Multiphase Flow Simulation

Optimizing Field Productivity

Intan Azian Binti Abd Aziz

Petronus Carigali Sdn Bhd
Kuala Lumpur, Malaysia
Ivar Brandt
Oslo, Norway
Dayal Gunasekera
Abingdon, England
Bjarte Hatveit
Kjetil Havre
Gjermund Weisz
Zheng Gang Xu
Kjeller, Norway
Steve Nas
Kuala Lumpur, Malaysia
Knut Erik Spilling
Sandvika, Norway
Ryosuke Yokote
Eni Australia
Perth, Western Australia, Australia
Shanhong Song
Chevron Project Resources Company
Beijing, China
Oileld Review 27, no. 1 (May 2015).
Copyright 2015 Schlumberger.
For help in preparation of this article, thanks to Bjrn-Tore
Annsen and Lars Magnus Nordeide, Bergen, Norway;
Dag Biberg, Kevin Andre Hermansen, Norbert Hoyer,
Bin Hu and Hans Marius With, Oslo, Norway; Rajesh Puri,
London, England; and Mack Shippen and Steve Smith,
Houston, Texas, USA.
Drillbench, OLGA, OVIP and PIPESIM are marks of
Schlumberger.

26

As oil and gas well construction and eld development become more complex, the
need for more sophisticated ow simulation methods increases. New generations of
multiphase ow simulation tools are helping operators construct wells, pipelines and
processing facilities safely and efciently and optimize long-term eld production at
minimal risk and maximal prot.

Increasingly sophisticated ow simulation models have been developed to meet the needs of
operators as they open new frontiers. These models are vital for helping drilling engineers overcome well design challenges and production and
facilities engineers to understand and anticipate
ow conditions as they seek to extract hydrocarbons from deeper, more remote and geologically
complex reservoirs.
Flow simulation is a well-established means
by which engineers approximate the multiphase
ow behavior in a well, production system or
pipeline. Using mathematical models built into
specialized software programs, ow simulations
yield representations of the steady-state and
transient ow of oil, gas and water that might be
encountered in a real-world network of wells,
owlines, pipelines and process equipment. The
output of these simulations guides operators
eld development decisions in determining the
number of wells to drill, the location of such wells

and how to complete each well to ensure optimal

long-term production from the eld.
Multiphase simulations predict ow behavior
at all stages in the life of a well and eld, from
drilling to downhole production to network to processing facilities.1 For example, simulations may
guide well control design and engineering decisions by helping understand the effects of gas
inux in HP/HT wells.2 Another area is planning
for reservoir sections prone to lost circulation or
kick events where managed pressure drilling
(MPD) may be the best option for development.3
Flow simulation is also a useful tool in developing contingency plans in case of well blowouts,
during which reservoir uids ow into the wellbore in an uncontrolled manner and may reach
the surface. Well control companies and operators have used ow simulations to understand the
expected ow rates during a blowout, information that is then used to calculate the volumes
and densities of well kill uids as well as the

1. Edwards DA, Gunasekera D, Morris J, Shaw G, Shaw K,

Walsh D, Fjerstad PA, Kikani J, Franco J, Hoang V and
Quettier L: Reservoir Simulation: Keeping Pace with
Oileld Complexity, Oileld Review 23, no. 4
(Winter 2011/2012): 415.
2. For more on lost circulation events: Cook J, Growcock F,
Guo Q, Hodder M and van Oort E: Stabilizing the
Wellbore to Prevent Lost Circulation, Oileld Review 23,
no. 4 (Winter 2011/2012): 2635.

3. Managed pressure drilling uses ow control devices to

precisely control the annular pressure prole throughout
the wellbore. Managed pressure drilling techniques are
commonly used to maintain wellbore control during
drilling by managing kicks or preventing an ingress of
drilling uids into the reservoir.

Oileld Review

May 2015

27

Tubing pressure gauge

Ptbg = BHP waterhead + P 1.
Annular pressure gauge
Pann = BHP P hyd + P 1.

Relief well

No control,
waterhead pressure = 0

Blowout well

Drillstring

Fracture pressure

BHP = waterhead pressure +

Phyd + P 1.

Reservoir
pressure

> Flow simulation for well control. To regain control of a blowout, operators
often use a dynamic kill operation. Well control specialists kill the well using a
uid density that will contain the well but not fracture the formation. While
keeping the annulus and the drillstring of the relief well lled with uid, the
BHPmonitored through annular and tubing pressure gaugesis controlled
through the uid ow rate into the relief well. Flowing frictional pressure
supplements the hydrostatic pressure of the kill uid injected through the relief
well (red arrows) and up the blowout well (blue arrows). Because these
operations include produced uids and kill uids, they can be modeled with
multiphase ow simulators. The Pann is annular pressure, Phyd is the
hydrostatic pressure created by the kill uid, P 1 is the frictional pressure drop
caused by ow in the annulus of the blowout well, Ptbg is the tubing pressure
in the relief well drillstring and waterhead pressure is the pressure exerted by
the weight of a column of water from surface to the relief point.

28

pumping rates required to bring the well back

under control (left). In addition, capping operations can be reviewed because the ow simulation
includes realistic pressures and temperatures for
situations in which a capping stack is used to
control a blowout.4
Flow simulations help engineers optimize the
design and operation of producing wells. Models
provide insight into well completion designs,
including choices about inow control methods,
well trajectory design, sand control and articial
lift.5 Production engineers use ow simulations to
estimate how producing layers of the reservoir
contribute to the total well production. They can
then use this information to determine how to
operate the wells for optimal recovery.
Simulation models are also used to optimize
operations across an entire oil and gas eld.
Design engineers use ow simulations during the
concept, front-end engineering and design and
detailed design phases to guide decisions on sizing and materials selection for piping, valves, vessels and processing facilities. Models, which can
also estimate the risks of hydrate and wax formation in the production system, guide the selection
of optimal chemical inhibition methods as well as
thermal control systems in the forms of insulation, bundling and heating. Flow simulations provide insight that systems designers use to counter
corrosion and erosion in pipeline transmission
and processing systems.
Production engineers implement ow simulation models to establish procedures for operational events such as pipeline startup, shut down
and blowdown, production rate changes, optimal
process equipment usage, pipeline pigging and
network debottlenecking. Model output guides
normal operational procedures for these events
and highlights safe operating limits, which can
be used to develop emergency procedures and
contingency plans.
Flow simulations may play a role in operator
training programs. Simulation models help operations personnel become familiar with initial
startup procedures and ow assurance considerations for new production systems. Simulations
also give less experienced personnel a means of
practicing safe processing equipment operation
and to run numerous what-if scenarios prior to
working in real-world operations.
This article describes the evolution of ow
simulation methodologies with emphasis on
advances in the simulation of upstream and midstream transient multiphase ow in wells and
pipeline networks. A brief history details how
ow simulators evolved from those that modeled

