SPE-196443-MS
Reservoir Performance Benchmarking to Unlock Further Development of
Malaysian Oil Fields
Rakesh Ranjan, Mas Rizal, Sumit Soni, and Rahim Masoudi, PETRONAS; Laurent Souche, Schlumberger
Copyright 2019, Society of Petroleum Engineers
This paper was prepared for presentation at the SPE/IATMI Asia Pacific Oil & Gas Conference and Exhibition held in Bali, Indonesia, 29-31 October 2019.
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
As resource owner and enabler, PETRONAS's Malaysia Petroleum Management (MPM) is entrusted to
ensure maximizing recovery efforts from more than 1000 oil reservoirs under production in its portfolio.
Performance and recovery from oil reservoirs depends on many factors that can be broadly classified into
Reservoir Complexity and how the reservoir has been developed and managed. To undertake a development
gap analysis and expectation setting the exercise was undertaken to benchmark reservoir performance
against reservoirs of similar complexity. The objective was to take the learnings from better performing
reservoirs and explore potential replication in poor performing reservoirs of similar complexity. The main
challenge was to establish a single term to define Reservoir Complexity. This term should encompass all the
factors like geological, petro physical, rock & fluid etc. that could potentially make the reservoir complex
and at the same time also decide on the relative weightage of these parameters posing recovery challenges.
Data analytics has been used to accomplish this task and the calibration with reference reservoirs has been
achieved. This benchmarking tool can help to internally set targets for all fields where the recoveries have
been lower than normally observed, help set EUR numbers for green fields and drive additional development
strategies to maximize recoveries in existing fields where they are falling short.
When reservoirs of similar complexities are grouped together, they show varying performance indicators
viz. recovery factor, decline rates etc. The gap analysis between the reservoirs of similar complexity
has helped in identifying poor performing reservoirs and the underlying reasons for underperformance.
Learnings from the better performing reservoirs have been incorporated and a detailed action plan has
been prepared to improve the performance of these reservoirs. Considering the various ways in which this
information can be used, a reservoir complexity benchmark would be a great asset to any major operator
or regulator.
The workflow has been developed to calculate complexity based on the parameters that are affecting
microscopic displacement efficiency, horizontal displacement efficiency and the vertical displacement
efficiency. Data analytics has been used to assign weightages to each component posing recovery challenges
and derivation of a single number defining complexity on a scale of 0 to 1. This is major improvement on
all previous works of this nature attempted in various parts of the world and provides the user with not only
the complexity per se but also its distribution. This benchmarking tool has been used for selected fields
2 SPE-196443-MS
and has enabled development gap analysis and helped in initiating course correction to unlock more values
from the underperforming reservoirs.
Introduction
PETRONAS Malaysia Petroleum Management (MPM) has been entrusted to manage all the E&P activity
in Malaysia as the resource owner, regulator and enabler. One of the key objectives for MPM is to maximize
the recovery from the Malaysian Oil & Gas fields thereby creating more value for all the stakeholders
involved. With more than 1000 oil reservoirs under production and many more under development or pre-
development in its portfolio, a quick exercise was done to assess how the different producing reservoirs
are performing in terms of Recovery Factors. The analysis showed that many reservoirs cumulating to
significant volumes are suffering from low recovery factors (Figure 1).
Figure 1—Distribution of Recovery Factor vs STOIIP
Low recovery from reservoirs can be attributed to several reasons which can be primarily grouped into
reservoir complexity, sub-optimal development or over-estimation of STOIIP. The issue of appropriate
estimation of STOIIP is being taken care through the periodic Full Field Reviews (FFR) incorporating all
the available information till date. Malaysian depositional environment varies from multilayered shallow
marine to deepwater clastics, fluvial and carbonates, heavily contributing to complexity and varying
recoveries. Recovery factor from the reservoirs of similar complexity should be within the same range
if a similar development effort has been made on both the reservoirs. If there are significant differences
in recovery factors of reservoirs of similar complexities, then the development learnings from the better
performing reservoirs can be applied to low RF reservoirs (Figure 2). This can help in identifying
the development gaps and course correction to bring the reservoir performances upto the benchmark
expectations.
