Sie sind auf Seite 1von 8

chemical engineering research and design 8 7 ( 2 0 0 9 ) 1596–1603

Contents lists available at ScienceDirect

Chemical Engineering Research and Design

journal homepage: www.elsevier.com/locate/cherd

Sensitivity of pipeline gas flow model to the selection


of the equation of state

Maciej Chaczykowski ∗
Warsaw University of Technology, Faculty of Environmental Engineering, Heating and Gas Systems Department, Nowowiejska 20,
00-653 Warszawa, Poland

a b s t r a c t

Real gas effects exert a significant influence on the hydraulics of natural gas transmission pipelines. In this article
the implications of the selection of the equation of state for the pipeline gas flow model are investigated. A non-
isothermal transient gas flow model with AGA-8 and SGERG-88 equations of state was studied. Models with Soave-
Redlich-Kwong and Benedict-Webb-Rubin equations of state were solved to illustrate the overall gas flow model
inaccuracies. The effect of the selection of different equations of state on the flow parameters is demonstrated and
discussed.
© 2009 The Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.

Keywords: Natural gas pipeline; Transient non-isothermal flow; Equation of state; Compressibility factor

1. Introduction preclude pressure values above permissible limits. Extra line-


pack is necessary to provide some flexibility to accommodate
Pipeline simulations are widely used by gas transmission oper- variations in gas demand and a safety margin in emergency
ators that are obliged to ensure that the system is balanced situations.
and that deliveries of the gas are maintained. According to EU Seeing that safe and efficient gas transmission requires
regulatory framework, the responsibility for the physical bal- physical balancing of the system as a necessary condition to
ance of the system is imposed on the pipeline operator and ensure correct technical operation of the network, pipeline
the balancing should be carried out on a daily and monthly operators, and in particular their dispatching centres, control
basis. This decision has many ramifications in the field of gas transmission parameters such as flowrates and pressures
flow measurement; among other things is an increased impor- using real time gas network simulators based on transient flow
tance of the accuracy of pipeline simulation, which is used for models. Gas flow models with literature review of their solu-
the determination of system line-pack (gas network accumu- tion methods are widely discussed by Thorley and Tiley (1987)
lation) on an hourly and daily basis. Pipeline leak detection and references therein. Many specific contributions were also
system based on volume balance methods is another exam- discussed by Osiadacz (1996).
ple of where the accuracy of simulation results is an important Modelling of pipeline flow transients behaviour requires an
matter. application of the equation of state. Modisette (2000) provided
Physical balancing of the system can be considered as a review of equations of state commonly used in the gas indus-
a management of system line-pack. The minimum pipeline try for compressibility factor calculations. Some equations
line-pack is the amount of gas in the pipeline required to reviewed included the NX-19, AGA-8, Benedict-Webb-Rubin
achieve the desired gas flow, and the required delivery pres- (BWR) and Soave-Redlich-Kwong (SRK) equation of state.
sure. Physical balancing of the system can be achieved by There are two different applications which affect the choice
an adequate amount of storage capacities as well as through of the equation of state when modelling flow of natural gas
the variations in the system line-pack. The latter supports in pipelines. These are custody transfer measurements and
the hourly modulation of gas delivery and supply rates and pipeline simulation. Custody transfer has its financial aspects
is determined by means of pipeline simulation in order to and legal regulations that impose restrictions on the selec-


Tel.: +48 22 234 50 57; fax: +48 22 825 29 92.
E-mail address: maciej.chaczykowski@is.pw.edu.pl.
Received 15 December 2008; Received in revised form 15 May 2009; Accepted 14 June 2009
0263-8762/$ – see front matter © 2009 The Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.
doi:10.1016/j.cherd.2009.06.008
chemical engineering research and design 8 7 ( 2 0 0 9 ) 1596–1603 1597

