You are on page 1of 12

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/311342613

Importance of conceptual geological models in 3D reservoir modelling

Article  in  First Break · July 2016


DOI: 10.3997/1365-2397.2016010

CITATIONS READS

2 380

5 authors, including:

J. Cavero Ney Orellana


College of Engineers of Peru Repsol
6 PUBLICATIONS   9 CITATIONS    9 PUBLICATIONS   65 CITATIONS   

SEE PROFILE SEE PROFILE

Ivan Yemez Virendra Singh


Repsol Repsol
24 PUBLICATIONS   21 CITATIONS    28 PUBLICATIONS   103 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Working in clastic and carbonate development projects. View project

Reservoir properties for oil and gas prospects, geothermal prospects and mineral resources. View project

All content following this page was uploaded by Virendra Singh on 15 December 2017.

The user has requested enhancement of the downloaded file.


first break volume 34, July 2016 technical article

Importance of conceptual geological models


in 3D reservoir modelling
J. Cavero1, N.H. Orellana1, I.Yemez1, V. Singh1* and E. Izaguirre1

Abstract
Conceptual Geological Models (CGM’s) represent geological knowledge and are traditionally visualized by 2D geologic cross-
sections. Construction of CGMs is a non-linear and complex process which involves application of geological rules and experi-
ence. These models describe essential features of geological situations, illustrate the principal processes of the petroleum system,
and provide important information about reservoir characteristics, pressure and fluid flow of the field under study. The CGMs
are being extensively used as the key input for 3D reservoir modelling and simulation at different stages of E&P projects. A fully
integrated CGM building process consists of at least four stages: construction of one or more structural models, identification
of depositional models, construction of sedimentary facies models and finally building of diagenetic facies models. This process
requires integration of geological, geochemical, geophysical, and petrophysical data along with information from well testing
and production into a comprehensive model describing the physical features of the geologic system.
This paper briefly describes the different components of CGM and model building process. The importance of integrating CGM
in the 3D reservoir models has been demonstrated through an example of siliciclastic reservoir. This reservoir has complex
internal sedimentary distribution and stratigraphy which leads to high subsurface uncertainties. The study shows that different
possible conceptual geologic scenarios provide significantly different Hydrocarbon-In-Place, recoverable resources and produc-
tion forecasts. Therefore, the successful application of appropriate CGM’s is critical for better 3D reservoir characterization
and modelling which enables us to identify and rank the key reservoir uncertainties and assess their impact, establish interde-
pendency of spatial property distribution, make volumetric assessments, plan the number and location of wells including their
drainage area, optimize recovery efficiency and obtain reliable production forecasts.

Introduction limited data and its quality, multiple possibilities of the


Nowadays, three-dimensional reservoir interpretation, geologic interpretations, reservoir geometry, areal extent
modelling, and flow-simulation studies are used as com- and thickness variations which lead to uncertain reservoir
mon tools to generate production forecasts and to economi- characterization (Singh et al., 2013a). Despite the boom of
cally support value-based E&P business decisions. One of stochastic simulation owing to computing advances in recent
the key challenges in 3D reservoir modelling is distributing years, most of the stochastic models suffer due to lack of
the identified facies and their properties, and integrating geologic soundness (Tamhane et al., 1999). Most of these
lateral and vertical variability in reservoir quality within 3D reservoir models either do not incorporate the CGM
and between different flow units in the defined 3D frame- or use only 2D models qualitatively. As a result, the oil and
work (Orellana et al., 2014). This requires incorporation of gas industry has witnessed mismatch between actual and
the CGM in the integrated 3D reservoir modelling work- simulated production profiles in both brown as well as green
flow through use of appropriate geostatistical techniques fields (Singh et al., 2013b). In some cases, geoscientific and
(Orellana et al., 2015). The CGM represent geological engineering data is not enough. In other cases, geostatistical
knowledge and is used as a guide to construct the 3D res- techniques are not appropriately utilized. All these aspects
ervoir model (s) at different stages of an E&P project based contribute to poor assessment of subsurface uncertainties
on real and or conceptual geological objects and processes and it remains a key issue for any 3D reservoir charac-
(Quinto et al., 2013). terization and modelling. Therefore, it is never possible to
The major challenges in building the CGMs are the guarantee only one accurate ‘Best Estimate’ even in more
availability of the subsurface information, data from differ- matured fields where abundant data are available. The best
ent disciplines with the variable scale of resolution (micro geoscientists and engineers can do is to provide a range of
to mega), lack of enough geological knowledge and proper equi-probable volumetric estimates.
data integration (Al-Khalifa et al., 2007). Some of the most To properly define and quality-control the 3D modelling
common uncertainties in the geological studies are: the workflows a detailed interpretation of geological aspects of

