MAY 1986 ISSN 0018-8190

Hydrocarbon
Processing

REFLUX AND SURFACE TENSION EFFECTS ON DISTILLATION

Reflux and surface tension
effects on distillation
Theoretical and experimental data suggest More difficult separation systems (relative volatility close
to 1.0) have more uniform separation efficiencies over the
how to set reflux ratio for the most efficient entire concentration range (i.e., X has a lesser influence).
distillation separation in packed towers. A difference in surface tension can explain some differ-
ences in efficiencies. Comparing two easy separation sys-
Surface tension effects also are considered tems, one with a negative surface tension gives a lower sepa-
ration efficiency than another with a positive surface
T. D. Koshy and F. Rukovena, Norton Co., Akron, Ohio tension.
EASY SEPARATION SYSTEMS (those with relative volatili- Operating efficiency. The operating efficiency of packed
ties of approximately 2.0 or greater) are influenced signifi- towers is generally expressed as Height Equivalent to a The-
cantly by reflux ratio. Theoretical relations suggest the best oretical Plate (HETP) or Height of a Transfer Unit (HTU).
separation efficiency occurs when X, the ratio of the slopes of But note, a minimum HETP or HTU represents a maxi-
the equilibrium and operating lines (mG\{/L\), is approxi- mum separation efficiency.
mately 1.0. Then, experimental data are obtained to relate X The effect of reflux ratio is shown in Fig. 1, a plot of
to its equivalent reflux ratio. HETP versus a vapor capacity factor CY These data are for
a methanol-water binary distillation at atmospheric pres-
2.0 sure. The upper plot is for total reflux. The lower plot is for
an internal reflux ratio, L\i/G\, of 0.5. The HETP is signif-
icantly larger for total reflux than for partial reflux.
o o These data were obtained at Norton Company's Cham-
0 a berlain Laboratories in Stow, Ohio. The distillation tower
1.5 had a diameter of 381 mm and was filled with a 3,048-mm-
o deep bed of No. 25 IMTP stainless steel packing. Additiona
0
runs were made to examine the relation between HETP ana*"
reflux ratio. Further, HETP measurements were conducted
at a fixed boil-up rate to eliminate the effect of Cy on HETP.
1.0
Theoretical analysis. A generalized (HTU) (X ; correlation
assuming a two-resistance theory 1 - 2 is given as follows:
(HTU) 0(; = (HTU) C; + X(HTU),. (1)
0.5
If the equilibrium curve and the operating line are as-
sumed to be straight lines over a single stage, the following
Total reflux relation is obtained:
Partial reflux, L/D = 1.0
\1
HETP = (HTU)oo(ln X)/(X - 1) (2)
0.2 0.3 0.4 Then define
Cs at midpoint, ft/s
-Packed tower rectification of methanol-water.
(3 = (HTU) L /(HTU) G
and
= HETP/(HTU)C; (4)
Combining the foregoing equations gives:
= ( l + X / 3 ) ( l n X ) / ( X - 1) (5)

Fig. 2 shows a plot of as a function of In X for various

values of /3. For a fixed (8, the minimum is determined by
setting the first differential of Eq. 5 to zero (i.e., d/dX = 0).
This gives:
/3 = (X - 1 - X In X)/(X In X - X 2 + X) (6)
Substituting Eq. 6 into Eq. 5 gives:
L,,, = (lnX) 2 /(X - 1 - In X)
2 3 456 810- 1 2 3 456 810 2 3 4 56 810
Slope ratio X Table 1 shows the variation of ,, with j3 and X.
Fig. 2Relationship among variables. Finding minimum HETP. If (HTU) ( , is assumed to be a con-
Reprinted from HYDROCARBON PROCESSING, May 1986
Copyright* 1986 by Gull Publishing Co., Houston, Texas.
All rights reserved. Used with permission.
 stant, it follows from Eq. 4 that having a minimum also 2.500
Methanol-water
means there is a minimum value for HETP. And since X, *- Water-DMF
and to some extent /?, varies over the range of concentration
2,000
for which HETP is measured, values of/? and X must be ob-
tained experimentally.
The righthand term of Eq. 2, (In X)/(X - 1), approaches
. 1.500
'.0 as X approaches 1.0. Therefore, Eq. 2 suggests HETP
^approaches (HTU) OG ; and it is feasible to correlate HETP as
a function of the physical properties. It is customary to use
published correlations 5 "' to evaluate (HTU)oo using Eq. 1.
Relative volatility, a, relates the vapor and liquid phase
compositions of the more volatile component in a binary
'*,, .,
separation, i.e., 0%