hydrocarbon gas is equal to the heating value of the natural


Nomenclature 1
gas. The calculation of molar mass, molar heating value and
mole fraction of the equivalent hydrocarbon gas is performed
A cross-section area of the pipe (m2 )
iteratively. The convergence criterion is the absolute differ-
B second virial coefficient (m3 /kmol)
ence between the calculated density of the gas mixture with
C third virial coefficient (m6 /kmol2 )
the equivalent hydrocarbon gas and the density of the natural
cp specific heat at constant pressure (J/(kg K))
gas at standard conditions.
cv specific heat at constant volume (J/(kg K))
Non-isothermal gas flow models are already widely used
D pipe diameter (mm)
in pipeline simulations. In regard to compressibility factor, it
f Fanning friction coefficient
is treated either as a constant parameter or as a function of
g the net body force per unit mass (the accelera-
temperature and pressure in the flow model. However, we are
tion of gravity) (m/s2 )
not aware of any published work in the field of single—phase
h specific enthalpy (J/kg)
natural gas pipeline simulation, with more complex and accu-
k heat-transfer coefficient (W/(m K))
rate equations of state than two-parameter correlations in the
L pipeline length (m)
flow model.
m heat-transfer element mass (kg)
The objective in this paper is to focus on a transient gas
p gas pressure (Pa)
flow modelling with the application of various equations of
q rate of heat-transfer per unit time and unit
state including those frequently used in gas and petroleum
mass of the gas (W/kg)
industry. A sensitivity analysis of transient pipeline gas flow
Q volumetric flowrate (m3 /s)
model to the selection of the equation of state has been car-
R specific gas constant (J/(kg K))
ried out. Non-isothermal gas flow model comprises: (a) AGA-8,
t time (s)
(b) SGERG-88, (c) BWR and (d) SRK equations of state, and their
T gas temperature (K)
influence on flow parameters, especially on the gas tempera-
Tsoil soil temperature (K)
ture and pipeline line-pack is presented.
u specific internal energy (J/kg)
v specific volume (m3 /kg)
2. Transient gas flow model
w flow velocity (m/s)
x spatial coordinate (m)
z compressibility factor The unsteady one-dimensional compressible flow within a gas
pipeline is described by a set of partial differential equations
Greek symbols expressing mass, momentum and energy conservation laws
˛ angle between the direction x and the horizon- as follows
tal
∂ ∂(w)
ε pipe roughness (mm) + =0 (1)
∂t ∂x
 thermal conductivity (W/(m K))
 viscosity of natural gas (N s/m2 )
∂(w) ∂(p + w2 ) 2fw|w|
 density of the gas (kg/m3 ) + =− − g sin ˛ (2)
∂t ∂x D
ω acentric factor
     
∂ w2 ∂ w2
Subscripts u+  + h+ w = q − wg sin ˛ (3)
∂t 2 ∂x 2
s standard conditions
c critical
Eqs. (1)–(3) may be rewritten in terms of pressure and the volu-
m molar
metric flowrate under standard conditions instead of density
amb ambience
and velocity, respectively. This is a matter of convenience,
Superscripts since these quantities are commonly measured and used in
k iteration index the gas industry. As a result, equation of state which would
express the density in terms of pressure and temperature is
needed to close the system of the above equations
tion of the equation of state. Natural gas composition and the
p
range of pressure and temperature values occurring in gas = (4)
zRT
transmission pipelines enable the use of equations of state
of moderate complexity in gas industry. Currently AGA-8 and Eq. (1) can be rearranged in the form
SGERG-88 equations of state are widely used by American and
European pipeline operators. In engineering practice, the gross 1 d ∂w
+ =0 (5)
characterization method of gas composition is widely used in  dt ∂x
pipeline simulations since a complete compositional analy-
sis of natural gas often is not available. The method does not Rewriting Eq. (4) in logarithmic form and differentiating
consider the detailed gas composition, but considers hydro- with respect to time yields
carbons collectively as an equivalent hydrocarbon gas. The
molar heating value of the gas mixture with the equivalent 1 d 1 dp 1 dT 1 dz
= − − (6)
 dt p dt T dt z dt

