Skip to Main Content

Abstract

Jubarte is an oil field located offshore of Campos Basin, southeastern Brazil. Discovered in 2001, it is currently the most important oil accumulation of Espirito Santo State. Today, Phase 1 of the development plan has already been implemented and is producing through four one-kilometer-reservoir-exposure horizontal wells connected to FPSO P-34 platform. The field cumulative oil production already has reached 67,000,000 BO, representing around 3% of the total reserve.

Phase 2 of the development plan is ongoing and eighteen new wells are being drilled at the moment. To support well planning, a new version of the 3D geological model, incorporating recent acquired data, is underway. This task has being accomplished by integrating several sources of information. New data coming from the new wells is associated to information from analog outcrops, from interpretation of seismic horizons corresponding to erosive surfaces of turbiditic channels (corresponding to relative sea level falls correlative surfaces), and from inversion of seismic attributes (VP-VS ratio) from a high density 3D (HD3D) cube.

Analog outcrop information comes from the Annot sandstone (Annot Basin, Eocene sequence, France). The vertical shape of facies sequence is inferred based on 3D models from this outcrop as well as the geometry of the channel filling, the vertical proportion curves (VPC), and facies variograms, which work as a complement of well data.

Mapped seismic surfaces were used to separate different channel-filling populations that were modeled as individual regions in geological grid using different VPC, variograms, and global statistics.

A good correlation between the VPC of turbidite-channel-data and Vp-Vs ratio, from elastic inversion performed in the HD3D seismic cube, was observed, making it possible to generate a 3D facies proportion map based on VP-VS ratio, wells, and outcrop data.

The integration allowed a simulation of facies that represents better the existing reservoir heterogeneity.

Introduction

The importance of reservoir 3D geologic models that match geological and geophysical data, in order to forecast the production of oil field, is widely recognized in industry. Otherwise, to build geologic models, which have a good production history match, is a challenge to petroleum engineers and geologists. This task has been accomplished by integrating several sources of information: (1) wells logs and cores, (2) analog outcrop, (3) inversion of seismic attributes from a HD3D (high density seismic tridimensional data) cube, (4) interpretation of seismic horizons corresponding to erosive surfaces of turbidite channels (can be correlated surfaces of relative sea level falls), and (5) dynamic information of producing wells. Each source of information offers a unique perspective and resolution of observation. Outcrops provide a scale of observation that is not equaled by wells, seismic attributes, and interpretation, and they provide important contributions to our understanding of the inner organization of the reservoir. Despite this benefit, traditional outcrop studies are commonly limited to addressing the facies and architecture of reservoir using 2D exposures. This paper presents a methodology for managing different scales of geologic and geophysical information in order to build a high resolution 3D geologic model that honors the dynamic, geologic and geophysical data of an oil field. The data are from the Maastrichtian turbidity reservoir of Jubarte Field.

Jubarte Field General Data

The Jubarte Oil Field is located 80 km offshore from Espírito Santo State coast, under water depths around 1300 m in the Campos Basin (Fig. 1). The reservoirs are Upper Maastrichtian deep-water sandstones containing reserves of 600 million barrels of heavy oil (Bezzera et al., 2004). The 17.1° API, 3,000 cP oil (dead oil at 20°C), is the most viscous oil at surface conditions ever produced in Brazil through a wet completion system at that time. Such fluid characteristics and the large discovered volumes in this area were considered as a big challenge and an opportunity to develop knowledge to exploit heavy offshore oil reservoirs in deep waters under economical basis (Daher Jr et al., 2007).

Figure 1.

Jubarte Field Location (red dashed polygon).

Figure 1.

Jubarte Field Location (red dashed polygon).

Jubarte Field was discovered in 2001 by the wildcat 1-ESS-100, the location of which was based on 2D seismic. The reserves were estimated originally at around 150 million BOE. A drill-stem test was performed in cased hole obtained a daily rate at 580 BOPD. The acquisition of 3D seismic made possible a better understanding of depositional/trapping mechanisms and conclusion that oil reserves were higher than the original estimate. In order to define the commercial viability of the Field it was decided to drill a 1000m horizontal well and start an extended well test, in October 2002. The excellent well productivity results, 16,500 BOPD by natural flow, made PETROBRAS issue the declaration of commerciality of Jubarte Field in December, 2002. The phased-approach field development plan is as follows:

  • Production Pilot—DP FPSO Seillean started production in October 2002 and finished three years later, reaching an oil rate peak of 22,700 BOPD from one producing horizontal well. Important information was acquired from this phase about vertical permeability, presence of a single oil-water contact, gas-oil ratio, and aquifer influence.

  • Phase 1—Early Production System. FPSO P-34 started in December 2006 and finished in 2012, having a production of peak 51,000 BOPD in 2007 from four producing horizontal wells. The pressure measured before production start up was very important to match the aquifer volume modeled in 3D geological model and concluded that the aquifer strength was very similar to that predicted by simulation model.

  • Phase 2—Definitive Production System. FPSO P-57 started production in December, 2010. The development concept was based on a 180,000 BOPD FPSO, having a large liquid handling capacity (300,000 BOPD) and gas compression capacity up to 3 million std m3/d. It was connected to 15 horizontal production wells and 7 horizontal water injection wells through 6” flexible risers and flow lines. The average length of the horizontal section was 1,000 m for both production and injection wells, all of which were completed by open-hole gravel packing.

