An evaluation of thermodynamic effects on leak detection and location estimation in natural gas pipelines was provided through non-isothermal process model. To estimate the process states (discretized pressures and mass flow rates) under varying thermodynamic conditions for leak detection and location estimation, a state and parameter estimation method (dual unscented Kalman filter) was proposed for the pipeline model. The dual unscented Kalman filter was adopted to estimate both the states and parameters in the presence of parameter uncertainties. Using the data generated from the nonisothermal model, the proposed new algorithm can detect the leak location efficiently and a real case study was performed to validate it.
Keywords Dual Unscented Kalman Filter, Parameter Estimation, Leak Detection, Non-Isothermal Modeling, Natural Gas Pipeline