1
Note: flowrate Qs is shown in the standard conditions of For a given gas composition, the compressibility factor
273.15 K, 0.1 MPa, Substantial derivatives are indicated as d/dt. can be expressed as a function of pressure and temperature
1598 chemical engineering research and design 8 7 ( 2 0 0 9 ) 1596–1603

z = z(p,T) thus Combining this with Eq. (2) gives the following derivation:
 ∂z   ∂z 
dz = dp + dT (7) dh dp 2fw3
∂p T ∂T p  − − = q (17)
dt dt D
Substituting Eq. (7) into Eq. (6) we obtain
By using the thermodynamic identities
1   dp     dT
1 d 1 ∂z 1 1 ∂z
= − − + (8) p   ∂p 
 dt p z ∂p dt T z ∂T dt
T p dh = du + d , du = cv dT + T − p dv
 ∂T v

The velocity in terms of pressure, temperature and volu-


metric flowrate at standard conditions the following form of energy equation is obtained

w=
s Qs zRT
(9)   ∂z  
pA ∂T s Qs zRT ∂T RT s Qs zRT 1 1
+ + zT +
∂t pA ∂x cv pA T z ∂T 
Differentiation of the logarithmic form of Eq. (9) with respect

1 ∂Q 1   ∂p     ∂T
to spatial coordinate yields s 1 ∂z 1 1 ∂z
× − − + +
Qs ∂x p z ∂p T ∂x T z ∂T p ∂x
1 ∂w
=
1 ∂Qs

1 ∂p
+
1 ∂T
+
1 ∂z
(10) 2f
 zRT |Q | 3 q
s s
w ∂x Qs ∂x p ∂x T ∂x z ∂x − − =0 (18)
cv D Ap cv
The last term on the right-hand side of Eq. (10) can be substi-
tuted using Eq. (7). Therefore The first and the second term of Eq. (18) represent the time
 ∂z  ∂p  ∂z  ∂T rate of change of the temperature of the gas as it flows along
1 ∂w 1 ∂Qs 1 ∂p 1 ∂T 1 1 the pipeline. The third term represents the real gas effects
= − + + +
w ∂x Qs ∂x p ∂x T ∂x z ∂p T ∂x z ∂T p ∂x resulting from the enthalpy dependence on pressure and tem-
(11) perature, and in the range of pressure and temperature values
representative for natural gas transmission pipelines it has a
Substituting Eqs. (8) and (11) into Eq. (5) gives the following positive value, causing the expanding gas to cool. The fourth
form of continuity equation expressed by quantities which are term represents heating of the gas from friction. The last term
directly measured in Eq. (18) represents the heat-transfer from the gas to the

1  ∂z  ∂p   ∂z   ∂T pipeline surroundings and has a significant effect on the gas


1 1 1 s zRT ∂Qs parameters obtained from the solution of the above model.
− − + + =0
p z ∂p T ∂t T z ∂T p ∂t pA ∂x Heat-transfer in gas pipelines was studied by Gersten et al.
(12) (2001). Based on a steady-state non-isothermal gas flow model
in both onshore and offshore pipelines, they showed that
considering heat-transfer reduces uncertainties in planned
Eq. (2) can be rewritten in the form transport capacities and pressure losses. Paper (Osiadacz and
Chaczykowski, 2001a) presents comparison of isothermal and
dw 1 ∂p 2fw|w| non-isothermal pipeline gas flow models. Non-isothermal
+ + + g sin ˛ = 0 (13)
dt  ∂x D model contained simplified form of energy equation with
enthalpy and internal energy calculated from ideal gas equa-
Differentiating the logarithmic form of Eq. (9) with respect
tions, and the steady-state heat-transfer term for calculation
to time we obtain
of the heat-transfer from the gas to the surrounding soil. Nev-
1 dw 1 dQs 1 d ertheless, it has been shown that there exists a significant
= − (14) difference in the pressure profile along the pipeline between
w dt Qs dt  dt
isothermal and non-isothermal flow processes. The use of an
Substituting Eq. (8) into Eq. (14) and combining this with Eq. isothermal model may lead to significant errors in calculation
(13) results in of the energy consumption of the drivers of the compressors.
The work of Modisette (2002) concluded that the accuracy of
∂Qs s Qs zRT ∂Qs 1 1
  ∂z   ∂p s Qs zRT ∂p