1
Repsol, Méndez Álvaro 44, 28045 Madrid, Spain.
*
Corresponding author, E-mail: vsingh@repsol.com

© 2016 EAGE www.firstbreak.org 39


technical article first break volume 34, July 2016

Figure 1 Schematic diagram showing input data


and output scenarios that can be generated and
analysed from combinations of critical components
of CGM. Number of CGM scenarios will depend on
available data and uncertainties of each component.
Depending on geological complexity, scanty informa-
tion, ambiguous position of the field in the geological
configuration, several scenarios for each component
can be generated. A combination of these scenarios
leads to multiple cases of deterministic CGMs which
can be used as the input for capturing the geological
uncertainties.

the reservoir – including environment, sequence stratigraphy, incorporate possible scenarios of structural/stratigraphic
areal extent, pore space characteristics, reservoir thickness, configurations, permeability systems, geochemical factors,
porosity, and permeability – is absolutely essential. In order evolution, temperature evolution and distribution, seal and
to capture a whole range of the uncertainties, it is important trap evolution etc. Construction of CGMs is in fact a non-
to understand the complexity of different components of linear and complex process which involves application of
the CGMs and integrate them in the 3D reservoir model. geological rules and geologist’s experience and in turn helps
Without a realistic geological framework and an appropri- to build a geologically consistent 3D reservoir model.
ate 3D modelling workflow, no 3D reservoir model can be Components of CGM: Building a fully integrated CGM
used reliably as a predictive tool for hydrocarbons In-place, involves at least four stages: construction of structural,
recoverable resources and production forecasts. stratigraphic, sedimentary and diagenetic facies models. This
To demonstrate the impact of CGMs, an example of a requires integration of geological, geochemical, geophysical
siliciclastic reservoir, deposited in distal-fluvial sequences, is and petrophysical data along with information from well
illustrated. Based on 3D seismic and well data, a trap was testing and production. CGM should be understood either
interpreted as a well-defined anticline. Available field data as a map or set of maps of geological concepts/components
indicate that the internal sedimentary configuration and and relationships, between them taking into account mul-
its continuity away from the wells in the field carry high tiple possible scenarios that may cover the uncertainties.
uncertainties. Several scenarios of well logs correlation were Depending on the complexity of the field, CGMs can relate
generated considering multiple criteria, keeping in view a to the components for each geological entity, singly or
complex internal sedimentary configuration and a wide integrally, to specific objectives and to a set of key uncertain-
range of feasible channel size with respect to field positions ties that are required to be analysed. The criticality and
for the distal fluvial system. These well correlation scenarios importance of the specific component of CGM and/or their
were used as the basis to define multiple possibilities of the combination will vary from field to field.
CGMs. These CGMs were used as input in different 3D CGM Building Process: The CGM building process
static reservoir models. Detailed results are discussed in this integrates geologists’ knowledge, experience and imagina-
paper. tion and probable geological objects to build the geological
scenarios. Each input used in the CGM building process has
Conceptual Geological Models (CGM) its own uncertainty. The uncertain structural configuration,
It is necessary to know: What is a CGM? What are differ- overlapping of the sedimentary environment, uncertain verti-
ent components and geological processes for CGMs? What cal and lateral continuity, uncertainty in post-depositional
are the data required for building CGM? What will be the changes etc. form an integral part of the conceptual model
benefit for having more than one model? What if the newly building process. Therefore, use of single deterministic geo-
created CGM’s are totally wrong? What is needed to reduce logical concepts, particularly during the appraisal and early
the risk of getting wrong CGMs? What are the most effective development phases, is too risky.
ways to utilize CGM for different applications? Recent technological advances, in terms of computing
CGM is an expression of geological knowledge which and 3D visualization, allow effective integration of avail-
represents geological objects spatially in terms of their able data, geological knowledge and geologists’ wisdom to
physical features based on evidences and or analogues. They visualize impact of different critical components (structural,

40 www.firstbreak.org © 2016 EAGE


first break volume 34, July 2016 technical article

stratigraphic, sedimentary, diagenetic and other processes), Example illustrating the importance of multiple
different cases and their combinations in the generation CGM scenarios
of multiple CGM (Figure 1) within shorter timeframe. To demonstrate the importance of the CGMs on 3D reservoir
This plays an important role in capturing the subsurface interpretation and modelling outcomes, a real field example
uncertainties in terms of reservoirs quality and their property was taken. The available data for this example is summarized
distribution. in Table 1.
During the CGM’s building process, it is recommended to The discovered sandstone reservoir’s gross thickness
take the following into account: varies from 25-35m and net pay from 10-25m. It has well-
n Analyse validated data and information from wells, defined structural boundaries. The interpreted oil water
seismic surveys, outcrops, regional geological setting, contact (OWC) for the field is at around 1371 m TVDSS.
basin analysis, play concept and analogues related to the To build fully integrated and representative 3D geomodels,
prospect, discovery or field to be evaluated. geophysical, geological, petrophysical and engineering data
n Ensure that the CGM represents the subsurface were used in two steps:
configuration with real geological objects avoiding n Conceptual geological models building
artefacts or software synthetic events. Consider n 3D geomodel building using CGMs as inputs
qualitative and quantitative descriptions of the geological Conceptual geological models building: In order to produce
objects or geological concepts, internal and external an improved 3D seismic volume with better signal to noise
geometrical characteristics of different scales, data and ratio and attenuated noise, a seismic data enhancement pro-
evidence from wells, field and or analogues. cedure was applied in the original 3D PSTM volume. The
n Consider description of static geometry and kinematic procedure for construction of conceptual geological models,
history of geological structure, rock bodies, tectonic described in the previous section, consists of four compo-
processes and/or assumptions. nents: (1) structural, (2) stratigraphic, (3) sedimentary, and
n Consider classification of range of rock sequences (4) post depositional processes.
into litho-stratigraphic, bio-stratigraphic, chrono-
stratigraphic, rock units, etc. Structural component
n Consider all possibilities of trends and interpretation This component incorporates the spatial positions of major
for the structural, stratigraphic, sedimentary, diagenetic boundaries of geobodies, faulting, folding, and erosion etc.
elements which affect the reservoir quality. to define the framework of the traps, seals and reservoirs.
n Consider possible reservoir compartmentalization which These structural components might have control on the
may have impact on hydrocarbon recovery. sedimentation trends or natural fracturing and may provide

