Model-based Leak Detection And Localization Of Gas Pipeline System

Pipeline transportation has its unique advantage in natural gas production. With thedevelopment of western oil fields and the West-East Gas Pipeline Project, the method ofpipeline transportation will be increasingly important in the national economy. Natural gaspipeline leakages not only lead to the loss of resources, but also significantly pollute theenvironment, even can ignite fire or explosion, being a serious threat to people’s lives andproperty. The technology of pipeline leaking detection and location is an importantguarantee for safety of pipeline transport operation. Therefore, the natural gas pipeline leakdetection is a research field of great significance.Model-based leak detection and localization of natural gas pipeline was establishedupon the model of internal flow. Through analyzing real-time pressure parameters of thepipeline, the method is able to detect leak incidence quickly and accurately. Afterthoroughly development, the method has been applied in actual production site, aiming toguarantee the normal production. At present, the model-based leak detection andlocalization use the model of liquid. Exclude the unique characteristics of gas, likecompressible factors, lowering positioning accuracy when detecting leakage. The density ofnatural gas is an important factor in model-based detect algorithm and in the previousmodel, the density was set to be a constant. In general, pressure is relatively big in pipelinetransportation. Considering the characteristics of mixed gas, natural gas having differentcompositions has distinct densities under varying circumstances. The changes, as a whole,may have a negative effect on the localization accuracy.Starting from the previous model-based leak detection algorithm, the paper firstestablish the steady state model of compressible flow. Based on that, the paper discuss theimpact of different compositions on natural gas’s density and localization accuracy. Inaddition, the research improve the model by incorporating a model to better describe thechanges in density of natural gas.The simulation and experiment results show that model-based leak detection andlocalization of gas pipeline is a fast and quick way to detect leak incidence. The outcomepresent a significant improvement of the new method in the leak location compared withthe former algorithm.