Quality assessment of material models for reinforced concrete flexural members

• Published in: Structural concrete : journal of the FIB Vol. 16; no. 1; pp. 125 - 136
• Main Authors: Jung, Bastian, Morgenthal, Guido, Xu, Dong, Schröter, Hendrik
• Format: Journal Article
• Language: English
• Published: Berlin WILEY-VCH Verlag 01.03.2015; WILEY‐VCH Verlag; Wiley Subscription Services, Inc
• Subjects: analysis and design methods; building materials / construction materials; Compressing; Concretes; Fracture mechanics; Materials selection; Mathematical models; model evaluation; model quality; model uncertainty; Nonlinearity; parameter uncertainty; Prognosis; Reinforced concrete; tension stiffening; Uncertainty
• ISSN: 1464-4177; 1751-7648
• DOI: 10.1002/suco.201300066

Non‐linear constitutive models for concrete in compression are frequently defined in design codes. The engineer generally uses either the linear (in SLS) or non‐linear (in ULS) compression model. However, a large variety of different approaches exists for describing the behaviour of the cracked concrete tension zone, and the selection of a corresponding model is usually based on qualitative engineering judgement. The aim of this paper is to assess the prediction quality of several concrete material models in order to provide a quantitative model selection. Therefore, uncertainty analysis is applied in order to investigate the model and parameter uncertainty in the bending stiffness prognosis for flexural members. The total uncertainty is converted into a prognosis model quality that allows a quantitative comparison between the material models considered. The consideration of the reinforced concrete in tension is based on the characterization of the tension stiffening effect, which describes the cracking in an average sense. In the interest of the practical applicability of the models considered, even for large structures, no discrete crack simulations based on fracture mechanics are considered. Finally, the assessment identifies that the prediction quality depends on the loading level and, furthermore, the quality across the models can be quantitatively similar as well as diverse.