Water alternating CO₂ gas injection (WAG CO₂) is one of the most promising enhanced oil recovery techniques. The optimization of this process requires performing many time‐consuming simulations. In this paper, an intelligent hybridization based on support vector regression (SVR) and genetic algorithm (GA) is introduced for the WAG process optimization in the presence of time‐dependent constraints. Multiple SVRs are used as dynamic proxy to mimic numerical simulator behavior in real time. Latin hypercube design (LHD) is applied to generate the proper runs to train the proxy and ten supplementary runs are randomly chosen to validate it. The goal of GA in this study is twofold. First, it is employed during the training of multiple SVRs to find their appropriate hyper‐parameters. Second, once the training and validation of the dynamic proxy are done, the GA is coupled with it to find the optimum WAG parameters which maximize field oil production total (FOPT) subject to time‐dependent water‐cut constraint and some domain constraints. The task is formulated as a non‐linear constrained optimization problem. A semi‐synthetic WAG CO₂ case is used to examine the reliability of the approach. The results show that the established dynamic proxy is fast and accurate in reproducing the simulator outputs. The hybridization proxy‐GA is demonstrated to be reliable for the real‐time optimization of the formulated WAG process. © 2020 The Authors. Greenhouse Gases: Science and Technology published by Society of Chemical Industry and John Wiley & Sons, Ltd.

Waterflooding is the most widely applied improved oil recovery technique. Recently, there has been growing interest in the chemistry and ionic composition of the injected water. Low-salinity waterflooding (LSWF) is a relatively recent enhanced oil recovery technique that has the ability to alter the crude oil/brine/rock interactions and improve oil recovery in both clastics and carbonates. In this paper, the increase in the recovery factor during LSWF was modeled based on the exchange of divalent cations (Ca²⁺ and Mg²⁺) between the aqueous phase and the carbonate rock surface. Numerical simulations were performed using laboratory coreflood data, and oil recovery and pressure drop from experimental works were successfully history matched. The ion exchange equivalent fractions, effluent ions concentrations, changes in mineral moles, and pH have also been examined. Besides, an investigation of multi-component ionic exchange as a mechanism responsible for wettability alteration during LSWF in heterogeneous low-permeability carbonate cores is presented. The results show that wettability alteration is responsible for the increase in oil recovery during LSWF, as reflected by the shift in the crossover points of the relative permeability curves. A sensitivity study done on many key parameters (e.g., timing of LSWF injection, injection rate and temperature) and the mechanistic modeling method revealed that they all have huge effects on the process.

Optimization studies are an important task in reservoir engineering practices such as production optimization and EOR (Enhanced Oil Recovery) assessments. However, they are extensive studies with many simulations that require huge computational effort and resources. In terms of EOR, CO₂ injection is one of the most common methods employed due to a high recovery potential and environmental benefits. To assess the feasibility of CO₂-EOR projects, a reservoir design study must be conducted before optimization is performed. Some studies have demonstrated the advantages of employing proxy models to perform this task in terms of saving huge amounts of computer memory space and time. In this study, proxy models were developed to solve a multi-objective optimization problem using NSGA-II (Non-dominated Sorting Genetic Algorithm II) in two selected reservoir models. The study was performed for a CO₂-WAG (Water Alternating Gas) application, where gas and water injection rates and half-cycle lengths were assessed to maximize the oil recovery and CO₂ stored in the reservoir. One model represents a simple geological model (the Egg Model), while the other represents a complex model (the Gullfaks Model). In this study, the good performance of the proxy models generated accurate results that could be improved by increasing the amount of sampling and segmenting the behavior of the reservoir model (depending on the complexity of the reservoir model). The developed proxies have an average error of less than 2% (compared with simulation results) and are concluded to be robust based on the blind test results. It has also been found that to reach the maximum oil recovery using CO₂-WAG, the maximum gas injection rate with the minimum water injection rate is required. However, this configuration may result in a reduction in the total CO₂ stored in the reservoir.

