Hydrocarbon Reservoir Modeling: Comparison Between Theoretical and Real Petrophysical Properties from the Namorado Field (Brazil) Case Study

Home , Reservoir modeling

Hydrocarbon reservoir modeling: comparison between theoretical and real petrophysical properties from the Namorado Field (Brazil) case study.

Hashimoto, Marcos Deguti, Student from the Master in Oil Engineering E-mail: [email protected]

1. Abstract In reservoir characterization and modeling, due to information-acquisition’s high costs, frequently only indirect measurements of the subsurface properties such as seismic reflection data is available. In the worst-case scenario, only regional geological information is at disposal. In an attempt to provide deeper insights over the study area, with low costs, modeling synthetic reservoirs has been a reliable tool to better characterize reservoir/prospects. In this work two synthetic hydrocarbons reservoirs were modelled recurring to two different approaches to characterize Earth’s subsurface petrophysical (facies, porosity and permeability) and elastic (P-wave, S-wave and density) properties. In the second half of 2013, during the IST (Instituto Superior Técnico) Internship, a synthetic reservoir was conceived and modeled using Namorado Field’s (Campos Basin, Rio de Janeiro, Brazil) as reference. During this intern public data, knowledge, papers, books and dissertations were gathered. In order to validate and certify this outcome, a new synthetic reservoir was proposed, but this time using real data for this field provided by the Brazilian Oil & Gas Agency (ANP). This dissertation addresses the comparison between the theoretical and real synthetic reservoir results, validating the first reservoir step-by-step. The major conclusion reached confirms that the theoretical synthetic reservoir outputs reliable results, however with caution in some of the modelled properties.

Keywords: Hydrocarbon synthetic reservoir, Reservoir Modeling, Rock Physics Model, Petrophysical properties, Namorado Field, Campos Basin (Brazil).

2. Introduction The objective of this work is to compare the results obtained by two different synthetic reservoirs models: a theoretical-approached built during the internship at CERENA, using public papers and books; and the real-approached using real data provided by ANP (Agência Nacional do Petróleo, Gás Natural e Biocombustíveis). This paper will present both reservoir modeling procedures step-by-step, including all related theory and parameters used. At last, all the obtained results from both models will be presented and compared to each other. Finally, the pertinent conclusions will be presented. The real data, necessary for modeling the second reservoir, were provided by The Brazilian Oil and Gas Agency (BDEP/ANP). The request was made through the Department of Mines and Petroleum Engineering of the Escola Politécnica (University of São Paulo - USP).

1 2.1. ANP/BDEP Data ANP provided the data used in this study to model the real-approached reservoir. The request involved core plugs’ descriptions and the following loggings: sonic (DT), gamma ray (GR), neutron (NPHI) and density (RHOB) from six wells: 1RJS 0019 RJ, 1RJS 0387 RJ, 3NA 0005A RJS, 3NA 0021B RJS, 3RJS 0393D RJ, 4RJS 0042 RJ. It is important to note that not all wells have continuous measurements or all the four loggings at the same depth. It was applied a data conditioning: first, it was correlated the loggings measurements and the Namorado’s Sandstone and shale. Other lithologies were not considered within this study. The second filter applied was the logging depth. In order to keep it as much real as possible it was used only data from the Namorado Fields depth, which ranges from 2940 to 3300 meters (Barboza, 2005).

3. Campos Basin Campos Basin is located between the northern coast of Rio de Janeiro (Brazil) and the southern coast of Espírito Santo (Brazil). This basin has approximately 100,000 km² and more than 1,600 wells drilled over nearly four decades. This Basin was formed during the breakup of the supercontinent Gondwana (about 140 Ma or millions of years ago) creating the Atlantic Ocean and dividing the current South American and African continent. This rupture results from the action of distensive forces producing a rift valleys system and developing horsts, grabens and half-grabens, bounded by synthetic and antithetic faults (Gabaglia, 1991). Two structural highs surrounds Campos Basin - one from the North (Vitória) and another from the South (Cabo Frio). Figure 1 exhibits the Campos Basin’s and Namorado Field’s location and shape.