+ − Qs − × + the heat-transfer model affects both line-pack and pressure
∂t pA ∂x p z ∂p T ∂t pA ∂x loss in gas pipelines. Recently, non-isothermal transient flow
     ∂T  of natural gas in a pipeline was studied by Abbaspour and
1 1 ∂z s Qs zRT ∂T
+ Qs + + Chapman (2008). They showed that the effect of cooling of the
T z ∂T p ∂t pA ∂x
gas due to expansion is significant on the temperature distri-
A ∂p 2fzRTs Qs |Qs | bution, and the effect of treating the gas in a non-isothermal
+ + =0 (15)
s ∂x DAp manner is very necessary for pipeline flow calculation accu-
racies, especially for rapid transient process.
Some terms in Eq. (3) can be cancelled out since the mass Various methods are used for estimation of the heat-
and the momentum are also conserved. Eq. (3) can be con- transfer term in the energy equation, most of which assume
verted using Eq. (1) to the following form: modification of steady-state heat flow expression. Generally,
models describing the heat-transfer to the surroundings in gas
dh dw ∂p pipelines are one-dimensional, due to the lack of accurate data
 − w − = q − wg sin ˛ (16) describing ground properties.
dt dt ∂t
chemical engineering research and design 8 7 ( 2 0 0 9 ) 1596–1603 1599

3. Heat-transfer model section of a pipeline, Ti is the element temperature and ki is


the heat-transfer coefficient between elements (i − 1) and i (k0
In the energy equation, the heat-transfer term q represents denotes heat-transfer coefficient between the gas and the first
the amount of heat exchanged between unit mass of gas and element). In case of one-dimensional approach, the process of
the surroundings per unit time. Application of Fourier’s law heat-transfer may be modelled by a minimum two cylindri-
to calculate the overall heat-transfer between the gas and the cal layers as heat capacitors. Assuming substantially different
ground, for a discretization section of a pipeline, yields heat capacity of the layers, so that their time constants were
different, the near and the remote surrounding of the pipeline
qAdx = −k(T − Tamb )dx
would therefore respond to temperature changes quickly and
where k is an overall heat-transfer coefficient and Tamb is the slowly, respectively. It has been assumed for the purpose of
ambient temperature at the same horizontal level as pipe axis, heat-transfer area discretization in this study, that every ele-
but at a sufficient lateral distance from the pipe. Therefore ment has the same thermal resistivity. Thus, the temperature
differences between consecutive ground sections (element
k surfaces) are equal in steady state, and the initial condition
q = − (T − Tamb ) (19)
A can be accurately modelled. The technique for heat-transfer
modelling presented above and its applicability to calculate
There exists an analytical steady-state solution for k for a
flow parameters in the gas pipeline has been evaluated in the
cylinder near a half-plane, which corresponds to the geometry
case study presented in this work.
of a buried pipeline. Nevertheless, it is a common practice to
calculate k as for a concentric cylindrical layer, with the dis-
4. Equations of state
tance between the outer boundary and the pipe equal to the
burial depth of the pipe. The ambient temperature is fixed and
equal to the ground temperature at the same horizontal level According to AGA-8/1992 (Compressibility Factor of Natural
as the pipe axis, and at a sufficient lateral distance from the Gas and Related Hydrocarbon Gases, AGA Report No. 8, Amer-
pipe. ican Gas Association, Arlington, VA.) and ISO 12213-3:1997
In this work, however, the process of heat-transfer from (Natural gas—calculation of compression factor—Part 3. Cal-
the gas to the surrounding environment is described using culation using physical properties), the equation of state for
unsteady heat-transfer model, so that the description of heat the calculation of compressibility factor of natural gas is in
flux could take into consideration the effect of heat capacity the form of the virial expansion:
of the surroundings of a pipeline. Using the element method,
2
one-dimensional axial-symmetric heat exchange process can Z = 1 + Bm + Cm (21)
be expressed by the following set of equations, representing
thermal balances of the elements—coaxial cylindrical layers where m is the molar density of the gas, kmol/m3 .
(Fig. 1): It is convenient to rewrite Eq. (21) as a series in powers
of pressure instead of molar density, which would be some-
k0 what better form considering the dependent variables of the
q = − (T − T1 )
A system of Eqs. (12), (15) and (18). An equivalent form used for
m1 cp1 ∂T1
= k0 (T − T1 ) − k1 (T1 − T2 ) calculation of the derivatives of compressibility factor is
dx ∂t
m2 cp2 ∂T2
= k1 (T1 − T2 ) − k2 (T2 − T3 ) (20) Z = 1 + B p + C p2 (22)
dx ∂t
..
. Virial coefficients in Eq. (22) are calculated from the original
mn cpn ∂Tn virial coefficients by equating (21) and (22) and solving the orig-
= kn−1 (Tn−1 − Tn ) − kn (Tn − Tamb )
dx ∂t inal virial expansion for p. The new virial coefficients in terms
of B, C are
where n is the number of discretization sections of heat-
transfer area (equal to number of elements), mi is element B
B =
mass (i = 1, . . ., n), cpi is the is the specific heat of element RT
i, mi cpi is the element heat capacity, dx is the discretization
C − B2
C = 2
(RT)

