Beruflich Dokumente
Kultur Dokumente
Equations and
Methods
ver. 5
rel. November 2002
This issue has a special position within the SIMONE documentation library. While the
purpose of all the other issues is to support the user by information practical for real use, this
text offers the information about basic equations which the SIMONE system is based on:
while all other is-sues treat the fruits of the SIMONE tree shown bellow, this issue is
dedicated to its roots.
It is definitely out of the possibilities of this material to describe the trunk of the tree the
theory behind SIMONE which is considerably complicated and abstract. SIMONE is based
on several original methods tailor-made for the pipeline networks. The main are the
integration method and the method for sophisticated exploitation of the sparseness of the
matrix.
Those who are interested only to know how it works but not why it works so well are
recommended to ignore this issue and to turn attention only to other sheets of the SIMONE
documentation library.
Logic
e
Mo m
ntum
Automatic
Prefrencial
Optimum
Optimiza- Financial
Design
tion
Optimization
ua
Eq
tion
He
at
Config.
Optimization
Set-Point
Optimization
d
ho
.
et
M
eth
nM
tion
gra ositio eth.
p
Inte
M
com tion
De imiza
t
Op
Error
Identific. Leak
State
Detection Reconstruction
Results
Monitoring
llan
ce
Mixing Compressor
Stations Stations
Topo
Visualization
Ba
Quality
Heat
Tracking
Dynamics
M
as
Network
Adjustment
State to State
Optimization
S ta
te
Bal
lanc
e
Equat
ion
Contents
1. Units ....................................................................................................................................... 7
1.1.
2.2.
2.3.
Friction factor............................................................................................................. 9
2.3.1.
Hofer formula..................................................................................................... 9
2.3.2.
Nikuradze formula............................................................................................ 10
2.3.3.
PMT-1025 formula........................................................................................... 10
2.3.4.
Pipe efficiency.................................................................................................. 10
2.3.5.
3.2.
3.3.
Redlich-Kwong equation.......................................................................................... 15
3.4.
3.4.1.
3.4.2.
3.5.
3.6.
3.6.1.
3.6.2.
3.6.3.
Joule-Thomson coefficient............................................................................... 20
3.7.
3.7.1.
Calculation gas relative density and volumetric heating value using gas
composition ...................................................................................................................... 21
3.7.2.
3.8.
3.8.1.
3.8.2.
4.2.
4.3.
5.2.
5.3.
6. Compressor Stations............................................................................................................. 34
6.1.
6.1.1.
6.1.2.
Adiabatic head.................................................................................................. 34
6.1.3.
6.1.4.
6.1.5.
Discharge temperature...................................................................................... 35
6.1.6.
6.1.7.
6.2.
6.3.
6.4.
6.5.
6.6.
6.7.
6.8.
Gas cooler................................................................................................................. 41
6.9.
7. Controlled valves.................................................................................................................. 43
7.1.
Preheating power...................................................................................................... 43
7.2.
7.2.1.
Characteristic A (Mokveld).............................................................................. 44
7.2.2.
Characteristic B (Argus)................................................................................... 45
7.3.
8. Resistor................................................................................................................................. 47
9. Solution methods.................................................................................................................. 48
10. State reconstruction ............................................................................................................ 50
10.1.
10.2.
10.3.
10.4.
11.2.
Constraints............................................................................................................ 55
11.3.
12.2.
6
12.2.1.
12.3.
12.3.1.
12.3.2.
Carson-Katz method......................................................................................... 60
12.3.3.
12.3.4.
12.4.
12.5.
Risk indicators...................................................................................................... 64
1. Units
1.1.
For low-pressure networks the local pressure is expressed in terms of overpressure in the
order of up to 5 kPa. The effect of lower gas density with respect to air density (i.e. lift force)
therefore leads to non-negligible changes of local overpressure with elevation.
SIMONE calculates all pressures as absolute ones. For correct simulation of low-pressure
systems the local air pressure depending on node elevation is used for conversion between
absolute and gauge value of pressure. This is done by selection of kPah pressure unit. This
unit can be used both for value visualization and for input of pressure conditions or set points.
The conversion is based on the following equation
p l = p p air (h )
Here
pl
p
p air (h )
h
Local overpressure
Absolute gas pressure
Local air pressure (absolute) depending on elevation
Elevation
[kPah]
[kPa]
[kPa]
[m]
The elevation change of air pressure is derived form the barometric equation using the
definition of International Standard Atmosphere:
gh
p air (h ) = p 0 exp
Rair T0
2. Flow equations
Simulation of dynamic processes running with gas transport and distribution is based on the
non-linear partial differential equations describing the dynamics continuity equation and
momentum equation.
For a more sophisticated and detailed description of the heat dynamic processes running with
the gas flow, another partial differential equation must be linked to those mentioned above
see sections 4 and 5.
2.1.
Continuity equation
+S
=0
t
x
Here
m
S
x
t
2.2.
[kg.s-1]
[m2]
[kg.m-3]
[m]
[s]
Momentum equation
1 m
dh
P
2w
w2
+
+ g
+ fR = 0
x x
S t
dx
t
Here
p
g
fR
h
S
m
w
x
t
Pressure
Gravity acceleration
Gas density
Pressure loss due to friction per unit of pipe length
Geodetic height
Pipe cross section
Mass flow rate
Flow velocity
Lengths coordinate
Time
[Pa]
[m.s-2]
[kg.m-3]
[Pa.m-1]
[m]
[m2]
[kg.s-1]
[m.s-1]
[m]
[s]
The unit hydraulic resistance in a pipeline with a circular cross section is described by DarcyWeisbach equation
fR =
| w| w
2D
Here
Friction factor
Flow velocity
Gas density
Pipe internal diameter
2.3.
[1]
[m.s-1]
[kg.m-3]
[m]
Friction factor
SIMONE is equipped with several formulae for the friction factor (being selected using the
LAMBDA scenario parameter):
Any other formula can be built into the system on users request.
4.518 Re
k
2 log
log +
7 3.71D
Re
Here
k
D
Re
Re = w
D
4 m
=
v D
Here
D
w
[m]
[m.s-1]
[m2 s-1]
[kg.s-1]
[kg.m-1.s-1]
The value of dynamic viscosity for natural gas is approximately 10-5 kg.m-1.s-1.
10
D
2 log + 1.138
k
Re D
0 .2
= 0.067
k = 0.03 mm
in the momentum
fR =
| w| w
2
2D
pipe
There is simple and straightforward link between flow rate and pipe efficiency for steady
flow: If one assumes the pipe efficiency 0.99 instead of 1.00, then the flow rate for constant
pressure drop is also reduced to 0.99 of its original value.
