PERFORMANCE O F AN AGITATOR ABSORBER REMOVING .

COz FROM A GAS

IN

A Statistical Investigation by Albert S. Moore and Sidney'Katell U. S; Bureau of Mines, Branch of Coal Gasification Morgantown, West Virginia

ABSTRACT
The e f f e c t s of p r o c e s s p r e s s u r e , carbon dioxide content of the feed g a s , solution throughput, and liquid-to-gas r a t i o on the performance of a n agitator absorber at levels of 200-300 p. s . i. g., 10-20 mol-percent, 60-90 gal. / h r . , and 16.5-55 gal. / h r . solution per 1000 std. cu. f t . g a s / h r . , respectively, have been determined. The absorbent used was 40 weightpercent diethanolamine i n water; the gas used was i n e r t gas with carbon dioxide added. The data and results f r o m t h r e e s e t s of sequential factorial experiments a r e presented. The experiments w e r e designed to t e s t each of the four factors a t three levels of operation.

Not f o r Publication P r e s e n t e d Before the Division of Gas and Fuel Chemistry American Chemical Society New York City Meeting, Sept. 8-13, 1957 "Performance of an Agitator Absorber in Removing C02 F r o m a Gas" A Statistical Investigation Albert S . Moore and Sidney Kate11 U. S. Bureau of Mines, Branch of Coal Gasification Morgantown. West Virginia During the l a s t fifteen y e a r s , p r o c e s s e s for removing carbon dioxide and hydrogen sulfide f r o m industrial g a s e s by scrubbing with various amines have become generally accepted. To date, these p r o c e s s e s have employed conventional g a s -liquid contactors, packed columns and bubble cap towers. It has been shown that the absorptive capacities of the a m i n e s i n c r e a s e markedly with i n c r e a s e d concentration, but because of the operating difficulties encountered when conventional scrubbers a r e used to p r o c e s s liquids of even moderate viscosity, the limiting concentration has been about 15 percent amine in water. Thus, for these p r o c e s s e s to realize f u l l commercialization, methods of gas -liquid contacting able t o tolerate m o r e viscous scrubbing media must be developed. This paper describes studies of the absorption of carbon dioxide in 40 weight-percent diethanolamine i n water utilizing an agitator-type contactor. Figure 1 shows a vertical c r o s s section of the a b s o r b e r . The action within the a b s o r b e r is a s follows: The lean solution is fed into the vessel through a s p a r g e r ring. The liquid is r a i s e d in the l i f t tube by the centrifugal action of the turbine and propelled f r o m the absorber i m p e l l e r in the f o r m of a fine spray. This s p r a y of liquid droplets moves a t a high speed relative to the gas s t r e a m , which passes through the s p r a y c u r t a i n in upward flow through the a b s o r b e r . Inasmuch as this is a relatively novel gas-liquid contactor, it i s to be expected that its mode of action would be distinctive.
\

The work described here was undertaken to determine the char a c t e r i s t i c s of the agitator-absorber with respect to: (1)P r o c e s s p r e s s u r e , ( 2 ) carbon dioxide content of the feed gas, ( 3 ) absorbent solution r a t e , and (4) liquid-to-gas ratio. DESCRIPTION OF GAS-PURIFICATION PILOT PLANT

"

Figure 2 is a flow d i a g r a m of the pilot plant. F o r this study the absorber was installed to operate with existing solution-regeneration facilities. The Absorber The a b s o r b e i is shown in figure 1. The essential internal p a r t s a r e the annular ring and lift tube and the turbine and impeller. The turbine i m p a r t s a centrifugal action to the absorbent solution, which is directed by the annular ring and lift tube onto the impeller, where the s p r a y is formed. The latticed faces of the impeller a r e beveled to about 3 8 O off horizontal S O

.
SAS 1. 1

.. .

N
1
I

I! i

FLOW DlbGRbH OF ( G I s -PURlFlGbTlON

PILOT PLbNT)

w m

a
0
I -

w
I-

1 .

