You are on page 1of 8

SPE 127235

Closing the Loop between Volumetric and Material Balance Estimates -


Naturally Fractured Carbonates
Rafi Mohammad Aziz and Qasem Dashti, SPE, Kuwait Oil Company, Manoch Limsukhon, Kassem Ghorayeb,
Subrata Chakraborty, and Hai Liu, SPE, Schlumberger

Copyright 2009, Society of Petroleum Engineers

This paper was prepared for presentation at the 2009 Kuwait International Petroleum Conference and Exhibition held in Kuwait City, Kuwait, 14–16 December 2009.

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
This paper shows a systematic approach to reduce uncertainty of volume, recovery factors and production forecasts with a
closed loop of static and dynamic information over the life of the reservoirs. A case study of a Middle Marrat reservoir block
in the North Kuwait Jurassic Complex (NKJC) is presented to illustrate the procedures. The NKJC comprises of six discovered
fields each consisting of at least four prospective producing units containing multiple fluid types at near critical conditions.
Intra-field compartmentalization due to fracture intensity, faulting and litho-facies variations resulting in porosity /
permeability pinch outs has further created isolated blocks with a high level of uncertainty in the reservoir parameters
describing the storage and flow characteristics necessary for forecasting well behaviors resulting from connected volumes. It is
vital to collect as much dynamic data as possible at an early stage of the development for reserve booking and production
forecast. Field development plan envisaged for the NKJC considered a three-phase development in order to mitigate CAPEX
risk, reduce uncertainty and optimize recovery. Each phase has sufficient gap (3-4 years) to drill wells, collect formation
evaluation and well performance data and adjust future plans as necessary. This paper describes the building blocks of
volumetric estimates and fluid flow characteristics, considering the uncertainty levels in litho-facies spreads, porosity /
permeability relationships, water saturations and fracture geometries etc. based on initial drilling and limited well tests. After
the start of production, fluid properties monitoring, collection of well surveillance data such as pressure vs. rate curves,
pressure transient analysis (PTA), production logging surveys (PLT), Modular Formation Dynamics Tester (MDT) and
material balance calculations indicate another set of storage and flow properties. Integrating these estimates with geology,
structural framework and fracture geometry is the key to obtaining a consistent reservoir description. Consistency in turn,
reduces uncertainty in the connected volumes and flow dynamics over time. Flow assurance calculations using simulation
model calibrated with the integrated set of parameters have been made with confidence to support a lifecycle reservoir
management strategy to optimize recovery.

Case study illustrates that Green Field development plans benefit greatly from early dynamic data collection integrated with
volumetric estimations by reducing uncertainty of reservoir description parameters that are used to calculate in-place
hydrocarbon volumes and reserves. Reservoir uncertainty in naturally fractured carbonates is maximum with consequent risks.
The methodology and work flow thus generated may be repeated for other blocks in the complex and for similar naturally
fractured reservoirs.

Introduction
In any oil company involved in Exploration and Production (E&P) activity, an exploration success creates an atmosphere of
optimism and expectations that is hard to contain. Professional technical staff of these companies is faced with the challenge of
assessing the true significance of a discovery, accurately estimating the discovered in-place volumes of hydrocarbons and
racing against time to develop and convert these reserves into financial success to maximize net present value of the
discovered assets. Unfortunately, uncertainty of subsurface reservoir descriptive parameters is highest in the early field life,
leading to a high risk of either under- or overestimating the in-place volumes and estimates of ultimate recoveries. Implications
of wrongly guessing on either side and inefficiently committing capital and human resources are an economic loss for the
project. Early development risk mitigation involves computing volumes with large uncertainty of input parameters, heavy use
of analogues to fill information gaps and multiple scenario analyses to arrive at a range of computed outcomes. Based on the
risk vs. cost of failure assessments, companies spend considerable time and resources to collect additional dynamic data and/or
adapt a phased approach to allocation of resources to mitigate risk. Either way, oilfield development is a journey on a learning
curve that concludes only with the last barrel of production from a reservoir.
2 SPE 127235

Initial histogram showing frequency of Range of HCPV Estimates after early Final range of HCPV estimates after
HCPV estimates dynamic data establishing decline trends and confirm
Reservoir Description parameters
Fig. 1—Progressive decrease in uncertainty level with time after production startup and integrating reservoir description with material
balance estimates during field life.

