Influencing factors analysis for drinking water steel pipe pitting corrosion using artificial neural network

Kibum Kim, Heechang Kang, Taehyeon Kim, David Thomas Iseley, Jaeho Choi, Jayong Koo

Research output: Contribution to journalArticlepeer-review

1 Scopus citations


Steel is a metal, and thus, it undergoes corrosion over time. The comprehensive analysis of the factors influencing corrosion can aid in developing strategies, such as new ways to avoid corrosive environments. This study explored the factors influencing pitting corrosion in steel water pipes in South Korea between 1988–2020, using artificial neural networks. Partial dependence plots and variable importance are used to identify the degree of influence of the 12 corrosion-influencing factors. Pipe age had the highest importance and strongest influence on corrosion among the corrosion-influencing factors. Soil resistivity strongly influenced external corrosion, especially at values less than 5,000 Ω-cm, and the influence of sulfide concentration on external corrosion was also relatively strong. Water alkalinity exhibited the strongest influence on internal corrosion. This study will serve as reference data for developing corrosion depth prediction models and will contribute to understanding corrosive environments when laying new pipelines and improving existing ones.

Original languageEnglish
Pages (from-to)550-563
Number of pages14
JournalUrban Water Journal
Issue number5
StatePublished - 2023


  • Artificial neural network
  • corrosion
  • machine learning
  • partial dependence plot
  • steel pipe corrosion


Dive into the research topics of 'Influencing factors analysis for drinking water steel pipe pitting corrosion using artificial neural network'. Together they form a unique fingerprint.

Cite this