AbstractA prime reason for the restricted use of high-strength steel (HSS) rebar as the main reinforcement in concrete columns in seismic regions is the lack of appropriate seismic design guidelines. Within the context of performance-based seismic design of concrete bridge columns, it is critical to identify drift ratios corresponding to the initiation of different types of damage states. Such drift ratios, which are often referred to as drift ratio limit states, do not seem to exist for concrete columns reinforced with HSS. In this study, a comprehensive analytical program with an overarching objective of proposing simplified expressions to predict drift ratio limit states for circular concrete columns reinforced with HSS was carried out. Factorial analysis was initially performed to quantify the effects of geometry-, section-, and material-related parameters on the drift ratio limit states, and then identify the parameters with significant contributions. To generate a sufficient amount of data for the development of the drift ratio limit states expressions, the Monte Carlo sampling technique was adopted. Three samples each consisting of 1,000 unique columns, were generated for three types of HSS. The columns were analyzed under displacement-controlled quasi-static cyclic loading, and the drift ratios at the onset of the damage states were recorded. The resulting data were first used to establish the drift ratio limit states on a probabilistic basis. Machine learning-based symbolic regression was then employed to establish relationships between the variable parameters and the drift ratio limit states. To maintain simplicity, the proposed expressions contained only parameters deemed significant following the factorial analysis. The proposed simplified expressions for concrete columns reinforced with ASTM A706 Grade 550 had the highest square of correlation coefficients. The proposed simplified expression provided adequate predictions of the drift ratio limit states obtained from the numerical analysis as well as those measured in previous experimental programs.

Source link

Leave a Reply

Your email address will not be published.