SPE-196443-MS 3
Figure 2—Estimated RF vs Reservoir Complexity Index (RCI)
The biggest challenge was to come up with a single term to reflect the Reservoir Complexity
encompassing all the different parameters that can impact the recovery factor. Realising the need for a
benchmarking exercise, it was decided to develop a methodology and tool to enable quick benchmarking
and gap analysis.
Literature review and preliminary considerations
A comprehensive literature survey was carried out on the published methodologies of calculating reservoir
complexity and estimation of corresponding recovery factors. This formed the basis for the preliminary
consideration for the benchmarking tool and the brief synopsis of some of the published methodology is
presented below.
The first structured publications relating reservoir complexity index (RCI) to ultimate reservoir recovery
was mandated by the Norwegian Petroleum Directorate (Figure 3) and published by Bygdevoll [2007].
Figure 3—Original RCI calculation parameters proposed by Bygdevoll 2007
4 SPE-196443-MS
The results suggest that the ultimate recovery factor (URF) of North Sea fields can be predicted with a
reasonable level of accuracy using six carefully selected parameters. These parameters were binned into
different groups to come up with a composite score of RCI. An in-depth review of the relevant literature
however reveals that reservoir complexity index calculations had to be adapted to the geological setting, the
type of fluid, etc. For example, [Wickens and Kelly, 2010] show that the correlation between RCI and URF
obtained using the same six parameters is much poorer when applied to fields located in the USA (Figure 4).
Figure 4—Correlation between RCI and Ultimate Recovery Factor using the methodology by [Bygdevoll 2007]
Consequently, several other authors have been developing specific RCIs to describe the complexity of
other types of reservoirs. Examples include [Horikx, 2013] which studied the complexity and recovery of
Chalk reservoirs and [Jia, 2016], tackling the issues specific to heavy oil reservoirs (Figure 5).
Figure 5—Different algorithms for assessing reservoir complexity in chalk reservoirs heavy oil reservoirs
Different algorithms for assessing reservoir complexity in chalk reservoirs from Horikx (2013) can be
seen on the left graph of Figure 5 and heavy oil reservoirs from Jia (2016) on the right.
It was quite evident that there is non-uniqueness in the approaches published and all the proposed
correlations work for specific regional settings. The methodology developed as part of this project was
therefore adapted to the complexity elements which are specific of the Malaysian fields and reservoirs,
while incorporating the lessons learnt from the published methods. It was designed to be flexible enough to
incorporate more parameters than those initially short-listed by Bygdevoll.
From an algorithmic standpoint, the developed benchmarking engine must be able to link multiple,
heterogeneous inputs (structural, sedimentological, petrophysical parameters, etc.) to multiple outputs
(reservoir complexity index, estimated ultimate recovery factor, various recovery challenges, etc.) in a non-
linear manner. Moreover, the tool needs to proceed from a limited number of uncertain parameters (a few
SPE-196443-MS 5
tens for each reservoir). The level of accuracy and uncertainty of the input parameters is expected to vary
greatly depending on the type of considered parameter. For example, while oil viscosity and gravity API are
commonly measured from lab data with a high degree of accuracy, the stratigraphic complexity of the field
has to be estimated from interpretation data (number of correlated tops, etc.) which may vary significantly
from one field to another. There was a conscious effort to select and design parameters for complexity
calculation to be more quantitative than qualitative. Moreover, it is expected that the database is not
complete, and the tool should cope with missing or incorrect data. The retained data analytics methodology
was therefore designed to be adapted to limited amounts of data and account for input data uncertainty.
A combination between analytical calculation, Bayesian Network and forward modeling approaches is
therefore chosen for benchmarking engine.
Finally, due to the limited number of input parameters used and single indicator produced (the RCI), the
standard methodology described by Bygdevoll does not enable performing much diagnostics beyond the
quantification of the remaining reservoir potential (i.e., difference between the estimated ultimate recovery
and the current recovery). The proposed methodology allows to identify more specifically the reasons for
poor recovery and the techniques that could lead to an improvement. This required the generation of more
output indicators in order to enable a more granular diagnostic.
Methodology
The calculation of a reservoir complexity index can be decomposed into the product of three main
components (Figure 6):
• The microscopic displacement efficiency, which characterizes the ability of the water to displace
oil into the rock pore volume. This parameter is essentially linked to the rock reservoir quality and
to the fluid properties (e.g., viscosity)
• The horizontal displacement efficiency, which quantifies the areal homogeneity of the sweep along
a particular reservoir layer. It is mostly impacted by the heterogeneity of the rock, the presence of
conductive fractures (macro-scale water fingering) or by sealing or partially sealing faults.