Most E&P organizations have well defined in-house processes to make assessments of volumes and recoveries at each stage of
the development. Capital allocation to the next phase depends upon satisfactory outcome and future expectations at key stages
in the development lifecycle. Fig. 1 shows the process of uncertainty reduction in hydrocarbon volume estimates at various
stages of reservoir development. Rose (2001) has described the systematic approach adapted by exploration companies in
developing probabilistic volumes and reserve estimates. After the start of production, continuous pressure data collection,
periodic shut-ins to obtain bottomhole pressure (BHP), PTA and other Decline Curve Analysis (DCA) methodologies better
reveal the extent of reservoir volumes. The final hydrocarbon pore volume (HCPV) estimates have a small uncertainty around
them, but the outcomes more often than not move asymmetrically around the expected (P50) pore volumes. The additional
dynamic data is integrated by a revisit of the original assumptions and adjustment of reservoir descriptive parameters such as
storage, shapes, block sizes, reservoir continuity, lateral variations and a host of other factors that combine to adjust the
original estimates. The final result generally does have a narrow range (±10%), but the value may move on either side of the
original probability frequency histogram. Kabir et al. (2005) showed a process of narrowing down the uncertainty with
multiple sandstone reservoirs, each having its own range of values.

Fig. 2—Initial range of matrix HCPV estimates in NKJC and its input sources to produce the multi-scenario analysis.

Description
Naturally fractured carbonate reservoirs present the most complexity to subsurface technical professional staff not only in
terms of lateral and vertical variations in the computed volumes, but also in the producing rates and recovery factors that
depend upon localized fracture intensity, orientation and matrix / fracture permeability contrast, among other parameters.
Given this level of complexity, it might not be possible to justifiably use analytical tools alone to arrive at credible estimates
and performance forecast scenarios. Geostatistical models and upscaled flow (dynamic) models are better suited at any stage
to handle this type of reservoirs with multiple interacting parameters. Bahar et al. (2005) and Bockel-Rebelle et al. (2005)
showed, using practical examples, a systematic approach to build carbonate reservoir models and multi-scenario analysis to
handle uncertainty. NKJC carbonate reservoirs spread over an area of 1800 km2. Sparse initial drilling combined with a less
than perfect seismic image, multiple facies, only the hydrocarbon down to (HDT) estimates from the wells, and resulting
porosity and water saturations variations made it virtually impossible to estimate the HCPV with a good deal of certainty.
Combining all these factors with depositional and diagenetic vagaries that are inherent in the nature of carbonate rocks, a range
of HCPV values were obtained as shown in Fig. 2 . The minimum (P90) estimate was at least one third the maximum (P10)
value.
SPE 127235 3

Fracture intensity in these carbonates also varies with the


regional and localized tectonic stress conditions and
lithological variations. Volume contained in these
fractures may be significant enough to cause high initial
production rates, which could decline exponentially once
fracture volumes are exhausted. On the other hand, sub-
seismic faulting and sealing capacity of faults may create
smaller than anticipated blocks. Fig. 3 shows the
distribution of fracture porosity derived from core
descriptions and limited fracture image logs run in the
Fig. 3—Fracture porosity distribution contributing towards
wells drilled with oil-based mud. Fracture intensity and
HCPV estimates in layers 1-20 of NKJC.
apertures in turn combine to produce the initial rates on
tests that are higher than predictable by matrix properties
alone.
(1)
Reservoir complexity of NKJC has been described in
many SPE papers and presentations showing the
(2)
probabilistic nature of input parameters for HCPV (7)
calculations by each reservoir and field (Ghorayeb et al.
2008, 2009; Limsukhon et al. 2009; Rao et al. 2009).

In addition to the HCPV, fracture geometries and matrix /


fracture permeability contrasts combined with fluid (6)
property variations yield a spectrum of recovery factors, (3)
and therefore Estimated Ultimately Recoverable (EUR) oil
and gas also vary significantly on the initial probability
curve. Dynamic data combined with data obtained from
further drilling programs is needed to narrow the range of (5) (4)
HCPV and EUR estimates. The process to verify volumes,
reserves and production forecasts over time and reduce
uncertainty is shown as a flow chart in Fig. 4.

Fig. 4—Process flow diagram illustrating the uncertainty


reduction process while updating the reservoir models.

