A simplified homogeneous approach for non-linear analysis of masonry infill panels under in-plane loads.

This paper aims to create a unified model that effectively combines continuous 2-dimensional elements and discrete components to capture the nonlinear characteristics and failure mechanisms of solid and perforated masonry infill panels. Given that masonry infill behavior is primarily influenced by shear deformations, an equivalent model is developed by using multiple small square panels arranged diagonally and interconnected by two-component springs, encompassing axial and shear behavior at their intersections. For the sake of simplicity, the divided panels are assumed to behave elastically, with plasticity concentrated only in the axial component of the connector springs. Plastic behavior in the boundary elements was considered to involve both flexural and shear plastic hinges to provide an accurate estimation of the entire infill panel's behavior. To validate this approach, the simplified model is benchmarked against eight experimental masonry infill panels surrounded by steel or reinforced concrete frames and with or without openings. The results including global behavior and crack pattern were compared with available numerical predictions based on finite element method from the literature in addition to experimental outcomes. Ultimately, this comparison demonstrated that the homogeneous model could effectively predict the non-linear lateral behavior of the panels and accurately forecast crack patterns. Additionally, the use of unidirectional non-linear springs and the appropriate arrangement of elastic panels significantly reduced both pre-processing and analysis time.

Predicting the Compressive Strength of Concrete Containing Binary Supplementary Cementitious Material Using Machine Learning Approach.

Several advantages of supplementary cementitious materials (SCMs) have led to widespread use in the concrete industry. Many various SCMs with different characteristics are used to produce sustainable concrete. Each of these materials has its specific properties and therefore plays a different role in enhancing the mechanical properties of concrete. Multiple and often conflicting demands of concrete properties can be addressed by using combinations of two or more SCMs. Thus, understanding the effect of each SCM, as well as their combination in concrete, may pave the way for further utilization. This study aims to develop a robust and time-saving method based on Machine Learning (ML) to predict the compressive strength of concrete containing binary SCMs at various ages. To do so, a database containing a mixture of design, physical, and chemical properties of pozzolan and age of specimens have been collected from literature. A total of 21 mix design containing binary mixes of fly ash, metakaolin, and zeolite were prepared and experimentally tests to fill the possible gap in the literature and to increase the efficiency and accuracy of the ML-based model. The accuracy of the proposed model was shown to be accurate and ML-based model is able to predict the compressive strength of concrete containing any arbitrary SCMs at ay ages precisely. By using the model, the optimum replacement level of any combination of SCMs, as well as the behavior of binary cementitious systems containing two different SCMs, can be determined.