Sie sind auf Seite 1von 8

Journal of Loss Prevention in the Process Industries 18 (2005) 55–62

www.elsevier.com/locate/jlp

Estimating sonic gas flow rates in pipelines


Jason M. Keith, Daniel A. Crowl*
Department of Chemical Engineering, Michigan Technological University, Houghton, MI 49931, USA
Received 28 July 2004; received in revised form 8 December 2004; accepted 20 December 2004

Abstract

This paper presents a detailed review and analysis of sonic gas flow in pipelines, adding considerably more detail to the analysis, particularly
for long pipelines. Our results show that (1) the mass flow rate is asymptotic as the velocity head pipe loss increases (2) the asymptotic value is
identical for both adiabatic and isothermal conditions and (3) a maximum is found in the gas flow, although this maximum is near the asymptotic
value. A graphical method for isothermal flows and a simple, shortcut formula is presented using asymptotic analysis which accurately
estimates gas flow rates in long pipelines under both adiabatic and isothermal conditions. Process safety applications are provided.
q 2005 Elsevier Ltd. All rights reserved.

Keywords: Sonic gas flow; Pipelines

1. Compressible flow theory of the pressure head and kinetic energy of the flowing gas,
both acting over the cross-section of the fluid element:
There have been many studies of the flow of compres-
sible gases in pipelines in textbooks (Churchill, 1980; Crowl pD2 pD2
tpDdx Z K dP K ur du (1)
& Louvar, 2002; Levenspeil, 1984), in articles (Cochran, 4 4
1996; Farina, 1997; Kumar, 2002; Levenspeil, 1977), and In industrial practice, long piping networks may not be
technical documents (Crane Co, 1986). Two limiting cases, horizontal. However, the contribution of gravitational poten-
adiabatic and isothermal, are often considered. Adiabatic tial to the mechanical energy balance is generally negligible.
flow conditions assume flow through an insulated pipe. The following steps are applied to Eq. (1):
These conditions are usually valid for short pipelines since
there is little heat transfer to or from the gas. Isothermal flow 1. Rewrite Eq. (1) in terms of the mass velocity (which is
conditions assume flow through a pipe held at a uniform also called the mass flux): GZur.
temperature; these conditions are commonly assumed when 2. Introduce the Fanning friction factor, f, from the relation-
studying the flow of a gas in an uninsulated pipeline. Most ship tZfru2/2ZfG2/2r.This results in the following:
natural gas pipelines are considered isothermal. f rD D dr
dx Z K 2 dP C (2)
For isothermal flow, the literature (Crane Co, 1986; 2 4G 4 r
Dodge & Thompson, 1937) states that flow under isothermal
3. Integrate Eq. (2) over the length of the pipe.
conditions is applicable to most industrial conditions, the
4. Introduce the sonic velocity, a, and the Mach number,
ideal example being a lengthy pipeline submerged under MaZu/a.
water. The analysis begins with a mechanical energy 5. Apply the ideal gas law.
balance about a fluid element of length dx within a 6. Apply mass conservation, which states that the mass
horizontal pipe of diameter D. The shear stress at the velocity, G, denoted by the product of the density and
exterior surface of the fluid element is balanced by the sum velocity, ru, must be a constant at steady-state. Thus,
r1u1Zr2u2. The exiting density can now be calculated as
r2Zr1(u1/u2) which can be written as r2Zr1(Ma1/Ma2).
* Corresponding author. Tel.: 906 487 3221; fax: 906 487 3213.
E-mail address: crowl@mtu.edu (D.A. Crowl). Similarly, the exit pressure is given by P2ZP1(Ma1/Ma2).

