Prediction of Pressure Gradients for Multiphase
Flow in Tubing
GEORGE H. FANCHER, JR.*
JUNIOR MEMBER AIME
KERMIT E. BROWN
MEMBER AIME
ABSTRACT
An 8, OOO-ft experimental field well was utilized
to conduct flowing pressure gradient tests under
conditions of continuous, multiphase flow through
2 Jl8-in. OD tubing. The well was equipped with
10 gas-lift valves and 10 Maihak electronic pressure
recorders, as well as instruments to accurately
measure the surface pressure, temperature, volume
of injected gas and fluid production.
These tests were conducted for flow rates ranging
from 75 to 936 BID at various gas-liquid ratios
from 105 to 9,433 scflbbl. An expanding-orifice
gas-lift valve allowed each flow rate to be produced
with a range of controlled gas-liquid ratios. From
these data an accurate pressure traverse has been
constructed for various flow rates and for various
gas-liquid ratios.
A comparison of these tests to Poettmann and
Carpenter's correlation indicates that deviations
occur for certain ranges of flow rates and gas-
liquid ratios. Numerous curves are presented
illustrating the comparison of this correlation
with the field data. Poettmann and Carpenter's
correlation deviates some for low flow rates and,
in particular, for gas-liquid ratios in excess of
3,000 scflbbl. These deviations are believed to
be rna inly due to the friction-factor correlation.
However, Poettmann and Carpenter's correlation
gives excellent agreement in those ranges of
higher density. This was as expected and predicted
by Poettmann. He pointed out that their method
was not intended to be extended to those ranges of
low densities whereby an extreme reversal in
curvature occurs.1
As a result of these experimental tests, correla-
tions using Poettmann and Carpenter's method
were established between the friction factors and
mass flow rates which are applicable for all gas-
liquid ratios and flow rates. Definite changing
Original manuscript received in Society of Petroleum
Engineers office Aug. 13, 1962. Revised manuscript received
Jan. 7, 1963. Paper presented at 37th Annual Fall Meeting
of SPE, Oct. 7-10, 1962, in Los Angeles, Calif.
lReferences given at end of paper.
*Now associated with The California Oil Co. in Casper,
Wyo.
MARCH, 1963
THE U. OF TEXAS
AUSTIN, TEX.
flow patterns do not allow anyone correlation to
be accurate for all ranges of flow.
INTRODUCTION
The ability to analytically predict the pressure
at any point in a flow string is essential in deter-
mining optimum production string dimensions and
in the design of gas-lift installations. This infor-
mation is also invaluable in predicting bottom-hole
pressures in flowing wells.
Although this problem is not new to industry, it
has by no means been solved completely for all
types of flow condit'ions. Versluys,2 Uren, et al,3
Gosline,4 May,S and Moore, et al, 6, 7 were all
early investigators of multiphase flow through
vertical conduits. However, all of these investiga-
tions and proposed methods were very limited as to
their range of application. Likewise, many are
extremely complicated and therefore not very useful
in the field.
Only in the last decade have any significant
methods been proposed which are generally applic-
able. The most widely accepted procedure in
industry at the present time is a semi-empirical
method developed from an energy balance, proposed
by Poettmann and Carpenter 8 in 1952. Their
correlation is based on actual pressure measure-
ments from field wells. Accurate predictions from
this correlation are limited to high flow rates and
low gas-liquid ratios.
Although this method will be discussed in
detail later, it should be pointed out that two
important parameters, namely the gas-liquid ratio
and the viscosity, were omitted in their correlation.
The viscosity was justifiably omitted since their
data was in the highly turbulent flow region for
both phases, and most wells fall in this category.
The gas -liquid ratio was incorporated to some
extent in the gas-density term. In 1954, Gilbert 9
presented numerous pressure gradient curves
obtained from field data for various flow rates and
gas-liquid ratios for the determination of optimum
flow strings. However, no method is presented for
predicting pressure gradients except bJ' comparison
to these curves. Baxendell, et aI, 1 proposed a
method based on P oettmann and Carpenter's procedure
59
for high-rate flowing wells. Similarly, Tek 11 pre-
sented a procedure for calculating pressure gradients
in two-phase flow by employing a two-phase Reynolds
number and friction-factor function in Poettmann's
correlation.
Undoubtedly, the most significant correlation
since Poettmann and Carpenter's is the one proposed
by Ros.12 This is a general; but complicated,
procedure which is based on a pressure balance
between any two points in the flow string. The
relationships between all variables w,~re investigated
by dimensional analysis from which ir:elevant groups
were eliminated. The resulting relationship consists
of a dimensionless pressure gradient expressed
in terms of a friction term and a liquid hold-up
term (depending on slip only) for various flow
patterns. However, this correlation is only for
two-phase flow but may be linearly interpolated on
the basis of percentage water-cut in the case of
oil, water and gas systems (considerable error may
be introduced by this interpolation). Using our
data, Ros reported agreement was excellent in all
ranges of flow rates and gas-liquid ratios (+0.60
bias using his improved method). It should be
emphasized that this method is extremely tedious
and complicated for ordinary field use unless a
large computer is available.
This paper represents the continuous-flow portion
of an experimental gas -lift project covering both
continuous and intermittent flow. The project was
conducted by The U. of Texas and was jointly
sponsored by the Marathon Oil Co. (formerly the
Ohio Oil Co.), Sun Oil Co. and Otis Engineering
Corp. A paper on intermittent flow was repotted by
Brown and Jessen. 13 This was also given in more
detail in a technical re,r0rt to Marathon Oil, Sun Oil
and Otis Engineering,1 as well as in a dissertation
by Brown. 1S
The purpose of these tests was to establish
pressure traverses in 2 3/8-in. OD tubing for vary-
ing liquid flow rates as produced with varying gas-
liquid ratios. The range of liquid flow rates was
between 75 and 936 BID. Each liquid flow rate was
produced with a gas-liquid ratio that varied from
105 to 9,433 scf/bbl, depending upon the total liquid
volume.
It was futther intended to establish the reliability
of known methods for calculating pressure traverses.
Poettmann 8 had mentioned the possible limitations
