Introduction

In recent years, CO2 injection has emerged as a significant enhanced oil recovery (EOR) technique due to the twin advantages of EOR and mitigating the impact of CO2 on climate. A cost-effective EOR can extend the production life of an oil field for several years (Muggeridge et al. 2014). In response to these reasons, the practice of CO2-EOR has increasingly attracted the policy makers and industries to implement it. Oil industries have been utilizing CO2 flooding successfully worldwide as a tertiary recovery mechanism for several years in which, CO2 is compressed and injected into the reservoir. Studies show that CO2-EOR in oil fields can improve the oil recovery significantly (Orr and Taber 1984; Bondor 1992; Akervoll and Bergmo 2010; Vuillaume et al. 2011; Dimri et al. 2012; Ganguli et al. 2014; Ganguli et al. 2016a). Nevertheless, reduction of injectivity is a serious threat to CO2 flooding and is reported in many fields (Stein et al. 1992; Rogers and Grigg 2000; Goodyear et al. 2003; Barati et al. 2016), which should be avoided by decreasing the water alternating with gas (WAG) ratio, increasing injection pressure, etc. The overall process of CO2-EOR involves efficient displacement of oil towards the production wells by overriding gas and under-riding water fronts. In practice, the CO2 is injected in the reservoir as a supercritical fluid (temp. 31.1 °C, pressure 74 bar), and hence, it can lower the viscosity of the oil and increase its mobility. Injected CO2 can displace oil either by miscible or immiscible displacement, which depends on minimum miscibility pressure (MMP). The MMP is defined as the lowest pressure at which multi-contact miscibility can be achieved. Immiscible displacement takes place at reservoir pressure below MMP, and miscible displacement takes place when reservoir pressure is above MMP. The miscible CO2-EOR works more efficiently than the immiscible one (Clark et al. 1958; Bondor 1992; Muggeridge et al. 2014). If initial reservoir pressure is less than attaining MMP by virtue of injection may affect reservoir health, hence implementation of CO2-EOR in an oil field needs a systematic approach and research.

This paper details the development of a conceptual CO2-EOR model based on Ankleshwar oil field, and a systematic study to estimate the incremental oil recovery efficiency using limited data provided by the operator. Further, we investigate the possibility of miscible displacement in Ankleshwar oil field of Cambay Basin by analysing various operational parameters.

Geology and reservoir description

The Ankleshwar oil field is situated in Cambay Basin, India, which is one of the main onshore Cenozoic oil basins of India. The field is being operated by Oil and Natural Gas Corporation of India Pvt. Ltd. (ONGC). The field was put on production on August 15, 1961, and subjected to peripheral water injection since 1966. It has produced approximately 49% of original-oil-in-place (OOIP) under natural aquifer drive and peripheral water injection (ONGC personal communication; ONGC report 2010). This is an arenaceous multi-layered reservoir structure, runs into the Gulf of Cambay in an approximately NNW-SSE direction (Fig. 1a). Age of the sediments ranges from Paleocene to recent (Mukherjee 1981; Mehdizadeh et al. 2010). Figure 1b depicts the stratigraphy of the study area, in which, the reservoir formation is of middle to upper Eocene age, comprised of thick sequence of sands (e.g., Ardol and Hazad) and shales (e.g., Telwa and Kanwa). The Telwa and Kanwa shale members are devoid of coarser clastics and act as a cap-rock to the Ardol and Hazad members, respectively. In total, the Eocene sandstones broadly divided into 11 layers (S1–S11), constitute the reservoir, where S1 to S5 layers represent the middle sand group and S6 to S11 represent the upper sand group (Srivastava et al. 2015). The potential layers identified for CO2-EOR are S3 and S4 layers, the most productive sand layers, were clubbed together in the simulation model as S3 + 4 (Ganguli et al. 2016b). It is noteworthy to mention that S3 + 4 layers are not continuous throughout the reservoir and some pinch-outs were observed. These discontinuities cause production challenges. The formation thickness of the target layers (S3 + 4) is around 30 m. The oil-water-contact (OWC) varies between 1190 to 1214 m, and gas oil contact (GOC) is around 1050 m. For conventional hydrostatic equilibrium, the datum was fixed at 1113 m with initial pressure of 115.5 bar.

