Sei sulla pagina 1di 12

SPE 116936

Workflow for Integrated Production Modelling of Gas Wells in the Northern


Cooper Basin
Tejaswi Shrestha, Suzanne Hunt, Paul Lyford/Santos Ltd; Hemanta Sarma/University of Adelaide

Copyright 2008, Society of Petroleum Engineers

This paper was prepared for presentation at the 2008 SPE Asia Pacific Oil & Gas Conference and Exhibition held in Perth, Australia, 20–22 October 2008.

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.

process of predicting the effects of changes through a


Abstract systematic analysis of individual components and the
Integrated production modelling is essential for well impact of their interaction on field performance1.
performance evaluation and enhancement of the
production system. It is a process of predicting the effects Gas well performance cannot be analysed without
of changes through a systematic analysis of individual considering the reservoir, the flowline and the processing
components and the impact of their interaction on field facility, as each of these components affect the operation
performance1. of the entire production network.

This paper describes a best-practice workflow that was General Allocation Package (GAP) software by
followed in an integrated production modelling study of a Petroleum Experts is a total system-modelling tool. It is a
region in the Cooper Basin, Australia. Procedures are powerful tool which can model and optimise a gas field
presented that improve time efficiency in development of network comprised of wells, pumps, compressors, chokes
a surface network model, including rigorous validation of and separators. Figure 1 is an overview of how the
tank, simulation, VLP, IPR and compressor models. production system network is modelled. GAP models the
surface network internally2. The MBAL and PROSPER
The broad objectives of this work were to improve speed tools are used to model the reservoir and well
of model development in order to study well inflow respectively.
performance in complex stacked fluvial sand systems and
optimise gas production through investigation of Figure 1 – Modelling of production system3
debottlenecking opportunities and additional drill
projects. Following the model build, four optimisation
scenarios effects were investigated: GAP
(1) Alternative flowline connection points for a
high rate well
(2) Alternative tie-in connections to different Production System
satellites (debottleneck)
(3) Additional drill projects
(4) Debottlenecking with additional drill
projects.
Reservoir Wells Facilities
The aim of this paper is to provide an example of best-
practice approach to network optimisation in complex
stacked sand multi-layered reservoirs where time MBAL PROSPER GAP
efficiency and model validation are critical.

1.0 Introduction
A GAP model of the region had previously been
Integrated production modelling provides an effective developed in 2005. However, the model required
understanding of wells and field performance. It is a substantial updating and reviewing to reflect the current
2 SPE 116936

conditions. New wells had been drilled, flowlines


changed and several wells brought into compression. In Figure 3 – Schematic of base case GAP model
addition, new production and pressure data for each well
had to be included and consequently, all well models
amended.

The region of study was the Patchawarra South West


(PSW) region in the Northern Cooper Basin. There are a
total of 11 fields (A, B, C, D, E, F, G, H, I, J, and K)
producing from mainly three formations (“R” sand,
“T”sand, and “U”sand). Figure 2 shows part of the region
with gas fields highlighted in green. The red lines
represent the existing flowlines.
Compressor 2
Figure 2 – Map of PSW region Key:
Separator
N

Pipe Well Reservoir


Compressor 1 (Nodal)

2.0 Workflow Description

2.1 General

Figure 4 illustrates the workflow that was followed in the


integrated production modelling process. A series of
preparation stages are suggested, this data may already be
readily available and should form part of regular gas
reservoir engineering files.

