Probabilistic Beam–Column Joint Model for Seismic Analysis of Concrete Frames

Abstract

Beam–column joints in concrete buildings are critical for structural integrity under earthquake loading. Thus, accurate beam–column joint nonlinear models are important in seismic assessment and design of buildings. Most joint models in the literature are deterministic; the majority of which either are based on a small data set or do not distinguish failure modes and types of joints. This paper presents a data-driven probabilistic nonlinear model for concrete beam–column joints with transverse reinforcement. The model is based on multilinear regression using a newly developed database of 395 reinforced (confined) joints. Based on this database, current US seismic assessment provisions for beam–column joints were found to be conservative and have inconsistent levels of probability of exceedance for various nonlinear modeling parameters. The proposed probabilistic model unifies the level of probability of exceedance at 50%, a level that is consistent with the recently updated nonlinear beam and column provisions to avoid the bias in nonlinear assessment procedures. The proposed model exhibited good correlation with the test database, validation data sets, and cyclic backbones curves. Improved performance-based seismic design accuracy and economy for ductile concrete frames may result if current seismic assessment standards are updated based on the proposed model.