PETROLEUM

SCIENCE
&
ENGINEERING
ELSEVIER

Journal of Petroleum

Science and Engineering 14 (1996) 137- I58

Steady-state gas flow in pipes

Liang-biao

Ouyang

**I,

Khalid Aziz

Petroleum Engineering Department, Stanford &ice&y,

Received

1 April 1995; accepted

Stanford, CA 94305-2220, USA

15 May 1995

Abstract
New general Bow equations of simple form are developed to account for the pressure drops due to friction. elevation and
kinetic energy change. Simplified forms are also presented for new flow equations for gas flow in pipelines or wells where
the kinetic energy term can be neglected. The new general flow equations and their simplified forms are compared with the
widely-used AGA equations and tested with field data. Results show that the new equations make excellent predictions of
flow rates or pressure drops, and that they are applicable over a much broader range of gas types and gas flow rates than the
AGA equation and old simplified flow equations. In addition, different empirical explicit correlations for the Fanning
friction factor are compared. It is found that different correlations give quite different values of the friction factor. For
smooth pipes, modified 1/9th power law, Blasius, Drew et al., and Panhandle equations are recommended for different
Reynolds number ranges. For rough pipes, Serghides (I) and (II), Zigrang-Sylvester
(I) and (II), Chen. and Haaland
equations can be employed with confidence. Other friction factor correlations reported in the literature should be avoided
because they can result in large errors.

1. Introduction
Many

factors

must

be considered

in designing

modern pipeline system or a gas well. These include

the nature and the volume of gas to be transmitted,
the length and the size of the pipeline, the depth of
the well, the operating temperature and pressure, the
type of terrain to be crossed, the capacity of producing wells, the type of gas produced, process plant
operating conditions,
plant location, the elevation
change over the route, and so on. Among these, the

* Corresponding
author.
E-mail: ouyang@pangea.stanford.edu.
Fax: (415) 725-2099.
E-mail: aziz@pangea.stanford.edu.
Fax: (415) 725-2099. _

Phone: (415) 725-2730.

Phone:

(415)

723-9116.

0920.4105/96/$15.00
0 1996 Elsevier Science B.V. All rights reserved
SSDI 092Ob4105(95)00042-9

pressure drop along the pipe and the quantity of

natural gas that flows through the pipe are the most
important of the first items of information required
for design.
Flow of natural gas in wells and pipelines is
dependent upon Reynolds number, friction factor,
pipe roughness, pipe diameter, pipe length, temperature, pressure, pressure drop and gas properties. Accurate predictions are required for optimum design.
During more than 100 years of gas production and
transportation
in pipes, dozens of flow equations
(e.g., AGA, Weymouth, Panhandle, Modified Panhandle, IGT, Cullender and Smith, Sukker and Cornell equations, etc.) have been proposed to relate the
gas volume transmitted through pipes to the various
factors that influence this rate. Many of the equations
and methods
presented
in the literature
have been

138

L.-b. Ouyang, K. Aziz/Journal of Petroleum Science and En@wering 14 (19961 137-158

oversimplified
by assumptions
and approximations
as well as by inclusion of inaccurate friction factor
correlations. These simplified equations can lead to
nontrivial errors, which are not justified in this day
and age of easy access to computers. Furthermore,
they are constantly
introduced
in textbooks and
widely used in industry without regard to their ranges
of applicability (Katz et al., 1959; Institute of Gas
Technology,
1965; Ikoku,
1980; Beggs,
1984).
Towler and Pope (1994) tried to clarify the application ranges of some of these equations. Unfortunately, it seems that this publication
resulted in
additional confusion, partly due to some typographical errors that appeared in their paper.
New general flow equations which take the pressure drops due to potential energy change (elevation),
kinetic energy change, and frictional pressure drop
into account, are proposed in the present paper to
describe the gas flow in pipes. They can be reduced
to the AGA equation and other simplified flow equations by applying certain simplifications and assumptions. The calculation of friction factor, which can be
a source of great confusion
among engineers, is
discussed in detail and applicable ranges of Reynolds
number and pipe roughness are identified for different correlations. Comparison and analysis are also
performed for various methods used by industry.

2. Pressure

or flow rate calculation

method

2.1. General flow equation in differential form

The general flow equation can be derived from a
total momentum balance around an element of fluid
flowing through a differential length of pipe under
the following assumptions:
1. Newtonian fluid,
2. Single phase flow,
3. Steady-state flow,
4. Isothermal flow,
5. No heat transfer to and from the gas to the
surroundings, and
6. No mechanical
work done on or by the fluid
during its passage through the pipe (No shaft
work or work of compression).
The resulting general flow equation is of the
form:

udu
dp
-+Rdz+-+2-dL=0
P
Yg,
g,

fu?
(1)

Q?,

where the correction factor (Y is introduced to compensate for the variation of gas velocity over the pipe
cross-section.
Its value depends upon the velocity
profile and ranges from 0.75 for laminar flow to
about 1.O for fully developed turbulent flow (White.
1986) . Aziz (1962) suggested that a value of 0.9 be
used for practical gas flow problems. The correction
factor is ignored in some textbooks and technical
papers (such as in Economides et al., 1994; Tian and
Adewumi, 1992; Young, 1967).
The assumption of isothermal flow is used in the
solution of Eq. 1. However, in some cases it is
important to perform simultaneous temperature and
pressure profile calculations.
This can be done by
considering the energy balance equation in addition
to Eq. 1 (Gregory et al., 1979).
2.2. Method

qf solution

Eq. 1 can be applied to determine the natural gas

flow rate in pipes by means of three different methods, i.e., the multi-step method (the segmented design method), the numerical integration method, and
the analytical
integration
method
(single-step
method).
2.2.1. Multistep or segmented design method
The differential form of Eq. 1 can be used directly
for numerical calculation
along the pipe without
making any further assumptions.
This is accomplished by dividing the pipe into small increments
and evaluating the gas properties according to the
pressure and the temperature at each segment.
-Ap=pgAz+KAu+
gc

yg,

2fPU2
-AL
Dg,

(2)

The procedure for calculating the pressure drop

over a pipe using the multistep method can be
outlined as follows (Beggs, 1984):

For 1/7th power-law

velocity profile, cy = 0.9804. For
1/8th power-law profile, LY= 0.9843. The (Y value is taken as
0.98 for all the calculations in this paper.

L.-O. Ouyang, K. Aziz/ Journal

Step
Step
Step
Step

Step
Step
Step

Step
Step

of Petroleum Science and Engineering 14 (1996) 137-158

I Starting with the known pressure, p,, at

location L, = 0, select a length segment, AL.
Ap, for the
2 Estimate a pressure increment,
pipe segment AL.
3 Compute the average pressure and the average temperature in the segment.
4 Determine density, viscosity and other properties of the gas at average pressure and
temperature of the segment.
5 Compute the pressure gradient in the segment
at the average conditions.
6 Compute the pressure increment in the se lected segment.
7Compare
the estimated and the calculated
values of the pressure increment obtained in
step 2 and step 6. If they are not sufficiently
close, choose a new pressure increment and
return to step 3. Repeat steps 3 through 7
until the error between the estimated and
calculated values are within the tolerance.
8 Set L = L, + XAL and p =p, + 2Ap.
9 If ZAL is less than the total pipe length,
return to step 2.