• The vertical displacement efficiency measures the impact of bypassed zones and of uneven vertical
sweep on the recovery. Elements like thief zones, water coning or vertical conductive fractures
influence this parameter.
Figure 6—Relationship between displacement efficiency, RCI and STRF
A fourth component, which measures the reservoir drive efficiency (i.e., the pressure support) needs
to be accounted for in order to estimate more accurately the ultimate recovery factor. The methodology
6 SPE-196443-MS
used in the reservoir benchmarking tool aims at estimating three components (microscopic displacement
efficiency, horizontal displacement efficiency, vertical displacement efficiency) in order to deduce the
reservoir complexity index and to add one additional component (the drive efficiency) for estimating the
ultimate recovery factor.
The petrophysical and engineering data being considered less uncertain than the descriptive geological
indicators, an analytical methodology can be used for computing the microscopic displacement efficiency.
A forward model, based on the relative importance of the various drive mechanism (aquifer strength,
solution gas drive, gas cap expansion) and on the observed pressure decline is used for estimating the drive
efficiency. Finally, due to the more qualitative nature of some of the parameters used for computing the
vertical and horizontal displacement efficiency (e.g., stratigraphic complexity indicators) and due to the
lack of analytical formula for computing some of the parameters e.g. the compartmentalization impact on
recovery, a Bayesian Network approach has been selected for computing those.
Displacement Efficiency & RCI Calculation
The microscopic displacement efficiency is based on the application of the Buckley Leverett theory for
immiscible fluid displacement in porous media and quantifies the proportion of in-place oil which is
effectively pushed out of the pore volume by the water-front. The drive efficiency is a model of the
proportion of in-place hydrocarbons likely to be displaced by the various drive mechanism. It accounts for
the greater effectiveness, in general, of a good aquifer over e.g., solution gas drive.
Finally, a rigorous, quantitative approach is followed for computing the horizontal and vertical
displacement efficiencies. It consists in estimating the proportion of hydrocarbons which is left unrecovered
because of various recovery challenges (e.g., water coning, bypassed zones) and subtracting it from the
estimated ultimate recovery (Figure 7). By estimating the impact of each recovery challenge on the reservoir
complexity and on the ultimate recovery, a much finer grained diagnostic of the reservoir performance can
be performed. For example, it is expected that reservoirs in which early water breakthrough is likely to
occur can be identified using the proposed benchmarking engine. Furthermore, situations in which early
water breakthrough can be avoided by maximizing the drilled section away from the water leg (e.g., via
horizontal drilling) can be distinguished from situations in which the best solution to avoid is to place
intelligent completions.
Figure 7—Link between recovery challenges, displacement efficiency, RCI and STRF
SPE-196443-MS 7
Recovery Challenges
The likelihood that various type of recovery challenges will occur is computed as a precursor to the
horizontal and vertical displacement efficiency estimates, based on the input data from the Malaysian oil
reservoirs. Some of the recovery challenges accounted for in the reservoir benchmarking engine are listed
in Figure 8. These diverse recovery challenges reflect the impact of several source of complexity: structural,
stratigraphic, oil column thickness, rock quality, fluid mobility, etc.
Figure 8—Examples of recovery challenges likely to impact recovery
The key advantage of using an approach based on recovery challenges rather than an approach based
on multiple abstract complexity indices (e.g., structural complexity index, stratigraphic complexity index,
fluid complexity index, etc.) are that:
• the selected recovery challenges are observable or measurable in mature fields and reservoirs, thus
providing an additional set of calibration points to the methodology
• predicting the probability of occurrence of recovery challenges is extremely valuable when
diagnosing the cause for which a field is underperforming, or for assessing the best development
scenario.
Calculation Methodology & Model Calibration
The microscopic displacement efficiency calculation is based on the Buckley Leverett theory and depends
both on the rock quality and on the fluid properties. In the proposed implementation, porosity and
permeability values are used as a proxy for categorizing the rock quality. A fractional flow curve is then
built, incorporating information on the initial oil saturation and oil viscosity. Finally, the oil saturation at
abandonment water cut is computed, and converted to a microscopic displacement efficiency.