Figure 1 - Location of the Campos Basin and Namorado Field (from Bacoccoli et al., 1980).

3.1. Namorado Field Namorado field (Figure 1) is located in west-central Campos Basin portion with an estimated area of 20 km2, approximately 80 km away from the coast and its water depth ranges 140 to 250 meters. Discovered in 1975, oil production began in June 1979 employing two platforms (PNA-1 and PNA-2) and the development itself began in December 1982. Its vertical depth ranges 2.940 and 3.300 meters (Barboza, 2005) with an average thickness of approximately 300 meters. Furthermore, according to Cruz (2003) the average porosity is 26%, the average oil saturation is 75% and the average permeability is 400 mD. The oil API degree is 28° and the viscosity is close to 1 centipoise (cP). Regarding the depositional evolution and according to Barboza (2005), Namorado Field has four different turbidite systems. In general, the depositional model proposes a turbidite stack, with narrow

2 channels stack at the base section, tending to amalgamated geometry channels stack at the top. Figure 2 presents the depositional evolution diagram.

Figure 2 - Namorado Field's depositional evolution diagram (from Barboza, 2005).

4. Synthetic Reservoir Workflow 4.1. Structural Model The first step of the reservoir modeling comprised building the structural model. Given the lack of real seismic, a structural model was developed from the start with the following attributes: siliciclastic channels system, anticlinal geometry and a normal fault (45° dip) perpendicular to the anticline’s axis plane. The model was divided into 151 x 200 x 300 cells grid (x, y, z), each one with 25 m x 25 m x 1 m resulting in a 3.75 Km x 5.00 Km x 300 m grid. The synthetic reservoir has three main layers (same 100-meter height). Figure 3 presents the structural model’s bulk volume.

Figure 3 - Reservoir's structural model.

4.2. Facies Model Lithofacies modeling was the next step. In order to reach maximum Namorado Field’s characteristics, the synthetic model tried to reproduce its depositional evolution. The sedimentary model was conceived using three different siliclastic channels pattern accordingly to the Figure 4 and Table 1 in which is presented the channels attributes parameters. Sand represents reservoir rock while shale, the non-reservoir rock. Figure 5 presents the final Facies model that comprises sand and shales. Table 1 - Silicilastic channels attibutes divided in layers Attribute/Layer Top Intermediate Base Amplitude (m) 500 600 800 Wavelength (m) 1500 1500 750 Width (m) 1200 500 150 Figure 4 - Channel’s attributes diagram Thickness (m) 10 10 10

Figure 5 - Both reservoir's facies model (yellow represents sand and grey represents shale).

The same structural and facies model were used to both modeling approaches. However, the following properties, accordingly to the theoretical and real approach and to the Facies Model, were differently calculated or simulated. In other words, cells filled with sand or shale suffers different simulation or calculations. The following sections will present each procedure and its results.

4.3. Properties Model 4.3.1. Porosity Both reservoir’s porosity were simulated through DSS (Direct Sequential Simulation) (Soares, 2011) algorithm, although, the conditional distributions used were different. The theoretical reservoir, or reservoir A, used statistical parameters data (mean, standard deviation, maximum and minimum) extracted from Fonseca (2011) while the real approach reservoir, or B, used effective porosity calculated employing RHOB log data provided by ANP (Equation 1).

ρ푚푎푡푟𝑖푥 − ρ푅퐻푂퐵 푙표푔 ρ푚푎푡푟𝑖푥 − ρ푠ℎ푎푙푒 휑푒푓푓푒푐푡𝑖푣푒 = − V푠ℎ푎푙푒 ( ) (1) ρ푚푎푡푟𝑖푥 − ρ푓푙푢𝑖푑 ρ푚푎푡푟𝑖푥 − ρ푓푙푢𝑖푑

