SPE 77502
Application of Transient Multiphase Compositional Tracking for Pipeline Flow Analysis
Allan Rydahl, SPE, Scandpower Petroleum Technology, Inc.
Copyright 2002, Society of Petroleum Engineers Inc.
This paper was prepared for presentation at the SPE Annual Technical Conference and
Exhibition held in San Antonio, Texas, 29 September–2 October 2002.
This paper was selected for presentation by an SPE Program Committee following review of
information contained in an abstract submitted by the author(s). Contents of the paper, as
presented, have not been reviewed by the Society of Petroleum Engineers and are subject to
correction by the author(s). The material, as presented, does not necessarily reflect any
position of the Society of Petroleum Engineers, its officers, or members. Papers presented at
SPE meetings are subject to publication review by Editorial Committees of the Society of
Petroleum Engineers. Electronic reproduction, distribution, or storage of any part of this paper
for commercial purposes without the written consent of the Society of Petroleum Engineers is
prohibited. Permission to reproduce in print is restricted to an abstract of not more than 300
words; illustrations may not be copied. The abstract must contain conspicuous
acknowledgment of where and by whom the paper was presented. Write Librarian, SPE, P.O.
Box 833836, Richardson, TX 75083-3836, U.S.A., fax 01-972-952-9435.
Abstract
This paper details the application of transient multiphase
compositional tracking in a pipeline flow simulation study.
Comparisons of results between the compositional modeling
approach and results obtained using a precalculated table-
based fluid property approach is presented. Both simulation
models are based on the dynamic two-fluid theory commonly
used to simulate transient multiphase flow in wells and
pipelines. When comparing the simulation results with
pipeline depressurization field data for one field case, the gas
outlet rate and pressure is simulated with greater accuracy
using the compositional tracking approach. For a second field
case the simulation results are comparable for gas outlet rates
and pressures although the compositional tracking overall is
better. For liquid volumes the results are of comparable
quality for both field cases.
Introduction
Transient multiphase flow in wellbores and pipeline systems is
typically analyzed using a dynamic two- or three-fluid
modeling technique (1.). To determine pressure drop,
temperature changes and flow regime, the model essentially
solves conservation equations for mass, momentum and
energy for the gas and liquid phase or phases as a function of
time. For hydrocarbon systems with water, the model can
handle water either as an integral part of the hydrocarbon
phase or as a separate liquid phase. The most important fluid
physical properties and transport properties that are required
by the model are density, viscosity, thermal conductivity and
heat capacity, surface tension and phase mass fractions.
The traditional representation of the required fluid
properties is through precalculated tabular values with an
upper and lower boundary for pressure (P) and temperature
(T). Based on an overall total fluid composition the properties
are tabulated at a number of (P, T) points. During the course
of the simulation the model interpolates in the properties table
as needed. This method is referred to as a table-based
approach. The primary purpose of using a table-based
approach is to save simulation time.
Recent advances of (2.) in the numerical handling of flash
calculations for multiphase flow simulations have enabled a
more rigorous simulation approach to be developed.
The approach, compositional tracking (3.), accurately takes
into account the fluid compositional changes in space and
time, and continuously calculates physical properties based on
the in-situ hydrocarbon and aqueous compositions. The
compositional tracking approach consequently does not have
the same limitations in application as the table-based
approach.
Potential application areas for the compositional tracking
approach are for simulations involving some degree of phase
separation of gas and liquid for example:
• Shut-in (liquid settles in low spots)
• Restart (potential slugging)
• Depressurization (gas removed from pipeline
potentially together with liquid slugs)
• Slugging (alternate flow of gas and liquid)
• Mixing of fluids from pipelines that individually
are in a transient state resulting in varying
transient state in the merge pipeline
• Hot oil circulation and well start-up
Whereas it is intuitively obvious that a compositional
modeling approach will allow a more accurate fluid
description in a system with varying mixing of different fluids
from different pipelines with time, it is less obvious that a
compositional model is required for a simpler system
consisting of, for instance, a single pipeline. Even if there are
recognized compositional effects during slugging, is it enough
to justify the more rigorous compositional approach
considering the uncertainty in the calculation of the properties
of the slug flow regime (4.) in the first place? Even if there are
compositional effects during the depressurization of a
pipeline, is it enough to justify the compositional tracking
approach to be applied for all such simulations or should it
only be mandated when certain extreme conditions apply?
2 ALLAN RYDAHL
Simulations with Table-based Approach
When pre-calculating the physical properties as a function of P
and T for one overall total fluid composition, one essentially