The first stage, highlighted in yellow, was essentially the


Many fields in this region comprise reservoirs consisting “model initiation”. The reservoir model, well model and
of a series of multiple stacked sands with differing surface network were built. Each of these models will be
properties. These reservoir sands were largely deposited discussed in detail in the following section.
in a fluvial system dominated by high sinuosity channels.
The next stage was “model validation”, which consisted
Due to intermitant constraints at the plant, wells can be of a three step procedure. The first step was to ensure the
choked back. Also, some wells produce significant compressor settings were valid. This was accomplished
condensate along with the gas. Condensate-gas ratio by checking the compressor performance, in particular
(CGR) ranges from 0 to 20 bbl/MMscf for the region, making sure the suction pressure did not go below the
based on liquid evaluation tests (LETs). limit specified on the performance curves. When this
occurred, the compressor type was changed from using
Figure 3 shows the schematic of the base case GAP performance curves to a fixed DP compressor type to
model. This case reflects the current conditions. All labels rectify the problem. The fixed DP type essentially fixed
have been removed, but the key on the bottom left corner the suction pressure by specifying a particular pressure
illustrates the icons that represent the pipe, well and drop:
reservoir respectively. The small white arrows indicate DP = Pd − Ps ,min ……………………….. (1)
the direction of flow in the network. There are total of 11 where DP is pressure drop (psi),
different fields producing from three formations. 25 wells Pd is discharge pressure (psi), and
are currently producing through a series of flowlines to a Ps,min is minimum suction pressure (psi).
separator of a fixed discharge pressure of 1015 psig at the
satellite. Three wells were currently offline, so they were Figure 5 illustrates compressor performance being
masked in the model (highlighted in grey in diagram). modelled by two compressor types: one performance
There are two compressors in the production system, one curve type (highlighted in yellow) and one fixed DP type
being a nodal compressor. The model assumes both (masked in gray with cross). The red coloured time
compressors are functioning continuously over time. The constraint allows the compressor to be modelled with
compressors and separator have been labelled on the performance curve until a specified time (when suction
schematic. pressure goes below compressor limit), after which the
SPE 116936 3

original compressor becomes masked and the fixed DP


type is used to model compressor performance.

Figure 4 – Integrated production modelling workflow with suggested pre-preparation work.

Figure 5 – Rectifying compressor performance through use of fixed DP type

Performance curve type (unmasked)

Fixed DP type (masked)


4 SPE 116936

The second step was to match individual well gas rates


to the present time. This was achieved using a process of 2.1.2 History Matching
history matching in MBAL followed by IPR and VLP This step involves matching the model to historical data.
validation in PROSPER The final step in the model The reservoir pressure data that is specified in the data
validation is to cross check the predicted line pressures input step are kept fixed and the program calculates fluid
against the observed and the corresponding total system production rates based on material balance equations.
gas rate. This involved checking flowline lengths, The OGIP was regressed to get a better history match.
adjusting pipe friction coefficient and varying pipe
elevation to replicate “sand dunes” in order to match Figure 6 (top) illustrates the history match. The dark
pressures. The last stage of the workflow was “model points represent the actual pressure data and the blue
prediction” in which various optimisation cases could be line is the modelled pressure over cumulative gas
run. This full workflow with pre-preparation efficiencies production. Looking at the horizontal separation of data
is described in the following sections. points to the blue line, it can be seen that a reasonably
good match is obtained. There is an outlier point
2.1 Reservoir Model Build evident, suggesting a dubious pressure data point.

The MBAL software was used to develop the reservoir As the reservoirs were being modelled as single tanks
model. Material balance applies the Law of Mass under volumetric depletion, we expect the pressure to
Conservation to a reservoir at large by considering it as decline linearly as gas is produced. However, in some
large tank at a uniform pressure4. By history-matching to reservoir models, it was found that initially, the reservoir
past performance, material balance provides information pressure declined quite rapidly; characteristic of a
about the reservoir drive mechanism and allows depletion drive system with fracture stimulation, but
estimation of OGIP. However, material balance does not then the pressure profile appeared to curve and level out.
account for the geometry of the reservoir, orientation of This flattening nature of the pressure profile suggests an
wells or the geology of the formation external pressure support or fluid influx, which requires
a two tank model. Two reservoirs were modelled as two
Material balance analysis requires average reservoir tank models:
pressure and cumulative gas production data. The (1) Reservoir model for Well 5 in Field A was
reservoir pressure data were gathered from p/z plots that modelled with a tight gas tank. This assumes
were available for each well. The p/z plot is a graphical pressure support is coming from a low
solution to the gas material balance. Gas production data permeability region that surrounds the main
for each well was obtained from a production database. reservoir.
Most of the reservoirs are assumed to be under (2) Reservoir model for Well 1 in Field D was
volumetric depletion. The following steps were used to modelled with aquifer support. A simple Hurst-
construct the updated reservoir model using MBAL van Everdingen radial aquifer model was used
software. These steps take approximately 30 minutes per – confirmed with surveillance data.
well if the preparation work has been done as shown in
Figure 4. 2.1.3 Simulation
As a final check to the history match, a production
2.1.1 Data Input history simulation was carried out. Figure 6 (bottom)
The region comprises multiple stacked sands, illustrates an example of the simulation plots obtained,
knowledge of pressure communication helps to break which is the opposite of the analytical plot in the history
down reservoirs into more representative units. matching stage. In simulation, the program uses the rates
However, identification of pressure changes in from the analytical history match to calculate the
individual sand units was not possible, so most wells reservoir pressure6. The blue lines represent actual data
were assumed to be producing from separate reservoirs. and the red line is the simulated pressure data. To
In three fields (G, H and K) some wells were identified evaluate the quality of the match, the vertical separation
as producing from the same reservoir pool. In these between the simulated pressure and historical pressure
cases, it was important to identify an average reservoir points should be compared. It can be seen that in this
pressure trend for the particular formation again this case, a reasonably good pressure match is obtained.
should be done in the preparation stages.

