SPE 95612

3D Geological Modeling in a Turbidite System With Complex Stratigraphic-Structural

FrameworkAn Example From Campos Basin, Brazil
G.A. Alberto and A.P. Grell, Petrobras; D. Badolato, Beicip-Franlab; and L.R. dos Santos, Schlumberger

Copyright 2005, Society of Petroleum Engineers

This paper was prepared for presentation at the 2005 SPE Annual Technical Conference and
Exhibition held in Dallas, Texas, U.S.A., 9 12 October 2005.
This paper was selected for presentation by an SPE Program Committee following review of
information contained in a proposal submitted by the author(s). Contents of the paper, as
presented, have not been reviewed by the Society of Petroleum Engineers and are subject to
correction by the author(s). The material, as presented, does not necessarily reflect any
position of the Society of Petroleum Engineers, its officers, or members. Papers presented at
SPE meetings are subject to publication review by Editorial Committees of the Society of
Petroleum Engineers. Electronic reproduction, distribution, or storage of any part of this paper
for commercial purposes without the written consent of the Society of Petroleum Engineers is
prohibited. Permission to reproduce in print is restricted to a proposal of not more than 300
words; illustrations may not be copied. The proposal must contain conspicuous
acknowledgment of where and by whom the paper was presented. Write Librarian, SPE, P.O.
Box 833836, Richardson, TX 75083-3836, U.S.A., fax 01-972-952-9435.

Abstract
After 3 years of production, results indicated necessity to
refine the geological model of Cretaceous turbidite reservoirs
in an off-shore oil field of Campos Basin, Brazil.
A stratigraphic analysis was developed to build a new
stratigraphic-structural framework. Seismic interpretation was
used to incorporate structural data by mapping the system of
normal faults. Using concepts of sequence stratigraphy,
seismic, biostratigraphic and lithologic data were used to
define 12 major depositional sequences, spanning from
Cenomanian to Maastrichtian. In 8 of these sequences, 12
turbidite systems were recognized and 9 of them focused for
the present study. These turbidite systems compose the
operational reservoir zones. Six of them were deposited during
the Turonian and three during the Santonian. Erosive surfaces
and regional unconformities affect the distribution of the
turbidite systems. Action of erosion and faulting results in
complex framework and lateral communications among
reservoirs of different ages. Several oil-water contacts are
present and controlled by some of the faults and the lateral
communications.
Modeling this complex system included many different
techniques and softwares in steps of the work. The first step
involved a topologic three-dimensional construction, including
all of the major geologic information; relationships among the
main turbidite systems were validated by production data.
Next was creating a refined grid to populate the model with
rock-properties; corner point geometry grid was built to honor
the direction of major faults. Third was facies characterization;
vertical proportion matrix was used to represent the horizontal
non-stationary of the data and seismic amplitude was used as a
constraint. Plurigaussian simulations were used to populate the
model; this type of simulation is suitable to represent the
multiple and non-sequential contact relations among facies.

The fifth step was using Monte Carlo simulation to fulfill the
model with porosity and permeability, based on petrophysic
histograms from well data, classified by facies. Upscaling the
model was the last step in order to finally transfer the
geological data to the flow simulator.
Preliminary results from the flow simulator reveal better
fitting between this new geological model and the present
production data.
Introduction
Along the years from 2000 to 2002, a series of perforations
and workovers in this studied field indicated a partial
inadequacy of the model then existent to explain the
deviations found in relation to the expected results. A
multidisciplinary
team
(petrophysicist,
geophysicist,
sedimentologist, reservoir geologist and engineer) was
composed to perform an integrated study to update the
geological model.
An extensive revision work of all the available data and
their internal organization into a three-dimensional (3D)
model, to respect the main complexities and heterogeneities
observed in the field reservoirs, was then put in practice. Part
of the methods and results is next exposed.
Preliminary Information
The studied field is composed of 6 different blocks distributed
in an area of 728 km2 with water depths varying between 800
and 1500m. The present work deals with its main producing
block.
The main reservoirs are Upper Cretaceous (Turonian and
Santonian) turbidite sandstones, with good quality oil (28
API). A large number of stacking and partially connected
reservoirs is recognized in different turbidite systems. These
turbidites are characterized as amalgamated channel
complexes. An increasing erosive character upwards causes,
especially in Santonian reservoirs, erosive superimposed
surfaces, including some unconformities. These complex
stratigraphic sequences were later disturbed by normal faults,
during episodic salt movements.
Main flow restrictions should be related (i) to structural
subjects (complex fault system) (ii) external geometry and
internal heterogeneity of the reservoirs and (iii) the structural
relationship among the different stratigraphic units.