.E .9

'

that the principal portion of the s p r a y i s directed toward the walls of the absorber v e s s e l r a t h e r than the top. A 1-hp. motor (1.750 r . p . m.) connected to the shaft through sheave pulleys i s u s e d to drive the t u r b i n e . A constant turbine speed of 750 r . p , m . was u s e d for these t e s t s . The Reactivator

$ '

'
J ,

The reactivator tower i s a 17-foot length of 10-inch, schedule 40 pipe, which i s packed to a depth of 15 feet with 314-inch Raschig rings and direct-connected to the reboiler by IO-inch d i a m e t e r flanges. The r e boiler is 10-feet in length and 16 inches in diameter. It i s fitted with a p r e s s u r e connection, connections for adding makeup condensate and f o r &raining the solution, and a thermocouple connection installed n e a r the end of the tube bundle. Steam for reactivation i s controlled by a p r e s s u r e controller and is m e a s u r e d b y weighing the condensate. The entire unit i s lagged with 2 inches of magnesia insulation and 1 / 4 inch of tlInsulcotetl. Auxiliarv Eouioment The solution heat exchanger, solution cooler, and acid-gas cooler a r e all s i m i l a r . Each consists of two Griscom-Russell twin G-fin sections connected i n s e r i e s . The heat exchanger and solution cooler each contain four, schedule 40, 3/4-inch, G-fin pipes about 10 feet long. The f i n pipes in the acid-gas cooler a r e abcut 5 feet long. Thermocouple connections a r e installed s o that a l l inlet and outlet t e m p e r a t u r e s c a n be recorded.
A 1-114 b y 4-inch triplex, single-acting power pump circulates the solution. The flow of the solution to the a b s o r b e r is controlled by an air-operated bypass valve.

The gas i s s t o r e d , before compression, in a wet-type gas holder having a capacity of 1,000 cubic feet. The 2-stage gas c o m p r e s s o r can deliver 7,200 standard cubic feet p e r hour a t 350 p. s.i. g. Solution and Gas Flow The flow of solution and gas i n the pilot plant is typical of most gas-purification s y s t e m s . Inert gas, varying in C 0 2 content f r o m 7 to 10 mol-percent, is taken f r o m the gas holder, c o m p r e s s e d t o about 350 p. s . i. g. , and fed to the absorber. The c a r b m dioxide n e c e s s a r y to bring the carbon diolEide content of the unpurified gas to the level of predetermined experimental conditions is added to the suction line of the c o m p r e s s o r . The lean diethanolamine solution e n t e r s through the top of the absorber. It i s distributed by a s p a r g e r ring directly above the liquid level and just below the top of the lift tube and leaves the a b s o r b e r through a side connection 6 inches f r o m the bottom. The level of solution i n the a b s o r b e r i s maintained to a depth of 12 inches (indicated by the broken line i n figure 1).

The diethanolamine is forced up into the lift tube b y the turbine; baffles in the lift tube eliminate any swirling action. The r i s i n g solution contacts the absorber i m p e l l e r , a deflector designed to give maximum s p r a y commens u r a t e with the dimensions of the absorber v e s s e l . T h r e e distinct s p r a y patterns a r e formed--(l) the initial s p r a y f r o m the a b s o r b e r impeller, consisting of small droplets moving at v e r y high velocities, ( 2 ) impact s p r a y formed when the initial s p r a y rebounds f r o m the walls of the vessel, and ( 3 ) conjunctive s p r a y f o r m e d when the initial s p r a y and impact s p r a y collide. The feed gas e n t e r s at the base of the absorber to the right of the lift tube. It bubbles through the diethanolamine, p a s s e s through the s p r a y pattern, and leaves the top of the a b s o r b e r into a n entrainment s e p a r a t o r as purified gas. It i s then r e t u r n e d to the gasholder f o r feed-gas makeup. The fouled amine solution leaving the bottom of the absorber i s heated by heat exchange with lean amine solution and regenerated i n the reactivator with indirect heat f r o m l o w - p r e s s u r e s t e a m . The acid g a s leaves the top of the reactivator and p a s s e s through a cooler where entrained s t e a m is condensed and returned to the column. The a c i d gas is then r e t u r n e d t o the holder for feed-gas makeup. The reactivated diethanolamine leaves the reactivator, flows through the heat exchanger where it gives up a portion of its heat t o the fouled solution leaving the a b s o r b e r , and is pumped through a water-cooled solution cooler back to the absorber.

\.

EXPERIMENTAL PROCEDURE
The agitator gas -liquid contactor is a novel piece of equipment.
To evaluate it in t e r m s of m o r e conventional gas -cleaning apparatus, packed

