F =

C/C„ Literature Cited

I0 =
Bessel function of zero order and first kind of imaginary
(1) Brut van, D. R., Ph.D. dissertation, Rensselaer Polytechnic
argument Institute, Troy, N. Y., 1958.
L =
length of test section, cm. (2) Cairns, E. J., Prausnitz, J. M., Chem. Eng. Sci. 12, 20 (1960).
A' =
dimensionless axial distance from tracer source =
(3) Carberry, J. J., Bretton, R. H., A.I.Ch.E. J. 4, 367 (1958).
r.v o, (4) Danckwerts, P. V., Chem. Eng. Sci. 2, 1 (1953).
r
A Pe U-tfdp/Dl (5) Ebach, E. A., White, R. R., A.I.Ch.E. J. 4, 161 (1958).
=

A Re =
dplJw'p/µ (6) Epstein, Norman, Can. J. Chem. Eng. 36, 210 (1958).
A».' =
dtJjw'p/ß (7) Gier, T. E., Hougen, J. O., Ind. Eng. Chem. 45, 1362 (1953).
A Re dpL· Pi µ (8) Hayworth, C. B., Treybal, R. E., Ibid., 42, 1174 (1950).
(9) Johnson, H. F.. Bliss, Harding, Trans. A.I.Ch.E. 42, 331
R

U =
interstitial velocity, cm./sec.
Ur =
interstitial velocity of ketone phase, cm./sec. (1946).
(10) Keith. F. W., Hixson, A. N.. Ind. Eng. Chem. 47, 258 (1955).
Uw' =
velocity of water based on empty tube, cm./sec. (11) Kreager, R. M., Geankoplis, C. J., Ibid., 45, 2156 (1953).
Uw =
interstitial velocity of water phase, cm./sec. (12) Levenspiel, Octave, Ibid., 50, 343 (1958).
VK =
flow rate of ketone, cc./min. (13) Licht, W., Conway, J. B., Ibid., 42, 1151 (1950).
Vw = flow rate of water, cc./min. (14) Licht, W., Pansing, W. F., Ibid., 45, 1885 (1953).
1* =
volume of test section of tower, cc. (15) Liles, A. W., Geankoplis, C. J., A.I.Ch.E. J. 6, 591 (I960).
v =
flow rate, cc./sec. (16) Miyauchi. Terukatsu. U. S. At. Energy Commission. Rept.
flow rate of water, cc./sec. UCRL-3911 (1957).
i’ir =

X =
Z =
t =
See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.
axial distance from point of tracer injection to point of (17) Perrv, J. H., “Chemical Engineers’ Handbook,” McGraw-
Hill, New York, 1950.
observation, cm. (18) Sleicher, C. A., Jr., A.I.Ch.E. J. 5, 145 (1959).
X -
ue (19) Streeter, V. L., “Handbook of Fluid Dynamics,” McGraw-
volume fraction or holdup of phase Hill, New York, 1961.
(20) "'Vogt, H. J., Geankoplis, C. J., Ind. Eng. Chem. 46, 1763

eK —
volume fraction or holdup of ketone phase
(1954).
Downloaded via UNIV OF WESTERN AUSTRALIA on October 15, 2022 at 14:31:40 (UTC).

eir =
volume fraction or holdup of water phase
=
time, sec. Received for review August 17, 1962
reduced time Accepted July 12, 1963
=
vwe/ewV =

µ =
viscosity, gram cm. sec. Financial assistance in the form of an Ohio State University
p
=
density, gram cc. fellowship is acknowledged.

PRESSURE DROP IN HORIZONTAL AND

VERTICAL COCURRENT GAS-LIQUID FLOW
G. A. HUGHMARK, Ethyl Corp., Baton Rouge, La.

A new "lumped type" pressure drop correlation for gas-liquid flow utilizes a lost work term from an energy
balance derived equation for gas-liquid flow. Experimental data for horizontal, vertical upward, and
vertical downward flow show that the lost work term is a function of the pipe orientation. The average
absolute deviation between experimental and calculated pressure drop data is 19% for horizontal flow in
0.5- to 0-inch pipe and 5% for vertical upward flow in equivalent diameters to 2!/2 inches. The correla-
1 1

tion is applicable to all gas-liquid flow regimes.

