PSI - Issue 17

Hayder Al-Salih et al. / Procedia Structural Integrity 17 (2019) 682–689 Al-Salih/ Structural Integrity Procedia 00 (2019) 000 – 000

684

3

uses a series of images taken during loading and compares the images to obtain relative displacement and strain. DIC has shown potential for detecting and characterizing fatigue cracks, but testing has been limited to simplified test setups. DIC has been used in material testing to measure deformation and strain, and DIC can serve as an alternative to traditional sensing technology, such as strain gauges. Using DIC to evaluate cracking has been applied in the calculations of stress intensity factors (Zhang and He 2012) and to detect cracks in a concrete structure (Küntz et al. 2006). While research on vision-based crack detection methods have been conducted, testing has primarily occurred under idealized conditions looking only at in-plane fatigue loading or at cracks in non-metallic materials. Very few studies have been conducted evaluating vision-based crack detection methods on out-of-plane loading with the complex geometries found on steel highway bridges. Researchers have theorized that out-of-plane test setups have not been evaluated due to the inherent complexity and sophistication required (Sutton et al. 2007). Obtaining accurate results from DIC is dependent on the preparation of the specimen, the camera setup, calibration, and image collection. The preparation of the specimen involves applying a high-contrast random speckle pattern to the material surface, creating points of reference for image comparison. The necessary camera setup is dependent on the complexity of the specimen being tested. In-plane fatigue specimens experiencing no out-of-plane deformations require only two-dimensional analyses, necessitating the use of only a single camera. When testing for out-of-plane displacement, however, two cameras are required to calculate a three-dimensional displacement field. To move towards automated fatigue crack inspection, crack characterization must be quantified. Initial single-camera DIC testing was performed on a compact (C(T)) specimen subjected to in-plane loading on a servo-hydraulic testing machine. The C(T) specimen was 6.35 mm (0.25 in.) thick with a width of 127 mm (5.0 in.). The large specimen was capable of accommodating extensive crack growth, allowing for testing at a variety of crack lengths. As bridge loading is highly variable, multiple load cases were defined and applied based on stress intensity ranges of 11, 22, 33, 44, and 55 MPa √ m (10, 20, 30, 40, and 50 ksi √ in). Crack tip plasticity was limited during testing by applying the lowest stress intensity first and increasing to the highest. DIC data was recorded for each load case at crack lengths ranging from 12.7 to 50.8 mm (0.5 to 2.0 in.) in 12.7 mm (0.5 in.) increments. Although the general crack location could be identified in the visualized DIC results, the goal of automation prompted the development of algorithms to determine crack length from the DIC data. The twenty in-plane data sets were used to develop a method for this quantification. 2.2. Crack Characterization Methodology Although edge detection algorithms often produce false positive results when attempting to identify fatigue cracks on steel bridges, they can work well for images analyzed with DIC. Images of DIC-produced surface displacement contours were analyzed using commercially-available edge detection algorithms to initially identify the crack path. The coordinates of the crack path were extended linearly beyond the end of the crack tip as identified by edge detection. This linear extension of the crack path was necessary due to the inability of the edge detection process to accurately identify the crack tip. After identification of the crack path, differential surface displacements across the crack and perpendicular to the idealized crack path were calculated. This is schematically represented in Fig. 1a. Relative displacement at each point along the crack path, Δ i , was then divided by the maximum relative displacement, Δ max . A convergence value was defined by subtracting this ratio of relative displacements from 100%, calculated as in Eq. (1) and plotted in Fig. 1b. = 100% − ∆ ∆ (1) 2. DIC Crack Characterization Methodology 2.1. Initial Testing and Methodology Development