SPE 77924
Investigation for Mature Minas Waterflood Optimization
Agus Rahardja Hendih, SPE, Rinaldi Imran, and Lawrence Lee Williams, SPE, Caltex Pacific Indonesia
Copyright 2002, Society of Petroleum Engineers Inc.
This paper was prepared for presentation at the SPE Asia Pacific Oil and Gas Conference and
Exhibition held in Melbourne, Australia, 8–10 October 2002.
This paper was selected for presentation by an SPE Program Committee following review of
information contained in an abstract submitted by the author(s). Contents of the paper, as
presented, have not been reviewed by the Society of Petroleum Engineers and are subject to
correction by the author(s). The material, as presented, does not necessarily reflect any
position of the Society of Petroleum Engineers, its officers, or members. Papers presented at
SPE meetings are subject to publication review by Editorial Committees of the Society of
Petroleum Engineers. Electronic reproduction, distribution, or storage of any part of this paper
for commercial purposes without the written consent of the Society of Petroleum Engineers is
prohibited. Permission to reproduce in print is restricted to an abstract of not more than 300
words; illustrations may not be copied. The abstract must contain conspicuous
acknowledgment of where and by whom the paper was presented. Write Librarian, SPE, P.O.
Box 833836, Richardson, TX 75083-3836, U.S.A., fax 01-972-952-9435.
Abstract
With a recovery factor of ~51% and watercut in excess of
97%, the Minas Field in Sumatra Indonesia can certainly be
categorized as mature field. Recently, the field has
experienced a steep production decline, along with increased
operating costs. As a result, a waterflood optimization study
was initiated. This paper discusses the reservoir simulation to
optimize the waterflood operations. The simulation, a sector
model for sections 8C, 8D, and 8E, was used to test various
prediction schemes such as pattern realignment and infill
drilling. A seven well infill program is in progress to test for
unswept oil.
Introduction
The Minas Field discovery was in 1944 and production start
up was in 1952. It is the largest oil field in Southeast Asia,
located in the Central Sumatera basin, a faulted anticline, 28
km long by 10 km wide as seen on Figure 1. The field has
produced to date ~51% of original oil in place (OOIP). The
field is divided into two areas, called the Main Segment (MS)
and Northwest Segment (NWS). A major and complex fault
separates the two segments. The Main Segment has ~80%
OOIP. The reservoir consists of five major sand bodies within
the Sihapas Formation, called the A-1, A-2, B-1, B-2, and D
sands. These sands lie within a 260 feet gross interval of
interbedded sand and shale at an average depth of 2,000 feet
sub-sea and relatively hot (210 oF). Minas crude is 36o API
and waxy with a viscosity of 3.5 centipoise (cp) at reservoir
condition. The formation water is fresh with salinities of
approximately 3,000 ppm and a viscosity of 0.32 cp. There
was no original gas cap.
Primary production was supported by a limited aquifer as
indicated by a significant reservoir pressure decline. Original
development well spacing was approximately on 214 acres. In
the early 1970’s, a peripheral waterflood for pressure
maintenance was started in the western flank. At the same
time, infill development drilling reduced producer well
spacing to 71 acres. In the late 1980’s well spacing was
further reduced to 24 acres in a limited area (two crestal
sections). A field wide pattern waterflood was initiated in
1993 using an inverted 7-spot pattern, 24-acre well spacing.
Despite the field water cut above 90% at startup, the initial
results of this pattern waterflood were encouraging.
Through the primary recovery phase (1952 – 1970), the
field produced 13.3% recovery factor (RF). During the
peripheral waterflood, 1970 to 1993, an additional 28% RF
was produced, of which 14.4% RF was considered
incremental production. The pattern waterflood, which began
in 1993, was expected to improve the volumetric sweep
efficiency. However, depending on the results of the
optimization efforts, it will contribute only 1.4% – 2.7% RF of
incremental reserves.
The Minas sands are characterized by high porosity,
permeability and relative mobility ratio, which limit the PWF
areal sweep efficiency. This problem was augmented by high
injection and production rates, creating preferential water
channels between the injectors and the producers. It is
believed that these extreme conditions may have created
localized areas of bypassed oil between the producing wells.
Recently, the field has experienced a steep production
decline and an increase in operating costs. Among the
challenges facing Minas today are:
- managing water production,
- reducing power requirement and cost,
- managing water disposal and rising reservoir pressure,
- arresting the steep production decline,
- improving sweep efficiencies.
A waterflood optimization study was initiated to identify,
field test, and develop a reservoir management strategy to
maximize the value of the Minas waterflood project. Both
conventional reservoir and production analyses were
performed and then compared to the reservoir simulation to
identify unswept areas.
2 A.R. HENDIH, R. IMRAN, L.L. WILLIAMS
Geological Modeling
Geologic description is the foundation for design, operation
and evaluation of waterfloods. The purpose of reservoir
characterization is to define the reservoir flow units and their
areal and vertical extent, and to quantitatively describe the
variation of reservoir rock properties and their impact on fluid
saturation and fluid flow.
Using an integrated team approach involving geoscience
and engineering professionals, Minas embarked on building a
geological model to:
- Use as the basis for evaluating alternative development