Material balance requires valid pressure and production,


but also needs good pressure-volume-temperature (PVT)
data. The ideal situation would be to have data from
laboratory studies done on fluid samples5.
Unfortunately, there were no PVT studies available to
match to models, therefore correlations were used.
SPE 116936 5

Figure 6 – Analytical history match (top) and amount of time was spent on “setting-up” the well
simulation plot (bottom) model, this can be reduced if preparation is done, eg.
surveillance data summarized in spreadsheets for field
with contemporary rate and pressure data, this would
allow fast IPR/VLP matching.

2.3.1 IPR

PROSPER offers a range of different models to use to


calculate the IPR. The models used to construct the IPRs
were Petroleum Experts and Multi-rate C and n types.
Table 1 describes these two models.

Table 1 – IPR model types7


Type Brief Description Input Data
Required
Petroleum • Uses multi-phase Permeability
Experts pseudo pressure Formation
function to allow thickness
for changing Drainage area
saturations Dietz shape
around the factor
wellbore Wellbore radius
• Assumes Perforated
condensate- interval
dropout is Porosity
produced Connate water
• Accounts for saturation
transient effects
on PI
Multi-rate • Fits rate and Gas rate
C and n pressure data to Flowing
the C and n bottom-hole
backpressure pressure
equation for gas

To verify each IPR, wellhead pressures and gas rates


from orifice metering data recorded on a database were
obtained at the time around the reservoir pressure
specified for the particular IPR. Careful attention was
2.3 Well Model Build given to ensure gas rates were taken at 100% choke
setting. Using a calculation option in PROSPER
Stewart et al. (2001) state that the largest effort in (“Calculate BHP from WHP”), the wellhead pressures
building a production model is the assessment of were converted to flowing bottom-hole pressures by
individual well performance based on the IPR and VLP. specifying the rate, water-gas-ratio (WGR) and
PROSPER was used to develop the well model, it is condensate-gas-ratio (CGR) as well as a flow
best-practice when using this model to match well IPR correlation. Figure 7 (top) shows the effect of changing
and VLP data whenever surveillance data is available. WGR and CGR on the bottom-hole pressure for Well 2
Matching was performed for a point in time that is the in Field G. It can be seen that changing CGR was
time of the surveillance test; if a full historical history insensitive to the bottom-hole pressure, whereas WGR
match is not required – this validation can be used as a had more impact on the pressure. For increasing WGR,
starting point to match to present day gas rates and the predicted flowing bottom-hole pressure increases as
BHFP, when recent surveillance data is unavailable. The well. Values for CGR and WGR for each well were
equipment set-up was verified by checking the well taken from the latest available liquid evaluation test
downhole diagrams. Well models had to be updated to (LET) which is a production test that looks at the
reflect any workovers. Reservoir properties, such as composition of the sample. Uncertainity around
permeability and formation thickness were checked variability in these ratios was accounted for by varying
against petrophysical interpretations and perforation the expected range and using a mean result.
summaries (from production database). A significant
6 SPE 116936

The effect of various flow correlations on the bottom- Figure 8 – Constructed IPR for well 2 in Field G
hole pressure was also investigated. Figure 7 (bottom)
shows the results for Well 2 in Field G. For each
wellhead pressure and corresponding rate, constant
values for CGR and WGR (10 and 5 respectively) were
used, while changing the flow correlation. In each case,
it can be seen that the Beggs and Brill correlation
consistently gave a much higher value compared to
other correlations. It was chosen to use Duns and Ros
Modified as this correlation was applicable to high gas-
oil ratio (GOR) and condensate wells; although where a
direct flowing gradient survey was available the best fit
correlation to the survey was used.