Oileld Review

on

Pressure information
Equation
of State

erg

yi

nfo

rm
ati

on

Energy
Equation

En

F lo

on
a ti
rm

in f

o
inf

orm
ati

Momentum
Equation

Flo

on
ati
rm

May 2015

Mass
Equation
nfo
si

A Brief History
The vast majority of produced uids do not
come to the surface in a steady, single-phase
stream. Rather, production is a complex and
ever-changing combination of hydrocarbon uids
and gases, water and solids owing together at
nonuniform rates.
The basis for multiphase ow design and
operation is uid dynamics.6 The driving force
behind the earliest oil industry simulation tools
was multiphase ow system designers need for
accurate estimates of the pressure, temperature
and liquid fractions in wells and along pipelines.
One fundamental approach to modeling ow
behavior in oil and gas systems is the two-uid
model, in which designers assume only two uid
phasestypically a liquid and a gasare present.7 Other models extend this treatment to
include uids that coexist in more than two
phases such as a gas, oil and water phase.
Separate phases can ow in a pipeline in three
stratied, continuous layersa gas layer on top,
an oil layer in the middle and a water layer at the
bottom of a pipeline. A phase can ow in each of
the three layers. For example, some of the gas is
transported through the pipeline in the upper gas
layer, while the rest is transported as gas bubbles
dispersed in the oil and water layers.
The multiuid model consists of mass,
momentum and energy conservation equations.8
Often, mass conservation equations are written
for each phase. Momentum conservation equations are written for each of the continuous
layers, whereas energy equations can be written for the total uid mixture or for each of
the layers. In the case of a two-phase, two-layer
ow model, a total of six differential equations
are written.

Solving this set of equations requires development of closure laws, which are necessary relations that must be added to the conservation
equations to allow their calculation (right). One
basic closure law is the equation of state of the
uid, which is a thermodynamic equation that
provides a mathematical relationship between
uid properties, such as density and viscosity, to
two or more state functions; state functions
include temperature, pressure, volume or internal energy associated with the uid.
This relationship can be obtained by consulting precalculated tables of uid properties as
functions of pressure and temperature, assuming
a constant total chemical composition at each
pipe location and at each point in time.9
Functional relationships are also afforded
through the study of black oil formulations, in
which uniform uid properties are used, or
through full compositional analysis of reservoir
uid samples, in which individual uid properties
are used for each hydrocarbon component.
Another set of basic closure laws includes laws or
equations that relate the friction factors to velocities, pipe geometry and physical properties of
the uid.
The rst simulations were performed in
steady-state models in which uid properties
such as ow rate, density, temperature and composition were assumed to remain constant over
time at a given point in the system. Steady-state
models thus perform a mass, energy and momentum balance of a stationary processone that is
in a local equilibrium state. While ow parameters may change upstream or downstream of the
particular point in the system, that point remains
in a state of local equilibrium if the uid always
has the same properties, regardless of time.
Since their introduction into the oil and gas
industry nearly 30 years ago, steady-state simulators have evolved signicantly. For example, the
PIPESIM steady-state multiphase ow simulator
allows engineers to predict a range of ow challenges that hinder production optimization, from
the occurrence or formation of asphaltenes, wax
and hydrates to carbon dioxide [CO2] -induced
corrosion and ow-induced erosion.
Steady-state simulations provide systems
designers a method for quickly estimating ow
results at a specic set of conditions and yield
near-immediate insight into how changes in system conditions will impact production. However,
because they operate on the fundamental principle that ow parameters do not vary with time,
steady-state simulators are not applicable for
transient ow phenomena simulation.

as
M

two-phase uid systems under steady-state

conditions to those able to model multiphase
systems in which uid and ow properties
change over time. The article also discusses the
derivation of mathematical models that represent a real-world ow system and includes a
review of the numerical methods used to solve
these models in a simulator.
Case studies highlight how the OLGA dynamic
multiphase ow simulator has helped optimize
well construction and production processes for
operators working off the coasts of West Africa,
the Middle East and Southeast Asia. An example
of hydraulic well control simulation of an exploration well offshore Malaysia is also included.

> The ow of information between single-phase

ow equations. In the case of single-phase ow
in a pipe, the conservation of momentum equation
solves for the ow or velocity of uid in the section
of pipe under study. This ow information is then
used as input to the conservation of mass and
energy equations to update the mass and energy
contents at that section. The new mass and energy
information is then used as input into the equation
of state to update pressure distribution. This new
pressure information as well as the updated uid
density and energy information are then used to
update the momentum equation for the next
section of pipe, and so on. This general relationship
between ow equations exists for every uid
phase present in a multiphase ow system.

