Sunteți pe pagina 1din 18

SPE-173308-MS

Modeling from Reservoir to Export: A Compositional Approach for


Integrated Asset Model of Different Gas Fields in North Kuwait Jurassic
Carbonate Reservoirs
Richard Torrens, Ahmed Daoud, Mustafa Amari, Ahmad Sharifzadeh, and Roshan Prakash, Schlumberger;
Bashayer Al-Enzi, and Qasem Dashti, Kuwait Oil Company

Copyright 2015, Society of Petroleum Engineers

This paper was prepared for presentation at the SPE Reservoir Simulation Symposium held in Houston, Texas, USA, 23–25 February 2015.

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 have not been reviewed by the Society of Petroleum Engineers and are subject to correction by the author(s). The material does not necessarily reflect
any position of the Society of Petroleum Engineers, its officers, or members. Electronic reproduction, distribution, or storage of any part of this paper 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 SPE copyright.

Abstract
A project was undertaken to construct an overview to build an integrated asset model (IAM) of an onshore
fractured carbonate gas condensate and volatile oil asset in Northern Kuwait that is considered the first
gas asset discovered in Kuwait. The asset has the potential to produce from six distributed fields producing
from four hydrocarbon-bearing structures. The development strategy calls for extensive drilling and
facilities expansion to increase and sustain production with the potential addition of depletion compres-
sion to further sustain the plateau. Because the reservoirs are highly compartmentalized, they are split into
19 separate models. Production is through three surface facilities, fluids vary significantly across the field
from sour gas condensate to volatile oil, and it is important to consider the impact of reservoir
deliverability, facilities capacity, and surface backpressure when evaluating different development sce-
narios.
A novel IAM was constructed that integrates reservoirs, wells, pipelines, and facilities models into an
integration platform. The IAM comprises 19 black oil dual porosity reservoir models coupled to a
compositional network model via black oil delumping to convert the subsurface rates into six-components
composition. A split table (compositional delumping) is then used to convert the six-components
composition to 35 surface components to be used in the equation-of-state (EOS) surface network models
to estimate the composition at each point at the surface (inlet and outlet of each facility). Then the network
model is coupled to surface facilities modeling to estimate the rates and composition at the export level.
This idea of mapping the subsurface fluid from black oil at subsurface to compositional at surface reduces
the subsurface running time and makes the IAM more feasible from the running time perspective.
The IAM has highlighted several differences versus the stand-alone modeling and the coupled
modeling at the surface only. First, more accurate accounting for backpressure results in an increase in the
plateau. Second, a production forecast for each facility gives a detailed analysis of production and the
number of wells for each facility. Finally, detailed compositional information becomes available at all
points in the surface network, which is important input to the facilities design.
2 SPE-173308-MS

Figure 1—Location of the six NKJ fields.

Introduction
The North Kuwait Jurassic (NKJ) asset consists of six fields, namely Raudhatain (RA), Northwest
Raudhatain (NWRA), Sabriyah (SA), Umm Niqqa (UN), Dhabi (DA), and Bahrah (BH), extending over
an area of approximately 1,772 km2 in the northern part of the state of Kuwait, as shown in Fig. 1. These
fields, the first free-gas fields in Kuwait, were put on early production during 2004. The NKJ asset is
located in a challenging exploration and development environment that consists of heterogeneous
reservoirs in which natural fractures contribute significantly to production. Portions of the Jurassic
reservoirs consist of a tight matrix with a high density of connected fractures. In other areas; fractures are
sparse and have limited connectivity. These fields are characterized by dual porosity, dominated by low
porosity and permeability, in deep high-pressure, high-temperature conditions, and have a wide variety of
hydrocarbon fluids, ranging from volatile oil to gas condensate with sour gas. Deep horizons (approxi-
mately16,500 ft), overpressured formations, and zones with narrow pore-fracture pressure windows make
drilling operations in these fields very challenging.
To improve development of these fields, an integrated study was conducted to build reliable models for
all reservoirs using all available static and dynamic data. A total of 19 models were generated to simulate
the production behavior for the entire NKG asset. These models are dual porosity, single permeability
models with fluids ranging from volatile oil to gas condensate. The plan is to produce these fields through
EPF50 (an existing facility) and EPF150 and Phase 3 (two planned new facilities) to boost production to
approximately 8 times the current level. Examining the ability of the subsurface reservoirs to deliver this
increase in production requires generating the expected production profile from these 19 models, either
by running each model as a stand-alone model and summing their production profiles, which proved to
be inefficient1 and not to represent the actual production from the asset, or by coupling all the models to
one single controller with one single production target for all the assets, which is currently being
implemented.
Currently, production from all wells in the NKJ fields is routed to a centrally located early production facility (EPF) via
multiphase well flowlines. This facility presents a bottleneck to production due to limited well slots and limited oil and
gas processing capacity. Expansion of the field will occur through the addition of two new facilities, which will receive
production from in-field gathering manifolds via multiphase trunklines.
SPE-173308-MS 3