Therefore the equations for the first derivative of the com-


pressibility factor with respect to temperature and pressure
are
 ∂z  p
 dB B
 p2
 dC dB
 2p2 2
= − + 2
− 2B + (B − C)
∂T p RT dT T (RT) dT dT R2 T 3

 ∂z  B 2(C − B2 )
= + 2
p
∂p T RT (RT)

For given gas composition the virial coefficients are


functions of temperature only. The AGA-8/1992 and ISO 12213-
3:1997 standards give constants, gas parameters and mixing
Fig. 1 – Heat-transfer area discretization scheme. rules for the calculation of the virial coefficients in Eq. (21).
1600 chemical engineering research and design 8 7 ( 2 0 0 9 ) 1596–1603

SRK equation of state and BWR equation of state were taken where j is the spatial coordinate discretization section index
for comparison of the flow models. The following form of SRK and X (Tj ) = ∂Tj /∂x, X (pj ) = ∂pj /∂x, X (Qnj ) = ∂Qnj /∂x. The five-
equation, allowing for convenient compressibility factor cal- point differentiation formula for spatial derivative of pressure
culations was used in this work: is given below as an example:
⎡ ⎤
Z3 − Z2 + Z(A − B − B2 ) − AB = 0 (23) dp(x0 )
⎢ dx ⎥
⎢ dp(x1 ) ⎥
where ⎢ ⎥
⎢ dx ⎥
⎢ ⎥
A=
a˛p ⎢ dp(x2 ) ⎥
2 ⎢ ⎥
(RT) ⎢ dx ⎥
⎢. ⎥
X (p) = ⎢
⎢ ..


