Water alternative gas (WAG) optimization for a heterogeneous Brazilian pre-salt carbonate reservoir

Abstract

Water alternating gas (WAG) is a cyclical process that involves alternating water and gas injections with the primary goal to improve sweep efficiency by maintaining initial high pressure, slowing water and gas breakthrough, and lowering oil viscosity. The objective of this work is to apply and optimize a WAG strategy on a carbonate field with light oil, compare it to the initially planned water-flooding strategy, and investigate the capability of WAG to improve field production. In this research, a compositional reservoir simulator was used to model a WAG process by injecting produced gas into the reservoir, using the same well structure as an optimized water-flooding strategy. Subsequently, a WAG strategy was created, optimizing the number and locations of wells, to facilitate a comparative analysis of the two recovery methods. The WAG optimization involved a detailed assessment of variables such as bottom hole pressure (BHP), WAG cycle duration, maximum gas oil ratio (GOR), and well positioning, to achieve a high net present value (NPV). The study focuses on the application of WAG optimization modeling in unconventional reservoirs, specifically pre-salt carbonate reservoirs, and investigates its implications on production strategy and forecast, emphasizing its potential for maximizing NPV and oil recovery in a recently producing field. The results showed that WAG improved reservoir performance when compared to water injection and produced a greater amount of oil. This solution showed potential to be tested under uncertainties (reservoir heterogeneity, faults, fractures, karsts, vugs, etc.) as future steps.

1. Introduction

Carbonate reservoirs are extremely important in the world scenario, but they are also challenging because of critical heterogeneities, folds, faults, fractures, karsts, and vugs. It is known that the presence of fractures affects the dynamic behavior of fluids and, consequently, their production. For this reason, there is a need to integrate characterization studies, geological modeling, and reservoir simulation for a better understanding of these phenomena, a more reliable production forecast, and to minimize risks and improve long-term decision-making.

Brazilian pre-salt, which currently represents one of the largest oil potentials, has complex challenges due to the high degree of heterogeneities of its carbonate rocks and because of the uncertainties in acquisition and modeling.

One of the main reasons for this complexity is the intrinsic dissolution processes and diagnosis of these rocks. This dissolution generates fractures and caves. In the exploration phase of the oilfield, these cavities also greatly affect the porosity and permeability of the reservoirs, creating anisotropies in the porous medium flow and, consequently, increasing the complexity of the reservoir modeling. Karst processes also influence the flow of fluids on a regional scale, between formations, and even hydrocarbon migration. The existence of these features at the sub-seismic scale makes their occurrence unpredictable and the modeling of carbonate reservoirs complex (Formigli et al.2009).

Carbonate reservoirs are well-known for their relatively high oil production opportunities. (Beltrão et al.2009). WAG injection is the main oil recovery method applied worldwide with oil recovery incremental up to 20% (over conventional water flooding) in field-scale of WAG projects (Campozana et al. 2000). Great effort is being made to develop new technologies, methodologies, and procedures to overcome these challenges presented in this type of fractured and karstified reservoir. It is, however, still difficult to estimate an increase in oil recovery from the WAG method although many studies demonstrate the advantages of the WAG technique (Sampaio et al.2020).

The low recovery factor associated with carbonate reservoirs combined with the availability of affordable gas and has heightened interest in a WAG strategy to improve hydrocarbon recovery for reservoirs at an early stage of production. The use of the WAG method enhances the efficiency of the sweep by reducing the potential for fingering and hydrocarbon retention at both the macroscopic scale through water injection and the microscopic scale through pore-level gas injection. (Saeeda et al.2020).

A unique analysis of an offshore immiscible HC-WAG pilot in the Al-Shaheen field offshore Qatar has been presented at the SPE EOR Conference at Oil and Gas West Asia, the authors point out the limited information available on optimizing HC-WAG operations in an offshore environment. They propose a methodology for evaluating the operational strategy for an offshore HC-WAG pilot and explore different geological and petrophysical properties to optimize HC-WAG recovery. The benefits of optimizing HC-WAG operations, in terms of incremental oil recovery and managed GOR, are quantified. The authors conclude that optimizing HC-WAG operations can significantly benefit an offshore immiscible HC-WAG pilot (Pal et al.2018).