This missing time element prompted the

development of dynamic multiphase ow simulations, which allow users to model the time-varying behavior of a system; as a result, predicting
multiphase ow variations that occur regularly
during normal oileld operations is possible. As
in steady-state simulations, dynamic simulations
comprise equations for conservation of mass,
momentum and energy. However, the local variables, including inlet and outlet conditions of the
4. A capping stack is used to control, divert ow and shut in
a well during containment operations. It is not part of the
standard drilling conguration and is deployed only as
necessary.
5. Articial lift refers to any system that adds energy to the
uid column in a wellbore; the objective is to initiate or
improve production from the well. As wells mature and
their natural reservoir pressure declines, most will need
to use some form of articial lift. For more on articial lift:
Fleshman R and Lekic O: Articial Lift for High-Volume
Production, Oileld Review 11, no. 1 (Spring 1999): 4963.
6. Brandt I: Multiphase Flow, Euroil (March/April 1991):
6263.
7. Ayala LF and Adewumi MA: Low-Liquid Loading
Multiphase Flow in Natural Gas Pipelines, Journal of
Energy Resources and Technology 125, no. 4
(December 2003): 284293.
8. Li C, Liu E-b and Yang Y-q: The Simulation of Steady Flow
in Condensate Gas Pipeline, in Naja M and Ma B (eds):
ICPTT 2009: Advances and Experiences with Pipelines
and Trenchless Technology for Water, Sewer, Gas, and
Oil Applications. Reston, Virginia, USA, American Society
of Civil Engineers, 733743.
9. A compositional tracking model may also be used to
provide more accurate compositions for transient ow,
particularly for networks that have different uids
and time-variable ow rates and thus time-dependent
local compositions.

29

Offshore Development

Subsea Tieback

Onshore Processing Facilities

> Models for the life of a project. The OLGA simulator models transient multiphase ow throughout a project life cycle. The software has become the
industry standard for all multiphase eld development, from drilling the rst wells for an offshore eld to developing subsea tiebacks to modeling ow into
onshore processing facilities.

system being modeledsuch as ow rates, inlet

pressure and local gas volume fractionsare
allowed to vary with time to more closely reect
real-world changes that occur in hydrocarbon
production systems.10
Dynamic uid models are used in a wide
range of applications in the simulation of multiphase ow systems, including aircraft design,
prediction of weather patterns and the analysis
of steam and water ow in the core of nuclear
reactors.11 In the early 1980s, uid dynamics
experts began to use such models to simulate oil,
gas and water ow in pipelines.
Development of a Dynamic Flow Simulator
One of the earliest such attempts began in 1980
as a joint research project between the Norway
state oil company, Statoil, and the Institutt for
energiteknikk (IFE), or Institute for Energy
Technology.12 The rst version of the simulation
tool, known as the OLGA dynamic multiphase
ow simulator, was released in 1983 out of this
research project.
The OLGA simulator models slow transients
time spans of ow uctuations ranging from a few
minutes to a few weeksassociated with mass
transport in oil and gas systems.13 Production
engineers use the simulator to model ow in networks of wells, owlines, pipelines and process
equipment (above).14
Starting in 1984, a joint research program
between IFE and SINTEF further advanced the
simulator.15 The program was supported by companies operating on the Norwegian Continental
Shelf, including Statoil, Conoco Norway, Esso
Norge, Mobil Exploration Norway, Norsk Hydro
A/S, Petro Canada, Saga Petroleum and Texaco
Exploration Norway. This research program aimed
to extend the empirical basis of the model and to
introduce new applications.16

30

Early attempts at modeling two-phase ow in

single pipes used separate empirical correlations
for volumetric gas fraction, pressure drop and
ow regimes even though these entities are physically interrelated.17 In the OLGA simulator, ow
regimes were treated as an integral part of the
two-uid system. In the late 1990s, the OLGA
simulator was extended to model three-phase
ow regimes, including the tracking of threephase slugs, during which the ow stream is
divided into intermittent segments of oil or water
that are separated by gas pockets.
Mathematical Models in the OLGA Simulator
A mathematical model within the dynamic ow
simulation space is a digital representation of a
real-world phenomenon. Mathematical models
tend to provide a macroscopic view of uid ow
in pipelines. This approach may simplify the ow
regimes by assuming uid composition within
small sections of the pipeline are uniform, velocity elds at the inlet and exit surfaces are normal
to these surfaces and that uid properties such
as density and pressure are uniform across the
entrance and exit cross sections.
The rst mathematical models within the
OLGA simulator were based on data from lowpressure air and water and steam-water ows in
pipes with an inside diameter range between
2.5 and 20 cm [1 and 8 in.]. The data from the
SINTEF laboratory, which included the addition
of hydrocarbon liquids owing in 20-cm diameter
pipes at a pressure of 20 to 90 bar [2 to 9 MPa;
290 to 1,300 psi]. Scientists used the data to
make several modications to the rst version of
the OLGA simulator. Further iterations of the
simulator have included eld data from pipe systems of up to 76 cm [30 in.] in diameter, which
expanded the tools extrapolation capabilities.18

The transient simulation from the OLGA simulator also accounts for the ow regime within
the modeled section of borehole or pipe.19 For
two-phase gas-liquid ow, the structure of multiphase ow falls into four basic ow regimes:
stratied ow, consisting of two separate and
continuous uid streams: a liquid stream owing at the bottom of the pipe and a gas stream
(usually with entrained liquid droplets) owing above the lower stream
annular ow, consisting of a regime in which a
thin liquid lm adheres to the pipe wall and a
gas stream containing entrained liquid droplets ows internal to this lm
dispersed bubble ow, consisting of a continuous liquid ow with entrained gas bubbles
hydrodynamic slug ow, consisting of stratied
ow punctuated by intermittent slugs of highly
turbulent liquid (next page).20
Initial testing of the mathematical model
using data supplied by SINTEF showed that the
simulator did an adequate job of describing bubble and slug regimes but was less accurate in predicting stratied and annular ows. In vertical
annular ow, the simulator predicted pressure
drops that were as much as 50% too high, whereas
in horizontal ow, the predicted liquid holdups
were too high by a factor of two in some cases.21
Scientists rened the model to account for the
presence of a droplet eld moving at approximately the same velocity as the gas phase, which
describes the ow regime in stratied or annular
mist ow. Mathematical models within the
dynamic multiphase ow simulator also include
continuity equations for three uid phases: a gas
phase; a liquid phase consisting of oil, condensate
or water; and a liquid droplet phase consisting
of hydrocarbon liquidoil or condensate
dispersed in water. These continuity equations