Fig. 1
figure 1

a Location map of the main oil and gas fields in the Cambay Basin, where the study area is marked by red ellipse. b Schematic distribution of the litho-stratigraphy of the study area along with the trajectory of the feasible CO2 injection well within the Ankleshwar formation

Full size image

Laboratory studies for CO2-EOR

Before going for a pilot study, the operator carried out laboratory experiments using different injection fluids such as CO2, N2, and hydrocarbon gas (HC) to evaluate the feasibility of tertiary gas injection in sand S3 + 4 for EOR. Berea cylindrical core with oil sample from the Ankleshwar reservoir have been used to evaluate the potential of various fluid (i.e., CO2, N2, HC gas, etc.) injection in mobilizing the residual oil within water flooded sand unit of S3 + 4. The injection rate for all the fluids was set to 1 cm3/h (ONGC Pvt. Ltd., personal communication). CO2 injection resulted in an incremental oil recovery of approximately 11.8% of hydrocarbon pore volume (HCPV) over water flooding as compared to N2 and HC gas injection, which were contributed to the oil recovery of about 4.8 and 4.0% of HCPV, respectively.

Estimation of MMP

Slim-tube simulations were performed by using Eclipse-300 software to estimate the MMP between the recombined Ankleshwar oil and pure CO2, and mixtures of CO2 and intermediate hydrocarbon gas components, all at reservoir temperature and pressure. We identified that the MMP is around 134 bar (Fig. 2a), suggesting that Ankleshwar oil is not miscible with pure CO2 at the reservoir temperature and pressure (102 bar, and 78 °C). Thus, to lower the MMP, we developed a new injection fluid composition, consisting of 40% mole volume of CO2, 10% methane, 20% ethane, 20% propane, and 10% butane. The combined condensing and vaporizing drive mechanism was used in MMP calculation at 78 °C. The slim-tube simulation results show that MMP was reduced to 93 bar from 134 bar by using this injection fluid (Fig. 2b).

Fig. 2
figure 2

The estimated MMP at reservoir conditions a before introducing the new injection fluid and b after introducing the new injection fluid composition, consisting of 40% mole volume of CO2, 10% methane, 20% ethane, 20% propane, and 10% butane, respectively

Full size image

Conceptual CO2-EOR model

To study the field in detail, we developed a conceptual CO2-EOR model, inspired by the generic sandstone reservoir (Fig. 3). The model consists of 38 × 34 × 23 cells representing six sand layers of Hazad and Ardol formation and five shale layers alternatively within the sands representing Telwa and Kanwa formations. The reservoir model is penetrated by two wells, one injector (I1) and one producer (P1). The production well is located in the up dip direction or at the crest of the model, while the injector is located in the down-flank. The average distance between the wells is around 920 m. The depth of reservoir model extends from 1075 to 1265 m. The reservoir parameters were used for the simulations as shown in Table 1 (ONGC report 2010). To conduct the simulations, we considered that the reservoir contains under-saturated oil (i.e., no gas-cap condition) with oil API gravity of 47. Aquifer lying below the reservoir has provided a strong pressure support for oil production.

Fig. 3
figure 3

Geometry, grid, depth, and well positions of the 3D conceptual model for CO2-EOR in the Cambay Basin. Color bar indicates depth; the model is exaggerated by a factor 7.5 in the vertical direction

Full size image
Table 1 Reservoir properties of the Ankleshwar oil field, Cambay Basin
Full size table

Petrophysical properties

Petrophysical properties such as porosity, permeability, rock compressibility, etc. populated in the conceptual model were provided by the operator. The total thickness of the reservoir is 26 ± 1.5 m. The S3 + 4 layers have average porosity of 23% and permeability of 1000 mD, respectively. The porosity and permeability assigned to the individual sublayers of the conceptual EOR model are summarized in Table 2 (ONGC report 2010). The shale layers were assigned 100% water saturation with negligible permeability. The rock compressibility of 2.167e−5 psi−1 was considered for all the simulations.

