Abstract
Accurate corrosion modeling allows engineers to predict when and where corrosion is likely to occur, and to which extent it can affect exposed structures. This allows improved design and materials selection, facilitates timely maintenance and repair, hence reducing downtime and preventing structural failures. Physics-based galvanic corrosion modeling has reached a high level of maturity and reliability, where the mechanics of the simulation is well proven and understood. However, there are still many challenges to address. One of them is that the solvers require accurate data in order to produce accurate predictions. The acquisition of such data for a meaningful simulation is not a trivial task. For example, corrosion and environmental sensors can collect relative humidity, temperature, electrolyte conductance, and accumulated charge among other parameters, but the required direct inputs for the model are polarization curves, film thickness and electrolyte conductivity. This paper presents the different techniques including reduced order models, polarization data deconvolution and corrosion modeling via finite elements currently developed for a digital twin system capable of collecting environmental data from sensors, transforming the acquired data into useful inputs for numerical models, generating representative load cases, and finally predicting accumulated corrosion damage over a certain period.