of two-phase gas-liquid pressure drop in pipe

rediction Ifa linear pressure change with length is assumed for the two-

subject of interest for many years. An ap- phase section, the mechanical energy equation derived by Lamb
* has been a

preciable amount of experimental data has been obtained and White and by Vohr (38) can be reduced to the form sug-
for isothermal two-phase flow with the greatest emphasis gested by Hughmark and Pressburg (21).
on horizontal and vertical upward flow. Many correlations (Wl + Wg)(H2 Hi) + -

have been proposed for these data. The correlations apply

Pi P2
—

+
= —

gc ( Wlvl + Wove)
with reasonable accuracy to the data on which they are based, Kl(V!, Vh) + Wg(V¡2 - -

VS,)
(2)
but are limited in some form of application. Either they do not 2gc( Wlvl + Wove)
extrapolate to data other than those on which they are based or It is apparent that Ca and are not identical. Cd can be
they are valid only for specific flow conditions. None of the considered as a drag coefficient applied to the pipe wall, which
correlations apply with reasonable accuracy to all of the experi- has a positive sign in most flow conditions but is negative for
mental data. conditions of vertical upward flow in which the net liquid flow
Lamb and White (26) recently presented derivations of at the pipe wall is downward. represents the irreversible
momentum and mechanical energy equations to show their energy loss through the two-phase section and must be positive
relation to pressure drop correlations in two-phase flow. The for all flow conditions.
derived momentum equation can be rearranged to the form The major part of the isothermal two-phase data has been
suggested by Martinelli and Nelson (28): obtained at conditions in which the momentum or kinetic

energy change is negligible. When this term is negligible, the

Pi -
Pt =
Cd+ (plRl + poRg)(H2 -

Hi) + momentum equation gives:

1 / 5!_ _

_|_
Gj _
Gg
(1) Pi -
P2 =
Cd+ i (plRl + pgRg)(H2 -

Hi) (3)
gc \plRl2 plRl, pg,Rg, PG-Rg,

VOL. 2 NO. 4 NOVEMBER 1 9 6 3 315

and the mechanical energy equation:
£ (WL + Wa)(H, -
Hi)
Pi —
Pi =
+ (4)
gc ( Wlvl + Wove)
The equations are further simplified for horizontal flow with
negligible momentum or kinetic energy change to:
Pi ~
Pi =
Cd
Pi —
Pi =
This is the only condition at which Cd and are equal.
The most generally accepted method for predicting Cd or
APTP for two-phase flow is that proposed by Lockhart and Mar-
tinelli (27). The method is based upon data for flow in 0.62-
to 1-inch horizontal pipes with a pressure range from atmos-
pheric to 50 p.s.i.g. It does not extrapolate well to larger
diameters or higher pressure. Chenoweth and Martin (77)
presented a method of correlation similar to that of Lockhart
and Martinelli in that a ratio of the two-phase pressure drop to
the pressure drop for the liquid is used. This applies to large
as well as small pipe diameters and to high pressure. It is
restricted to turbulent flow in horizontal pipe. Bertuzzi,
Tek, and Poettman (7) proposed a correlation for horizontal
flow that uses a two-phase / factor. The / factor is similar to
the friction factor for single-phase flow with homogeneous flow
assumed for calculating the velocity and the density to be used
with the correlation. This correlation is applicable to all
conditions of flow except when the gas mass flow rate per unit
time is greater than that of the liquid. Hoogendoorn (78)
has suggested individual correlations for the specific flow re-
gimes. Pressure drop for horizontal annular flow has been
considered by Aziz and Govier (3) and Wicks and Dukler
(47).
For vertical upward flow Galegar, Stovall, and Huntington
(74) and Tek and Chan (36) have proposed friction factor-
type correlations for the combined potential energy and APTP
terms. Hughmark and Pressburg (27) predicted APTP
from the slip velocity and a combination of the physical prop-
erties and the mass velocity. Calvert and Williams (70),
Anderson and Mantzouranis (2), and Collier and Hewitt (12)
have analyzed pressure drop for annular flow.
Prediction of pressure drop for gas-liquid flow depends upon
prediction of the terms Cd or APTP. For horizontal flow with
negligible kinetic energy, the two terms have been shown to be
identical. For vertical flow, the terms are not equal to each
Figure 1. Horizontal flow pressure drop
316 I & E C FUNDAMENTALS
other and are not equal to those for horizontal flow under
identical flow conditions. Therefore, these terms must be
considered as a function of pipe orientation as well as flow rates
and physical properties of the phases. Correlation of the term
APTP was selected for this work.
The irreversible energy loss for single-phase fluid flow is
(5) designated as lost work and can be expressed in units of foot
pounds force per pound mass. This lost work for single-phase
(6)
flow is the product of the friction pressure drop and the fluid
specific volume. By analogy the lost work for two-phase flow-
can be considered to be the product of the two-phase pressure
drop, APTP, and the specific volume of the mixture in the pipe
section. Thus a correlation can be used for the two-phase lost
work if the two-phase specific volume can be predicted. This is
the correlation method used in this paper.
In gas-liquid two-phase flow, the difference in the physical
properties of the two phases causes the phases to flow at differ-
ent average velocities in the pipe. Therefore, a knowledge of
the mass flow- rates per unit time and the specific volume of each
phase does not define the specific volume of the mixture in the
pipe. Hughmark (79) has recently presented a method for the
determination of holdup in gas-liquid flow which is based upon
the variable density single fluid model proposed by Bankoff
(5). The holdup is the fraction of the pipe occupied by a phase
and can be used to calculate the specific volume of the mixture
in the pipe for the lost work correlation.
Comparison of Two-Phase Lost Work for
Horizontal and Vertical Flow
The lost work terms for horizontal, vertical upward, and
vertical downward cocurrent isothermal flow were calculated
with APTP defined by Equation 2 and the two-phase density
obtained from the holdup. Figures 1 through 4 show these
lost work terms as a function of the liquid volume fraction, yL,
and the Froude number. These are defined as:
Wlvl
VL
Wlvl + Weva
Frrp ^WLVL±_Wovay
Id
The lost work term as shown on the plots is modified by sub-
traction of the friction lost work for the total mass flowing at the
Figure 2. Correlation for vertical upward flow pressure
drop
Frrp > 1
Figure 3. Correlation for vertical upward flow pressure
drop
Fr tp < Vertical downward flow pressure drop
Figure 4.
1

