Slugging in large diameter pipelines: field measurements, experiments and simulation
Abstract
This work has been performed primarily utilizing field measurement holdup and pressure data provided by
Saudi Aramco Oil Company. The holdup data was obtained using gamma measurement technique and
simulated using two slug prediction models; slug capturing in LedaFlow and slug tracking in OLGA.
The work presents a unique set of large diameter, high pressure data which can be utilized to develop
and/or improve multiphase flow simulation tools for industrial oil and gas applications. The data were
presented to the OLGA Verification and Improvement Project (OVIP) as part of Saudi Aramco contribution
to the improvement program.
Recent OVIP reports which were published by OLGA development team in the 3rd and 4th quarters of 2014
confirmed the author conclusion which indicates a significant over-prediction of the pipelines pressure
drop when slug tracking is enabled in most of the cases. The OLGA development team explored this issue
and reported a critical flaw in the slug tracking module which is currently being tackled by the team in
order to resolve it in a timely manner. A temporary solution was implemented which improved the results
as per the published reports, however a more thorough analysis is required before any final solution is
implemented. The discovery of this major flaw in OLGA slug capturing module proves the importance of
field measurement data in developing accurate multiphase flow simulation models.
LedaFlow on the other hand required the use of a fine mesh in order to obtain good results using the slug
capturing module. In most of the cases, a fine mesh which was constructed using (10 * Diameter) was
found to provide reasonable results. A finer mesh could have been used but that would require a
significant increase in computational time, especially for long pipeline cases which sometimes exceeded
(50) Km.
In addition to the simulation work, a detailed statistical analysis was carried out to extract the most
important information from the holdup data. This mostly included information about the slugs lengths, tail
and front velocities, amplitude and frequency. The data was obtained using a Matlab code which was
customized specifically for each case.
The analyzed data consists of two main parts:
•The first part, Field-A, was obtained by Mikal Espedal in 1999 as part of a field measurement project.
The work was focused on extremely high hilly terrain multiphase flow pipelines which were 16 inch and
20 inch in diameter. These pipelines were operated in a very high pressure mode, 30 to 60 bar. The
high hilly terrain, 100 meter high hills, implies that the dominant slugging mechanism is terrainslugging
and as such accurate predictions of pressure drop and pressure cycles were expected from the
simulation tools. However, the combination of terrain slugging with choked valves introduced a new
phenomenon which provided an interesting and challenging simulation problem. The simulation work
indicates a negative impact of the choke on most of the slugging cases.
This is counter-intuitive and has not been addressed by any researcher in large diameter oil and gas
pipelines prior to this work. Most of the earlier work on choke-slug interactions indicates a positive
impact on slug reduction/elimination in offshore pipeline-riser systems. However, the present work
indicates that with terrain-induced slugs, the slugging behavior is amplified and a careful investigation
of the pipeline system is required before any installation of choked valves. The new phenomenon was further confirmed by the experimental work carried out at NTNU multiphase flow lab. The lab
experiments indicate that the additional force created by the choked valves does not help to overcome
the compressibility force that exists in the pipeline system. As a result, slugs are still created in the
pipeline system and they are amplified by the accumulation of liquid at the tight valve opening. This
liquid accumulation compresses the gas upstream and causes a further pressurizing of the system,
which eventually leads to larger pressure fluctuations.
In addition, OLGA simulation results indicated that, with exception of one pipeline, G2NT1, the nonslug
tracking option provided the best results as expected in the cases were large terrain-induced slugs
dominate the multiphase flow behavior. Utilizing the slug tracking option in OLGA did not improve the
results and in some cases caused larger deviations from the field measurements. Slug tracking also
was not able to correctly predict any of the small hydrodynamic slugs measured during holdup
measurements.
On the other hand, LedaFlow simulation results provided similar results to OLGA as expected. Both
multiphase flow codes utilize the same unit cell model to predict large terrain-induced slugs with
possibly minor differences in the correlations used in each code. The non-slug capturing cases
provided the best results with LedaFlow and the use of slug capturing module did not significantly
improve the results as these pipelines where mostly exhibiting large terrain slugging. Slug capturing
with relatively fine grid, (5) meters sections, was evaluated to simulate the small hydrodynamic slugs
observed between the large slugs. However, LedaFlow could not correctly predict these
hydrodynamics slugs using the utilized fine mesh.