History of Jubarte Field 3D Geological Modeling

In 2002, after the wildcat reached the Maastrichtian reservoir of Jubarte Field and appraisal wells had been drilled, the first 3D geological model was built based on exploratory interpretation. Available data were top and base of reservoir mapped on 3D seismic cube and petrophysical data from logs and sidewall cores. This model was used to define the pilot project. The static properties were populated using constant mean values obtained from petrophysical analyses and drill-stem tests (DST) (Fig. 2).

Figure 2.

First 3D geological model built based on exploratory interpretation in 2002.

Figure 2.

First 3D geological model built based on exploratory interpretation in 2002.

A new model was constructed in 2003 trying to incorporate soft information associated to the inner heterogeneities of the reservoir. That information came from turbidite channel facies association observed in cores and stratigraphic amplitude slices from a seismic cube that suggest channelized shape. In order to reach this objective, a Boolean algorithm was utilized to simulate the discrete channels inside the 350 m thick trough that confined the turbidite deposits. Channels fill included porous sandstones and encompassing non-porous heterolithic rocks. An extended well test (EWT) was performed during this time in order to gather information on productivity, connected volumes, reservoir barriers, water cut behavior, flow assurance, and fluids processing, among others. As expected, the heterogeneities incorporated into the model proved useful to fit the dynamic data of the EWT. Based on the matched model, a large amount of information was obtained, helping to define Phase 1 (Fig. 3).

Figure 3.

Second 3D geological model (2003) using Boolean algorithm in order to incorporate the inner heterogeneities of the reservoir that suggest channelized shape.

Figure 3.

Second 3D geological model (2003) using Boolean algorithm in order to incorporate the inner heterogeneities of the reservoir that suggest channelized shape.

During 2004, an updated model was built using data from cores, new wells, reprocessed seismic data, and reinterpretation of the seismic cube acquired during phase 1. Core analyses were used to define a reservoir facies model conditioning by petrophysical data. Facies distribution was modeled using a pluri-Gaussian algorithm. This algorithm reproduces different body geometries for each facies and use non-stationary matrix of vertical proportion curves (VPC), in order to try to honor the vertical and horizontal variations of the depositional system. Well production history, EWT, and DST information were fitted locally by using multiplying factors to petrophysical data. The phase 2 project was planned using this model (Fig. 4).

Figure 4.

Third 3D geological model constructed in 2004 using a multi-Gaussian algorithm to reproduce different body geometries, using a non-stationary matrix of vertical proportion curve (VPC).

Figure 4.

Third 3D geological model constructed in 2004 using a multi-Gaussian algorithm to reproduce different body geometries, using a non-stationary matrix of vertical proportion curve (VPC).

A new seismic interpretation, focusing on the architecture of the reservoir, was made in 2007. The work was aimed to correlate horizons interpreted to be correlative surfaces of relative sea level falls with turbidite channel complexes observed on log and core data. In order to use this information in the geological model, an analog outcrop data was needed to help volumetrically distributing the facies in the model. The Annot outcrops located in French Alps were selected as the analog. The information collected was used for facies control and distribution of the channel complexes that determine the petrophysical proprieties. The model built with the new soft and hard data had a better fitting with the dynamic data obtained during Phase I (Fig. 5).

Figure 5.

Fourth 3D geological model constructed in 2004 using new seismic interpretation, focusing on the architecture of the reservoir and information of the Annot outcrop analog data.

Figure 5.

Fourth 3D geological model constructed in 2004 using new seismic interpretation, focusing on the architecture of the reservoir and information of the Annot outcrop analog data.

The last model was built by using all data acquired during the implementation of phase 2 and integrated several sources of information: (1) well logs and cores, (2) Annot analog outcrop, (3) seismic attributes from an elastic inversion of a HD 3D cube, (4) interpretation of seismic horizons corresponding to correlative surfaces of relative sea level falls related to turbidite channel complexes, and (5) dynamic information of producer wells. The method used in this last model will be described in this paper.

Jubarte Field Geological Setting

The Campos basin is a passive margin basin resulting from Gondowana supercontinent breakup and the opening of South Atlantic Ocean. It is located on the eastern Brazilian margin along Rio de Janeiro and south of Espírito Santo States, between 21° and 23° S latitude. The northern limit is the Vitoria basement high; the southern boundary is the Cabo Frio highs. The sedimentary history can be summarized by five depositional mega-sequences: continental rift (early Neocomian-early Aptian), transitional evaporitic (middle Aptian-early Albian), shallow marine (early-middle Albian), marine transgressive (late Albian-early Tertiary) and marine regressive (early Tertiary–Recent).

The sandy and conglomeratic reservoirs of Jubarte Field are turbidites deposited during the marine transgressive mega-sequence. That sequence was developed in response to a combined effect of thermal subsidence, sedimentary load, and a first-order sea-level rise. Onlapping, deepening-upward sedimentation throughout the eastern Brazilian margin characterize this tectonic process. Several turbiditic successions are found within the marine transgressive mega-sequence, the most important of which occurred in Albian, Cenomanian, Coniacian-Turonian, Maastrichtian, and Paleocene strata.

