Real-time health monitoring of a new above-ground pipeline using the Kalman Estimator-based system/LQG

  • Latifa A. Al Ghailani

Student thesis: Master's Thesis


Pipelines are widely used for distribution and transportation of many products such as water, oil, and natural gas. Due to the long distances of the piping systems that transport oil and natural gas, they possibly will pass through different topography of the earth which affects their integrity. The safe operation of pipelines system can be affected by causing rupture leaks caused by pressure surges, corrosion and/or scaling. The challenging task is to predict failure in the pipeline before occurring or at least when they occur and subsequently locate them. Therefore, an a priori prediction of failure is of particular interest to oil and gas industry due to the severe economic and environmental implications associated with such failure. The method proposed in this work presents a fresh look at the problem, relying mainly on the pipeline vibration signature to predict failure, or identify ongoing ones. It takes sensor placement difficulties into consideration and, instead, relies mainly on estimation of failure using minimal feedback from the pipeline structure. Therefore, this report will present a novel technique for predicting failure in pipeline using a dual Linear Quadratic Gaussian (LQG) and hardware in-the-loop approach. The approach requires minimal sensory feedback and calculations cost to operate and can target many failure modes such as scaling, leak, significant corrosion and any other structural related defects. It provides a real-time prediction and early warning using Kalman Estimator-based system by using MATLAB in conjunction with dSPACE. This estimator can predict vibration at any location along the pipeline utilizing both a structural dynamic model and measurements from sensors placed at strategic locations on the pipeline. The method is based on estimating the vibration of the pipeline using Linear Quadratic Estimator (LQE) with minimal feedback measurement combined with Linear Quadratic Regulator (LQR). An optimal output feedback controller is created through coupling LQE with LQR. Tuning of LQG is used to reduce the error between the actual measurement at certain locations of the pipe and the estimated state values. ANSYS has been used to create a finite element model (FEM) of a pipe section to determine the first 10 mode shapes and natural frequencies. A series of experiments were conducted to verify the outcome of the modeling work. The results indicate that modeling and experimental results are in good agreement when using LQE and LQR controller. Then differences of natural frequencies have been found between clean pipe and when applying different masses on different locations on the pipe span. Finally, 2kfactorial experiment is designed t establish predictive relationships between mass, location, and direction on the variation of modal properties of the structure. ANOVA methods are utilized to establish the failure prediction model.
Date of AwardDec 2016
Original languageAmerican English
SupervisorAmeen Hussein El-Sinawi (Supervisor)


  • Applied sciences
  • Health monitoring
  • Kalman estimator
  • Pipeline vibration
  • Mechanical engineering
  • 0548:Mechanical engineering

Cite this