of his method in the range of high gas-liquid ratios
and very low densities. Also, it has been anticipated
that at high gas-liquid ratios a reversal in curvature
occurred in the pressure traverse. This investiga-
tion afforded an oppottunity to establish the exact
gas-liquid ratio at which reversal occurred.
Because the well was making 95 per cent water,
the effect of viscosity could not be ascertained.
However, it has been pointed out by other authors 8 , 16
that viscous shear is negligible. In addition, be-
cause water is more difficult to lift than oil, the
water-gradient curves can be used with confidence
for the purpose of artificial-lift design.
Another objective was to obtain sufficient pressure-
60
traverse data to allow accurate prediction of the
pressure traverse in 2-in. tubing for any liquid flow
rate, and for any gas-liquid ratio in the range of
these tests.
EXPERIMENT AL TESTING
EQUIPMENT AND PROCEDURE
The details of the experimental set-up and testing
procedure has been previously outlined in a paper
by Brown and J essen. 13 , 14 The test well was capa-
ble of making 1,000 BID of total liquid (95 per cent
salt water). The physical properties of the oil,
water and gas are shown in Table l.
The production rate was controlled by a bottom-
hole choke located at 6,077 ft. Once a particular
flow rate had been set by a predetermined bottom-
hole choke size, this particular liquid flow rate was
maintained over a range of gas volume rates. This
technique made possible the production of each
barrel of liquid at different gas-liquid ratios. These
gas-liquid ratios changed from 9,433 scf/bbl for the
lowest liquid rates, to as low as 250 scf/bbl for
the highest liquid rates. The upper range of gas-
liquid ratios was limited only by the maximum rate
at which gas could be supplied. Fortunately, a
peak gas supply of over 1.4 MMscf/D was available.
In order to be able to lift each liquid rate with
varying volumes of gas, a flexible-sleeve gas-lift
valve was used. It operates essentially as an
expanding-orifice regulator (Fig. 1). The resilient
element acts as an expanding orifice, and will open
to pass the desired gas. This, in turn, allows one
predetermined liquid rate to be produced with vary-
ing gas volumes. For example, the resilient element
could pass both the maximum rate of 1.4 MMscf/D and
the minimum rate of 34 Mscf/D. A valve of this type
was necessary for these tests because a fixed-
orif~ce valve would not allow this flexibility.
FIg. 2 shows all of the down-hole equipment. As
can be seen, some 10 flexible-sleeve gas-lift valves
were installed along with 10 Maihak electronic
pressure transmitters. These pressure transmitters
have been described previously in detail;17 there-
fore, only a brief discussion will be presented.
These instruments measure the frequency of a
stressed wire which is correlated with pressure.
TABLE 1 - PHYSICAL PROPERTIES OF
OIL, WATER AND GAS
API Gravity at 60 F ____________________________ 34.0 0
Farmation Volume Factor
at 3,375 psig ---------------------------------- 1.313
Gas in Solution at 3,375 psig ____________ 620 scf/bbl
Oil Viscosity at 3,375 psig ________________ 0.46 cp
Salt Water-Chlorides __________________________ 60,000 ppm
Estimated Specific Gravity ________________ 1.065
Specific Gravity of Injected Gas
at Standard Cond it ions __________________ 0.57
Specific Gravity of So1ution Gas
at Standard Conditions _________________ 0.65
Bottom-Hole Pressure, March 14, 1961-- 3,375 psig
Original Reservoir Pressure---------------- 3,575 psig
Original Reservoir Temperature ------------ 193 F
Saturation Pressure ____________________________ 3,575 psig
Tubing Size ------------------------------------- 2 3/B-in. OD
Plastic-coated
SOCIETY OF PETROLEUM ENGINEERS JOURNAL
Each transmitter was connected by one conductor to METHODS OF PREDICTING
the surface with the tubing acting as ground to PRESSURE GRADIENTS
complete the circuit. By individually swit.ching from
Many methods have been developed for predicting
one transmitter to the other, a recordlOg of the
pressure gradients in vertical, multiphase flow.
pressure at each transmitter location could be
However, this discussion is limited to Poettmann
obtained. Calibration both before and after each
and Carpenter's method 8 because it seems to be
test indicated a maximum error of 0.62 per cent for
the most reliable and is the generally accepted
anyone instrument. Most of the continuous-flow
method. Also, a correlation which is based on
tests were conducted with gas being injected at a
Poettmann and Carpenter's work has been developed.
valve located at 2,774 ft. The valves above and
It is shown that their method can be extended to
below were closed to prevent the installation from
include other ranges of flow.
moving from one valve to the other as the rate of
gas injection was changed. POETTMANN AND CARPENTER'S METHOD
As can be seen from the surface installation (Fig.
This semi-empirical correlation of Poettmann and
3) an input-gas regulator was used to maintain a
Carpenter is based on field data from 49 wells (34
c;nstant downstream pressure and a uniform rate of
flowing wells and 15 gas-lift wells) covering only a
gas injection. The liquid and gas production was
limited range of flow rates and gas-liquid ratios.
measured to the nearest 1 14 gal by a Rollo Test
The gas-oil ratios or gas-liquid ratios were con-
Separator. Production was also measured at another
stant for each flow rate. Most of the tests on the
calibrated test separator and liquid test meter before
gas-lift wells were in the range of total liquid flow
entering the tanks.
rates between 300 and 800 BID and gas-liquid ratios
The pressure was recorded at all points above
between 100 and 800 scf/bbl.
the point of injection. Gas was injected through
An energy balance between any two points in the
Valve No.9 or 10 at 2,774 ft or 2,182 ft, respectively
flowing string in which the flowing fluid was treated
(Fig. 2). Therefore, pressure recordings were made
as a single homogeneous phase was developed as
using Maihak Pressure Transmitters 7, 8, 9 and 10 the basis for their correlation. The irreversible
located at 477,969, 1,685 and 2,493 ft, respectively,
energy losses were incorporated in a Fanning-type
and at the surface. From these recordings (all above
friction-factor term. A correlation was developed by
the point of gas injection), the pressure traverse
back-calculating the friction term from field data
down the hole was obtained for a controlled gas-
and plotting it against the numerator of the Reynolds
liquid ratio.
number. Viscosity effects were not included in this
correlation due to the high degree of turbulence of
both phases. Viscous shear is negligible if both

