Damage Localization and Uncertainty Analysis of Weld Cracks in Portal Steel Frame Based on Vibration and Electromechanical Impedance Technique

Steel structures widely used in modern infrastructure development due to their fast assembly and easy of repair abilities, often rely on welded connections. However, these connections are prone to damage due to corrosion and dynamic loads, leading to progres- sive cracks and sudden structural failure. Hence, these connections should be continu- ously monitored for damage to ensure structural safety and sustainable use. Structural health monitoring methodologies using electromechanical impedance (EMI) have been proven to be useful and reliable techniques for evaluating the state of a structure at early stages of damage. These investigations uses strain gauge for acquiring vibration data, and PZT sensor for getting impedance data through electromechanical coupling. The test frame is a two-story asymmetrical steel portal frame with welded connections at the joints. The connections are subsequently damaged using a manual saw cut to sim- ulate the crack in the welded connection. At first, the strain data has been recorded for healthy and various damage conditions. Damage localization, along with uncertainty quantification, has been implemented using vibration time history obtained from strain sensors. Subsequently, the conductance signature for healthy and different damage states has been monitored. Different statistical damage parameters like modified root mean square deviation (mRMSD) and Average Canberra Distance (ACD), and Discrete wavelet transform based parameters like wavelet coefficient mean, skewness, and kur- tosis are calculated. These damage parameters are then used to formulate an objective function which is optimized using particle swarm optimization (PSO) for damage sever- ity estimation, considering the uncertainty in conductance signatures. The depth of weld damage has been estimated accurately using the proposed methodology, considering the effect of uncertainty. The devised methodology achieved more than 90% accuracy in the severity estimation of the saw cut damage and can be used as a low-cost continuous SHM solution for steel frame structures.