Machine learning-based Surrogate Reservoir Models (SRMs) can replace/augment multi-physics numerical simulations by replicating the reservoir simulation results with reduced computational effort while maintaining accuracy compared with numerical simulations. This research will demonstrate SRMs’ potential in long-term simulations and optimization of geological carbon storage in a real-world geological setting and address challenges in big data curation and model training. The present study focuses on CO₂ storage in the Smeaheia saline aquifer. Two SRMs were created using Deep Neural Networks (DNNs) to predict CO₂ saturation and pressure over all grid blocks for 50 years. 18 million samples and 31 features, including reservoir static and dynamic properties, build the input data. Models comprise 3–5 hidden layers with 128–512 units apiece. SRMs showed a runtime improvement of 300 times and an accuracy of 99% compared to the 3D numerical simulator. The genetic algorithm was then employed to determine the optimal rate and duration of CO₂ injection, which maximizes the volume of injected CO₂ while ensuring storage operations’ safety through constraints. The optimization continued for the reproduction of 100 generations, each containing 100 individuals, without any hyperparameter tuning. Finally, the optimization results confirm the significant potential of Smeaheia for storing 170 Mt CO₂.

Waterflooding is one of the methods used for increased hydrocarbon production. Waterflooding optimization can be computationally prohibitive if the reservoir model or the optimization problem is complex. Hence, proxy modeling can yield a faster solution than numerical reservoir simulation. This fast solution provides insights to better formulate field development plans. Due to technological advancements, machine learning increasingly contributes to the designing and building of proxy models. Thus, in this work, we have proposed the application of the two-stage proxy modeling, namely global and local components, to generate useful insights. We have established global proxy models and coupled them with optimization algorithms to produce a new database. In this paper, the machine learning technique used is a multilayer perceptron. The optimization algorithms comprise the Genetic Algorithm and the Particle Swarm Optimization. We then implemented the newly generated database to build local proxy models to yield solutions that are close to the “ground truth”. The results obtained demonstrate that conducting global and local proxy modeling can produce results with acceptable accuracy. For the optimized rate profiles, the R² metric overall exceeds 0.96. The range of Absolute Percentage Error of the local proxy models generally reduces to 0–3% as compared to the global proxy models which has a 0–5% error range. We achieved a reduction in computational time by six times as compared with optimization by only using a numerical reservoir simulator.

This work uses the MATLAB Reservoir Simulation Toolbox (MRST) to reduce the 3D reservoir model into a 2D version in order to investigate CO₂ storage in the Aurora model using the vertical equilibrium (VE) model. For this purpose, we used an open-source reservoir simulator, MATLAB Reservoir Simulation Toolbox (MRST). MRST is an open-source reservoir simulator, with supplementary modules added to enhance its versatility in addition to a core set of procedures. A fully implicit discretization is used in the numerical formulation of MRST-co2lab enabling the integration of simulators with vertical equilibrium (VE) models to create hybrid models. This model is then compared with the Eclipse model in terms of properties and simulation results. The relative permeability of water and gas can be compared to verify that the model fits the original Eclipse model. Comparing the fluid viscosities used in MRST and Eclipse also reveals comparable tendencies. However, reservoir heterogeneity is the reason for variations in CO₂ plume morphologies. The upper layers of the Eclipse model have lower permeability than the averaged MRST model, which has a substantial impact on CO₂ transport. According to the study, after 530 years, about 17 MT of CO₂ might be stored, whereas 28 MT might escape the reservoir, since after 530 years CO₂ plume reaches completely the open northern boundary. Additionally, a sensitivity analysis study has been conducted on permeability, porosity, residual gas saturation, rock compressibility, and relative permeability curves which are the five uncertain factors in this model. Although plume migration is highly sensitive to permeability, porosity, and rock compressibility variation, it shows a slight change with residual gas saturation and relative permeability curve in this study.