Table 2 organizes their main statistical parameters. Table 2 - Porosity Conditional Distribution's Statistical parameters Reservoir A Reservoir B Porosity (%) Sand Channel Shale Sand Channel Shale Fonseca Distribution Fonseca Distribution Average 24,96 24,86 3,81 5,44 14,52 8,26 Std. Deviation 2,80 4,85 4,88 3,30 6,92 4,74 Minimum 0,01 12,22 0,01 0,01 0,10 0,06 Maximum 33,53 33,30 20,42 16,05 36,27 14,99

4.3.2. Permeability Both reservoirs’ permeability were simulated through Co-DSS (Direct Sequential Co-Simulation with Joint Distribution) (Horta et al., 2010) algorithm using porosity as secondary variable. Assuming the correlation between permeability and porosity, the Kozeny-Carman equation was used to calculate

4 permeability using porosity (as shown on Equation 2). The reservoir A used statistical parameters data (mean, standard deviation, maximum and minimum) from Fonseca (2011) study. 1 Φ3 퐶 Φ3 퐾 = 푑2 퐾 = 푑2 (2) 72 (1 − Φ)2 ∙ 휏 72 (1 − Φ)2 ∙ 휏

Where Ф is the porosity, d is the average grain diameter and 휏 is the tortuosity. Regarding both reservoirs, a coefficient C was added to infer more realism (Equation 2) to the simulation, according to parameters extracted from Fonseca (2011) (Table 3).

Table 3 - Permeability Conditional Distribution's Statistical parameters Reservoir A Reservoir B Permeability Sand Channel Shale (mD) Sand Channel Shale Fonseca Distribution Fonseca Distribution Average 562,40 562,75 1,39 1,42 242,09 1,40 Std. Deviation 418,98 321,48 3,61 3,56 328,58 1,64 Minimum 0,10 37,12 0,10 0,01 0,01 0,01 Maximum 3000 1530,14 58,98 22,07 3264,08 5,92

4.3.3. Density The reservoir A’s density was calculated using the Equation 3 and Table 4.

휌 = Φ ∙ 휌푓푙푢𝑖푑 + (1 − Φ) ∙ 휌푚푎푡푟𝑖푥 (3)

Table 4 - Mineral, density and proportion used on reservoir A Mineral Density (g/cm3) Sand Channel (%) Shale (%) Clay 2,40 4 70 Quartz 2,65 56 20 Feldspar 2,63 40 10

On the other hand, the reservoir B’s density was simulated through DSS (Direct Sequential Simulation) algorithm, using the RHOB data provided by ANP as conditional distribution (Table 5).

Table 5 - Reservoir B Density Conditional Distribution's Statistical parameters Reservoir B Density (g/cm³) Sand Channel Shale Average 2,3123 2,4212 Std. Deviation 0,1295 0,1092 Minimum 2,0134 1,8804 Maximum 2,6976 2,6689

5 4.3.4. Compressional Velocity The theoretical compressional velocity (reservoir A) was calculated individually to sand and shale. To sands velocity, it was used the Constant Cement Model from Dvorkin and Nur (1996) which calculates

푉푝 considering the cement between the grains (Equation 4). 4 퐾 + 퐺 (4) 푉 = √ 3 푝 휌

Where 퐾 is the bulk modulus, 퐺 is the shear modulus and 휌 is the density. Moreover, regarding reservoir A shale, it was used the empirical relationship from Oliveira et al (2005) shown in Equation 5.

푉푝 = −0,0582 ∙ 휑 − 0,0145 ∙ 푉푎푟푔 + 4,7634 (퐾푚/푠) (5)

Regarding reservoir B, compressional velocity (sand and shale) was simulated through DSS (Direct Sequential Simulation). The conditional distribution used sonic log data (DT) provided by ANP. Table 6 presents the conditional distribution's main statistical parameters.

Table 6 - Reservoir B Compressional Velocity Conditional Distribution's Statistical parameters Reservoir B Compressional Velocity (m/s) Sand Channel Shale Average 3549,45 3457,37 Std. Deviation 309,32 442,42 Minimum 2513,99 2973,19 Maximum 4788,63 5076,57

4.3.5. Shear Velocity The reservoir A’s sand shear velocity were calculated applying the physical relationship and using both bulk and shear modulus obtained in Dvorkin and Nur (1996) model (Equation 6). Regarding the shale from reservoir A, it was applied the Castagna (1985) empirical relationship (Equation 7).

