Quasi-Static and Dynamic Crack Initiation and Growth in Soda-lime Glass: Full-field Optical Investigations
Type of DegreePhD Dissertation
Restriction TypeAuburn University Users
MetadataShow full item record
Soda-lime glass (SLG) is a widely used structural material with numerous advantages in terms of thermal, physical, and mechanical properties besides sustainability and recyclability. It is widely used for structural facades, in automobile windshields, as transparent military armor, and in medical packaging, as observation ports, in home appliances, etc. Being a highly brittle material with very low-toughness and high-stiffness, understanding its failure behavior in general and fracture mechanics in particular is extremely challenging yet critical for mechanical design and structural performance. In this context, this dissertation deals with characterization of quasi-static and dynamic crack initiation and growth behaviors of SLG using full-field optical methods. The first part of this dissertation dynamic fracture of SLG using full-field optical methods are assessed. Here the focus is on crack initiation and growth by overcoming spatio-temporal challenges posed by SLG as crack speeds typically exceed mile-a-second accompanied by nanoscale deformations. Yet, it is tacitly assumed that all optical methods are equally capable of dealing with this challenging problem though none are demonstrated to date. Towards this end, three prevalent optical techniques of experimental mechanics - transmission photoelasticity, 2-D Digital Image Correlation (DIC) and transmission Digital Gradient Sensing (DGS) - were concurrently implemented to visualize crack tip fields and quantify fracture characteristics during crack initiation and growth in SLG plates subjected to identical loading. Each method used ultrahigh-speed (1 Mfps) photography, flash/pulse visible light illumination, and a modified-Hopkinson pressure bar for impact loading. The feasibility of measuring fracture parameters along with the pros and cons of each approach for this material were critically examined. The photoelastic recordings allowed precise visualization and quantification of crack length history. The stress intensity factor histories, however, could not be evaluated due to insufficient stress-optic response of SLG. Hence, photoelasticity was found suitable only to test velocity-based fracture criteria; others based on stress intensity factors and/or energy release rate cannot not be tested. The 2-D DIC approach did not allow direct visualization of the crack tip at different time instants. The identification of the crack tip via dominant displacement fields did not fare any better and resulted in unrealistic instantaneous velocities and stress intensity factors. The DGS method, though relies on principles of 2-D DIC to quantify angular deflections of light rays proportional to stress gradients in the whole field, performed well on both these fronts. That is, visualization and quantification of crack length and crack velocity histories with high precision was possible. Thus, both velocity, stress intensity factor and energy release rate-based criteria could be readily tested. Based on the comparative investigation on SLG using different optical methods, dynamic mixed-mode fracture was experimentally investigated next using DGS, again in conjunction with ultrahigh-speed photography. Single edge-notched specimens subjected to inertial loading using a modified Hopkinson pressure bar apparatus were examined. A simple, eccentrically loaded specimen geometry was designed to generate a wide range of mode-mixities, from mode-I to nearly mode-II condition, at initiation. Two time-resolved orthogonal angular deflection fields of light rays proportional to the respective full-field stress gradients were measured during experiments. Mode-I and -II stress intensity factor histories spanning pre- and post-initiation behaviors were evaluated using measured full-field data in conjunction with the prevailing crack tip fields. By considering the critical stress intensity factors at crack initiation, a fracture envelope for SLG encompassing various mode-mixities was generated and compared with predictions from prevailing approaches for brittle solids. The measured crack kink angles were compared with two popular fracture criteria; both predicted the kink angles reasonably well. The effective critical stress intensity factors for SLG were found to be independent of mode-mixity over a large range of values but rapidly decrease as mode-II conditions become dominant; a critical mode-II stress intensity factor of 0.37 MPam, approx. one-half of its mode-I counterpart, was estimated. In the next part of the dissertation, unprovoked crack branching events during dynamic crack growth in SLG was assessed via direct measurement of fracture mechanics-based precursors. Towards this end, time-resolved stress gradients were measured in SLG plates of two different geometries, first one producing a single crack bifurcation event and the second one a two-tier cascading crack bifurcations in dynamic wedge-loaded specimens. The measurements were then used to extract precursors based on crack velocity, stress intensity factors, and higher order coefficients of an asymptotic crack tip field. A significant drop in the macroscale crack velocity was observed consistently when crack branching event is imminent. The cracks bifurcated when the effective stress intensity factor reached twice its value at initiation in both geometries. The fracture surface roughness and other features were also separately quantified via fractography to corroborate them with macroscale measurements. There was a lack correlation between surface roughness (Ra) and crack speed, predicted in the literature. However, a linear variation of log(Ra) with the effective stress intensity factor was seen. Hairline cracks in SLG occasionally heal when the structure is unloaded, become optically undetectable and cause catastrophic failure upon reloading during service. This was addressed in the final part of the dissertation by recreating crack initiation and slow crack growth from a self-healed crack in a SLG plate. A wedge-splitting test (WST) geometry was adopted for this study to generate a hairline crack and let it heal without external stimulus. An opto-mechanical study of crack (re-) initiation and growth was carried out by mapping the crack tip stress gradients in the whole field using DGS. An abrupt re-initiation and extension of the self-healed crack, subsequent initiation of the natural crack tip followed by a slow crack growth were all captured both in the far- and near-fields and the stress intensity factor history for the entire event was extracted. The results indicate a 50% reduction in critical stress intensity factor at re-initiation of the self-healed crack relative to the virgin counterpart. A companion finite element (FE) model was developed; the simulations suggest a contact stiffness of the self-healed crack to be about 60% of virgin SLG.