0950-4230/$ - see front matter q 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jlp.2004.12.004
56 J.M. Keith, D.A. Crowl / Journal of Loss Prevention in the Process Industries 18 (2005) 55–62

Nomenclature
a Speed of sound Ma Mach number
A Area P Pressure
CP Heat capacity R Ideal gas constant
D Diameter T Temperature
F Fanning friction factor u Velocity
gc Gravitational constant x Distance
G Mass flux Y Expansion factor defined by Eq. (28)
h Enthalpy
Symbols
K Velocity head losses
3 Pipe roughness
L Length
g Heat capacity ratio
m_ Mass flow
r Density
M Molecular weight
t shear stress

The following equation results after completion of steps 2. The gas temperature is determined by the conservation of
1 through 6 above: thermal energy:
   
4fL 1 1 1 Ma1 udu Z Kdh Z KCp dT
Z K C 2 ln (3)
D g Ma21 Ma22 Ma2

Under choked flow conditions, the speed of the exiting 3. The sonic velocity is introduced in the form:
gas reaches a maximum (as shown by Churchill, 1980), such
g K1
that the exiting Mach number Ma22 Z 1=g. This gives: Yi Z 1 C Ma2i
    2
4fL 1 1
C ln C 1K Z0 (4) Differentiation and integration of the resulting equation,
D gMa21 gMa21
with considerable algebraic manipulation, results in the
such that the exit velocity is u2Za/g1/2, the exit pressure is following equation for the adiabatic case:
P2ZP1Ma1g1/2 and the exit density is r2Zr1Ma1g1/2. A  2   
g C1 Ma2 Y1 1 1 4fL
detailed derivation of Eq. (4) is provided in Keith and Crowl ln K K Cg Z0 (6)
2 Ma21 Y2 Ma21 Ma22 D
(2004).
The mass velocity can also be written in terms of the Under choked flow conditions, the maximum velocity of
Mach number at the inlet and exit conditions: the exit gas is at the speed of sound (as shown by Churchill,
rffiffiffiffiffiffiffiffiffiffiffiffi 1980), such that the Mach number is unity. This simplifies
ggc M (6) to the form:
G Z r1 u1 Z r1 Ma1 a Z Ma1 P1 Z r2 u2
RT    
g C1 2Y1 1 4fL
rffiffiffiffiffiffiffiffiffiffiffiffi ln K K1 Cg Z0 (7)
ggc M 2 ðg C 1ÞMa21 Ma21 D
Z r2 Ma2 a Z Ma2 P2 (5)
RT such that the temperature ratio T2/T1Z2Y1/(gC1), the
A more complicated situation exists under adiabatic flow pressure ratio P2/P1ZMa1(2Y1/(gC1))1/2, and the density
conditions because the temperature changes as the gas flows ratio r2/r1ZMa1((gC1)/2Y1)1/2. A detailed derivation of
towards the tube exit. The temperature may either increase Eq. (7) is provided in Keith and Crowl (2004).
or decrease depending upon the relative effects of gas The assumptions that are inherent in the above analysis
expansion and friction. for both the adiabatic and isothermal flow cases are:
The following steps are applied to Eq. (1) for the
1. Either isothermal or adiabatic conditions along the entire
adiabatic case:
pipe length.
1. The gas velocities at the inlet and exit are introduced: 2. The fluid is an ideal gas
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 3. Potential energy contributions, Dz, are negligible
gc RT1 g compared to kinetic energy and pressure
u1 Z Ma1 a Z Ma1 and
M head/compressibility
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 4. Constant compressibility ratio, g, along the entire pipe
gc RT2 g length
u2 Z Ma2 a Z Ma2
M 5. Constant friction factor, f, along the entire pipe length.
J.M. Keith, D.A. Crowl / Journal of Loss Prevention in the Process Industries 18 (2005) 55–62 57

The validity of each of these assumptions will be 1


discussed in detail later.