liquid condition in the pipe. This modification forces the phase region if the two-phase friction factor is divided by a
Froude number parameters to approach zero at total liquid friction factor for single-phase flow corresponding to a Reyn-
flow. olds number for the mixture. One form of a Reynolds number
Figure 1 is obtained from the data of Taylor (35) and John- for a mixture is:
son and Abou-Sabe (25) for horizontal flow in 0.87- to 1-inch
diameter pipe. At a constant value of yL, it is observed that Rlpl + (1
—

RL)µß
the lost work increases with increasing Froude number.
This meets the requirement of reducing to the Reynolds num-
Figures 2 and 3 represent the data of Hughmark (20), Sher
ber for the single phase at the conditions of total liquid and
and Green (33), and Bennett and Thornton (6) for vertical
total gas flow. Liquid volume fraction, yL, derived from the
upward flow. At a constant value ofyL and at very low Froude
mass flow rates per unit time and the densities of each phase
numbers, the lost work decreases to a minimum corresponding
can be used to define the range from total liquid, yL 1, =

to a Froude number of 1. The lost work then increases with

to total gas 0. =

increasing Froude numbers above 1.

Figure 5 shows the correlation obtained for the friction factor
Figure 4 is based on the vertical downflow data of Bryant (9)
ratio as a function of yL and the parameter (Rerp) (p,·/pL) X
for a 20-foot section of 1-inch pipe. The lost work at a constant
value of yL passes through a minimum as for vertical upward (0.085/73). (The coordinate points for these correlations can
be obtained from the author.) The term (0.085/73) is the
flow, but here the minimum corresponds to a Froude number
of 100. ratio of a 0.085-foot I.D. pipe to the experimental pipe inside
diameter in feet. Data are available at values of yL less than
Comparison of the four figures definitely shows that the
0.001 but are too limited in number to justify extending the
lost work for gas-liquid flow is dependent upon the pipe orien-
tation. At Froude numbers above 5000 the lost work for the plot into this region.
three orientations approach each other, apparently because the
Table I shows the range of the experimental test conditions
effect of gravity is negligible at these high mass velocity condi-
for the data represented by the horizontal two-phase flow
tions.
correlation (Figure 5). Table II gives the average absolute

Correlation for Horizontal Flow

The lost work correlation shown by Figure 1 applies to data

for a smooth pipe of about 1-inch diameter but does not apply
to data for pipe with other diameters and roughness conditions.
A general correlation was obtained using a two-phase friction
factor:
2 geDlv
J1P (7)
LV2pTp

