Recently there has been an increasing interest in Enhanced Oil Recovery (EOR) from shale oil reservoirs, including CO2 and field gas injection. For the performance assessment and optimization of CO2 and gas injection processes, compositional simulation is a powerful and versatile tool because of the capability to incorporate reservoir heterogeneity, complex fracture geometry, multi-phase and multi-component effects in nano-porous rocks. However, flow simulation accounting for such complex physics can be computationally expensive. In particular, field scale optimization studies requiring large number of high resolution compositional simulations can be challenging and sometimes computationally prohibitive. In this paper, we present a rapid and efficient approach for optimization of CO2 and gas injection EOR in unconventional reservoirs using the Fast Marching Method (FMM)-based flow simulation.
The FMM-based simulation is analogous to streamline simulation and utilizes the concept of ‘Diffusive Time-of-Flight (DTOF)’. The DTOF is a representation of the travel time of pressure ‘front’ propagation and accounts for geological heterogeneity, well architecture and complex fracture geometry. The DTOF can be efficiently obtained by solving the ‘Eikonal equation’ using the FMM. The 3-D flow equation is then decoupled into equivalent 1-D equation using the DTOF as a spatial coordinate, leading to orders of magnitude faster computation for high-resolution and compositional models as compared to full 3-D simulations. The speed of computation enables the use of robust population-based optimization techniques such as genetic or evolutionary-based algorithm that typically require large number simulation runs to optimize the operational and process parameters.
We demonstrated the efficiency and robustness of our proposed approach using synthetic and field scale examples. We first illustrate the validation of FMM-based simulation approach using an example of CO2 Huff-n-Puff for a synthetic dual-porosity and heterogeneous model with a multi-stage hydraulically fractured well. In the field-scale application, we present an optimization of operating strategies for gas injection EOR for a depleted shale oil reservoir in the Eagle Ford formation. The rapid computation of the FMM-based approach enabled intensive simulation study involving high-resolution geological models with million cells resulting in a comprehensive evaluation of the EOR project including sensitivity studies, parameter importance analysis and optimal operating strategies.
This study shows the novelty and efficiency of the systematic optimization workflow incorporating the FMM-based compositional simulation for the field-scale modeling of CO2 and gas injection in shale oil reservoirs. Not only can it account for relevant physics such as reservoir heterogeneity, fracture geometry and fluid phase behavior but also lead to orders of magnitude saving in computational time over commercial finite difference simulators.
Reconciling geological models to the available dynamic information, commonly known as history matching, is an essential step for optimizing reservoir management and field development strategies, including improved recovery methods. There are several challenges in the current history matching workflow, particularly for high resolution geologic models with multimillion cells and complex geologic architecture. Streamline-based inverse modeling has shown great promise in this respect because of computational efficiency and analytic calculation of sensitivity of production response to reservoir properties. However, the current streamline-based approach is mostly restricted to history matching water-cut and tracer response in two-phase flow.
In this paper we present a novel approach to extend the streamline-based history matching to three-phase flow by incorporating water-cut, gas-oil ratio and bottomhole pressure data while updating high resolution geologic models. The crux of our approach lies in the analytic computation of bottomhole pressure and gas-oil ratio sensitivities which allows for efficient inversion of production and pressure data. Thus, our approach overcomes one of the major limitations of the current state-of-the-art while preserving the computational efficiency and the intuitive appeal of the streamline method. The streamline-based approach can also be used in conjunction with finite difference simulators, further generalizing its applicability to enhanced oil recovery methods. We validate the accuracy and efficiency of the streamline-based sensitivities by comparison with adjoint or numerical methods using finite-difference simulators. In history matching, we incorporate the novel streamline-based method with multiscale approach to account for the disparity in resolution of different types of history data. This method leads to capturing of the large- and fine-scale heterogeneity and reproducing the pressure and production responses efficiently.
We demonstrate the power and utility of our approach using synthetic and field applications. The synthetic example involves the SPE9 benchmark field case with waterflooding and aquifer drive. The field example involves full-field history matching of the Norne Field in the North Sea using water-cut, gas-oil ratio and bottomhole pressure data and subsequent design of a polymer flood. A novel multiscale workflow demonstrates the efficiency and advantage of our proposed approach in achieving geologically consistent history matching at the full-field level.