y = ax/[I + (a - l)x] (8) I I I III

10-' 2 3 4 5 6 7 8910 2 3 4 5 6 7 8910
Here, a is not necessarily a constant. Slope ratio X
On the other hand, if a is assumed to be constant over a Fig. 3Two binary systems compared.
small range of composition, the slope of the equilibrium
curve can be given as follows:
dy/dx = m = a / [ l + (a - \)x]2 (9)
Then when x > 1, m * I/a (10) E
E
Q.-200
and when x > 0, m > a (11)
Thus, for liquid phase compositions between x = 0 and
x = 1.0, m lies between ex and I/a.
Difficult separations (those binary systems whose a. val- 10-' 2 3 4 56 810 2 3 4 5 6 8 10' 2 3 4 56 8102
ues are not significantly larger than 1.0), have values of m Slope ratio X

that do not greatly differ from 1.0. This also implies the in- Fig. 4Literature data for isopropanol-water.10
ternal reflux ratio, L\i/G\i, which is the slope of the operat-
ing line either in the rectifying or stripping sections, for these
difficult separations will not vary significantly from 1.0. TABLE 1Variation of mm with 0 and X
This is the case because the operating line is enclosed within (3 X min
boundaries created by the equilibrium curve and the diago- 0 00 0
,nal. 0.1 28 0.469
As a result, X = mG\IL\, or the ratio of the two slopes, 0.3 5.963 1.003
__^emains close to 1.0 for such systems. One should not expect 0.5 2.817 1.373
X to be an important variable in the performance efficiency 0.7 1.706 1.660
of distillation towers making difficult separations. Under 1.0 1.0 2.0
such conditions HETP = (HTU)oo, and HETP is an ap- 3.0 0.01954 3.219
propriate design tool. 5.0 0.0936 3.836
7.0 0.0583 4.250
Easy separations, on the other hand, where a > > 1.0, 9.0 0.0412 4.560
X can have values much less than 1.0 to much greater than oo 0 00

1.0. Therefore, X becomes an important variable in the per-

formance of a distillation tower making easy separations, es-
pecially when making very pure products.
Surface tension effects can be considered using a defini-
tion given by Zuiderweg and Harmens. 8 They termed distil-
lation systems positive or negative on the basis of whether
the surface tension of the reflux liquid increases or decreases
as it flows down the column.
Experimental data. Two binary systems that can be charac-
terized as easy separations were chosen in order to compare
the performance of two high a systems and, at the same
time, contrast the performance of a system with a positive
surface tension with one having a negative surface tension.
The first system was the methanol-water binary men-
tioned earlier in connection with Fig. 1. The a of this system
ranged from 2.6 to 8.1 and the m ranged from 0.38 to 8.1, at
atmospheric pressure. This is a strongly surface-tension-pos-
itive system. The second binary system was a water-di-
methylformamide (DMF) system whose a varied from 3.6 to
6.8 and m varied from 0.28 to 6.8, at atmospheric pressure.
This latter is a strongly surface-tension-negative system.
The data for HETP versus log X for both systems are illus-
,crated in Fig. 3. The HETP values were calculated using a
modified McCabe-Thiele method taking into account the
enthalpy difference between the components.'' Tower opera-
tions were held within a narrow range of boil-up rates in or-
der to eliminate the effect of tower loading on efficiency. For
both systems, HETP goes through a minimum at a X value
of approximately 1, although boil-up limitations prevented
getting data for the water-DMF system at X values greater
than 1.0.
Other data for an isopropanol-water system reported by
Deed, et al., are shown in Fig. 4.' They performed atmo-
spheric rectification of the binary below azeotropic condition
in a 152-mm-diameter brass column with 1,194 mrn of 13-
mm ceramic Raschig Rings. These data also indicate HETP
goes through a minimum.
Based on the observation that (HTU)oo increased rapidly
as X was raised, Deed, et al., concluded that the principal
resistance to mass transfer was in the liquid phase. This ex-
planation implies that, at large values of X, /? defined by Eq.
3 increases.
On the other hand, Norman, et al., 11 conclude that large
variations in HETP are probably the result of poor liquid
distribution over the packing and variations in the wetted
area. This conclusion was based on wetted wall column data
determined by other researchers in the field who found large
variations in HETP in experiments with packed columns.
However, based on recent experience with liquid distributors
providing a widely varying number of distribution points
and liquid flowrates, there are no indications that liquid dis-

