You are on page 1of 8

SIMULTANEOUS FLOW of LIQUID and GAS

through HORIZONTAL PIPE

A. F. BERTUZZI
M. R. TEK PHILLIPS PETROLEUM CO.
F. H. POHlMANN* BARTLESVILLE, OKLA.
MEMBER AIME

T. P. 4203

ABSTRACT The method has been reduced to a niques has created the need for re-
simplified graphical procedure, suit- view and re-design of many surface
A method is presented for predict- able for field use.
ing pressure drop for two-phase fluid gathering lines for properly handling
A two-phase f factor is defined the increased production. Optimum
flow in horizontal pipes.
and correlated with parameters in- pipe size for the situations described
A set of 267 experimental meas- volving the flowing gas-liquid mass above has become an important fac-
urements randomly sampled from ap- ratio, a Reynolds number for the gas tor.
proximately 1,000 measurements phase, and a Reynolds number for
from various literature sources was the liquid phase. The choice of The problem which is of interest
used. Pressure gradients calculated parameters allows the correlation to here involves the ability to predict
by the procedure develooed and reduce to the usual f factor plot for the relationship between pressure
compared with the expe~imental the limiting conditions of all gas drop, fluid properties, fluid rates,
values showed a bias of +0.82 per or all liquid. pipe diameter, and pipe length.
cent and a standard deviation of 20.8 Although the literature" r., 10, "', l', '"
per cent. INTRODUCTION contains various articles dealing with
The advantages of this method specific cases of the problem, Lock-
over other available methods o.f pre- The mechanics and characteristics hart and Martinelli" proposed the
dicting two-phase flow pressure drop of two-phase flow systems have been most general solution. Their method
are (a) its comparative simplicity of of interest throughout the industry has been later modified by Baker'.
application, (b) its relative indepen- for some time. In numerous engi- Alves' demonstrated that different
dence of flow patterns, (c) its accur- neering installations such as pipe flow patterns are possible for a flow
acy, which on the basis of a statis- lines, chemical reactors, and heat ex- mechanism defined by Martinelli
tical evaluation predicts pressure changers, two-phase flow conditions such as gas turbulent, liquid turbu-
drop closer than other available are of every day occurrence. In oil lent. Baker' explained deviations ex-
methods, and, (d) its ability to satis- production operations it has been de- perienced with Martinelli's correla-
factorily correlate laboratory data sirable, in some cases, to consider tions on the basis that different flow
from various sources while the cor- transporting gas and oil together in patterns could exist for the same flow
relations from these sources do not a common pipe from oil field to mechanism. Using Martinelli's meth-
appear to agree with one another. process plant. The trend toward cen- od of correlation, Baker correlated
trally located stock tank batteries in data for each flow pattern for the
:::1". H. Poettmann is now with Ohio Oil Co.,
oil fields has resulted in longer gath- turbulent-turbulent flow mechanism.
Denver, Colo. ering pipelines in which more than Bergelin and Gazley" 7 presented
Original manuscript received in Petroleum one fluid phase is flowing. The in-
Branch office on May 6, 1955. Revised manu- a correlation for predicting gas phase
script received on Dec. I, 1955. Paper pre- crease in the producing capacity of
sented at Petroleum Branch Fall Meeting in
New Orleans, Oct. 2-5. oil wells due to new production tech- lReferences given at end of paper.
SPE 544-G
VOI~. 207,1956 17
pressure drop for stratified and annu- well established theory of single- flowing mixture based on the cross
lar flow patterns. phase flow behavior with experimen- section of the pipe; V is the specific
Schneider" presented a method of tal data collected on two-phase flow volume of the flowing mixture. The
correlating two-phase pressure drop systems. problem then revolves around the
which appears limited to the flow The development which follows is ability to correlate f in terms of the
conditions covered by his data. based on a total energy balance for physical properties of the fluids, pipe
The purpose of this study was to the flowing two-phase mixture under diameter and roughness, and fluid
apply the concept of a two-phase f steady state conditions. Energy dis- rates.
factor, which may be called the en- sipated to heat due to irreversibilities In two-phase flow the flowing mix-
ergy dissipation function, in an at- in flow is expressed in terms of a ture density or specific volume can
tempt to correlate horizontal flow two-phase f factor which is analo- be different from the in situ density
data. This concept has been success- gous to the resistance factor used for or specific volume because of slip-
fully employed in correlating multi- single-phase flow. page or liquid hangup. Slippage oc-
