Managing external corrosion, especially for underground assets, is a significant challenge dating back to the first underground pipeline in 1865. The very first issue of the journal, CORROSION, featured a headline story on this subject. This subject is fundamental for corrosion engineers and pipeline operators.

Estimating corrosion growth rate for underground pipelines is a non-linear multivariate problem. There are many potential confounding variables such as soil parameters cathodic protection AC/DC interference seasonal / climate conditions and proximity to unique geographic features such as wetlands or polluted environments. The work presented provides an approach for estimating underground corrosion growth rates using a dataset of observations from a North American pipeline operator. Extensive geospatial data is utilized that has been obtained from public and private sources and extrapolated using Inverse distance weighted (IDW) interpolation. This work presents a model using IDW to estimate parameters involving soil interference geography and climate factors for any location in North America.Using this data this work then presents several different machine learning approaches including Generalized Linear Models eXtreme Boosted Trees and Neural Networks. All three provide an accurate estimation for corrosion growth rates for an underground asset at any latitude and longitude pair in North America. Each method comes with potential benefits and pitfalls specifically; trade-offs between model accuracy and transparency. This work presents a framework for comparing geo-spatial and machine learning estimates. Findings and a framework are provided for owners to assess how to think machine learning on their own assets.

Estimating corrosion growth rates for underground pipelines is a challenging problem. There are confounding variables with complex interaction effects that may result in unexpected outcomes. For instance, the relationship between soil conditions and AC interference is highly non-linear and challenging to model. This work expands upon prior work using a suite of machine learning tools to estimate corrosion rates. However, instead of estimating a single corrosion growth rate for a single girth weld address (GWA), this work estimates a distribution of potential corrosion growth rates. Modeling distributions provide a more effective risk-measurement framework, especially concerning high volatility or areas of severe tail risk. 

This work relies heavily on machine learning and geospatial tools - particularly artificial neural networks and gradient boosted trees to estimate the corrosion rates and non-linear processes. Building upon prior work using data from a North American Operator, the models in this paper use additional variables from recent research in AC interference and microbiologically influenced corrosion to construct a higher accuracy and distribution-based model of pipeline corrosion risk.