Figure 2—Facilities schematic.

Figure 2 shows the layout of the current facility and one of the new surface gathering systems and
facilities.
The current facility routes individual well production through separate flowlines to the central EPF for
initial processing in a three-stage separator. The two planned facilities will route production from
wellheads via individual well flowlines to in-field gathering manifolds. These manifolds will then be
routed to the slug catcher at the plant inlet, after which three-stage separations occurs. Note that while the
EPFs are colocated, they are completely separate entities with separate processing capacities. The current
EPF has a gas processing capacity approximately three times less that that of the first planned EPF
(EPF150). The third facility (Phase 3) has a conceptual design capacity similar to that of EPF150.
The current modeling approach to generate the production forecast for the whole NKJ asset is based
on a reservoir coupling approach. An external controller application is used to connect the 19 subsurface
black oil simulation models and apply a common production constraint that is distributed among all fields.
This method proved to be efficient but it still has some key shortcomings: (a) its inability to accurately
predict changes in facility backpressure due to changes in produced composition, (b) an absence of facility
level reporting of production forecasts and number of wells required for each facility, and (c) a lack of
detail in surface compositional reporting for each facility.
In order to address these shortcomings, a new modeling approach based on IAM has been used to
couple a full composition model of the surface gathering system and facilities to the reservoir simulation
models. Thus, the proposed approach is able to improve the accuracy of production forecasts and give
more detailed production and composition for each facility. The remainder of this paper describes the key
components of IAM, which start with fluid modeling and how it is mapped from subsurface to surface,
followed by the network models built for each facility and a description of the workflow for mapping the
wells among the different facilities. Finally, it presents a comparison of the key forecast scenario using
both the current modeling approach and the new IAM approach.

Fluid Modeling
Subsurface Fluid Models
Black oil fluid modeling has been adopted in lieu of full composition reservoir simulation, due to the
complexity of the subsurface modeling. As discussed, the reservoirs are highly heterogeneous and
production is driven by fracture dominated flow, resulting in the need for dual porosity modeling.
Consequently, black oil fluid modeling has been adopted as a pragmatic approach to curb simulation run
times.
The reservoir fluids in the various formations are defined by a total of 10 separate fluid samples. The
experimental data developed from these samples was used to tune 10, six-component EOS models. These
EOS models were in turn used to generate 10 sets of pseudocompositional black oil tables, which form
the basis for modeling the fluid behavior in the reservoir simulation models. The black oil fluid models
use the formulation defined by Cook, Jacoby, and Ramesh2 that allows for five phases, including oil,
4 SPE-173308-MS

Figure 3—Global EOS tuning.

