Abstract
Digitisation is making huge progress, and it is not stopping at automotive corrosion either. Within the entire automotive material life cycle, computer-aided approaches can already assist corrosion engineering and management today. From constructive corrosion protection on galvanically active hybrid constructions to the virtual design of active or passive corrosion protection systems, everything is possible. We are already very close to the goal of a continuous and realisable digital corrosion twin, but the complete integration into existing value chains is far from complete. This article provides an insight into current research and development and discusses the bottlenecks that still exist. The role of data or data collection and the smart combination of data- and physics-based modelling approaches are discussed. The possibilities and scope of applications of artificial intelligence methods for automotive corrosion topics are addressed. Concrete application scenarios are outlined by using examples, and the next work steps are derived.
Zusammenfassung
Die Digitalisierung macht riesige Fortschritte, und sie macht auch nicht vor der Korrosion am Auto halt. Innerhalb des gesamten automotiven Materiallebens können bereits heute computergestützte Ansätze das Korrosionsmanagement unterstützen. Vom konstruktiven Korrosionsschutz an galvanisch aktiven Hybridkonstruktionen bis hin zum virtuellen Design von aktiven bzw. passiven Korrosionsschutzsystemen ist alles grundsätzlich möglich. Dem Ziel eines durchgängigen und auch umsetzbaren digitalen Korrosionszwillings ist man bereits sehr nah, jedoch ist die vollständige Integration in bestehende Wertschöpfungsketten bei Weitem nicht vollzogen. Dieser Beitrag gibt einen Einblick in die aktuelle Forschung und Entwicklung und diskutiert weiterhin bestehende Hürden an. Die Rolle von Daten bzw. Datenerhebung und die geschickte Kombination von daten- und physikbasierten Modellierungsansätzen werden erörtert. Die Möglichkeiten und Anwendungsbereiche der Methoden der Künstlichen Intelligenz für automobile Korrosionsthemen werden addressiert. An Beispielen werden konkrete Anwendungsszenarien skizziert und die nächsten Arbeitsschritte abgeleitet.
Similar content being viewed by others
Explore related subjects
Find the latest articles, discoveries, and news in related topics.Avoid common mistakes on your manuscript.
1 Introduction
Computer-aided engineering is state of the art within the automotive value chain. However, the maturity level of respective contributions at the different construction and design elements varies a lot. Digital corrosion engineering is far away from digital mechanical or crash engineering. Nevertheless, the capabilities for reasonable and reliable digital corrosion engineering contributions for accelerated automotive construction and decision-making have been improved. Artificial intelligence (AI), respectively machine learning (ML) methods, as well as new modelling strategies together with the availability and accessibility of High Performance Computing (HPC) lead to an enhanced uptake of corrosion simulations within industrial stakeholder groups [13].
The main expectations are related to shorting the long duration of corrosion quality assessment and protection concept development times. Additionally, the amount of required testing might be reduced in the near future as accelerated corrosion testing protocols are currently being developed. New design concepts can be tested virtually including its integration ability into existing processing routes along existing quality specifications sheets. Economical and ecological advantages are intrinsically expected, as well as game changing innovation within today’s corrosion engineering.
Digital twins typically address a kind of smart data tracking from the raw material along the process and construction phase until materials (or cars) performance within the field and, at the end of the life cycle, their scrapping, which might be substituted by recycling in the future. For corrosion and corrosion protection, the implementation of these approaches requires tracking from data with respect to corrosion and corrosion protection until corrosion initiation within the field and its propagation, including the assessment of safety, warranty, and maintainability. Since data from the field is unique for each car and since the projection back onto minimisation of corrosion protection efforts is not suitable, automotive industry has defined a number of in-house testing and construction guidelines to manage the corrosion issue. In the future, such efforts might be reduced if the digital (corrosion) twins allow reliable predictions as well as tailored (reverse) engineering and innovation.
The initiation and propagation of corrosion and coating defects depends on multivariate data. Such data is not always available and makes e.g. reliable extrapolations with a pure data-based approach impossible. In cases like galvanic corrosion, crevice corrosion, edge retention, or coating delamination, physics-based corrosion simulations can enrich the data space and support decision-making for defined construction systems and corrosive exposure scenarios. In the following, methodical aspects will be discussed. Use-cases will give insights into the topic and act as a basis for the conclusions outlined in this contribution.