phase flow in vertical strings" and curs when the gas flows at a greater
DEVELOPMENT OF CORRELATION
has been used for predicting pressure linear velocity than the liquid. The
drop for the flow of a flashing mix- It is well known that a certain flowing mixture density was used
ture of steam and water in pipes. ' 5, 10 amount of energy is necessary to rather than an in situ density because
Another objective was to eliminate overcome the resistance which any the energy entering and leaving the
the necessity of establishing a flow flowing fluid encounters in its path. tubing by virtue of the flowing fluid
pattern for predicting pressure drop, The basic energy relationship of any is a function of the pressure-volume
since the same fundamental variables fluid flow process stems from the law properties of the fluid entering and
that determine flow pattern should of conservation of energy which leaving the tubing and not of the
also determine pressure drop. Fin- merely states that the energy of fluid fluid in place.
ally, and probably foremost, upon entering the conduit minus the energy For single phase flow in smooth
attaining some degree of success in dissipated in the conduit through ir- pipe, f can be expressed in terms of
the first two objectives, it was desired reversible effects is equal to the en- the Reynolds number which involves
to present the results of the correla- ergy of the fluid leaving the conduit. inertia and viscous forces. In two-
tion in such a manner to be useful to The above is usually formulated by phase flow the interfacial and gravity
the practicing field engineer. the following equation based on a forces undoubtedly also enter the pic-
unit mass of fluid flowing. ture. After an unsuccessful attempt
ANALYSIS AND SOLUTION P, at correlating f in terms of dimen-
OF THE PROBELM
The general problem of predicting
./v
P,
dP + 6[~~]
sionless numbers involving all four
forces, another approach was made
which proved adequate for most en-
the pressure drop in a multiphase + 6h + W, + W, = 0 (1) gineering calculations. Since Lock-
flow system is a very complicated hart and Martinelli" attained a fair
one. One of the main difficulties is For a horizontal pipe, the differ-
ence in elevation, 6h, is zero. When degree of success in correlating data
due to the fact that numerous flow for a wide range of conditions neg-
patterns of widely different geometry no work is done on or by the fluid,
W, is zero. In most cases the kinetic lecting interfacial and gravity forces,
and mechanics may exist. These flow our development proceeded on this
patterns, usually referred to as bub- energy term can be neglected. The
equation then reduces to: basis. For the data used, the two-
ble, plug, stratified, wave, slug, annu- phase f factor was found to correlate
lar, spray, froth, etc., not only impose P,
unstable and intricate geometry to
the system but also critically affect
the relative magnitudes of several
f VdP + W, = 0 . (2)
best in terms of Reynolds numbers of
the liquid and gas phases, each based
on the pipe diameter, and the gas-
P, liquid mass ratio. The functional re-
force systems active to varying ex- where W, is the energy dissipated lationship hence took the following
tents. to heat in the system due to irreversi- form:
The classical approach of attempt-
ing to solve the Navier-Stokes equa-
bilities.
For single-phase flow the energy
f = cJ> [(DW")
Ft<
a (DWL)
/-LL
b]. (5)
tions becomes hopelessly devoid of dissipation term, W f, is expressed as where a and b were functions of the
any promise, not only due to ana- part of a dimensionless number called gas-liquid mass ratio, K.
lytical difficulties in setting up the the resistance factor, f.
boundary conditions but also because K 1
a = + K ; b - -eO.-
of the non-linear nature of the equa- = 2g, W,~ (3) 1K
f 4v'L .
tions. The necessity to include, in a (6)
general formulation of the problem, The resistance factor is correlated and
the interfacial and gravitational forces for single phase flow with Reynolds
along with viscous, inertia, and pres- number, DW I /-L, and relative rough-
sure forces further complicates the ness, EID."
theoretical approach. f = cJ> (DWI/-L) (EID) . (4) (7)
This equation can be obtained by As K approaches CJJ (all gas)
Appreciating the above complica- a approaches 1
tions and since the direct mathemati- dimensional analysis.
b approaches 0
cal approach seemed to fail to give A two-phase f factor is defined
(R:) (R~) approaches Rg
any hope of practical success, it was using Eq. 3 and is used with Eq. 2
thought best to resort to experimen- for making the energy balance on a As K approaches 0 (all liquid)
tal facts and use available data. Thus horizontal two-phase flow system. a approaches 0
a semi-empirical approach was For two-phase flow, in Eq. 2 and 3, b approaches 1
adopted in combining ideas from the v is the superficial velocity of the (R:) (R~) approaches RL