Figure 7 – Effect of changing CGR and WGR on


BHP (top) and effect of various vertical lift
correlation on BHP (bottom) 2.3.2 VLP
Figure 2.14 - Effect of changing CGR and WGR on BHP
2150
In constructing the VLP for a well model, the VLP/IPR
Changing CGR Matching option was attempted to match the well model
Flowing Bottomhole Pressure (psig)

Changing WGR to a test where possible. A flowing gradient survey


2100
(FGS) measures the pressure and temperature profile in
2050 the tubing and this particular type of test can be used in
correlation comparison. This helps to choose the best
2000
correlation to match the well flow regime and represent
1950
the pressure drops in the well. PROSPER offers a wide
range of vertical flow correlations for use, some of
1900 which are listed in Table 2a. Each correlation would
give different results, so it best to match to field data to
1850
select the optimum correlation.
0 5 10 15 20
CGR or WGR (stb/MMscf)
Tables 2a and b – Different vertical flow correlations
Figure 2.15 - Effect of Various Vertical Lift Correlations on BHP available7 and those providing best match.
2800
Gray Hagedorn Brown Duns and Ros Modified
Mukerjee Brill Beggs and Brill Petex 2
2700
Calculated BHFP (psig)

2600

2500

2400

2300

2200

2100
0 1 2 3 4 5

Figure 8 shows the IPR for well 2 in Field G. It used the


Petroleum Experts model. The blue points represent the
test data that were plotted to verify the IPR curve.
Parameters such as permeability were adjusted to get the
curve to fit around the data points.
SPE 116936 7

Figure 9 – Correlation comparison (top) and Final


VLP/IPR match (bottom)

Figure 9 (top) shows the correlation comparison for


Well 1 in Field H. It is a plot of depth versus pressure.
Data from a FGS was plotted against the various
correlations and it can be seen that the closest match was
the GRE correlation. It is important to note that the
closest match is not always the best choice, because it
may not apply for future conditions (GOR, watercut,
etc.); this should be considered if conditions are 2.4 Surface Network
expected to change significantly over the prediction
period. GAP software was used to model the surface network
which comprised of wells, flowlines, compressors and a
After the flow correlation for the VLP was selected, it separator. Flowline lengths and diameters were verified
was matched to the FGS. The IPR was tuned by from a database and engineering drawings, this work
adjusting the reservoir pressure so that the intersection should be done during the preparation stages of the
of the VLP and IPR fit the well test rate measurement. model build.
Figure 9 (bottom) as an example illustrates the final
VLP/IPR match for Well 1 in Field H. When matching Performance curves and fixed DP compressor types
the VLP to the FGS, it was important to make sure the were used to model the compressors. In the performance
IPR was constructed at the reservoir pressure at the time curve type, known performance curve data (such as rate)
around which the FGS was dated. were entered. The rate at the inlet is used to look up the
head and power of the compressor6. A one stage
compressor with 100% polytropic efficiency was
assumed for both models. The polytropic efficiency is
used to calculate the discharge temperature and
pressure6. The fixed DP compressor type was only
employed when the suction pressure went below the
limit specified on the compressor performance curve. It
essentially fixes the suction pressure by specifying a
particular pressure drop across the compressor.

2.5 Compressors

Figure 10 and Figure 11 show the performance curves


(flow rate vs suction pressure) of Compressor 1 and
8 SPE 116936

Compressor 2 respectively. The red data points represent 3.0 Model Validation
historical data and show how the compressor has been
working in the past. The blue points show the predicted After the compressor settings were checked for validity,
performance for a 10 year model. In both models, some individual well gas rates and line pressures were
red points lie outside the curves. These points represent matched to the present time.
when the compressors have tripped and stopped
working. For compressor 1, it can be seen that the 3.1 Matching Rate
predicted performance is a bit varied. In the field, this
compressor has had some operational problems and has To obtain a perfect match to the wellhead gas rates,
effectively been working half the time. For compressor either the reservoir model or well model had to be
2, it can be seen that these points lie within the specified modified slightly. Figure 11 illustrates the effects of
performance curves indicating that the compressor is some changes on the well rate. The variations were
operating within its capacity. Also, later points were made so that any adjustments were within the expected
situated on a better curve at a lower suction pressure, range of uncertainity for each parameter.
which indicates an improvement in compressor
performance. A lower suction pressure means a larger Figure 12 – Methods used to match well rates
pressure drawdown, so the reservoir can drain more and
flow rate improves.