The Jubarte field is located in water depths of 2700 and 3150 m. The reservoir is a late Maastrichtian, deep-water, 400 m thick stacked succession of sandstone turbidite channeled complexes (Fig. 4). These turbidites were deposited in a southwest-northeast elongated, 8–14 km long and 1.5–5.5 km wide trough. This trough was created by salt tectonics and produced by the combined effects of subsidence along listric faults rooted in salt pillows and erosion by turbidity currents. (Gontijo et al., 2005). Normal growth faults that were triggered during halokinesis controlled the evolution of the trough-filling pattern. This pattern is composed of a series of stacked composite channels filled by layers of sandstone interbedded with thin heterolithic layers, having a high net-to-gross ratio of about 80%. The average porosity is around 25%. and average permeability about 900 mD.

The sandstone beds are mostly massive or possess low-angle stratification. Occasionally, 2–21 m thick cross-bedded bodies occur of very coarse-grained (sometimes conglomeratic) to medium-grained, fining-upward sandstones and parallel and rippled cross-laminated, fine to very fine-grained sandstones. Marls and mudstones are subordinate ( Bezzera et al., 2004 and Gontijo et al., 2005). Above the reservoir, there is a chaotic diamictite (from several debris flows) containing abundant glauconite.

The Maastrichtian succession was subdivided into a series of sequences bounded by unconformities, using a combination of core and wireline log data (Fontanelli et al., 2009). The third-order sequence (sensu Mitchum and Van Wagoner, 1991; i.e., about 0.5-5Ma), which corresponds to the 350 m thick sandstone package, was deposited in about one million years. Three fourth-order sequences (I, II, and III) were defined on the basis of seismic data by Fontanelli et al. (2009). Those authors described a 162 m cored interval of the well referred as JUA and defined eight fifth-order sequences (a, b, c, d e, f, g, h). The fifth-order sequences from “a” to “d” were part of fourth-order Sequence II; fifth-order sequences from “e” to “h” were part of fourth-order Sequence III. These sequences were correlated from well JUA to JUB, an 18m cored well, located about 600 m northwest from JUA (Fig. 6). Fontanelli et al. (2009) could not confidently extend to more distant wells in the field the fourth- and fifth-order stratigraphic subdivisions observed between these two close wells.

Figure 6.

East-West cross-section of the Jubarte Field showing the Maastrichtian reservoirs and location of the study area and wells JUA and JUB studied by Fontanelli et al. (2009).

Figure 6.

East-West cross-section of the Jubarte Field showing the Maastrichtian reservoirs and location of the study area and wells JUA and JUB studied by Fontanelli et al. (2009).

Gontijo et al. (2005) interpreted the mineralogical and textural immaturity and the presence of coal fragments in the sandstones as the product of hyperpicnal flows promoted by major fluvial floods. These flows would have been captured by submarine canyons, driving the sediments to deep marine settings.

There are two main possible directions of sedimentary supply to the Jubarte deep-water systems: one is from the southwest, mainly from the erosion of the Precambrian Ribeira terrain in the Rio de Janeiro State; the other is from to the northwest, from the erosion of Precambrian Araçuaí terrain in the Espírito Santo State, and controlled by the Colatina shear zone (Fontanelli et al., 2009).

Grès d´Annot Geological as an Analog Turbidite System

Outcrop analogs to improve subsurface reservoir models are widely applied in oil reservoir studies. The objective is to help distributing rock facies and petrophysical properties throughout the model, using subsurface data (rock and log) as hard data. However, to select an appropriate group of outcrops that have similar deposition geometry to a corresponding sedimentological and stratigraphic setting is not a trivial exercise (Moraes et al., 2004). The Grés dˋAnnot outcrops, located in the French Southern Alps, northwards of Nice are considered an adequate analog for Jubarte field reservoirs based on the following:

  1. The Jubarte Field sandstones are confined in an elongated trough of similar dimensions (a few kilometers wide).

  2. Both systems exhibit similar facies associations, dominated by thick, massive, coarse-grained to gravel sandstones in association with medium- to coarse-grained sandstones in low angle stratification bodies interlayered by thin and discontinuous heterolithic layers.

  3. The Grés dˋAnnot third-order sequences of Late Eocene tend to be the equivalent to the third order sequence of upper Maastrichtian in Jubarte Field in sedimentological and stratigraphical behavior.

The Annot sandstone outcrops in the French Southern Alps are deposits of an Eocene-Oligocene marine basin highly deformed by Mio-Pliocene tectonics that is preserved only in syncline zones. The sandstones represent a thick sequence of gravity deposits settled during the late Eocene (Priabonian) to Oligocene (Joseph et al., 2000).

The African/European plate convergence during the Eocene developed a complex tectonic regime, in which both extensional and compressional movements induced the initiation of several large scale trough trending northwest-southeast. The troughs later evolved as sub-basins and were filled by sediments supplied from the south of the Corsiga-Sardinia massif. Due to the progressive development of the foreland system from east to west, the sub-basins were successively deformed and filled by sand-rich turbiditic systems onlapping its margins. The sediment influx was mainly longitudinal from southeast to northwest (Joseph et al., 2000).

The thickness of the sandstones deposits is very variable, ranging from a few hundreds of meters on structural highs to more than one thousand meters in deep sub-basins (Joseph et al., 2000).

Well exposed cliffs located near the town of Annot are known as the Scaffarels or Chambre du Roi outcrops and correspond to the proximal part of Annot sub-basin. This proximal area is very sand-rich and characterized by the development of deeply incised ephemeral channels, filled by pebbly to coarse-grained high density sandy turbidites. The outcrop thicknesses can reach 120 m. At their base, they pinch-out by onlap of the successive sand units onto Blue Marls paleoslope.