퐺 푠푎푡 (6) 푉푠 푠푎푡 = √ 휌푠푎푡

Shale: 푉푠 = 0,862 푉푝 − 1,172 (퐾푚/푠) (Castagna,1985) (7)

On the other hand, regarding reservoir B, the shear velocity was calculated using Castagna (1993) empirical relationship (Equation 7 for shale and Equation 8 for sandstone). Sandstone: 푉푠 = 0,804 푉푝 − 0,856 (퐾푚/푠) (Castagna, 1993) (8)

4.3.6. Gassmann Fluid Substitution In sand regions above the Oil-Water-Contact, it was performed the Gassmann Fluid Substitution (Smith et al., 2003), introduced in 1951, which allows to recalculate the property saturated with another fluid as the Equation 9 shows.

6 4 퐾 + 퐺 퐺 휌 = 휌 + 휑 ∙ (휌 − 휌 ) 푉2 = 3 푉2 = (9) 2 1 푓푙2 푓푙1 푝 휌 푠 휌

In order to apply Gassmann Substitution, an oil and water contact was considered, as shown on Figure 6 (green represents sand channels saturated with 80% of oil, blue represents sand channels saturated with brine and gray, the shale). Gassmann Fluid Substitution was performed only in Reservoir A.

Figure 6 - Reservoir A saturation map

5. Results and Comparison This section presents both reservoir’s results achieved (statistical parameters as well as histogram). The blue color represents the reservoir A and the green refers to the reservoir B.

5.1. Porosity Comparing Table 7 and Figure 7, it is observed that there are, indeed, some differences in the porosity statistical parameters and histogram. In general, the reservoir B porosity was underestimated. Regarding histogram, both reservoirs has a bimodal distribution.

Table 7 - Porosity statistical parameters

Porosity (%) A B Average 15,54 11,54 Std. Deviation 10,45 6,29 Minimum 0,01 0,06 Maximum 33,53 36,27

Figure 7 - Simulated porosity histogram

5.2. Permeability Analyzing Table 8, it is possible to conclude some divergences in permeability’s statistical parameters. Reservoir B’s permeability is, in general, underestimated, which is a result of porosity’s underestimation, given that both properties are correlated (Kozeny-Carman). Another difference is the

7 maximum value of 3264 mD (sands), however this value is equal to the one presented by Fonseca (2011) (Table 3).

Table 8 - Permeability statistical parameters

Permeability (mD) A B Average 292,81 119,71 Std. Deviation 363,75 256,43 Minimum 0,01 0,01 Maximum 1530,14 3264,08

Figure 8 - Co-simulated permeability histogram

5.3. Density Although similar statistical parameters, as shown in Table 9, there is certain difference in the histogram (Figure 9): while reservoir A presents two families of data, the reservoir B has only one. One reason could be the exclusive application of Gassmann Fluid Substitution in the reservoir A. The second possible reason would be the density of the composition and mineralogy assumed for each facies in the reservoir A density. In general, the reservoir A density is sub estimated.

Table 9 - Density statistical parameters

Density (g/cm³) A B Average 2,3123 2,4212 Std. Deviation 0,1295 0,1092 Minimum 2,0134 1,8804 Maximum 2,6976 2,6689

Figure 9 - Density histogram

5.4. Compressional Velocity Analyzing Figure 10 and Table 10, it is noted a very high degree of results compatibility. This effect is observed not only in the average, minimum and maximum parameters but also in the histogram. This fact reflects the success of the theoretical compressional velocity calculus methodology. In other words, the Constant Cement Model (1996) allied with Oliveira et al (2005) empirical method may be one of the best method to calculate the Namorado Field compressional velocity.