Multistep methods require that calculations

be
performed over very small segments of the pipe and
that iterations be employed to obtain the change in
pressure and temperature over each segment. The
segment lengths can be chosen such that their sum is
exactly equal to the total pipe length and so interpolation is not necessary in the last step. Every segment
length can be adjusted to ensure that the pressure
drop over the segment is below a specified maximum. The integration is easy to program and can be
made as precise as desired by reducing the segment
length and the tolerance for the iterative process. The
procedure can be applied from upstream to downstream or in the reverse direction.
2.2.2. Numerical integration method
Eq. 1 can be expressed in integral form provided
the kinetic energy term can be neglected (see Appendix Al:

(3)

139

This is the form used by Cullender and Smith (1956).

It cannot be integrated analytically
because of the
interdependence
of the variables p, Z, p, p and T.
There are many numerical integration methods available to calculate pressure changes in wells and
pipelines using this form of the flow equation (e.g.,
Aziz, 1967). It is also possible to simplify the integrand by making different assumptions. Rzasa and
Katz (1945) integrated Eq. 3 by assuming that the
gas deviation factor is a constant evaluated at average pressure and average temperature. Sukker and
Cornell (19551, and Poettmann (1951) integrated the
equation based on the assumption that the gas deviation factor varies only with pressure. More widelyused approaches are those of Cullender and Smith
(19561, and Crawford and Fancher (19591, where the
gas deviation factor is taken as a function of both
pressure and temperature.
It should be noted that all methods mentioned
above are fairly accurate for natural gas flow calculations in pipes.
2.2.3. Single-step or analytical integration method
It is possible to integrate Eq. 1 analytically if one
assumes that the temperature,
the compressibility
factor, the friction factor as well as the gas viscosity
can be assumed constant at some average values, but
updated iteratively. This method is called the singlestep method. A number of single-step methods have
been published in the literature based on different
assumptions (refer to the series of articles by Aziz,
1962-1963). One of the assumptions, which is employed by most of these methods, is that the change
in kinetic energy can be neglected. This assumption
may lead to significant errors for some special gas
flows in pipes (Young, 1967) though it is valid for
most engineering applications.
New general flow equations and their simplified
forms are presented later in this paper. At this stage,
we will discuss different forms of the friction facto1
correlation that are used in the gas industry.
2.3. Friction ,factor
The Fanning friction factor, one of the most important variables in flow equations, is used to determine the pressure loss due to friction in pipes. The
variation of the friction factor with Reynolds number

L.-b. Ouyang, K. Aziz/ Journul of Petroleum Science and Engineering

140

14 (1996) 137-158

0.0040.004 0.0035 -

0.0002
0.0001
QOOOOS

0.003 0.00250.002250.002 -t+f++t+I03

3456B

104
Reynolds Number, Re

Fig. 1. Regimes of laminar and turbulent tlows (after Govier and Ark,

and pipe roughness for circular pipes can be divided

into different regimes (Govier and Aziz, 1972; see
Fig. 1): laminar flow, smooth wall turbulent flow,
partially rough wall turbulent flow and fully rough
wall turbulent flow 4. Transitions from smooth wall
turbulent flow to partially rough wall turbulent flow
and from partially rough wall turbulent flow to fully
rough wall turbulent flow are determined by the
roughness Reynolds number Re, which is defined as
follows:

$Re i

Re,=

For laminar flow, the friction factor can be shown

to be a simple function of Reynolds number:
16

f=Re
The friction factor is only a function of Reynolds
number for smooth wall turbulent flow, and a function of relative pipe roughness for fully rough wall

Partially rough wall turbulent flow and fully rough wall

turbulent flow are also named as partially developed turbulent
flow and fully developed turbulent flow.

1972).

turbulent flow, whereas it depends upon both the

Reynolds number and relative pipe roughness in
partially rough wall turbulent flow.
In most practical situations, gas flow in pipes is
turbulent. Different formulae have been reported to
determine the Fanning friction factor for turbulent
pipe flow. These methods can be classified as smooth
pipe correlations and rough pipe correlations.

3. Friction factor calculation

3. I. Smooth pipes
From an examination
of the measurements
available, Blasius (19 11) proposed:
f= 0.079&-

2s

then

(6)

and showed that this equation correlates pressure

loss data up to Reynolds numbers of approximately
10. Since then, several modifications
of Eq. 6,
including
1/7th power law (Ward-Smith,
19801,
1/8th power law (Ward-Smith, 1980), 1/9th power
law (Ward-Smith,
19801, modified 1/9th power law
(Knudsen
and Katz, 1958), l/lOth
power law
(Ward-Smith, 1980), Panhandle equation (also known

L.-h. Ouynng, K. Aziz/Journal

Table 1
Constants

141

of Petroleum Science and En,qineering 14 (19961 1.37-158

r and t in Eq. 7

Equation

Blasius

1/7th

l/&h

1/9th

Mod. 1/9th

I/lOth

Panhandle

Mod. Panhandle

IGT

,t

0.079

02.5

0.0763
0.25

0.0563
0.222

0.0437
0.20

0.046
0.20

0.0347s
0.182

0.02118
0.1461

0.003678
0.03922

0.04675
0.20

as Panhandle A equation) (Institute of Gas Technology, 1965; Beggs, I984), modified Panhandle equation (also known as Panhandle B equation) (Institute
of Gas Technology,
1965; Beggs, 1984), IGT (Beggs, 1984), have been used. All of them are known
as the Blasius form or power-law relationships, and
can be expressed as follows:

,f =

rRe_

(7)

where the values of r and t are listed in Table 1, and

the Reynolds number can be calculated from:

(8)
for

any

consistent

units.

The

b =

constant

11.5.88/rrR,
and for the field and SI units in the
Nomenclature
the constant becomes 0.7104 and
4.4364 X lo-?, respectively.
Though the Blasius forms or the power-law relationships have the merit of simplicity, they also have
certain disadvantages,
one of which being that the
relations can only be applied over a limited range of
Reynolds numbers. Extrapolations beyond this range
cannot be made with confidence. Another criticism is
that the Blasius forms or the power-law relationships
do not adequately describe the conditions that arise
in real situations with flow either close to the wall or
near the pipe axis (Ward-Smith,
1980).
Two additional equations of this type are also
reported in the literature:

b
,.;;

(4
-7
_,_,_....
,__--~.y.z?.

. ..- ..._..._._

--,~+_-.-----

o_,

::

,,_,.;;::

..__ _-_. .. .. .

7.

,.(~.

1
s
s

.L.,

;Z.

...

,I

,/

,I
..;(

,f

--_

..
. . .f
i::::_..,,

,l

..,
.\

-20 ~

_/-

,
,.,,.

,i

;,<

b:

,,i

.-;~;;_---7--__

....,

.. ,

.:;..
..I..,

+.,
....
.:..

.....

..
...

. ,,

,/.

. .,

./

..,
.....

,i

.....,,

,/
.,.
.I.
.i
,/

Fig. 2. (a) Comparison of different

different correlations for calculating

...,

I. ,.

c-w
EL&us(l9rr,
.---.---,hfcditied lmhPoW*,Law

----,
_.-.-..
. . . . .

Panhmdis
hfomsd Panhandls
,GT

-----

mewera,

formulae for calculating the smooth pipe friction factor with Prantdls
the rough pipe friction factor (relative pipe roughness = O.OOOS).

equation.

(b) Comparison

of

Drew
et al.