Figure 10 – Performance curves of compressor 1


(flow rate vs suction pressure)
800
0% VVCP
Compressor 1 (Nodal)
25% VVCP
(Pd = 2,500kPag; 1,200rpm)
700 50% VVCP
75% VVCP
100% VVCP
600
Actual Data
Model (10 yrs)
500
Poly (100
Flow (E3m3/D)

400

300

200

100

0
0 250 500 750 1,000 1,250 1,500 1,750 2,000 2,250 2,500 2,750 3,000 3.2 Matching Line Pressure
Ps, kPag

Matching line pressures was achieved by increasing the


pipe friction coefficient, increasing pipe length and
Figure 11 – Performance curves of compressor 2 decreasing pipe diameter all increase the pressure drop
(flow rate vs suction pressure) across the pipe. A further adjustment is the pipe
1,800
0" VVCP
1" VVCP Compressor 2 elevations to replicate ‘sand dunes’, this was applicable
1,600
2" VVCP
3" VVCP
4" VVCP
(Pd = 7,500kPag; 900rpm) for some pipeline directions.
5" VVCP
6" VVCP
1,400 7" VVCP
Actual Data
Model (10 yrs)
0" VVCP 2
1" VVCP 2
3.3 Base Case Results
1,200
Poly. (0" VVCP)
Poly. (1" VVCP)
Poly. (2" VVCP)
Flow (E3m3/D)

1,000 Poly. (3" VVCP)


Poly. (4" VVCP)
Poly. (5" VVCP)
Table 3 shows the final matching results for gas rate and
Poly. (6" VVCP)
800
Poly. (7" VVCP)
Linear (0" VVCP 2)
line pressure for the base case model.
Linear (1" VVCP 2)