18 PETQ.OLEUM TRANSACTIONS, AIME


therefore in the limiting conditions,
all gas or all liquid, the correlation
reduces to that for single phase flow,
neglecting relative roughness.
, +-_. , , j ;! .:.,
Two-phase f factors have been cor- i'
i'
related against (R;) (R~) for a range
': ::!:
of gas-liquid mass ratios. i -,-
- --=---:----X: ....
The selection of the exponent func-
tions a and b was arbitrary except
that they meet the requirements
stated above at the limiting condi-
tions. The factor of 0.1 used in ex- "''
.,-
ponent b insures significance of the 1'-
liquid phase in computing the func-
tion (R:) (R~) for gas-liquid mass
ratios as high as 50 to 1.
DISCUSSION OF CORRELATION "

The correlation is presented as


FIG. 2- ENERGY DISSIPATION FUNCTION FOR GAs-LIQUID MASS RATIOS
Figs. 1, 2, 3, and 4. Each figure is
0.2 TO 004.
for a range of gas-liquid mass ratios,
as indicated. The nature of the data
did not permit correlating for specific pendent of flow pattern for a given literature. '" A set of 267 experimental
gas-liquid mass ratios. Fig. 5 is a gas-liquid mass ratio. On the basis of measurements randomly sampled
composite of the four previous fig- the information available, the lower from approximately 1,000 measure-
ures and shows the single phase curve (1) on each figure seems to ments from various literature sources
curves for smooth pipes. The nature apply to stratified, wave, and semi-
was used. The range of the variables
of the correlating curves, their rela- annular flow; whereas the upper
curve applies to the other types of involved, therefore the limitations of
tion with one another and with the the correlation, is summarized in
single phase curves results, in part, flow. As is indicated on each figure
the transition from the lower curve Table 1. Although available field
from the selection of the exponent data' were not used directly in the
functions, a and b. (1) to the upper curve (2) occurs
for liquid Reynolds numbers between correlation, they were used as a guide
This method of correlating shows in extrapolation, indicated by dashed
500 to 10,000. More data are needed
that it is possible to have more than lines on Figs. 1, 2, 3, 4 and 5.
in the lower range of Reynolds num-
one value of f for particular values A comparison of field measure-
ber function with Rg less than 2,000
of (R:) (R~) and gas-liquid mass ments' with predicted values is given
to see if there is a similar transition
ratio over a range of (R;) (R~) val- region. Representation of the corre- in Table 2. Only 10 of Baker's 27
ues. The reason is attributed to the lation by straight lines where possible sets of data were used in the com-
fact that it is possible to have several was for convenience in further sim- parison presented as Table 2. The
flow patterns for the same value of plification to working charts. Point other data were beyond the gas-
the function over this range. It ap- scatter in the correlation can be at- liquid mass ratio limit of the correla-
pears, however, that more than one tributed partly to the fact that a par- tion which is equivalent to a gas-
flow pattern can be correlated by a ticular curve applies to a range of liquid volume ratio (cu ft/bbl) of
single curve in this range, and at high gas-liquid mass ratios. approximately 5,000. We have found,
values of the Reynolds number func- The correlation was developed however, that for gas-liquid volume
tion the data can be correlated inde- from laboratory data available in the ratios greater than 50,000 cu ft/bbl
the gas flow equations" can be
used. This method assumes the liquid
to be uniformly distributed through-
out the gas phase, and an adjustment
is made to the gas gravity to account
for the liquid. The calculations are
then made assuming only gas flow.
Table 3 gives a comparison of Bak-
er's field measured pressure drops for
gas-liquid volume ratios greater than
50,000 with those calculated by the
method just described.
Data by Gazley 12 for stratified flow
were not used. Gazley measured gas
phase pressure drops and showed
that there could be considerable dif-
ference between gas phase pressure
drop and liquid phase pressure drop
because of change in head of the liq-

FIG. 1- ENERGY DISSIPATION FUNCTION FOR GAs-LIQUID MASS RATIOS "'The tabulated data were omitted to re-
duce the length of paper, but can be obtained
UP TO 0.2 from the authors.

