The fluid injection in sedimentary formations may generate geochemical interactions between the fluids and the rock minerals, e.g., CO 2 storage in a depleted reservoir or a saline aquifer. To simulate such reactive transfer processes, geochemical equations (equilibrium and kinetics equations) are coupled with compositional flows in porous media in order to represent, for example, precipitation/dissolution phenomena. The aim of the decoupled approach proposed consists in replacing the geochemical equilibrium solver with a substitute method to bypass the huge consuming time required to balance the geochemical system while keeping an accurate equilibrium calculation. This paper focuses on the use of artificial neural networks (ANN) to determine the geochemical equilibrium instead of solving geochemical equations system. To illustrate the proposed workflow, a 3D case study of CO 2 storage in geological formation is presented.
The emergence of liquid-rich and gas shale reservoirs presents major strategic opportunities and challenges for the oil and gas industry. Accurate estimation of Stock Tank Oil and Gas Initially In Place (STOIIP & GIIP) is one of the priority tasks before defining the reserves. An accurate method is proposed to calculate hydrocarbon volumes using high-resolution geological models taking advantage of huge improvements made during last decade in the field of characterization and geological modeling of unconventional reservoirs. This exact method provides fluids in place in reservoir and surface conditions with an extended black-oil formulation including condensates. The physic including the equilibrium between gravity and eventually capillary forces and adsorbed gas is fully respected using, for each lithofacies, the most accurate available geological description with 3D porosity distributions, Langmuir isotherms (Langmuir, 1918), capillary pressure curves, and thermodynamic data. Adsorbed and liquid-rich gases are considered. This method calculating hydrocarbons in place is the natural endpoint of any workflow devoted to the geological modeling of newly discovered reservoirs, particularly suited to heterogeneous reservoirs. The knowledge generated by this calculation has significant impact on fracturation programs to increase the recovery rate and field development planning.
In a reservoir, gravity forces lead fluids to stratify according to their density, and transitions zones appear around contacts (water-oil and gas-oil contacts). These transitions zones are gradual variations of saturation of each phase. These saturations are often generated using geostatistical methods leading to errors in the calculations of fluid in place. They must be calculated in a physical manner using capillary pressure curves associated with lithofacies. In this paper, a physical volumetric method to calculate the hydrocarbon in place with a high-resolution geological model is used on data coming from a synthetic field: the Brugge field. The uncertainties on hydrocarbons in place are evaluated using the Monte-Carlo method.
History matching of production data is a process requiring a large number of reservoir simulations that are huge time-consuming. To reduce the computation time, an approach consists of using a meta-model to replace the reservoir simulator. In this paper, a proxy model based on an artificial intelligence technique (artificial neural network) is evaluated and compared to proxy models conventionally used for history matching such as polynomials or kriging methods. The proposed approach provides accurate prediction results to speed up and improve the history matching process. An application to the Brugge field is presented.
Decisions for field development of oil and gas reservoirs are often based on uncertainties assessment on forecast productions and other variables which are highly impacted by the uncertainties on the reservoir characteristics. Using geostatistical models, it would require thousands of flow simulations of several hours each to consider the geological uncertainties. Each of these simulations would require several hours even with current high power computers. To bypass this restriction due to the computation time, one approach consists to replace the simulator by an approximation of it, also called proxy. This paper focuses on the use of Artificial Neural Networks (ANN) proposing an innovative method to build an optimal ANN.
This paper recall the Compositional Dual Mesh Method, an extension to the concept of dual mesh for reactive transport modeling. This approach involves two meshes, a low-resolution mesh to resolve the pressure equation and a high-resolution mesh to transport the species and to calculate the geochemical equilibrium. Geochemical equilibrium being very sensitive to the concentration, preserving the fine heterogeneities leads to a more accurate field behavior simulation than conventional approach which consist in performing simulations on a coarser mesh. The method is applied to a simulation of CO2 storage in a geological model representing a fluvial deposit with a complex realistic architecture that keep a high resolution of the heterogeneities.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.