Oileld Review

are coupled through interfacial friction, interfacial mass transfer and dispersions such as oil in
water. Modelers track dispersions by means of a
slip relation, which is the dimensionless ratio of
the velocity of the gas phase to the velocity of the
liquid phase.22
The conservation of mass equations can be
written to account for several components and
uid types, including full chemical composition
tracking, the presence of scale and corrosion
inhibitors, drilling uids, wax, isotopic tracers
and solid particles. A model capable of simulating ow in particle beds was introduced in the
2014 release of the OLGA simulator.23
The OLGA simulator also expresses the conservation of momentum for three continuous layers, yielding separate momentum equations for
the gas layer, which may contain liquid droplets,
and for oil and water layers. One conservation of
energy equation in this model treats the energy
of the system in terms of the combined mixture of
the uid phases and assumes that each phase is
at the same temperature. An equation of state,
one for each uid layer, provides the functional
relationship between the uid volume and its
pressure, temperature and composition.
The simulator selects the particular ow
regime for the model based on the minimum slip
criterion.24 For given supercial velocities, the
simulator selects the ow regime that gives the
lowest difference, or minimum slip, between the
gas and liquid linear velocities. In the 2000s, the
OLGA High Denition (HD) model was developed
by starting with 3D models for frictional forces for
stratied water, oil and gas ow in a circular pipe
and deriving 1D wall friction models as well as 1D
interfacial friction models that have 3D accuracy.25
These mathematical models, applied together,
account for the real-world complexities of multiphase ow in production systems that may include
multilateral wells, pipelines, articial lift systems,
processing facilities and ow control equipment
such as chokes and sand control devices.
Analysts use mathematical models to compute
solutions using numerical methods or algorithms.
These methods take advantage of advances in
computer processing power and speed to create
digital solutions that simulate real-world ow
phenomena at a ne level of detail.
Numerical methods begin by dividing the
overall uid stream in the pipe into small, discrete grids or cells. Each cell has its own values
of pressure, temperatures, uid compositions,
densities, ow rates and uxes.26 Solving the

May 2015

Separated

Stratified flow

Annular flow
Distributed

Dispersed bubble flow

Hydrodynamic slug flow

> Flow regimes as categorized by multiphase ow simulators. Separated ow

regimes are broadly categorized as stratied or annular (top), whereas
distributed ow regimes are either dispersed bubble ow or hydrodynamic
slug ow (bottom). These categories can be further divided based on whether
the uid stream is two phase or three phase and whether the pipe sections
are horizontal, vertical, straight or bent.

conservation equations for each cell begins by

rewriting continuous equations into discrete
counterparts by applying discretization concepts
such as upwind weighting, which uses uid properties of the upstream cells in ow calculations.
This process results in a set of ordinary differen-

tial equations and algebraic equations that represent the model. However, since the equations
may exhibit strong nonlinearity and have to be
constrainedthe total uid volume must be
equal to the pipe volumethe solution methods
must be designed carefully.

10. In a production pipeline, time-dependent phenomena

include changes to ow dynamics caused by pipeline
topography such as terrain-induced slugging, pipeline
startup and shut-in, variable production rates of gas
versus liquids and pigging.
11. Bendiksen KH, Malnes D, Moe R and Nuland S:
The Dynamic Two-Fluid Model OLGA: Theory and
Application, SPE Production Engineering 6, no. 2
(May 1991): 171180.
12. Institutt for energiteknikk (IFE) is a Norway-based
independent research foundation for energy and
nuclear technology.
13. Bendiksen et al, reference 11.
14. Flow in the reservoir itself is modeled by a number of
reservoir simulators, which consider the ow of multiple
components in a reservoir divided into a large number of
3D components known as grid cells. For more on
reservoir simulation: Edwards et al, reference 1.
15. Stiftelsen for industriell og teknisk forskning (SINTEF), or
the Foundation for Scientica nd Industrial Research, is
an independent, noncommercial research organization
based in Scandinavia.
16. SINTEF had been running an experimental program at
the large-scale Tiller Laboratories in Trondheim, Norway,
since 1980. The funding is provided by the same
companies that supported the IFE/SINTEF joint research
program in 1984. Output from the extended OLGA
simulator was tested against the datasets acquired from
this initial work.
17. Bendiksen et al, reference 11.
18. Brandt, reference 6.
19. Flow regime refers to the large-scale variation in the
physical distribution of the gas and liquid phases in a
ow conduit.

20. Danielson TJ, Brown LD and Bansal KM: Flow

Management: Steady-State and Transient Multiphase
Pipeline Simulation, paper OTC 11965, presented
at the Offshore Technology Conference, Houston,
May 14, 2000.
21. Liquid holdup refers to a condition in two-phase pipeline
ow in which the gas ows at a greater linear velocity
than the liquid. The slower moving liquid collects at
low-lying areas of a pipe section.
22. In homogeneous models of two-phase ow, the slip
relation is 1, because the gas and liquid phases are
assumed to be traveling at the same velocity. In many
real-world situations, the velocities of the two phases
can be signicantly different, depending on the ow
regime of the system under study.
23. Brandt I: Some Aspects of Particle Flow Modeling
Within a Commercial, Transient, Multiphase Flow
Simulator, presented at the Geoff Hewitt Celebration
Conference, Multiphase Flow, London, July 2325, 2014.
24. Bendiksen et al, reference 11.
25. Biberg D, Holms H, Staff G, Sira T, Nossen J,
Andersson P, Lawrence C, Hu B and Holms K:
Basic Flow Modelling for Long Distance Transport of
Wellstream Fluids, presented at the 14th International
Conference on Multiphase Production Technology,
Cannes, France, June 1719, 2009.
26. The number of cells dened for a given pipeline is
limited only by complexity of the particular pipeline
being modeled. Additional cells can be dened around
areas of the pipeline requiring greater simulation
scrutiny, such as around valves or inow control
devices. The optimal number of cells for a given pipeline
is often a compromise between the processing time and
accuracy required from the numerical simulation.