r- -
,-----
---+- PROOUCTION

~
:NJECTION GAS----+-~

.",",._. ,JI
,.25 ""
OTIS SLEEVE VALVE#IQ-218Z' 1
r
[
. " , . , PRESSUR' , . . . .",rTfR _

hlAIHAK PRESS TRANS IIIf 9

."HA,.RESS ,.ANS _ . . ,'

-9~
,o-m'

(600P51)
r;.-'MA1HAK PRESS TRANS -#1-249"

QTISSLEEVE VALVE # 9-2174'

(57~ PSI)

OTIS SLEEVE VALVE" 8 - 3269' ,.,.-'MAIHAK PRESS TRANS #6 -3302'

(~!>O?SI)

ons SLEE'V vALvt_7-4072'

(5~PSI)
r+-MA1HAK PRESS TRANS #5- 4290'

OTIS SLEEVE VALVE #6-4785'

(500 P5 1)

ons SLfEYE VALli[

(4~PSI)
'II' -5377'
r+-MAIHAK pRESS T'RA"" ..... -5t.O

r;.- .....tHAI( TfMP. TRANS ..... '-5n7'

CAMeo fiIIM YAHOREl - 5914'
ons SlEEVE VALVE.3 -'924' (45O PSI VI6" PORT )
ons SLEEVE VALVE # Z-'936'
(4SOPSI- 1/2" PORT)
OTIS GAS Uf'T VALVE #!-!948'
(4~ PSt-! POftT) f,.. IIlAtHAI< PRSS TRANS .. 2-!956'
80TTOM HOLE CHOKE - 6077' -
~ ,.,.. MA'HAII: PRESS TRMS .. I -6UO'

2 roo TUBING----- 5f CA$(NG

CtOSED ~fNfO

... ""eKE" - 7840'

PERFQftATEO - 7870 73'-f

FIG. 1 - FLEXIBLE SLEEVE VALVE. FIG. 2 - DOWN -HOLE EQUIPMENT.

MARCH, 1963
61
phases are in a high degree of turbulence. However, was assumed to be linear and the average flowing
for highly viscous fluids, a calculation of the temperature was used. In several tests, the flowing
Reynolds number may show one fluid to be in the surface temperature was obtained from the corre-
viscous flow region. It is known that, if either lation between flowing surface temperature flow
fluid is in the viscous flow region, then viscosity rate and gas-liquid ratio. (Fig. 4) Also, the compres-
cannot be neglected. Poettmann has pointed out sibility factor of the gas was assumed equal to one.
that their correlation is not valid for those regions This should not be neglected for high pressures and
of flow where either fluid is in the viscous flow gases with numerous impurities. In these tests, a
region.! Also, the slippage of the gas by the liquid dry gas (specific gravity = 0.57) and pressures less
was not separated from wall friction in their corre- than 600 psig were encountered. Two comparisons
lation. were made using the program procedure and the
However, this irreversible energy loss is inher- original method, with no assumptions. The maximum
ently incorporated in their friction-factor correlation deviation between the two procedures was 0.80
because it was obtained empirically from field data per cent.
in which slippage undoubtedly occurred.
In programming Poettmann and Carpenter's method PROPOSED CORRELATION
for the 1604 Control Data Corp. computer, it was When Poettmann and Carpenter's correlation was
necessary to make several assumptions and ap- applied to the field data, deviations were noted in
proximations. Equations were written for the so- several ranges of flow rates and gas-liquid ratios
lution gas and formation volume factor as a function (Figs. 24 through 32). Because of these deviations,
of pressure. Two simplifying assumptions were an empirical correlation based on their method was
also necessary. The flowing temperature gradient developed to fit the field data for the particular

fi-
_.r TUB'NG .... ESSURE

GAGE

MANUAL
RECEIVER

CENTRAL CONTROL PANEL

~------------------------
FIG. 3 - SURFACE TESTING EQUIPMENT.

I
.. ~
_.10 --- -- . I----

i- f-- _.- 1-- -

I
_+- _____ 1'0

t=:=: --- r---

0 ~

--r-=::-----
100 . , . :-.:::::: I i

'Xl
eo.,. i-----':: 1 .0
I ,

'20
>00 810

~~.ID
,
t- r----
r-- ;::::lt--
t-- -~ -. 20

o~-.-
--=
--
. '0

100 810

I-- .
~. .