This is suggested by the relationship with Froude number

shown by Figure 1 when
Equation 7 is rearranged in the form
,
=
2g lw 1
JTP - 5-
—

gi Lprp rrrp

As the velocity and Froude number represent the total flow

rate (liquid plus gas), the friction factor at the extreme condi-
tions of total liquid or total gas flow is that for the single-phase
fluid. The two-phase friction factor can be divided by the
friction factor for single-phase flow and the resulting ratio will
go to unity for all conditions of total liquid or total gas flow. A Figure 5. Correlation for horizontal flow pressure
friction factor ratio can be applied through the entire two- drop

VOL. 2 NO. 4 NOVEMBER 19 6 3 317

 Table 1. Experimental Test Conditions for Data Included in Correlation for Horizontal Flow
Av.
Series Reference Ripe L.D., Pressure, Liquid Flow Rate, Gas Flow Rate,
No. No. Inches System P.S.I.A. Lb./Hr. Lb./Hr.
1
(11) 1.602 Water—air 15-30 178-31,600 56-1.780
2 (11) 1.602 Water-air 100 100-31,600 56-3,160
3 (11) 3.072 Water-air 15-30 560-31,600 56-3,160
4 (11) 3.072 Water-air 100 316-31.600 56-3.160
5 (35) 1 .017 Water-air 15-20 180-8,300 2.3-230
6 (35) 1.017 Kerosine-air 15-20 115-5.200 3-250
7 (35) 1 .017 Oil blend 1-air 15-21 170-3,800 21-250
8 (35) 1.017 Oil blend 2-air 15-21 180-1,800 22-200
9 (35) 1 .017 SAE 40 oil-air 16-22 170-1,200 20-195
10 (35) 1 .017 Diesel fuel oil-air 15-19 140-9,500 3-250
11 (37) 0,622 Benzene-air 49-56 50-270 17-95
12 (37) 0.622 Kerosine-air 46-56 62-280 18-100
13 (37) 0.622 W ater-air 43-54 28-300 13-110
14 (25) 0.87 Water-air 16-65 975-15.000 5-790
15 (24) 1.0 Water-air 15 167-2.940 8.3-247
16 (31) 4 Water-air 21-31 50,000-100,000 460-1.850
17 (31) 6 Water-air 27-30 96,000-250.000 460-2.200
18 (18) 0.945 Gas oil-air 17-33 1.400-14.000 0.002-0.2
19 (15) 1.026 Water-air 36 12.7-64,800 0.9-195
20 (1) 1.049 Water-air 15-53 155-8.400 2.4-195
21 (1) 1 .049 SAE 10 oil-air 15-50 87-6,600 2.1-225
22 (4) 7.75 Oil-gas 900 750-75.000 8,200-50,000
23 (4) 10.14 Oil-gas 900 2,300-75,000 8.200-50,000
24 (40) 1 .025 Water-air 17 520-3.126 127-516
25 (3) 1.049 Water-gas 135 61-3.270 350-2,300
26 (3) 2.067 W ater-gas 135 16-2,760 720-4.000
27 (22,30) 1.062 Water-steam 30 910-2.450 25-750
28 (22,30) 1 .062 Water-steam 60 500-3.800 130-1.170
29 (22,30) 1 .062 Water-steam 100 215-2.925 92-1,210
30 (22,39) 0.484 Water-steam 415 180-4.150 31-1.365
31 (22,39) 0.484 Water-steam 615 115-2.900 23-1.430
32 (22,39) 0.484 Water-steam 815 100-3.400 17-1,380
33 (22,29) 0.484 Water-steam 1015 28-3.400 30-1,400
34 (22,29) 0.484 Water-steam 1215 148-2.520 25-1,350
35 (22,29) 0.484 Water-steam 1315 310-2,840 15-1,250
36 (22,29) 0.484 Water-steam 1415 82-2.700 56-1.395

