Statistical Parameter Estimation to Find Oil Pipeline Leakage Using a Fiber Optic Sensor

Abstract

We present a new method for enhancing the detection of leakage points on an oil pipeline using a fiber optic sensor stretched along the pipeline. The method is based on Maximum Likelihood Estimation (MLE) and is used with an Optical Time Domain Reflectometer (OTDR). The system depends on a series of fuse blocks, which respond to the presence of oil. The fuse blocks bend the fiber, when the fuse blocks are immersed in oil. The location of the bend in the fiber can be detected using an OTDR. This approach, while very effective in general, can become less successful in certain conditions including: low signal-to-noise ratio, long lengths of fiber (> 200km) and low signal strength. In this work we modify the basic method to add steps including the Maximum Likelihood Estimation of the parameters in the method. This new modified method is tested using various signals and the approach is suggested to be effective for realistic situations in which the fiber is only very slightly bent or in the presence of high noise. The results suggest that the new method is very effective and could be implemented in field trials.

Keywords: Distributed fiber sensors, Maximum Likelihood Estimation, oil pipeline leakage, OTDR signal analysis