The sand units generally present a more or less tabular shape; however, Joseph et al. (2000) in analyzing of photo-panels notes that these tabular sand units contain large incised erosional channels (several meters deep) which truncate previous deposits. The filling of the major erosional channels is composed of massive sand units, which may up to 40 m thick, without obvious diastems. These units are composed of pebbly to very coarse sandstone 1 to 8 m thick, organized in amalgamated beds. Frequently, the base of the beds is strongly erosive, but the amalgamation may be marked only by the existence of thin horizons of granules and small pebbles. The basal part of the beds includes sometimes coarse-grained traction carpets and more frequently decimeter-thick mud pebbles of shale or marls. The middle part of the beds is well-graded and marked by a slight increase in grain size. The upper part of the beds is commonly composed of a coarse to medium-grained sandstone having parallel and convolute laminations. No shale breaks and no heterolithic levels have been found. This facies association is associated to high-density turbidites of ephemeral erosive channels.

More tabular sand beds can occur with thickness commonly lesser than 3m, rarely erosive at the base. Heterolithic levels occur interbedded with tabular sand beds and are composed of an alternation of decimeter-thick layers of marls and silts with thin-bedded low-density turbidites (5 to 40 cm thick). These sand turbidites are fine- to medium-grained, often bioturbated, and display a lot of parallel ripples and convolute laminations. The thickness of individual beds varies rapidly and packages of several beds can disappear in a few tens of meters. This facies association may reach a thickness of a few meters that corresponds to a strong decrease in the sand supply or towards overbank deposits (Joseph et al., 2000).

Stratigraphic Framework

A stratigraphic framework is essential to any reservoir characterization study, for a correct definition of facies modeling as well as the positioning of the most important heterogeneities. Stratigraphic surfaces are references to determine the framework of the 3D grid model. The stratigraphic framework of Jubarte rocks has been created using cores descriptions, integrated log interpretation, and intra-reservoir seismic surfaces. Those different scales of information define sedimentary elements as bed, bed-set, and complex of channels. Figure 7 shows the JUA core description that presents the different scale sedimentary elements.

Figure 7.

Lithological and gamma-ray logs showing a core of JUA description that presents the different scale sedimentary elements, such as bed, bed-set and channel-complex.

Figure 7.

Lithological and gamma-ray logs showing a core of JUA description that presents the different scale sedimentary elements, such as bed, bed-set and channel-complex.

The beds observed in cores and logs are dominated by a facies association having a thickness in the order of tens of meters and lateral extensions of hundreds of meters. Three facies association were recognized and named according to the dominant facies: (Cgl) coarse massive sandstone and conglomerate; (Ss) medium- to coarse-grained sandstone having parallel low angle stratification and (Ht) Heterolithics—alternation of fine- to medium-grained sandstones and mudstones. Those facies associations classes have been used to fill discrete elementary cells of the 3D model.

The bed-sets have been focused as channel deposits in the 3D model. Information about this stratigraphic element is gathered from core, log and data from analog outcrops. Basic bed-sets in Jubarte Field are 20m thick and exhibit both fining- and thinning-upward patterns without obvious diastems. Frequently, thin (less than 1 m thick) conglomeratic beds occur at the base. The basal part occasionally includes coarsegrained traction carpets and, more frequently, decimeter-thick mud pebbles of shale or silt. The middle part of the bed-sets is well-graded and marked by a slight decrease in grain size. The upper part is commonly observed to be coarse- to medium-grained sandstone having parallel and convolute laminations. It is common that heterolithic layers end the bed-sets cycles is common. This information has been used to construct vertical proportion curves (VPC) and define the geometry and dimensions of the beds for define the variograms.

The channel complexes are limited by erosive intra-reservoir surfaces mapped in seismic data (Fig. 8). These surfaces correspond to the base of a fifth order sequence observed in cores and logs and represent the basal erosive event of a fourth order major event. Twelve erosive events, have been identified and numbered from base to top between 100 and 1100, in one hundred steps. The event named 700 corresponds to a fourth-order sequence boundary defined by biostratigraphy (nannofossil data). The erosive surface when observed in seismic sections is channel shaped and, for this reason, named channel complex.

Figure 8.

Stratigraphic section showing channel complexes that are limited by the erosive intra-reservoir surfaces mapped by seismic data. Those surfaces correspond to the base of a fifth order sequence observed in cores and logs and at least one of them represents the basal erosive event of a fourth order major event. The twelve erosive events numbered from base to top between 100 and 1100 in one hundred steps are labeled. The event named 700 corresponds to a fourth order sequence boundary defined by biostratigraphy (nannofossil data).

Figure 8.

Stratigraphic section showing channel complexes that are limited by the erosive intra-reservoir surfaces mapped by seismic data. Those surfaces correspond to the base of a fifth order sequence observed in cores and logs and at least one of them represents the basal erosive event of a fourth order major event. The twelve erosive events numbered from base to top between 100 and 1100 in one hundred steps are labeled. The event named 700 corresponds to a fourth order sequence boundary defined by biostratigraphy (nannofossil data).

In general, the channel complexes dimensions decrease from the base to the top of the reservoir. Following this trend, the thickness of the bed-sets also decreases. The observed variations of the Jubarte channel complexes dimensions have a good match with the channels complexes present in the 3D geological model of Annot outcrops, where 10 erosive channel complex surfaces have been mapped. Based on those similarities, the geometry and the vertical and lateral facies distributions of Annot 3D model has been used to help volumetrically distribute the facies in the Jubarte 3D model. One important observation about the Annot outcrops is that layers deposited directly above the erosive surfaces are parallel (show an onlap filling geometry) to the less confined deposits at the top of the deposits. This geometry has been considered when building the layering of the structural grid do the Jubarte 3D model parallel to the top of the reservoir.