builds any compositional dependency directly into the table. If
the composition changes, the compositional effect on any fluid
property will be incorrectly calculated. But how much does
fluid properties change with composition? What is the
compositional effect on the fluid properties required in
transient multiphase flow simulation?
Phase behavior – an example
Consider an example where a well is producing the fluid
composition listed as “Char Fluid” in Table 3 into a pipeline at
800 psia and 80F. Further, assume that the well is slugging
resulting in the relative amounts of liquid and gas to vary with
time. The “Oil” and “Gas” compositions in equilibrium at the
inlet conditions are also listed in Table 3.
In the table-based approach a pre-calculated table of
physical properties would be based on the overall composition
for an average gas-oil flow ratio. In Figure 1 the liquid density
calculated from the total composition – “Char Fluid” is
compared to the “Oil” density calculated using the “Oil” phase
composition only. Isocurves are introduced to illustrate the
difference in density in percent as a function of pressure and
temperature.
At 800 psia and 80F the density match is perfect. The
question is, how much does the density change in the pressure
and temperature region of interest? The “Char Fluid”
composition is the same as used in the shut-in and
depressurization case described elsewhere in this paper, and a
good assumption of minimum temperature during a shut-in is
thus ambient temperature (60F). From Figure 1 it can be
derived that, had the production into the pipeline at the time
shortly before a shut-in been dominated by “Oil” only, the
difference in the liquid density at the end of a shut-in would be
in the order of 0-5% (assuming the shut-in conditions are ~720
psia and ~60F). Had the pipeline been subsea the ambient
temperature would have been ~40F and the difference would
be in the order of 5-10%. For a depressurization case the
temperatures may get even colder and it may be relevant to
include a wider P, T area.
In Figure 2 the gas density calculated from the total
composition – “Char Fluid” is compared to the “Gas” density
calculated using the oil phase composition only. Isocurves are
introduced to illustrate the difference in density in percent. It
is evident that the difference in gas density is not great for
these compositions at any P, T point. Almost throughout the P,
T range the difference is between 0-3%.
In summary, even if the overall composition of the fluid
changes as a function of time, the physical properties for a
given phase at a given (P, T) point may remain fairly constant.
Phase behavior – another example
In the dynamic two-fluid model, the phase fractions
computed during a simulation are normally based on the initial
gas mass fraction. When building a pre-calculated fluid
properties table based on an overall composition, a gas mass
SPE 77502
fraction at all (P, T) points is included in the table. However,
it is very common to modify the table-defined default gas
mass fraction to a different gas mass fraction. There are
several situations where this is common practice for instance
when simulating the effect of a change in gas-oil ratio with
time, when mixing multiple production streams or when
simulating gas lift scenarios. How is this possible? The mass
transfer is handled through derivatives only (1.) and thus
requires only an initial value. Once initiated only the relative
change in the phase fractions are computed.
An important implicit assumption is made when the gas
mass fraction is set different from the one calculated by the
overall composition: it is implicitly assumed that all inner
lines, or quality lines, in the phase diagram are equidistant. In
Figure 3, the phase diagram for “Char Fluid” is shown with
some inner quality (volume) lines in the range from 0.95 to 1
(saturation line). A phase diagram with quality lines is a great
tool to evaluate the potential transfer of mass between phases
before a simulation is performed. Estimating the operating
pressures and temperatures, it is easy to read in the diagram to
get an overview of the likely phase mass transfer as the
production passes through the pipeline.
Before initiating transient multiphase flow simulations, it
is worthwhile to investigate the phase behavior of the fluid to
better understand the compositional effects in the relevant
pressure and temperature region.
Case study: Shut-in and depressurization
The following example illustrates some similarities and
differences in simulation results for a typical shut-in and
depressurization test. The field study is previously described
in more detail in (5.).
In brief, the system is a 5.3 km, 3.688-inch pipeline
producing a lean gas condensate. The composition is given in
Table 1. At the outlet a separator is separating the produced
liquid and gas. Metering systems at various places measures
pressures and gas flow rates. Estimates of volumes are
presented based on partly measured, partly estimated values.
The field test data consisted of two depressurization tests
through a 0.5-inch and a 1.0-inch valve, respectively. The
system was previously simulated using the table-based
approach and compared to field test data. Since the original
simulation model was not available, a new model was build
for the purpose of this paper. Minor differences are seen in the
results with this table-based simulation model compared to
that of the original study. Some of the differences may be due
to updates in the overall simulation model; other may be the
