US 12,449,557 B2Granted

Prediction of permeability distribution of deep buried sandstones using machine learning

AssigneeSaudi Arabian Oil Company

InventorsAhmed Khidir, Ahmed Alghamdi, Ali Al-Masrahy

Life of this patent

Estimating the term…

Loading patent…

Abstract

Example computer-implemented methods, media, and systems for predicting permeability distribution of deep buried sandstones of a subsurface sandstone reservoir using machine learning. One example method includes receiving multiple wireline well log responses of multiple cored well sections of a subsurface sandstone reservoir. A respective authigenic clay type associated with each of the multiple cored well sections is received. The multiple wireline well log responses are labeled based on the respective authigenic clay type associated with each of the multiple cored well sections. A machine learning (ML) model for predicting a permeability distribution of multiple uncored well sections of the subsurface sandstone reservoir is trained based on the multiple labeled wireline well log responses of the multiple cored well sections. Multiple wireline well log responses of the multiple uncored well sections are received. The permeability distribution of the multiple uncored well sections is predicted using the trained ML model.

Claims20

Dependent claims

A computer-implemented method, comprising:

receiving a plurality of wireline well log responses of a plurality of cored well sections of a subsurface sandstone reservoir;
receiving a respective authigenic clay type associated with each of the plurality of cored well sections of the subsurface sandstone reservoir, wherein the respective authigenic clay type comprises a respective type of authigenic clay of each of the plurality of cored well sections of the subsurface sandstone reservoir;
labeling, based on the respective authigenic clay type associated with each of the plurality of cored well sections of the subsurface sandstone reservoir, the plurality of wireline well log responses of the plurality of cored well sections;
training, based on the plurality of labeled wireline well log responses of the plurality of cored well sections, a machine learning (ML) model for predicting a permeability distribution of a plurality of uncored well sections of the subsurface sandstone reservoir;
receiving a plurality of wireline well log responses of the plurality of uncored well sections of the subsurface sandstone reservoir;
predicting, using the trained ML model and based on the plurality of wireline well log responses of the plurality of uncored well sections, the permeability distribution of the plurality of uncored well sections of the subsurface sandstone reservoir;
mapping the predicted permeability distribution to a 3D sweet spot map of the subsurface sandstone reservoir;
and
identifying, using the 3D sweet spot map, one or more new wellbore locations within the subsurface sandstone reservoir.

6 more elements

A non-transitory, computer-readable medium storing one or more instructions executable by a computer system to perform operations comprising:

receiving a plurality of wireline well log responses of a plurality of cored well sections of a subsurface sandstone reservoir;
receiving a respective authigenic clay type associated with each of the plurality of cored well sections of the subsurface sandstone reservoir, wherein the respective authigenic clay type comprises a respective type of authigenic clay of each of the plurality of cored well sections of the subsurface sandstone reservoir;
labeling, based on the respective authigenic clay type associated with each of the plurality of cored well sections of the subsurface sandstone reservoir, the plurality of wireline well log responses of the plurality of cored well sections;
training, based on the plurality of labeled wireline well log responses of the plurality of cored well sections, a machine learning (ML) model for predicting a permeability distribution of a plurality of uncored well sections of the subsurface sandstone reservoir;
receiving a plurality of wireline well log responses of the plurality of uncored well sections of the subsurface sandstone reservoir;
predicting, using the trained ML model and based on the plurality of wireline well log responses of the plurality of uncored well sections, the permeability distribution of the plurality of uncored well sections of the subsurface sandstone reservoir;
mapping the predicted permeability distribution to a 3D sweet spot map of the subsurface sandstone reservoir;
and
identifying, using the 3D sweet spot map, one or more new wellbore locations within the subsurface sandstone reservoir.

6 more elements

A computer-implemented system, comprising:

one or more computers;
and
one or more computer memory devices interoperably coupled with the one or more computers and having tangible, non-transitory, machine-readable media storing one or more instructions that, when executed by the one or more computers, perform one or more operations comprising:
receiving a plurality of wireline well log responses of a plurality of cored well sections of a subsurface sandstone reservoir;
receiving a respective authigenic clay type associated with each of the plurality of cored well sections of the subsurface sandstone reservoir, wherein the respective authigenic clay type comprises a respective type of authigenic clay of each of the plurality of cored well sections of the subsurface sandstone reservoir;
labeling, based on the respective authigenic clay type associated with each of the plurality of cored well sections of the subsurface sandstone reservoir, the plurality of wireline well log responses of the plurality of cored well sections;
training, based on the plurality of labeled wireline well log responses of the plurality of cored well sections, a machine learning (ML) model for predicting a permeability distribution of a plurality of uncored well sections of the subsurface sandstone reservoir;
receiving a plurality of wireline well log responses of the plurality of uncored well sections of the subsurface sandstone reservoir;
predicting, using the trained ML model and based on the plurality of wireline well log responses of the plurality of uncored well sections, the permeability distribution of the plurality of uncored well sections of the subsurface sandstone reservoir;
mapping the predicted permeability distribution to a 3D sweet spot map of the subsurface sandstone reservoir;
and
identifying, using the 3D sweet spot map, one or more new wellbore locations within the subsurface sandstone reservoir.

9 more elements

Dependencies are inferred from the claim text.