Steady-State Flow of Gas through Pipes

Introduction
Pipes provide an economic means of producing (through tubing or casing) and transporting (via flow
line or pipeline) fluids In large volumes or giant distances. They are convenient to fabricate and
Install, and provide an almost indefinite life span. Because flow is continuous, minimal storage
facilities ire required at either end (held supply end, and the consumer end). Operating costs are
very Ion, and flow ¡s guaranteed under all conditions of weather, with good control (an Installed
pipeline can usually handle a wide range of flow rater). There are no spillages or other handling
losses unless the line develops a leak, which can be easily located and fixed lot surface.
The flow of gases through piping systems Involves flow In horizontal, In-clined, and vertical
orientation, and through constrictions such as chokes for flow- control. This chapter introducer some
basic concepts foe the flow

Gas Flow Fundamentals

All fluid flow equations are derived from a bask energy balance which, For a steady-state system (no
time-dependence of flow parameters), can be expressed as:

Change in Internal energy + Change In kinetic energy + Change In potential energy +work done on
the fluid + Heat energy added to the fluid - Shaft work done by fluid on the surrounding = 0

Thus, on a unit mass basis, the energy balance for a fluid under steady-state flow conditions can be
written as:

dv 2 g
dU + + dz +d ( pV )+ dQ−d w s=0 (7-1)
2 g c gc

where

U = Internal energy, ft-lbf/lbm

v — fluid velocity, ft/sec
z — elevation above a given datum plane, ft
p = pressure, lbf/ft2
V = volume of a unit mass of the fluid, ft3llbni
Q = heat added to the fluid, ft-lbf/lbm
ws shaft work done by the fluid on the surroundings, ft-lbf/lbm
g — gravitational acceleration, ft/sec!
gc — conversion factor relating mass and weight

This basic relationship can be manipulated In several different ways. Com monly, It is converted Into
a mechanical energy balance using the well-known thermodynamic relations for enthalpy (h):
dU+d(pV)=dh=Tds+Vdp

where h — specific fluid enthalpy, ft-lbf/lbm

T — temperature, °R
s = specific fluid entropy, ft-lbf/lbm

Equation 7-1 now becomes:

d v2 g
Tds+Vdp+ + dz+ dQ−d w s=0
2 gc gc

−dQ
For an ideal process, ds= ,since no process ¡s ideal (or reversible), ds ≥−dQ+d l w
T

Tds=−dQ+ d l w

where lw is the lost work due to lrreversibilities such as friction.

On this further substitution, Equation 7-1 becomes:

d v2 g
Vdp+ + dz +d l w −d w s=0 (7-2)
2 g gc

Neglecting the shaft work w, and multiplying throughout by the fluid density,ρ:

ρd v 2 g
dp+ + ρdz+ ρd l w =0 (7-3)
2 g c gc

All the terms in Equation 7-3 have units of pressure. Equation 7-3 can also be written as:

ρ ∆ v2 g
∆ p+ + g ∆ z+ ρ ∆l w =0
2 gc g c

Or

ρ ∆ v2 g
∆ p+ + g ∆ z+ ∆ p f =0 (7.4)
2 gc g c

Where Δpf represents the pressure drop due to friction, and is dependent upon the prevailing flow
conditions.
Types of Single-Phase Flow Regimes and Reynolds Number

Four types of single-phase flow are possible: laminar, critical, transition, and turbulent (see Figure 7-
1). Reynolds applied dimensional analysis to flow phenomena, and concluded that the flow regime
that will prevail is a function of the following dimensionless group known as the Reynolds number,
NRe:

Inertia forces dv ρ
N ℜ .= = (7-5)
viscous forces µ

Where

d = (inside) diameter of the conduit through which the fluid Is flowing

v = velocity of the fluid
ρ = density of the fluid
µ = viscosity of the fluid

For cross-sections other than circular, an equivalent diameter, d, defined as four times the hydraulic
radius, Rh, is used instead of d:

[ area of flow ]
d e =4 Rh=4 (7-6)
[ wetted perimeter ]
For example, for a flow conduit with a square cross-section (a x a):

Cross-sectional area of flow =a2

Wetted perimeter = 4a

Thus de= 4(a2/4a)=a

For flow through a casing-tubing annulus with casing inside diameter equal to d ci and tubing outside
diameter equal to dto:

π 2 2
Cross-sectional area of flow =( )(d ci −d ¿ )
4

Wetted perimeter = π ¿

π
4 ( )(d ci2−d ¿2)
Thus 4
de=
π ¿¿

To verify the applicability of this approach, consider the case of a pipe with a circular cross section of
diameter d:
 π d2
Cross-sectional area of flow =
4

Wetted perimeter = πd

π d2
Thus, d e =4
4 πd ( )
=d

The units for parameters in the Reynolds number should be consistent so that a dimensionless
number is obtained. One such consistent set would have d in ft, v in ft/sec, ρ in lbm/ft3, and viscosity
µ, frequently given in centipoise (cp), can be converted into lbm/(ft sec) using the converstion factor
of 1 cp = 6.7197 x 10-4 lbm/ft sec). Thus:

ft lbm ft lbm
d ( ft ) v( ) ρ( 3 ) d ( ft ) v ( ) ρ( 3 )
sec ft sec ft (7-7)
N ℜ= =1488
μ ( cp ) × 6.7197× 10 −4
μ ( cp )

For gases, the flow rate is commonly expressed in Mscfd (thousands of standard cubic feet per day).
This volumetric rate, qsc (Mscfd), at standard cinditions of pressure (p sc, psia) and temperature (Tsc,
°R), can be converted into mass flow rate, m in lbm/sec. as follows:

( 1000 q sc ) M ρsc 3.1243 ×10−2 q sc γ g psc

m= Avρ= =
( 24 )( 3600 ) Z sc R T sc T sc

Where

A = cross sectional area of flow, ft2

M = molecular weight of the gas

R = gas constant (=10.732 psia-ft3/lbmole-°R)

psc = pressure at standard conditions, psia

Tsc = temperature at standard conditions, °R

Zsc = gas compressibility at standard conditions (= 1)

ϒg = gas gravity (air=1)

Thus, for a circular pipe of diameter d(ft), vρ is given by:

( 4 ) ( 3.1243 ×10−2 ) qsc γ g p sc

vρ=
T sc πd 2

( 3.9780 ×10−2 ) q sc γ g psc

vρ=
T sc d 2

Substituting for vρ in Equation 7-9 for NRe:

 d (3.9780 ×10−2) q sc γ g psc 59.1991 q sc γ g p
N ℜ= = sc

T sc d 2( μ × 6.7197 ×10− 4) T sc dμ

Generally, d is used in inches for most gas flow equations. For d in inches, N Re becomes:

p sc ( psia )

N ℜ=
710.39
[ ] q (Mscfd )γ g
T sc ( ° R ) sc (7-8)
d (¿ .)μ (cp)

The Factor 710.39psc/Tsc for some common standard condition is as follows:

psc, psia Tsc, °R 710.39psc/Tsc

14.4 520 (60°F) 19.672
14.65 520 (60°F) 20.014
14.73 520 (60°F) 20.123
15.025 520 (60°F) 20.526

Thus, for most practical applications, the Reynolds number for a gas is given by:

20 q sc γ g
N ℜ= (7-9)
μd

Where qsc is in Mscfd, µ is in cp, and d is in inches.

As shown in the Moody friction factor chart (Figure 7-1), flow regime is related to Reynolds number
as follows:

Flow Type Nℜ
Laminar (or viscous) <2000
Critical 2000-3000
Transition (or intermediate) 3000-4000 (or 10000)
Turbulent >4000 (or 10000)

Pipe Roughness

Friction to flow through a pipe is affected by pipe-wall roughness. However, pipe roughness is not
easily or directly measurable, and absolute pipe roughness e is, therefore, defined as the mean
height of protrusions in uniformly sized, tightly packed sand grains that give the same pressure
gradient as the given pipe. This roughness may change with pipe use and exposure to fluids. Initially,
the pipe contains mill scale that may be removed by fluids flowing inside the pipe. The fluids may
also increase roughness by erosion or corrosion, or by precipitating materials that stick to the pipe
wall. Thus, estimating pipe roughness is quite difficult. Usually, absolute roughness is determined by
comparing the observed friction £4ìctor to that given in Moody’s chart (Figure 7-1). If no roughness
data arc available, a value of e = 0.0006 in. can be used. Some typical values for roughness are
shown be low (Chemical Engineers’ Handbook, 1984):
 Type of pipe ε , in.
Drawn tubing (brass, lead, glass) 0.00006
Aluminum pipe 0.0002
Plastic-lined or sand blasted 0.0002—0.0003
Commercial steel or wrought iron 0.0018
Asphalted cast iron 0.0048
Galvanized iron 0.006
Cast iron 0.0102
Cement-lined 0.012-0.12