Regional n Regional geological studies: Geological history of the basin, structural evolution, stratigraphic
information correlation.
n Basin analysis and play concept: Source (volume, richness, maturation), Timing, migration
and preservation (timing of closure, expulsion, effective migration fairway, and preservation),
Reservoir (presence, quality and reservoir performance, trap geometry, closure: map reliability
and control, trap model confidence) and Seal (top, lateral and base seal).
n Rock outcrops associated to the fluvial and tidal environments with high variability. Facies
association in the siliciclastic sequences deposited in proximal and distal fluvial environments
leading to highly uncertain vertical and lateral continuity.
n Analogue fields in and around the area. These analogues with similar geologic characteristics
indicate that field performance is well below the production forecasts used in defining the field
development plan.
n Public domain database for structural, stratigraphic and sedimentary configuration including
geological parameters ranges (published literature, third party such as ERGO CGG, C&C, etc.).
Drilled wells n Reports: Well, geological studies, mud-logging reports, core sample description image logs,
(data and reports) conventional and core analysis reports. DST (2 wells) and MDT (2 wells) Analyses reports.
n Digital data: Open hole and cased hole wireline logs from 11 wells, image logs from 4 wells, and
mud logs from 11 wells, well tops and geological markers, conventional and special core analysis
from 4 wells, DSTs and MDTs (2 wells each).

3D seismic, check n Reports: 3D Acquisition, processing and interpretation, Check shots and zero offset VSP reports.
shots & VSPs n Digital data: T-D curves from 2 wells, VSP T-D curves and corridor stacks from 2 wells, 3D seismic
(data and reports) pre-stack time migrated cube covering around 60 km2 (full fold), integrated 3D velocity cube.
Table 1 Summary of available digital data and reports used for this study.

© 2016 EAGE www.firstbreak.org 41


technical article first break volume 34, July 2016

Figure 2 Depth maps for (a) reservoir top, and (b)


reservoir base. Top and Base reservoir maps, gener-
ated using the 3D seismic velocity model and well
data, show similar structural trends.

Figure 3 Illustration showing the tectonic factors,


structural framework and elements to define the
type of traps. A real structural object from the out-
crop (Source: Wilson, 2010) is used as analogue to
make correlation with 3D seismic data for creating
the 3D structural framework of geological model.

explanation for particular characteristic of part or whole Stratigraphic component


of the reservoir. Integration of regional tectonic setting, Identifying the stratigraphic sequences variability and com-
outcrop evidences, structural configuration from 2D/3D plexity, their hieratical orders and stacking patterns require
seismic, play and analogues fields and geologic data from more specific data and the robust methodologies to make
wells define the possible structural scenarios for the field flexible stratigraphic interpretation (Catuneanu, 2011).
or area under study. Velocity is one of the key parameters Establishing the link between lithology and stratigraphic
which affect the seismic imaging of the subsurface in time sequences is primarily essential whether they are siliciclastic;
or depth. Velocity uncertainty together with reservoir top/ a carbonate system, intercalation of carbonate and siliciclas-
base interpretation uncertainty can have high impact on tic or even volcanic sedimentary sequences. Additionally, the
the structural configuration, gross and net reservoir thick- stratigraphic sequences types and their areal distribution are
ness away from well control areas and hence gross rock also the most critical information required for defining the
volume estimation (Singh et al., 2009). Depending on 2D/3D geological models.
the velocity uncertainty combined with prior geological The integration of the regional stratigraphic setting,
knowledge, single or multiple structural scenarios can be geological outcrops, seismic sequence stratigraphic inter-
defined. pretation and well correlation (Singh et al., 2013b) allows
For this example, using enhanced 3D Pre-stack Time for capturing the time framework for correlating and
Migrated (PSTM) seismic data, welt-to-seismic tie and VSP mapping of sedimentary strata, fosters integration between
data at key wells, seismic interpretation was carried out stratigraphy, paleontology, geochemistry, petroleum geol-
for defining the structural configuration. The structural ogy and helps in defining the stratigraphic scenarios. It is
trap appears to be a coffer-shaped-antiform with four-way absolutely necessary to include the history of geological
closure calibrated by 11 drilled wells tied to the seismic. evolution for the basin, play and field, mechanism of basin
The interpreted reservoir top and base time horizons were filling, definition of the mega sequences, seismic sequences
depth converted (Figure 2) using a 3D velocity model built and unconformities, type of internal sequences and possibil-
integrating Check-shots/VSP and seismic velocities. The ity of multiple interpretations. Sequence stratigraphy helps
conceptual structural model (Figure 3) was interpreted as us to understand systematic and orderly arrangement of
strike slip faults bounded by a structure generated by com- facies and their locations.
pressional stress. Seismic evidences are very well correlated The enhanced 3D PSTM volume was used for sequence
with outcrop analogue. Due to good number of drilled wells stratigraphy study. Seismic data shows plane parallel strati-
and available velocity data, the structural component does graphic sequences close to the top and the base of the
not carry high uncertainties. Therefore, the only single struc- reservoir. The base shows a minor erosional unconformity
tural framework was used as input in creating the 3D static probably because of action of the fluvial channels in the
model. sequence. As reservoir thickness variation (25-35m) con-

