Multiscale corrosion Modelling for the Oil & Gas Industry

The infrastructure used by the oil & gas and the petrochemical industries is often submitted to very corrosive environments, both because of the location of these infrastructures and because of the nature of the chemical processes that must be used. To limit costly and potential catastrophic failures, the industry is currently forced to select expensive stainless steels and other corrosion resistant alloys based on very conservative estimates integrated in industry-wide standards largely developed through empirical try-and-learn processes. Yet, as conditions evolve in specific environments, even these conservative choices do not guarantee long-lasting use and pitting corrosion often arise in nominally passive materials, forcing costly repairs and downtime and demonstrating the lack of fundamental understanding of these degrading processes. Corrosion is a typical multiscale phenomenon that has its origins at the atomic level while potentially affecting the global macroscopic properties of the infrastructure; this relation between length scales is complex. However, the time evolution of corrosion is a close balance between macroscopic and microscopic conditions, requiring, to be fully understood, models that couple scale in both directions. Today, understanding the static properties of small systems, typically a few hundred atoms or less, is well under control with current ab initio approaches that achieve remarkable precision and prediction regarding structural, electronic and optical properties. Ab initio methods, however, are typically unavailable, with current computer power, for systems larger than 400 or 500 atoms and for timescale longer than the nanosecond. To go beyond these limitations, it is necessary to turn to various levels of approximations for describing the atomic scale such as quantum mechanical/molecular mechanical (QM/MM) description, tight-binding approximation and empirical potentials. At the lightest computational level, we find empirical potentials that allow the study of millions of atoms on timescale that can reach, with molecular dynamics (MD), the microsecond, and with Kinetic Monte Carlo approaches, the second and more. These potentials, however, suffer from significant limitations for describing self-defects, surface and multi element environments. Roughly, they are almost systematically unable to provide coherent structural and kinetic information beyond the standard crystal and a few fitted environments. This means that, as we move to higher scales, using methods such as phase-field and finite-element approaches, parameters are more and more disconnected from fundamental knowledge as they rely instead on the capacity to reproduce ad-hoc sets of specific results, reducing the range of reliability of these methods. This missing link between fundamental unbiased physical properties and higher scales has significantly hampered our capability at understanding atomistic and larger scale phenomena across material science - from metallurgy to semiconducting industry. This limitation represents, today, the most significant hurdle to fundamental advances in computational materials science in general and corrosion in particular. Our project intends to lift this limitation by coupling a unique set of state of the art approaches developed and used by the PIs. This proposal aims at developing a computational multiscale framework to link fundamental atomistic processes responsible for corrosion to our current macroscopic-scale empirical knowledge with the long-time goal of optimizing the choice of materials for specific corrosive environments. With a very general structure, this framework will present a range of applicability that will be relevant to a much wider set of problems both in the oil & gas industry and beyond.

Hamad Bin Khalifa University (HBKU)