Stratigraphic Analysis
General Information. The stratigraphic interval approached
in the present work is represented by sedimentary sequences
that
were
deposited
from
the
Upper
Albian
Superior/Cenomanian to the Lower Tertiary. That section
represents the Marine Transgressive Megasequence (Bruhn,
1998). During this period Campos Basin was submitted to a
combination of thermal subsidence, amplified by sedimentary
load, with general trend of eustatic elevation of the sea level
(Chang et al., 1988). Except for the systems deposited during
the Cenomanian, the preponderant factor in the deposition of
those sequences would be the tectonic activity (Bruhn, 1988).
General conditions of starving basin controlled a
deposition of coarse turbidites in thick amalgamated
successions, but confined in narrow troughs formed by
halokinetic faults.
Several works approach the study of sequences (sensu
Mitchum, 1977) between the Cenomanian and the
Maastrichtian periods in Campos Basin. General concepts and
data are developed by Martins and Moreira (1994), and, more
specifically for the studied area, Becker et al. (1999).
The present study suggests the subdivision of the
Cenomanian to the Maastrichtian periods into 12 possibly 3rd
order depositional sequences (table 1), since each sequence
spans a time interval of about 1 to 4 (predominantly 2 and 3)
millions of years. These sequences were defined by their fossil
content (especially the nannofossil biostratigraphy determined
by Martins et al., 1994-2003; see table 1); in practice some
stratigraphic markers, using log-correlation, were also used to
define the sequences.
Table 1 Relation of 12 depositional sequences, determined with
the support of biostratigraphy for the studied area. Oil-bearing
reservoir zones are identified in 6 sequences.
Petrobras
Reservoir
DEPOSITIONAL SEQUENCES
nannofossil
zones
Biozones
UPPER MAASTRICHTIAN
290
--LOWER MAASTRICHTIAN
280
--CAMPANIAN-MAASTRICHTIAN
270
--CAMPANIAN
265
--LOWER CAMPANIAN
260.9
--SANTONIAN-CAMPANIANO
260.9
CRP-100
SANTONIAN
260.7
CRP-200
CONIACIAN-SANTONIAN
260.5
CRP-300
CONIACIAN
260.4
ESP-90
TURONIAN
260.3
ESP-100
ESP-200 to
TURONIANO
260.2
ESP-500
CENOMANIAN
260.1
---

In the studied area, 12 turbidite systems are identified in 8

of the 3rd order depositional sequences. Nevertheless only 9 of
these turbidite systems are oil-bearing reservoirs and represent
the operational reservoir zones, mapped in the present study.
These 9 zones occur in 6 depositional sequences and were
deposited during the Turonian and the Santonian (Tables 1 and
2). Cenomanian and Maastrichtian depositional sequences,
since they do not present hydrocarbon accumulations, were
not detailed in the present work.
Turonian (Turonian to Lower Coniacian indeed) turbidites
are locally named Espadarte Sandstones and Santonian
(Coniacian to Lower Campanian indeed) are locally named

SPE 95612

Carapeba Sandstones (table 2). Espadarte (ESP) and Carapeba

(CRP) Sandstones are subdivided into 9 reservoir zones (table
1). Espadarte Sandstones are composed by 6 reservoir zones
while Carapeba Sandstones are composed by 3 reservoir zones
(table 2).
Table 2 Classification of reservoir zones of the studied field.
These code names will be used in the present work.
MAJOR SYSTEMS

CARAPEBA SANDSTONES
(mainly Santonian in age)

ESPADARTE SANDSTONES
(mainly Turonian in age)