0.8

Pressure Drop Ratio (P1 - P2 )/P1


2. Solution techniques for choked flow conditions
γ = 1.67
0.6
In this section, solution procedures for estimating
mass flow rates under choked conditions will be outlined. γ = 1.4
These include trial-and-error methods, graphical methods,
0.4
and a new shortcut method formulated using asymptotic
γ = 1.2
theory.
0.2
2.1. Trial-and-error methods

For problems involving the estimation of choked 0


isothermal flow rate conditions, the feed temperature T, 0.01 0.10 1 10 100 1000
feed pressure P1, pipe length L, and inside diameter D are Velocity Head Loss, Σ K
known. In order to computer the mass velocity, one needs to
Fig. 1. Pressure drop ratio as a function of the excess velocity head pipe
first determine the value of the Fanning friction factor f.
losses for adiabatic flow.
Generally, it is safe to assume fully developed turbulent
flow at high Reynolds numbers when considering choked a weak function of the heat capacity ratio, g, but this
flow. Then, the friction factor is given by:
difference is too small to show on Fig. 2.
 
1 D A similar graphical procedure can be developed for
pffiffiffi Z 4 log 3:7 (8) isothermal conditions. The only differences to the procedure
f 3
is that Eq. (4) is used to determine Ma1, and now
where 3 is the pipe roughness. P2ZP1Ma1g1/2. Fig. 3 shows the resulting pressure ratio
For isothermal flows, Eq. (4) is solved first using a trial and Fig. 4 shows the gas expansion factor. As expected, the
and error procedure to determine the Mach number at the pressure ratio is independent of the heat capacity ratio g,
inlet. The mass velocity is then determined from Eq. (5). For which affects the temperature of the gas during adiabatic
adiabatic pipe flows, Eq. (7) must be solved iteratively. The
conditions.
mass velocity is then determined from Eq. (5) with TZT1.
The procedure for determining the mass velocity, given
the type and diameter of the pipe, the pipe length, the type of
2.2. Graphical methods-Lupo correlation (Crowl, 1999)
0.8
Crowl (1999) suggests using the Darcy formula to solve
for the mass velocity under adiabatic flow conditions: 0.7
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
m_ 2gc r1 ðP1 K P2 Þ 0.6
G Z ZY P (9)
A K
Expansion Factor,Y

P 0.5
where Y is a unitless gas expansion factor and K is the
sum of frictional losses, including pipe friction, entrance and 0.4
exit losses, and fittings, which areP written in terms of an
equivalent friction factor given by K Z 4fL=D. 0.3
For adiabatic flows, the gas expansion factor Y is
determined by equating Eqs. (9) and (5) with TZT1 to yield: 0.2
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
P
g K P1 0.1
Y Z Ma1 (10)
2 ðP1 K P2 Þ
0
0.01 0.10 1 10 100 1000
Figs. 1 and 2 are determined by repeated trial-and-error
P
or direct solutions of Eq. (7), given a value for K and for Velocity Head Loss, Σ K
g. Then the exit pressure is determined from P2Z Fig. 2. Gas expansion factor as a function of the excess velocity head pipe
P1Ma1(2Y1/(gC1))1/2 and the gas expansion factor, Y, is losses for adiabatic flow. The results are a weak function of heat capacity,
determined from Eq. (10). The gas expansion factor is but the difference is less than can be shown on this figure.
58 J.M. Keith, D.A. Crowl / Journal of Loss Prevention in the Process Industries 18 (2005) 55–62

1 Table 1
Correlations for the expansion factor Y and the sonic pressure drop ratio
P
(P1KP2)/P1 as a function of the pipe loss K for adiabatic flow conditions
Pressure Drop Ratio (P1 - P 2)/P1