VOL. 207, 1956 19


I"'.", '.""1' .. '.
-;;b
I I' , 'I' dP 2fv'
dL + = O. (10)
Since W" = pv, and the gradient is
negative,
dP _ 2f W~
(11)
dL - g,. D p
---j---~
One need go no further than Eq.
+-----' .-+-

11 to make the pressure gradient cal-


culation. However, to effect addi-
tional simplification for graphical
representation, f has been expressed
as part of an equation of the form

f = C

where d is the slope of a straight lin~


[CR:) (Rn r (12)

and C is the intercept at unity for


the function (R:) (R~) on log-log
paper. The slope, d, is constant and
equal to -0.12 for curves (2) on
Figs. 1, 2, 3, and 4. The intercept, C,
can be expressed in terms of the gas-
Fro. 3 - ENERGY DISSIPATION FUNCTION FOR GAs-LIQUID MASS RATIOS liquid mass ratio, K.
FROM 0.4 TO 0.6. By substituting Eq. 12 into Eq. 11
and collecting terms,
uid phase. Since the gas phase pres-
sure drop does not necessarily repre-
cent, and the standard deviation, CT,
from the algebraic average was 27.5
~~ = [Z :}t12(Hb)]
(K

sent the two phase pressure drop for per cent. The algebraic average devia- W'" ". . 12(a+b)] [ftc0.12" ft L0'''b1
[-D-1+o. 12(U-"-:w ~-p--
. 12b
(13)
this type of flow and range of condi- tion represents the bias of the corre-
tions, the data were not used. Some lation. For a normal distribution, Expressing Wo in terms of QM,
of his data, where the gas phase pres- 6S.26 per cent of the values will be mass flow per day,
sure drop is approximately equal to included within plus or minus one dP
the two phase pressure drop, appear standard deviation; 95.46 per cent dL
to correlate by the methods described within plus or minus two standard "~'''(''+'') (K + 1) o.,,(a+b)]
here and fall within the range of deviations: 99.73 per cent within [
(l.4737xlO') (Z) KO"o
plus or minus three standard devia-
Schneider's" data which consisted pri-
marily of wave flow. Jenkin's" data tions. [(O Lf),.O.",,,+b.'] r ""0--r --
. .. iV1
- 6-'-:'o~'.!(a+b)
O.1ObJ
-
ftg
... -
P'L

were not available at the time of this


APPLICATION whereZ = 0.6217 C
investigation. Van Wingen's'" data
were not complete for the type of DEVELOPMENT OF WORKING CHARTS K = 0.000219 (Sg/o) (Gg/GrJ
(15)
analysis made here. There are much To facilitate prediction of pressure
all exponents are functions of the
more data in the literature that were drop, Eq. 2 and 3 have been com-
gas-liquid weight ratio and are ex-
not used since they covered approxi- bined with the correlation and re-
pressed on Fig. S as
mately the same range as the data duced to graphical form in Figs. S
r = 0.12a
used in the correlation. and SA. The development which fol-
s = 0.12 b
An evaluation of the correlation is lows is the basis for the graphical
t = 0.12(a+b)
presented as Fig. 6. A comparison of presentation. Putting Eq. 2 in differ-
u = 2-0.12(a+b)
experimental pressure drops with ential form
v = 5-0.12(a+b) .. (16)
pressure drops predicted by Marti-
nelli's correlation is presented as Fig. -dP + dW f = 0 (S) Eq. 14 has been put in graphical
P form on Figs. S and SA. Eq. 14
7. For the correlation developed in equates pressure gradient with the
and combining with Eq. 3 in differ-
this report the algebraic average de- product of three terms, the first of
ential form one obtains
viation was +0.S2 per cent, and the which involves gas-liquid mass ratio,
standard deviation, CT, from the alge- -dP + ---.-
4fv'dL
= O. (9) the second involves fluid rate and
braic average was 20.S per cent. For p 2gc D
pipe diameter, and the third involves
Martinelli's correlation17 the algebraic By expressing in terms of a pressure the physical properties of the flow-
average deviation was + 14.4 per gradient, ing fluids. It should be n8ted that
the third term is the only one which
TABLE I - RANGE OF VARIABLES
is a function of pressure, thus sim-
Diameter in. ,(, To 2 plifying the calculations.
Temperature OF 50 To 100
Pressure psia 15 To 55
Liquid phase Reynolds numbers,
liquid Rate Ib/sq fl/sec 2 To 1,000 R L , should be greater than 10,000
Gas Rate Ib/sq ft/sec 0.5 To 15
liquid Viscosity Ib/ft/sec 0.00025 To 0.190 when using Figs. S and SA. For most
Gas Viscosity Ib/ft/sec 6.5 X 10. 4 To 13 X 10"
liquid Reynolds No. Rr. 10 To 135,000
practical field cases RL will be greater
Gas Reynolds No. Rg
0
1,000 To 100.000 than 10,000 for gas-liquid volume
liquid Gravity GL [60 F/60 F) 0.813 To 1.00
Gas Gravity G g [Air = 1.0) 0.71 To 1.00 ratios less than 5,000. This limita-