Case Study
The closed loop process can best be illustrated with an example of one of the NKJC reservoirs identified as SA-MM, for which
initial estimation of a range of input parameters and computational results were modified in several stages through volumetric
and dynamic data integration to increase confidence level and decrease uncertainty of the HCPV and EUR estimates. The
following is a description of certain key parameters and their range of values.

Fluid Property Estimation


The reservoir contains a gas–condensate fluid at an
original reservoir pressure of 10,900 psi. Several reservoir
fluid samples were taken from three wells initially drilled
in the field. The variation of key components / pseudo-
component groups (C1, C2-C6, C7+ and NHC) of these
reservoir fluid samples is shown in Fig. 5. It can be seen
that the variation of the fluid composition amongst these
three wells are in line with each other (excluding three
suspicious samples: SA-1 RSS1, SA-1 RSS2 and SA-3
WHS). The variation could possibly be explained by
different sampling depths between wells as well as
different sampling locations from the same well. The
compositions, phase diagram and Constant Volume
Depletion (CVD) experiment results suggest that the
reservoir fluid is near critical gas-condensate. In-situ fluid
behavior is more complex and near-wellbore condensate
Fig. 5—Compositional variation in SA_MM between the
banking might have a significant effect on well
original three wells.
productivities (Fig. 6).
4 SPE 127235

Volumetric Uncertainty
At the initial estimation stage, static reservoir model in the structure was built from an openhole log and production tests from
a single well. The porosity and water saturation distributions from logs are shown in Fig. 7 and Fig. 8, respectively. The
discrete fracture network (DFN) was built from core descriptions, drill time curves and seismic attributes to estimate the
fracture distribution and properties. Initial distribution of fracture porosity is shown in Fig. 3.

a b
Fig. 6—Phase diagram (a) and liquid saturation from a CVD experiment (b) of a representative sample from SA MM.

Fig. 7— Matrix porosity distribution. Fig. 8— Water saturation distribution.

The main SA structure is a narrow steeply folded, doubly plunging anticlinal


feature, with a high surface curvature which suggests high fracture intensity in
this area. This is consistent with fracture observations from cores and logs. Fig. 9
shows the original reservoir extent considered to be connected to the main
discovery wells on this structure.

The original three wells in the structure were put on Long Term Test (LTT) for a
minimum flowing period of two months each followed by a pressure buildup test
(PBU). Fig. 10 shows pressure difference and derivative responses from pressure
buildups of these three wells on log-log plots. It is clearly shown that the flow SA‐2

mechanism of these three wells is fracture dominated, in which the dual-porosity Discovery Well     
SA‐1
behavior is observed within 2 hours of the onset of pressure buildup period. This
indicates that the wells are intersecting with a large fracture network. Multiple
PBUs were performed on well SA-1 as shown in Fig. 11. The dual-porosity
SA‐3
behavior in all three PBUs acquired in SA-1 occurs almost at the same time
(around 0.1 hrs). The dual-porosity behavior finished around 10 hrs, after which
the total system pseudo-radial flow is observed. Note that the total system
permeability-thickness (kh) was reduced in the third PBU period which may Fig. 9—SA-MM structure with its
suggest that either the system permeability or the contributing interval was bounding faults and non-interference
reduced with time. In either case, this suggests that the flow properties of this between SA-1 & SA-2 on the Western
area are largely dependent on the geomechanical properties of the fracture downthrown block.
system.

The integration of petrophysical logs and facies distribution for volumetric estimate of the main SA structure showed that the
normalized value of initial volumetric estimate ranged between -1.5 to +1.8 with an expected value of 1.0, i.e. the maximum
reservoir volume due to uncertainties in the porosity, water saturation in matrix and fractures given the container size was
fixed by faulting on both sides of the structure.
SPE 127235 5

LTT was performed on the well, followed by bottomhole shut-in pressure, PTA tests and material balances to confirm the
connected volume in well SA-1. Two scenarios were considered in estimating reservoir pressures for material balance
calculations:

1. Reservoir pressure is equal to the final shut-in pressure after a long shut-in period.
2. Reservoir pressure is equal to average pressure (Pavg) calibrated from extrapolated pressures (P*) from Horner's plot.

SA-1

SA-2

SA-3
Fig. 10—Pressure responses of original three wells in Middle
Marrat.

Fig. 11—Pressure build-up responses of Well SA-1 at three


different periods.