I ,
t-_
tI ~ ~. -
t
1----+ t-. -.---
,
,
>-
t
0 i i I .
I
100 )()()
"00 XlO
."'" '100 1000 2'00 nco zooo Z70D 2900 3100 00 >000 "00 ""'"
.1 00

GAS- LIQUID ... TID

FIG. 4 - CORRELATION FOR SURFACE FLOWING TEMPERATURE.

62 SOCIETY OF PETROLEUM ENGINEERS JOURNAL

ranges of flow rates and gas-liquid ratios involved.
Viscous shear was also found to be negligible
in these tests due to the high degree of turbulence
of both phases. However, the original Poettmann
and Carpenter correlation was extended to cover
the lower density ranges (in particular less than
10 lb / cu ft for 2 - in. tubing), which Poettmann
stated was outside the range of their original data'!
This lower density range was further correlated by
using the producing gas-liquid ratio as an addition-
al parameter.
The irreversible-energy-loss term was evaluated
by back-calculation using field data to determine
the pressure gradient from the following equation.
f = 7.413 x 10 10 (p)(d)5 x 144[(dp/dH)
_ p]!q2 x M2 . . .
where f = Fanning-type friction factor,
.3.---------------,
..
FRICTION FACTOR' 7.4131f 'Nt 0
(Iv.
P = flowing density, lb/ cu ft,
d
jfJ = pressure gradient, psi/ft,
q oil flow rate, B/D,
d diameter of pipe, ft.
The Fanning-type friction factor was plotted
against the numerator of the Reynolds number (Fig.
5). The scattering of points in this correlation sug-
gests that an important parameter (or parameters)
was neglected. By employing the gas-liquid ratio
as an additional parameter, a correlation was de-
veloped between the Fanning friction factor and
the numerator of the Reynolds number for three
ranges of gas-liquid ratios (Figs. 7, 8 and 9). In
comparing this correlation to that of Poettmann and
(1) Carpenter (Fig. 6), only a small deviation is found
for gas-liquid ratios below 1. 5 Mscf/bbl and qM/ d's
between 15 and 50. This is to be expected because
Poettmann's correlation was based on data in these
,. .
ranges. Also, it should be noted that, as the gas-
liquid ratio increases, the wall friction decreases
(Fig. 6). Although the friction factor is decreasing,
(h.-h,) the pressure gradient increases at high gas-liquid
ratios due to a decrease in the flowing density as
shown in the following equation.

...
I
03

. . .. ...... . .

.. .... . .".... 0.1

... .. ..... ~.

.01
. . .. . ...... 000 0 0
I

10 40

1,4737.IO"MQ/O
FIG. 5 - BACK-CALCULATED FRICTION
FACTOR.
,,0..-----,---------------,
10 50
1.4737110"110/0

FIG. 7 - FRICTION-FACTOR CORRELA-

TION.
,I
~.--------------------,

..:
'"u
0
FRICTION FACTOR
CAl.CULATiON FROM
FIE'LD IlIEI'A FOR

..~
!l
1500<GIl < 3000

01
e:

COl

OOIL-~~~~~L_~~~~~~~~~~~ ,OC'~S--~~~~~IO~---~--~-~~J60
~ m 00
0pV'1.47371 10"M0/0 1.4737 liD" MOlD

FIG. 6 -- COMPARISON OF FRICTION-FACTOR FIG. 8 - FRICTION-FACTOR CORRELA-

CORRELATION. TION.

