Cellular Automata-Based Deduction of Remaining Oil Distribution

Abstract

AbstractThe proportion of oilfields developed by water injection is high in China. After several decades of water injection development, the distribution of reservoir fluids has become more complex, leaving a large amount of unmined remaining oil underground. The study on remaining oil distribution has important practical significance for the improvement of oil recovery and development efficiency of mature oilfields. Numerical reservoir simulation is a common method to study the distribution of remaining oil, but it is complex and needs huge amount of calculation which cause large time overhead. In this paper, we propose a method of deducing remaining oil distribution based on cellular automata that is a lightweight algorithm. A horizontal reservoir is divided into a rectangular grid. Each grid is regarded as a cell with the oil saturation as the states, and porosity, permeability, oil saturation, pressure, and other parameters as the attributes. Its 8 surrounding nodes adjacent to it are called neighbor nodes which form its neighborhood, and all the cells with their neighborhoods form the cell space. Combined with the actual development of the oil field, the deduction rules are set, and a cellular automata model for residual oil distribution is constructed. Each cell node updates its oil saturation state according to the state of its own, the states of its neighbors at the previous moment, and the deduction rule. Then the trend of oil saturation evolution on the target layer is deduced through the state updating of the cellular automaton at each moment. The dynamic evolution of remaining oil distribution can be analyzed according the deduction. Taking a certain reservoir of a block in * * oilfield as an example, we deduce oil saturation trend of target reservoir from July 1965 to May 2019, simultaneously visualizing the evolution process in real time. The accuracy of our method is similar to that of numerical simulation and the deduction time on the target layer is reduced to less than 8 min. The remaining oil distribution deduction model proposed in this paper is significantly improves the timeliness of deduction and provides timely and powerful technical support for efficient and accurate tapping potential of remaining oil in mature oilfields.KeywordsRemaining oil distributionCellular automataOil saturationDeductionNumerical simulation