water, gas, vaporized oil, and solution gas, to be modeled separately, which permits modeling of fluids
exhibiting critical behavior.
The validity of this approach is presented by Ghorayeb and Holmes3 and Ghorayeb et al4, who show
good agreement between a fully compositional approach and the black oil approach to reservoir fluid
modeling.
Global Surface Fluid Model
Central to the success of any simulation is effective modeling of fluid behavior under varying pressure-
volume-temperature (PVT) conditions. It is of particular significance for the Jurassic integrated asset
model for a number of reasons: First, the design requirements for the new facility necessitate a detailed
knowledge of the variation of fluid composition over time; second, some of the reservoirs produce fluids
with relatively high H2S and CO2 content; third, many of the fluids being produced exhibit critical
behavior and, therefore, will exhibit phase changes in surface trunklines that may result in flow assurance
issues; and finally, quite different fluids from a variety of sources are being commingled in the surface
gathering system. These facts combine to give four key requirements for the surface fluid model:
1. It must be capable of delivering a detailed compositional description for consumption by facilities
design.
2. It must be able to accurately predict phase changes due to PVT variation over long flowlines.
3. It must be able to reproduce the fluid behavior of the well streams from any of reservoirs with
differing fluids.
4. It must be able to effectively commingle the wells’ streams.
To address these needs, a compositional fluid model was used for surface fluid behavior. The starting
point for this surface EOS was a model that had been developed for the facilities design process. This was
a 50-component EOS model, which was tuned to match the 10 key reservoir fluid samples representing
the variation of formation fluids over the extent of the asset. To illustrate, consider Fig. 3, which shows
three samples representing three different reservoir fluids. The data from lab analysis of these samples
provide input for the simultaneously matching process, using a proprietary algorithm to tune the central,
global EOS model so that it can reproduce the PVT behavior of any of the fluids at surface conditions.
While the 50-component model was ideal for facilities modeling, due to the level of complexity
required for process design calculations, it was considered too complex for trunkline pressure loss
calculations. It was therefore necessary to simply the 50-component EOS through pragmatic lumping of
component groups, which reduced the number of components from 50 to 35. To validate the predictive
capability of the new surface EOS at surface conditions, it was used to predict phase behavior for all fluid
samples and the results were compared with experimental data.
Figure 4 shows the comparison of the 35-component EOS-generated phase envelopes with experi-
mental data for three representative samples representing different fluids in the NKJ asset (volatile oil, gas
condensate, and sour gas). It can be seen that for the volatile oil sample below reservoir temperature
SPE-173308-MS 5

Figure 4 —(a) Volatile oil sample, (b) gas condensate sample, and (c) and sour gas sample EOS performance versus experimental data.

(approximately 250 degF), the experimental data agrees reasonable well with the global EOS predictions.
The same level of agreement was achieved for the other two samples with respect to both phase envelope
and fluid property prediction (i.e., density, viscosity, etc.). Therefore, it was concluded that the 35-
component EOS model was fit for purpose.

Translation of the black oil to compositional well streams at surface


As outlined previously, each black oil fluid model is derived from a six-component EOS model that has
been tuned to match the behavior of experimental data from a specific fluid sample. Consequently, the
black oil model is representative of that fluid sample. Similarly, the global EOS model has been
simultaneously tuned to match all experimental data derived from the key fluids samples. Therefore, the
fluid behavior for any well can be modeled using the 35-component EOS, a six-component EOS, or a
black oil fluid model.
The reservoir models deliver well streams described using the pseudocompositional black oil model
and the surface network model uses the global EOS model, as explained in the previous section.
Therefore, a delumping method must be used to translate the black oil well streams into the compositional
global fluid model. The basis for this translation is the common experimental data from which all fluid
models are defined.
The translation method used by the IAM is a two-step process illustrated in Fig. 5. The first step
involves conversion of the reservoir simulation black oil well streams to the original six-component model
using black oil delumping3,4. The second step involves delumping each six-component well stream into
the global 35-component well stream for the surface network model.
6 SPE-173308-MS

Figure 5—Fluid translation workflow from subsurface to surface,

Step 1: Black oil delumping


The black oil delumping process uses the method
outlined by Ghorayeb and Holmes3 and Ghorayeb et
al4, which utilizes depletion experiments to derive a
relationship between the black oil well streams and
the tuned six-component EOS models from which it
was derived. This is achieved by performing a series
of flash experiments using a PVT modeling package Figure 6 —Compositional split parameter example for NHC component.
(for each six-component EOS model) to generate
tables of pressure versus liquid and vapor composi-
tion. Therefore, at a given pressure, the liquid and vapor compositions are known. While the simulation
is running, the following information is queried for each well at every time step:
1. Free oil volumetric rate and dissolved gas volumetric rate.
2. Free gas volumetric rate and vaporized oil volumetric rate.
3. Liquid and vapor densities.
4. Phase pressures (e.g., liquid and vapor pressures).
Items 1, 2, and 3 deliver the well stream liquid and vapor mass rates. Item 4 is used in combination
with the aforementioned liquid and vapor mole fraction tables to look up liquid and vapor composition at
the queried pressures. The compositions and mass rates are then recombined to deliver a well stream mole
fraction. Note that as the values of items 1 through 4 are dynamic, the composition may vary with time.
For example, consider a gas condensate reservoir initially above dew point; there will be no liquid at
reservoir conditions and, therefore, item 1 will be zero. Consequently, the total stream composition will
conform to the vapor composition. As the field blows down, the reservoir pressure will drop below dew
point and item 1 will increase. Consequently, the liquid composition will become relevant.