600

400

200

0
0 500 1,000 1,500 2,000 2,500 3,000 3,500 4,000
Ps, kPag
SPE 116936 9

Table 3 – Final rate and pressure match of base case 4.0 Optimisation Scenarios
model
Gas Rate (MMscf/d) Line Pressure (psig) Bottlenecks can occur in the production process which
Field Well
Actual Model Difference % Diff Actual Model Difference % Diff
A 1 0.80 0.63 0.17 22% 647 669 -22 -3% limits the total satellite throughput. Debottlenecking
A 2 1.80 1.61 0.19 11% 641 658 -17 -3% refers to the process of increasing the production
A 3 0.02 0.00 0.02 620 674 -54 -9%
A 4 1.20 0.95 0.25 21% 637 669 -32 -5% capacity of existing facilities through the modification of
A
A
5
6
0.42
3.80
0.66
3.60
-0.24
0.20
-57%
5%
648
649
638
669
10
-20
2%
-3%
existing equipment to remove throughput restrictions8.
B 1 0.41 0.39 0.02 6% 645 697 -52 -8% The cost of debottlenecking is usually very small
C
C
1
2
0.28
0.32
0.37
0.42
-0.09
-0.10
-33%
-31%
751
675
743
725
8
-50
1%
-7%
compared to the cost of building new facilities8.
D 1 1.88 1.76 0.12 7% 735 776 -41 -6%
D 2 Offline -
D 3 0.86 0.79 0.07 8% 734 776 -42 -6% Four following optimisation scenarios were
D 4 0.01 0.00 0.01 696 776 -80 -11% investigated:
E 1 1.10 0.96 0.14 13% 777 816 -39 -5%
F 1 0.20 0.33 -0.13 -66% 815 847 -32 -4% (1) Effect of switching flowline for a high rate well
G 1 8.03 8.04 -0.01 0% 863 882 -19 -2%
G 2 4.44 4.28 0.16 4% 543 548 -5 -1%
(2) Effect of connecting a tie-in to different satellite
G 3 2.84 2.77 0.07 2% 580 548 32 6% (debottleneck)
G 4 Offline -
H 1 0.50 0.48 0.02 4% 180 149 31 17% (3) Effect of two additional drill projects
H 2 1.00 0.96 0.04 4% 181 150 31 17% (4) Effect of debottleneck with additional drill
I 1 0.90 1.01 -0.11 -12% 205 150 55 27%
J 1 1.10 0.93 0.17 16% 160 148 12 7% projects.
J 2 0.60 0.54 0.06 10% 180 148 32 18%
K 1 1.40 1.52 -0.12 -9% 242 157 85 35%
K 2 0.70 0.62 0.08 12% 190 148 42 22% 5.0 Results
K 3 0.80 0.70 0.10 12% 207 151 56 27%
K 4 Offline -
Total Flow 34.50 34.31 0.19 1% 5.1 Base Case
Flow to Nodal 6.70 6.76 -0.06 -1%
Notes:
1 Nodal compressor assumed to be in operation. In reality,
Figure 3 showed the schematic of the base case GAP
this has not been fully operational. model, which reflected the current field operating
2 Actual gas rates and line pressures for post-april period conditions.
(when both compressors have been operating).
3 Model gas rates and line pressures taken at 30/07/2007. A prediction (production forecast) was run from early
4 For wells flowing into nodal compressor (H1, H2, J1, J2, 2007 to end of 2035 (28 years). In prediction mode, the
K1, K2, K3 and K4), average of three pressures was GAP program uses pressures and saturations from the
calculated for model results as line pressures were quite MBAL reservoir model to calculate well inflow6. GAP
variable due to compressor performance. recalculates all well IPRs for current reservoir
5 Total actual gas rates obtained from compressor
performance charts on database.
conditions (i.e. at the start of the prediction run) and
calculates well production rates. Using well rates,
Key: cumulative production for each well and reservoir tank
is calculated. The reservoir model is then used to find
Match Quality Gas Rate Pressure
the reservoir pressure at the end of each time step, and
(MMscf/d) (psig) the process repeated stepwise until the end of the
Good 0 - 0.2 0 - 35 prediction period6.
Reasonable 0.2 - 0.4 35 - 70
Poor 0.4 + 70 + Figure 13 is a graph of total gas rate and cumulative raw
gas production for the base case over the time period
The colour key indicates the quality of the match. Green from April 2007 to end 2035.
is considered good, yellow is reasonable and orange
deemed a poor match. It can be seen that the well gas Figure 13 – Total gas rate and cumulative gas
rates matched quite well, however, line pressures production for base case
matched reasonably well with two poor results. 40
Base Case Prediction Results (2007 to 2035)
140
Base Case Total Gas Rate
34.6 MMscf/d Base Case Cumulative Raw Gas Production
35
It was observed that compressor suction pressures were 120
Cumulative Raw Gas Production (Bcf)

highly variable even in small time periods (in particular, 30


114.8 Bcf 100
Total Gas Rate (MMscf/d)

(2P Reserves)
for Compressor 1, which differed by 70 psi) and this 25
80
gave quite variable line pressures. For this reason, a 20

difference in pressure of up to 70 psi was considered 60

reasonable. Total flow to the satellite is currently 34.5 15

40
MMscf/d and the base case model has predicted 34.3 10

MMscf/d. This is a difference of 0.19 MMscf/d which 5


20

represents a mere 1% difference. Total flow capacity of


0 0
the satellite is roughly 42.4 MMscf/d. 30/04/07 29/04/09 30/04/11 29/04/13 30/04/15 29/04/17 30/04/19 29/04/21 30/04/23 29/04/25 30/04/27 29/04/29 30/04/31 29/04/33 30/04/35

