Comprehensive analysis of predictive efficiency of stress-strain models for concrete under uniaxial compression

Most studies on stress-strain relationships in compression have relied on limited experimental data, resulting in models that often lack general applicability. This study aimed to evaluate the relevance and accuracy of existing stress-strain models across a range of concrete grades. Graphical, analytical and statistical methods were used to compare the predicted and measured stress-strain responses. Whereby ten stress-strain curves for normal to high strength concrete, with peak compressive stresses ranging between 18 MPa to 121 MPa, were selected and applied to selected established stress-strain models from the literature. The graphical comparison assessed ability of each model to replicate both the ascending and descending branches of the stress-strain curve in a normalized format. To further assess predictive performance analytically, each selected model was evaluated based on normalized toughness and ductility index. Additionally, root mean square error and coefficient of variation were calculated to statistically validate the accuracy of the predictions. The conclusion of this study indicate that most models were found to inadequately capture both the ascending and descending branches of the stress-strain curve across all concrete grades. While the best-performing model achieved higher accuracy, involved multiple equations, making it complex to implement. These findings highlight the need for a novel model that incorporates easily measurable parameters, such as compressive strength and uses a single equation to predict both branches of the curve, ensuring broad applicability across concrete grades.