Step 2: Mass-based split from six-component to 35-component model


The six-component model consists of nonhydrocarbon gas components (NHC), methane, ethane, C3–C6,
C7⫹, and C20⫹. The 35-component model has far more detail, as outlined in the previous section. For
example, the NHC gas components are fully described (N2, CO2, and H2S). It is therefore necessary to
translate the simple to the more complex description. Figure 6 shows this translation for the NHC
component.
This split is performed by specifying mass-based split parameters, such as those denoted by A, B, and
C in Fig. 6. These are determined using the compositions from the PVT lab report. For example, the mole
fractions of the N2, CO2, and H2S components for one of the volatile oil samples are shown in Table 1:
SPE-173308-MS 7

Table 1—Split Factors for Nonhydrocarbon Components (NHC)


Component Mole % (Zi)

N2 0.03
CO2 1.4
H2S 3.06

Figure 7—Mechanistic model for delumping QC

The group component NHC mole fraction for the sample is the sum of these fractions (i.e., 4.49%).
Split parameters are determined by calculating the mass of the NHC component and the mass of the
individual components as shown below:
(1)

(2)

(3)

(4)

The split parameters are then evaluated based on the ratio to individual components to grouped
component mass as shown below:
(5)

(6)

(7)

The split parameters are normalized for each such grouped component (i.e., A ⫹ B ⫹ C ⫽ 1). All other
grouped components are split on the same basis into constituent components and as the sample
composition and grouping properties vary from fluid to fluid, so the split parameters are also different.
Validation of the delumping method
Mechanistic models were used to perform a quality check (QC) of the delumping for each of the PVT
models used in the simulation. These models consisted of a single reservoir model coupled to a single well
in the simplified surface network shown in Fig. 7. The objective of the QC exercise was to show that the
volumes reported by the reservoir simulator (resulting from black oil tables) agreed reasonably well (i.e.,
within 10%) with the volumes reported by the surface network model (resulting from flash calculation of
the global 35-component EOS model.
These QC models were constructed for each of the key reservoir samples to assess the viability of the
delumping process. In all cases, the postseparator volumes predicted by the 35-component EOS model
8 SPE-173308-MS

Figure 8 —Three facility network models.

agreed with the black oil predictions from the reservoir simulator within the stated tolerance. The average
difference between the stock-tank gas and oil volumes predicted by EOS and black oil models was
approximately 2.5% and 6%, respectively. The delumping method was therefore considered to be
sufficiently robust and accurate to provide translation of all back oil well streams into the global EOS.

Surface Network Models


The use of a full composition surface network model in the integrated asset model delivered the following
key value and insight:
● A means to dynamically calculate the backpressure response on the wells subject to changing
composition and flowline conditions.
● The ability to monitor and control production at the facility level and to control drilling based on
facility-level demand requirements.
● Delivery of gas and liquid composition at the outlet of the three-stage separation process.

Three steady-state multiphase flowlines models were constructed to represent the current and future
EPFs. The model of the current facility (EPF50) consisted of the existing wells, flowlines and separation
facilities. The model of the new facilities (EPF150 and Phase 3) consisted of in-field manifolds, trunklines
and the processing facility (three-stage separator).
The EPF50 model consisted of flowing wells connected directly to the facility via 6-in flowlines up to
approximately 20 km long. The well models were built with the 35-component EOS described previously
and a vertical lift profile that matched well with either the Hagedorn and Brown or Gray multiphase flow
correlation. The existing EPF50 well and pipeline models are connected to a real-time system where they
are updated with surface pressure and temperature sensors and used to estimate the flow rates hourly.
The EPF150 and Phase 3 model consists of wells connected to in-field manifolds via 6-in flowlines
rather than directly to the facility. The manifolds are connected to the processing facility via large-
diameter (10- to 14-in nominal size) trunklines. The three network models are shown in Fig. 8.
SPE-173308-MS 9

Figure 9 —Separator train used in the network models.