The connected volume obtained in scenario 1 confirmed the minimum volume while the one obtained from scenario 2
represented the maximum volume. The PTA analyses of all three wells showed matrix / fracture interaction and dual-porosity
behavior. However, a short term interference test did not show a connection between well SA-1 and SA-2 well blocks (Fig.
11).
6 SPE 127235

Well SA-1 was subsequently connected to the first phase production facility and continuously produced for more than 9
months. Two more PTA data sets and reservoir Pavg values were obtained that led to confirmation of fracture contributions, but
the computed volumes from an extrapolation of P/Z data became progressively larger (Fig. 12). The last P/Z plot showed at
least three times the maximum normalized HCPV (Fig. 12 - b). This additional connected volume suggests that the secondary
storage may exist that supports the pressures.

a- Initial P/Z, well SA-1


b - P/Z plot, well SA-1, after Startup

c - P/Z plot, well SA-2 d - P/Z plot, well SA-3


Fig. 12—P/Z plot for original three wells in SA-MM structure.

Production Decline Curve Analysis


The production rates of well SA-1 were plotted against time in semi-log scale to estimate the ultimate recovery as shown in
Fig. 13. Well performance during LTT (Fig. 13) indicates a very small decline after the production was quickly stabilized.
Production decline from dynamic model (full field model) is shown in Fig. 14. The forecasted production shows that based on
the current assumptions, the structure would not be able to produce the estimated EUR as per DCA. This suggests that either
the structural closure is larger than initially thought or there may be some sort of pressure support (water drive?) in the
reservoir.

Fig. 13—Gas production rate vs. cumulative gas production of Fig. 14—Comparison of production rates from well SA-1 vs.
well SA-1. model prediction.
SPE 127235 7

There are many scenarios under which a well would show a similar P/Z vs. cumulative production behavior:
a) An active aquifer supports the pressure. However, the known structure's HDT depth from drilling do not support this
assertion. Fig. 15 is a structural cross section through the discovery well. None of the wells on the narrow structure
encountered a gas / water contact.
b) The matrix porosity and water saturation computed from logs and cores may be too conservative. A revisit of the logs
from discovery and other wells and comparison with core analyses confirms that the range of numbers cannot support
the huge increase in pore volume computed from material balance analysis.
c) Fracture intensity, lengths and apertures would support some increase.
d) Rock matrix compressibility would initially support a higher estimate.
e) The bounding faults on east and west flank, if not sealing, will also behave as a conduit for moving hydrocarbons
and/or water to the formation.
f) Another well drilled to target the eastern downthrown flank of the structure proved that movable hydrocarbons exist
in that flank. However, there was no pressure depletion observed in the MDT data collected in the MM layers.
g) Porosity cutoff used in the calculation of original HCPV was 2%. Removal of this cutoff with some increase in
fracture intensity will also support the larger computed volumes from material balance analysis.

In short, the worst case scenario may be narrowed down to water influx and the best volume computation involves removal of
cutoffs and increase of fracture intensity. Further production and pressure data collection is expected to narrow down the
actual reason for the increase of connected volumes. In turn, Expected Recovery Factors and reservoir management strategy of
either outcome will be different. Fig. 16 shows the latest structural configuration of MM reservoir after incorporation of
dynamic data, well pressure behavior and confirmation of hydrocarbons on both East and West flank of the structure. Even
though the discovery well is on the top of the structure, a gas-water contact has not been observed in subsequent saddle wells.

Fig. 15—SA-MM cross section through the discovery well and


showing faults on both flanks. Fig. 16—SA-MM structure after drilling four additional wells
and after production and dynamic data collection.

Conclusions
1. Naturally fractured carbonate reservoirs are the most complex systems. Original computations of HCPV have a wide
range of uncertainty. Multi-scenario analysis and modeling techniques as well as heavy use of analogs are required to
arrive at the range of production forecasts at the initial stage.
2. Initial probabilistic range of values for both the HCPV and EUR can be narrowed down after collection of additional
pressure and production data and incorporation of up to date reservoir geometry, based on material balance and PTA
indicators.
3. Multiple scenario analysis using reservoir models facilitates the narrow down process of numbers, but often there is
no substitute for time. Extrapolation of short term data may result in erroneous results.