bp ⎢ ⎥
B= ⎢ dp(xN−2 ) ⎥
RT ⎢ ⎥
⎢ dx ⎥
⎢ dp(xN−1 ) ⎥
R2 Tc2 ⎢ ⎥
a = 0.42747 ⎢ dx ⎥
pc ⎣ ⎦
dp(xN )
dx
b = 0.08664
RTc ⎡ ⎤
pc −25 48 −36 16 −3 ··· 0
⎢ −3 −10 −6 ⎥
⎢ 18 1 ... 0 ⎥
 2 ⎢ ⎥
⎢1 −8 0 8 −1 ... 0 ⎥
˛= 1+m 1−
T ⎢ ⎥
1 ⎢. .. .. .. .. .. .. ⎥
Tc = ⎢.
12 x ⎢ . . . . . . . ⎥⎥
⎢ ⎥
⎢0 ... 1 −8 0 8 −1 ⎥
m = 0.480 + 1.574ω − 0.176ω2 ⎢ ⎥
⎣0 ... −1 6 −18 10 3 ⎦
For iterative compressibility factor calculations Newton’s 0 ... 3 −16 36 −48 25
method was used. ⎡ ⎤
p(x0 )
BWR equation of state is in the following form:
⎢ p(x1 ) ⎥
  ⎢ ⎥
C0 ⎢ ⎥
p = RT + B0 RT − A0 − 2 + (bRT − a)3 + a˛6 ⎢ p(x2 ) ⎥
T2 ⎢ ⎥
⎢. ⎥
× ⎢ .. ⎥ + O( x4 )
c3
(1 + 2 )e−
2 ⎢ ⎥
+ ⎢ ⎥
T2 ⎢ p(xN−2 ) ⎥
⎢ ⎥
⎣ p(xN−1 ) ⎦
The values of eight coefficients: A0 , B0 , C0 , a, b, c, ˛ and
 depend on gas composition only. Application of Newton’s p(xN )
method enables iterative density calculation. Once the density
is known, compressibility factor is calculated from Eq. (4). This approximation is fourth-order correct, i.e. the trunca-
tion error is proportional to x4 . The derivative p(x) approx-
5. Solution method imated at point xj is based on function values at grid points
xj−2 , xj−1 , xj , xj+1 , xj+2 . For a detailed description of the solution
Eqs. (11), (14) and (17) comprise the set of hyperbolic partial algorithm see paper (Osiadacz and Chaczykowski, 2001b).
differential equations with pressure, flow and temperature as
a function of time and location. It is solved by the method of 6. Simulation results and discussion
lines with a five-point biased upwind approximation scheme
for spatial derivatives In the case study, different equations of state were incor-
porated into the gas flow model in order to determine the
  ∂z  
∂T n Qnj zRTj RTj n Qnj zRTj 1 1
=− X (Tj ) − zT +
∂t pj A cv pj A Tj z ∂T 
       ∂z    3
1 1 1 ∂z 1 1 2f zRTj n |Qnj | q
× X (Qnj ) − − X (pj ) + + X (Tj ) + + = 0, j = 0, 1, . . . , N
Qnj pj z ∂p T Tj z ∂T p cv D Apj cv

  ∂z    1  ∂z  −1 ∂T   ∂z  −1
∂p 1 1 1 n zRTj 1 1
= + − − − X (Qnj ), j = 0, 1, . . . , N (24)
∂t Tj z ∂T p pj z ∂p T ∂t pj A pj z ∂p T

  ∂z    ∂p 
∂Qn n Qnj zRTj 1 1 n Qnj zRTj
=− X (Qnj ) + Qnj − × + X (pj )
∂t pj A pj z ∂p T ∂t pj A
     ∂T 
1 1 ∂z n Qnj zRTj ∂T A 2fzRTj n Qnj |Qnj |
− Qnj + + − X (pj ) − , j = 0, 1, . . . , N
Tj z ∂T p ∂t pj A ∂x n DApj
chemical engineering research and design 8 7 ( 2 0 0 9 ) 1596–1603 1601

Fig. 2 – Section of the gas transportation system.

influence on the flow parameters such as pressure, temper-


ature, flowrate and line-pack. Line-pack calculations enable
the pipeline operator to justify for how long the supplies
of the gas could be continued in case of production stops,
and optimize its operation strategy. The Yamal-West Europe
pipeline on Polish territory was selected as a test network,
in particularly the 363 km pipe section between Kondratki Fig. 3 – Change of flowrate at x = L (boundary condition).
and Wloclawek compressor stations (Fig. 2). This is a typical
onshore gas transmission system with a maximum operating
pressure 8.4 MPa.
In the numerical calculations the following data were
used: Gas: The gas is a 9-component mixture with a
molar composition x of CH4 : 98.3455, C2 H6 : 0.6104, C3 H8 :
0.1572, i-C4 H10 : 0.0299, n-C4 H10 : 0.0253, i-C5 H12 : 0.0055, n-
C5 H12 : 0.0040, N2 : 0.0303 and CO2 : 0.7918. The density
n = 0.695 kg/m3 , dynamic viscosity  = 12.59 ␮Pa s, thermal
conductivity  = 0.0396 W/m K. Pipe: The distance between
the compressor stations is L = 363 km and the pipe diam-
eter do = 1422 mm. The average roughness of the pipe
ε = 0.0015 mm. The properties of the pipe wall are listed in
Table 1. Soil: The thermal conductivity  = 1.0 W/m K, den-
sity  = 1640 kg/m3 , specific heat at constant pressure cp = 1530
(J/kg K) and the pipe depth z = 1.5 m. The soil temperature is
3.1 ◦ C.
The boundary conditions are