2 Digital (Corrosion) Twins
A digital representation of automotive corrosion and conceptual data of corrosion protection requires detailed knowledge along the value chain as shown in Fig. 1. The engineering and design phase interacts with the in-service phase and the other way around. As a consequence, the digital (corrosion) twin links data between both phases and enables knowledge based design improvements (incl. materials selection), early error detection within the processing chain, or early virtual testing at in-service conditions. Seen the other way around, it enables enhanced diagnoses and repair, anomaly detection, and a system optimisation [15], meaning car corrosion behaviour in-service vs. engineering and design parameters.
The obvious problem for such a digital approach is the lack of data [2]. Therefore it is very important to collect field data, to establish testing protocols at laboratory scale that approximate the field data at accurate level, or to have reliable models ready for validation that allow the computation of corrosion initiation and its propagation. Moreover, it has to have tools that capture the statistics as required by the entire quality assessment. These complex data interactions require a computational methodology as outlined in the following paragraphs.
3 Methodical Aspects
Novel data science approaches enable the management of very complex data dependencies that are very typical for corrosion and corrosion protection due to its multidisciplinary nature. By combining data science approaches with classical physics-based corrosion simulations, new and more robust capabilities of digital (corrosion) twins as part of existing value chains emerge. Finally, they enable corrosion simulations as examplarly illustrated in Fig. 2.
Principles of corrosion simulations including corrosion initiation and propagation stage for the in-service phase of components and its construction materials. Corrosion damage descriptors A are e.g. corrosion rate or crevice depth; Mechanism descriptors B are e.g. Cl ion concentrations/corrosivity. \(\mathrm{B}_{1,2}\) are local threshold values linked to a mechanistic change at respective components e.g. loss of corrosion protection
3.1 Machine Learning in the Context of Corrosion
Machine learning has become a hidden companion in our daily life, while, most of the time, we do not even notice the algorithms that make our life more convenient. Concerning transport applications and automotive in particular, machine learning approaches can be employed to facilitate an automated detection of traffic signs and other vehicles that are on the road during commuting. They can also be applied to detect and evaluate corrosion-induced damages in car parts and to accelerate the discovery of protective systems to prevent them (e.g. passivating surface treatments, coatings, corrosion inhibiting agents). The latter is a prime example for the application of quantitative structure-property (QSP) relationships that can accelerate the discovery of components for protective coatings by creating short-lists of promising compounds that mitigate corrosion. Moreover, process-structure-property (PSP) relationship concepts can be employed to optimize manufacturing processes to support the development of sustainable surface treatments (e.g. anodizing) As discussed in the following, sophisticated computer simulations can provide a deeper understanding of the underlying corrosion mechanisms and can be employed to evaluate approaches that are a prerequisite to inhibit them. However, the associated computational costs for highly parameterised numerical models are usually substantial, and data-driven approaches constitute efficient frameworks to develop surrogate models that can provide reliable estimates for the targeted property. Image recognition approaches can be used to detect and evaluate corrosion-induced damages in cars (e.g., rust) or to investigate the reliability of welded joints which can be subsequently used to provide estimate for the severity of the detected defects and when maintenance is required. In this context, sensor data that is collected throughout the service-life of a car to detect damages at an early stage cannot only be used to gain deeper insights into the corrosion behaviour of the materials in different conditions but can also be leveraged to realise predictive maintenance strategies (e.g. effect of environmental conditions and driving behaviour on the distance between service intervals). All of the aspects mentioned above may be linked to develop digital twins for corrosion aspects along the complete value chain of a car (from component selection to recycling) thereby supporting a sustainable circular economy.
3.2 Physics-based Corrosion Modelling
Corrosion simulations based on a mathematical modelling can be grouped into three classes [18]. Heuristic/semi-empirical models describe the propagation of a corrosion descriptor (e.g. weight loss, delamination depth, etc.) as a function of time. The functional typically consist of two or three fitting parameters that describe initiation and propagation phase [5]. The second class can be named numerical models. They are based on an equation set to be solved in time and space. The danger of such models belongs to the lack of the ability to capture localisation effects as well as the lack of adequate mathematical model and system data (e.g. local pH, conductivity, U‑I relationships, etc.). Thus, a more complex third class of modelling following a multi-scale simulation strategy has to be defined. Within a computational workflow, two or more mechanisms are captured. Their interaction might simply be a parameter passing but can even become a hard solver coupling problem.
4 Use Cases
To provide a clearer picture of the previously described theoretical approaches, multiple use cases are presented in the following section. For detailed insights into the topics and the techniques used, the reader is encouraged to explore the provided references. Topics related to hydrogen embrittlement, respectively environmental assisted cracking, are not addressed within this short overview.
