Artificial neural network and RSM-based optimization of window openings in trapezoidal infilled RC frames
摘要
For many centuries, researchers have been working on easily adaptable frame structures, such as square and rectangular infilled frames, which are infilled frames with straight columns. Because high-rise structures are increasing frequently, regular frame structures are found to be insufficient to withstand lateral forces. Therefore, a novel geometry of frame structure, such as an infilled frame with inclined columns, known as a trapezoidal shape building, can withstand more lateral forces. Studies on infilled frames with inclined columns have also been limited. Using Finite Element Software (ABAQUS), this study examined an RC single-bay, one-story infilled frame with various interface materials used to connect the frame and infill. This analysis was further investigated by optimizing various infill opening percentages under static, monotonic, lateral loading. Different interface materials, such as lead, cork, and conventional interface material cement mortar, were used for comparison. In this study, we compared the elastic behavior of a 2D infilled frame with different openings that have been optimized and different interface characteristics. Different parameters, such as the bending moment, principal stresses in the infill, lateral stiffness, bending moment, shear force, axial force, and stresses in the infill, are specifically discussed in this paper. This study presents two major novelties: an investigation of infilled frames with inclined columns and the inclusion of soft materials as interfaces. The results showed that using soft materials as the interface reduced the stiffness of the frame compared with the conventional interface material. The regression results effectively illustrate the mean R2 values for different window opening sizes. When examining the R2 values, it was observed that for values up to 30%, they consistently exceeded 0.999. The performance of the intermediate size, which fell between 20 and 30%, was enhanced and aligned with a value of 0.999. The displacement of reinforced concrete infill frames was evaluated by comparing analytical research with an artificial neural network.