Investigations of copper pitting corrosion in drinking water using statistical methods

摘要

Pitting corrosion of copper in domestic drinking water installations is a rare but costly problem if pits cause an in-service-failure of the installation. The pits depth distribu-tion may be an important characteristic of the extent of such damage. The pitting likelihood of copper is determined by the material, the processing, the operation and the corrosion characteristics of the drinking-water. But the corrosion damage is generally influenced by an unfavourable combination of all these factors. The influence of material and processing mainly refers to the initiation process of pitting. Defects of the primary oxide layer on the inside wall of the pipe, inhomogeneities or the formation of copper(II) oxide layers by extreme heat treatment (for example hard soldering) cause a susceptibility to local attacks.The surface characteristics also determine the pitting potential. The initial pitting potential is the lowest value of the corrosion potential at which pit initiation is possible on a passive surface in a given corrosive environment. Until today, not all limiting conditions of pitting corrosion of copper in drinking-water are yet known, e.g. initiation, stabilization and repassivation of pitting corrosion, par-ticularly with regard to activation of repassivated pitting at aged copper tubes when the drinking-water quality is changed significantly. Under corrosion promoting conditions which were introduced by local heat treatment of copper tubes simulating hard soldering, the copper corrosion behaviour was stud-ied in different field corrosion tests for a period of two years. The test conditions were similar to those in a domestic installation with respect to stagnant and flow con-ditions. After different exposition times copper tubes were removed and examined with re-spect to morphological examinations of localized attack, measurements of surface roughness, measurements of all pitting attacks in a defined area, maximum corrosion depth measurements and statistical characterization. The group of data presenting maximum corrosion depth were analyzed by extreme value statistical analysis. The combination of the measurement of all pitting attacks and the extreme value statistical analysis showed as a result the differentiation between pits representing 1 typical uniform corrosion 2 active pitting 3 repassivated pitting. This method enables an assessment of the likelihood of pitting corrosion by changing the corrosion conditions, e.g. changing the water quality.