strategies through flow simulation.
- Identify near term opportunities to improve production
performance through workovers and infill drilling.
Minas’ reservoir modeling strategy was to first build a
model over a smaller portion of the field, areas 8C, 8D and 8E,
henceforth referred to as the 8CDE area. The 8CDE area is a
focus for waterflood optimization activities, an area of the
field where significant reserves are thought to remain. The
results of the 8CDE modeling effort was intended to support
the selection of a field trial site for testing pattern realignment
or alternative patterns as indicated by flow simulation tests.
This area, shown in Figure 1, contains about 230 of the ~1400
wells in the Minas field. A 9.8 million cell faulted,
geostatistical model was built for a portion of the Minas field,
covering the 8CDE areas, 245 by 108 by 372 layers with a cell
size of 40m by 40m by 1 feet.
The key components fundamental to the construction of a
geostatistical model for reservoir characterization and flow
simulation are sequence stratigraphy, structural framework
construction from 3D seismic interpretation, petrophysical
analysis of reservoir rock and fluid properties and production
data. The interpretation of faults and horizons from 3D
seismic data forms the constraining structural framework
within which flow-unit geometry (constrained by sequence
stratigraphic concepts and petrophysical analysis) is defined.
In establishing the sequence stratigraphic framework, four
major flooding surfaces and five significant sequence
boundaries were correlated across the Minas field. These
flooding surfaces constitute effective, continuous permeability
barriers that, in most instances, vertically compartmentalize
the Minas reservoirs. They were used to define four separate
stratigraphic grids (sgrids) that were split into layers using the
sequence boundaries. These stratigraphic surfaces were then
tied to the interpretation of horizons and faults from a low-
resolution 3D seismic survey recorded over the Minas field in
1994 to establish the structural framework.
In this workflow, the definition of flow unit geometry
within isolated stratigraphic sequences is based on mineral
model analysis of core samples and on depositional facies
maps that reflect magnitude and orientation of the flow unit
geobodies.
Seven rock types were defined from the mineral model
analysis in this field. Of these, five are defined as effective
sand in terms of oil production. Simulations of flow units and
static reservoir properties were accomplished using multi
binary sequential indicator simulation (MBSIS) and sequential
SPE 77924
Gaussian simulation (SGS) respectively on a layer-by-layer,
flow unit by flow unit basis. Thus, it honors the geometric
characteristics of the pay sands modeled in the variogram
analysis. After establishing the static reservoir properties in
this way, the problem of modeling saturation was addressed.
This field could not be history matched from initial
condition because of lacking early completion and production
data sufficient to the task. Consequently, the estimate of water
saturation circa 1992 was derived to act as the starting point
for history matching.
Reservoir Simulation Modeling
Basic Input Data. Static rock properties: porosity, per-
meability, irreducible saturation, and dynamic properties of
current (as of 1992) water and oil saturations were input into
the model upon the result of geostatistical model.
As mentioned before, a simulation model for the entire
producing life was not possible because of a lack of reliable
well production history or completion for the first 11 years the
field was on production (1953-1964). Fortunately, many new
wells were drilled in the early years of 1990’s in preparation
for the pattern waterflood. The logs from these wells were
used as a basis to determine the fluid saturations across the
model area. Thus, the simulation model was run from that
point forward.
In addition, a Minas light oil steamflood special core
analysis was used to facilitate the model properties.
The well completion data from well history records starting
from 1992 through 2000 was incorporated into the model.
During the history matching process, some well completion
modifications were necessary as indicated by the actual well
performance. In other words, failing cup packer assemblies
and leaking cement squeezed in wet intervals must have
occurred to get the observed production.
Model Scale-up. Scaling the geologic model up to a manage-
able size prior to flow simulation permits rapid iteration
through the history matching process and encourages the
evaluation of numerous development scenarios to optimize
field performance. Normally, this process considers the
distribution of permeability in the geological model and
accumulates layers according to specified limits of total flux
(vertical sum of permeability) and the maximum number of
accumulated layers.
An early scaled-up Minas 8CDE model had very limited
connectivity across the area. It consisted of 38 layers, some
quite thick. There were many actual well completions located
in the inactive cells. The scale up process used simplifying
shale interval across the area, therefore the older model missed
some thin sand intervals. It became inactive cells upon
combining with thicker interval of shale. The model could not
perform as expected without an unreasonable permeability
multiplier factor.
In the second pass, the scale-up of the geologic model was
performed in the Immersive Visualization and Communication
SPE 77924 INVESTIGATION FOR MATURE MINAS WATERFLOOD OPTIMIZATION
Center (IVCC) so that team members were able to ensure that
the objectives of the modeling project were honored in the
scale-up process. Viewing the flow unit definition and other