Schiozer et al. (2019) proposed a twelve-step process known as closed-loop field development and management (CLFDM), as shown in Fig. 1. Steps 6 to 11 of this process (depicted by the blue line) aim to enhance production strategy selection while minimizing computational and human effort. The CLFDM process, as defined by (Mirzaei et al.2021), involves updating an uncertain field model with the most recent measurements (data assimilation) at regular intervals, and then optimizing production to maximize the economic value of the field. The CLFDM process can be broken down into four primary phases, each represented by a different color in Fig. 1: the first phase (green line) involves collecting data, defining uncertainties, and creating the field model. The second phase (red line) entails data assimilation and history matching. The third phase (blue line) focuses on life-cycle production optimization under uncertainty. The final phase (black line) concerns real-time or short-term decisions based on data.

This research is in line with CLFDM blue area with more focus on step 6, which involves the production strategy.

The principle of WAG injection is the alternation of the injection fluids during the process. Consequently, the hysteresis of relative permeability and capillary pressure are mostly associated with this method. For this reason, when a reservoir model with WAG injection is simulated, it is recommended to implement the hysteresis phenomenon. (Ligero et al.2013).

Similar to WAG, the oil recovery from simultaneous water and gas injection (SWAG) processes is achieved through mechanisms such as miscibility, reduction of interfacial tension, gravity drainage, and control of fluid mobility (Masalmeh et al.2010; Tunio et al.2011; Heidari et al.2013). SWAG can be performed in two ways: (i) by mixing the gas and water at the wellhead and injecting it into the reservoir through highly deviated wells, or (ii) by injecting the gas and water simultaneously through various horizontal wells. This process is widely used in reservoirs in the North Sea (Awan et al.2008).

Gaspar et al. (2016) proposed a new classification system for evaluating variables when selecting a production strategy, organized into three groups based on their nature and significance: project variables (G1) from the development phase, which include factors such as the number and location of wells, platform capacity, and drilling schedule; control variables (G2) from the management phase, which further refine field operations; and revitalization variables (G3), which are relevant to later stages of the field life, such as well completion.

Huang et al. (2021) establishes a long short-term memory (LSTM) neural network model based on 15 years of data for predicting oil production, gas oil ratio (GOR), and water cut (WC) of WAG flooding. LSTM outperforms traditional reservoir numerical simulation in computational efficiency and prediction accuracy. The study classifies producers into three types and proposes optimization measures to mitigate risks and challenges. Field implementation shows promising outcomes with better reservoir support, lower GOR, lower WC, and stabler oil production, indicating a novel direction for AI in WAG flooding development (Huang et al.2021).

Mayuor Pal (2022) proposes a methodology for evaluating the immiscible hydrocarbon water alternating gas (WAG) injection process for enhanced oil recovery in a giant Middle Eastern carbonate reservoir. While WAG injection has been a proven technique for enhancing oil recovery for some time, there is limited published literature available on the complete evaluation of immiscible WAG processes in field applications. This article aims to bridge the gap by presenting a detailed study of the immiscible water alternating gas evaluation process, including the optimization of various parameters such as WAG cycle length, WAG ratio, production constraints, injected gas compositions, WAG duration, and WAG slug size. The presented methodology can help oil field operators understand the impact of different parameters on the immiscible WAG injection process and potentially optimize production from their fields.

The target of the swept area will be the bottom of the reservoir if we inject only water and, if only gas is injected, the impact will be on the top of the reservoir, but in WAG it will have an effect both on the top and bottom of the reservoir and the swept area will increase. The main purpose of WAG injection is to improve both macroscopic and microscopic sweep efficiency and the recovery factor will be increased accordingly based on equation (1).

where the recovery factor (RF) is a metric that measures the volume of oil recovered from a reservoir as a proportion of the total volume of oil initially in place (OIIP). Microscopic displacement efficiency (EPS) is a measure of the fraction of oil that is displaced from the pores when water is injected. The macroscopic sweep efficiency (ES) is a measure of the proportion of the connected reservoir volume that is swept by the injected fluid(s). This is largely influenced by the heterogeneity of rock permeability and the effects of gravity on fluid segregation. The connected volume factor (ED) represents the proportion of the total reservoir volume that is connected to the wellbores. This means that if there are sealing faults or other low-permeability barriers in the reservoir, there may be compartments of oil that are not in pressure communication with the rest of the reservoir. Last, the economic efficiency factor (EC) represents the physical and commercial constraints that impact the life of the field, such as the lifespan of facilities, the ability to handle produced gas and water, and the energy of the reservoir (i.e. if the reservoir pressure becomes too low, fluids may not be producible). (Tarek et al. 2019).