The pipe efficiency can be altered by setting CORLAM scenario parameter for subsystem or
individual pipe element (the default value is 1).
11
In the fist picture, the line is calculated by Hofer formula and the last point is calculated by
Nikuradze formula (case Re + ):
Hofer - Nikuradze
0.05
D/k=60
0.04
120
250
0.03
500
1000
0.02
2500
5000
0.01
10000
100000
0
1.00E+06
1.00E+07
1.00E+08
1.00E+09
Re
Percentage differences between Hofer and Nikuradze formulae are depicted in the next figure:
8
7
D/k=100000
10000
5000
2500
1000
500
1
0
1.00E+06
1.00E+07
1.00E+08
Re
1.00E+09
12
To get rough figures about flow rate and Reynolds number see the following graph (drawn for
natural gas of normal density 0.73 kg/m3, i.e., relative density 0.565):
Q[1000m3/h]
100000
10000
D=1400 mm
1200 mm
1000
900 mm
500 mm
300 mm
100
10
1.00E+06
1.00E+07
1.00E+08
1.00E+09
Re
For D=1000 mm, k=0.01 mm, and usual load of about 106 m3/h results Re=2.4107 and
therefore the non-negligible difference between Hofer and Nikuradze can be seen!
13
p = RTz
Here
R
T
z
Pressure
Gas density
Gas constant
Absolute temperature
Compressibility factor
[Pa]
[kg.m-3]
[J.kg-1.K-1]
[K]
[1]
The gas constant is related to molar weight or relative density (specific gravity) of gas
~
R
R=
M
r=
Here
~
R
M
r
z
Subscript 0
Subscript air
[J.kmol-1.K-1]
[kg.kmol-1]
[1]
[kg.m-3]
[1]
The non-ideal behavior of the gas is expressed by the compressibility coefficient z . The
compressibility coefficient is function of pressure, temperature and composition of gas. The
following choices are available in SIMONE:
Here
p
pc
T
Tr =
Tc
pr =
[1]
[1]
14
N
p c = x i p c ,i
Tc = xi Tc ,i
x1 K x N
p c ,i
[1]
[Pa]
Tc ,i
[K]
[Pa]
i =1
N
i =1
Here
N
= xi i
[1]
[1]
i =1
z = z ( p, T , x1 K x N )
More complex mixing rules are used than the above mentioned.
The following equations of state are currently available in SIMONE (being selected by the
ZET scenario parameter):
AGA formula
Papay formula (default)
Redlich-Kwong equation
BWR in simplified form (Fasold et al.)
BWR
AGA8 DC92 (ISO 12213-2)
3.1.
z = 1 + 0.257 p r 0.533
pr
Tr
3.2.
15
Papay formula
3.3.
Redlich-Kwong equation
p= ~
~ ~
V b
T V (V + b)
Here
p
T
~
R
~
V
~
R 2Tc2.5
a = a
pc
~
R Tc
b = b
pc
1
a =
1/ 3
9(2 1)
a =
Pressure
Temperature
Universal gas constant
Molar volume of gas
[Pa]
[K]
[J.mol-1.K-1]
[m3.mol-1]
21 / 3 1
3
For pure gases, Redlich-Kwong equation is the most precise 2-constant equation of state.
Redich-Kwong equation is currently used in simplified form with the simple definition of
pseudo critical pressure and temperature:
N
p c = x i p c ,i
i =1
Tc = xi Tc ,i
i =1
16
3.4.
BWR equation
) (
Here
p
T
~
R
~
A0 , B0 , C 0 , a , b , c , ,
Pressure
Temperature
Universal gas constant
Molar density of gas
BWR equation constants (traditional units)
[atm]
[K]
[atm.dm3 mol-1.K-1]
[mol.dm-3]
A0 = 0.06000000 c + 0.6900000
B0 = 0.00797872 c + 0.0781383
C 0 = 0.02000000 c + 0.0340000
a = 0.04083330 c 0.2128330
a = 0.01700000 c 0.0706000
a = 0.02400000 c 0.0980000
= 0.00183333 c + 0.0138333
= 0.01400000 c + 0.1332000
p c = x i p c ,i
Tc = xi Tc ,i
[atm]
i =1
N
i =1
17
c = x i c , i
c,i
x1 K x N
[1]
i =1
Critical Riedel factor is calculated using the acentric factor by equation of Edmister
c = 5.7839 + 4.8763
As a result, this simplification leads to 3-parametric equation of the form z = z ( p r , Tr , ) .
A0 = xi A0,i
i =1
B0 = xi B0,i
i =1
C 0 = x i C 0 ,i
i =1
a = xi 3 ai
i =1
b = xi 3 bi
i =1
c = x i 3 ci
i =1
= xi 3 i
i =1
= xi i
i =1
Here
x1 K x N
Subscript i
[1]
The constants of BWR equation for basic 21 gas components listed in ISO 12213 are
predefined in SIMONE. The user has the possibility to add new gas component and enter the
constants for it.
18
3.5.
The expansion of compressibility factor according to AGA8 DC92 (ISO 12213-2) equation is
18
58
n =13
n =13
Here
r
bn , c n , k n
Compressibility factor
Second virial coefficient (function of temperature and
composition)
Molar gas density
Dimensionless reduced density
Constants
[mol.dm-3]
[1]
[1]
C n
[1]
z
B
~
[1]
[dm3.mol-1]
r = K 3 ~
Here
Mixture size parameter (function of composition)
K3
[dm3.mol-1]
3.6.
All important thermodynamic properties of gas can be derived from the equation of state,
namely
Specific heat
Isentropic exponent
Joule-Thomson coefficient
Here
c~p
19
[J.mol-1.K-1]
[J.mol-1.K-1]
cp =
cp
M
The ideal gas specific heat is a function of temperature and in SIMONE is approximated by
second-order parabola:
c~p0 (T ) = A + B T + C T 2
For gas mixture the coefficients are simply mixed form pure component values
A N Ai
B = x i Bi
C i =1 C
i
Here
x1 K x N
Subscript i
Here
~
R
z
20
z
~
R
Z p ~ Z T2
cp
Here
p
~
V
S
z
Pressure
Molar volume
Entropy
Compressibility factor
[Pa]
[m3.mol-1]
[J.mol-1.K-1]
[1]
z
Z p = z p
p T , x1Kx N
[1]
[1]
z
ZT = z + T
T p , x1Kx N
~
R
c~
p
JT =
p H
and can be calculated from the equation
JT
~
T R
= ~ (Z T z )
p cp
Here
JT
p
T
H
z
z
ZT = z + T
T p , x1Kx N
~
R
c~
p
Joule-Thomson coefficient
Pressure
Temperature
Enthalpy
Compressibility factor
[K.Pa-1]
[Pa]
[K]
[J.mol-1.K-1]
[1]
[1]
The integral Joule-Thomson effect, i.e. temperature change caused by pressure drop during
isenthalpic process, can be described by the initial-value problem
SIMONE Research Group, s.r.o.