Table II. Comparison of Experimental Horizontal Flow Data with Correlations

Average Absolute Deviation, APtp, %
Lockhart- Chenoweth- No. of Tests for
Series No. Martinelli Martin Poettman Hughmark Hughmark Method
1 19.2 14.9 68 15.5 59
2 38.6 19.9 32.0 14.3 77
3 21 .9 23.2 72 21.6 61
4 75.3 18.5 35.2 24.3 46
5 18.6 20.1 30 26.4 22
6 21.7 15.6 45 28.8 33
7 19.4 19.3 25.3 59
8 18.7 22.3 12.3 33
9 31 25.5 12.2 20
10 17.2 22.3 43 26.7 50
11 18.2 25.2 20.7 20.8 18
12 29.6 14.3 27.9 22
13 25.7 20.8 17.5 21.6 28
14 28 26.7 55 27.5 25
15 24.1 49 83 18.9 228
16 13.0 6.3 30.5 11 .7 15
17 19.7 8.0 25.5 6.0 18
18 25.2 4.8 23.5 9.2 18
19 42.3 19.6 49.5 24.5 29
20 22.7 22.6 27.4 10.1 40
21 32.6 46.5 16.9 31
22 90.5 43 55 42.5 15
23 63 45 54 38 8
24 8.2 12.7 35.2 6 9 57
25 31 .3 37.8 35.4 24 30
26 28.4 34.8 14.4 24.6 25
27 18.5 18.1 85.5 23 23
28 22.9 17.8 69.5 18.2 26
29 24.7 22.4 47 15.3 27
30 60.2 26.6 32.1 19.5 48
31 63.1 20.1 24.9 13.3 31
32 92 24.7 26.6 11 .4 33
33 45 15.3 24 11.6 52
34 61 .6 20.8 35.2 10.6 30
35 69 15.2 22.9 8.3 20
36 67 23.4 45.5 11.2 22
Av. 36.3 22.4 38.4 18.9 Total 1379

318 l&EC FUNDAMENTALS

 Table III. Coordinates for Flow Parameter K in High
Viscosity Liquids
(79, Figure 1)
K
z 70-cp. viscosity 200-cp. viscosity
3.0 0.52
6.0 0.61 0.50
8.0 0.66 0.56
10 0.69 0.60
15 0.76 0.69
20 0.82 0.77
30 0.90 0.88

deviation for the two-phase pressure drop, , between the

horizontal experimental data and the proposed method. Ab-
solute deviations for three of the published methods are also
shown. The method appears to give about the same deviation
as the published methods where these methods are applicable
and has the advantage that it appears to be applicable to all of
the available data. Figure 6 compares the calculated and
experimental values of fTP/f for representative horizontal two-
phase flow data.
The pressure drop correlation requires knowledge of the Figure 6. Comparison of calculated and experimental
holdup in gas-liquid flow to calculate the two-phase density for /f for horizontal flow
the two-phase friction factor. This requires use of two correla-
tions to obtain the two-phase pressure drop. To determine the
reliability of this two-correlation technique, the two-phase Table IV' shows the range of test conditions for the data used
friction factor for all systems with a liquid viscosity less than in obtaining the vertical flow correlation. The data for ver-
30 centipoises was calculated by the Flughmark holdup method tical flow do not cover as wide a range of liquid viscosity (30-cp.
(79) even though experimental holdup data were available. maximum) or pipe diameter as the data for horizontal flow but
Experimental holdup data were used for the few systems with appear to be valid at 2000 p.s.i.a. for a channel section. Equiv-
liquid at a greater than 30 cp. viscosity because the single line alent diameter was used for the channel calculations. Table
holdup correlation is not applicable. It was later found that V gives the average absolute deviation of the calculated two-
the Hughmark holdup correlation could be extended to the phase pressure drop from the experimental values for this
high viscosity systems by additional high viscosity lines to the method. The table also shows the average absolute deviation
K vs. Z plot (Figure 1, 79). The coordinates of these lines are for pressure drop by the Hughmark-Pressburg method.
given in Table III. The Hughmark holdup method was used to calculate all
two-phase densities for vertical upward flow.
Correlation for Vertical Upward Flow
Vertical Downward Flow
Figures 2appear to represent all of the available
and 3
vertical upward flow data and, therefore, represent the pro- The Bryant data (9) are for a 20-foot pipe section. Dukler
posed correlation. (The coordinate points for these correla- data (13) for a 14-foot section show the same general trend as
tions can be obtained from the author.) the Bryant data but show a value of / about 60% of