42 www.firstbreak.org © 2016 EAGE


first break volume 34, July 2016 technical article

forms to the sedimentation trends, only one stratigraphic excellent example showing alternate sedimentological mod-
scenario (Figure 4) is considered for CGM building. els, for clastic reservoirs deposited in a fluvial environment,
generated using same data with different assumptions.
Sedimentary components The well logs correlation of 11 drilled wells in the field
Regional sedimentary trends, 2D/3D seismic data, analogues show complex sedimentary trends. The sandstone reservoirs
and public databases (e.g. ERGO Robertson CGG) are used encountered in wells show intercalation of sand and shale
to interpret the sedimentary trends including facies and their layers with different configurations and apparently short-
association. However, this interpretation may not be unique distance vertical and lateral variations. Based on analysis
particularly when data is limited and sparse. It becomes even of conventional cores from four wells, several sedimentary
more complicated when the reservoir bodies have smaller facies associated to the fluvial and tidal associations were
dimensions and are discrete in nature. Conventional core defined indicating different geometry and distribution of
data can provide clear evidences for some sedimentary facies channels owing to possible variation of the gradient in the
but only at the cored location. Away from the core interval, distal part of the fluvial system. Cross-bedding with different
it is possible to find facies of some other environment(s) geometry indicates different energy and different granu-
of variable vertical/horizontal scales and geological time. lometric distribution within similar facies with different
Seismically derived facies have inherent limitations owing matrix and mineralogical composition (Figure 5). Intervals
to vertical resolution and may lead to several scenarios. of bipolar cross bedding, mud-drapes and thin bed of red-
Sometimes, outcrop evidences provide meaningful informa- black shales in the well and core suggest small periods of
tion about facies distribution and the environment. non-reservoir sedimentation with tidal influences. Reduced
The litho-stratigraphic unit can consist of several sequenc- presence of paleosols and mud floodplain explains that the
es and/or several reservoir units within it and may be isolated mechanism of sedimentation was progradation of the point
owing to genetic depositional architecture of sedimentary bars and sands lobules reworked over the time with less
distribution (Yu et al., 2013). Sedimentary components in aggradation influences.
siliciclastic sequences, can significantly affect the mobility The offset analogue fields are located far from this field
of fluid and reservoir productivity. The partial lateral and and they do not provide enough confidence. The sedimentary
vertical sealing concept could be an important issue to be facies evidence suggests a possible complex configuration in
considered in the sedimentary studies. Uncertainty in sedi- term of vertical/lateral variation and dynamic factors but
mentary evolution interpretation can lead to several possible their assumptions can help in defining the distribution of the
scenarios using the same data. Miall (2006) has published an quality and quantity of the good rock reservoirs.

Figure 4 Seismic section (2) passing through the field shows plane parallel sequences and it is corroborated with the formations top and base obtained from
the well logs correlation (3). Outcrop image from the Colorado canyon (1) shows similar plane parallel stratigraphic configuration (Source: Seo-Websitedesigner,
2012) which is used as an analogue to build a conceptual stratigraphic model.

Figure 5 Core samples (1 to 9) showing different


sedimentary structures and facies in the fluvial
environment. Samples (1 to 4) show different types
of fluvial cross-bedding owing to different energy
associated from close continental to tidal zones.
Samples 5 and 8 represent laminated facies in fine
grain sands with draping of limolites, clay and
carbonaceous lamination of quiet and low energy
of floodplain and levee zones. Sample 6 shows bio-
turbation and remainant red soils. Sample 7 shows
the variation of type of sand sedimentation from
fine cross bedding to massive blocky sand. Sample
9 shows bioturbation roots.