Commonly used well tubing and line pipe:

New pipe 0.0005—00007
12-months old 0.00150
24-months old 0.00175

From dimensional analysis, It has been deduced that relative roughness, the ratio of the absolute
roughness and inside pipe diameter, e/d, rather than absolute roughness, affects flow through pipes.

Friction Factor

For convenience, friction factor f’, defined as the ratio of the shear stress at the fluid-solid Interface
and the kinetic energy of the fluid per unit volume, Is used in computing the magnitude of the
pressure drop due to friction. For steady-state flow in a uniform circular conduit such as a pipe, this
results in the well known Fanning equation:
2 f ' L ρ v2
∆ pf= (7-10)
gc d

where d is the inside pipe diameter.

Any consistent set of units can be used In the previous equations. In customary units, Δpf is in psi, L is
in ft, ρ is in lbm/ft3, v is in ft/sec, d is in ft, and g is therefore equal to 32.17 lbm-ft/lbf-sec 2.

Friction factor f’ is called the Fanning friction factor. Usually, the Moody (also called Blasius, or
Darcy-Weisbach) friction factor, equal to 4f’, is used. In terms of the Moody friction factor, f, the
Fanning equation becomes:

fL ρ v 2
∆ pf= (7-11)
2 gc d

The friction factor includes, besides roughness, the flow characteristics of the flow regime. It is
therefore a function of Reynolds number and relative roughness:

f = f(NRe, ε/d)
Laminar Single-Phase Flow

The pressure drop for laminar flow Is given by the analytic Hagen Poiseuille relationship as follows:
μvL
f =32 (7-12)
gc d2

Equating Equations 7-11 and 7-12:

64 μ 64
f= = (13)
dvρ N ℜ

Thus, the friction factor is independent of pipe roughness in the laminar flow regime.

Partially Turbulent and Fully Turbulent Single-Phase Flow

For partially-turbulent flow, friction factor is a function of both Reynolds number and pipe
roughness. For fully turbulent flow, however, the friction factor is only very slightly dependent upon
Reynolds number (see Figure 7-1).

Generally, intermediate or partially turbulent flow is included in turbulent flow for purposes of
developing correlations. Several correlations have been reported for the dependence of friction
factors on Reynolds number and pipe wall roughness; only some of the most accurate ones are
presented here.

For smooth pipes, the following relationships are applicable:

f =0.5676 N −0.3192
ℜ for intermediate flow (7-14)

N ℜ f 0.5
f =16 lo g (
0.7063 )
for partially turbulent flow (7-15)

N ℜ f 0.5
f −0.5=2 log ( 0.628 )
for fully turbulent flow (Prandtl’s Formula) (7-16)

f =0.3164 N ℜ−0.25 for NRe upto 105 (Blasius formula) (7-17)

f =0.0056+ 0.5 N ℜ−0.32 for 3000<NRe<3 x 108 (Drew, Koo and McAdams) (7-18)

For rough pipes, Colebrook’s equation is the basis for most modern frictional factor charts:

ϵ 0.628
f −0.5=−2 log
[ +
3.7 d N ℜ f 0.5 ] (7-19)

For very rough pipes, Colebrook’s equation reduces to:

f −0.5=−2 log ( 3.73 d ) (7-20)

Whereas for smooth pipes, Colebrook’s equations reduces to Prandtl’s Equations (Equations 7-16).

Swamee and Jain (1976) have presented an explicit correlation for frictional factors as follows:

ϵ 21.25
f −0.5=1.14−2 log
[ +
d N ℜ0.9 ] (7-21)

Equation 7-21 is applicable for 10-6 < ε/d < 10-2, and 5,000<NRe < 108, with errors within 1 % when
compared with Colebrook’s equation.

