Concrete is the most widely used construction material in the modern world, but plain concrete fails in a brittle manner with little warning. Research on the fracture of composites has led to the development of Strain-Hardening Cementitious Composites (SHCCs) that exhibit high tensile strain capacity and multiple cracking behaviors. At the ultimate state under tension, the strain of SHCCs can reach 38%, with the crack width typically self-controlled to less than 100??m before final failure. To ensure durability of concrete structures, design codes would limit crack widths to below 0.2-0.3mm. This criterion can be satisfied by strategically integrating SHCCs in modern structures such as using SHCC to make permanent formwork that acts like an external jacket of a concrete member to prevent water/chemical penetration.
To characterize the durability of a SHCC component, it is necessary to know the crack spacing and opening at a particular strain level. As the relation between the rate of water/chemical transport and crack opening is not linear, the distribution of crack opening, rather than the average value, also needs to be obtained. In the laboratory, the member can be investigated with hand-held microscope. To improve the contrast so cracks can be easily revealed, the member surface needs to be pre-treated by a white paint. This limits the application of these methodologies in real engineering structures. In this research work, we explore the feasibility of using modern computer vision techniques to develop a more general technique for characterization of cracks in SHCCs. To do so, experiments have to be designed to create and label a set of images with cracks formed under various test conditions (such as tension and bending tests). The images will be taken at various load levels and angles using different test equipment (including drones). Algorithms will then be developed to detect the cracking state. The level of complexity is slightly challenging.
Students will be involved in the preparation and testing of samples, collection of images as well as computer programming.
To expose students to mechanical testing as well as new approaches for damage detection.