0.80 Function value A B C D Range of


validity, K
Expansion factor Y 0.00129 K0.0216 0.116 K0.528 0.2–1000
0.60 Sonic pressure 0.943 0.00727 1.12 – 0.01–1000
drop ratio gZ1.2
Sonic pressure 0.965 0.00461 0.944 – 0.2–1000
0.40 drop ratio gZ1.4
Sonic pressure 0.989 0.00178 0.767 – 0.01–1000
drop ratio gZ1.67
0.20 The equation used to fit the functions are of the form ln YZA (ln K)3CB
(ln K)2CC (ln K)CD for the expansion factor and fðP1 K P2 Þ=P1 gK1Z
AC Bðln KÞ2 C C=K 0:5 for the pressure drop ratio.
0
0.01 0.10 1 10 100 1000
Velocity Head Loss, Σ K users a library of several thousand functional forms-all of
these equations are applied to find the best fit. Here, a ‘best
Fig. 3. Pressure drop ratio as a function of the excess velocity head pipe
losses for isothermal flow. The results are independent of the heat capacity fit’ is defined as the simplest equation with the fewest
ratio. number of parameters. For this work we used an arbitrary
criterion of 1% for the equation fit. That is, the resulting
gas, and the upstream and downstream pressures, is as equation must fit the computed trial and error values to
follows: within 1% over the range specified. The gas expansion
factor shown in Figs. 2 and 4 can be fit to a curve of the
1. Given P
the inlet conditions, determine the sum of friction following form: ln YZA (ln K)3CB (ln K)2CC (ln K)CD.
terms K The pressure drop ratio is best fit using an equation of the
2. Determine
P the pressure ratio (P1KP2)/P1 as a function of form: {(P1KP2)/P1}K1ZACB(ln K)2CC/K0.5. Applying
K from Figs. 1 or 3 this to the data in Figs. 1–4 yields the Lupo correlations,
3. P
Determine the gas expansion factor Y as a function of
which are summarized in Table 1 for adiabatic flow and in
K from Figs. 2 or 4
Table 2 for isothermal flow.
4. Calculate the mass velocity from Eq. (9)
P Inspection of Figs. 1 and 3 show that as the loss term
A program called TableCurve2D (TableCurve2D, 2004) K increases, the pressure drop ratio reaches an
was used to determine the best fit equation for both the gas asymptotic value of (P2KP1)/P1Z1. Also, Figs. 2 and 4
expansion factor and the pressure drop ratio. TableCurve show that the gas expansion factor Y reaches an asymptotic
value of about 0.7 for both the adiabatic and isothermal
cases. The asymptotic value is also evident from the graphs
0.80
of Crowl and Louvar (2002). It is also interesting to note that
0.70 under all conditions, our P analysis identified a maximum
value for Y as a function of K. Table 3 shows the location
0.60 of this maximum value. For adiabatic flow, the maximum is
Expansion Factor,Y

a function of the heat capacity ratio. For isothermal flow, the


0.50
maximum is independent of the heat capacity ratio, but
0.40
Table 2
Correlations for the expansion factor Y and thePsonic pressure drop ratio
0.30
(P1KP2)/P1 as a function of the pipe loss K for isothermal flow
conditions
0.20
Function value A B C D Range of
0.10 validity, K
Expansion factor 0.00130 K0.0216 0.111 K0.502 0.2–1000
0 Y
0.01 0.10 1 10 100 1000 Sonic pressure 0.911 0.0118 1.38 – 0.1–1000
Velocity Head Loss, Σ K drop ratio, all g

Fig. 4. Gas expansion factor as a function of the excess velocity head pipe The equation used to fit the functions are of the form ln YZA (ln K)3CB (ln
losses for isothermal flow. The results are independent of the heat capacity K)2CC (ln K)CD for the expansion factor and fðP1 K P2 Þ=P1 gK1 Z AC
ratio. Bðln KÞ2 C C=K 0:5 for the pressure drop ratio.
J.M. Keith, D.A. Crowl / Journal of Loss Prevention in the Process Industries 18 (2005) 55–62 59