columns and bubble -cap towers, optimum operating conditions for the agitator absorber p r o c e s s m u s t f i r s t be found. Eight obvious factors will affect performance of the absorber: (1) P r o c e s s p r e s s u r e , (2) carbon dioxide content of the feed gas, ( 3 ) absorbentsolution throughput, (4) liquid-to-gas ratio, (5) concentration of the absorbent solution, (6) feed-gas throughput, (7) the speed o the turbine, and (8) carbon f dioxide content of the lean absorbent solution. Although the operational levels of all of these f a c t o r s a r e easily controlled within relatively narrow l i m i t s , t h e r e i s no r e a s o n to suppose that the efficiency of the p r o c e s s is a n additive function of any two of them. Consequently, the complete exploration of the causal relations among these factors would r e q u i r e that the effect of each factor be observed under a l l combinations of values of the other f a c t o r s . Such a procedure would require a relatively l a r g e number of tests. To overcome this difficulty, a s e r i e s of sequential factorial experi m e n t s , each concerned with a group of relevant f a c t o r s , was initiated. Such a procedure allows for maximum flexibility in planning and h a s the inherent advantage that, a s each factorial is developed, the factors involved m a y be directed toward optimum conditions relevant to all other factors. P r o c e s s p r e s s u r e : carbon dioxide content of the feed g a s , a b s o r bent solution throughput, and liquid-to-gas r a t i o w e r e chosen as the independent variahles for the initial experiment; the levels chosen for these factors were 2 0 0 - 3 0 0 p. s . i. g . , 10-20 mol-percent, 60-90 g.p. h., and 16.4-55 respectively.

I,

',

\
1,

Liquid-to-gas r a t i o i n this instance is defined as gallons of solution p e r 1 , 0 0 0 s t d . cu.ft. of feed gas per hour. The dependent variable in the p r o l c e s s i s the total carbon dioxide absorbed, since a l other variables for these experiments were predetermined.
A s dictated by subsequent operation, a second f a c t o r i a l was designed at new operational levels f o r the four f a c t o r s . The purpose of the second s e t of experiments w a s t o investigate further the interactions among the various factors revealed in the initial s e t of t e s t s . The block design for each factorial is presented in Appendix 1.

DATA A N D RESULTS Table 1 presents typical operating conditidns prevalent throughout the 30 tests performed f o r this study; table 2 presents the cumulative data f r o m these tests. Table 1.

- Typical operating range f o r all runs

G a s temperature, OF.
Before agitator A f t e r agitator Solution temperature, Before agitator After agitator Reboiler p r e s s u r e , p. s . i . g .
0

75-85 140-155 F.
100-110 140-160

6.5-8
265-275

Reboiler temperature, OF. Agitator motor Ammeter, amps Voltmeter, volts S t e a m p r e s s u r e , p. s . i. g. Steam consumption, lb. / g a l . amine Turbine speed, r . p. m.

1.2-1.4
470 (constant)

22-30

0.6-0.8
750 (constant)

A s t a t i s t i c a l analysis of the data f r o m experiment "A" is made and an analysis of the cumulative data f r o m experiments "A" and IlBl' is presented.
An e m p i r i c a l equation relating the independent variables t o the performance of the agitator absorber is developed. The effect of the s a m e variables on the performance of the agitator is presented graphically i n chart 1.

\\

'1

'\
\
\, \

The following nomenclature i s used i n these presentations :

Y = total C 0 2 absorbed, cubic feet p e r hour L c L / C = liquid-to-gas r a t i o , gallons per hour solution
C = CO2 content of feed g a s , mol-percent P = p r o c e s s p r e s s u r e , p. s. i. g. d / 1 6 = s u m of the s q u a r e s of the dependent variable accounted for by a given f a c t o r , divided b y the number of t e s t s involved. n = number of tests, contingent t o a given calculation.
EVALUATIONOF DATA
Table 3 presents an analysis for significance of the 16 runs of experiment "A" i n which the factors involved w e r e tested a t the high and low levels. T h i s analysis was made using the method of Yates. The e r r o r mean square was determined using data f r o m the runs a t which the f a c t o r s were tested at the median levels. The calculation was made using this relationship:

S = solution throughput, gallons / h r .

I, 000 std. cu. ft. gas / h r .

E.M.S. = f,YZ - ( Y ' L) n = 33.2 n-1

The standard deviation by definition is the s q u a r e root of the e r r o r m e a n s q u a r e . In this instance the standard deviation was 2 6 c u . f t . / h r . of CO2 absorbed. The interactions between the carbon dioxide content and solution throughput, and between the carbon dioxide content and liquid-to-gas r a t i o revealed i n the analysis o the data f r o m experiment "A", w e r e evaluated f f u r t h e r by solving the normal equations using m a t r i c e s s e t up f r o m the cumulative data of experiments "A" and ItB1'. Using the methods of Davig, the coefficients of table 4 were determined and a n empirical equation relating the significant f a c t o r s to the dependent variable w a s s e t up.

T2ble
P C S L

3.