MARCH, 1963 63

dp =_1_
dH 144
Therefore, as the gas-liquid ratio increases and
the liquid rate decreases, pressure gradients calcu-
lated from Eq. 2 are influenced more and more by
the second term of Eq. 2.
Correlations were developed for the producing
gas-liquid ratio (Figs. 6 through 9). Correlations
have been previously developed for in situ gas-
liquid ratios. However, it was found that excellent
agreement was obtained by using the producing
gas-liquid ratio as a parameter.
DISCUSSION 0 F RESULTS
In discussing the results of this investigation,
the reliability and range of application of both cor-
relations, viscosity, the effect of gas-liquid ratio,
heading phenomena and reverse curvature will be
discussed.
POETTMANN AND CARPENTER'S CORRELATION
Pressure traverses predicted by this method were
in close agreement to field data for flow rates of
936 to 420 B/D in the range of gas-liquid ratios
possible to obtain (Figs. 13 through 22). For flow
rates greater than 600 B/D in all ranges of gas-
liquid ratios, the average deviations were less than
5.0 per cent. For flow rates of 552 and 504 B/D
and gas-liquid ratios of 1.7 and 1.9 Mscf/bbl, the
average per cent deviations were - 10.3 and - 11.7,
respectively. Except for these two flow rates, all
~~------------------------,
FRICTION FACTOR
CAI.llAA1'tD . -
FIELD DATA FOR
GIL>sooo
.oaI L -~-'-~~......l10---~-~---:50
3
1.4737,10-'MQlO
FIG. 9 -- FRICTION-FACTOR CORRELA-
TION.
~r-------------------------------,
FRICTION FACTOR
CALCULA)EO
01 FROM FIELO OATA
FOR
0< GIL < 1500
01
003'L------~--~10---~~~~100~--~~-----
1.4737,10-' MO
0,.,,0-
FIG. 10 -- FRICTION-FACTOR CORRELA-
TION INCLUDING THE LIQUID VISCOSITY.
64
deviations were less than 10 per cent in all ranges
(2) of gas-liquid ratios for flow rates greater than 414
B/D. In general, for flow rates less than 414 B/D,
considerable deviation was found, regardless of
the gas -liquid ratio. In particular, for flow rates
less than 312 B/D, average deviations were greater
than - 20 per cent. For gas-liquid ratios greater
than 1 Mscf/bbl and for flow rates less than 192
B/D, the average deviations were greater than - 80
per cent (Figs. 27, 28, 31 and 32).
These results are to be expected from Poettmann
and Carpenter's correlation since, as mentioned
previously, all irreversible energy losses are in-
corporated in a friction-factor term, for the range
of their data, and fail to hold for the low density
ranges. This correlation shows that the original
Poettmann and Carpenter correlation can be used
to give greater accuracy in the low density ranges
by using the gas-liquid ratio as a parameter (Fig. 6).
PROPOSED CORRELATION
Predicted pressure traverses were in excellent
agree me nt for all flow rates and gas-liquid ratios
for liquid flow rates greater than 330 B/D (Figs.
13 through 23). The maximum error was 5.8 per
cent. For flow rates between 192 and 330 B/D,
the maximum average deviation was - 22.2 per cent;
however, in all but four tests, the deviation was
less than 10.8 per cent. In general, for flow rates
below 192 B/D, agreement was good for all ranges
of gas-liquid ratios (Figs. 26 through 32). The av-
erage deviation exceeded 10 per cent in only five
tests in which the maximum deviation was - 26.7
per cent. In this range of flow rates and gas-liquid
QIO-,------------------,
FRICTION 'ACTOR
CALCULATED
'ROM FIELD DoCTA
'OR
1500 < GIL <3000
' .
.01
.OOSL--~-~~~~~I:':-O--~~-~--~-'::!IOO
1.4737,10"'MQ
01'0,.
FIG. 11 -- FRICTION-FACTOR CORRELA-
TION INCLUDING THE LIQUID VISCOSITY.
.10,..------------------,
FRICTION FACTOR
CALCULATED
~If
FROM FIElD DilTA
~ro.
O!
;. .:. GIL >3000
"
0
.
01
.006
10 100
1.4737,10-'MO
01'''
FIG. 12 -- FRICTION-FACTOR CORRELA-
TION INCLUDING THE LIQUID VISCOSITY.
SOCIETY OF PETROLEUM ENGINEERS JOURNAl.
 ratios, Poettmann's correlation resulted in devi-
tests that the viscosity of the liquid mixture was
ations greater than - 100 per cent for most tests negligible (Figs. 10, 11 and 12).
(Figs. 31 and 32). Again, it should be emphasized
Poettmann and Carpenter 8 found that viscous
that Poettmann and Carpenter did not expect their
shear was negligible. This is true if both phases
correlation to hold for all these ranges.
are in highly turbulent flow, as was the case for
VISCOSITY their correlation. However, Ros12 indicated that
the viscosity has a significant effect on pressure-
The effect of viscosity on the friction-factor cor-
traverse predictions for fluids with viscosities
reI ation was investigated. It was determined in these
greater than 6 cstk. This undoubtedly places the
o
liquid in a viscous or partial-viscous flow region.
~
I
0.936 BID

\\ GIL- 324 SCFI BBl

~
\ o - 792 BID
1------------
6 FIELD DATA
~~ G/l-e90 SCF/BBl
"2
.
POETTlIANH 8 CARPENTER
I\, I

6
2
1::
-L
Ii:
~ ~
CALCULATED FROM FRICTION
FACTOR CORRELAnON
-

12
_\ 6. FIELD DATA

POETTMANN 8 CARPENTER
~

_\ 16
"0
.
..... '\ 0 CALCULATED
"<e"
FRICTION
FACTOR OORREl..ATION

\
20
t-
~ o ~ ----- ------

\
24

24 -- -

0
aN 00 600 eoo
PMSSlIIE. PSI

o 200 400 6 0 800

FIG. 13 - COMPARISON OF PRESSURE TRAVERSES. PRESSURE. PSI

o FIG. 16 - COMPARISON OF PRESSURE TRAVERSES.

\ I o

\
o - 936 1110
GIL 525 SCF/BEL \ o 696 BID
l"ELD DATA
\
,
GIL - 643 SCF/BEL

1\ a POETT....N CARPENTER 8

\
I
6 FIELD DATA

,
12 o CALCULATED fROM FRICTION -
"2

\
a POE TTMAHN 8 CARPlNTER
FACTOR CCIIRLATION.

I. t" 2

1\
o CALQJLATED FROM FRAC110H
FACTOf' CORRELATlON.
-

20
\ 6
~ .. as {COUftTES"I' OF SHELL}

\ . 0

~

0
_zoo 400 100 eoo
4

0
PRESSURE,PSI
200 ~ )Q

FIG. 14 - COMP A RISON OF PRESSURE TRAVERSES.

FIG. 17 - COMPARISON OF PRESSURE TRAVERSES.
o

\
o

\

\ Q .799 BID
GIL '400 SCF/BBL
~
~
Q 696 BID

\,
GIL' 1195 SCF/BBL
'" FIELD DlU --

~
t::. FIELD DATA
a POETTMAHN CARPENTER POETTMAHN CARPENTER
12 -

~
o CALCULATED ftROM .... CTI(III 12 o CAL..QI..AT'tD FROM FRtCTlON -

.'."'
"2
\\
0 FACTOR CClfWtELATIOH
FACTOR CORRELATION.
I. 16
t:'

II)
~.
I!I
\ ~
.
Ii:
i!i
\\
20

\~
4

~
o 200 400 500 800 0
200 400 00 BOO
I'IIDSUIIl. " ' ~PSI

FIG. 15 - COMPARISON OF PRESSURE TRAVERSES. FIG. 18 - COMPARISON OF PRESSURE TRAVERSES.

