Granted Patent B2

US 12,536,431 · App. 17/547,112 · Granted Jan 27, 2026

Inventors: Chicheng Xu (Houston, TX); Tao Lin (Katy, TX); Lei Fu (Houston, TX); Weichang Li (Katy, TX); Yaser Alzayer (Dhahran, SA)

Assignee: Saudi Arabian Oil Company

G06N3/08G06F16/9035G06F16/909G06F18/211G06N3/045

Abstract

Claims (58)

1 . A computer-implemented method of managing training wells for a target well, the computer-implemented method comprising:

for each training well of a plurality of training wells,

building a training network for the training well based on well log data of the training well;

generating a predicted target well log of the target well using the training network built for the training well;

determining a relevancy level between the training well and the target well based on the predicted target well log of the target well and a measured well log of the training well;

selecting relevant training wells among the plurality of training wells based on the relevancy levels associated with the plurality of training wells;

determining whether the selected relevant training wells satisfy a coverage criteria for the target well; and

selecting a drilling operation based on well attributes associated with the relevant training wells satisfying the coverage criteria for the target well.

2 . The computer-implemented method of claim 1 , wherein determining the relevancy level between the training well and the target well comprises:

calculating a correlation coefficient between the predicted target well log of the target well and the measured target well log of the target well, and

determining the calculated correlation coefficient to be the relevancy level between the training well and the target well.

3 . The computer-implemented method of claim 1 , wherein selecting the relevant training wells among the plurality of training wells based on the relevancy levels associated with the plurality of training wells comprises:

comparing the relevancy levels associated with the plurality of training wells with a predetermined relevancy threshold; and

selecting, among the plurality of training wells, training wells having corresponding relevancy levels greater than or equal to the predetermined relevancy threshold to be the relevant training wells for the target well.

4 . The computer-implemented method of claim 1 , wherein building the training network for the training well based on the well log data of the training well comprises:

training the training network using multiple input well logs of the training well as input parameters and an output well log of the training well as an output parameter, the well log data of the training well comprising the multiple input well logs and the output well log.

5 . The computer-implemented method of claim 4 , wherein the multiple input well logs of the training well comprise two or more of a list of well logs comprising permeability, porosity, oil saturation, water saturation, lithology, matrix density, and clay content, and

wherein the output well log of the training well comprises at least one of a list of well logs comprising bulk density, resistivity, velocity, gamma ray, deep induction, neutron porosity, and density porosity.

6 . The computer-implemented method of claim 4 , wherein predicting the target well log of the target well using the training network for the training well comprises:

providing multiple well logs of the target well as inputs of the training network, the multiple well logs of the target well corresponding to the multiple input well logs of the training well, and

obtaining an output of the respective training network based on the multiple well logs of the target well as the predicted target well log of the target well.

7 . The computer-implemented method of claim 1 , wherein the training network comprises a single layer neural network having an input layer, a hidden layer, and an output layer.

8 . The computer-implemented method of claim 1 , further comprising:

obtaining the plurality of training wells by filtering a multi-well database storing well data of multiple wells.

9 . The computer-implemented method of claim 8 , wherein filtering the multi-well database comprises:

selecting wells located within a predetermined proximity of the target well in a field, wherein the plurality of training wells are within the selected wells.

10 . The computer-implemented method of claim 8 , wherein filtering the multi-well database comprises:

selecting wells for the target well based on stratigraphic zonation, wherein the selected wells have a common set of geological properties with the target well in multiple zones of a field, and wherein the plurality of training wells are within the selected wells.

11 . The computer-implemented method of claim 8 , wherein filtering the multi-well database comprises:

selecting wells for the target well based on one or more operational settings, wherein the plurality of training wells are within the selected wells.

12 . The computer-implemented method of claim 11 , wherein the one or more operational settings comprise well type, drilling mud properties, and logging survey types.

13 . The computer-implemented method of claim 8 , wherein the well data of the multiple wells comprises at least one of well attributes, well logs, or core data in a field where the multiple wells are located.

14 . The computer-implemented method of claim 1 , wherein determining whether the selected relevant training wells satisfy a coverage criteria for the target well comprises:

obtaining, for each of a plurality of pairs of well log samples in the target well and the selected relevant training wells, a specified norm of distance between a well log sample in the target well and corresponding well log samples in the selected relevant training wells,

averaging the specified norms of distance of the plurality of pairs of well log samples in the target well and the selected relevant training wells to obtain an average coverage, and

determining whether the average coverage exceeds a predetermined coverage threshold.

15 . The computer-implemented method of claim 1 , further comprising:

in response to determining that the selected relevant training wells satisfies the coverage criteria for the target well, providing the selected relevant training wells as a training set of an artificial intelligence (AI) network, wherein the AI network is configured to be trained using the training set based on at least one machine learning algorithm.

16 . The computer-implemented method of claim 1 , wherein the AI network comprises a capsule convolutional neural network and a deep belief neural network that are interconnected with each other.

17 . The computer-implemented method of claim 1 , wherein the target well is a well to be drilled, and wherein the target well and the plurality of training wells are within a same reservoir.

18 . A computing system comprising:

at least one processor; and

at least one non-transitory machine readable storage medium coupled to the at least one processor having machine-executable instructions stored thereon that, when executed by the at least one processor, cause the at least one processor to perform operations comprising:

for each training well of a plurality of training wells,

building a training network for the training well based on well log data of the training well;

generating a predicted target well log of a target well using the training network built for the training well;

determining a relevancy level between the training well and the target well based on the predicted target well log of the target well and a measured well log of the training well;

selecting relevant training wells among the plurality of training wells based on the relevancy levels associated with the plurality of training wells;

determining whether the selected relevant training wells satisfy a coverage criteria for the target well; and

selecting a drilling operation based on well attributes associated with the relevant training wells satisfying the coverage criteria for the target well.

19 . A non-transitory machine readable storage medium coupled to at least one processor having machine-executable instructions stored thereon that, when executed by the at least one processor, cause the at least one processor to perform operations comprising:

for each training well of a plurality of training wells,

building a training network for the training well based on well log data of the training well;

generating a predicted target well log of a target well using the training network built for the training well;

a relevancy level between the training well and the target well based on the predicted target well log of the target well and a measured well log of the training well;

selecting relevant training wells among the plurality of training wells based on the relevancy levels associated with the plurality of training wells;

determining whether the selected relevant training wells satisfy a coverage criteria for the target well; and

selecting a drilling operation based on well attributes associated with the relevant training wells satisfying the coverage criteria for the target well.

Assignments (4)

Continuity (1)

References Cited (192)