result of a different simulation setup.
The original composition was tuned with the Soave-
Redlich-Kwong equation of state (6.) to match two saturation
points as found in Table 2. After regression the saturation
points match exactly. The resulting composition including
pseudo fractions is found in Table 3.
Steady state results
In Figure 4 and Figure 5, the steady state results when
using the table-based approach and the compositional tracking
SPE 77502 APPLICATION OF TRANSIENT MULTIPHASE COMPOSITIONAL TRACKING FOR PIPELINE FLOW ANALYSIS
approach are presented. Very little discrepancy is seen in
simulated inlet pressure, outlet temperature and liquid holdup.
The liquid content in the pipeline is in both cases ~38 bbl
which compares reasonably to an estimated 40-55 bbl from the
field test.
Shut-in results
In Figure 6 and Figure 7, the shut-in results when using the
table-based approach and the compositional tracking approach
are presented. It is seen that the temperature profiles are
identical, but the pressure profile is slightly different – about 3
psia. This might be due to different in-situ densities of the
settling liquid phase in the pipeline low spots as previously
discussed. The holdup profiles are essentially the same.
Depressurization through 0.5-inch valve
In Figure 8 and Figure 9, the depressurization results for
the 0.5-inch valve when using the table-based approach and
the compositional tracking approach are presented. The plots
show the difference between the simulation results and the
field data in percent deviation as a function of depressurization
time. The results for the gas flow show that the table-based
approach seems to first overestimate then underestimate the
measured values whereas the compositional tracking model
seems to overestimate the gas flow out at all points. The
maximum deviation between simulations and measurements
for the table-based approach is -35%, and for the
compositional tracking approach only 15%.
The inlet pressure is more accurately simulated with the
compositional tracking approach with a maximum error of less
than 8% whereas the table-based approach show a maximum
difference as high as –25%.
In Table 4 the simulated liquid volume outlet flow is
compared to the measured volume. The two approaches give
very similar results.
Depressurization through 1.0-inch valve
In Figure 10 and Figure 11, the depressurization results for
the 1.0-inch valve when using the table-based approach and
the compositional tracking approach are presented. The plots
show the difference between the simulation results and the
field data in percent deviation as a function of depressurization
time. The results show that the compositional tracking and
table-based approach seem to follow the same trends.
However, the compositional tracking outperforms the table-
based model for almost all data points. Both approaches in
general underestimate the gas flow out as well as the inlet
pressure during the depressurization.
In Table 5 the simulated liquid volume outlet flow is
compared to the measured volume. Also the remaining liquid
volume is compared to field data. The two approaches give
similar overall results.
3
Conclusion
Transient multiphase flow compositional tracking simulations
were successfully applied to simulate two pipeline
depressurization tests. In compositional tracking, the fluid
properties are calculated continuously during the simulation
using the in-situ composition The simulation results were
compared to the field data and to a traditional table-based
approach in which fluid properties are pre-calculated based on
an overall composition.
The compositional tracking approach outperforms the
table-based approach especially for the 0.5-inch case. For the
1-inch case where the trend in the results is similar, the
compositional tracking approach still is more accurate.
The advantages of a compositional tracking model are very
evident from a theoretical point of view, but the results in this
particular case are not unambiguous considering measurement
uncertainty. It is likely that the compositional effects are small
in this case, which is probably due to the relatively low
maximum pressure in the system. With small pressure changes
only small errors may be expected in the fluid properties
calculations between the two approaches. It must be
anticipated that larger differences between the approaches will
be seen in systems with higher pressures.
Acknowledgements
The author would like to thank Calsep Inc. for allowing access
to a simulation tool able to handle the PVT issues described in
this paper.
References
1) Bendiksen, K.H., Malnes, D., Moe, R & Nuland, S: “The
Dynamic Two-fluid Model OLGA: Theory and
Application”, SPE 19451, SPE Prod. Eng. May 1992, 171-
180.
2) Rasmussen, C. and Michelsen, M., Submitted to OTC
2003.
3) Xu, Z.G., Bjurstroem, K.E. and Rasmussen, C.P.: “A
Compositional Model for Transient Simulation in
Pipelines”, 3rd North American Conf. On Multiphase
Technology, Banff, Canada, June 6-7, 2002, BHR Group.
4) Nossen, J., Shea, R. and Rasmussen J.: “New
Developments in Flow Modelling and Field Data
verification”, 2nd North American Conf. On Multiphase
Technology, Banff, Canada, June 21-23, 2000, BHR
Group.
5) Shoup, G., Xiao, J.J. and Romma, J.O.: “Multiphase
pipeline blowdown simulation and comparison to field
data”, Multiphase Technology #31, BHR Group, Banff
June 1998.
6) Soave, G., Equilibrium constants from a modified Redlich-
Kwong equation of state, Chem. Eng. Sci., 27, pp. 1197-
1203, (1972).
4 ALLAN RYDAHL SPE 77502

