摘要：Osawa et al. (2016) developed a two-dimensional cellular automaton (CA) for under-film corrosion analysis and succeeded in consistent analysis of coating film deterioration and corrosion depletion. In this CA analysis, the CA parameters for coating performance and metallic corrosion are given by random number fields without spatial covariance structures. In this report, a corrosion test of coated steel panels with double scribes was performed, showing non-uniformity in corrosion surface profiles. The under-film corrosion simulations of double scribed coated panels were carried out by using the improved CA in which CA parameters were given by lognormal random number fields with spatial covariance structures. As the result, the probability that the non-symmetric corrosion surface profile in the experiment could be reproduced by simulation was improved by 10.5 times by considering the covariance structures. Based on this result, the effectiveness of the developed analysis method was verified.