In the IAM modeling approach, there is currently no fixed assignment of wells to facilities, which
means that when a new well is opened it may be routed to either the existing facility or a new facility.
Well routing would be determined based on facility demand requirement. For example, if production from
the existing facility drops below its demand capacity, a new well would be opened from the bank of
available wells and routed to that facility.
The three-stage separation process was incorporated into the network models using three sequential
two-phase separators as shown in Fig. 9. This three-stage separation provides detailed compositional
information on both separated liquid and vapor streams at the outlet of the model, which is a critical input
for facilities design.
The connection between the surface network models and the subsurface models was made at the well
level so that every well in each reservoir simulation model was mapped to its corresponding node in the
surface network. The interface point between the models was the tubing head. The inflow and drawdown
across the sandface, the top perforation, and the well are handled by the reservoir simulator, with the wells
being modeled with vertical flow performance (VFP) tables. From the tophole and subsequent down-
stream gathering system to the outlet of the production separators, they are handled by the production
system model.
The surface to subsurface connection must achieve
● mass and composition conservative transfer of well stream information from the reservoir
simulator to the surface network simulator (and vice versa for injectors)
● volumetric consistency to ensure correct backpressure response
● pressure balance to ensure that the reservoir can deliver the requested production and receive the
required injection.
Mass, composition, and volumetric consistency are achieved through dynamic transfer of boundary
condition information for each coupled well from the reservoir simulator to the surface network model at
every time step. Pressure balance is achieved by employing a variation of the iterative explicit coupling
scheme outlined by Ghorayeb et al5 and as described by Okafor6.
Wells Mapping Within the Facilities
The facilities need to be operated at their design capacities for as long as possible in order to sustain the
production plateau through the addition of new wells. The timing and allocation of wells to facilities is
not predetermined; rather, wells will be added to facilities on an “as needed” basis. Consequently, the well
drilling sequence and facility routing is a critical piece of information, which must be delivered by the
forecast model as part of the asset business planning process.
10 SPE-173308-MS

Figure 10 —Logic for the wells mapping between facilities.

When adding new wells, there are two key questions to be answered: (1) “When should a new well be
added?” and (2) “Which facility will the new well be routed to?”. The current modeling approach is able
to answer only the first question; it was not able to delineate to which facility the new well is to be
connected. Adoption of IAM has provided a means to complete the picture by not only defining when new
wells will be required, but also to which facility they will be connected.
This is achieved through the use of centralized drilling queues to service the demand of the three
facilities. The IAM implements a set of rules to monitor the deliverability of each facility and to add new
wells to a facility when the facility deliverability drops below its design capacity. An example of these
rules is

Each facility was monitored by a separate rule, and when the deliverability of the wells producing into
the facility fell below the demand capacity, a new well was opened from the centralized queue and added
to that facility to maintain production. Figure 10 illustrates the logic used for the well mapping to the
EPF50 and EPF150 facilities. As time passes, the deliverability of each facility is reduced due to field
blowdown. Once deliverability reaches the facility capacity, it will no longer be possible to meet its target
and a well from the drilling queue is added to the facility.
In order to optimize condensate production, the wells in the centralized queue were prioritized to favor
wells with higher condensate gas ratios (CGR). Prioritizing the higher CGR wells accelerated oil and
condensate production.
In summary, the IAM wells’ mapping logic provides important information about the well timing and
well routing to each facility, which is critical to the asset business planning process and in the design of
the new facilities to incorporate the available well slots, based on the number of wells required for each
facility.

Forecast Scenarios Comparison: Proposed Versus Current Modeling


Approach
There are three major scenarios generated for reporting the production forecast for the NKJ asset: planned,
potential, and possible. In the planned scenario, the asset has already secured budget for implementation.
The potential scenario is for assets that see value in implementation, but for which the budget is not yet
SPE-173308-MS 11