In practice, solutions to the equations presented for friction factors are cumbersome, so a composite
friction factor chart is used as shown inn Figure 7-1

Allowable Working Pressures for Pipes

It is desirable to operate a pipe at a high pressure in order to achieve higher throughputs. This is,
however, limited by the maximum stress the pipe can handle. The maximum allowable Internal
working pressure can be determined using the following ANSI (1976) specIfication:
2 ( t−c ) S E
Pmax = (22)
d o−2 ( t−c ) Y

Where

Pmax = maximum allowable internal pressure, psig

t = pipe thickness, in.
c = sum of mechanical allowances (thread and groove depth), corrosion, erosion, etc., in.
S = allowable stress (minimum yield strength) for the pipe material, psi

E = longitudinal weld joint factor (i.e., the anomaly due to weld seam), equal to 1.0 for
seamless pIpes, 0.8 for fusion-welded and spiral-welded, and 0.6 for butt-welded pipes
do= outside diameter of the pipe, in.
Y = temperature derating factor, equal to 0.4 up to 900°F, 0.5 for 950°F, and 0.7 for 1,000°F
and greater

Allowable Flow Velocity in Pipes

High flow velocities in pipes can cause pipe erosion problems, especially for gases that may have a
flow velocity exceeding 70 ft/sec. The velocity at which erosion begins to occur is dependent upon
the presence of solid particles, their shape, etc., and ¡s, therefore, difficult to determine precisely.
The following equation can be used as a simple approach to this problem (Beggs, 1984):

C
v e= (23)
ρ0.5

where
 v = erosional velocity, ft/sec
p = fluid density, lbm/ft3
C = a constant ranging between 75 and 150

In most cases, C is taken to be 100 (Beggc, 1984). Substituting for C and the gas density (p =
pM/ZRT), EquatIon 7-23 can be written as:

100 100 (ZRT )0.5

v e= =
pM 0.5 ( 28.97 p γ g ) 0.5
( )
ZRT

where ϒg is the gas gravity (air =1 basis).

The gas flow rate at standard conditions for erosion to occur, (q e)sc, can be obtained as follows:

p T sc π d2
( q e )sc =v e ( Z Tpsc )( ) 4

100( pR)0.5 T sc π d2
¿
( (28.97 γ g ZT )0.5 )( )( )
P sc 4

where

(qe)sc = gas flow rate for onset of erosion, scf/sec

d = diameter of the pipe, ft
p = flowing pressure, psia
T = flowing temperature, °R
R = gas constant ( = 10.732 psia-ft3/lbmole-°R)
Z = gas compressibility factor at pressure p and temperature T

Horizontal Flow
Many pipeline equations have been developed from the basic mechanical energy balance (Equation
7-3):

dp + (p/2gc)dv2 + (g/gc)ρdz + ρdIw = 0

Assuming horizontal, steady-state, adiabatic, Isothermal flow of gas, with negligible kinetic-energy
change, Equation 7-3 becomes:

dp+ρdlw=0

Frictional losses for a length dL of pipe are given by (Equation7-11):

fρ v 2
ρd l w = ( )
2 gc d
dL
Substituting for frictional losses:

fρ v 2
dp+ ( )2 gc d
dL=0

Substituting for gas density ρ:

pM
ρ=
ZRT

And gas velocity v:

ZT p sc 4
v=q sc ( pT sc )( )
π d2

We obtain:

f pM 16 q2sc Z 2 T 2 p2sc
−dp= ( 2 gc d )( )(
ZRT p2 T 2sc π 2 d 4 ) dL

Or,

p 8 fMT P2sc q 2sc

−∫ dp= ∫ dL (7-25)
Z R π 2 d 5 g c T sc2

Note that T is constant (or, independent of length) since isothermal flow is assumed. Otherwise, and
average temperature, Tav, is commonly used instead of T in the previous relationship. The two types
of averages used are the arithmetic average,

T 1+ T 2
T av = (7-26)
2

And the log-mean temperature,

T 1−T 2
T av =
T1 (7-27)
ln
( )
T2

In practise, both the averages are quite close since temperatures T 1 and T2 are used as absolute
temperatures. Using an average temperature is practically expedient, because analytic descripting of
the variation of temperature along the pipeline length is rather difficult and introduces some
complexity. The gas compressibility factor, Z, is made independent of temperature and pressure by
using an average compressibility factor, Z av, for simplicity. Integrating over the pipe length from 0 to
L and pressure p1 (at L=0, at the upstream) to p2 (at L=L, at the downstream end), Equations 7-25
becomes:
−p 21− p22 ( 8 )( 28.97 ) p2sc q2sc γ g Z av TfL
2
=
(
R π 2 g c T 2sc )( d5 )
Or