20 PETROLEUM TR.\'iSACTIO'lS, AIME


function locates point 3 on
Fig. 8.
Sg/o (Gg/G L ) = 1,000
(0.65/0.80) = 812.5
LOCKHAJll ...URTlNHLI (11) Step 4. This step involves location of
o $01NIIOH. "1111''' NUNTINGTClM In!

HLOW ........ UUOF_FOI!IR"'IllIt l


point 4 which establishes the
Vilot.UnO# .....IIEG." ... TE.fHAtI_
pressure gradient for a given
=:=;i:=t==~j
:=:: -- "'IOVE
T ......un
....... lION 0# CUlt
F.OM ".. .! FOIl 1"'1aJVE
(I) TO (tl
III IXCI,III$ FOIl V.... UU OF .~ It .... CMNG pressure. Point 4 is deter-
- ::~(: C::V~":- ;~L~llr: IA":O,
FOIIItLGltUnll m ... Ml. . . . CUIIV(
mined from the physical
(31 ~OUlO'" U.o.
property function, (fl/)
(flL S) / p. Since this function
varies with pressure it is best
to calculate the function for
several pressures and tabu-
late corresponding pressure
gradients in preparation for
Steps 5 and 6. As a first ap-
proximation fluid viscosities
at the outlet conditions were
used. Usually if the pressure
drop is less than 100 psi,
change in fluid viscosities
with pressure can be neg-
FIG. 4 - ENERGY DISSIPATION FUNCTION FOR GAs-LIQUID MASS RATIOS lected. Also if the pressure
FROM 0.6 TO 1.0. level is low gas solubility can
be neglected and Eq. 20 can
Table 1. Field data are needed to G L = 0.80 be used for calculating V m'
further verify the applicability of cor- flg = 0.014 cp = 9.4 X 10-0 K
relation for pressures greater than 1b/ft/sec @ 75 and a=K+l=
100 psi and pipe diameters larger 50 psia
than 2 in. flL = 0.6 cp = 4.0 X 10-4 812.5 (0.000219) = 0 151
Evaluation of M and p can be 1b/ft/sec @ 75 and (812.5) (0.000219) + 1 .
made from the following relation- 50 psia 1 1
ships: b = eO. 1 Ie = eO. 017S = 0.980
M = (5.61) (62.4) (Gd + SOLUTION
(0.0764) (G g ) (Sg/o) Step 1. Determine the product, QM r = 0.12a = 0.018 : s = 0.12b
(17) Q = 400 BID =-- 0.118
M M = (350.1) G L + (0.0764) fL:-
= (9.4 x 10-')0018 = 0.812
p= (18) (G g ) (Sg/o) (16) flL S = (4.0 x 10- 4) 0.118 = 0.397
Vln M = 329.7 Ib/bbl fl/ fLL' = 0.322
Vm = 5.61 F + QM = 400 (329.7)
M 329.7
P a Tnyg Z 131,880 lb/day p=-=--
(Sg/o - S,). (19)
This value establishes the Vm Vm
P Ta
If the gas solubility is nearly zero starting point on Fig. 8. Vm = 5.61 +
Step 2. The pipe diameter, 2 in.
~)
such as in the case of air and water
mixtures then: (1.995 ID) locates point 2 ( Po J,av g
) (Sg/o) ( (19)
on Fig. 8.
Vm = 5.61 + V'll = 5.61 +
Step 3. The gas-liquid weight ratio
P a T.v. Z S I (20)
P Ta go