reservoir properties in the geologic model as a backdrop to the
layer selection process in scale-up ensured that sufficient
resolution was preserved where it is needed.
The new scale up process was done together with earth
scientists. The current model has 58 layers as compared to 38
layers on the previous model. Thus, it has more vertical
definition honoring thin sand intervals and they remain as
active cells. By having more vertical definition, the well
completion intervals in the model have aligned with the actual
well data. Figure 2 compares the first and second models.
The permeability distribution was re-ran based on the most
recent core analysis from nearby area of the light oil steam
flood (LOSF) pilot project. In general, it resulted in higher
permeability values across the model area.
The final flow simulation model contained 267,000 cells
(98 by 47 by 58 layers) with a cell size of 80m by 80m by 3 to
10 feet thickness. This grid system was deemed sufficient to
retain reservoir detail and run in a reasonable time. It was
about one hour CPU time per year history period.
Model Validation (History Matching). The model validation
or history matching process consisted of two main variables,
fluid saturation (water cut) and reservoir pressure.
Pressure Matching. The Minas 8CDE is a sector model
and in fact the actual production far exceeded the injection in
this area, thus pressure declined in the model. Since field
pressure was actually going up, fluid must have been coming
into the 8CDE area from outside the model area. Large pore
volume multipliers on the sides of the model were used to
mimic the extended reservoir. The largest was to the west,
where the reservoir is connected to the aquifer. Horizontal
well injectors were also used to mimic influx from
surrounding reservoir. See Figure 3. The pseudo injector well
rates were adjusted to reflect the actual reservoir pressure. A
good pressure match was obtained using these techniques as
can be seen in Figure 4.
Saturation Matching. As mentioned before, the early
1990’s wells were used as a basis to determine the fluid
saturations across the model area.
The total liquid production rate (oil+water) was input in
the flow simulation model. Thus, the model “tried” to produce
the historical liquid withdrawal rates. Once the pressure match
was achieved, the model oil and water production would be
matched the actual field data as well. The final (1992) fluid
distribution was fine-tuned using an approach which adjusted
the 1992 water saturation based on the ratio between actual
and calculated cumulative oil well by well.
Production Matching. In addition to adjusting the
saturation, well completion intervals were modified to match
the actual well performance after performing well workovers,
especially leaking cup packer assemblies and bad cement
3
squeezed-off perforation intervals. A method to view the
success of the changes on individual well basis was developed
and is shown in Figure 5. The ratio of the actual cumulative oil
produced to the model cumulative oil was plotted for every
well along the x-axis. Ratio’s less than one indicate model
wells producing more oil than actual. Ratios greater than 1
indicate the model wells producing less oil than actual. The
plot shows that nearly 80 out of 160 producers produced at
acceptable ratios. It also shows how fine-tuning the
saturations, as discussed earlier, helped improve the match for
a number of wells. Figure 6 shows the production rate of oil
and water area-wide. Figure 7 and 8 show the cumulative
production plots and bubble maps of historical data versus the
model calculations.
Resulting Products. Important information extracted from
the reservoir flow simulation was a fluid distribution map to
develop future infill wells. Figures 9 to 11 show the remaining
oil saturation (porosity x thickness x So) as of end of year
2000; composite all sands, A2 sand only, and other individual
sand. These maps were used in conjunction with the prediction
runs. The objective was to test several waterflood re-alignment
alternatives to select a site in the 8CDE area. It will be
followed by a field trial of the optimum development
alternative to optimize the areal sweep efficiency.
Prediction. Upon having a reasonable history match
(production and reservoir pressure), numerous prediction
schemes were run to meet the study objectives. Basically four
prediction cases were performed:
1. Do nothing
2. Infill well drilling at low cost and low rate
3. Pattern water flood (PWF) re-alignment
4. Shut in some of the PWF injectors
In addition to the above scenarios, investigations were
performed at various infill wells’ production rates, with and
without voidage balance in order to develop the optimum
future recovery of the interest area.
Common prediction conditions applied to any cases were:
- Use the late 2000 production and injection rate for the
existing producers and injectors
- Minimum oil rate 10 BOPD
- Average minimum BHP of producers 235 psi.
- Maximum BHP of injectors 900 psi.
- Maximum water cut 99.5%
Table 1 summarizes all prediction cases of sensitivity runs
for various scenarios. It indicates a production loss for the
Pattern Water Flood Realignment (PWFR) Case and the Shut
In (some) PWF Injectors (SI4PWFI) Case as compared to the
Base Case (as is operating conditions) despite on oil gain from
some existing producers. Higher incremental oil was seen in
the Seven-Infill Well Cases as compared to any other cases.
An average per infill well of about 150 to 200 MBO
cumulative oil within 10 years period, a total of additional
reserves of 2 MMBO from seven infill wells. It was observed
that the infills robbed oil from the current producers but that
the net production was economic. Figure 12 shows a cartoon
4 A.R. HENDIH, R. IMRAN, L.L. WILLIAMS SPE 77924

