Browse Prior Art Database

Publication Date: 2016-May-27
Document File: 5 page(s) / 145K

Publishing Venue

The Prior Art Database


This disclosure describes a statistical anomaly detection method for leak detection in risers and pipelines. The method is selected to deal with a variable background acoustic signature present on risers and a variable nature of the acoustic signatures of leaks themselves. The anomaly detection method models a nominal acoustic behavior of the riser and then quantifies deviations from it. These deviations quantified as anomaly scores are then be used to detect unusual behaviors that may be due to leaks. Anomaly detection is carried out using a spectral approach, where energy in frequency bands is modeled as a multimodal multidimensional Gaussian random variable. Model training using a baseline data with no leak determines normal statistical variation of acoustic signature for each section of the pipe. Then, in monitoring mode, statistically significant deviations of measured acoustic signature from the learned model are flagged as outliers and leaks.

This text was extracted from a Microsoft Word document.
At least one non-text object (such as an image or picture) has been suppressed.
This is the abbreviated version, containing approximately 34% of the total text.



The present disclosure relates generally to a risers and pipelines used in oil and gas industry and more particularly to a method for anomaly detection to detect leaks in risers and pipelines.

Risers are fluid conduits that are used in oil and gas applications to guide a product from sea bed to a surface. A typical riser in an oil field can be several kilometers long. Different sections of the riser may observe different acoustic characteristics during normal operation. For instance, sections close to a floating platform may observe a significant acoustic response from equipment placed on the platform. Leaks in risers adversely affect surrounding environment and add cost to drilling operation. Therefore, it is necessary to monitor and detect leaks in risers early. Fiber optic sensing cables are usually used for monitoring the risers due to distributed nature of the fiber optic sensing cables. One fiber can be used to measure multiple parameters along length of the riser.

In sub-sea risers, there are several possible modes of failure, for example, fatigue, over stress, pipeline movement, among others. Such modes of failure may be manifested through different abnormalities in common mechanical and thermal characteristics of a pipe throughout pipe length. Deviations of these parameters from normal expected values and patterns can be captured with distributed optical fiber sensing technologies. Anomalies, for example, due to leaks, show up as change in local acoustic behavior of the riser. Such changes in acoustic behaviors are picked-up by the fiber optic sensors. It is required to detect and isolate anomalous behaviors under uncertain environments and background noise. Moreover, nominal local acoustic characteristic may not be the same along the length of the riser, thereby, making detection of the leakage in the riser even more difficult.

Several techniques are known in the art for detecting leaks in pipelines. For example, in a conventional technique, leak in a pipeline or conduit is identified by determining occurrence of an acoustic anomaly along with temperature anomaly in a processed coherent Rayleigh noise. Backscatter is collected as a result of passing a plurality of pulses and analyzed to determine spectral characteristics or power spectral density (PSD) of signal at each point of interest.

Another conventional technique employs distributed fiber optic temperature sensor. The sensor works in conjunction with acoustic sensors (DAS) for detecting temperature changes on exterior of pipeline simultaneously with acoustic events. Output of acoustic monitoring is compared with normal background acoustic noise for anomalies and presence of an acoustic anomaly is selected as an acoustic event of interest (leaks).

One other conventional technique employs at least two detection offset modules for detecting and localizing mechanical impacts or transient or leaks in a ...