This study aims to model anisotropic damage (i.e. increase of porosity and loss of stiffness) and healing (i.e. recovery of stiffness) in salt rock subject to microcrack initiation, propagation, and rebonding. We introduce enriched fabric tensors in a Continuum Damage Mechanics model to link micro-crack evolution with macroscopic deformation rates. We carry out creep tests on granular salt assemblies to infer the form of fabric descriptors. We use moments of probability of fabric descriptors to find relationships between microstructural and phenomenological variables. Creep processes in salt include glide, cross-slip, diffusion, and dynamic recrystallization. We assume that healing is predominantly governed by diffusive mass transfer. We model the corresponding crack cusp propagation on grain faces by means of a two-dimensional diffusion equation. We calibrate this grain-scale healing model against experimental measures of crack cusp propagation distance. We simulate the opening, closure and rebonding of three orthogonal families of micro-cracks during a compression-tension loading cycle. Multi-scale model predictions illustrate the evolution of stiffness, deformation, and crack geometry during the anisotropic damage and healing process, and highlight the increased healing efficiency with time. We expect that the proposed modeling approach will provide more precise and reliable performance assessments on geological storage facilities in salt rock.


Because of its low permeability, water solubility and self-healing properties, salt rock is a favorable host rock for the geological storage of nuclear waste, oil and natural gas, compressed air (CAES), and hydrogen. Salt properties depend on the state of damage and recovery of the rock, which undergoes temperature and stress cycles around storage facilities. In order to understand and predict the microscopic mechanisms at the origin of damage and healing, we propose to model anisotropic thermo-mechanical damage and healing processes by means of alternative fabric descriptors.

Thermo-mechanical damage and healing models of salt rock proposed in continuum mechanics rely on the concept of dilatancy boundary [1]. Anisotropic healing models based on Continuum Damage Mechanics (CDM) usually resort to the concept of "net damage", which allows modeling stiffness degradation (damage) and recovery (healing). These theoretical frameworks are purely hypothetical and do not allow the prediction of all coupled processes that occur in actual rock materials. Rock damage models that distinguish closure and rebonding conveniently model all dissipation processes with rate-dependent evolution laws, which avoids implementing threshold-based yield functions in numerical codes. Unfortunately, those models do not properly represent the brittle behavior resulting from rate-independent crack opening and closure. In such models, healing is defined as a particular form of crack closure detected by an increment in wave velocity, rather than crack rebonding.

This content is only available via PDF.
You can access this article if you purchase or spend a download.