Construction of the 3D Geologic Grid

The first step to construct a 3D geological model is the definition of the boundaries of the reservoir, the second step is to establish how the layers are correlated inside the boundary surfaces, and the third step is to group the complexes of bed-sets having similar sedimentary and diagenetic processes.

The boundaries of the modeled reservoir were its structural top and base mapped on seismic data. The areal limit was defined by the pinchout of the sandstone deposits to the north, south, and west, and by a normal fault located in the northeast area of the field.

Layering was defined based on observation of the Annot outcrops, in which the layers are deposited independently of the basal limits of the complex of bed-sets (Fig. 9). The most representative stratigraphic surface observed in seismic was the top of the reservoir. The layering style chosen for the model was conformable to that surface using cells of constant thickness. The final grid has 100 by 100 by 1 meter cells.

Figure 9.

3D geological model (made by the IFP group, Joseph et al., 2000) of the Annot outcrops located near the town of Annot, known as the Scaffarels or Chambre du Roi outcrops correspond to the proximal part of Annot sub-basin. This model was used as “soft data” to fill the gaps in Jubarte Field.

Figure 9.

3D geological model (made by the IFP group, Joseph et al., 2000) of the Annot outcrops located near the town of Annot, known as the Scaffarels or Chambre du Roi outcrops correspond to the proximal part of Annot sub-basin. This model was used as “soft data” to fill the gaps in Jubarte Field.

Each channel complex was bounded by an erosive surface mapped at its base and a composite surface of the upper channel complex (Fig. 10).

Figure 10.

3D geological model showing each channel complex.

Figure 10.

3D geological model showing each channel complex.

Filling the 3D Geological Grid with Facies

Sedimentary and stratigraphic information is incorporated into the geological grid based on soft and hard data. Hard data comes from the wells, basically cores, and interpreted facies from image and conventional logs (density, neutron, gamma ray, and resistivity). Soft data are obtained from seismic attributes and outcrops.

Grid cells intersected by wells having core were assigned facies data from the cores, whereas grid cells crossed by wells without core were assigned facies based on well log interpretation (Fig. 11). Each channel complex must have a global proportion and a vertical variogram for the three facies association. They were calculated using information from wells, when enough data are available, and information from seismic and outcrops when data from wells were not available

Figure 11.

Grid cells intersected by wells with facies data.

Figure 11.

Grid cells intersected by wells with facies data.

The vertical proportions curves were estimated based on the VP/VS seismic attribute (i.e., the ratio of compressional and shear sonic velocities). The VP/VS ratio can be used to assess reservoir quality and discriminate porous sand (2.5-3.75) strata from shale (1.73-3.0) based in the different VP/VS ratios of those lithologies (Ross, 2010) (Fig. 12). As the criterion for estimates of 100% sand or shale based on VP/VS ratio is not absolute, the range 1.7-2.0 was used for shale/heterolithics and 2.3-2.7 for sandstones. Each channel complex facies proportion was calculated using VP/VS values for shale and sandstones that better fit global proportion observed in the well data (Fig. 13). Where the channel complex was not intersected by any well, the variogram and the global proportion were estimated using analog data.

Figure 12.

VP/VS ratio ranges of commonly encountered lithologies. VP/VS modified from Ross (2010).

Figure 12.

VP/VS ratio ranges of commonly encountered lithologies. VP/VS modified from Ross (2010).

Figure 13.

VP/VS Ratio distribution for a channel complex and correspondent estimated VPC. The vertical proportions curves have been estimated based on VP/VS (compressional/shear sonic velocities) seismic attribute. The VP/VS ratio can be used to assess reservoir quality and discriminate sand porous (2.5-3.75) strata from shale (1.73-3.0) based in the different VP/VS ratios of those lithologies.

Figure 13.

VP/VS Ratio distribution for a channel complex and correspondent estimated VPC. The vertical proportions curves have been estimated based on VP/VS (compressional/shear sonic velocities) seismic attribute. The VP/VS ratio can be used to assess reservoir quality and discriminate sand porous (2.5-3.75) strata from shale (1.73-3.0) based in the different VP/VS ratios of those lithologies.

The final facies distribution was modeled using a sequential indicator simulation algorithm considering the simple kriging with non-stationary proportion. The results show, as it was expected, similarities, both in facies association (core and logs) and in geometry of the bodies between the Jubarte Field and the Annot outcrops (Fig. 14).

Figure 14.

3D geological model showing the final facies distribution modeled using sequential indicator simulation algorithm considering the simple Krieging with non-stationary proportion based in the VP/VS ratios and analog outcrop data.

Figure 14.

3D geological model showing the final facies distribution modeled using sequential indicator simulation algorithm considering the simple Krieging with non-stationary proportion based in the VP/VS ratios and analog outcrop data.

Distribution of Petrophysical Properties

Petrophysical properties were modeled stochastically in order to capture the variations and heterogeneities of the reservoir. The porosity for each facies association was assigned considering the distribution calculated on log data corrected by measurements made in core plugs. Well data were up-scaled to the geological grid using arithmetic mean and the vertical and horizontal variograms were defined based in those data. Sequential Gaussian simulation algorithm with simple krieging was adopted as the interpolation method (Fig.15).