Recommendations
1. Naturally fractured carbonates are too complex to venture on conventional volumetric computations and analytical
modeling. Geostatistical modeling and probabilistic methods should be applied to compute the range of possibilities
under various scenarios even when initially data sources are limited. Further appraisal and delineation work can then
be systematically performed to obtain additional data.
2. Analogues may have a limited value when using localized structural and lithological complexities to compute initial
volumes. A well thought out strategy comprising of a combination of static well data and location picks with
interference testing will help to narrow down the ranges.
3. Continuous dynamic data collection and calibration of volumetric input parameters with knowledge of drive
mechanisms are a must to narrow the range of HCPV and EUR estimates.
8 SPE 127235

Acknowledgements
The authors would like to thank Kuwait Oil Company for permission to publish this paper. The North Kuwait Jurassic
Project’s team members are gratefully acknowledged for contributing to the paper.

Nomenclature
BHP = bottom hole pressure, m/Lt2, psi
DCA = decline curve analysis
DFN = discrete fracture network
PBU = pressure build up
PTA = pressure transient analysis
HCPV = hydrocarbon pore volume
HDT = hydrocarbon down to
EUR = estimated ultimately recoverable
K = permeability, L2, md
MDT = modular dynamic formation tester
MM = middle Marrat
LTT = long term test
NKJC = North Kuwait Jurassic Complex
P = pressure, m/Lt2, psi
P10 = less that 10% probability
P50 = 50 % probability
P90 = more than 90% probability
Phi = porosity
Q = flow rate, L3/t, Mscf/D
Z = compressibility factor
Subscripts
avg = average
References
Rose, P.: “Risk Analysis and Management of Petroleum Exploration Ventures”, AAPG methods in Exploration Series, No. 12.

Kabir, C. S., Gorell, S. B., Portillo, M E., Cullick, A. S., 2005. Decision Making with Uncertainty While Developing Multiple
Gas/Condensate Reservoirs, paper SPE 95528 presented at the 2005 SPE Annual Technical Conference and Exhibition, Dallas, Texas,
9-12 October.

Bahar A., Ates H., Al-Deeb, M. S., Salem E. S., Badaam H. and Mohan K., 2003. Practical Approach in Modeling Naturally Fractured
Reservoir: A Field Case Study, paper SPE 84078 presented at the 2003 SPE Annual Technical Conference and Exhibition, Denver,
Colorado, 5-8 October.

Bockel-Rebelle M.O., Dabbour Y., El Abd Salem S.,Vesseron M., Silva F.P., 2005. Fault and Fracture Corridors-How to Reduce the
Structural Uncertainty for Reservoir Management Optimization, paper SPE 93752 presented at SPE Middle East Oil and Gas Show and
Conference, Kingdom of Bahrain, 12-15 March.

Ghorayeb K., Limsukhon M., Dashti Q., and Aziz R. M. 2008. Multiple Reservoir Simulations Integration: 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, Denver, Colorado, 21–24 September.

Ghorayeb K., Limsukhon M., Dashti Q., and Aziz R. M. 2009. Black Oil Delumping: Running Black Oil Reservoir Simulations and Getting
Compositional Wellstreams in the North Kuwait Jurassic Complex, paper SPE 118850 presented at the 2009 SPE Reservoir Simulation
Symposium, The Woodlands, Texas, 2–4 February.

Limsukhon M., Ghorayeb K., Aziz R. M., Narhari S. R., and Chakraborty S. 2009. Calibration of a Discrete Fracture Network Model with
Well Test Data – A Case Study of the North Kuwait Jurassic Complex, paper SPE 115566 presented at the 2009 SPE/EAGE Reservoir
Characterization & Simulation Conference, Abu Dhabi, 19-21 October.

Ghorayeb K., Tan L. H., Limsukhon M., and Aziz R. M. 2009. Ensuring Water Saturation Consistency between Static (Fine Grid) and
Dynamic (Upscaled) Models – A Case Study of the North Kuwait Jurassic Complex, presented at the 2009 SPE/EAGE Reservoir
Characterization & Simulation Conference, Abu Dhabi, 19-21 October.

Chakraborty S., Rao N. S., Al-Awadi, M. A., Kidambi, V. K., AL-Houli, M. M, Al-Ateeqi, K. A., Perrin, C. and Ghorayeb, K., 2009. 3D
Geological modeling of a tight naturally fractured carbonate reservoir as an input to reservoir simulation – a case study from North
Kuwait Jurassic fields, paper SPE 127559 to be presented at 2009 Kuwait International Petroleum Conference and Exhibition, Kuwait
City, Kuwait, 14–16 December.

You might also like