MARCH. 1963
65
 However, it was impossible to establish the effect For any mass flow rate, the friction term de-
of viscosity on the pressure traverse because the creases as the gas-liquid ratio increases. Due to
well was producing 95 per cent salt water. this fact, separate friction-factor correlations were
developed for three ranges of gas-liquid ratios -
GAS-LIQUID RATIO
o to 1.5 Mscf/bbl, 1.5 to 3.0 Mscf/bbl, and greater
The results indicate that the gas-liquid ratio is than 3.0 Mscf/bbl. Using the gas-liquid ratio as an
a very important parameter. In particular, it must be additional parameter, accurate pressure traverses
considered for flow rates less than 300 BID, re- were predicted for all flow rates and gas-liquid ratios.
gardless of the gas-liquid ratio.
o

8
\1\.
, o 600 BID
GIL' 178 SCFI BBL
I I:. FIELD DATA

"
t:. FIELD DATA
o POETTMANN Eli CARPENTER
a POETTN ANN S CARPENTER 12
2 - o CALCULATED FROM FRICTION
Q CALCULATED FROM FlICTtON

. I\.
0 FACTOR CORRELATION
FACTOR CCARELATION.
16
.
>-
16

~,
%
I-
o.
l'l 20~~---~--~~~-------+-----~---~
0

4
"- ~
!
24f------+_

200 400 600 o 200 400 600 800

PRESSuFE, PSI PRESSURE, PSI

FIG. 19 - COMPARISON OF PRESSURE TRAVERSES. FIG. 22 - COMPARISON OF PRESSURE TRAVERSES.

o
o I .~ I

\ ~
0, - 336 BID
GI L - 2 180 SCF/BBL
400
4
I
f\
0-600B/D
GIL' I 0~6 SCFI BBL
I 800 \"1\ A FIELD OITA -

\ ~ \
A FIELD DATA o POETT..... a CARPENT1t
c POETTMANN a CAfl=lENTER o CALCULATED FROM
12 - 1200 FRICTION FACTOR CCJItRE-

16
~

t
. \
o CALC\,l...ATEO ~OM

fACTOR CORRELATION
FRICTION

1600 1\ 1\ LATtON

\
"'\ " 1\
X
I-
o.
20
l!! 2000

240 0
\~
1\1
24
\':
o 200 400 600 o 100 200 300
I
400 500 600 100 800
PRESSURE. PSI PRESSURE. PSI

FIG. 23 - COOPARISON OF PRESSURE TRAVERSES.

FIG. 20 - COMPARISON OF PRESSURE TRAVERSES.

o 0

\ ~

12
-- \ , o - ~04 BID
GIL 790 SCF/BBL

o
I

o
FIELD DATA

POETTMA~ "

CALCULATED fROM
CARPENTA

fRICTIQN-
4

12
\\
\ 1\
o - 312
GIL -2 21~ SCFI BBL
I
BIO

A FIELD DATA

a POETTMANN a CAfFENTR
-

~ \ \
o CALCULATED !=HOM FRICTlON
1> FACTOR OORRElATICIriI
0
RtCTOR COMELATION
16 t: 16 ..... ~ )( ROS (COURTESY OF SHELL) -

l
"\\
-i
0
Ii:
l!! ~
4

FIG. 21 -
PRESSURE.PSI
200 400

COMPARISON OF PRESSURE TRAVERSES.

000 800
24

o
PRESSURE. PSI

FIG. 24 -
200
"
400
'\
COMPARISON OF PRESSURE TRAVERSES.
600

66 SOCIETY OF PETROLEUM ENGINEERS JOURNAL

 HEADING sure traverses recorded. These pressure variations
Severe heading occurred in several tests (Figs. cannot be attributed to end-effects because the fluid
26, 27 and 32). In these figures, all pressures re- level was well above the bottom-hole choke and
corded at 477, 969, 1,685 and 2,493 ft were plotted critical flow existed through the gas-lift valve. Great
and indicate the pressure variations; i.e., the pres- care was taken also to minimize surface restrictions
sure did not stabilize as in the other tests. The which might affect the flow. The casing pressure
dash lines indicate the maximum and minimum pres- remained constant during each test, but the tubing
pressure fluctuated (Fig. 33). No correlation could
oF=~==~====~====~======~==~ be found between heading, flow rate and gas-liquid

4~--~~~~-- ______~________L-______~________~
Q. 192 BID 0~=4~~=====+======~====~======
GIL' 1035 SCF/BBL

l'::. fIELD DATA

a POETTWAHN a CARPENTER
12
0 o CALCULATED FROM FRICTION b. FiELD DATA
FACTOR CORRELATION

......
o POETTMANN a CARPENTER
16 )C ROS (COURTESY OF SHEt..L) 12 f - - - - - + - \ \
o CALCULATED FROM FRICTION

0
~ FACTOR CORRELATION

20 16 ::::_-------I-~------I---~ )0( ROS (eOURTE SY OF SHEL Ll

...x
2< 20 ~
~-r-----
200 400 .00 800
2<f-----f--~~-~ -----+1--
PRESSURE, PSI

FIG. 25 -- COMPARISON OF PRESSURE TRAVERSES. 200 <00 600 800

PRESSURE, PSI

FIG. 28 -- COMPARISON OF PRESSURE TRAVERSES.