Blasius
(1911)
1/7th
law

power

I /Sth power
law

(a) Average deviation from the Prantdls Universal Fricfion Law for smooth pipes
2000-4000
1.190
2.366
5.703
13.060
4000- 1.Oe4
2.716
1.253
2.309
7.832
1.Oe4- 1.Oe5
1.795
1.288
2.672
3.044
l.Oe5-l.Oe6
0.422
9.387
12.484
7.071
I .Oe6- 1.Oe7
2.443
24.952
27.517
17.944
1.Oe7- I .Oe8
12.446
41.380
43.384
31.659
Average in the whole Rr range
3.502
13.438
15.678
13.435
(b) Maximum deviation from the Prandfls Uniurcwl Friction Law for smooth pipes
Drew
Blasius
1/Xth power
Re range
1/7th power
law
et al.
(1911)
law
2000-4000
1.988
- 4.543
~ 7.806
~ IS.838
4000-I .Oe4
3.002
2.135
- 3.925
- 10.576
1.Oe4- 1.Oe5
3.013
2.236
-4.661
- 5.706
1.Oe5- I .Oe6
0.876
- 14.235
- 17.166
- 10.011
1.Oe6- 1.Oe7
4.574
~ 30.680
- 33.049
- 22.422
l.Oe7-l.OeX
17.404
-46.X27
- 48.644
~ 36.529
Maximum in the whole Re range 17.404
- 46.827
-4X.644
- 36.529

Re Range

Table 2

Mod. 1/9th
power law
- 18.719
- 12.310
- 5.652
2.372
- 9.637
- 22.228
- 22.228

1/9th power
law
- 22.783
~ 16.695
- 10.369
~ 5.341
- 14.155
-26.116
-26.116

Mod. 1/9th
power law

15.377
X.648
1.919
1.179
6.128
17.776
8.505

power

19.608
13.216
4.703
3.961
10.821
21.887
12.366

I /9th
law
1 / 10th power

1, 10th power
law
- 29.656
-23.157
- 15.947
- 5.087
- 8.839
~ 18.220
- 29.656

26.292
19.243
X.137
3.541
6.667
14.806
13.114

law

43.626
36.965
28.632
12.467
~ 3.393
~ 3.352
- 43.626

Panhandle

40.168
32.475
17.559
6.045
1.309
2.191
16.624

Panhandle

Modified
Panhandle
- 77.941
- 73.396
-66.X33
~ 47.970
- 26.553
20.204
~ 77.94 I

75.669
70.003
54.662
34.056
11.538
12.719
43.108

Modified
Panhandle

~ 17.393
- 10.881
-4.114
4.04 I
-8.164
~ 20.960
- 20.960

IGT

13.997
7.159
2.850
2.742
4.850
16.436
8.006

IGT

L.-b. Ouyang, K. Aziz/ Journal of Petroleum Science and Engineering

Weymouth
f=

equation

(Weymouth,

and McAdams equation for Re = 2000-4000

and
4.0 X lo-4.0 X 106, and the IGT equation for 8.0
X lo-2.0
X 106. Although the IGT equation is very
close to the modified 1/9th power law, the latter is a
better approximation to the Prantdls universal friction law (Fig. 2a). Other equations should not be
used for any Reynolds number because of the high
error involved.
Another correlation for calculating
the friction
factor in smooth pipes is used in the literature (Brill
and Beggs, 1991; Aziz, 1994):

1913):

O.O08D-

(9)

where the units for pipe diameter

Drew et al. (1932):
,f= 0.00140 + O.l25Re-

are inches.

143

14 (1996) 137-IS8

(10)

The results of many measurements of the pressure

drop in smooth wall turbulent pipe flow were analyzed by Prandtl (1935). He found that the plot of
,f- against log{Ref0-5} is linear and can be represented by:

Re
,fO. = 41og{ Ref,}

- 0.4

(11)

Eq. 11 is known as Prantdls universal friction

law for smooth pipes and has been verified as the
best correlation of the experimental data over the full
Reynolds number range of turbulent flow (WardSmith, 1980).
Inasmuch as the friction factor is given implicity,
Eq. 11 is sometimes inconvenient
for engineering
applications. To test the errors resulting from the use
of explicit equations, i.e., the Blasius form or the
power-law relationships, we have calculated the Fanning friction factors based on different correlations
and compared the values obtained to those calculated
from the Prantdls universal friction law. Table 2 and
Fig. 2a show the results. It is found that the 1/7th
power law. modified
1/9th power law, Blasius,
Panhandle,
IGT, and Drew, Koo and McAdams
equations can all be used in place of Eq. 11 for
different ranges of Reynolds numbers, but none of
them is suitable for the full range (Table 3). In other
words, the Blasius equation should be used for Re =
2000-lo,
the modified 1/9th power law for Re =
IO- lo, the Pandhandle equation for Re = 106-lo8
(This is the range of Reynolds number most commonly encountered in gas pipelines), the Drew, Koo

Table 3
Application
Equation

range of Reynolds
Drew
et al.

Reynolds

2000-4000

number
application
range

4.Oe + 4-4.0
C+6

number for smooth pipes (accuracy

Blasius
(I 9 1 I)

and

1/7th

= 410g

4.5223 log Re - 3.8215

Re

= 410g

3.5922 log( 0.1429Re)

-0.4

(12)

This equation happens to be a very good approximation to the Prantdls universal friction law, with the
average error for the Reynolds number range of 2000
to 10 of only 0.157%. But it seems impossible to
derive the simplified flow equation based on this
correlation, so it will not be discussed in detail in
this paper.
3.2. Rough pipes
The Colebrook-White
equation (Colebrook and
White, 1937; Colebrook, 1939) is widely used to
predict the Fanning friction factor for rough pipes:
j-o.5 = _410g

1.255
~
f,Re

E
3.70

=3.477-410g[;+$Z]
In some

textbooks

(13)
and technical

papers

(e.g.,

= f 2%)

i/9th

power
law

1/8th
power
law

power
law

3000

7000

None

None

-I.Oe + 5

-40 000

Mod.
power

I /9th

iaw

20 000
-1.Oe + 6

1 /lOth
power
law

Panhandle

None

2.0e + 6
-l.Oe + 8

Modified
Panhandle

None

IGT

8.Oe + 5 -2.Oe + 6

144

L.-b. Ouyang, K. Aziz/Journal

qf Petroleum Science and Engineering 14 (19961 137-15X

Govier and Aziz, 1972; Beggs, 1984; Baker et al.,

1988; El-Oun, 1990; Beggs, 1991), another form of
the Colebrook-White
equation is employed:
j-~ = 3.48 - 410g [D2E + $&)

Jain (1976):
,f,

( 14)

Eq. 14 is slightly different from Eq. 13 and it is

named here as the modified
Colebrook-White
equation .
It is also worth noting that Eq. 11 is very close to
the Colebrook-White
equation applied to smooth
pipes (the average relative error for the Reynolds
number range of 2000 to 10 is 0.068%). The only
difference is that the constant
-0.4
changes to
-0.395 (if Eq. 13 is used) or - 0.403 (if Eq. 14 is
used).
The Colebrook-White
equation and its modified
form are implicit in friction factor, and must be
solved by iteration. Though a solution can be obtained easily and quickly by numerical methods to
any desired degree of precision, explicit formulae are
more welcome and more likely to be used by engineers in industry.
Several explicit forms for computing friction factors for rough pipes have been proposed. Some of
them are derived based on approximations applied to
the Colebrook-White
equation.
Moody (1947):

= 2.28 - 41og (DL + S)

Churchill

(18)

(1977):

1
l/l?