21
T ( p1 ) = T1
dT
= JT ( p, T )
dp
T2 = T ( p 2 )
Here
p1
T1
p2
T ( p)
T2
Starting pressure
Starting temperature
Final pressure
Solution of the initial-value problem
Final temperature
3.7.
[Pa]
[K]
[Pa]
[K]
In gas transport & distribution, volumetric quantities and units are widely used:
Flow rates and line pack are expressed in standard volumetric units (defined for
standard condition pressure 101.325 kPa and temperature typically 0, 15 or 20C)
Volumetric heating value per unit volume in standard condition; thermal billing is
based on volumetric flow rate and volumetric heating value
Relative density is used to convert volume of gas to mass.
The composition of gas can be expressed in volumetric fractions (defined for standard
condition) rather then in molar or mass fractions.
On the other hand, all equations mentioned above (flow equation, equations of state) are
based on mass or molar quantities.
SIMONE supports the conversion of molar quantities to volumetric and vice versa using
formulae consistent, whenever possible, with ISO 6976 norm.
z air ,0 M
z 0 M air
Here
z air ,0
[1]
22
p0 = 101.325 kPa
[kg.kmol-1]
M air
z0
M
M = xi M i
i =1
xi
Mi
[1]
[kg.kmol-1]
Molar gas heating value is the amount of heat released by burning 1 kmol of gas at constant
pressure p0 = 101.325 kPa while the temperature of gas and air before combustion and the
temperature of combustion products are the same:
~
H U [TH ] [MJ.kmol-1]
TH
Temperature of combustion products (0, 15, 20 or 25oC)
SIMONE uses the upper (superior) heating value (i.e. all water in combustion products is
assumed to condense) at the temperature TH = 25C .
Molar heating value for gas mixture is given by simple weighted sum of molar heating values
of individual mixture components:
N
~
~
H U = xi H U ,i
i =1
Volumetric gas heating value is the amount of heat released by burning 1 m3 std. of gas at
constant pressure p0 = 101.325 kPa (while the temperature of gas and air before combustion
and the temperature of combustion products are the same):
z 0 = 1 xi bi
i =1
bi
23
[1]
~
The data for individual components (heating values H U ,i ; summation factors bi for 0, 15 and
20C) are taken from ISO 6976:1995(E), Table 1-3.
yi
z 0 ,i
xi = N
, i = 1, K , N
yj
z
j =1
yi =
0, j
x i z 0 ,i
N
x
j =1
z 0, j
, i = 1, K , N
Here
z 0,i
3.8.
Gas viscosity
By default, the gas viscosity is taken as constant entered within the network editor. However
for the case of e.g. high-pressure systems with coated pipes the effect of varying viscosity can
be included.
SIMONE supports currently two classic correlations for gas viscosity:
= ( , T , x ) or = ( p, T , x )
can be easily included as well. Here
LIWACOM Informationstechnik GmbH
24
p
T
x
Dynamic viscosity
Gas density (mean value for pipe element)
Pressure (mean value for pipe element)
Gas temperature (mean value from temperature profile for pipe element)
Gas composition vector of molar fractions
[Pa.s]
[kg.m-3]
[bar]
[K]
[1]
= K exp(X Y )
K=
122.4 + 12.9 M + T
1914.5
X = 2.57 +
+ 0.0095M
T
Y = 1.11 + 0.04 X
Here
T
M
[Poise=10-7Pa.s]
[g.cm-3]
[R]
[g.mol-1]
= K exp(X Y )
K=
209.2 + 19.26 M + T
986.4
X = 3.448 +
+ 0.01009 M
T
Y = 2.4 0.2 X
Here again
T
M
[Poise=10-7Pa.s]
[g.cm-3]
[R]
[g.mol-1]
25
4. Quality Tracking
To describe the dynamics of particular components of a gas mixture within the pipeline
system in an exact way, the mathematical model must contain the mass balance equation for
each component:
mci
ci
+S
= 0, i = 1K N
x
t
Here
m
S
ci K c N
x
t
[kg.s-1]
[m2]
[kg.m-3]
[1]
[m]
[s]
4.1.
Solution procedure
The number of the above mentioned additional partial differential equations would complicate
the solution essentially. Therefore, an approximate solution has been adopted, based on the
following presumptions:
The quality signal is discrete, i.e., its value is known only for the discrete time of
sampling and no information exists within the sampling period;
Neither the time step of sampling need not be equal nor to the time step of calculation
nor to the sampling period of other variables.
The value of the sample of quality parameter is related to an infinitesimal volume of gas just
present at the sampling time in the place of sampling quality flag. The quality flag is
characterized by its position and by a set of values corresponding to the respective tracked
quality parameters.
The quality tracking is then formulated as a task to
Follow the movement of all quality flags entering the network in the supply points
until they leave the system
Simulate the mixing of quality values in all nodes of a network to which more than
one gas stream is linked to.
26
flags, the mixing of quality parameters in the crossings of the network, and the
interpolated quality values in nodes.
2. As some parameters of the simulation model itself depend on the gas parameters, (gas
composition, relative density, heating value, critical temperature and critical pressure
etc.), all these values are used in the next step of simulation to calculate the new
values of corresponding coefficients.
4.2.
Quality parameters
Uniform gas quality the network is filled by gas of one composition only.
Gas physical properties tracking various gases at individual supplies can enter the
pipeline network. To describe the gas behavior, only the set of gas physical properties
is used and tracked over the network. This implies that only 2- or 3-parametric
equations of state can be used.
Gas composition tracking various gases at individual supplies can enter the pipeline
network and both gas physical properties and gas composition are tracked over the
network.
4.3.
Mixing rules
(1)
(2)
(k)
27
The mixing of different gas streams is described by the following balance equations:
Mass balance:
k
m = m( j)
j =1
Here
m
M
Superscript ( j )
Mass flux
[kg.s-1]
Gas molar weight [kg.kmol-1]
Gas stream j
Therefore the mixing rule for all gas quality parameters can be written in a compact form:
M
~
HU
Tc
pc
=
B
x1
M
x
N
~
1
m( j )
( j)
j =1 M
k
M ( j)
~
H U( j )
( j)
Tc
( j)
pc
j )
k
m ( j ) A( j )
( j) ( j)
j =1 M
B
C ( j)
x1( j )
x( j)
N
~ ( j )
Here
Molar mass of gas
Molar heating value of gas
[kg.kmol-1]
[MJ.kmol-1]
M
~
HU
Tc
pc
A, B, C
xi
[1]
28
[x.kmol-1]
29
5. Heat Dynamics
In the standard SIMONE package, heat dynamics are respected by calculating the temperature
changes in a pipeline behind a compressor station or pressure reducer using a fixed
exponential model. The temperature of each node can be entered individually for a particular
time period.