0~~==4======+======~====~====~
~--~~~-------+----~-~~------~------~
Q. 144 BID
GIL' 795 SCF/BBL

l'::. FIELD DATA

o POE TTMANN a CARPENTER

6. FJELD DATA
121-----+--\.-*----';-----\:---+---
o CALCULATED FROM FRICTION
o POHlMANN 8 CARPENTER
FACTOR CORRELATION
12
CALCULATED FROM FRlCTOli
" :---~-~~~-~J------,-----y----~ <:>
..."- FACTOR CORRELATIC:W
i
:. 16
...x >< ROS (COURTESY OF SHELL)

20~O~_ _~~-+\-~~-+-\,---~~--.---+------ ~
20
24'------+\-_\_
2<

o 200 <00 .00 800

PRESSURE, PSI 600 800
PRESSUfIE,PSI
FIG. 26 -- COMPARISON OF PRESSURE TRAVERSES.
FIG. 29 -- COMPARISON OF PRESSURE TRAVERSES.

Q. 144 BID
G/L'1541 SCF/BBL
O' 114 BID
GIL' 1000 SCF/BBL

l'::. FIELD DATA

..:, FIELD DATA
D PO E TTMANN a CAR PENT[R

,
12 c: POETTMANN a CARPENTER
0
o CALCULATED FROM FRICTION. 12 f-------\A-
FACTOR CORRELATION o CALCULATED FROM FRICTION

...... )0( ReS (COURTESY OF SHELL)

FACTOR CORRELATION

" ~ -------+_lM_

--+---+--
16

~
\ ,.. ROS (eOURTES Y OF SHELL)

20
0 \ I
\ 201---"'------~-+

\ I i

: I 1
24 \--. t
I
2< --~-t------+--
I

200 a 200 400 600 BOO

PRESSURE, PS I PRESSURE, PSJ

FIG. 27 -- COMPARISON OF PRESSURE TRAVERSES. FIG. 30 -- COMPARISON OF PRESSURE TRAVERSES.

MARCH, 1963 67
ratio. However, heading always occurred at flow
rates below 192 BID.
This heading phenomenon would make the design
of a gas-lift installation somewhat more difficult.
A safe design would be to use the maximum pres-
sure traverse occurring, whereas the use of the av-
erage traver se might result in a stymied condition
in attempting to reach the operating gas-lift valve.
REVERSE CURVATURE
At very high gas-liquid ratios, a reversal in cur-
vature in the pressure traverse (9) was expected.
However, the reversal in curvature appears to be
not only a function of the gas-liquid ratio, but also
of the flow rate and pressure. In general, as the
flow rate decreases and the gas-liquid ratio increases,
a reversal in curvature occurs at a particular gas-
liquid ratio for flow rates less than 400 BID (Figs.
23, 28 and 32). In particular, for flow rates of 336,
312 to 192, 144 and 114 BID, a reversal in curvature
occurred for gas-liquid ratios greater than 2.0, 3.0,
4.0 and 4.5 Mscf/bbl, respectively. Both correla-
tions predict a reversal in curvature for flow rates
below 400 BID, regardless of the gas-liquid ratio.
A reversal in curvature is expected to occur at the
very low pressures and low density ranges.
CONCLUSIONS AND RECOMMENDATIONS
Foettmann and Carpenter's method of correlation
is shown to be applicable to predict pressure gra-
dients outside the range of their original data. This
has been done by including the gas-liquid ratio as
an additional parameter. Their original correlation
has been shown to give excellent agreement in the
medium-to-high density range. Deviations begin to
occur at the point in their correlation where low
densities occur. This was the point where a reversal
in curvature was shown to occur and the point at
which Poettmann indicated that their correlation
may not hold. 1
CONCL USIONS
1. Poettmann and Carpenter's correlation shows
excellent agreement for high flow rates, but results
in large deviations for low flow rates and low density
ranges.
8 ~
6. FIELD DATA
12
16
0
~
----i o POETTMANN
o CALCULATED
fACTOR
a
FROM
CARPENTER
CORRELATION
FRICTION
x ROS (COURTESY Of SHELL 1
~ ~
20
24f---
000
PRESSURE, PSI
FIG. 31 - COMPARISON OF PRESSURE TRAVERSES.
68
2. In general, the proposed correlation is valid
for all flow rates and gas-liquid ratios in the range
of these tests with an accuracy of 10 per cent.
3. The gas-liquid ratio is definitely a significant
parameter in the friction-factor correlation.
4. Severe heading was noted in several ranges of
gas-liquid ratios for flow rates less than 192 BID;
however, it was impossible to determine any cor-
relation between heading, flow rate and gas-liquid
ratio.
5. A reversal in curvature occurs in pressure
traverses for flow rates in the low pressure region
and low density regions.
6. It is believed that these data will allow ad-
ditional correlations to be checked.
RECOMMENDA nONS
This work shou'rd be extended to cover the fol-
lowing factor s.
1. These tests should be extended to cover other
tubing sizes because it might be erroneous to at-
tempt to correlate these results to any other size
of pipe.
2. A complete set of pressure traverses should
be established for small tubing sizes.
3. The effect of viscosity should be established
for all tubing sizes.
4. A complete set of working curves should be
prepared to eliminate tedious calculations. (Fig. 34)
ACKNOWLEDGMENTS
The authors wish to thank the managements of
the Marathon Oil Co. (formerly The Ohio Oil Co.),
Sun Oil Co. and Otis Engineering Corp. for furnish-
ing the experimental well, surface equipment, down-
hole equipment and tubing jobs. In particular, they
extend thanks to Messrs. Wheeler, Grimes, Shearer,
Spellman and Herring of Marathon's Houston office;
to Bill Howard, Doyle Jones, Gilbert Naert, Cris
Murry and the men of the gang and gas plant; to Jim
Evans, petroleum engineer with Marathon; to Carlos
Canalizo, Don Taylor and Wally Robertson of the
Dallas office of Otis Engineering; to Bill Bertman,
Harold Menke and Walter Groth of the Otis Engineer-
ing Bay City (Tex.) area; and to Messrs. Hodges
and Weiss of Sun Oil.
In particular, the authors would like to thank
12
16
--
0
~-
a POETTMANN
o CALCULATED
FACTOR
a
FROM
CARPENTER
CORRELATION
FRICTION
)( RDS (COURTESY OF SHE:lL)
20f--"'--\--*--'.--
+-~t--
24 f------t--+I\c-
200 800
PRESSURE, PSI
FIG. 32 - COMPARISON OF PRESSURE TRAVERSES.
SOCIETY OF PETROLEUM ENGINEERS JOURNAL

