We attended the NAFEMS 2025 Simulation Engineering in the Automotive Industry, and as always, I view the NAFEMS conference as the “thermometer” of the CAE industry. Unlike other conferences, the NAFEMS conference stands out because it is more focused on “listening to the pain of the main players” and serving as a melting pot for companies that focus on solving industry problems. This allows for in-depth discussions about possible avenues, solutions, and alternatives. We also see many CAE players presenting new solutions and technologies to address the problems customers face. Instead of presenting ideas in isolated silos, NAFEMS conferences feel like everybody is examining the same issue from different perspectives. If you attend with the right mindset, you will gain a lot of insight! In addition to presenting our latest work on “A Computational strategy for corrosion mitigation and prevention in automotive design,” I also attended the keynote and took some notes that I would like to share in this brief blog.
We are immersed in an AI (Artificial Intelligence) trend, but most companies, even large corporations, still face data chaos. They have scattered information, silos, and wasted efforts, although fortunately, data intelligence is gaining attention, forcing engineers to think thoroughly about metadata and tagging. So, even if your solution is not AI-powered, you can still make a huge contribution by making your solution AI Data Ready. A major challenge is combining physics-based and machine-learning models. As VIAS3D noted, the key aspect is the integration of data from multiple sources without losing physical accuracy. Ram Bhandarkar from Stellantis delivered a phenomenal presentation, highlighting digitalization as a necessity. I certainly agree with him that we have moved from experimental to theoretical to computational. Currently, we are more in the data-driven science. He mentioned that for Stellantis, digitalization is not optional; it’s fundamental. Their goal for 2039 is to reduce testing by 70% and speed simulations by 40%. Certainly, this is only possible with a clear digital twin, where multiple disciplines integrate their solutions in an AI-data-ready fashion. Dan Woodman from Dow Chemical emphasized the importance of starting simulations early – before CAD is finalized – to influence design when it matters most. He also noted that not every single system needs a high-fidelity model; the goal is a fit-for-purpose simulation that drives real design decisions. I couldn’t agree more; we have been working extensively in this direction, creating various fidelity tools for our four-tier galvanic corrosion simulation. Certainly, we are going in the same direction. From the panel discussion, the main takeaway is that all panelists agreed that “Simulation now precedes prototyping”, and the fact that simulation is critical for meeting cost, safety, and timing goals. However, adoption still faces resistance – “a social and mental battle” against entrenched tools and workflows. It’s clear we are making a step forward in the right direction, working on computational workflows to identify galvanic corrosion risk early in the process, even when no CAD is available. We also leverage full 3D CFD Multiphysics simulation with stochastic models to increase predictive capability. All the while, we use the same source of truth – our Corrosion Database. It is fantastic to see that simulation is now the first step, rather than just a tool used to solve problems when it is already too late.
This year, we presented our four-tier solutions for the Automotive industry, and more importantly, our approach to help them reduce more than $ 20 billion/year in corrosion-related issues, just for the automotive industry alone. Just because it is a “natural” process doesn’t mean we are unable to manage and reduce it. In the results we presented at the conference, we utilized automotive-relevant material that resulted from our collaboration with the USAMP (Ford and GM). As pointed out by Dan Woodman, we don’t need high-fidelity models for every analysis; instead, accessing tools that are CAD agnostic will make a profound impact on the project timeline. Here is where our Electrochemical database Corrosion Djinn solution will be adequate. Corrosion Djinn needs only the materials from the system. This means that you can start resolving galvanic corrosion problems early in the engineering project, right at the conceptual stage. Once CADs are available, you can use the same electrochemical database, but now at the CAD level. You can interrogate hundreds or thousands of interfaces in minutes and get a list of the interfaces and parts sorted based on the galvanic corrosion rate (or risk). Figure 1 shows the Corrosion Risk Assessment tool, where our Electrochemical Database has been directly integrated into Siemens NX CAD software. Note that after a click, the tool shows with a traffic light the spots where the galvanic corrosion rate is high (red), medium(yellow), and low (green). This option enables you to resolve 80% of potential galvanic corrosion problems in minutes, while leveraging a single source of truth that can be integrated with your PLM solution. Corrdesa’s Corrosion Djinn allows you to manage your own database and update the material library from Siemens NX and Star-CCM+.
Returning to the fidelity discussion from Dan Woodman (again), we know that in some instances you need higher fidelity tools, and the reason is obvious: With Corrosion Djinn and Siemens NX, you are considering only the interface between two face pairs, while the area ratio, namely the anode-cathode ratio, is kept constant. In other words, you are not solving transport equations, are not taking into consideration the exact geometry, or allowing for an IR drop that will drastically affect gradients. All these aspects are important, but they are inherently complex because you need a CFD solver. However, the idea is to use the higher fidelity tool, only in those complex scenarios where multiple materials, wide aspect ratios, and applications with complex materials and coatings. In other words, for those scenarios where we know the lower fidelity models do not perform well. For this scenario, we have the Simcenter Star-CCM+, where Corrosion Djinn has now been incorporated to the material database and you can perform Galvanic Corrosion analysis with our validated electrochemical database.