Figure 15.

3D geological model showing the final porosity using up-scaled data and vertical and horizontal variograms considering each facies association. Sequential Gaussian simulation algorithm with simple Krieging was adopted as the interpolation method.

Figure 15.

3D geological model showing the final porosity using up-scaled data and vertical and horizontal variograms considering each facies association. Sequential Gaussian simulation algorithm with simple Krieging was adopted as the interpolation method.

The permeability was also modeled stochastically using the correlation with porosity. There was no permeability value from logs. Therefore, the estimation was based on empirical correlation between petrophysical measurements from cores and multiple log statistical linear regression. The data were up-scaled to the geological model using arithmetic mean and extrapolated using sequential Gaussian simulation followed by co-krieging with porosity for each facies (Fig. 16).

Figure 16.

3D geological model showing the final permeability up-scaled from wells to the geological model using arithmetic mean and extrapolated using Sequential Gaussian simulation in association with co-Krieging for porosity of each facies.

Figure 16.

3D geological model showing the final permeability up-scaled from wells to the geological model using arithmetic mean and extrapolated using Sequential Gaussian simulation in association with co-Krieging for porosity of each facies.

Simulation grid and upscaling

The geological grid cell was up-scaled from 100 x 100 x 1 m to 200 x_200 x_5m, resulting in a flow simulation model with nearly 50,000 cells. Porosity up-scaling was made using arithmetic average. Permeability up-scaling was made using a simple two-step approach combining arithmetic average for the horizontal direction and an empirical reduction factor based on net to gross ratio for the vertical direction. To obtain the effective horizontal permeability of a coarse cell, it was used an arithmetic average for each row of finer cells. The vertical permeability up-scaling was calculated using an empirical reduction factor based on an exponential correlation with net to gross ratio obtained from flow simulation of micro-models.

Simulation results

The estimated volume of oil in place based on the model differed from the volume calculated based on 2D maps by less than 3%. However, our model was based on a set of different sources of information that better represent the spatial distribution of heterogeneities and local buffers. Therefore, the history match of production data (i.e., oil rate, pressure, and water breakthrough), was reached globally. Starting from a model adjusted globally, an accurate match for well history was obtained more quickly after inclusion of dynamic parameters, such as relative permeability curve and transmissibility of mapped surface. The resulting simulation model shows major improvements over previous models.

Conclusions

The use of various sources of data allows the enhancement of the geological representation of a subsurface reservoir model. It is important, however, to correctly understand how information can be gathered to improve our knowledge on the distribution of heterogeneities and on the geometry of different geological bodies.

Two different sources of data were important to improve the 3D model of Jubarte: the analogue information from Annot outcrops and the seismic attribute VP/VS ratio. The similarities observed between the litho-stratigraphy of the Annot sandstone and the cores of the studied reservoir made it possible to use information from outcrops data to populate the model with the interpreted facies association and the petrophysical properties. Facies continuity and proportion estimated from outcrops were applied to the 3D stochastic model controlled by the stratigraphic framework. VP/VS ratio proved to be a good seismic attribute to infer the expected probability of occurrence of different rocks in the 3D model. However, it is important to study better how to correlate that attribute with the rocks. The flow simulation model constructed on basis of this new integrated geological and geophysical approach produced a better history match mainly on water cut values and, consequently, on the quality of the production forecast for the field.

References

Bezzera
,
M.F.C.
,
C.
Pedroso
Jr.
,
A.C.C.
Pinto
, and
C.H.L.
Bruhn
,
2004
,
The appraisal and development plan for the heavy oil Jubarte Field, deep-water Campos Basin, Brazil
:
Offshore Technology Conference
,
Houston, Texas, USA
, ID 16301-MS,
5
p.
Daher
Jr,
B.
,
C.A.M.
Siqueira
,
I.
Nascimento
,
I.A.
Pinto
,
J.B.
Farias
,
R.A.B.
Vieira
, and
R.O.
Goulart
,
2007
,
Jubarte Field – Development Strategy: Offshore Technology Conference
,
Houston, Texas, U.S.A.
, OTC 19088-PP,
5
p.
Fontanelli
,
P.D.R.
,
L.F.
De Ros
, and
M.V.D.
Remus
,
2009
,
Provenance of deep-water reservoir sandstones from the Jubarte oil field, Campos Basin, Eastern Brazilian Margin
:
Marine and Petroleum Geology
  v.
26
, p.
1274
1298
.
Gontijo
,
R.C.
Gontijo
,
C.E.
Souza Cruz
,
J.L.L.
Caldas
,
L.M.
Arienti
, and
R.S.F.
D'Avila
,
2005
,
Structurally controlled sand-rich gravity deposits of the Jubarte Oil field (Brazil deep-water sedimentation on the Southeast Brazilian margin project)
:
AAPG Search and Discovery Article #90039
 .
Grassi
,
A.D.A.
,
A.H.A.
Castro
, and
G.A.
Albertão
,
2004
,
Bacia de Campos
:
PHOENIX
 , v.
65
, no.
6
, p.
1
6
.
Joseph
,
P.
,
N.
Babonneau
,
A.
Bourgeois
,
G.
Cotteret
,
R.
Eschard
,
B.
Garin
,
D.
Granjeon
,
O.
Lerat
,
C.
Ravenne
,
O.G.
Souza
,
F.
Guillocheu
, and
J.M.
Quemener
,
2000
,
The Annot Sandstone Outcrops (French Alpes): Architecture description as input for quantification and 3D reservoir modeling
:
GCSSEPM Foundation 20Th Annual Bob F. Perkins Research Conference
, p.
442
449
.
Mitchum
,
R.M.
Jr
, and
J.C.
Van Wagoner
,
1991
,
High-frequency sequences and their stacking patterns: sequence-stratigraphic evidence of high-frequency eustatic cycles
:
Sedimentary Geology
 , v.
70
, p.
131
160
Moraes
,
M.A.S.
,
P.R.
Blaskovishy
, and
P.
Joseph
,
2004
, The Grès dˊAnnot as an analogue for Brazilian Cretaceous sandstone reservoirs: comparing convergent to passive-margin confined turbidites, in
P.
Joseph
, and
S.A.
Lomas
, eds.,
Deep-Water Sedimentation in the Alpine Basin of SE France: New perspectives on the Grès dˊAnnot and related systems: Geological Society London Special Publication
 