Table 3 the kinetic energy terms. This is because the change in


Maximum values in the expansion factors for adiabatic and isothermal temperature within the pipe affects the fluid density (and
sonic flows
hence kinetic energy). Nevertheless, as the frictional pipe
Adiabatic flow: losses grow, Eq. (12) can be used to sufficiently describe its
P
Heat capacity ratio g Velocity Head loss K Expansion factor Y relationship with the inlet Mach number Ma1. Thus, for
1.0 56.3 0.7248 large pipelines this theory predicts that there is no
1.1 64.0 0.7277 fundamental difference between adiabatic and isothermal
1.2 72.5 0.7210 flows! The literature claims that ‘real gas flow’ lies in
1.3 81.3 0.7195 between those predicted by the isothermal and adiabatic
1.4 90.0 0.7182
1.5 101.1 0.7172 models (Churchill, 1980; Crowl & Louvar, 2002; Leven-
1.6 112.0 0.7162 speil, 1984).
1.7 123.5 0.7154 Nevertheless, one can insert Eq. (12) into the relationship
1.8 135.8 0.7147 between inlet and outlet pressure: P2ZP1Ma1g1/2 for
1.9 148.7 0.7141 isothermal conditions or P2ZP1Ma1(2Y1/(gC1))1/2 for
2.0 162.4 0.7135
Isothermal flow 56.3 0.7248
adiabatic
P conditions to show, for large pipelines when
(all values of g): K /N, P2Z0, such that the exiting gas has no gauge
pressure (and thus cannot speed up by expansion). Inserting
this relationship, along with Eq. (12), into Eq. (10) gives a
interestingly, it is the same as for the gZ1.0 case in simplified formula for the asymptotic value for the gas
adiabatic flow. In all cases, this maximum is within 3% of expansion:
the asymptotic value. We will now analytically obtain these
asymptotic limits and use them to develop a simple shortcut 1
Y Z pffiffiffi Z 0:7071 (14)
formula for predicting choked gas flows. 2
which can subsequently be inserted into Eq. (9) to give a
2.3. Shortcut method simple, analytical estimate for the gas flowrate under sonic
flow conditions, which does not require using cumbersome
The goal of the shortcut method is to develop a simple figures or trial-and-error solutions:
formula to predict the maximum flowrates through very sffiffiffiffiffiffiffiffiffiffiffiffiffiffi
long pipelines. Under isothermal conditions the inlet Mach g c r 1 P1
m_ Z A P (15)
number Ma1 is P determined from Eq. (4), where the frictional K
pipe losses K are balanced by kinetic energy,
½lnð1=gMa21 Þ, and compressibility ½1K 1=gMa21 : The asymptotic value of Y and the pressure drop ratio are
    evident in Figs. 1–4. Eq. (15) is equivalent to the asymptotic
X 1 1
K C ln C 1 K Z0 (11) formula obtained by Churchill (1980).
gMa21 gMa21 The validity of the five assumptions following Eq. (7)
will now be discussed in detail. The first concerns itself with
For very small inlet gas velocities (and thus small Ma1),
having adiabatic or isothermal conditions along the entire
the gas compressibility term grows more rapidly (scaling as
pipe length. However, the literature has suggested that ‘real’
MaK21 ) than the kinetic energy (scaling as K2 ln Ma1). From
pipe flow lies somewhere in between the adiabatic and
a physical standpoint, the gas expands as the pressure
isothermal limits (Churchill, 1980; Cochran, 1996; Crowl &
changes down the pipeline while the kinetic energy is
Louvar, 2002; Levenspeil, 1984). The results presented
relatively unimportant. In fact, as the frictional pipe losses
here, in particular the derivation of Eq. (15), shows that this
approach infinity, the inlet Mach number is expected to vary
is quantitatively true, especially for long pipelines. There-
as:
fore, maintaining an entire pipe under isothermal conditions
1 may not be necessary for the theory to hold.
Ma1 Z pffiffiffiffiffiffiffiffiffiffiffiffiffi
P (12)
g K The second assumption addresses whether or not the
pressure-volume relationship of the expanding gas can be
A similar argument can be made during adiabatic described by the ideal gas equation of state, and depends
conditions, where the inlet Mach number can be computed mainly on the upstream gauge pressure P1. In many
from Eq. (7) industrial situations, the upstream pressure is dictated by a
X    
g C1 2Y1 1 storage tank pressure or reactor operating pressure. On the
g KC ln C 1K 2 Z0 other hand, the downstream gauge pressure is equal to P1
2 ðg C 1ÞMa21 Ma1
minus the pressure drop DP with the pressurized discharge
(13)
entering the atmosphere. However, during choked flow
Comparing Eq. (11) with Eq. (13) shows that the only conditions the pressure drop in theP pipeline DP approaches
difference between the isothermal and adiabatic model are P1 for very long pipelines (where K [ 10, as seen from
60 J.M. Keith, D.A. Crowl / Journal of Loss Prevention in the Process Industries 18 (2005) 55–62