8 Table 10 - Compressional velocity statistical parameters

Vp (m/s) A B Average 3432,68 3511,89 Std. Deviation 272,92 344,10 Minimum 2567,31 2513,99 Maximum 5045,78 5076,57

Figure 10 - Compressional velocity histogram

5.5. Shear Velocity The shear velocity was calculated based on the compressional velocity. As so, it presents as well a high degree of results compatibility when analyzing both Table 11 and Figure 11: average, standard deviation, minimum and maximum are very similar and the histogram presents high result consistency.

Table 11 - Shear velocity statistical parameters

Vs (m/s) A B Average 1985,10 1912,40 Std. Deviation 273,71 303,72 Minimum 1101,05 1130,30 Maximum 3199,83 3300,09

Figure 11 - Shear velocity histogram

6. Conclusion The major conclusion is that the theoretical reservoir, or A, is very reliable when considering the realism of its outputs/results, however with some caution in some properties results. Although, porosity and permeability present some results divergences, the theoretical reservoir presents reliable results concluding that the theory used to model this synthetic was indeed capable to reproduce the real Namorado Field properties pattern and statistical parameters. The evidence to conclude this is the very similar results from density (statistical parameters) and both compressional and shear velocities (consider porosity) presented when comparing their statistical parameters and histograms. Briefly, the conclusion reached is that the reservoir A is very similar to reality and its future use is allowed. However, some properties (porosity and permeability) demands special attention and some caution in their future use depending on the future objective. As future study, seismic inversion may be interesting for this work. In the context of this work, the seismic inversion could be performed in order to model acoustic impedance and therefore the shape and size of the facies (sand channels).

9 7. Reference

Bacoccoli, G., Moraies, R.G., Campos, O.A.J. The Namorado Oil Field: A Major Oil Discovery in the Campos Basin, Brazil. In: Giant Oil and Gas Fields of the Decade: 1968-1978, AAPG Memoir 30. Tusla: American Association of Petroleum Geologists, p. 329-338, 1980.

Barboza, E.G. Análise Estratigráfica do Campo de Namorado (Bacia de Campos) com base na Interpretação Sísmica Tridimensional. PhD Thesis, Universidade Federal do Rio Grande do Sul, 230p. 2005.

Castagna, J.P, Batzle, M.L., and Eastwood, R.L, 1985. Relationships between compressional-wave and shear-wave velocities in clastic silicate rocks. Geophysics, 50, 571-581;

Cruz M. M. 2003. Aplicação de perfilagem geofísica e sísmica na caracterização da faciologia do reservatório de Namorado. Master Dissertation, Universidade Federal Fluminense, 121p.

Dias, J.L., Scarton, J.C., Esteves, F.R., Carminatti, M., Guardado, L.R. 1990. Aspectos da evolução tectono-sedimentar e a ocorrência de hidrocarbonetos na Bacia de Campos. In: Raja Gabaglia, G.P., Milani, E.J., 1991. Origem e Evolução de Bacias Sedimentares. Petrobras. Rio de Janeiro, Brasil, p. 333-360.

Dvorkin, J., and Nur, A., 1996. Elasticity of high-porosity sandstones: Theory for two North Sea datasets. Geophysics, 61, 1363-1370;

Fonseca, P.P., 2011, Métodos Geoestatísticos de Co-Estimativas: Aplicações aos Dados do Campo Escola de Namorado. Master Dissertation. São Paulo, Brasil;

Horta, A., Soares, A., 2010. Direct Sequential Co-simulation with Joint Probability Distributions. Mathematical Geosciences (2010) 42: 269–292

Oliveira, J.K., Soares, J.A., Martins, J.L., 2005, Influência da argilosidade, porosidade efetiva e densidade nas velocidades compressionais de distintas litologias do Campo de Namorado;

Smith, T. M., Sondergeld, C. H., Rai, C. S., 2003, Gassmann fluid substitutions: A tutorial: Geophysics, 68, 430-440.

Soares, A., 2001, Direct sequential simulation and cosimulation. Mathematical Geology, Vol. 33, No. 8, November 2001