Table 2—NKJ Asset Production Forecast Scenarios—Definitions and Constraints

secured. In the possible scenario, the asset recognizes the value of the scenario, but it is considered highly
uncertain and needs additional work to confirm the possibility of securing budget.
Both planned and potential scenarios are producing from the Middle Marrat (MM) reservoir in five
fields (hence, five subsurface models) and both are producing from two facilities, EPF50 and EPF150. The
main difference is that the potential scenario assumes the implementation of compression to be able to
produce at lower wellhead pressure while the planned scenario does not. The possible scenario assumed
the asset will produce through all three facilities, with the third facility (Phase 3) mainly dedicated to
production from non-Middle Marrat reservoirs. These reservoirs are Najmah Sargelu (NS), Upper Marrat
(UM), and Lower Marrat (LM). Hence, this scenario involves using all 19 of the existing subsurface
models for the NKJ asset.
Table 2 summarizes the definition and the constraints imposed for each scenario, using the proposed
IAM approach. For the current modeling approach, the same constraints mentioned in Table 2 exist, but
they are imposed at the wellhead rather than the facility inlet pressure because the facility is not integrated
with the subsurface models in the current modeling approach. For the planned scenario, the minimum
wellhead pressure imposed for each well is 1,500 psi; for the potential and possible scenarios, the
minimum wellhead pressure is 700 psi after starting of compression.
Figure 11 compares the total gas, oil, and water production profile for the whole NKJ asset using the
current modeling approach with the proposed IAM approach after coupling surface facilities with
subsurface models for the planned scenario. As seen from Fig. 11, there is a considerable difference
between the current approach and IAM, which results in a plateau approximately two years longer and a
significantly slower decline in production. The reason for this is the pessimistic backpressure assumption
taken in the current modeling approach, where the reservoir simulation models account for system
12 SPE-173308-MS

Figure 11—Total gas, oil, and water production profile from proposed IAM and current modeling approach of the planned scenario.

backpressure using an approximate minimum tubinghead pressure (THP) of 1,500 psi. When the field
blows down to the point that the THP meets this limit, well rates are cut back aggressively to ensure the
limit is not violated. The IAM approach does not use this assumption. Instead, it relies on dynamic
coupling with the surface network model to calculate the backpressure response based on the fixed
pressure at the first stage of the production separator (inlet pressure to the facility), as mentioned in Table
2. In virtually all cases, wells are capable of producing well below the previously imposed limit of 1,500
psi and, therefore they can produce more fluid for a longer period, which results in the observed longer
plateau and less steep decline.
This phenomenon can be explained at the well level, as shown in Fig. 12, for the planned scenario. The
upper chart shows the well pressure variation while the lower shows the well oil production rate variation
as obtained from the IAM approach. The pressure chart contains three separate plots: the THP, the
flowline pressure (FLP), which is the backpressure response experienced by the well, and the fixed
1,500-psi limit imposed in the reservoir simulation models in the current modeling approach.
The THP starts at approximately 1,900 psi and quickly declines. Initially, the THP is greater than the
FLP, so the reservoir has sufficient energy to deliver the fluid through the surface system. During this
period, the difference in pressure is taken up by a choke at the wellhead. After some time, the THP and
the FLP intersect, and it is at this point that the production of the well is impacted by network
backpressure. This is reflected in the oil production rate plot, where the oil rate starts to decline from its
peak of approximately 1,100 stock-tank bbl/d. Shutting the well during the planned maintenance
shutdown in 2014 allowed the reservoir pressure to recover slightly, resulting in an increase of THP.
Following this, however, the well’s production is again curtailed due to backpressure when the THP
coincides with the FLP. The behavior observed in this chart is in stark contrast to that of the current
SPE-173308-MS 13

Figure 12—(a) Wellhead and flowline pressures and (b) oil rate for one of the existing wells as example.

Table 3—Observed Pressure for the Example Well


Measurement Pressure (psi)

Tubinghead pressure 1,250


Flowline pressure 1,123

modeling approach, where the 1,500-psi fixed limit in the pressure plot in Fig. 12 is the cutoff threshold
at which the current approach would have significantly curtailed production.
Table 3 shows the reality for the example well described previously. The well in question is equipped
with wllhead pressure (WHP) sensors upstream and downstream of the wellhead choke, giving both THP
and FLP. The values shown in the table were taken in Q4 2013.
The actual operating THP for this well is approximately 250 psi lower than the 1,500-psi fixed limit
used in the current modeling approach, which highlights the extremely pessimistic nature of this assumed
14 SPE-173308-MS

Figure 13—(a) Gas rate, (b) oil rate, and (c) water rate for each facility for the planned scenario based on the IAM approach.