Table 1 Input molar composition Table 3 Characterized production fluid with pseudo fraction
definition and equilibrium gas and liquid compositions obtained
when flashing at 800 psia and 80 F.
Mole% Molweight Density
[-] [g/mol] [g/cm3] Gas
N2 0.442 28.014 Char. @ 800 psia, Oil
Fluid 80F @ 800 psia, 80F
CO2 1.476 44.01
[Mole%] [Mole%] [Mole%]
C1 78.503 16.043 N2 0.442 0.450 0.038
C2 13.501 30.07 CO2 1.476 1.488 0.858
C3 3.551 44.097 C1 78.505 79.695 19.338
iC4 0.376 58.124 C2 13.501 13.493 13.915
C3 3.551 3.427 9.753
C4 0.663 58.124
iC4 0.376 0.345 1.933
iC5 0.233 72.151 C4 0.663 0.586 4.522
C5 0.194 72.151 iC5 0.233 0.180 2.874
C6 0.209 86.178 0.664 C5 0.194 0.141 2.854
C7 0.255 96 0.738 C6 0.209 0.106 5.307
C8 0.172 107 0.765 C7 0.256 0.057 10.112
C9 0.116 121 0.781 C8 0.172 0.022 7.646
C10 0.073 134 0.792 C9 0.116 0.007 5.555
C11 0.044 147 0.796 C10 0.073 0.002 3.594
C12 0.03 161 0.81 C11 0.044 0.001 2.218
C13 0.022 175 0.825 C12 0.030 0.000 1.511
C14 0.016 190 0.836 C13 - C14 0.038 0.000 1.940
C15 0.012 206 0.842 C15 - C18 0.034 0.000 1.721
C16 0.009 222 0.849 C19 - C20 0.029 0.000 1.459
C17 0.007 237 0.845 C21 0.017 0.000 0.872
C18 0.006 251 0.848 C22 - C23 0.020 0.000 1.026
C19 0.004 263 0.858 C24 - C49 0.019 0.000 0.954
C20 0.083 300 0.910
Table 2 Saturation points
Pressure Temperature
[psia] [F]
4253 50
4406 90
SPE 77502 APPLICATION OF TRANSIENT MULTIPHASE COMPOSITIONAL TRACKING FOR PIPELINE FLOW ANALYSIS 5