p(0, t) = 8.4 MPa Fig. 4 – Change of suction pressure at Wloclawek (x = L).


T(0, t) = 296.65 K
Qn (L, t) = f (t)

where f(t) is depicted in Fig. 3 with a time interval, t ∈ [0,240].


The period function is arbitrarily selected. In case of AGA,
BWR and SRK equations of state, values of the derivatives of
compressibility factor were calculated based on polynomial
regression with degree two. Van der Waals (classical quadratic)
mixing rules were used to calculate the coefficients of SRK
equation of state for gas mixture. Tests have shown that the
convergence in the solutions of SRK and BWR equations is fast,
and the results are obtained within three iterations in this par-
ticular case study. Fanning friction coefficient was calculated
using Colebrook-White equation.
Figs. 4–6 present a graphical interpretation of the calculated
flow parameters at the inlet and outlet nodes of the pipeline.
Figs. 4 and 5 clearly show a minor effect of the type of the equa-
tion of state in gas flow model on pressure and flowrate values
Fig. 5 – Change of flowrate at Kondratki (x = 0).
in the pipeline. Changes of the results obtained from differ-

Table 1 – Properties of pipe wall.


Pipe wall structure Thickness (mm)  (W/m K)  (kg/m3 ) cp (J/kg K)

Internal coating 0.5 0.52 1800 1050


Steel L480MB (70×) 19.22 45.3 7830 500
External coating (polyethylene) 3.0 0.4 940 1900
1602 chemical engineering research and design 8 7 ( 2 0 0 9 ) 1596–1603

7. Conclusion

This paper has considered the possible improvements in the


simulation accuracy by studying the influence of the equa-
tion of state on the non-isothermal flow model, which is used
in the evaluation of control and operating strategies of the
typical onshore gas pipeline. The results of comparison of
flow parameters and pipeline line-pack values show relatively
small influence of the type of the equation of state on the
simulation results. The character of the results cannot be eas-
ily generalized, but closer observation suggests that the form
of the equation of state might be regarded as a contributory
factor in leak detection systems based on volume balance
methods.
There are many parameters associated with a mathemat-
ical description of fluid flow transients within the pipeline,
for which it is difficult to determine appropriate values: fric-
Fig. 6 – Change of temperature at x = L. tion factor, heat-transfer coefficient, speed of sound, ambient
temperature, actual pipeline geometry. By using measured
response data for the pipeline under a range of operation con-
ditions and regimes, comparison with the simulation results
ent equations of state are insignificant. It can be seen from
can be made. It is then possible to validate the model with
Fig. 6 that the value of the gas temperature at the outlet node
estimated values of these parameters, and provide a means of
drops below the surrounding soil temperature (3.1 ◦ C), which
establishing an improved efficiency of the overall operation of
is a demonstration of the effect of cooling of the gas due to
the pipeline.
expansion. The value of the gas temperature decreases when
increasing the flowrate, because larger pressure drop in the
pipeline and larger expansion of the gas downstream of com- Appendix A. Detailed derivation of Eq. (18)
pressor station is observed. The temperature tends to be lower
in the case of SGERG and AGA equations of state. From this we Substituting h = u + (p/), Eq. (3) takes the form
can assume that compressor station power, calculated based
on the simulation results obtained from these equations, will
   
