Detecting leakage of fluid from a conduit

• US 5,355,324 A
• Filed: 08/10/1992
• Issued: 10/11/1994
• Est. Priority Date: 11/06/1991
• Status: Expired due to Term

• Current Assignee: Shell Oil Company (Shell Plc)
• Original Assignee: Shell Oil Company (Shell Plc)
• Inventors: Zhang, Xue Jun
• Primary Examiner(s): Black, Thomas G.
• Assistant Examiner(s): ZANELLI, MICHAEL J
• Application Number:
  US07/926,428
• Time in Patent Office: 792 Days
• Field of Search: 364/509, 364/510, 364/554, 73/40, 73/40.5 R, 340/605
• US Class Current: 702/45
• CPC Class Codes: G01M 3/2807 for pipes G01M3/2892, G01M3/30 take precedence

Abstract

A method is provided to detect leakage from a conduit through which a stream of fluid flows and to indicate when there is a leakage condition or a no leakage condition. The method includes measuring at least one variable of a group consisting of fluid pressure and fluid flow rate over a period of time and at least one location of the stream to provide a data set representing the variable at the location and over the period of time. A statistical probability of obtaining the data set from the measurement under a first hypothesis that no leakage occurs is determined and a statistical probability of obtaining the data set from the measurement under a second hypothesis that leakage occurs is determined. One of the hypotheses is selected as being true if the probability of obtaining the data set from the measurement under the selected hypothesis is larger than the probability of obtaining the data set from the measurement under the other hypothesis. Should the second hypothesis be selected, a signal is initiated which provides an alarm to indicate that a leakage condition exists. Should the first hypothesis be selected a signal is initiated which indicates that a no leakage condition exists. In some embodiments of the invention, sequential probability ratio tests are used to determine the selected hypothesis.

Claims

1. A method to detect leakage from a conduit through which a stream of fluid flows and to indicate when there is a leakage condition or a no leakage condition on the conduit, comprising:

measuring at least one variable of a group consisting of fluid pressure and fluid flow rate between a first and a second time and at a location of the stream flowing through the conduit to provide a data set representing the variable at said location and between said first and second times;

determining a statistical probability of obtaining said data set from the measurement under a first hypothesis that no leakage occurs from the conduit and a statistical probability of obtaining said data set from the measurement under a second hypothesis that leakage occurs from the conduit;

selecting one of said hypotheses as being true if the probability of obtaining said data set from the measurement under said selected hypothesis is larger than the probability of obtaining said data set from the measurement under the other hypothesis; and

creating a signal indicative of said selected hypothesis, wherein the signal initiates an alarm when the second hypothesis is selected to indicate the leakage condition on the conduit or said signal initiates a notice of normal operation for the conduit when the first hypothesis is selected to indicate the no leakage condition.

2. The method of claim 1 wherein the step of determining said probabilities of obtaining the data set comprises selecting a statistical distribution of the variable at the location pertaining to the data set under the first hypothesis that no leakage occurs and selecting a statistical distribution of the variable at said location pertaining to the data set under the second hypothesis that leakage occurs, the statistical distributions being similar to each other but having different mean values.

3. The method of claim 2, wherein said statistical distributions are gaussian distributions having substantially equal variances.

4. The method of claim 1, wherein said second hypothesis is selected as being true if the difference between the probability of obtaining the data set under said second hypothesis and the probability of obtaining the data set under the first hypothesis exceeds a minimum difference which is related to a selected false alarm probability.

5. The method of claim 1, wherein said first hypothesis is selected as being true if the difference between the probability of obtaining the data set under said first hypothesis and the probability of obtaining the data set under the second hypothesis exceeds a minimum difference which is related to a selected missed alarm probability.

6. The method of claim 1, wherein the steps of determining said probabilities and selecting said one hypothesis are carried out using a sequential probability ratio test.

7. The method of claim 6 wherein the sequential probability ratio test includes determining a logarithm of a ratio of the probability of the second hypothesis to the probability of the first hypothesis and selecting one of the two hypotheses as true when the logarithm exceeds values established by a strength test having preselected probabilities for a false alarm and a missed alarm, the second hypothesis being selected as true when the logarithm exceeds a positive value established by the strength test, and the first hypothesis being selected as true when the logarithm exceeds, in a negative direction, a negative value established by the strength test.

8. The method of claim 1, wherein if a leakage condition is indicated, a location of leakage is determined by solving a steady state equation relating said measured variable to said location of leakage, using a least squares algorithm.

9. The method of claim 1, wherein said location defines a first location, said data set forms a first data set, the method further comprising:

measuring said at least one variable between said first and second times at a second location of the stream to provide a second data set representing the variable at the second location and between said first and second times;

determining a statistical probability of obtaining the second data set under the first hypothesis that no leakage occurs and a statistical probability of obtaining the second data set under the second hypothesis that leakage occurs; and

wherein the selecting step includes selecting said one hypothesis as being true if the probability of obtaining the first and second data sets under said selected hypothesis is larger than the probability of obtaining the data sets under the other hypothesis.

10. The method of claim 9 wherein the step of determining said probabilities of obtaining the first and second data sets comprises selecting a statistical distribution of the variable at each location pertaining to the location'"'"'s data set under the first hypothesis that no leakage occurs and selecting a statistical distribution of the variable at each location pertaining to the location'"'"'s data set under the second hypothesis that leakage occurs, the statistical distributions being similar to each other but having different mean values.

11. The method of claim 10 wherein said statistical distributions are gaussian distributions having substantially equal variances.

12. The method of claim 9 wherein said second hypothesis is selected as being true if the difference between the probability of obtaining each locations data set under said second hypothesis and the probability of obtaining each locations data set under the first hypothesis exceeds a minimum difference at each location which is related to a selected false alarm probability for the location.

13. The method of claim 9 wherein said first hypothesis is selected as being true if the difference between the probability of obtaining each location'"'"'s data set under said first hypothesis and the probability of obtaining each location'"'"'s data set under the second hypothesis exceeds a minimum difference at each location which is related to a selected missed alarm probability for the location.

14. The method of claim 9 wherein the steps of determining said probabilities and selecting said one hypothesis are carried out using sequential probability ratio tests.

15. The method of claim 14 wherein the sequential probability ratio tests include determining a logarithm of a ratio of the probability of the second hypothesis to the probability of the first hypothesis at each location and selecting one of the two hypotheses as true when the logarithm at each location exceeds values established by a strength test having preselected probabilities for a false alarm and a missed alarm at each location, the second hypothesis being selected as true when the logarithm at each location exceeds a positive value established by the strength test for that location, and the first hypothesis being selected as true when the logarithm at each location exceeds, in a negative direction, a negative value established by the strength test for that location.

16. The method of claim 9, wherein said first location is at an inlet of the conduit, and said second location is at an outlet of the conduit.

17. The method of claim 9, wherein if a leakage condition is indicated, a location of leakage is determined by solving a steady state equation relating said measured variable to said location of leakage, using a least squares algorithm.