Time
10 SPE 116936

In Figure 13, it can be seen that the initial gas rate (at 5.2.3 Additional Drills
start of prediction) was 34.6 MMscf/d. Cumulative raw Drilling additional wells in a field allows the geometry
gas production at the end of 2035 is predicted to be of the reservoir and its bulk volume to be become more
approximately 114.8 Bcf. clearly defined. In addition to increasing the total gas
rate, it can extend the proved reserves of a reservoir.
5.2 Optimisation Scenarios Figure 14c illustrates the model with two additional
development drills. These were Well 7 and Well 8 in
5.2.1 Flowline Switch Field A. All parameters used for the reservoir and well
Looking at individual well rates for the region, Well 1 in models for these two new wells were taken as the same
Field G produced at a high rate (currently 8 MMscf/d). as for Well 6 in the field. Both wells were assumed to
Field G also had two other wells (Well 2 and Well 3) come online in January 2009 at approximately 4.0
that produced at lower rates (4.4 MMscf/d and 2.8 MMscf/d each.
MMscf/d respectively). Unlike these two wells which
flowed directly to Compressor 2 at the satellite, Well 1 5.2.4 Debottleneck with Additional Drills
was currently being flowed along a separate path to tie-
in with well 1 Field F which flowed through a line that The final optimisation scenario investigated the effect of
connected other wells from Fields A, B, C, D and E. The a debottlenecck project with additional drills and Well
next model investigated the effect of this high rate well G1 flowing direct to the compressor at satellite 1.
being flowed directly to Compressor 2 (same flow path
as Well 2 and Well 3 in the Field). There is a twin Figure 15 is a graph of total gas rate and cumulative raw
flowline in place from Field G, so switching the flowline gas production for all four optimisation scenarios over
for Well 1 is easily possible. Figure 14 illustrates the the time period from April 2007 to end 2035.
schematic of the Well G1 flowline switch model.

In Figure 14a, the dashed purple line represents the Figure 14a,b,c - a) Schematic of well G1 flowline
current flowline path for Well G1. The green line switch GAP Model b)Schematic of debottleneck and
represents the proposed flowline switch where the well G1 direct GAP model c) Schematic of additional
flows direct to Compressor 2. The distance from Well drills and G1 direct GAP model
G1 to the tie-in with Well F1 is roughly the same as the
distance direct to the compressor at the satellite.

5.2.2 Debottleneck

Figure 14b illustrates the model with a debottleneck


project and Well G1 flowing direct to the compressor at
the satellite. In this model, we have a flowline from a
separator in different satellite (separator 2) tie-ing to the
flowline in Field A, highlighted in the schematic.

Satellite B already has a number of wells from various


fields producing to it. Currently, 25 wells are producing
to the separator at Satellite A. As a result of
debottlenecking, there are now 11 wells from 5 fields
producing to this satellite. The remaining 14 wells are
flowing to the separator at Satellite B.

The flowline from Separator 2 to the tie-in point is infact


similar in distance from the tie-in to the Compressor 2.
Initially, the debottleneck case was modelled using a
fixed line pressure equal to that of the current suction
pressure of the compressor at Satellite B. However, this
was found to be an inadequate assumption. Another
compressor (Compressor 3) was included to model the
expected improvement in suction pressure in the new
flowline with time. Operational conditions for this new
compressor were set as per the set of compressors
currently in use at the new satellite.
SPE 116936 11

Table 4 – Summary of results


Model Description Initial End Comments
Total Gas Cumulative
Rate Prod (Bcf)
(MMscf/d
)
1 Base case 34.6 114.8 -
2 Well G1 38.0 114.1 Flowline Switch
direct to (New Base Case)
Compressor 1
3 Debottleneck 42.0 116.6 + 4.0* MMscf/d
with new base on New Base
4 Additional 38.0 122.5 +3.1 MMscf/d**
drills with but loss of 4.9
new base MMscf/d,
incremental prod
of 8.4 Bcf
5 Debottleneck 42.0 129.5 +4.2 MMscf/d**
and additional but loss of 3.8
drills with MMscf/d,
Figure 15 – Total gas rate and cumulative gas
new base incremental prod
production for optimisation scenarios of 15.4 Bcf
45
New Wells
Optimisation Scenario Prediction Results (2007 to 2035)
140 * Does not include backout along new flowline
40
Online ** When two new drills come online
120

35
Cumulative Raw Gas Production (Bcf)

100
30
6.0 Conclusion
Total Gas Rate (MMscf/d)

Base Case G1 Direct Total Gas Rate


80
25 Debottleneck G1 Direct Total Gas Rate
Base Case G1 Direct with Additional Drills Total Gas Rate

