EUVIMEDEUVIMED entdecken
Crossref

Machine Learning-Based Surrogate Modeling of High-Fidelity Multiphysics CO2 Corrosion Model Predictions

Ethan Proudlove, Harvey M. Thompson, Richard C. Woollam, Michael Jones, Richard Barker

Corrosion · 01.07.2026 · 82 (7) · 626-639

DOI: 10.5006/4888 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Corrosion, online erschienen am 06.05.2026. Verfasst von Ethan Proudlove und 4 weiteren Autor:innen. Die Arbeit nennt 46 Literaturangaben.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

The University of Leeds has developed a high-fidelity mechanistic CO2 corrosion model that simulates the underlying bulk equilibrium, mass transport, and electrochemical processes using multiphysics simulation software. This high-fidelity model (the Leeds Model) provides accurate forecasting of corrosion rate (CR), surface pH, and the precipitation tendency of protective corrosion products within mild steel pipelines. However, there are several barriers hindering the effective use of the outputs from this model, including the requirement for trained users, licensing costs, and the timescale and complexity of achieving satisfactory convergence when performing extended run-time, large-scale parametric sweeps. In this work, we investigate the use of machine learning (ML)-based surrogate modeling to combat these limitations, to develop a surrogate model to emulate Leeds Model predictions in a much more convenient format. Six ML models were trained on over 160,000 CO2 CR predictions made by the Leeds Model, each with hyperparameters optimised accordingly. Of the models examined, the Deep Neural Network (DNN) and extra trees regressor exhibit most accurate test set predictions, achieving a MAPE of 0.44% and 0.39%, respectively. Further trade-offs between these two surrogates are discussed, highlighting significant improvements in simulation time, computational cost, and usability compared to the original model. During analysis, 47 anomalous datapoints are discovered within the training dataset, caused by poor convergence in the Leeds Model. DNN predictions are shown to be insensitive to poorly converged datapoints and prove effective in detecting convergence issues in the Leeds Model’s CR predictions.

Zitieren

APA

Proudlove, E., Thompson, H. M., Woollam, R. C., Jones, M., & Barker, R. (2026). Machine Learning-Based Surrogate Modeling of High-Fidelity Multiphysics CO2 Corrosion Model Predictions. Corrosion, 82(7), 626-639. https://doi.org/10.5006/4888

Vancouver

Proudlove E, Thompson HM, Woollam RC, Jones M, Barker R. Machine Learning-Based Surrogate Modeling of High-Fidelity Multiphysics CO2 Corrosion Model Predictions. Corrosion. 2026;82(7):626-639. doi:10.5006/4888

BibTeX
@article{Proudlove2026,
  author = {Proudlove, Ethan and Thompson, Harvey M. and Woollam, Richard C. and Jones, Michael and Barker, Richard},
  title = {Machine Learning-Based Surrogate Modeling of High-Fidelity Multiphysics CO2 Corrosion Model Predictions},
  journal = {Corrosion},
  year = {2026},
  volume = {82},
  number = {7},
  pages = {626--639},
  doi = {10.5006/4888}
}
Kostenlos starten →