Modeling passage of a tool through a well

Patent metadata

Field	Value
Publication number	US-9507754-B2
Application number	US-201213677091-A
Country	US
Kind code	B2
Filing date	Nov 14, 2012
Priority date	Nov 15, 2011
Publication date	Nov 29, 2016
Grant date	Nov 29, 2016

Abstract

In modeling passage of an elongate well tool through an interval of a well an adaptive machine learning model executed on a computing system receives a first set of inputs representing a plurality of characteristics of the well tool and a second set of inputs representing a plurality of characteristics of the well. The adaptive machine learning model also receives historical data representing a plurality of other well tools passed through a plurality of other wells and a plurality of characteristics of the other well tools and the other wells. The adaptive machine learning model matches the historical data with at least a portion of the first and second sets of inputs, and determines, based on the matching whether the well tool can pass through the interval of the well.

First claim

What is claimed is:

1. A computer-implemented method for modeling passage of an elongate well tool through an interval of a well, the method comprising:
receiving, with an adaptive machine learning model executed on a computing system, a first set of inputs representing a plurality of characteristics of the well tool and a second set of inputs representing a plurality of characteristics of the well, the well comprising a wellbore formed from the Earth's surface through one or more geologic formations to a subterranean zone, and the well tool positioned on a downhole tool string residing in the wellbore;
receiving, with the adaptive machine learning model, historical data representing a plurality of other well tools passed through a plurality of other wells and a plurality of characteristics of the other well tools and the other wells;
matching, with the adaptive machine learning model, the historical data with at least a portion of the first and second sets of inputs;
determining, with the adaptive machine learning model, whether the well tool can pass through an open space proximate the interval of the well based on matching the historical data with the portion of the first and second sets of inputs;
receiving, in real time as the well tool is conveyed through the wellbore, a determination from a mathematical model of whether the well tool can pass through the open space proximate the interval of the well, the mathematical model operating separately and in parallel to the adaptive machine learning model, such that the determination from the mathematical model is independent of the determination from the adaptive machine learning model;
comparing the real-time mathematical model determination of whether the well tool can pass through the open space proximate the interval of the well with the adaptive machine learning model determination of whether the well tool can pass through the open space proximate the interval of the well; and
based on the comparison, outputting a probability value of whether the well tool can pass through the open space proximate the interval of the well.

2. The computer-implemented method of claim 1 , further comprising:
determining, with the adaptive machine learning model, a predicted reaction force on a portion of the well tool due to contact between a surface associated with the well tool and a surface of the well.

3. The computer-implemented method of claim 1 , where the adaptive machine learning model comprises a neural network.

4. The computer-implemented method of claim 1 , further comprising:
receiving, with the adaptive machine learning model, an input representing a specified failure force of the well tool;
receiving, with the adaptive machine learning model, historical data representing a plurality of forces applied to the plurality of other well tools passed through the plurality of other wells;
matching, with the adaptive machine learning model, the historical data representing the plurality of forces and the input representing the specified failure force of the well tool; and
determining, with the adaptive machine learning model, whether the well tool can pass through the interval of the well based on matching the historical data representing the plurality of forces and the input representing the specified failure force of the well tool.

5. The computer-implemented method of claim 1 , where the well tool is a first well tool, the method further comprising:
receiving, with an adaptive machine learning model executed on a computing system, a third set of inputs representing a plurality of characteristics of a second well tool;
matching, with the adaptive machine learning model, the historical data with at least a portion of the third and second sets of inputs; and
determining, with the adaptive machine learning model, whether the second well tool can pass through the interval of the well based on matching the historical data with the portion of the third and second sets of inputs.

6. The computer-implemented method of claim 1 , further comprising:
determining, with the adaptive machine learning model, a first probability of whether the first well tool can pass through the interval of the well based on matching the historical data with the portion of the first and second sets of inputs;
determining, with the adaptive machine learning model, a second probability of whether the second well tool can pass through the interval of the well based on matching the historical data with the portion of the third and second sets of inputs; and
suggesting, with the adaptive machine learning model, one of the first or second well tools based on a greater of the first and second probabilities.

7. The computer-implemented method of claim 1 , further comprising:
receiving a determination from a 3D geometric model of whether the well tool can pass through the open space proximate the interval of the well, the determination from the 3D geometric model independent of the determination from the adaptive machine learning model and the mathematical model; and
comparing the determination from the 3D geometric model to at least one of the determination from the adaptive machine learning module and the determination from mathematical model.

8. The computer-implemented method of claim 7 , where the determination from the 3D geometric model and the determination from the mathematical model are based on respective inputs including different well tool characteristics.

9. The computer-implemented method of claim 1 , further comprising:
operating the mathematical model in real time, as the well tool is conveyed through the wellbore, based on a third set of inputs that vary with time.

10. The computer-implemented method of claim 9 , wherein the third set of inputs comprises one or more inputs related to a current position and orientation of the well tool within the wellbore.

11. The computer-implemented method of claim 9 , wherein the third set of inputs comprises one or more inputs related to characteristics of fluid flowing through the wellbore as the well tool is conveyed through the well.

12. Non-transitory computer-readable media embodying instructions that, when executed by a computing system, cause the computing system to perform operations comprising:
receiving, with an adaptive machine learning model, a first set of inputs representing a plurality of characteristics of a well tool and a second set of inputs representing a plurality of characteristics of a well, the well comprising a wellbore formed from the Earth's surface through one or more geologic formations to a subterranean zone, and the well tool positioned on a downhole tool string residing in the wellbore;
receiving, with the adaptive machine learning model, historical data representing a plurality of other well tools passed through a plurality of other wells and a plurality of characteristics of the other well tools and the other wells;
matching, with the adaptive machine learning model, the historical data with at least a portion of the first and second sets of inputs;
determining, with the adaptive machine learning model, whether the well tool can pass through an open space proximate an interval of the well based on matching the historical data with the portion of the first and second sets of inputs;
receiving, in real time as the well tool is conveyed through the wellbore, a determination from a mathematical model of whether the well tool can pass through the open space proximate the interval of the well, the mathematical model operating separately and in parallel to the adaptive machine learning model, such that the determination from the mathematical model is independent of th

Assignees

• Halliburton Energy Services Inc

Inventors

• Fox Philip Edmund
• Kennedy Ronnie
• Gary Ben
• Tran Dominic Anh
• Wood Josiah

Classifications

• G06F17/00 Primary

  Digital computing or data processing equipment or methods, specially adapted for specific functions (information retrieval, database structures or file system structures therefor G06F16/00) · CPC title

• E21B44/00 Primary

  Automatic control systems specially adapted for drilling operations, i.e. self-operating systems which function to carry out or modify a drilling operation without intervention of a human operator, e.g. computer-controlled drilling systems; Systems specially adapted for monitoring a plurality of drilling variables or conditions · CPC title

• E21B45/00

  Measuring the drilling time or rate of penetration · CPC title