Issue 50

S. R. Pereira et alii, Frattura ed Integrità Strutturale, 50 (2019) 242-250; DOI: 10.3221/IGF-ESIS.50.20 243 The ABNT NBR 9062:2017 [2] recalls the ABNT NBR 6118:2014 [3] standard, which deals with the design of concrete structures in general. It contemplates the verification of structural elements, precast or not, with any cross-section. In addition to usual design procedures, evaluation of serviceability and ultimate limit states by means of physical model tests, performing statistically based validation or the adoption of non-linear physical analysis (for which basic hypotheses to be followed are indicated) are also permitted. In the study of precast concrete columns, an effort of the scientific community is observed due to the great efficiency of the constructive method, reduction in the construction time and consequently in the cost of the building. However, the shapes of structural elements may be diverse in order to facilitate assembly between the elements. Because of this, the design elements cannot be covered by the conventional procedures presented in current reinforced concrete standards. Thus, it is necessary to evaluate experimentally or numerically the behaviours and resistance of columns with non-conventional cross- section. Faced with this motivation, Feng et al. (2019) [4] developed numerical simulations of the progressive collapse performance of precast reinforced concrete. Li et al. (2017) [5] numerically evaluated segmental columns under blast loads. Ghayeb et al. (2017) [6] experimentally evaluated a new type of precast concrete beam-to-column connections. Wibowo et al. (2009) [7] developed numerical models of precast reinforced concrete columns and the results were compared with field test results. Regarding numerical modelling of concrete structures, due to the complexities in the material’s behaviour, finding a suitable mathematical model that enables the prediction of structural behaviour is a challenge, requiring the use of simplifications. In the past, the development of stress-strain relationships for concrete was limited to formulations in the uniaxial and biaxial stress state. Increased use of large triaxially loaded structures, such as reactor pressure vessels, containments, damns, offshore platforms, marine tanks, etc. have led to the need for formulations of constitutive relations for concrete subjected to three-dimensional stress states. Several authors have been engaged in the search for a realistic physical model, which would enable simulation of the linear mechanical behaviour and the nonlinear behaviour caused by plasticity and cracking. Several studies on the applicable failure criteria in a triaxial stress state available in the literature present good correlation with experimental studies and enable prediction of ultimate limit state after the occurrence of cracking. Chen and Chen (1973, 1975) proposed constitutive relationships representing the nonlinear behaviour of concrete submitted to monotonic loading [8, 9]. Willam & Warnke (1975) suggested an elastoplastic failure criterion based on five control parameters [10]. Ottosen (1977) proposed a failure criterion based on four parameters, considering the application of short-term monotonic loads [11]. Hsieh et al. (1982) developed an elastoplastic failure criterion with fracture based on four parameters that allow the occurrence of large-scale flow [12]. Boswell and Chen (1985) proposed a failure criterion based on eight parameters with a strength surface for the case in which concrete is considered as linear elastic with fracture or a yield surface for concrete considered elastic and perfectly plastic [13]. Currently, users of the ANSYS commercial software [14] often utilize the Willam & Warnke (1975) [10] criterion because this model is the only one available in the material library for physically nonlinear analysis involving concrete. Studies addressing the behaviour of concrete structures have been based on this failure criterion. Barbosa and Ribeiro (1998) used different stress-strain relationships for concrete and obtained good agreement with experimental results by using this model [15]. Wolanski (2004) simulated the behaviour of reinforced and prestressed concrete slabs submitted to simple bending [16]. Queiroz et al. (2014) developed numerical models of steel and concrete composite beams with flexible connectors to evaluate the contribution of friction at the steel-concrete interface [17]. Another line involving finite element models in the ANSYS software is based on the development of new finite elements (solid and plane) using user programmable features, which allow the inclusion of new material models that are not available in the library. Lazzari et al . (2017) used these features to develop elastoplastic and viscoelastoplastic constitutive models, combined with the Ottosen (1997) failure criterion for prediction of the behaviour of reinforced and prestressed concrete [18] [19]. The present work presents a case study of a column with non-usual cross-section used in a Brazilian popular housing project of buildings with up to four floors. This project presents a system with prefabricated columns, beams and slabs that are developed in modules. For an adequate sealing of the wall sealing panels, the column presents recesses and protrusions in the section shape that provide tight connection regions along its height, resulting in an element with non-conventional shape. These columns were analysed in three different ways. First, an experimental program was developed, comprising four tests on real-scale columns, in order to verify their behaviour both in serviceability and in ultimate limit state. Secondly, a previously developed theoretical model was applied to verify the columns behaviour in serviceability. Finally, finite element numerical models of the tested columns were developed in the commercial code ANSYS to verify their behaviour.