Optimum Temperature Gradient in Selective olvent Extrac E. J. REEVES Magnolia Petroleum Company, Beaumont, I k x .
T h e relations ,suggested by Kalichevsky for estimating yields and properties of products from the solvent refining of lubricating oils are expanded to include the temperature gradient used to increase the efficiency of counterflowextraction towers. Analysis of the new functions proves the existence of optimum raffinate and extract reject temperature which give the greatest raffinate yield. The numerical values of these optima are found by substitution of experimentally determined constants in simple algebraic expressions.
S
ELECTIVE solvent extraction offers a useful method of separating liquid mixtures and has been employed for the past twenty years in petroleum refining. Because of the extremely large number of organic compounds contained in such mixtures, theoretical knowledge of the extraction equilibria which would be of assistance in the solution of practical problems has been meager. The first theoretical approaches to the solvent extraction of lubricating oils were based on the classical laws of physical chemistry, ?;ash (IO) assumed the oil to consist of two components and attackrd the problem by means of distribution coefficients. Similar equations were later published by Hunter and Nash (55 ) . Saal and van Dijck (If) considered extraction analogous to distillation and proposed a diagram of the LlcCabe and Thiele type. Successive treatmmt of solute with solvent containing portions of solute was analyzed by Underwood ( I $ ) , and Fisher (2) discussed treatment in orifices and packed towers. Rleulen (9) discussed the use of triangular diagrams and pointed out the difficulties of theoretically rxpressing rstractmn equilibria for complex hydrocarbon mixtures The errors of ext,rapolation or interpolation 0.80 in most of these theories 0.a can be traced to the use of distribution coeffici0.40 ents and the assumption t,hat the extract and rafinate fractions are 0.20 pure compounds. KaliW chevslry (6) reasoned that lubricating oil fracF 0.10 tions should be con5 0.80 eidered as a continuous E 0.00 phase .\Those properties s h o ~gradual changes, 0.40 and derived from experimental data empirical solvent extraction equao.eo tious expressing the extract, yicld as
Since both solvent and oil act as solvents for each other, a modified form of Equation 1 was presented in a subsequent publioation (8) for calculating t,he solvent, fraction in t,ho raffinate. This relation may be expressed as log S R = (a
SOLVENT TO OIL RATIO IN EXTRACT LAYER, S,
Figure 1. Effect of S E on E i n Extraction of Naphthenic Distillate withvarious Solvents
(2)
(C
Counterflow extraction towers are favored by many designers in preference to the early-stage ext,ractors and are widely used by industry. These tovers are operated by applying a temperature gradient between the extract and raffinate exit ends for providing internal reflux and thus raising the extraction efficiency. All the data used in derivation of the foregoing relations were obtained from laborat,ory constaiit-t,emperature extractions. To express accurately the solvent extraction equations for systems in which the raffinate is withdrawn at one temperat,ure and the extract at. another, new relations must be found. Analysis of constant-temperature extraction data available in the literat,ure shows that t,he extract yield may be expressed as log E
=
c log S E + C'
(3)
and fraction of total solvent loss to raffinat'e as log
XR = K
log R'
+ K'
(4)
Pertinent data substantiating these relations, published by Ferris et, al. ( I ) , are shown in Table I and Figures 1 to 4. Assuming a temperature effect similar to that found by Kalichevsky, Equations 3 and 4 become
TABLEI. SINGLE-STAGE
p>XTRACTlON O F P A R A F F I N I C n I S T I L I A T E S WITH VARIOUS
Solvent Sitrobenoene a t 100 c.
hniline a t 65" C .
0-
0.10
+ b'l')lag R' + + fT)
Furfural 980 C.
at
Pyridine at 0' C.
Solvent- Solvent- Oil-Solvent Oil Oil Ratio Ratio in Charge in Ext. ltaffinate Ratio Layer Layer Naphthenio Distillate 0.60 0.93 2.70
NlPIITHENIC SOLVENTS ( I )
AND
Yield
Fraction of Total Solvent in Raffinate Layer
Ext.
0.75 1.00 1.50 2.00
1.23 1.74 2.39 3.23
3.29 3.87 5.25 5.71
0.41 0.48 0.60 0.59 0.63
0.36 0.21 0.13 0.052 0.032
0 .BO 1.00 2.00 3.00
2.18 3.31 5.17 6.19
2.93 3.92 4.49 6.63
0.14 0.24 0.36 0.47
0.49 0.19 0.071 0.027
0.60 1.00
2.13 2.86 4.55 7.69
3.46 4.00 6.09 6.87
0.18 0.29 0.42 0.64
0.39 0.18
1.00
1.65 2.43 2.84 4.88 6.40
2.01 3.00 3.08 9.76 14.38
0.44 0.55 0.66 0.71 0.86
0.27 0.10 0.056 0.0066 0,0020
2.00 5.00
1.50 2.00 3.00 5.00
0.048
0.010
Paraffinic Distillate
nT)X + ++ PT) (1)
log E = ( m l o g 8 (P
for a constant number of stages.
1490
Xitrobenzene a t 10' C.