of the interference effect between the existing and the infill

producers.
The reservoir simulation model has identified unswept
spots despite of high water cut field performance. A seven
infill drilling well located between the existing producers will
validate the concept for field wide implementation. An
interference effect of the infill wells to the existing wells is
incorporated into the economic evaluation.

Conclusions
The Minas 8CDE sector model has been reasonably history
matched on production and pressure parameters. Based on that
result, several production schemes have been investigated.
The model indicated a production loss in pattern water
flood realignment from the current inverted seven spot to any
other pattern, either triangular (four spot) pattern or line drive
pattern.
The seven-infill well prediction case shows a result of
promising oil incremental. A seven infill well program is in
progress to validate the unswept oil in certain area.

Acknowledgements
The authors would like to express deep appreciation to the
Management of Caltex Pacific Indonesia and Pertamina for
permission to submit this paper for publication. We would like
to acknowledge Gerald Schmit, Dominique Frizzell and many
CPI colleagues for their contribution in part to the content of
this paper.

References
1. Schmit, G. H., Logan, J. P., Rahardja, A., Williams, L. L.:
“Enhancement of Field Development Practices with Geostatistical
Reservoir Modeling at Minas Area 8CDE, Central Sumatera,”
Proceeding of the Indonesian Petroleum Association Conference,
Jakarta, Indonesia (2001).
2. Bou-Mikael, S., Schmit, G. H., Fanandi A.: “Application of
Reservoir Management Techniques to Improve Recovery from a
Mature Waterflood Project in Sumatera,” Proceeding of the
Indonesian Petroleum Association Conference, Jakarta, Indonesia
(2001).
SPE 77924 INVESTIGATION FOR MATURE MINAS WATERFLOOD OPTIMIZATION 5

Fig. 1–Minas Field Location Fig. 4–Pressure matching, model calculated vs. historical data

Fig. 2–Early model vs. current model Fig. 5–Well Oil Production Ratio-Actual and Model

Fig. 3–Dummy injectors to mimic influx from surrounding Fig. 6–Oil and Water Production Rate–Actual vs. Model
reservoir
6 A.R. HENDIH, R. IMRAN, L.L. WILLIAMS SPE 77924

Fig. 7–Oil and Water Cumulative–Actual vs. Model Fig. 10–Remaining So Thickness @ end 2000 Map–A2 Sand

Fig. 8–Oil and Water Cumulative Bubble Map–Actual vs. Model Fig. 11–Remaining So Thickness @ end 2000 Map–Other Sands

Fig. 9–Remaining So Thickness @ end 2000 Map–All Sands Fig. 12–Cumulative Oil of Infill with Interference effect
SPE 77924 INVESTIGATION FOR MATURE MINAS WATERFLOOD OPTIMIZATION 7

Table 1–Prediction Cases Summary

Minas 8CDE Model

Sensitivity runs - with voidage balance of 1

Infill Well Cum. Oil Loss by Net Oil Avg. Net

Production Limit for Oil Production Existing Wells Production for Production Interference
Case Each Infill Well (10 years) (interference) Infills per Infill Fraction
Oil+Water b/d MMSTB MMSTB MMSTB MSTB fract.
Base -- 0.0 0.0 0.0 0 0

Infill Case 1 500 1.01 0.35 0.66 94 0.35

Infill Case 2 1000 1.44 0.49 0.95 134 0.34

Infill Case 3 2000 2.03 0.74 1.29 184 0.36

Infill Case 4 4000 2.73 1.13 1.60 228 0.41

Production Gain/Loss from WF

Realignment & SI, MMBO Infill Constraints:
PWF
Realign (1.87) Min. Oil Prod. = 10 bopd
Voidage ratio of 1.
SI 4
PWFInj. (1.16)