Figs. 1 and 3). Thus, the exit gauge pressure approaches Eq. (17) is used to determine the mass flow for the first
zero. Under these circumstances, the upstream pressures can case.
be high enough to suggest the possibility of non-ideal For the second case, the solution can be found directly
behavior. In fact, for very long pipelines simulation of from Eq. (16). Solving for D,
Eq. (2) suggests there is a small pressure drop over most of sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
the pipe. Near the pipe exit, the pressure drop is much higher 5 64m_ 2 fL
DZ (18)
as the gas is now expanding at a rapid rate. Therefore this is p 2 gc r1 P 1
the most restrictive assumption of the theory. Nevertheless,
ideal gas solutions are usually used as a back-of-the- If we further substitute the ideal gas law for the upstream
envelope tool for predicting the pressure-volume relation- density, then,
ship. The ideal gas law thus can be used here to provide a sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
5 64m _ 2 fLRT1
similar tool for predicting choked flow. DZ (19)
The assumption of constant friction factor will now be p gc MP21
2

discussed. Simulation of Eq. (2) shows that the local gas The solution to Eq. (19) is by trial and error since the
Mach number (and hence the velocity) at any position x friction factor, f, depends on the pipe diameter.
slowly increases until near the pipe exit, where it quickly
increases to a value of unity. Although the velocity is
changing along the pipe length, its rate of change is small
and a constant friction factor is probably a good assumption. 4. Some examples
The remaining two assumptions concerning horizontal
pipes and constant gas compressibility ratio will now be We now use these results to estimate mass flow rates
discussed. In many industrial situations, pipelines are not under sonic conditions for two examples. The first problem
horizontal; however, even for cases with significant is number 4–14 from the second edition of Crowl and
potential energy
P changes these losses can be lumped into Louvar’s (2002) textbook on chemical process safety.
an effective K value. The heat capacity ratio is a material
property and is relatively independent of pressure. There- Example 1. A laboratory apparatus uses nitrogen at 250
fore, the assumption of constant heat capacity ratio should psig. The nitrogen is supplied from a cylinder, through a
not affect the solution. regulator, and to the apparatus via 15 feet of 1/4-inch ID
drawn copper tubing. If the tubing separates from the
apparatus, estimate the flow of nitrogen from the tubing.
The nitrogen in the tank is at 75 8F. Assume that gZ1.4 and
3. Special cases a friction factor of fZ0.00355, which is estimated for fully
developed turbulent flow. The gas density at the supply
Two special cases of choked flow occur in practice which pressure is determined from the ideal gas law and is
can use the asymptotic solution of Eq. (15). In the first case, 1.2846 lbm/ft3.
the pipe diameter and length are known, and the mass flow is Solution: Adiabatic flow is assumed. The 1/4-in ID is
required. This is equivalent to the rupture of a pipeline at a equivalent to 2.08!10K2 ft, and the cross-sectional area of
specified location. In the second case, the mass flow is the tube can be calculated as 3.41!10K4Pft2. Thus the
specified over a fixed distance and the pipeline diameter friction losses can be calculated as K Z 4fL=DZ
required to transport the gas is required. In both cases it is ð4 !0:00355 !15ftÞ=ð2:08 !10K2 ftÞZ 10:24. A (lengthy!)
assumed that the frictional loss is due entirely to the pipe trial and error computation using Eq. (7) yields Ma1Z0.231.
friction due to the pipe length. Eq. (5) can be used to compute the mass flowrate:
For the first case, for a circular pipe AZpD2/4 and sffiffiffiffiffiffiffiffiffiffiffiffi
assuming that the frictional
P losses are due entirely to the ggc M
pipes length results in K Z 4fL=D. Substituting into _
mZAMa 1 P1
RT1
Eq. (15) results in
  