© 2016 EAGE www.firstbreak.org 43


technical article first break volume 34, July 2016

According to the regional setting, the main trend of ferent types of channels were visualized in the distal fluvial
fluvial deposition was SE-NW, overlying basement with environment as possible scenarios and four different types of
constant and moderate rate of sediment discharge into a well log correlation were performed based on Gamma Ray
relatively quiet tectonic basin with minimum subsidence and DT patterns.
and inclination. Rambling local trends, variable distribu- Consideration of different sedimentary environments
tion of the channel and crevasse splays, internal facies allows for the capturing of high uncertainties in the predic-
configuration, different sedimentation mechanism and tion of reservoir quality. Likewise, several other criteria
sediment loading could be responsible for the complex including internal correlation, geometry and facies distribu-
distribution of reservoir facies and associated properties. tion can be applied to create conceptual models, as any
Fields can have been located in a distal fluvial system single deterministic sedimentary model is neither enough nor
where the reservoir properties distribution can be different adequate to capture geologic uncertainties.
(Figure 6). Four scenarios of well correlation (1: Following top,
Different distal fluvial facies types in this siliciclastic 2: Following base, 3: Following internal slope, and 4:
sequence can be present vertically and laterally in a relatively Proportional layering from reservoir Top to Base) were
short distance and the sediments can change the proportion considered (Figure 7) and respective 3D grid layering was
from blocky sand to inter-bedding of sand and clay sequences used for further modelling. 3D Seismic data considered in
and can also have variation of shape and thickness from flat the interpretation has provided some possible evidences to
to sigmoidal sedimentation (Galloway, 1981). A set of dif- support the well correlations.

Figure 6 Uncertainties of the position of the field in the sedimentary environment can modify the areal and vertical distribution of the reservoir properties (e.g.
porosity and permeability). This variation can positively or negatively affect the production rate and final recovery of the resources.

Figure 7 Illustration of four different well-log correlations with the same data to cover the uncertainties owing to the complexity of distal fluvial sandstone
deposition. Scenario-1: Following the reservoir top, Scenario-2: Following the reservoir base, Scenario 3: Following the internal slope different from the dip of
the reservoir top and supported by seismic evidences, and Scenario-4: Considering proportional layering from reservoir Top to Base. Internal configuration of
sand and shale distribution will change according to the specific scenario.

44 www.firstbreak.org © 2016 EAGE


first break volume 34, July 2016 technical article

Figure 8 The core images and thin sections show the diagenetic effects on the reservoir interval. This shale-sand intercalation and or overlapping of sedimentary
sequences along with diagenesis can affect the overall performance of the reservoir/field, if these uncertainties are not accounted for properly.

Figure 9 ERGO Public Data Base of clastic depositional environment used as reference to define: (a) lithology relationship between thickness and width, (b)
fluvial channels facies association relationship between length and width (i.e., size and geometries).

Post depositional geologic processes (diagenesis) components but some meniscate filled burrows and roots are observed in
Diagenesis has been identified as one of the most compli- the cores. The most predominant mineral is Quartz (75%),
cated and lesser addressed components. Sometimes, it is even Feldspars (10%) and rock fragments and clays (10-15%)
completely ignored while building the 3D geological model. that composed the matrix (Figure 8). Clay minerals are
Neglecting the uncertainty of diagenetic processes on the infiltrated between the grain quartz pores partially, affecting
reservoir properties (porosity and permeability) could be one the effective porosity. Likewise, porosity reduction is also due
of the main causes that can severely affect the commercial to the compaction of rock fragment and feldspars alteration.
development of the fields (Ali et al., 2010). Microscopic and Scanner Electronic Microscopic (SEM)
Detailed petrographic analysis together with burial analysis showed natural compaction, quartz overgrowths
history helps to define the most important diagenetic and clay coating. This evidence suggests that diagenesis
processes that might have affected the reservoir qual- reduced the pore space. Based on core samples, it is assumed
ity. Core analysis, outcrops and offset analogues can also that the diagenetic processes have affected the reservoir
provide useful trends for diagenetic processes. Volumetric homogeneously. Therefore, only one scenario for this com-
significance and relative timing of occurrence are important ponent can be considered.
constraints to building several possible diagenetic models. Based on the above analysis, one scenario each from
However, unless statistically meaningful data (e.g. high structural, stratigraphic and diagenetic components and
quality petrographic point counts/grain size distribution/ four discrete scenarios from sedimentary components (as
grain coating per sample), burial history with time (burial illustrated in Figure 7) were considered for defining the four
depth, temperature, overpressure and effective stress) is CGMs which were used as inputs for building the four 3D
available, it is extremely difficult to build multiple scenarios geological models.
of this component.
The formation is composed of sandstones (75%), silt- 3D Geomodels building using four CGMs as input
stone (10%) and shale (15%). Sandstones are composed To build a 3D reservoir model, a 3D structural framework
by coarse grain (25%), medium grain (35%) and fine grain is built first. After 3D structural/stratigraphic modelling,
(40%) sands which are slightly argillaceous with good the four CGMs discussed above were used as the input
roundness. Abundant cross bedding and base of channel within the defined stratigraphic framework. Each CGM
surfaces are present in these sandstones. Bioturbation is rare contains a mixture of sedimentary facies. Porosity and

