top of page

Unlocking the Future: The Fascinating World of Corrosion Prediction

Materials.Business Newsletter ⚙️ January 31th, 2024


Modeling, simulation, and corrosion prediction pay 

A permanent challenge
Tackling the intricate challenge of predicting material behavior has long been at the heart of the corrosion prevention field, touching upon business, safety, health, aesthetics, and beyond. Initially relying on empirical knowledge and heuristic engineering methods, the quest for more precise estimates with minimal uncertainty has driven the evolution of corrosion prediction into a scientific endeavor.

However, more precise estimates, with the least possible uncertainty, have always been needed, and predicting corrosion behavior has become increasingly scientific. Simulation by physical modeling has been a helpful tool for corrosion studies. Just like we reviewed in the previous podcast, where a Corrosion Model is being used to help scientists model life on Saturn's Moon Enceladus. Most of the works developed in the lab deal with the simulation of authentic situations but under controlled conditions. This means that some known variables are fixed while the remainder are systematically tracked (usually, there are unknown, uncontrolled variables, too). Experimental simulation is based on the scientific method, and the results are just an approach to the actual field situation. Nonetheless, simulation is a simplification, and real cases are complex, as the world is. Theoretical modeling has been a complementary tool, trying to solve limitations such as accurate control of variables and cost and time-consumption requirements in simulation experiments. Modeling is a mathematical approach to any situation, and this is one of its more substantial facts. Quality of modeling depends on the initially given information, almost always empirical, with all the risks that such information carries and the degree of adjustment of the mathematical tool to the fact under consideration. 
The quality of information available to modeling tools has significantly improved with the advancements in corrosion science and technology. A deeper understanding of materials, environments, and their interactions is essential for more accurate modeling, resulting in enhanced predictions. Additionally, the nature of corrosion products plays a crucial role in shaping the information required for modeling corrosion and anti-corrosion scenarios.
Valuable tools 
In the beginning, mathematical equations were the best way to predict corrosion behavior, either by interpolation among known situations or extrapolating beyond these. Thus, linear processes like iron corrosion in strong acids, where corrosion products go away immediately, can be modeled. Besides, phenomena like atmospheric corrosion of engineering alloys (Fe, Cu, Al, and Zn) behavior are not linear, and, instead, equations as bi-logarithmic have been applied successfully many times. Equations that try to adjust to the thermodynamic and kinetic features happening in real situations. For example, some equations distinguish corrosive attack at the beginning of exposure from progression afterward. Many studies have been reported in the literature, including equations predicting the behavior of several building materials deduced from specific sites across the planet in particular periods. Yet there are no universal equations, valid wherever and whenever. Fortunately, more advanced mathematical tools than the simpler equations have been developed and applied to better model corrosion processes. These are statistical methods, including conceptual, analytical, probability-based, and knowledge-based models. In particular, the last ones have been applied to corrosion studies. Knowledge-based modeling approaches have evolved from the last decades of the twentieth century to the powerful tool today, named machine learning – ML, as a prominent constituent of Artificial Intelligence tools. ML is one of the radical innovations shaping the Fourth Industrial Revolution. In such a condition, it is possible to solve more complex, multi-objective problems. Some of the essential methods available today that must be considered by Corrosionists dealing with the prediction of the corrosion phenomena include Markov chains, Monte Carlo simulation, grey relational analysis, artificial neural networks, and their several derivate methods, support vector regression, expert systems, fuzzy logic, genetic models, and, obviously, machine learning where the previous methods merge. A plethora of options are often combinable or complementary and may be considered depending on the requirements of a situation and the possibilities of data processing. 
On the other side, it is crucial to consider adding different approaches to the corrosion studies. Future corrosion and anticorrosion engineering must pass through a more multidisciplinary and complex approach. For example, corrosion processes, including the flow of fluids, such as weather or ocean situations, water flow in a pipe, airflow around a building, etc., can be studied by the mechanics of fluids. So, modeling with equations like the Navier-Stokes could be very useful because it is possible to handle variables such as temperature and pollutant distribution. 
In all cases, numerical techniques such as the finite element method, finite difference method, boundary element method, and computational fluid dynamics help process vast amounts of data. In addition, recent improvements in computing capability open a new horizon for computationally intensive modeling options. Currently, supercomputers are available for processing enormous amounts of information in a relatively short time. Besides, quantum computers are being developed. They will achieve instantaneous calculations that would take hundreds of millions of years to complete in a supercomputer. 
The future of corrosion and anti-corrosion engineering demands a shift towards a more scientific and less empirical approach. Predicting material and asset behavior necessitates leveraging the most advanced mathematical tools available. In this context, machine learning emerges as a game-changer, capable of avoiding financial losses, environmental pollution, fires, and structural failures. The mantra for corrosionists is clear: utilizing the best tools, such as machine learning, ensures the profitable business of protecting materials and equipment. The journey into the future of corrosion prediction is nothing short of captivating.

Special thank you to Carlos Arroyave for his contribution to this edition!  


This event is almost here - in less than a week, make sure to RSVP!

The Women in Science Week 2024 is an effort in that direction, enabling women, young girls, and minorities to interact with scientists and researchers who have already established themselves in their respective fields. This event is aimed to be organized as a virtual event, in collaboration with the Young EFC - European Federation of Corrosion, Coatings - Open Access Journal, and Alexander von Humboldt Foundation. Register here!

(This space could be you, your company or, your services - contact us for info on how to appear here!)

Listen to our podcast episodes on Apple Podcast, Spotify, or YouTube, and make sure to Follow US!


Next Podcast Episode - "the last one of 2023 and perfect to close the year"...


Industry and Leadership events on the lookout

We are the Knowledge Partners for MECOC 2024:

Did you know we are the MEDIA PARTNER for IPEIA 2024?

(This space could be you, your event - contact us for info on how to participate!)

Materials.Business Newsletter brought to you exclusively by Infinity Growth!


bottom of page