In most situations, this concept is quite satisfactory. Nevertheless, there are occasions when
the description of heat dynamics is required pipe sections downstream of compressor
stations, underwater lines, and long sections with high-pressure drop. In all of these cases, the
omission of heat balance results in a loss of accuracy.
Therefore the heat dynamics model is available in SIMONE.
5.1.
Differential Equations
Heat transients in the gas resulting from the Joule-Thompson effect longitudinal
dynamics.
Heat transients resulting from the heat exchange between the flowing gas and the
enveloping tube together with its further environment axial dynamics.
+ Swg
+ QE = 0
Sw
S cP
+ w S 1 +
dx
z T p x
z T p t
x
t
Here
S
cp
T
t
w
x
p
z
g
h
QE
[m2]
[kg.m-3]
[J.kg-1.K-1]
Gas temperature
Time
Flow velocity
Lengths coordinate
Pressure
Compressibility factor
Gravity acceleration
Geodetic height
Heat flux from gas through the inner surface of
pipe to the surrounding soil per unit of length
[K]
[s]
[m.s-1]
[m]
[Pa]
[1]
[m.s-2]
[m]
[J.s-1.m-1]
The description of the heat flux QE should describe the effect of heat capacity of pipesurrounding soil, e.g. if gas cooler than near soil enters suddenly the pipe, the back heat flux
from soil to gas (until the near soil is cooled) occurs and should be modeled properly.
LIWACOM Informationstechnik GmbH
30
The character of the dynamic processes in the pipe and the surrounding soil is stiff it
consists of a quick component corresponding to the near soil and a very slow one
corresponding to the remote mass of soil. Therefore the process needs to be modeled by a
minimum of two capacitors with substantially different time constants:
d Ts1
= 1(T Ts1 ) 2 (Ts1 Ts 2 )
dt
dT
C 2 s 2 = 2 (Ts1 Ts 2 ) 3 (Ts 2 Tsoil )
dt
C1
Here
Gas temperature
Near soil temperature at the surface of first layer (capacitor)
Near soil temperature at the surface of second layer (capacitor)
Far soil temperature being not influenced by temperature changes of
gas; this value is referred in SIMONE scenario as Ground Temperature
(GT)
Heat capacity of first layer (per unit of pipe length)
Heat capacity of second layer (per unit of pipe length)
Heat transfer coefficient through first layer (per unit of pipe length)
Heat transfer coefficient through second layer (per unit of pipe length)
Heat transfer coefficient between the surface of second layer and the
surface of constant far soil temperature (per unit of pipe length)
T
Ts1
Ts 2
Tsoil
C1
C2
1
2
3
[K]
[K]
[K]
[K]
[J.m-1.K-1]
[J.m-1.K-1]
[J.m-1.K-1.s-1]
[J.m-1.K-1.s-1]
[J.m-1.K-1.s-1]
The heat flux QE is then given by the first right-hand side term in the equation for first layer
temperature:
QE = 1(T Ts1 )
The steady-state solution of axial heat dynamic leads to formula
QE (t + ) = Dk (T Tsoil )
Here
D
k
[m]
[J.m-2.K-1.s-1]
The parameters of 2-layer model of axial heat dynamics must therefore obey the steady-state
condition
1
1
Dk
31
The dynamics of 2-layer model can be fully described by the following 4 parameters:
1
Dk
k3 = 3
Dk
k1 =
1 =
C1
1 + 2
2 =
C2
2 +3
Here
Dimensionless parameters describing the splitting of overall heat transfer [1]
coefficient into 3 resistors. They can be altered using SIMONE
parameters LEEKK1, LEEKK3
Time constants of the 2 capacitors. They can be altered using SIMONE [s]
parameters LEEKT1, LEEKT2.
k1 , k 3
1 , 2
SIMPLE only the steady-state heat flux between pipe and soil is used. This model is
for use in steady-state scenarios.
FULL the full axial dynamic is respected. However, the initial condition is to be
carefully prepared the initial gas temperature distribution should correspond to
steady-state profile in order to initialize reasonably the axial model.
5.2.
In the following paragraph, the basic recommendations for the value of heat transfer
coefficient are summarized.
Heat flow between gas and surrounding environment (in a steady-state situation) is described
in SIMONE by equation
QE = D HTC (T Tsoil )
Here
QE
D
T
HTC
Heat flux from gas through the inner surface of pipe to the surrounding
soil per unit of length
Internal pipe diameter
Gas temperature
Heat transfer coefficient between gas and far soil (per unit of internal
[W.m-1]
[m]
[K]
[W.m-2.K-1]
32
Tsoil
Let we suppose cylindrical model of n layers of different materials where the last layer has
constant temperature equal to Tsoil . The heat flow between gas and surrounding environment is
described then as
QE =
2
D
ln i +1
Di
i =1 i
n
(T Tsoil )
Here
Di
[m]
[W.m-1.K-1]
Comparing of both equations one obtains the equation for the Heat Transfer Coefficient
HTC =
D1
i =1
2
1
ln
Di +1
Di
Due to high conductivity and small thickness of pipe wall the layer of pipe itself can be
neglected. Typical situation can be represented by two layers, bitumen and clay. Because
for bitumen is roughly the same as for wet clay only one layer is sufficient and equation for
HTC can be rearranged as
HTC =
2
2
D ln1 +
Here
Thickness of the clay layer; at the outer surface of clay the temperature is assumed [m]
to be equal to Tsoil
33
To show how thickness of clay layer and pipe diameter D influence the relation between
HTC and see the following picture. The typical values for are summarized below:
Material
Concrete
Dry clay
Wet clay
Bitumen
[W.m-1.K-1]
1.28 1.5
0.14
0.65
0.6
HTC/lambda
clay thickness 500 mm
HTC/lambda
5
4
3
2
1
0
200
400
600
800
1000
1200
1400
Diameter [mm]
5.3.
p
p
T
dh
T
+ w S (1 + c p JT )
Sw c p JT
+ Swg
+ QE = 0
S cP
x
dx
t
x
t
then neglecting Joule-Thomson effect leads to
p
T
dh
T
+w S
S cP
+ Swg
+ QE = 0
x
dx
t
t
34
6. Compressor Stations
6.1.