The objective of this study is to apply a WAG and optimize a recovery method that can improve the field production, instead of water flooding, on the carbonate model named C0-MFM-2018 using the base model of water flooding—G1 strategy. The research investigates the novelty of WAG optimization modeling for unconventional reservoirs, particularly pre-salt carbonate reservoirs, and explores its implications for production strategy and forecast. A new methodology was developed to optimize the WAG strategy, including the WAG injection parameters and selection of the number and locations of wells, enabling a comprehensive comparison between different recovery methods. The study focuses on a recently producing field, where WAG holds potential as a production strategy for maximizing the net present value (NPV) and oil recovery.

2. Methodology

A fractured dual-permeability model (called C0-MFM-2018) is used as a base model. It initially only uses the water-flooding wells strategy and later considers wells drilled specifically for the WAG strategy; because the WAG needs a compositional model, we first converted the black oil to a compositional reservoir model, as in Fig. 2. In this study, a compositional reservoir simulator GEM was used to generate the reservoir model. The compositional PVT model was generated by Winprop considering the three methods and steps of the component compositional, saturation pressure, and differential liberation for the conversion from black oil to compositional PVT model. Table 1 shows the gas component used for the WAG model. The WAG process model was built as a preliminary test and then a comparison between water injection and WAG was performed. Following this, a detailed WAG optimization was conducted using CMOST AI, and many variables such as BHP, WAG cycle duration, GOR, and well position were considered as well as economic parameters to achieve the high NPV as an objective function for this study. Figure 3 provides the workflow for modeling input and output.

In a nominal approach, (Gaspar et al.2016) proposed an assisted technique to optimize several variables to select a production strategy, where the most significant variables associated with infrastructure and wells are defined as (i) design variables (G1) to be determined before field development reflecting the option of configuration and equipment, which requires high investments; (ii) control variables (G2), which determine how the oilfield operates and are usually not subject to significant additional costs; and (iii) revitalization variables (G3), which are applicable in later stages such as infill drilling.

The main assumption for the WAG model is the re-injection of all produced gas. The WAG optimization process was performed as per the following steps:

• Step 1: Compare base case (water flooding vs. WAG).

• Step 2: Fix G1 – optimization of G2L variables.

• Step 3: Optimization of G1 well position.

• Step 4: Optimization of G3 well position.

• Step 5: Finally compare the results of water flooding and WAG.

The computer assisted history matching, optimization, and uncertainty assessment tool– (CMOST) was used to determine the best set of parameters for each oil production scheme compared in terms of RF, cumulative oil, or NPV. The DECE optimizer is an iterative technique that consecutively uses the designed exploration and controlled evolution (DECE) stages and it was used here to assist in the optimization of the well control parameters for each production strategy.

During the designed exploration stage, space is searched in a designed random manner to collect as much information about the solution space as possible. Experiment design strategies are used at this step to select parameter values and generate representative simulation datasets. Statistical studies of the simulation findings produced in the designed exploration stage are undertaken in the controlled evolution stage. Based on the results of the studies, the DECE algorithm analyses each candidate value of each parameter and assesses whether there is a better possibility of improving the solution quality if specific candidate values are rejected (banned) from being chosen again. The algorithm records these rejected candidate values and will not use them in the next controlled exploration step. To reduce the potential of becoming stuck in local minima, the algorithm periodically verifies rejected candidate values to ensure that earlier rejection decisions are still valid. If the algorithm concludes that certain rejection judgments are invalid, those choices are recalled and the associated candidate values are used again with a solution space of 100–500 simulations for each hysteresis parameter configuration (CMG 2021).

3. Reservoir simulation model

The reservoir's dimension of the model used in this study is 64 × 160 × 500 m and the total number of blocks is 5 120 000, including 105 726 active blocks; the model is unconventional with dual porosity and permeability. 2D relative permeability is used in the model as shown in Fig. 4 and the porosity and permeability distribution for the base model are shown in Fig. 5, which includes the parameters in Table 2.

Table 3 provides information on well, group, and platform control for the WAG optimization study. The WAG model parameters include the assumption of the limitation for the efficiency of the well, manifold, and platform based on the actual field data, the study uses the C0-MFM-2018 (RM0) model with G1 strategy, which is based on a pre-salt real field. The GEM simulator is employed for the WAG process. The field consists of eight wells, including three injectors and five producers. The simulation starts with 8 years of historical testing data before the field enters the production phase. The assumption for this study is that the model is already matched and ready for modeling and WAG optimization during the development phase. The field development process spans over 24 years, the model constraints for the history matched and WAG models can be found in (Tables 4–6).

