers, but no trace could be found in the water leaving. The coolers were then revamped to permit continuous recirculation of cooling water with indirect external cooling by means of shell and tube exchangers. After this change had been effected, corrosion in the plant was reduced materially.
Relative Stability of Amines toward Oxidation An accelerated oxidation test was devised to study the stability of the various amines toward oxidation. The solutions (500 cc.) are placed in glass tubes containing 20 grams of steel wool and held a t a constant temperature of 185' * 2" F. while oxygen is passed through the solution at the rate of 5.0 liters per hour. The tests are continued for 160 hours, and the final solutions are then examined for specific gravity, alkalinity, color, and iron content. This test showed that monoethanolamine had the greatest resistance toward oxidation, triethanolamine was next, and Dapol showed the least resistance. If all traces of oxygen are removed, the oxidation tendency of these amines remains only a potential source of corrosion and loss of amine, and on this basis the use of Dapol in the large plant appeared justifiable. Another angle of approach to the control of oxidation and corrosion difficulties of the amines has been proposed by the Girdler Company in the use of various oxidation inhibitors. Satisfactory results have been claimed but, to date, inhibitors have not been used a t the Baton Rouge plant and consequently no data are available on their use. R~CEIYRID January 27, 1937. Presented before the symposium under the title "Some New Solutions for Absorption of Carbon Dioxide from Gases."