p0
pi
Here
pi
po
Pressure ratio
[1]
Inlet (suction) pressure
[Pa]
Outlet (discharge) pressure [Pa]
1
RTi z i 1
1
Here
H ad
R
Ti
z i = z ( pi , Ti )
Adiabatic head
Isentropic exponent (mean value)
Gas constant
Inlet gas temperature
Compressibility factor at inlet condition
Pressure ratio
H ad
Ho Hi
Here
ad
H ad
Hi
Ho
Adiabatic efficiency
Adiabatic head
Gas enthalpy at inlet condition
Gas enthalpy at outlet condition
H ad
ad
[1]
[kJ.kg-1]
[kJ.kg-1]
[kJ.kg-1]
[kJ.kg-1]
[1]
[kJ.kg-1.K-1]
[K]
[1]
[1]
35
Here
m
ad
H ad
Hi
Ho
[kg.s-1]
[1]
[kJ.kg-1]
[kJ.kg-1]
[kJ.kg-1]
Tx = Ti
1
ad
1
+ 1
Tx = Ti
ad
Isentropic equation
Tx = Ti
2. The temperature Tx obtained in previous step is corrected to the real gas behavior
using formula selected by the THETACOR (Theta correction) scenario parameter:
Basic (default)
To = Tx
zi
z ( p o , Tx )
To = Tx
LIWACOM Informationstechnik GmbH
36
Here
Ti
ad
Tx
z
To
[K]
[1]
[1]
[1]
[K]
[1]
[K]
Ti + To
2
z
R
Z p Z T2
cp
Here
Compressibility factor
[1]
z
Z p = z p
p T , x1Kx N
[1]
z
ZT = z + T
T p , x1Kx N
R
cp
[1]
Gas constant
Real gas specific heat (for constant pressure)
[J.kg-1.K-1]
[J.kg-1.K-1]
37
i +o
2
Qvol ,i =
=m
RTi z i
pi
Here
Qvol ,i
m
R
Ti
zi
[m3.s-1]
[kg.s-1]
[kg.m-3]
[J.kg-1.K-1]
[K]
[1]
6.2.
Centrifugal compressor
SIMONE uses the working envelope (wheel map) of centrifugal compressor in the
coordinates (inlet volumetric flow rate Qvol .i adiabatic head H ad ).
The envelope is approximated by set of curves expressing adiabatic head and adiabatic
efficiency as biquadratic polynomials in compressor speed and volumetric flow rate:
H ad
ad
a1
= 1 n n a2
a
3
a4
b1
= 1 n n b2
b
3
b4
(
(
)
)
a5
a6
b5
b6
a 7 1
a8 Qvol ,i
2
a9 Qvol
,i
b7 1
b8 Qvol ,i
2
b9 Qvol
,i
Here
ad
Adiabatic efficiency
H ad
n
Qvol ,i
Adiabatic head
[kJ.kg-1]
Compressor revolutions (speed in r.p.m.) [min-1]
Volumetric flow rate at inlet condition
[m3.s-1]
a1 K a9 , b1 Kb9
Coefficients
[1]
38
6.3.
Reciprocating compressor
Qvol .i = VW
n
60
Here
[m3.s-1]
Qvol ,i
VW
n
PS = 2
n
MT
60
Here
PS
n
MT
Combining the above equations with the general formula for compressor shaft power, one
obtains the relation between adiabatic head and torque:
MT =
VW
2 ad
H ad i
Here
VW
ad
H ad
[m3]
[1]
[kJ.kg-1]
[kg.m-3]
6.4.
39
Gas turbine
Pmax
c1
= 1 n n 2 c2
c
3
c4
c5
c6
c 7 1
c8 t amb
2
c9 t amb
Here
Pmax
n
t amb
c1 K c9
PF = d1 + d 2 PS + d 3 PS2
Here
PF
PS
d1 K d 3
6.5.
Gas engine
Pmax = c1 + c 2 n + c3 n 2
Here
Pmax
n
c1 K c3
40
PF = d1 + d 2 PS + d 3 PS2
Here
PF
PE
n
d1 K d 3
6.6.
Electro drive
Pmax
c1
= 1 n n c2
c
3
c4
c5
c6
c 7 1
c8 t amb
2
c9 t amb
Here
Pmax
n
t amb
c1 K c9
Pmax = c1 + c 2 n + c3 n 2
Here
Pmax
n
c1 K c3
41
PE = d1 + d 2 PS + d 3 PS2
PE
PS
d1 K d 3
6.7.
For modeling of combined-cycle drives (primary gas turbine, steam boiler utilizing exhaust
heat and secondary steam turbine) the correlation of total energy consumption of primary
turbines and shaft power of secondary turbine is used:
2
FFP = a1 + a 2 PSS + a3 PSS2 or PSS = b1 + b2 FFP + b3 FFP
Here
FFP
PSS
a1 K a3 , b1 Kb3
Started only if the fuel gas consumption of primary turbines is sufficiently high
Stopped only if the fuel gas consumption is lower than prescribed switch-off limit
6.8.
Gas cooler
k
To = Tc ,i + (Ti Tc ,i )exp
m
Here
Ti
To
Tc ,i
42
m
k
[kg.s-1]
[kg.s-1]
6.9.
For the compressor station element, the pressure loss in inlet and outlet parts of piping yard of
the compressor station can be modeled. In addition, the local pressure drop between stages
can be taken into account.
Local pressure loss is calculated using local pressure loss coefficient
1
pi p o = i wi2
2
Here
pi
po
wi
[Pa]
[Pa]
[1]
[kg.m-3]
[m.s-1]
The gas velocity is calculated using representative internal diameter of compressor station D ,
entered in network element data:
wi =
4m
D 2 i
Here
m
D
43
7. Controlled valves
7.1.
Preheating power
Due to Joule-Thomson effect, the pressure drop in control valve is accompanied with
temperature drop. To avoid dangerously low temperatures at the outlet of controlled valve, the
gas is being preheated before it enters the valve.
The temperature drop is calculated from integral Joule-Thomson effect:
T ( pi ) = Ti
dT
= JT ( p, T )
dp
To = T ( p o )
Here
pi
Ti
po
JT
T ( p)
To
Inlet pressure
Inlet temperature
Outlet pressure
Joule-Thomson coefficient
Solution of the initial-value problem
Outlet temperature
[Pa]
[K]
[Pa]
[K.bar-1]
[K]
P = m(T2 T1 )
c p ( p1,T 1 ) + c p ( p 2 ,T 2 )
2
Here
P
m
cp
p
T
Subscript 1
Subscript 2
7.2.
