Structures located in high seismic zones often utilize reinforced concrete(RC)frame-wall systems for improved lateral strength and stiffness,whereby the structural walls serve as a critical component of the lateral lo...Structures located in high seismic zones often utilize reinforced concrete(RC)frame-wall systems for improved lateral strength and stiffness,whereby the structural walls serve as a critical component of the lateral load resisting system.To effectively assess the potential vulnerability of structural systems across different levels of seismic demands,it is important to establish clear,quantitative thresholds for specific damage states,especially for the critical structural components within a building system.The currently available damage state definitions for RC structural walls are based on empirical limits and do not provide predictions for damage thresholds based on key design characteristics of a wall.To address this challenge,the present study employs genetic programming(GP),a form of artificial intelligence,to formulate accurate expressions for drift prediction for various damage states,using a dataset of 8,125 analytically studied specimens of RC structural walls.These expressions take into account the effects of various design characteristics,such as wall aspect ratio,axial load ratio,boundary element longitudinal reinforcement ratio,web longitudinal reinforcement ratio,and ratio of boundary element length to wall length in determining deformation limits.The developed prediction models have been evaluated for accuracy and validity using various statistical measures.In addition,the proposed equations have been compared with other available deformation limits in relevant design standards and the available literature to predict experimental results of RC wall components.The findings of these analyses indicate that the developed expressions provide significantly higher accuracy and superior predictions compared to existing empirical damage state definitions.展开更多
文摘Structures located in high seismic zones often utilize reinforced concrete(RC)frame-wall systems for improved lateral strength and stiffness,whereby the structural walls serve as a critical component of the lateral load resisting system.To effectively assess the potential vulnerability of structural systems across different levels of seismic demands,it is important to establish clear,quantitative thresholds for specific damage states,especially for the critical structural components within a building system.The currently available damage state definitions for RC structural walls are based on empirical limits and do not provide predictions for damage thresholds based on key design characteristics of a wall.To address this challenge,the present study employs genetic programming(GP),a form of artificial intelligence,to formulate accurate expressions for drift prediction for various damage states,using a dataset of 8,125 analytically studied specimens of RC structural walls.These expressions take into account the effects of various design characteristics,such as wall aspect ratio,axial load ratio,boundary element longitudinal reinforcement ratio,web longitudinal reinforcement ratio,and ratio of boundary element length to wall length in determining deformation limits.The developed prediction models have been evaluated for accuracy and validity using various statistical measures.In addition,the proposed equations have been compared with other available deformation limits in relevant design standards and the available literature to predict experimental results of RC wall components.The findings of these analyses indicate that the developed expressions provide significantly higher accuracy and superior predictions compared to existing empirical damage state definitions.