w2 d dh d w2
be a little smaller than in the case of the SRK and BWR equa- h+ + +
2 dt dt dt 2
tions of state (reduced cost and/or higher throughput).  
Fig. 7 shows the influence of different equations of state ∂p w2 ∂w
− + h+  = q − wg sin ˛ (A1)
on the pipeline line-pack. Differences between the equa- ∂t 2 ∂x
tions exist but are relatively moderate. The largest difference
between line-pack value with AGA and BWR equations Substituting Eq. (5), the first and the last term on the left-hand
amounts to 118 000 m3 which corresponds to 5,9% of the actual side of Eq. (A1) can be canceled out. Combining this with Eq.
flowrate (2 000 000 m3 /h) at the outlet node of the pipeline, (2) we obtain:
and is equal to 0.23% of the average value of the line-pack
obtained from the solutions of the all analysed equations of dh ∂p 2fw3 ∂p
state.  −w − − = q (A2)
dt ∂x D ∂t

Substituting dh = du + d(p/), Eq. (A2) can be rewritten in the


form:

du p d 2fw3
 − − = q (A3)
dt  dt D

By using the thermodynamic identity

  ∂p 
du = cv dT + T − p dv (A4)
∂T v

we obtain

dT ∂p
  1 d 2fw3
cv −T − = q (A5)
dt ∂T   dt D

Substituting Eq. (5) into Eq. (A5) we have

dT ∂p
  ∂w 2fw3
cv +T − = q (A6)
Fig. 7 – Change of pipeline line-pack. dt ∂T  ∂x D
chemical engineering research and design 8 7 ( 2 0 0 9 ) 1596–1603 1603

From the equation of state References


 ∂p    ∂Z  
Abbaspour, M. and Chapman, K.S., 2008, Nonisothermal transient
= R z + T (A7)
∂T  ∂T  flow in natural gas pipeline. J. Appl. Mech., Trans. ASME, 75(3):
0310181–0310188.
From Eq. (11) we have Gersten, K., Papenfuss, H.D., Kurschat, T., Genillon, F., Fernandez,
P. and Ravell, N., 2001, Heat transfer in gas pipelines. Oil Gas
  ∂z  ∂p   ∂z   ∂T Eur. Mag., 27(1): 30–34.
1 ∂w 1 ∂Qs 1 1 1 1 Modisette, J.L., 2000, Equations of state tutorial, In Proceedings of
= − − + +
w ∂x Qs ∂x p z ∂p T ∂x T z ∂T p ∂x the PSIG—The 32nd Annual Meeting Savannach,
(A8) Modisette, J., 2002, Pipeline thermal models, In Proceedings of the
PSIG—The 34th Annual Meeting Portland,
Finally, Eq. (18) is obtained by substituting Eqs. (A7) and (A8) Osiadacz, A.J., 1996, Different transient models—limitations,
into Eq. (A6): advantages and disadvantages, In Proceedings of the PSIG—The
28th Annual Meeting San Francisco,
  ∂z   Osiadacz, A.J. and Chaczykowski, M., 2001, Comparison of
∂T s Qs zRT ∂T RT s Qs zRT 1 1 isothermal and non-isothermal pipeline gas flow models.
+ + zT +
∂t pA ∂x cv pA T z ∂T  Chem. Eng. J., 81(1–3): 41–51.

1 ∂Q 1   ∂p     ∂T Osiadacz, A. and Chaczykowski, M., 2001, The thermodynamics
s 1 ∂z 1 1 ∂z of pipeline gas flow. Arch. Thermodyn., 22(1–2): 51–75.
× − − + +
Qs ∂x p z ∂p T ∂x T z ∂T p ∂x Thorley, A.R.D. and Tiley, C.H., 1987, Unsteady and transient flow

2f
 zRT |Q | 3 q
of compressible fluids in pipelines—a review of theoretical

s s
− =0 and some experimental studies. Int. J. Heat Fluid Flow, 8(1):
cv D Ap cv 3–15.

Das könnte Ihnen auch gefallen