© 2016 EAGE www.firstbreak.org 45


technical article first break volume 34, July 2016

Figure 10 Vertical sections and 3D views of internal


configuration of facies of four 3D static models built
using four CGMs in Petrel Software. These models
show: (a) Internal correlation following top configur-
ing a braided system, (b) Internal correlation follows
base configuring anastomosing channels system, (c)
Internal correlation following the internal slope con-
figuring stacking migrant meandering systems (d)
Internal correlation of channels proportional to the
top and base reservoir target configuring non-sequen-
tial meandering channelization. Those scenarios show
the wide range of variability of the bodies deposited
in fluvial environment where the fluvial sandstone
reservoir can be located. Difference of sand and shale
distribution and connectivity can be observed in each
scenario, this internal configuration will affect the
static and dynamic models in term of volumetric and
final recovery.

permeability within each facies are typically well cor- Figure  9), and (C) variogram modelling (Orellana et al.,
related in all static models. These four 3D static models 2014) honouring well data and seismic stratigraphic configu-
would be capturing uncertainties of the distal fluvial sys- ration in order to obtain realistic references of geometry and
tem and also different conceptual sedimentary models, size of the fluvial bodies.
structural, stratigraphic and diagenetic components. Object Based Simulation ‘OBS’ (in ‘A’ and ‘B’ cases) and
These considerations in the 3D geological models provide Sequential Indicator Simulation ‘SIS’ (in C case) techniques
criteria for rock type identification. A semi-log permeability- for facies and Sequential Gaussian Simulation ‘SGS’ (in
porosity cross-plot was used to define different rock types all cases) for rock properties distribution with porosity-
incorporating pore throat variations (Winland R35 function permeability functions were used. The vertical sections and
relationship, Amaefule et al., 1993). 3D views extracted from the four 3D static models (Figure
Five rock types (R1 to R5, R1 best reservoir and R5 10) clearly show vertical and lateral variability of the facies
non-reservoir) were identified. This was done in two steps: distribution. Therefore, a total of 12 independent 3D static
(1) Analysis of well log responses (GR, RHOB, NPHI, DT) models (four CGMs and three cases of facies and property
in association with core description to define electro-facies. distribution in each) were generated.
(2) Integration of petrophysical information from conven- Using these 3D static models as input, around 30
tional and special core analysis to establish the link between realizations were made for each model to capture the
electro-facies and rock types. Average capillary functions lateral and vertical variability of facies and property distri-
for each rock type was used to build the saturation profile bution. The number of iterations was defined based on the
and for defining irreducible water saturation. To validate Central Limit Theorem (Johnson, 2004) in order to have
the rock-type reservoir characteristics, the calculated satura- a reasonable precision of probabilistic results. According
tion was compared with those derived from petrophysical to this theorem, the sampling distribution of the sampling
interpretation of logs (e.g. effective porosity, Vshale and water means approaches a normal distribution as the sample size
saturation). gets larger – no matter what the shape of the population
Stochastic realizations based on four deterministic distribution.
conceptual geological models are created to cover the
wide range of possibilities in facies distribution and their Results and discussion
association, considering the parameterizations for the three Several criteria of well logs correlation and type of channel
input cases: (A) Analogues (geometries from a modern systems were conceptualized in the distal fluvial environ-
distal fluvial depositional system), (B) Public Domain Fluvial ment. Reservoir continuity of the channelized reservoir and
Sedimentary DataBase (Geometries from ERGO CGG, 2015, shale relocation owing to varying grid layering and channel

46 www.firstbreak.org © 2016 EAGE


first break volume 34, July 2016 technical article

Figure 11 Diagram showing hydrocarbon saturation, OOIP probabilistic distribution, recoverable resources and production forecasts ranges from different 3D
reservoir models built using different CGMs. (Abbreviations: PLLT – Parallel to reservoir top, PLLB-Parallel to reservoir base, SIGM – Sigmoidal following internal
slope, PRP – Proportional layering to the top and base).

geometry allows for the capturing of a reasonable range of property distribution) provide the most conservative OOIP
Original Oil In-Place (OOIP) uncertainties. value of 37 Million BO and the realizations based on the
360 stochastic realizations created using 12 3D static scenario 4 (using analogue as input for facies and property
models provide a full range of OOIP distribution using distribution) with the most optimistic OOIP value of 46 mil-
single oil water contact. These volumes plotted as a lion BO.
Probability Distribution Function (PDF) show the narrow Different 3D static models were used for generating the
range of variation (Figure 11). This is mainly due to good recoverable resources and production forecasts which led to
lateral continuity of the channelized reservoir with shale variable plateau (most pessimistic forecast of around three
layers close to top and base (see Figures 7 and 10). The P50 years to scenario 3 and most optimistic forecast of around five
OOIP value (see Table 2) from these 3D static models (12 years to scenario 4). Around 50% variance in the cumulative
PDFs) was used for estimating the recoverable resources production between extreme cases was observed indicating the
and production forecasts. importance of incorporating different CGMs.
In terms of conceptual geological models, the realizations The recovery factor obtained from different 3D reservoir
based on Scenario 3 (using variograms as input for facies and models ranges from 30% to 43%. The difference between