In general OLGA and LedaFlow produced large deviations in average pressure drops which ranged
between 15% and 25%, especially in the longer pipelines, G3ST1 pipeline (18) Km and G3ST2 pipeline
(20) Km. On the other hand, the frequency and amplitude of slugs were better predicted by both
multiphase flow transient codes. The deviations cannot be solely attributed to the multiphase flow
simulation codes as uncertainties with field measurements exist and particularly the ones related to
fluid flow rates, which were estimated based on historical well production data.
•The second part, field measurement of Field-B, C, D and E, was obtained by the author and members of
the flow assurance team at Saudi Aramco in 2012, with the help of a gamma-licensed contractor,
Tracerco Ltd., who carried out the physical gamma measurement based on the project requirements
set by Saudi Aramco. The work covered variety of onshore and offshore pipelines with very large
diameters that ranges from 24 inch to 42 inch. The analysis of the data was performed by the author
and are presented in the present work. Although slugging was not experienced in most of these
pipelines, a different set of challenges were encountered in each one of them as explained separately
in each chapter. The hydrodynamic-slugs initiated differently by the simulation codes OLGA and
LedaFlow, seem to have a significant impact on the pressure drop results obtained in each case and as
such should be used with caution when simulating multiphase flow pipelines.
OLGA simulation indicated an overall good agreement with pressure and holdup field measurements
when slug tracking option was disabled. A significant over-prediction in pressure drop was observed in
OLGA predictions when slug tracking option was enabled, as was also noticed by OLGA development
team in their latest reports. However, there was an exception for this overall good agreement which
was observed at Field-E TL-12. At TL-12 pipeline, the oil wells where connected to the trunkline over a
long distance along the pipeline, which created a situation where the flow rates were considerably low
at the start of the pipeline. The superficial gas and liquid velocities ranges from 0.15 m/sec and 0.14 m/sec at the beginning of the pipeline to 10 m/sec and 0.75 m/sec at the end of the pipeline,
respectively. The OLGA pressure under-prediction in the non-slug tracking case was approximately
(75) psi, which is equivalent to 20% error.
LedaFlow on the other hand showed a similar behavior to OLGA predictions without slug capturing.
When slug capturing is enabled, then a fine grid is required to obtain accurate predictions. The slug
capturing generally provided better results in terms of pressure drop and holdup if a fine grid is
utilized. The one exception where LedaFlow did not perform very well was the same Field-E TL-12
remote header case where OLGA also did not perform very well. In that particular case, both options
in LedaFlow, the slug capturing and non-slug capturing cases proved equally poor with pressure underpredictions
of approximately (60) psi which is equivalent to approximately 16% error.
The large pressure discrepancy noted for Field-E TL-12 pipeline case could partially be attributed to the
uncertainties with pipeline profile which is closely related to the static head pressure component
known to be the dominant phenomena in low multiphase flow rates cases. However, a large pressure
discrepancy was also noted about two years ago in a 36 inch pipeline which was operating in the same
area. OLGA simulation at that time was carried out using an approximated pipeline profile and
resulted in a pressure under-prediction of more than 100%. When using a more accurate pipeline
profile, the pressure predictions where greatly improved but still showed significant under-prediction
in pressure of approximately 35%. Therefore, there seems to be a general tendency for multiphase
simulation packages to under-predict pressure drop in very low flow rates pipelines such as TL-12.
Finally, the work included a thorough analysis of the gamma measurement technique. It showed how a
careful and deep understanding of the methodology is required in order to obtain accurate holdup results
especially in a field measurement environment were uncertainties are normally higher than the controlled
experimental laboratory environment. In particular, there was a focus on the gamma calibration
techniques and the unavailability of calibration pipeline spools with the same wall thickness during the
field measurements. This issue created a great challenge that had to be pursued carefully by correcting for
the additional wall thickness in order to obtain the right mixture density.