Capturing fracture initiation and crack propagation of car windshields

Abstract

To facilitate the development of accurate material models for glass, there is a need for reliable experimental procedures that capture the stochastic nature of the material. The experimental setup must be easy to model numerically and output relevant data for validation of important material characteristics. We conducted an experimental study on three types of windshields exposed to quasi-static loading. A robust and repeatable testing procedure was designed with the aim of revealing the stochastic behavior of windshields. 3D-Digital Image Correlation (3D-DIC) was used to monitor the windshield displacement field and we were able to capture fracture initiation locations and crack propagation data by using high-speed cameras combined with a pose estimation-based post-processing tool implemented in Python. We performed finite element (FE) simulations in LS-DYNA to estimate major principal stress values at first fracture. The results from the experimental study provided new insight and valuable data for the development of numerical models of the problem at hand.