Study Examining Idealised Corrosion Pits in Metal Pipes Under Longitudinal Bending

Abstract

Cast iron represents one of the oldest materials used in water distribution networks across North America. However, aging cast iron water pipes breaking and/or leaking is an increasingly important issue as most are beyond their expected service life. One potential reason for these leaks is the presence of corrosion reducing the wall area. To explore the impact of corrosion and bending on pipe behaviour, a control specimen and a deteriorated specimen were exposed to differential ground motion in the split-box facility at Queen’s University. The data indicated that the curvature of the two specimens was not linearly related to the ground displacement and the effect of a simulated corrosion pit in the form of a hole was localized and had a negligible effect on the curvature on the moving side while causing an increase of 4.3% in the curvature and 61.0% in the maximum strain at the hole location on the stationary side. The axial strains along the neutral axis featured zones of compression and tension near the fault line where it was expected that zero strain would develop (i.e. pure bending was expected), requiring further investigation. Since the effects of deterioration appeared to be localized to the deteriorated area, a series of four-point bending tests were conducted on a control pipe with no deterioration and three other specimens with idealised deterioration. Distributed fibre optic strain measurements were used to explore both the behaviour of the pipe and the localized strain distribution in the vicinity of the simulated corrosion pits. The test results indicated that strain at the invert of the intact sample was higher than the calculated strain based on the beam theory, and the strain at the centroid (springline) of the pipes was non-zero contrary to what elastic beam theory would suggest. These deviations from the expected theoretical behaviour will require further study. A method of installing optical fibres inside and outside of the pits was developed. The measured strain distributions around each simulated corrosion pit were compared to the Kirsch solution. The Kirsch solution provided good visual agreement with the measured strains in the case of a full thickness hole but the overall agreement was not as good when the corrosion pit did not fully penetrate the wall section. However, the Kirsch solution did provide accurate estimates of the maximum strain in all cases suggesting that it could be used to estimate peak stresses if the corrosion geometry is known.