Table IV. Experimental Test Conditions for Data Included in Correlation for Vertical Upward Flow
Series Reference Pipe I.D., Av. Pressure, Liquid Flow Rate, Gas Flow Rate,
No. No. Inches System P.S.I.A. Lb./Hr. Lb./Hr.
40 (79) 1 .049 Water-air 16-25 108-3.240 0.006-180
41 (79) 1 .049 NaCO.i soln.-air 16-24 108-3,240 0.06-175
42 (79) 1 .049 Varsol-air 16-23 108-3.240 0.063-185
43 (79) 1 .049 Oil blend 1-air 16-25 108-3,240 0.063-180
44 (79) 1 .049 Oil blend 2-air 19-25 540-3,240 0.3-180
45 (79) 1 .049 Triclene-air 18-24 810-3,240 0.75-135
46 (16) 1 .025 Water-air 36 90-9,450 4.3-90
47 (8) 1.25 Water-air 18-110 2,400 7.2-440
48 (77) 0.63-2.5 Water-air 36 350-10.500 3-190
49 (74) 2.0 Water-air 16-22 525-8,600 12-320
50 (74) 0.52 Water-air 14.7-17 6-460 1-16
51 (23) 0.87 Water-steam 15-19 1.200-2,500 5-55
52 (32) 2.34 Water-steam 295-1180 10.000-18,500 355-860
53 (33) »
Water-steam 1100 1,200-1,340 17.5-165
54 (33) Water-steam 2000 270-460 10-210
55 (33) “
Water-steam 2000 1.750-3.380 33-260
56 (34) 0.52 Kerosine—air 15-22 i.25-420 0.16-19
57 (6) 1.25 Water-air 16 14-1.280 158-704
58 (9) 6
Water-air 16 17-128 160-305
59 (2) 0.427 Water-air 16 10-1,000 2-47
a
1 X 0.097 inch channel. >J
1-inck OD by 1.5~inch ID annulus.

VOL 2 NO. 4 NOVEMBER 1963 319

 Nomenclature
Table V. Comparison of Experimental Vertical Upward Flow
Data with Correlations A =
pipe cross section, sq. ft.
No. of Cd =
two-phase drag coefficient, lb. f/(sq. ft.)
Av. Absolute Deviation, , % Tests
D =
pipe ED., ft.
for
Hughmark- Hughmark FrrP =
two-phase Froude number
Series No. Method G =
mass velocity, lb. ,„/ (sq. ft.)(sec.)
Pressburg Hughmark
g =
acceleration due to gravity, ft./(sec.)2
40 37.6 9.5 63
41 19.7 6.5 46 gc
=
32.17 (ft.)(lb. ,„),.·'(lb. ,)(sec.)2
42 23.8 10.3 51
=
elevation, ft.
43 22.5 9.5 52 K =
dimensionless flow parameter for holdup (20)
44 12.4 18.5 39 L =
pipe length, ft.
45 17.1 15.7 25 lw =
lost work, lb. f/ (sq. ft.
46 27.3 20 26 P =
pressure, lb. f/(sq. ft.)
47 27.4 20.4 25 APTP =
two-phase pressure drop, lb. f/'(sq. ft.)
48 23.1 17.5 57 R =
volume fraction of phase in place in pipe
49 35.5 22 32
50 9.1 30 15 Rerp =
two-phase Reynolds number
51 14.4 15.5 20 ^¿Í~L
V =
velocity, ^GÍ"C, ft./'sec.
52 24 12 10
53 o
19.8 11 v =
mean specific volume, cu. ft./lb. m