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2 lbf 144in2
p gc r1 P 1 D 5 Zð3:41!10 ft Þð0:231Þ!ð250C14:7Þ
K4

m_ Z (16) in2 ft2


8 fL sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ð1:4Þð32:17ft–lbm =lbf –s2 Þð28lbm =lbmolÞ
!
Furthermore, by the ideal gas law (r1ZP1M/RT1) it ð1545ft–lbf =lbmol–8RÞð75C460Þð8RÞ
follows that,
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi Z0:118lbm =s
p gc P21 MD5
m_ Z (17) An almost identical answer can be obtained using Eq. (9)
8 RT1 fL and the charts in Figs. 1 and 2 or the Lupo correlation in
J.M. Keith, D.A. Crowl / Journal of Loss Prevention in the Process Industries 18 (2005) 55–62 61

Table 1. For this case the pressure drop ratio (P1KP2)/P1Z Now guess DZ6.18 cm. Then fZ0.00457, and the
0.778 and YZ0.70. computed DZ6.32 cm
Alternatively, substituting into Eq. (15) yields the Now guess DZ6.32 cm. Then fZ0.00455 and the
asymptotic solution: computed DZ6.32 cm.
sffiffiffiffiffiffiffiffiffiffiffiffiffiffi rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
gc r1 P 1 K4 2 ð32:17 ft–lbm =lbf –s2 Þð1:2846 lbm =ft3 Þð250 C 14:7Þðlbf =in2 Þð144 in2 =ft2 Þ
m_ Z A P Z ð3:41 !10 ft Þ !
K 10:24
Z 0:134 lbm =s