value. That is why the IAM approach, based on a realistic inlet pressure of 1,100 psi obtained from the
first-stage separator for the planned scenario, let this well produce at a higher rate for a longer period. This
proves why the IAM approach shows more production and a longer plateau than the forecast based on the
current approach.
Another important advantage of the IAM compared to the current modeling approach is the ability of
the IAM to provide the production rate for each facility, which cannot be obtained from the current
modeling approach that couples the subsurface models at the surface, without including the surface
network models. Figure 13 shows the gas, oil, and water production for each facility generated from the
IAM. This information is considered important to understanding the period the facility will work under
its maximum capacity per the outcome from the subsurface production capacity.
As can be seen from Fig. 13, for the planned scenario, the EPF50 capacity can work at its maximum
capacity until 2025 (about 17 years from its startup in 2008) while the EPF150 facility can operate at
maximum capacity for six years. This information is important from an economic point of view, enabling
the decision makers to compare the cost of building the new EPF150 facility to the revenue that will be
generated from the gain in production if the facility is on line by that date.
Figure 14 compares the total number of wells obtained from the current modeling approach and the
IAM approach for the planned scenario. The IAM approach shows fewer wells are required at the early
stage of production and more wells are needed at the later stage to sustain the production plateau longer
than the current modeling approach. This leads to 135 wells able to produce in the IAM approach
SPE-173308-MS 15

Figure 14 —Comparison between current approach and IAM approach of (a) number of wells per year and (b) total number of wells per year for
the planned scenario.

Figure 15—(a) Number of wells per year and (b) total number of wells per year for both EPF50 and EPF150 facilities based on IAM approach of
the planned scenario.

compared to 124 wells for the current modeling approach. The main reason for this is the realistic
minimum pressure constraint imposed at the inlet of the facility of 1,100 psi in the IAM approach
compared to the unrealistically high value of the constraint on the minimum wellhead pressure of each
well of 1,500 psi imposed in the current approach. The IAM approach leads to more wells to produce and,
hence, a longer production plateau and higher cumulative production than the current modeling approach.
Figure 15 shows another important advantage of the IAM approach compared to the current modeling
approach—the ability of IAM to quantify the number of wells required for each facility. This information
is very useful both for designing the new EPF150 facility to accommodate the number of wells required
to be connected to it and for planning the number of wells that need to be added in the future for the
existing EPF50 facility to maintain its production capacity. As shown, the planned scenario requires an
additional 27 wells to be connected to EPF50 while EPF150 requires about 108 wells.
Finally, one of the major advantages of the IAM approach compared to the current modeling approach
is the ability of the IAM approach to report detailed composition for each facility at its inlet and outlet;
16 SPE-173308-MS

Figure 16 —(a) H2S mole fraction and (b) CO2 mole fraction for the current approach (total mole fraction) compared to IAM mole fraction at the
inlet and outlet of each facility for the planned scenario.

the current approach reports just the total composition of the production stream. This detailed composition
reporting plays an important role in the design of the new planned EPF150 facility that the current
approach cannot provide. Figure 16 shows the mole fractions for H2S and CO2 obtained from the current
modeling approach and those obtained from the IAM approach at the inlets and outlets of EPF50 and
EPF150 for the planned scenario. The H2S and CO2 components were selected to be displayed because
they are the two most critical compositions that affect facilities design. Both the IAM and current
approaches show the same trend, but their values differ. The highest values of H2S from the two modeling
SPE-173308-MS 17

approaches are close, and the maximum value reached is 4%. For CO2, the highest values obtained differ,
1.8% from the current approach and 2.5 % from the IAM approach. The detailed estimates of the different
compositions obtained from the IAM approach provided crucial information for the planned EPF150
facility design that could not be inferred from the current modeling approach.