R g c T sc2 ( p 12− p 22 ) d 5
q sc2 =
( 46.9644 p sc2 )( γ g Z av TfL ) (7-28)

Any consistent set of units can be used in Equation 7-28. In common units, with q sc in Mscfd, p in
psia, T in °R, d in in., Lin ft, and with R = 10.372 psia ft 3/lb mol-°R and gc=32.17lbm ft/lbf-sec2,
Equation 7-28 becomes:

5
1
103
2

( ( 10.372 ×144 )( 32.17 ) ( ) 12

)( T 2sc ( p 12 − p 22 ) d 5
[ 3600 ×24 ]
q 2sc =
46.9644 P2sc )( γ g Z av TfL )
Thus,
0.5
T ( p12 −p 22 ) d 5
q sc =5.6353821 sc
p sc ( )( γ g Z av TfL ) (7-29)

Where

qsc = gas flow rate measured at standard conditions, Mscfd

psc = pressure at standard conditions, psia

Tsc = temperature at standard conditions, °R

P1 = upstream pressure, psia

P2 = downstream pressure, psia

d = diameter of the pipe, ft

ϒg = gas gravity, (air =1 basis)

T = flowing temperature, °R

Zav= average gas compressibility factor

f = Moody friction factor

L = Length of the pipe, ft

Equations 7-29, attributed to Weymouth, is the general equation for steady state isothermal flow of
gas through a horizontal pipe. Implications of the various assumptions made in the development of
Equation 7-29 are as follows:
 1. No mechanical work: It Is assumed that no work is done on the gas between the points at
which the pressures are measured. This condition can be satisfied easily by putting pressure
measurement stations such that no mechanical energy Is added (by compressors or pumps)
between these two points.
2. Steady-state flow: Rarely, if ever, encountered in practice, this assumption Is the major
cause of discrepancies In pipeline calculations. Reasons for unsteady behavior Include:
pressure/flow rate pulsations or surges, liquids in the pipeline, variations in operating
conditions, variations in withdrawal or supply rates, etc.
3. Isothermal flow: This assumption Is usually met because buried pipelines are used which are
not affected much by atmospheric temperature variations. Heat of compression is also
dissipated rapidly, usually within a few miles downstream of the compressor station. For
small temperature changes, the average temperature given by Equation 7-26 or 7-27 is
generally satisfactory.
4. Negligible kinetic energy change: This assumption is justified because kinetic energy changes
are negligible, compared to changes In pressure, for very long pipelines, such as commercial
transmission lines.
5. Constant (average) gas compressibility factor: This is a reasonable approximation, especially
If Z, is computed at the average pressure given by Equation 7-32.
6. Horizontal pipeline: In practice, flow is never truly horizontal. Equations developed to
account for elevation changes will be discussed in the section on inclined flow.

Average Pressure ¡n a Gas Pipeline

For an incompressible fluid, the average pressure Is simply the arithmetic average of the inlet (p 1)
and outlet (p2) pressures:
p 1+ p 2
Pav =
2
For a gas, a compressible fluid, this is not true. Consider a pipeline AB shown In Figure 7-2. Using
Equation 7-29 for gas flow, one can derive a simple formula to determine the pressure at any point C
along the pipeline at a fractional distance x from the inlet end. For a flow rate q sc. In the pipe, the
following two relationships are true (from Equation 7-29):
0.5
T ( p 12− p 22 ) d 5
q sc =0.0775543 sc
psc( )( γ g Z av TfL ) (7-30)
And
0.5
T ( p 12− p 22 ) d 5
q sc =5.6353821 sc
( )(
p sc γ g Z av TfL (1−x ) ) (7-31)

Equating these Equations 7-30 and 7-31 (wIth the assumption that the difference in Z av, for the two
pipe sections is negligible):
p 21− p2x p 2x − p22
=
x (1−x)

Solving for px:

0.5
p x = p21−x ( p21− p22 )
[ ]
This suggests a pressure profile as shown In Figure 7-3. The mean pressure is given by:
1 1
0.5
pav =∫ p x dx=∫ p 21−x ( p 21− p22 )
[ ] dx
0 0