© 2016 EAGE www.firstbreak.org 47


technical article first break volume 34, July 2016

Summary of production forecasts from dynamic simulation of different 3D static models

CGM Scenarios input data used for facies OOIP range P50 OOIP production recovery
and properties distribution (million BO) (million BO) (million BO) factor (%)
P99-P1

Scenario 1 analogue 39-48 44 18.3 42%


(parallel to reservoir top
ergo database 36-44 41 15.9 39%
(PLLT))
variogram 38-42 40 16.3 41%

Scenario 2 analogue 35-47 42 15.9 38%


(parallel to reservoir base
ergo database 35-42 38 14.6 38%
(PLLB))
variogram 37-42 40 15.1 38%

Scenario 3 analogue 38-44 42 12.5 30%


(sigmoidal configuration
ergo database 33-45 40 13.3 33%
(SIGM))
variogram 33-39 37 15.2 41%

Scenario 4 analogue 42-48 46 18.7 41%


(proportional layering
ergo database 34-43 39 16.7 43%
from top and base (PRP))
variogram 34-43 39 16,0 41%
Table 2 Output summary of P50 OOIP, production forecasts and recovery factor generated from different 3D reservoir models and dynamic simulation.

the lowest and highest recovery factor is around 13% due The proposed workflow for the creation of adequate
to different internal architecture in different models achieved CGM integrates qualitatively the boundary conditions from
from using CGMs. The difference in the cumulative produc- basin to the reservoir scale facies modelling and provides an
tion and recovery factor of the Scenarios 3 and 4 is mainly improved understanding of reservoir depositional environ-
due to the internal layering distribution of sandstone and ments, geometry and connectivity. Therefore, CGM is a
shales in those scenarios which affect the oil flow and exhibit critical input for the facies and property distribution in 3D
different flow behaviour in the reservoir. static reservoir model building and allows for bridging the
This exercise provides higher confidence on possible gap between reservoir geology and 3D numerical reservoir
scenarios and ranges of volumetric estimation and produc- modelling activity. Visualization of simple 2D vertical cross
tion forecasts. Probability distribution functions obtained sections and maps showing structural/stratigraphic configu-
from these 12 3D geological models provide a much more ration, lithology/facies and property distributions has proven
reasonable range of OOIP, recoverable resources and to be very effective tool to analyse and quality control the
production forecasts. This study reinforces the point that quality of 3D reservoir models.
if the appropriate data quality control, mapping principles The example shows that the integration of multiple
and geostatistics are not applied in the light of the rules CGMs with 3D reservoir model-building processes has
implied by fundamental geological processes during the allowed for the capturing of the most likely development
conceptual geological models and 3D static reservoir model scenario with a full range of geological possibilities and
building process, the 3D numerical and automatic models improved confidence in the estimated oil-in-place, recover-
will not reasonably represent the geology regardless of the able resources and production forecasts. These multiple
sophistication of the software and algorithms deployed. 3D static models were used to history match the existing
production and DST data. The most likely CGM-guided
Conclusions static model required very minor or negligible modifications
The CGM building workflow is a nonlinear process which to achieve a reasonable history match against previous
involves integration of geologic and engineering data/ models built without the CGM workflow. The history
interpretation studies of geological aspects from the basin matching process was used to rank models and ascertain the
scale to reservoir or field scale, pore space characteristics at most likely scenario. Such models, without ad-hoc changes
reservoir scale using field and offset analogues, application and with better history match, have improved the confidence
of geologic rules and geologists’ experience. level of the most likely development scenario.

