Modelling and quantification of measurement uncertainty for leak localisation in a gas pipeline

Abstract

To assess the accuracy of the estimated leak location along the gas pipeline, measurement uncertainty (MU) must be taken into consideration. The MU of the estimation of leak location mainly consists of five contributors: flow medium temperature, flow velocity, time difference of arrival (TDOA) of negative pressure wave (NPW), and pipeline length. A model has been proposed to determine the MU of the estimation of leak location based on the Guide to the Expression of Uncertainty in Measurement (GUM). It was found that the uncertainty propagation from measurement of dynamic pressure to the TDOA is implicit. In order to resolve this issue, Monte Carlo method is employed. The contribution of each factor to the MU of the estimation of leak location has also been analyzed. The denoising technique of Butterworth filter can minimize the uncertainty of the TDOA, which is associated with the MU of dynamic pressure sensors. The results show that the manual measurement of the length of pipeline and actual leak locations play the most crucial role in the MU of the estimation of leak location. When a leak is nearer to one extremity of the pipeline, the localization MU is influenced more greatly by the flow and temperature MUs at the other extremity of the pipeline rather than at the same end. When the leak location is moving away from the inlet of pipeline, the expanded uncertainties will increase in the range of 0.0522% to 0.0806%.

Keywords

• Leak localization
• Measurement uncertainty
• Monte Carlo method
• Negative pressure wave