Table 2 Reservoir rock properties for the sublayers of S3 + 4 sands, the major pay zone of Ankleshwar oil field
Full size table

Fluid properties and flow parameters

As per the information obtained from the operator of this mature hydrocarbon field, the live oil viscosity and oil-formation-volume factor (Bo) as a function of increasing pressure are summarized in Table 3 (ONGC Pvt. Ltd., personal communication). Capillary pressure is assumed to be neglected since no reliable measured data were provided to us. The relative permeability functions were derived by using Corey relative permeability correlations, and the water-oil and gas-oil relative permeability curves are illustrated in Fig. 4.

Table 3 Live oil properties as a function of pressure, which were used for reservoir simulations (ONGC, personal communication)
Full size table
Fig. 4
figure 4

a Drainage oil/water relative permeability curves as a function of water saturation. b Drainage gas/oil relative permeability curves as a function of gas saturation

Full size image

Selection of optimum operational parameters

Before doing reservoir simulations, we carried out sensitivity analysis of various reservoir parameters to understand the reservoir performance under CO2-EOR. The simulations were carried out by using commercial software such as Eclipse 300 (E-300) and E-100. It is well known that E 300 is more reliable as it honors compositional oil, but it takes more computational time. Hence, we also made an attempt to recommend a faster black oil model in E-100, which can be comparable with the compositional model.

Miscible CO2 injection was assumed and CO2 injection rate was controlled by the production rate target. The bottom-hole-pressure (BHP) of the producer was maintained at 102.9 bar, which is above the bubble point pressure. To mimic the reservoir conditions, the conceptual model was subjected to water flooding for about 50 years followed by continuous gas injection for next 30 years.

Grid sensitivity analysis

We know in simulations that there is a tradeoff between computational time and accuracy. In general, very fine grid simulations are more accurate than coarser ones, but computational time is more for fine grid models. Hence, to select optimum grid size for simulation, we consider four grid sizes, ranging from very fine scale, viz., 12.5 m (155 × 140 × 23) to coarse scale, i.e., 100 m (19 × 17 × 23). The petrophysical properties like permeability and porosity were upscaled accordingly for each grid resolution. Our aim was to recommend an optimum grid size for the model, which can adequately represent the reservoir geometry and correctly describe the reservoir behavior with a good agreement with the fine scale compositional oil model.

The field oil production rate (FOPR) for various grid sizes is shown in Fig. 5a. The production curve for 50 m grid black oil model (blue line) is comparable with the production curve for 25 m grid compositional oil model. It is also seen that after the gas-breakthrough (2016), the oil production increases rapidly for 12.5 m grid model (green line) and 25 m grid size grid model (red line), but these models are very expensive in terms of the computational cost. The field oil recovery efficiency (FOE) also follows the trend similar to oil production rate as shown in Fig. 5b. We found that simulations results obtained by using 50 m grid black oil model are in good agreement with the 25 m compositional model (Fig. 6), which is widely accepted for its accuracy. Therefore, we selected 50 m grid black oil model for further analysis.

Fig. 5
figure 5

The grid resolution sensitivity plots for the conceptual CO2-EOR model of the Cambay Basin: a the field oil production rate as a function of time, b field oil recovery efficiency. Legend represents the different values of grid size in x and y directions, where the green curve, red curve, blue curve, and black curve represent 12.5, 25, 50, and 100 m grid resolution, respectively

Full size image
Fig. 6
figure 6

Geomodel validation including the field oil recovery efficiency for fine grid compositional (25 m) and medium grid black oil simulation model (50 m)