4.1 Crevice Corrosion
Surface treatments via e‑coat technology is state of the art in automotive. Defects in the top-coat system can be the nuclei for severe crevice corrosion along the steel, zinc, atmosphere triple junction [1]. Thus, numerical modelling has been utilised to compute the corrosion progress in order to reproduce climate chamber test results.
4.2 Coating Application and Adhesion
Electronic or atomistic models can contribute to the better understanding of e.g. polymer based top coats and their delamination properties [3]. The application of coatings itself can be multifaceted, and computational studies can assist to optimise the processing itself [10]. In this context, machine learning based concepts can contribute to process control [9].
4.3 Coating Chemistry
Corrosion inhibitors are a mandatory component of any protective coating system used by aerospace, consumer goods, and automotive industries. Small organic molecules have shown great potential to control and mitigate corrosion reactions in lightweight materials based on Aluminium [16] and Magnesium [7]. However, finding novel inhibitors by experimental high-throughput techniques alone is intractable as small organic compound space is essentially infinite. Data-driven approaches may create great efficiencies by screening databases of commercially available compounds based on quantitative structure-property relationship modelling approaches and the subsequent generation of short-lists of promising corrosion inhibitors prior to testing [17].
4.4 Edge Retention
For each corrosion engineer, edges are a challenge. Short curvature radii at edges to be coated should be avoided. As a consequence, constructive and active corrosion protection concepts must be aligned. In the study of Waibel et al. [14], a physics [6] and data based approach have been combined to describe edge retention effects. The model allows to predict delamination width at various positions of a 3D body-in-white structure.
4.5 Field Data Utilisation
Each car has somehow a unique service-life and at its end a corrosive exposure history. For a more efficient corrosion protection conceptualisation, it would be very beneficial to consider real-life statistically approved field data. Combining sensors and data science within a hybrid approach might allow more efficient and reliable corrosion protection concepts in the future [8].
4.6 Fuel Induced Corrosion
Salt water and its aerosols are not the only critical electrolytes. Fuels, e.g. ethanol based liquids, can lead to severe pitting corrosion if the wrong materials are used [4]. Modelling can support the selection of most adequate material to be used for fuel pipes or the climate control system of a car.
4.7 Galvanic Corrosion
The assessment of galvanic couples by a computational simulation approach is feasible. Minimising corrosion currents by tailored washer design and geometrical joint layout has already been shown [11]. The effect of various electrolyte exposure scenarios can be studied as well [12]. Combining these studies enables most proper joint layout during the car design phase.
5 Conclusions
Digital (corrosion) twins and the role of corrosion simulations for automotive applications were introduced. Individual and partial computational models (pieces of the puzzle) exist for various corrosion or material interface problems and scales. An improved synergy between simulation and experiment (especially for mobility) is a major requirement for progress within digital automotive corrosion engineering. Currently, data science tools are successful in use (damage assessment, image recognition, inhibitor screening, etc.), although the synergistic potential of these approaches has not been fully unlocked yet. Digital concepts contribute to the acceleration of the development of corrosion protection concepts as for edge retention problems, galvanic couple design or for the handling of crevice corrosion.
Technology transfer still needs improvement as standards and certification constraints might not allow a fast uptake and the integration into existing value chains is not straightforward. The entire digital twin will cover surface (finishing) technologies, pretreatments, anodising, and accelerated testing. Smart monitoring and maintenance systems will be enabled. On the software side, workflow builders, uncertainty quantification tools and data managers which link to automation and robotics and testing data bases and repositories are under development. They will lower the barrier for utilisation of corrosion simulations as part of automotive engineering and business decision making.