RESERVOIR ZONES
CRP-100
CRP-200
CRP-300
ESP-90
ESP-100
ESP-200
ESP-300
ESP-400
ESP-500

Sedimentologic Data. Producing zones were described by

Voelcker et al. (2002) in terms of lithofacies, mineralogical
composition and petrophysic properties.
Espadarte Sandstones. Two lithofacies with reservoir
characteristics were identified: fine massive sandstone and
very fine to medium laminated sandstone. They are interpreted
as transition channel-lobe (massive), channel fill (massive)
and spill (laminated) deposits. Nevertheless, Mutti (2002,
verbal information) suggested the possibility of the action of
bottom currents to originate the laminated sandstones. Main
accessory facies is interlaminated and bioturbated
sandstone/siltstone/shale.
The Espadarte Sandstones are arkosian in composition;
micas, organic matter, bioclasts and glauconite are found in
laminated and bioturbated portions. They present excellent
reservoir characteristics with the following averages,
according facies: (i) massive sandstones with 1200 mD of
permeability and 30% of porosity; (ii) laminated sandstones
with 600-1200 mD and 26-28% for the same petrophysic
properties. Primary porosity is predominant.
These sandstones bear most of the volume of oil-in-place
in the studied field and for these reason Espadarte Sandstones
are economically more important.
Carapeba Sandstones. Three main reservoir lithofacies
were identified: (i) medium to fine massive sandstone, (ii)
conglomeratic medium sandstone and (iii) intraclastic shaly
sandstone. Amalgamation is observed, forming 0.8 to 1.7m
thick layers, with shale intraclasts at the top of cycles. They
are interpreted as low to high density turbidites, reworked by
bottom currents. Accessory facies are interlaminated
sandstone/siltstone/shale and slump deposits.
Their mineralogical composition is very similar to the one
of the Espadarte Sandstones. They also present excellent
petrophysic properties, with porosities varying from 21 to 33%
and permeabilities around 1000 mD. Some secondary porosity
is also observed (dissolution and fracturing of feldspars and
cement dissolution).
Massive sandstones present greater homogeneity in terms
of petrophysic properties distribution, with no marked
anisotropy between vertical and horizontal permeabilities.

SPE 95612

Stratigraphy, tectonics and sedimentation. Two major

turbidite systems, Carapeba and Espadarte Sandstones (tables
1 and 2) are associated to tectonic systems: the first is the
Turonian-Coniacian and the younger is the SantonianMaastrichtian. The limit between both systems is marked by
conglomerates and sandstones deformed by gravitational
action as well as by intensive igneous activity, indicating the
maximum tectono and halokinetic actions (Aguiar and Destro,
2000).
The base of the sequence deposited during 260.5 biozone
is a regional erosive surface and marks an important
unconformity, which separates CRP from ESP systems. Other
important erosive surfaces are associated to the base of
sequences deposited during 260.7 and 260.9 biozones (see
table 2).
The Espadarte Sandstones are, in general, amalgamated,
with maximum thickness of 60m and were deposited along the
direction NW-SE. According to the facies association
previously described and to the external geometry of
sandstone bodies, mapped in ESP-90 to ESP-500 reservoir
zones, we may interpret these sedimentary deposits as
confined channel complexes. Some transition to lobes is
observed, since the troughs become filled and the confinement
is reduced; the effective spreading of lobes seems not to
happen indeed, because migration of the channel axis, with
associated sandstone deposition, is observed. Some bottom
currents reworking may have acted, and it was responsible for
the formation of heterolitic facies in some reservoirs.
An intensive tectonic activity occurred during the time
period between the Albo-Cenomanian and the unconformity at
the base of sequence 260.5. This activity is probably
associated to salt movement and was responsible for the space
creation. The final of the tectonic activity tilted the underlying
units (Espadarte Sandstones) and permitted the entrance of
coarse sediments, generating high density turbidites, which
characterize the following sequence (Carapeba Sandstones).
Carapeba Sandstones are characterized for thick sandstone
cycles mainly related to confined channel complexes. These
complexes are very confined (more than the previous system)
in northeastern, and more spread in northern and western areas
of the studied field.

Fig. 1 Schematic workflow used in modeling the reservoirs of

the studied field.

Stratigraphic-Structural Framework
Preliminary actions. The initial step for the construction
of the framework was the stratigraphic interpretation starting
from the well log analysis, through identification and
correlation of the main stratigraphic markers. The support for
those interpretations came from data of previous studies
concerning the interpretation of sequences, biostratigraphic
zoning, reservoir zoning, structural mapping and seismic
interpretation (Martins and Moreira, 1994, Martins et al.,
1994-2003, Becker et al, 1999 and Aguiar and Destro, 2000).
The main result of this correlation was the establishment of the
depositional sequences and reservoir zones described in the
item Stratigraphic Analysis.
General illustration for the structures is present in figures 2
and 3: stacking of the sedimentary sequences is observed, as
well as the superimposed fault system. In the strike section
(Fig. 3) is possible to observe the relationships of lateral
contacts among different turbidite systems (reservoir zones of
Table 1). The main observed lateral communications are
among CRP-300 and ESP-100 zones to south of the field (left
side of Fig. 3) and CRP-100 and ESP-200 to northeast of the
field (right side of Fig. 3).

Three-Dimensional (3D) Geological Modeling

Workflow. The workflow used in the present work (Fig. 1)
was a strategy we have chosen to reflect the concern in trying
a more effective representativeness of the structural and
stratigraphic complexities of the studied field. For that, a long
list of different softwares was used in different work stages.

1 Km

Fig. 2 Seismic dip (NW-SE) section of impedance; ESP-100 and

ESP-200 zones appear amalgamated in green/red color. Total
horizontal distance is about 5 km and total vertical depth is about
400m.