Full size image

Corey exponent for oil and water

Relative permeability, between constrained endpoints, is controlled by the Corey exponents, Nw (water) and No (oil). In general, Corey’s exponents are obtained from relative permeability curves generated by using laboratory studies. In case of nonavailability of laboratory data, two-phase relative permeability curves can be generated by using empirical correlations (Corey 1954; Stone 1970; Sigmund and McCaffery 1979). For unconsolidated sands, oil-water Corey exponents of 3.0 and 3.5 have been proposed in literature (Honarpour et al. 19867). It is noteworthy that lower Corey exponent values result in more concave relative permeability curve, thus lower relative permeability, indicating more sand heterogeneity, while higher exponent values result in comparatively a less concave curve, indicating more homogeneous sand (Kevin 2002). Corey’s exponents are reservoir specific; hence, its value must be adjusted based on simulation results. To analyze the effect of Corey water exponent (Nw) and oil exponent (No) on reservoir performance, we selected values of Nw and No typically as 3, 4, and 5 in a consistent manner by keeping one fixed at a time, which covers wide range of heterogeneity of the sand layers. Hence, we assumed that the wetting phase is water and nonwetting phase is oil.

We observed that the field-oil-production rate (FOPR) and field-oil-recovery-efficiency (FOE) decreases drastically with the increase of value of No (Fig. 7a, b). These results are reasonable as previous studies suggest that oil permeability and recovery decrease with an increase in No (Corey 1954). However, an opposite scenario is seen for Nw in this case. We observed increase in FOPT and FOE with increase in the exponent (Fig. 7c, d). We selected the values of No = 3 (red solid line) and Nw = 5 (green curve) as for these values reservoir performance was better.

Fig. 7
figure 7

The impact of different Corey exponents for oil (No) and water (Nw) on the field oil production rate and the field oil recovery efficiency. Red curve, blue curve, and green curve represent the value of Corey exponents for oil and water as 3, 4, and 5, respectively

Full size image

Todd-Longstaff (T&L) parameters on reservoir response

Todd and Longstaff (1972) have proposed an empirical mixing parameter (ω), known as T&L mixing parameters, particularly for viscosity and density calculations to define the effective properties during miscible displacement. These parameters are generally used for field-scale miscible flood simulations, particularly, for CO2 flooding in the reservoir. Use of these parameters can circumvent intensive computations for the compositional simulation, without compromising the accuracy level. The values of ω lie between 0 and 1 and control the degree of injected fluid mixing within each grid cell. The value ω = 1 suggests that the fluids are miscible in each grid cell and if ω = 0, the fluids are immiscible (Todd and Longstaff 1972). These parameters are also adjusted on the basis of simulation results. Thus, to analyze all possible scenarios, we considered different combinations of values of ω for viscosity and density computations, which are tabulated in Table 4.

Table 4 Different T&L mixing parameters (ω) used to calculate the viscosity and density for miscible CO2 flooding
Full size table

Figure 8 depicts the sensitivity of T&L mixing parameters on reservoir performance. We observed promising results for “ω” = 1 and 0.67 for viscosity and density computations, respectively (brown curve). However, for ω = 1 and 0.33, the field oil production peaked during gas injection period (pink curve), but the field-oil-recovery-efficiency curve was not satisfactory. Hence, the optimum value of ω for viscosity and density computation were selected as 1 and 0.67, respectively. Miscible CO2 injection can be possible by considering the optimum T&L parameters suggested by this sensitivity analysis. This allows more injection of gas into the reservoir, and hence results in incremental oil recovery.

Fig. 8
figure 8

Effect of various T&L parameters on a the field oil production rate and b the field oil recovery efficiency for the conceptual CO2-EOR model. Color bar represents the different combination of the T&L mixing parameter for viscosity and density calculations

Full size image

Estimation of CO2-EOR potential

After performing the simulations of the 3D conceptual model, we estimated that about 10.4% of additional oil recovery can be achieved from this field as a result of CO2-EOR. This has been validated from Fig. 9, which illustrates the difference in recovery due to two different injection schemes, continuous water injection (blue solid curve) and continuous CO2 injection (blue dash-dot curve).