(19)

where:

A=

{2.4571n](-7-~y+0.27tl:)6

Chen (1979):
5.0452
- 7

0.2698;

I.1098

5.8506

+---&,0.898
Haaland

Ii(20)

(1981)

l/3
2 x 104;

+ g

fos=

-3.61og

(15)

[ Re
6.9 + (&)

(21)

1
Zigrang and Sylvester

(1982)(I):

Wood (1966):
+ O.,,,(i)

+ 22( $ jRF
E
x log p+p
3.70

(16)
where:

13.0

Zigrang and Sylvester

f05

= -4.Olog

(17)

E
x log -+3.70

(1982)(U):
5.02
- ---&og

&
i

Eck (1973):

(22)

Re

.
13.0
Re

111

(23)

of Petroleum Science and Engineering 14 (1996) 137-1.58

L.-h. Ouyang, K. Aziz/Journul

Serghides

(1984)(I):

(24)
where:
A=

-2log[-&+;)

Serghides

(1984MI):

(A - 4.781)2
.f= 0.25 4.781 - B _ 2A + 4,781

-
(25)
I

Most of the explicit equations listed above have

their own validity range of Reynolds numbers and
relative pipe roughnesses
(Zigrang and Sylvester,
1982; Serghides, 1984). Table 4 lists the overall
average relative errors and the maximum errors of
the Fanning friction factor values obtained from
different explicit equations compared with those from
the Colebrook-White
equation. The Reynolds number ranges considered are 2000-4000,
4000-104,
lo-lo,
lo-lo,
lo-lo,
and lo-lo*,
and the
relative pipe roughness values considered are 10P6,
lo-,
lo-, 5.0x 10-j, 0.001, 0.005, 0.01, 0.05
and 0.1
Calculations
show that absolute relative deviations for different ranges of Reynolds number and
pipe roughness are within 5% except for the Moody,
Wood, Churchill and Eck equations. The errors in
the friction factor calculated from the Moody equa-

145

tion are very high for large Reynolds numbers or

large relative pipe roughness, whereas the errors by
the Wood equation are high for very large pipe
roughnesses or for very small pipe roughnesses. For
the whole range of Reynolds numbers and relative
pipe roughnesses, the average errors are very small
for the Serghides(1) and (II) (1984), Zigrang-Sylvester (I) and (II) (19821, and Chen (1979) equations
(Table 4 and Fig. 2b). The average and the maximum errors for these methods are all less than 1%.
This is not surprising because both the Serghides
equations and the Zigrang-Sylvester
equations are
actually obtained from the Colebrook-White
equation by two or three iterations of direct-substitution
based on different accelerated-convergence
techniques. Noting the fact that the Colebrook-White
equation is also an empirical approximation of experimental data, it can be concluded that the equations
mentioned above [Serghides (I) and (II) (1984), Zigrang-Sylvester
(I) and (II) (1982), Chen (1979)].
and the Haaland (1981) equation provide an adequate correlation for the Fanning friction factor for
turbulent flow in rough pipes.

4. Flow rate equation

New general flow equations (Eq. A8 for inclined
pipes and Eq. A9 for horizontal pipes) which account
for the pressure drop not only due to friction and
elevation, but also due to the kinetic energy change,
are derived in Appendix A. These equations can be
used to derive the AGA (Eq. A12), the Cullender
and Smith (Eq. Al 2), and other commonly used
equations.
If one substitutes different correlations for the
Fanning friction factor for smooth pipes in Eq. Al2

Table 4
Deviation from the Colebrook- White
roughness)
Error

Average
Error
Maximum
Error

Serghides
(1) (1984)
0.00037
- 0.003

Z-S (II)
(1982)
0.0287
0.206

(1939) equations

Serghides
(II) (1984)
0.0359
- 0.355

Chen
(1979)

for rough

Z-S (I)
(1982)

pipes (For the whole range of Reynolds

Haaland
(1981)

Jain
(1976)

Eck
(1973)

Churchill
(1977)

number

and relative

Wood
(I 966)

pipe

Moody
(1947)

0.137

0.234

0.582

0.929

3.010

4.092

5.107

6.276

- 0.689

- 1.060

2.952

4.598

11.962

- 70.730

- 32.553

- 26.790

146

L.-h. Ouyang, K. Aziz/Journal of Petroleum Science and Engineering

and then solves the equation for flow rate, the simplified form of the flow equation given below is
obtained (Institute of Gas Technology, 1965; Ikoku,
1980; Beggs, 1984):

(26)
where the efficiency factor, E, is introduced to account for the effect of pipe roughness,
and the
constant c = 0.0375 for field units and c = 0.00697
for SI units. The values of the ai constants resulting
from various friction factor correlations (Panhandle,
modified Panhandle, Weymouth, and IGT) for both
field and SI units are given in Table 5a.
When the elevation term is neglected, Eq. 26
becomes:

14 (1996) 137-158

Eqs. 26 and 27 are usually named as the Panhandle equation, the modified Panhandle equation, the
IGT equation and the Weymouth equation according
to the values of the a, constants used (i.e., the
correlation chosen for calculating the friction factor).
In order to avoid confusion, they are all classified as
old simplified flow equations to distinguish from
new simplified flow equations to be presented later
in this paper.
Eqs. 26 and 27 are widely used in the gas industry
and in some textbooks (Katz et al., 1959; Institute of
Gas Technology,
1965; Ikoku, 1980; Beggs, 1984).
It appears that these simplified flow equations are
often used without regard to their ranges of applicability. In fact, Eqs. 26 and 27 are based on the
Reynolds number obtained by (IGT, 1965; Ikoku,
1980; Beggs, 19841:

where with D in inches, q in scf/day, the constant

G = 1.91286; and with D in meters, q in m3/sec @
s. c., G = 1.4858 X 105. This constant is obtained by
taking pg = 0.0105 cP, p,, = 14.7 psia, T,, = 520R
(Field units) or ~~ = 1.05 X 1O-5 Pa. s, p,, =
101,325 Pa, q, = 288.15K (SI units).

Table 5
Friction factor
equation
(a) Values of coefficients
Panhandle
Mod Panhandle
IGT h
Weymouth
(b) Values of coefficients
Friction factor equation
Blasius (1911)
i/7th power law
1/8th power law
1/9th power law
Modified 1/9th
I/ 10th power law
Panhandle
Modified Panhandle
IGT
Weymouth

a, (SI units)

in Eqs. 26 and 27 a
157.92
152.92
169.09
137.19
in Eqs. 29 and 30
a, (SI units)
17.1541
17.4983
21.6136
25.6034
24.8841
29.6754
40.3120
108.1291
24.6615
137.1902

a, (Field units)

435.73
737.50
343.28
433.50
a, (fields units)
265.5812
270.9 114
295.871 I
3 19.0366
3 10.0735
342.9768
402.7467
722.5812
307.3000
433.5065

a The values of a, will change slightly if different standard conditions

b a, and a4 are set as 337.90 and 0.4000 in Beggs (1984).
The values of a, will change slightly if different standard conditions

a2

1.0788
1.0200
1.1110

I .oooo
a2

0.5714
0.5714
0.5624
0.5556
0.5556
0.5501
0.5394
0.5100
0.5556
0.5000

0.5394
0.5100
0.5560
0.5000

0.4604
0.4900
0.4444
0.5000

2.6182
2.5300
2.6667
2.6667

a3
0.4286
0.4286
0.4376
0.4444
0.4444
0.4499
0.4606
0.4900
0.4444
0.5000

a4
2.7143
2.7143
2.6873
2.6661
2.6667
2.6502
2.6182
2.5300
2.6667
2.6667

a5
0.1429
0.1429
0.1249
0.1111
0.1111
0.1001
0.0788 1
0.0200
0.1111
0.0000

and the general gas constant are used.

and the general gas constant

are used.

