3D Peridynamic Models for Pitting Corrosion and Stress Corrosion Cracking

Abstract

Corrosion causes significant reduction in mechanical strength of metals via embrittlement, material loss, and surface roughening [1]. In the presence of mechanical loading, corrosion can lead to catastrophic failure of structures used in various industries: naval and aerospace, power plants, bio-implants, etc. Predictive models for localized corrosion and stress corrosion cracking (SCC) are of significant interest since they can assist us in material selection and design. Moreover, 3D predictive models can facilitate investigation of corrosion and stress corrosion cracking (SCC) at time scales and size scales that are out of reach of existing experimental techniques of investigation. In this study, we introduce 3D peridynamic (PD) models to simulate pitting corrosion and SCC under applied mechanical loading. Dissolution at the corrosion front, propagation of the corrosion damage in time, and transport of the dissolved ions are all captured in this nonlocal model within a single integro-differential equation defined on the bi-phase metal/electrolyte domain [2,3]. We include passivation and salt layer presence [4], to allow for simulating autonomous formation of lacy covers in pitting corrosion of stainless steel with realistic morphologies, as well as merger of pits, all with remarkable agreement with experimental observations. The PD corrosion model can be easily coupled with PD fracture models to predict SCC. For the first time, we simulate crack initiation and growth in 3D, in a case where two neighboring pits are growing. The fracture pattern and crack surfaces in the simulation are very similar to experimental 3D tomograms and SEM images. The left side of Figure 1 shows an SEM image and an internal tomogram of two pits that evolved and were joined by a crack, as observed in the experiments in reference [5]. The peridynamic simulations for this experiment are shown on the right side of Figure 1: the view from above the pits, after cracks start to form (shown on the top figure) and the inside view of the bottom of the pits and the crack connecting them. Compared with other attempts of modeling corrosion damage, peridynamic models have the following advantages: 1) Propagation of the corrosion front is autonomous and does not need to be explicitly tracked; 2) The model is general and easily modifiable for various types of corrosion (pitting [2,4], intergranular [3], etc.); 3) The partially damaged layer near corrosion front, known as the diffuse corrosion layer, with degraded mechanical properties is captured automatically in this model [1]; 4) The presence of mechanical damage in the PD formulation of corrosion allows for natural coupling to fracture models, thus accessing the exceptionally capabilities of peridynamic models in predicting fracture and failure.