Adaptive filter prediction method and system for detecting drill bit failure and signaling surface operator

• US 6,722,450 B2
• Filed: 10/26/2001
• Issued: 04/20/2004
• Est. Priority Date: 11/07/2000
• Status: Expired due to Term

• Current Assignee: Halliburton Energy Services Incorporated (Halliburton Company)
• Original Assignee: Halliburton Energy Services Incorporated (Halliburton Company)
• Inventors: Schultz, Roger L., Osborne, Andrew J. Jr., De Jesus, Orlando
• Primary Examiner(s): DANG, HOANG C
• Application Number:
  US10/040,301
• Publication Number:
  US 20020148646A1
• Time in Patent Office: 907 Days
• Field of Search: 175/40, 175/39, 702/9
• US Class Current: 175/39
• CPC Class Codes: E21B 12/02 Wear indicators
• CPC Class Codes: E21B 2200/22 Fuzzy logic, artificial intelligence, neural networks or the like
• CPC Class Codes: E21B 41/0085 Adaptations of electric power generating means for use in boreholes
• CPC Class Codes: E21B 44/00 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 Class Codes: E21B 47/18 through the well fluid , e.g. mud pressure pulse telemetry

Abstract

An apparatus and method for monitoring and reporting downhole bit failure. Sensors are located on a sub assembly (which is removable from the drill bit) and send data to neural net or other adaptive filter. The neural net uses past sensor readings to predict future sensor readings. The value predicted for the sensors is subtracted from the actual value to produce a prediction error. Increases in prediction error are used to indicate bit failure. The results of this are transmitted to the operator by varying the pressure in the drilling mud flow.

Claims

1. A system for detecting drill bit failure, comprising:

a drill string having a drill bit and a sub assembly, said sub assembly having no electrical communication with said till bit;

one or more sensors on said sub assembly;

a detection platform on said sub assembly, connected to receive data from said sensors;

wherein said detection platform adaptively models said data.

2. The system of claim 1, wherein drill bit condition is characterized based on the accuracy of said detection platform'"'"'s adaptive modelling.

3. The system of claim 1, wherein said adaptive model comprises an adaptive neural network.

4. The system of claim 1, wherein drill bit failure is predicted when the accuracy of said detection platform'"'"'s adaptive modelling degrades significantly.

5. The system of claim 1, wherein till bit failure is predicted when the accuracy of said detection platform'"'"'s adaptive modelling falls below a predetermined minimum threshold.

6. A system for detecting drill bit failure, comprising:

an adaptive model which fits noise filtered measurements from one or more downhole sensors to within a predetermined tolerance during normal drilling;

wherein bit failure is indicated when said model no longer fits said sensor measurements within said tolerance.

7. The system of claim 6, wherein said adaptive model predicts future sensor readings based on past sensor readings.

8. The system of claim 6, wherein said adaptive model estimates intermediate sensor readings based on previous and later sensor readings.

9. A system for detecting drill bit failure, comprising:

a drill string having a drill bit and a sub assembly;

one or more sensors located on said sub assembly;

an adaptive filter connected to receive signals from at least one of said sensors;

wherein said adaptive filter predicts future sensor signals based on past sensor signals; and

wherein bit failure is conditionally indicated when an error measurement of said filter exceeds a predetermined threshold.

10. The system of claim 9, wherein bit failure is indicated in dependence on said error measurement and on at least one other condition.

11. The system of claim 9, wherein said adaptive filter is an adaptive neural network.

12. The system of claim 9, wherein said adaptive filter has an infinite impulse response configuration.

13. The system of claim 9, wherein drill bit failure is predicted when the accuracy of said adaptive filter degrades significantly.

14. The system of claim 9, wherein said sub assembly has no signal communication with any sensors which may be in said drill bit.

15. A system for detecting drill bit failure, comprising:

an adaptive model which uses past sensor measurements to predict future sensor measurements;

wherein drill bit failure is conditionally indicated when predicted sensor measurements differ from actual sensor measurements in a predetermined way.

16. The system of claim 15, wherein said predetermined way comprises a calculated prediction error exceeding a predetermined threshold.

17. The system of claim 15, wherein said predetermined way comprises a calculated prediction error exceeding a predetermined threshold with a predetermined frequency.

18. The system of claim 15, wherein said predetermined way comprises a standard deviation of a calculated prediction error exceeding a predetermined threshold.

19. The system of claim 15, wherein said predetermined way comprises a predetermined change in a calculated prediction error.

20. A method of predicting failure of a drill bit, comprising the steps of:

adaptively modeling data from one or more downhole sensors; and

conditionally signalling the surface when the results of said adaptively modeling step deviate significantly from actual sensor measurements, to thereby predict failure of a drill bit.

21. The method of claim 20, wherein said conditionally signalling step depends on whether said results have deviated beyond a predetermined level.

22. A method of predicting failure of a drill bit, comprising the steps of:

adaptively modeling data from sensors in a sub assembly located on a drill string above a drill bit;

said adaptively modeling step having response characteristics which track said data during normal drilling; and

signalling a failing bit state conditionally, when said adaptively modelling step ceases to track said data to within a predetermined goodness-of-fit.

23. The method of claim 22, wherein said adaptively modeling step is performed by an adaptive neural network.

24. The method of claim 22, wherein said step of signalling is performed by varying downhole pressure of drilling fluid.

25. A method of determining drill bit failure, comprising the steps of:

using part of a time series of downhole sensor data to estimate intermediate elements of said time series, and calculating an error therefrom; and

signalling a failing bit state conditionally in dependence on said error.

26. A method of determining drill bit failure, comprising the steps of:

providing a drill string with a sub assembly, said sub assembly located on said string above a drill bit and having one or more sensors which collect data;

providing sensor data to an adaptive filter;

using part of a time series of said sensor data to model other elements of said time series, and calculating a modeling error therefrom; and

signalling a failing bit state conditionally in dependence on said error.

27. The method of claim 26, wherein said adaptive filter is an adaptive neural network.

28. The method of claim 26, wherein said test-failing bit state is signaled when said error exceeds a threshold value.

29. The method of claim 26, wherein said test-failing bit state is signaled when said error exceeds a threshold value with a predetermined frequency.

30. The method of claim 26, wherein said failing bit state is signaled when the standard deviation of said error exceeds a threshold value.