Char Fluid vs Oil density difference

1000

859
0.00-5.00
-5.00-0.00
719
-10.00--5.00
-15.00--10.00

Pressure (psia)
-20.00--15.00
578
-25.00--20.00
-30.00--25.00
-35.00--30.00 437
-40.00--35.00
-45.00--40.00
-50.00--45.00 296
-55.00--50.00
-60.00--55.00
-65.00--60.00 156
-70.00--65.00

15
-40 -23 -6 11 29 46 63 80
Temperature (F)

Figure 1: Density difference between “Char Fluid” and “Oil”.

Isocurves illustrates the change in difference as function of pressure and temperature

Char Fluid vs Gas density difference

1000

859

719
Pressure (psia)

9.00-12.00 578
6.00-9.00
3.00-6.00
0.00-3.00 437
-3.00-0.00
-6.00--3.00
296

156

15
-40 -23 -6 11 29 46 63 80
Temperature (F)

Figure 2: Density difference between “Char Fluid” and “Gas”.

Isocurves illustrates the change in difference as function of pressure and temperature
6 ALLAN RYDAHL SPE 77502

Char Fluid - Phase diagram

4500

4000

3500

3000
Vap/liq volume frac 1.000
Vap/liq volume frac 0.990
2500 Vap/liq volume frac 0.980
Vap/liq volume frac 0.970
Vap/liq volume frac 0.960
2000 Vap/liq volume frac 0.950

1500

1000

500

0
-100 0 100 200 300 400 500
Temperature/°F

Figure 3: Phase diagram “Char Fluid” with inner quality volume lines
SPE 77502 APPLICATION OF TRANSIENT MULTIPHASE COMPOSITIONAL TRACKING FOR PIPELINE FLOW ANALYSIS 7

Tre nd data
Inlet pres s u re TB [ps ia ]
Inlet pres s u re C T [ps ia ]
Outlet tem pe ra tu re TB [F]
Outlet tem pe ra tu re C T [F]
755 62
754.5 61.99
61.98
754
61.97
753.5 61.96
psia

753 61.95
F

752.5 61.94
61.93
752
61.92
751.5 61.91
751 61.9
0 100 200 300 400 500 600
T ime [m]
Figure 4: Steady state results. Inlet pressure and outlet temperature.

Profile data
H o ldu p TB
H o ldu p C T
0.5
0.45
0.4
0.35
0.3
0.25
-

0.2
0.15
0.1
0.05
0
0 0.5 1 1.5 2 2.5 3 3.5 4
Le ngth [mile s]

Figure 5: Steady state results. Pipeline holdup profile.

8 ALLAN RYDAHL SPE 77502

Shut-in afte r 2 hours

P res s ure TB
Te m perature TB
Te m perature C T
P res s ure C T
750 80
745
740 75
735
psia

730 70
F

725
720 65
715
710 60
0 1 2 3 4
Le ngth [mile s]
Figure 6: Shut-in results. Comparison of pressure and temperature profiles for
table-based (TB) and compositional tracking (CT) approach

Shut-in 2 hours
H oldu p TB
H oldu p C T
1
0.9
0.8
0.7
0.6
0.5
-

0.4
0.3
0.2
0.1
0
0 0.5 1 1.5 2 2.5 3 3.5 4
Length [mile s]
Figure 7: Shut-in results. Comparison of holdup profile for
table-based (TB) and compositional tracking (CT) approach
SPE 77502 APPLICATION OF TRANSIENT MULTIPHASE COMPOSITIONAL TRACKING FOR PIPELINE FLOW ANALYSIS 9

Gas outlet rate

20

10

0
Deviation %

-10
TB
-20
CT
-30

-40
0 10 20 30 40 50 60
Time (min)
Figure 8: Difference in gas outlet flow rate for 0.5-inch depressurization case using TB and CT approach compared to field data

Inlet pressure - site 1

5
0
-5
Deviation %

-10

-15
TB
-20
CT
-25

-30
0 10 20 30 40 50 60
Time (min)

Figure 9: Difference in inlet pressure for 0.5-inch depressurization case using TB and CT approach compared to field data

0.5 inch case Measured TB CT

[Std. bbl] [Std. bbl] [Std. bbl]
Liquid volume 24 25 24
remaining
Liquid volume 7 6 6
outlet
Table 4: Absolute liquid volumes for 0.5-inch depressurization case using TB and CT
approach compared to field data
10 ALLAN RYDAHL SPE 77502

Gas outlet rate

40
30
20
TB
10
CT
Deviation (%)

0
-10
-20
-30
-40
-50
-60
0 5 10 15 20 25
Time (min)

Figure 10: Difference in gas outlet flow rate for 1.0-inch depressurization case using TB and CT approach compared to field data

Inlet pressure - site 1

10

0 TB
CT
-10
Deviation %

-20

-30

-40

-50
0 5 10 15 20 25
Time (min)

Figure 11: Difference in inlet pressure for 1.0-inch depressurization case using TB and CT approach compared to field data

1.0 inch case Measured TB CT

[Std. bbl] [Std. bbl] [Std. bbl]
Liquid volume 22 18 18
remaining
Liquid volume 10 11 11
outlet
Table 5: Absolute liquid volumes for 1.0-inch depressurization case using TB and CT
approach compared to field data