Factorial analysis

Ekpriment A

E
A-11

Y
296

Fzc t o r

&
1,211,100

Significance

-1 -1 -1 -1
-1 -1 -1 1

Total
L/G
S

A-4
A-10

96 373
132

136,900
41,820

0.001
001 .0

-1 -1 1 -1 -1 -1 1 1 -1 1 -1 -1 -1 1 -1 1
-1 1 1 -1 -1 1 1 1 1 -1 -1 -1

A-13

L/G

2,601 51,984 812

.

00 .5
0.001

A-3
A-llI

350
l@

C '
CL

nil
0.01

A-19

46 9
303
260

cs
CSL

5,625 462 2 @

A-7
A-9
A-15
A-16 A-2 A-18

nil nil nil nil nil

P
PL
PS

1 -1 -1 1
1 -1 1 -1
I

91
342 155

756
324
20 1

1 -1 1 1
1 1 -1 -1
1 1 -1 I .

PSL PC PCL PCS PCSL

?
i

jcg
201

5 6
225 2

nil
nil nil nil

1:-6
A-1

1 1 1 -1
~

515

1 1 1 1
0

A-17

294
263 270 271 275

676
1,453,842
1,2ll 100
J

0 0

A-5
A-8
A-12 A-20

0 0 0 G

o o o o
3

242 742

0 0 0

..

Table- 4.

- Empirically determined coefficients

Coefficient 267.3 t57.0 +57.125 -92.5 -12.75 +18.75

Factor Constant C S

L /G
S L/G CS

B y substituting these r e a l f o r the coded factors:

P = P-250

5 0
C

= C-15.2 5
-IF7

s = s - 7 5

L = L / G . - 35.7 19. 3 and by affixing the proper coefficients and reducing, equation (1) resulted.

(1) Y = 1 7 8 . 3 - 7 . 3 5 C + l . l 8 S - 1 . 4 9 L - 0 . 0 4 4 S L f 0 . 2 5 C S
An analysis f o r variance was made b y solving the inverted mat r i x s e t up f r o m the cumulative data of experiments A and B. again using the methods of D a v i s . This revealed that the e m p i r i c a l equation would account for 98.5 percent of the total s u m of s q u a r e s of the dependent variable.

A further analysis f o r significance was made using Student's 'IT1' t e s t a s presented by Villars. B y this calculation, the e m p i r i c a l equation contains only those t e r m s significant at l e s s than the 5 percent level.
The comparison of observed and c a l c d a t e d values f o r the r a t e of adsorption of C 0 2 i n the agitator a b s o r b e r , included in table 2. was made employing equation (1). C h a r t 1p r e s e n t s graphically the absorbing capacity of the agitator as a function of solution throughput and GO2 content of f e e d gas with gas r a t e s and liquid-to-gas r a t i o s as p a r a m e t e r s . The data f o r plotting c h a r t 1 was evolved by solving equation (1).
-.

DISCUSSION O F RESLILTS

The dependent variable for these experiments was taken a s total carbon dioxide absorbed p e r hour. The effect of the independent variables-p r e s s u r e , carbon dioxide content of the feed gas. solution throughput, and liquid-to-gas r a t i o on the performance of the agitator a b s o r b e r was d e t e r mined on that basis.

. .. ..

..

..

.. ..

_ .

. ...-.

-_.

. .

_ _ . I

-. .- .. .

. .

_. .

, .

..

Significant Effects The f a c t o r s of these experiments a r e tabulated i n the fourth column of table 3; the coefficients that give these f a c t o r s quantitative meaning a r e tabulated i n table 4. Examination of these tables shows that the performance of the agitator a b s o r b e r w i l l appreciate with higher solution throughput, higher C 0 2 content i n the feed gas and lower liquidto-gas ratios. The effect of these f a c t o r s and of the interactions among theni on the performance of the agitator absorber is given explicit quantitative meaning in the correlation afforded by equation (1). Using this equation the performance of the agitator, i n t e r m s of cubic feet p e r hour of carbon dioxide absorbed, can be predicted to a n accuracy of plus o r minus 13 cubic feet p e r hour within the range of these t e s t s . Operating requirements commensurate with feed gas composition may be observed d i r e c t l y b y reference to c h a r t 1. Comparison with Conventional Scrubbers Using the methods' of Coulson and Richardson, the over -all mass transfer coefficients f o r the agitator absorber w e r e calculated a t t h r e e s e t s of conditions, runs A-1, A-2. and A-12.