SPE 95612

Fig. 3 Schematic geologic strike (SW-NE) section; there are

relationships of lateral contact among different zones.

A total of 8 dip sections and 6 strike sections was

interpreted and permitted the definition of the major turbidite
systems (ESP and CRP) and their internal subdivisions (ESP90 to ESP500 and CRP-100 to CRP-300).
Building a topologic, stratigraphic-structural framework.
2D interpretation and data, including the mentioned
correlation, were used to build a 3D model. Interpreted
sections, with their respective "well picks" (tops and bases),
support seismic sections and the fault system, already
converted to depth, were the data basis to be imported in a 3D
platform.
The construction of the model based on first place in the
establishment of the fault hierarchy. The faulting system was
divided into main and secondary faults.
The sequences of the studied area were built according the
order of geological events, from the base to the top, and over
the established structural framework. Each control surface was
built based on the seismic surface reference, with control of
well picks (tops and bases) and additional picks from the 14
interpreted geological sections. After building top and base
surfaces of each sequence, pinch-outs were represented from
interpreted seismic and geological sections, and additional
well control. In practice, control points were used to represent
the pinch-outs modeled in 2D sections.
The 9 reservoir zones were mapped and then modeled. In
terms of the technique to build the model, the zones were
defined as depositional, except the bases of CRP-100, CRP200 and CRP-300 zones, more strongly erosive.
The complexity of the stacked reservoir zones and the
superimposed faulting may be observed at the block-diagram
(Fig. 4). In spite of small throws (about 20m) of most of the
faults inside the model, there are evidences they are important
to control the fluid flow. We can also observe relationships of
lateral contact among CRP and ESP zones, in northeastern
portion of the field (right side of Fig. 4).

Fig. 4 - Three-dimensional model view with two orthogonal

sections. Y indicates north direction.

Preliminary analysis of the topologic framework. The

macro-geometry of ESP-90 to ESP-500 zones indicates the
predominant occurrence of channel complexes and of channellobe transition. The back-stepping pattern is dominant,
especially in ESP-100 and ESP-200 zones. Next discussion on
reservoir zones may be better realized by observing Fig. 5.
ESP-200 zone, besides presenting the largest distribution
in area among the Espadarte sandstones, has also the thickest
sedimentary section, better reservoir conditions as well as the
largest hydrocarbon volumes; lithologically it prevails fine to
medium, massive sandstones. ESP-100 is the second in terms
of economical importance, among the Espadarte Sandstones.
ESP-300 presents a more restricted distribution and it seems to
represent channel spills probably genetically related to the
channel complexes of ESP-100 and ESP-200 zones. In spite of
its large distribution in area, ESP-400 was not studied in detail
since it occurs, in most parts, below the oil-water contacts of
the field; seemingly it is formed by the same genetic elements
of ESP-100 and 200 zones. ESP-500 zone always occurs
below the oil-water contacts and in an area of very restricted
distribution. ESP-90 zone occurs with more strongly erosive
character over ESP-100 sandstones, maybe already
anticipating the character of the events that were responsible
for the deposition of the Carapeba Sandstones.
The macro-geometry of CRP-100, CRP-200 and CRP-300
zones indicate the predominance of channel complexes,
particularly in the eastern side of the field. The preferential
channeling in this area apparently took place with some lateral
or frontal migration of the channels. To west, the lobes prevail
in the deposition of the sands. We consider that to east the
paleotopography seems to have conditioned the confined
character of the sand distribution, while to west there is one
point which might have acted as by-pass and the sands have
been deposited in lobes, more distally to south (southwestern
of the field).

SPE 95612

Fig. 5 presents the geometry discussed in the previous

paragraphs; each zone is individually represented.

(a) CRP-100

(b) CRP-200

(c) CRP-300

(d) ESP-90

(e) ESP-100

(f) ESP-200

The description of building this framework in previous

item corresponds to a topologic one. After that it is necessary
to transform it into an explicitly gridded structural model. This
type of operation was a problematic one due to the very nature
of softwares we have used. Although not a preliminary
intention, it was necessary to review many times the imported
and further gridded surfaces, since we observed differences on
the data recovery when transferring the topologic model to the
explicitly gridded model (Fig. 6). On the other hand, faults,
wells and well-logs were recovered without problems.

(a)

(b)

Fig. 6 Example of problems in transferring the 3D structuralstratigraphic framework (example in CRP-200 zone): (a) topologic
model, (b) gridded model with some missing blocks. See Fig. 5b
for figure orientation and scales.