Preheating power
Mass flow rate of gas
Gas specific heat
Pressure
Temperature
Exchanger inlet condition
Exchanger outlet condition
[kW]
[kg.s-1]
[kJ.kg-1.K-1]
[Pa]
[K]
The valve in an intermediate position can be modeled as controlled valve element with
particular mode and setpoint equal to the opening percentage (using the SCVO scenario
parameter).
The flow rate through the valve in an intermediate position is a function of inlet pressure,
outlet pressure and the position of the valve. The valve position is expressed by opening
LIWACOM Informationstechnik GmbH
44
percentage (closed 0% and fully open 100%). An equation individual for each valve producer
& type describes this function:
m = m( pi , p o , o,K)
Here
m
pi
po
o
K
[kg.s-1]
[Pa]
[Pa]
[1]
Two such formulae are currently implemented in SIMONE; other valve characteristics can be
implemented upon request.
241F1 p1 Y 0.148Y 3
1.63
Y = min1.5;
F1
rT1 z1
p1 p 2
p1
Here
V0
CV 0
o
Volumetric flow rate at standard condition (101.325 kPa and 0C) [m3.h-1]
Sizing coefficient
[-]
Opening percentage, o 0,100
[1]
[1]
[1]
[bar]
[bar]
[K]
[1]
[1]
The valve behavior is therefore described by the sizing coefficient CV 0 and functions CVR (o ) ,
F1 (o ) (being defined by table values).
SIMONE Research Group, s.r.o.
45
Q0
514
0T1
( p1 p 2 ) p 2
p1
2
p2
K VS K VR (o ) =
Q0
257 p1
0T1
Here
K VS
o
Sizing coefficient
Opening percentage, o 0,100
[m3.h-1]
[1]
[1]
[m3.h-1]
[bar]
[bar]
[K]
[kg.m-3]
The valve behavior is therefore described by the sizing coefficient K VS and function K VR (o )
(being defined by table values).
7.3.
For the control valve element, the pressure loss in inlet and outlet parts of piping yard of the
control station can be modeled.
Local pressure loss is calculated using local pressure loss coefficient
1
pi p o = i wi2
2
46
Here
pi
po
wi
[Pa]
[Pa]
[1]
[kg.m-3]
[m.s-1]
The gas velocity is calculated using representative internal diameter of controlled valve D ,
entered in network element data:
wi =
4m
D 2 i
Here
m
D
47
8. Resistor
A resistor is an element representing the lumped resistance of an armature, partially closed
valve, filtering equipment etc. The hydraulic resistance is characterized by the local pressure
loss coefficient. The resulting pressure drop is described by the equation
1
pi p o = i wi2
2
Here
pi
po
wi
[Pa]
[Pa]
[1]
[kg.m-3]
[m.s-1]
The gas velocity is calculated using representative internal diameter of resistor D , entered in
network element data:
wi =
4m
D 2 i
Here
pi
Ti
po
JT
T ( p)
To
Inlet pressure
Inlet temperature
Outlet pressure
Joule-Thomson coefficient
Solution of the initial-value problem
Outlet temperature
[Pa]
[K]
[Pa]
[K.bar-1]
[K]
48
9. Solution methods
The set of original methods developed within the SIMONE project is what makes the power
and quality of SIMONE system.
As a rule, the description of a pipeline network consists of thousands of non-linear partial
differential equations and of hundreds of inequalities and constraints. In principle, the
requirements on a dynamic simulation model result from the great extent of reality being
described, and from the great complexity of its mathematical description:
1. The transients in network elements must be described with high accuracy. Otherwise,
due to the accumulation of errors in a great complex network, an unacceptable
inaccuracy may result.
2. High speed of simulation is required even for very extensive networks. This is
important to get high value of the ratio (real time)/(simulation time). The greater this
ratio, the wider the operation possibilities in practical, particularly on-line use.
Another essential benefit of the high speed of simulation will be appreciated in
building the (optimum) control system.
3. To bring down the memory demands is required, because due to the great number of
network elements and variables, they may exceed the possibilities of even powerful
computers.
4. In the network, both extremely slow transients and quick ones (after a rupture of a
tube or opening a valve) may occur. The model must simulate both with high accuracy
being simultaneously safe against any oscillations of the numeric solution.
5. The simulation model must copy the reality in all aspects. The model must therefore
simulate, besides the transients in pipes, the behavior of all other elements, namely
compressor stations, in full variety of their technological realization.
Besides all the technical reasons, there exists one good reason more why all these
requirements must be respected: the dispatcher will quickly abandon the system which will
not be quick, exact, user friendly, and will not allow him to do everything what can be done
with the real network and in a most similar manner.
To satisfy the above-mentioned requirements, SIMONE simulation kernel is based on the
following keystones:
Modified implicit integration method was developed to ensure safe numerical stability
of solution (item 4) in all transient conditions. It is safe against parametric oscillations,
which could rise due to the stepwise linearization of nonlinear equations, and exactly
maintains the mass balance condition.
The demands on small memory consumption and high speed of calculation (items 2
and 3) are satisfied by means of a proprietary decomposition method that the
simulation system is based on. This method takes advantage of sparseness of the
system matrix and of its specific structure.
Proprietary algorithmic procedure has been developed which allows the solution of
combined set of nonlinear partial differential equations (pipes) together with the
characteristics and constraints of non-pipe elements (compressor stations and
controlled valves). It satisfies the demand of realistic description of all aspects of
49
simulated reality (item 5) by an easy and flexible linkage of non-pipe elements into the
model of a pipeline network.
50
Input
System
err_2
err_1
real
state
estimated
state
Model
measurement
-
Controller
What is above dashed line is for us unknown. We have only Model of the System and input
and restricted number of measurements that are corrupted by errors. If Model is good
SIMONE Research Group, s.r.o.
51
approximation of System and Controller makes the closed loop stable then Observer (laying
bellow dashed line) is able to fulfill the task of state reconstruction.
The output of the dynamic model is compared with measured variables. The difference is
used by the controller that actuates on the model to minimize the difference i.e., to get the
state of the simulated system close to reality. The task is to find such a gain of the controller,
which corresponds to an optimum solution. If the optimum solutionis defined by the
minimum variance of state reconstruction error, the solution is known as Kalman filter.
10.1.
In the case of pipeline networks, such a solution is not practical due to the high dimension of
the system. Instead of an optimum, a sub-optimum solution combining the mathematical
approach with engineering aspects has been elaborated. The result is the real-time state
reconstruction.