L.-b. Ouyang, K. Aziz/ Journal of Petroleum Science and Erqineering 14 (1996) 137-158

a) Change Pipeline Length

-

.....

..---.-----.-.-.

Modified 1/9th
Panhandle
ModPanhandls
IGT
Weymooth

b) Change Outlet Pressure

150 -

Modified 147th

-...---.
----.-.-.

ModPanhandle
IGT
Weyrnouth

panhand,*

IW c

Fig. 3. Gas flow rate based on different friction factor formulae (horizontal

.....

. ..-.....

pipeline #I).

Modibed 1/9th
Panhandle
ModPanhandle

b) Change Welihsad Pmssure

..
..----.
----.-.-.

Modified V9h
Panhandle
ModPanhandle
IGT
Weymouth

Fig. 4. Gas flow rate based on different friction factor formulae (vertical gas well #2).

147

148

L.-b. Ouyang,

K. Aziz/Journal

ofPetr&um

Science and Engineering

14 (1996) 137-158

Table 6
Test data for gas pipelines and gas wells
Data set

#l

#2

Type
Pipe ID (in)
Pipe roughness (micro-in)
Pipeline length (mile)
Well depth (feet)
Inclination angle
Gas gravity
Gas Viscosity (cp)
Pseudo-critical
P (psia)
Pseudo-critical
T CR)
Average temperature (Fl
Inlet pressure (psia)
Outlet pressure (psial

Gas Pipeline
15.437
0.0
15.452

Vertical

0.0
0.57
0.0121
672.5
350.0
75.0
798.24
729.72

Gas Well

2.992
0.0
_

#3

#4

Gas Pipeline

Slant Gus Well

4.0
600.0
0.568
_

11029.0
90.0
0.6997
0.012
801.2
410.9
202.0
4123.2
2990.4

4.0
600.0
8000.0
70.0
0.75
0.018
661.0
411.0
560.0
600.0
63.0

0.0
0.75
0.018
661.0
411.0
545.0
200.0
30.0

pipe flow where the gas has a viscosity of about

0.0105 CP or 1.05 X IO- Pa. s.
(b) A specific value of the term T,,/p,, is used in

Three aspects related to Eqs. 26 and 27 should be

noted:
(a) Eqs. 26 and 27 should be used only for gas

&

80 _
a) tforizonfal
-

b) Gas Well #Z

Pipeline#I

Eq.27
Eq, 30, Mug=O.OiZCp
Eq, 30, Mug=O.O2&4cp
Eq, 30, Mugd.036cp

.......

. ..----- - -

70 _

Eq.26

.............
Eq, 29,

----..--.
----.

Mug=O.OlZcp
Eq. 29, MJg=o.o24cp
Eq.8 Mug=O.O36cp

60_

2
2 so_
3
.Y
2
6
G

40

<
G

30 _

20

IO

I
60

80

Pipeline Length (miles)

Fig. 5. Effects of viscosity on gas flow rates with changing

m
Well Depth (feet)

pipeline length or well depth

IGXO

l2mJ

I4003

L.-h. Ouyang, K. A:iz/Journal

calculating
constant G for the Reynolds number.
Using different values of this ratio in Eqs. 26 and 27
can lead to errors.
(cl The widely-used correlations for the friction
factor (the Panhandle, the modified Panhandle, the
IGT, and the Weymouth) are not the best choices to
calculate gas flow in a pipe. This observation results
from the comparison of friction factor calculations
for smooth pipes summarized in Table 2.
In fact. a more general form of simplified flow
equations can be derived from Eq. A10 by replacing
the Fanning friction factor with different correlations:

D4

x-

(29)

/q

.----.-..
- - - -.

Eq.27
Eq. 30, Mug=O.OlZcp
Eq, 30, Mug=O.O24cP
Eq. 30, Mug=O.O%cp

149

of Petroleum Science and Engineering 14 (1996) 137-158

When the elevation term is neglected, the simplified flow equation for horizontal pipes is obtained
from Eq. 29:

Pf-P22

02

I I i-

(30)

TZL

where the values of a, corresponding

to different
friction factor correlations in both sets of units are
given in Table 5b (an equation of the same form as
Eq. 30 was first published by Towler and Pope,
1994).
Eqs. 29 and 30 can be utilized for flow calculation with any type of gas and at any rate provided
that the correlation for the friction factor is selected
so that it applies for the Reynolds number range of
interest (see Table 31, and the efficiency factor is
chosen to correctly account for pipe roughness.

70

................
...-.--..
----.

20 1

Eq. 26
Eq. 29, M,,@.O12cp
Eq. 29, Mug=O.O24cp
Eq. 29, Mug--o.O36cp
\

500

low

I XXI

2m

Wellhead Pressure (psia)

Fig. 6. Effects of viscosity on gas flow rates with changing

outlet or wellhead pressure.

zsw

3m

3mJ

150

5. Comparison

L.-b. Ouyang, K. Ariz/.lournal

of Petroleum Science and Engineering

of flow equations

5. I. Sensitiuity analysis
(a) Friction factor correlation
As stated in Section 3.1, the Fanning friction
factors obtained from different correlations can differ
greatly, thus the gas flow rates predicted will also be
different. Fig. 3 shows the gas flow rates for the
smooth pipeline # 1 in Table 6 with changing pipeline
length (Fig. 3a) and changing outlet pressure (Fig.
3b). The gas flow rates are computed from the new
simplified flow equation (Eq. 29) where the friction
factor is calculated from the modified 1/9th power
law, Panhandle, modified Panhandle, IGT, as well as
Weymouth equations. In both situations, there are
significant differences between the gas flow rates
from the Weymouth and other equations. The largest
difference is as high as 30%, which occurs in the
case of changing the outlet pressure (Fig. 3b). The
other four equations show only small differences in
gas flow rate with changing pipeline length, but still

Fig. 7. Kinetic energy and pipe roughness

14 (19%) 137-158

give appreciable differences for the case of changing

outlet pressure. Obviously, it is very important to
choose an appropriate correlation for the friction
factor. Otherwise, calculated gas flow rates can be in
error. Usually gas pipelines are designed for Reynolds
numbers in the range of lo6 to 107, so the Panhandle
equation is recommended for gas flow rate calculations in pipelines based on the comparison in Section
3.1. Because the gas flow rates change substantially
in gas wells, the correlation for Fanning friction
factor should be chosen according to the Reynolds
number as stated in Section 3.1. Calculation of gas
flow rates in a vertical well (Well #2 in Table 6)
supports this comment (Fig. 4).
(b) Gas uiscosi~
When the gas viscosity is altered, the gas flow
rates will change. Figs. 5 and 6 compare the gas flow
rates obtained from the new simplified flow equations (Eqs. 29 and 30) where viscosity affects flow
rates and from the old simplified flow equations
(Eqs. 26 and 27) which are independent
of gas
viscosity for the gas pipeline # I and the gas well #2

effects (horizontal

pipeline #3).

