Peak drift ratio estimation for RC moment frames using multifractal dimensions of surface crack patterns

Abstract

In this paper, a novel computer-vision based methodology is developed for predicting the seismic peak drift ratio of damaged reinforced concrete moment frames using surface crack patterns. A comprehensive database comprising 974 surface crack images from cyclic test results of 256 beam-column joint specimens at various drift ratio levels is collected. The database covers a broad range of concrete compressive strengths, rebar and stirrup strengths, longitudinal and transverse reinforcement ratios, beam and column length to depth ratios, in-plane configurations, and failure modes. Multifractal dimensions of damaged beam-column subassembly images are obtained by the box-counting algorithm to capture the complexity and irregularity of the surface crack patterns. The variation of multifractal dimensions with the peak drift ratio is investigated for specimens with strong and weak joints. Empirical equations are developed that predict the peak drift ratio during an earthquake according to the corresponding crack patterns. Four scenarios are considered based on the availability of the structural properties as the input data, and for each scenario, a unique predictive equation is obtained. Finally, the application of the proposed methodology is described for an example specimen at various drift ratios. Interestingly, the multifractal dimensions of the crack pattern alone as the input data lead to an acceptable accuracy level for the peak drift ratio estimation. The peak drift ratio predicted by the proposed equations can finally be used to identify the damage state and performance level of the reinforced concrete moment frames in accordance with relevant specifications.