Issue 50

S. R. Pereira et alii, Frattura ed Integrità Strutturale, 50 (2019) 242-250; DOI: 10.3221/IGF-ESIS.50.20 248 Figure 9 : Load vs. displacement curves obtained with the four different methodologies for model E4. It was possible to note that the numerical and theoretical models showed similar behaviour throughout the analyses, with small differences after the occurrence of cracking. It should be highlighted that in most cases, there is a formation of a plateau in the load-displacement curves both for the numerical and analytical assessments, corresponding to the beginning of cracking. This behaviour, although theoretically expected, was not observed in the experimental analysis. For the theoretical and numerical models of specimen E1, an initial overestimation of stiffness (up to about 110 kN) was observed until the beginning of cracking. From this point on, a sharp decrease in stiffness was observed. All models, except for model E3, which presented an excessive estimation of the stiffness, behaved similarly to the experimental model up to the serviceability limit state, estimated as the ultimate strength divided by 1.4. After this limit, a difference between predictions was observed. For the theoretical model, this distance can be caused by the inability of the formulation of analysing concrete with severe cracking, in ultimate-limit state. For the numerical models, after the serviceability limit, excessive loss of stiffness was observed due to the amplification of stresses caused by geometric nonlinearity, together with generalized cracking. Numerical models for tests E3 and E4 were not able to predict adequately the strength of the column. While model E3 led to an overestimation of strength, model E4 was unable to reach convergence due to excessive cracking. For the theoretical and numerical models of specimen E2, a coincidence in behaviours was observed up to the beginning of cracking. After this point, both models presented loss of stiffness higher than the one presented on the experimental model, with the behaviour described by an almost straight line. At this point, it was observed that a considerable part of the section was cracked, and only the reinforcements resisted the tensile stress in that portion of the column. Even with loss of stiffness, both models presented a behaviour close to that of the experimental model up to the serviceability limit state (70 kN). For this column, the strength obtained in the numerical model presented a difference of less than 5% of the one obtained experimentally. Figure 10 : Specimen E1 at the time of rupture in the experimental testing.