Crack network monitoring upon curing and drying of high-alumina MgO-containing refractory castable via digital image correlation

Abstract

MgO particles are added to high-alumina refractory castable formulations to form spinel at high temperatures. However, these particles react with water forming brucite, which causes heterogeneous expansions in the material that usually induces damage. Damage was studied by Impulse Excitation Techniques, which gives access to a global Young’s modulus. Digital Image Correlation (DIC) is a full-field technique that can provide field information about damage. In this dissertation, DIC was applied to cubic and bar-shaped specimens of the refractory castable mentioned, during curing and drying, in an in-house climatic chamber. This application was validated by tomographic volumes that showed cracks initiating on the surface and propagating to the center of the specimen. The images were analyzed using the software Correli-3.0, which implements a global approach (i.e., FE-based DIC). The Mechanical Regularization and the Brightness and Contrast Correction tools were applied to enhance the use of fine discretizations. The cracks were identified and quantified using maximum principal strain fields, which is the basis of other parameters defined in the dissertation (e.g., the Mean Crack Opening Displacement and the Surface Crack Density). The importance of a Representative Elementary Volume was highlighted by com- paring the results of cubic and bar-shaped specimens. Further, the Principal Component Analysis was applied to the displacement and the maximum eigen strain fields obtained via DIC. It revealed the crack network as the most relevant component with a temporal development of a sigmoidal curve where a two-parameter Weibull law was satisfactorily fitted. The approach allows the need for user-defined thresholds to be avoided for crack quantification Finally, an Adaptive Meshing (AM) procedure was implemented to locally refine the mesh on regions with cracks.