Masonry-infilled reinforced concrete (MI-RC) frames are widely used in building construction due to their enhanced stiffness and load-carrying capacity under lateral loads. However, their seismic performance is often governed by failure mechanisms such as diagonal cracking, crushing, and frame–infill separation. This study investigates the effectiveness of geogrid surface reinforcement in improving the strength of masonry prisms through experimental investigations. Further, the influence of geogrid surface reinforcement on the seismic response of MI-RC frames is studied numerically. Experimental investigations were conducted to determine the mechanical properties of constituent materials, including clay bricks, cement mortar, and brick masonry prisms with and without geogrid reinforcement. The results indicate that geogrid confinement enhances the compressive strength of masonry prisms by 16.67%, while also reducing crack propagation and improving stress distribution characteristics. The experimentally obtained nonlinear stress–strain properties were incorporated into numerical models developed in OpenSees, where masonry infill panels were represented using equivalent diagonal strut elements. In-plane nonlinear static pushover analysis was performed on three models: a bare RC frame, an RC frame with unreinforced masonry infill, and an RC frame with geogrid-reinforced masonry infill. The results demonstrate that the inclusion of masonry infill significantly increases the lateral load capacity and stiffness of RC frames, while geogrid reinforcement further enhances structural performance. Specifically, geogrid-reinforced infill improved lateral load capacity by 8.2%, initial stiffness by 59.2%, and ductility by 48% compared to unreinforced infill systems. This study highlights that geogrid surface reinforcement provides an efficient strengthening technique for masonry infill systems, leading to improved strength, stiffness, ductility, and energy dissipation characteristics, particularly in the early deformation phase. Moreover, the comparison of damage using the expended energy-based damage index confirms the benefit of using geogrid surface reinforcement in controlling the progression of damage during lateral deformation. The findings contribute to a better understanding of the seismic response of masonry-infilled RC frames and support the adoption of geosynthetic-based retrofitting strategies for enhanced structural safety.

This study investigates the prediction of fundamental time periods of reinforced concrete (RC) bare-frame buildings using machine learning (ML) techniques as an alternative to empirical height-based code formulas. A synthetic dataset consisting of 292 RC frame configurations was generated through parametric modelling using the ETABS API, varying key structural parameters including building height, base width, height-to-base ratio, mass, and stiffness. Seven supervised ML regression algorithms—Linear Regression, Support Vector Regression, Decision Tree Regressor, Random Forest Regressor, Adaptive Boosting, Gradient Boosting, and XGBoost—were trained using 80% of the dataset and validated with 10-fold cross-validation. Model performance was evaluated using R², RMSE, MAE, and the a20 index and results compared against empirical formulations from IS 1893:2016 and prominent literature-based models. Ensemble boosting models exhibited superior accuracy, with AdaBoost achieving the highest predictive performance (R² = 0.969, MAE = 0.050s, a20 = 0.90). SHapley Additive exPlanations (SHAP) analysis revealed mass and stiffness as the most influential parameters controlling the time period, unlike code formulations dependent solely on height. Comparative analysis demonstrated that empirical models significantly overestimate or underestimate period values with large scatter, whereas ML-based predictions closely match calculated values with minimal deviation. The findings confirm the capability of ML models, particularly AdaBoost, to improve reliability in seismic design and risk assessment of existing RC frames. Future extensions may include neural network architectures and irregular structural configurations.