L.-h. Ouyang, K. Aziz/Joumal

of Petroleum Science and Engineering 14 (19961 137-158

Table 6. The pipe roughness is 600.0 micro-inches

(relative roughness = 0.00015). The friction factor is
obtained directly by an iterative solution of the Colebrook-White
(1939) equation. The gas flow rates
predicted by using new simplified flow equations
(Eqs. 29 and 30) and calculating friction factor from
the Panhandle equation with E = 1.0 and E = 0.92
are also plotted in Fig. 7. Obviously the differences
between the gas flow rates with and without the
consideration of pipe roughness are large, especially
for the case of changing outlet pressure, where the
maximum differences reach about 28%. Although
0.92 is a typical value for the efficiency factor E in
the Panhandle
equation to account for the pipe
roughness effect, the gas flow rates predicted are still
much higher than those obtained from rough pipe
equations. The differences can be reduced if an
appropriate efficiency factor is used, but its value for
different practical pipe types and gas flow rates can
be quite different. Based on the Colebrook-White

(Table 6). The gas viscosity values of 0.012 cP,

0.024 CP and 0.036 CP are considered and the Panhandle equation is used for friction factor calculation. Recall that the gas viscosity is assumed to be
0.0105 CP and absorbed in Reynolds number calculation in the old simplified flow equation. The gas
flow rates based on Eqs. 29 and 30 for gas viscosity
= 0.012 CP are close to the values calculated from
Eqs. 26 and 27. With the increase in gas viscosity,
the differences between the gas flow rates predicted
by new simplified equations and by old simplified
equations become larger and larger. For gas flow
with viscosities different from 0.0105 cP, the old
simplified flow equations should not be used.
(cl Pipe roughness
In addition to the Reynolds number, the pipe
roughness also affects the Fanning friction factor for
turbulent Bow in rough pipes. The gas flow rates
calculated from the new general flow equation (Eq.
A8) are shown in Fig. 7 for horizontal pipeline #3 in

--.______
..T

-...
-.
*\

\\

0.9 ~

\
\

\
\

\
\

\
\.,

$
2
\

...
-..,

L.

..

\
\

Y.
\

\
\

Relative

Roughness

__ __----.

0
,,0e_,
,,0e_6

----.
-._.-..
_._._._._....
_._.--._

l.Oe-5
0,000,
0.00,
0.0,

*...
*...
.\,
XI,

.
xx.

Id
Reynolds

Fig. 8. Efficiency

151

106
Number

factor for the use of Panhandle

friction factor correlation.

152

L.-b. Ouyang, K. Aziz / Journal of Petroleum Science

Fig. 9. Correction

and Engineering14 (I Y96i 137-158

factor for the kinetic energy change

equation, the authors have calculated the efficiency

factor value corresponding to the Panhandle friction
factor correlation (Fig. 8). Note that the efficiency
factor is very sensitive to both the pipe roughness
and the Reynolds number. For Re = 10h, the efficiency factor decreases from 0.97 to 0.52 when the
relative pipe roughness increases from lop7 to 0.01.
Cd) Kinetic energy change
The effect of kinetic energy change on gas flow
rates in gas pipeline #3 are shown in Figs. 7 and 9
and 10. As expected, these results indicate that the
kinetic energy changes are very small (the correction
factor for the kinetic energy change E, is very close
to 1.0, see Fig. 9) and can be neglected in most
situations of practical interest. But there appears a
maximum gas production rate for horizontal pipeline
#3 with the change of outlet pressure if the kinetic
energy term is considered (Fig. 7b) . In some cases,
the error caused by neglecting the kinetic energy

change can reach about 3.37% (Fig. 9). Young (1967)

stated that the error may reach 9%. Tian and
Adewumi (1992) reported 28% and 43% errors in
outlet pressure calculation for special gas flow problems (the pressure drops due to the kinetic energy
change are 6.8% and 5.0% of the total pressure
drops) . These conditions are, however, not likely to
be encountered in practice.
(e) Eleuation change
For gas flow in a slant well (gas well #4 in Table
6) with elevation change, the new general flow
equation (Eq. A8) and the AGA equation (Eq. A12)
are both employed to calculate gas flow rates (Fig.
10). The results are different. The deeper the well,
the larger the difference between them. This is reasonable since the AGA equation is derived by assuming that p* in the elevation term remains unchanged along the pipe, which is not true for gas
wells where the pressure may change rapidly.

The maximum value observed in Fig. 7b when kinetic energy

term is included can be confirmed analytically.

hth
correction

Young (1967) and Tian and Adewumi

factor u is ignored.

(1992).

the

L.-b. Ouyang, K. Azi:/Journal

of Perroleum Science and Engineering 14 119961 137-158

--------.

153

New Eq., KE Change Neglected

New Eq., KE Change Considered
AGA Equation with Newtion Change

i
8000

lOax

12OiM

1400

16000

Well Length (feet)

Fig. 10. Comparison

5.2. Application

of results from new equation

derived in this work (Eq. A8) and the AGA equation

to field data

The new general flow equation (Eq. A8) and the

AGA general flow equation (Eq. A12), along with
the new simplified flow equation (Eq. 29) which
uses the Panhandle friction factor correlation, are

(Eq. AIZ) (slant gas well #4).

used to calculate the gas flow rates in practical gas

pipelines. The field data, Test A, Test C, Test F-l,
Test Q, Test YA-1, are chosen randomly from IGT
Report No. 10 (Institute of Gas Technology, 1965).
The pipe length, the pipe ID, the inclination angle
and the number of data points are listed in Table 7a.

Table 7
Source

Test A

Test C

Test F- 1

Test Q

Test V-l

Test YA-I

152.4
19.44
- 0.09
11
gas flow rates

16.16
29.3 I
0.019
10

10.63
25.5
-0.02

9.12
25.375
0.013
9

Test F-I
0.678
2.694
2.603
2.593

Test Q
5.946
1.538
1.536
1537

Test V-I
0.973
2.463
2.462
2.461

Test YA-

(a) Test pipeline data distribution

Length (mile)
93.69
36.82
Pipe ID (inch)
12.0
15.375
Inclination (1
0.076
0.067
Number of data points
II
8
(b) Average relative error (%o) between the predicted and the measured
Equation
Test A
Test C
New Panhandle (E = 0.92)
7.224
19.881
AGA Equation
4.830
0.49 1
New Equation (No k term)
4.995
0.492
New Equation (With k term)
4.99 1
0.506

I I.023
0.428
0.425
0.422

154

L.-h. Ouyzq,

K. Aziz/Journal

of Petroleum

The pipe length varies from 9.12 miles to 152.4

miles, the pipe ID from 12.0 to 29.31 inches, and the
inclination
angle from - 0.09 to +0.076. All of
the results are obtained by iteration using the singlestep method. The compressibility
factor is calculated
from the Dranchuk and Abou-Kassen (1975) method
at the average temperature and pressure. The average
pressure is updated during each iteration. Fig. 11
compares the calculated gas flow rates with measured values. The differences between the predicted
and the measured gas flow rates are quite small for
both the new equation and the AGA equation (Table
7b). Even for Test F- 1, where the pipeline length is
152.4 miles, the differences are only about 2.603%
and 2.694%. In general, the new equation (Eq. A8)
predicts a slightly better gas flow rate than the AGA
equation (Eq. A12). However, over the range of

Science and Engineering

14 (19961 137-158

conditions considered, the AGA equation yields good

results. The main reasons for this are: (a) The pipe
inclination is very small (smaller than + 0.1 degree),
so the elevation term is much smaller than the
friction term and has little effect on the total pressure
drop. (b) The pressure change along the pipelines is
small. (c) The kinetic energy effects are insignificant. Notwithstanding
these results, it must be emphasized that the equations based on approximate
expressions in the elevation term are not as reliable
as the equations where no such assumptions
are
made.
As demonstrated earlier, the new simplified flow
equations (Eq. 29 with E = 0.92 and the Panhandle
correlation used for the friction factor calculation)
give poor predictions of the gas flow rates for some
pipelines but good predictions for other pipelines.

