Pipe condition assessment provides useful information to assist rehabilitation and replacement decisions. Due to high cost or difficult pipe access, condition assessment must be limited to specific locations. To address this limitation, this paper presents a methodology in which values from discrete sampling positions are used to quantify systematic and random variation of pipe condition, and the statistical basis of applying the information to determine the failure probability in the network.
The methodology includes development of a sampling program on a network zoned according to the soil type, using extreme value statistics to describe the distribution of data from limited samples within each zone and where appropriate extrapolation of the distributions to larger areas of the network. Structural reliability methods, specifically Level II first-order-second moment reliability analysis, are then applied with distributions of pipe conditions as input in order to estimate pipe failure probability and failure rate predictions. A practical example of a mild steel pipeline subjected to external corrosion is used to illustrate the value of this technique in practice.
An overview of commercially available and emerging techniques suitable for condition assessment of pipes is presented. This includes direct assessment by electromagnetic, ultrasonic and acoustic techniques and indirect assessment through measurement of soil electrical and chemical properties.