54 a
17.2 14 IF =
mass of flow, lb. m/sec.
rate
55 a
9.6 8 y
=
volume fraction of phase that would exist if there were
56 54 12.4 39 no slip between phases
57 a
11.8 78 Z =
dimensionless correlating factor for holdup (20)
58 a
14.4 20
59 a
10.5 46 p
=
density, lb. m/'(cu. ft.)
Av. 24.8 15.1 Total 677
pTP
=
two-phase density, plRl + pgRg, lb. m/(cu. ft.)
µ
=
viscosity, lb. m/(ft.) (sec.)
1
Outside of range of correlation.
Subscripts
G =
gas or vapor
L =
liquid
that for the Bryant data. Figure 4 cannot be regarded as a
1 =
inlet to pipe section
general correlation because entrance effects and pipe length 2 =
outlet to pipe section
must also be considered. Figure 4 is based entirely on the
experimental holdup data. The Hughmark holdup method
appears to apply only for the flow conditions in which the gas Literature Cited
velocity is significantly greater than the liquid velocity and does
not apply in the reverse (1) Alves, G. E., Chem. Eng. Progr. 50, 449 (1954).
situation which represents a large
(2) Anderson, G. H., Mantzouranís, B. G., Chem. Eng. SÁ. 12,
part of the downward flow data. 109 (1960).
(3) Aziz, K., Govier, G. W., Can. J. Chem. Eng. 40, 51 (1962).
(4) Baker, Ovid, Oil Gas J. 53, 185 (1954).
Summary (5) Bankoff, S. G., J. Heat Transfer 82, 265 (1960).
Two general equations can be derived to represent pressure (6) Bennett, J. A. R., Thornton. J. D., Trans. Inst. Chem. Engrs.
(London) 39, 101 (1961).
drop for gas-liquid cocurrent flow. The equation derived (7) Bertuzzi, A. F., Tek. M. R., Poettman, F. H., J. Petrol. Technol.
from a momentum balance results in a form of drag coefficient 8, 17 (1956).
(8) Brown, R. A. S., Sullivan, G. A., Govier, G. W., Can. J.
which may be either positive or negative in sign. The energy Chem. Eng. 32, 62 (1960).
balance derived equation yields a lost work term for the ir- (9) Bryant, P. A., M.S. thesis, Lousiana State Univ., Baton
reversible energy loss which is always positive. Experimental Rouge, 1961.
(10) Calvert, S., Williams, B., A.I.Ch.E. J., 1, 78 (1955).
data for horizontal, vertical upward, and vertical downward (11) Chenoweth, J. M., Martin, M. W., Petrol. Refiner 34, No. 10,
two-phase flow show that the lost work term is dependent on 151 (1955).
the pipe orientation. (12) Collier, J. G.. Hewitt, G. F.. Trans. Inst. Chem. Engrs. (London)
'

39, 127 (1961).

The lost work for the energy balance derived equation can be
related to the two-phase pressure drop and the density of the
mixture in place in the pipe to provide a correlation method.
Pressure drop for horizontal gas-liquid flow is found to be
dependent upon: liquid volume fraction determined from the
mass flow rates per unit time and the average density of each
phase, two-phase Reynolds number, gas to liquid density ratio,
and pipe diameter and roughness.
An average absolute deviation of about 19% is obtained
between experimental and predicted pressure drop for 1379
data points over the following range of conditions:
to 1400
Pressure, p.s.i.a. Atmospheric
Pipe size, inches 0.484 to 10
Liquid viscosity, cp. 0.09 to 500
The lost work term for vertical upward gas-liquid flow is a
function of only the liquid volume fraction and a Froude
number. An average deviation of 15% is obtained between
the experimental and predicted lost work term for 677 data
points over the following range of conditions:
Liquid, p.s.i.a. Atmospheric to 2000
Pipe size 1 X 0.097 inch channel to 2.5-inch pipe
Liquid viscosity, cp. 0.08 to 30
(13) Dukler, A. E., Ph.D. thesis, Univ. of Delaware, Newark,
1951.
(14) Galegar, W. C., Stovall. W. B., Huntington, R. L., Petrol.
Refiner 33, 208 (1954).
(15) Govier, G. W., Omer, . M., Can. J. Chem. Eng. 40, 93
(1962).
(16) Govier, G. W., Radford, B. A., Dunn, J. B., Ibid., 35, 58
(August 1957).
(17) Govier, G. W., Short, W. L., Ibid.. 36, 195 (October 1958).
(18) Hoogendoorn, C. J., Chem. Eng. Sci. 9, 205 (1959).
(19) Hughmark, G. A., Chem. Eng. Progr. 58, 62 (April 1962).
(20) Hughmark, G. A., Ph.D. dissertation, Louisiana State Univ.,
Baton Rouge, 1959.
(21) Hughmark, G. A., Pressburg, B. S., A.I.Ch.E. J. 7, 677
(1961).
(22) Isbin, H. S., Moen, R. H., Wickey, R. O., Mosher,
D. R.,
Larson, H. C., Chem. Eng. Progr. Symp. Ser. 55, 75 (1959).
(23) Isbin, H. S., Sher, N. C., Eddy, K. C., A.I.Ch.E. J. 3,
136
(1957).
(24) Jenkins, R., M. S. thesis, Univ. of Delaware, Newark, 1947.
(25) Johnson, H. A., Abou-Sabe, A. H., Trans. ASME 74, 977
(1952).
(26) Lamb, D. E., White, J. L., A.I.Ch.E. J. 8, 281 (1962).
(27) Lockhart, R. W., Martinelli, R. C., Chem. Eng. Progr. 45,
39
(1949).
(28) Martinelli, R. C., Nelson, D. B., Trans. ASME 70, 695
(1948).
(29) Moen, R. H., Ph.D. thesis, Univ. of Minnesota, Minneapolis,
1956.