The state reconstruction is performed in two subsequent steps:
1. In the first step, the simulation is calculated. Within the step, the corrections of intakes and off-takes are done in the nodes with pressure measurements. The correction
is done by the additional artificial off-takes, the quantity of which are stated on the
basis of pressure measurements considering the accuracy of pressure and flow rate
data.
Input pressure, output pressure, and flow rate are used to be measured for the
controlled elements (compressor stations and control valve). From these values the
special control mode is constructed to minimize for each controlled element the
objective
(p
pi
pi2
) + (p
2
p o
2
po
) + (m m ) + (m m )
2
m2
s2
pi , pi
p o , po
m , m
ms , s
52
2. In the second step, the distances between measured and estimated pressures are
minimized respecting the uncertainty of pressure drops on tube elements computed in
the first step. The minimized objective is
(P P )
* T
Here
P
P_
P*
P_
R P
R_1
RP
10.2.
The SIMONE concept is based on suppressing of outliers instead of omitting. The reason is to
avoid the possible loss of observability in case that all pressure measurements are omitted in a
sub-network. In SIMONE, the following method is used:
If the difference between the results of the first reconstruction step and the pressure
measurement is greater than rel p p , where rel p is coefficient and p is accuracy
of pressure measurement, then the difference value is locally used as modified
accuracy of the measurement.
If the difference between the first reconstruction step and flow rate measurement is
greater than rel m m , where rel m is coefficient and m is accuracy of flow rate
measurement, then this number is locally used as modified accuracy of the
measurement.
The results of this approach evidently depend on parameters rel p and rel m .
10.3.
relQP
53
Real number modifying artificial off-takes (these are defined in all supply/offtake nodes where pressure measurement is available) as
QP =
Here
Q * relQP
(p p* )
p
Q is accuracy of supply/off-take
p is accuracy of pressure measurement.
Coefficient reflecting uncertainty of pipe line pressure drops in the first step of
reconstruction (see matrix RP ).
.
Recommended value is 0.03;010
rel p
rel m
For starting reconstruction run use values of parameters near to upper bound and for
normal reconstruction run to lower bound.
10.4.
For scanning and checking the state reconstruction results, the following values can be
displayed by SIMONE graphic support:
nodename.phat
54
to database
The reason for using the mean value of flow rate since the last writing is the calculation of
mass balance under the circumstances of different computation time step and writing time
step.
There are two integrals in the graphic arithmetic: INTG for integrating by trapezoidal rule and
INTG2 for rectangular rule. To compute balances use INTG only for supplies and off-takes
where continuous linear interpolation between samples is assumed. For all other flow rates
use INTG2.
There is possible to use in topology visualization the statement tv qrec to have quick
look over the situation in network at one time instant.
Using subsystem TOTAL, the above data for the whole network are displayed.
Because gasses of different quality may stream within the network, the above subsystem
variables are expressed in volumetric units using the default gas defined in topoeditor.
55
11.1.
Objective Function
The objective function is represented as a sum of different penalty functions (for fuel
consumption, for price of purchased gas, for pressure condition, etc.). The components of
penalty functions can be classified into two groups:
Terms expressing the basic aims of optimization (fuel consumption, prices of gas etc.).
Terms substituting some constraints by penalty functions (e.g., for pressure and flow
rate conditions) to reach the feasibility.
If the components of the above two classes are merged together in one objective function, the
valley problem may emerge what decreases the speed of convergence. Therefore, we
recommend to use only the terms from the second group for finding a feasible solution, and
then apply the first group (completed by all constraints) for the optimization itself.
11.2.
Constraints
11.3.
The method is based on the generalized reduced gradient method and generalized projection
gradient method with the use of a steady-state simulation model.
The set of constraints can be divided into two basic groups: the equality constraints (group 1)
and the inequality ones (groups 2 and 3).
The constraints read after the local linearization at the working point x0 , for the first group:
Ax 0 x = a x 0
56
B x 0 x bx 0
The number of equality constraints can be great (it is determined by the network extent), but
the following transformation can simply be computed (using the knowledge of the structure of
Ax 0 ):
x = Dx0 y
where y is a choice of linearly independent variables from x and matrix D x 0 has a special
structure given by the configuration of the network. From this follows, that it is advantageous
to respect these constraints in the computation of a new step of optimization by gradient
reduction.
The inequality constraints will then be transformed to:
M x 0 y bx 0
M x0 = Bx0 Dx0
It should be noted that the vector y represents the set of control variables. The set points of
non-pipe elements need not necessarily correspond to real control variables used for the
optimization it is better to interpret them as a way how to describe the final optimum
solution (may be completed by some additional information about the pressure and flow rate
distribution).
The number of active constraints belonging to the second group is in comparison to the
number of constraints of the first group much smaller, but it changes within iterations. The
integration of these constraints in the first group would significantly complicate the
computation as well as the structure of the matrix D x 0 .Moreover, the procedure of
determining the D x 0 matrix would have to be repeated any time the number of active
constraints changes! For these reasons, the method of gradient projection was chosen to
respect the constraints. It enables to change the number of active constraints very simply. This
method also allows solving the problem of linearly dependent constraints (this problem does
not occur in the first group of constraints).
57
Equilibrium condition
This can be seen on the following picture showing the longitudinal profiles of pressure and
temperatures along a pipeline.
T<Td
Free water
present
Td(p,WC)
Th(p)
T
T<Td
T<Th
Gas
inlet
Hydrate formation
risk area
Gas
outlet
The water content of gas WC [g/m3 std. dry] is being tracked over the network (as a gas
quality parameter). For known water content and given pressure, the water dew point
temperature can be calculated anywhere in the network.
The hydrate equilibrium temperature calculation can take into account the effect of hydrate
inhibitor (methanol, glycols). For this purpose, the inhibitor content of gas IC [g/m3 std. dry]
is being tracked over the network.
58
12.2.
The water dew point temperature Td is calculated from the water content of gas
Td = Td ( p, WC , x ) .
Here
p
WC
x
Pressure
[Pa]
Water content of gas
[g/m3 std. dry]
Gas composition using molar fractions (dry gas) [1]
12.2.1.
Bukacek aprroximation
The equilibrium water content of gas is expressed as a function of pressure and temperature:
WC 0 =
A
+B
p
Here
WC 0
p
A,B
(
)
B = 0.0448739 exp(0.053570327t 0.00019866041t )
A = 4.9265573 exp 0.0737433t 0.00030673912t 2
t + 40C,+90C :
(
)
B = 0.0405058 exp(0.055042799t 0.00016553793t )
A = 6.1569611 exp 0.0620686t 0.00013096345t 2
t + 90C,+130C :
A = 6.1569611exp(0.0620686t 0.00013096345t 2 )
B = 0.0700737 exp(0.032637894t )
Bukacek approximation is suitable for sweet natural gases ( r 0.8 , xCO2 < 2% , x H 2 S < 1.5% ).