0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
Fig. 5Vapor-liquid equilibrium diagram.
tribution was a limiting factor for our test data.
Surface tension effects. According to Zuiderweg and Har-
mens,8 a binary system is positive if the less volatile compo-
nent has the higher surface tension. They demonstrated
that, in apparatus where the interfacial area is mainly of the
film type, mass transfer rates for positive systems (with stabi-
lized film) may be much higher than for negative systems.
This generalization, however, may fail when applied to
negative systems, where the reduction in contact angles may
result in spray formation. The interfacial surface area in
such formations is not significantly affected by surface ten-
sion changes.
Fig. 3 illustrated that for the range of X values for which
comparison can be made, the HETP for the methanol-water
(positive) system is consistently lower than that for the water-
DMF (negative) system. Meier, et al., 12 who performed dis-
tillation at total reflux on both systems in a structured stain-
less steel tower packing, found that the HETP for the
water-DMF system is approximately twice that of the metha-
nol-water system.
For these tests, the concentration range of the methanol-
water mixture was 10% methanol at the bottom and 98%
methanol at the top. The range of the water-DMF system
was 30% water at the bottom and 99.5% water at the top.
The authors attribute the lower HETP for the methanol-wa-
ter system to its good wettability. The present study, how-
ever, demonstrates that wettability per se does not explain
the widely differing values of HETP obtained for essentially
the same range of separation at different reflux ratios. The
data indicate that at X values much larger or smaller than
1.0, HETP increases dramatically.
Application. The practical importance of our findings can
be illustrated in Fig. 5 and the following:
Case I: X is much less than 1. The circumstances when
this happens are:
high purity of the more volatile component, i.e., m * I/
a.
high reflux ratio, i.e., L_\/G_\i ~* 1.0.
both foregoing conditions mean X * I/a.
Case II. X = 1. The following situations will produce
this condition:
very low reflux ratio for high purity rectification, i.e., x
-* i.o
very high reflux ratio for high purity stripping, i.e., x
-" 0
total reflux for a symmetric separation. Note, the term
"symmetric separation" is used here to mean that on a
McCabe-Thiele diagram, the liquid phase compositions
of the overhead product and bottom product are roughly
equidistant from 0.5.
Case III. X is much greater than 1. The circumstances
when this occurs are:
high purity of bottoms products, i.e., x - 0 and m -
low L/V approaching total reflux, i.e., LIV * 1.0
both foregoing conditions mean that X -* a.
NOMENCLATURE
Cs - capacity factor V s vpc/(pL ~ PG)> m/s
GM - gas flowrate, kg-moles/h
HETP - height equivalent to a theoretical plate, m
HTU - height of a transfer unit, m
LM - liquid rate, kg-moles/h
m - slope of the equilibrium line
Vs - superficial gas flow velocity based on tower area, m/s
x - mole fraction of the more volatile component in the
liquid phase
y - mole fraction of the more volatile component in the
vapor phase
Greek symbols
a - relative volatility between components 1 and 2
/3 - as defined in Eq. 3
X - m GM/LM, ratio of the slopes of equilibrium and
operating lines
- as defined in Eq. 4
p - density, kg/m 3
Subscripts
G - gas phase
L - liquid phase
M - moles of gas or liquid
OG - overall gas phase
LITERATURE CITED
1 Lewis, W. K., and Whitman, W. G., Industrial and Engineering Chemistry, Vol. 16, p.,
1215, 1924.
: Whitman. W. G., Chemical and Metallurgical Engineering, Vol. 29, p. 247, 1923.
:i Shulman, H. L., Ullrich, C. F, and Wells, N., AlChE Journal, Vol. 1, p "
1955.
4 Shulman, H. L., Ullrich, C. F, Proulx, A. Z., and Zimmerman, J. O., AlChE
Journal, Vol. 1, p. 259, 1955.
~ ' Shulman, H. L., Ullrich. C. F., Wells, N., Proulx, A. Z., AlChE Journal, Vol. 1, p.
259, 1955.
6 Onda, K., Takeuchi, H., and Okumoto, Y. J., Chemical Engineering Japan, Vol. 1, p.
56, 1968.
7 Onda, K., Mem. Fac. Eng., Nagoya Univ., 24 (2), p. 64, 1972.
R Zuiderweg, F. J., and Harmens, A., Chemical Engineering Science, Vol. 9, Nos. 2/3,
p. 89, 1958.
y Neretnieks, I., Ericson, I., Eriksson, S., British Chemical Engineering, Vol. 14. No.
12, p. 653, 1969.
1(1 Deed, D. W., Schutz, P. W., Drew, T. B., Industrial and Engineering Chemistry, Vol.
39, No. 6, p. 766, 1947.
11 Norman, W. S., Cakaioz, T., Fresco, A. Z., Sutcliffe, D. H., Transaction of the Insti-
tution of Chemical Engineers, Vol. 41, p. 61, 1963.
12 Meier, W., Hunkeler, R., Stocker, D., Institution of Chemical Engineers Symposium Se-
ries No. 56, 3.3, p. 1, 1979.
The authors
T. Dan Koshy is manager, mass transfer devel-
opment, Norton Chemical Process Products,
Akron, Oft/o, where he manages pilot plant re-
search and data correlation. Earlier he was a
design and development engineer (distillation)
with Union Carbide Corp., Linde Division. He
holds a BS degree from Nagpur University, In-
dia, an MS degree from Wayne State University,
Mich., both in chemical engineering, and an
AlChE membership.
Frank Rukovena, Jr., is manager, mass transfer
engineering and development, Norton Chemical
Process Products, Akron, Ohio, where he man-
ages mass transfer development, packed tower
application and mechanical engineering. Earlier
he was a research development engineer wit
Koppers Co. and a design and project engine*, ..
with General Tire. Holder of a BS in chemical en-
gineering from \bungstown University, he is a
member of AlChE.
Printed in United States