Horizontal and vertical efficiency are essentially computed using a Bayesian network type of approach.
Bayesian networks rely on conditional probabilities to link a set of inter-related phenomena with each other.
More specifically, they quantify the probability that an output event occurs based on the values of a set
of input parameters which can be correlated with each other or independent. In the scope of the reservoir
benchmarking engine, the Bayesian networks (Figure 9) aim at assessing the probability that recovery
challenges will occur (output parameter) based on the value of input parameters (average porosity, structural
complexity, etc.).
8 SPE-196443-MS
Figure 9—Conditional probability table link to output categories
For each reservoir, the probability for each recovery challenge impacting vertical and horizontal
displacement efficiency to occur given the input parameter sets is computed using Bayesian networks.
Finally, both the reservoir complexity index (RCI) and the simulated recovery factor (STRF) are deduced
from the estimated displacement efficiencies. The calculation used in computing uncalibrated RCIs and
STRFs is shown in Figure 10.
Figure 10—Calculation of uncalibrated STRF and RCI from displacement efficiencies
All reservoir complexity indices proposed in the literature (e.g., [Bygdevoll 2007], [Horikx 2013], [Jia
2016]) have been calibrated to some actual reservoir data. The purpose of the calibration is to ensure that the
reservoir complexity indices computed by the reservoir benchmarking engine are linearly correlated with
ultimate recovery factors for a subset of reservoirs for which the ultimate recovery factor is known with
a reasonable degree of accuracy (the calibration reference dataset). One can then be relatively confident
that when the calibrated engine is applied to reservoirs for which the ultimate recovery factor is unknown
(i.e., in prediction mode) the computed RCIs will correlate reasonably well with the actual URF of these
fields (Figure 11). Once it is calibrated on selected brown fields, the RCI calculation engine can be used in
prediction mode on both green and brown fields and is expected to yield a greater degree of accuracy.
SPE-196443-MS 9
Figure 11—Principle of the calibration of simulated recovery factor to simulated or observed recovery factor
The ultimate recovery factor needs to be known with a reasonable degree of accuracy for reservoirs
selected in the calibration reference dataset. Typically, this dataset is therefore composed of well managed
brown fields from which detailed reservoir models exist (allowing to compute the end of field life recovery
factor), possibly complemented with fields for which detailed field study (including predictive flow
simulations) have been performed, allowing to estimate the ultimate recovery factor. The calibration phase
consists of updating some parameters used for computing the calibrated RCI model until a satisfying linear
correlation between the computed RCIs and known URFs of the calibration reference dataset is reached.
Gap analysis and upside potential evaluation
Several methods are combined to facilitate gap analysis, underperformance diagnostic and upside potential
evaluation. The first methodology enables the identification and comparison of analogue fields and
reservoirs. By comparing production data and Estimated Ultimate Recovery Factors for fields with similar
complexity or simulated recovery factors, it is possible to identify under-performing fields. Furthermore, by
comparing development decisions (e.g., pressure maintenance strategy, type of injected fluid, well density)
of analogue fields, it is possible to identify the most successful development strategies for a given set of
reservoir parameters.
The second methodology enables plotting simulated recovery factor or reservoir complexity index against
estimated ultimate recovery factor for any selection of fields or reservoirs. This plotting facilitates the
identification of under-performing reservoirs (EURF<<STRF). Part of the follow-up analysis, however,
should be completed outside of the reservoir benchmarking tool as the reasons for under-performance may
be due to operational issues rather than subsurface issues.
The tool also provides a breakdown of the recovery challenges likely to impact ultimate recovery in each
field and reservoir. Understanding the main recovery challenges is key to de-risking a development strategy.
Finally, the tool enables a pre-screening of potential development options by comparing their relevance
across fields and reservoirs.
Case Example
The Reservoir Performance Benchmarking tool has been applied to evaluate the remaining potential from
the field "D". Field D is one of the largest field in terms of STOIIP in offshore Peninsular Malaysia. The
field has been under production for more than 25 years and has achieved around 28% Recovery factor.
10 SPE-196443-MS
The field has always been considered as "Complex" because of its geology and several vertical and lateral
compartments. The analysis of the field is also made difficult because of commingled production and well
integrity issues making production allocation to different reservoirs quite uncertain. Several attempts to
come up with reservoir simulation models were found to be challenging and resulting in mixed outcomes.