Unit S/me Straiaht Line

E9. 29 (Panhacdle, EzO.92)
New Eq., KE Change Considered
AGA Equation

zw

AJU

Measured Gas Flow Kate (MMscf/D)

Fig. 11. Comparison

of measured and calculated

gas flow rates.

L.-h. Ouyang,

K. A&,/

Journal

of PetroleumScience

The difficulty related to simplified flow equations is

the selection of the efficiency factor and the friction
factor correlation.

6. Recommendations

and conclusions

The new general flow equations which account

for the effects of friction, elevation and kinetic energy change can be easily used for any single phase
gas flow in pipelines and wells. It is more accurate
than the AGA equation or other simplified forms.
The AGA equation with the elevation change
introduces an error for gas flow calculations, especially for the cases where the elevation change is
large. For slightly inclined gas pipelines, the AGA
equation can be utilized provided the kinetic energy
change is negligible.
Kinetic energy change may be neglected for most
industrial applications. In some special cases, such as
where pressure drop over the pipe is very large, the
kinetic energy change can lead to appreciable errors
in gas flow rate and pressure drop calculations. For
these cases, using an equation, such as the new
general flow equations can lead to improved accuracy.
For gas flow where the kinetic energy change is
negligible,
the new general flow equation can be
simplified into specific flow equations (Eqs. 29 and
30). The simplified flow equations thus derived can
be used for steady-state gas flow with any flow rate
and gas type, provided that the correct correlation for
the friction factor and the correct value of the efficiency factor are chosen. In contrast, the old simplified flow equations can only be applied to gas flow
with gas viscosities of about 0.0105 cP, and thus can
lead to particularly large errors for dense phase flow
in pipelines where the viscosity may be as high as
0.06 CP (Gregory et al., 1979).
Correlation for the Fanning friction factor should
be selected according to the Reynolds number and
the relative pipe roughness of interest. For single
phase gas flow in smooth pipelines or wells, the
Blasius equation can be used for the Reynolds number range of 3000 to lo, the modified 1/9th power
law for 2 X 10 to IO, the Drew et al. equation for
2000 to 4000 and 4.0 X IO to 4.0 X 106, and the
Panhandle equation for 2 X lo6 to 10. For rough

and Engineering

14 (19961 337-158

15s

and smooth pipelines

or wells, Chen (1979).
Serghides (1984) (I) and (II), Zigrang and Sylvester
(1982) (I) and (II), as well as Colebrook-White
(1939) equations can be used. Because the Colebrook-White
equation also gives good results of the
Fanning friction factor evaluation for smooth pipes.
it is strongly recommended to be used for both rough
and smooth pipe flows.
Nomenclature
constant. Eqs. 26, 27, 29, 30
constant, Eq. 8
pipe diameter (m; inch)
efficiency factor, Eqs. 26, 27, 29, 30
correction factor, Eq. A9
Fanning friction factor
acceleration due to gravity (m/s; ft/s)
conversion factor
constant, Eq. 28
pipe length (m; mile)
pressure (Pa; psia)
average pressure (Pa, psia)
pressure at standard conditions (Pa; psia)
gas flow rate (m /s @ s. c.; scf/day)
constant. Eq. 7
universal gas constant (J . Kgmol~ K- ;
psi . ft lbmole- . R- )
Reynolds number, Eqs. 8 and 28
dimensionless elevation factor with the kinetic
energy change ignored
dimensional elevation factor
dimensionless
elevation and kinetic energy
change factor
constant, Eq. 7
temperature (K; R)
temperature at standard condition (K; R)
gas velocity (m/s; ft/s)
elevation (m; ft)
compressibility
factor
correction factor to compensate for variations
in the velocity profile over the pipe cross-section
specific gravity of gas, which is defined as the
ratio of the density of the gas to the density of
dry air with both at standard temperature and
pressure
absolute pipe roughness (m;inch)
gas viscosity (Pa . s, cP)
gas density (kg/m ; lbm/ft)

156

L.-b. Oqang,

Appendix
Equation

A. Derivation

K. Aziz/Journal

of Petroleum

udu

dp
P

a Tc,

Rc

+2cdL=O
Dg,

(A)

There are four terms appearing in Eq. Al, i.e., the

pressure drop term, the elevation term, the kinetic
energy term, and the frictional pressure drop term.
For real gases:
pV = nZRT

A=-

4fBC
DS,

s = 2 B ?!-sin 8
g,
It is convenient

to rewrite Eq. A5 as:

p2 - (BC/ag,)
p2(s'p2+4

dp2=

28.97~~
-gjr~=B~

(42)

where M is the gas molecular weight, and y,, is the

specific gravity of the gas.
If 9 is the volumetric flow rate at standard conditions, then the local gas velocity is:

(A31

A + spf
A + sp;

Often, the compressibility

factor Z does not vary
greatly over the range of pressures and temperatures
existing in pipelines so that it can be assumed constant, thus yielding:

L-Sk =

2af
1
-+-

lnPZ
PI
D
4af

and:
s = SL

Eq. A7 can be rearranged

-yZ($j(~)($U))dp=

(A?

with:

du=

W)

-dL

This equation can be numerically integrated as discussed in Aziz (1967) for both wells and pipelines,
or it can be integrated analytically provide A, B and
C are assumed to be constant:

or:
PM
P=E=

14 (1996) 137-158

of the General Gas Flow

The momentum balance equation for natural gas

flow in pipes has the following differential form:

-+Ldz+-

Science and Engineering

to solve for the flow rate:

--$f

(Ad)
0.5

D2.5

Substituting Eqs. A2, A3 and Eq. A4 into Eq. Al,

the following equation can be obtained:
2BC dp
+ ( sp2 + A)dL = 0
-ag,
P

2pdp-

where:
B=---

28.97~~
ZRT

(-45)

(A81

This equation is the most general integrated flow

equation published so far. It accounts for the effects
of elevation, friction, and kinetic energy change.
Note that for horizontal flow:
s = 0 *

s = 0 ads,

= 0

This makes the last term in Eq. A8 singular.

The

L.-h. Ouyang, K. Aziz/ Journal of Petroleum Science and Er+wering

singularity can be removed with the application of

LHopitals rule or by directly substituting s = 0 in
Eqs. A5 or A6. In either case the result is:

(Al41

lny

l--

E,=

for the
where:
28.97g
c, = ~
Rg,

-0.5

D
2a.P

PI

157

It is also interesting to note that the widely-used

Cullender
and Smith (1956) equation for use in
vertical wells can also be obtained by neglecting the
kinetic energy term in Eq. A6 and rewritting it as
follows:

(A91
where E, is called the correction factor
kinetic energy change and it is defined by:

14 (19961 137-158

Inasmuch as the kinetic energy effect is usually

small, it can often be neglected in Eq. Al, which is
equivalent to setting sk = s in Eq. A8:

0.5
(AlO)

Another approach is to neglect the kinetic energy

term and rewrite Eq. A5 as follows:
2pdp+(.+$,+A)dL=O

Other methods, such as Sukker and Cornell (19551,

Rzasa and Katz (1945), are all special cases of Eq.
A6
Note that the equations in this Appendix are valid
for any consistent set of units. For example, in the SI
system, the units for mass, time, and length are
kilogram, second, and meter, respectively. The units
for all other quantities must be derived from these
three base units.