20 Debottleneck G1 Direct with Additional Drills Total Gas Rate


60
This paper describes a workflow used to create an
15
Base Case G1 Direct Cum Prod
Debottleneck G1 Direct Cum Prod
integrated production model (using GAP software) for a
Base Case G1 Direct with Additional Drills Cum Prod
Debottleneck G1 Direct With Additional Drills Cum Prod
40 number of fields through a relatively complex gathering
10
system. A systematic approach was used for the model
20
5
build which included data preparation stages and best
0
30/04/2007 29/04/2009 30/04/2011 29/04/2013 30/04/2015 29/04/2017 30/04/2019 29/04/2021 30/04/2023 29/04/2025 30/04/2027 29/04/2029 30/04/2031 29/04/2033 30/04/2035
0 practice model validation procedures. It is hoped that
Time
using procedures as outlined herein a model match can
be achieved rapidly and with a small number of
Table 4 summarises the results for all 5 models. Flowing iterations from the initial starting model, thus saving
Well G1 direct to the compressor at the satellite provides engineering time.
a significant initial rate increase of 3.4 MMscf/d.
7.0 Nomenclature
The debottleneck option predicted a rate increase of 4.0
MMscf/d and a reserves increase of 2.5 Bcf compared to BHP: Bottom-hole Pressure
the new base case (Well G1 direct). These results did not CGR: Condensate-Gas Ratio
include backout through the flowline to the second FGS: Flowing Gradient Survey
satellite, but included compression along the line. The GAP: General Allocation Package (software by
addition of new wells provided a smaller total gas rate Petroleum Experts)
gain due to line pressures and backout along flowlines. GOR: Gas-Oil-Ratio
This incremental rate was dependent on whether or not IPR: Inflow Performance Relationship
adebottleneck was included. However, new drills LET: Liquid Evaluation Test
enhanced the recoverable reserves. MBAL: Material Balance (software by Petroleum
Experts)
The above predictions were based on history matches of OGIP: Original-Gas-In-Place
well performance. These matches are not unique. PI: Productivity Index
Bitsindou and Kelkar9 state that there are many other PROSPER: Production and System Performance
(software by Petroleum Experts)
sets of system parameters that can match the past
PSW: Patchawarra South West
performance of a well. So, it is important to realise there
PVT: Pressure-Volume-Temperature
is always some uncertainty associated with the model.
VLP: Vertical Lift Performance
WGR: Water-Gas Ratio
WHP: Wellhead Pressure
12 SPE 116936

8.0 Acknowledgements 4. Hagoort, J., 1988. Fundamentals of Gas


Reservoir Engineering, Elsevier, Amsterdam.
The authors wish to thank Santos Ltd and their Joint 5. Petroleum Experts, 2005b. MBAL A Reservoir
Venture Parties for permission to publish this paper. Engineering Toolkit User Guide, Manual,
Special thanks to Adam Kremer, Santos Ltd for Edinburgh.
assistance in the modelling of compressors and to Nasir 6. Petroleum Experts, 2003a. General Allocation
Subhani, Santos Ltd for help with PROSPER software. Program (GAP) User Guide, Manual,
Edinburgh.
9.0 References 7. Petroleum Experts, 2003b. Prosper Single Well
Systems Analysis User Guide, Manual,
1. Bikoro, F., 2005. Unlocking the Value of Edinburgh.
Marginal Assets: Production Optimisation 8. ConocoPhillips, 2007. Glossary [online,
through Effective Integrated Production System accessed on 12 July 2007]
Modelling [online, accessed 2 August 2007] http://www.conocophillips.com/newsroom/othe
http://alrdc.org/workshops/2005_Spring2005Ga r_resources/energyglossary/glossary_d.htm
sLift/presentations/Reviews/2%20- 9. Bitsindou, A. B., and Kelkar, M. G., 1999. Gas
%20PCL%20--- Well Production Optimization Using Dynamic
%20Production%20Optimization.pdf Nodal Analysis, SPE 52170, presented at 1999
2. Petroleum Experts, 2005a. An Introduction to SPE Mid-Continent Operations Symposium
PROSPER, MBAL and GAP, Training Course held in Oklahoma, 28-31 March 1999.
Notes Integrated Production Modelling,
Edinburgh.
3. Behrenbruch, P., 2006. Production
Optimisation (Lecture Notes), PE 3020
Production Optimisation and Project Course,
The University of Adelaide, Semester 1, 2006.