EXAMPLE PROBLEM
A well is being produced by gas-
lift. The well is producing 400 BOPD
with a total GOR of 1,000 cu ft/bbl.
The separator which is 1,000 ft from
the wellhead is operated at a pres-
sure of 50 psia and the fluids are
transported to the separator through
a 2-in. (1.995 ID) horizontal line.
In gas-lift operations it is desirable
to maintain a low wellhead pressure
for efficient lifting; therefore, it
would be of value to know the well-
head pressure for the 2-in. pipe.
Additional pertinent data are as
follows: -~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
T.,.g = 75F <?tRtl - ItETJrIOUlS MlI/I8ERFtllCTlCII

G g = 0.65 FIG. 5 - COMPOSITE ENERGY DISSIPATION CHART.

VOL. 207, 1956 21


TABLE 2 - COMPARISON OF FielD MEASURED PRESSURES' WITH CALCULATED PRESSURES USING P a = base pressure 14.65 psia.
OUR CORRELATION FOR GASlIQUID RATIOS UP TO 5,000 CU FT/BBL
Pipe Dia. Outlet Press. Outlet Press. Q = barrels of liquid flowing per
Gas Rate Liquid Rate Pipe Length 1.0. line Temp. Inlet Press. Meas. Calc.
M.S.C.F.D. Bbl/Day Feet Inches OF Psig Psig. Psig.
day.
25,552 5,484 11 ,317 7.750 80 1,007 975
962
--954-
957
Rg =
gas Reynolds number, di-
12,050 4,167 11 ,317 7.750 f,Q 972
11 ,886 6,592 11 ,317 7.7'0 78 977 960 965 mensionless.
6,474 4,970 5,534 7.750 66 964 9.#)0 959
4,348 5,420 11 ,317 7.750 112 940 930 933 RL = liquid Reynolds number, di-
25,552 5,484 41,333 10.136 80 975 946 925 mensionless.
12,050 4,167 31,11 ; 10.136 69 962 948 951
11 ,886 6,592 41,333 10.136 78 960 936 940 Sg! 0 = cu ft of gas!bbl of liquid
6,474 4,970 41,33J 10.136 66 952 936 944
4,348 5,420 41,333 10.136 82 930 912 923 at std. condo (14.65 psi
60F) .
(14.65) (535) (1,000) Z A plot of P vs L cumulative gives S, = solubility of gas in liquid at
a value of 97 psia. Since the pres- pressure, P.
(520) P
sure drop from separator to well Ta = base temperature 60F.
V m = 5.61 + 15,073
zp head is only 47 psi, taking into ac- T''''g = Arithmetic average of the
count variation in fluid viscosities inlet and outlet pipe temper-
The NGSMA manual can be used and gas solubility with pressure will atures.
for establishing Z, the compressibil- give no significant difference in pres- v = velocity of flowing fluid,
ity factor for the gas. If it is neces- sure drop. ft! sec.
sary to use Eq. 19 for V m, the gen- V = specific volume of flowing
eralized correlations of Standing" can NOMENCLATURE fluids, cu ft/lb.
be used for estimating gas solubility V m = cu ft of gas and liquid at
and formation volume factor. Esti- D = inside diameter of pipe, ft. pressure, P,/bbl of liquid
mates of liquid viscosity may be f = dimensionless energy dissi- based on the ratio of the
made by the method of Katz and pation function, resistance
factor for single-phase flow. fluids flowing into and out
Bicher' or by the Beal Method.' Es-
F = ratio of barrels of liquid at of the pipe.
timates of gas viscosity may be made
by the method of Kobayashi" as pressure, P, to barrel of W = mass throughput, lb! sec! sq
seen in Table 4. liquid at standard condi- ft.
Step 5. This step involves a correc- tions, formation volume fac- Wa = Wg + WI.'
tion to the estimated gradi- tor. W f = energy losses resulting from
ents determined by Step 4. ge = constant = 32.174. irreversibilities of the fluid
Fig. 8A is used for this cor- G g = gas gravity (air = 1.0). in flow.
rection. G L = liquid specific gravity. W g = gas mass through, based on
K = gas-liquid mass ratio, di- pipe diameter.
QM = 131 880(~) mensionless. WL = liquid mass throughput,
D '\1.995 L = length of pipe, ft.
= 793,260 based on pipe diameter.
M = total mass of liquid and gas
Sg/o (Gg/Gr.) = 812.5 W, = work done by or on the
associated with one barrel
Correction = 1.21 of liquid flowing into and fluid while in flow.
Step. 6. Table 5 shows the incre- out of the pipe, lb. Z = compressibility factor of gas.
mental integration to give P = absolute pressure in consis- e = 2.7218.
the pressure at the well head. tent units. p = density at pressure P and

