PREDICTION OF THE BEHAVIOUR OF REINFORCED CONCRETE SECTIONS BY CALIBRATING WITH EXPERIMENTAL TESTS

Abstract

ABSTRACT: The development of numerical models that can calculate the yield and ultimate states of a member's section is essential to adequately know the seismic behavior of reinforced concrete structures. Unlike other applications, these models are directly involved in the definition of the hysteretic behavior of plastic hinges developed in the structure when subjected to seismic loading. These models must accurately reproduce the real behavior of structures with high computational efficiency. In this sense, using theoretical-empirical expressions that reduce the computational time might be suitable. The main objective of this research is the calibration and improvement of some existing expressions that are capable of modeling several parameters related to yield and ultimate states of reinforced concrete rectangular sections (yield moment, yield chord rotation and ultimate chord rotation). These expressions are calibrated with a set of tests selected from an existing database of more than 1000 tests: Only those elements complying with the constructional and seismic requirements imposed by the main codes are selected. Due to their robustness and numerical features, optimization techniques based on genetic algorithms are used in order to calibrate the expressions with the selection of tests. Results are compared with those obtained: (i) by previous authors, and (ii) using the expressions included in EC-8, showing an improvement in some statistical parameters. The calibration with structural elements complying with the codes requirements implies that these expressions are adequate for use in standard buildings. Keywords: section behaviour, reinforced concrete, genetic algorithms, calibration with tests