'

Until s t r i c t c r i t e r i a f o r the choice of agitator volume a r e r e solved, i t is not possible to make 'a generalized comparison of the performance of the agitator absorber with conventional scrubbers on the b a s i s of over-all mass-transfer coefficients. The principle of operation, and hence the flow I \ mechanics i n the two s y s t e m s , are quite different; indeed, the flow mechanics i t is possible to compare in xt (, thethe agitator a r e ether e m e l y complex. However, specific operating conditions performance of two types of equipment f o r , \ and to describe the performance of the agitator i n t e r m s of a physical tower ' required to effect the s a m e degree of scrubbing at those operating conditions. '. This comparison has been made and is presented i n table 5. It shows the s i z e \ towers that would have to be employed in place of the agitator a b s o r b e r for the \specified runs.
I

Table 5.

4:

Data for comparing agitator absorber to packed tower Packing 314 Raschig rings 3 / 4 Raschig r i n g s 314 Raschig rings C r o s s Tower section, s q . f t . Packing height, ft.

No. AK &

Run

A - 1 0.227 A-2 .262 A-12 .111

0.5

.5
-5

6
10

+KgA for agitator absorber = lb. -mol of CO2 absar bed (hr. ) (atm.CO2 p a r t i a l pressure)(cu. f t . solution) Cubic feet of solution is taken a s that volume of the absorber a c tu'ally occupied b y solution, irrespective .of the s p r a y . F o r these calculations, . that volume was considered t o b e a cylinder 18 inches in diameter b y 12 inches in height. The actual absorber has a c r o s s section of 1.42 sq. ft. and an inside height of 3. 5 ft.

\ '

. . ...

__

. . .

. .-

FACTORS T O B E INVESTIGATED It has been shown that the performance of a n absorber of centrifugal design is g r e a t l y affected by the c h a r a c t e r i s t i c s of the s p r a y p a t t e r n it develops. The s p r a y pattern i n t u r n is a l m o s t totally a function of the turbine design and speed, the i m p e l l e r design, a n d the viscosity of the a b s o r bent. The viscosity of the absorbent is dependent on the concentration of the diethanolamine and on the heat of absorption developed. Consequently, this factorial will be expanded to include the f a c t o r s (1) turbine speed, (2) concentration of absorbent, and (3) heat of absorption. APPENDIX I THE BLOCK DESIGN FOR FACTORIAL EXPERIMENTS A AND B F a c t o r i a l Experiment A
7

F a c t o r i a l Experiment B

i K -1 1 -1 1 1 1 -1 1
0

Set B

-1 -1 1 1 -1 -1 1 1
0

1.
1 1 1 1
1

-1 -1 -1 -1 1
1

32
0 1
O 1

1 1

O/
1

O
O
0

0
0 ~ 0 ! 0 1

O I

01

0 0 0 -- -

1 o i
LEVELS

Factorial A Interactions

Factorial B Curved Effects Low High Low

High

Medium

300 250 200 321 179 A = ressure, p.s.i.g. B = 8 0 2 content of foul g a s , mol-percent 20 15 10 22.4 8 C = solution throughput gal. /hr. 90 75 60 96.2 53.8 D = liquid-to-gas r a t i o , gal. /lo00 std. cu. f t . / h r . 55 35. 7 16.4 58.8 12.6 The l e v e l designations: 1, -1, 0, -y>, 1.2 and -1.2 may be defined a s the value of the variable in question minus the mean value of that variable divided by the interval for that variable. The interval i n t u r n may be defined as the value of the variable designated a s 1 minus the value of that variable designated as 0.

i,c,

LITERATURE CITED

1. Coulson, J. M. and Richardson, J. F., Physical P r i n c i p l e s of Chemical P r o c e s s Design, McCraw-Hill, N. Y., Vol ZI, (1956).
2. 3. Davies, 0. L. , Design and Analysis of Industrial Experiments ; Hafner Publishing Co., N. Y., Appendix 11-A, pp 552-561, (1954). Dixon, B. E . , Some Aspects of the Absorptive Mechanism in Centrifugal A b s o r b e r s , T r a n s . Instn. Chem. E n g r g . , Vol. 32. supplement, pp S - 8 5 4 - 9 5 , (1954). Villars, D. S., Statistical Design and Analysis of E ~ e r i m e n t s for Development R e s e a r c h , W m . C. Brown G o . , Dubuque, Iowa, pp 191 and 378-379. Yates, F., The Design and Analysis of F a c t o r i a l E x p e r i m e n t s , I m p e r i a l B u r e a u of Soil Science, Harpenden, (1937), as found i n Davies, 0. L . , (Ref. 2 above) pp 263-286.

4.

5.

'\