(g) ESP-300

(h) ESP-400

(i) ESP-500

Fig. 5 Individual representation of each reservoir zone, for both

Espadarte (ESP) and Carapeba (CRP) systems. We may see the
horizontal distribution of the zones. North is upper part of the
maps; the distances in each map are 9 km in X (horizontal) and
about 10 km in Y (vertical) directions.

Locally it is possible to observe vertical communications

among the zones, principally between ESP-90 and ESP-100,
and between ESP-100 and ESP-200 zones, being this last an
important factor for the flow of fluids in the field.
Lateral communication among zones is also observed and
it occurs as result of a complex action of faults and
unconformities (Figs. 3 and 4). That happens eventually
among Espadarte zones, but it occurs more frequently among
some zones of Carapeba Sandstones with others of Espadarte
Sandstones (see item Stratigraphic-Structural Framework
Preliminary Actions). More specifically, in the northeastern
area, CRP-100 communicates with ESP-200, and still CRP200 with ESP-400; in the southern area of the field, there is
communication between ESP-100 and CRP-300 zones (Fig.
3).
Reservoir Quality Model
Structural modeling. The stratigraphic-structural model is
composed by the group of surfaces (tops and bases) of mapped
zones, faults and well picks (stratigraphic markers, tops and
bases) as well as of the basic electric logs (lithologic, porosity
and of corresponding electrofacies). The connections and
correspondences among all those objects should be coherent.

Coarse grid for dynamic flow simulation

Structural grid. A usual workflow would not go into this
step; nevertheless given the complexity of the created grid, we
opted to advance it to evaluate the grid in flow simulator.
After (re)building the stratigraphic-structural model in a
gridded model (previous item), defining the grid blocks
(structural grid) and building the coarse grid for dynamic
simulation was an automatic operation.
We used a Corner Point Geometry (CPG) grid in order to
respect the zone geometries and main depositional and faulting
orientations. This would optimize both geostatistical
simulations of facies and petrophysic properties as well the
dynamic flow simulation. In a CPG grid, cells may have any
geometry, but it is advisable to preserve at maximum their
orthogonal character. Therefore, the structural grid was
defined in a fine plenty way to respect the majority of faults
(directions preferentially northeast; see Fig. 7), without
deforming the final grid too much; nevertheless, some few
secondary faults with oblique direction were simplified (as
zigzag faults, according the grid).
A XY grid was built in each "block" of the structural grid;
88 cells were used in i (X) and 87 cells in j (Y) directions. The
size of each cell of the obtained final grid was variable, given
the nature of the CPG grid, but we can consider an average
cell dimension of the order of 100x100 m. The final structural
grid was used to define the grid mesh.
Model Sequences. This step considers the main vertical
subdivisions of the reservoir model. It is advisable that
stratigraphic sequences (e.g. reservoir zones) coincide with
model sequences.

For this reason we defined the reservoir zones as the units

of the model sequences. So the two main systems CRP and
ESP were divided into: CRP-100 to 300 (three units) and ESP
90 to 500 (six units). Each shale between every pair of
reservoir zones were also considered as one model sequence
as well as one of the main unconformities (base of CRP-300),
which represents the limit between CRP and ESP turbidite
systems. By this way, a total of 18 sequences were created: 9
of reservoirs, 8 of shales and 1 of unconformity.

SPE 95612

in the flow simulator, with the objective of verifying the

adaptation of that grid. Certain inadequacies were observed,
still relative to the geometry of the cells, as well as the
subjects of lateral communication among areas commented on
previously. Latter was solved in other step of the workflow
and former demanded an immediate return for cell geometry
adjustment.

(a)

(b)

Fig. 8 - Representation of the gridded model: (a) Stacked

reservoirs (model sequences); (b) Same as (a) including faults.
Wells are also shown. Colors represent different reservoir zones.

Fig. 7 - Perspective view of the structural grid (top and base of

the whole model); fault planes and some wells (in yellow) are
represented. Northeast direction is top-right corner of the figure.

