Data-Driven Study of Atmospheric Corrosion Under Multi-Droplet Conditions
An End-To-End Experimental-Computational Multi-Modal Framework For Electrolyte-Resolved Corrosion Kinetics Investigation
K. Zhang (TU Delft - Mechanical Engineering)
J.M.C. Mol – Promotor (TU Delft - Mechanical Engineering)
Y. Gonzalez Garcia – Copromotor (TU Delft - Mechanical Engineering)
More Info
expand_more
Other than for strictly personal use, it is not permitted to download, forward or distribute the text or part of it, without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license such as Creative Commons.
Abstract
Atmospheric corrosion is a major challenge for infrastructure and industry, yet predicting corrosion under realistic environmental conditions remains difficult. Existing empirical and machine-learning approaches typically relate environmental conditions directly to corrosion outcomes without capturing the electrolyte layer that physically mediates the corrosion process. This thesis develops an end-to-end experimental and computational framework to link the dynamics of realistic multi-droplet electrolytes to macroscopic corrosion kinetics.
-A custom climate chamber and electrical resistance sensor system were developed to continuously monitor corrosion under discontinuous, droplet-based conditions.
-An automated computer vision pipeline was developed to track droplet geometry and corrosion product formation across thousands of individual droplets.
-Larger droplets showed earlier onset and faster corrosion, with two distinct spatial patterns of attack identified and quantified.
-A weakly supervised machine learning framework was developed to infer individual droplet corrosion kinetics from the global sensor signal without requiring droplet-level ground-truth labels.
-Surface roughness was shown to influence corrosion by promoting larger, more elongated droplets through enhanced pinning and coalescence.
Together, these results establish a framework for making the electrolyte population visible, measurable, and directly linkable to atmospheric corrosion kinetics.
help_outline