References
Bösch, N.C., Höche, D., Mittelbach, A., Kainer, K.: Predictive modeling of long-time crevice evolution at e‑coat defects under climate chamber test conditions. Mater. Corros. 68(7), 699–710 (2017)
Coelho, L.B., Zhang, D., Van Ingelgem, Y., Steckelmacher, D., Nowé, A., Terryn, H.: Reviewing machine learning of corrosion prediction in a data-oriented perspective. npj Mater. Degrad. 6(1), 1–16 (2022)
Dagdag, O., Berisha, A., Mehmeti, V., Haldhar, R., Berdimurodov, E., Hamed, O., Jodeh, S., Lgaz, H., Sherif, E.S.M., Ebenso, E.E.: Epoxy coating as effective anti-corrosive polymeric material for aluminum alloys: Formulation, electrochemical and computational approaches. J. Mol. Liq. 346, 117886 (2022)
Gazenbiller, E., Arya, V., Reitz, R., Engler, T., Oechsner, M., Höche, D.: Statistical analysis of AA-1050 localized corrosion in anhydrous ethanol. Corros. Sci. 179, 109137 (2021)
Gießgen, T., Mittelbach, A., Höche, D., Zheludkevich, M., Kainer, K.U.: Enhanced predictive corrosion modeling with implicit corrosion products. Mater. Corros. 70(12), 2247–2255 (2019)
Kapfer, K., Mandel, M., Mittelbach, A., Krüger, L.: Investigation and modelling of edge corrosion of e‑coated galvanized steel with respect to the spatial edge orientation. Mater. Corros. 73(1), 55–67 (2022)
Lamaka, S.V., Vaghefinazari, B., Mei, D., Petrauskas, R.P., Höche, D., Zheludkevich, M.L.: Comprehensive screening of Mg corrosion inhibitors. Corros. Sci. 128, 224–240 (2017)
Mizuno, D., Suzuki, S., Fujita, S., Hara, N.: Corrosion monitoring and materials selection for automotive environments by using Atmospheric Corrosion Monitor (ACM) sensor. Corros. Sci. 83, 217–225 (2014)
Paturi, U.M.R., Reddy, N., Cheruku, S., Narala, S.K.R., Cho, K.K., Reddy, M.: Estimation of coating thickness in electrostatic spray deposition by machine learning and response surface methodology. Surf. Coat. Technol. 422, 127559 (2021)
Pendar, M.R., Páscoa, J.C., Lima, R.: Numerical investigation of automotive paint oven for improving the thermal efficiency. In: Fluids Engineering Division Summer Meeting, vol. 85840, p. V002T05A032. American Society of Mechanical Engineers, (2022)
Ruiz-Garcia, A., Mayen-Mondragon, R., Genesca, J., Morales-Murillo, E., Taha-Tijerina, J., Montoya, R.: Numerical simulation of the aluminum–zinc–steel galvanic system for new designs of automotive chassis. Mater. Corros. 71(3), 401–408 (2020)
Vega, J., García-Lecina, E., Genesca, J., Montoya, R.: The AA2024/CFRP galvanic couple under a dynamic electrolyte drop: modeling and experimental. Electrochim. Acta. 432, 141137 (2022)
Verma, K., Ayuso, L., Wille, R.: Parallel simulation of electrophoretic deposition for industrial automotive applications. In: 2018 International Conference on High Performance Computing & Simulation (HPCS), pp. 468–475. IEEE, (2018)
Waibel, L., Kapfer, K., Hepfner, C., Mittelbach, A., Funken, S., Krüger, L.: Edge delamination width prediction of 3D body-in-white part by finite element-based corrosion simulation and neural networks. Mater. Corros. 73(1), 68–81 (2022)
Waibel, L., Mittelbach, A., Funken, S.: Semantic segmentation of corrosive critical designs in body-in-white structures for corrosion simulation. Mater. Corros. 72(5), 788–804 (2021)
Winkler, D.A., Breedon, M., White, P., Hughes, A.E., Sapper, E.D., Cole, I.: Using high throughput experimental data and in silico models to discover alternatives to toxic chromate corrosion inhibitors. Corros. Sci. 106, 229–235 (2016)
Würger, T., Mei, D., Vaghefinazari, B., Winkler, D.A., Lamaka, S.V., Zheludkevich, M.L., Meißner, R.H., Feiler, C.: Exploring structure-property relationships in magnesium dissolution modulators. npj Mater. Degrad. 5(1), 2 (2021)
Zander, D., Höche, D., Deconinck, J., Hack, T.: Corrosion and its context in service life, chap. 2.10. In: Handbook of Software Solutions for ICME, pp. 227–245. John Wiley & Sons Ltd, Chichester (2016)
Acknowledgements
The authors would like to acknowledge the invitation by the ASMET team.
Funding
Open Access funding enabled and organized by Projekt DEAL.
Author information
Authors and Affiliations
Corresponding author
Additional information
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Höche, D., Feiler, C. Corrosion Simulations for Automotive Applications. Berg Huettenmaenn Monatsh 168, 3–7 (2023). https://doi.org/10.1007/s00501-022-01309-0
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s00501-022-01309-0