which has a difference of 13% from the rigorous solution. The answer is DZ6.32 cm. For this problem 4fL/DZ
Clearly, the velocity head pipe loss of 10.24 is too small for 2880 so the asymptotic solution does apply.
the asymptotic solution to apply. Nevertheless, it provides a
good estimate of the mass flow, which is within reasonable
accuracy for process safety applications.
P
It is worth mentioning that if K Z 50, the full solution
of Eq. (5) for adiabatic flow gives a mass flow rate of 5. Discussion and conclusions
0.058 lbm/s while the asymptotic theory of Eq. (15) yields a
mass flow rate ofP0.061 lbm/s, an error of 3.9%. Further- This paper has presented significantly more detail on
more, increasing K to 100 and 500 give an error of 2.2 sonic gas flow than has been presented elsewhere,
and 0.2%, respectively. Thus, the asymptotic theory particularly for the case of a long pipeline. In this case,
becomes more accurate with increasing the pipe length. both the adiabatic and isothermal equations show maxi-
mum and asymptotic values in the gas expansion
Example 2. You wish to construct a pipeline to deliver
parameter. The asymptotic values for the adiabatic and
1 kg/s of nitrogen to a location 10 km away. The pipeline is
isothermal cases are the same. A shortcut formula is
constructed from commercial steel pipe. If the upstream
derived from the asymptotic values, enabling easy
pressure of the nitrogen is 50 barg, what is the required
calculation of the flow.
pipeline diameter? Assume a temperature of 25 8C, that
nitrogen is an ideal gas, and the gas exits at choked The key to the success of the shortcut formula lies mainly
conditions. in the Y parameter, and somewhat in the P pressure ratio.
Figs. 1–4 show that for pipelines with K Z 5 the gas
Solution: The asymptotic solution, provided in Eq. (19), expansion parameter, Y, is within 5% or less of the
applies for this very long pipeline. The procedure is as asymptotic value of 0.707. In addition, the pressure ratio
follows: is
Pwithin 30% or less of the asymptotic value of 1.0. For
K Z 10 the error in the gas expansion factor and pressure
1. Guess a pipe diameter, D.
ratio are 2 and 20%, P respectively. As most industrial
2. Use Eq. (8) for fully developed turbulent flow to
pipelines have large K values due to their length, the
determine f.
asymptotic equation is very good at predicting the
3. Compute a new diameter using Eq. (19).
maximum gas flow rate. The argument of isothermal vs.
4. Continue steps 1 through 3 until guessed D equals
adiabatic flow is also unimportant.
computed D.
The equations presented here are useful for conse-
The procedure converges rapidly if the computed D is quence modeling of releases of gases. A simple, direct
used as the next guess. solution is possible rather than a complicated trial and
For this case, P1Z50 bargZ51 baraZ5.1!106 kg/m s2. error procedure.
Guess DZ10 cmZ0.1 m. For commercial steel pipe,
3Z0.046 mm. From Eq. (8), fZ0.004098. Substituting into
Eq. (19)
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
5 64m _ 2 fLRT1
DZ
p gc MP21
2

sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
5 ð64Þð1 kg=sÞ ð0:004098Þð10; 000 mÞð8:3143 kPa–m =kg–mole–KÞ½ð1000 kg=m–sÞ=ð1 kPaÞð298 KÞ
2 3
DZ
ð3:14Þ2 ð1 kg–m=s2 =NÞð28 kg=kg–moleÞð5:1 !106 kg=m–s2 Þ2

D Z 0:0618 m Z 6:18 cm
62 J.M. Keith, D.A. Crowl / Journal of Loss Prevention in the Process Industries 18 (2005) 55–62

References Farina, I. H. (1997). Critical length helps calculate compressible flow.


Chemical Engineering, 104(2), 88.
Churchill, S. (1980). Inertial flows. Thornton, PA: Etaner Press pp. 53–58. Keith, J.M., Crowl, D.A., New methods for estimating sonic flows in
Cochran, T. W. (1996). Calculate the pipeline flow of compressible fluids. pipelines, AICHE Loss Prevention Symposium, New Orleans,
Chemical Engineering, 103(2), 115. 2004.
Crane Co. (1986). Flow of fluids through valves, fittings, and pipe. Kumar, S. (2002). Spreadsheet calculates critical flow. Chemical
Technical Paper No. 410, New York, NY Engineering, 109(10), 62.
Crowl, D.A. (1999). Guidelines for consequence Analysis of Chemical Levenspeil, O. (1977). The discharge of gases from a reservoir through a
Releases. NY: American Institute of Chemical Engineers, pp. 29–35. pipe. AIChE Journal, 23, 402.
Crowl, D. A., & Louvar, J. F. (2002). Chemical process safety: Fundamentals Levenspeil, O. (1984). Engineering flow and heat exchange. New York,
with applications. Upper Saddle River, NJ: Prentice Hall PTR pp. 136–151. NY: Plenum Press pp. 41–47.
Dodge, R. A., & Thompson, M. J. (1937). Fluid mechanics. New York, NY: TableCurve2D, Version 2.0, Systat Software, Inc., 501 Canal Blvd., Point
McGraw-Hill Book Company pp. 193, 288, and 407. Richmond, CA 94804-2028, www.systat.com (2004).

Das könnte Ihnen auch gefallen