0,50 1.00 3.00 5.00
1.40 2.35 5.41 6.51
3.85 4.74 7.00 6.87
0.20 0.37 0.56 8.76
0.13
Eulfur dioxide a t -74 c.
1.00 3.00 7.00 10.00
9.00 19.00 49.0 48.5
9.00 5.66 4.26
0.11 0 .14 0.17
0.10 0.05 0.027
4.00
0.22
0.019
0.41 0.021 0.007
July 1949
INDUSTRIAL AND ENGINEERING CHEMISTRY
1491
LT 0.60 VI
s’ 0.40
5
g;
0.20
These relations are of the utmost importance in that they give optimum treating temperatures for highest raffinate yield and solvent. efficiency when the number of stages is constant.
0.10
e
I 9
0.08 0.06
0.04
c
2
$
IMPORTANCE OF TEMPERATURE GRADIENT
Laboratory experimental data ( 7 ) indicate that a temperature gradient between extraction stages is necessary only when the number of stages is small. As the number of extraction stages increases, the differential temperature required for highest extractor efficiency decreases. When a large number of stages is used, the extraction may be conducted a t constant temperature without product loss.
0.02
22 0.01
E P
0.008 0,006
0,004
0.002
,
10.
0,001 1.0
2.0
4.0
W
6.0
0 0 150 2.0
0 . 4
60 8 0
20
4.0 B.0 B,O
0.8
w 0.6
OIL TO SOLVENT R A T 0 IN RAFFINATE LAYER, R‘
0-
d-
Figure 2. Plot of R’ against SR in Extraction of Naphthenic Distillate with Various Solvents
0.4
>
5 ,0.2 W X
log E
log S
(m’ R =
+ n’T)lOg AYE f
(p’
+ b’P‘)lOg R’ + (e
(a’
+ p’T)
(5)
f’T)
(6)
Equation 5 represents the extract and Equation 6 the raffinate terminal compositions. For an extractor operated with a temperature gradient, these relations are expressed as log E’
(W
+ zT1)log + (Y +
log S R = (r
xE
ZT1)
+ sT2)log R’ + (u+ V T Z )
0.1
1.0
2.0 4.0 6.0 6.0 10.0 20.0 4QO 60.0 SOLVENT TO OIL RATIO IN EXTRACT LAYER, SE
Figure 3. Effect of SE on E i n Extraction of Paraffinic Distillate with Two Solvents
(7 1
Commercial extraction towers are designed t o balance cost and
(8)
. number of stages economically. For this reason they usually have
Equations 7 and 8 give terminal conditions for the extractor. Temperatures of the intermediate st,ages may effect the over-all separation; however, for this article it is assumed that the terminal conditions define the over-all equilibria. Limited data available to the author substantiate the assumption:
a relatively small number of stages and must be oyerated with a temperature gradient to achieve maximum efficiency. Since the extraction stages vary from tower to tower and are affected slightly by severity of the solvent treatment, the optimum temperature gradient necessary to obtain greatest raffinate yield, a s defined by Equation 16 and 17, would vary somewhat with the equipment under consideration.
Relation 9 holds for a constant number of stages. Since
R‘ =
1 x
E = 1 - R
(10)
(12)
Equation 9 becomes
By analysis of the partial derivatives it is possible to determine the most effective treating temperatures. Partial derivatives of R are: Figure 4. Plot of R’ against S R in Extraction of Paraffinic Dis-
dR
tillate with Two Solvents
dR _
(w
1 - R
+ zT1) (16)
Equating these derivatives to zero and solving for Ta and TI,
I n the equations presented, only yield relations have been conkidered; nevertheless, from the proces6 viewpoint raffinate quality is of equal importance. It is believed that quality relation8 similar to the yield equations shown in this paper can be d e veloped.
INDUSTRIAL AND ENGINEERING CHEMISTRY
1492
CONCLUSIONS
5':
Mathematical analysis of solvent extraction equilibria shows that the temperature gradient has an optimum which depends on the quantities and nature of solvent and oil used and on the number of estraction stages, other conditions being equal. For a tower of fixed efficiency. the most efficient utilization of solvent to obtain the highest yield requires maintenance of optimum raffinate and extract reject temperatures which are expressed, respectively, by the relations
T,
=
- (.
Ti =
+ l)/s
-W/X
ACKNOWLEDGMENT
The rviter n-ould like to express appreciat'ion to J. IT;.Newton, \I7,IT,Leach, P. L. Smit'h, and V. i\. Kalichevslry for permission to publish this nianuscript. I-Ielpful suggestions nnd criticisms bv A. J. Hartsook, Rice Institut'e, and J. H. Lee arc gratcfully acknowledged. SOMENCLATURE
E
= = = SR = = E'
S T
Vol. 41, No. 7
extract yield solvent to charge oil ratio extraction temperature fraction of total so!vent in raffinate phase at T P ratio of oil to solvent in raffinate phase a t 57,
= solvent to oil ratio in raffinate phase a t 2'2 1'2 = raffinate reject temperature TI = extract reject temperature R = raffinate yield S E = ratio of solvent to oil in extraFt phase a t TI m, n,p , q , a,b, c,S, C, C', IC,K ' , nz', n , p ' , q', a ' , b', e ' , f', w,x, y , P , r , sI ti, v = constants
LITERATURE CITED (1) Ferris, S. W., et al., IKD.ENG.CHmi., 23, 753 (1931). (2) Fisher, Raoul, J . tech. P h y s i k , 10, 163 (1929). ( 3 ) Hunter, T. G., and Nash, A. W., W-odd Petroleum Cungr., London, f 933, Proc. 2, 340 (1933). (4) Hunter, T. G., and Nash, A. W.,IND. EXG. CHEM.,27, 836 (1935). (5) Hunter, T. G., and Nash, A. W., J . Inst. Petroleum Tech., 22, 49 (1936). (6) Kalichevsky, V. A., ENG.CHEM.,38, 1009 (1946).