Layering. After defining the sequences it is necessary to

subdivide them into layers. In order to better represent the
internal geometries of the reservoir zones and their
depositional differences we opted for different types of
subdivision in ESP and CRP reservoirs. This is described
ahead (see the item Fine Geological Grid) in more detail,
since this action in the present step of the workflow was just to
perform a previous analysis of the grid quality.
Adjustments. In order to correct some cell distortions,
manual adjustments had to be done, after verifying geometric
parameters (orthogonal character, angles among faces, etc.) of
the generated cells. This would avoid major problems
concerning flow in dynamic simulator. Figs. 8 and 9 show the
complete representation of the gridded sequences (reservoir
zones).
As mentioned before, in the eastern portion of the field, the
unconformity created an erosive depression (conduit), which
was fulfilled with CRP sandstones. Along the deep erosive
wall, a lateral communication is observed among ESP and
CRP reservoirs (see discussion and figures in item
Stratigraphic-Structural Framework). The representation of
this stratigraphic character was not well transferred from the
topologic to the gridded model. The necessary adjustment was
performed in pre-processor of the flow simulator.
Block and actnum properties were created to represent
limits of the blocks as well as active and inactive cells (values
of 0 or 1).
This previous coarse gridded model was provisorily and
experimentally populated with average petrophysic properties

(a)

(b)

Fig. 9 - (a) Oblique 3D view of the gridded model, with the

complete set of zones; (b) 2D section view of gridded zones:
sequences and their internal subdivision into layers.

Fine geological grid. Previous step, when we defined the

coarse geological grid, facilitated the generation of the fine
one because it assured the geometric coherence to do the
upscaling (next step).
A fine grid was built for each zone. In XY the density grid
was increased in about twice and the new average size was
about 50 x 50m. In Z, the individual layers had their thickness
reduced to detail the reservoir appropriately.
ESP zones were internally subdivided into proportional
layers, to respect the predominantly depositional character of
such reservoirs, except for ESP-300 and ESP-500 which were
not subdivided. In that way, the thickness was variable, but it
was determined such proportion to generate 5m of maximum
thickness, to preserve the detail of geological information.
On the other hand, CRP zones were internally subdivided
with parallel structure to the top of the unit, reinforcing the
confined geometry of these reservoirs and the erosive base
character. In that way the thickness was determined as 2m for
each layer, resulting in a great number of layers for each unit.
Data preparation for geological simulation. We used in
the model the results of a detailed previous petrophysic study,
including electrofacies, corrected curves for porosity, and
permeability estimation (Franco Filho, in press). Lithofacies
(see item Sedimentologic Data) were grouped into the
following electrofacies: 1 (massive sandstone), 2 (intercalated

SPE 95612

sandstone), 3 (cemented sandstone), 4 (shale). Therefore 1 and

2 represent reservoir facies, while 3 and 4 represent nonreservoir facies.
Electrofacies, porosity and permeability were discretized
into the grid, upscaling the log high resolution of about 20 cm
for about 2 m in the fine grid.
We searched for possible correlations of some seismic
attributes with net-to-gross (NTG), for each reservoir zone
with seismic resolution. NTG means the facies proportion:

represent non-sequential contacts among different facies. Fig.

12 schematically illustrates the difference between truncated
mono and plurigaussian facies simulations. Two parameters
must be defined: the contact-type among facies and variogram
models for the facies contacts.

NTG = (1+2) / (1+2+3+4) (Eq. 1)

Or, in other words:
NTG = (reservoir facies) / (reservoir + non-reservoir
facies) .. (Eq. 2)
It was observed the possibility to use seismic amplitude as
an external drift to map NTG in CRP-100, CRP-200, ESP-100
and ESP-200 zones; for similarity, amplitude of CRP-200 was
also used to map NTG in CRP-300 zone. These five NTG
maps were later used as constraints for facies simulation (Fig.
10).

Fig. 11 VPM for ESP-200 zone; this is a simplified matrix, and

each cell is grouping 5 cells of the original matrix.

Fig. 10 Example of NTG map, obtained by kriging well data and

using seismic amplitude as external drift (example of ESP-200
zone).

Facies and petrophysic geostatistical simulations

Vertical Proportion Curves (VPC) and Vertical Proportion
Matrix (VPM) for Facies. A fine analyses of facies (the log
facies determined before) was performed for each layer,
comparing and eventually adjusting the gridded data with the
raw log data. Vertical proportion curves were determined for
every well in each zone. Individual VPCs were then grouped
to determine regions; we also determined synthetic VPC to
represent some other geological knowledge which is not
explicit in the wells (e.g. trends of pinch-outs).
VPCs, as raw data, and the NTG, as conditional data, were
combined to map the distribution of facies proportion in area
(by external drift kriging and normalization). The result is a
Vertical Proportion Matrix (VPM) of facies (Fig. 11), which
indicates the horizontal variability of the VPCs. This process
was used for CRP-100, 200 and 300, and for ESP-100 and
200. For other zones, without NTG maps, the VPM was
conditioned only with VPCs.
The importance of building a matrix of different VPCs
(VPM) is exactly to represent the horizontal non-stationary of
the vertical proportions.
Simulations. The facies simulations were performed using
algorithm of Truncated Plurigaussian Simulation. The most
important feature in this type of algorithm is that we may

monogaussian

contacttype

sequential contacts

plurigaussian

contacttype

non-sequential contacts

Fig. 12 Theoretical example of the larger variability allowed by

plurigaussian simulation. In the case of monogaussian, transition
between green and red facies can only occur through the yellow
facies; such limitation does not appear in plurigaussian
simulation.