Figure 2 illustrates the spatial distribution of the corrosion rate, utilizing the same electrochemical database, with the electrolyte thickness assumed to be 100 microns. Although the case we ran on Siemens NX was under a bulk assumption, we expect to see a higher corrosion rate under the thin film assumption. Nevertheless, the trend is the same for both workflows; the higher and lower corrosion rates are found in the same locations within the computational domain, with the difference being the magnitude of the corrosion rate. How different are they? We will see!
Figure 3 compares the solution between Siemens NX and Simcenter Star-CCM+. The table shown on the left-hand side of Figure 3 shows the output from Siemens NX. The maximum corrosion rate is predicted at 197,898 microns per year, followed by 59,847 microns/year under the bulk assumption. On the other hand, the CFD solution under the thin film assumption, where we expect a higher corrosion rate due to the thinner electrolyte and higher oxygen diffusion, is predicted at 4,230 microns/year, followed by 1,195 microns/year, in the same part and area as predicted by NX. The difference we see in the magnitude is the result of using the exact geometry, which drastically impacts the actual Anode-Cathode ratio.
In addition to this, we showed the new capability we are bringing to our CFD workflow, incorporating the ToW (Time of Wetness), as well as the Corrosivity Category, both of which are defined in the international standard ISO9223. The Time of Wetness accounts for the actual time the system was under a condition prone to corrosion; in other words, it represents the period during which the system was exposed to relative humidity and temperature that yield deliquescence and, consequently, corrosion. As you might imagine, this is unique for each city, state, and even varies by zip code. The Corrosivity Category, on the other hand, accounts for the “aggressiveness” of the environment. For example, something will corrode a lot faster in Panama City, FL, than in Newnan, GA. If two systems are exposed to the same ToW but in different locations, each of these two systems will corrode at a different rate due to environmental effects.
Figure 4 shows the weather spectrum for 95 days of exposure for two different locations, Warner Robins, GA, and Lancaster, CA. On the left side of Figure 4 we have the temperature from Lancaster (top), and Warner Robins (Bottom), while similarly, on the right we have the relative humidity. Anything that is above the orange line simultaneously on both plots (temperature and relative humidity) is considered wet, and it will be prone to corrosion. We clearly see that the temperature is high for both locations, but the system exposed in Warner Robins will be exposed to a longer Time of Wetness, since relative humidity crosses the orange line (relative humidity threshold) more times, and hence the mass loss will be higher. Although we need to obtain the weather spectrum from historical data, the corrosivity category is determined from maps, such as the one shown in Figure 5.
After incorporating these two new models and assumptions in our 3D CFD workflow, we are improving our prediction capability by considering weather data and local weather conditions. One important comment, this is the first attempt to account for corrosivity categories. We are exploring additional ideas to address temperature, Cl ion concentration, and other pollutants. These parameters influence the chemistry of the electrolyte, and hence the kinetics, resulting in different corrosion rates.
Figure 6 shows the spatial distribution of the penetration or material loss. Note that the units are microns, and more importantly, that both solutions have different spatial distributions (gradients), and their values differ drastically. Although this is what we expect, without incorporating the Corrosivity category, the spatial distribution would have looked the same in terms of gradients. That is because incorporating only the ToW results in different mass loss, due to the time integral of the corrosion rate over time, but the gradients remain the same. This was the first thing we noted if we did not include the Corrosivity Category, and that makes sense! It is clear that these two cars will not only have different mass loss, but the Electric Current Density field must differ for both scenarios. We can only accurately capture the latter if we can change the “kinetics,” although both have the same boundary conditions, namely, Polarization Curves.
Here, we demonstrated that you can use the same boundary conditions (Electrochemical database), which are fully compliant with MIL-STD-889D, that have been widely validated in different fidelity solutions, which we refer to as our 4-tier solutions. More importantly, you can indeed start resolving galvanic corrosion issues way before you have your first CAD ready, meaning you start a lot early in the engineering project, pushing solutions far to the left, which translates into a more proactive attitude. Our electrochemical database management platform, Corrosion Djinn®, enables you to manage your own database if you work with unique materials, while making it readily available in Siemens NX and Star-CCM+. We strongly believe that there is no silver bullet, and instead, we believe that a fit-for-purpose technique is not only more cost-effective, but it also puts the right tool in the hands of the right personnel, making CAE democratization a reality. There is something else I would like to highlight about our Stochastic Corrosion Modeling Technique, but I’ll leave that for a future blog. I think we have covered a lot in this one!
As always, Happy Computations and “Muchas Gracias”
Julio Mendez