TND.
(7) Kalichevsky, V. A., 'Modern Methods of Refining Lubricating Oils," A.C.S. Monograph 76, p. 131, New York, Reinhold Publishing Corp., 1938. ( 8 ) Kalichevsky, V. A , , XutZ. PetroZezim S e w , 38, R-013 (1946). (9) hleulon, H., J . Inst. Petroleum, 29, 237 (1943). Oil Gas J . , 31, No. 45, 62 (1933). (10) Nash, A. W., (11) S a d , R. N. J., and Dijck, W.J. D. van, TYorZd Petroleum Congr., London, 1933,Proc. 2, 354 (1933). (12) Underwood, I n d . Chemist, 10, 128 (1934). RECEIVEDOctober 11, 1947. Presented before t h e Southwest Regional Meeting of t h e AMERICAN CHENICAL SOCIETY, Houston, Texas, Dcceinber 13, 1'247.
AND KENNETH A. EARNARTI 0'. S. Industrial Chemicals, Inc., .Vewark 5 , N . J .
CATHERINE COSGROVE
T h e maleic anh5 dride adduct of cyclopentadiene reacts with linseed oil at elevated temperatures to give a product similar to those obtained when maleic anhydride or fumaric acid reacts w-ith linseed oil. The maleic anhydride adducts of butadiene, isoprene, 2-methj lpentadiene, and piperylene do not react with linseed oil under the same conditions. A tentative explanation is offered for this diff'erence in behavior of these maleic anhydride adducts.
ALEIC anhydride and similar unsaturated acids have long been known to react with the double bonds of linseed oil and its alcoholysis products a t fairly high temperatures--that is, above 150' C. (6, 9, 10, I @ , I n this investigation it was desired to determine vhether several unsaturated acid anhydrides, which had been prepared by a Diels-Alder reaction between maleic anhydride and unsaturated, conjugated dienes, would react with linseed oil in a manner similar to maleic anhydride. Biclcford and eo-workers (3) devised experiments on the reaction between maleic anhydride and the methyl esters of several unsaturated, long-chain fatty acids. They concluded that the first molecule of anhydride reacting with methyl linoleate and the first two molecules reacting with methyl linolenate react mostly to saturate one double bond per mole of anhydride reacted, and that the ratio between the number of double bonds of the methyl ester saturated to the moles of anhydride reacted approaches unity as less anhydride is used. Furthermore, they found that maleic anhydride reacts more rapidly with the linolenic and linoleic esters than with the oleic ester. 1 Present address, U. S. Industrial Chemicals, Inc., P.O. Box 1956, Baltimore 3, Md.
I n the present experiments 0.602 mole of unsaturated acid anhydride was used (except for the maleic anhydride adduct of substituted cyclopentadiene) for each mole of linseed oil. If we assume that any addition of these anhydrides t o the linseed oil would be similar to the addition of maleic anhydride to the methyl esters, as determined by Bickford et al. from their Figure 1 (S), we can conclude that practically all of the anhydride reacted with the linoleic and linolenic acid chains; from their Figure 4 (3)we can conclude that the use of 0.602 mole of anhydride for approximately 2.04 molw of linoleic and linolenic acid collectively in 1 mole of linseed oil should result in the average saturation of nearly one double bond for each mole of anhydride which reacted. Table I shows the unsaturated acid anhydrides used in this investigation. Fumaric acid, maleic anhydride, and phthalic anhydride mere used as controls. These acid anhydrides 1% ere reacted with the unsaturated, fatty acid esters by thrcc different methods. TREATMENT I
Each unsaturated acid anhydride (0.62 mole), linseed oil (1.03 moles), and glycerol (0.51 mole) were heated together a t 260" C. under a blanket of carbon dioxide. Maleic anhydride, fumaric acid, 1,2,3,6-tetrahydrophthalic anhydride (butadiene M.A.A.), 1,2,3,6-tetrahydro-3,6-endomethylenephthalic anhydride (cyclopentadiene M.A.A.), and the substituted cyclopentadiene M.S.A. produced two phases before reaching 260 C. 1,2,3,6-Tetrahydro-4-methylphthalic anhydride (isoprene M. A.A.), 1,2,3,6-tetrahydro-3-methylphthalioanhydride (piperylene M.A.A.), and 1,2,3,6-tetrahydrc-3,5-dimethylphthalican-