Damage mechanics based model for low-rise infilled RC frames incorporating neural networks

Abstract

Purpose

Separately, nonlinear finite element analysis, artificial neural networks (ANNs) and continuous damage mechanics (CDM) attracted many investigators to model masonry infilled frames. The purpose of this paper is to pursue four phases to develop a versatile model for partially and fully low-rise infilled RC frames using these tools.

Design/methodology/approach

The first phase included the study of the behavior of 1,620 low-rise infilled reinforced concrete frames using macro-scale nonlinear pushover finite element analysis. The approach helped to explore the effects of imposing different masonry infill distributions for one of the typical models of school buildings in Egypt. The outputs of this phase were used in the second phase for the development of an ANN model where input neurons included number of stories, continuity conditions, frame geometry, infill distribution and properties of RC sections. The third phase included the employment of the notions of CDM on the structural scale. Monitoring frames’ stiffness degradation allowed for damage variables identification. In the fourth phase, the simpler equivalent static lateral load (ESLL) for elastic analysis was employed in conjunction with ANN and CDM to obtain the capacity curves for partially and fully low-rise infilled RC frames.

Findings

The obtained capacity curves were compared with the nonlinear finite element results. The close agreement of all curves indicated how rigorous, yet simple, the suggested solution procedure is.

Social implications

The study is concerned with an important type of service buildings. These are the school buildings of Egypt.

Originality/value

The paper presents a combination of four phases that include FE analysis, ANNs, ESLL, and CDM to obtain the capacity curves for partially and fully low-rise infilled RC frames. Such a combination of approaches in tackling a practical problem related to service buildings is innovative and deserves research interest.

Citation

Abou El-Ftooh, K., Atta, A., Seleemah, A.A. and Taher, S.E.-D.F. (2016), "Damage mechanics based model for low-rise infilled RC frames incorporating neural networks", Engineering Computations, Vol. 33 No. 4, pp. 1114-1140. https://doi.org/10.1108/EC-05-2015-0140