221
, p.
419
436
.
Roos
,
C.P.
,
2010
,
AVO ritualization and functionalism (then and now)
:
The Leading Edge
 , v.
29
, (May), p.
532
538
.

Acknowledgments

The authors wish to thank Petrobras for the permission to publish this paper. Discussions with Marco A. S. Moares and Philippe Joseph improved our understanding of Annot analogs outcrops. We are grateful to Petrobras managers Carlos Pedroso Jr and Carlos H.L Bruhn for support this study. We thank reviewer José Mauricio Bento and GCSSEPM editors for suggestions that greatly improved the manuscript.

Figures & Tables

Figure 1.

Jubarte Field Location (red dashed polygon).

Figure 1.

Jubarte Field Location (red dashed polygon).

Figure 2.

First 3D geological model built based on exploratory interpretation in 2002.

Figure 2.

First 3D geological model built based on exploratory interpretation in 2002.

Figure 3.

Second 3D geological model (2003) using Boolean algorithm in order to incorporate the inner heterogeneities of the reservoir that suggest channelized shape.

Figure 3.

Second 3D geological model (2003) using Boolean algorithm in order to incorporate the inner heterogeneities of the reservoir that suggest channelized shape.

Figure 4.

Third 3D geological model constructed in 2004 using a multi-Gaussian algorithm to reproduce different body geometries, using a non-stationary matrix of vertical proportion curve (VPC).

Figure 4.

Third 3D geological model constructed in 2004 using a multi-Gaussian algorithm to reproduce different body geometries, using a non-stationary matrix of vertical proportion curve (VPC).

Figure 5.

Fourth 3D geological model constructed in 2004 using new seismic interpretation, focusing on the architecture of the reservoir and information of the Annot outcrop analog data.

Figure 5.

Fourth 3D geological model constructed in 2004 using new seismic interpretation, focusing on the architecture of the reservoir and information of the Annot outcrop analog data.

Figure 6.

East-West cross-section of the Jubarte Field showing the Maastrichtian reservoirs and location of the study area and wells JUA and JUB studied by Fontanelli et al. (2009).

Figure 6.

East-West cross-section of the Jubarte Field showing the Maastrichtian reservoirs and location of the study area and wells JUA and JUB studied by Fontanelli et al. (2009).

Figure 7.

Lithological and gamma-ray logs showing a core of JUA description that presents the different scale sedimentary elements, such as bed, bed-set and channel-complex.

Figure 7.

Lithological and gamma-ray logs showing a core of JUA description that presents the different scale sedimentary elements, such as bed, bed-set and channel-complex.

Figure 8.

Stratigraphic section showing channel complexes that are limited by the erosive intra-reservoir surfaces mapped by seismic data. Those surfaces correspond to the base of a fifth order sequence observed in cores and logs and at least one of them represents the basal erosive event of a fourth order major event. The twelve erosive events numbered from base to top between 100 and 1100 in one hundred steps are labeled. The event named 700 corresponds to a fourth order sequence boundary defined by biostratigraphy (nannofossil data).

Figure 8.

Stratigraphic section showing channel complexes that are limited by the erosive intra-reservoir surfaces mapped by seismic data. Those surfaces correspond to the base of a fifth order sequence observed in cores and logs and at least one of them represents the basal erosive event of a fourth order major event. The twelve erosive events numbered from base to top between 100 and 1100 in one hundred steps are labeled. The event named 700 corresponds to a fourth order sequence boundary defined by biostratigraphy (nannofossil data).

Figure 9.

3D geological model (made by the IFP group, Joseph et al., 2000) of the Annot outcrops located near the town of Annot, known as the Scaffarels or Chambre du Roi outcrops correspond to the proximal part of Annot sub-basin. This model was used as “soft data” to fill the gaps in Jubarte Field.

Figure 9.

3D geological model (made by the IFP group, Joseph et al., 2000) of the Annot outcrops located near the town of Annot, known as the Scaffarels or Chambre du Roi outcrops correspond to the proximal part of Annot sub-basin. This model was used as “soft data” to fill the gaps in Jubarte Field.

Figure 10.

3D geological model showing each channel complex.

Figure 10.

3D geological model showing each channel complex.

Figure 11.

Grid cells intersected by wells with facies data.

Figure 11.

Grid cells intersected by wells with facies data.

Figure 12.

VP/VS ratio ranges of commonly encountered lithologies. VP/VS modified from Ross (2010).

Figure 12.