31

The entire equation set is then grouped into

subsets according to the characteristics or properties of the equations. The subsets are solved in
stages, one stage followed by the next at the same
time step. The stages are coupled together explic-

itly at the time step. The equations are solved

numerically using iterative techniques until convergence is reached for the entire system. These
methods can be applied to study ow in both
steady-state and dynamic conditions (below).27

Start

Start

Divide the pipeline into grids

Input parameters

Set grid section number: N = 0

Mesh length and time into grids

Input boundary conditions for

P, T, g and L at initial point of
pipeline (N = 0)

Get steady-state solutions

t = t + t.
Calculate the thermophysical
parameters at N section
Designate boundary conditions

No

Calculate the liquid holdup

at N section

Calculate physical parameters

Solve conservation of mass,

momentum and energy equations
to calculate P, T, g and L
at N + 1 section

Solve conservation of mass,

momentum and energy equations
to calculate P, T, g and L
at time t + t

Save results

Calculate the water holdup

N = N + 1.

Save the solutions

Is N = Nmax?
Yes

No

Is t tmax?
Yes

Output the results

Output the solutions

End

End

> Steps for solving two-phase ow models for stratied ow in a condensate-gas pipeline. Steady-state
models (left) begin by dividing the pipe section into smaller sections (N) and inputting the boundary
conditions for pressure (P ), temperature (T ), liquid velocity (L) and gas velocity (g) at the initial point
of the pipeline. The model uses these initial conditions to solve the continuity equations in the rst
section (N = 0) and to calculate values for pressure, temperature and uid velocities in that section.
These values are used as inputs into the next section (N + 1), and the process repeats until the nal
section the other end of the pipeline, Nmax is reached. A similar process is followed in the dynamic
model (right), but an additional iterative step accounts for changes to uid properties and ow
parameters and boundary conditions with time (t ). Because the ow equations are nonlinear,
performing iterations to reach a solution with acceptable accuracy is usually necessary.

32

Principles in Action
When applied to eld operations, dynamic multiphase ow simulation plays an important role in
scientists understanding the likelihood and severity of generating uid-related by-products such as
hydrates, wax, scale and emulsions in a production system. Design engineers use such simulations to predict the occurrence of these species in
the actual eld system and then test various
design alternatives that are aimed at minimizing
their impact. Ideally, such simulations are performed before the production system is built, thus
allowing the operator to design and construct a
production systemfrom the wellbore to the surface processing equipmentthat will keep these
ow assurance challenges to a minimum.
Chevron used dynamic multiphase ow simulation to help manage ow assurance and operational risks in its Lobito-Tomboco subsea eld
in Block 14, offshore Angola (next page).28 The
development comprises three subsea centers
tied back to the Benguela Belize (BB) compliant
tower.29 Both the Lobito and Tomboco reservoirs
contain light (31 to 32 degree API), low-sulfur and
low-acid crude that has little asphaltene and
naphtha content.
Chevron engineers were challenged to design
a robust production system for this eld that
would economically transport the produced uids
from subsea wells to topside while sufciently
mitigating anticipated operational and ow assurance risks. To maximize production from each
subsea center, water injection would be required
to sweep and provide pressure support to each
well. In addition, as water cut increased, the operator would have to implement gas lift. Additional
challenges were expected in the form of ow
assurance risks that included the formation of
hydrates, scale, wax and corrosion by-products as
well as the occurrence of sanding and slugging.
The operator rst assessed its exposure to
ow assurance risks by collecting and analyzing
oil and water samples from the reservoirs. This
analysis included uid pressure-volume-temperature characterizations and a comprehensive
assessment of the uid compositions from each
reservoir.30 Using the OLGA ow simulator, modelers employed the resulting output of this analysis to develop various thermohydraulic models,
which study hydraulic ow in thermal systems.
The operator produced the following thermohydraulic models: individual wellbore and owline,
production system of each subsea center and of
the entire integrated production system with the
wells and owlines.

Oileld Review

Simulations were run under steady-state

conditions, which the operator dened as any
condition in which produced uids were owing
in a fairly uniform and uninterrupted manner
through the wellbore, owlines and surface processing lines. A number of transient simulations
were also run to determine how the production stream would react to dynamic situations
that included system commissioning, startup,
shutdown (planned-versus-unplanned and shortversus-long), initial well ramp up, pigging, deadoil circulation and slugging.
This extensive modeling effort allowed
Chevron to make informed decisions that lowered
the companys initial capital expenditures for the
project while ensuring more reliable production
with a low risk of upsets or unplanned shutdowns.31 For hydrate mitigation, the operator was
able to design the optimal thermal insulation
thickness for subsea owlines and irregularly
shaped components such as connectors and pipeline end terminations. The simulations also
guided the optimal use and injection rate of treatment chemicals such as methanol for hydrate
inhibition, corrosion inhibitor, biocide and oxygen
scavenger to mitigate corrosive attacks.
To mitigate sanding risk, the simulation output guided the operators decision to complete all
producing wells with robust gravel packs and limit
maximum drawdown across the completion,
which also minimized the risk of nes migration.
The dynamic ow simulations suggested a slugging risk mitigation strategy that included proper
well ramp up and ramp down procedures, minimum ow rates in each owline to keep ow outside the slugging regime and ideal topside choke
settings to control slugging during pigging and
dead-oil circulation.
To further mitigate ow assurance and operational risks at the Lobito-Tomboco eld, the operator used OLGA simulator training to help eld
personnel gain familiarity with how the subsea
production system would interact with the topside processing system under various production
conditions. The dynamic ow simulator was also
implemented as part of the operators pipeline
management system to model the real-time
behavior of the multiphase ow in the subsea
production system. The management system can
be used in three different modes: an online realtime application to monitor the current state of
the production system, an online look-ahead
application to predict future operation based on
planned changes to the production system and an
ofine what-if application for planning and engineering studies.

