How do we assess the condition of the pipe cost effectively?

Associate Professor Jaime Valls Miro and Professor Gamini Dissanayake of the University of Technology Sydney are co-leading a team of researchers and students for this activity. The aim of this activity is to advance knowledge and improve levels of confidence of direct methods for condition assessment using advanced data interpretation techniques which have already been successfully employed in fields such as aerospace, cargo handling, undersea ecology, land vehicles and mining. Sydney Water has provided a decommissioned 600 mm diameter cement-lined cast iron pipe of 1.5 km length in Strathfield for this purpose, but it will support the other two activities also.

The outcome of Activity 2 will be a method of accurately predicting sensor readings for a given geometric description of a buried large water main, and obtaining the best estimate of the pipe geometry from a set of measurements based on maximum likelihood principles.

Current work includes:

  • Sensor models have been developed using Finite Element Analysis (FEA) techniques for sensor technologies based on magnetic flux leakage (MFL) and broadband electromagnetic technologies (BEM).
  •  A highly precise and accurate description of actual pipe surface (internal and external) condition (known as the ‘ground truth’) was developed using 3D laser mapping technologies for selected pipes.
  • Using both real and simulated data machine learning algorithms were employed to model features and allow inferences about the depth of pipe defects to be made with confidence. The current algorithms produce a measure of certainty on the interpretation of the sensor output.
  • Simulation models for two additional condition assessment technologies are currently being developed and investigated
    1. Acoustic Wave Propagation technology applied to a fluid filled pipe is being modeled using FEA
    2. Remote Field Technology which is based on the theory of remote field eddy currents is currently being modelled using FEA techniques in conjunction with the ground truth to improve sensor localisation. At this stage, a detection and verification of pipe construction features such as joints has been developed using machine learning.
Magnetic flux leakage, sensor modelling . Comsol simulations

For information on data exchange for Activity 2 contact uts-water-mains@lists.feit.uts.edu.au