There was a need for MPM to complete a high level of the true potential of field "D" in terms of recovery
factor within a short period of time. The objective was to establish an independent assessment to represent
the Host Authority's point of view (POV) in order to challenge the Operator's findings.
The basic data of reservoirs of "D" field was populated in the tool for it to calculate the reservoir
complexity index. Then the benchmarking of D reservoir recovery factor was done against other producing
reservoirs of Malaysia (Figure 12). Based on the benchmarking exercise it was clear that D reservoirs were
underperforming against the reservoirs of other fields of similar complexity.
Figure 12—Benchmarking of "D" Reservoirs with other Malaysian Reservoirs
By comparing reservoirs with similar complexity, it was possible to dissect each reservoir in terms of
the factors enabling it to achieve higher recovery factor as far as development was concerned. Few things
that came out very clearly helping in higher recoveries were well density, optimized secondary recovery
and producer-injector ratio. These became the recommendation for further evaluation for replication in poor
performing reservoirs. The project team carried out the simulation modeling work and incorporation of
the proposed recommendation coming out of the benchmarking exercise resulted in significantly higher
potential recoveries than earlier thought of.
Contrary to conventional belief, the "D" field did not entirely consist of complex reservoirs, but also
simple-textbook reservoirs that were capable of achieving a significantly higher recovery factor compared
to the field average. The tool enabled the Host Authority to recommend additional efforts to achieve higher
recovery compared to the Operator's initial findings. This was achieved by overcoming the perceived barrier
to higher recovery caused by years of underachieving drilling results.
During the exercise it was also recognized that few other things that needs to be incorporated in tool
to account for operational complexity, commercial considerations and sub-optimal historical reservoir
management to get a more holistic view on the remaining potential from a particular reservoir. These
learnings have been incorporated in the bigger effort of Recovery Factor Improvement Plan (RFIP) to
improve the overall Recovery Factors from the Malaysian oil fields. In addition to benchmarking exercise,
SPE-196443-MS 11
RFIP considers the validation of reserves & resources, performance gap analysis, FDP analogue and upside
management potential to come up with a comprehensive plan for actions to improve recoveries. The
methodology and workflow developed as part of RFIP is being tested for the selected fields and will be
published later.
Conclusion
There have been several attempts made to come up with the methodology of calculating Reservoir
Complexity Index and its relationship in predicting Recovery Factor from these reservoirs. The published
methodologies are developed for and thereby applicable to certain geological and regional setting. The
benchmarking tool developed for Malaysian oil reservoirs uses the combination of analytical equations and
data driven methodology to compute Reservoir Complexity Index and benchmark recovery factor. This
methodology has been successfully tested and helped in identifying development gap in the studied field
leading to formulation of Recovery Factor Improvement Plan.
Acknowledgement
The authors would like to thank PETRONAS Management for the permission to publish this paper, and
gratefully acknowledge the input from Schlumberger staff who helped build the tool and in completing the
paper.
References
1. Jan Bygdevoll, "How to find field candidates for enhanced recovery by water additives on the
NCS"., Enhanced Recovery by water additives FORCE Seminar 2007
2. SPE-184101-MS "Novel Benchmark and Analogue Method to Evaluate Heavy Oil Project"s; L.
Jia, A. Kumar, R. Bialas, T.P. Lanson, and X.D. Jing, Shell International E&P Inc.
3. SPE-187780-MS "New Approach to Estimate Reservoir Complexity Index for West Siberial
Field"s; M.V. Naugolnov, M. Bolshakov, Gazpromneft; R. Mijnarends, Salym Petroleum.
4. SPE-134450 "Rapid Assessment of Potential Recovery Factor: A New Correlation Demonstrated
on UK and USA Field"s; L.M. Wickens, R. Kelly, RPS Energy.
5. SPE-136139-MS "Express Method of Oil Recovery Ratio Estimation On the Basis of Oil
Reservoir Statistical Characteristic"s; A. Roschektaev, A. Yakasov, V. Krasnov, Toropov K.
6. Xiao Qi Yeoh in collaboration with GCA, "Thin Oil Rim Reservoir Developmen"t.
7. Reservoir Benchmarking Tool (RBT) – User Manual; Laurent Souche & Nor Mahira Mahmod,
Schlumberger