12.3.
59
Hydrate equilibrium
Th = Th ( p, x )
Here
p
x
Pressure
[Pa]
Gas composition using molar fractions (dry gas) [1]
12.3.1.
Motiee approximation
Simple relation involving gas relative density only (published in form of charts)
Th = Th ( p, r )
has been approximated by Motie using the equation
t
p
r
b1 Kb6
Temperature
[F]
Pressure
[psia]
Gas relative density [1]
Coefficients
i
1
2
3
4
5
6
bi
-238.24469
78.996674
-5.352544
349.473877
-150.854675
-27.604065
The original charts exhibit relatively high uncertainty, especially for gas relative density
around 0.6. However Motiee approximation seems to perform well for sweet, methane-rich
gas.
60
12.3.2.
Carson-Katz method
xi
K ( p, T ) = 1
i
vs ,i
Here
K vs ,i
xi
Subscript i
[1]
ln K vs ,i = Ai + Bi t + C i p + Di t 1 + Ei p 1 + Fi pt + Gi t 2 + H i p 2 + I i pt 1 + J i ln pt 1 +
Ki p
+ Li tp + M i t p + N i pt
2
+ Oi tp
3 2
+ Pi t + Qi p t
3
+ Ri t
Here
t
p
Ai K Ri
Temperature [F]
Pressure
[psia]
Coefficients
12.3.3.
Ponomarev approximation
Here
p
P0 , P1
r x
r =
x
i
Pressure
Coefficients dependent on reduced relative density r
[at]
[1]
[1]
[1]
[1]
ri
xi
Subscript i
61
P0
r
0,56 24,25
0,57 21,80
0,58 20,00
0,59 18,53
0,60 17,67
0,61 17,00
0,62 16,45
0,63 15,93
0,64 15,47
0,65 15,07
0,66 14,76
0,67 14,51
0,68 14,34
0,69 14,16
0,70 14,00
P1
77,4
70,2
64,2
59,5
56,1
53,6
51,6
50,0
48,6
47,6
46,9
46,2
45,6
45,0
44,5
12.3.4.
P0
r
0,71 13,85
0,72 13,72
0,73 13,57
0,74 13,44
0,75 13,32
0,76 13,20
0,77 13,08
0,78 12,97
0,79 12,85
0,80 12,74
0,81 12,62
0,82 12,50
0,83 12,40
0,84 12,28
0,85 12,18
P1
43,9
43,4
42,9
42,4
42,0
41,6
41,2
40,7
40,3
39,9
39,5
39,1
38,7
38,3
37,9
P0
r
0,86 12,07
0,87 11,97
0,88 11,87
0,89 11,77
0,90 11,66
0,91 11,57
0,92 11,47
0,93 11,37
0,94 11,27
0,95 11,17
0,96 11,10
0,97 11,00
0,98 10,92
0,99 10,85
1,00 10,77
P1
37,6
37,2
36,8
36,5
36,2
35,8
35,4
35,1
34,8
34,5
34,2
33,9
33,6
33,3
33,1
These models for hydrate equilibrium temperature calculation have been verified against the
codes CSMHYD (detailed thermodynamic model of hydrate equilibrium) and HYDK
(implementation of K vs -model) of Colorado School of Mines for the pressure range up to
about 100 bar (1500 psia) for three typical composition of natural gas:
Component [%mol.]
L-gas GroningenCH4-rich RussianH-gas North Sea
CH4
81,29
96,52
85,91
N2
14,32
0,26
1,00
CO2
0,89
0,60
1,49
C2H6
2,87
1,82
8,49
C3H8
0,38
0,46
2,31
H2O
0,00
0,00
0,00
H2S
0,00
0,00
0,00
H2
0,00
0,00
0,00
CO
0,00
0,00
0,00
O2
0,01
0,00
0,00
i-C4H10
0,00
0,10
0,35
n-C4H10
0,15
0,10
0,35
i-C5H12
0,00
0,05
0,05
n-C5H12
0,04
0,03
0,05
n-C6H14
0,05
0,07
0,00
Relative density r
0,645
0,581
0,650
Reduced relative density r
0,589
0,578
0,646
62
The results of CSMHYD, HYDK, Motiee, Ponomarev and our implementation of K vs -model
(referred below as HETKVS) are shown on the following pictures:
Hydrate equilibrium curve for "Groningen gas"
1000
CSMHYD p->T
CSMHYD T->p
p [bar]
100
CSM K-code
Motie
10
Ponomarev
HETKVS
1
-5
10
15
20
25
30
t [C]
p [bar]
100
CSM K-code
Motie
10
Ponomarev
HETKVS
1
-5
10
15
20
25
30
t [C]
p [bar]
100
10
1
-5
10
15
t [C]
20
25
30
63
The results of hydrate equilibrium temperature calculation for given pressure may be
summarized in the following points:
12.4.
Effect of inhibitors
Th =
X
1297
M 100 X
Here
Th
M
X
WC = WC 0 ( p, T , x ) + WC l
IC = MC v ( p, T )X + WC l
X =
IC
100
WC l + IC
X
100 X
for methanol
for glycols
Here
WC
WC 0
WC l
64
IC
Total (tracked) inhibitor content of gas
[g/m3 std. dry]
MC v ( p, T ) Methanol losses to the hydrocarbon phase
g/m 3std.
amount of methanol vaporized into gas per 1%
MC = 1, t , t b2
b
3
v
b4
b5
b6
b7 1
b8 log p
b9 log 2 p
Here
MC v ( p, T )
t
p
Temperature
Pressure
12.5.
Risk indicators
The results of the equilibrium analysis over pipe element are derived from the detailed
internal temperature profile used by the SIMONE Heat Dynamics module (see chapter 5). The
following parameters show, how far the local gas condition are from the equilibrium ones. For
each pipe element the worst case, i.e. minimum of these sub cooling values over the element
length, is stored:
Gas sub cooling with respect to the three-phase equilibrium boundary temperature:
DTHF = min (T min (Td , Th ))
[0; L ]
Here
65
Pipe length
[m]
Longitudinal coordinate [m]
Note: If the profiles of all temperatures along the pipe element are monotone (what the most
frequent case), these three indicators show only two different values:
DTHF = max(DTDP, DTHE )
However the monotonicity of all profiles along pipe element is generally not guaranteed
(assume for example the possibility of a batch of gas of different water content moving
inside the pipe element), so all three indicators have to be stored.