SPE 95612

The contact-type is function of the type of sedimentary

deposit and the well data. In Espadarte field we wanted to
represent two different facies relationships: first Gaussian
direction indicates the contacts between both reservoir facies
(facies 1 and 2); second Gaussian direction indicates contacts
among reservoir (facies 1 and 2) and non-reservoir facies
(facies 3 and 4). We adopted the contact-type of Fig. 13.

(a)

Fig. 13 Contact-type among facies adopted in the Truncated

Plurigaussian Simulation; it was used for every reservoir zones.

In each zone, VPM, variograms and the chosen contacttype were used to perform several simulations. For this
specific study just one realization was chosen upon subjective
criterion, in agreement with the intrinsic geological knowledge
of the conceptual depositional model of the area.
Simple Monte Carlo simulations were performed for
porosity and permeability in each zone and conditioned by
facies, using histograms of the well data parameters. The final
models for facies, porosity and permeability are illustrated in
Fig. 14).
Upscaling of the fine geological model to the coarse grid
and exporting it to the dynamic flow simulator. The fine
geological model properties were upscaled to be transferred to
the dynamic simulation model. Instead of the classic
geometric upscaling we used a topologic model, where each
layer of the coarse grid is manually associated to the
corresponding layers in the fine grid. This type of vertical
association improves the pertinence of calculated values after
the scale change.
Vertically, we grouped original units (layers of the fine grid)
in units to be recognized after the scale change (Table 3). This
was done after recognizing the main heterogeneities among
layers of the fine grid, which were observed in representative
VPCs of each area (and still according the conceptual
depositional model) and, secondarily, considering the different
oil-water contacts present in these reservoirs.

(b)

(c)
Fig. 14 Longitudinal dip section representing the simulation
results for (a) facies (Plurigaussian), (b) porosity and (c)
permeability in ESP-200 zone. Porosity (phi) and permeability
(Kx) are results of facies (Lithotype) conditioned Monte-Carlo
simulations.

SPE 95612

According to petrophysic information (Franco Filho, in

press), a porosity cutoff of 14% was considered to estimate
NTG of each coarse cell.
Final coarse grid representation exhibits of course a
decreased variability of parameters (see Fig. 16).

Table 3 Vertical grouping of fine geological grid into coarse

flow simulator grid.

ZONES
CRP-100

CRP-200
CRP-300
ESP-90
ESP-100

ESP-200
ESP-300
ESP-400
ESP-500

COARSE GRID
1
2
3
6
7
8
10
11
13
16
17
18
20
21
22
24
26
27
29

FINE GRID
1-5
6-15
16-48
1-8
8-15
16-56
1-9
9-48
1-10
1-12
13-18
19-20
1-20
21-40
41-50
1-25
1-13
13-25
1-20

The following criteria were used for properties:

- facies: the most frequent
- porosity: the average of the fine cells porosity
(weighted by the cell volumes)
- permeability: the classic method of Cardwell &
Parsons, that allows to respect eventual anisotropies
in the fine model distribution of values, according the
formula :

Kx = K min K max

(Eq. 3)

Kmin and Kmax are calculated for each direction (X and

Y), using a combination of harmonic and arithmetic averages
(see Fig. 15).

Fig. 15 Determination of Kmax and Kmin, to be used in Cardwell

& Parsons method of permeability upscaling.

(a)

(b)

(c)

Fig. 16 - Porosity in the fine grid (a), and after the upscaling (b); a
comparison between both grids is in (c).

After the upscaling, the resulting data were exported for

the dynamic flow simulator: coarse grid with all of the
properties (areas, NTG, porosity, permeability), discretized
wells (list of crossing well cells), discretized faults (main and
zigzag faults), and perforation cells.
Additional Remarks
It is important to relate some of the main difficulties and
problems found during the execution of the work:
-Complexity of the model
Still in way previous to the own geological model conception,
we already knew about some of the difficulties for 3D model
construction in the studied field: large number of faults,
presence of antithetic secondary faults, their small throws, and
the overlapping of at least three main erosive events. The large
number of reservoirs, related to different turbidite sequences,
is also a complex factor to be added to the others. Results of
these geological complexities have been observed in terms of
production: several different oil-water contacts, unexpected
water production in certain wells and possible interaction
between formation and injected waters.
-Resources for model representation and workflow choice
As a consequence of trying to build a more representative
model, to include most of structural and stratigraphic
complexities as well as reservoir heterogeneities, and still
according the available softwares and hardwares, we have
decided for the mentioned and previously presented workflow.
Best software resources were considered and an intricate net
was used, resulting in link additional problems. A final robust,
but complex and not-so-easy-to-update model was resulted.
-Problems in representing lateral and vertical reservoir
zones communications
Although the representation of the lateral and vertical
communications, among different stratigraphic zones, is
explicited in topologic structural model, this didnt happen on
gridded structural model. That was accentuated when
transposing that representation to the coarse flow simulator
grid. The final adjustments had to be done inside the own preprocessor of the dynamic flow simulator.