2
2 p2
(
¿ p 1+
3 p1 + p2 )
Rearranging and multiplying both numerator and denominator by (p 1-p2):
 2 2
2 p1 + p1 p 2 + p 2 p1 −p 2
pav =
3 ( p 1+ p2 )( p1 −p 2 )
2 p 13− p 23
Pav = (
3 p 12− p 22 ) (7-32)

Where

pav = average pressure, psia

p1 = upstream pressure, psia

p2 = downstream pressure, psia

Therefore Zav must be computed using Equation 7-32 for average pressure.

Non-Iterative Equations for Horizontal Gas Flow

Several equations for gas flow have been derived from Equation 7-29 assuming a friction factor
relationship to avoid a trial and error computation. These equations differ only in the friction factor
relationship assumed. As we have seen earlier, friction factors vary over a wide range with Reynolds
number and pipe roughness. Thus, none of these equations Is universally applicable. In gas field
operations, engineers use the equation that best suits their conditions, or use their own modified
versions.

Weymouth Equation

Weymouth proposed the following relationship for friction factor as a function of pipe diameter d in
Inches:

f =0.032/d 1 /3 (7-33)

Substituting for f from Equation 7-33 into Equation 7-29:

0.5
T ( p 12− p 22 ) d 16/3
q sc =31.5027 sc ( )(
p sc γ g Z av T av L ) (7-34)

This is known as the Weymouth equation for horizontal flow. It Is used most often for designing gas
transmission systems because It generally maximizes pipe diameter requirements for a given flow
rate and pressure drop.

Panhandle (Panhandle A) Equation

This equation assumes that f is a function of Reynolds number only as follows:
0.0768
f= (7-35)
N ℜ0.1461
Substituting for f from Equation 7-35 into Equation 7-29:
1.07881 0.53940
p 12−p 22 0.46060
T sc 1 d 2.61821
q sc =32.6491 ( ) (
psc Z av T av L ) ( ) γg μ g0.07881
(7-36)

The Panhandle A equation is most applicable to large diameter pipelines, at high flow rates.

Modified Pan Handle (Panhandle B) Equation

One of the most widely used equations for long transmission lines, the Panhandle B equation
assumes that f is a function of Reynolds number as follows:
f =0.00359 /N ℜ0.03922 (7-37)

The pipeline flow equation is thus given as follows:

1.020 0.510
p 12 − p 22 0.490
T sc 1 d 2.530
q sc =109.364
p sc ( ) ( Z av T av L ) ( ) γg μ g0.020
(7-38)

The Panhandle B equation Is most applicable to large diameter pipelines, at high values of Reynolds
number.

A More Precise Equation for Horizontal Gas Flow: The Clinedinst Equation

The Clinedinst equation rigorously accounts for the deviation of natural gas from ideal behavior (an
average gas compressibility factor, Zav, is not used in this method), and the dependence of friction
factor, f, on Reynolds number and pipe roughness, leading to a trial and error solution scheme. The
integral ʃ(p/Z) dp must be evaluated. In order to generalize this evaluation, it is converted into the
following:

p pc pr p
∫ Zp dp=∫ [( )
Z ]
p pc d pr = p2pc ∫ r d p r
Z
where

ppc = pseudo critical pressure of the flowing gas mixture, psia

pr = reduced pressure (pr = p/ppc)

Values for the integral function ʃ(pr/Z) dp are given in Thble 7-1. For this case, Equation 7-25
becomes:

pr 1 pr 2
pr p 8 fMT p 2sc q 2sc L
p 2
pc
[∫ (
0 Z )d p r−∫ r d p r =
0 Z ( ) ]
R π 2 d 5 gc T 2sc

Rearranging this equation, and using common units, with q In Mscfd, p in psia, T in °R, d in In., L in ft,
and substituting for R ( = 10.732 psia-ft3/lbmole-°R) and g ( = 32.17 lbm-ft/lbf-sec 2):

0.5 pr 1 p r2 0.5
p T
q sc =7.969634 pc sc
psc [ d5
γ g T av Lf ][ ] [∫ 0
pr
Z
p
dpr−∫ r dpr
0 Z
] (7-39)
This is known as the Clinedinst equation for horizontal flow