Fig. 9
figure 9

Quantitative estimation of CO2-EOR potential for Ankleshwar oil field in Cambay Basin, India. The field oil production rate (magenta curve) and oil recovery efficiency (blue curve) has been plotted as a function of time. The solid line represents the results from continuous CO2 flooding, while dashed-dotted line represents continuous water flooding. The difference in results from continuous CO2 flooding and water flooding helps to estimate the incremental oil recovery from this field

Full size image

CO2 distribution in the reservoir

Once the operational parameters are adjusted, we carried out simulations using E-100 (black oil simulator). The conceptual model was subjected to water flooding for about 50 years followed by continuous gas injection for next 30 years. Changes in the lateral spreading of CO2 with time in the reservoir can provide qualitative insights into the plume dynamics. Simulation results indicate patchy CO2 distribution, with highest saturation in the top-most layer of reservoir (Fig. 10). The saturation of reservoir fluids at different stages, i.e., from the beginning to till the end of the CO2 injection period is shown in Fig. 11. This study reveals that the oil saturation is comparatively less near the high gas saturated zones, which suggests that CO2 has successfully pushed the residual oil towards the production well and resulted in incremental oil recovery. Results from the simulation not only demarcated reservoir areas with high oil saturation but also revealed that the mobility ratio needs to be improved for better incremental oil recovery. The problem of unfavorable mobility ratio is quite common with CO2 flooding, leading to poor sweep efficiency and low oil recovery due to viscous fingering. This type of issue has been well taken by using polyelectrolytes and polyelectrolyte complex nanoparticles in addition to CO2 foam (Kalyanaraman et al. 2015; Kalyanaraman et al. 2016). This type of information can be useful for the production engineers to plan the drilling strategy for optimum tertiary oil recovery.

Fig. 10
figure 10

Time lapse CO2 saturation in the reservoir as a consequence of CO2 flooding in the reservoir for EOR. The color bar represents the CO2 saturation where red and pink represent maximum and minimum CO2 saturation, respectively

Full size image
Fig. 11
figure 11

Time lapse ternary diagram of saturation of reservoir fluids at different time scales due to CO2 flooding in the reservoir: a after gas breakthrough, b after 10 years of CO2 injection, c after 20 years of CO2 injection, and d at the end of CO2 injection. The color bar represents the saturation of various reservoir fluids, where red, green, and blue represent CO2 saturation, oil saturation, and water saturation, respectively

Full size image

Conclusions

The present study shows interesting and optimistic results for the possible CO2-EOR in Ankleshwar oil field situated in Cambay Basin, India. In this work, a dynamic 3D model is developed to simulate the behavior of the CO2-EOR process over the time. In order to study the influence of various reservoir parameters on reservoir performance, the sensitivity analysis of various reservoir parameters is performed, and optimum reservoir parameters are recommended for improved CO2-EOR for this mature field.

We propose a 50-m grid size (horizontal x and y directions) black oil model with the optimized parameters for industrial-scale simulations. This model is in good agreement with the fine-scale (25 m grid) compositional simulation model of high accuracy. Sensitivity studies on Corey exponent for oil (No) and water (Nw) were performed, and we found that the reservoir responded very well for No = 3 and Nw = 5, which are recommended for conducting further numerical analysis on improved oil recovery in this mature field. For miscible displacement, we propose the optimum values of T&L mixing parameter for viscosity and density calculations should be 1 and 0.67, respectively. We also synthesized a new injection fluid, which can reduce the MMP for miscible, and more efficient displacement of CO2. Thus, we can conclude that this reservoir has CO2 EOR potential, but keeping in mind the age of platform, the operator should evaluate the proposal very carefully before initializing a pilot study. Moreover, the present estimation of CO2-EOR potential were made possible by considering zero capillary pressure and the quantification of tertiary oil production will differ if capillary pressure from reliable source is considered, which is beyond the scope of the present study.