10

SPE 95612

Concluding Remarks
With this study we have searched the extensive incorporation
of available data and a very detailed representation of the
reservoir complexities and heterogeneities observed in the
field. For that, an extensive study of the stratigraphy and
structural geology of the area was performed, as well as an
analytical use of core, log, production and seismic data.
The final result was a robust three-dimensional model of
immediate application for the dynamic flow simulator and
subsequent reservoir studies. Detailed and final compatibility
analyses of all different oil-water contacts, and of volume
estimation still were not ended for final validation of the
model. Eventual adjustments should still be necessary.
Nevertheless, first results obtained from dynamic
simulation reveals a much better production history matching,
when comparing with the previous geological model. This is
valid not only for the general matching, when considering the
field as a whole, but also when considering individual
production zones and even individual producing wells.

Bibliographic References

Acknowledgements
We thank Marcelo Blauth (consultant geologist, Reservoir
Division of the Petrobras E&P Department), for important
technical advices, suggestions and discussions during the
studies; we also thank the technical team of the Reservoir Area
of the Central Production Asset (Petrobras Campos Basin
Business Unit), for valuable discussion and previous
interpretation during the field exploitation phase. Finally we
thank Petrobras, for permission to present this paper.

6.

1.

2.

3.

4.

5.

7.

8.

9.

Bruhn, C.H.L. 1998. Petroleum Geology of Rift and Passive

Margin Turbidite Systems: Brazilian and Worldwide Examples,
Part 2: Deep-Water Reservoirs from the Eastern Brazilian Rift
and Passive Margin Basins. AAPG International Conference &
Exhibition, Rio de Janeiro, Brasil. Short Course Notes, 187p.
Chang, H.K., Kowsmann, R.O., Figueiredo, A.M.F. 1988. New
Concepts on the Development of East Brazilian Marginal
Basins. Episodes 11, 194-202.
Mitchum, R.M. Jr. 1977. Seismic Stratigraphy and Global
Changes of Sea Level, Part II: Glossary of Terms Used in
Seismic Stratigraphy Applications to Hydrocarbon
Exploration. AAPG Memoir 26, 205-212.
Martins, F.A.L., Moreira, J.L.P. Estratigrafia Fsica do
Cretceo Superior com Base em Correlaes de Perfis Eltricos
de Poos, na Bacia de Campos. II Seminrio de Interpretao
Exploratria (SINTEX) da Petrobras. Rio de Janeiro, 1994.
Anais...45-50.
Becker, M.R., Carrasco, B.N., Antunes, R. L. 1999.
Sedimentologia e Estratigrafia do Cretceo Superior na rea
do Campo em Estudo. Relatrio Interno da Petrobras/Cenpes.
Servio Tcnico Especial 032611009. 19p.
Martins, G.A.R., Costa, S.O., Antunes, R.L., Maldonado, G.P.,
Shimabukuro, S., Cunha, A.A.S., Grassi, A.A., Strohschoen Jr.,
O., Moreira, C. 1994-2003. Anlises Bioestratigrficas dos
Poos do Campo em Estudo.
Voelcker, H.E., Franco, M.P., Magalhes, J.L. 2002. Resumo
das
Caractersticas
Sedimentolgicas,
Petrofsicas
e
Petrogrficas dos Reservatrios Santoniano e Turoniano do
Campo em Estudo. Relatrio Interno da Petrobras/UNBC/ATEX/LG. Com. Tecn. 328.022/2002, 4p.
Aguiar, L.A.M., Destro, N. 2000. Reconstruo Estrutural 2D
em Perfil Ssmico do Campo em Estudo. Relatrio Interno da
Petrobras/Cenpes. CT-DIVEX 148/2000, 3p.
Franco Filho, N.P. in press. Avaliao Quantitativa de Perfis
dos Reservatrios Carapeba e Espadarte do Campo em Estudo.
Relatrio Interno da Petrobras/ST/CER.