320 l&EC FUNDAMENTALS

(30) Mosher, D. R., M.S. thesis, Univ. of Minnesota. Minneapolis,
1954.
(31) Reid, R. C., Reynolds, A. B., Diglio. A. J., Spiewak, I.,
Klipstein, D. H., A.I.Ch.E. J. 3, 321 (1957).
(32) Schwarz, K., VDI Forschungsheft 445, 1 (1954).
(33) Sher, N. C., Green, S. J., Chem. Eng. Progr. Symp. Ser. 55,
61 (1959).
(34) Stovall, W. B., M.S. thesis, Univ. of Oklahoma, Norman,
1953.
(35) Taylor, T. . M., M.S. thesis, Univ. of California, Berkeley,
'
1942.
(36) Tek, M. R., Chan, W. J., A.I.Ch.E. meeting, San Francisco,
Calif., December 1959.
(37) Thomsen, E. G., M.S. thesis, Univ. of California, Berkeley,
1941.
(38) Vohr, John, A.I.Ch.E. J. 8, 280 (1962).
(39) Wickey, R. O., M.S. thesis, Univ. of Minnesota, Minne-
apolis, 1956.
(40) Wicks, M., M.Ch.E. thesis, Univ. of Houston, Texas, 1958.
(41) Wicks, M., Dukler, A. E., A.I.Ch.E. J. 6, 463 (1960).
Received for review December 14, 1962
Accepted August 30, 1963

COMMUNICATION

MAXIMIZATION OF YIELD IN ISOTHERMAL GAS

PHASE REACTIONS IN THE PRESENCE OF INERTS
General expressions are derived for the distribution of initial mole fractions of reactants required to maxi-
mize the yield in isothermal gas phase reactions carried to equilibrium in the presence of inerts. If the inerts
are present at a fixed concentration, the initial mole fractions should be proportional to the stoichiometric
coefficients. However, if one or more inerts are carried along with a reactant in fixed ratio, the optimum
composition of reactants differs in a predictable way from the classic distribution. Significant gains in equi-
librium yield may be realized by utilization of the modified distribution.

donder and Van Lerberghe (I) and Prigogine and In the initial reactant mixture, the fixed constant ratio of a
Qe
Defay (6) have shown that the maximum equilibrium given inert, i, to that of a given reactant, j, will be denoted by
yield in an isothermal perfect gas reaction can be obtained by '

bringing the reactants together so that their original mole vt° =

ru°yi°
fractions are proportional to the stoichiometric coefficients, as
i <; i <c j (3)
given by the following expression:
1 < 3 < L
We will admit only one such equation for each inert; how-
ever, a reactant may determine the concentration of several
Pings (5) has shown that this result remains valid as long as
the system conforms to ideal solution laws, but requires slight inerts. For example, in the case of air, regarding oxygen as the
modification if one accounts for the interaction terms in the reactant, and nitrogen, argon, etc., as inerts, the following equa-
second virial coefficient. Pings (3) has separately reported the tions would apply:
modification of that initial composition distribution necessary
to maximize profit when given the knowledge of a value distri-
bution of the reactants and products. It is the purpose of this
note to consider two cases of isothermal gas phase reactions
carried out in the presence of inerts, one of which requires no
correction to the previous simple answer and one of which
etc.
requires a correction, which we provide. The normalization of the mole fractions to unity in the initial
Suppose the reactants are combined with inerts at a fixed reactantmixture may be expressed as follows:
concentration. A straightforward maximization of the yield
reveals that the reactants should still be combined in the stoichi-
=1
ometric ratio, the normalized distribution for which is given f1 + /
(5>
y=i X
by the following equation:
L
(! + V)l° = 1 (6)
j = 1

where Y ° is the original fixed mole fraction of the inerts. where Rj° has been defined as
Consider now the more interesting case where a given react-
j
ant may carry with it one or more inerts in fixed ratio. Let
j be an index denoting a reactant or product, i be an index R,° =
rii° 1 < i ^ L (7)
; = i
denoting an inert, and J, L, and M be integers equal to the
total number of inerts, reactants, and products. Products of Thus defined, R° represents the total concentration of inerts
the reaction are assumed absent from the original mixture. carried along by any given reactant.

VOL 2 NO. 4 NOVEMBER 1963 321