~ ILl

__ 1_ ~-f
T
~, I

--J
-1
tf
1.' 10.D
fI~flll .. eoTAl - .OIJl;DS .U SOUA!I[ tOOT .U fOOl
! I 1.
FIG. 7 - EXPERIMENTAL VS CALCULATED GRADIENTS
FIG. 6 - EXPERIMENTAL VS CALCULATED PRESSURE
GRADIENTS - THIS CORRELATION (FOR GAs-LIQUID - MARTINELLI CORRELATION (FOR GAs-LIQUID MASS
MASS RATIOS TO 1.0). RATIOS TO 1.0).

22 PETROLEUM TRA;>ISACTIO,,"S, AIME


TABLE 3 - COMPARISON OF FIELD MEASURED PRESSURES' WITH CALCULATED PRESSURES USING temperature Toy", lb/cu ft.
GAS FLOW EQUATIONS" FOR GASlIQUID RATIOS GREATER THAN 50,000 CU FT/BBL
Pipe Die. Outlet Press. Outlet Press. p.g = gas viscosity, lb/ftlsec.
Gas Rate
M.S.C.F.D.
Liquid Rote
Bbl/Doy
Pipe Length '-D. line Temp. Inlet Press.
OF
Meas. Calc.
Psig.
jJ.L = liquid viscosity, lb/ ftl sec.
. _Feet
--- Inches ~ Psig.
26,970 514 11,317 7.750 75 983 964 970
26,970 514 41,333 10.136 75 964 945 951
11 ,767 192 11,427 7.750 65 712 703 70B ACKNOWLEDGMENT
6,484 136 11,427 7.750 65 705.5 703 704.2
4,441 76 11,427 7.750 69 1,075.0 1,075.0 1,075.1
6,484 136 10,617 7.750 65 703 701.5 701.8 The authors wish to thank the
6,668 107 10,617 7.750 70 1,067 1,065.5 1,066.2
6,798 102 10,617 7.750 69 1,075 1,074 1,074.3
management of Phillips Petroleum
11,950 236 11,313 7.750 70 1,064 1,062 1,061.5 Co. for permission to publish this
6,484 136 22,044 7.750 66 705.5 701.5 703.1
11,950 236 41,317 10.136 70 1,063 1,055.5 1,060.7 paper.

REFERENCES
TABLE 4
OM 131,880 131,880 131,880 131,880 131,880 131,880 1. Allen, W. F., Jr.: Trans. ASME.
D 1.995 1.995 1.995 1.995 1.995 1.995
$g/o (Gg/GI.i 812.5 812.5 812.5 812.5 812.5 812.5 (1951), 73, 257.
j.l~/' j.lLt>>f. 0.322 0.322 0.322 0.322 0.322 0.322
M 329.7 329.7 329.7 329.7 329.7 329.7
2. Alves, G. E.: "Co-Current
Z 0.99 0.99 0.99 0.985 0.985 0.98 Liquid-Gas Flow in a Pipeline
P psia 50 60 70 80 90 100
Vm 304.05 254.31 218.79 191.19 170.57 153.32 Contactor," Paper presented at
P 1.084 1.296 1.507 1.724 1.933 2.150 AIChE Meeting, San Francisco
J.lgr J.LL~ 0.297 0.248 0.214 0.187 0.167 0.150
dP/dL psi/lOO It 5.7 4.6 4.0 3.6 3.25 3.0 (Sept. 14, 1953).
*Fluid viscosity at outlet conditions were used. 3. Baker, 0.: Oil and Gas Jour.
(July 26,1954),185.
4. Beal, C.: Trans. AIME (1946),
TABLE 5 165, 94.
P 50 60 70 80 90 100 5. Benjamin, M. W., and Miller, J.
dp/dl psi/l00 It 5.7 4.6 4.0 3.6 3.25 3.0 G.: Trans. ASME (1942), 64,
Correction 1.21 1.21 1.21 1.21 1.21 1.21
dp/dl psi/laO It 6.9 5.6 4.8 4.4 3.9 3.6 657.
6. Bergelin, O. P., and Gazley.

