Oil-water relative permeability curve calculation method based on machine learning

Abstract

The invention discloses an oil-water relative permeability curve calculation method based on machine learning. The oil-water relative permeability curve calculation method comprises the following steps: establishing a model representing characteristics of an oil-water relative permeability curve; pretreating an oil-water relative permeability curve sample; training a sample and an inspection modelby adopting a neural network Turing machine; and predicting an oil-water relative permeability curve. The oil-water relative permeability curve is obtained by utilizing a machine learning calculationmethod, the speed is high, the cost is low, dynamic and static factors such as oil-water viscosity, porosity, permeability, a pore structure, mineral components, a sedimentary facies belt and displacement conditions are considered, the method conforms to actual mine field application, and a new idea is provided for research of the oil-water relative permeability curve.

Description

technical field [0001] The invention belongs to the technical field of oilfield development data mining, and in particular relates to a method for calculating oil-water relative permeability curves based on machine learning. Background technique [0002] At present, artificial intelligence technology is developing rapidly at home and abroad, and machine learning technology, as one of the general directions, is being continuously applied to the field of oil and gas field development. The oil-water relative permeability curve is an important curve reflecting the characteristics of oil-water seepage in reservoirs, and it is an indispensable basic data for oilfield development plan design, oilfield dynamic calculation, and reservoir numerical simulation, so its research value is very great. There are two main methods for calculating oil-water relative permeability curves: [0003] First, most of the current methods for studying oil-water relative permeability curves are mostly ...

AI Technical Summary

Problems solved by technology

Judging from the data obtained from the core, it can accurately reflect the formation conditions around the bottom of the well, but due to the complexity of the reservoir, it cannot fully reflect the actual conditions of the entire block

At the same time, the experimental method to calculate the relative permeability curve takes a long time, the cost is high, and the amount of data is limited by the number of cores.

[0004] Second, the oil-water relative permeability curve obtained from the experiment has its limitations, so some people propose to use production data such as oilfield production dynamic data to calculate the oil-water relative permeability curve that can reflect the situation of the entire reservoir

Although this method can calculate the oil-water relative permeability curve, it cannot reflect the relevant factors that affect the oil-water relative permeability curve, and lacks certain basis

[0005] In summary, the current methods for studying oil-water relative permeability curves are relatively limited and have their corresponding limitations, and further research is needed; in addition, the method of combining machine learning methods to study oil-water relative permeability curves is blank

Images