4. Results and discussion

Preliminarily WAG model was built using the base GEM model, defining the wells, group, and platform constraints shown in Table 3 by applying WAG control parameters in Table 6, then comparing the results with the water flooding from the base model considering that the same production strategy with results shows WAG is much better compared to WI, considering the following assumption of the economic parameters:

• Oil Price (US$/m³) = 314.

• Oil production cost (US$/m³) = −35.73

• Water production cost (US$/m³) = −3.58

• Water injection cost (US$/m³) = −3.58

• Gas injection cost (US$/m³) = −0.0096

• Gas production cost (US$/m³) = −0.0103

From the field performance analysis, we can see some interruptions in the production of liquid rate and oil rate when it comes to well stoppage due to well constraint. The WAG shows a higher oil rate with lower water production compared to the water-flooding model Figs 6 and 7. For the gas production from the WAG base model, it can be observed that almost all produced gas has been recycled as gas injection as shown in Figs 8 and 9. The summary of the comparison between the WAG preliminary test and water-flooding model is given in Fig. 10, and the cumulative oil production, RF, and NPV improved from 40.47 MM m³, 12.62%, and 3.13 E + 9$, to 69.69 MM m³, 18.17%, and 4.10 E + 9$, respectively, with less water production than the WAG model. In general, the results show that the WAG from the preliminary test can increase the RF and improve the NPV, which is promising for the next step, which will be conducting a detailed WAG optimization.

4.1. WAG optimization

In this section, the results will be discussed considering using the WAG model for the preliminary test as the base model for the detailed optimization of the (BHP, WAG cycle length, and well position) variables and the assumption of the economic parameters will be the same as used before.

4.1.1. WAG cycle length optimization

The WAG base model had three injectors and the base cycle length was 6 months. To guarantee that all gas produced had been recycled, at least two injector wells were considered for gas injection and one for water injection, as provided in Table 7, which shows the WAG cycle length designed from 3–24 months. And from the optimization model results of around 100 models, as in Fig. 11, the results show that the highest NPV achieved by WAG cycle length was of 12 months.

The proxy Sobol analysis used to study the effect of the optimization variable on the objective function (NPV) found that the BHP for wells (HI1_w, HI1, Inj13_w, and IRK150) and the WAG cycle length has the highest impact on the NPV, as in Fig. 12.

The results of WAG cycle length optimization in Fig. 13 shows that the cycle length of 12 months with a WAG ratio of 2:1 achieved the highest NPV, considering that the results are very close in cycle lengths of 3, 9, and 15 months. It would be interesting to have a deeper understanding of the cycle length impact but, for this research, we will continue the next steps with the highest NPV achieved with 12 months of water and gas injection.

The optimized WAG model was selected based on the objective function of NPV and Fig. 14. The results show the range of cumulative oil ranging from 4.46E + 7 to 7.62E + 7 m³, and RF ranging from 11.63 to 19.87%.

The optimal parameters selected were compared to a WAG base case, which was adjusted using standard commercial tools based on industry best practices. The results obtained for the current reservoir demonstrate the effectiveness of the optimization algorithm applied to the WAG strategy. WAG improves the RF and increases oil production, which is crucial for enhancing the NPV according to the economic model compared to continuous water injection. By optimizing WAG cycles, it is possible to identify the best hydrocarbon production while considering the constraints imposed by the facilities and gas availability.

Comparing the pressure and oil saturation for WI and WAG at the end of the simulation, as in Figs 15 and 16, in the highlighted area the impact of WAG is very clear and the well connectivity could be the main factor for better sweep efficiency. From the comparison, the initial results of WAG optimization confirmed that using gas injection alongside the water injection can help increase sweep efficiency, which would sweep more oil compared to using only water injection.

4.1.2. WAG optimization results

The second phase of WAG optimization for the model was to do optimization of G2L (control rules and variables of WAG cycle duration, BHP) using the same G1 strategy as the base model, and the objective function was to maximize the NPV. The results are:

• WAG cycle length (12 months Gas and 12 months Water)

• HP1 (2035) and HP6 (2024) shut-in (GOR) limit

• BHP (Kpa)

• HI1 = 62149.63

• HI1_w = 65827.121

• inj13 = 55530.149

• Inj13w = 71159.482

• IRK150 = 66930.368

• IRK150_w = 72262.729

The production performance results of the WAG optimized model are shown in Fig. 17 and, from Fig. 18, The WAG achieves higher sweep efficiency, resulting in more oil produced on both the top and bottom of the reservoir, as well as an increase in swept area in line with the primary goal of WAG injection, which is to increase both the macroscopic and microscopic sweep efficiency and recovery.