May 2015

BB compliant tower

Center C
Center B

Center A

> Chevron subsea eld. In the Lobito-Tomboco subsea eld, three subsea centers (A, B and C) are
located in approximately 396 to 550 m [1,300 to 1,800 ft] of water and are tied back to the Benguela
Belize (BB) compliant tower, which is in approximately 390 m [1,280 ft] of water. Each center is
connected to the tower via one 10-in. owline (green), which carries production from the wells to
the tower, and via one 8-in. test line (red). The subsea system capacity is 115,000 bbl/d [18,300 m3/d].
This volume lls the Lobito-Tomboco production module on the BB platform and the available
80,000-bbl/d [12,700-m3/d] light oil production train on the BB platform. The subsea system capacity
also allows for production from future subsea centers in the nearby development areas. (Adapted
from Song, reference 28.)

Once the system was built and production

started owing from the three subsea centers,
the eld data were compared with the simulated
results obtained from the OLGA simulator. The
actual and simulated datasets correlated in each
instance, indicating that the thermohydraulic
models used to develop the operations procedures for the eld were accurate. The steady-

state owing pressures calculated using the

OLGA simulator were within 90% of the eld
pressures, while the calculated temperatures
were within 95% of those measured in the eld.
The actual and simulated tree and manifold cool
down times also matched well. The systems were
robust for operations and not over- or underdesigned for maintaining effective ow assurance.

27. Changjun L, Wenlong J and Xia W: Modeling and

Simulation for Steady State and Transient Pipe Flow
of Condensate Gas, in Moreno-Pirajn JC (ed):
ThermodynamicsKinetics of Dynamic Systems.
Rijeka, Croatia: Intech Books (2011): 6584.
28. For more on this eld development: Song S: Managing
Flow Assurance and Operation Risks in Subsea Tie-Back
System, paper OTC 19139, presented at the Offshore
Technology Conference, Houston, May 58, 2008.
29. A compliant tower is a xed-rig structure used for
deepwater oil and gas production. The tower consists of
exible (compliant) legs that reduce resonance and

minimize forces caused by ocean waves. A deck sits

atop the legs to accommodate drilling and production
operations.
30. A pressure-volume-temperature characterization is a
means of characterizing reservoir uid systems through
laboratory experiments and equation of state modeling.
The resulting uid parameters are then used as input for
various reservoir, pipeline and process simulations.
31. An upset in a produced uid stream occurs when
physical conditions such as pressure, temperature or
ow rates in the ow stream give rise to the formation
of precipitates or emulsions.

33

EUROPE
ASIA

AFRICA

Ras Laffan

0
0

100

200 km
75

150 mi

QATAR
Sharjah
Dubai
Fujairah

Doha
Production platform
Processing plant
Previously existing pipelines
Dolphin constructed pipelines

Jebel Ali
Taweelah
Abu Dhabi
OMAN
Maqta

SAUDI
ARABIA

Al Ain

UNITED ARAB EMIRATES

> Dolphin Energy in Qatar and the UAE. Dolphin Energy is involved in every stage of the gas value chain,
from production of raw natural gas at its two offshore platforms to processing at its onshore Ras Laffan
plant to transmission of natural gas by export pipeline to the UAE. The company also distributes the
gas to customers across the UAE and to Oman.

Dynamic Production Management of a Gas

Condensate Field
Today, it is common for operators to implement
production management systems that incorporate dynamic ow simulation tools to optimize
their eld operations. Dolphin Energy used such
a system on its Dolphin Gas Project, which comprises two offshore production platforms 80 km
[50 mi] off the coast of Qatar (above). These platforms produce wet natural gas from the Khuff
Formation; production ows to an onshore processing plant through two 91-cm [36-in.] subsea
owlines. The processing plant separates the
hydrocarbon liquidscondensate and liqueed
petroleum gasfor sale and processes the
remaining natural gas for compression and transportation by pipeline to the UAE.
To meet the challenge of managing liquids
production in a gas condensate eld such as the
Dolphin Gas Project, designers must properly
size pipelines during the project planning phase.
During operations, challenges include managing
rate changes, pigging operations, shut-ins and

34

restarts. Additionally, accurately predicting ow

regimes and the onset of slugging in gas condensate elds is difcult; hydrate mitigation requires
the operator to select the optimal type and thickness of insulation for subsea owlines and the
proper type and dosage of hydrate inhibitor to be
deployed during production.
A pipeline management system was installed
on the project to address these ow assurance
concerns. The system included the OLGA Online
dynamic production support system, which is an
online simulator that generates real-time models
designed to match eld conditions and supports
the reliable operation of the multiphase pipelines from the wellhead to the onshore receiving
and processing plant.
The online simulator incorporates data from
in-eld monitoring and sensor systems, which
measure uid pressure, temperature, ow rate
and liquid holdup in the pipeline. The simulator
then runs real-time models to provide information that supports or adds to what is available
from the existing control system. Such real-time

simulation results help the operator detect leaks,

calculate hydrate risks, understand the likelihood of pipeline slugging and track the progress
of pigs during pigging operations (next page, top).
The dynamic online simulator can also be run
in a mode that allows it to forecast future production or potential ow assurance problems. For
example, an operator can simulate ve hours into
the future at regular intervals to gain an early
warning of situations that might generate a shutdown alarm. The simulator may also be used in a
planning mode, which allows the operator to
understand the impact of any planned design
changes to the operation of the pipeline and processing facility.
For the Dolphin Gas Project, the pipeline
management system generated models for the
two 85-km [53-mi] long, 91-cm diameter multiphase pipelines. It also modeled the operation of
several offshore systems, including two platforms
containing 15 production wells and pig launchers, and the injection and tracking of a hydrate
inhibitor and a scale inhibitor. Onshore, the system provided real-time model updates on the
operation of pig receivers and slug catchers.
The system has been used in daily pipeline
operations since it was installed at the project at
the end of 2007. Through continuous monitoring
of the risk of hydrate formation, pipeline management has helped ensure optimal injection of
hydrate inhibitor. The system is used for active
liquid inventory management and tracking of pigging operations. Dolphin Energy pipeline integrity experts have also used the management
system to track the use of corrosion inhibitor,
providing input to calculate the viable operating
life of the pipelines.
Well Cleanup and Startup in the Kitan Field
Dynamic simulations are also applied to well
cleanup and startup operations.32 Eni Australia
used multiphase, numerical transient simulation
to guide decisions on well cleanup on the Kitan
oil eld (next page, bottom).33 Located approximately 200 km [124 mi] off the southern coast of
32. Well cleanup is a period of controlled production,
typically following a stimulation treatment, during which
time the treatment uids return from the formation and
are produced to the surface. The duration of the cleanup
generally depends on the complexity of the stimulation
treatment; operations such as gravel packing and
hydraulic fracturing require slower and more careful
cleanup to avoid jeopardizing the long-term efciency of
the treatment.
33. For more on this cleanup operation: Yokote R,
Donagemma V and Mantecon JC: Dynamic Simulation
Applications to Support Challenging Offshore
Operations: A Kitan Oil Field Offshore East Timor Case
Study, paper SPE 156146, presented at the SPE Annual
Technical Conference and Exhibition, San Antonio,
Texas, USA, October 810, 2012.