(All)

where the authors have assumed that the elevation

term can be evaluated by setting p2 to be constant at
its average value, pi,. The resulting equation is the
well-known AGA equation (Institute of Gas Technology, 1965):

For horizontal pipes, with s = 0 and AZ = 0, both

Eqs. Al 0 and A12 reduce to:

(Al31

References
Aziz, K., 1962-1963. Ways to calculate gas flow and static head.
Petroleum Engineer, Series from Nov. 1962 through Sept.
1963. Vols. 34-35.
Aziz. K., 1967. Calculation of bottom-hole pressure in gas wells.
J. Pet. Technol.. 19(7): 897-899.
Aziz, K., 1994. Multiphase Flow Technology for the Development
of Oil and Gas Reservoirs. Course note, Stanford University.
Calif.
Baker, A.. Nielsen, K. and Gabb, A., 1988. Pressure loss. liquid
holdup calculations developed. Oil Gas J.. 86( 1 I J: 55-59.
Beggs, H.D., 1984. Gas Production Operations. Gulf Publishing
Co, Houston, Tex., 287 pp.
Beggs. H.D.. 1991. Production
Optimization
IJsing NODAL
Analysis. OGCI Inc.. Tulsa. Okla.. 41 1 pp.
Blasius, H., 1911. Das Ahnlichkeitsgesetr
bei Reibuogsvorgangen. Physik. Zeitschr., XII: 117.5-l 177.
Brill. J.P. and Beggs, D.P., 1991. Two-phase
Flow in Pipes.
Course note, University of Tulsa, Tex.
Chen. N.H.. 1979. An explicit equation for friction factor in pipe.
Ind. Eng. Chem. Fund., lS(3): 2966297.

158

L.-b. Ouyang, K. Aziz/ Journal of Petroleum Science and Engineering 14 (1996) 137-158

Churchill, S.W., 1977. Friction-factor

equation spans all fluid
flow regimes. Chem. Eng., 84(24): 91-92.
Colebrook, C.F., 1939. Turbulent flow in pipes, with particular
reference to the transition region between the smooth and
rough pipe laws, J. Inst. Civ. Eng. London, 12: 133-156.
Colebrook, C.F. and White, CM., 1937. Experiments with fluid
friction in roughened pipes. Proc. R. Sot. London, 16lA:
367-38 1.
Crawford, P.B. and Fancher, G.H., 1959. Calculation of flowing
and static bottomhole
pressures of natural gas wells from
surface measurements.
Tex. Pet. Res. Comm., Austin, Tex.,
Bull. 72.
Cullender, M.H. and Smith, R.V., 1956. Practical solution of
gas-flow equations for wells and pipelines with large temperature gradients. Trans. AIME, 207: 281-287.
Dranchuk, P.M. and Abou-Kassen,
J.H., 1975. Calculation of Z
factors for natural gases using equations of state. J. Can. Pet.
Technol., 14(31: 34-36.
Drew, T.B., Koo. E.C. and McAdams, W.H., 1932. The friction
factor for clean round pipes. Trans AIChE, 28: 56-72.
Eck. B., 1973. Technische Stromungslehre.
Springer, New York,
N.Y.. 142 pp.
Economides, M.J., Hill, A.D. and Ehlig-Economides,
C., 1994.
Petroleum Production Systems. PTR Prentice Hall, Englewood
Cliffs, N.J., 61 1 pp.
El-Oun, Z., 1990. Gas-liquid
two-phase flow in pipelines. 65th
Annu. Tech. Conf. Exhib., Sot. Pet. Eng. AIME. New Orleans, La., SPE 20645.
Govier, G.W. and Aziz, K., 1972. The Flow of Complex Mixtures
in Pipes. Van Nostrand Reinhold Co.. New York, N.Y., 792
PP.
Gregory, G.A., Aziz. K. and Moore, R.G.. 1979. Computer design
of dense phase pipelines. J. Pet. Technol., 31(l): 40-50.
Haaland, S.E., 1981. Simple and explicit formula for the friction
factor in turbulent pipe flow including natural gas pipelines
IFAG B- 13 1. Div. Aero- and Gas Dynamics, The Norwegian
Inst. Technol.
Ikoku, C.U., 1980. Natural Gas Engineering: A System Approach.
Penn Well Publishing Co., Tulsa, Okla., 788 pp.
Institute of Gas Technology,
1965. IGT Report No 10: Steady
Flows in Gas Pipelines. PRC Project NB-13, Am. Gas Assoc..
Inc., New York. N.Y.

Jai, A.K., 1976. Accurate explicit equation for friction factor.

ASCE Hydraul. Div. J., 102(HY5): 674-677.
Katz, D.L., Cornell, D., Vary, J.A., Kobayashi, R., Elenbaas. J.R.,
Poettmann, F.H. and Weinaug, C.F., 1959. Handbook of Natural Gas Engineering. McGraw-Hill Book Co. New York, N.Y..
802 pp.
Knudsen, J.G. and Katz, D.L., 1958. Fluid Dynamics and Heat
Transfer. McGraw-Hill Book Co., New York, N.Y., 576 pp.
Moody, M.L., 1947. An approximate formula for pipe friction
factors. Trans. ASME, 69: 1005- 1011,
Poettmann, F.H., 1951. The calculation of pressure drop in the
flow of natural gas through pipe. Tram AIME, 192: 3 17-326.
Prandtl, L., 1935. The mechanics of viscous fluids. In: W.F.
Durand (Editor). Aerodynamic
Theory. III(Division G): 344
208.
Rzasa, M.J. and Katz, D.L., 1945. Calculation of static pressure
gradients in gas wells. Trans. AIME, 160: 100-l 13.
Serghides, T.K., 1984. Estimate friction factor accurately. Chem.
Eng., 91(5): 63-64.
Sukker. Y.K. and Cornell, D., 1955. Direct calculation of bottomhole pressures in natural gas wells. Trans. AIME, 204: 43-48.
Tian. S. and Adewumi. M.A., 1992. Development of analytical
design equation for gas pipelines. 67th Annu. Tech. Conf.
Exhibition, Sot. Pet. Eng. AIME., Washington,
D.C.. SPE
24861.
Towler, B.F. and Pope, T.L., 1994. New equation for friction-factor approximation developed. Oil Gas J., 92(14): 55-58.
Ward-Smith.
A.J., 1980. Internal Fluid Flow. Clarendon Press,
Oxford, 566 pp.
Weymouth,
T.R.. 1913. Problems in natural gas engineering.
Trans. ASME, 34: 185-231.
White, F.M., 1986. Fluid Mechanics. McGraw-Hill Book Company. New York. N.Y.. 732 pp.
Wood. D.J., 1966. An explicit friction factor relationship. Civ.
Eng., 36( 12): 60-6 I.
Young, K.L., 1967. Effect of assumptions used to calculate bottom-hole pressures in gas wells. J. Pet. Technol., 19(41: 547550.
Zigrang, D.J. and Sylvester, N.D., 1982. Explicit approximations
to the solution of Colebrooks friction factor equation. AIChEJ,
2X(3): 514-515.