The effect of manufacturing flaws in the meso-structure of cast glass on the structural performance

Abstract

The success of projects such as the Crystal Houses façade in Amsterdam has triggered an increasing interest from architects, engineers and glass producers in the development and application of structural cast glass components. This interest raises, simultaneously, the needs for a controlled manufacturing process, a system for quality control and structural validation, to guarantee the production of safe components. Manufacturing-related flaws, such as stones, cord inclusions, or air-bubbles, occurring in the mesostructure of the components, form weak zones within the material and may lead to “spontaneous” cracking. The casting parameters such as the forming temperature and corresponding glass viscosity, the dwell time at this temperature and the cooling rate, largely determine the homogeneity of the final product. Additional complexity arises once the use of waste/recycled glass is considered, due to the probable presence of variable glass compositions and miscellaneous contaminants in the initial batch. The risk of inhomogeneity and resulting eventual mechanical failure, indicates the necessity of understanding the causes of flaw-formation and the impact of the developed flaws on the structural performance of the cast components. Therefore, a series of 50mm cubic glass components are cast at the TU Delft Glass Lab, using a selection of already formulated discarded soda-lime glasses from different commercial applications. The cubes’ meso-structure is documented and- when required- scanned employing a Computer Tomography scanner and a polariscope to identify possible density differentials and internal stresses respectively. Then the cubes are tested for splitting strength and their performance is analyzed in relation to the previously documented flaws. The destructive tests suggest that there is a correlation between the meso-structure, structural performance and failure pattern of the cast glass components.