Oileld Review

> Pig Tracking Advisor in OLGA Online. In the Pig Tracking Advisor module of OLGA Online, the operator is able to see a display of a subsea production loop
and the subsea template (top, yellow). The connection to the topside processing facility includes the pig launcher receiver. When a pig is launched into one
of the legs of the production loop, its location is marked by an icon visible along the production loop. Operators are also able to monitor pipeline proles
(bottom left), including liquid holdups, elevation proles and calculated variables (bottom right) such as estimated arrival time at the receiver and current
location and velocity of the pig.

EAST TIMOR
Kitan field

AUSTRALIA

0
0

2,000
1,000

4,000 km
2,000 mi

> Kitan eld. The Kitan eld (left ) is located about 200 km [124 mi] southeast of East Timor and 500 km [310 mi] northwest of Australia. Oil and gas ow from
Kitan subsea Wells 5 and 3 and Well 2sidetrack 1 to a oating production, storage and ofoading (FPSO) vessel (right). Oil and gas ow from the seaoor
to the FPSO (top right) via exible production lines (black, bottom ), whereas gas-lift gas is delivered through separate owlines (red) from the FPSO to the
wellheads (black). The control unit (yellow) distributes commands from the FPSO to the well centers via a main umbilical cable (yellow and black).

May 2015

35

East Timor, the Kitan eld consists of three subsea intelligent wells, subsea owlines, risers and
one oating production, storage and ofoading
(FPSO) vessel. Three wells were completed and
cleaned up prior to the FPSO arrival on location.
Intelligent completions were installed at similar depths in three wells to control ow from an
upper and lower zone. The upper zone for each
well was perforated at a measured depth of
between 3,344 and 3,367 m [10,971 and 11,047 ft];
the lower zone was perforated between 3,384 and
3,394 m [11,102 and 11,135 ft]. Downhole ow
control valves with eight choke positions
fully opened, fully closed and six intermediate
control ow from each zone. Downhole gauges
were deployed to monitor pressure and temperature for each zone.
Because of the remoteness of the eld location, the operator needed assurances that the
cleaned up wells would perform as required prior
to deploying the FPSO. The dynamic ow simulator was used to model the intelligent completion
and several preselected well cleanup scenarios in
which cleanup time, pressure and temperature at
various points of interest and ow rates were
altered to determine their impact on cleanup.
The objectives of the study were to estimate the
required ow rate and duration to unload the

base oil and brine during well cleanup and to use

these estimates as a guide for the actual cleanup
program in the eld.34 In addition, sensitivities to
reservoir parameters such as permeability, pressure and temperature were simulated to estimate
the effect on pressure, temperature and ow rate
at the downhole gauges and upstream of the
choke manifold. These simulations provided the
rig engineers with the information they needed
to predict ow conditions in the wells before conducting the cleanup operations and bringing the
wells onto production.
The simulations dened well cleanup as complete when the amount of brine and base oil in the
produced reservoir oil, measured at the surface,
was less than 1% by mass. The optimal oil ow rate
was estimated to be 7,000 bbl/d [1,100 m3/d],
and the estimated owing wellhead pressure
and temperature were 1,200 to 1,400 psi [8.3 to
9.7 MPa] and 43C [109F]. Pressure and temperature were also predicted dynamically at locations of interest such as the downhole gauges and
upstream of the choke manifold.
The operator used the values from the simulations as guideposts for the actual well cleanup
and well test operations. A comparison of the
model data with the actual well data after
cleanup and testing found that matching was

4,800
4,790
4,780

Pressure, psi

4,770
4,760
4,750
4,740
4,730
Measured

Simulated

16

20

4,720
4,710
4,700

10

12
Time, h

14

18

22

24

> History matching of pressure. The actual measured pressures recorded at the lower downhole gauge
in one of the Kitan eld wells (green) are in reasonable agreement with the pressures obtained by the
OLGA simulation (red); the largest difference measured between them is 1%.
34. A base oil is the continuous phase in oil-base drilling
uids. In the case of well cleanup at Kitan, the base oil
was pumped downhole to displace the brine that had
been used during the well completions operation.
35. For more on dynamic modeling: Nordin NAB, Umar L,
Aziz IABA, Nas S and Woo WK: Dynamic Modeling of
Wellbore Pressures Allows Successful Drilling of a

36

Narrow Margin HPHT Exploration Well in Malaysia,

paper IADC/SPE 155580, presented at the IADC/SPE Asia
Pacic Drilling Technology Conference and Exhibition,
Tianjin, China, July 911, 2012.
36. For more on drilling windows, pore pressure and
fracture gradients: Cook et al, reference 2.