48 www.firstbreak.org © 2016 EAGE


first break volume 34, July 2016 technical article

Even though, the CGM building process is time con- Quinto S., Weltje, G.J. and Salina, B.E. [2013]. How to integrate basin-scale
suming and labour intensive, in-depth understanding and information into reservoir models. 75th Annual EAGE Conference &
incorporation of various geologic aspects and use of experts’ Exhibition, Extended Abstracts.
knowledge from different disciplines in the 3D integrated 3D Seo-Websitedesigner [2012]. Top pictures of Grand Canyon. http://www.
reservoir interpretation and modelling process is essential traveltourismblog.com/grand-canyon.php.
to improve the reliability of the 3D numerical modelling Singh, V., Hegazy, M. and Fontanelli, L. [2009]. Assessment of reservoir
outcomes and assessment of the risks. uncertainties for development evaluation and risk analysis. The
Leading Edge, 28 (3), 260-268.
Acknowledgements Singh, V., Yemez, I. and Sotomayor, J. [2013a]. Key factors affecting 3D
The authors would like to thank Repsol management for reservoir interpretation and modelling outcomes: Industry perspec-
providing the facilities, guidance and encouragement during tives. British Journal of Applied Science & Technology, 3 (3), 376-405.
the execution of this work and permission to publish this Singh, V., Yemez, I. and Sotomayor, J. [2013b]. Integrated 3D reservoir
material. The suggestions from the colleagues from Dirección interpretation and modelling: Lessons learned and proposed solutions.
Ejecutiva de Desarrollo Técnico (DEDT), DG E&P are also The Leading Edge, 32 (11), 1340-1353.
acknowledged for their valuable discussion and interaction. Tamhane, D., Wang, L. and Wong, P.M. [1999]. The role of geology in
Sincere thanks goes to Tapan Mukerji (Stanford University, stochastic reservoir modelling: The future Trends. SPE Asia Pacific Oil
USA) and Malleshwar Yenugu (Icon Science Americas, and Gas Conference and Exhibition, 54307.
Houston) for their valuable comments which were very use- Wilson, M. [2010]. Anticline in the Barstow Formation (Miocene) at
ful in improving the quality of the paper. Calico Ghost Town near Barstow, California. Wikimedia Commons.
USA, www.wikipedia.org/wiki/Anticlinal#/media/File:BarstowFormati
References onAnticlineMarch2010.jpg.
Ali, S.A., Clark, W.J., Moore, W.R. and Dribus, J.R. [2010]. Diagenesis Yu, X., Li, S., and Chen, B. [2013]. Depositional Architecture-based
and Reservoir Quality. Oilfield Review, 22 (2), 87-93. Correlation Techniques for Deltaic and Fluvial Reservoirs in Lacustrine
Al-Khalifa, T., Tobias, M.A., Payenberg, H.D. and Lang, S.C. [2007]. Basins. AAPG 2013 Annual Convention and Exhibition, #10513.
Overcoming the Challenges of Building 3D Stochastic Reservoir
Models Using Conceptual Geological Models – A case study. Received: 27 July 2015; Accepted 11 March 2016.
SPE Middle East Oil and Gas Show and Conference, SPE- Doi: 10.3997/1365-2397.2016010
104496.
Amaefule, J.O., Altunbay, M., Tiab, D., Kersey, D.G. and Keelan, D.
[1993]. Enhanced reservoir description: Using core and Log data to
identify hydraulic (flow) units and predict Permeability in uncored Colorado School of Mines
intervals/wells. SPE Annual Technical Conference and Exhibition, Department of Geophysics
SPE-26436. Department Head
Catuneanu, O., Galloway, W., Kendall, C., Miall, A., Posamentier,
The Department of Geophysics at Mines seeks a dynamic,
H., Strasser, A. and Tucker, M. [2011]. Sequence Stratigraphy: enthusiastic leader to head the Department. We seek
Methodology & Nomenclature. Newsletters on Stratigraphy, 44 (3), candidates excited to share in our mission to address the
challenges of a sustainable global society by educating the
173–245.
next generation of leading engineers and scientists, and by
ERGO Robertson CGG [2015]. Database of reservoir analogues and expanding the frontiers of knowledge through research.
knowledgebase of geologic information for E&P projects. http:// Department heads at Mines generally continue to teach
robertson.cgg.com/products/ergo. undergraduate and graduate courses, mentor graduate
Galloway, W.E. [1981]. Depositional Architecture of Cenozoic Gulf students, and carry out research. Further, the successful
candidate will be expected to work effectively in a
Coastal Plain Fluvial systems. SEPM Special publication, 31 (4),
collaborative, interdisciplinary environment within the
127-155. Department and across Mines.
Johnson, O.T. [2004]. Information theory & the central limit theorem. Requirements include a Ph.D. in Geophysics or related
Imperial College Press, London. field, and a proven track record in research and service.
Miall, A.D. [2006]. Reconstructing the architecture and sequence Effective interpersonal, communication and organizational
skills are a must. Must have demonstrated high ethical
stratigraphy of the preserved fluvial record as a toll for reservoir
standards, operating in a transparent, collegial way. The
development: A reality check. AAPG Bulletin, 90 (7), 989-1002. successful candidate should be responsive to the needs of
Orellana, N., Cavero, J., Yemez, I., Singh, V. and Sotomayor, J. [2014]. the Department faculty, staff, and students and have a
strong commitment to enhance diversity.
Influence of Variograms in 3D reservoir Modelling Outcomes: An
Example. The Leading Edge, 33 (8), 890-902. For the complete job announcement, full statement of
qualifications and directions on how to apply, visit:
Orellana, N., Yemez, I., Cavero, J., Singh, V. and Izaguirre, E. [2015].
http://inside.mines.edu/HR-Academic-Faculty
Role of geostatistical techniques on the evolution of 3D interpreta-
tion and modelling outcomes: An Example. Petroleum Geostatistics Mines is an EEO/AA employer.
Conference, Extended Abstracts.

© 2016 EAGE www.firstbreak.org 49


CC01317-MA020.indd 1 10/05/16 10:18
View publication stats

You might also like