4~-~~--~~~------+--~~---t-----,
~5!
6 ----~~~---~~~------------~----------~
..
>'
%
~-------~~--~~~----
I!I
PMSSUft GRADIENT
'0
PRESSURE. PSI
FIG. 33 - EXAMPLE OF HEADING.
Fred Poettmann of Marathon for his original interest
in the project and for his n:any comments and sug-
gestions in preparing this paper. . 257.
It was from his original suggestiOn - that his
correlation may not be valid in the very low density
ranges - that this project was instigated.
Thanks are due to Ralph Schilthuis of Humble
Oil & Refining Co. for his suggestions and interest
in this project.
The authors are grateful to Nick Ros and H. E.
Gray of Shell Oil Co. for their contributions to the
production of this paper.
Appreciation is also extended to Lloyd Johnson
and Osmar Abib for their assistance in programming
the computer.
REFERENCES
1. Poettmann, F. H.: Private Communication.
2. Versluys, J.: "Mathematical Development of the
Theory of Flowing Oil Wells", Trans., AIME (1930)
Vol. 86, 192.
3. Uren, L. C., et al: Oil and Gas Jour. (Oct. 3, 1929)
Vol. 28, 148.
4. Gosline, J. E.: "Experiments on the Vertical Flow
of Gas-Liquid Mixtures in Glass Pipes", Trans., AIME
(1936) Vol. 118, 56.
5. May, C. J.: "Efficiency of Flowing Wells", Trans.,
AIME (1935) Vol. 114, 99.
6. Moore, T. V. and Wilde, H. D.: "Experimental Meas-
urements of Slippage in Flow through Vertical Pipes",
Trans., AIME (1931) Vol. 92, 296.
MARCH, 1963
Q 600 BID ~_ _ _ _ _~
WOR 19/1
'f 11,!:t F
---
- - - G I L -'00 SCf/BBL
_--ll<-~~_+_--=----,--"",-_G:::./:::.L~'2::50::::.,:S::C:..:F/:::.BBL~_ _ _---1
1000 4000
FIG. 34 - WORKING CURVES.
7. Moore, T. V. and Schilthuis, R. J.: "Calculations of
Pressure Drops in Flowing Wells", Trans . AIME
(1933) Vol. 103, 170.
8. Poettmann, F. H. and Carpenter, P. G.: "The Multi-
phase Flow of Gas, Oil and Water through Vertical
Flow Strings with Application to the Design of Gas-
Lift Installations", Drill. and Prod. Prac., API (1952)
9. Gilbert, W. E.: "Flowing and Gas Lift Performance",
Drill. and Prod. Prac .. API (1954) 126.
10. Baxendell, P. B. and Thomas, R.: "The Calculation
of Pressure of Gradients in High-Rate Flowing Wells",
Trans., AIME (1961) Vol. 222, 1023.
11. Tek, M. Rasin: "Multiphase Flow of Water, Oil and
Natural Gas through Vertical Flow Strings", Trans.,
AIME (1961) Vol. 222, 1029.
12. Ros, N. J. C.: "Simultaneous Flow of Gas and Liquid
as Encountered in Well Tubing", Trans., AIME (1961)
Vol. 222, 1037.
13. Brown, K. E. and Jessen, F. W.: "Evaluation of
Valve Port Size, Surface Chokes, and Fluid Fall
Back in Intermittent Gas-Lift Installations", Jour.
Pet. Tech. (March, 1962) 315.
14. Brown, K. E. and Jessen, F. W.: "Experimental Gas
Lift Project-The Vertical Flow of Liquid Slugs by
Intermittent Gas Lift", Tech. Report to the Marathon
Oil Co. (then Ohio Oil Co.), Sun Oil Co. and Otis
Engineering Corp. (May 1, 1962) 18.
15. Brown, K. E.: "The Vertical Flow of Liquid Slugs
by Intermittent Gas Lift", Dissertation, The U. of
Texas (June, 1962).
16. Brown, G. G. and Associates: Unit Operations, John
Wiley & Sons, Inc., N. Y. (1950).
17. Lozano, G. and Harthorn, W. A.: "Field Tests Con-
firm Accuracy of New Bottom-Hole Pressure Gauge",
Jour. Pet. Tech. (Feb., 1949) Vol. XI, No.2, 26.
18. Ros, N. C. J.: Private Communication.
***
69