achieved at all downhole gauges with a maximum

difference of less than 1% (below left). Matching,
which was also achieved upstream of the choke
manifold, had a maximum 1% error during the
commingled ow period.
The validated well models were subsequently
integrated with the operators owline models,
which were run to provide information to the
eld startup team. The team used the dynamic
owline simulation results to estimate the optimal position of the downhole ow control valves
without exceeding the systems production limitations. These simulations also helped the operator set the proposed ramp up schedule, estimate
pressure and temperature at various places in
the owline and FPSO and estimate the production uids arrival time at the FPSO. The production system model was validated with actual
production data.
Drilling a Narrow-Margin Well Offshore
Malaysia
Dynamic modeling methodologies also prove useful during drilling operations, particularly when
an operator plans a drilling program in narrowmargin offshore wells. Petronas Carigali Sdn Bhd
faced such a situation ahead of drilling an exploration well in the SB eld, located in the PM block on
the west side of the Malay basin, Malaysia.35 This
basin is characterized by interbedded sand, coal
and shale formations. These conditions, coupled
with a high-pressure, high-temperature (HPHT)
environment and a steeply rising pressure ramp
presented numerous drilling challenges, including
reduced kick tolerances, narrow drilling windows
between the pore pressure and fracture gradient,
high drilling uid densities and equivalent circulating density (ECD) effects of the uids.36
An initial exploratory well was drilled in the
area, and although the operator used MPD techniques, an inux of reservoir uids into the wellbore exceeded kick tolerances and the fracture
gradient of the reservoir, resulting in complete
uid losses and loss of the well. As a result, the
drilling operation failed.
The operator planned a second exploratory
well just 50 m [160 ft] from the rst well, but
with a more rigorous approach to wellbore pressure management that included the use of the
Drillbench dynamic drilling simulator software
(next page). This simulator uses a modeling
methodology similar to that used by the OLGA
ow simulator but focuses on predicting dynamic
downhole conditions that pertain to maintaining
well control while drilling. The Drillbench simulator provides prole plots simulating pressure

Oileld Review

3,500

3,700

3,900

Depth, m

4,100

4,300

4,500

4,700

4,900

5,100
1.3

1.4

1.5
1.6
1.7
1.8
1.9
2.0
Equivalent circulating density, specific gravity

2.1

2.2

> Pore pressurefracture pressure drilling window. The Drillbench simulator

provides prole plots showing pressure conditions, including the pore
pressure (blue line) and fracture pressure (green line) for the entire wellbore
at any time. Understanding the entire system ensures that the wellbore
equivalent circulating density (ECD, red line) stays within the operational
window dened by the formation pressure and strength. Maximum and
minimum values (dotted lines) are updated and stored during the simulation,
providing a useful tool for examining the downhole boundaries to ensure that
no zone of the well is in danger of being fractured by the proposed ECD.

conditions for the entire wellbore at any time and

has a particular focus on identifying potential
weak zones in the formation and intervals of narrow drilling margins. The operator can then
adjust the drilling program to minimize the risk
of a kick or other well control event prior to
reaching potential trouble zones.
Engineers began planning the second well by
gathering offset data from the rst well, which
they used to validate data input to the simulator.
Data inputs included the planned well geometry,
pore and fracture pressure predictions, casing
setting depths, predicted bottomhole-to-wellhead temperature prole, drilling mud weight
and rheology parameters at various sections of
the well.
Dynamic drilling simulations were then run
for each hole section to determine ECDs for various mud ow rates and weights. Tripping simulations were performed to investigate the effects of
changing wellbore temperatures and potential
surge and swab pressure concerns, particularly
in deeper sections of the well where the pore
pressurefracture pressure window was narrow.
Kick tolerance calculations were also completed

May 2015

for each hole section; the dynamic simulations

produced estimates of the impact of uid circulation rates on the kick margin in the well.
The simulations allowed the driller to drill
each section of the well using appropriate drillstring running and tripping speeds, mud circulation rates and surface back pressures to prevent
well control events. Using these dynamic ow
simulations and MPD techniques, the operator
drilled the exploratory well within the narrow
drilling margin to its target depth of 2,800 m
[9,200 ft]. Petronas plans to use the workow
established by this hydraulic well control simulation effort as the blueprint for future appraisal
and development wells in the region.
Future Directions in Flow Simulation
To meet operator demands for models with
greater accuracy and ner details, multiphase
ow simulators must continually evolve. To that
end, Schlumberger has gained experience
through several joint industry projects (JIPs)
that focused on extending the simulators physical and numerical models.

The OVIP OLGA verication and improvement project, for example, began in 1996 as a
three-year study designed to verify the simulators output against eld data provided by oil
company participants, which included Statoil,
Saga, Norsk Hydro, BP, Elf, Total, Agip, Exxon,
Conoco and Chevron. The success of this initial
project, which included ne-tuning the models to
more closely match eld realities, led to a series
of subsequent OVIP project JIPs. The project has
run continuously since its inception. The 2013 to
2015 OVIP members include BG Group, BP,
ExxonMobil, Gassco, Eni, Repsol, Saudi Aramco,
Shell, Statoil, Total, Woodside and PEMEX.
The OVIP project main objective is to serve
as a platform for sharing knowledge about how
OLGA simulator predictions compare with eld
and laboratory data. The project also serves
as a means of sharing ow assurance expertise
between its member oil companies. Members
provide, from their OLGA Online systems, eldwide operational data that has been collected
over long time spans. Last year, one member
supplied the OVIP project group with detailed
measurements from eight onshore and offshore
pipelines. At present, another member is planning experiments covering the entire operational
range of a 34-in. [86-cm] diameter offshore gas
condensate pipeline.
Another JIP, known as HORIZON I, started in
2004 with industry participants including IFE,
Chevron, Eni, ExxonMobil, Statoil and Shell. The
project develops models to better simulate ow
conditions in greater reservoir and water depths,
longer owlines and more challenging temperature and pressure environments. This project
was followed by the HORIZON II JIP, which ran
from 2008 to 2012. The original JIP participants
were joined by Total and ConocoPhillips in
Horizon II. HORIZON II was aimed at expanding the modeling capacity of the OLGA simulator
for long distance gas condensate transport and
long distance transport of oilwell streams. These
projects resulted in new software modules that
have expanded features for the OLGA simulator;
these modules are in use today for longer and
deeper pipelines and processing systems around
the world.
Future developments promise to extend the
reach of multiphase ow simulators even further
by tying them into reservoir modeling, drilling
and production optimization software systems.
The ultimate objective of this work is to provide
operators with a seamless, start-to-nish view of
their production systems for better control of
long-term eld development costs and production potential.
RvF/TM

37