VP/VS ratio ranges of commonly encountered lithologies. VP/VS modified from Ross (2010).

Figure 13.

VP/VS Ratio distribution for a channel complex and correspondent estimated VPC. The vertical proportions curves have been estimated based on VP/VS (compressional/shear sonic velocities) seismic attribute. The VP/VS ratio can be used to assess reservoir quality and discriminate sand porous (2.5-3.75) strata from shale (1.73-3.0) based in the different VP/VS ratios of those lithologies.

Figure 13.

VP/VS Ratio distribution for a channel complex and correspondent estimated VPC. The vertical proportions curves have been estimated based on VP/VS (compressional/shear sonic velocities) seismic attribute. The VP/VS ratio can be used to assess reservoir quality and discriminate sand porous (2.5-3.75) strata from shale (1.73-3.0) based in the different VP/VS ratios of those lithologies.

Figure 14.

3D geological model showing the final facies distribution modeled using sequential indicator simulation algorithm considering the simple Krieging with non-stationary proportion based in the VP/VS ratios and analog outcrop data.

Figure 14.

3D geological model showing the final facies distribution modeled using sequential indicator simulation algorithm considering the simple Krieging with non-stationary proportion based in the VP/VS ratios and analog outcrop data.

Figure 15.

3D geological model showing the final porosity using up-scaled data and vertical and horizontal variograms considering each facies association. Sequential Gaussian simulation algorithm with simple Krieging was adopted as the interpolation method.

Figure 15.

3D geological model showing the final porosity using up-scaled data and vertical and horizontal variograms considering each facies association. Sequential Gaussian simulation algorithm with simple Krieging was adopted as the interpolation method.

Figure 16.

3D geological model showing the final permeability up-scaled from wells to the geological model using arithmetic mean and extrapolated using Sequential Gaussian simulation in association with co-Krieging for porosity of each facies.

Figure 16.

3D geological model showing the final permeability up-scaled from wells to the geological model using arithmetic mean and extrapolated using Sequential Gaussian simulation in association with co-Krieging for porosity of each facies.

Contents

References

References

Bezzera
,
M.F.C.
,
C.
Pedroso
Jr.
,
A.C.C.
Pinto
, and
C.H.L.
Bruhn
,
2004
,
The appraisal and development plan for the heavy oil Jubarte Field, deep-water Campos Basin, Brazil
:
Offshore Technology Conference
,
Houston, Texas, USA
, ID 16301-MS,
5
p.
Daher
Jr,
B.
,
C.A.M.
Siqueira
,
I.
Nascimento
,
I.A.
Pinto
,
J.B.
Farias
,
R.A.B.
Vieira
, and
R.O.
Goulart
,
2007
,
Jubarte Field – Development Strategy: Offshore Technology Conference
,
Houston, Texas, U.S.A.
, OTC 19088-PP,
5
p.
Fontanelli
,
P.D.R.
,
L.F.
De Ros
, and
M.V.D.
Remus
,
2009
,
Provenance of deep-water reservoir sandstones from the Jubarte oil field, Campos Basin, Eastern Brazilian Margin
:
Marine and Petroleum Geology
  v.
26
, p.
1274
1298
.
Gontijo
,
R.C.
Gontijo
,
C.E.
Souza Cruz
,
J.L.L.
Caldas
,
L.M.
Arienti
, and
R.S.F.
D'Avila
,
2005
,
Structurally controlled sand-rich gravity deposits of the Jubarte Oil field (Brazil deep-water sedimentation on the Southeast Brazilian margin project)
:
AAPG Search and Discovery Article #90039
 .
Grassi
,
A.D.A.
,
A.H.A.
Castro
, and
G.A.
Albertão
,
2004
,
Bacia de Campos
:
PHOENIX
 , v.
65
, no.
6
, p.
1
6
.
Joseph
,
P.
,
N.
Babonneau
,
A.
Bourgeois
,
G.
Cotteret
,
R.
Eschard
,
B.
Garin
,
D.
Granjeon
,
O.
Lerat
,
C.
Ravenne
,
O.G.
Souza
,
F.
Guillocheu
, and
J.M.
Quemener
,
2000
,
The Annot Sandstone Outcrops (French Alpes): Architecture description as input for quantification and 3D reservoir modeling
:
GCSSEPM Foundation 20Th Annual Bob F. Perkins Research Conference
, p.
442
449
.
Mitchum
,
R.M.
Jr
, and
J.C.
Van Wagoner
,
1991
,
High-frequency sequences and their stacking patterns: sequence-stratigraphic evidence of high-frequency eustatic cycles
:
Sedimentary Geology
 , v.
70
, p.
131
160
Moraes
,
M.A.S.
,
P.R.
Blaskovishy
, and
P.
Joseph
,
2004
, The Grès dˊAnnot as an analogue for Brazilian Cretaceous sandstone reservoirs: comparing convergent to passive-margin confined turbidites, in
P.
Joseph
, and
S.A.
Lomas
, eds.,
Deep-Water Sedimentation in the Alpine Basin of SE France: New perspectives on the Grès dˊAnnot and related systems: Geological Society London Special Publication
 
221
, p.
419
436
.
Roos
,
C.P.
,
2010
,
AVO ritualization and functionalism (then and now)
:
The Leading Edge
 , v.
29
, (May), p.
532
538
.

Related

Citing Books via

Close Modal
This Feature Is Available To Subscribers Only

Sign In or Create an Account

Close Modal
Close Modal