WAG cycle optimization enables the best hydrocarbon production strategy under the constraints imposed by facilities and gas supply. WAG seems to be much better for water injection in terms of cumulative oil, RF, and NPV (Figs 19 and 20). When compared to the base case, the optimal field development plan increased the NVP by 63.7%.

When the GOR limit has been considered in the WAG optimization, the WAG model manages to produce up to the end of the simulation with good rates for all wells as in Figs 21–23, and the final results show an incremental of the RF up to 22.97% and NPV of 5.88E + 9$. This is a 14.92% increase compared to previous results of the WAG optimization with the limitation of GOR.

4.2. WAG optimization of G1 and G3 (well position).

In this phase, WAG optimization was done for the well position by optimizing the location of the existing wells from the

G1 strategy and considering new wells. The results show that location optimization of wells from the G1 strategy shows higher NPV compared to the option of adding new wells for the G2L and G3 strategy as in Table 8, and it is highly recommended to use the same G1 strategy with the same number of wells (five producers and three injectors): the new optimized location can be found in Table 9. The final results show that optimization of the well position of the base strategy (G1) is much better than adding new wells, which cause well interference, and this optimization impacts the field's cumulative oil production. The NPV and RF improved up to 24.31 and 6.01E + 09 after optimization of the well position.

From the detailed analysis of the well's positions and impact on the production process, we found that the optimization of the location of the well (Inj13) has a high contribution and impact on RF and NPV compared to other wells (Fig. 24).

Detailed optimization of WAG shows improvement in RF and NPV and the optimization of well location with base strategy (five producers, three injectors) achieved high RF and NPV compared to adding new wells.

The best options were compared to a WAG base case adjusted using standard commercial tools following best practices. The current reservoir results highlight key performance characteristics of the optimization algorithm used in the WAG strategy. Compared with continuous WI, WAG increases the RF and improves oil production (which is equivalent to improving the NPV according to the economic model). WAG cycle optimization enables the best hydrocarbon production strategy under the constraints imposed by facilities and gas availability.

5. Conclusions and recommendations

In this study, the application and optimization of a WAG strategy in a carbonate field with light oil demonstrated improved reservoir performance and enhanced oil recovery compared to water injection. The WAG process effectively maintained high pressure, delayed breakthrough, and reduced oil viscosity, improving sweep efficiency.

A compositional reservoir simulator (GEM) was used to generate the reservoir model, and the WAG strategy was modeled using the same well structure as the optimized water-flooding strategy. Detailed WAG optimization was conducted, considering variables such as bottom hole pressure (BHP), WAG cycle duration, maximum GOR, and well positioning, with a focus on achieving a high NPV as an objective function for this study.

The results of the WAG optimization showed a maximum NPV of 5.12E + 09$ and a RF of 19.32% with a WAG cycle length of 12 months. By considering GOR limit optimization, the RF can be further increased to 22.9%. The WAG model successfully produced oil at good rates for all wells throughout the simulation.

Further analysis revealed that implementing WAG injection with a 12-month cycle length, optimizing BHP, and strategically positioning well (Inj13) in the G1 strategy improved NPV and RF up to 6.01E + 09$ and 24.31%, respectively. The study emphasized the significant contribution and impact of optimizing the location of wells on RF and NPV.

Overall, the research demonstrated the effectiveness of WAG optimization in improving reservoir performance and maximizing oil recovery in pre-salt carbonate reservoirs. It provided valuable insights for the application of WAG strategies in similar fields, highlighting the potential for increased NPV and overall production. The study also recommended conducting a detailed optimization study, investigating various cases for WAG cycle length, water and gas injection rates, initiation time, WAG ratio, and gas slug size. Additionally, it suggested simultaneous optimization with new well locations, considering uncertainties related to reservoir heterogeneity, faults, fractures, karsts, and vugs in future studies.

Acknowledgements

The authors would like to express their gratitude to the Center for Petroleum Studies (CEPETRO-UNICAMP/Brazil), the Department of Energy (DE-FEM-UNICAMP/Brazil), and the Reservoir Management and Simulation Research Group (UNISIM-UNICAMP/Brazil) for their assistance. Special thanks also go to Shell and CMG for providing software licenses.

Conflict of interest statement. None declared.

References