Reliability and Model Accuracy for High-Strength Concrete Column Design

Abstract

Ultimate strength limit state provisions of the LRFD code ACI 318-95 are based largely on code-calibration considerations. Calibration exercises to date have been restricted to normal-strength concretes (NSC). The present paper calculates structural reliabilities of NSC and high-strength concrete (HSC) short columns sized to existing and proposed concrete rectangular stress block design models. Model errors for existing and proposed design models are presented also. It was found that the reliabilities of NSC and HSC short columns designed to ACI 318-95 are relatively consistent, and that the rectangular stress block design model proposed by the writers is the most accurate design model. Compared with all proposed design models, the reliabilities of NSC and HSC short columns designed to the writers' design model are the most consistent, though at a reliability slightly higher than that currently obtained for NSC columns designed to ACI 318-95. A strength reduction factor increase to 0.8 is proposed for the writer's design model so that existing safety indices would be met. The alternative is to use the existing ACI 318 rectangular stress block (which is a poorer predictor model) with the current strength reduction factor of 0.7. Both give consistent safety indices. The results can be subject to a number of interpretations, particularly since they relate to potential change to an existing code and the various implications that this may have. As such, issues relating to code development are discussed.