Summary and Conclusions


This paper has presented an overview of a full-field integrated asset model of the North Kuwait Jurassic
gas asset. The model consists of multiple reservoir simulation models coupled to a full composition
surface network, which has been used to resolve shortcomings of the current modeling approach that
couples the subsurface simulation models to only one controller at the surface without taking the facilities
network model into consideration. The following points summarize the major outcome from this work.
1. Using the black oil delumping technique to convert the black oil wells stream production using
black oil simulation models to a 35-component compositional well stream reduces the subsurface
running time and makes the IAM approach more feasible from the running time perspective.
2. Coupling the subsurface simulation models to the facilities network models using the IAM
approach replaces the unrealistic high minimum wellhead pressure imposed on the wells based on
the current modeling approach with a realistic known pressure at the inlet of the facility. This
improves the accuracy of the estimated production forecast for the whole NKJ asset.
3. As a consequence of the above, the production gas rate plateau was extended by two years and the
rate of production decline reduced considerably, as shown by the planned scenario presented in
this paper.
4. The IAM approach has resolved many shortcomings of the current modeling approach by
providing detailed insight into production forecasting at the facility level, including
a. phase rates for each facility
b. number of wells required for each facility
c. Composition of liquid and gas at the inlet and outlet for each facility, which is considered an
important input for the design of the planned facilities.
This new IAM approach provides a robust, accurate means to deliver a long-term production forecast
with sufficient detail to provide key information for multiple processes, including long-term planning,
short-term operational reservoir management, and facilities design.

Acknowledgement
The authors would like to thank Kuwait Oil Company for permission to publish this paper. The
contribution from the North Kuwait Jurassic team members to this paper is gratefully acknowledged.

Nomenclature
MWCO2 ⫽Molecular weight of carbon dioxide (CO2)
MWH2S ⫽ Molecular weight of hydrogen sulphide (H2S)
MWN2 ⫽ Molecular weight of nitrogen (N2)
MWNHC ⫽ Molecular weight of the nonhydrocarbon components (N2, CO2, and H2S) based on
six-components EOS model
ZCO2 ⫽ Mole fraction of carbon dioxide (CO2)
ZH2S ⫽ Mole fraction of hydrogen sulphide (H2S)
ZN2 ⫽ Mole fraction of nitrogen (N2)
ZNHC ⫽ Mole fraction of nonhydrocarbon (N2, CO2, and H2S) from the six-components EOS model
wtCO2 ⫽ Mass per total mole of carbon dioxide (CO2)
18 SPE-173308-MS

wtH2S ⫽ Mass per total mole of hydrogen sulphide (H2S)


wtN2 ⫽ Mass per total mole of nitrogen (N2)
wtNHC ⫽ Mass per total mole of nonhydrocarbon components (N2, CO2, and H2S) from the six-
components EOS model
A ⫽ Split parameter for N2 in the 35-components surface EOS model
B ⫽ Split parameter for CO2 in the 35-components surface EOS model
C ⫽ Split parameter for H2S in the 35-components surface EOS model

References
1. Ghorayeb K., Limsukhon M., Dashti Q., and Aziz R.M.: “Multiple Reservoir Simulation Inte-
gration: An Alternative to Full Field Simulation in the North Kuwait Jurassic Complex,” paper
SPE 115881, presented at the 2008 SPE Annual Technical Conference and Exhibition.
2. Cook, R.E., Jacoby, R.H., and Ramesh, A.B.: “A Beta-type Reservoir Simulator for Approxi-
mating Compositional Effects During Gas Injection,” paper SPE 4272, Society of Petroleum
Engineers Journal 14, no. 5 (1974): 471–481.
3. Ghorayeb, K. and Holems, J.: “Black Oil Delumping Techniques Based on Compositional
Information from Depletion Processes,” SPE Res Eval & Eng 5, no. 10 (October 2007): 489 –499.
4. Ghorayeb, K., Aziz, R., Limsukhon, M., and Al-Anzi, E.: “Innovative Alternative to Full-Field
Compositional Modeling-Case Study of North Kuwait Jurassic Complex,” SPE Res Eval & Eng
(June 2011) 332–344.
5. Ghorayeb, K., Holmes, J., Torrens, R., et alet al.: “A General Purpose Controller for Coupling
Multiple Reservoir Simulations and Surface Facility Networks,” paper SPE-79702-MS, presented
at SPE Reservoir Simulation Symposium, Houston, Texas, USA, February 3–5, 2003.
6. Okafor, C.C.: “Breaking the Frontiers for Effective Flow Assurance Using Integrated Asset
Models (IAM),” paper SPE-149537-MS, presented at the SPE Arctic and Extreme Environments
Conference, Moscow, Russia, February 18 –20, 2011.

S-ar putea să vă placă și