Defect Detection in Pipeline Using Acoustic Pulse Reflectometry

UoM administered thesis: Phd

  • Authors:
  • Mohd Shahrul Azmi Mohamad Yusoff

Abstract

Early detection of defects in the commercial piping system is important in preventing large and expensive repairs due to huge unprecedented failure caused by corrosion, blockage or structural deformation. Over the years, various non-destructive inspection methods were used within the multimillion oil and gas industry to prevent losses in both people and money. Studies have been done on stable isotropic media by means of studying the ultrasonic guided test. The dispersive nature of propagation within solid media requires diagnoses from technicians with great experiences, in interpreting the signals acquired through the available sensors and limited to a short distance inspection due to the high attenuation effect. Meanwhile, acoustic pulse reflectometry (APR) within the travelling fluid medium offers an on-line long-range detection potential in a simpler plane wave solution. Although the use of current APR method for locating the non-conformity within a pipeline is straight forward using flight time analysis, the reconstruction knowledge of the blockages or leakages from the response is limited. The incoming input and the response signature are in need of greater understanding. Therefore, in this research, the time domain reflectometry was investigated for various acoustic wave reflections from defects within a pipeline system by means of analytical solution, experimental procedures, and finite element analysis. The travelling potential of input pulse was initially studied for various frequency and amplitude using analytical model, and later modelled in fluid-solid-interface module within 3-D finite element software. The cut off limit of plane wave frequency before higher order mode excitation was determined. Based on the plane wave excitation frequency limit, suitable frequency range of input pulse for long pipeline was further investigated by designing sensitivity analysis test rig. A modification was introduced to the governing attenuation equation for plane wave model of various diameters by exponentially fitting the experimental peak decay of propagating input pulse results, and later adopted into the analytical reconstruction of blockage geometry. Blockages of different sizes, porosity and length were introduced in the experiment to understand the propagation of acoustic wave within the pipe and its echoed signals from non-conformity. Classifications of blockage and length based on the acquired responses were developed using time and frequency domain analysis and compared to the analytical model. Responses from open end excitation were then used as the input to the classification of blockage parameters and trained using neural network for faster processing. This thesis narrowed the gap of knowledge by establishing the methodology for attenuation coefficient using the curve fitting method of multiple pipe reflection, the classifying method of blockage within a pipeline and the training of neural network data from finite element analysis simulation. The results show a great improvement to the detection and re-construction techniques for blockage percentage and length based on the signature of responses for pipeline inspection using APR. The research established different input signals responses, experimental model for determination of defects in pipe and new empirical determination for attenuation coefficient for propagating sound pulse in pipe using single sensor. All these findings will be great input to field application procedures for pipeline inspection.

Details

Original languageEnglish
Awarding Institution
Supervisors/Advisors
Award date1 Aug 2019