15

FIG. 8-
*PS.I./IOOFEET

CHART FOR ESTIMATING PRESSURE DROP FOR HORIZONTAL MULTIPHASE FLUID FLOW IN PIPES.

VOL. 207, 1956 23


TABLE 6 8. Boelter, L. M., and Kepner, R.
p
50 60 70 80 90 100
!:.p 10 10 10 10 10 H.: Ind. Engl'. Chern. (1939),
dp/dl 6.9 5.6 4.8 4.4 3.9 3.6
clo/dl (ave) 6.25 5.2 4.6 4.15 3.75
31, 426.
~"cremental -- 160 192 217 241 267 9. Bicher, L. B., and Katz, D. L.:
~ cumulative 0 160 352 569 810 1077
Ind. Engl'. Chern. (1943), 35,
754.
10. Davidson, W. F., et al: Trans.
ASME (1943),65,553.
c., Jr.: "Co-Current Gas-Liquid Carpenter, F. G., and Gazley, 11. Dittus, F. W., and Hilderbrand,
Flow," (I) "Flow in Horizon- c., J r.: "Co-Current Gas-Liquid A.: Trans. ASME (1942), 64,
tal Tubes," Heat Transfer and Flow," (II) "Flow in Vertical 185.
Fluid Mechanics Institute, Berk- Tubes," Heat Transfer and Fluid 12. Gazley, Carl Jr.: "Interfacial
eley, Calif. (1949). (Available Mechanics Institute, Berkeley, Shear and Stability in Two-
from ASME.) Calif. (1949). (Available from Phase Flow," PhD Thesis, Uni-
7. Bergelin, O. P., Kegel, P. K., ASME.) versity of Delaware (1948).
13. Holmes: Perry's Chemical En-
gineers Handbook, Third Edi-
tion, McGraw-Hill Book Co.,
Inc., New York. Figs. 17 and
-- 18, 686.
-- 14. Jenkins, Rodman: "Two-Phase
Two-Component Flow of Water
--- - 0
0
~
_0

~
_ 0 0

~
- ~ -r - "1 --

~
f--o
~ ~ -"
2
f--
g and Air," MChE Thesis, Uni-
versity of Delaware (1947).
15. Johnson, H. A., and Abou-
r-- r-- 'I Sabe, A. H.: Trans. ASME
- -- -- - --- r= =r i+- F
(1952), 74, 977.
-
1 FF
1- rr
16. Kobayashi, R., Carr, N. L., and
-- f 1/- -1 Burrows, D. B.: Trans. AIME,
(1954) 201,264.
17. Lockhart, R. W., and Martinelli,
0'..... +
1--1- IF F-I-
fl + R. c.: Chern. Engl'. Prog.
8 II -I 7- (1949),45, No.1, 39.

::;\
~o
10'
VI
~
1if I 711 -- 18. McAdams, W. H., et al: Trans.
ASME (1941),63,545.
19. McAdams, W. H., et al.: Trans.
, ASME (1942), 64, 193.
I
20. Moody, L. F.: Trans. ASME
f--- -IIi f- (1944), 66, 671.
10' 1rryrr 21. Poettmann, F. H., and Carpen-
ter, P. G.: Drill. and Prod.
Prac. (1952) 257.
r- t-- f- H -
+-- 22. Schneider, F. N., et al: Dallas
i-- ---I- --\-I-- -1- -c- - l-f f ~

+-- Meeting of AIME (Oct., 1953).

10'
7f ~f ---- -- - 23. Standing, M. B.: Volumetric
and Phase Behavior ot Oil Field
Hydrocarbon Systems, Reinhold
Publishing Corp., New York
(1952).
-- t---I--
24. Van Wingen, N.: World Oil
III (Oct. 1949) 156.
10'
0.4 0.. 0.6 to 1.2 1.4 25. Clinedinst, W. D.: Oil and Gas
CORRECTION "0 PRES5URE CRADlE NT OF FIGURE
Jour. (April 7, 1945) 79.
FIG. 8A - MULTIPLICATION CORRECTION TO FIG. 8.
***

24 PETROLEUM TRANSACTIONS, AIME

You might also like