Proof of Concept of Sequentially Coupling a GPU-Based Reservoir Simulator with Cpu-Based Geochemical Reaction Software to Model Reactive Transport

Abstract In subsurface rocks, the brine can act as a carrier where non-hydrocarbon components, either natively present or introduced externally, are transported and possibly react with each other as well as with the porous medium. The proposed paper describes the integration of PHREEQC, a leading open-source software for 0D/1D geochemical modeling, with a commercial GPU-based reservoir simulator. This will provide the reservoir engineering community with a user-friendly tool incorporating reactive transport into reservoir studies. PHREEQC is widely used for batch geochemical modeling, thanks to its proven robustness as well as compatibility with extensive databases. We present a non-intrusive approach to enabling reactive transport in reservoir simulation by sequentially coupling a GPU-based reservoir simulator with PHREEQC. The former solves reservoir equations and the transport of chemical master species as aqueous tracers, and the latter computes solutes and rock composition at the end of each time step, based on the master species content in each cell. Then porosity, absolute and relative permeability are explicitly updated by the reservoir simulator as a function of the geochemical state. In this work, two application examples of reactive transport modeling (RTM) using the proposed approach are shown; first RTM is used to simulate in a mechanistic manner a low-salinity EOR process, and second to study injection impairment due to scale deposition. Note that chemical reactions of hydrocarbon/light components dissolved in the aqueous phase are not supported in the current implementation. First, results include an assessment of the method accuracy, as well as a discussion of practical advantages compared to alternative coupling approaches; because all relevant chemical reactions information is kept within PHREEQC datafiles, seamless integration between reservoir and geochemical studies is ensured. Second, results include a set of performance benchmarks to assess the computational cost of such explicit coupling, in particular considering that while the simulator runs on GPUs, chemical reactions are computed on CPUs. This paper shows the first steps towards the integration of RTM capabilities in a commercial GPU reservoir simulator, with the objective of developing the ability to account for advanced chemical processes required for both hydrocarbon recovery and energy transition problems, minimizing the impact on simulator infrastructure and leveraging capabilities of state-of-the-art CPU-based geochemical codes.